Process System Innovation by Design

Towards a Sustainable Petrochemical Industry

Process System Innovation by Design

Towards a Sustainable Petrochemical Industry

Proefschrift

ter verkrijging van de graad van doctor

aan de Technische Universiteit Delft,
op gezag van de Rector Magnificus prof. dr .ir. J.T. Fokkema,
voorzitter van het College voor Promoties,
in het openbaar te verdedigen op maandag 13 september 2004 om 13:00 uur

door

Gerhard Pieter Jan DIJKEMA

ingenieur in de chemische technologie

geboren te Norg
Dit proefschrift is goedgekeurd door de promotoren:
Prof. dr. ir. M.P.C. Weijnen
Prof. ir. J. Grievink

Samenstelling promotiecommissie:

Rector Magnificus, voorzitter

Prof. ir. J. Grievink, Technische Universiteit Delft, promotor
Prof. dr. ir. M.P.C. Weijnen, Technische Universiteit Delft, promotor
Prof. dr. M.A. Reuter, Technische Universiteit Delft, Universiteit Stellenbosch, SA
Prof. dr. ir. A.A.H. Drinkenburg, Technische Universiteit Eindhoven
Prof. A.W. Westerberg PhD, Carnegie-Mellon University, US
Prof. D. Bogle PhD, University College, London, UK
dr. ir. S.A. Lemkowitz, Technische Universiteit Delft

Published and distributed by:

G.P.J. Dijkema
p/a Delft University of Technology
P.O. Box 5015
2601 GA Delft
The Netherlands
e-mail: [email protected]

Library of Congress Catalogue Data:

ISBN 90-5638-127-X

Keywords: fuel cells, functional modelling, innovation, olefins, petrochemical industry,

process systems engineering, sustainable development

Cover design: M. Mallee/P. Rüpp, www.kunstopmaat.info.

Copyright © 2004 G.P.J. Dijkema, Voorburg.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval
system, or transmitted, in any form or by any means, electronic, mechanical, photocopying,
recording, or otherwise, without the prior permission in writing of the proprietor.

Printed in the Netherlands

Preface
This dissertation is a synthesis of research that was completed in the better part of a
decade. The origins of some of the concepts and case studies date back to my stay
with Interduct. Most of the research presented was conducted in the
multidisciplinary, system-oriented academic environment of the faculty of
Technology, Policy and Management.
I would like to thank Margot Weijnen for offering me the opportunity to change
career-path, for her patience in awaiting the forthcoming of this thesis and for the
stimulating discussions and suggestions that set the stage for this work. I am most
grateful to Johan Grievink for his encouraging in-depth reviews and his suggestions
on formalisation of the work and its embedding in process systems engineering. To
me, a brew of green tea will for always be associated with a discussion in Johan's
office.
I would like to thank all my committee members for their positive feedback and
constructive criticisms. I owe special thanks to Art Westerberg for his painstaking
review of the content, my English and punctuation. Markus Reuter I thank for many
intriguing discussions and for his moral support.
Many people have helped me to shape and sharpen the concepts presented. This
thesis is a tribute to Han Jonk of SIPM, Odd Asbjørnsen of NTNU, Aad Montfoort
of TU Delft and Pieter-Jan Jongman of the Port of Rotterdam.
My colleagues at Interduct, the E&I section and the faculty I thank for creating a
stimulating, pleasant and supportive working environment. Rob Stikkelman involved
me in the project on fuel cell vehicles and in many creative sessions. Kas Hemmes I
thank for sharing his knowledge on fuel cells and his encouragement to complete
this work. In the office we shared, Paulien Herder put up with me and my
sometimes messy filing. I appreciated the in-depth discussions on life, research and
education with Zofia Verwater-Lukszo. I would like to hear Petra Heijnen sing more
often and to learn more Indian wisdom from Harish Goel. I expect to have more
discussions on electricity with Laurens de Vries. I am grateful for the swift help of
Ivo and Mirjam Bouwmans with the wording and translation of my propositions. I
acknowledge the indispensable support of Connie van Dop, Angelique Nauta and
Rachel Kievits of our secretariat.
My parents I thank for their love and for creating the conditions for my personal
development.
Finally, I treasure Elly's support and Roeland and Jenske's appreciation.

Gerard Dijkema
Voorburg, July 2004
Contents

PREFACE...................................................................................................................V

CONTENTS ...........................................................................................................VII

1 INTRODUCTION................................................................................................1
1.1 OVERVIEW: WHAT, WHY AND HOW?....................................................................1
1.2 READER'S GUIDE .................................................................................................1
1.3 A SYSTEM IMAGE OF THE PETROCHEMICAL INDUSTRY ..................................... 2
1.4 INDUSTRY DEVELOPMENT, INNOVATION AND SUSTAINABILITY ..................... 17
1.5 A NEED FOR PROCESS SYSTEM INNOVATION? ..................................................29
1.6 RESEARCH APPROACH AND STRUCTURE OF THESIS .........................................34

2 PROCESS SYSTEM INNOVATION SOURCES..............................................37

2.1 INTRODUCTION.................................................................................................37
2.2 SOURCES OF INNOVATIONS ..............................................................................37
2.3 SYSTEM THINKING FOUNDATIONS ...................................................................46
2.4 PROCESS SYSTEM ENGINEERING ...................................................................... 51
2.5 SYSTEM DECOMPOSITION .................................................................................59
2.6 CONCLUSIONS ...................................................................................................63

3 PROCESS SYSTEM INNOVATION BY DESIGN ..........................................65

3.1 INTRODUCTION.................................................................................................65
3.2 SYSTEMATICALLY TOWARDS INNOVATION CONTENT? ....................................66
3.3 FUNCTIONAL MODELLING FOR PROCESS SYSTEM INNOVATION .....................93
3.4 CONCLUSIONS ................................................................................................. 112

4 INNOVATION AROUND OLEFINS ............................................................ 115

4.1 INTRODUCTION............................................................................................... 115
4.2 THE INDUSTRIAL OLEFINS SYSTEM ................................................................ 117
4.3 SYSTEMATICALLY TOWARDS INNOVATIONS?.................................................. 123
4.4 FUNCTIONAL MODELLING OF THE OLEFINS SYSTEM .................................... 137
4.5 DISCUSSION AND CONCLUSIONS .................................................................... 151

5 FUEL CELLS AND TRIGENERATION ....................................................... 153

5.1 INTRODUCTION............................................................................................... 153
5.2 FUEL CELLS IN THE CHEMICAL INDUSTRY .................................................... 154
5.3 THE DEVELOPMENT OF TRIGENERATION SYSTEM CONCEPTS ................... 175
5.4 DISCUSSION AND CONCLUSION ...................................................................... 187
viii Process System Innovation by Design

6 FUEL CELL VEHICLES AND INDUSTRY DEVELOPMENT .................. 188

6.1 ABSTRACT ........................................................................................................ 188
6.2 INTRODUCTION .............................................................................................. 189
6.3 FUNCTIONAL MODELLING.............................................................................. 190
6.4 SCENARIO DEVELOPMENT AND MODELLING ................................................. 197
6.5 THE 'SUPERIOR PEMFC VEHICLES' SCENARIO.............................................203
6.6 PROCESS SYSTEM INNOVATIONS FOR INDUSTRY DEVELOPMENT.................. 211
6.7 DISCUSSION AND CONCLUSION ...................................................................... 218

7 CONCLUSIONS AND RECOMMENDATIONS ..........................................223

7.1 INTRODUCTION...............................................................................................223
7.2 PROCESS SYSTEM INNOVATIONS FOR SUSTAINABILITY ..................................223
7.3 (ENGINEERING) SCIENCE CONTRIBUTIONS ..................................................224
7.4 RESEARCH APPROACH.....................................................................................225
7.5 RECOMMENDATIONS ......................................................................................226

REFERENCES .......................................................................................................227

SYMBOLS ............................................................................................................... 241

ABBREVIATIONS .................................................................................................245

APPENDICES.........................................................................................................249

SUMMARY ..............................................................................................................279

SAMENVATTING..................................................................................................283

CURRICULUM VITAE..........................................................................................289
 1 Introduction

1.1 Overview: what, why and how?

What is this petrochemical industry that is the subject of this thesis? What are its
economic and ecological characteristics? To elucidate these questions, this
introductory chapter gives an overview of the petrochemical industry by presenting
it as a system in its surroundings. It is a layered networked system embedded in and
linked to the global material cycles. The system image is presented by addressing
system definition, boundary selection, system content and surroundings (§1.3).
Why do the historic development of the petrochemical industry, its current status
and resource horizon merit a reconsideration of its technologies, process systems
and petrochemical networks for sustainability? Why must we contemplate 'process
system innovations', which “… are defined as changes in the system structure
or system design of the petrochemical industry, its industrial complexes, or
individual plants (…) enabled by technological inventions or vice versa”
(Dijkema et al. 2003)? To address these questions, the meaning and implication of
sustainability for industry is discussed. The development of the chemical industry is
analysed with respect to drivers of and barriers to change, the process of change,
R&D and the scope for innovation (§1.4).
How can innovation for sustainability be fostered in the largely mature petrochemical
industry that consists of technologically advanced systems? The dilemma of lack of
economic incentive for much needed innovation for sustainability has given cause to
our central research theme - how to specify process system innovation1
content for a sustainable petrochemical industry (§1.5).
The meta-model of the research (§1.6) summarizes how the above questions have
been addressed and converted into the present thesis.

1.2 Reader's guide

The system image of the petrochemical industry (§1.3) is included to introduce the
object of the research and to serve as basis for those unfamiliar with this important
sector of the industrial economy. Readers with a chemistry or chemical engineering
background may find the system representation to be a useful refresher.
In 'industry development, innovation processes and sustainability' (§1.4) insights
from technology management, innovation theory and economics are combined and
linked to the petrochemical industry system image and scientific and chemical
engineering development.
'The need for process system innovation' (§1.5) is the core section of this chapter.
The dilemma's and problems associated with change and innovation for
sustainability are identified, which leads to the research theme, research questions
and hypothesis.

1 Where in process system engineering (PSE) the focus is on single plant process systems, the

definition of 'Process system innovation' is based on the notion of layered networked process
systems, which structure and content can be modified at each level.
2 Process System Innovation by Design

1.3 A system image of the petrochemical industry

1.3.1 SYSTEM CONTENT, INTERFACES AND SURROUNDINGS

Since the beginning of the 20th century, petrochemical facilities have been
continuously improved and major breakthroughs have been realized by the
invention and application of novel catalysts, reactor concepts, separation technology
and process system concepts. In many cases, the interplay of technological
development, economies-of-scale and market-growth made older, smaller and less
efficient plants obsolete. In many locations petrochemical complexes have
developed consisting of multiple interconnected and interdependent plants.
Today, petrochemicals are produced in bulk quantities in large-scale out-door
facilities that have become the blueprint image of the chemical industry in the
industrialised countries. Whilst these are complex installations in appearance, they
can all be represented as a system wherein a number of physical inputs is transformed
into a number of outputs, e.g. a steam cracker transforms naphtha to olefins and
aromatics (e.g. Rudd et al. 1973; Smith 1995).

Reserves and Renewables?

Earth Timely and economic exploration?
Carrying Capacity?

New Resources
Resource Improved Extraction
Novel Deposits

Technology Timely process innovation?

Systems More use of proven technology?

Products
Product New or improved
Services

Captive or end-user market?

Market Market growth or decline
Sustainability awareness?

RPMT-
Combination

Figure 1-1: The Pmt-R concept.

Any such system is part of a Product-Market-Technology combination (Pmt)
(Dijkema and Stikkelman 1999). The existence of any Pmt, however, depends on the
availability of resources (Figure 1-1). The steam cracker, for example, requires the
Introduction 3

availability of crude oil for the production of naphtha. Since system structure and
technology may change because of changing market demand and because of
changing resource availability, we prefer the label Pmt-R to emphasize the
relationship between product markets, technology and resource use for the system.
In the petrochemical industry, many Pmt-R's are linked, as it is a complex system of
interconnected processes and process routes. The entire petrochemical industry,
however, can be modelled as a black box that is part of and contributes to our
industrial society via the inputs it requires and the outputs it generates (Figure 1-2).
Inputs are required for day-to-day system operation, for the erection of new facilities
and for rejuvenation or end-of-life system abatement (Grievink 1994). Physical inputs
comprise feedstock and utilities, construction material and equipment. Other inputs
are market information, skilled labour, technological know-how, system design,
project organisation and management skills, investments and working capital. Physical
outputs include products, by-products, waste and emissions as well as industrial plants
and complexes. Other outputs are know-how and services rendered, salaries, fees, taxes
and dividends paid.

Non-Physical Surroundings
Operating Cost

Skilled Labour

Regulations
Investment

Know-how

Emissions
Feedstock

Products

Utilities
Waste
System: The Petrochemical Industry
(Dis)comfort
Experience

Interface
Revenue

Salaries

Surroundings

Non-Physical Surroundings

Figure 1-2: System image of the petrochemical industry.

Together, inputs and outputs represent the interface between the petrochemical
industry and its surroundings over the system boundary. Development of system
content is driven by market demand, feedstock availability, production cost, and
legislation. These are the non-physical -economic, societal, managerial, scientific,
regulatory- system 'surroundings' (e.g. Stobaugh 1988; Kuipers 1999; Dijkema and
Kuipers 2001). Feedstock cost, product revenue, capital flow, direct labour and
4 Process System Innovation by Design

services affect system economics (e.g. Peters and Timmerhaus 1991). Inputs from
and outputs to the physical surroundings result in adverse ecological effects, such as
resource depletion, toxic waste and harmful emissions. These have resulted in
societal concern, economic effects, environmental regulation, scientific interest,
technology development and shifting design criteria and changing design space (e.g.
Carson and Darling 1962; World Commission on Economic Development 1987;
Smith 1995; van Breda and Dijkema 1998; Wood 2001).
The industry's present ecologic impact is largely determined by its system content
and characteristics, which results from past decisions on product and technology
development, process system design, management, economics and regulation. In
competitive markets, the economic health of individual companies depends on
information and knowledge on how and when to properly bring novel products and
innovative production systems into being (e.g. Wei et al. 1979; Pisano 1997; Linse
2002; van Klaveren 2003). Company prosperity requires effective use of current
assets -technology, systems, skills and organisation- and strategic capabilities to
anticipate market dynamics, shifting societal concerns and changing regulation to
ensure sustainability. The system image of the petrochemical industry therefore
includes a summary of present developments in its surroundings.

1.3.2 A DEFINITION OF 'PETROCHEMICAL INDUSTRY' AND 'PETROCHEMICAL'

Obviously, the petrochemical industry is part of the chemical industry. The prefix
'petro' serves to further limit the system to those establishments that use petroleum-
derived feedstock. Typically, the petrochemical industry employs large-scale,
continuous process plants. Relatively small facilities that produce dyes, resins,
insecticides or pharmaceuticals from petroleum-derived feedstock are ‘chemical
industry’ but not petrochemical industry because of their limited scale of operation.
The large-scale manufacture of fertilizer and detergents respectively are chemical
industry because these industries predominantly use non-petroleum-derived
feedstock.
A common characteristic of ‘the (petro)chemical industry’ is that chemical reactions
are effectuated in a commercial fashion in specifically designed chemical reactors
that are the heart of any chemical process plant (see Figure 1-4, p.13). Using this
description, however, many industrial operations would be classified as
‘petrochemical industry’ that generally are not perceived as such. One can think of,
for example, steel mills and aluminium smelters that use petroleum coke. The
Standard Industrial Classification (SIC) by the US bureau of the Budget circumvents
this problem by defining SIC 28 as “the major group that includes establishments
producing basic chemicals, and establishments manufacturing products by
predominantly chemical processes (Wei et al. 1979: 25).” Thus, one has developed a
classification code that is both product-oriented and process-oriented. In the
professional community, the petrochemical industry is synonym for the large-scale
chemical industry that uses petroleum products as feedstock or materials that are
primarily obtained from petroleum products, e.g. olefins and aromatics from the
steam cracker and refinery operations.
Using the systems view presented in Figure 1-2, and using the above definitions we
define the petrochemical industry as 'the subsystem of industry where large-scale
Introduction 5

chemical processes are employed to convert petroleum-derived feedstock into base

chemicals'. A petrochemical then is a base chemical that is predominantly produced
by the petrochemical industry.

1.3.3 MARKETS AND SOCIETY

The outputs of the petrochemical industry are used in important industrial sectors
such as agriculture, building and construction, electric power, automobiles and fuels,
electronics, food and health care, textiles, consumer durables and packaging. (Wei et
al. 1979; Weissermel and Arpe 1994; Mol 1995). Thus, this industry plays an essential
part in the fulfilment of all categories of human needs in the hierarchy of Maslow:
food, clothing, shelter, transport, entertainment and the need for self-expression
(Maslow 1954)2.
Consumers purchase a myriad of industrial products and services from a variety of
industrial sectors to fulfil their needs (Figure 1-3). Hardly any consumer knowingly
buys petrochemicals, however, and to the general public this industry is mainly
known because of its perceived safety hazards, toxic waste and harmful emissions.
Its role as a major enabler and contributor to our current standard-of-living remains
largely unknown and unappreciated. Many of its products do reach consumers,
however, via various industries that incorporate synthetic polymers in their products
(Figure 1-3). Generally known as plastics, these materials and their building blocks
have enabled successful introduction of a myriad of innovative products. Presently,
polymerisation is used to produce thermoplastics, thermoset resins, resins and
coatings, rubbers, foams and industrial fibres. All these polymer categories comprise
chemically stable products. With the exception of some special resins for adhesive
and coatings applications, the products sold are characterised by non-existent or very
low chemical reactivity.
Since the World War II the demand for synthetic polymeric materials has grown
dramatically. As a consequence, today the majority of products of the petrochemical
industry are polymer building blocks. Ethylene and propylene, for example, are used
to manufacture polyethylene and polypropylene for a variety of packaging
applications, toys, synthetic carpets, cutlery and automobile parts. Copolymers of
styrene, acrylonitrile and butadiene are produced in high-grade semi-bulk for the
electronics and automobile industry. These industries also use a myriad of specialty
polymers or engineering plastics that are produced in small volumes, for example for
mobile phone exteriors, switchboards and car parts. Apart from polymers,
petrochemicals are used as precursors for fine chemicals and pharmaceuticals,
electronics, fertilizer as well as automotive fuels and de-icers. Thus, the
petrochemical industry has become an essential supplier of our modern industrial
society. It can be labelled the industry’s industry (Brennan 1998).

1.3.4 PRODUCTS
The chemical industry produces intermediates that are used in many industrial
sectors to cater for consumer demand (Figure 1-3). Most chemical products are not
sold to consumers, however, and the product spectrum of the industry remains
relatively unknown to the general public.

2 Cited in (van der Graaf 1980)

6 Process System Innovation by Design

The products of the chemical industry can be categorised as undifferentiated vs.

differentiated (Wei et al. 1979; Stobaugh 1988; Brennan 1998:14-16) and as large,-
volume versus small and very small volumes produced (Table 1.1).
Differentiated products are also known as performance products. These possess at
least one unique characteristic. The large-volume segment mainly consists of bulk -
polymers (e.g. PE, PP, PS, PVC). These pseudo-commodity plastics are produced in
many different grades for specific applications. The small-volume segment by
definition includes thousands of specialty chemical products, resins and specialty
plastics and active ingredients of crop protection chemicals, veterinary medicines
and pharmaceuticals.
Table 1.1: Classification of petrochemical industry products (after Brennan 1998)
Product classification
Production volume Undifferentiated Differentiated
Large Commodity Pseudo-Commodity
Small Fine Chemical Specialty Chemical
Very Small Pharmaceutical

An undifferentiated product has a specific chemical formula, and a standard

particular specification, regardless of its producer. The commodities category
includes products that are traded internationally under stringent AA-quality
specifications such as ammonia, sulphuric acid, ethylene, styrene and standard grade
plastics such as HDPE, LLDPE and PP. 'Fine chemicals' includes agrochemicals,
food additives and 'commoditized' medicines such as aspirin. Most petrochemicals
are undifferentiated commodity products. Worldwide, annual production capacity
for some 60-100 petrochemicals exceeds a million metric tons each (Appendix A. 1).

1.3.5 FEEDSTOCK AND ENERGY USE

The carbon-hydrogen skeleton of petrochemicals originates from crude oil. Oxygen
from the air is incorporated in petrochemicals via partial oxidation. The chlorine in
PVC is produced electrochemically from common rock salt (NaCl).
Once, coal used to be the chemical industry's most important feedstock. Presently,
however, coal use is largely restricted to electric power generation and coke
production for steel works. In coke ovens limited quantities of aromatics are
produced from the coal tar by-product (Franck and Stadelhofer 1988). The
exception is Sasol in South Africa, where it is still economic to convert coal to
logistic fuels (gasoline, diesel, fuel oil) via partial oxidation to syngas and subsequent
Fischer-Tropsch synthesis.
The primary carbon and hydrogen sources for the petrochemical industry today
comprise oil products, natural gas liquids and natural gas. Since crude oil cannot be
converted directly into petrochemicals economically, the industry's feedstock is
obtained primarily from oil refineries. These separate crude oil into a range of so-
called petroleum ‘cuts’, of which naphtha and gas oil are suitable and economic
feedstock for steam crackers that produce ethylene and propylene, other olefins and
aromatics (Figure 1-3 and Ch. 4).
 Refining Bitumen
Fuels Transport /
Industry Lubricants
Energy
Introduction

Production
Pharma
Oil Network
Winning
Refineries Fine Chemicals Food

Textile
Olefins Polym. Moulding Paper
Consumers /
Packaging Households
Steam- Polym. Shaping
Gasoil
Cracker Electronics
Production Monomers
Complexe network Automobiles
Naphtha
s
Building
Aromatics Polym. Process
Materials
Machinery
Petrochemical industry Other

Production network Polym. Process

Industry
Natural
Methanol & Derivatives Polymer Industry Sectors
Winning Gas

Production network Farming

Fertiliser Industry
Agriculture
Chemical Industry

Figure 1-3: 'Chemical industry', 'petrochemical industry', 'polymer Industry' and society.
7
8 Process System Innovation by Design

Many oil fields co-produce associated gas or 'condensate' -ethane, propane and
butanes-. Natural gas wells often co-produce natural gas liquids (NGL) -butanes to
heptanes-. Both condensate and NGL are premium feedstock for steam crackers.
The butanes are the preferred feedstock for MTBE.
Natural gas3 is important as a carbon and hydrogen source for industry, as it is the
dominant feedstock for methanol via the production of synthesis gas (Figure 1-3).
In the Netherlands the large-scale process industry - petroleum refining,
petrochemical industry, polymer industry, fertilizer production and base metal
production - together account for over 30% of the end-use of fossil energy resources
(Dijkema et al. 1995). Combined, these sectors represent no less than 67% of annual
Dutch industrial energy end-use, which includes feedstock use and net energy use
for production. Worldwide some 10-15% of annual fossil resource use is consumed
by these industries. These data indicate that the Dutch economy is characterised by a
relatively large process industry sector, which accounts for some 1020 PJ4 end-use
on a total Dutch energy account of some 2700 PJ in 1990 (3100 PJ in 2001) (CBS
1992; 2002)56.
With the exception of natural gas, some offshore oil production and a limited share
of renewables, the Netherlands is completely dependent on imported crude oil and
coal for its internal energy use and the associated industrial production.

1.3.6 UNWANTED BY-PRODUCTS: WASTE AND EMISSIONS

Apart from desirable outputs, petrochemical operations also result in by-products,
emissions and waste that can be detrimental to our environment. Smokestacks of
industrial (boiler) furnaces have been the trademark of industry since the early days
of the industrial revolution. It was after World War II, however, that smokestacks
taller than 100 metres where erected to combat local pollution. In the 1960-70's, a
period of rapid economic growth, the petrochemical industry obtained the label
‘Dirty, Damaging, and Dangerous’. This was largely due to the rapid expansion of
petrochemical production, associated increasing waste volumes and Rachel Carson's
'wake-up call' that even very small quantities of chemicals can have a devastating
effect on the environment (Carson and Darling 1962).

3 Hydrogen and nitrogen are converted to ammonia (NH3) in the Haber-Bosch process (Anonymous
1991; Jennings 1991). This is the industrial route for nitrogen fixation from air, and the basis of the
world nitrogen fertiliser industry. The primary source for hydrogen is natural gas (CH4).
4 Pèta Joule = 1015 Joule; One PJ is the energy content of 24 Thousand Tons Oil Equivalent (TOE).

Thus the net end-use of fossil resources in the Dutch process industry equates to some 24 Million
TOE; Total Dutch fossil resource use equates to some 74 Million TOE.
5 Since for many products of the process industry, fewer than three companies operate facilities in the

Netherlands, the Dutch Statistics Office CBS cannot disclose a detailed break-down between
feedstock, energy-use and production. The industry associations (VNCI, FediChem, FOI) do report
total sector energy use, but these figures exclude feedstock consumption. Based on the case studies
done in 'de grondstoffenstudie' (Dijkema et al. 1994a; 1995) we estimate the total amount of fossil
resources that does not end up in petrochemical products in the Dutch petrochemical industry at
350-400PJ.
6 Other use of fossil fuels in the Netherlands concerns energy conversion for electric power

generation (some 480-500PJ for 300-320 PJe), mobility (380-400PJ), space-heating (400-420PJ) and
miscellaneous industry and services (600 PJ).
Introduction 9

The environmental catastrophes thus exposed and a number of serious incidents in

the 1970's gave cause to general public concern on safety, health and environmental
risks of chemicals and the chemical industry. At the time, for example, untreated
effluents polluted watersheds and toxic waste was dumped in uncontrolled,
unprotected landfills. Without change, the industry was at peril of loosing its 'license-
to-operate' in the 1970-1980's.
The pre-1960 environmental ‘disasters’ exposed the chemical industry. Not only did
these result in growing public concern during the 60's, 70's and 80's, also
governments responded by environmental policy development that led to extensive
and more stringent regulation on environmental pollution (e.g. VROM 1972). The
chemical industry focused on improving its environmental performance with respect
to harmful emissions to air, water and soil. Initially, this focus on environmental
pollution caused by the industries’ inadvertent outputs led to better operating
procedures and so-called end-of-pipe solutions such as flue-gas cleaning and
chemical waste incineration. Later improved designs for novel and existing plants
emerged that include process-integrated solutions for emission-control and waste-
abatement. In addition, a framework for environmental management was developed
and adopted.
With respect to environmental management the chemical industry appears to
positively distinguish itself from a number of other sectors. In a recent survey the
European Environmental Agency concluded that the status of our environment is
not improving anymore. Amongst others, the agency attributes this to the many
sectors of the European Economy that to date have only implemented end-of-pipe
solutions (Anonymous 2003b). In the chemical industry, nowadays environmental
management is usually amongst the responsibility of senior management (Mol 1995),
and for many a company in the petrochemical industry “the environment” has
become a strategic issue. Major refining and chemical corporations have included
Responsible Care for the environment in their mission statement. As a result, the
chemical industry has ensured its continued license-to-operate (Mayer and Dijkema
1999).
In the petrochemical industry waste quantities per unit product are generally low, as any
significant amount of hydrocarbons can be (re)utilised as feedstock or converted
into power or heat. Weber (1909) distinguished between “Verlustrohstoffe” and
“Reinrohstoffe” (Weber 1909). The former type of resource carries a high amount
of useless material. Upon processing, inevitably a significant amount of waste
material must be discarded. By contrast, a Reinrohstoff constitutes primarily usable
components.
Crude oil changed from a Verlustrohstoff to a Reinrohstoff as a consequence of the
late 19th century developments in crude oil refining, transportation and the chemical
industry. In accordance with Weberian location theory, this not only affected the
location pattern of refineries worldwide, but also implied that the absolute amount
of waste generated in the refining and petrochemical industry is limited.
In many industrial regions where petrochemical industries are concentrated, these
have jointly realised facilities for final treatment of hazardous waste, for example
AVR in Rotterdam, Indaver in Antwerp. In the refining industry, however, waste
potentially may become a problem when heavy residues and asphalt cannot be fully
10 Process System Innovation by Design

marketed anymore for heavy transportation fuel (ships) or road construction. In the
future, it can be expected that additional conversion technology is required to
further upgrade these streams, and capture sulphur, heavy metals, and ashes in an
environmentally acceptable way. In other large-scale processes, notably base metal,
low-value or waste by-product generation remains the rule, as well as in coal
processing.
Today, in response to global treaties some classes of harmful products, such as
chlorinated hydrocarbons and CFC's, have largely been phased out. Inadvertent
emission losses of volatile organics (VOC) have been reduced. Design and
performance of fired-equipment has been improved to reduce VOC, NOx, CO and
soot emissions. Thus, in the petrochemical industry, the environmental “sanitation”
of the production systems is considered to be on track, and harmful emissions are
continuously monitored, whilst process performance is continuously improved.
As in waste management, three decades of development and corporate experience
with environmental management systems should enable the industry to respond
appropriately to whatever environmental demand or regulation on harmful
emissions. The reduction of global human CO2 emissions, however, presents a
relatively new and formidable challenge. Production and combustion of oil products,
natural gas, petrochemicals or polymers inevitably yields CO2. Thus its formation is
linked to the industry's feedstock and energy supply and to the use and final
degradation of its products. Hence, allowed CO2 emission can become an active
constraint, and CO2 emission control a key to the industry's license to grow.

1.3.7 ECONOMICS
The traditional and major chemical economies are the US, the EU and Japan. In
general, OECD countries “have rather similar chemical industries with only minor
variations between these industrialized nations” (Mol 1995: 127). Today, growth
regions are the Middle East, South-East Asia and to a lesser extent Eastern Europe.
Around the globe, the petrochemical industry is characterised by
- the use of large-scale to very large-scale continuously operated facilities
- large sunk costs - a huge amount of capital invested
- links with both cyclical and largely non-cyclical markets
- oligopolistic market structure
- a high level of internalisation, both in trade and production sites.
- a dependence on abundant, low-priced resources
- being capital intensive, labour extensive and knowledge intensive
- a relatively low return on investment (ROI) or return on asset capital employed
(ROACE) (Linse 2002)
(e.g. Wei et al. 1979; Stobaugh 1988; Brennan 1998; Whitehead 2000).
The volumes of many undifferentiated commodities merit continuously operated
facilities over batch wise operation because of achievable process efficiencies, asset
utilisation and investment per unit product. Investment for new continuous chemical
plants generally is given by I = I0 * (C/C0)p, where I is total plant Investment, C is
plant capacity, and suffix 0 denotes some known investment I0 at a known capacity
Introduction 11

C0. Since p has a value of around 0.6, a chemical plant of twice a nominal capacity
will cost only 50% more (Chauvel 1981). Generally it will require the same number
of operating personnel, and have similar overhead costs.
Whilst in (pseudo)-commodities, investments are huge (100-1500 million US$ per
facility), the profit margin or value added per unit product over feedstock cost is
relatively low (e.g. Wei et al. 1979). This margin, however, must be sufficient to allow
recovery of all operating costs and the investment-related costs, i.e. average costs.
Since petrochemical plants often exhibit a long technical lifespan -30-50 years-
would-be competitors in an established petrochemical market do face ‘marginal
competition’. Established manufacturers ('incumbents') compete on the basis of the
marginal cost of their facilities only because the original investment is a 'sunk cost'
that either has been recovered or written off in previous operating years.
Investments in new facilities, however, are justified only when sufficient return over
total cost can be realised. As a consequence, incumbents may operate economically
at efficiencies considerably below state-of-the-art.
The gap between marginal revenue and average revenue often can only be closed by
combining innovative process technology and economies-of-scale. Existing facilities
are expanded and modernized or phased out; in newly built plants state-of-the-art
technology is used and capacity is at the technically feasible design limit. Thus,
knowledge intensity, scale-of-operation and investments continuously increase to
bring down cost per kilogram of product.
The typically low margins of petrochemical operations must be sufficient to realise
return-on-investment targets. These target levels are relatively low in the
petrochemical industry - around 10% -. The financial risk, however, is substantial
due to the cyclical nature of markets, the risk of overcapacity and the technically and
economically limited flexibility of petrochemical plants in capacity-utilisation
(Kuipers 1999).
A typical petrochemical plant has a physical turndown ratio of some 50%, whilst the
average economic turndown ratio of a new plant may be as small as only 10%. Thus,
usually utilisation-factors at or below 80% indicate that the owner is not recovering
average cost. In the early 1990's, for example, many companies invested in plants
for PET - the plastic used for sodas sport drinks bottles, 'single service disposable
containers' - and its precursor, purified terephtalic acid because of expected high
market-growth. A few years later, however, the resulting capacity build-up grossly
exceeded real market growth. As a consequence, PET/TPA producers suffered
from low capacity utilisation and depressed market prices around the turn of the
century.
The PET/TPA market is an example of cyclicality, where the supply-demand
balance can only be restored by continued market growth or by phase-out of existing
capacity. At times of limited market growth -or decline- closure of facilities may only
be expected when their marginal cost cannot be recovered anymore and revamping
cannot be justified. Since capacity build-up was realised in a few years, most capacity
is in modern plants with similar performance. The inevitable result is price erosion
from which all players suffer.
In order to reduce risk, over time many major petrochemical companies have stated
they prefer to be number one or two in their respective markets or else exit.
12 Process System Innovation by Design

Thereby, these companies apparently prefer to be active in oligopolistic markets -

markets with only a few major players. It is well known that in such markets
individual companies possess market power and have flexibility to set prices to
recover cost and realise target ROI or ROACE (Lipsey and Steiner 1991). The
oligopolistic market structure and the characteristics of petrochemical production
represent formidable ‘entry-barriers’ for new players. These must have sufficient
capital, access to resources and markets or create their own outlets. A competitive
edge must be established with respect to established producers (Wei et al. 1979),
which implies a company must have the technical and organisational skills to build,
finance and operate modern large-scale facilities and to market and sell its products.
Vertical integration is another strategy to achieve economy-of-scale and create entry
barriers. Vertically integrated companies control significant parts of supply chains or
production networks wherein many petrochemical plants produce for captive use. At
the company boundaries, feedstock supply, utilities and product sales generally are
arranged under long-term contracts with reliable local suppliers, which results in only
small volumes to be obtained via global sourcing or to be sold on the free-market
respectively.
Whilst the petrochemical industry is a truly global business with exchange of
technology and operating practices through multinational corporations, the actual
volume of petrochemicals traded between the geographic regions mentioned above
remains relatively small compared to internal trade (e.g. Anonymous 2001b)). Apart
from vertical integration, apparently the global exchange of market information and
adaptation of prices leads to a situation where transport at volumes at or above the
level of a single plant of typical world-scale is not competitive. Thus, a sufficient
local market for low-priced commodity chemicals is important and has led to the
formation of huge petrochemical complexes in for example Antwerp, Singapore,
Houston and Rotterdam. These enable operating companies to reduce transport
costs and risks.
The Rotterdam Chemical Cluster, for example represents a vested investment of
some 10 billion euros, only 13.000 direct jobs and some 60.000 indirect jobs, and is
largely based on low-priced international oil-supply (Kuipers 1999). Once
established, these agglomerations represent huge advantages because of the very
availability of product outlet and feedstock, hard- and soft infrastructure - pipelines
and utilities; knowledge and service providers and professional authorities. As a
consequence, the establishment of new “grass-root” complexes in the developed
countries is hardly to be expected because of the economic cluster advantages rather
than because of the fear of “NIMBY” sentiments.
The very existence of petrochemical complexes contributes to the industry's
inflexibility, which is caused by sunk cost, capital intensity, ROI/ROACE targets,
technical lifespan of facilities and their interdependence if not vertical integration.

1.3.8 TECHNOLOGY AND PROCESS SYSTEMS

Specialty chemicals often are produced in batch-wise multi-purpose production
facilities. As these are often located within a building, they are generally unnoticed,
except for odorous smells or cases of pollution. Commodities and pseudo-
commodities, however, are produced in bulk quantities in large-scale out-door
Introduction 13

continuous operations that have become the blueprint image of the petrochemical
industry in the industrialised countries.
Such large-scale continuous plants are complex installations in appearance. Indeed at
a detailed technical level, they consist of a variety of interconnected apparatus,
instruments, pipes, and construction. As stated, however, any petrochemical plant
can be represented as a system where a number of physical inputs are transformed
into a number of outputs. As far as the representation of system content or design is
concerned, each petrochemical plant can be completely characterised by a single
configuration of a limited number of process sections or functional units, such as
feed-preparation, reaction, separation, final product purification, process flows,
process recycle flows and utilities (Montfoort et al. 1989; Dijkema et al. 1998). In
Figure 1-4 this is illustrated by a simple general implementation of a chemical
process. The core of the process is some reaction subsystem, which often consists of
multiple reactors. Since most chemical reactions do not give 100% yield of the
desired product only, a separation subsystem is required to achieve product
specification. As a matter of course, intermediate separation in multi-reactor systems
is a possibility.
A particular characteristic of continuous processing is the recycle structure, which in
its simplest form is the recycling of unconverted feedstock back to the reactor. In
multi-reactor / multi- separation systems, the recycle structure is an important
degree-of-freedom in system design.

Feed T Product

Fuel Feed Puri-

preparation Reaction Separation fication
Purge

Recycle Flow

Figure 1-4: General representation of a petrochemical process system.

Since in refining and the petrochemical industry distillation is the dominant choice
for product separation, the image of this industry is formed by the well-known arrays
of distillation towers. Especially at night petrochemical plants offer a spectacular
sight when the pipes, distillation tower, reactors, vessels and a whole range of other
equipment are illuminated. Second in dimension to the distillation towers are the
huge cracking reactors in the refining industry, the largest being the Exxon
Flexicoker and Shell Hycon reaction systems, followed by continuous cat crackers
14 Process System Innovation by Design

and hydro crackers. In the petrochemical industry, industrial furnaces together with
their chimneys have developed into installations of enormous physical dimensions.
These also are the prime exception to the relatively small physical size of reactors in
the petrochemical industry: steam cracker furnaces and reformer furnaces are fired
reactors for the production of crude ethylene and synthesis gas respectively. In
contrast, a plant for the production of styrene, the reactor only is of limited size.

Chemical plant

Inputs Chemical plant Chemical plant

Outputs

Outputs >> Inputs

Exchange

Chemical plant Flow

Chemical plant
Info

Category

System
Chemical plant Chemical plant Element

System
Surroundings The Petrochemical Industry Boundary

Figure 1-5: System model of the petrochemical industry.

Modern continuous chemical plant designs often exhibit complex structures to
optimise the internal process structure for plant performance. Multiple reaction and
separation subsystems may be present, as well as multiple recycles. Feed preparation
and product polishing may be integrated with reaction and separation, respectively.
In advanced designs, reaction and separation may be integrated. Since operating
conditions can be selected to meet a range of objectives, the design of any
commercial petrochemical plant is only one out of a large set of feasible
configurations (Montfoort 1992). A number of alternative process systems for the
production of, for example, propylene oxide and methanol therefore are currently in
use.
Over time, in the industry many petrochemical processes have been developed to
upgrade novel feedstock, waste and by-products. The result is a petrochemical
network that is characterised by interconnectivity and interdependence.
Introduction 15

Cadre 1-1: System analysis and Input /Output modelling

Both in thermodynamic analysis and in system analysis, any process may be

divided into the system and the surroundings by drawing a hypothetical envelope
or system boundary around the operating units or system elements studied. The
surroundings are everything that is not in the system (Seader 1982), i.e. the
complement set to the system set. Inputs and outputs may cross the system
boundary. A trivial system model of the petrochemical industry, for example,
consists of individual chemical plants that exchange intermediate products (Figure
1-5, Table 1.2). A net amount of feedstock and utilities cross the system boundary,
as well as a net amount of industry products and waste material.
Irrespective of what happens inside a system's boundary, a production system's
performance can be characterised completely by its inputs and outputs. An
analysis based on this notion is commonly labelled a 'black-box' approach (e.g. Blair
and Whitston 1971). In economics, this has become known as input-output analysis,
or I/O analysis, and the associated type of modelling is Input-Output modelling.
Upon opening the box, any system consist of a structured collection of system
elements or sub-systems, which itself forms a new black-box or system
(Asbjørnsen 1992). Not only the system can be fully described by its inputs and
outputs, the same holds true for a system element or network node, in our case an
individual chemical plant. When the system boundary is expanded, a hierarchy of
systems appears, where at each level a system may become an element of a higher-
level system. The 'Surroundings' and 'System Elements' labels then change from
'Petrochemical Industry' and 'Chemical Plant' respectively (see Figure 1-5). Each
system aggregation level thus has its unique labels, as given in Table 1.2. Only a
limited number of labels are required for all aggregate system levels to describe
inputs and outputs, interconnecting flows and the exchange with the system'
surroundings The entire petrochemical industry, any of its petrochemical
complexes or any of its individual process plants or parts thereof can be
characterised by their ingoing and outgoing streams only. It may thus be seen that
I/O modelling provides a sound basis to provide a quantitative image of the
petrochemical industry and to assess the performance of its constituting elements
to the level of detail required. Does resource utilisation improvement in propylene
oxide production, for example, require a focus on improvement of existing
industrial technologies, a focus on the development of alternative production
routes, on the expansion of the existing networked systems with additional
conversion steps or maybe a focus on the development of alternative polyurethane
precursors? I/O modelling allows one both to select a suitable aggregate level and
to change level to help evaluating resource utilisation from a variety of
perspectives.
16 Process System Innovation by Design

A petrochemical plant, complex and the entire industry can be represented as a

structured configuration of individual system elements that are connected by some
kind of flow (Figure 1-5, Table 1.2, p. 16). These convert oil products, other fossil
resources and intermediates to suitable feedstock for the polymer industry and other
sectors (Figure 1-3).

Table 1.2: Overview of aggregation levels in and around the petrochemical industry7
System Chemical Plant Petrochemical (Inter)National
Boundary Industry Industry
Surroundings Petrochemical National Industry World Material
(dominant) Industry Cycles
System element Unit Operation Chemical Plant Industry Sectors
Flows: Materials, Energy, (Exergy)
Inputs, Outputs: Resources, Products, Waste, Emissions
Exchange Flows: Intermediate Products, By-Products, Energy
Intensive properties of Flows: Composition, Pressure, Temperature

As with the designs employed for methanol or propylene oxide production, the
present configuration is only one instance of a very large set of possible system
structures (Table 1.3). Where single plants are characterised by a combination of
'once through' interconnectivity and intra-connectivity via recycling between reaction
and separation, the petrochemical complexes and the industry predominantly exhibit
'once-through' intra-connectivity from feedstock to product. The majority of chains
originate from the steam cracker and re-unite somewhere in the polymer industry,
e.g. where polyurethanes are produced and moulded into car seat-cushions (Ch.4).
In the petrochemical industry, the major chemical activities are the chemical
rearrangement and partial oxidation of hydrocarbons. Thus it has become an
extremely fast element of the world's fast carbon cycle8 that is driven primarily by
life-on-earth. Biomass grows and decays, which maintains a balance of carbon as
CO2 in the atmosphere. A limited amount of biomass is temporarily extracted from
the cycle by fossilisation. Fossil resource extraction for use in power generation,
industry, transportation and households has resulted in a rapid reintroduction of
carbon fossilized over millions of years and which has induced a steep increase in
atmospheric CO2 content (IPCC 2001).

7
Currently, a great many (petro-)chemical plants are part of an industrial cluster. Thus, the
petrochemical industry' system element may also be defined as a petrochemical complex, with the
possibility in some cases the element or 'complex' only comprises of a single petrochemical plant.
8 The slow carbon cycle concerns geochemical processes where over millions of years erosion releases

carbon. In this slow cycle invertebrae sequester carbon by the formation of shells composed of lime
(CaCO3). The calcination of lime for cement production thus has become part of this slow carbon-
cycle.
Introduction 17

Table 1.3: System levels in the petrochemical industry

System System Intra -Connectivity Degree-of-freedom
elements
Petrochemical Complexes / Mass flow, Selection & configuration of elements;
Industry Plants infrastructures Feedstock/product state and composition
Complex Chemical Mass & energy Selection & configuration of elements;
Plants flows; utility Feedstock/product state and composition
infrastructure
Chemical Unit Mass & energy Selection & configuration of elements;
Plant Operations flows; utility Conditions & flow composition
network
Unit Apparatus Direct transfer of Selection & configuration of elements;
Operation mass, heat, power, Principle & design
impulse;
Apparatus “Internals“ Mechanical, Selection & configuration of elements;
Electrical, Chemical Operating conditions

Industrial development has created novel interconnectivity between material cycles.

The fast carbon cycle is linked with other material cycles via fertiliser production and
through the contaminants present in coal and crude oil that range from heavy metals
to sulphur, phosphorus and nitrogen. The production of iron, for example, requires
carbon, aluminium and zinc plants use carbon electrodes, and many non-
hydrocarbon components are added to commercial plastics (TiO2, ZnO, Chlorine),.
Since the majority of fossil fuels are burnt, and only a fraction of the plastics
produced is subject to back-to-feedstock recycling, the majority of fossilized carbon
is dissipated into the atmosphere as CO2.

1.4 Industry development, innovation and sustainability

1.4.1 DEVELOPMENT OF THE CHEMICAL INDUSTRY

The Dutch petrochemical industry has developed to its present state over some 350
years. In his overview Mol (1995) distinguishes four periods of development in the
chemical industry. In the subsequent overview a subdivision of the period after
World War II results in five characteristic periods of development.
The 17th and 18th century. In the 17th century the Netherlands held a prominent position
in the international chemical and metallurgical industry of the time; activities
included the production of mercury compounds, madder dyes, such as lead white,
blue, soap work, litmus and volatile oils. During the 18th century the industry's
importance decreased, with the exception of madder dyes, the candle industry, lead-
white and soap.
The 19th century until World War I. In the 19th century, major developments in inorganic
chemistry took place, e.g. development of alkali, sulphuric and phosphate processes.
Large breakthroughs were realized internationally through the work of renowned
chemists. The industry prospered and grew, albeit not in the Netherlands, which Mol
largely attributes to the role of universities, where teaching in chemistry was largely
absent. Notably, in other historical studies, the importance of scientific research and
education is also labelled as an important factor (Homburg et al. 1998).
 18

Solar Radiation

Sources
Interconnected
World >> Economic System Material Cycles
Inputs
'Extracted'

Europe >> National Economies

Natural Physical, Chemical,
Biological and Geochemical
Processes

Sources
Nations >> Industrial Complexes
Outputs
'Emitted or
Wasted'

Complexes >> Plants

Natural Physical, Chemical,
Sinks

Biological and Geochemical

Processes

Plants >> Process Units

Figure 1-6: Worldwide Interconnected material cycles.

System Levels System Boundary
Sinks

Space Occupied
Process System Innovation by Design
Introduction 19

In addition, economic growth stagnated, and Dutch industry was still focused on the
maritime industry and trade. By some it was considered ‘backward’ in comparison
with the rest of Europe, notably Germany, France and Great Britain. Gas
production from coal, and the work-up of the coal-tar by-products were important
activities, with around 180 factories in operation around 1900. The products, coal
tars and its aromatic distillates, were largely exported because vested interests of
natural dye producers prevented the set up of a Dutch synthetics dye industry. The
only surviving coal-tar refinery today, Cindu in Uithoorn was established in 1922.
The 20th century - interbellum. The development towards the petrochemical industry
known today started after World War I, when a transformation from agriculture-
based products to ‘chemical-based products’ started. In the Netherlands chemical
operations of well-known companies such as DSM, Akzo, and Shell Chemicals were
started. Before World War II, the Netherlands remained a net importer of chemical
intermediates.
World War II - second oil crisis. After the war, industry successfully developed because
of the Marshall plan, the national industrial policy adopted to rebuild the
Netherlands and the forced reduction of competing German coal tar refineries. In
order to understand the present structure of what is known today as the
petrochemical industry, the post World War II period is most important.
The war initiated a large number of product innovations demanded by the military.
As a consequence, both production facilities and new products became available for
the civil sector. A second major development was the rapid-transformation from
coal-based to crude-oil based operations.
According to Chauvel and Lefebvre, the transition from coal to crude oil was mainly
initiated by the continuously growing demand for ethylene, the precursor of three
important plastics, viz. polyethylene, polyvinyl chloride and polystyrene (cf. Ch. 4).
Initially, ethylene was produced from coal, mainly via the processing of coke-oven
gas, a by-product of large-scale steel production (Chauvel and Lefebvre 1989a).
Although the pyrolysis of light petroleum fractions9 was developed in the U.S. for
the production of ethylene as early as 1920, coal-based supply for ethylene and other
chemical products became rapidly insufficient after the World War II. In addition,
economics were favourable at the time and large newly discovered oil reserves led to
an abundant supply of cheap raw material. Industrial plants could be realised at
lower investment costs for the installations whilst delivering better quality products.
At the time, chemical science and engineering rapidly developed and provided novel
chemical synthesis routes enabled by new or improved catalysts. Industrial process
plant development benefited from advances in process technology that range from
reactor technology to separation and rotating equipment. As a consequence, for
most petrochemical products, by the end of the sixties this transition was complete.
Since the Netherlands did not have a large coal-based chemical industry, the growing
demand for petrochemical products opened an opportunity for competition with the
German industry, which historically was largely based on coal, and continued to be
based on coal-derived aromatic products for a long time (Molle and Wever 1984). In
the sixties and early seventies, double-digit growth continued in the chemical

9 This technique is nowadays is generally known as 'steam cracking'

20 Process System Innovation by Design

industry, largely as a result of worldwide economic growth. The availability of fossil

resources, both for feedstock and energy supply, despite yearly consumption
continued to increase due to new discoveries.
During the sixties, an important stimulus for the Dutch industry was the discovery
of large resources of natural gas, notably the Slochteren field. A number of industrial
projects were launched to further the industrialisation of the Netherlands. In
Delfzijl, for example the methanol-plant started operation in 1973, and the
Aluminium-smelter of Aldel was built - later acquired by Hoogovens, and presently
part of Corus. In Geleen and Sluiskil, large fertiliser complexes were constructed that
used natural gas for the production of ammonia. The purpose of these investments
was both to quickly reap the benefits of the natural gas and to create local jobs.
The introduction and market growth of new products increased demand for fuels
and chemicals, where sometimes a large amount of low-value by-products were
generated. In the entire petrochemical industry the availability of such previously
low-valued by-products provided an incentive to develop products and processes
that allowed upgrading of these by-products. The existence of previously
unexploited economies-of-scale, combined with a relatively low-cost of
transportation for most chemicals led to the formation of petrochemical complexes
and a continuous increase in plant capacities, a development that continues today.
After the second oil crisis. The second oil crisis of 1979 severely impacted the world
economy and is seen to be the prelude to the worldwide economic recession in the
'80s (Woltjer 2002). In 1979, concerted OPEC actions resulted in a tripling of oil
prices. Meanwhile, due to the monetary policy of the Reagan administration real
interest rates steeply increased worldwide. This exposed the weak financial and
competitive position of companies and countries. After three decades of continuous
high growth, most petrochemical markets decelerated, and overcapacity in the early
80’s resulted from new plants coming on-stream for non-existent market demand.
In the last two decades of the century, the industry went through a number of 'boom
and bust' cycles: it had become a 'cyclical industry'. In times of saturated markets the
risk of constructing overcapacity is very real. Thus, the establishment of a new
facility in an existing product market is difficult to justify. Moreover, it is uncertain
whether the cost of process innovations can be recovered (after Stobaugh 1988).
The response of a number of large firms in the petrochemical industry has been to
reassess their core competences and core business (Hamel and Pralahad 1994). Some
have decided to move out of the (semi)-commodities market and to move
increasingly into specialties, fine chemicals and the highly specialized production of
pharmaceuticals that are perceived as attractive because of their high profit-margins
(viz. Asselbergs 1998). These companies leave value-extraction from existing
facilities thus to the new owners and management, who more often than not aspire
for economy-of-scale through mergers and acquisitions (Dijkema 2000). The result is
an oligopolistic structure of many chemical markets, i.e. markets dominated by only
a few firms (Wei et al. 1979: 251).
In the nineties, the companies in the petrochemicals business have restructured their
organisations, often at the expense of corporate functions, administrative overhead,
internal services and support, and R&D and engineering departments. These have
been labelled 'cost-cutters (Whitehead 2000). In construction or expansion projects,
Introduction 21

there has been a strong focus on minimising investment and initial time-to-market,
in order to obtain revenues as quickly as possible. Thus there has been a strong
emphasis on efficient organisation of the project execution from conceptual design
to start-up. Concepts used include, amongst others, 'concurrent engineering' to
shorten project execution elapsed time, 'front-end-loading, which is a strategy to
amass all crucial information and elucidate white-spots as early as possible and not to
allow any changes after an early date in the project and 'value engineering', where
project items are scrutinised for value-creation potential versus cost (Herder 1999).

1.4.2 THE PROCESS OF CHANGE

Between periods of dramatic change and 'outbursts' of innovation The mechanism
of continuous change operates via the complex interplay of markets and
competition, policy and regulation, strategic management, technology and system
development, public image and scientific development at large (e.g. Twiss 1992).
This complex interplay often is reduced to technology-push versus market-pull (e.g.
Grunert et al. 1997), which reflects two diametrically opposed views on the origin of
innovations. In developing an analysis framework, Grunert et al. combined
technology-change fuelled by R&D and market orientation respectively (Figure 1-7).

R& D

Invention

Feedback
Feedforward

Innovation
Market
Public <development> Market Orientation
<commercialisation> Orientation
Orientation <Business-to-Business >
<Consumers>

Diffusion
<customer acceptance>
<societal acceptance>

Feedback Feedback
Performance

Figure 1-7: Framework for the innovation process in the petrochemical industry; adapted
from (Grunert et al. 1997: 3).

Market orientation is a proactive attitude towards the detection of novel, as yet

unfulfilled customer needs that can lead to successful innovations. The
petrochemical industry, however, is a provider of intermediates to a myriad of
industrial customers, i.e. it largely operates in a business-to-business market where
innovation through market orientation gets the form of co-makership and co-
22 Process System Innovation by Design

development. Its products remain largely invisible and unknown to the general
public.
Where consumers thus interact only indirectly with the industry, a variety of other
stakeholders directly influence its change and innovation. Apart from shareholders,
competitors and would-be competitors include the public at large, governments,
non-governmental organisations, pressure groups and universities. Necessitating a
'Public orientation' (Figure 1-7), petrochemical companies have realised they must
ensure their license-to-operate and respond to environmental, safety and
sustainability concerns. Together with sustained value creation, these factors
determine long-term competitiveness and continuity and thereby have become
drivers for innovation by the industry and its knowledge and service-providers such
as universities, technology licensors and engineering contractors.
The changes or innovations in the industry vary in level of detail, scope and time-
scale. Production volume change, for example involves day-to-day utilisation
adjustment, small plant changes, incremental capacity increase by revamping, and
closure or erection of facilities, while for many products the market stimulates
sustained capacity increases. Short-term competition drives day-to-day operation and
organisation; long-term competition involves process innovation, which leads to
phase-out of currently employed facilities. Increasing public awareness of impact of
suspect chemicals combined with cumulating scientific knowledge and evidence may
call for the phase-out of particular products, production processes or production
routes. In other cases, product innovation leads to successful substitute competitive
products, and total phase-out of current facilities.

1.4.3 INNOVATION AND THE PETROCHEMICAL INDUSTRY

“An innovation refers to any good, service, or idea that is perceived by someone as
new. The idea may have long history, but it is an innovation to the person who sees
it as new” (Kotler 1991)10, or in more general terms, the audience or the stakeholder
who perceives it as new. This implies, for example, that past process innovations by
some chemical company that become publicised must be perceived as an innovation
by for example public policy makers or competitors!
The process of change was emphasized and decomposed by Schumpeter into
invention, innovation and diffusion (Schumpeter 1942; cited in Ferguson and
Ferguson 1988). Invention, the generation of any new idea, must be followed by
innovation, where amongst others technological development, manufacturing
organisation and scale-up enable commercialisation to bring the invention to the
market. The final stage is diffusion, acceptance by consumers (products) or
employment by producers (processes). Considerable time may lapse between these
phases. Inventors may protect their ideas from competitors, to prevent them from
subsequent innovation and commercialisation. The scientific and patent literature
may serve as a database of ideas awaiting innovation. Notably, Schumpeter argued
that invention is not limited to developments that require scientific advance, i.e.
inventions may originate from new ideas that combine existing technologies in new
process system designs (Schumpeter 1942; cited in Ferguson and Ferguson 1988).

10 cited in [Grunert, 1997 #67

Introduction 23

Products (including services), production systems, and organisations can be

innovated, and are typically referred to as product innovation, process innovation,
organisation (private sector) and institutional (public domain) innovation ,
respectively. In addition, Schumpeter identified the development of new markets
and sources of supply (feedstock!) as distinct categories of innovation. Product
innovation, process innovation and organisational innovation are closely related. A
new product or service often depends on an innovative production process or
organisation, respectively. Many innovative services such as 'Air-Miles' have only
emerged through ICT-enabled organisation; many new material products can only
be produced because of new production processes.
The primary objective of a process innovation is to further manufacturing in a
particular sector, whilst a new product can affect a multitude of sectors. Taking this
perspective on 'type of use' to the extreme, an innovation that is used outside its
sector of invention must be labelled a product innovation; only if it is used within a
sector itself it is labelled a process innovation (Pavitt 1984). In the engineering
sciences, however, a common understanding of a process innovation is 'any
improvement of the process of production or the production system through the
development or application of new technology.' For example mechanical engineers
consider the application of gas turbines in electric power generation from 1963 and
later a process innovation, whilst according to Pavitt’s definition it is a product
innovation because the gas turbine is used outside the turbine machinery industry.
For the purpose of this introduction we focus on process innovations, which we
define as ‘any improvement of a manufacturing system, be it a sector, industrial
complex, individual plant or unit operation'.

Cadre 1-2: Petrochemicals - a mature industry

Since the early nineties, the petrochemical industry's development is characterised

by limited growth rates and maturation (see § 1.4.1 below). At a time of limited or
non-existent market-growth, a new facility that comes on stream results in
overcapacity. As a consequence, the market price for its products can be expected
to deteriorate for all producers, who also will suffer from a decrease in average
plant utilisation. Meanwhile, each individual plant has its own break-even
utilisation, which is a function of historic investment, current margin between
product and feedstock, energy-efficiency as a function of plant capacity, flexibility
and fixed operating cost (operating personnel, insurance, maintenance, overhead).
Only when there is no technically feasible utilisation where a net marginal profit
results, an existing facility will be shut down completely. In all other cases,
ownership of existing facility can be transferred to optimise return on asset capital
employed by each individual company.
A mature industry is now characterised by the fact that possible innovations do
not provide sufficient edge to make existing installations economically obsolete.
24 Process System Innovation by Design

1.4.4 INNOVATION AND CHANGE

In a time frame of 50 years the petrochemical industry has gone through a typical S-
shaped growth curve. Around World-War II, product innovations initiated its
development. Rapid market growth led to the invention of a myriad of new products
and to a continuous demand for new and more effective processes. Subsequently,
during two or three decades, it dramatically improved its performance by process
innovations that allowed, amongst others, the exploitation of economies-of-scale.
Since the beginning of the nineties, the industry's development is characterised by
limited growth rates and maturation. Thus it exhibited the general pattern where
process innovation follows product innovations (Abernathy and Utterback 1978).
The petrochemical industry has been classified a science-based industry (Pavitt 1984)
and has an impressive record of process innovation at or below the level of single
plants, where fundamental and applied research have fuelled continuous
improvement of e.g. catalysts, reactors, separation technology and scale-up of
operations (e.g. Weijnen and Drinkenburg 1993). Science-based’ industry's “main
sources of technology are the R&D activities of firms in the sectors, based on the
rapid development of the underlying sciences in the universities and elsewhere.”
Pavitt argues, however, that for “large-scale producers, particular inventions are not
in general of great significance, (…). Technological leads are reflected in the capacity
to design, build and operate large-scale continuous processes … Technological leads
are maintained through know-how and secrecy around process innovations, and
through inevitable lags in technical imitation, as well as through patent protection”
(Pavitt 1984: 359).
Based on elaborate studies into US petrochemical companies, Stobaugh observed
that successful companies use process innovation through scientific advancements
to maintain their competitiveness in predominantly oligopolistic markets (Stobaugh
1988). He demonstrated that innovation must be part of overall company strategy to
protect market-share by effectively discouraging potential competitors from market-
entry. Others have shown that firms in the petrochemical industry produced a
relatively high proportion of their own process technology and product technology
(Pavitt 1984: 362; Hutcheson et al. 1995).
Schumpeter's observation that science often is not a necessary requirement for
innovation and Pavitt and Hutcheson's findings explain why petrochemical
companies tend to be very secretive, e.g. about their operating practices and project
execution tactics: know-how and know-to together bring a competitive edge in
design, construction, financing and operating chemical plants. Indeed, secrecy has a
long tradition in this industry where companies compete by the performance of their
process technology (van Breda and Dijkema 1998).
In the 90's, 3rd generation R&D management was adopted to match existing
portfolios to present and future business (Roussel et al. 1991). The adopted
strategies differ somewhat among companies, the effect being the same: major
reorganisation and budget-cuts, stripping and even closure of R&D facilities. As an
exception to the rule, a company like Air Products continues to recognize one of its
strengths being its proprietary technological know-how in large-scale air separation
(Govaerts 1999). The general trend, however, has been to cut in-house R&D
Introduction 25

directed at large-scale processing. As a result, rarely do real breakthrough process

innovations make it to the market place.
Meanwhile, there appears to be a strong drive on 'reinventing' organisations. If
product and process innovation do not offer sufficient scope for improvement, what
else is there to face competition than to cut cost by streamlining organisation? Thus
petrochemical corporations went from vertically integrated corporations to 'loose'
conglomerates of strategic business units focused on oligopolistic product markets.
As of the last decade of the 20th century, they are in the process of organising their
way of doing business around the previously unexploited benefits of ICT-enabled
organisation (Dijkema 2001a).

1.4.5 SUSTAINABILITY
With the 1989 publication of ‘Our common future’, the worries about the status of
our planet and worldwide society were coined by the caption of a need for
‘sustainable development’, which in Brundtlandt’s definition is (economic)
development and consumption that does not hamper the needs of future
generations (World Commission on Economic Development 1987). Welford argues
that ever since
There exists a strange and fruitless search for a single definition of
sustainable development amongst people who do not fully understand
that we are really talking here of a process rather than a tangible
outcome. This search is most apparent amongst positivist researchers
who grope for a hard core of definitions and data that they can
manipulate to produce simple solutions and singular answers to very
complex concepts. Such simplifications cannot exist in the post modern
world and they simply hide a scientific research bias, which is not
appropriate to a highly political issue such as sustainable development
(Welford 1997).
Although the above is valid for the economy at large, we do believe that for the
petrochemical industry some criteria can be formulated. As we have argued, today
harmful emissions and the wastes produced by the petrochemical industry have
become 'managed problems', with the exception maybe of SO2 and NOx emissions
associated with burning fuel oil and natural gas. It is the large quantities of fossil
resources processed that will become the future problem of the industry. At the
output side, large quantities of CO2 are emitted to atmosphere. At the input side the
industry depends on crude oil and natural gas. In the near term future, we expect
society to require continued availability of petrochemical products and derivatives. It
may not before long, however, before these will have to be produced from much
more expensive fossil resources, under more stringent arrangements with respect to
the fate of CO2 generated, or from alternative feedstock altogether.
But how has the petrochemical industry responded to the demand for sustainability?
In retrospect, the petrochemical industry has done a good job in improving its day-
to-day operations as well as technology and systems portfolio. The benefits from
increased resource utilisation provided sufficient incentive for R&D efforts, which
indeed have led to a substantial reduction in the consumption of fossil resources per
unit of product. Meanwhile, however, the industry has demonstrated almost
26 Process System Innovation by Design

continuous growth since 1950, and as a consequence net fossil resource use
increased dramatically.
Although we labelled the petrochemical industry as being furthest in embarking on
strategies for sustainable development(Mayer and Dijkema 1999), there is no
evidence that suggests that these strategies involve drastic process innovation. On
the contrary, whilst from 1989 onwards there has been a renewed growth in the
world-wide petrochemical industry and its consumption of fossil resources, the
number of successful petrochemical process innovations expressed as novel
industrial processes that were commercialised has declined for three decades
(Satterfield 1991). Despite the continued interest in the use of biomass and coal, the
increased demand for the building-blocks of commodity-plastics has largely been
met by the construction of an increasing number of steam crackers that use naphtha
or gas oil for feedstock. Exceptions are the first methanol-to-olefins cracker project
(Dijkema and Kuipers 2001) and Fischer-Tropsch synthesis that uses coal- or natural
gas-derived synthesis gas. Since their economic life-time is long, the fossil resource
demand and the products that result from these installations will dominate the
industry for decades. Despite process and product improvements, e.g. by the use of
new catalysts or separation technology, the net industries' share of fossil resource
consumption will remain at a high level.
Presently, in the commodity markets of the petrochemical industry, it is hard to
foresee that innovative products will be developed short-term that will completely
replace current products. Quite the contrary, many long-established products still
demonstrate considerable growth rates. As a consequence, improvement of the
sector largely has to be fuelled by process innovation. But according to Abernathy
and Utterback, the rate of such innovations eventually will fall off as prospective
incomes fall (Abernathy and Utterback 1978). Indeed, in the last decade the sector
appears to increasingly emphasize organisational and even inter-organisational
innovation. Product-chain management and sustainability are elements of the
Responsible Care (RC) programme, a strategic communication programme that first
has been developed by the petrochemical industry of Canada. The concept of RC
expresses the notion that a chemical company is part of its environment.
Consequently it must act responsibly and precautionary towards the environment
and society. The RC-programme encompasses the aspects of health, safety,
environment, welfare and responsible entrepeneuring. As stated, however,
Responsible Care is primarily a strategy for adequate communication with all the
stakeholders of a chemical company, in society, politics, and business. It also helps,
however, to provide arguments such as that “we have done enough, others outside
our company must now take action first“, or as one respondent remarked, a
mechanism to buy time, so that investments required for environmental
improvement can be matched with the right time for business (Mayer and Dijkema
1999). Thus, Responsible Care and similar programmes can create both incentives
and disincentive for the much needed process innovations.
In the Netherlands, the petrochemical industry has negotiated voluntary agreements
on energy-efficiency improvement, and it appears that the petrochemical industry is
on-target with respect to the objective of 20% energy reduction (VNCI 1998). These
agreements largely focus on the net ‘utility’ energy efficiency. As stated, however, in
Introduction 27

contrast to the other manufacturing industries, in the petrochemical industry fossil

resources are both feedstock and 'utility' energy supplier. Feedstock efficiency,
however, appears largely to be left outside the agreements and the industry structure
remains unchallenged. Sustainability, however, cannot be achieved by optimisation
of utility energy use alone.

1.4.6 RESOURCE OUTLOOK

While fuels of fossil origin are used for energy supply throughout our industrial
society, the use of these carbonaceous materials as feedstock is largely limited to oil
refineries, electric power plants, coke-ovens and the petrochemical industry and
related chemical industry (Figure 1-3). In base metal production, for example,
carbon-based products are of primary importance to enable the production of iron,
zinc and aluminium. Blast-ovens for steel manufacture consume coke for energy
supply and the reduction of iron ore. Aluminium smelters, zinc plants etc. use
carbon electrodes. In these cases, however, no significant amount of carbon ends up
in the product; the feedstock of these industries is metal ore. Coke and carbon
electrodes are required utilities to enable production.
The energy need of refineries and petrochemical plants generally is covered by
combustion of product waste, fossil fuels and electricity. Thus, oil refining and the
petrochemical industry share a unique characteristic: the use of fossil fuels for both
energy-supply and feedstock purposes. Since oil and gas power the world, the
resource situation of these industries is intrinsically linked with global energy supply
and demand and the amount of 'proven reserves'. Presently, world proven crude-oil
reserves are estimated to suffice for some 45 years of current total global
consumption. The ratio of resource to yearly production (and consumption), the
R/P ratio, is 65 and 250 years for natural gas and coal, respectively (Anonymous
2004). Net annual consumption, however, is expected to grow significantly as the
result of population growth, increased standards of living and economic growth.
Especially a large impact of developments in South-East Asia is expected. As a
consequence, the real horizons may turnout to be much shorter than predicted on
the basis of the R/P ratios calculated using simple data.
Indeed, as early as in 1972 the Club of Rome first warned for the rapid worldwide
depletion of resources based on a system dynamics model that predicted patterns of
resource availability, human lifespan and welfare based on the interrelated dynamics
of population, standard-of-living, healthcare, resource consumption, economics of
resource exploration, technological advancement, land-use and environmental stress
(Meadows et al. 1972). Although events have not yet taken the dramatic turn
('system collapse') as predicted by this model in a number of scenario runs, it is
obvious that total fossil resources of the system Earth are being depleted
continuously.
Reassuringly, since the discovery of crude oil in the United States, its R/P ratio grew
almost continuously as a result of new discoveries, market changes and technological
development. Since the mid-90's, however, combined developments only have
compensated for current use, despite the recent discoveries of previously unknown
large reserves of oil and natural gas. With global energy-demand expected to grow
28 Process System Innovation by Design

significantly, the OECD predicts that around 2015-2020 the time horizons for
availability of oil and gas resources will start to shrink (OECD 1999).
After the Gulf War (1991), abundant oil-supply by eager producing countries
combined with free market conditions drove down oil prices to historically low
levels. This development appears to support Grubbard's scenario where oil-price
remains relatively low, until the moment where the time-horizon of perceived
shortage becomes comprehensible (Grubbard 1992). Then oil-prices rise to levels
never seen before, largely because the perceived abundance of oil not only results in
low energy prices and limited exploration and development of oil fields, but also to a
lack of economic incentive for R&D and innovation to reduce and change our
energy use. Indeed, currently OPEC and other producing countries control their oil
production to manage the supply-demand balance to maintain oil-prices and
optimise revenues. Other scenarios have been developed, however, wherein such
managed oil prices rise slowly but steadily in response to awareness of a definite
dwindling of known economic oil reserves. These scenarios recognise that price
increases create additional incentive for exploration and for the development of
previously uneconomic resources. Managed crude oil availability and managed price-
levels, however, still adversely affect investments in and R&D into improved energy
resource utilisation and alternative energy sources. Long-term, however, it is certain
we will run out of relatively cheap oil and gas and that alternative energy sources
must be developed.
As a matter of fact, crude oil already is being replaced by natural gas as the world's
most important energy source. Researchers of IIASA11 investigated and modelled
the evolution of the use of energy carriers by humankind. They modelled the use of
a particular energy carrier by it's share f as a function of time on a logarithmic scale,
where (f/(1-f)) exhibits linear growth for a number of decades, peaks, and
subsequently declines linearly (Marchetti and Nakicenovic 1979; cited in Spiro and
Stigliani 1996: 54). Historic data for the market share of wood, coal, crude oil and
natural gas were found to be in fair accordance with this model.
During the industrial revolution coal replaced wood as major energy carrier. In the
20th century oil replaced coal. Currently, the share of natural gas is growing to replace
crude oil as the major energy carrier. History indicates that energy carriers start being
replaced when the most economic sources have been depleted. Thus, due to the
availability of fossil resources and ongoing biomass depletion, the share of biomass
in worldwide energy supply has dwindled. Initially, during the industrial revolution it
was replaced by coal. Since World-war II, the use of biomass and coal declined due
to substitution by oil products and natural gas. Nuclear fission for electric power
generation started to develop in the sixties and seventies, and at the time nuclear
fission and fusion were expected to become the next major energy source for
humankind.
The inevitable changes in fossil resource availability and price represent major
incentives for innovation in the petrochemical industry. To ensure its continuity as
the industry's infrastructure, current feedstock must be used even more efficiently;

11 The International Institute for Applied Systems Analysis, Laxenburg, Austria

Introduction 29

novel process systems must be developed to utilise alternative feedstock. Changes in

product demand and product slate must be anticipated and catered for.
At a worldwide share of 8-12 %, the petrochemical industry is only a modest
consumer of crude oil, the major share of oil products being used in automotive
transport and household fuel. The industry presently can only react to crude oil price
fluctuations. Its activities have limited to no influence on long-term feedstock
exploration (e.g. Stobaugh 1988; Fahy 1990). Ironically, however, it is only in the
petrochemical industry that some advantage is taken from complex molecular
structure of the chemical substances in crude oil and oil products. In retrospect, a
relatively young petrochemical industry could grow rapidly in a period that lasted
roughly from 1960 to 1980 on the basis of uninhibited depletion of scarce resource.
In the 21st century, however, the petrochemical industry may have to face accelerated
depletion of its current resource base and economic conditions unfavourable for its
innovation. This is a situation unanticipated at the time of its initial development and
in the period of massive growth of the industry, when basically a 'once-through
processing' structure was adopted throughout the industry. Combined with the long-
lead times of the development of new technology or new processes in the
petrochemical industry, this justifies the economic significance of timely efforts to
improve technology! In addition, strategies must be developed to avoid
entrenchment in technology and systems that are characterised by insufficient
resource efficiency.

1.5 A need for process system innovation?

1.5.1 TOWARDS A SUSTAINABLE PETROCHEMICAL INDUSTRY?

The global industrial economy has been dependent on a glut of oil, gas and coal for
more than a century. The present use of fossil resources by far exceeds the use of
renewable energy sources. As a consequence, oil and natural gas are expected to
become scarce resources in maybe as early as 2020 (OECD 1999). The CO2 liberated
in fossil fuel conversion for the greater part has accumulated in our atmosphere and
oceans. It has caused a steep increase in atmospheric CO2 concentration never seen
before in the Earth's geological history. Evidence gathered in a variety of scientific
studies has led the UN Intergovernmental Panel on Climate Change to acknowledge
that emissions of human industrial society cause an enhanced greenhouse effect that
drives global climate change (IPCC 2001).
Today the developed countries depend on and benefit most from the use of the
world's fossil energy resources. The OECD countries have consumed the major
share of oil and natural gas resources. While the economic development of China
and India will foster continued growth in world energy-use, world consumption
patterns will remain far from equitable. These ecological and societal concerns have
been coined in the early definition of sustainable development “Development that
meets the needs of the present without compromising the ability of future
generations to fulfil their own needs” (World Commission on Economic
Development 1987), which also implies suitable economic development that
stipulates equity throughout the globe.
30 Process System Innovation by Design

Whilst the depletion of fossil resources and the threat of global-warming have led
the public, governments and NGOs to believe that dramatic change is required in
industry to achieve sustainability, only few companies have acknowledged that
sustainability is crucial for their long-term prosperity (Dijkema and Mayer 2001a).
The Dutch Scientific Council for Government Policy recently has categorized CO2
emission, climate change, biodiversity and the continuously growing energy-use as
wicked problems (Wetenschappelijke Raad voor het Regeringsbeleid 2003). These are
labelled wicked because to date these have largely escaped (inter)national policy and
regulation. Upon a series of wake-up calls on environmental pollution (Carson and
Darling 1962), in the past 30 years many tough environmental problems have been
turned into manageable problems. The interplay of end-of-pipe measures, technology
development, process improvements, investments, environmental legislation and
covenants has fuelled a continuous process of improvement. Wicked problems also
must be addressed effectively to enable sustainable development. Effective
communication on wicked problems is difficult, however because there is no direct
exposure to the associated adverse effects that develop gradually over decades in a
global context. Moreover, vested interests are huge, access to cheap energy being
pivotal to the prosperity of many an economy or private company.
Drastic reduction of resource consumption and CO2 emission is a wicked problem
for the petrochemical industry because the core business of the industry is the
conversion of petroleum products, its CO2 production is linked to its present use of
these fossil feedstock, and most companies compete in global markets where
opinions on CO2 and global-warming vary. Thus, despite efficiency gains, the
industry's CO2 emission has remained high largely because of the continuous growth
of production volume, which is illustrated by the doubling of ethylene production in
the past 25 years (see Ch. 4).
A sustainable petrochemical industry is characterised by appropriate resource
selection, effective resource utilisation, the avoidance of waste and emissions and a
product slate that fosters a sustainable society.
While “innovative design and clever process modifications are essential to the
economic health of the producer of commodity chemicals” (Wei et al. 1979: 255),
companies or sectors differ in their degree and development with regard to
innovation strategies for sustainability. Amongst others via environmental
management and technological development the petrochemical industry has reduced
its harmful emissions. In addition, the industry has realised that it must continuously
renew its license-to-operate obtained from local communities and other
stakeholders. Through the Responsible Care programme, therefore, the industry has
been developing its communication and image with local communities and the
general public (VNCI 1998). There are a few 'early adopters' or trendsetters that
have realised and acknowledged that sustainability is crucial for their long-term
continuation.
Some companies have formulated a sustainability strategy in interaction with external
stakeholders, others perceive sustainability as a continuous development process, in
which the specific goals repeatedly need redefinition and reformulation (Mayer and
Dijkema 1999). In the petrochemical industry, notably substantial CO2 emission
reduction, however, cannot be realised by environmental management only. Good-
Introduction 31

housekeeping or end-of-pipe measures do not suffice because its CO2 production is

linked to its present use of fossil feedstock and its maximum efficiency of operations
achievable. Therefore, changes in feedstock, process systems, process routes and
products are required. In addition, it has been argued that these changes must,
somehow, be enabled by technological development, if not technological
breakthroughs.
The need for sustainable development provides a major incentive to elucidate the
question of process system innovation in the petrochemical industry. Can changes in
system structure or system design of the industry, its industrial complexes, or
individual plants that are enabled by technological inventions or vice versa foster
bringing into being a sustainable petrochemical industry?

1.5.2 CENTRAL RESEARCH THEME AND RESEARCH QUESTION

The emergent need for sustainability at a time where the economics of process
innovation are unfavourable poses a dilemma. There is a need for a resurrection of
process innovation to anticipate changes in resource availability and to deal with
wicked environmental problems such as global climate change. Grossmann and
Westerberg suggested to broaden the scope of process system engineering to span a
chemical supply chain -from mastering the molecules to chemical enterprises-
(Grossmann and Westerberg 2000: 1700).
We conjectured that in order to contribute to sustainable development, industrial
complexes, industrial sectors need to be modelled, or possibly worldwide economies
and material cycles (Dijkema and Reuter 1999). Any new approach to innovation in
the petrochemical industry however, must have sound technological and system
content in order to have a chance to achieve some impact in industry. The questions
arise therefore as to what innovations are required to meet the challenge of a
sustainable petrochemical industry and from where can the innovations required for
the petrochemical industry originate. The third and main question addressed in
this thesis is

how can technological or systemic content of innovation required for

sustainability be specified?

We argued that while end-user markets are catered for by newly innovated products,
the products of the petrochemical industry are probably here to stay for a prolonged
period. To meet sustainability demands, therefore, and to produce alternative
products, innovation of the 'T', the process systems, is required. Presently, however,
economically, the industry is in a mature phase. In a response to perceived
shareholder pressure, shareholder-driven CEOs lead the industry on an ever-
accelerating pace of mergers and acquisitions to capture the benefits of economy-of-
scale. At the same time R&D budgets are being cut or activities in
(pseudo)commodity markets are divested.
An image of current business climate emerges that is unfavourable for process
innovation initiated from within the industry. This dichotomy between industry
32 Process System Innovation by Design

climate and need for global sustainability demonstrates the relevance of our central
research theme:
The systemic, top-down specification of the desired technological content
for innovation in the petrochemical industry in relation with business
and societal needs.

The central research question addressed is

Is it possible and worthwhile to devise some procedure to structure the
search of the process system innovation space and to foster the
specification of innovation content?

Our research objective was

The development of a systemic approach to specify the desired content
of process system innovations. “These are defined as changes in the
system structure or system design of the petrochemical industry, its
industrial complexes, or individual plants. These can be enabled by
technological inventions or vice versa” (Dijkema et al. 2003).

Preferably, an approach results that includes and links system analysis and
subsequent synthesis and conceptual specification of options for R&D: process
system innovation by design.

1.5.3 CONJECTURE AND HYPOTHESIS

In this thesis the focus is on the technology and the structure of the industrial
systems in the petrochemical industry. How to achieve sustainability is an open
problem that cannot be solved by planning and optimisation of system structure or
traditional R&D and technology development alone.
In the majority of the scientific community the system design of the industry has
been considered fixed and the effect of most research and development (R&D) on
industry structure is not planned. Over time petrochemical clusters have developed
that are characterised by interconnectivity, interdependence and inflexibility12
(Kuipers 1999).
With the maturation of the industry (see Cadre 1-2), changes or extensions of the
industry's structure have become incidental consequences of successful fundamental
R&D. Thus it may be seen that one cannot be sure that the largely ‘bottom-up’
approach employed in the scientific community will provide sufficient and timely
solutions to support sustainability of the petrochemical industry.

12 Each chemical plant in a particular cluster may be seen as the result of an individual company's
strategy. Each company innovates its slate of technologies employed and its organisation for efficient
realisation of large-scale facilities. Once completed, these represent huge sunk-costs, have very long
technical lifespan, and require large number of highly trained and skilled personnel. The facilities are
'sticky', and while expansions or additions to the cluster may cease sometime, existing facilities remain
in operation ofter for decades before closure. Thus, the dynamics-of-change for the cluster are slow.
Introduction 33

We conjecture that in order to prepare for the future of the petrochemical industry,
integrated innovation at all system levels (Table 1.3) is required, from material to
apparatus to plant to petrochemical complex and industry configuration.
Furthermore, the industry system structure must be considered a degree-of-freedom
in the search and specification of yet unknown technological developments, as well
as in the adoption of extra-sector innovations.
Our hypothesis is that there exists scope for improvement of the petrochemical
industry by process innovation, not withstanding the dramatic improvement of the
chemical industry’s performance with respect to economic and ecological efficiency.
Secondly, our proposition is that this scope for improvement cannot be reaped by
progress in the existing disciplines of chemistry, physics and engineering alone, but
rather that a large source of innovation remains unexplored: process system
innovations.

1.5.4 RESEARCH QUESTIONS

It is the general belief of many an environmentalist and policy-maker that the scope
for improvement within the petrochemical industry is substantial. Meanwhile, the
chemical industry appears to be entrenched in accepted, accredited methods of
production that are advocated and supported by its management.
The problem, thus is twofold: how to assess the scope, and subsequently, how to
arrive at a specification or identification of options for process innovation that go
beyond the normal trajectories of innovation in the petrochemical industry and that
merits the allocation of substantial R&D capacity. This leads to three research
questions:
(1) is it possible to develop methods for the assessment of the resource
utilisation and the scope for improvement of the production systems of the
petrochemical industry that are objective, systemic, effective, versatile and
data-proof?
(2) can the search of the innovation space be structured to foster the
specification of process system innovation content - is process system
innovation by design feasible?

The research by Pavitt (1984) demonstrated that traditionally innovation has come
from within the sector, whilst presently the capacity for R&D within the sector is
reduced. This dilemma of stimulation of process innovation in a mature industry has
been explored from a technological perspective. This has been addressed under the
third research question:
(3) what is the usefulness and scope of application of the methods developed?
Do the methods result in innovative concepts at industry, complex or plant
level? Is the application of outside sector innovations supported or
facilitated?

1.5.5 RELEVANCE
In the past, systems in the industry , such as entire chemical plants or steam cracker
complexes (see Table 1.3, p.17), were dramatically improved. Presently, however, the
34 Process System Innovation by Design

scope for improvement within the system boundary of many petrochemical

processes appears to be limited by the absence of major breakthroughs. With the
maturation of the industry, the added value and the adoption of process innovations
depend more and more on the industry system structure. However, there is not a
single stakeholder that decides on innovation or evolution of the industry system
structure. Moreover, at the level of the industry opportunities for system
improvement are hard to elucidate because of its structural and technological
complexity. Thus, the invention or specification of process system innovations is not
straightforward. This thesis is an attempt to address this problem, with the intent to
help all stakeholders involved to bring-into-being yet unknown process system
innovations!

1.6 Research approach and structure of thesis

1.6.1 META-MODEL OF RESEARCH ACTIVITIES

The methodology for 'process system innovation by design' (Ch.3) has been based
on the body-of-knowledge of system theory, thermodynamics, process system
engineering and computer systems modelling, respectively.
An overview of the body-of-knowledge of system theory can be obtained from (e.g.
Boulding 1956; Truxal 1972; Asbjørnsen 1992; Blanchard and Fabrycky 1998). On
computer systems modelling notably (Baylin, 1990) has been used. A useful
introduction to applied thermodynamics comprises (e.g. Denbigh 1956; Seader 1982;
Kotas 1985; Smith et al. 2001). An overview of relevant process system engineering
knowledge is offered by (e.g. Rudd and Watson 1968; Douglas 1988; Siirola 1996;
Seider et al. 1999; Grossmann and Westerberg 2000).
The methodology has been specified and refined via case studies (Ch. 4, 5 and 6)
that cover a range of system aggregation levels. Whilst its actual development
required many iterations and adjustments, often along erratic trajectories, the
activities involved and results reported can be represented in a meta-model (Figure
1-8) that matches the classic framework for conducting scientific research. Thus,
further to the research theme and questions formulated, the 'meta' question
addressed is whether it is possible and worthwhile to develop, underpin, test and
validate some systemic innovation procedure that enables 'top-down' process system
innovation by design.
In the meta-model, a modelling strategy is required to allow proper set up of assessment
criteria and a reproducible assessment procedure of the industry or parts thereof. Its
completion must lead to increased understanding of system design and operation
and thereby provide starting points for the innovation procedure. Finally, the approach
can be subject to validation:: does the entire approach provide useful results and is its
scope sufficient? As indicated, the outcome of the validation may be fed back to the
various activities in the meta-model.
Introduction 35

5.2

1.1 Straightforward
Model
1.2 Modeling &
5.1 Decomposition
Strategy

1. Select / Develop

Model
4.1 1.2

5. Validation 4.5 4. Develop 2. Select / Define

Process System Innovation Assessment

Innovation for Feedback
procedure Criteria
Sustainability?

5.4

3.4 2.3
4.1. Systematic
Search R&D
options
3. Design / Compile 2.1 Losses,
4.2. Specification
of Innovation Efficiency
Assessment
Content
Procedure

3.1 Priority list

5.3 Weak Elements

x.y
Legenda: Activity Objective Result Info-label Information
from x to y

Figure 1-8: Meta-model of activities.

1.6.2 STRUCTURE OF THESIS

The work reported in this thesis is the result of two completions of the circle of
activities in the meta-model.
In chapter 2, an overview of innovation sources for the petrochemical industry is
given, and review of relevant process system engineering literature is presented.
Combined with other sources, this serves as basis for the method developed for the
specification of process system innovation content in the petrochemical industry.
In chapter 3 the methods developed for petrochemical complex assessment and the
specification of process system innovation content are presented and illustrated by
the case of industrial aromatics production and consumption. The first part of the
chapter represents the first completion of the meta-model activities, the second part
together with chapters 4, 5, and 6 represent the second completion with an emphasis
on illustration and validation.
36 Process System Innovation by Design

Chapter 4 is a case study into the worldwide system around olefins, where ethylene
ranks first, with global production capacity exceeding 100 Million Tons per annum.
Use of the methods developed enables the specification of process system
innovations.
Chapter 5 demonstrates the use of the method in investigating innovative use of fuel
cells in the chemical industry. Notably, the early evaluation of system concepts
developed is addressed. This is expanded into a treatise on the design of
trigeneration systems.
Chapter 6 is a case study into the development of an industrial cluster beyond the
traditional petrochemical industry. The advent of novel technology, again fuel cells,
is used to assess the impact on petrochemicals and the global platinum material
cycle. A number of process system innovations are specified.
Chapter 7 contains final conclusions and recommendations.
 2 Process System Innovation Sources

2.1 Introduction
The objective of our research was the development of a systemic approach to
specify the desired content of process system innovations, which “… are defined as
changes in the system structure or system design of the petrochemical industry, its
industrial complexes, or individual plants. These can be enabled by technological
inventions or vice versa” (Dijkema et al. 2003). In this chapter, the scope and
foundations are explored of such 'process system innovation by design'.
First the past and present sources for inventions relevant for the chemical industry
are reviewed. These range from basic science, chemistry and chemical engineering to
top-down system and policy studies. Since it is impossible to review this massive
literature base, an eclectic overview is presented. Thus, the usefulness and potential
is analysed of relevant knowledge domains.
Subsequently because process system engineering was identified as a source of
building blocks for said systemic approach, the vast body-of-knowledge of system
theory and process system engineering is explored to underpin and define the
concepts of 'system', system elements, system analysis and system design.
Third, results from a literature search are reported to unearth concepts and methods
useful for process system innovation by design. Specific attention is given to system
decomposition and to process synthesis.

2.2 Sources of innovations

2.2.1 OVERVIEW
Past innovations in the chemical industry may appear to be the result of a chaotic
process. In reality, the process of change and innovation is driven by public R&D
policy and funding, corporate R&D programmes to ensure continuity by sustained
competitiveness and by individual scientist's curiosity and preferences.
Scientific literature often only reveals the results of research programmes, projects
and initiatives. In addition, corporate R&D strategies and goals generally remain
beyond scrutiny, although some results do land in the literature or patents.
What follows, therefore is a structured overview of 'innovation sources' classified in
categories relevant for process system innovation in the petrochemical industry.
Apart from serendipity, innovations relevant for the petrochemical industry may be
seen to originate or emerge from
1. basic scientific developments
Virtually all disciplines of the natural sciences and the engineering sciences may
yield inventions that allow conceptualisation of new unit operations,
improvement of existing technologies and realisation of novel reactions towards
new or existing products. Within specific scientific or application domains,
exploratory research attempts to go beyond what is already known.
38 Process System Innovation by Design

2. R&D in chemistry, chemical engineering and related disciplines

In general, more focused R&D provides a path towards utilisation of scientific
knowledge. Common denominators are the focus on development and
realisation of new technology, the creation of novel realisations of unit
operations and new products. Much R&D is directed to improve known
technology or products. As a matter of course, all sub disciplines of chemical
engineering and related engineering disciplines play an important role.
Process system engineering (PSE) is a body-of-knowledge to bring industrial
process plants into being, which requires a system focus to integrate information
and knowledge from a variety of engineering (sub) disciplines. In a
Schumpeterian view on innovation, each combination or plant design is an
invention, and when built, an innovation.

3. System, policy or management studies

The (petro) chemical industry contributes significantly to GNP in many
countries. It also has a history of serious environmental impact and is
characterised by oligopoly. To no surprise, therefore it has been the subject of
many a policy and management study. In addition to elucidating incentives and
barriers for innovation, these may represent valuable sources of systemic
innovations for the sector.

2.2.2 BOTTOM-UP: CHEMISTRY AND CHEMICAL ENGINEERING

The petrochemical industry is a ‘science-based’ industry (Pavitt 1984). Historically,
R&D expenses accrue to 3 to 4 percent of sales, and a large number of patents
administrated are claimed by the chemical industry. As stated, traditionally much
R&D is focused on process technology, and a small improvement can have a
marked effect on processing cost (Wei et al. 1979; Brennan 1998).
The first companies that were involved in chemical operations in the 17th century,
however, were small, and did not exploit scientific knowledge. The modern
profession of the chemist developed around 1840. A major process innovation,
however, had then already been developed: the Leblanc process for the production
of alkali, which replaced the traditional potash process that used timber.
Subsequently, the Solvay process succeeded the Leblanc process. The first synthetic
dye was developed in 1856 (Freshwater 1997), which led to the foundation of the
Badische Anilin und Soda Fabrik, or BASF.
Today, chemistry, amongst other scientific disciplines such as physics and biology,
allows us to understand how both the living and non-living world functions at a
fundamental (sub-microscopic) level. Chemistry is focused on the micro-scale of
molecular structures and change involving chemical reactions, for example the
development of synthesis paths, catalysts, membranes for separation etc. Notably the
co-evolution of chemical engineering together with chemistry has proven to be an
enormous source of innovation.
Technical chemistry provides the toolkit for the development of new (chemical)
products, the elucidation of reaction mechanisms on a sub-microscopic scale, and
the basic phenomena for separation equipment. Also in the petrochemical industry
Process System Innovation Sources 39

chemistry plays a pivotal role, but it is chemical engineering science that has shaped
the industry through the development of catalyst systems, chemical reactors,
separation equipment, and integrated continuous process designs.
Innovation of chemical factories at first proceeded at a slow pace. The products
developed by 17-19th century chemists were directly put to work in simple chemical
factories. From 1880 onwards the development of chemical processes and plants
rapidly evolved ‘from art to science’. In Europe and the US chemical engineering
programs were started, and in 1923 chemical engineering was firmly established by
the publication of the “classic text ‘Principles of Chemical Engineering’ (Walker and
Lewis, 1923)” (cited in Freshwater 1997). Academic chemical (process) engineering
science was rapidly recognized as essential to carry out chemical processes safely and
competitively. Its application for the transformation of laboratory-scale synthesis
routes to industrial scale facilities became common practice.
Today, the Institution of Chemical Engineers defines chemical engineering as 'the
design, development and management of a wide and varied spectrum of industrial
processes' (IChemE 2003). In chemical engineering, methods and tools have been
developed for scientific underpinning of the understanding, analysis and design of
chemical operations at industrial scale. Continuously operated systems were
developed, initially in crude oil refining, ammonia production, sulphuric acid and
other processes. Today, chemical engineering has developed into a myriad of sub
disciplines which range from the microscopic level of underlying transport
phenomena in reactors, to the macroscopic level of equipment and plant
specification. The Ullmann encyclopaedia of chemical technology (Anonymous
1995b) now comprises 26 volumes, and hundreds of journals have been established
on every sub discipline of chemical engineering.
One fundamental concept of chemical engineering is that any process plant is
considered a structured collection of unit operations, the first paradigm for the
discipline (NRC 1988). Sub disciplines of chemical engineering focus on single
classes, or even subclasses, of unit operations, such as reactors or separation
systems, which have specific links with their underlying basic sciences. Examination
of unit operations from a more fundamental point of view became a second
paradigm for the discipline. In addition, some unit operations remain largely the
realm of mechanical engineers, e.g. 'rotating equipment' such as compressors and
pumps, whilst other such as industrial furnace and heat-exchanger design require
specialization and are increasingly supported by Computational Fluid Dynamics
models.
In the NRC report one distinguished between the microscale, mesoscale and
macroscale. At the microscale, among other molecular reactions and heat and mass
transfer phenomena are studied. At the mesoscale, the selection of a particular set of
unit operations, the specification of each unit operation and the design of the
structure between the unit operations is the realm of process design.
Conceptual process system design involves the analysis and anticipation of process
plant performance, as well as process synthesis to meet plant design objectives in a
sound and safe combination of unit operations. At the macroscale, entire plants
must be developed that satisfy the needs and requirements of a complex
environment that may require integration of multiple plants and synthesis of supply-
40 Process System Innovation by Design

chains, enterprise modelling and life-cycle models (NRC 1988; Li and Kraslawski
2004).
Notably the conceptual design of chemical plants was established as a topical sub
discipline after the seminal publication by Douglas, who put forward a structured
method to arrive at a design of a chemical plant (Douglas 1988) based on and
bundling the experience of generations of process design engineers. Today
conceptual process design is part of the body-of-knowledge of process system
engineering, which is elaborated in section 2.4.

2.2.3 TOP-DOWN: SYSTEM AND POLICY STUDIES

Studies relating to the entire petrochemical industry fall into the following categories:

1. Studies on the development of the industry

Academic research with relevance for the entire petrochemical industry includes
studies by industrial historians, regional economists or economic geographers
on location, regional, national and sometimes continental or global development
of the chemical industry, its establishment and business. Important contributions
to the theory on location development for the petrochemical industry are (Weber
1909; Molle and Wever 1984; Witlox 1998; and Kuipers 1999). These studies
largely address the economic and ecologic drivers in the “surroundings” of the
petrochemical industry or parts thereof.
The technology and system structure are merely seen as enablers to achieve
economic or ecologic objectives; the global or local 'organisation' or structure of
the industry changes over time, however, as a result of the implemented
strategies. Weber postulated that processing of Verlustrohstoffe is most
economic at or close to the resource deposit, whilst Reinrohstoffe best be
processed near the end-user markets. Thus, the early location pattern of
refineries and petrochemical industry could be explained. Molle and Wever
(1984) took a descriptive approach “to picture the patterns of growth and
stagnation of one of Europe’s major industries“, viz. oil refining and
petrochemicals production, with a major emphasis on location adaptation of the
industry.
Of particular relevance is their verification of the extent of complex formation in
the petrochemical industry, which confirms regional development theory and
location theory that industries linked in a product chain tend to locate in one
another’s vicinity to save on transport costs. A complex is then a joint location
of linked industries that exchange feedstock and products. In their chapter
heading, this concept is labelled functional linkage, which is defined as
'connection of production units by feedstock and product streams'. Interesting
enough, they found that ‘vicinity linkage’ often is preferred over ‘concern
linkage’.
A range of specialised consultants also monitors the development of sectors or
sub sectors of the chemical industry, its establishment and business. More often
than not, however, their focus is on providing reliable data, while the breadth
and depth of their analysis is often limited (e.g. Anonymous 1995a).
Process System Innovation Sources 41

2. Studies that use economic Input/Output analysis

According to (Wei, 1979) the “very powerful bookkeeping technique of input-
output analysis“, hereafter I/O analysis, was developed by Leontieff in the 1930s
to account for inter-industry interaction, and to allow segregation of the value-
added contribution to GNP by industry sectors. I/O analysis provides a
mechanism to determine the specific contribution to, for example, GNP,
resource use and emissions of the various sub sectors of an economy. Thus,
inter-industry studies “are useful for both structural analysis and policy guidance
” (Chenery and Clark 1959: 7). The technique may also be used to assess the
performance of the industry or parts thereof by valuation of the I/O analysis
results. A specific example for the US chemical industry is reported by
Thompson (Thompson et al. 1978). In 1993, a Dutch Research consortium13
compiled input-output tables of the Dutch economy for the purpose of
monitoring material flows at a meso-level (Konijn 1994; Smits and Dijkema
1994). The results of these studies repeatedly emphasize the importance of the
chemical industry as a supplier to agriculture. The results appear to be useful to
set priorities for industry, environmental and innovation policy.

3. Studies that employ mathematical optimisation techniques

As early as 1951, Linear Programming (LP)14 was developed and used for activity
analysis in single plants, firms and in inter-industry economics (Chenery and
Clark 1959). The planning and optimisation of refinery- and petrochemical
complexes, however, has been largely the domain of Operations Research. LP-
models by process system engineers are usually limited to a single level of
aggregation, and limited to the modelling of some existing process network for
optimisation. Stadtherr and co-workers were the first to report the use of LP to
investigate the performance of the US chemical industry. (Stadtherr and Rudd
1976; 1978). The results of the studies initiated as early as 1972 were also
published in a book (Rudd et al. 1981). The objective of their work was
Defining the inherent technical structure within which the world-wide
petrochemical industry must function. The structure is formed by the
large but limited number of chemical syntheses that are available on the
commercial scale and by the rigid feedstock, by-product, and energy
requirements of these chemical syntheses. The products of one segment
of the industry becomes the feedstock for other segments of the
industry, thereby defining a network of material and energy flows that
constrains business activities.
Their assessment focused on giving insight into the questions “what are the
ultimate performance characteristics of the industry, in terms of feedstock,
energy, and investment capital use, (…) is it possible to anticipate the effects on
the industry of changing product lines and raw material resources” and “what are

13 The Dutch Office of Statistics, the CBS executed this study that was commissioned by the Ministry
of Vrom and RIVM. The study was completed with the help of DHV, TNO and Interduct, the Delft
University Clean Technology Institute.
14 The use of LP and related techniques in conceptual process system design is addressed in Ch. 5.
42 Process System Innovation by Design

the characteristics of new processes and products that lead to their eventual
adoption in the economy.” In order to answer these questions, they developed
and employed a linear, economic input/output model that was used to establish
the optimal structure of the petrochemical industry with respect to production
costs. Out of an extensive data catalogue on industrially applied processes, the
1977 (and 1940, 1950, 1960, 1970) structure of the industry could be
reconstructed. By changing the available technology catalogue, the impact of
novel processes becoming available was assessed. In all cases, the existing
industry proved to be a barrier to selection of novel processes, which was
modelled by applying capacity constraints to the model selections.
We adapted their model to investigate the use of novel technologies in existing
process networks (Dijkema et al. 1997). Optimisation of C4 networks in
retrospect confirmed maturity of this part of the industry. We concluded that the
existing industrial process network has a large flexibility towards feedstock
economics, which implies that new processes must offer significant advantage to
overcome this barrier to implementation. These kind of studies appeared
particularly useful to establish the scope for innovation and for some
prioritisation of R&D efforts.
4. Scenario studies to elucidate the need for resource and energy conservation.
The work reported by the ‘Club of Rome’ (1972) was the first to address the
resource situation of our global industrial society using the system dynamics
modelling approach. The energy sector, base metals, and the petrochemical
industry were explicitly recognized and modelled as an important sector
(Meadows et al. 1972). The Club of Rome warned for major resources becoming
unavailable for society in the foreseeable future. In contrast with common belief,
however, the report did not state a definite horizon but merely pointed out that
the modelling exercises gave similar scenario patterns over a very wide range of
parameter for birth rates, economic-growth, welfare etc. The authors conclude
that a drastic change in lifestyle would be needed for sustainability. Partly
responding to criticism on the original model, the authors published the results
of an improved global model in 1992 (Meadows et al. 1992), where they argued
that the model supported that a status of sustainability at the time was still
achievable, however that the 'escape'-window had become much smaller, which
implied more drastic adaptations to our global society were needed.
Governments and major oil companies have developed their own scenarios for
world energy developments to help developing business strategies. In the
Netherlands, the energy research centre ECN has developed various scenario
models, the most well-known “Nederland in Drievoud” (Gielen 1999).
5. Sector-wide studies to find general means of improvement of resource and
energy efficiency
In the Netherlands, around the time of the Club of Rome report, a first Cabinet
letter on the environment was issued (VROM 1972). Not much later results of
sector wide policy studies on efficiency improvement were published. Examples
of studies that addressed energy improvement potential are (Boot and van Wees
1982; Melman et al. 1991; Worrell 1994). In all of these studies feedstock supply
and sector structure are taken for granted, and the possibilities of improvements
Process System Innovation Sources 43

are addressed largely at a technological level. An exception is the report issued by

Quakernaat et al., who proposed a dramatic strategy: to initiate a complete
transition of the energy sector to use hydrogen derived from coal. This would
open the way towards development and use of solar hydrogen plants. It was
suggested to locate these e.g. in the Sahara desert and connect Western Europe
by pipeline (Quakernaat 1981). Boot and van Wees (1982) were the first in the
Netherlands to investigate the utilisation of fuel in the Dutch industry. From the
database on industrial emissions they computed an estimation of the net use of
heat at various temperature levels in the industry. Their basic intent was to
investigate the possibilities for cogeneration capacity expansion. Others have
reported attempts to identify so-called 'cross-cutting' technologies, that would be
applicable in many sectors of industry to achieve resource and energy
conservation (e.g. Melman et al. 1991; Worrell 1994).
6. Studies with an emphasis on environmental impact
Notably Life-Cycle-Analyses fall into this category. Early LCA- studies were
primarily product oriented and not so much process oriented. An exception is
the work done by CE and others on the evaluation of processing routes for
mixed plastic waste. (Dijkema and Stougie 1994; Sas et al. 1994a). In order to
elucidate the effects of a particular product, often large parts of the industries
involved have to be accounted for.
Through journal and Setac publications LCA has evolved into Lifecycle
Inventory (LCI) followed by Life Cycle Assessment, LCA. Not only emissions
and waste discarded into the environment are accounted, but also the net
withdrawal of resources. The valuation of these inputs and outputs, however,
remains a controversial issue. Objective allocation of emissions and withdrawals
in networked production systems is difficult. In case LCAs are conducted as part
of the formulation or evaluation of public policy and regulation, valuation and
allocation often are the result of negotiation between stakeholders involved
(Dijkema and Stougie 1994; Sas et al. 1994a; Bretz and Fankhauser 1997; Bras-
Klapwijk 2001). In addition, on the basis of interviews in eight industry sectors,
we concluded that only few LCAs are completed, primarily because the benefits
of the exercise do not justify the cost (Dijkema and Mayer 2001b). Product
LCAs, however, can yield insights into which products to innovate or where
production systems must be improved. Too often, however, the interconnection
between systems, mutually dependent performance, and technological content
and constraints are neglected in formulating solutions (Reuter and Dijkema
2003).
7. Studies with a policy or management focus.
Stobaugh investigated and analysed the mechanisms of the development of
global petrochemical business largely for an industry perspective: how-to ensure
continued success in this business (Stobaugh 1988). One element, for example, is
vertical integration, which was taken one step further in the concept of
Integrated-substance-chain management (VNCI 1989). The latter developed
from the notion that all outputs from an industrial system must originate from
some input. It may be seen as an early attempt to develop some system
perspective that combines system surroundings (economy, ecology), structure
44 Process System Innovation by Design

(substance-chain) and organisation and management. Based on results of a

number of consultant studies into the performance of chemical companies
Whitehead argues that it is not portfolio restructuring through (de) mergers and
acquisition, nor cost cutting and reorganisation, but value creation that should be
the objective of any company's strategy. Innovation and innovativeness are the
most important ingredients (Whitehead 2000).
8. Studies with a holistic character
Recently, the concepts of Industrial Ecology (Graedel and Allenby 1995; Ayres
and Ayres 1996) and Industrial Ecosystems (Allenby and Richards 1994) have
been developed to contribute to sustainable development by mimicking the
functioning of nature’s ecosystems in industrial operations. Addressing the
aggregation levels beyond a single plant and cross-sectoral exchange, Industrial
Ecologists develop a body-of-knowledge on how to use analogies and examples
from natural ecosystems for the analysis and conceptual design of industrial
networks and industrial society at large.
An early example addressed by the industrial ecologists was the investigation of
inter-company networks (Frosch and Gallopoulos 1989). A much cited example
is the Kalundborg industrial estate in Denmark (Verhoef et al. 2004b). As has
been the experience of a great many (petro)chemical companies indeed the
nearby location of similar industrial operations can offer synergies by exchange
of mass and energy flows, utilities and other site services and infrastructure. This
has been used to explain the large degree of clustering and the very emergence of
petrochemical complexes (e.g. Molle and Wever 1984). Today, the availability of
such amenities is a conditio-sine-qua-non for many a petrochemical company
when confronted with the task of site-selection for new process plants (Dijkema
and Kuipers 2001). Whilst existing clusters continue to evolve, to bring-into-
being new clusters is tedious, amongst others because the demonstration and
allocation of the benefits of industrial cluster concepts is not trivial. . We
developed a decision support system based on I/O techniques to assess
industrial park performance (Dijkema and Stikkelman 1999). To support the
Regional Development agency, van Zanten constructed an I/O model of part of
the Rotterdam Chemical Cluster and applied Linear Programming to identify
potential candidates for improvement of the cluster (van Zanten and Dijkema
2002). Quantitative I/O analysis and Linear Programming to assess the
optimality of an industrial network from a variety of perspectives, however,
appears to be largely lacking in Industrial Ecology.
The need for closure of material cycles is another example that has been
advocated in the industrial ecology community (e.g. Ayres and Ayres 1996).
Modelling of heavy metals in Zn, Cu, Al, Si, Mn, and Mg production and their
interconnected material cycles appears natural as, except for nuclear reactions
and space travel, the chemical elements never vanish from the earth (Reuter
1998). Indeed, many scholars have addressed these metal cycles, and a large
inventory of Cu and Zn flows around the world has recently been completed
(Graedel 2003). Metal production, however, is not only characterised by
interconnectivity, but also by interdependence, and for many metals no primary
production exists, only co-production (Verhoef et al. 2004a). Presently, largely
Process System Innovation Sources 45

dissipative once-through supply-chains cater for consumer demand for short-

lived products and packaging. These span the metal industries, crude oil refining,
the petrochemical industry and polymer production (Figure 1-3). A recent survey
of the European Environmental Agency, for example, revealed that per capita
waste production in the European Union is 415 kg/yr and continues to rise
(Anonymous 2003b). Meanwhile, society has developed and depends on end-of-
pipe waste management infrastructure, land-filling and incineration (Dijkema et
al. 2000). The result is loss of material and loss of quality material.
In the nineties, sustainability has also been address from the perspective “how
much do we need to improve societies effectiveness in using resources, the
environment and use of natural ecosystems.” Factor-10 or Factor-20 have been
coined. The Factor 4 approach (von Weiszacker 1996) starts at the needs-side,
which leads to an orientation to product development: what concepts are
required in the future to satisfy human needs with a limited input of resources?
Therein it is implicitly assumed that change of the production systems will follow
because of the interest and commitment of parties involved. In the Netherlands,
a sustainable technology program ('DTO') was launched and executed. In the
program scenario methodology and back casting were combined to develop both
images of a sustainable future and a roadmap or multi-stakeholder process to get
there (Mulderink et al. 1997).
Both in Industrial Ecology and in holistic sustainability studies, such as
(Mulderink et al. 1997), the entrenchment caused by the presence of existing
production systems appears to be largely neglected (Verhoef et al. 2004a). In
addition, both approaches appear to suffer from lack-of-content (Boulding
1956). The analogies from ecology and the scenario/back casting approach,
however, do offer significant potential to develop useful concepts for innovation
towards sustainability, such as using the integration of recycling and production
systems in closed material cycles (Verhoef et al. 2004a).

2.2.4 ANALYSIS AND OBSERVATIONS

Generally speaking, holistic and system, policy and management approaches suffer
from lack-of-content (Boulding 1956). A few exceptions noted, in most cases a
largely descriptive approach is used to capture features of the industry, the
petrochemical industry or industrial complexes. A number of models have a macro-
economic orientation and use I/O analysis to predict performance of industry and
technology used. These subsequently provide a basis of descriptions of what a -
more- sustainable future should look like and what should be done to get there.
Only in a limited number of cases, however, ingredients with (technological)
substance or analysis of existing production systems are linked with the images
created. The analysis results do not relate to single plants or particular technology
because they do not include aspects that are required for meaningful assessment of
technology such as, for example, a sound thermodynamic basis. As a consequence,
the usability to formulate R&D strategies appears to be limited as there is no
coupling with ‘technological reality’.
On the other end of the spectrum, “Traditionally, chemical engineers have been
rather more micro-oriented in their engineering approach to the process industries,
46 Process System Innovation by Design

typically working in a plant or focused business area as a technical or technological

specialist” (Brennan 1998: 255). This even appears to be largely the case in process
system engineering, where many congress and journal papers appear that focus on
mathematically sophisticated solutions to problems of limited scope.
Thus, our literature survey indicates there is a gap between system and policy studies
centred around industrial sectors or beyond, and the work done in chemistry and
chemical engineering on products and industrial processes.
In publicly funded chemistry, chemical engineering and related disciplines the
selection of R&D focus is largely left to accredited specialists from academia who
focus funding on fundamentals at micro- and molecular scale. Adequate R&D
portfolio management across the various scales - from molecule to process to
industrial network to industry - to foster sustainable development appears to be
lacking. Industrial R&D addresses relevant product sectors and all relevant scale of
process engineering.
No methodology or recipe for the effectuation of sustainability in the large-scale
process industry presently exists. In the scientific community at large the system
design of the industry has been considered fixed. With the maturation of the
industry (Ch. 1), changes or extensions of the industry's structure have become
incidental consequences of successful fundamental R&D. Systemic innovations,
however, do require a long-time period for development and implementation
irrespective of the mechanism by which they have been identified. Thus it may be
seen that one cannot be sure that the largely ‘bottom-up’ approach employed in the
scientific community will provide sufficient and timely solutions to support a
transition to a sustainable petrochemical industry.
The gap identified represents a niche for 'systems thinking', notably process system
engineering to effectuate and combine industrial ecology concepts and policy
analysis and management theories.

2.3 System thinking foundations

2.3.1 OVERVIEW
This section firstly introduces (general) system theory and system engineering.
Subsequently we address and define 'system', 'system elements', system assessment,
design and analysis.

2.3.2 SYSTEM THEORY AND SYSTEM ENGINEERING

In many industrial sectors scientific, knowledge and advancement is turned into an
advantage through subsequent applied R&D and system engineering. Based upon
this recognition, general system theory was coined (Boulding 1956), and general system
engineering principles were developed that augment detailed subject knowledge by
offering methodologies to bring complex systems into being (Asbjørnsen 1992).
Contributions originate from aerospace and defence industries (e.g. Corrigan and
Kaufman 1965), the automobile industry, electronics industry, computer systems,
information, communication and telecom infrastructure (e.g. Blanchard and
Fabrycky 1998; Arnbak and Weber 1999).
Process System Innovation Sources 47

The chemical engineering community has been very successful in adopting system
concepts (Villermaux 1996). Process system engineering has emerged as a sub discipline
where methods for optimal design of chemical plants have been developed, both
with respect to the selection of technology content and of system structure.
According to Prof. Villermaux, it is the mere training of chemical engineering
students in systems thinking that allows them to be productive in so many sectors of
science, engineering and business (Villermaux 1996).
After the original publications of (Boulding 1956) and (von Bertalanffy 1951), system
theory has grown into a vast area with sub disciplines that address hard- vs. soft
systems; static and dynamic systems, system analysis and system design. (Corrigan
and Kaufman 1965: 29) describe the activities of system engineering as “a progressive
definition and integration of functions and design decisions at every level of system
complexity.”
Nelson addresses system engineering in terms of functional activities performed “in
the conduct of a system definition/design (Nelson 1972 216: vi). INCOSE has
defined system engineering as “an engineering discipline whose responsibility is
creating and executing an interdisciplinary process to ensure that the customer's and
the stakeholder's needs are satisfied in a high quality, trustworthy, cost efficient and
schedule compliant manner throughout a system's entire life cycle” (Incose 2004).
According to Asbjørnsen, “systems engineering and systems thinking are formalized
common sense”(Asbjørnsen 1995).
Blanchard & Fabrycky coined the phrase “bringing systems into being” as the
emphasis in the overall activity of systems engineering (Blanchard and Fabrycky
1998: xiii). System engineering thus includes the definition of needs, requirements
analysis, functional analysis, design synthesis and finally system evaluation. The latter
is the objective of systems analysis, which prime domain of use is the improvement
of systems that are already in existence.
(Hartmann and Kaplick 1990) also distinguish between system analysis and system
synthesis steps. According to Sargent, David Rippin would have described the
objective of the systems engineering approach “as the provision of a systematic
framework for tackling the solution of complex problems” (Sargent 1998). “Process
synthesis is the invention of conceptual chemical process designs” (Siirola 1996: 10),
and the question whether “the invention of chemical process designs can be
organised, systematised, or even automated” is at the focus of process synthesis
research. (Siirola 1996: 3).
From these definitions , we conjectured that system theory, and more specifically
(process) systems engineering would provide useful insight and procedures to
address complex industrial systems.
In our case, the system constitutes the whole petrochemical industry. Its present
structure is the result of a long process that by some has been labelled chaotic. In
addition, the utilisation phase of commercial chemical plants often is very long,
“much longer than the life of artefacts15 and artefact-making machinery” (Siirola

15An 'artifact' is “… an ornament, tool, or other object that is made by a human being …” [Sinclair,
1995 #59 :82]. By using this term, Siirola (1996) succinctly captures the differences and analogy
between the production of consumer goods (including intermediates) and capital goods. The output
48 Process System Innovation by Design

1996: 2). Indeed, the economic lifespan of many a chemical facility is extended over
and over again by multiple rejuvenation phases (Grievink 1994). It is to be expected,
therefore, that elements of the present petrochemical industry will be in existence for
a long time to come. Referring to industrial practice, Siirola hints at the relatively
scarcely addressed problem of having to generate flow sheets from scratch, as often
flow sheets from previously realised plants are available (Siirola 1996: 13). Genuinely
novel process system concepts, such as distributed chemical manufacturing (Rowe et
al. 1997) are scarcely reported. It may thus be seen that system engineering per se is out
of the question because it is focused on the technology acquisition phase (Blanchard
and Fabrycky 1998:13). A more realistic objective appears to be to provide a
contribution to system re-engineering or a re-direction of the system’s general
direction of development or transition towards sustainability. The initial activity then
comprises a thorough analysis and assessment of the present system such as
presented in chapter 3, to be followed by a creative but possibly systematised phase
during which novel, innovative system concepts are specified.

2.3.3 DEFINITION OF 'SYSTEM'

We have freely used the concept of ‘system’, which to some is natural, or even trivial.
Dictionary descriptions vary from “a system is a way of working, organizing, or
doing something which follows a fixed plan or set of rules (..) the English legal
system“, “a device or set of devices powered by electricity (…) a computer system“,
or “your system is your body’s organs and other parts that together perform
particular functions” (Sinclair 1995). Thus, the interpretation and use of the term
system exhibits many variations. How then to define ‘system’ unambiguously for the
purpose of devising a suitable modelling strategy for the petrochemical industry?
Nelson defines a system as “a combination of elements that work together to
perform a preconceived mission” (Nelson 1972). Blair and Whitston define a system
as “a set of objects, ideas, activities, or a combination thereof with a unifying
organization, and distinguish between ‘real-world’ systems (objects) and conceptual
systems (ideas). They mention the hierarchical nature of systems. In addition, they
state (Blair and Whitston 1971: 66) that a system can be “analytically described in
terms of (1) the attributes and functions of its individual unit parts and (2) various
relationships of these components with each other. The function or functions of the
overall system are the resultant of the component interactions.”
After quoting a dictionary definition “A system is an assemblage or combination of
elements or parts forming a complex or unitary whole,” Blanchard and Fabrycky
(1998) do not coin a definition of ‘system’, but. merely restrict themselves to a list of
features that distinguishes a system from a random set of objects: systems are
composed of components, attributes, and relationships, that together work toward a
common objective or purpose. “A common system function is that of altering
material, energy, or information” (Blanchard and Fabrycky 1998). In addition, the
hierarchical nature of systems is emphasized: every component or subsystem can be
broken down into smaller components, subsystems or system elements.

of continuous chemical process plants, mass production or the assembly facilities for consumer
products are artefacts, so are the production facilities.
Process System Innovation Sources 49

Using the foundations laid out by these and other authors, Asbjørnsen (1992) coined
a definition of system that appears to avoid ambiguity and to be suitable for the
purpose of devising a modelling strategy for the petrochemical industry. Thus, a
system is defined as “a structured assemblage of elements and subsystems,
which interact through interfaces. The interaction occurs between system
elements and between the system and its environment. The elements and their
interactions constitute a total system, which satisfies functional, operational, and
physical characteristics, as defined by the user and customer needs and requirements,
over a defined total system life cycle of the system existence, including the life cycle
of bringing the system into being” (Asbjørnsen 1992).

2.3.4 SYSTEM ELEMENTS

The lowest level of detail of a system structure consists of the system elements,
which Nelson defines as “fundamental building blocks that comprise a system”
(Nelson 1972). He indicated both hardware (equipments, facilities) as well as
software (skilled personnel, procedural data) as possible system elements. According
to Asbjørnsen (1992), the functional characteristics of a system element are technology-
free. The element is fully described by its inputs and outputs, irrespective of how the
relationship between inputs and outputs has come about, notably irrespective of
what technology (hardware) and procedures (software) bring about the
transformation between inputs and outputs.
This relates to the common ‘black-box’ approach in engineering, where the
treatment is concerned with what goes into the unit, what comes out of it, and what
happens to anything while inside, without being concerned with the details of how it
happens (Blair and Whitston 1971: 67). This very principle allows the engineer to
design, construct and control complex systems, as for example in plant design not
every detail of the operation of a pump needs to be known. In addition, gaps-in-
knowledge can be filled by establishing empirical correlations, proper assumptions,
and observing system behaviour.
Not only a system element or network node can be fully described by it’s inputs and
outputs, the same is true for the system. Thus it may be seen that input-output
characterisation can be performed at any aggregate system level. In the
petrochemical industry, for example, the levels suggested in (Table 1.2, p.16). It must
be noted that at each level the analysis can be performed irrespective of the further
decomposition selected of the system under analysis. As a corollary, whenever a
system is modelled as structured collection of system elements, the modelling result
is dependent on the decomposition selected!
In process system flow sheet simulation, for example, a process system is considered
a structured collection of unit operations, which are technology-free representations
of the process system' elements. Process flow sheet simulators are commonly
constructed around these unit operations. The user must define the inputs and
required output of the system, and the connection structure between unit operations.
The simulator then provides unit operation models and thermodynamic property
estimators to proceed from the input of a single unit operation to the outputs.
Finally, a solver routine ensures that all internal flows in the system will match.
50 Process System Innovation by Design

Another example is the successful use of the ‘black-box’ approach found in control
system design (e.g. Aström and Wittenmark 1984)), where only the black-box’
response to input changes and/or measured disturbances need to be known over the
range of operating conditions for most common types of controllers. Nothing of the
physical realisation of the black-box needs to be known, its contents are incognito.
The system can equally comprise an area inhabited by rabbits and foxes, a stirred-
tank, or a ship. The dynamic response patterns can be established by observation,
analysis or experiments.
In a ‘grey-box’ approach, knowledge about the contents of the box is used to
conjecture an expected dynamic transfer function, which is validated and calibrated
by experiments. In a ‘white-box’ approach, finally, a model is constructed from first
principles and the available knowledge on the box’s technological contents. It
follows that process flow sheet simulators follow a white-box approach at the level
of unit operations, as the input-outputs relations obey the thermodynamic models
thought applicable. The actual apparatus that would constitute the unit operation,
however, is still considered a black box.
The definition of system elements by Asbjørnsen provides a means to identify
analogies, and explains the impact of information systems engineering on the
development of general systems engineering theory. “Like materials and energy,
information is acquired, transmitted, stored, retrieved, and processed (...) Like
materials, information is processed by separating, combining or otherwise
transforming its parts. (...). Information flow and processing constitute, along with
flow and processing of materials and energy, fundamental aspects of systems
operation” (Blair and Whitston 1971: 28-33).
In our modelling strategy, therefore we have adopted the system concept as
phrased by Asbjørnsen, with the specific notion that the system’s lowest level
is an element that is only defined by its functional characteristics. This then
opens the way to develop new technological concepts to bring about the
transformation between inputs and outputs.

2.3.5 SYSTEM ASSESSMENT, SYSTEM ANALYSIS AND SYSTEM DESIGN

(Corrigan and Kaufman 1965 :19) define the process of 'system analysis' as “the
process of reducing or breaking down a whole [system] into an organization of parts
while relating the parts to each other and to the whole.” This definition relates to the
above definition of 'system'. Others, (e.g. Nauman 1988), perceive system analysis to
be a specific domain of applied mathematics.
To avoid this ambiguity, we prefer the term system assessment, which we define as a
combination of quantitative analysis to assess performance and qualitative analysis,
which includes both (sub)system boundary selection and system decomposition, to
understand structure and assignment of sub functions to understand the system's
purpose.
A number of authors quote the iterative nature of system engineering activity, where
system analysis precedes and follows upon system design (Corrigan and Kaufman
1965; Blair and Whitston 1971; Blanchard and Fabrycky 1998). This is best
illustrated in the General Design Paradigm that was first postulated by French in
Process System Innovation Sources 51

1971 (French 1999). An extended graphical version of this Paradigm is given in

Figure 2-1 (p. 55).
According to Asbjørnsen, in system engineering theory a technology-free specified
system element is the building block in system design. The functionality of each
block and the connection structure between blocks together yield the complete
system (Asbjørnsen 1992). As a corollary, in systems analysis it is the smallest item of
which the performance is assessed.
In process system engineering, 'process system design' generally is decomposed into
systems analysis and system synthesis (e.g. Douglas 1988; Hartmann and Kaplick
1990; Biegler et al. 1997). “System analysis is the formal decomposition of a process
into its constituents and their investigation with respect to the different hierarchic
levels. A term which actually means the same thing is process analysis, which is
understood as the scientific, technical and economic investigation of a chemical
process system. The purpose of such studies consists in finding bottlenecks,
showing possible improvements and extensions and finding measures for their
implementation” (Hartmann and Kaplick 1990: 17).
In process synthesis, the objective is to generate the optimal system design out of a
given set of available and tuneable building blocks. The synthesis thus involves
assembling alternative process configurations and final selection. As of late, process
synthesis techniques have been used for the development of new chemicals (e.g.
Marcoulaki and Kokossis 2000) and the importance of the relation between product
and process research has been emphasised (Charpentier 2002).

2.4 Process system engineering

2.4.1 OVERVIEW
Inventions may originate from basic science or R&D in chemistry, chemical
engineering and related disciplines and from studies with a system, policy or
management focus. The latter necessarily employ a meso- or macro-perspective
opposed to the micro-oriented approach in science and engineering. Any invention
becomes an innovation only when realized in an industrial plant or product.
Therefore, process system engineering is crucial for innovation in the petrochemical
industry.
Indeed, reflecting on and summarising the advancements of thirty years of process
synthesis efforts in PSE, Siirola “describes a framework for the industrial chemical
plant innovation process, showing how process synthesis fits into that structure and
how that framework has in turn influenced the development of systematic process
synthesis methods” (Siirola 1996: 2).
As a result of the work of and interaction among academia, the chemical and related
process industry and its knowledge providers, PSE has evolved as a sub discipline of
chemical engineering. Its focus is bringing into being chemical plants that exhibit
excellent operation characteristics and performance with respect to economy,
ecology and safety. The quality of a design, however, is not uniquely quantifiable.
There exists not “the quality“, but many quality factors, some of which can be
exactly quantified, while others lack quantitative precision. The result is a
“fingerprint” of quality factors of a design (Herder 1999).
52 Process System Innovation by Design

Over time, elaborate knowledge of chemistry, thermodynamics and technology has

been incorporated into a range of unit operation models. Sophisticated numerical
solver routines enable computations around complex systems that may incorporate
multiple recycles whilst respecting a variety of constraints. Thus, today PSE allows
ex-ante evaluation, synthesis and simulation of innovative plant designs. It's body-of-
knowledge enables us to establish the feasibility of new plant designs as well as the
potential of novel, presently technically or economically infeasible technology or unit
operations (e.g. Dijkema 1998).
Traditionally PSE is limited to analysis, synthesis and design of individual chemical
plants, which may involve selection and structures of unit operations (e.g. Smith
1995). According to Grossmann and Westerberg, however, “PSE is concerned with
the understanding and development of systematic procedures for the design and
operation of chemical process systems, ranging from microsystems to industrial-
scale, continuous and batch processes ” (Grossmann and Westerberg 2000: 1700).

Cadre 2-1: The CAPE community

In Europe, research groups on chemical process system design collaborate in the
Working Party on Computer Aided Process Engineering (CAPE) of the European
Federation of Chemical Engineering EFChE. For a number of years, this Working
Party has organised a symposium series, known as Escape as of 1992 (Grievink
and Schijndel 2002). Its equivalent in the United States is the Annual AIChE
symposium series. In the US every five years an Engineering Foundation
Symposium is organised respectively for process design (FOCAPD), process
control (CPC) and process operations (FOCAPO). Every three years a world-wide
symposium is organised - PSE-, which circulates over three continents -America,
Europe and Asia When landed in Europe, PSE is combined with the Escape
symposium.
Refereed contributions to these conferences have appeared in Computers and
Chemical Engineering, an Elsevier Science publication. In addition, from these
research communities a number of books have emerged, notably (Douglas 1988;
Hartmann and Kaplick 1990; Smith 1995, Anderson, 1996; Biegler et al. 1997; and
Seider et al. 1999), which were preceded by the founding work of Rudd and
Watson (Rudd and Watson 1968).
The most relevant themes addressed in the Escape-12 symposium were Integrated
Product and Process Design, Process Synthesis and Plant Design and Sustainable
CAPE education and careers for chemical engineers (Grievink and Schijndel
2002).

The body-of-knowledge of PSE has been developed over some thirty-five years
(Edgar et al. 1999). Today it includes (1) process systems analysis through mass and
energy balancing, second-law analysis and flow sheeting (Denbigh 1956; Rudd et al.
1973; Kotas 1985; Seider et al. 1999 respectively) (2) process system design
supported by elaborate design knowledge formalised in heuristics for conceptual
design (Sherwood 1963; Rudd and Watson 1968; Douglas 1988; Siirola 1996) and by
algorithms for process synthesis that have been developed for specific sub domains
Process System Innovation Sources 53

of chemical process system design such as heat exchanger networks, separation

networks and reactor networks (Floudas 1995; Westerberg and Wahnschafft 1996;
Grossmann et al. 2000) (3) Methods and algorithms for control, optimisation and
scheduling to enable and support the operation (Stephanopoulos 1984; Biegler et al.
1997; Edgar et al. 2001; Heijnen and Verwater-Lukszo 2003).
The emphasis of PSE efforts in R&D, however, is the support of design of chemical
plants using known (process) technology. Where Rudd and Watson advocate to start
from a primitive problem - 'an ill-defined statement of a need' (Rudd and Watson
1968: 4) -, at present often a known basic system design is used.
Referring to industrial practice, Siirola hints at the relatively scarcely addressed
problem of having to generate flow sheets from scratch, as often flow sheets from
previously realised plants are available (Siirola 1996). Genuinely novel process system
concepts, such as distributed chemical manufacturing (Rowe et al. 1997) are scarcely
reported. This matches the conservatism in the chemical industry, which can be
explained by its high-capital intensity and the associated risk. Common tactics are to
use proven concepts for novel chemical plants, to change these in small steps in
conceptual design, and to adopt a minimum-change standard in detailed design, no-
change during construction.
Despite the suggestions by Grossmann and Westerberg (Grossmann and
Westerberg 2000), in process system engineering work concerning process systems
that consist of multiple plants have been largely limited to optimisation, notably
(MINLP) approaches to the optimisation of refinery operation planning, and
investment selection, the work of Stadtherr and Rudd being one of the exceptions
(see section 2.2.3). Another subject area where considerable work has been reported
is the work on the synthesis of reactor and reactor/separation networks (e.g.
Papalexandri and Pistikopoulos 1996; Glasser and Hildebrandt 1997; Sargent 1998).
Ismail et al. reported some work on exploring some solution-space beyond existing
system concepts where process alternatives have not been pre-postulated (Ismail et
al. 2001).
Literature reports on the conceptual design of petrochemical complexes comprising
multiple plants are lacking, let-alone true multiscale approaches that link the
industrial complex, single plant and the unit operation level. With respect to what
can be known about corporate R&D, it appears that for a long time true process
system innovations were hardly known or recognized and have not been capitalised
upon in corporate R&D, which suspicion is confirmed by findings of a study into
DuPont's R&D (Hounshell and Smith 1989). Some successful 'dual-scale' process
system innovations are known, such as the optimisation of the Refinery-Steam
Cracker interface and the development of integrated unit operations for reactive-
distillation. Generally speaking, however, the potential of networked process system
innovations is a largely unexplored 'white spot' in PSE.
In the past decade, the inclusion of environmental impact minimisation into process
synthesis was addressed (Buxton et al. 1997; Yang and Shi 2000). The inclusion of
environmental criteria into the design process was elucidated (Herder 1999; Sharratt
1999). The inherent multicriteria design problem of pollution prevention and waste
management was addressed (Steffens et al. 1999; Mellor et al. 2002; Linke and
Kokossis 2004) as well as the incorporation of Life-Cycle Assessment into optimal
54 Process System Innovation by Design

chemical process system design (Azapagic 1999; Burgess and Brennan 2001; Khan et
al. 2001). Others incorporated engineering thermodynamics into process design and
optimisation and suggested thorough consideration of the available energy resources
and their conversion to useful work (Kalitventzeff et al. 2001; Bakshi 2002). While
the PSE community has realised that “The process industries are facing a range of
growing commercial pressures and challenges” (Bogle and Perris 1999), PSE
research contributions that address sustainability are relatively few in number.

2.4.2 PROCESS SYNTHESIS

“Process synthesis is the invention of conceptual chemical process designs” (Siirola
1996) and the selection of the better ones based on incomplete information
(Westerberg 1989).
The question whether “the invention of chemical process designs can be organised,
systematised, or even automated” is at the focus of process synthesis research.
(Siirola 1996: 10, 3). Daichendt and Grossmann, however, equate process synthesis
to conceptual process design (!) and describe it as being “concerned with the
identification of the best flow sheet structure (process system) to perform a given
task (Daichendt and Grossmann 1997). They distinguish two major approaches to
address the process synthesis problem: hierarchical decomposition developed by Douglas
and mathematical programming using Mixed Integer Non Linear Programming
(MINLP). In the former approach one appears to focus on conceptual design whilst
in the latter approach the value of rigorous simulation is emphasized.
Kaibel and Schoenmakers distinguish between the heuristic approach and the
superstructure/thermodynamic principles approach (Kaibel and Schoenmakers 2002).
According to Daichendt and Grossmann, Fonyo and Mizsey (1990) were among the
first to attempt and combine the two approaches, hierarchical decomposition and
mathematical programming, in a procedure starting with (1) hierarchical
decomposition, followed by (2) user-driven synthesis techniques to reduce the design
space and finally (3) rigorous profit calculations based on flow sheet simulations
(Fonyo and Mizsey 1990)16. Daichendt and Grossmann present an attempt to
combine the two via the construction of a high-level process system superstructure,
but restricted to a process system that has been (pre)decomposed hierarchically into
the subsystems reaction, separation and heat integration. They present a
computational procedure that includes a multi-level tree-search and algorithm for
selecting optimal process system options at the conceptual design stage, i.e. to
support step (1), hierarchical decomposition of the design problem into successive
levels of system detailing and system design solution. In such approaches, however,
the possible outcome is limited by the system boundary selected, by the type of
system element available for selection and modification, and by the original set of
alternative elements included. In addition, the solution selected may represent a
compromise to deal with multiple objectives.
Given the huge number of scientific papers on development and case studies using
methods of either category, both are well established in the CAPE17 community.

16 (cited in Daichendt and Grossmann 1997)

17 Computer Aided Process Engineering- see Cadre 2-1, p. 52.
Process System Innovation Sources 55

Need

Define Primitive
Problem
Analysis of
Problem
Gather Cognate
Facts

Statement Systematic
Create Specific
of Problem Generation
Problems
of superstructure

Screen Specific Evolutionary

Problems Conceptual
Design Modification
of superstructure
Feedback

Optimisation
Selected Structure and Design
Schemes
Critique / Evaluate
of Alternatives
Embodiment of
Schemes

Detailing

Working
Drawings
etc.

Rudd & Watson French Siirola

(1968) (1971) (1996)

Figure 2-1: Expansion of the General Design Paradigm for process synthesis (after Rudd
and Watson 1968; Siirola 1996; Herder 1999).
56 Process System Innovation by Design

(Kaibel and Schoenmakers 2002: 11) of BASF, however, comment that above
restrictions are disadvantages of both approaches. They recommend that “the
inclusion of a process in the company's process chain and the site conditions must
be specified. Finally, they state that “both methods are used in industry (…),
however, that neither of these methods is used regularly.” Indeed, Sargent reiterates
Rippin's concern “Will computer-aided synthesis of complete plants be used by
industry in due time (…) Are the methods too simple to cope with real industry
problems-or too complex for everyday use by industry engineers?” (Sargent 1998).
Notably Pistikopoulos et al. have addressed chemical process design, integrating four
domains of interests, viz. (a) the process system (b) the control (c) the operations -
static and dynamic optimisation thereof - (d) the production planning and scheduling
in a supply-chain (Bansal et al. 2000; Bansal et al. 2002; Ryu et al. 2004). For any pair
of domains {a,b}, {b,c}, {c,d} and {a,c} convincing studies have been published
concerning the feasibility of such approaches for realistic sections of a process plant.
Dealing with any of these domain-pairs for a full scale plant appears to be beyond
current computational capability. Similarly, whilst the concept is there, actually
dealing with all four domains for a realistic plant section is computationally too
demanding at present.
Kussi et al. of Bayer formulate a work-flow procedure for the effective use of
computer tools in the optimisation of existing chemical plants. They stipulate that
the interlinkage between the different types of modelling approaches and their tools
is a major obstacle to the effective use in industry of computer-aided process design
and optimisation. Whilst the vision has been coined to integrate simulation, design,
control and optimisation capabilities into a single model to support chemical
operations over a plant's life-cycle, they observe that rigorous modelling, black box
modelling and hybrid modelling are still 'living apart together' (Kussi et al. 2002).
Sherwood defined 'engineering design' as “the process of applying the various
techniques and scientific principles for the purpose of defining a device, a process,
or a system in sufficient detail to permit its physical realization” (Sherwood 1963). In
teaching design his approach was to continually remove biases suggested by the
terms used to get students to think outside the box (Westerberg 2004). “A
complete process design includes a critical review of the idea that there should be
any process, the invention or selection of an appropriate process (…).” (Sherwood
1963).
Rudd and Watson also emphasize the synthesis of plausible alternatives and
elaborate on generating ideas prior to engineering solutions. From an ill-defined,
primitive problem they suggest to arrive at a limited set of specific problems that can
be subject to engineering solutions (Rudd and Watson 1968). No use of
mathematical models is made in this phase; only relevant knowledge, information
and data is extracted from those concerned, experts and scientific information. This
approach fits with the General Design Paradigm (French 1999), which was expanded
by Siirola for the conceptual design or synthesis of chemical processes. The relations
are depicted in Figure 2-1. Siirola distinguishes four fundamental phases in a process
synthesis strategy, viz. (1) systematic generation, (2) evolutionary modification 3)
optimisation of structure and design parameters and (4) evaluation of best
alternatives with respect to criteria not in the optimisation problem (Siirola 1996). In
Process System Innovation Sources 57

this strategy, the first two phases have found both heuristic and algorithmic
implementations, the third is the natural domain of mathematical programming,
notably MINLP-techniques and the fourth stipulates to evaluate the mathematical
programming results in a wider context. Indeed, through the execution of every
design process, not only the particular design at hand may be improved, but also the
methods employed in each phase. Difficulties, however, arise when one seeks to
assess the quality of a particular design (Herder and Weijnen 1999).
Chemical processes can be modelled as objects that transform materials from an
initial state to a goal state. Usually, property differences are observed between raw
materials, products and intermediates. According to Siirola, the purpose of “an
industrial chemical process is to apply technologies in such sequence that these
property differences are systematically eliminated and the raw materials thereby
become transformed to the desired product” (Siirola 1996). With regard to process
synthesis steps, Seider et al. appear to define the main objective in chemical process
design as to arrive at elimination of property differences (between input,
intermediate and output materials in the process). The correction of such difference
can be seen to be effected by operators, the essence of means-ends analysis (Simon
1969). According to Seider, such formal, logic based strategies for process synthesis,
means-end analysed have only been developed for simple processes (Seider et al.
1999: 44). Meanwhile, the informal or heuristic approach initially introduced by
Rudd has been applied widely (Rudd et al. 1973). It involves the rules for structuring
process plant operations as indicated in Table 2.1.
Table 2.1: Process synthesis procedure (adapted from Seider et al. 1999: 44);
underlining added; original italics
Synthesis Step Process Operations
1 eliminate difference in molecular types Chemical Reactions
2 distribute the chemicals by matching sources and Mixing
sinks
3 eliminate differences in composition Separation
4 eliminate differences in temperature, pressure and T, P, phase change
phase
5 integrate tasks, i.e. combine operations into unit
processes
These steps appear in recommended order of execution (ibid: 112).

Elaborating upon means-ends analysis, Siirola distinguishes between synthetic,

forward and opportunistic directed operators used in engineering, in contrast with
the retro synthetic or backward oriented operators (from desired products) used by
research chemists. He points out that problems emerge when property-changing
operators can only be applied within a pre-specified range for other properties,
which creates sub problems, and which explains the natural hierarchy found in
successful design. Means-ends analysis, however, is recursive but not iterative, which
creates problems with process recycles for mass and energy as commonly exploited
in chemical process plants.
58 Process System Innovation by Design

Case-based reasoning (CBR) is an algorithmic technique to capture past experience

and appears to augment heuristics by reusing past solutions to new problems, albeit
in a limited problem and solution space (Avramenko et al. 2002). According to
Avramenko et al., establishment of similarity is a main problem. In their case-based
reasoning for decision support of the heuristic approach to conceptual design of a
reactive distillation unit they use symbolic data divided into subclasses simple,
structuring and classifying. Their representation of the CBR process can be
expanded to represent the functional modelling process for expansion of the
solution space.
In his functional approach to process synthesis, after (Kondili et al. 1993) Sargent
suggests to represent a “process” as a “state-task-network,” and to equate each unit
operation to a task or device “for transforming material from one state to another”
(Sargent 1998). Conceptual process design to effect a required transformation then
implies making choices from a finite and known collection of chemical reactions and
technology that can be interlinked in only a finite number of ways: a closed
combinatorial problem of considerable dimension. He subsequently points out that
in this solution space many combinations must be excluded because there is strong
linkage of decisions and because of the hierarchical nature of design decisions (after
Douglas 1988), with “broad choices at one level devolving into subsidiary choices at
lower levels.”
Sargent strongly advocates to make the choices in whatever design approach explicit,
and suggests the use of two type of models. Performance models define the relationship
between inputs and outputs for the task's range of feasible operation, and thus must
inform the designer whether a process or task is fit-for-purpose, feasible and meets
additionally the requirements. Value models assign a net “value” (Sargent's quotations
marks) to the execution of each task. It allows to choose between alternative tasks
that perform the same function. Both types of models can be very simple in early
phases of design, or extensive for use in rigorous simulation during detailed design,
but Sargent points out that to allow their use in subsequent design phases, the
confidence regions of more refined models should be fully contained in those of
simpler models, an objective which is hard to realise in practice (Sargent 1998).
In a state-task network there must be a connected path from each feedstock to some
product and vice-versa, and each intermediate state must be on one such path. Using
these rules and models, for any given problem all possible state-task-network
realisations can be enumerated, evaluated and possibly subjected to more refined
analysis. Based on a case study of distillation network synthesis, Sargent concludes
that Douglas' hierarchical approach can be embedded in a rigorous enumeration
procedure, which provides the option to select refinement and accuracy of the
models used as required, and that feasible networks can be generated by defining a
set of elementary tasks and assigning a purpose or function to each of these. The
more tasks that are available for use, the greater the variety of flow sheets generated!
Seider et al. discern among process invention, algorithmic process synthesis and
detailed design including design optimisation (Seider et al. 1999: v). They note that
design problems are open-ended and that “design is the most creative of engineering
activities, with many opportunities to invent imaginative new processes. It is also the
essence of engineering, differentiating an engineer from a scientist” (Seider et al.
Process System Innovation Sources 59

1999: 6). In process invention, the focus is on process creation, i.e. the formulation
and preliminary analysis of a promising base case design. The importance of
simulation tools and heuristics in process analysis and synthesis respectively is
emphasized. Whilst process invention is not defined, the goal of the procedures
outlined appears to be to create an attractive, economically feasible, environmentally
sound and safe process design package. This is in accordance with the
Schumpeterian definition of innovation (Ch. 1). In this view process synthesis
activities lead to innovation when the object of the synthesis activities, the process
plant, is realised. Process synthesis itself, however, is also the subject of innovation:
each novel method of conceptual process creation, if adopted in the community,
represents an innovation.
Beyond single plants, a vast body of process synthesis research concerns the design
and optimisation of energy-exchange or utility networks. The pinch analysis ideas
were pioneered by Hohmann and rediscovered by Linnhoff in 1978, who
subsequently pursued its adoption by industry (Hohmann 1971; Linnhoff 1993). A
massive body-of-knowledge has been developed on the optimum exchange of heat -
using temperature level and pinch based decision methods - and later the exchange
of water, hydrogen and solvents - using purity level and pinch. On an alternate,
applied thermodynamics and optimisation track, exergy analysis and exergy-based
optimisation of energy systems were developed (e.g. Kotas 1985; Yee and
Grossmann 1990; Yee et al. 1990; Cornelissen 1997).
In a recent review paper, however, Li and Kraslawski conclude that the picture of
conceptual process synthesis given in (NRC 1988) is still current (Li and Kraslawski
2004). They argue that process synthesis “at the meso-scale has attained a high
degree of scientific maturation” and that “a strong demand for new, innovative
products and processes has attracted a lot of attention to the phenomena at the
microscopic level”.
Despite the suggestions by Grossmann and Westerberg (Grossmann and
Westerberg 2000), however, process synthesis research at the macro-scale and
conscious analysis or re-design of the industry system structure beyond single plants
has only attracted limited attention.

2.5 System decomposition

2.5.1 OVERVIEW
In the overview of system engineering and PSE (section 2.1) definitions have been
given of 'system', 'system element', 'system engineering', and 'system assessment',
'system analysis' and 'system design'. These appear to be sufficient for a modelling
strategy that supports process system innovation of the petrochemical industry.
What is missing, however, is a conscious decision process on selection of the system
boundaries, formulation of the boundary conditions for inputs and outputs, and
system decomposition.
Since the global petrochemical industry has never been designed as a single system,
difficulties can be expected in decomposition. What is a suitable system boundary
and aggregation level for the elements of the system studied? How to break down
60 Process System Innovation by Design

the petrochemical industry system into progressively smaller units, so that the model
meets the objectives stated?
Difficulties in devising a conceptual classification scheme or typology of systems are
encountered in many sciences. In summing up the problems encountered when
addressing world agriculture, for example, Grigg mentions that criteria for
assignment to a certain class must both be developed and tested (Grigg 1974).
Reviewing the system engineering literature and the process system engineering
literature, we have searched for elements and theories that would help develop a
solution for this problem. Texts on system engineering address the topic of system
boundary selection and system decomposition in general terms only, reiterating that
in system design it is not only useful to discern between systems and system
elements, but also to group system elements into subsystems that together constitute
the total system (e.g. Blair and Whitston 1971; Blanchard and Fabrycky 1998). Where
(Corrigan and Kaufman 1965) use the analogy of the magnification setting of a
microscope as a means to increase the level of detail of analysis, Asbjørnsen quotes
the familiar top-down decomposition in trees, which says that each system can be
decomposed into ever smaller level of detail (Asbjørnsen 1992). No indication is
given, however, on how to achieve an optimum magnification-setting, when to
abstained from opening increasingly smaller boxes or when the system content has
become known in sufficient detail. In other words, no stop-criterion is given for the
decomposition-depth.

2.5.2 PROCESS SYSTEM ENGINEERING

In the process systems engineering literature, decomposition more often than not is
addressed in general terms only. Hartmann and Kaplick for example, state that “The
decomposition and aggregation (…) are decisive in the methodical consideration of
the objects under study” but hardly address decomposition strategies of existing
systems” (Hartmann and Kaplick 1990: 14). The exception, of course is hierarchical
decomposition by Douglas (1988), which is an established approach to break down
the conceptual process design and conceptual design procedure. In general,
however, it appears that “pragmatism prevails over principle” (Grigg 1974). In flow
sheet simulation for process design, the unit operation concept has been adopted
(Ch. 1). In a well-known cost-estimation method for novel plants a process system is
decomposed in functional units (Zevnik and Buchanan 1963). Where Hartmann and
Kaplick introduce the process stage, Douglas uses the concept of a process unit
(Douglas 1988; Hartmann and Kaplick 1990). In order to get an up-to- date
impression of the work done in the CAPE community (Ch. 1, Cadre 1-2), we have
searched the Escape symposium series (Pierucci 2000; Grievink and Schijndel 2002),
as well as the Web-of-Science with respect to literature on decomposition methods.
In line with earlier observations, (Dijkema and Reuter 1999), it appeared that system
decomposition is only addressed sparsely, superficially or implicitly in the
symposium proceedings and other publications. No clear-cut guidelines are given
when to label a set of apparatus as a single piece of equipment, a unit operation, a
process stage or a process unit and subdivision of processes or grouping of
subsystems by use of these concepts appears to be largely anecdotic. Douglas,
however, prescribes a procedure for the decomposition of existing chemical
processes with the objective to modify the subsystems, without limiting the solution
Process System Innovation Sources 61

space of the decomposition to arrangements of unit operations (Douglas 1988: 17).

Biegler et al. address the decomposition of decisions for process design, with the
objective to limit the solution space for the process synthesis problem (Biegler et al.
1997). This is the commonality of all literature cited: the system modelling strategy is
developed for a relatively clearly defined domain of application, and solutions are
found in a fairly delineated space. Siirola appears to be among the few exceptions,
when addressing systematic generation of process system alternatives through
means-ends analysis (Siirola 1996: 13-15). In developing a computer implementation
of means-end analysis, a set of useful rules are presented which resemble some form
of decomposition:
1. keep task identification distinct from task integration and equipment design.
“Form follows function.”
2. do not adhere inflexibly (to the property hierarchy)
3. be wary of convenience
These three rules developed by Siirola for conceptual process design are useful to
understand and guide the development of methods for specification of system
innovations because they (1) relate to the concept of 'technology-free' system
characterisation and specification, (2) prompt consideration of inflexibilities or lock-
in caused by current industrial practice and (3) stipulated to be wary of risk
avoidance, for example when accepting only minor deviations from proven
technology or concepts. Siirola's work is reassuring, as he gives an example of a real
process system innovation attributed to employing this methodology: a novel
process system concept for methyl acetate production. “It is not clear how such a
design might have been conceived without the explicit separation of task
identification, task integration and equipment design” (Siirola 1996: 28). Finally,
Siirola envisages a process synthesis paradigm of the future where existing
approaches -systematic generation of alternatives, evolutionary modification of
existing designs and structural optimisation- are combined whilst exploiting all
available methods (algorithmic, heuristic, mathematical programming), labelling it
“generate-evolve-optimise-critique.”

2.5.3 OTHER KNOWLEDGE DOMAINS

The literature reviewed for innovation sources (§2.2) was readdressed for system
decomposition methodology and theory. In economic input/output analysis, as
employed by economic statisticians, the system element has been selected on the
basis of the availability of data, and a logical classification of coherent economic
activities (CBS 1992). Thus, the I/O analysis yields a suitable picture of the
economic activities in a country. The economic statistics may also underpin
environmental policy making, e.g. by calculation of the cumulative consumption and
emissions for single types of products. To allow reliable use of the statistics,
however, implies a retrofit of the models and system decomposition. Konijn, for
example, used 'economic activity' - a subset of the industrial activities in a single SBI-
code - to disaggregate economic sectors and to correctly relate economic data and
physical reality (Konijn 1994). In the retrofit process it was revealed that parts of
coherent 'economic activities' had been included in the statistics of a variety of
62 Process System Innovation by Design

economic sectors. No specific treatise, however, was found on the problems of

decomposition and (re-use) of input/output analysis results.
In system thinking, the function of an apparatus, plant, or product is the prime
concern. In Life-Cycle-Assessment (LCA) this has been formalised in the
requirement for a clear definition of the functional unit to which the net resource
input and waste and emission generated in a production-consumption chain are to
be related. In such a way, the net balance of the alternative packaging of for example
a litre of milk can be compared for the cases of using a recycled glass bottle and a
carton respectively. The case for LCA is argued as that it should open the way for
(both product and process) innovation. The system elements in many an LCA study,
however, have been selected for the purpose of analysis of emissions to the
environment, and the net withdrawals (resource use) from that same environment, in
order to relate everything to a single functional unit, e.g. packaging. The formal
decomposition of products or systems subject to LCA appears not to have been
addressed. The solution space following an LCA is limited to either product
substitution, which implies replacement of the related production system, or
improvement of elements of said production system. It remains questionable,
however, whether the method offers useful information for process technology
innovation because LCA takes the system structure for granted at any level of
aggregation18.
In computer science, system decomposition is addressed under the heading of
“cohesion methods”, which according to Baylin is a well-known subject in computer
system design and programming literature (Baylin 1990: 349). The subject relates
primarily to how to arrive at a sophisticated and useful breakdown of (computer)
programs or information system. Cohesion thus assumes the system decomposition
is executed with the objective to arrive at a set of highly cohesive, loosely coupled set
of modules. A number of cohesion principles or types can be distinguished,
however. When cohesion is coincidental the system elements have no function, data or
procedure in common, i.e. the decomposition does not serve a prior purpose. When
a system decomposition exhibits logical cohesion, the elements perform certain logical
classes of functions or procedures. In a temporal cohesive decomposition the
modules or elements are related logically and by time of execution, but they do not
necessarily share a common procedure. A decomposition exhibits sequential cohesion
when no other relation exists than that output of one element provides input to the
next. Procedural cohesion is achieved when all system elements share a similar
procedural context, in other words when all elements exhibit similarity in how
activities are performed. In a communicational cohesive system the elements all operate
on the same set of data (object-orientation). Details and examples on these cohesion
methods are given in Appendix A. 3. Finally, a decomposition of a system's
functions exhibits functional cohesion when it suits to keep together related
elements, keeps apart unrelated elements while at the same time avoiding duplication

18 The improvement of ‘the refining of crude oil’, for example, requires improved and new system
elements and networked process system design. Its optimisation, modification or substitution for its
performance in a single product-chain will have an effect on all other product-chains that depend on
this operation.
Process System Innovation Sources 63

and conflict. Thus functional cohesion underpins an abstract, iterative method for
the conceptualisation of a system's functional structure19.

2.6 Conclusions
In academia presently a gap exists between studies on industrial sectors or
economies, and chemical R&D and engineering on products and industrial
processes. Integrated conceptual design of the petrochemical industry's systems
above the level of single plants is a largely unexplored 'white spot'. Conscious
attempts at structured or planned innovation for sustainability at and beyond the
level of single plants appear to be largely absent.
Process system engineering was identified as a resourceful body-of-knowledge to
bridge this gap and address the research question elaborated in this thesis: how can
technological or systemic innovation content be specified that enables a transition
towards a sustainable petrochemical industry? PSE, however, focuses on the
invention and realisation of single plants, and the central question addressed in
process synthesis research is whether “the invention of chemical process designs can
be organised, systematised, or even automated” (Siirola 1996: 3). In process synthesis
research indeed methods are being developed to automate some parts of process
invention. To date, however, the PSE-body-of-knowledge does not offer a
procedure to specify process system innovation content20. Despite the suggestions
by Grossmann and Westerberg (Grossmann and Westerberg 2000), conscious
analysis or re-design of the industry system structure beyond single plants has hardly
attracted attention.
A literature review confirmed that the conceptual design of petrochemical
complexes comprising multiple plants is a largely unexplored 'white spot' in process
synthesis R&D. Thus, it is no surprise that true multi-scale process synthesis
approaches that link the complex, single plant, unit operation and molecular level
appear only to have been suggested (Charpentier 2002). In process synthesis the
computer-based automation of the generation of and selection among design
alternatives receives much attention. It is the work on conscious articulation of
primitive problems into specific design problems (Rudd and Watson 1968; Douglas
1988) (Siirola 1996), however, that offer some building blocks for a methodology to
explore process system innovation.
From a limited literature-search focused on development of system simulation
models and tools for process system design and synthesis, we have only found
superficial coverage of the problem of devising suitable system decompositions. In
none of the literature sources is a stop-criterion given for the opening of the box, i.e.
the decomposition and the 'depth' of the model. Careful consideration of system
decomposition, however, is of utmost importance to avoid unwanted ‘lock-in’
situations that limit the applicability of the model (such as in an economic I/O
analysis and LCA). In PSE, the system decomposition of the industry adopted

19 Functional cohesion is one of the foundations of the method reported in §3.3 Functional
modelling for process system innovation.
20 These observations are largely based on the literature reviewed to unearth previous work on

methodologies useful for the specification of process system innovation content. This work is
presented in detail in Chapter 3, this thesis and has been published in part as (Dijkema et al. 2003).
64 Process System Innovation by Design

appears to be considered trivial and the aggregate level of single process plants taken
for granted. Cohesion methods, a topic in computer science, appear useful for
system decomposition and modelling. In academic R&D the industry system
structure is considered rather fixed and has remained largely unchallenged, except
for the notion that the tasks represented in this structure may be integrated (Douglas
1988; Siirola 1996). As a consequence, the search for improvements implicitly is
confined to the system boundary of the battery limits of individual plants, and the
potential of networked process system innovations has remained largely unexplored.
 3 Process System Innovation by Design

Somewhere, (…) between the specific that has no meaning and the general
that has no content there must be, for each purpose and at each level of
abstraction, an optimum degree of generality (Boulding 1956: 302).

For the sanity of the engineer and his chances of success in design, we
would prefer the simplest possible mathematical model (Truxal 1972).

3.1 Introduction
While 'sustainability' has been adopted as a strategic principle for business, it has
been difficult to translate this notion into guidelines or recipes for technology and
business development (e.g. Scholes 1999; Dijkema and Mayer 2001b).
We conjectured that a transition towards a sustainable petrochemical industry21
requires innovation at all system levels, from reaction path to process system, from
chemical plant to petrochemical complex or industry (Ch. 1). Such process system
innovation reaches beyond traditional R&D and beyond the optimisation of proven
technology, proven system concepts and established process networks.
The overview of innovation sources (Ch. 2) revealed that there presently exists a
'gap' between the aggregation level of the economy or its sectors, and the topics
addressed in chemistry, chemical engineering and process system engineering.
According to Brennan “The chemical engineering profession has thus far been slow
to develop a techno-economic taxonomy or classifications system, useful for
understanding the structure and performance of the process industry sectors”
(Brennan 1998: 255).
We conjectured that innovation opportunities must exist that have never been
explored systematically - process system innovations where explicitly the network
structure is considered a degree-of-freedom.
In this chapter, methods are presented to explore this 'innovation-space'. Thus, the
second part of this thesis' central research question is addressed: how can
technological or systemic innovation content be specified that enables a transition
towards a sustainable petrochemical industry?
The methods presented are based on system representation and characterisation
using the input-output paradigm (Appendix A. 4). It is assumed that the function of
a system, (Σ), is to generate a behaviour B to transform a set of inputs (Su) into a set
of outputs (Sy), which relate the system to its surroundings (the external world). The
performance (Ω) of a system is assessed by means of criteria or value functions (P).

21 A sustainable petrochemical industry is characterised by appropriate resource selection, effective

resource utilisation, the avoidance of waste and emissions and a product slate that fosters a
sustainable society (Dijkema et al. 2003).
66 Process System Innovation by Design

In section 3.2 the focus is on system modelling and characterisation for assessment
of resource utilisation, the main indicator for sustainability (see Cadre 3-3: Effective
resource utilisation). Input/output modelling, engineering thermodynamics and the
stream-valuation concept allow formulation of value functions used in a
performance assessment procedure. A priority list of weak performing elements
provides a starting point for the search and specification of promising R&D themes.
Subsequently, in section 3.3, a new method for the specification of innovation
content is reported based on the concepts of functional modelling (Baylin 1990). Re-
addressing system modelling and system decomposition (meta-model, Ch.1, p. 35)
has led to a new modelling-decomposition-synthesis strategy, which serves as a basis
for a procedure for the specification of process system innovation content.
Throughout the chapter the industrial systems around aromatics (benzene, toluene
and xylenes) are used to illustrate and test the methods.

3.2 Systematically towards innovation content?

3.2.1 MODELLING AND STREAM VALUATION

When the Earth, its interconnected material cycles, and parts thereof are modelled as
a system (see Cadre 1-1), in a single analysis the depletion of resources may be
assessed as well as the fate of toxic materials such as heavy metals and harmful
organic components. Using systems representation and characterisation based on the
input-output paradigm (Appendix A. 4), its performance may be assessed with
respect to resource utilisation and, preferably, world sustainability (Cadre 3-3 and
Figure 1-6) (after Dijkema and Reuter 1999).
The Club of Rome attempted such a modelling exercise using a systems dynamics
approach (Meadows et al. 1972). A characteristic of the system dynamics approach,
however, is that only aggregate (dynamic) relations are established for the underlying
performance of technology. The structure and content of interconnected industrial
systems is lacking. In refining, petrochemicals and metals, however, systems have
evolved to a varying state of interconnectivity (e.g. Weissermel and Arpe 1994;
Reuter 1998). Their content, structure, performance and utilisation are influenced by
(1) present economics and know-how - the markets of feedstock, products, capital,
labour, technology and knowledge (2) continuous change - induced by economic
prospects, changing demands, emerging environmental problems, shifting
regulation, strategies for sustainability, product innovation and the emergence of
feasible novel technology and plant designs through process innovation. Technology
required to realise conversions for the closure of material cycles, however, may not
be feasible or available, which for example is the case for mixed plastic waste and for
various grades of zinc- containing residues (Sas et al. 1994b; Südholter et al. 1996).
The scope for change and the speed of adaptation are constrained by industry
structure, company strategies including innovation, and the technology, economics
and ecology of existing facilities, notably the scale-of-operations. The very existence
of integrated petrochemical complexes implies interdependence or even entrapment.
As a consequence, exceptions noted, trajectories for change of the petrochemical
industry span decades. It is characterised by relatively slow dynamics of change. In
the assessment, therefore, steady state can be assumed.
Process System Innovation by Design 67

System Decomposition
Together, the process plants and complexes in the petrochemical industry are a
single system (see Ch. 1, Figure 1-5). This can be modelled as a collection of system
elements between which material and energy (exergy) are exchanged. Within each
element a set of inputs (Su) is manipulated and converted to a set of outputs (Sy). In
studies of the petrochemical industry, a single commercial plant within battery limits
appears (see Cadre 3.1) to be the implicit system element of choice (cf. Dijkema and
Reuter 1999). The system image of the petrochemical industry thus comprises a
networked system that is decomposed into nodes and arcs. Single plants are the
transformation nodes and the material flows are the connecting arcs. This is the
model that has been adopted implicitly in chemical engineering and in most of the
work in process system engineering.
Cadre 3-1: Individual (petro) chemical plants

In a logical decomposition of the petrochemical industry, the individual chemical

plant appears to be a suitable system element candidate. In industrial practice the
‘battery limits’ represent a clear distinction between individual plants or processes
and their surroundings. The ‘inside battery limits’ of a safe, environmentally
sound, and economic industrial plant consists of many system elements. In
chemical engineering these have been logically grouped in classes of unit
operations, which include operations such as reaction, mixing, pumping, heating,
cooling, distillation, extraction, flashing, compression, expansion etc. The
classification is based on the recognition that one can model any industrial plant
by a combination of unit operations. Their selection and physical implementation
is a function of the transformation or Pmt-R required (Ch.1, Figure 1-1). A
particular selection constitutes the physical embedding of any particular plant.
Many industrial plants produce substances that are generally perceived chemical
products, without the application of any chemical process. A traditional hydro
skimming oil-refinery, for example, is not a chemical plant because it involves
physical operations only, notably distillation. A chemical plant or process thus is
characterized by the presence of at least one reactor. Its unique characteristic
appears to be ‘transformation by chemical reaction of substances A, B, C etc. into
products, substances D, E, F etc.’ Many types of industrial plants, however,
employ chemical processes. Examples include power plants, waste incineration
facilities, flue gas cleaning installations and some types of waste-water treatment.
Some do include chemical reactors, such as margarine-hardening by
hydrogenation in the food industry. In energy conversion, fuel cell systems include
an electrochemical reactor while in cogeneration facilities the gas turbine
combustion chamber is yet another type of chemical reactor. Traditionally,
however, these systems are not labelled chemical plants because their purpose is
not the generation of a chemical product.
A precise definition therefore is: 'a chemical plant is a facility for the production of
a chemical and includes at least one chemical reactor'. A petrochemical plant is a
special class of chemical plant that uses petroleum-derived feedstock.
68 Process System Innovation by Design

Many chemical plants are part of industrial complexes, which together with 'stand-
alone' individual plants make up the petrochemical industry. The existence of
integrated petrochemical sites may both enhance and hamper the assessment of the
petrochemical industry with respect to resource utilisation. The aggregate data
available on complex performance provides a means to check the data of individual
plants for consistency. As a result of integration the performance of petrochemical
complexes is expected to be better than the simple sum of the individual plants. The
results of assessments based on literature data of individual plants therefore can be
biased. On the other hand, options for performance improvement can be masked by
the very existence of petrochemical complexes, the structure of which might change
considerably in case novel technology for existing plants is offered. The site-wide
integration of technologies, however, in many cases is not so much governed by the
scope of technology, but more by the type organisation, the quality of management,
and by the prevailing economic incentives. Thus one may argue that rather than
assessing the industry as an assemblage of individual plants, the industry assessment
must also address petrochemical complexes, which can be decomposed into a set of
interconnected plants. Thus, a more-or-less intuitive, or logical decomposition of
the petrochemical industry has been formulated as a set of individual petrochemical
plants, some of which are more closely linked by their location in a petrochemical
complex.
Cadre 3-2: Petrochemical complexes

A petrochemical process site or petrochemical complex is a system element or

system aggregate level that seems to neatly fit into to the hierarchy of system
aggregation levels (Ch. 1, Table 1.2, p.16). The grouping of various chemical plants
in a petrochemical complex offers significant economic and organisational
advantages. Investment and operating costs may be reduced by shared utility
systems (power, heat, cooling water, industrial gases) shared maintenance services,
administrative and technological support, transfer of feedstock and products and
reuse of waste via direct process linkages, site energy optimisation, e.g. by pinch
analysis and site emission and waste management. As a consequence, in the
petrochemical industry plants are hardly ever built as so-called stand-alone units
because more margin can be generated within a petrochemical complex. Since the
facilities investment represent considerable 'sunk cost', the increased earning
power per facility may be at the expense of long-term flexibility caused by the
introduced interdependencies, however. In regional development theory and
location theory a complex is defined as a joint location of industries linked in a
product chain. A petrochemical complex can then be defined as 'a number of
petrochemical plants that are located in one another’s vicinity', where ‘vicinity’ is
determined by practice, e.g. the existing pipeline infrastructure (Molle and Wever
1984: 98-99). Further to vicinity linkages, company linkages provide incentive for
the formation of complexes and may justify a considerable geographical distance
between production facilities.
Process System Innovation by Design 69

I/O modelling, mass and energy balancing

Any system can be viewed as a network of system elements connected internally by
relations or flows and connected to the external world via its set of inputs and
outputs (Su, Sy). A stream of material can be characterised by its mass flow, its
composition, energy flow or exergy flow. Since energy and exergy are
thermodynamic state properties of a material, their value depends only on present
conditions, notably material composition, pressure and temperature (see Appendix
A. 2). A material flow's energy content and other properties such as carbon-content
cannot be calculated, therefore, if mass flow and composition are unknown.
In a node of the system or system element both mass and energy are conserved. The
Law of Conservation of matter states that in the absence of nuclear reactions the
mass of a closed system is constant. It applies to complex materials and individual
chemical element (Carbon, Oxygen, Nitrogen etc.) alike. The First Law of
Thermodynamics states that the energy of the Universe remains constant. Thus, the
total mass and energy of an open system and its surroundings are constant. In other
words, neither mass nor energy can be lost or generated, and for an open system
such as a continuous chemical process, which is in steady state, the sum of the
inputs equals the sum of the outputs22. Thus around each system several balances
can be set up and completed for mass per chemical element, total mass and energy
respectively. Variations in accumulation in the system over time can be included in
each balance, so that finally the formula results:

κ• = ∫tot1 {u(t) •κuT(t) - y(t) •κyT(t)} dt (Eq. 3.1)

Where u(t) = vector of input flows [flow unit/time]
y(t) = vector of output flows [flow unit/time]
t = time [time unit]
κu,y(t) = vectors of flow conversion factors
to chemical element, mass, or energy [unit/flow unit ]
•
κ = net accumulation
of chemical element, mass or energy [unit]
•
In steady state, κ = 0.
The completion of an energy balance for a system or system element requires mass
and composition of all inputs and outputs. A proper mass balance therefore is key to
all other computations for system characterisation.

Stream valuation by labelling

An experienced process engineer can quickly get an impression of a plant's
performance using information from a mass and energy balance. Molle (1984) used
mass balances to explain the location of refineries and petrochemical complexes
effectively (Molle and Wever 1984). SRI (1989) published an overview of mass
balances of industrial processes, which quickly gives an impression on the selectivity
of the industrial chemistry involved (SRI 1989). Sheldon (1993) interpreted mass

22 In case the system is not in steady state, the effects of temporary accumulation or depletion of

mass or energy stored in the system must be taken into account.

70 Process System Innovation by Design

balances to assess the real function of fine chemical plants (Sheldon 1993). In all
these cases, however, implicitly some stream valuation is used.
Generally, two perspectives are used to assess the performance of a particular
production system or part thereof: economy and ecology. In the Brundtlandt definition
of Sustainability, economy, ecology and society are related (World Commission on
Economic Development 1987). In a much used conception of sustainability, 'People,
Planet, Profit', these three dimensions return. We have left out the societal
dimension explicitly in the assessment of production system performance because it
would require a thorough analysis of system surroundings, economic structure,
market dynamics, division-of-wealth (equity) and ecological impact. The economy
and ecology do impact peoples lives, however - if economics are favourable,
production systems will be realised and continue operations, jobs will be created and
consumer needs will be fulfilled. If ecological characteristics are unfavourable,
people will react either because their health or environment is at risk and demand
change.
The set of stream labels presented in Figure 3-1 is an attempt to combine the
valuation of economic and ecological impacts. To allow operation of the system
material and/or energy resources are extracted from natural cycles to provide
feedstock and utilities. Their cost depends on availability or scarcity, market demand,
taxes levied and regulation. The associated ecological impact depends on the origin,
extent and characteristic of such extraction. Ubiquities such as seawater or air are
neither scarce nor does their extraction pose a significant ecological impact. A long-
term risk is involved, however because these may become subject to taxation. The
processing of waste has a positive ecological impact and results in a fee. Using an
economic perspective, in case a market demand exists, inputs are obtained at a cost,
ubiquities are for free, and waste processing brings a fee.
At the output side, main products and co-products are sold at revenue, which must
lead to gross profits, i.e. revenue exceeds average total cost. 'Wanted' by-products
yield at least marginal profits, i.e. revenue exceeds feedstock and operating cost.
'Unwanted' by-products include those sold at revenue less than marginal cost and
those that must be disposed off at a cost. Main products that are in demand bring
such revenue that a profit results under appropriate market conditions. Wanted by-
products add to the margin realised by operating the system. Unwanted but accepted
by-products remain in the cycle but do not add any margin for the operator of the
system. Emissions in many cases cannot be avoided. In case regulatory restrictions
apply, disposal requires treatment at a cost.
The ecological impact of the output categories depends on their use and end-of-life
characteristics. Unaccepted, unwanted by-products represent a loss from the system,
will leave the industrial material cycle, and thus will directly impact the environment.
Finally, emissions can be labelled constrained to limit their adverse ecological impact
at a cost, and unlimited because of their perceived lack-of-impact, which represents a
risk.
The scheme illustrates that between inputs and outputs the economic and ecological
perspectives often do not align: in the manufacture of many products, some adverse
emissions or waste to the environment are accepted.
 Input Labels Output Labels

Feedstocks Main Products

Scarce Reference Profit f(u, ch,eol)
Resources
f (o,ch) Cost Primary Co-Products Profit f(u, ch,eol)

f (o,ch) Cost Co-feeds

Process System Innovation by Design

ByProducts
f (o,ch) Cost Incidental
Industrial Wanted + Margin f(u, ch,eol)
System
none Risk Ubiquties Not Desired

f (o,ch) Fee Waste Accepted - Margin f(u, ch,eol)

Wasted Cost impact

Legenda
Environmental Impact
function of characteristics, Emissions
origin, use, or end-of-life)
Limited Cost impact
Economic

Figure 3-1: Labels for streams around an industrial system.

Impact Allowed Risk impact
Stream Label
71
72 Process System Innovation by Design

In electric power production, for example scarce resources are used to generate a
wanted product, electricity, at the expense of unwanted thermal waste - cooling
water effluent-, CO2 emission and solid waste - flue gas dust and slags-. Slags of
coal-fired power plants are applied as road foundation material, however. Thus these
have been 'upgraded' to a status somewhere in between unwanted waste and wanted
by-product, as there is no revenue or net disposal cost.
The classification in Table 3.1 has been based on the notion that any industrial
process has at least one main product, the reference product. To enable
reproducible calculations all process data are expressed as ratios: the quantity
required for the production of 1 kg reference product. Usually, with the reference
product a primary feed is associated which delivers the bulk of the product. Often, a
process has been specifically designed for co-products and by-products result for
which the system has not been intentionally designed. These may be chemicals that
are produced in small quantities, and that can be sold at a net profit margin, such as
for example the C4 stream in a steam cracker. Over time, stream labels around a
process system may change. By-products, for example, may become key to process
economics and thus be labelled co-products.
Table 3.1: Overview of stream labels around a system or system element
Stream Label Symbol Suffixe │ Symbol Suffixe
Inputs I Outputs O
Scarce resources S Main Product M
Primary f Reference r
Co-feed c Co-product c
Incidental i By Product B
UbiquiTies T Wanted/Desired d
Waste W Not desired
Accepted n
Wasted w
Emission E
Limited l
Allowed a
Grey-box approach -see text
Process-related p Process-related p
Utility-related u Utility-related u

Note: The set of Inputs I, for example has subset Scarce resources S, Ubiquities T
and Waste W. Adopting a grey-box model( Figure 3-2, p.74), these may be process
related or utility related.

Illustration: Aromatics production in a steam cracker complex

The classification summarized in Table 3.1 can be applied to a typical aromatics
plant that processes pyrolysis gasoline or C5+ -gasoline. Pyrolysis gasoline is a by-
Process System Innovation by Design 73

product of steam cracking (see Chapter 4) and consists of simple aromatics, non-
aromatics, olefins (C5+) and complex aromatics (C9+). The non-aromatics, olefins and
diolefins present are recovered and recycled to the steam cracker. To that end,
olefins/diolefins are selectively reacted with hydrogen and recovered as a C5-cut by-
product. A mixture of simple aromatics is isolated and a C9+ fraction is recovered as
by-product that has fuel value. Often, aromatics plants are designed for the
production of Benzene and para-Xylene only. Sometimes, styrene recovery is
feasible. This is due to lack-of-demand for the other aromatics present in pyrolysis
gasoline, viz. toluene, m-xylene and o-xylene. Therefore, after separation of benzene
and p-xylene, these simple aromatics are converted to benzene and p-xylene. To
operate, the aromatics plant requires fuel, steam, electricity and cooling agent
(Chauvel and Lefebvre 1989a). In the past decade, in some aromatics facilities
recovery units for styrene have been implemented. From this description, the labels
given in Table 3.2 seem appropriate.
Table 3.2: Stream labels for an industrial aromatics complex
Inputs I Outputs O
Pyrolysis Gasoline S f p Benzene M r p
Hydrogen S c p p-Xylene M c p
Fuel S c u Styrene B d p
Electricity S c u C5-C8-non-aromatics B d p
Cooling agent T c u C9+ B n p
Steam S c u Condens Water B n u
Catalysts S i p Spent Catalyst B w p

In every system, some energy conversion takes place to complete the system's
operation in a finite time-span. In a chemical plant, the chemical transformation of
feedstock to desired chemical product also involves an energy conversion. The
Second Law of Thermodynamics dictates that part of the Gibb's free energy ∆G is
converted to waste-heat. When the conversion of feedstock turns some chemical
energy ∆G into power and heat and the generated amounts exceed the plant's
internal demand it becomes a net exporter of heat or power. An industrial ammonia
plant, for example, has been labelled a power plant with ammonia as a lucky by-
product (Montfoort 1993).
A distinction between 'process' and 'utility' related streams therefore appears useful,
where utility includes non-energy related streams such as lubrication, inert gas,
process water and organic solvents. Thereby a system is decomposed into a process
subsystem and utility subsystem or subsystems.
At the process-side streams are subject to some chemical reaction or some physical
operation. A significant part of the input ends up in the output. Utility streams enter
and leave the process system to supply or withdraw energy from the system or to
provide other utility functions, e.g. to facilitate separation or product purification via
solvent extraction. They are not chemically converted with the process streams.
The utility category consists of streams that enter and leave the process-side
unchanged except for their energy and related physical state (P, T, composition,
74 Process System Innovation by Design

phase, momentum). An example is steam that is used in a great many industrial

reactors as a diluent and energy supplier. This classification is a 'grey-box' approach
(Figure 3-2) because it requires limited knowledge of the inner workings of the
system. The additional classification comprises two mutually exclusive stream
categories as included in Table 3.1. Since the other labels remain valid, this sub
classification applies to both the generators of industrial products and of the
generators of utilities!

Process
Process input 1...m 1...n Process Outputs
Side

Energy transfer

Utilities
Utilities input 1...p Side 1...q Utilities output

Mass Circulation

Figure 3-2: A process system represented as a 'grey-box.'

In a number of industrial operations the distinction between process and utility is
ambiguous. In cement kilns, for example, co-fed organic or polymer waste reduces
and replaces fuel input (utility) and feedstock (process). The kiln utilizes the polymer
waste energy and minerals content simultaneously (e.g. Vanderborght and Dijkema
1996; Delissen 1997; Verhoef et al. 2002a). In the grey-box approach the waste co-
feed to a cement kiln would be labelled a process-stream: in the petrochemical
industry and in other process industry energy supply through exothermic reaction of
part of the feedstock is common practice.
The grey-box approach allows an additional check on the systems energy balance
and allows one to distinguish between two categories of process system
improvements. These are (1) improvements that do not directly affect the process,
such as end-of-pipe and utility system improvements; Examples are optimal heat-
exchanger networks and the closure of industrial water-cycles and (2) improvements
that do affect the process, such as changes in feedstock quality or composition,
catalysts system, process technology employed, process system design including
recycle structure as a function of feedstock and product labelling.

3.2.2 ASSESSMENT CRITERIA: LOSSES AND EFFICIENCIES

In this section the quantitative criteria are formulated that can be used in the
assessment procedure using input/output modelling and stream labelling. Preferably,
these criteria can be applied to each type of system element, ranging from unit
operation to national economy (Table 1.2).
Process System Innovation by Design 75

Effective resource utilisation expresses the notion that depletable resources that have
been extracted from the earth or from its ecosystems in order to satisfy human need
must be utilised to their full potential, both with respect to quantity and quality (see
Cadre 3-3). The development or selection of criteria suitable for an assessment of
resource utilisation is not trivial. In many (petro-) chemical operations hydrocarbon
material of fossil origin is both feedstock and energy supplier. Resources and
material flows exhibit properties that are conserved in the Universe or part thereof,
such as mass and energy. Entropy, which relates to quality, however, is a non-
conserved property. Thus, while mass and energy cannot be lost in the Universe, it is
implicit that something is 'lost' by ineffective resource utilisation: loss of quality and loss
from the system to the surroundings.

Operation performance criteria or a life-cycle approach?

Life-cycle costing and environmentally oriented life-cycle analysis are so-called
cradle-to-grave analysis where it is conjectured that for proper analysis and
comparison of systems and their alternatives one should take into account bringing-
into-being the system studied, expected life-span of operation and its end-of-life.
Boustead (1979) has coined the terms primary, secondary, tertiary and quaternary
resource/energy use. Primary resource/energy use is directly related to production -
to create the product and to sustain system operation. Secondary denotes the
resource/energy use for creation of the system's inputs. Tertiary and quaternary
denote resource/energy use to bring the system into being and producing the
construction materials respectively. (Boustead and Hancock 1979).
A petrochemical plant consists of thousands of tons of concrete, steel and other
material from energy/resource intensive production. In petrochemical operation, at
typical annual throughputs of hundreds of thousand tons over a lifespan that
typically exceeds 20 years, millions of tons of materials are converted in these
facilities. Thus, in cradle-to-grave analysis, total energy/resource use of a
petrochemical production facility is dominated by its primary material/energy use.
Efficiency improvement at the expense of increased tertiary and quaternary
energy/resource is favourable except for extreme cases because at end-of-life,
concrete, steel and aluminium can be recycled using the available industrial
infrastructure for primary production obtaining recycled material at a fraction of the
primary resource and energy use for virgin material (Verhoef et al. 2001). Secondary
resource/energy use is significant for many petrochemical plants. The evaluation
must focus on primary resource/energy use, however because the assessment
objective is to identify and prioritise system elements with respect to non-renewable
fossil resource utilisation.

Physical or economic assessment criteria?

Since we are after the specification of innovation content for improved
sustainability, both ecology and economy appear to be required in some assessment.
Economic criteria, however, appear unsuitable for quantitative assessment due to:
1) Unavailability of data. Petrochemical plants produce undifferentiated commodity
products. Thus, one competes by cost of goods sold. Cost leadership can be
established through economy-of-scale, which amongst others can be achieved by
76 Process System Innovation by Design

Cadre 3-3: Effective resource utilisation

The future of the petrochemical industry depends on the continued availability of

and access to suitable resources. In our present industrial economy, however, for
the greater part non-renewables are used. High quality fossil resources represent a
state of low entropy. When used in petroleum refineries and the petrochemical
industry these are degraded eventually to CO2, water, and heat. This 'downhill'
conversion process implies a loss of well defined and highly ordered structures of
the components present in fossil feedstock. The reverse operation, the 'uphill'
conversion of CO2 and water to suitable structured petrochemical feedstock or
energy resource, requires that part of the energy restored in the components
formed is (shaft) Work done on the “uphill” conversion system.
According to the Second Law of Thermodynamics, in every energy conversion
that takes place in a closed system in interaction with it's surrounding, inevitably
some quality of energy is lost, i.e. exergy, the potential to do Work. According to
Georgescu-Roegen (1991) there is a true misunderstanding about “the entropy of
matter. In mechanical engineering, for example, one studies the engines or
apparatus for energy conversion to minimise quality loss or the degradation of
energy resources. There is a tendency, however, to study energy conversion in
isolation from the chemical conversions involved. For the greater part, however,
energy supply for heat engines is by the combustion of fossil fuels, a chemical
process.
In the combustion of fuels derived from crude oil complex organic compounds
are destroyed and forever reduced to CO2 and H2O, which is an extreme example
of degradation of material resources. The combustion process itself and the
associated emissions into the environment of CO2, SO2 etc. are an example of
dissipation of matter, a term first introduced by Planck (1897). In the combustion
of natural gas, a fuel that may consist of pure methane, for example, the gas is
mixed with air and converted to a mixture of nitrogen, oxygen, CO2, H2O and a
host of other products of (incomplete) combustion and compounds present in the
air used. Another example is the metals extracted from natural ores. These will
also become completely dissipated in the environment eventually. A rather
extreme example is dispersion of lead by the use of ethyl-lead compounds in
motor-gasoline, a practice that has presently been phased out in large parts of the
world.
Both the dissipation of energy and of matter are forms of degradation, or entropy
generation. As the entropy increases, the quality of energy or matter is reduced. A
consequence of, for example, the mixing of essentially pure components, a not
uncommon operation in the petrochemical industry, is an increase of entropy or
degradation. Effective use of resources thus implies that one strives for minimal
dissipation of matter (Planck, 1897) (cited in Planck 1926; Georgescu-Roegen
1991). Effective resource utilisation therefore is defined as the prevention and
postponement of all forms of degradation of resources to the maximum extent.
Process System Innovation by Design 77

continued process innovation, by efficient organisation and by a sound

investment strategy. Confidentiality of the underlying economic data is the norm,
and it may be expected that improving process economics has been a main
incentive to unearth process innovations throughout the industry.
2) Market dynamics. The flexibility of petrochemical clusters to meet market
fluctuations is very limited (Kuipers 1999). Petrochemical economics dictate
plant utilisation of 75% or higher to create a net return on asset capital
employed. Individual plants, however, will only be phased out in case they are
worn-out or do not create sufficient marginal revenue (Ch.1). Petrochemical
markets exhibit cyclicality in capacity and supply-demand balance. The cost of
capital equipment, labour and the value of products are influenced by internal
market structure and dynamics. The price of its dominant feedstock, crude oil, is
severely affected by the OPEC, which attempts to maintain a favourable world
supply-demand balance. Thus, market prices are no suitable indicator of the
prospect of future resource scarcity, environmental impact or sustainability.
3) Mismatch of economic incentives and conditions for innovation. In the life cycle of a
petrochemical plant bringing-into-being the facility is an important phase. The
initial capital-invested must be recovered through the operational margin
throughout the plant's life span. In a mature market, however, realisation of a
new facility is only advantageous when the total cost of production in the new
plant is less than the marginal operating cost in existing plants (Ch.1). Older
possibly less efficient plants are only phased out gradually, which is confirmed by
a recent overview of worldwide polypropylene production (Linse 2002). Thus,
considerable time may pass before one may reap the benefits of increased
economy-of-scale and efficiency realised through innovation. In addition, to
reduce risk and time-to-plant completion, a rule-of-thumb in many an
investment project is to stick to 'proven technology', i.e. technology that has
been proven to function reliably in at least one other industrial application of
similar scale.
Innovation activity - as measured by number of patents filed - is high throughout the
chemical industry compared to other sectors (Stobaugh 1988). The number of
fundamentally new processes introduced to the market, however, is limited, some
processes having been around since the beginning of the past century (e.g. Jennings
1991; Satterfield 1991). This may be due to market dynamics and economic
incentives addressed. A large time-to-market characterises fundamental innovations,
such as a new synthesis route, and slow diffusion to the market of major process
innovations, such as entire new plant concepts. Shareholders, however, increasingly
demand short-term profit increases. As a consequence, many companies have opted
to move away from exploratory R&D on petrochemical production systems. As a
corollary, the analysis of current economic conditions hints at the existence of an
untapped reservoir of major process system innovations!

A Second-Law based criterion?

Any system, including chemical process systems or reaction systems, can be
considered a black-box where the transformation effected involves a change in
78 Process System Innovation by Design

Internal energy, ∆U23, between input and output materials and some exchange of
Work w and Heat q with the system's surroundings (Figure 3-3). Internal energy is a
state property (Appendix A. 2). According to the first law of Thermodynamics,
∆U = q + w (Eq. 3.2)
Enthalpy is a state property defined as
∆H = ∆U + ∆(PV). (Eq. 3.3)
It follows from eq. 3.1 and 3.2 that in case only PV work is allowed on a system (i.e.
no shaft Work, w=0) that ∆H = qp. According to the Second Law of
Thermodynamics the sum of the entropy changes in the system and its surroundings
can only be positive or zero. Entropy, or disorder, is defined as
∆S = q / T. (Eq. 3.4)
Whenever a transformation process proceeds, changes occur both in the entropy of
the materials involved and the entropy of the system surroundings. According to the
Second Law, the created entropy σ is given by
σ = ∆Siirr., created = ∆Ssystem + ∆Ssurrounding >= 0 (Eq. 3.5)
The Gibb's free energy of a system is defined as:

∆G = ∆H - T∆S (Eq. 3.6)

It has been shown that the total Work requirement of a process is given by Eq. 3.7
(after Denbigh 1956):
wt = T0 •σ + ∆G0 (Eq. 3.7)
Thus, for a process where materials enter and leave at standard conditions (Po,To) the
total Work requirement equals the change in Gibb's free energy plus the lost work
due to irreversibilities. This is illustrated in Figure 3-3.

The formal definition of Exergy or Availability A resembles that of G24:

Ex = A = (H-Ho) – To (S-So) (Eq. 3.8)
B is a state property that can be computed for any stream. According to Eq. 3.7, B
equals zero for heat qP exchanged at the reference temperature T0 (H=Ho; S=So).
The exergy of a heat reservoir or flow is given by the familiar Carnot factor:
Ex = A = q (1 – To/T) (Eq. 3.9)
Due to unavoidable irreversibilities, the exergy balance around any real process
system cannot be completed. A proper exergy balance of the petrochemical industry

23 In eqn. 3.1 - 3.10, the change in property X (X = Ex, G, H, S or U) between input and outputs is
denoted as ∆X = Xout - Xin
24 Gibb's free energy is a state variable that combines enthalpy and entropy change, accounting for

any change in conditions (P,T,x) associated with the transition; exergy or availability denotes the work
available or required when transforming from the present state to standard conditions.
Process System Innovation by Design 79

or parts thereof can be set up, however, to calculate exergy loss, ∆Exloss. The exergy
loss that is irrevocably related to the system's implementation and operation is
calculated from the exergy change represented by the material transformation,
combined with and the net other exergy inputs or outputs of the system. The exergy
of materials relates to the Gibb’s free energy at standard conditions (Po,To). Since G is a
state property (see Appendix A. 2), whether a chemical conversion takes place in the
system or whatever its exact route is irrelevant for the analysis25.

Process System
Inputs Outputs
G

To Sirr. Heat (q)

Work G>0 G<0 Work

Figure 3-3: Work and a process system. In case ∆G > 0, Work needs to be delivered to
the process; in case ∆G < 0 the process can deliver Work.

Results from inventorying feedstock, products and utilities for mass and energy
balancing can be used to calculate the exergy value of all streams around a process.
Exergy loss is obtained as

∆Exloss = ΣExin - ΣExout (Eq. 3.10)

∆Exloss = Σ(φm •∆G0) in - Σ (φm •∆G0 )out

+ (Σwin - Σwout )

+ (Σ qin (1 – To/Tin) - Σ qout (1 – To/Tout) (Eq. 3.11)

where φm indicates the material streams.

According to Eq. 3.7, the theoretical amount of Work associated with a chemical
transformation can be determined from a Gibb's free energy input/output balance
of feedstock and products by assuming 100 % reversible conversion. The delivered
actual amount of work, wact , will be less due to the process implementation selected

25
It is safe to assume for most industrial processes the final state of both inputs and outputs of a
process system are standard conditions P0 and T0 Notable exceptions of course are steam,
cogeneration and heating systems
80 Process System Innovation by Design

and irreversibilities. The minimum amount of work required for an endothermic

process equals ∆G0, the actual amount wact supplied to the process must be larger:

wact = ∆(Σ (φm •∆G0)) + ∆Exloss ; ∆Exloss >0 (Eq. 3.12)

Thus it appears exergy loss is a basis for a suitable criterion for the assessment of
resource utilisation. In the assessment of individual plants or system elements a
more pronounced 'landscape' impression of the petrochemical industry is obtained
when the indicator is normalised by division with the minimum achievable change in
exergy
θ =∆Exloss / ∆(Σ (φm •∆G0)) (Eq. 3.13)
θ could be labelled a 'Thermodynamic distance to target'.
An appropriate thermodynamic efficiency for a process is

η = ∆(Σ (φm •∆G0))/ wact (Eq. 3.14)

Substituting Eq. 3.12
η = ∆(Σ (φm •∆G0))/ (∆(Σ (φm •∆G0)) + ∆Exloss ) (Eq. 3.15)
In industrial practice this thermodynamic efficiency may be down to only a few
percent due to irreversibilities. In exothermic processes, θ can assume a negative
value. In such a case, a transformation that theoretically could supply work has been
implemented such that a net input of work is required for the operation to complete!

Assessment criteria: losses and efficiencies

In the previous section we have identified exergy as an indicator energy quality. Due
to the internal irreversible operation of a system there is always some internal exergy
loss associated with the working of any system The First Law of Thermodynamics
implies that there cannot be an internal energy loss, The Law of Conservation of
matter implies there cannot exist an internal mass loss in a process system. Using
stream valuation, however, external losses can be defined as mass, energy or exergy
transferred to the system surroundings that cannot be utilised in the system or in its
surroundings without further processing. Using the set of stream labels summarized
in Table 3.1 (p. 72), industrial systems can be identified where mass loss from the
system occurs, for example in the form of waste material or CO2. Similarly, energy can
be 'lost' or wasted in the form of heat that is rejected to the environment, which is a
common approach in large-scale electric power generation. External mass loss and
external energy loss can be calculated. An overview is given in Table 3.3. Examples of
external mass loss are the flue gas from a process furnace, light-gases that are being
flared, waste, spills, and fugitive emissions. Associated with these streams is an
external energy loss. Other external energy losses from closed systems are radiation
losses, and notably heat removed from the system by cooling water or in cooling
towers. The set can be used in the petrochemical industry and beyond at any
aggregate system level (Table 1.2, p.16); its application is not limited to this industry.
It helps to understand the (mis)use of the term 'energy consumption' such energy
either ends up in the product or goes unnoticed as external system loss.
Process System Innovation by Design 81

Table 3.3: Formulation of a sound criterion is possible yes / no

Type of Loss
Criterion Internal External
Mass No Yes
Energy No Yes
Exergy Yes Yes

For streams around a process system or part thereof (Figure 3-1, p.71) all (sub)sets
of streams as indicated in Table 3.1 (p.72) can be completed by assigning valuation
labels to all streams around a system. External losses per unit main product and
system efficiencies can be defined for any conserved property. Thereby, a set of
performance indicators is defined whereby an impression or “fingerprint” is
obtained of system performance in relation to its surroundings. Each performance
indicator has a unique value function:
Reference Loss - any scarce resource that does not end up in main product:

La = ((Σ I(S) - Σ O(M)) / Mr) (Eq. 3.16)

Product Loss - any scarce resource that does not end up in 'desired' products:

Lb = (Σ I(S) - Σ O(M, Bd ) / Mr) (Eq. 3.17)

Emission Loss: any scarce resource that does not end up in products

Lc ((Σ I(S) - Σ O(M, B) / Mr) (Eq. 3.18)

Total Emission Loss: any input that does not end up in product or product utilities

Ld = ((Σ I - Σ O (M, B)) / Mr (Eq. 3.19)

Balance Loss:

Le = (Σ I - Σ O)/ Mr (Eq. 3.20)

It follows from the system balance equation that for mass and energy Le = 0, whilst
for exergy Le> 0
By selecting Scarce Resources S as a basis for calculation La, Lb and Lc, a net negative
loss may result for systems that upgrade Ubiquities such as air or Waste material by
incorporating part of this material in some of their products. In some systems, this
of course is a basis for their economic feasibility ('make money out of thin air'), but it
also offers a means to identify the dependence of systems on the free use of
ubiquities and their vulnerability with respect to changes in the applicable regime. In
addition, it provides a quantitative affirmation of the reuse of waste material, which
may also become scarce resource for other production processes .
To obtain values for Losses that relate to Scarce Resources only, the incorporation
of Ubiquities and Waste Material in Products must be corrected for. This requires a
further opening of the black box and additional assumptions because this
82 Process System Innovation by Design

information often can only be estimated from a system mass and energy balance and
the applicable overall stoechiometric chemical reaction26.
The indicators La…Ld can be calculated for the entire system or for the process or
utility-side respectively. Notably comparison of La…d and La…d, p will yield information
on the amount of utilities required to operate a particular process system, and where
the utility material or energy supplied does end up.
As a matter of course, the stream sets can also be used to define system efficiencies
for materials and energy use, such as:
- the reference efficiency:

ηa = Σ O(Mr) /Σ I(S) (Eq. 3.21)

- the product efficiency:

ηb = Σ O(M, Ba) /Σ I(S) (Eq. 3.22)

- The emission efficiency:

ηc = Σ O(M, B) /Σ I(S) (Eq. 3.23)

- The total emission efficiency:

ηd = Σ O (M, B) /Σ I (Eq. 3.24)

La and ηa directly relate to the primary objective of the system analysed: most
continuous chemical plants are designed for the production of a single chemical
product, a plant with a number of co-products being the exception. When these
indicators are low, a substantial loss from the system occurs.
Lb and ηb relate to economic objective of a system, i.e. a process may be designed
for a main product and a range of valuable by-products to realise sufficient margin.
Loss b is the amount of feedstock material or energy that is either lost to unwanted
products that must be sold at a negative margin or as outputs classified as emissions.
Lc and ηc are indicators of the total flow of material or energy required to effect the
transformation in the system assessed. The material efficiency ηc must be close to or
equal 1, and Emission loss Lc must be close to or equal zero, except for cases where
there are large process-related emissions or downgrading of by-products to utility
material.
Ld and ηd are meaningful to single out system elements (chemical plants, chemical
complexes) that require a large flow of utilities to sustain their operation. This utility
use can be a cause of a low energetic efficiency of these process systems because all
these utilities need to be flow around the system and recovery is required of spent
utilities. Similar to loss Lc, Total emission loss Ld must be close or equal to zero. A
significant value of Total Emission Loss d implies that there are large process-

26 In the petrochemical industry, oxygen from air for (partial) oxidation processes is the most
commonly used Ubiquity. Waste materials are hardly used. In the assessment of the current system,
therefore, to obtain corrected values for L can be based on an element balance for oxygen. Assuming
that any organic feedstock not accounted for in O is converted to Emission of CO2 and H2O allows
to complete the material balance for each system.
Process System Innovation by Design 83

related, utility-related emissions or that for example unwanted organic products are
downgraded to utilities, where they are eventually converted to CO2 and H2O.
Together, ηa…d and La….d provide a “fingerprint” of the system element with respect
to resource utilisation. For any higher system level (complex, industry), combined
with capacities installed the La….d provides an impression of relative importance of
system element losses.

3.2.3 ASSESSMENT PROCEDURE

The modelling and the assessment criteria have been integrated in a logical
procedure for assessment of the petrochemical industry:
1) Inventory and define the system assessed
a. System: select system boundary and identify the petrochemical plants within
the boundary and estimate the nameplate capacities installed.
b. System elements: assemble the information required per petrochemical plant
as for a 'typical' plant, viz. (1) Process description (2) Set of input / output
(3) Capacity range commercially applied;.
c. Inputs and outputs: Except for heat and electricity these are substance flows.
A set of input/output data set that represents a typical plant must be
assembled, validated and completed. Underlying information must be
assembled that is required for the determination of mass, energy and exergy
properties of inputs and outputs.
2) Obtain quantitative system image for system evaluation:
d. System performance: After assignment of stream labels, system losses and
system performance indicators can be calculated. Using installed capacities,
an indication of Losses can be calculated. The procedure can be applied to a
petrochemical plant, a subsystem thereof, or a sector of the industry or the
total system.
e. Networked industrial systems: A total material and energy balance of
interconnected petrochemical systems can be computed from the capacities
installed and the input/output characteristics of typical plants.
3) System evaluation and analysis interpretation
f. 'Fingerprint': a comparison between the four efficiencies will provide a first
insight in the effectiveness and selectivity of the process system and system
elements assessed.
g. Prioritisation of the elements of each system analysed. This can be done with
respect to various indicators.

Upon completion of this procedure a snapshot of industry structure, content and

performance may be obtained, as well as a priority-list of weak system elements and
indication of root causes.

Ad 1) Inventory and define the system assessed

In this procedure a logical static system model of the petrochemical industry is
adopted. The industry is viewed as a structured collection of networked chemical
84 Process System Innovation by Design

plants. At both these aggregate levels the Laws of Conservation of Matter and
Energy apply.
The system boundary must be selected to include large-scale petrochemical plants
for undifferentiated, commodity products. Since the focus is on improvement of
current systems and technologies, the product range is fixed. Some of these products,
however, are directly extracted from intermediate oil refinery products. Therefore,
system boundary and feedstock restrictions must accommodate inclusion of, for
example, refinery production of aromatics from reformed naphtha and propylene
from cat-cracked gases.
Often in system assessments, only information is available from open literature.
Such data sets often consist of vendor-supplied data or aggregated input/output data
for a 'typical' plant. The limitation to open literature sources is hard to overcome, as
current performance data on existing petrochemical plants is considered strategic
information, because it is process innovation by which the mature petrochemical
industry competes (Ch.1). This kind of information is assembled, aggregated, made
anonymous and exchanged via consultants for benchmarking purposes and the
industry is extremely cautious to expand the exchange of quantitative information
(Dijkema and Mayer 2001a).
Using open source data a somewhat dated image of the industry's plants
performance is obtained, certainly not an image of current 'Dutch industry', nor the
performance of an individual company. Open source data on petrochemical
complexes, the industry and parts thereof is even more scarce. Construction from
individual plant data is possible, however, assuming that synergies achieved in
industrial chemical complexes are exploited and reflected in the labelling of streams
adopted (§ 3.2.1, p.66). A fairly extensive publication of 1990 name-plate27 capacities
is available (Anonymous 1990), which can be augmented with material from various
other sources. Thus, soundness, speed and simplicity can be matched to the
magnitude of the system addressed, time and data available. Using the data available,
a sufficient 'proxy' to industry performance is obtained for first prioritisation
between system elements (see § 4.3.2, p.123 for an example). To validate the results,
the data used can be compared with company or consultant data and industrial
complex or national economy aggregate data.
Complete data sets on industrial plants are extremely hard to find and there is no
clear standard of representation of performance data. Thus, Input-Output analysis
combined with the valuation of streams as introduced in this chapter is to be
adopted as a standard of representation. Data sets from the literature must be
converted prior to their use in the assessment. Data reconciliation involves a check
for data-consistency and completeness per data set on an individual plant (an
example is given in Appendix A. 6).
To construct a complete the process for useful and sufficiently reliable results the
following may serve as a guideline:

27 The so-called name-plate capacity is the production capacity for which a chemical plant originally is

constructed. Over the years, so-called 'capacity-creep' occurs for a great many plants through
operating experience, plant changes, debottlenecking and/or revamping. Such progress often remains
the proprietary knowledge of plant owners.
Process System Innovation by Design 85

- with respect to corrections, assumptions must be underpinned on a case-by case

basis by intricate process knowledge.
- with respect to additions for completion of a petrochemical plant's mass
balances, assume that a deficiency (i.e. more mass going into the process than
going out) is caused by the generation of light gases that are emitted as CO2 and
H2O. This assumption can be considered valid, since most petrochemical
process schemes include an off-gas treatment section that remains otherwise
unspecified, whilst 'hydrocarbons' can always be utilised as fuel after proper
treatment. The carbon and hydrogen thus lost will eventually end up as CO2 and
H2O.
- the specific energy content of streams not reported in a particular data set can be
estimated. However, since often nothing is known about the energy content or
the application of these streams, for the analysis assume that these are zero.
Mass and energy balancing can be used to check the validity of data obtained; with a
few assumptions the mass and energy balances can be completed. However, mass
loss La, Lb and mass efficiency ηa, ηb do not require complete datasets, they
immediately give an impression of the performance of the process. Calculation of
these indicators does not require an assumption with respect to the fate of the
external mass loss!

Ad 2) Obtain quantitative system image for system evaluation

Simple criteria are to be preferred in the assessment because the system analysed is
extensive, its dynamics are relatively slow and data availability is limited. The
modelling approach developed allows the use of conserved properties (mass and
energy) and for one to derive the related indicators La…e and ηa..d. Combined these
provide a fingerprint of system performance. The calculations can be based on
input/output data, a few assumptions to satisfy mass balances and system capacities
installed28. In addition, losses of type a,b, c and d for energy expressed as Lower
Heating Value were calculated. These calculations require the combination of
material flow and composition data with conserved properties (mass, energy). Under
the assumption that most products and resources are at standard conditions,
information from thermodynamic property tables suffices29 (Stull et al. 1969).

Should exergy be used as a criterion in the procedure?

With proper precautions, exergy, a non-conserved property, can be used to assess
system resource utilisation (Table 3.3, p.81). Exergy loss or efficiency appear to be
perfect indicators for one to obtain a quantitative system image because exergy-
based indicators express the system's internal and external utilisation of
thermodynamic potential. In a number of early case studies on styrene and natural
gas derivatives, however, it was found that the use of exergy-based indicators in the

28 Since it was neither the objective nor possible to judge the actual situation in individual production
locations, the data used relate to the situation in or before 1990 are considered to be suitable.
29 Of course, many processes and product flows in the petrochemical industry are at elevated

pressure. The energy-change with respect to the reference state is small, however, even for significant
pressure increase. Care has been taken to use data of the appropriate phase (gas, liquid, solid).
86 Process System Innovation by Design

assessment of networked process systems is difficult and data-intensive and yields

subjective results.
It appeared to be impossible to establish accurate exergy values of non-pure
substances such as oil products. These are composed of many chemical components
and the precise composition simply is not available. Thus the exergy of a steam
cracker feed such as naphtha must be estimated. Typical mass and energy values (as
LHV), however, almost always are available.
At the level of 'a network of petrochemical plants' the results of the case studies
were similar when using either (labels + energy)-based or exergy-based indicators.
Inspection of these results revealed that many petrochemical processes exhibit
similar exergy and energy losses. These processes involve the rearrangement only of
hydrocarbons, which results in a significant enthalpy-change but a low entropy
creation. When prioritising the element's performance of the system analysed (step
3), the use of exergy indicators only results in a different priority-list when the
enthalpy change and the entropy effect of a reaction are comparable. Notably partial
oxidation processes do possess this characteristic.
Finally, whilst exergy appears to be an objective criterion based on the Second Law
of Thermodynamics, in a system's context it is not. “But what has made the problem
of entropy truly exasperating has been the general fashion to extend the denotation
of that term to problems other than the transformation of energy” (Georgescu-
Roegen 1991). Exergy values shaft-Work above all other manifestations of energy,
for example complex organic chemical compounds. The heat liberated due to
oxidation cannot be converted entirely to Work but is still exported from the system
as a valuable by-product (cf. Denbigh 1956). While exergy is an objective measure of
the potential of a substance or substance flow to do Work, our industrial society
requires and utilises a significant amount of substances and energy carriers that
represent a low potential to do Work. Labelling streams thus appears to be more
suitable to reflect current status of our industrial societal needs than the 'single but
thermodynamically sound' property exergy. In the subsequent search for
improvements qualitative exergy analysis can help unearth the root cause of low
resource utilisation and help specify novel system concepts. On the basis of early
case study results, we concluded that in the assessment of networked system
resource utilisation, a proxy of exergy is to be preferred that can be reliably and
reproducibly computed and that the indicators La…e and ηa..d. are such proxies of
exergy. These indicators are based on mass and energy balancing and stream
labelling. Together, they provide a fingerprint of system and system element
performance with respect to resource utilisation.

Ad 3) System evaluation and analysis interpretation

The objective of the analysis is to develop a quantitative image of the production
system and its parts. A common objective of industrial process systems analysis is to
establish the process’ bottlenecks that limit plant capacity or plant performance.
Once established, these provide starting points for plant changes, revamp studies
and R&D. By analogy, at the aggregate level of networked industrial systems also
bottlenecks or weak elements may be present where resource utilisation is less
effective than in some or all other elements of the system. The assessment objective
Process System Innovation by Design 87

thus is to determine the position of processes relative to each other, in order to

obtain a global priority of weak elements as starting points for the search for
improvements for improved resource utilisation.

Illustration: Assessment of the Dutch Aromatics System

Assessment of the Dutch industrial aromatics system in 1990, using name-plate
capacities reported in (Anonymous 1990) and process information available in
literature data (Chauvel and Lefebvre 1989a) serves to illustrate the approach. At the
time, the Dutch industrial aromatics system consisted of some 40 petrochemical
plants. An aggregate overview of the system is given in Figure 3-4. In the BTXProc
subsystem, Benzene (B), Toluene (T) and Xylenes (ortho-Xylene, oX and para-
Xylene, pX) are produced from pyrolysis gasoline and reformed naphtha. These
intermediate are co-products of steam cracking of naphtha and crude oil refining
respectively. The purified individual aromatics serve as intermediates for a number
of petrochemical production chains that lead to, amongst others, polystyrene,
styrene copolymers such as ABS, nylon and polyethyleneterephtalate (PET).

B
Crude oil
Refinery BTX plant T

St. Cracker

Cokes
Coke oven oX
pX
BTXProd

B Process Process Polymerisation

PS
T ABS
Nylon
etc.

oX Process Process Polymerisation

pX
BTXCon

Figure 3-4: Overview of the Dutch industrial aromatics system.

88 Process System Innovation by Design

Analysis of the industrial aromatics system

The overall mass efficiency ηb of the Dutch aromatics system is some 89%, and
annually some total 1.3 Mton of material is ‘lost’ by degradation to inferior,
unwanted by-products and emissions - mainly as CO2 and water -. The overall
energy efficiency ηb is some 77%, annually some total 110 PJ is ‘lost’. The associated
energy is removed from the system as waste-heat by cooling, via flue gases or in the
form of unwanted by-product flows. The results for the aromatics system are
presented in Table 3.4: Results for the aromatics system.
Table 3.4: Results for the aromatics system

Process ηb,mass Lb,mass ηb,energy Lb,energy

[%] [MTA] [%] [PJ/yr]
BTX-Prod 93 0.65 86 53
BTX-Con 86 0.69 68 63
BTX-Tot 89 1.34 77 116

100 Potential 40
External
% Energy Loss PJ/yr

80
30
BTXCon

60
BTXProd
20

40

10
20

0
0
PyGsl BTXExtr. RefGsl CaproLtm Phenol1 PhtcAnh
HDA oXylSep BTXSep Stryren1 Styren2 Phenol2 Cumene
Scope for improvement (100-ηb) %
External energy loss per process

Figure 3-5: Results of energy analysis of the Dutch aromatics system; processes ranked
according to scope for improvement (100-ηb).
The total losses of the analysed aromatics system BTX-Tot are evenly spread over
the two parts. As part of the assessment procedure, however, also the performance
and loss of individual petrochemical plants is determined. Ranking based on the
results yields a number of relatively weak performing chemical processes, notably
Caprolactam, Phenol and Phtalic Anhydride production, as illustrated in the bar
Process System Innovation by Design 89

graph in Figure 3-5. This illustrates that over 90% of the energy loss reported for the
aromatics system can be attributed to the 13 processes in the bar graph. Of these 13,
only 8 exhibit significant scope for improvement. Thus, albeit the Dutch aromatics
system includes some 40 process-plants installed, all else being equal, in order to
improve resource utilisation as a start we only need to examine 20% of these
processes.

Small Scope [%]

0
BTXSep
BTXExtr.
oXylSep
RefGsl
Effective 10 Effective
but high Loss Cumene

20
40 PyGsl 30 20 10 0
High Styren1 Small
Loss Loss
[PJ/Yr] 30 Styren2 [PJ/Yr]
Phenol2

Ineffective HDA
40
Ineffective
but small Loss
PhtcAnh
50 CaproLtm Phenol1
Large Scope [%]

Figure 3-6: Aromatics case study - assessment of processes in the system.

An alternative representation of the results is given in Figure 3-6. In this drawing,
the axes have been chosen to let effective elements of the system assessed end up in
the upper right quadrant. These elements (industrial plants) are relatively efficient
and do generate a moderate external loss. This kind of drawing helps to identify
weak elements and may help to visualize two strategies for improvement: to move
from the lower-left quadrant upwards via efficiency improvement, and to the right
by reducing the volumes processed. As a matter of course the classification shown
depends on ranges of both axes, and only serves to provide a first indication where
to look for improvements or possible innovations. The preliminary conclusion for
the aromatics case is that large-scale processes (exceptions noted) appear to have
reached a high degree of effectiveness over time. A number of plants that are
operated on a small-scale provide scope for improvement, albeit a limited absolute
effect).
In the 'BTXProd' subsystem analysis it was impossible for mixed substances such as
naphtha or pyrolysis gasoline to obtain an exergy-value consistent with the ∆G based
exergy for pure substances. As a consequence, a thermodynamically sound and
90 Process System Innovation by Design

reproducible assessment of this system cannot be completed without sophisticated

methods to estimate the thermodynamic properties of the mixtures involved. The
criteria based on stream labelling and energy balancing represent a proxy to exergy,
by which the analysis of the entire system and its subsystems could be completed.

3.2.4 SEARCH FOR R&D OPPORTUNITIES

Starting points for the search

Assume that (1) there will be a continued demand for the product spectrum of the
petrochemical industry and that (2) the availability of fossil resources is ensured in
the mid- to long-term. The major question then appears to be “What technological
opportunities exist to produce the current set of products of the petrochemical
industry by using less energy and fewer fossil resources.”
Under the two assumptions given above, the functions of the petrochemical industry
remain unquestioned and the results expected from some assessment of resource
utilisation provide starting points for the exploration of R&D opportunities within
the existing system boundary.
The ranking of relatively poor performing petrochemical processes is the starting-
point for the procedure to specify directions for R&D and possible innovation
content.

Analysing weak elements

Establishment of the causes for the losses calculated is half the solution. Inspection of
the “fingerprint” of the losses and efficiencies from the assessment gives a first clue
in what form mass or energy is lost from the system. Detailed process analysis,
design studies and expert knowledge can help to elucidate the causes of internal
exergy loss and external material, energy and exergy loss from the system. In the
industrial aromatics system for example, the caprolactam process produces a large
amount of ammonium sulphate. On the basis of literature (e.g. Chauvel and
Lefebvre 1989a) this was labelled an unwanted by-product. Hydrodealkylation is
implemented largely as a thermal process. As a consequence, its selectivity is
relatively poor and unwanted by-products result.
Thermodynamic inefficiencies are caused not only by the use of non-ideal
conversion routes, but also by system designs that result in operation at much lower
efficiencies than theoretically achievable. Phenol production, for example, involves
partial oxidation where chemical energy is largely converted to low-quality heat. The
large number of steps in the complicated caprolactam process leads to a substantial
internal exergy loss. In other processes, available catalysts and reaction conditions
favour only limited conversion per reactor pass. The associated separation and
recycle result in losses and inefficiencies, in particular when these require significant
volumes of gas to be recompressed.

Systematic search
Three main classes of improvement options exist for a petrochemical process that
has been labelled a weak system element:
Process System Innovation by Design 91

1. modification of system content

In any process system, the selection and arrangement of system elements as
well as their principle, design and operating conditions may be altered. This
can affect system performance by quantitative improvement (yield,
selectivity) or by an upgrade of output quality to match with product material
specifications that enable a stream valuation change.
2. a shift in the classification of streams around a particular process
Albeit determined by economy and ecological considerations, and sometimes
formalised by Law or regulations, the classification adopted of both inputs
and outputs is subjective. Such a change in stream valuation can be effected
by development and adoption of alternative material applications in the
enveloping system. As a consequence, while the system internals remain
unchanged, its performance indicators do change ( except Le and ηe).
3. Modification of system content of the petrochemical industry
Technology development for the reduction of the margin between ideal
reversibility and irreversible practice must include alternative process routes.
Opportunities that involve system change outside the chemical plant system
boundary must be explored with the explicit condition that technology and
structure of subsystems within each chemical plant are a degree-of-freedom.
4. Generalisation of options identified
The specific weak elements identified allow identification of specific
innovation content for a single plant. The opportunities thus identified may
have wider applicability, possibly to all weak elements of the petrochemical
industry identified. Combined, these may be of sufficient ‘global’ character to
allow the formulation of more widely applicable concepts and directions for
R&D.

In the mature petrochemical industry, process innovation is a much-applied element

of competitive strategies. Process innovation predominantly focuses on everything
within the system boundary of an existing plant or plant conceptual design. Indeed,
open literature gives proof that improvement opportunities for almost every
commercially applied chemical processes are being investigated. Process innovation
can be realised by a combination of expert knowledge, literature search, process
simulation and laboratory work. Options considered can involve changes in process
system design, the set of unit operations applied, change in operating conditions, the
selection of the technology for a particular unit operation, and last but not least the
catalyst system employed. However, the thermodynamic efficiency limit of industrial
processes is constrained not only by the technology selected and catalyst available,
but also by the particular chemical reaction completed. Indeed, after (Denbigh 1956),
it is the responsibility of the process engineer to make “a suitable choice of the path
between the given initial and final states, i.e. the choice of the details of operation.”
The 'grey-box' approach (§3.2.1), for example, is widespread and entrenched in
chemical engineering. In the analysis of the petrochemical industry it gives useful
clues of what is going wrong. In the search for innovations, however, it must be
realised always that the distinction between process and utility is an artificial one, one
92 Process System Innovation by Design

determined by technological skill and accepted or best technological practice at the

time of construction of some facility.
Relatively straightforward opportunities can be identified by looking beyond the system
boundary of the particular process studied and considering options for change of the
stream labels around individual petrochemical plants, for example to reconsider the
fate of unwanted by-products. In many plants their utilisation is within the plant
system-boundary by recovering the fuel-value only. To upgrade their utilisation
requires fuel replacement and identification of some destination of the by-products.
Nevertheless, to upgrade by-products and to reap the associated benefits is a
longstanding tradition in the petrochemical industry, which in many cases has led to
new business development. In modern styrene units, for example, the hydrogen
present in recycle off-gas is captured for pure hydrogen production. In caprolactam,
the large exergy loss and ammonium sulphate production were an incentive for the
development of a new generation of caprolactam processes.
In catalyst research traditionally, much research and development is devoted to
finding the ultimate catalyst that combines high selectivity and conversion. Looking
beyond the boundary of a single plant, however, suggest the development of
effective catalysts for “high conversion” that are not necessarily selective to a single
product. These catalysts must be employed in intensive process systems where the
combination of unselective reaction and product separation can be effectuated at the
expense of limited exergy loss. Concurrently, the utilisation of all products and the
associated business must be developed to ensure their being labelled co-products or
wanted by-products.
In the petrochemical industry structure and content changes when new types of
plants are included or existing ones are phased out or changed. Therefore,
technology development for the reduction of the margin between ideal reversibility
and irreversible practice must include modification or replacement of process routes.
Thereby, an under performing process can be phased out. In the production chain
for PET, for example, developing a one-step process from ortho-Xylene to polymer-
grade terephthalic acid that avoids the production of intermediate DMT has reduced
energy-loss.
A multiplier-effect may be achieved when general but sufficiently specific R&D
themes can be formulated. This was the central conjecture of 'de grondstoffenstudie'
(Weijnen 1991). The development of trigeneration systems (Ch. 6, Dijkema 2001b) is
such a theme that is a generalisation of the integration of energy conversion
technology in chemical plant design. Trigeneration implies a change of the structure
and content of the petrochemical industry. In trigeneration designs utility outputs no
longer are considered 'unwanted but accepted' by-products, but 'wanted' co-products.
As the industry had been assessed and explored, upon completion of the methods
presented in § 3.2, however, only a limited number of promising R&D directions
emerged (Dijkema et al. 1995), possibly because of the long process of development,
evolution and maturation that led to a petrochemical industry characterised by
strong procedural cohesion. Whilst in the analysis, the adherence to procedural
cohesion could only be determined after examination of sufficient systems, the
search results indicate that for the petrochemical industry the supposed multiplier
effect of procedural cohesion is limited. In retrospect, this may be explained by the
Process System Innovation by Design 93

industry's evolution. While the scientific knowledge of chemistry, chemical

engineering and related disciplines expanded continuously, process economics and
innovation strategies contributed to the emergence of 'best-practices'. In response to
some company's (procedural) process innovation, the other players in the industry
will adopt and apply these to their advantage and competitive. Meanwhile,
regulations, environmental and quality increasingly have been translated into
procedural standards.
Similar developments have shaped many ‘engineering disciplines or sub disciplines’,
which exhibit some procedural cohesion that allows their use in various domains of
application. Anyone trained in chemical engineering, for example, can apply his/her
knowledge not only in the chemical industry but, amongst others, also in power
systems, oil refining and pulp and paper where the concept of unit operation has
proven to be very useful. In the base metal industry, a similar transition is being
effected making use of the unit operation concept, progress, however is tedious
because of the complexity of the physical and chemical processes in many
metallurgical plants (Reuter 1998).

3.3 Functional modelling for process system innovation

3.3.1 INTRODUCTION
In every design of a novel industrial plant for the production of commodities, e.g.
ethyl benzene, the particular selection of unit operations, their implementation and
interconnections are degrees-of-freedom. In actual practice, however, a ‘best’ or
common design appears to have been developed for a great many plants. The
design-space is limited and process flow sheets are hardly ever produced from
scratch (Siirola 1996). Designs of industrial chemical plants thus show considerable
procedural cohesion. As a consequence, by outsiders the development of the
petrochemical industry is perceived to have entered ‘the end-of-the S-curve’, where
there is only scope for 'incremental' improvements when adhering to existing plant
structures. Meanwhile, process engineers are hardly ever confronted with 'ill-defined
statements of a need', primitive problems (Rudd and Watson 1968). Despite the
research agenda's suggested (Ch.2), the focus of process system engineering still is
individual chemical plants. To ‘get out of’ this lock-in and foster process system
innovation, functional modelling is introduced, and related to 'technology-free'
systems representation (section 2.3.3) using the input-output paradigm. Functional
cohesion is used as a basis for system decomposition because modelling using logical
decomposition (§3.2) suffers from difficulties in system boundary selection and lacks
a stop-criterion for decomposition-depth.

3.3.2 FUNCTION AND MODELLING

The description of function in a contemporary English Dictionary is “The function
of something or someone is the useful thing that they do or are intended to do”
(Sinclair 1995). This captures two aspects of ‘function’, viz. the organisational notion
of function, for example ‘the R&D function“, and the objective of some entity or
artefact that combines behaviour and performance valuation, for example, ‘the
94 Process System Innovation by Design

function of R&D is timely development of new products’, or ‘the function of a clock

is to show the right time’.
In technical objects, a function is synonymous with ‘a black-box filled with
operations leading to the fulfilment of an objective’. Hence, a system with function
X may serve to achieve objective Y. The function is generally concerned with what
should be achieved, not how, and thus opens the way to various means of action.
This relates to the definition by Asbjørnsen, who essentially argued that the
specification of a system element should be technology-free (see §2.5) (Asbjørnsen
1992).
In a system representation and characterisation using the input-output paradigm, we
assume that the function Φ of a system (Σ) is to generate a behaviour (B) to transform
a set of inputs (Su) into a set of outputs (Sy), which relate the system to the external
world. The performance of the system (Ω) is assessed by means of value functions (P)
(see Appendix A. 4). The completion of this transformation at suitable performance
then is the system's objective. This representation is a ‘technology-free’ specification
of a system or system element.
Let's define the system functions and technology30.
A system (Σ) is characterised by:
B: SU => SY (Eq. 3.25)

P: SU * SY => Ω (Eq. 3.26)

Σ = { SU , SY , Ω, B, P }. (Eq. 3.27)

The function of the system (Φ) is defined by the selection of { SU , SY , Ω }. The

system boundary (Λ) is defined by the selection of { SU , SY }.

Φ = { SU , SY , Ω }. (Eq. 3.28)

Λ = { SU , SY }. (Eq. 3.29)

Functional decomposition is based on the ability to decompose the system with

function Φ into sub-systems with sub-functions {Φk}, where k counts the sub-
systems.
This representation captures the objective-defined functions introduced by Baylin as
a basis for decomposition. “The mission (...) of each system is accomplished by
achieving a number of intermediate objectives, up to and including fulfilment of the
final system objective(s) An objective defined function is “a sub-set of the operations of a
system, consisting of only those operations directly necessary to the achievement of
the same specific objective (…)” (Baylin, 1990: 92). A physical system thus is
completely specified by its objective-defined functions Φ. Since neither the

30This section is largely based on personal communications (Grievink 1998) and internal research
notes (Grievink 2003) that elaborate on (Dijkema et al. 2003).
Process System Innovation by Design 95

technological means nor implementation for achieving behaviour B have been

specified, the description is technology-free.
It may be seen that the technology issue is very much related to the characterisation
of the system behaviour and its realisation. The system behaviour is the effect of
transformations induced by technological actions that obey the Laws of conservation
of mass, energy, momentum and electrical charge. These technological actions
always involve the transformations of physical and information variables. A
technological action can be characterised by the sources and the sinks of the
transformation and the associated rates of change. In the petrochemical industry and
chemical engineering practice the obvious transformations encountered are the
conversion of chemical species, which are characterised by stoichiometry and rate of
the reactions. In addition and often simultaneously, transformation of energy occurs
(chemical -> thermal; chemical -> electrical; thermal -> mechanical, …). Other
transformations, for example, are physical quantity into an information signal
(measurements) and the transformation of an information signal into mechanical
action (e.g. change of valve position). Given these considerations, it may be seen that
the behaviour (B) can be defined as the joint result of conservation laws of nature
(N) and the transformations (Ψ) within the system boundary (Λ), induced by
application of technologies (T) and the state of the stream(s) σ, on which T are
acting:

B = [ N | ΨΛ(T, σ) ] (Eq. 3.30)

A technology T, is defined as a mapping from an initial set of stream states, σ, in a

domain DT , to target stream states, ϕ ∈ DT (in the same domain):
T: DT => DT (Eq. 3.31)
Algebraically:

ϕ = T (σ) (Eq. 3.32)

The transformation can be expressed as:

Ψ Λ (T, σ) = ϕ - σ (Eq. 3.33)

The above description is analysis driven. Given σ and T one can determine Ψ, and
hence B. In a synthesis situation one has to deal with the inverse case. No unique
inverse relationship, however, exists to map from a specified transformation (Ψ’Λ)
onto a particular technology T’:

{Ψ’ Λ, σ} ≠ T’ (Eq. 3.34)

This implies there exists a design or innovation space that can be explored.
96 Process System Innovation by Design

Usually, one will set targets for the transformation from system inputs to outputs;
then, it will be possible to set targets for the required behaviour (Btarget ) and the
associated transformations: Ψ Λ , target .

Btarget = [ N | Ψ Λ , target ] (Eq. 3.35)

There can be multiple technologies {T} and stream states {σ} whereby the target
transformation Ψ Λ ,target can be achieved within the system boundary Λ selected.
In process synthesis it is recognized that often the candidate technologies {T’} will
belong to some superset of technologies capable of acting on certain type domains
of stream vectors. Whilst the synthesis problem then can have multiple feasible
solutions, the multi-objective optimisation of system performance can provide a
unique mapping or a non-unique mapping of a Pareto set of solutions.
Process synthesis can be represented as smart search strategy over the available
technologies to minimise the distance between the achievable transformations
Ψ(T’Λ, σ) and the targets Ψ Λ ,target :

Minimise || Ψ Λ (T’, σ) - Ψ Λ ,target || (Eq. 3.36)

{T’, σ}

The above equation expresses that known transformations in the petrochemical

industry are optimised in a design space that is limited by available or known
technologies T.
Indeed, our literature review indicated that the PSE community largely focuses on
the development of “new flow sheets for existing problems” (Ch.2, §2.4). In
exploring possible innovations, however, one must avoid being limited, or
entrenched in the current way things are being done, i.e. avoid 'lock-in' in a 'closed'
problem formulation for optimisation only.
The functional approach may focus, foster and enrich the discussion to address the
question of 'what innovative concepts', not 'how to improve realisation of known
concepts'.
By specifying functions Φ it gives decision-makers with a non-technical expertise a
chance to participate and to specify new system options by (re) defining system
functions or expansion of the system boundary. Subsequently, identification and
specification of the corresponding transformations Ψ must include expansion of the
set of technologies and streams {T’, σ} in concert with the system boundary Λ
selected.
In summary, for process system innovation a system Σtarget is defined by selection of
its system boundary and specification of the:
function Φ = { SU , SY , Ω }
system boundary Λ = { SU , SY}
targets for behaviour: Btarget
targets for performance: Ptarget
Hence: Σtarget = {Φ, Btarget, Ptarget, Λ }.
Process System Innovation by Design 97

The above treatise is focused on a formal representation of a system with a potential

realisation of its functions by means of transformation technologies T. In the
subsequent section, the focus shifts towards the constituting parts of a system, the
system elements, which are specifications of technologies T.

3.3.3 FUNCTIONAL COHESION

“Cohesion methods” is a well-known subject in computer system design and
programming literature (Baylin 1990: 349). The subject relates primarily to how to
arrive at a sophisticated and useful breakdown of (computer) programs or
information system. A number of cohesion principles31 or types exist, which we
assumed can also be applied to chemical process systems modelling.
Traditional or logical cohesion covers the signalled trivial but logical breakdown of
the petrochemical industry into in complexes and single plants. In the design of a
chemical plant, sequential cohesion must be achieved between its unit operations. At
the aggregate level of the industry, however, the decomposition is not obvious:
where does crude oil refining end, where does the petrochemical industries’
operations start. How must an ‘aromatics’ refinery be classified. In petrochemical
complexes the exact subsystem boundaries are not always clear. Within the boundary
of a single petrochemical plant it is often difficult to exactly identify or decompose
the unit operations. The stop criterion for the decomposition appears to be based on
pragmatism - down at the level of unit operations the number of alternatives
explodes- rather than having been embedded in some theoretical basis.
We found little treatise of 'decomposition' in the literature reviewed (§2.5 System
decomposition, p.59). Baylin indicated that ‘existing texts give a merely superficial
coverage of how to conceptualise a system’s functional (operational) structure” (Baylin
1990: xi)(italics added).
In functional modelling a system is broken down in such a way that each subsystem
corresponds to the sub-set of system operations that achieve one of the intermediate
or the final (system) objective-defined functions. This involves system abstraction
and decomposition in a trial and error procedure - but when is an adequately
structured model built and a suitable decomposition achieved?
The functional cohesion principle, as developed for (information) system
decomposition, involves “…keeping together related system elements, and keeping
apart unrelated system elements while at the same time avoiding both duplication
and conflict” (Baylin 1990: 57). This definition indicates that decomposition is in
part subjective.
Recognizing that a single system may have multiple functionality and may serve
multiple objectives, Baylin extended his definition of an objective-defined function
to “a sub-set of the operations of a system, consisting of only those operations
directly necessary to the achievement of the same specific objective, or the same
‘mutually contingent’ [see §3.3.3] specific set of objectives” (Baylin, 1990: 92).

31 See §2.5.3p.61 and Appendix A. 3, Overview of cohesion types for definitions and details on
logical, procedural and other types of cohesion.
98 Process System Innovation by Design

This concept of mutual contingency thus offers a more objective criterion for
adherence to the functional cohesion principle. The selection of objective-defined
functions, however, remains subject to interpretation.
When in functional modelling the condition of mutual contingency is met the
decomposition adopted is functionally cohesive. Functional modelling of the system
is successfully completed when functional cohesion has been achieved at each (sub)
system level. If not, the model and the decomposition must be modified. Thus,
mutual contingency provides both the foundation for a definition of functional
cohesion and a stop-criterion for the functional decomposition procedure.

A set S of functional objectives Φ is said to be mutually contingent in case

∇ (ΦX, ΦY) ∈ S (for all possible pairs of objectives)
ΦXY = (ΦX AND ΦY ); (both achieved in parallel)
OR
ΦXY = ( (ΦX OR ΦY) AND NOT (ΦX AND ΦY ) );
(mutual exclusion, either one achieved, but not both.)
OR
ΦXY = (ΦX →ΦY) (achievement of ΦX leads to achievement of ΦY).

In Figure 3-7 this is illustrated: if the complete set of objectives S is decomposed

into subsets S1 and S2, mutual contingency is achieved; the decomposition result
(subset S1 and S3), however, is not mutually contingent.
Consider for example a system Σ that is characterised by three objective-defined
functions, ΦX, ΦY and ΦZ. Each ΦX..Z can be realised (True) or not (False). The
system Σ then has a limited set of system states as listed in Table 3.5 that is not
mutually contingent because instances 1 and 3: {1 (ΦX AND ΦY), 3 (ΦY)} do not
comply with the condition for mutual exclusiveness. The subset (ΦY, ΦZ), however,
does meet the criterion of mutually contingency.
Table 3.5: Instances of example system characterised by three objective-defined
functions X, Y and Z
Objective ΦX ΦY ΦZ
Instance 1 True True
Instance 2 True
Instance 3 True

A (pseudo) algorithm has been written for the test for the criterion for mutual
contingency in functional system decomposition (Appendix A. 5).
Process System Innovation by Design 99

Objective-Defined Functions
{S1}
Specific Specific
Objective Objective
A B

Specific
Objective
C

Objective-Defined
Functions {S3}
Objective-Defined Functions
{S2}
Specific
Objective Specific
D Objective
E

Figure 3-7: Mutual contingency of sets of objective-defined functions ΦA….E.

The above definition opens the way to a decomposition method for the breakdown
of systems by selecting appropriate objective-defined functions. Each objective-
defined function must relate to “a sub-set of the operations of a system”, i.e. a
subsystem, which objectives serve to fulfil the general objective or mission of the
system in accordance with the system definition of Blair quoted above (Blair and
Whitston 1971).
A hierarchy of functions appears, where objective-defined functions are composed
of a number of functional items, which can be either intermediate functions, or
functional elements (Figure 3-8). Intermediate functions are conceptual components
of an objective-defined function. At least one intermediate objective equates to the
system objective.
These decomposition principles, including adherence to functional cohesion, must
be applicable to all three selection features of a system's function Φ, which has been
defined as { SU , SY , Ω } (Eq. 3.28) , i.e. the set of inputs, outputs and performance.
This is not much of an issue in selection of inputs and outputs, as long as:
a) The input and output streams are consistently characterised by the same attributes
(“stream vector“).
b) The merging and splitting of streams (arcs) is only done in “nodes“, which qualify
to be part of a (sub-) system.

Illustration: Styrene production

Let us consider a petrochemical complex where important precursors for plastics are
manufactured, viz. styrene and propylene oxide starting from naphtha as a feedstock.
In addition, TBA is produced, a gasoline additive (Figure 3-9). The owner of the
complex has some flexibility in catering for the markets of these three products. Is a
100 Process System Innovation by Design

logical decomposition of a production site by the processes shown also a mutual

contingent one?

Objective Defined Function

Intermediate functional item

Functional Functional
Element Element

Intermediate functional item

Functional Functional
Element Element

Figure 3-8: A hierarchy of objective-defined functions.

The overview of the system (Figure 3-9) is only a summary that lists objective
outputs only. To check the decomposition shown for mutual contingency, the
possible states or instances of the system are related to the production of a single
molecule of the objective output or outputs. Thus, in the test a molecule of
propylene produced can only be used for a single objective. To circumvent the
problem of defining the composition of petroleum cuts and the variety in product
spectrum, in this example it is assumed that out of ‘1 pseudo molecule’ of naphtha,
simultaneously 1 molecule of ethylene, propylene, isobutane and benzene are
produced. The system thus is defined by its objective-defined functions: produce
PO, styrene, and TBA. The subsystems are defined by the intermediate objective
outputs: produce ethyl benzene, or petrochemical ethylene, propylene, benzene or
isobutane. The instances that will realise each pair of objectives are listed in Table
3.6. Each instance is a set of processes by which a single molecule of naphtha can be
transformed. The objectives to produce a single molecule of ethyl benzene and PO,
for example, are achieved in parallel, when EB and PO-TBA are operated, or
sequential, when EB produced is used in the PO-Styrene process.
The decomposition of the system for the production of propylene-oxide, styrene
and TBA presented is a mutual contingent set of objectives (Table 3.6). Especially
the objectives to produce styrene and TBA are mutually exclusive because out of a
single molecule of naphtha a single molecule of propylene results, which must be
sent to either the PO-TBA process, or to the PO-SM process. From the scheme a
‘mutual contingent’ option for process integration is obvious, i.e. the production of
EB as an integrated part in the PO-SM process.
Let us now consider an industrial complex where additional commercially available
technology for the production of these chemicals is to be incorporated. This leads to
a system where novel technology has been added for the production of PO and
Process System Innovation by Design 101

styrene respectively (Figure 3-10). In a model of the complex, these technologies can
be grouped to yield a subsystem that is similar to the PO-Styrene plant, a logical
approach for the comparison of system alternatives.
The decomposition of the complete industrial complex by these subsystems,
however, does not meet the criterion of mutual contingency! In the new subsystem,
the objectives of ‘produce PO, ‘produce styrene’ are not mutually contingent: when
the subsystem is part of the larger system, depending on the route of both PO and
EB, either one can be achieved, or both! The possible instances for the processing of
a single molecule of naphtha are listed in Table 3.7.
The solution, of course, is to modify the decomposition, and to separate the PO and
Styrene-process again. This decomposition is mutually contingent.

From these examples we see that:

• Production systems can be objective-defined by their products
• A logical decomposition based on similar activities may not always result in a
mutually-contingent set of objectives
• Domain-specific knowledge is required to arrive at such decomposition.
• A mutually contingent set of objectives is arrived by iteration: a first attempt at
decomposition is checked for mutual-contingency. Based on the results, the
decomposition is either accepted or improved.
• Some subsystems generate ‘non-objective’ outputs.

Cohesion in process system design

Baylin states that the functional cohesion method can be applied to any kind of
business. We conjectured that -with modification and extension- the method of
functional decomposition increases our understanding of large-scale manufacturing
systems as found in the petrochemical industry. The characteristics that determine
the dynamics and the scope for change differ considerably between physical
production systems and information system. These, however, are outside the scope
of this thesis. Relevant differences are found in the coupling between system
elements and the integration of functions respectively.
Cohesion methods and system decomposition methods aspire to arrive at an
adequate degree of ‘loose-coupling’ between subsystems, or system elements. Types
of coupling are no coupling, pooled or indirect coupling, sequential coupling and
reciprocal coupling. Keeping unrelated things apart then mainly is concerned with
the activity to study coupling, and eliminating the undesirable ones (Baylin 1990: 35).
In business information systems, these concern couplings between data, and
procedures that act on data.
In physical production systems, couplings have a physical implementation: some
intermediate storage facility and transportation by road, rail, water or air, or, in the
continuous process industry, direct coupling via pipelines. In production systems,
however, at any one time, there will be found ‘reciprocal couplings’ in the form of
recycle streams, that involve a number of subsystems or system elements identified.
 102

Isobutane
Naphtha steam cracking
Oxygen per-oxidation
Propylene
separation Oxide
Propylene epoxidation

Steam Cracker TBA

PO-TBA process
Ethylene Benzene

Oxygen

alkylation Ethylbenzene per-oxidation

Ethylbenzene process
Propylene
Propylene epoxidation
Oxide

dehydration Styrene
Legenda
Intermediate
Objective- Objective-
Input output PO-Styrene process
Output

Figure 3-9: Overview of the production system for styrene and propylene oxide.
Process System Innovation by Design
Process System Innovation by Design 103

Table 3.6: Instances of the styrene /propylene oxide production system

Objective output Petro- Ethyl Propylene- Styrene
ΦX / ΦY chemicals benzene oxide
Petrochemicals -
Ethyl benzene A1 (NAF,EB) -
Propylene-oxide A1 (NAF, Pl ((EB, PO- -
EB, PO-SM) TBA) or A1
or (NAF,PO- (EB, PO-
TBA) SM))
Styrene A1 (NAF, A1 (EB,PO- Pl (PO-SM) -
EB, PO-SM) SM) or Mx (PO-
TBA)
TBA A1 Pl (EB, PO- Pl (PO-TBA) M (PO-SM,
(NAF,PO- TBA) or Mx (PO- PO-TBA)
TBA) SM)

Notation used in Table 3.6 and Table 3.7 (p.105)

Petrochemicals = (ethylene, propylene, butylenes, benzene)

Labels of instances:
Pl = (ΦX AND ΦY) (parallel)
M = (ΦX XOR ΦY) (mutual exclusive OR);
Mx = only ΦX can be achieved; both (ΦX AND ΦY) and (only ΦY) are not
possible (singularly exclusive towards ΦX - condition M is not met,
only condition Mx is met).
Ax = ΦX implies ΦY (achievement of ΦY through ΦX)
Ay = ΦY implies ΦX (achievement of ΦX through ΦY)
O = ΦX OR ΦY (inclusive OR: ΦX, ΦY or ΦX AND ΦY)

Parameter values for X and Y:

When X or Y indicate operate process …
naphtha cracker' = NAF
ethyl benzene = EB
PO-Styrene = PO-SM
PO-TBA = PO-TBA
 104

Isobutane
per-oxidation
Oxygen Propylene
Naphtha steam cracking Oxide
epoxidation
Propylene
TBA
separation PO-TBA process
Ethylbenzene

Oxygen per-oxidation
Steam Cracker
Propylene
Propylene epoxidation
Oxide
Ethylene Benzene
dehydration Styrene

Chlorine PO-Styrene process

alkylation Propylene HCl

hydrochlorination
Water Propylene
Ethylbenzene process
Oxide
NaOH dehydrochlorination
NaCl

Figure 3-10: Instances of the extended SM/PO system.

Ethylbenzene
Legenda de-hydrogenation Styrene
Intermediate
Objective- Objective- Hydrogen
Input Output output PO / Styrene processes
Process System Innovation by Design
Process System Innovation by Design 105

Table 3.7: Overview of instances in the extended Styrene /PO production system
Objective output Petro- Ethyl Propylene- Styrene
ΦX / ΦY chemicals benzene oxide
Petrochemicals -
Ethyl benzene A1 (NAF, -
EB)
Propylene-oxide A1 ((NAF, Pl (EB, PO- -
PO-SM) or TBA) or
(NAF,PO- A1((EB, PO-
TBA) or PO- SM2) ((EB,
SM2) PO-SM),
Styrene A1 (NAF, A1(EB, PO- Pl ((PO-SM) -
PO-SM) or SM) or (EB,or (PO-SM2)
(NAF, PO- PO-SM2) or (PO-SM2,
SM2) PO-TBA)
TBA A1 Mx(EB, PO- Pl (PO-TBA) M (PO-SM,
(NAF,PO- TBA, PO- PO-TBA)
TBA) SM2)

Note: for explanation of table and symbols, see notes below Table 3.6, p.103

These result in a reversal of activity flow and introduce feedback loops, which may
introduce complex dynamics that affect system performance. An equivalent reversal
of information flow appears to be largely avoided in business information systems.
Information and control systems do create direct couplings in physical production
and couplings between production control systems, scheduling and management
information systems. At the bottom level control system feedback loops represent
reciprocal couplings that may span multiple unit operations or plant sections. The
feedback from management information systems, scheduling and optimisation also
may introduce reciprocal coupling.
The ‘process recycle’ must be considered an important part of a great many (petro)
chemical plants. Its functions often are multiple: -reprocess unconverted material;
provide initiator/seed for reaction; circulate catalyst; allow safe process conditions;
intensify the use of reactors etc. Indeed, an important step in conceptual design is
the definition of the recycle structure (Douglas 1988). At the level of industrial
clusters, industrial networks, or industrial infrastructure notably materials are
recycled through a series of networked plants, the major problem in the metals
industry being how to keep metals within the cycle, maintain their quality and
availability (Verhoef et al. 2004a).
The recycles create interdependency between the system elements. Using functional
decomposition of process systems, it may be seen that a process recycle that results
in a reciprocal coupling is only a technological implementation of some function in
the black-box of a chemical plant that purposely must convert substances A and B
to product C. In Chapter 5, Fuel cells and Trigeneration, it is demonstrated that
106 Process System Innovation by Design

considering the recycle as a crucial functional system element leads to dramatic

improvements in system design. At the level of industrial networks, it is a long-
standing tradition in the petrochemical industry to not only upgrade by-products as
much as possible, but also to allow 'reciprocal couplings' over a multitude of plants.
One well-known example is the treatment of aromatic mixtures in an aromatics
complex to Benzene, Toluene and Xylene, which is implemented as a series of plants
that includes a number of intra-process recycles (Chauvel and Lefebvre 1989a). In
industrial ecology, these concepts have been labelled 'interconnecting and cascading'
material and energy flows; however, this emerging inter-discipline does not offer a
unified approach to recycles in interconnected systems (Graedel and Allenby 1995;
Ayres and Ayres 1996; 2002).
As we have shown, the use of objective-defined functions is in accordance with the
above definition quoted from (Asbjørnsen 1992), where a system element
specification is technology-free. What is imminent from the case in Ch. 5 is that in a
great many production systems, at various levels, there is an increasing level of
integration of system functions, i.e. in the technological specification of systems,
single items achieve seemingly inseparable multiple functions. Reactive distillation,
for example, offers an example where chemical reaction and subsequent
separation/purification are combined in a single apparatus. At first glance, in
information systems there does not appear to be the equivalent of incorporating
multiple functionalities by combining functions in a single item.
Of course, ‘system integration’ is a well-known aspect of both hard- and software
but this largely is concerned with gluing system elements together to a properly
function system. Thus is this a unique characteristic of manufacturing systems in the
process industry?
Is it only in appearance that computer systems combine functions in single items,
while apart from parallel supercomputer, transputers and neural networks most
microprocessors only allow sequential execution of system functions? Without
answering these questions, the cases presented in this thesis support that functional
modelling is a very effective means to identify, illustrate and manipulate multiple
functionality of technological systems or parts thereof.

3.3.4 DECOMPOSITION PROCEDURE

In the preceding sections the functional cohesion concept has been introduced,
formalised and illustrated. The analogy and difference between physical systems and
information systems were discussed. In this section, first a procedure is presented to
arrive at a functionally cohesive decomposition using the mutual contingency
criterion. The initial functional modelling strategy subsequently is improved by
analysing model decompositions employed for the petrochemical industry or parts
thereof. To that end, the literature discussed (chapter 2) was readdressed with the
question 'has functional cohesion been achieved in the models of the chemical
industry or have other types of cohesion been achieved'. Finally, the functional
decomposition procedure is embedded in a framework of cohesion methods.
Process System Innovation by Design 107

Functional decomposition procedure

To achieve functional cohesion in system decomposition, all system elements that
are closely related, either directly or indirectly, to the same intermediate or final
mutually contingent set of objectives must be clustered together within the same
subsystem. This can be tested for. At the system level, there are also some
exceptions:
1. Some subsystems may not be adequately coupled
2. Subsystem selection would result in duplication of system elements
3. The selection would result in subsystem conflict
Duplication or subsystem conflict can result when some system elements are closely
related to more than one intermediate or final set of system objectives. To avoid
duplication, the elements must not be placed in more than one subsystem. In order
to avoid subsystem conflict the elements involved must not be placed in anyone of
the subsystems that correspond to the common objective, but, rather, into a
separate, distinct subsystem of their own. By obeying these principles, functional
cohesion attempts to achieve optimal independence of different subsystems.
Using these concepts, a list of approximate steps to follow in actual practice can be
given, as illustrated by Figure 3-11:
#1 “Identify intermediate objectives of the parent system, and then group all
(cluster) together all the system elements closely related to each intermediate
objective, or to the final objective“
#2 “eliminate from each group (identified in #1) those elements not closely related
to the objective(s) of the group (cluster).”
#3 separate sub-groups of system elements into distinct subsystems of their own
which are:
• Insufficiently coupled within the group
• Common to a number of different groups already identified
• Tightly coupled to a number of different groups already identified
#4 Check for functional cohesion at system, subsystem(s) and system element levels.
#Repeat steps #1-3 depending on the outcome of step #4.
An initial step is required where the functional objective of the parent system is
identified. As indicated, an iterative procedure results.
Describing the functions of systems, subsystems and system elements requires
domain specific knowledge and multidisciplinary input. The procedure implies some
‘trial-and-error’ because execution of step 1 already assumes some consideration of
the systems objectives and possible formulation of system intermediate objectives.
To start the iteration, it is suggested to use an initial guess that can be based on
present knowledge of the decomposition of a particular system. Depending on the
result of #2-#4 the decomposition must be modified, i.e. begin at step 1. In this
way, a complete procedure for the functional decomposition of systems has been
defined.
108 Process System Innovation by Design

Decomposition in Industry Studies and Design

In this section, decompositions of industry or parts thereof as applied in studies
reported in the literature are analysed with respect to functional cohesion. The
examples align with the presentation of studies of relevance to 'petrochemical
industry' adopted in Chapter 2, Process System Innovation Sources, i.e. from 'top-
down' to process design of chemical plants.

Identify
1
Intermediate Objectives

Eliminate
2
Unrelated Elements

Separate
3 Modify / Adjust
Sub-groups

Check for
4
Functional Cohesion

Complete

Figure 3-11: The functional decomposition procedure.

In 1909, Weber published his theory on the location of industrial operations (Weber
1909). He decomposed the processing industry into two subsystems, viz. one where
“Reinrohstoffe” are being processed and another where “Verlustrohstoffe” are
being processed. The petrochemical industry falls completely into the first category,
since crude oil and natural gas must be considered a Weberian Reinrohstoff, i.e. a
resource that can almost completely be converted into useful products at the
expense of only limited generation of solid-waste. On the contrary, most ores are
Weberian Verlustrohstoffe, since for the greater part they consist of rock that is
considered ballast.
Weber's decomposition is still useful for the explanation of the general location
pattern of process’ industry facilities. In retrospect we conjecture that it is a
functionally cohesive decomposition: the process industries’ activities were
decomposed in a subsystem where activities are grouped together that completely
Process System Innovation by Design 109

process feedstock to saleable products, and a subsystem where the operations

generate a large ‘non-objective’ output.
In explaining the location patterns of oil refineries and the petrochemical industry in
Europe, (Molle and Wever 1984) distinguished 4 categories of complexes. Coastal
complexes are located in seaport regions, which offer tremendous advantage for the
shipping of bulk feedstock and product. Today, their nucleus more often than not is
an oil refinery that provides feedstock for petrochemical plants, although some argue
that at least a naphtha cracker is required for long-term complex viability (Dijkema
and Kuipers 2001). Older established inland complexes are found in the regions where the
chemical industry historically developed based on traditional feedstock, coal or river
access. Over time, the facilities on these sites gradually were adapted or replaced by
new plants that use oil products as feedstock. Hybrid complexes developed inland
around newly established oil-refineries as a response to local market development.
Subsequently, these could also meet feedstock demand of new petrochemical plants.
Growth-pole complexes developed where public authorities stimulated joint construction
of both oil refineries and the petrochemical industry largely by support beyond the
provision of 'public infrastructure'.
Molle and Wever’s is a logical classification that is of value to understand current
location patterns. It is not a functionally cohesive decomposition, but largely a
description of the result of a chaotic decision process under changing market
conditions, availability of transport modalities, and the role of public authorities. The
result of these evolutions exhibit largely ‘coincidental’ cohesion only. Their study
supports the impact of entrenchment in the petrochemical industry: new facilities
preferentially are constructed at or in the vicinity of existing petrochemical sites.
Thus, all petrochemical complexes exhibit 'relational cohesion', as all industrial
operations relate to the object-class ‘manufacture of chemicals’
The objective-defined functions we can extract from Molle and Wever's
classification are (#1-Produce oil products; #2-Produce petrochemical feedstock;
#3-Produce (petro) chemicals for local markets; #4-Provide jobs to stimulate local
job market). In order to provide economic feasibility, relations with the system
environment must include (#5-Access to large-scale transport facilities or cheap
connection to global resource market; #6-Closeness to local market or optimal
position with respect to regional markets and transportation costs and #7-access to
service industry).
In conceptual process system design the system element is the unit operation. The
composition of an appropriate set of unit operations is the basis of current
conceptual process design (Douglas 1988; Biegler et al. 1997). The unit operation
classification is procedural cohesive. Unit operation specification is not completely
technology-free, as by the mere training of chemical (process) engineers, each unit
operation has become associated with a series of methods of realisation. In addition,
when realising process functions the number of alternative combinations of unit
operations is virtually unlimited. The simple function ‘move a substance from
reactor A to B’ for example can be performed for a liquid by the single unit
operation ‘pumping’. Generally, requirement and realisation of the unit operation
‘pipeline’ conveniently is taken for granted. An alternative realisation of moving a
liquid from A to B would be the series of unit operations ‘evaporate’, ‘compress’,
110 Process System Innovation by Design

‘pipeline’, ‘condense’. In this case, or in the case of moving vapour, the unit
operation 'compress' is considered early on in the design process because of the
substantial cost of compressors.
In conceptual plant design, pipelines are not normally considered a unit operation.
'To move substances from A to B', however, is an important functional element in
industrial process plants. This function can be implemented using current state-of-
the art technology and system or by realisation of the concept of a pipeless plant
(Drinkenburg 1999). Douglas emphasized that in conceptual design not only the
selection, conditions and arrangement of unit operations are important, but also the
recycle structure (Douglas 1988; Biegler et al. 1997). We suggested considering
'recycle' in a chemical plant to be a unit operation (Ch. 5).
In the (strategic) planning process for the erection of new chemical plants early
estimation of investment costs is of imminent importance. Due to the large-scale of
operations, it is well known that the actual investment will dominate the net earning
power of the project for a long time. Over time, various methods for the generation
of investment cost estimates have been developed (Chauvel 1981; Peters and
Timmerhaus 1991). In estimating the costs for a new plant, the design of plant must
be broken down into individual units. In their founding article on their method for
the estimation of the investment costs for a chemical plant, Zevnik and Buchanan
actually defined the system element of a chemical process system to be a functional
unit and labelled their method functional unit method (FUM) (Zevnik and Buchanan
1963). At DSM, this cost estimation method has been expanded upon the
recognition that required functions in a process plant can be identified. Each
function can be given an estimated cost prior to knowing or specifying its
technological content. (Drinkenburg 1994; van Geem 1994). Thus, even in the very
early phase of R&D, if some new technology appears to be useful to create a new
production process, a cost estimate can be generated. A drawback of the method is
that in process system design, it is often difficult to clearly define a functional unit,
which very often turns out to consist of more than one unit operation. Possibly, the
procedure outlined in Figure 3-11 offers a workable recipe for the functional unit
method for cost engineering, especially when a basic, expandable set of functional
units that is applicable to all chemical plants can be identified. When this is the case,
a (historic) database of possibly a context dependent cost factor for each functional
unit can be derived.
In simple product assessment, the performance or fit-for-purpose of two alternatives
are compared. In energy analysis, primary, secondary, tertiary and quaternary
resource/energy use may be accounted for (Boustead and Hancock 1979) (§3.2.2)
In Life Cycle Analysis (LCA), such analysis is extended to the complete production
system that is required to bring a particular product into being. Modern LCA or
cradle-to-grave analysis attempts to determine and allocate to a single product or
'functional unit' the effect of mining, operations, production, consumption and
waste processing, recycling and disposal (Setac 1994).
Whilst this decomposition commonly adopted in LCA analysis probably is
functionally cohesive at the level of production systems, the entire approach of many
an LCA-study is not. LCAs are often executed to obtain a fingerprint of a particular
product and its related production chain or network. The result is claimed to offer a
Process System Innovation by Design 111

starting point for improvement by selecting other products or production processes .

Where in LCA analysis, assessment of production systems and products, however, are
tightly coupled, in searching for solutions they should be considered loosely
connected. The process network that currently creates a particular product's LCA
fingerprint is only one out of many possible system realisations. In addition, in many
an LCA study it is neglected that production systems for a variety of products are
not only linked but also interdependent (Verhoef et al. 2002b). This interdependence
does cause problems when allocation of environmental effects for product
assessment is required, and introduces complexity when developing solutions. Thus,
often process system innovations are either excluded from the solution space or
modifications are proposed that do not take into account the interdependencies in
the process network. LCA thus suffers from over specification of the production
systems analysed.
The studies evaluated indicate that the decomposition is largely implicitly adopted.
As a consequence, most underlying classifications are useful only for the specific
study. Relational cohesion has explanatory power for the location of industrial
operations, and visualise the importance of the relations with the systems
environment. Procedural cohesion in retrospect explains a large part of the current
status in chemical process systems engineering, where the unit operation concept is
widely used and allows the breakdown of process systems in elements, thereby
creating the opportunity to rearrange the process system configuration and reselect
the technology implemented. Its procedurally cohesive character, however, also
presents a barrier to process system innovation because it is difficult to step away
from 'best practice' both in process system design and unit operation
implementation. Functional cohesion explains the success some companies have had
with extended versions of the functional unit methodology in cost estimation for
novel chemical plants.

A recipe for Innovation and sustainability?

By specifying systems, subsystems and system elements by objective-defined
functions one opens the way to a technology-free system description. The other
cohesion methods, except relational cohesion, all require some classification or
typology of technology, and inadvertently will bring in the limitations associated with
those technologies, and the classification scheme. Often, logical classification
schemes are also functionally cohesive. They, however, do not open the way to
innovative concepts, as equipment always must be classified within an existing
category. The introduction of a new category often is a lengthy process and there is
always technology that does not fit in any category.
Another drawback of logical and procedural classification schemes is that when
these cohesion methods are applied in high-level system decomposition, subsystem
duplication or conflict may result. This may lead to endless reconsideration,
reordering and restructuring of classification schemes. In the process the focus on
and scope for improvement is lost.
The use of logical classification at a lower subsystem level, e.g. in equipment
engineering, appears natural to engineers. It seems logical, for example, to
distinguish between pumps and heat exchangers. Procedural cohesion can be useful
112 Process System Innovation by Design

in developing new equipment by examining similarity. In mechanical engineering, for

example, ‘rotating equipment’ can be considered a procedural category that includes
well-known equipment, such as compressors, turbines and pumps.
A hypothesis of a major study into more effective resource utilisation (Weijnen
1991) was that industrial operations in process industries, oil refining, base metals
and chemicals, exhibit procedural cohesion. The objective of the study was to
elucidate the 'bottlenecks' and derive innovative solutions that because of procedural
cohesion could be transferred to other operations in the industry.
While logical and procedural cohesion help in the initial stages of both modelling
and assessment, upon re-evaluation of the results (Dijkema et al. 1995) we conclude
that functional cohesion is to be preferred for the search and specification of
process system innovation content. Subsequently, procedural cohesion can play a
role in the transposition of solutions found in one sector to solutions in the other
sector.
A complete recipe to improve the ‘sustainability’ of a sector by assessing what the
activities within the sector are thus can be formulated as:
• Qualitatively investigate a suitable ‘top-level’, preferably by functional analysis,
and determine the appropriate system level.
• Look for possible procedural or logical decompositions, and search the literature
along the lines of these decomposition schemes
• Use the information thus gathered to develop a functionally cohesive scheme for
decomposition of the system under study. Define the lowest (sub) system level,
the system element, down to which functional cohesion has to be met.
• Identify subsystems where in current practice strong procedural cohesion is
manifest. When these subsystem coincides with subsystems with a relatively low
performance, the functional decomposition can be used as a stepping-stone
towards novel procedural = technology implementations to meet the objectives
identified.

3.4 Conclusions
In this chapter, the second part of this thesis' central research question was
elaborated: how can technological or systemic innovation content be specified that
enables a transition towards a sustainable petrochemical industry?
Firstly, a method has been developed for system modelling and characterisation for
assessment of resource utilisation, the main indicator for sustainability. Its
foundations are system representation using the input/output paradigm and
straightforward system decomposition where the petrochemical industry is
considered a structured collection of single chemical plants. Mass and energy
balancing, thermodynamics and system engineering were combined in a suitable,
reliable and versatile assessment procedure, which provides a quantitative image of
the petrochemical industry or parts thereof from information that is available in
open literature. The criteria or indicators calculated in the procedure could be
developed employing subjective but robust valuation by labelling system
input/output flows. These labels combine economic and ecological considerations.
Process System Innovation by Design 113

Together with energy values, these provide a sufficient 'proxy' for exergy. In
addition, the stream labelling approach presented provides an interface between all
stakeholders involved.
The assessment results can be condensed into an overview of 'scope for
improvement' versus 'current loss generated'. Obviously, petrochemical plants or
complexes that exhibit a large scope for improvement and a high current loss must
be high on the innovation priority list. Starting from such a priority list, a systematic
search for innovations must include identification of causes within the 'weak
element' combined with inspection of its system surroundings. Since inefficiencies
and losses are manifested as material or energy streams with inferior quality labels;
these labels may be changed to reduce the perceived loss and efficiency of process
systems or petrochemical complex involved. Thus a technology-oriented procedure
has been developed to systematically explore the 'innovation-space' for the
petrochemical industry. This includes the network structure, which is explicitly
considered a degree-of-freedom. We suspected that such a procedure would be most
effective to foster the science- and technology-fuelled innovation process (Figure
1-7), as well as help to balance short-term consumer needs, mid-term business
objectives and long-term sustainability of our industrial society at large. The
illustrations for the industrial aromatics system are no proof, but do demonstrate the
usefulness of the method compiled. Application of the procedure does yield options
for improvement viz. conceptual specification of innovation content. Elsewhere,
more extensive results have been reported that cover the entire petrochemical
industry (Dijkema et al. 1995). Overall, as stated, both the specific results per process
or complex, and more generic results, however, were meagre. This may be due to
very character of the petrochemical industry, a science-based industry with an
impressive track record of technological innovation driven by economic and
environmental incentives. Alternatively, the results can be viewed as confirmation
that reduction of resource consumption and CO2 emission indeed is a wicked
problem for the petrochemical industry (Ch. 1).
Section 3.2 represents the first completion of the circle of the meta-model of
activities (Figure 1-8) where the subsystem decomposition and boundary selection
remained unexplored. In short, a relatively straightforward systemic approach was
compiled for the structured identification of innovation opportunities in the
petrochemical industry, notably R&D options for improved resource utilisation.
In section 3.3, the modelling strategy itself was considered a degree-of-freedom,
notably system characterisation and system decomposition. Functional modelling
(Baylin 1990) was linked and formally embedded into the system representation
using the input/output paradigm. Functional cohesion was adopted as the principle
for decomposition and modelling strategy. A decomposition procedure was
developed using mutual contingency as a stop-criterion.
Our review of process system engineering literature (Ch.2) indicated that until 2002
the process system engineering community has been largely focused on the
development of “new flow sheets for existing problems,” thereby optimising known
transformations in the petrochemical industry while operating in a limited design and
solution space. Functional modelling, however, provides a mechanism for
abstraction that is expected to open the way to redefine functions of the existing
114 Process System Innovation by Design

industry and its production systems, explore novel means of achieving said functions
through system rearrangement and/or novel technology, and finally to explore
creative applications of available technology. In providing a means to decouple from
current practice it resembles and relates to state-task representation: functional
modelling allows the proper formulation of objective-defined functions of existing
operations based on past design. Hidden or new, emergent objective-defined
functions required for sustainable process systems may be identified. Thus, the
functional modelling strategy presented may become an activity that precedes
process synthesis, which mainly assists in primitive problem formulation and
specification of the design-space to be explored in process synthesis. To use the
approach requires a thorough consideration of the functional objectives of the
(production) system studied, and in essence will yield a subjective decomposition
and model.
The general system representation explains the limits of process synthesis for
process system innovation. System boundary selection and functional modelling are
a means to identify and specify novel or alternative system functions and include or
specify novel technology in the innovation space. Thus, re-addressing system
modelling and system decomposition (meta-model, Ch.1) has led to a new
modelling-decomposition-synthesis strategy or a procedure for the specification
process system innovation content.
 4 Innovation around Olefins

4.1 Introduction
In this chapter, the focus is on innovation for improved resource utilisation in the
industrial system for the manufacture of olefins32 and their derivatives. The system
around olefins was selected because ethylene and propylene are the key products of
the petrochemical industry because olefins represent the major non-fuel application
of crude oil products and because the mature 'olefins' industry represents a 'perfect'
case to elucidate the specification and the potential of process system innovations.
Olefins are key products of the petrochemical industry. Since these are the building
block of many plastics33 they are of major importance to our industrial economy.
Today polymers34 are used in over 20,000 grades in a myriad of applications, which
include packaging almost everything, synthetic clothing and carpets, toys and
consumer electronics, construction materials and automobile parts. The availability
of olefins is essential for the manufacture of many of these polymers. They are the
single building block in polyethylene (PE), polypropylene (PP) and polyvinyl
chloride (PVC), while polystyrene (PS) is produced from ethylene and benzene. In
addition, olefins are essential components for the manufacture of specialty polymers
such as polyesters (e.g. PET-bottles), polyurethanes (insulation foam, cushions),
polyacetates (paints), polyacrylamides (clothing) and a vast range of intermediate
chemical products. Some 70-80% of world ethylene and propylene production is
used for polymer production (Stanley 2001). The total 79 Million Ton per Year
(MTA) world polyolefin35 production represents approximately 63% of the global
plastics business (CMAI 2001b). The group of olefin chemicals thus includes the
most important building blocks manufactured in the petrochemical industry, viz.
ethylene and propylene. In fact, only sulphuric acid and ammonia are produced in
quantities that exceed ethylene production (Brennan 1998).
Olefins represent the major non-fuel application of crude oil products. Current
nameplate capacity for ethylene is estimated at some 100 MTA worldwide, with
current capacity utilisation around 90% at prices that fluctuate between US$380 and
$550 per ton in the last decade (CMAI 2000a). These units have a capacity for co-
producing some 40 MTA of propylene. Since production quantities are large, the
associated use of feedstock and fossil energy sources is high, roughly 200 MTA.

32 'Olefins' is the common trivial name for the group of mono-unsaturated hydrocarbons with a
carbon skeleton that consist of 2 atoms (ethylene) to 5 (isoprene) or even 8 atoms (1-octene). The
presence of a double C=C bond in these molecules causes them to exhibit significant reactivity,
which makes them a suitable feedstock for a host of chemical processes. The double carbon-carbon
bond is labelled 'unsaturated' as it can be saturated by reaction with a single molecule of hydrogen.
33 'Plastic' is a trivial name for polymers that originates from the label 'thermoplastic resins', polymers

such as polyethylene that slowly weaken and melt at elevated temperature. Thermoset resins do not
exhibit this behaviour. Today, main categories of polymer use are plastics (PE, PP, PS, SAN), resins
(paints, epoxy resins), foam (PS, PUR), rubber (EPDM, SBR) and fibers (Nylon, Kevlar).
34 Polymers is the common name of molecules that are formed from many small monomer

molecules.
35 Polyolefins is the commercial group label for polymers derived from olefins, viz. polyethylenes

(LDPE, LLDPE, HDPE), polypropylenes and polybutylenes.

116 Process System Innovation by Design

Today, these solely comprise non-renewables, notably oil products and natural gas
liquids (NGL). These are estimated to account for some 5-8% of global oil
equivalent consumption (Anonymous 2001a). Thus, only small improvements of the
global system around olefins (production, consumption, end-use) could result in
avoidance of considerable CO2 emissions from this system, which are roughly
estimated to total some 500 MTA36. Since steam crackers for ethylene and propylene
production exhibit an average net energy efficiency of some 70% there appears to be
scope for improvement with respect to resource utilisation.
Olefins represent a 'perfect' case to demonstrate the specification of process system
innovation content because not only is the industry's state-of-the-art rooted in a
large body-of-knowledge that was developed over a number of decades, it is also
largely a mature industry where the steam crackers built or revamped today are built
for decades of operation (Zeppenfeld and Walzl 2001).
Over time the industry has become linked and integrated at both the supply-side and
product side. Polyethylene and propylene plants do appear to be primarily located in
the vicinity of steam crackers, or ethylene / propylene pipelines. Styrene plants are
located near producers of ethylene and benzene, the exceptions obtaining their
benzene from other locations of the same company. The same holds for cumene
production plants. Vinyl chloride presents an interesting case, as both ethylene and
chlorine are costly to transport. Most sites, however, were also found to be located
near ethylene producers. The exceptions were plants that had switched from
acetylene based production, and hooked on an ethylene pipeline. These observations
led Molle and Wever to conclude that “there appears to be a very strong spatial
linkage among functionally related chemical activities” (Molle and Wever 1984).
Determining factors for the location of olefins facilities are the difficult transport of
olefins, the vertical integration in the petrochemical industry, and the pull that an
existing complex exerts on new settlers because of the agglomeration economy, the
availability of existing infrastructure, services etc. The present structure of the
industry and the technology applied are a result of a process that was driven by
economic motives, technological innovation and regulation (Dijkema and Kuipers
2001).
Olefins production by steam cracking is considered “a fairly mature technology”
(Stanley 2001). A characteristic of this maturity is that the potential impact and
associated earning power of R&D expenditure is perceived to be low compared to
other industry sectors. It is well known that such a situation is detrimental for the
rate of process innovation and that gradually the “quest for profitability
improvements” focuses on organisational restructuring to reduce cost (Hutcheson et
al. 1995). Although the industry around olefins has a history of continuous process
technology improvement, real process system innovations have been largely lacking.
This is illustrated by the structure of the industry that is dominated by steam
crackers and where novel processes adopted in commercial practice have been few.
In ethylene production, for example, production capacity has double from 50 MTA
in 1984 to 100 MTA in 2001 (Chauvel and Lefebvre 1989a; CMAI 2000a). Despite

36 Assume (1) that each ton of ethylene represents a total use of two tons of hydrocarbon as
feedstock and utility source for all steam cracker products (2) that, eventually, all hydrocarbon /
polymer derivatives degrade to CO2. Then per ton of hydrocarbon feedstock 2.5 tons of CO2 result.
Innovation around Olefins 117

this dramatic expansion, “the process chemistry and fundamental flow sheet
configuration {of the steam cracker} have remained relatively unchanged” during
the past 25 years (Stanley 2001).
The historic average growth-rate of some 4% per year of ethylene production
capacity is a trend that is expected to continue up to 2004, when world capacity is
expected to increase by 20 MTA to a total of 120 MTA ethylene installed (CMAI
2000a), and total system-related CO2 emission increases by 100 MTA to 600 MTA.
As we have argued elsewhere, however, the petrochemical industry must respond to
the call for reduction of greenhouse gas emissions, and timely anticipate the
depletion of fossil resources (Dijkema et al. 2001). Therefore, the procedures
developed to identify weak system elements and to specify process system
innovation were applied to the industrial system for olefins and their derivatives,
notably ethylene and propylene.
The olefins system in the Netherlands was used to develop the case study where it
was assumed initially that the product spectrum of the petrochemical industry
remains relatively unchanged. First, the entire system was assessed for weak elements
with respect to resource utilisation. The results provide the starting point for a
structured search for possible innovations in the olefins system: options addressed
within the boundary of single plants and changes that imply a change in network
structure or feedstock. The structured search for innovation options for the steam
cracker is reported to illustrate the scope and limitations of the methods reported in
chapter 3. Secondly, a functional model was constructed on the basis of all
information gathered in the assessment, structured search and literature review. This
model was used to categorise the innovations reported in the literature reviewed and
to explore additional process system innovation options to cater for the set of
functions of the olefins system. Thereby, changes within or beyond the boundary of
single plants are considered that rely on a change in network structure or feedstock.
Finally, economic and ecological aspects in evaluating the innovations reported in
the literature and proposed in this study are discussed and conclusions drawn.

4.2 The industrial olefins system

4.2.1 OVERVIEW OF THE PRODUCTION OF OLEFINS

Olefins today are produced primarily by steam cracking of preferably aliphatic
hydrocarbons37. This operation yields a product-mix of light olefins, with an average
ethylene to propylene ratio of 10:4, and a small share of butylenes (C4) and C5 -
olefins. In Figure 4-1 an overview is presented of the olefins production system.
About 70% of propylene is a co-product of ethylene manufacture by steam crackers,
28% is recovered as a by-product of refinery catalytic cracking operations (CMAI
2001a). Refineries located in Rotterdam, for example, have erected facilities to
increase their propylene recovery from cracking operations. As indicated, the
remaining 2% of propylene is produced by catalytic dehydrogenation of propane,

37 Aliphatic or saturated hydrocarbons are linear or branched hydrocarbons such as methane, ethane,
propane, octane, iso-octane etc. of the general formula CnH2n+2. These do not contain double or
triple bonds, nor special reactive side-groups attached to the carbon skeleton. The larger the
molecule, the more susceptible it is to cleavage and abstraction of hydrogen (dehydrogenation).
118 Process System Innovation by Design

which may originate from natural gas winning and production or from crude oil
refining.

Figure 4-1: The present olefins production system.

When light olefins are considered the major product of steam cracking, its major by-
product is pyrolysis gasoline. This co-product has a boiling-range similar to naphtha,
the main constituent of motor gasoline. Pyrolysis gasoline is rich in aromatics38.
Many steam cracker complexes therefore include an aromatics extraction facility to
co-produce pure aromatics from pyrolysis gasoline. In a refinery, reforming of

38 Aromatics are hydrocarbons that contain at least one triple-unsaturated C6 ring (C6H6-n)R1..n. In the
simplest aromatic, benzene, the C6H6-skeleton, is both heavily unsaturated and more stable than the
corresponding aliphatic hexane, C6H12. Aromatics in steam cracker feedstock largely yield undesired
unwanted products because of the relative stability and unfavourable C/H ratio of the aromatic
structures.
Innovation around Olefins 119

straight-run naphtha is used to increase the aromatics content by the abstraction of

hydrogen, and catalytic cracking is an operation where the aromatics content of the
product streams (cat cracked-naphtha, -gas oil, - kerosene) is much higher than that
of their 'straight-run' equivalents as obtained from the crude-distiller. Thus, also in a
refinery, aromatics-rich streams are sourced for on-site co-production of pure
aromatics.

4.2.2 FEEDSTOCK & RESOURCES

The present system for olefins production, and consequently olefins consumption
including polymer production, depends on the availability of fossil resources (Figure
4-1). Straight-run naphtha and gas oil from crude-distillation in refineries are the
dominant steam cracker feedstock because of their low aromatics content. Another
attractive feedstock is the Natural Gas Liquids (NGLs) that largely consist of the
preferred lower aliphatic hydrocarbons. Examples are ethane/propane mixtures, gas
condensate that contain ethane, propane up to hexanes and Liquefied Petroleum
Gas (LPG) fractions that largely consist of propane and butanes. Other sources (not
shown in Figure 4-1) are synthetic crude-oil fractions from Fischer-Tropsch
synthesis plants that use synthesis gas that may originate from Coal (Sasol), natural
gas (SMDS by Shell) or Mixed Plastic Waste processing (Veba Oel) (Dijkema and
Stougie 1994). According to CMAI (CMAI 2000b), “By 2002, (…) naphtha/gas-oil
based capacity will make-up 57 % of global ethylene capacity.” Large regional
differences do exist, however. The available data confirms that where European
capacity is largely naphtha-based, in the US ethane and LPG have remained the
preferred feedstock over the years (Table 4.1). The one significant change over the
years being a preference for naphtha over gas oil cracking.
Table 4.1: Steam cracker feedstock use. Data sources: 2000 (Chauvel and Lefebvre
1989a; Kvisle et al. 2001).

2000 1986
Feedstock Europe US World Europe US World
[%] [%] [$] [%] [%] [%]
Ethane 5 55 29 8.0 57.5 30.5
LPG 10 17 11 11.0 19.0 11.0
Naphtha 75 23 54 69.0 9.5 49.0
Gas oil 9 4 6 12.0 14.0 8.5
Others 1 1 1 - - 1.0

4.2.3 ETHYLENE AND DERIVATIVES

Ethylene is the most important building block in the petrochemical industry. Many
chemical products are derived from ethylene (overview: Figure 4-2) (based on
information from Kniel et al. 1980; Chauvel and Lefebvre 1989a; Anonymous 1990;
2000). The reactivity of the double bond in ethylene is used in polymerisation, in
halogenations that lead to a vast range of chlorinated products, in hydratations to
produce industrial alcohols, and in oxidation routes to produce ethylene oxide and
120 Process System Innovation by Design

acetaldehyde. In these products ethylene becomes the major constituent. In

alkylation processes, ethylene is a functional (chemical) additive to, for example,
aromatics. As is illustrated by this drawing, the route towards a finished product can
be rather short (polyolefins), or long (coatings ingredients). The data in Table 4.2
gives an impression of the percentage ethylene production used for important
applications. Since the early 80s, notably the production of linear low-density
polyethylene, LLDPE has grown significantly.
Table 4.2: Worldwide ethylene application (after Chauvel and Lefebvre 1989a)

Production of USA, 1983 Europe, 1983 Japan, 1983

[MTA] [MTA] [MTA]
Ethylene 13.0 9.5 3.7
[%] [%] [%]
LLDPE-LDPE 28.5 35.5 29.8
Ethylene oxide 17.0 12.0 11.2
HDPE 20.4 15.2 19.8
Ethylene dichloride 13.9 18.8 18.2
Ethyl benzene 6.9 8.2 9.1
Ethanol/Acetaldehyde 3.6 5.6 4.7
Vinyl acetate 2.5 * 3.1
Other 7.2 4.7 4.1

4.2.4 PROPYLENE AND DERIVATIVES

Similarly to ethylene, the amount of products derived from propylene is huge (Figure
4-3). Superficially, the only chemical difference between ethylene and propylene is
the methyl group (-CH3) present in propylene. The different size and structure of the
propylene molecule, however, results in distinctly different chemical reactivity
characteristics. In polymerisation, four different organic groups are attached to the
central propylene monomer's carbon-atom. This implies that so-called stereo
isomers are found in the polymer product. Whilst ethylene is polymerised to a
suitable product at high pressure (HDPE-process) since 1938, the industrial
production of isotactic polypropylene (the favoured stereo-isomer of PP) only took
off 20 years later after the catalyst development by Ziegler and Natta for stereo-
specific polymerisation and the subsequent development of a commercial process by
Stauffer Chem. Co. (Piduhn 1999). While propylene finds similar application in
polymerisation and partial oxidation, there are also striking differences in their
respective patterns of application. In alkylation its most important application is the
production of cumene, which is the most important precursor of phenol.
Chlorination of propylene is relatively unimportant. In addition to propylene glycol a
range of commercially important products are derived from propylene oxide (PO).
Since a direct oxidation process for PO has not been successfully commercialised
Innovation around Olefins 121

yet39, hydrohalogenation (with HCl) of propylene is still a relatively important route

to PO compared to the indirect oxidation routes towards styrene/PO and tert-butyl-
alcohol (TBA) and PO.
Propylene production capacities are intrinsically linked to ethylene production
capacity. The exception is refinery propylene production (see Figure 4-1). Propylene
consumption, however, grows at a higher rate than consumption of ethylene (CMAI
2001a). Notably polypropylene demand is growing faster than any other polyolefin.
The result is a gradual shift in the demand ratio between ethylene and propylene.
Table 4.3: Worldwide propylene use (after Chauvel and Lefebvre 1989a; Anonymous
1995b)

1990 1980
Production of [MTA] [MTA]
Propylene 30.320 18.600
% %
Polypropylene 44 30
Acrylonitrile 14 19
Oxo-alcohols 15 19
Propylene oxide 9 9
Cumene 8 10
Oligomers 4 8
Other 6 5

4.2.5 C4'S AND C5'S

Over the years, steam cracker installations have been extended with facilities to
extract and upgrade C4 and C5 olefins, notably butadiene, 1-butylene, 2-butylene, 1-
pentylene and isoprene. Quantities available from the steam cracker are dependent
on steam cracker feedstock selection and operation conditions. For some of these
products, specific markets grew rapidly, however: 1-butylene is sought after, as a
grafting monomer in LLDPE productions, 2-butylene is a precursor of MTBE, a
gasoline additive40. Therefore, alternative feedstock is exploited, notably natural gas
liquids (NGLs). A summary overview of the system has been published elsewhere
(Dijkema et al. 1997).

39 Lyondell recently announced construction of pilot facilities, however, to test said direct conversion

of propylene to propylene oxide (www.lyondell.com/news/09.10.02).

40 MTBE, methyl-tert-butyl-ether, has succeeded tetra-ethyl-Lead (TeL) in gasoline as an additive for

the improvement of gasoline quality - notably research octane number, which indicates the gasolines
anti-knock characteristics. TeL has been phased out because of health and environmental risks
associated with distributed Lead emissions from vehicles. In the US, MTBE has become the subject
of a controversy because of leakage from storage tanks and MTBE ending up in groundwater. These
spills, however, appear to have little to do with the actual use of MTBE in gasoline.
122 Process System Innovation by Design

4.2.6 OUTLOOK
Today, the production of olefins, notably ethylene, is considered a mature market
that however exhibits moderate growth rates of some 4 to 5% per year for ethylene
and propylene (CMAI 2001a; Stanley 2001). The sector is in its 'systemic phase' of
innovation (Hutcheson et al. 1995) where the focus is on cost reduction by
reorganisation of the business. Presently, this is manifested in the trend towards
global-sized sites (Dijkema and Kuipers 2001), where larger individual plants for
production and consumption of olefins are combined at lower cost per ton of
product. A second trend is the shift of the ownership of petrochemicals operations
to refiners41.
The development of competing products for the market niches of ethylene-derived
products (e.g. propylene!) has not yet resulted in a decline in demand for ethylene.
On the contrary, with the growth-rates anticipated, by 2018 ethylene production
capacity will have doubled to 200 MTA! Thus ethylene remains a keystone to the
petrochemical industry while at the same time the importance of propylene
continues to increase.
Today, a single type of commercial process dominates ethylene production, the
steam cracker, which also produces the majority of propylene. Their continued
operation worldwide depends on the availability of crude-oil and NGLs. A major
share of propylene is being produced in refineries. The increased world gas reserves
and the launch of Lurgi's MegaMethanol technology has initiated renewed interest in
methanol based routes to olefins, which have lead to successful operation of a
demonstration facility and plans for the erection of at least one commercial plant in
the coming 5 years (Bekkum 2001).

4.2.7 CONCLUSION
On the basis of the olefins system overview and assessment, we conclude that long-
term 'sustainability' is not yet the major concern in this industry today. Where health,
safety and environment issues are high on the industry's agenda to ensure
continuation of its license-to-operate, the still limited efficiency of its major process,
steam cracking, appears to be largely accepted. Its consequence, a significant
contribution to CO2 emission worldwide , is hardly penalized. Major petrochemical
corporations, however, have recognized that CO2 regulation at some point in the
future will be put into effect. Therefore, they experiment with emission trading and
other clean development mechanisms. More important, however, ethylene producers
appear to be trapped in the systemic phase of innovation, where competition
requires continuous cost reduction, which are realised amongst others by increasing
the scale-of-operations. This reduces the industry's flexibility with respect to
developments in feedstock availability, market demand and alternative technology
development.
To sustain current and future olefin production levels at some point in time
alternative feedstock must be sourced and the associated processes must be
developed and employed. Most imminent at the demand side is the changing ratio

41This is illustrated by the recent takeover of DSM Petrochemicals by Sabic and the refocusing of
Shell Chemicals on olefins ethylene and propylene.
Innovation around Olefins 123

between propylene and ethylene demand, which cannot be met by exploiting

additional NGLs because these simply are not available. It appears that mid-term,
more refinery feedstock will be employed and possibly natural gas via the methanol-
to-olefins route. Long-term the industry must anticipate and adapt its systems for
resource scarcity and a ban on CO2 emission.
In the next section, we will demonstrate that ample options do exist for this industry
to contribute to sustainability by increasing its resource utilisation and avoiding CO2
emissions.

4.3 Systematically towards innovations?

4.3.1 INTRODUCTION
The 1990 Olefins system in the Netherlands has been assessed using the method
described in chapter 3. In this analysis, the steam cracker has been identified as the
major weak system element with respect to resource utilisation. Therefore, after an
overview of the olefins system analysis is given, the focus is on innovations in or
around the steam cracker. The steam cracker is described by a summary status of the
technology and technology development. A systematic search for
innovation/improvement options is completed for the steam cracker as per the
approach presented in chapter 3.

4.3.2 ASSESSMENT

The case: olefins in the Netherlands

In the past century, in the Netherlands three companies developed and operated
Olefins complexes, viz. Dow Chemical, DSM42 and Shell Chemicals. Over time,
these have been connected by ethylene pipelines and propylene pipelines (partly).
Both Dow and DSM operated two distinct cracker complexes, while Shell operates a
single complex. In 1990 the joint production Dutch capacity amounted to some 2.55
MTA ethylene and (Anonymous 1990) and some 1.0 MTA propylene. This situation
has changed dramatically over the past years, as both Dow and Shell Chemicals have
completed large revamp and expansion projects on their steam cracker sites. Notably
Dow has expanded its existing facilities in Terneuzen, and effectively operates three
cracker complexes with a combined capacity of 1.7 MTA ethylene (0.9 MTA name-
plate capacity in 1990). Its feedstock includes NGLs apart from naphtha. At the
Shell Moerdijk's 1973 cracker complex also a major revamp and debottlenecking
project has recently been completed. It is scheduled for another 25 years of
operation at a capacity of 0.9 MTA ethylene with feedstock flexibility towards
naphtha and gas oil. The DSM crackers have been overhauled and expanded in the
mid-90's, where also some new functionality has been added. Combined, these
projects have increased total Dutch ethylene capacity to some 3.8 MTA ethylene and
1.5 MTA propylene. The basic outline of these complexes, however, has not
changed.

42The DSM steam cracker installations in Geleen were sold to Saudi Aramco Basic Industries (Sabic)
as part of DSM's spin-off of its petrochemicals activities in 2002.
 124

Vinylidene
PdVC Food-packaging Films
Chloride

Chlorine
EDC VCM PVC Pipes, Film, Wire Insulation

Ethylene
Diamines

Chloro- Tetrafluoro
PTFE Smooth & Medical Products
form Ethylene

Vinyl PE/VAc
Acetic Coatings, Resins
Acetate copolymer
Acid

Terephtalic
p-Xylene
Acid

Ethylene Ethylene PET PET PET-Bottles, Fleece

Oxide Glycol
Organic Polyesters Boat Hulls, Clothing
(Di-)Acids

LDPE Packaging, Furniture, Toys

Ethylene
HDPE Packaging, Toys

Ethyl-
Styrene PS Disposables,Insulation
Benzene Benzene

1-butylene; LLDPE Packaging, Furniture, Toys

1-octene

Ethyl- Styrene Ethylene Ethylene

Benzene Glycol

Figure 4-2: Industrial production of ethylene derivatives and example applications.

Process System Innovation by Design

From C2 to C3 subsystem
 From C2 to C3 subsystem

Ethyl- Ethylene
Styrene Ethylene
Benzene Glycol

EPDM
Car door & window seals
PG Ether Rubber
Innovation around Olefins

Styrene

Propylene Propylene
Oxide Glycol
Poly-
Isobutane TBA Foams; Car cushions
Butanediol Urethanes
(BDO)
ABS Car parts; Electronics
Polyether
Polyols
Iso-
SAN Transparent Car parts
cyanates

Poly
Acrylonitril Fiber clothing
Acrylamides

Propylene PP Packaging; Furniture; Toys

Isopropyl Gasolline
Acetone MIBK MTBE
Alcohol Additive

MMA
Artificial windows -
PMMA
Plexiglas

Cumene Bisphenol Poly-

Benzene Phenol CD's; Artificial windows
A Carbonates

Allyl- Epichloro- Glycidyl Epoxyr

Figure 4-3: Industrial production of propylene derivatives and example applications.

Electronics; Adhesives
Chloride Hydrin ethers Resins
125
126 Process System Innovation by Design

System (boundary) selection

Combined, Figure 4-2 and Figure 4-3 represent the 1990 Olefins system in the
Netherlands. They provide an overview of the industrially employed conversions of
ethylene, propylene and their derivatives respectively. The system boundary selection
for an olefins system is not trivial. Propylene oxide is produced in combination with
either styrene or tert-butyl-alcohol (TBA). Other products manufactured from
olefins lead to end products that only include a small portion of material of olefin
origin (such as styrene that is used for the (co-) polymers PS; ABS, SAN). Therefore,
a selection was based on predominance of material content of olefin origin. In
Figure 4-2 and Figure 4-3, industrial processes included in the olefins system are
indicated by a grey-fill.

Method applied, information sources, data input and manipulation

The assessment has been developed to arrive at a satisfactory starting point for the
search for innovations in this complex industry. A priority weak element analysis has
been completed for the 1990 Olefins system in the Netherlands according to the
method described in chapter 3. An overview of the industrial plants in operation in
1990 is given in Appendix A. 6 and summarized in Figure 4-2 and Figure 4-3.
In order to enable the assessment, data from various sources had to be gathered,
checked and properly combined. To establish the input/output characteristic of
industrial petrochemical plants use has been made of performance data of 'typical
plants' available in the open literature (e.g. Chauvel and Lefebvre 1989a). For a
limited number of plants in operation a data set was compiled by using confidential
sources. The sets of stream data on industrial plants had to be reconciled without
exception. In order to arrive at closed mass balances, corrections were made based
on component balances and sound engineering judgement. Additional assumptions
were made with respect to unreported energy losses from the systems assessed.
The corrected plant input/output data were combined with thermodynamic data on
the substances involved to allow the calculation of process system energy balances.
This included adoption of the reference state 'standard conditions', i.e. P0 = 1 Bar, T0
= 298 K. Thus, per plant, efficiencies and losses could be calculated according to the
equations given in chapter 3. System losses were obtained by proper combination
with plant capacity data as reported in reputable sources: (Anonymous 1990), open
literature, permit requests, commercial bulletins.
Input/output data of the production and consumption subsystem and the total
olefins system were obtained by linear addition of individual plant data, under the
assumption of maximum internal exchange and the absence of synergistic effects on
plant performance.

Interpretation of results
The result of the assessment is not a performance assessment of the Dutch olefins
industry as it existed in 1990 or yesterday, let alone an assessment of a single
Innovation around Olefins 127

company or facility43. Although an attempt was made to compile reliable datasets

that represent the industrial state-of-affairs, the assessment results presented must be
interpreted with care. Information on industrial plant performance available often
lacks a time-stamp and may be out-of-date. As a matter of course, actual plant design
and operation can differ from 'typical installations'.
The results obtained, however, are of sufficient quality beyond an 'order-of-
magnitude' to merit an initial ranking of weak elements and a guestimate scope for
improvement. The information incorporated in this data set thus primarily serves the
objective of this study: the development and underpinning of some systematic
approach to process system innovation for the benefit of a complex olefins industry
that is largely mature.
Even with the limitations of the raw data available, the conclusions based on the
case study results are believed to be valid. Consequently, the results are of
significance to the further development of the olefins industry, both with respect to
the weak elements identified and the innovation options discussed.

Assessment results
The system analysed is decomposed in olefins production and olefins consumption.
Production is the subsystem illustrated in Figure 4-1; consumption predominantly
comprises conversion into polymers, often via a production chain of chemical
derivatives (Figure 4-2 and Figure 4-3), where strictly speaking polymer production is
outside the system boundary of the petrochemical industry. A complete overview of
the system analysed and results is given in Appendix A. 6.
Table 4.4: Key results of the olefin system analysis (1990 data)
Capacity Mass Loss Energy Loss
Lb Le Lb Le
[MTA] [MTA] [MTA] [PJ/Yr] [PJ/Yr]
Production Olefins 3.57 0.70 1.21 47.5 47.5
Consumption Ethylene 2.28 0.44 3.01 16.7 17.0
Propylene 1.67 0.42 4.74 32.8 32.9
Total 3.95 0.86 7.74 49.6 49.9
Polymerisation Ethylene 1.57 0.04 1.71 11.3 12.0
Propylene 0.54 0.04 0.35 4.8 4.8
Total 2.11 0.08 2.06 16.1 16.9
Olefins Total 8.61 1.29 11.01 89.4 90.5
Notes to Table 4.4:
1. Olefins production capacity is only ethylene and propylene; efficiencies and losses are listed
for a steam cracker that co-produces C4/C5 and aromatics.
2. The capacity of the olefins consumption subsystem exceeds that of ethylene and propylene
production because some derivatives produced include oxygen and chlorine.
3. The relatively small capacity of polymerisation indicates other use and export of monomers
from the Netherlands.

43In case one is particularly interested in such assessment, it is suggested to contact one of the many
consultancy firms that specialise in tracking the global (petro)chemical industry's development and
performance.
128 Process System Innovation by Design

The key results of the analysis are summarised in Table 4.4 for process net mass and
energy losses b (material/energy supplied that ends up elsewhere than wanted main
or by-products) and total mass and energy loss e (process energy/material + utilities
supplied that does not end up in products/by-products). Losses b and e have been
calculated as defined in chapter 3 (Eq. 3.21-3.24).
The entire system around olefins requires a total feedstock and energy input that is
equivalent to some 6 [MTA] naphtha, or some 250 PJ/Yr. Its chemical and polymer
represent some 135 PJ/Yr, and some 115 PJ is lost from the system annually. It
follows that the gross energy efficiency (on the basis of LHV) of the system around
Olefins is some 54%. Thus, the energy efficiency of this complex chemical
production system exceeds the efficiency of Dutch electricity generation (which
currently is some 45%).

Weak elements?
The results of the system analysis as presented in Table 4.4 can be interpreted in
various ways. Olefins production and consumption for conversion into
petrochemicals require attention when the results are analysed using the aggregate
system level perspective where the system around olefins is thought to be
decomposed into a subsystem for olefins production, conversion into
petrochemicals and final consumption for polymer production. The results of the
analysis show that both mass and energy loss Lb of olefins consumption for
conversion just exceeds the losses associated with of olefins production. The losses
associated with polymerisation are the smallest. These results should be interpreted
with care, however, as the results on polymer production do not include the losses
associated with PS and other polymers that include material with an olefin origin. In
addition, the Dutch capacity installed for polymer production does not match the
quantity of monomers produced. A considerable amount of petrochemicals is
exported. In case the exported monomers would have been converted in the
Netherlands, the associated losses would total some two-thirds of the losses in
olefins production and associated conversion respectively. The energy incorporated
into the olefins by steam cracking is largely retained when olefins are converted to
more complex monomers, as the energy-rich double bond remains unchanged in the
conversion or is converted into an epoxy structure. Eventually, however, the greater
part of the monomer's increased energy level is used to effect polymerisation, which
results in energy loss from the system as waste-heat.
Inspection of Figure 4-2 and Figure 4-3 reveals that while olefins production (Figure
4-1) only involves the steam cracker and propylene from catalytic cracking, olefin
conversion involves some 20 distinct industrial conversion processes and some 10
polymerisation processes that each contribute to their respective subsystem losses.
Therefore, a proper conclusion is to take special notice of the large loss of the steam
cracker, and to inspect the results for olefin conversion in more detail.
Indeed, when the analysis results are inspected at the aggregate level of individual
plants and when all processes in the olefins systems are ranked, the steam cracker
Innovation around Olefins 129

ranks first with respect to net energy loss and mass loss . In 1990, steam cracker
plants only had an overall energy efficiency of some 65%44.
The results can also be interpreted using an ethylene and propylene subsystem
division. In the subsystem around ethylene the total energy loss, the loss allocated to
steam cracker ethylene production, equals the combined losses associated with
EDC/VCM, ethylene oxide and ethylene glycol. In the propylene consumption
subsystem, especially propylene oxide production is the exception because its net
loss exceeds the loss allocated to steam cracker propylene production.
The energy efficiencies vary from a 39% low (ethylene diamine) to a 93% high
(PE/Vac copolymerisation). Especially endothermic processes (steam cracker),
partial oxidation (ethylene oxide, propylene oxide) and other exothermic processes
(ethylene glycol) have a relatively high scope for improvement index. In all these
processes there is a relatively large energy transformation between heat and chemical
energy where the Second Law of Thermodynamics dictates that retention of energy
quality is impossible and maximum retention of quantity is economically infeasible.

4.3.3 IN-DEPTH ANALYSIS OF INDUSTRIAL STEAM CRACKING

In the assessment of the industrial olefins system, the steam cracker has been
identified as a major weak system element because of its net mass and energy loss.
The indices calculated (Appendix A. 6) merit a thorough investigation of some 6
weak elements identified. Since our purpose is to test the possibilities and limitations
of the methods proposed in Ch. 3, we have opted to focus on the analysis of
industrial olefins production by steam cracking in this chapter.
In the method described in Ch. 3 one must use weak elements identified as a starting
point for the search for improvements. This search, however, must be preceded by a
thorough analysis of the causes for the losses calculated. Therefore in this section an
introductory overview of steam cracking essentials and development is given. In the
process, the 'black-box' of steam cracking is opened, and the subsections of this
industrial system, its unit operations and internal interconnectivity are analysed (see
Ch. 3, Table 3.1). Using the knowledge thus assembled, intra process causes are
identified and options for improvement postulated. These are limited to intra-
process; potential modifications involving re-labelling of process outputs and the
petrochemical system around the cracker will be addressed in the subsequent section
using functional modelling.
Since the steam cracker is the feedstock generator for the petrochemical industry,
the body-of-knowledge and the available literature on this process is immense.
Extensive introduction and background information can be found (e.g. Kniel et al.
1980; Chauvel and Lefebvre 1989a; Anonymous 2000; and Stanley 2001).

44 It is not the objective of this analysis to assess technology improvement, i.e. the effect of process
innovation during the past decade. Benchmarking has revealed, however, that the Dutch steam
crackers rank amongst the top-10% of steam cracker installations worldwide. Because of the
confidentiality of performance data on individual installed plants, however, it is not possible to verify
and compare current performance data. Suffice to say that if, through process innovation,
performance has indeed risen to some 70-75%, this is a tremendous achievement and proof of the
technological capabilities of the industry and its technology suppliers.
130 Process System Innovation by Design

Steam cracking is pivotal to the petrochemical industry because it serves to create

reactive chemical components from relatively unreactive hydrocarbons present in
fossil feedstock. Thus it allows the utilisation of fossil resources for the production
of durable and consumer goods. Steam cracking exhibits a variety of characteristics
that are typical of the processing industry: a large-scale of operations (current single-
train capacities up to one MTA ethylene per year), integrated production complexes
where a variety of products and co-products are further processed or recycled to
maximise the net revenue of the steam cracker site, high capital requirements, high
knowledge intensity, relatively low labour-intensity.
We have argued that each chemical process can be described as a system that
comprises feedstock preparation, reaction, separation, a process recycle, and final
product purification (Dijkema et al. 1998). In addition, we argued that the key
function of the process recycle is often misunderstood. The steam cracker also fits
into this general model, however, with an important deviation that the process
recycle is sent to a special reactor for recycle processing. This is a system design
option that also has been developed and applied elsewhere, for example in cumene
production (Dow Chemical Benelux 1994). In the literature a steam cracker is usually
described as comprising two or sometimes three major system elements, viz. a hot
reactor section, a compression section, and a cold separation section. The latter two
are often combined and labelled the separation section, while quenching and
stabilising are seen as part of the front-end or hot reaction section (Figure 4-4).
Another important system element is the recycle of ethane and propane, which
consists of unconverted feedstock or co-product formed in the hot section. Today
these are usually recycled to a dedicated reaction furnace. When the recovery of co-
produced pure aromatics is included, the so-called raffinate from aromatics
extraction is an additional feedstock recycle to the cracker.
The hot reaction section consists of a series of furnaces where the feedstock is
diluted with steam and heated to some 800 to 900 oC to effect the cracking reactions.
Immediately after the cracking furnaces, the reaction mixture is cooled rapidly
('quenched'). Water and heavy components ('tar, pyrolysis gasoline') are removed in a
stabiliser. In the separation section, the lighter products (up to C6) are compressed
and sent to the cold separation section that yields the individual light olefins as
essentially pure products.

Hot section: the cracking furnace

Apart from paraffins and iso-paraffins, the feedstock of a steam cracker can contain
significant amounts of aromatics and naphtenes (polycyclic aromatics. In the
process, these are broken-down and rearranged via a large array of chemical
reactions. The worlds most used kinetic model of a steam cracker, SpyroTM includes
over 2000 chemical reaction steps (Overwater 2001). These reactions can be divided
in primary and secondary reactions.
Innovation around Olefins 131

Figure 4-4: Steam cracker overview.

132 Process System Innovation by Design

Major primary reactions in the steam cracker are:

• Dehydrogenation,
An organic molecule loses two hydrogen atoms, e.g. from ethane to ethylene:
H3C-CH3 Æ H2C=CH2 + H2
• Cracking
An organic molecule is cleaved, e.g. hexane yields propane and propylene:
C-C-C-C-C-C Æ C-C-C + C-C=C.

Important secondary reactions in the steam cracker are:

• Dehydrocyclisation
In this case, a ring is formed from one or more olefins, after which hydrogen is
lost to yield an aromatic ring
C=C-C=C + C=C Æ C6H10 Æ C6H6 + 2H2
• Dealkylation
This is a reaction where aliphatic side-chains of aromatic rings are removed. The
end product is benzene or naphthalene (double ring)
C6H5-C2H5 Æ C6H6 + C=C

The most important desired reactions are the primary reactions, where light olefins
are formed. The secondary reactions largely occur following the primary reactions.
They are responsible for the heavier components in the cracked-gas such as benzene
and butadiene. Tertiary reactions include the formation of coke on the furnace pipe
surface. Coke formation occurs via a variety of reactions, by which large carbon
networks are formed. These reactions are favoured by high-pressure. In order to
suppress coke formation, in the steam cracking process the feedstock is diluted with
steam to lower the reactants partial pressure.
The steam cracking reactions occur largely in the gas phase. The reaction mechanism
is via free organic radicals (Chauvel and Lefebvre 1989a). The reactions are initiated
by bond-cleavage that yields two organic free radicals. In propagation, these radicals
can react with the other components to yield desired products + additional free
radicals. Finally, in a gas mixture where the radical concentration increases,
recombination of radicals terminates the chain of steam cracking reactions. An
inevitable consequence of this mechanism is the formation of a substantial amount
of hydrogen and methane in the product mix. All of the individual reaction steps
involved have different thermodynamic and kinetic characteristics, which are a
function of Pressure, Temperature and chemical composition. The product mix of a
steam cracker furnace thus is a function of feedstock composition, furnace radial
tube temperature distribution and longitudinal profile and the residence time. In
SpyroTM all these effects are accounted for to achieve an optimal design.
The steam cracking furnaces today largely have been designed as and operate as
single-pass reactors because the substantial energy requirements to operate the cold
separation dictate that a limited share of recycles must be sent through the process.
Innovation around Olefins 133

To achieve favourable process economics, therefore thermodynamic equilibrium

conditions achieved must be such that product yield is 50-60% or higher per furnace
pass. Thermodynamics determines the equilibrium reaction mixture composition per
chemical reaction. Its kinetics determines the speed by which this equilibrium is
approached. The equilibrium of cracking reactions is favourable from 300 oC and
beyond, while the equilibrium for dehydrogenation is favourable from 700oC onwards.
The rate of these reactions becomes significant above 700o (cracking) and 800o C
(dehydrogenation). These characteristics have large consequences for the 'facility' of
cracking of a particular feedstock. For example, ethane or propane cracking requires
a high temperature (850 oC because substantial dehydrogenation must occur. In gas
oil cracking, however, cracking reactions dominate, and a lower temperature (750 oC)
can be selected in furnace operation. A catalyst can suppress or accelerate the speed
of a particular reaction. A consequence of the thermodynamics, however, is that the
use of a catalyst appears to be relatively useless to improve the process. Economics
and the thermodynamics dictate high operating temperatures, where the required
reactions occur at high speed spontaneously. The use of catalyst, such as in catalytic
cracking in a crude-oil refinery, would then require both an expensive catalyst and an
expensive reactor / regenerator system to burn-off coke deposits.
To match the thermodynamics of the steam cracker reactions, the reaction furnaces
are laid out such that an optimal temperature profile in the furnace tubes is achieved.
In the entry-zone the temperature is some 400 oC, while the highest temperatures are
reached at the exit. Thus, cracking can occur first, followed by dehydrogenation.
Most primary and secondary reactions that yield olefins are strongly endothermic.
Thus, a large supply of energy to the feedstock / reaction mixture is required . The
energy is supplied in cracking furnaces where the heat liberated by burning natural
gas, liquid fuels and co-products of the cracking process is transferred to the
reaction mixture in the furnace pipes. The furnace and furnace-tubes have such a
layout that the specific heat-transfer requirements for the cracking and
dehydrogenation reactions are matched: feedstock entry in the convection zone, final
exit after passing the hot radiant zone. To freeze the reaction mixture, it is rapidly
cooled in a quench system, which is a combination of in-line heat exchangers that
generate high-pressure steam, and in-line quenching for direct heat-transfer to a
quench-oil. In each of these operations a significant part of the potential to do
Work, exergy, is lost. In transferring energy to the reaction mixture the chemically
stored exergy in the fuel is transformed to heat. According to the Second-Law, the
heat then can only be partly converted back into chemical exergy, the remainder is
lost as waste heat. Subsequently, the high-temperature reaction mixture is rapidly
cooled down, where the energy is recovered only as heat, and again potential to do
Work is lost. In short, the process employs crude, large-driving forces, which
thermodynamically inevitably lead to irreversible losses of exergy. Since the exergy
required presently is supplied by burning fossil fuels, there is a direct link with the
process' CO2 emission.<Cold> separation and purification
The heavier products (pyrolysis gasoline, tar), water and quench-oil are removed in a
stabiliser column that operates near atmospheric pressure. Olefins and light gases are
sent over the top of the stabiliser to the compressor. A number of preparations are
required to allow smooth cryogenic distillation, which is the preferred method of
134 Process System Innovation by Design

separation. First, the cracked-gas is compressed to some 35 bar to allow cryogenic

separation at acceptable temperature levels. Given the capacity, gas composition and
pressure difference, the cracked-gas compressor rank amongst the most advanced
centrifugal compressors in the world. It consists of four to five stages with gas-
intercooling to prevent oligomerisation, and is usually driven by steam turbines. At
each interstage, liquids are collected.
Sulphur compounds in the cracked gas originate from the feed and from trace
sulphur components added to the steam cracker feed to avoid furnace-tube
deterioration. Therefore, just before the final compression stage, the gas is
desulphurised and CO2 removed by an absorption process (employing caustic soda
or ethanol amines). Finally, the cracked gas is dried over molecular sieves to very low
moisture content in order to avoid the formation of ice.
In the cold separation section, essentially the following operations must be
completed:
- Removal of non-condensable gases
These include hydrogen and methane. In a traditional cracker design, this is
done in the first column, the demethaniser, where the lowest temperature is
achieved (-100oC)
- Removal of acetylenic compounds.
These compounds are removed by selective dehydrogenation from the
ethylene produced, and sometimes also from propylene.
- Production of essentially pure ethylene
This involves separating the C2's (ethane/ethylene), followed by
ethane/ethylene separation
- Production of essentially pure propylene
This involves separating the C3's (propane/propylene) followed by
propane/propylene separation.
+
The C4 fraction may be subject to C4/C5 splitting and butylenes, butadiene and C5
olefins recovery. The remainder C5+ gasoline is rich in aromatics, especially benzene,
and thus may be sent to an aromatics plant. Depending on the composition of the
cracked-gas and local economics, a variety of process schemes may be considered.
Optimisation of the separation section configuration has led to considerable energy
saving over the years.
Similar to supplying heat, cold represents a deviation from the reference state. A
driving force is required to create it. In effect, the cycles used to supply cold have a
thermodynamic upper limit to their efficiency. Work must be supplied to pump heat
from a lower temperature to a higher temperature, part of which is lost as waste
heat. Subsequently, the cold streams created provide the proper conditions and part
of the driving force to the separation process. A large part of the exergy supplied via
the cold streams is lost, however.

Intra process causes

The steam cracker is a weak element in the olefin-system because of its large energy
'loss' to the environment. In addition, it is characterised by a considerable mass loss:
Innovation around Olefins 135

part of the feedstock is eventually used as fuel and converted to CO2 and water
instead of desired products. This is caused by a variety of process characteristics: (1)
the reactions towards olefins are strongly endothermic, and are effected at elevated
temperatures in milliseconds. As a consequence, the energy efficiency of the furnaces
is limited to some 85-90% today. In addition, the net transfer of heat liberated from
fuel combustion to the reactor mix is limited, the remainder can only be recovered as
heat. (2) Thermal cracking leads to a mixture of products that inevitably require
separation. In this particular case, some of the products (hydrogen, methane,
ethylene) are separated in commercial installations at high pressure and very low
temperature, in large cryogenic distillation columns. Both the thermodynamic
efficiency of such columns is limited, as well as the efficiency of the cryogenic
operations. Finally, the maximum thermodynamic efficiency achievable in gas
compression is only some 35%. (3) The product spectrum of the cracking section is
such that a relatively large part of the product mix must be transported through a
large part of the separation system, which increases the section's energy-use.

Intra process solutions

Over the past 40 years, steam-cracking technology has development into an
advanced but mature technology (Stanley 2001). Within the constraints set by
thermodynamics, reaction kinetics, feedstock availability and process economics, a
host of innovations has been developed and implemented in new or existing cracker
installations. Hutcheson et al. (1995) give an overview of some 50 cracker
innovations which all can be characterised as intra-process (Hutcheson et al. 1995).
The fundamental problems or limits described above, however, are still valid today.
A non-exhaustive overview is presented below.
The crux of steam cracking is to rapidly achieve and freeze a favourable reaction
mix. The progression of undesired reactions to their thermodynamic equilibrium
must be prevented. Thus, where 'quenching' enabled the first successful steam
cracking operations, the development of the Milliseconds Technology represents a
major process innovation. Short residence times in the Milliseconds furnace result in
higher selectivity towards ethylene, but require a higher furnace temperature. Other
developments, such as in burner and tube design as well as furnace layout contribute
to continued improvement of thermal cracking selectivity. Steam cracker furnace
technology therefore is an extremely important area of research. Its scope, however,
is largely within the boundaries and limits of the technological concept of steam
cracking.
When one is confronted with a thermal and therefore by definition unselective
process, the very first option to be considered is the development of some catalyst
that offers selectivity at acceptable yield. It is well known that cracking reactions can
be accelerated by the use of catalysts. At high operating temperatures, excessive
coke-formation presents a problem. In catalytic cracking in refining, this problem
has been addressed by applying a reactor-regenerator system, where the deactivated
catalyst is reactivated in the regenerator by burning of the coke. Analogous catalyst /
reactor / regenerator combinations for steam cracking, however, are not in
commercial use. This may be because dehydrogenation is one crucial and final
reaction step in the production of ethylene from heavy organics. The
136 Process System Innovation by Design

thermodynamics of this reaction step prohibits the use of a catalyst for

dehydrogenation at low temperature for the conversion of ethane to ethylene.
Alternative reaction paths to avoid said thermodynamic limitation of ethane
dehydrogenation have been extensively investigated. Oxidative dehydrogenation is
one possibility, but any catalyst must prevent the oxidation from running to
completion. Eastman et al. (1990) summarize the developments and conclude,
however, that this route was not commercially viable because at the time no suitable
catalyst had been found (Eastman et al. 1990).

Energy use
Thermodynamically, the 'process path' in a steam cracker deviates considerably from
the optimum, reversible path. One major indication for this is the extreme
differences between the temperature levels of operation. A second indication is the
requirement of rapid heating, followed by rapid cooling. Thermodynamically this
equates to the use of large driving forces, which imply large losses (Prigogine and
Stengers 1984). Apparently, economic trade-offs drives one into this particular
direction. The consequence of this is that a large amount of Work, or exergy, is used
in the hot reaction section, the compressor and the cold separation section
respectively. This is observable in the difference between feedstock and product
exergy. The difference is liberated as heat, only part of which can be utilised.
Improvement, i.e. reduction of the net energy supplied to a steam cracker of existing
design, can be achieved via three major routes:
1. Optimal matching of heat demand and heat generated as a consequence of the
process paths selected.
This is the objective in the use of Pinch technology, to achieve the
thermodynamic optimum in heat integration.
2. Modification of the temperature levels by adding equipment in various heat
streams that convert part of those streams into work: process & energy
integration.
One option is the use of a gas turbine that provides part of the heat of reaction
to the furnaces, whilst producing shaft-work (Dijkema et al. 1998).
3. Modifications in the process system design, e.g. in the cold separation section or
recycle structure.
These modifications can be enabled by the advent of new technology, e.g. in cold
separation technology. Their adoption by industry is subject to the total expected
lifecycle cost trade-off between energy-saved and additional investment. Examples
are adsorption-cooling and open cooling systems that use product streams with
direct contact.
Modifications to the system design also can comprise a combination of 'utility
optimisation' and feedstock/product utilisation. For example, styrene in the
aromatics fraction can be hydrogenated and extracted as ethyl benzene, but it can
also be separated as a by-product. In the latter case, further processing steps from
ethyl benzene are avoided, and utility hydrogen otherwise required can be saved for
other applications.
Innovation around Olefins 137

4.3.4 CONCLUSION
The use of the systematic assessment procedure and systematic search for options
for improvement (Ch. 3) has been illustrated for the case of steam cracker
technology. Since the scope of the analysis and search for improvement was limited
within the steam cracker system boundary, only intra-process improvements were
suggested. It is well known, however, that the upgrading of waste products to by
product (C4/C5 olefins, aromatics) and eventually co product (propylene) has been a
successful strategy of stream cracker operators. The concept, however, can still be
useful today with respect to ethyl benzene and styrene isolation, and the recovery of
methyl acetylenics instead of their saturation. Some of the options thus identified
will require a market to develop and further changes in the petrochemical industry to
materialise.

4.4 Functional modelling of the olefins system

4.4.1 FUNCTIONAL MODELLING

The previous section illustrates that an in-depth review of intra process causes
initially leads to improvement options within the scope of the existing process
concept. As a matter of course a brainstorming session could be organised to try and
arrive at novel alternative concepts of the steam cracker or parts thereof. Our
present objective, however, is to demonstrate that functional modelling (Ch. 4)
provides a method to structure 'rethinking and reinventing' systems while linking
various system aggregate levels. Functional modelling can be employed to specify the
content of process system innovations of the technological part of systems, in this
case the steam cracker process for the production of olefins. Thus, although it may
be argued that one of the objective-defined functions (ODF) of every privately
owned technological system is to generate cash flow, profit, or a return on
investment, this is not the only starting point to look for innovations. As a matter of
fact, in a mature industry like that of olefins production, cost probably has been used
a million times as a starting point for innovations. Therefore one can better use the
(strategic) perspective of the 'end-user' of the system's function involved, which can
then be easily expanded to include societal goals such as sustainability. The
procedure presented in chapter 4 around functional modelling consists of
(1) Generalise at some top system level (…)
(2) Visualise the link of this subsystem of our industrial economy to other industrial
sectors or parts thereof, thereby eventually allowing some linkage with needs of
society catered for by ‘industry’ (…)
(3) Arrive at a suitable functional decomposition of the petrochemical industry that
finally allows the specification of technological or system content of innovations.
(…)
(4) Question and adapt system boundaries. Because functional decomposition is based
on system and subsystem boundary selection it opens the way for consideration
of alternative systems by expanding the set of functions in the model at a
particular system level, or to view the system as a part of a larger system”
(Dijkema et al. 2001).
138 Process System Innovation by Design

The Polymer Industry

Function Function
1 S

Mo
s
fin

nom
le
O

ers
The Petrochemical Industry

Olefins BTX
Production Consumption
Function Function
Function M Functio
n1 P
1
Py sol

B, T , m
Ga
ro ine

oX, pX
l ys
is

BTX

X,
Production
Gas ha
LPG

Oil
ht

Function Function
Nap

1 N

ed
rm a
fo hth
Re ap
N
Refining
Function Function
1 R

Figure 4-5: Decomposition around the olefins system (reprinted from Dijkema et al.
2003).
Since the concept of functions allows technology-free specification of systems, in the
process alternative methods of fulfilment - innovations - can be explored. Thus for
example at the level of 'packaging' being the major driver of olefins production and
consumption, one may consider devoting some attention at the 'reinvention' of
packaging and the associated systems to fulfil our (basic) human needs.

4.4.2 GENERALISATION AND VISUALISATION

The olefins system is part of the petrochemical industry, which primary ODF may
be described as “to produce suitable feedstock for polymer production” (Figure 4-5).
The generalisation achieved by the model presented in Figure 4-5, however, does not
offer clues for specification of innovation. Thus the functional decomposition and
system boundary were questioned. As a result of inspection and interpretation of
Innovation around Olefins 139

Figure 4-2 and Figure 4-3, a slightly different decomposition and boundary selection
was adopted to include the system level beyond the petrochemical industry. Thereby
it may be seen that the 'olefins' or 'polyolefins system' is part of an industrial
infrastructure, which major ODF is to provide packaging material, construction
material and raw material for more durable consumer goods (Figure 4-6). With this
model we can consider the whole chain from crude oil or gas to fulfilling the
function of 'packaging' by a polyolefin. In this system the net energy consumption is
very high. All the energy put into incorporating reactivity into the olefin is lost again
in polymerisation. Subsequently, after end-use, most polymer material is either land
filled or incinerated in the waste management infrastructure. The net result is that
exergy; the potential to do work is lost.
The functional model of our present system given in Figure 4-6 also indicates that
recycling to each other element may be realised possibly via some recycling industry.
Exceptions noted, however, present implementations of this functional element
'recycling' are relatively scarce. The conversion of plastic waste to feedstock, for
example, to date has not been commercially exploited on a large-scale for the
petrochemical industry nor the polymer industry.
A continuous process of scientific and technological development has enabled the
production of thinner and thinner packaging material, thereby consuming fewer
kilograms of polymer per packaging application45. For several decades, however,
packaging and consumer product supply chains have co-evolved, which for example
led to the supermarket of today that sells pre-packed fresh pizza's in competition
with pizza-restaurants. Such development illustrates that the improvements in
material have lead to an increase in the application spectrum. The net result is a
growth in the end-use of plastic, which trickles through to, amongst others, the
production capacity employed of ethylene and propylene: the doubling of ethylene
capacity in the past 16 years cannot be attributed to the population increase and
welfare increase in emerging economies alone. This growth is accompanied by
growth of two 'non-objective outputs' (Ch 4): resource depletion and CO2 emission
(Figure 4-6). This illustrates that sustainability of the petrochemical industry requires
innovation at higher system levels.
In natural ecosystems, the 'decomposers' sustain complex ecosystems such as
rainforests by enabling the high rate of nutrient cycling at a relatively constant (solar)
energy input. Long-term innovation of the petrochemical industry should enable
realisation of a similar industrial society, where essential materials are retained and
recycled at constant net energy input and acceptable CO2 emission.

4.4.3 FUNCTIONAL REPRESENTATION OF THE OLEFINS SYSTEM

Similarly to the case of aromatics (Dijkema et al. 2001), and as a matter of fact to any
chemical that is not produced for end-use by consumers, we can decompose the
industrial olefins system into a production system and a consumption system. The
main ODF of the Olefins production system today is to cater for the demand for
ethylene, propylene and butylenes. An additional function is the provision of
aromatics.

45 This is an example of progressisve 'dematerialization' - to fulfil functions with less or no material.

140 Process System Innovation by Design

The ODF of the olefins consumption system is NOT olefins consumption, but this is a
consequence of fulfilling its ODFs.

Satisfaction /
CO2-emission End Use Fulfilment
(non-ODF)
FE 1 FE N
FE 2

Manufacturing Industry

FE 1 FE N
FE 2

ds
in .
ag or..

u er

oo
od m

G
Du cts
g
Waste

Pr nsu
Pa ials

e
bl
Co
er

ra
ck
at
M
Polymer production
Recycling
FE 1 FE N
Industry
FE 2
s
er
m

FE 1
o
on
M

Petrochemical Industry FE 2
Fuels for...
FE 1 FE N
FE 2
he
po

Fee

FE N
ati
we

d s to
r

Fuel
g
tr a

Ge
ns

ck
po
rt

Energy Industry
Waste
Depletion of Resources (Non-ODF)
FE 1 FE N
FE 2 (non_ODF)

Figure 4-6: Functional decomposition of industrial society around the petrochemical

industry.

Olefins production system

The main functions of the Olefins production system today are to cater for the
demand for ethylene, propylene and butylenes. An additional function is the
provision of aromatics. These functions presently are performed by its major system
element, the steam cracker. A significant part of the provision of propylene as well
as that of aromatics is by refineries through their cat cracking operations.
At this system level, the functional difference between steam cracking and cat
cracking is that the primary ODF is 'produce olefins' and 'produce a high-value
naphtha cut for use in the gasoline-pool' respectively. An additional difference is that
the steam cracking operation is limited to feedstock up to gas oil, whilst cat cracking
uses heavy-gas oil, vacuum gas oil or light residue. The realisation of the ODF '
produce olefins' thus appears to be a combination of feedstock and technology,
Innovation around Olefins 141

which is a common characteristic of production processes in the petrochemical

industry.
The actual selection of the preferred (feedstock, technology) combination to execute
these ODFs depends on a host of factors, including system investment cost, time-to-
investment, match of the product spectrum to the relative demand for ethylene to
propylene to butylenes, feedstock availability, operating cost, availability of proven
technology and last-but-not least, regulation. The petrochemical industry is
characterised as a 'regulatory-driven industry' (Stobaugh 1988), which means that its
present state is optimised with respect to prevailing regulation. A change therein can
severely impact process economics. Changes may directly impact the economy of
new and existing installations, e.g. when emission limits are reduced or a CO2 tax is
levied. More often, regulation indirectly impacts part of the industry. For example
the EU Auto Oil regulation on the aromatics content in gasoline, for example, has a
significant impact on pre-Auto Oil steam cracker economics because a reduction in
gasoline-allowed aromatics content will lead to a surplus of aromatics and lower their
value. This requires changes in steam cracker operation or technology to maintain
the process' profitability.
We will address these issues again when discussing evaluation of improvement options
identified. In the remainder of this section, we will develop and use functional
modelling to specify possible content of process system innovation. The functional
model of olefins production is summarised in Figure 4-7. The main operation of
steam cracking and cat cracking is to breakdown large molecules into (preferably
unique) smaller ones. The precise set of ODFs is largely dependent on the particular
feedstock available and the system context. An overview is presented in Table 4.5.

Table 4.5: Functional characteristics of olefins production

Feedstock Product ODF (s) Context
1 Ethane Ethylene, Convert light, aliphatic
propylene feedstock to Ethylene,
Propylene
2 Ethane/Propane Idem Idem Petro-
3 LPG or butanes + Butylenes Idem, and butylenes Chemical
4 Naphtha + Aromatics Convert heavier, Complex
predominantly aliphatic
feedstock to Olefins and
aromatics
5 Gas oil Idem Idem
6 Vacuum Gas oil CC-Naphtha + Convert heavy feedstock
propylene into high-value gasoline
blending component &
Refinery
propylene
7 Residue CC-Naphtha + Idem
propylene
142 Process System Innovation by Design

The major by-product of steam cracking is an aromatics-rich stream. In the drawing

it is indicated that these are also produced from refineries and as a by-product from
coke-oven production.
In Figure 4-7these main functions have been visualised. In addition to the primary
functions of a refinery, fuels production, and a steam cracker complex, olefins
production, a novel function has been introduced over time, which is generalised as
co-product recovery. With the development of cat crackers with increased propylene
production, the cat cracking may become more a part of olefins production than
refining (Figure 4-7). This drawing also explains the present move of refiners and
petrochemicals producers to integrate their business or function(s).

CO2-emission
(non-ODO)

Olefins Production System

,X
B,T
e
ylen
Nat. Ga Convert Recover Eth
s Liquids Light- Aromatic
Aliphatic Byproducts ne
Hydro- Propyle
carbons to
) Olefins 1-Buty
PG
(L lene
tane
/ Bu
p ane
Pro
Convert
Naphtha Heavier Convert
Hydro-carbons Heavy Oil
to Olefins
Fractions to
Gasoline
oil blending
Gas component;
Recover
Propylene
oil
Gas ue
y id
e av es
H R
c.
Va

Figure 4-7: Functional model of olefins production.

At this functional level, thus it is obvious that not only refinery fractions are eligible
feedstock, but also other sources shown, such as synthetic crude-oil fractions from
Fischer-Tropsch synthesis plants that use synthesis gas that either originates from
Innovation around Olefins 143

Coal (Sasol) or natural gas (SMDS). The processing of (mixed) polymer waste in for
example the Veba process would also provide a suitable feedstock because its main
product is a synthetic crude that contains large and aliphatic naphtha and gas oil
fractions (Dijkema and Stougie 1994). Thus, the present fulfilment of packaging
visualised in Figure 4-6 indeed can be realised in a material cycle.
As stated above, the steam cracker can be decomposed into two system elements,
viz. a hot section and a cold separation/purification section.
If we inspect the steam cracker system diagram in Figure 4-4 a bit more closely,
however, we see that a cracker complex consists of a number of distinct functional
elements, viz.
1. Feed selection and feed preparation
2. Fuel selection, preparation and application
3. Various types of cracking furnaces (optimised for various feedstock)
4. A quench operation to provide rapid-cooling of the reaction mixture
5. A stabiliser to separate gases (olefins + saturated hydrocarbons + light gases)
and liquids (aromatics, non-aromatics)
6. A compression section to bring the gases up to suitable pressure for subsequent
cold separation
7. A cold separation section that produces individual olefins and a number of by-
products
8. By product recycle(s), respectively to cracker feedstock and energy supply
9. A separation section that produces individual aromatics and a number of by-
products

Function 1 and 2 are a consequence of the operation and feedstock selected in the
cracking furnaces, while 5 and 6 are a requirement for 7 and result in 8 and 9.
(Figure 4-8).
A logical, high-level decomposition therefore would be one where this system has
four main functional elements, viz. a reactor, separation section, recycle and product
purification . Thus, we are back at the 'standard' model of any chemical plant
(Dijkema et al. 1998) . In order to enable more specific starting points for system
innovation, at the next level of detail, these three functional system elements can be
further decomposed.
From the descriptions above it may be seen that in the reactor section the range of
functions is performed consecutively and/or simultaneously as listed in Table 4.6.
Following the flow of the feedstock / reactants through the installation, functions 1
to 3 take place largely consecutively. On the level of the chemical reactions taking
place in the tubes, both decoking and coking takes place continuously, but as coking
outweighs decoking, a special furnace operation mode of decoking is required.
Obviously, functions (1,4), (2,5) and (3,6) are paired.
144 Process System Innovation by Design

CO2-emission
(non-ODF)

Aromatics
,X
B,T
Recovery Heavies
Conversion
FE 1
(non-ODF)
FE 1

FE N
Product
e
ylen
FE 2 Purification
Eth
Recycle FE 1
e
Propylen
FE N
Feed
FE 2
Nat Selection &
. Ga Preparation Product 1-Butylene
s Liq Separation
uid s
FE 1 FE N
Propan FE 1
e / Buta
ne (LP
G)
FE 2 Fuel
Naphtha Selection & FE N
Preparation
oil FE N
Gas
FE 1

FE 2
il
aso Light gases
av yG
He Natu (non-ODF)
r al G
FE N as

Figure 4-8: Functional model of the steam cracker.

Table 4.6: Objective-defined functions in the reactor section

Operation ODF
1 Heat-up Create favourable conditions
2 Chemical reaction Generate desired products
3 Cool-down Prevent product degeneration
4 Supply Energy for Heating Transfer energy to reaction mixture
5 Supply Energy for Cracking Idem
6 Withdraw Energy for Cooling Transfer energy from reaction mixture
7 Supply energy for Flow Transfer energy to reaction mixture
8 Decoke (Remove coke deposit) Ensure continued correct operation
Innovation around Olefins 145

In the general model of a chemical process, separation is a consequence of the

imperfection of the reactions carried out. In the steam cracker, this is profoundly the
case because a complex mix of products is obtained from the cracking furnaces that
require a substantial range of operations to achieve the final ODF of the installation,
the production of essentially pure olefins. The sections' major ODFs have been
summarised in Table 4.7.
Table 4.7: Functional elements of the Purification / Separation section
Operation ODF
1a Split light-gases / liquids
1b Compress the light-gas Create favourable conditions for cryogenic
separation
1c Cool the light gas
2 Olefins purification Remove unconverted feedstock, undesired
co-products (acetylenics, CH4, H2)
3 Ethylene isolation Produce pure ethylene
4 Propylene isolation Produce pure propylene
5 C4 isolation Produce C4 mixture
6 Withdraw Energy by Cooling Transfer energy from gas/liquid flow
7 Supply Energy for Heating Transfer energy to liquid/gas flow
8 Supply energy for Flow Transfer energy to gas or liquids

This is largely a mutually contingent set of ODFs in the case of steam cracker cold
separation, possibly with the exception of ODFs (1b,8) because gas-compression
serves both to increase gas pressure, and to let the gas flow through the installation.
From the table it is clear that the operation selected for olefins isolation, cryogenic
separation requires a host of supporting functional elements.

Olefins consumption system

The function of the Olefins consumption system in the petrochemical industry is to
provide monomer material that can be used by the polymer industry. In addition, it
supplies some solvents and disposables (notably anti-freeze and cleaning agents).
Over time, the realisation of conversion of olefins to alternate monomers has
developed into the industrial network depicted in Figure 4-2 and Figure 4-3. The
existence of this network and its ODFs are intrinsically related to the ODFs to be
met by polymeric materials such as polyurethanes. Since the special performance
characteristics of polyurethanes are much appreciated, their production volume
growth has created considerable demand for propylene oxide, the precursor of
propylene glycol and propylene glycol ethers that are essential polyurethane
ingredients. The petrochemical industry, notably Daicel, Oxirane (later Arco, now
Lyondell) and Shell Chemicals has responded to this growth in demand and
environmental concerns on the chlorohydrin route by developing alternate
realisations of PO production. Of the many possible realisations (hydrogen
peroxide route, organic peroxide routes, organic peroxy acid routes), only the
146 Process System Innovation by Design

TBA/PO and Styrene/PO routes appeared to both chemically and industrially

feasible. Despite process improvements, the original chlorohydrin route (Dow
Chemical) became environmentally suspect. Direct oxidation still has not been
successfully realised. In case alternative fulfilment of the ODFs of polyurethanes
would be developed, however, over time the entire ODF to meet polyurethane
monomer material demand would disappear. Thus, the ODFs of olefins
consumption are 'meet the demand of a set of monomer materials'.

4.4.4 TOWARDS PROCESS SYSTEM INNOVATIONS

Now that we have formulated a functional model, we can (1) categorize innovations
reported/found in the literature and validate the model by identifying matching
conventional and new ODFs and (2) try and prove our methodology by the
specification of original innovations.
The functional model is a technology-free specification of systems around olefins,
and “because the decomposition is based on system and subsystem boundary
selection it opens the way for consideration of alternative systems by expanding the
set of functions in the model at a particular system level, or to view the system as a
part of a larger system” (Dijkema et al. 2001).
As the objective of this case study is to demonstrate the usefulness of functional
modelling and to 'prove' the methodology for process system innovation by design,
rather than to provide an exhaustive review and outlook for olefins production
overview, only a limited number literature resources have been used to categorise
some of the 'present state-of-the- art'. Notably the proceedings of the DGMK-
tagung 2001 (Emig et al. 2001) and Albright (Albright et al. 1992) provide a good
overview of the state-of-affairs and the change therein that is suitable for our
purposes. At the system level beyond the steam cracker, also some of our work on
waste management and industrial ecology has been incorporated.
In the methodology innovations comprise the specification of novel realisations of
existing ODFs or the identification of novel ODFs and their possible realisations.
Both types of innovations have been categorized and specified for the system within
the boundaries of the steam cracker model as well as for a larger system, e.g. when
the system boundary is expanded to span olefins production or the entire olefins
system.
In this section firstly innovations are categorised and specified to fulfil the ODFs
within the system boundary of the steam cracker. Subsequently, the system boundary
is expanded to olefins production and finally to span the entire system around
olefins.

Innovations within the steam cracker process system

Functional model. Over the years, the set of ODFs of steam cracking complexes has
expanded. Initially, ethane/propane crackers were built for ethylene and propylene
production, and were seen as a good opportunity to upgrade these co-products from
'wet-gas' production. The main ODF for these plants thus was to' upgrade wet-gas
co-products'. With the growth of the ethylene and propylene markets, notably in
Europe, crackers operating on naphtha feedstock were built to cater for this
demand. The main ODF for these plants thus was (and is) to 'produce sufficient
Innovation around Olefins 147

ethylene and propylene'. One of the co-products of these plants, aromatic pyrolysis-
gasoline was soon recognized as providing a source for functions, the production of
essentially pure B, T and X. Another function introduced was the isolation of pure
C4-olefins. Lately, C5-olefin isolation and hydrogen recovery have been introduced,
and processes for the isolation of ethyl benzene and styrene have been developed.
Today, the major ODF of steam crackers appears to be 'produce an economic mix
of ethylene and propylene'.
(1) Categorize/validate. Within the steam cracker flow sheet, major innovations have
been developed and implemented over the years (Hutcheson et al. 1995). In the hot
section, amongst others, these include:
• Improved furnace-tube materials and design, which led to higher olefin
selectivity, yield and reduced coking (fulfilment of functions 1&2)
• Improved furnace design for greater capacity (process intensification) and
greater furnace efficiency
• Milliseconds technology for improved selectivity and yield

In the separation section these include:

• Improved compressor design allowing greater cracked-gas compressor single-
train capacity at improved efficiency
• Improved distillation tower design, allowing greater capacity and separation
efficiency
• New 'cold-box' technology that leads to significant reduction in power
consumption for cooling

All these changes and improvements fall under the umbrella of the functional model
of the cracker and its elements presented, as they deal with improved 'fulfilment' of
said functions. As stated above, these have increased the efficiency of operations
considerably, with the possible exception of the milliseconds technology. It is easily
calculated, however, that today's efficiency of a steam cracker is still only some 65-
75% (based on LHV of feedstock and (co-) products.
Process system innovations have been developed that must be categorised as 'novel
ODFs/functional elements. At the level of the flow sheet, these include the concept
of 'side-cracking' furnaces for dedicated ethane/propane recycle, the isolation of
individual butylenes and the inclusion of metathesis to shift the relative yield of
olefins (Stanley 2001)).
(2) 'Proof' of the methodology. The core of this thesis is a methodology to arrive at
specification of the content of innovations in complex systems. Using functional
modelling and decomposition the ODFs of the steam cracker have been established.
Subsequently, the expert knowledge and results combined in (Dijkema et al. 1995)
and literature cited was combined. The functional model was used to identify
alternative ODFs for the system and to identify alternatives for fulfilment of the
ODFs in the steam cracker model. Together, these have led to the following
suggestion for innovations within the system boundary of the steam cracker complex:
148 Process System Innovation by Design

1. The transformation of a steam cracker to a trigeneration concept (Dijkema

2001b).
We have argued that chemical reactors that operate/require energy supply at a
high-temperature offer a large possible heat sink for energy conversion devices
for the production of electric power. Similarly to the implementation of
cogeneration, a cracker-furnace could thus be equipped with gas turbine(s) where
heat is supply to the cracking furnace from the GT-'s exhaust (Loffel and Schulz
1984). Although this option has been implemented in a few cracker units
worldwide, it has not been part of the recent cracker revamps. The concept
should be re-evaluated in the present liberalised electricity market and CO2
emission-trading schemes whilst taking into account the present state of furnace
technology.
2. Introduction or enhancement of steam cracker feed preparation (of naphtha and
gas oil) to eliminate (TX-) aromatics being sent to the cracker furnaces.
This would reduce the amount of methane due to dealkylation in the cracked
gas, and would allow direct utilisation of these aromatics. To absorb and
effectively utilise the 'glut' of aromatics that results after Auto Oil I and II,
similar to an ethane/propane side cracker, an aromatics side cracker or aromatics
extraction + conversion (hydrogenation) could be developed. Alternatively, in-
situ selective aromatics hydrogenation may be an attractive option despite the
requirement for hydrogen.
3. Steam cracker process system for flexible operation
The main ODF of any steam cracker is “to produce an economic mix of
ethylene and propylene”. This implies that operation flexibility with respect to
feedstock diet and product spectrum must be catered for. Apart from increased
furnace operational flexibility, the inclusion or addition of novel functional
elements can help achieve this. Examples are 'Side-cracking' furnaces for
ethane/propane, 'metathesis' reactors and aromatics converters (see below).
4. The exploration of reactor concepts where (the functions of) dehydrogenation
and cracking regimes are implemented largely separately.
By physically separating dehydrogenation and cracking, the former and possibly
the latter then may be conducted catalytically at greater yield and selectivity.
5. Enhanced-recovery of steam cracking co-products
The steam cracking operation produces a variety of chemicals in the cracked-gas
that presently are labelled 'impurities' or as invaluable. Both the resource
utilisation and economics of steam crackers could improve if one would consider
• Co-production of various-grades of hydrogen / methane mixture
• Co-production of styrene by isolation of styrene
• Co-production of di-olefins, acetylene and acetylenics for use in specialty
chemicals production
Innovation around Olefins 149

6. Extraction of olefins, use of aliphatic hydrocarbons produced for syngas

generation.
It is well known that syngas (CO/H2-mixtures) are can be produced from higher
aliphatic hydrocarbons under relatively mild conditions. Thus, instead of using
the higher aliphatic hydrocarbons (C5-C8) that are present in straight-run naphtha
as steam cracker feed, these may be selectively converted to syngas and C3-C4.
Such an option, however, also requires effective separation of C3/C4 from the
syngas mixtures produced. In other words a number of process innovations are
required to bring said process system innovation into being.

Innovations beyond the steam cracker system

Functional model. In the currently available systems for olefins production, notably
the yield of ethylene and propylene does not match the demand of these products.
Propylene demand is growing much faster than ethylene demand; even to an extent
refinery propylene is not expected to be able to match these demands.
Thus, the past ten years (Stanley 2001), companies worldwide have been active in
developing new ways to manufacture propylene. These options represent novel
functional elements in the olefins production system model.
(1) Categorize/validate. An important category is the processes that exploit methanol
as a feedstock:
• Methanol-To-Olefins (MTO)
A process to convert methanol to largely ethylene and propylene. After initial
development in the eighties (van Geem 1994), currently a scheme via the
intermediate production of dimethyl ether (DME) is favoured. This process is
nearly commercialised, with the first plant being constructed.
Functional element ODF: convert MeOH to ethylene and propylene
Important additional Function: produce water!
Actually, from some 1.7 MTA MeOH, the plant will produce 0.9 MTA water,
and 0.59 MTA propylene
• Methanol-To-Propylene (MTP)
A process that is developed by Lurgi that is in its demonstration phase.
Important additional Function: produce (process) water.
Actually, from some 1.7 MTA MeOH, the plant will produce 0.9 MTA water,
and 0.59 MTA propylene. Where Lurgi (Rothaemel and Holtmann 2001) argues
the process water can be made suitable for agricultural purposes, we would
suggest that both an MTO or an MTP plant could play an important role in the
water (cycle) management of an industrial site. In both cases trace chemical
species in the water must be taken care of.
Another option is a function-shift of cat cracking, such as developed by UOP
(Greer et al. 2001). Where the latter is probably limited by the amount of
aromatics by product, the routes via methanol have largely been developed on
the incentive of 'monetizing' remote natural gas-fields that otherwise must be
considered stranded. Combined with the launch of the MegaMethanol
150 Process System Innovation by Design

technology by Lurgi, a route that is competitive with oil-based olefins appears to

be in reach.
Metathesis of olefins is a function that is best described as 'conversion of a particular
olefin into a spectrum of other olefins'. Thus, metathesis has been developed for a
variety of feedstock, notably butylenes and higher for the production of additional
ethylene and propylene. Presently, the function is advocated for inclusion in the
steam cracker flow sheet because of the need of subsequent separation of the olefins
produced. It is expected, however, that such flow sheets exhibit a lower resource
utilisation than present ones. This is because realisation of this function requires
additional operations with inevitable losses due to irreversibilities. Another chemical
conversion of use in the olefins systems comprise 'Deep-FCC' for propylene
production: refinery cat crackers can be equipped with modified catalysts and be
operated at conditions optimised for propylene yield. Aromatics may be subject to
hydrogenation followed by ring opening and use as steam cracker feedstock. This
requires the availability of hydrogen, however. Oxidative coupling is a 'holy-grail' in
industrial organic chemistry: combine two molecules of methane and a single
molecule of oxygen to yield ethylene and water. Unfortunately, to favourably execute
this reaction has been impossible for the past three decades. Oxidative
dehydrogenation of ethane is another candidate to realise ethylene production.
(2) 'Proof' of methodology. The inclusion of novel functions in a system preferably
should help exclude those functions within the system that can only be implemented
at relatively low resource utilisation. Such functions can be entire processes
(functions) in a network or parts thereof such as the steam cracker separation
section.
Upon first inspection the options given above do not meet this criterion. In the
olefins system, however, the development of suitable polymers and polymerisation
of ethylene/propylene mixtures would greatly simplify the cracker separation
section and contribute to system flexibility.
Options that span multiple systems could also result in the avoidance of costly
separation, e.g. cannot an ethylene / ethane mixture be used for polymerisation,
where eventually ethane is recycled back to a steam cracker? At the feedstock end,
mild cracking could be introduced as an additional feed preparation to off-load
(inefficient) cracking furnaces. When waste-heat in such installations would be used,
this present a case where efficiency is increased by splitting an existing function into
two complementary functions.

Innovations beyond the petrochemical industry

Functional model. Moving beyond olefins production and consumption, it may be
seen that these functional elements are part of a longer chain, the fast carbon
material cycle (Ch.1). To date the route back from consumer goods to polymer
feedstock appears to be via CO2 emissions to the atmosphere. Obviously, for
polymers direct mechanical recycling is to be preferred because this does not involve
chemical conversion - cracking or depolymerisation, purification and
repolymerisation. In many cases such direct recycling cannot be realized, however,
because no suitable technology exists, e.g. to recycle polymers in non-ferro
automobile shredder waste. In other cases, the recovery chain is economically
Innovation around Olefins 151

infeasible. As a consequence, the implementations of recycling routes that include

industrial chemical processing are largely lacking, with the exception of large-scale
incineration.
(1) Categorize/validate. The R&D effort on the recycling of plastics involving some
industrial processing has been immense. As indicated in Figure 4-6, either an existing
facility for the production of olefins can use polymers, or it may convert polymer
waste-derived feedstock. In both cases, a more efficient exploitation of the industrial
part of our global carbon cycle would result. A number of ODFs can be
distinguished: 'enable recycling of polymeric material' ('back-to-polymer'), 'convert
waste polymer to useful monomer materials' ('back-to-monomer') and 'convert waste
polymer to suitable hydrocarbon mixtures' ('back-to-feedstock').
A few commercial plants are in operation that employ depolymerisation and back-to-
feedstock processing. Around packaging, one option that was heavily debated in the
Netherlands after implementation of the 'Duales System Deutschland' in Germany
was the recycling of mixed plastic waste. In a system study on the Veba-Oel route
for the processing of Mixed Plastic Waste (MPW), a back-to-feedstock route, the
system's ecology appeared to be favourable but the economics were not. In addition,
the organisational aspects and public acceptance of MPW collection were considered
questionable. At the time, however, olefin and polyolefin manufacturers feared that
successful recycling would be detrimental to the economy of their operations, as
they anticipated a loss of market share for the production of virgin material
(Dijkema and Stougie 1994; Sas et al. 1994a). In the study cited, only one option to
fulfil the ODF 'convert MPW' back-to-feedstock has been evaluated. In the near
future, however, olefin and polyolefin manufacturers may be threatened by a
competing system for provision of the ODF 'packaging' that is based on
biopolymers, especially when these can be produced with a CO2 neutral - label.
On a different track, in the present system for the ODF 'packaging' a lot progress
has been made to 'dematerialise packaging' by reducing thickness of the polymer
sheet used.
(2) 'Proof' of methodology. Functional modelling indicates a system solution is
required where needs are fulfilled instead of materials being supplied. Since a finite
amount of materials always will be used, recycling technology is required. A novel
functional element in the material cycle around olefins and polymers would be 'back-
to-infrastructure'. Rather than using recycling technology on location for a specific
product, the end-of-life conversion of polymers can be connected to electricity,
natural gas, synthesis gas or hydrogen infrastructure to realise effective conversion,
recycling and distribution of the recycle products. Specific realisations will comprise
of integrated system designs that can make use of depolymerisation, pyrolysis,
gasification, product purification and incineration with energy-recovery.

4.5 Discussion and Conclusions

In addressing possible innovation around olefins, the value of a structured approach
has been illustrated. Assessment based on a 'trivial' decomposition of the industry
provides a starting point for a systematic analysis and search for options for R&D.
The results, however, are largely confined to the initial process concept. Functional
modelling provides a structured method of abstraction of the current method of
152 Process System Innovation by Design

'fulfilment' of the present ODFs of the industry. In addition, it allows an integration

between system levels and innovation thereof.
In the case of the steam cracker, it has been demonstrated that categorisation of
innovations reported in the literature using the functional model developed provides
a validation of methodology. The novel functions and novel methods of realisation
of the functions categorized or specified are 'proof' of the methodology for process
system innovation by design.
Thus, functional modelling does indeed provide a route to specify the content of
process system innovations in the industrial systems around olefins.
 5 Fuel cells and Trigeneration

5.1 Introduction
This thesis includes three chapters on application of the methodology for 'process
system innovation by design' as developed and presented in Ch. 3. In the previous
chapter, a case study on an important sector of the petrochemical industry, olefins,
has been reported (Ch.4). Process system innovation content was specified assuming
a largely unchanged system external world. In Ch. 6 the impact and opportunities are
addressed that are related to a dramatic change in the external world of the
petrochemical industry, the emergence of PEM fuel cell vehicles in automotive
transport.
This chapter focuses on using the methodology to explore and specify innovations
for chemical process systems that make intelligent use of an extra-sector innovation,
fuel cells. These are a novel class of electrochemical energy-conversion technologies,
which are being developed for the energy sector and automotive transport (see also
Ch. 6).
The crux of this chapter is the recognition that it is the combination of fuel cell
functions that must be exploited in process system innovations. The first part of this
chapter consists of an in-depth analysis of the use of fuel cells in the petrochemical
industry at three system aggregate levels - unit operation/reactor, process system
and industrial complex. This work has been presented at the 1999 Joint International
Meeting of the Electrochemical Society46.
In the second part of this chapter, the concept of trigeneration is explored.
Trigeneration is the combined production of chemicals, electric power and heat. Functional
modelling is used to develop innovative system concepts to realise trigeneration. A
paper on this work has been presented at the 6th World Congress on Chemical
Engineering, 200147.

46 Dijkema, G.P.J. (1999a,b), Fuel Cells in the Chemical Industry, Proc. Electrochem. Soc., 196, 1013
(Abstract); Carbonate Fuel Cell Technology V, 306-321 (Paper); Hawaii, The Electrochemical
Society, Inc.
47 Dijkema, G.P.J. (2001) The Development of Trigeneration System Concepts, Proc. 6th World Congress on

Chemical Engineering, Melbourne, Australia.

154 Process System Innovation by Design

5.2 Fuel Cells in the Chemical industry

5.2.1 ABSTRACT
The use of fuel cells in the chemical industry leads to trigeneration systems that
produce useful chemical products, electricity, and heat simultaneously.
Firstly, stand-alone fuel cell cogeneration systems can be used to upgrade a chemical
plant’s by-product hydrogen or CO (Figure 5-1). Secondly, fuel cell stacks can be
employed as a multifunctional unit operation in chemical process system design,
which results in process-integrated applications (Figure 5-2). Finally, fuel cell
reactors combine the production of chemicals, electric power and heat in a single
electrochemical apparatus (Figure 5-3). In retrospect, the use of fuel cell technology
can thus be reviewed and categorised. In the chemical industry, however, the interest
in fuel cell application appears to be limited to date. In the paper therefore the focus
is on early identification, adequate representation, and early economic evaluation of
the possible applications of fuel cell technology in the chemical industry.
A systematic approach for the identification of fuel cell applications in the chemical
industry comprises three steps: (i) identify all functions of a fuel cell device; (ii)
identify all functions required of auxiliary system equipment; (iii) select suitable
chemical processes that can offer some of the auxiliaries’ functionality (Dijkema et
al. 1998).
A novel schematising technique is introduced that supports this method by
adequately representing the fuel cell system and the functions of its elements. The
method can be combined with elementary mass-balancing and thermodynamic
calculations.
A trigeneration system was developed for a new MCFC with separate CO2 supply
(Hemmes and Dijkema 1998). The result demonstrates that the approach developed
is a powerful tool to screen the potential of a technical innovation.
The early assessment of the scope for fuel cell application in the chemical industry is
completed by the development of an early economic evaluation method. The
additional profits from trigeneration largely determine the allowable investment for
the fuel cell (Vayenas et al. 1991b). In addition, the leverage effect on chemical plant
cost must be accounted for. The results revealed that the economics of fuel cell use
in the chemical industry are favourable compared to stand-alone fuel cell systems for
electric power generation.
In the upgrading of hydrogen by-product the fuel cell only offers a competitive
advantage when the electric power revenue exceeds the value-added of the recovery
and the subsequent conversion of the hydrogen to chemical products.
The use of process-integrated fuel cell modules is attractive because no alternative
technology exists that offers similar functionality. In addition, employing fuel cells
this way improves the performance of the chemical plant, which can provide
leverage to a fuel cell project.
Since fuel cells lack economy-of-scale, it was expected that fuel cell reactors could be
competitive only in suitable small-scale applications. Based on calculations on
potential electric power generation from a number of industrially important
Fuel Cells and Trigeneration 155

reactions, however, it was concluded that only relatively small-scale fuel cell units (6-
40MWe) are required for common large chemical plants. In addition, fuel cell
reactors appear economically competitive for most partial oxidation reactions.
Notably the partial oxidation of toluene to phenol stands out as a prime candidate
for further exploratory research.

Chemical plant
Feed Reactor Separator Product

Byproduct H2 / CO

Fuel cell
Electricity
system

Heat

Figure 5-1: Trigeneration through by-product utilisation with a cogeneration fuel cell
system.

Chemical plant
Feed Reactor Separator Product

Fuel cell
Electricity
stack

Heat

Figure 5-2: Trigeneration through process-integrated use of a fuel cell stack.

Chemical plant

Fuel cell Separator Product

Feed
reactor
Electricity

Heat

Figure 5-3: Trigeneration by implementation of a fuel cell reactor.

156 Process System Innovation by Design

5.2.2 INTRODUCTION
Traditionally, fuel cells have been promoted because of their high thermodynamic
efficiency for the generation of electric power. Their second advantage proclaimed is
their cleanliness (Hirschofer et al. 1994).
As of late, it has been realised, however, that the best fuel cell system not necessarily
is the thermodynamically most efficient system. Rather fuel cell systems must
uniquely meet customer demands at acceptable cost.
In the chemical industry ample opportunities exist for such fuel cell applications.
Not only because equipment can be shared, but also because the fuel cell application
can have an effect on the performance of the entire plant and thereby provides
economic leverage to a fuel cell project (Dijkema et al. 1994a; Schinkel et al. 1994).
The overall impression of the literature on fuel cell application in the chemical
industry is one of incidental, fragmented publishing of possible uses. The
trigeneration system concept therefore is used in this chapter to present all
prospective fuel cell applications in the chemical industry under a single systematic
framework.
Firstly fuel cell system design is addressed and the “functional scheme technique” is
presented. Subsequently, an overview of the literature is given per category of
prospective fuel cell use in the chemical industry (see Figure 5-1, Figure 5-2and
Figure 5-3). Thirdly, the functional scheme technique is used to augment the
identification of fuel cell applications (Dijkema et al. 1998). The merits of the
combination are illustrated by the development of a trigeneration system for a the
newly invented improved MCFC with separated CO2 supply (Dijkema 1998;
Hemmes et al. 1998). Finally, a method for early economic evaluation of fuel cell
applications in the chemical industry is presented and results are reported.

5.2.3 FUEL CELL SYSTEM DESIGN

(Kordesch and Simader 1996) give an extensive overview of fuel cell development
for stand-alone power generation with a high thermodynamic efficiency. In the
preface of their book, they state that the prime development goal for fuel cell
technology is ‘dispersed, efficient power production’ and ‘... that a long life is
needed for power plants ...’. As a revelation, the authors quote that the real
bottleneck in fuel cell technology is the fuel to be used, i.e. that hydrogen is far too
costly. This kind of conclusion is typical for the outcome of system design studies,
as they must provide guidance for R&D.
In many publications fuel cell systems are presented as process flow schemes, with a
varying amount of detail. Usually, however, important individual equipment items
are shown to the level of individual apparatus such as heat exchangers, pumps,
compressors, burners etc. The account, however, of the actual design process
unfortunately is often limited. More often than not, the focus is on system
simulation and design calculations to evaluate the system’s performance, rather than
on the arguments for the decisions reached on the basis of design.
In addressing fuel cell system design for the guidance of R&D, we can draw from
the literature on system engineering, which has been defined as “a progressive definition
and integration of functions and design decisions at every level of system complexity”
Fuel Cells and Trigeneration 157

(Corrigan and Kaufman 1965). Systems can be decomposed into progressively

smaller subsystems, where system elements are the lowest level of detail of a system
structure.
System complexity can be dealt with appropriately in design, when the focus is on
system functions because the functional characteristics of a system element are
technology-free (Asbjørnsen 1992; Dijkema and Reuter 1999). This means that when a
system element is described by it’s inputs and outputs, and labelled by one or more
functions it performs, the description is complete and independent from the actual
technical realisation of the element (Figure 5-4). The technical content of the
element only determines how the relationship between inputs and outputs has come
about. Thus it may be seen that such a technology-free description opens the way to
search for alternative technologies in system design.
Consider, for example a system that must use natural gas for the generation of
electric power. When we consider a conventional large-scale power plant, we see
that the system’s inputs are natural gas and oxygen, its outputs electric power, flue
gas, and a waste stream. If we leave out the technical details, we see that the system
representation has become technology free. The next step is to consider other
options to bring about the required function, power generation from natural gas. In
Figure 5-4 two alternative options are shown, the use of a gas turbine and a fuel cell
respectively, where in both systems it appears that we require fewer apparatus.

Air supply
Air Exhaust
Energy
Air Energy transformation
Exhaust
supply transformation Air
Step IIa
E
Fuel T Power
E Power

Abstraction
Fuel

Step Alternative Realisation I

IIb
Step I

Air Energy transformation

supply Exhaust
Air Energy
supply transformation
Air Fuel Anode
Exhaust
Cathode

E Power
T Air
Power

Fuel
Current Realisation Alternative Realisation II

Figure 5-4: System development – fuel to electric power (redrawn).

By focusing on the functional description, however, in this case we have applied
rather sloppy descriptions for the inputs and outputs. In the case of the GT, we still
produce flue gas, but it also contains waste heat around some 450-500 oC. In the
158 Process System Innovation by Design

case of the fuel cell, the flue gas will contain unconverted feedstock. In all cases we
have assumed the feed is at the right condition (P, T).
We suggest, therefore, that at each level of system complexity, the following
procedure can be applied:
1. Review an existing (similar) system design and identify the crucial system
functions;
2. Draw a functional scheme
3. Briefly describe each system element
4. Explore alternative technical realization options per system elements
5. Explore possible combinations of functions in a single apparatus
6. Consider alternative recycle schemes to achieve 5&6; the flow sheet recycle
structure needs careful attention (Douglas 1988).

In Figure 5-5 an early fuel cell system design is represented that operates on natural
gas. In this drawing, the system decomposition has been visualised. The system
elements have been enclosed by grey boxes. Each box’ title represents an element’s
function. The representation essentially is technology-free: the drawing is still valid
when the detail in the boxes is left out. The content depicted represent only one of
the possible technical realisations of the system elements. The particular
arrangement of system elements also represents a single possibility to ‘materialize’
the fuel cell system. Indeed, the second important feature of the drawing is the
system boundary combined with its title, which is a system element at one
aggregation level up. Note that all feed and product streams have been labelled
outside the system boundary.

Feed Exhaust
saturation Off-gas
Fuel cell module
processor

Water reformer
Anode
Power
burner Cathode
Fuel
Fuel processor

Compression
Air
Cathode
cooling
Cold Hot
medium medium

Figure 5-5: Example: functional scheme for a stand-alone fuel cell system.
A third important feature of this drawing technique is that system elements where
multiple, physically unseparated functions are performed have been enclosed by an
equal number of system element boundaries (effectuation of step 6). In this way
Fuel Cells and Trigeneration 159

alternative arrangements for the design of the system offered by the integration of
functions are visualised.
Besides the fuel cell module, the stand-alone system includes 4 other elements,
steam supply, fuel processor, oxidant supply, cell cathode cooling, and off-gas
processor.
Steam supply is a function implied by the particular selection of fuel processing in
this system, steam-reforming. Prior to reforming, methane needs to be mixed with
steam. The option shown is to integrate the steam supply in the fuel processor by
installing a feed-saturator that employs waste-heat.
Fuel processing consists of hydrogen generation and fuel purification. Since no
natural resources of hydrogen are known, it must be generated from hydrogen-
containing species.

Visualization of system improvements

An advanced fuel cell system is the Smarter system (Fellows et al. 1998). In this
system concept, gas-gas heat exchangers have been totally eliminated (Figure 5-6).
The main feature of the Smarter system is the use of a hot-recycle blower, which
allows the arrangement of a smart recycle scheme that replaces a number of system
elements by the integration of system functions.

Fuel cell module

Fuel
Anode
Cathode Hot exhaust
Feed preheat
Feed compression
Fuel processor

burner Power
Air

Off-gas processor
Air supply O2 supply
CO2 supply
Cathode preheat

Figure 5-6: Functional scheme of the Smarter system.

Recycling part of the anode off gas back to the inlet of an internal reforming fuel
cell eliminates the feed saturation system, since the off-gas will contain the water
160 Process System Innovation by Design

required for the steam-reforming reaction. In an external reforming fuel cell system
the off gas can be recycled to the reformer inlet.
The other essential feature of the Smarter system concept is the integration of the
off-gas processor and cathode oxidant and CO2 supply. The outlet of the off-gas
processor, a catalytic burner, is sent to the cathode of the direct internal reforming
MCFC together with excess air for quenching.
In retrospect, the system design procedure developed clearly illustrates the
advantages of the Smarter System compared to a common stand-alone fuel cell
system. The integration of multiple functions in a single apparatus and the
usefulness of recycles are apparent. The Smarter system is a dramatic breakthrough
in system simplification, and is a case in point for the use of the system design
approach presented.

5.2.4 FUEL CELLS IN THE CHEMICAL INDUSTRY

The use of fuel cells to upgrade by-product hydrogen

The possibility of utilising low-BTU off-gases in the process industry by
implementing fuel cells on-site has been identified as a possible high added-value
application in the fuel cell research and development community (Hirschofer et al.
1994).
Hydrogen-containing by-product streams are produced in a large number of plants
in the process industry indeed. Examples include oil refineries, steel blast furnaces,
and chemical plants that produce methanol, ammonia, ethylene, or aromatics.
The conversion of by-product hydrogen or carbon monoxide into electric power by
fuel cell systems that were developed for stand-alone power generation can be
effected as an add-on to existing chemical plants, and effectively is an end-of-pipe
solution.
A patent was issued to Haldor Topsøe on the use of ammonia-synthesis off-gas as a
fuel for an MCFC (Frydenlund 1993). ICI received a patent on the similar use of
methanol synthesis off-gas for the utilisation of industrial by product hydrogen-rich
gas (Pinto 1977). The utilisation of hydrogen by-product in the manufacture of
chlorine has been studied also (Lance 1984). Recently, Ansaldo-ClC reported the
design of hydrogen-based PAFC unit, specifically intended for the upgrading of
hydrogen by-product (Caserza et al. 1996).
The concept of by-product utilisation, however, appears not to have been
commercially applied. This may be due to the limited availability of commercial fuel
cell stacks and the limited track record of fuel cells in industrial service, which
together impede acceptance by the chemical industry as ‘proven technology’. In
addition, the development track selected by the fuel cell companies has been
focused on packaged systems for stand-alone power generation. More important,
however, is the fact that the economics of upgrading by-product hydrogen through
the use of fuel cells is apparently not good enough, which is illustrated by results
reported on the use of hydrogen by-product in chlorine electrolysis (Lance 1984),
and on the use of phosphorus-oven off-gas in an MCFC (Goossens 1995).
Fuel Cells and Trigeneration 161

On the basis of a preliminary investigation of the use of fuel cells in conventional

ammonia production, we found that using fuel cell stacks to utilise off-gas only is
attractive when a considerable electricity revenue is obtained, and when the cost of
the fuel cell equipment is relatively low (Dijkema et al. 1996). Pressure-swing-
adsorption technology (PSA) provides an economic solution for removal of the
greater part of the hydrogen from the synthesis-loop off-gases, allowing the
hydrogen to be recycled to the ammonia process. As a consequence, the incentive to
develop fuel cell applications is limited.

Process-integrated application of fuel cells

At the process system level, any chemical plant is designed as an interconnected set
of unit operations, i.e., reactors, separators, compressors, pumps etc. A stand-alone
fuel cell system design consists of a fuel cell stack and the so-called ‘balance-of-
plant’, which can also be viewed as an interconnected set of unit operations. A fuel
cell stack not only converts hydrogen into water by electrochemical reaction with
oxygen, which produces DC-power, but also acts as an active device for hydrogen
removal, oxygen depletion of air, etc. (Schinkel et al. 1994; Dijkema 1996).
Thus a fuel cell stack can be viewed as a multi-functional unit operation that can be
a part of any chemical plant, in the same as any other unit operation, such as a
pump, compressor, etc. Since this use of fuel cells involves the complete integration
of a fuel cell stack in a chemical plant, it is called a process-integrated application.
The aforementioned use of an MCFC in a conventional ammonia plant (Frydenlund
1993) is to be considered such a process-integrated application. Hydrogen-rich
ammonia synthesis loop off-gas is fed to an MCFC-anode. The pure CO2 produced
in the ammonia plant is mixed with air and sent to the MCFC-cathode. Anode and
cathode off-gases are sent to the reformer furnace. The amount of auxiliary
equipment used thus remains limited.
Synthesis gas, the common label for process gases that consist of a mixture of
hydrogen and CO, is an important intermediate in the chemical industry, e.g. for the
production of methanol and higher alcohols, and the production of ammonia.
Synthesis gas production is also the first step in most hydrogen plants. The
production of synthesis gas for fuel cells in a “fuel processor” is analogous to these
systems, except for the scale-of-operation, and to some extent the desired syngas
composition. The combination of these two therefore appears to be a natural one.
A patent was issued to CF Braun & Co on the integration of a fuel cell in an
ammonia plant based on the Braun Purifier concept (Jungerhans 1984). The
patented process system includes ammonia make-gas production by reforming, HT-
CO-shift, and purification of the syngas by PSA for the removal of CO2, residual
methane and water. The fuel cell is fed by part of the syngas produced. The anode
and cathode-gases are subsequently combined in a deoxygenating unit to produce an
oxygen-free nitrogen-rich, inert stream that is used for purging the PSA unit, thereby
producing waste-fuel. (Sederquist et al. 1993) claim a fuel cell arrangement for the
production of such pure nitrogen-gas, whereby the fuel cell operating mode is
optimised for the maximum consumption of oxygen.
(Drenckhahn et al. 1995) reported on the prospects for the use of the PAFC in the
chemical industry, in particular those chemical complexes that include the large-scale
162 Process System Innovation by Design

generation of synthesis gas by reforming. Their objective was to investigate whether

the PAFC could play a role in consuming the intermediate product hydrogen gas
from the reforming section, when the demand for the end plants suffers a reduction
which forces a reduction of operating capacity. From the paper we deduce that they
investigated PAFC use in an ammonia plant, which includes a liquid nitrogen-wash
for the removal of residual methane, which is the underlying principle of the Braun-
Purifier process for ammonia. They conclude that the most promising take-off point
for PAFC feed is where the synthesis gas has been fully treated to ammonia make-
gas.
We investigated the design of an ammonia-MCFC-based trigeneration system, where
a single syngas production facility is used to feed both an ammonia synthesis loop
and an arrangement of fuel cell stacks (Dijkema et al. 1996). The fuel cell off-gases
are sent to the primary reformer furnace as fuel and high temperature oxidant. The
economics of this system are comparable with the economics of the use of the
hydrogen-rich purge stream. The results indicate, however, that the production of
ammonia alone is more profitable than trigeneration, as electricity revenue is low.
Another illustration is the use of a molten carbonate fuel cell stack in the much-
applied ICI low-pressure methanol process. In this process, the reformer produces
synthesis gas that contains too much hydrogen for the methanol synthesis.
Normally, this excess hydrogen is purged from the methanol reactor loop. By the
integration of a MCFC-stack located directly after the primary reformer the syngas
composition can be corrected before it enters the reactor loop, prior to syngas
make-up compression (Farooque 1988). It was demonstrated that another option,
viz. to direct only part of the hydrogen to the fuel cell and to reroute the CO2-rich
anode off-gas to the process, provides sufficient leverage to plant performance to
make this application economically attractive (Dijkema et al. 1994b).

Fuel cell reactors

It is a well-known theorem in thermodynamics that, for a reaction that releases
chemical energy, the theoretical maximum of available chemical energy would be
converted to useful work in the case where the reaction is performed in a reversible
manner in a fuel cell device. Though a fuel cell can never operate completely
reversibly in practice, the essence of employing a fuel cell would be that in the fuel
cell the energy conversion to useful work is not limited by the well-known Carnot
limitation of a heat engine.
Ketelaar investigated the potential of combining the generation of electric power
and useful chemical products. In principle, any chemical reaction that runs
spontaneously, i.e. ∆Gr < 0, can be used for the generation of electric power
(Ketelaar 1968). He distinguished between electrochemical reactors, where the
chemical produced is the main product, and fuel cells, which main product is electric
power. In essence, he concludes that only industrial partial oxidation reactions are
candidates for electrochemical reactors.
Electrochemical reactors have long been applied on an industrial scale, notably for
the production of aluminium, zinc, steel, chlorine and adipic acid. Vayenas and co-
workers labelled this use of fuel cells “chemical cogeneration” (Vayenas 1988). We
prefer the term fuel cell reactors, to avoid confusion with the term “cogeneration”.
Fuel Cells and Trigeneration 163

Electrochemical reactors have long been applied on an industrial scale, notably for
the production of aluminium, zinc, steel, chlorine and adipic acid. According to
Yentekakis et al., however, the feasibility of electrochemical fuel cell reactors was
demonstrated for the first time in 1980, notably for the conversion of ammonia into
nitric oxide, one of the prime steps of nitric acid in fertiliser production, in a SOFC
(Vayenas et al. 1992). Other industrially important reactions include the oxidative
dehydrogenation of ethyl benzene to styrene, and of 1-butene to butadiene
respectively, the ammoxidation of methane to hydro cyanic acid, the partial
oxidation of methanol to formaldehyde, and the oxidation of H2S to SO2(Yentekakis
and Vayenas 1989). (Malhotra and Dattra 1996) mention tentative development of a
novel fuel cell for the trigeneration of electricity and acetaldehyde from ethanol.
The applications of the electrochemical reactors described above comprise
trigeneration of electricity and a useful chemical by controlled partial oxidation.
Synthesis gas-proper is another candidate product that can be produced by partial
oxidation. (Alqahtany et al. 1993) investigated the production of synthesis gas in a
solid electrolyte cell that incorporates electrochemical oxygen pumping (EOP). They
conclude that EOP allows control of CO-formation. Since H2 formation remains
largely unaffected, the H2 /CO ratio can be controlled by EOP. Finally, they suggest
to use their results for the development of an in-situ methane reforming SOFC,
which the authors claim to perform equally or better than existing industrial
methods under the co-production of electrical power.
Mazanec et al. were issued a patent on the use of certain materials in membrane
reactor or fuel cell arrangements for oxidation reactors. They list examples of
methane to synthesis gas or unsaturated compounds, the partial oxidation of ethane,
substitution of aromatic compounds, and, notably the extraction of oxygen from
oxygen-containing gases which they deem useful for exhaust gas cleanup (Mazanec
1993).

Conclusions from the literature review

Various examples of by-product hydrogen use in fuel cell systems have been
reported (see §0, p.160). In all of these cases, the valorisation of by-product
hydrogen in fuel cells competes with the simple use of hydrogen as co-fuel in
process furnaces or nearby power plants. In other cases the purification of the
hydrogen may be feasible. The recovered hydrogen then can be recycled to the plant
or be sold via pipeline or bottling station.
Process system concepts that integrate fuel cell modules in the chemical plant design
have been suggested for ammonia, methanol and pure nitrogen production (§0,
p.161). Generally speaking, fuel cell modules may be integrated in chemical plants
that include the generation of synthesis gas, pure hydrogen or pure CO streams. as
Since the economics favour chemicals production over electricity generation, system
energy efficiency improvement must compensate the production amount shifted
from chemical to electricity.
Many “proof-of-principle” of fuel cell reactors have been reported (see §Fuel cell
reactors, p.162. To date, however, no commercial application of fuel cell reactors is
known.
164 Process System Innovation by Design

5.2.5 IDENTIFICATION OF FUEL CELL APPLICATIONS IN THE CHEMICAL

INDUSTRY: THE IMCFC
As fuel cells and other energy conversion devices are expected to contribute to an
improvement in resource utilisation in the chemical industry, we developed a
systematic method to identify promising trigeneration options for fuel cells and gas
turbines (Dijkema et al. 1998).
The essence of the method is that the arrangement of fuel cell modules is
considered a multifunctional unit operation that can be used in any process system
design. In addition, it is considered which fuel cell system functions can be
performed by the process plant, which results in equipment sharing between the
production process or processes and the fuel cell system. The method will be
illustrated by the design of a trigeneration system for the iMCFC (Hemmes et al.
1998).

The iMCFC
During the initial testing of the iMCFC concept, the question emerged “Do market-
niches exist where only the iMCFC would fit, and what is the scope of the iMCFC
with respect to system improvement.”

recycle

CO / H2

Anode
CO2 / H2O / inert
Matrix
<CO, H2>
Cathode

N2 / O2 N2 / O2

recycle

Figure 5-7: iMCFC-module, 2 inputs - 2 outputs.

A trigeneration system was specifically designed to exploit the unique features of the
iMCFC, and to co produce electric power and synthesis gas. Other system designs
for the iMCFC are reported elsewhere (Dijkema 1998).
The main characteristic that distinguishes the iMCFC from a conventional MCFC is
the CO2 supply to the fuel cell matrix (Figure 5-7). No CO2 needs to be present in
the cathode feed. Hence the CO2 and oxygen supply to the fuel cell have been
decoupled, which offers an additional degree-of-freedom in system design: both the
CO2-supply and the oxygen supply can be designed for optimal cell performance and
cell cooling.
Fuel Cells and Trigeneration 165

Not only the cell’s performance improves, however, but also a 3 input - 3 output
fuel cell results. These unique features must be exploited in the design of a system to
fill some market-niche unique to the iMCFC. To be competitive, any system design
for the iMCFC must consist of a minimum amount of equipment, notably gas-gas
heat exchangers.

iMCFC sandwich system design

To introduce the iMCFC and to illustrate these concepts, considered the system design
in Figure 5-7. A recycle of anode off-gas to both anode inlet and matrix inlet is used,
whereby the number of both inputs and outputs is reduced to two again (Figure 5-7).
The required amount of additional equipment is limited. Such a module can
effectively be used in a stand-alone fuel cells system design replacing a conventional
MCFC stack, or in process-integrated applications.
In the system design that we have labelled the ‘sandwich system’, an iMCFC is
located between a reformer and partial oxidation (Figure 5-8). The core of the
Sandwich System comprises a gas heated reformer (GHR) with a product gas outlet
temperature that matches the MCFC inlet temperature, viz. 600-700 °C, the iMCFC,
and a partial oxidation reactor (POX) that converts anode and cathode off-gases of
the MCFC. The hot POX product gas provides the heat required to operate the
GHR. Both the iMCFC anode- and cathode off-gas are sent to the POX directly. A
system study has shown that the sandwich system is thermodynamically feasible
(Dijkema 1998).
As indicated by the functional scheme technique, a number of functions around the
iMCFC have been integrated into a single apparatus, viz. off gas processing and
syngas manufacture. The iMCFC, however, also requires additional system
functions, such as the supply of CO2 to the matrix (not shown in Figure 5-8). This
can be realised by a separate gas-loop, or by a bleed-stream from the anode off-gas
(Figure 5-7).

Off-gas cooling
Feed preparation Syngas production Fuel Cell Module Syngas
production
Feed
Fuel Anode
saturation
Gas-heated
reformer Matrix POX
Water
Cathode

Oxydant
preparation

Air Oxygen
plant
Syngas

Figure 5-8: The iMCFC sandwich system.

The particular combination of functions was derived from knowledge of the
thermodynamic equilibria in the reformer, that dictate that methane conversion is
166 Process System Innovation by Design

favoured by low operating pressure and high operating temperatures. Increasing the
pressure from 1 to 10 bar dramatically reduces the conversion of methane at
otherwise equal conditions. At 10 bar and iMCFC inlet temperature of 650 °C and a
Steam-to-carbon (SC) ratio of 2 the methane conversion is as low as 35%. As a
consequence, however, the heat load of the GHR is also reduced! The remainder of
the methane is converted in the partial oxidation reactor (POX), or autothermal
secondary reformer: oxidation of part of the gas drives the reforming of the
methane present. Thus, by sandwiching the iMCFC between GHR and POX heat-
load is shifted away from the GHR. Because the GHR involves physical heat-
transfer, and the POX ‘chemical’ energy transfer, this shift increases system
efficiency. The net result is that with the same amount of feedstock, synthesis gas
can be made and electricity produced. The iMCFC operation at an elevated pressure
is advantageous for the sandwich system design. An elevated pressure previously
was limited by cathode dissolution, but is no issue in the iMCFC because of the
decoupling of CO2 and oxygen supply to the cell. pCO2 can be kept at a very low
value (10-3 atm.) to reduce NiO dissolution.

Conclusion
The early identification method combined with the functional scheme technique has
proven to be useful to exploit fuel cell innovations by the early development of a
trigeneration system concept.

5.2.6 EARLY ECONOMIC EVALUATION OF FUEL CELL APPLICATIONS IN THE

CHEMICAL INDUSTRY
Technology selection for new power systems is done in the early phases of a project
on the basis of investment cost and margin expected. The prospects for the use of
stand-alone fuel cell systems therefore are not good because to date investment
costs for stand-alone fuel cell systems and gas turbine differ by an order of
magnitude (1996: ONSI-PAFC $6000/kW, GT cogeneration system $800/kW).
Lately, therefore, in the fuel cell community it has been realised that probably only
in niche markets where the unique features of fuel cells are exploited the costs
associated can be justified. Generally the cost-structure of a project is broken down
into capital related cost (capex), and operation-related cost (opex). The revenues
from the operation of the facility must compensate for both capex and opex. In
power generation projects the initial investment - cost per kW installed- largely
determines project profitability because it represents a large fixed annual cost
throughout the expected lifetime of the project. The electricity revenue must exceed
the total of the fixed annual investment cost, fuel cost and other operating cost in
order to realise some return-on-investment.
In capital-intensive industries such as the chemical industry and the power industry,
also a large part of the capex is the investment cost for the new facility. The exact
quantity of these costs, however, is hard to estimate. At the initial (screening) phases
of a project it is therefore accepted that the estimate still has wide margins. In the
chemical industry at the basis of design phase a +/- 40% margin in the investment
estimate is generally accepted (Chauvel 1981).
Fuel Cells and Trigeneration 167

Finally, it may be seen that in all the economic methods for early project evaluation
the problem is reduced to a closed problem by sufficiently defining the system. In
the early economic evaluation of fuel cell applications in the chemical industry in
general we are after some general insights into the attractiveness of fuel cell
technology in the chemical industry, an open problem where we attempt to explore
an endless solution-space to help define promising directions for R&D!

Development of a simple equation for fuel cell evaluation

Spillmann and co-workers developed an economic evaluation methodology that they
labelled differential economic analysis (Spillman et al. 1984) and applied it to fuel cell
reactors. In the differential analysis of chemical production, the reaction is
considered to be a source for additional revenue by the direct cogeneration of
electricity. The economic evaluation only involves the cost difference between
(chemical transformation + electricity generation) and chemical transformation
alone. The time-value of money is not taken into account.
Based on the basic notion of thermodynamics that the maximum amount of useful
work or electric power that can be extracted from a chemical reaction is equal to the
change in Gibb’s free energy ∆G, an indicator λth of fuel cell reactor attractiveness
was derived by Spillmann (1981).

λth = ε • ((-∆G0)/M ) • (re / (rp - rf)) (Eq. 5.1)

The equation for λth is an expression for the electricity revenue from the chemical
reaction divided by the value added of the chemical operation per se. Trigeneration
becomes more attractive in case λth increases relative to chemical transformation
alone. The indicator particularly quantifies the leverage of trigeneration; it does not,
however, yield a direct clue on the investment allowed for the changeover from
conventional chemical reaction to trigeneration. Spillmann concluded that
trigeneration in a fuel cell reactor would be attractive for reactions that are carried
out in industry on a bulk-scale, where the margin per unit product is small. He
argued that, in such cases, the additional generation of electricity could have a large
effect on the net operational margin.
Notably Vayenas et al. used and extended this method for the comparison of the
economics of a fuel cell reactor and the use of a conventional reactor. A simple
equation is derived that gives a first impression of the attractiveness of a fuel cell
reactors (Vayenas et al. 1991b; 1991a). This is summarised and expanded below.
According to Vayenas, the early economic analysis need only focus on the alternate
energy conversion because the chemical output of the two types of reactors is the
same. In the early economic analysis of a fuel cell reactor two systems are compared,
a convential reactor and the fuel cell reactor. These process the same feedstock to
the same product slate. In case the reaction is exothermic, in the conventional
reactor only heat is produced from the chemical reaction, while in an
electrochemical reactor both electric power and heat are produced. A fuel cell
reactor is then more profitable in case the following relationship is satisfied,

(Rf, e+ Rf,h) •θf - Rc,h•θc > If - Ic (Eq. 5.2)

168 Process System Innovation by Design

In words: the total additional revenue from trigeneration throughout the technical
lifespan of the fuel cell must exceed the difference in initial investment.

Evaluation of using by-product hydrogen

When we use the ‘fuel cell’ perspective, the use of by-product hydrogen eliminates
the need for a fuel processor, which may represent some 20-30% of system cost
which is the common argument to promote such an application. Differential
economic analysis helps to understand by-product hydrogen utilisation in the
chemical industry. It appears, however, that no equivalent single equation for by-
product use can be derived because there is more than one alternative to be taken
into account, for example
1. By-product hydrogen to-flare
2. Simple firing in furnace (heat value = lowest)
3. Utilise in gas turbine-based cogeneration system
4. Upgrade to electric power and heat in fuel cell
5. Recover and reprocess in plant
6. Recover and sell as product
We can safely assume, however, there is an economic incentive for the conversion
of hydrogen into electric power or heat rather than to waste it to a flare (option 1).
In addition, the use in a GT-based system is limited, as a GT can only accommodate
fuel gas with limited hydrogen content. Recovery and reprocess in the plant
apparently offers no economic incentive. Finally, option six probably in most cases
requires substantial investments and organisation to realise either delivery of pure
hydrogen to a pipeline grid, or delivery of bottled hydrogen.
A proper first indication of the merits of upgrading by-product hydrogen in a fuel
cell system therefore can be obtained by comparison of the options: recovery of the
hydrogen heat value in some furnace, and recovery of electric power and heat in a
fuel cell system.
The modified relationship (5.2) holds:

(Rf, e + Rf,h) •θf - Rc,h, (H2)•θc > (1-γ)•If - Ic (Eq. 5.3)

θc and Ic represent the lifespan and investment of furnace modifications for
hydrogen firing. Coefficient γ allows one to apply a correction for the investment in
the fuel cell system, e.g. to account for special provisions required to use by-product
hydrogen.
The actual incentive for the use of a fuel cell system to upgrade by-product
hydrogen in cases where all other options except simple firing are exhausted
depends on the magnitude of the investments avoided in furnace equipment, net
fuel cell investment, and the margin between electricity revenue and fuel value. The
equation shows that a trigeneration system may offer an advantage both because of
savings in plant investment and savings in fuel cell system investment.

Evaluation of process-integrated fuel cells

Process-integrated applications of fuel cells are expected to require only a fraction of
the cost of a stand-alone fuel cell system per kWh because fewer auxiliaries are
Fuel Cells and Trigeneration 169

required, in some cases not even a reformer. This type of fuel cell application can
provide an additional competitive edge over stand-alone fuel cell systems by
improving chemical plant efficiency and/or increasing plant capacity. In other
words, the fuel cell module incorporation results in a leverage effect on plant
performance. In the methanol plant, for example, the energy consumption of the
syngas compressors is significantly reduced, and plant capacity is increased by some
8%.
As we have shown in earlier work (Dijkema et al. 1994b; Dijkema et al. 1998), the
main advantage of a process-integrated fuel cell originates from the leverage effect
on chemical plant performance.
When we consider the function of hydrogen in chemical plants there are four
possibilities:
A) Hydrogen is a feedstock
B) Hydrogen is part of an (intermediate) feedstock in the plant, for example in
ammonia or methanol production via synthesis gas.
C) Hydrogen is a by-product of the chemical reactions only, for example in
dehydrogenation processes or dealkylation of toluene to benzene.
D) Hydrogen is a process agent (catalyst, diluent, promotor, reductor etc.)

But do all these options require their own modified equation (6.2)? In case A and D
it may be argued that there is no case for trigeneration using process-integrated fuel
cells: hydrogen is required for the chemical reactions themselves, any consumption
by the fuel cell would only add to the consumption of the plant. Note that in case D
usually some waste-hydrogen stream results, which then can be used in a fuel cell
system as has been analysed in the previous section. The same applies to case C.
In case B, however, an important incentive for trigeneration is the sharing of
facilities, which can offer advantages of economy-of-scale. In commercial plants,
however, hydrogen-production facilities often already are scaled to the maximum
technically feasible size. Only if there is a mismatch between maximum-size of
hydrogen generation and hydrogen consumption part of a chemical plant is there
scope for trigeneration, e.g. in ammonia production (Dijkema et al. 1996).
What remains is the incentive created by a leverage effect on plant performance, as
illustrated by the integration of a fuel cell stack in a methanol plant. In this case, (Eq.
5.2) can be extended with a term that quantifies this incentive, Rchem, trigen – Rchem, plant
The use of fuel cells to boost the performance of the plant at present appear the
most attractive market niche for fuel cells in the chemical industry. The exact
incentive, however, can only be estimated when performance data are available from
a preliminary process system design.

Evaluation of fuel cell reactors

As stated, the thermodynamically most attractive route for reactions where ∆Gr < 0
is a reversible fuel cell reactor. This implies, however, at a technical level a specific
(catalytic) fuel cell must replace each conventional reactor. But which reactions are
economically the most favourable?
170 Process System Innovation by Design

Because the investment cost normally quoted for fuel cells is in [$/kW], whilst that
for chemical reactors is quoted in [$/(ton/day)], we introduce α to relate the two:

α = ε • (∆Gr) • 4.1828 [kcal/mol] • [J/Cal] (Eq. 5.4)

/ (M•10-6 • 3600 • 24 [g/mol] • [ton/g] • [s/day]

α = 48.4 ε • (∆Gr) / M [kW/(ton/day)] (Eq. 5.5)

α is an expression for the electric power produced per ton capacity of a chemical
[kW/(ton/day)] as a function of fuel cell reactor electrical efficiency ε.

Ic = αIe,f (Eq. 5.6)

Equation (5.2) then becomes

(Rf, e + Rf,h) • θf - Rc,h • θc > α • Ie,f - Ic (Eq. 5.7)

For an initial estimate, the revenues can be calculated from the enthalpy and free
energy change of reaction at standard conditions:

Rc,h = β • rh • (-∆H0)/M (Eq. 5.8)

Rf, e = β • re • ε • (-∆G0)/M (Eq. 5.9)

Rf,h = β • rh • ((-∆H0)/M) - ε • ((-∆G0)/M)) (Eq. 5.10)

β is a unit conversion factor. Under the (optimistic) assumption that the lifetime of a
cogenerative fuel cell equals that of a conventional reactor, θf = θc, the total profit
difference V between the investment in the fuel cell or conventional reactor can be
calculated as

V = 24 α • (re – rh) • θf / (α • Ie,f - Ic) > 1 (Eq. 5.11)

Assuming a worst case, Ir = 0, i.e. there is no investment required for a conventional
reactor, this reduces to

V = 24 (re – rh) • θf / Ie,f > 1 (Eq. 5.12)

or Ie,f < 24 (re – rh) • θf (Eq. 5.13)

If, in addition to Vayenas, we assume that the revenue for heat equals its fuel
equivalent, whilst electricity represents double that value (as a worst case) the
inequality reduces to

Ie,f < 12 re • θf (Eq. 5.14)

Thus a simple equation has been derived to estimate the allowable additional
investment for fuel cell reactor. This implies, for example that at a technical lifespan
of 5 years, for example, and an electricity revenue of 5 ¢/kWh, the investment in the
fuel cell is allowed to be 1095 $/kW!
Fuel Cells and Trigeneration 171

Application to fuel cells in the chemical industry

A first selection of industrial reactions that appeared promising for trigeneration in a
fuel cell reactor was obtained by Ketelaar, who calculated its maximum electric
output at standard conditions, assuming a maximum thermodynamic efficiency of
such an electrochemical reactor of 70% (Ketelaar 1968). In essence, he concluded
that only industrial partial oxidation reactions display both a high open-cell voltage
and a significant electric power output per unit product. In addition, he expected
high investment costs for such reactors.
In the development of equation (5.14) the perspective of a fuel cell developer has
been used, and the result has been expressed in a [$/kW] figure. In order to
compare the fuel cell reactor with a conventional reactor, the above formulas must
be transformed into an expression in chemical product capacity. This is
straightforward using α.

If = α • Ie,f< 12 re • θf • α (Eq. 5.15)

The allowable investment in a fuel cell reactor is a linear function of electricity
revenue, fuel cell lifespan, and α. The equation represents a worst case because the
investment of a conventional reactor is assumed to be zero, and the electrical
efficiency only 50%.
In Table 1, heading Fuel Cell Reactor, both the input data for equation (5.14) are
listed, and the results, notably the calculated maximum allowable investment for the
fuel cell reactor, expressed as 1000$ per [ton/day] capacity. The results apply to a
fuel cell reactor with a minimum technical lifespan of 5 years, and an electricity
revenue 5 ¢/kWh.
Under the heading ‘Trigeneration system data’ these results have been related to data
that apply for chemical plants of typical commercial production capacity. Per
chemical reaction, the maximum allowable investment in fuel cell reactors has been
calculated by multiplication of Ic and C. In addition, the power rating P has been
calculated by multiplication of α and C. Finally, the maximum allowable investment
for a fuel cell reactor has been expressed as a percentage of typical plant investment.

Example evaluation
Consider, for example, the direct electrochemical partial oxidation of propylene to
propylene oxide, PO (Table 5.1, reaction 12). From the thermodynamic data in (Stull
et al. 1969), the Gibb’s free energy change of the reaction at standard conditions can
be calculated, -21.2 kCal/mole. Using equation (5.5) results in α = 8.8
kW/(ton/day).
The allowable investment at a technical lifespan of 5 years (5*365 days) and at an
electricity revenue of 5 ¢/kWh follows from equation (5.14). The allowable
investment is 10 k$ per (ton/day) of propylene oxide production capacity. By
division with α, we obtain the familiar result of 1095 [$/kW] allowed investment for
the fuel cell reactor.
Table 5.1: Fuel cell reactor evaluation for the chemical industry
Chemical Operation Fuel Cell Reactor Trigeneration system data
Max. Typical Fuel Cell Fuel Cell/ Typical ∆I as % of
Allowed plant Power Conventional Plant Inv. Plant Inv. I
Reactor Max ∆I
∆Gr M α If Co Rating
Reaction Kcal g/mole k$/ ton/day MW Million $ Million $ %
/mole (ton/day)
1 Hydrogen oxidation H2 ½ O2 → H2O -54.6 18 73.4 80 n.a. n.a. 0.0 0 0
2 Carbon partial oxidation C + ½ O2 → CO -32.8 28 28.3 31 n.a. n.a. 0.0 0 0
3 Carbon oxidation C + O2 → CO2 -94.3 44 51.8 57 n.a. n.a. 0.0 0 0
4 Syngas by partial oxidation CH4 + ½ O2 →CO + 2 H2 -20.7 30 16.7 18 2000 33.3 36.5 80 46
5 Syngas by reforming CH4 + H2O → CO + 3 H2 34.0 70 -11.7 -13 2000 -23.5 -25.7 0 0
6 Methanol by direct partial CH4 + ½ O2 → CH3OH -26.7 32 20.2 22 1800 36.3 39.7 130 31
oxidation
7 Formaldehyde CH4 + O2 → CH2O + H2O -14.1 30 11.4 12 285 3.2 3.6 6 59
8 Acetaldehyde C2H5OH + O2 → C2H4O H2O -46.3 44 25.4 28 285 7.2 7.9 19 42
9 Acetic acid via CH3OH + CO → CH3COOH -18.4 44 10.1 11 230 2.3 2.5 28 9
carbonylation
10 Ethyleneoxide C2H4 + ½ O2 → C2H4O -19.4 44 10.7 12 400 4.3 4.7 58 8
11 Ethyleneglycol C2H4O + H2O → C2H5(OH)2 -34.4 62 13.4 15 285 3.8 4.2 12 35
12 Propyleneoxide C3H6 + ½ O2 → C3H6O -21.2 58 8.8 10 285 2.5 2.8 102 3
13 Butadiene C4H8 + ½ O2 → C4H6 + H2O -34.4 54 15.4 17 140 2.2 2.4 26 9
14 Phenol by toluene C6H5(CH3) + 2O2 → C6H5OH + CO2 + -131.3 78 40.7 45 140 5.7 6.2 39 16
oxidation H2O
15 Styrene by C6H5(C2H5) + ½O2 → C6H5(C2H3) + -34.8 104 8.1 9 1000 8.1 8.8 52 17
oxydehydrogenation H2O
16 TPA C6H4(CH3)2 + 3O2 → C6H4(COOH)2 + -279.6 166 40.7 45 285 11.6 12.7 85 15
2H2O
17 Nitrous oxide NH3 + 1 ¼ O2 → NO + 1½ H2O 20.7 30 0.0 0 0 0.0 0.0 0 0
Fuel Cells and Trigeneration 173

A common capacity for a commercial industrial facility is some 285 ton per day PO,
or 100.000 ton per annum (Chauvel and Lefebvre 1989b), which implies that the
total allowable investment for the fuel cell reactor amounts to 2.85 Million dollars
(assumptions: investment conventional PO reactor zero; electricity revenue 5
¢/kWh, minimum lifespan fuel cell 5 years).
Any company involved in PO-production can easily insert the correct figures for
refinement of the result, for example:
◊ Actual feasible single train conventional reactor capacity must be used
◊ A portion of conventional reactor investment can be added
◊ Fuel cell lifespan can be corrected to compare with conventional catalyst
replacement cost and frequency

Evaluation of results
The calculation was completed for a number of important partial oxidation reactions
as they are carried out in the petrochemical industry today (Chauvel and Lefebvre
1989b). In addition, results are listed for candidate reactions to replace presently
employed reactions, for example methanol by direct partial oxidation.
A high value of α indicates that along with chemical production a high amount of
electric power can be potentially extracted. The combination with economic figures
allows additional assessment of the fuel cell reactor’s attractiveness: the higher the
maximum allowable investment, the higher the possible benefits. In the table, not
only reactions are included that result in some useful product, but also the
conventional fuel cell reaction, the oxidation of hydrogen appear favourable, as
shown by their high values of α. The following observations can be made by
inspection of these results:
1. The conversion of light molecules leads to high values of α.
2. CO2 formation in partial oxidation yields high α’s
3. The power rating of a trigeneration fuel cell reactor calculated, 6-40MW is
reasonable for world scale, chemical plants.
In addition, the allowable investment calculated for the fuel cell reactors offers
substantial scope for development of the technique for a number of reactions,
notably the partial oxidation of toluene to phenol. The electrochemical trigeneration
of formaldehyde by partial oxidation of methane appears attractive because of the
allowable investment represents a high portion of the plant’s total investment.
Since partial oxidation reactions involve the use of oxygen, the production of pure
oxygen is avoided in the case of employing a fuel cell reactor that uses air. This must
be taken into account, e.g. by adding an additional term in equation 5.2, (Rc,O2 –
Rr,O2), where R is negative, i.e. a cost.
The results also support the current focus on cogeneration fuel cell reactors, as the
figures for hydrogen and CO oxidation are most favourable. These are precisely the
overall reactions carried out in currently developed fuel cells. One could think of
selling pure water or CO2 as a trigeneration product of the fuel cell reactor. Note
174 Process System Innovation by Design

that the equation developed remains valid because of the assumed zero-cost of the
chemical reactor.
Please note that in the calculation of all these results, it was assumed that the cost of
a conventional reactor equals zero, i.e. economy-of-scale does not play a role!

5.2.7 CONCLUSIONS
The trigeneration concept is a suitable basis for a systematic framework to evaluate
fuel cell application in the chemical industry. The development of a trigeneration
system for a novel fuel cell type, the iMCFC has demonstrated the value of system
design in the earliest phase of R&D.
The functional scheme technique developed has proven to be valuable in the
identification and development of novel fuel cell systems. It is expected that it will
also be useful in the development of process systems, and that it may serve as a
communication framework for system development and system integration.
We expected that only qualitative early economic assessment of fuel cell use in the
chemical industry would be possible, and that the analysis would confirm the meagre
prospects of fuel cell use in the chemical industry. On the contrary, however, a
simple criterion could be derived for the evaluation of fuel cell reactors as the heart
of a chemical plant. The results show that the prospects for fuel cell reactors as the
heart of trigeneration systems are worthwhile exploring. Trigeneration yields
additional profits even when substantial investment is required, the allowable
investment in fuel cell reactors calculated is in the range of 15-60% of total plant
investment.
The use of by-product hydrogen has been shown to be only a last resort, when all
other options are exhausted. Finally, process-integrated use of fuel cells appears to
be most attractive when some leverage effect on plant performance or cost can be
realized.
Fuel Cells and Trigeneration 175

5.3 The Development of Trigeneration System Concepts 48

5.3.1 ABSTRACT
Trigeneration is the simultaneous production of chemicals, heat, and power. A
procedure is presented and illustrated for the development of trigeneration system
concepts from the industry level down to the level of individual plants and parts
thereof. It is shown for trigenerate petrochemical complexes that the selection of
system elements for minimal fuel consumption depends on the amount of
trigenerate electric power that can be exported. The avoidance of some 190 kg CO2
per MWh can be achieved for a typical complex. This net reduction in the total CO2
emission related to chemicals, heat and power production can be achieved because
the chemical industry acts as a heat sink. Finally, the realisation of trigeneration in
the chemical industry involves a shift in perception of the industry: rather than
considering it a threat because of pollution, safety and health risks, it must be
considered an enabler of a reduction of total human CO2 emission.

5.3.2 INTRODUCTION
Trigeneration systems are facilities for the simultaneous production of useful
(chemical) products, electricity, and heat. One may argue that today such facilities do
not exist, but actually, they do! Today, for example, national economies or even the
EU are systems where a net amount of chemicals, electricity and heat are produced.
At numerous petrochemical sites the initial application of cogeneration systems has
led to the development of the on-site utility island concept.
Current petrochemical systems, however, have not been 'designed' for trigeneration.
In addition, these systems do not operate somewhere close to the maximum
thermodynamic efficiency achievable. In accordance with the Second Law of
Thermodynamics, in each process the total entropy of a system and its environment
increases -some minimum amount of Exergy is lost- and a net amount of feedstock,
fuel or heat is expended to provide a process' driving-force. Since the economic
operation of a great many petrochemical processes requires operation at elevated
temperature and pressure, the total exergy loss increases. Thus, whilst a large part of
the chemical reactions completed in industry is exothermic and the net change in
free energy could allow production of Work, a net input of heat and Work is
required. Trigeneration offers a means to minimise the combined inputs for
electricity, chemicals and heat production, albeit at the expense of additional
investment.
Today, the threat of global warming has led many people to believe that some
reduction of human CO2 emissions into the global atmosphere is urgently required.
Therefore we explored whether the adoption of trigeneration in the petrochemical
industry can help reduce total human CO2 emissions by improving total fossil
resource utilisation.

48 The contents of this section have been published as (Dijkema 2001b).

176 Process System Innovation by Design

Outline of section
In the past edition of the World congress, we presented a paper on the integration of
energy-conversion devices in chemical plants (Dijkema 1996). This was largely
'technology-oriented' and focused on single chemical plants. Since then we have
been elaborating the concept of 'trigeneration' and developed functional modelling
(Dijkema and Reuter 1999; Dijkema and Grievink 2000) to help identify system
innovation options. This method was successfully applied to develop and evaluate
trigeneration system concepts for a range of chemical processes and a number of
energy technologies, notably fuel cells (Hemmes and Dijkema 1998; Dijkema 1999),
gas turbines and cogeneration systems.
In this chapter, we will address the development of trigeneration system concepts
for the level of industrial complex to individual plant level. Firstly, some background,
the problem addressed and the definition of trigeneration will be elaborated, and the
research approach summarised. Subsequently, functional modelling will be
introduced to help understand and develop trigeneration concepts. Third,
trigeneration as 'leitmotiv' for an industrial complex or area is investigated from an
ecological and economic perspective, and the potential of CO2 emission reduction
will be analysed. The prospects of trigeneration in the European Electricity markets
are briefly discussed. Finally, some conclusions are presented.

5.3.3 TOWARDS TRIGENERATION CONCEPTS?

Background & context of the work

Elsewhere we have reported some of our work on the development of methods to
support the conceptual development of process system innovations (Dijkema et al. 2001).
These are defined as technological innovations that enable or rely upon changes in
the system structure or design of the petrochemical industry, its industrial
complexes, or individual plants. It may be seen that trigeneration represents a class
of such process system innovations.
According to Wei, “(… ) innovative design and clever process modifications are
essential to the economic health of the producer of commodity chemicals” (Wei et
al. 1979). Indeed, the petrochemical industry has an impressive record of
technological innovation within single plants. However, in contrast with science-
based innovation within chemical plants, the concept of trigeneration suggests
innovation beyond the boundary of individual plants, complexes or even the
industry. Thus trigeneration is a concept to consider while exploring integrated
conceptual design of the petrochemical industry's systems, which to date represents a
largely unexplored 'white spot' in R&D (Dijkema et al. 2001).

Definition
We have defined trigeneration systems as facilities for the simultaneous production
of useful (chemical) products, electricity and heat (Dijkema 1996). Notably in utility
engineering, sometimes the term trigeneration is used for the simultaneous
generation of electric power, heat and cold. In chemical engineering and R&D, the
term chemical cogeneration has been used to describe systems that produce electric
power and chemical products, notably fuel cell reactors (Vayenas 1988).
Fuel Cells and Trigeneration 177

In our view, however, trigeneration best coins the view of any production plant as a
possible supplier of not only products, but also energy products viz. electric power
and heat.

Research approach
We conjectured that the development of trigeneration may contribute to achieve
sustainability, notably by avoidance of CO2 emissions and by increasing total
resource utilisation. Our first objective was to apply the systemic approach
developed (Dijkema 1996) for identification of trigeneration options and the
specification of desired system content. Our second objective was to demonstrate
that options for system improvement in the petrochemical industry exist beyond the
straightforward application of cogeneration.
Thus the research questions addressed are (1) whether this method is applicable and
(2) whether trigeneration is worthwhile with respect to sustainability not only
ecologically but also economically.
We used functional modelling to develop trigeneration concepts. A straightforward
LP-model was constructed to demonstrate the effect of trigeneration on the
selection of energy conversion facilities. The model was also used to assess the
extent of CO2 reduction when effecting trigeneration in an industrial complex. Early
economic evaluation was used to obtain a first impression of the economics
involved.

5.3.4 TRIGENERATION SYSTEMS AND THE CHEMICAL INDUSTRY

Classes of trigeneration systems

The combination of chemical reaction and energy conversion is fundamental to
chemical engineering and crucial to achieving proper plant design. Surprisingly,
however, trigeneration is not, and only has emerged recently under the label of
'multifunctional reactors'. Notably at the level of unit operations fuel cell reactors
include some ceramic or polymeric membrane to effect a chemical reaction and
separation simultaneously. Trigeneration, however, can also be effected by
combining systems beyond this level. This is illustrated by presenting some classes
of trigeneration systems in the petrochemical industry below.
1. The Petrochemical industry
At this level, the industry can be considered a trigeneration system for the
production of base-chemicals, base-load electric power, and heat at various
temperature levels. All other options mentioned in this list are in fact
technology/system specifications that 'bring-into-being' a trigenerate
petrochemical industry!
2. Petrochemical cluster
At the level of a geographically clustered number of petrochemical complexes or
plant, cogeneration facilities can be employed to supply power and heat both to
industry and society, employing the industry as a sink for high-level heat (power
production), and a source for low-level heat. In the Rotterdam Cluster, for
example, a number of joint Cogeneration facilities have been or are under
178 Process System Innovation by Design

construction that supply electric power and steam to a number of customers in

the cluster, whilst exporting electric power to the grid.
3. Petrochemical complexes
For many petrochemical complexes -single site or part thereof-, the combination
of the so-called utility-islands and petrochemical plants can be considered
trigeneration systems in cases where they are net exporters of electric power and
heat.
4. Petrochemical plants
At this level, a number of options exist. Firstly, cogeneration systems can be
used to upgrade a chemical plant’s off-gases; the system is extended and/or the
system boundary shifted. Secondly, within the existing system boundary of a
petrochemical plant, energy conversion technologies can be employed as unit
operations in chemical process system design. Thirdly, chemical reactors can be
designed for optimal product-to-heat ratios for inclusion in traditional power
cycles: the heat of reaction drives some cycle for heat-to-power conversion.
5. Functional elements
New functional elements for use in chemical plant design are crucial for
achieving optimal material and energy balances in trigeneration. At this level,
some options must be developed to (a) employ the petrochemical industry as a
heat-sink for power generation, (b) create leverage by creating novel functional
units to enable trigeneration and improve plant performance simultaneously, (c)
allow a shift in the balance between power and chemical production costs and
revenues and (d) directly convert feedstock to chemical products whilst
capturing part of the free energy change of reaction as electric power, thereby
approaching the thermodynamic optimum.

5.3.5 DEVELOPMENT OF TRIGENERATION CONCEPTS

Functional modelling
Applied thermodynamics teaches us that trigenerate conversions must be direct, in
order to obtain the highest efficiency, and traditionally an ideal fuel cell reactor is
assumed to prove the case (e.g. Denbigh 1956)). At the other end of the spectrum is
isolated or stand-alone conversion of fossil feedstock into chemicals, electric power
or heat respectively. Present petrochemical systems are somewhere in-between,
largely through the longstanding tradition of process- and energy-integration, while
the introduction and spread of the use of cogeneration also has improved its fuel
utilisation. This use of cogeneration can be considered a form of trigeneration.
In addressing the analysis and formulation of trigeneration system concepts we have
used functional modelling. The central idea of functional modelling is to enable
'opening-up' established or proven systems or technology concepts. System
engineering has been defined as “a progressive definition and integration of functions and
design decisions at every level of system complexity” (Corrigan and Kaufman 1965).
Systems can be decomposed into progressively smaller subsystems, where system
elements are the lowest level of detail of a system structure. System complexity can
now be dealt with effectively when the focus is on system functions rather than
system content because the functional characteristics of a system element are
Fuel Cells and Trigeneration 179

technology-free (Asbjørnsen 1992; Dijkema and Reuter 1999). In other words, a system
or system element can be described completely by its inputs and outputs and the
functions it performs. Specification of its technological content is not required;
rather the functional description is complete and independent from the actual
technical realisation of the element. For complex systems, functional modelling
offers a means to arrive at some functional decomposition (Dijkema and Grievink
2000). It may be seen that a technology-free description enabled by functional
modelling opens the way to search for alternatives in trigeneration system design.

Functional scheme technique

A power system's functional decomposition has been visualised in the top left
diagram in Figure 5-4: System development – fuel to electric power (redrawn), page
157, where the functional system elements have been enclosed by grey boxes. Each
box title represents an element’s function. The representation essentially is
technology-free when the system has been properly decomposed: note that the
drawing is still valid when no technological details are included in the boxes.
At the level of single elements, the content depicted represents only one possibility
that 'realises' the system elements. At the system level, the particular selection and
arrangement of system elements also represents a single possibility to ‘realise’, for
example, a fuel cell system. At this level, the system inputs and outputs, its boundary
and its title describe it as a system element at the aggregation level of 'complexes' or
clusters. Note that all feed and product streams have been labelled outside the
system boundary.
A third important feature of this drawing technique is that system elements, where
multiple, physically unseparated functions are performed, must be enclosed by an
equal number of system element-boundaries. Thus alternative arrangements for the
design of the system offered by the integration of functions are visualised.

Formulation of system concepts

Let us assume that we must develop a system concept for the use of natural gas for
the generation of electric power (Figure 5-4). When we consider a conventional
large-scale power plant, we see that the system’s inputs are natural gas and oxygen
(in air); its outputs are electric power, flue gas exhaust, and a waste-heat stream
(from 'cold' to 'hot').
As stated, the central idea of functional modelling is to enable 'opening-up'
established or proven systems or technology concepts. When we leave out the
technical details of the system, as indicated by arrow I, it may be seen that the system
representation has become technology free.
The next step in system design is to consider other options to bring about the
required functions, in this case power generation from natural gas. The technology-
free system representation opens the way to alternative realisations depicted by
arrow IIa and IIb in Figure 5-4, the use of a gas turbine and a fuel cell respectively.
When focusing on the functional description, rather general descriptions for the
inputs and outputs suffice initially. More detail can be added once a system concept
has been established.
180 Process System Innovation by Design

In the case of the gas turbine, for example, flue gas is produced that contains a
significant amount of waste heat at a temperature of some 450-500 oC, instead of
transferring heat to some cold stream - e.g. cooling water. In the case of the fuel cell,
apart from energy the flue gas will contain unconverted feedstock and some cell
cooling may be required. For the sake of simplicity, in all cases we have assumed that
the feed is at the right conditions (Pressure, Temperature, composition).
We suggested that in process system design at each level of system complexity, the
same procedure can be applied: (Dijkema 1999). Trigeneration system development,
however, implies that there must be an initial focus on what possible alternative
realisations of a particular system (step 4 and 5 in the procedure given below) may
exist involving trigenerate functions. Thus, such a procedure must be extended with
(1) determination of the aggregate system level used to start the analysis, (2)
assessment of the (possible) system environment and (3) selection of the most
appropriate system boundary and system products.
In marketing studies, the latter is certainly done for a plant's main product, however,
a mechanism for identifying and evaluating other possible products should be
included. In trigeneration in the petrochemical industry, for example, CO2-credits
may be obtained that can be marketed under an Emission-trading system.
A procedure for trigeneration system development thus may look something like:
I. Review an existing system design by functional modelling:
1. Determine the aggregate system level
2. Identify the crucial system functions
3. Draw a functional scheme
4. Briefly describe each system element
II. Explore the options for trigeneration by
5. Assessment of the (possible) system environment
6. Addition of system outputs
7. Modifying the system boundary
8. Addition of system elements
9. Alternative realisations per system element or combination thereof
III. Specify and select system concept(s) for early evaluation

5.3.6 ILLUSTRATION: TRIGENERATE PETROCHEMICAL COMPLEXES

From the above classification it may be seen that in the chemical industry ample
opportunities exist for trigeneration. Elsewhere we have used the trigeneration
system concept to present prospective fuel cell applications in the chemical industry
under a single systematic framework @(Dijkema 1999) and the previous section of
this thesis.
In petrochemical plants, heat is traditionally supplied in industrial furnaces and
reaction furnaces such as the naphtha-cracker- and the ethylene dichloride (EDC)
cracking furnaces. A study of the heat demand of the Dutch industry (Boot and van
Wees 1982) concluded that the chemical industry largely requires heat at a
temperature level between 300 and 800 oC at a total fossil fuel consumption of 164
PJ for energy supply (excluding naphtha cracking). These results led them to the
remark that ample opportunity existed for cogeneration in this industry. Indeed most
Fuel Cells and Trigeneration 181

petrochemical sites in the Netherlands either already included cogeneration at the

time or such facilities have been erected since. Often, however, due to lack of a heat-
demand from residential areas these appear to have been designed on the basis of
industrial - process - heat demand with the surplus power being transported to the
grid. With the liberalisation of the European power markets, however, the dynamics
of this market has increased, and, for any new cogeneration project in the Dutch
petrochemical industry, the long-term marketability of yet more base-load must be
thoroughly assessed. Our objective, however, is to support identification of system
innovations. Therefore, in this section, as an example, the procedure outlined above
will be used.

Use of the procedure for development of trigeneration system

I. Review an existing system design by functional modelling:
This part of the procedure for trigeneration system development comprises three
steps, viz. (1) Determine the aggregate system level, (2) Identify the crucial system
functions, (3) Draw a functional scheme and (4) Briefly describe each system
element
Step 1: The aggregate level is that of individual plants, or plant sections.
Step 2: The crucial system functions are the production of various chemicals,
electric power, and heat at various temperature levels. The power has 'base-
load' characteristics, and has limited technical and economic scope for
'turn-down'
Step 3: The scheme is similar to the ones given in Figure 5-4
Step 4: At this aggregate level, system elements are chemical plants (inside battery-
limits), power plants and heat conversion plants. The latter energy
conversions can be integrated sections of a single power plant, energy
complex or petrochemical complex.

II. Explore the options for trigeneration

The following steps must be completed to ensure some systematic exploration of
possible options:
1. Assessment of the (possible) system environment
2. Addition of system outputs
3. Modification of the system boundary
4. Addition of system elements
5. Alternative realisations per system element or combination thereof

In addition to the global chemical and polymer industry that largely represent the
markets for petrochemicals, the system environment assessed for the case of
trigeneration must include the electric power consumers or traders, and possible
consumers of heat. Apart from surplus electric power and heat, system outputs to be
considered are tradable CO2 emission rights. The system boundary of the petrochemical
complex changes because facilities for the generation of power and heat are
included. These include stand-alone power and heat generation respectively,
cogeneration options and others. At the single plant level, subsystem boundaries
182 Process System Innovation by Design

may shift, e.g. because of the integration of gas turbines in furnace design for
optimal preheat and power generation. Finally, gas turbines, steam turbines, fuel cells
etc. represent possible added system elements and these may offer new options for
realisation or combination of system elements.

III. Selection of system concept for early evaluation

In this particular case, our objective was to evaluate the impact of converting a
petrochemical complex to a trigenerate petrochemical complex.
Thus we compared two situations: (1) a petrochemical complex which power and
heat demand are met internally, whilst external power demand is catered for by
stand-alone generation and (2) a trigenerate complex which must cater for both
internal and external power demand.
To allow for this comparison, in the elementary evaluation presented we have added
to a hypothetical petrochemical complex the possibility of inclusion of
- stand-alone power generation (Power plant)
- cogeneration type 3, heat-to-power ratio 3:1 (Cogen-3)
- cogeneration type 4, heat-to-power ratio 4:1 (Cogen-4)
- steam turbine operating between 600 oC and 200 oC (StmTurbine)
- steam generator, 600 oC (StmGen-1)
- steam generator, 200 oC (StmGen-2)
These are common options available to meet the utility demand of any industrial
complex. The particular selection is not exhaustive, but provides a minimum set of
typical alternatives that are currently employed throughout the chemical industry.
Any set of options can be selected to meet the particular heat and power demand of
an industrial complex. Thereby a first impression of the possible selection of system
elements for a trigenerate industrial complex is obtained. It may be seen that this list
can be expanded for a more detailed analysis (e.g. more temperature levels) as the
need arises.
Notably the 600 oC level produced by cogen-3 or cogen-4 can also be perceived as
the temperature level of the heat sink presented by industrial furnaces or industrial
reaction furnaces. Thus, the analysis also leads to insights about the merits of
trigenerate petrochemical plants, by coupling of gas turbines to industrial furnaces.

LP-modelling for system element selection

A Linear Programming or LP-model was constructed to select the optimal
combination of system elements for heat and power generation. This was
implemented in Microsoft Excel. The optimal solutions were calculated using the
Excel Solver functionality.
The model (after Stadtherr and Rudd 1976; 1978)
Acc + r - p = 0 Eq. 5.16
The elements of matrix Ac are the coefficients that define each input-output relation
per system element.
c is the vector of capacities; c ≥ 0;
Fuel Cells and Trigeneration 183

r is the vector of feedstocks; r ≤ s(upply);

p is the vector of products; p ≥ d(emand);
An LP-model in standard form (Ax = 0) is obtained by using
A = [Ac I - I] Eq. 5.17

x = [c r p ]T Eq. 5.18
The optimal network design variables x0 are found by optimisation.
minx fT x subject to Ax = 0 Eq. 5.19
f is a vector of some objective function’s coefficients

The coefficients of the A-matrix listed in Table 5.2 reflect the performance or
efficiency of 'state-of-the-art' technology. These have been expressed in megawatts
(MW). A negative sign indicates an input, a positive sign an output. To avoid to
optimistic results with respect to trigeneration, the efficiency of stand-alone power
generation has been set at 50%. In this case, f is set such that r, which represents
total fuel input, is minimised. Thus, under the assumption that only fossil fuel
combustion drives these energy conversions in the installations listed, the net CO2
emission will also be minimised in this optimisation. Obviously, the actual CO2
emission depends on the fuel-mix employed. A first indication is obtained in all
cases by considering natural gas (CH4) as the fuel of choice.
Table 5.2: Coefficients used in the A-matrix. Includes net loss per element (MW)
System elements
Energy carriers Cogen-3 Cogen-4 Power StmGen-1 StmGen-2 Stmtrbine
Fuel -1 -1 -1 -1 -1 0
Electricity 0.24 0.19 0.5 -0.02 -0.02 0.14
Heat - 600 oC 0.71 0.76 0 0.85 0 -1
Heat - 200oC 0 0 0 0 0.85 0.76

Loss 0.05 0.05 0.5 0.17 0.17 0.1

LP model Results
In Figure 5-9 the results of the optimisation are given for the case where there is a
net demand in the Petrochemical complex of heat at 600 oC and 200 oC respectively
in a fixed ratio of 4:8:E (E=electric power) [MW].
From these figures it may be seen that:
- when export of electricity is not allowed, Cogen-4 caters for internal electricity
demand (0.5 MW), and there is stand-alone generation of 600 oC (HT) and 200
o
C (LT) steam
- when electricity export is allowed but limited, first Cogen-4 capacity is expanded,
subsequently Steam Turbine capacity is added
- allowing more electricity export brings Cogen-3 into being, thereby providing the
optimal match between electric power and heat required
184 Process System Innovation by Design

Above an export demand of 6 MW (per 12 MW of heat HT:LT = 4:8), only Cogen-

3, a steam turbine and Stand-alone Power Plant capacity are selected for optimal
trigeneration. In other words, anywhere in the region of 0 to 6 MW of electricity
export, trigeneration is preferred over stand-alone power generation.
The analysis clearly shows that the optimal selection of system elements picked for
minimum fuel consumption varies with the amount of electric power exported from
an industrial complex while satisfying its internal heat demand. The minimum fuel
consumption calculated for trigeneration demonstrates the possibility of using an
industrial complex as a heat sink for electric power production.

Total Power Generation [MW]

12

10

8
StmTurbine
6 Power Plant
4 Cogen-4
Cogen-3
2

0
0 1 2 3 4 5 6 7 8 9 10 11 12
-2
Total Electric Power Export Allowed [MW]

Fuel Consumption [MW]

35
30
25 StmGen2
20 StmGen1
15 Power Plant
Cogen-4
10
Cogen-3
5
0
-5 0 1 2 3 4 5 6 7 8 9 10 11 12

Total Electric Power Export Allowed (MW)

Figure 5-9: System element selection for minimal fuel consumption.

Fuel Cells and Trigeneration 185

5.3.7 EARLY ECONOMIC EVALUATION OF TRIGENERATION SYSTEMS

Generally the cost-structure of a project is broken down into capital related cost or
capital expenditure, and operation-related cost. Throughout the life cycle of the
project, the revenues obtained from the operation of the facility must compensate
for both types of expenditure. In addition, many companies require a minimum
return-on-investment or maximum payback period for an investment to be accepted.
In power generation projects the initial investment - cost per kW installed- largely
determines project profitability because it represents a large fixed annual cost
throughout the expected lifetime of the project. The electricity revenue must exceed
the total of the fixed annual investment cost, fuel cost and other operating cost in
order to realise some return-on-investment. Technology selection for new power
systems is done in the early phases of a project on the basis of investment cost and
margin expected.
Also in capital-intensive industries such as the chemical industry and the power
industry, a large part of the capital expenditure is the investment costs for the new
facility. The exact quantity of these costs, however, is hard to estimate. At the initial
(screening) phases of a project it is therefore accepted that the estimate still has wide
margins. In the chemical industry, a +/- 40% margin in the investment estimate is
generally accepted as the basis for the design phase (Chauvel 1981).
By analogy with (Vayenas et al. 1991b), our early economic analysis only focuses on
the alternate energy conversion because the chemical output of the systems is the
same i.e. the trigenerate petrochemical complex must process the same feedstock to
the same petrochemical product state. The trigenerate system is then more profitable
when the following relationship is satisfied:

(Rt, e+ Rt,h) • θt - Rs,e • θe > It - Is Eq. 5.20

In words: the total (discounted) additional revenue flux R from the trigenerate
petrochemical complex throughout its technical lifespan θ must exceed the
difference in initial investment I for trigeneration and stand-alone power generation.
The initial investment for the trigenerate petrochemical complex may include both
energy-conversion equipment and cost of adaptation of petrochemical installations.

Minimisation of CO2 emission

The analysis results demonstrate that in trigeneration more efficient energy
conversion can be selected to meet total power and heat demand. As a consequence,
transformation to a trigenerate petrochemical complex leads to avoidance of CO2
emission. This may become an additional source of income under some scheme for
emission trading. In table 2 the amount of CO2 avoided is listed per extra megawatt
(MW) of trigenerate electricity exported on top of internally produced 8:4 LT:HT
heat when optimal selection of system elements for trigeneration was allowed
(Figure 5-9). The exact figures are valid assuming all fuel is natural gas (methane)
with a LHV of 50 [MJ/kg], which represents a 'worst case' with respect to the fossil
fuel mix because CO2 emissions of oil product or coal exceed those of natural gas.
These figures confirm that whenever a combination of power and heat production is
186 Process System Innovation by Design

feasible, total CO2 emissions are reduced, which provides a clear incentive to convert
chemical complexes to trigenerate chemical complexes.

Transition of electricity markets

In trigeneration system development, one is faced with uncertainties with respect to
initial capital expenditure, whilst the margins to be expected depend on both
electricity, heat and chemicals revenue. It must be considered, however, that
trigeneration is largely an option for long-term 'base-load' because the utilisation
level of facilities must usually be above 85% in the petrochemical industry to ensure
some positive margin.
Until recently, in Europe electricity production companies were state-owned and
total capacity was centrally planned per country. In the Netherlands, for example, the
parent company of all production companies, SEP was obliged to issue electricity
plans on a regular basis. Currently, however, the sector is in transition to allow 'more
market' by privatising, liberalising and re-regulating the sector or parts thereof.
Trigeneration then represents only one of the innovation options for the new
electricity companies. In the market, a price will be set for the 'base-load' electric
power from the trigenerate petrochemical industry. This price may increase when
trigenerate electricity is labelled 'green electricity', a product that presently is in short
supply in the Netherlands, whereas of July 1st 2001 every consumer is free to select
its green electricity supplier. Finally, the net amount of CO2 emission avoided by
adoption of trigeneration represents an economic output when a system of tradable
CO2 emission rights is established.
Table 5.3: Avoided amount of CO2 [kg/MWh] per interval of extra trigenerated electricity
exported [MW]

Trigenerate Export CO2 Avoided

[MW] [kg/MWh]
1st 276
2nd 191
3rd 191
4th 191
5th 191
6th 190
7th 73
8 - 12th
th
0

5.3.8 CONCLUSIONS
By introducing functional modelling, the options for the simultaneous production of
chemicals, power and heat can be explored at each aggregate level by translating
existing system concepts to a technology-free specification. Thereby, an opening is
created towards development of innovative concepts. A procedure to systematically
complete this task was formulated and illustrated for the case of trigenerate
petrochemical complexes.
Fuel Cells and Trigeneration 187

By employing a straightforward LP-model, we demonstrated that the selection of

trigeneration system elements for minimal fuel consumption depends on the ratio
between the heat sink available and the amount of power that can be exported from
the system. For an industrial complex with a typical heat ratio for low level to high-
level heat of 4:8, the first 6 MW electricity exported per 12MW heat (LT+HT)
results in avoidance of some 190 kg CO2/MWh. Thus, parts of the large-scale
petrochemical industry can act as a large heat sink enabling a net global CO2
emission reduction by trigeneration.
These results support the proposed paradigm-shift in the perception of the
petrochemical industry. Rather than considering the industry a possible and
threatening source of pollution, safety and health risks, it can be an enabler for
reduction in global CO2 emission by increased fossil resource utilisation. This can
help to effect a change in perception of (petro)chemical plants and/or complexes by
all its stakeholders in society, and may help to ensure the industry's license-to-
operate.

5.4 Discussion and conclusion

As stated in the introduction, the crux of this chapter is the recognition that it is the
combination of fuel cell functions that must be exploited in developing process
system innovations. In addition, the system boundaries can be suggested but altered
to see the effect.
As a matter of course, both the choice of functions to describe a system and the
selection of the system boundary rely on the experience of the engineer or a
development team. Functional modelling, however, provides a codification of that
very experience and a route towards a first 'primitive' specification of a process
system innovation such as trigeneration.
The subsequent early economic analysis of fuel cell reactors and the effects of
trigeneration gives rise to consequent innovations once the initial proposal has been
formulated.
In conclusion, innovative use of fuel cells in the chemical industry were categorised
and used to develop functional modelling. Subsequently, 'proof' of the methodology
of process system innovation is given by the development of innovative system
concepts to realise trigeneration.
 6 Fuel Cell Vehicles and Industry Development

6.1 Abstract
Functional modelling has been applied to develop innovative system concepts for
the chemical industry to anticipate and exploit changes in logistic fuel demand
related to the market penetration of fuel cell vehicles. Within a decade polymer
electrolyte membrane (PEM) fuel cell vehicles may enable a true transition in
'automotive transport'. We conjectured that part of these will be methanol-fuelled.
Scenarios were constructed for automotive PEM fuel cell market developments,
technological progress and fuel diet. Simulations up to the year 2030 visualise the
dynamics and scope relevant for the chemical industry. In the conservative 'Superior
PEM vehicles' scenario, 10% of the annual sales of passenger vehicles and heavy-
duty vehicles are fuel cell powered by 2020. In addition, all combustion engine
passenger vehicles sold have a 42V PEM fuel cell auxiliary system on board. The
share of methanol in the fuel diet is expected to drop from an initial 100% to only
50% in all these markets. Technology advancement yields a 70% reduction in PEM
platinum-loading and an extension the economic lifespan of fuel cell stacks to 5
years. Technology to recycle fuel cells and recover platinum is developed.
In this scenario the production of the PEM fuel cell systems causes a peak in gross
annual platinum (Pt) demand of some 105 tons/year. Since fuel cell stacks are being
recycled, net demand for virgin Pt peaks at 95 tons per year around 2018, after
which it drops to a negligible level. Sustainable Pt use requires that a stack/Pt
recycling industry is developed in time. The main business opportunities lie in the
development of a stack platinum recycling industry and the associated transport
infrastructure because the long term dynamics of Pt-recycling appear much more
favourable than the dynamics of virgin Pt production.
In the chemical industry the main opportunity is the supply of methanol. The
scenario results in an annual demand of 70 MTA methanol after a period of high
demand growth. Using the methodology for process system innovation by design,
new objective-defined functions of the petrochemical industry and possible system
concepts for implementation in the Rotterdam industrial cluster were explored. One
option is the realisation of a unique integrated industrial complex around the flexible
manufacture and use of synthesis gas. The initial feedstock can be natural gas, but oil
residue, coal, biomass and biomass waste can also be converted to synthesis gas.
Methanol can consume a major share of the synthesis gas, but hydrogen
manufacture, metals processing, olefins and other chemical processes also benefit
from the variety of syngas qualities available in such a 'green' complex.
The only real barrier to these developments is the advancement of PEM fuel cell
technology and its final breakthrough to the marketplace.
The networked process system innovations presented illustrate the use of functional
modelling. Combined with scenario-building and dynamic modelling, a powerful set
of methods results to anticipate and respond to changes in the world outside the
industry. Thus, it enables the chemical industry to reap-the-benefits of a transition in
automotive transport.
Fuel Cell Vehicles and Industry Development 189

6.2 Introduction
The automotive transport sector worldwide today includes automobiles in all their
varieties, low- and heavy-duty trucks, buses, vans, and utility vehicles. Transport
accounts for roughly one-third of world energy resource use (WRI 2003).
Worldwide 20 million passenger vehicles and 600,000 heavy-duty vehicles (trucks,
buses) are sold each year (Anonymous 2003a; van den Hoed 2004).
The automotive sector represents a major market to the chemical industry. Today,
synthetic materials such as polymers, fibres and ceramics increasingly replace steel
bodywork and engine parts. Automotive fuel additives such as oxygenates are
produced in bulk quantities to augment gasoline quality. Its development continues
to create incentives for innovation and growth of the (petro)chemical industry.
Apart from the continuous stream of largely 'incremental' innovations, a transition
of 'automotive transport' initiated by the availability and breakthrough of polymer
electrolyte membrane fuel cell (PEMFC) technology could be realised in the near
future. Such transition will have major consequences for the petrochemical and
related industry. By first inspection of the requirements of PEM fuel cell vehicles
and the associated industrial activities, we conjectured that development of PEMFC
powered transport markets will open novel opportunities for networked process
system innovations around methanol49, hydrogen and carbon dioxide (CO2)
production. Available technologies may become part of an integrated system that
caters for transport, electricity, and product needs. These products are manufactured
via the generation of synthesis gas, which provides a generic link to a range of
conventional and novel feedstock, such as natural gas, coal, crude oil and biomass.
The use of biomass and CO2 sequestration contributes to sustainability. Thus, a
route towards a greener chemical industry can be opened.
Within a decade polymer electrolyte membrane (PEM) fuel cell vehicles may enable
a true transition in automotive transport, which is one of the most important outlets
of the chemical industry. We conjectured that alternative fulfilment of the objective-
defined function 'automotive transport' may be (in part) by methanol-fuelled PEM
fuel cell vehicles.
In this chapter, we report the use of functional modelling (Ch. 3) to explore the
possibilities, implications and opportunities related to the transition. This represents
a test case for the functional modelling approach and procedures developed:
Does functional modelling provide an adequate means for abstraction
from currently employed system concepts and do the procedures enable
or facilitate the specification of innovative system content?
The qualitative analysis is augmented with a quantitative assessment based on system
dynamics modelling and scenario analysis, in order to obtain a first impression of
the impact and constraints in potential transition scenarios.

49Methanol can be used in combustion engines but has never penetrated the logistic fuel markets on
a significant scale. Amongst others the Association of Methanol Producers for many years has
advocated the use of methanol as a suitable fuel for passenger vehicles. Such use of methanol
represents a major opportunity for those involved in the methanol supply-chain - natural gas
producers, methanol companies and suppliers of technology and equipment - to grow their business.
190 Process System Innovation by Design

Firstly, a functional model is developed that captures the possible outcome of

industry and automotive developments (§ 6.3). To this end, developments in
automotive transport around PEM fuel cell innovation are summarised. The options
for PEM fuel cell use in 'automotive transport' are identified, described and
categorised. Secondly, scenarios for 'automotive transport' realisation by methanol-
fuelled PEM fuel cell technology are constructed (§ 6.4). A dynamic model is
presented that allows visualisation and forecasting of the effects of PEM fuel cell
developments. Thirdly, scenario results are presented, their impact on the world
methanol industry and the global platinum material cycle analysed and PEMFC
related opportunities discussed (§ 6.5). Fourth, realisations of the functional image
thus created are conceptualised through integrated network system design based on
functional modelling and scenario analysis (§ 6.6). Elements for a transition strategy
for part of the petrochemical industry are combined in networked process system
innovations, notably the system concept for an integrated 'green' industrial complex
that could be located in the Port or Rotterdam. Finally, the results are briefly
discussed and conclusions drawn with respect to functional modelling (§ 6.7).

6.3 Functional modelling

6.3.1 'AUTOMOTIVE TRANSPORT' SYSTEMS

The objective-defined function of all transport systems used by mankind is the
realisation of 'automotive transport'. In the Twentieth century, the possibility to
realise affordable 'automotive transport' amongst others has led to the creation of an
immense industrial infrastructure to cater for automobile manufacture and
automotive fuel supply respectively (Figure 6-1).

Figure 6-1: The present industrial infrastructure for 'automotive transport'.

'Automotive transport' today comprises 'individual' vehicles that are able to move as
stand-alone items and 'system-linked' vehicles that are always attached to some
infrastructure. Existing and future realisations of individual vehicles (aircraft, marine
vessels) can be categorised around (1) the conventional combustion engines, (2)
electrochemical fuel cells, (3) batteries, (4) pneumatic systems and (5) solar or wind-
powered systems. System-linked vehicles comprise (6) combustion engine powered
and (7) electrically powered. The system-link may be rail tracks (diesel trains),
electricity supply from the grid (trolley-buses), or both (electric trains, metro
systems, city trams). Special system-links are sky trains and automatically guided
vehicle systems (AGVs).
As a matter of course, (1) represents current realisation of ' individual vehicles' such
as automobiles, trucks and auto buses. Presently, gasoline, diesel and LPG are
employed worldwide as fuel, while ethanol, bio-diesel and natural gas are employed
in various regions at a limited-scale only. The use of methanol, diesel/methanol
Fuel Cell Vehicles and Industry Development 191

mixtures and dimethylether (DME) have been demonstrated. Only recently, BMW
has developed a prototype of their 730 model that operates through internal
combustion of pure hydrogen.
Options (3) and (4) appear to be developments for transport with limited range.
Vehicles for city transport have been developed and tested that use compressed air
to provide shaft power. The use of battery-powered vehicles is largely limited to
utility vehicles. The weight to energy-storage ratio is less favourable for long-range
transport using the current and emerging generation of batteries. Advanced battery
subsystems are being incorporated in PEM fuel cell systems to cater for peak- and
start-up power. Options (6) and (7) are the well-known realisations of public
transport.
Compared to transport based on (1) combustion engines, options (2) around
electrochemical fuel cells are expected to achieve much higher 'well-to-wheel'
efficiencies whilst largely avoiding harmful emissions (Geissler et al. 2001). Initial
developments on alkaline fuel cells and phosphoric acid fuel cells were not
successful. Today, polymeric electrolyte membrane, PEM, appears to be the fuel cell
technology most suited for automotive application. Various demonstrations have
been launched, which all featured conventional vehicles propelled by PEM fuel cell
systems instead of combustion engines.
Since PEM fuel cells electrochemically convert hydrogen, on-board storage of
hydrogen and an infrastructure for hydrogen distribution is required. Many studies
and efforts of governments, fuel cell technology developers and automobile
companies focus on on-board storage of hydrogen and the associated infrastructure
for hydrogen supply. In the refining and petrochemical industry, however, it is well
known that hydrogen is expensive to manufacture, to handle and to transport. In
contrast, the use of logistic fuels for fuel cell vehicles that incorporate on-board
conversion to hydrogen would require only limited adaptation of the present
industrial infrastructure (refining, gasoline stations etc.). Therefore, developments
are under way to demonstrate that on-board fuel conversion to hydrogen is
technically and commercially feasible. Fuel candidates are (clean) gasoline, diesel,
LPG, ethanol, methanol, and DME. The best result would be the realisation of so-
called Flexible Fuel Reformer (FFR) technology, because this would open the way to
economy-of-scale in the mass production of only a single type of reformer for
various fuels50.

6.3.2 A FUNCTIONAL MODEL FOR 'AUTOMOTIVE TRANSPORT'

The abstraction 'from well-to-wheel' appears to be a functional model of the
industrial infrastructure and automobile industry that together allow us to realise
'automotive transport' through individual vehicles. However, it does not indicate
that intermediate functional elements are required in between a 'well' and the final
realisation of transport via 'wheels'. More importantly, it is not a technology-free
specification, as it hints at the use of an (oil or gas) well and at a transport concept
that employs wheels. Thus, the concept is not suited to open the way to think of
historic or innovative transport concepts that do not use (oil and gas) wells, wheels
50R&D is being conducted into the realisation of systems where methanol may be used without prior
conversion to hydrogen (and CO2) in a 'direct methanol PEM fuel cell'.
192 Process System Innovation by Design

or both. One may think of, for example, bicycles, solar powered vehicles, aircraft,
marine vessels, sailing ships, magnetic trains, hot-air balloons, rockets, zeppelins,
snow gliders and sky trains.
Therefore, we prefer the concept of 'automotive transport'. Its functional model
essentially consists of the functional elements (1) 'realising availability': some
(energy) resource is made available to be used for automotive transport and (2)
'realise automotive transport': employ or convert the energy resource to (shaft)
power that drives a suitable vehicle. Thus, the functional model for automotive
transport is similarly decomposed as the model for petrochemical systems around
aromatics (Ch. 3) and Olefins (Ch. 4), where 'production and consumption' systems
were discerned.

Emissions(non-ODF)

FE 1
FE 2
Fossil reso
Provide suitable
Provide Transport
urces energy source
automotive of goods
transport
Renewable FE 1
resources Energy Transport
FE 1
Source of people
FE 2
es FE 2
sourc
Flow Transpo
rt
FE N of livesto
FE N ck

Figure 6-2: Functional model around 'automotive transport.'

On a sailing ship or a solar power driven vehicle, these two functional elements
suffice to realise automotive transport, whenever there is a sufficient breeze or
sunlight respectively. In all other cases, functional element (1) will include substantial
infrastructure to distribute the energy source to be used. Thus, functional element
(1) may be further decomposed into (1.1) initial extraction -by depletion of a
reservoir or by adequately tapping a flow source (such as hydropower, solar power) -
, (1.2) resource processing (such as oil processing in a refinery, or the conversion of
biomass to ethanol) and (1.3) supply/distribution (through some wholesale/retail
infrastructure).
Functional element 2 is the “front-end” or market end of the total industrial
infrastructure that offers the possibility to realise (individual) automotive transport.
It may be further decomposed into all functional sub elements that together
constitute a vehicle (of various type and utilisation), and also includes the entire
industrial infrastructure to bring said vehicles into being. (Figure 6-2).
Fuel Cell Vehicles and Industry Development 193

The above model and decomposition match the current implementation of

'automotive transport'. A top-level decomposition between (1) 'make 'automotive
power' available', (2) 'distribute 'automotive power' to users' and (3) 'provide
transport' represents an improved abstraction. These functional elements may
overlap. Thus, an improved model is obtained that emphasizes that on the one hand
an industrial infrastructure must exist to create availability of motive power, on the
other hand some 'hardware' to realise transport itself, whilst in between somehow
motive power must be made available through distribution (Figure 6-3).

Emissions(non-ODF)

FE 1 FE 2
Foss FE 3
resou il
Convert to Distribute Provide Transport
rces suitable power power source automotive of goods
source
transport
FE 1
FE 1 FE 1 Transport
Energy Source of people

ble
ewa FE 2 FE 2 FE 2
Ren urces Trans
p
reso of live ort
stock
FE N FE N FE N
so low
s
ce
F
ur

Figure 6-3: Improved functional model of 'automotive transport.'

Currently, two rather extreme type of realisations of 'automotive transport' exist that
are captured by this model: (1) railroad- or other system-bound transport (trams,
metro, sky trains), which combines a fixed track with an infrastructure for motive
power supply and (2) road-based vehicles that depend on a system of motive power
supply from crude oil refining and distributed through a network of stations. Trolley
buses are somewhat of an exception, as they do employ a fixed infrastructure for
power supply. The model also captures novel concepts, such as magnetic trains.

6.3.3 AUTOMOTIVE TRANSPORT AND INNOVATION

The drive to capture and keep a share of the huge market of road vehicles has
supported a continuous innovative development of the automobile industry and its
suppliers. The demand for acceptable cost had to be met whilst realising enhanced
vehicle safety and comfort, whilst reducing the environmental impact of 'automotive
transport' that is subject to increasingly stringent regulation.
From a functional systems perspective, however, the systems in the automobile
industry, its co-makers and suppliers are largely characterised by maturity. Although
new market niches continue to be found, notably in the automobile sector, and
products and technology applied continue to be improved and optimised, the
essentials of 'automotive transport' have not changed for a century, i.e. since the
194 Process System Innovation by Design

introduction of the Model T Ford. At a (networked) systems level, innovations in

fuel supply and car manufacture are largely incremental in character. Combined,
these have led to automobiles that make optimal use of the system concept around
combustion engine technology. The maximum achievable well-to-wheel efficiency,
however, is low when using combustion technology. The hydrocarbon fuels used
result in continued emission of CO2, even from so-called ultra-low or zero-emission
vehicles. Thus there are many similarities between the development and status of
the automobile sector and one of its suppliers, the petrochemical industry (Ch. 1).
In the global quest for sustainability and in response to climate change concerns,
many advocate a transition towards a hydrogen economy. In the automobile sector,
this transition is to be enabled by fuel cell technology.
'Automotive transport' appears to be one of the promising applications for fuel cells
because of the limited efficiency of currently employed gasoline and diesel engines, the
potential size of the market and the potential ecological advantages.
Since both diesel and gasoline combustion engines convert fuel to heat and heat to
shaft-power, their performance is limited by the maximum achievable Carnot-efficiency.
Current engine efficiency is some 15-25%. In theory, the achievable efficiency of
direct electrochemical conversion of chemical fuel energy to shaft-Work is about
twice as large, some 40-50%.
Capturing a share of the market of vehicles produced would allow realisation of mass
production of fuel cell units, which would greatly reduce cost per kW power unit.
Apart from protecting vested interests in the current automotive fuel infrastructure,
these are the main incentives that have led to a number of global alliances between
oil companies, fuel cell developers and automobile manufacturers such as Shell-
Ballard-Siemens-Daimler-Chrysler and ExxonMobil-Ballard-General Motors (e.g.
van den Hoed 2004).
The ecological advantages of fuel cell powered vehicles are a third incentive for fuel cell
application in 'automotive transport'. A system solution around PEMFC vehicles
prevents diffusive emissions. Conventional engines suffer from a host of emissions
that are detrimental to local and regional air quality. These comprise soot, aerosol
particles, NOx, SO2 and volatile organics (Spiro and Stigliani 1996). Although
innovative engine technology and exhaust catalysts have successfully reduced many
of these emissions, a significant level of emission remains. Where combustion
engines can be equipped with exhaust catalysts, a classic example of an 'end-of-pipe'
solution, one cannot avoid diffusive CO2 emission without a change of fuel.
Hydrogen powered vehicles that use combustion engines would emit water and
some NOx. Vehicles that employ hydrogen and fuel cell technology would emit
water only.
Presently hydrogen is produced from fossil fuels. When using fuel cell technology,
however, the total CO2 emission of 'automotive transport' decreases because of the
improved 'well-to-wheel' efficiency that reduces net system fossil fuel input (Geissler
et al. 2001). In addition, hydrogen can be generated in central production facilities
that use fossil resources and sequester the CO2 by-product.
Fuel Cell Vehicles and Industry Development 195

Reduction of fossil fuel use and CO2 emission of 'automotive transport' requires a
transition in vehicle technology (FE 3) and an alternative realization of resource
conversion and distribution (FE 1 and 2; see Figure 6-3).
Cadre 6-1: From Well-to-wheel with methanol and PEM fuel cells

To realise a 'sustainable', notably clean, quiet and cost-effective system for

automotive transport, an alternative system may be implemented, where
polymeric electrolyte (PEM) fuel cells systems are used to propel vehicles and
where methanol is used as a logistic fuel. Methanol is already available from the
petrochemical industry, where it is produced in bulk quantities from natural gas.
Since it is a liquid, small-scale distribution costs are relatively low. Methanol is a
safe and sound source of hydrogen via relatively mild on-board fuel reforming.
An infrastructure around Methanol-fuelled PEMFC propelled vehicles represents
a realisation of the functional model described (Figure 6-3). Functional element
FE 1 is realised through the winning of natural gas and conversion to methanol,
possibly combined with the use of biomass. FE 2 is the supply/distribution of
methanol to customers through some wholesale/retail infrastructure. As winning
and conversion in FE 1 may not be at the same location, nor conversion at the
location of entry of FE 2, intermediate transport functions are required, such as
pipeline infrastructure for natural gas, liquefaction facilities and shipping of LNG
and shipping and storage of methanol. An overview of the system from 'well-to-
wheel' is given in (Figure 6-4).

One option is the fulfilment of the ODF 'automotive transport' by PEM fuel cell
vehicles. Since on-board conversion of methanol to hydrogen is technically and
commercially feasible, PEMFC vehicles may be built with on-board storage of
hydrogen or methanol. In both cases, the associated fuel supply and distribution
infrastructure must be developed. In contrast, the use of logistic fuels - (clean)
gasoline, diesel, LPG - would require only minimum changes in supply and
distribution.

Figure 6-4: 'Automotive transport' realisation by PEM fuel cells and methanol.
196 Process System Innovation by Design

As a matter of course, the use of methanol and logistic fuels will result in
production- related CO2 emission and diffusive emission from the vehicles. The use
of methanol is expected to yield the smallest amount of total emission. In addition,
where presently almost all methanol is produced from natural gas, an efficient
system would result when combining the production of methanol from natural gas
and biomass (Stikkelman 2001).

6.3.4 NETWORKED PROCESS SYSTEM INNOVATIONS

The existing realisations of automotive transport serve to illustrate that system
content and system input can be varied to yield similar output to meet the same
objective-defined function, 'automotive transport'. The alternatives may involve the
entire system or they may be limited to the 'front-end' functional element FE 3
(vehicle) or 'back-end' functional elements FE 2 and 3 (vehicle and distribution
system), respectively (Figure 6-3).
The realisation of 'automotive transport' through PEM fuel cell technology (FE 3,
Figure 6-3 and Figure 6-4) in combination with the use of existing logistic fuels
(gasoline, diesel, LPG) appears to be completely within the currently employed
system concept, with innovation remaining largely limited to the engine concept
used beneath the hood. Indeed, the industry has demonstrated that (prototype) fuel
cell cars do not have to look any different from their conventional counterparts.
Thus, fuel cell driven cars would be introduced as a new fuel channel and
technology beneath the hood, however with no consequences to car exterior, nor to
its objective function, the provision of automotive transport. Whenever this concept
for PEM fuel cell vehicles becomes reality, however, the product spectrum of
refineries must change, depending on the characteristic of FFR technology. It is, for
example, well known that reforming of aromatics is difficult. Thus, FFR
introduction may require that the composition of gasoline and diesel be modified. In
case this is to be done for substantial volumes, process system innovation of
refineries may well be required.
In case hydrogen or methanol would become the preferred fuel for distribution to
PEM vehicles, an alternative realisation of FE 2 is added to the current situation, i.e.
hydrogen and methanol become a logistic fuel. Upon first inspection, the extent of
networked process system innovation appears to be limited. The adoption of
methanol, however, does imply the use of natural gas instead of crude oil. Initiation
and subsequent growth of a methanol-fuelled fleet of PEM fuel cell vehicles will
require substantial growth of industrial methanol capacity. This may finally imply a
change in characterization of methanol from chemical to fuel (Fahy 1990). This
opens the way to innovative networked process systems (see § 6.6).
Advocates of a transition to a 'hydrogen economy' envisage production and
distribution of hydrogen on a large-scale. Various options for system realisation
have been proposed and are being tested, such as centralized hydrogen production
and compression, trucking the hydrogen to distribution stations and retail (BMW
demonstration, Germany). At the other end of the spectrum is decentralised
electrolysis of water and compression of the hydrogen using electric power from the
grid. Alternatively, an integrated industrial infrastructure for the supply of hydrogen
Fuel Cell Vehicles and Industry Development 197

can be envisaged, where hydrogen is distributed via some pipeline infrastructure,

which may or may not be integrated with the natural gas distribution grid (Weijnen
1999).
A system for distributed generation of hydrogen via from some logistic fuel or
methanol also appears to be feasible, in which case one may opt for hydrogen
production/compression at retail stations or for on-board hydrogen generation via
methanol reforming, as demonstrated by the Daimler/BASF NeCar 5. A transition
towards PEM fuel cell propelled 'automotive transport' thus requires fuel cell
technology and may require networked process system innovations to realise feasible
alternative fuel infrastructure and associated industry.

6.4 Scenario development and modelling

Within a decade polymer electrolyte membrane fuel cell vehicles may enable a true
transition in automotive transport, which is one of the most important outlets of the
chemical industry. We conjectured that alternative fulfilment of the objective-
defined function 'automotive transport' may be (in part) by methanol-fuelled PEM
fuel cell vehicles.
In order to help industry to anticipate and shape this transition in 'automotive
transport', quantitative scenarios were constructed that combine automotive PEM
fuel cell market developments, technological progress and fuel diet evolution.
Simulations up to the year 2030 visualise the dynamics and scope relevant for the
chemical industry.

6.4.1 TECHNOLOGICAL PROGRESS

On the basis of an authorative review (Arthur D. Little 2000a) - summarized in
Appendix A. 7, four key aspects were identified that must be incorporated in the
model used to calculate the scenario results:
1. Dominant system characteristics.
The net PEMFC system efficiency determines the yearly fuel requirement for a
typical vehicle and its gross power rating.
2. Status and outlook of PEM technology development.
The required PEM platinum-loading per kW stack output and the gross power
rating determine the net amount of platinum required for the production of a
new system.
3. PEMFC system lifespan and recycling.
The system operating lifespan and the development of stack recycling for at least
precious metals (platinum, ruthenium) recovery will determine the development
of net Pt demand for the PEMFC automotive markets.
4. Fuel selection
In PEMFCs hydrogen is converted electrochemically with oxygen to water.
Thus, on-board hydrogen storage or on-board hydrogen production is required,
for example from LPG or methanol.
198 Process System Innovation by Design

6.4.2 DETERMINANTS OF MARKET SCENARIOS

Our objective of sketching possible futures for PEMFC vehicles, scenarios, was to
visualise their consequences to the external world. With suitable scenarios and
modelling a projection can be made of the scope and development of methanol
demand, platinum demand and the impact on total transport related CO2 emission.
We have opted to use the development of PEMFC economics and PEMFC
competitiveness as the main determinants of market scenarios. This leads to the four
scenarios that have been summarised in Figure 6-5.
The economics of PEMFC vehicle use and production are related. The total cost of
ownership and use, TCOU, of a vehicle is determined by the required investment -
the annualised cost of acquiring the vehicle - and by the subsequent operational
costs: fuel, maintenance, repair, spare parts and taxes levied on car ownership and
fuel. The cost to manufacture a PEMFC vehicle depends on production volume and
technology status. Higher production volumes result in lower development and
assembly cost per vehicle. Meanwhile, automobile development and mass
production facilities require massive investments, which can only be recovered at
sufficient output levels.
The status of PEMFC technology influences TCOU via the impact on cost per
vehicle. At present, major cost items are materials needed to manufacture PEMFC
systems (platinum) and the combined cost of many system components and system
complexity. Thus, at $1000/kW, the of a 30 kW PEMFC engine is 20 times that of a
conventional 30kW gasoline or diesel engine purchased at some $1500/engine.
Similarly important, however, technology status determines cost of operation of the
vehicle. Targets and status with respect to technology development have been
summarised in Appendix A. 9, based on (Arthur D. Little 2000a). Major items are
the limited PEM fuel cell stack lifetime, which requires costly replacement after 4-5
years of operation.
PEMFC competitiveness, its cost and benefits compared to other types of vehicles,
determine its adoption in the market place. Apart from social and psychologically
determined preferences, which are important in automobile sales, it is safe to assume
that PEMFC automobiles and trucks will offer competitive performance
characteristics. Then market development will depend primarily on PEMFC
sustained competitiveness over the full life cycle of car ownership and usage, i.e.
TCOU. Governments may influence the TCOU by tax differentiation between car
type and fuel - logistic fuel, natural gas, biogas, hydrogen or methanol. In addition,
PEMFC competitiveness may be improved by the creation of a level-playing-field
with respect to emission levels in vehicle exhausts (CO, NOx, SO2, soot,
hydrocarbons, lead, metals) and the total well-to-wheel CO2-emission. Finally, tax
incentives may be created to favour investment in new vehicle production facilities.
Finally, an important parameter is well-to-wheel efficiency and net amount of energy
resources used for automotive transport. In combination with fuel characteristics of
logistic fuels replaced by natural gas via methanol, this gives an impression of the
total amount of CO2 avoided. Vogel (2001) has calculated that for methanol -
PEMFC this efficiency is moderately higher than the well-to-wheel efficiency of
gasoline /diesel operated automotive transport @(Vogel 2001). In case other
Fuel Cell Vehicles and Industry Development 199

sources of methanol, e.g. biomass, are employed a significant improvement may be

expected (Stikkelman 2001).

6.4.3 MARKET SCENARIOS

In Figure 6-5, four possible scenarios have been depicted for different combinations
of development of PEMFC economics and PEMFC competitiveness respectively.

High
Competitiveness

'Superior PEMFC 'PEMFC

Vehicles' breakthrough'
become available Rapid fleet
for a limited market replacement by
share PEMFC technology'

Unimproved Improved
Economics Economics

'PEMFC, 'PEMFC
the eternal promise' Outcompeted'
Science & technology Market power of
cannot deliver feasible vested players
solutions remains
unchallenged

Low
Competitiveness

Figure 6-5: Scenarios for PEM in automotive transport.

In 'PEMFC, the eternal promise', effectively no introduction of PEMFC vehicles of any

kind occurs, as both economics and competitiveness remain as is. This scenario has
been reality for the last decade, where science and technology breakthroughs,
though anticipated, have not materialised into an affordable automobile; a number
of demonstrations (cars, automobiles, van) have been launched though.
'PEMFC outcompeted' is a scenario where PEMFC does not make it successfully to
the market despite technological advancement and improved economics. This is
largely the scenario exhibited by steady-state fuel cell applications. These systems
have been dramatically improved. Meanwhile, however, cogeneration technology
also has been improved. Both in stationary power generation and in automotive
transport, vested interests and past investments are immense. Therefore, vested
players are reluctant to favour a level playing field or to allow temporary incentives
for PEMFC systems.
200 Process System Innovation by Design

In 'Superior PEM vehicles' the economics are sufficiently improved and incentives
have been created so that PEMFC vehicles are considered superior goods. Thus,
they will obtain a share of the market that depends on economic prosperity.
Conventional gasoline and diesel vehicles continue to dominate the world's vehicle
fleets. Many automobile companies do introduce novel models with PEM fuel cells
to deliver auxiliary power.
'PEM Breakthrough' materializes when general awareness of the threat of Global
Warming, technological progress, venture capital investment and clever marketing
result in a sweeping introduction and acceptance of PEMFC vehicles worldwide.
Subsequently, in OECD economies, in a short period of time legislation is passed
and enforced to phase out gasoline and diesel operated vehicles. The time of
abundant, inefficient use of oil products in transport ends relatively abruptly. A
massive restructuring of the refining industry worldwide follows to supply logistic
fuels suitable for use in PEMFC vehicles. This includes natural gas and crude oil
conversion into methanol and hydrogen combined with carbon fixation or CO2
sequestration.
At present, it may be seen that PEMFC vehicle development resides in the lower-
left quadrant. Indeed, many are of the opinion that fuel cell technology will remain
an eternal promise. Recently, however, considerable investments in R&D
programmes, demonstration and commercialisation have been launched to move the
technology to the upper-left quadrant. Whether the 'Superior PEMFC vehicles' is
the most plausible scenario remains unknown. There is, however, considerable risk
that said programmes are not sufficient, which would imply that 'PEMFC
outcompeted' is the near future where 'PEMFC-the eternal promise' remains the
perception of many believers. A dramatic success of R&D would open the way to
realise the ultimate scenario, 'PEMFC breakthrough', which is also the most
dramatic of possible futures.

6.4.4 FUEL DIET AND VEHICLE INTRODUCTION

PEMFCs and other types of fuel cells eventually convert hydrogen electrochemically
with oxygen to water. PEMFC vehicles may be equipped with on-board storage of
hydrogen or with a fuel processor. In the former case, the associated hydrogen
supply and distribution infrastructure must be developed. In a variety of (PEM) fuel
cell markets, indeed hydrogen is expected to be one of the fuels of choice.
Air/hydrogen mixtures, however, are highly explosive when an ignition-source is
available. Although worldwide the chemical industry has a good track record in
dealing with hydrogen, people are naturally concerned about the risks of hydrogen.
The cost of pipeline transport of hydrogen is high. Therefore, it may be attractive to
develop on-site generation of hydrogen, which is possible via dissociation of
methanol at low-temperatures or the reforming of logistic fuels. Incorporation of
any of these conversion processes in the vehicles themselves requires miniaturisation
and economic down-scaling. At fuel stations, however, the conversion process can
be erected as a medium-scale stand-alone stationary facility. Thus, a fuel service
station that wants to sell hydrogen in the future may opt for either methanol
dissociation, FFR of logistic fuels or the electrolysis of water. Similarly, on a local
scale a neighbourhood that wants to produce PEMFC power can be supplied with
Fuel Cell Vehicles and Industry Development 201

hydrogen via a shared methanol reformer (Arthur D. Little 2000b). In a liberalised

fuel market methanol may thus provide a route towards distribution of hydrogen.
Bringing-into-being 'automotive transport by methanol', as illustrated in Figure 6-4,
p. 195, is subject to competition from the vested players in the logistics fuel
infrastructure markets. In addition, entrenchment caused by currently preferred
automobile characteristics may slow down alternative fuel market penetration.
Therefore, a 'thinkable' test case for methanol PEM fuel cell vehicle demonstration
was conceptualised and modelled, automatically-guided vehicles (AGVs).
In consultation with some parties in and around the Port of Rotterdam, the case of
internal transport on container terminals was selected. Around the world, two major
types of container terminals are in operation, viz. terminals where containers are
lifted and transported by movable cranes, and terminals where lifting is by fixed
cranes. In the latter case internal transport is by 'container trains' pulled by
(automatically-guided) tractor vehicles. These AGVs, which are also used at airports,
are high-power diesel engine driven. Related emissions and noise present a problem.
The extent of applicable vehicle legislation is limited, however, because the AGVs
are used on private property only. Therefore, real tests may be conducted in an early
phase of development.

6.4.5 A DYNAMIC MODEL FOR SCENARIO SIMULATION

PEM fuel cell technology is being developed for both stationary power generation
and for automotive applications (see Appendix A. 7). The model includes the 5
market segments for PEM fuel cell technology as identified in the literature (e.g.
Arthur D. Little 2000a). Per market segment, the model is set up to characterise and
visualise the dynamics of a fleet of systems (see Appendix A. 8).
In the model the platinum required for new systems, the methanol used as fuel and
the platinum recovered from recycled systems are calculated as a function of
technological progress, pattern of use and market development51.
Each year the composition of the running fleet is calculated as a function of the
number of systems that have been built in all previous years and the phase-out
pattern of systems of a particular production-year.
At the input side, the development of the fleet is driven by the market demand
foreseen in a particular scenario, which is nothing more than the number of new
systems produced per year to cater for expected demand. Market scenarios are
defined by two parameters: the total demand per year, and the fraction of systems
that are methanol fuelled.
The output or removal of systems from the running fleet is determined by the
phase-out pattern, which is a function of the technical lifespan, the economic
lifespan and the operational use. The characteristic distribution of end-of-life
vehicles recently has been determined (van Schaik and Reuter 2004). In our scenario
model these parameters change each production year due to technological progress.
An approximation of the phase-out pattern is included in the model as a normal
distribution around average technical lifespan.

51The model equations for each market segment are the same; only the model parameters differ
between the market segments. The model is given in Appendix A. 8.
202 Process System Innovation by Design

The balance of inputs and outputs is used to calculate the 'presence' of systems
produced in each previous year. In the 'Superior PEMFC vehicles' scenario (see
below), for example, in 2003 the first 10 light-duty passenger vehicle 30 kW systems
are introduced. These have an expected technical lifespan of 1000hrs of operation.
They are used 200 hrs per year to drive some 16,000 kilometres. Thus their average
technical lifespan is five years. One of the systems, however, breaks down to a status
beyond repair in 2003, the others are in operation successfully until 2007 (1), 2008
(2), 2009(4) and 2010 (2). Thus the actual operational lifespan shows some variation
around the 1000hrs design value, which reflects the effects of variations in style of
use.
The phase-out or 'end-of-life' scenario and the technology status at time of
production determine the net volume of platinum to be recycled. In the model,
therefore, the variation by which systems break-down can be specified. Also the
fraction can be specified of systems that are removed from the world-wide fleet in
operation due to other reasons than PEM fuel cell breakdown (accidents, fire,
change of lifestyle, misuse etc.).
The combination of scenarios for market demand (input), system use and end-of-life
(output) provides a projection of the running fleet. From this dynamic forecast of
the size of the running fleet, its characteristics and use, the implications for
methanol and platinum demand can be computed. In addition, the net amount of
CO2 and other emissions can be calculated.
In order to get a realistic impression of the dynamics and scope for platinum and
methanol, the key parameters that are affected by technological progress have been
incorporated in the model. These are the total Pt-loading required (expressed as g Pt
per net kW system output), the system expected operational-economic lifespan (<
5% power degradation), and the overall system efficiency of methanol-to-mileage.
From the typical system power-rating and use pattern the net demand for methanol
is calculated. The total methanol demand is determined by the sales volume. In the
model, the fraction of methanol-fuelled PEMFC can be varied.
The typical power-rating and technology status with respect to required Pt-loading
determine the net Pt-demand for a system produced in a particular year. Once
produced, the net amount of Pt remains unchanged until the system ceases
operation and thus becomes available for recycling. In the calculations, a 100%
platinum recovery from these systems has been assumed.
In summary, the dynamics of future PEMFC-related methanol demand, the net
'virgin' platinum demand, the demand for Pt-recycling and the net amount of Pt-
stored in the PEMFC fleet are calculated. The model can be extended to visualise
future (avoided) CO2 and other emissions, the net content of Pt (g/kg) in systems
recycled (requires forecast of development of PEMFC specific weight development),
and the amount of hydrogen required for the hydrogen-fuelled fraction of the
PEMFC fleet.
Fuel Cell Vehicles and Industry Development 203

6.5 The 'Superior PEMFC vehicles' scenario

We have opted to visualise the consequences for platinum and methanol if the
'Superior PEM vehicles' scenario would become reality. An overview of this scenario
is given in Cadre 6-2.

6.5.1 DESCRIPTION OF THE SCENARIO

Technological development in this scenario affects PEM platinum loading and stack
lifespan (Figure 6-6). The dependency of the PEM fuel cell operation on platinum-
coated membranes, both anode and cathode side, is seen as a serious constraint for
market introduction. In a recent cost estimation study for PEM fuel cell systems,
Arthur D. Little (2000a) concluded that at 2000 technology status, the fuel processor
and tail gas burner together require 30 g of Pt as catalyst ingredient. At a 50-kW net
output, a PEM fuel cell system requires some 180 g of platinum loading for the fuel
cell stack alone (Arthur D. Little 2000a). Massive R&D efforts have led to this
current status in Pt-loading of the membrane electrode assembly (MEA). Hampden-
Smith claims to have developed a MEA with only a third of the Pt-Loading quoted
by Arthur D. Little (Hampden-Smith 2000). No proof, however, has been found
that this technology is used in commercial fuel cells. A typical heavy-duty truck using
PEM technology available today in a 300kW system would include a total of 1.2 kg
of Pt at current cost of $18.00052. This cost could drop to some $7000/system if Pt-
loading reduction for MEA becomes a reality.
Cadre 6-2: Superior PEM vehicles scenario summary

Market development for Automotive only:

• Total number of cars sold worldwide = 20 Million[systems/yr]. (2003-2030)
1. total penetration of fuel cell auxiliary power: 20 Million [systems/year]
2. PEM fuel cell powered vehicles: up to 2 Million [systems/year]
3. Heavy duty PEMFC vehicles (trucks, buses): 60,000 [systems/Year]
4. PEM fuel cell AGV's: 2000 [systems/Year]
• In each category, initial market share of methanol is 100%; mature market share
is 50%
• Technology development:
1. Pt-loading: present PEMFC technology status = 3.8 [g of Pt/kW]
2. Pt-loading improves to only 0.8 [g of Pt/kW] (2020-2030).
3. Stack lifespan increases from 1 000 to 40 000 [hrs/system].
4. System efficiency is constant at some 40% LHV
• Material cycle development
5. Spent PEM's are processed to completely recover platinum

52 At a typical Pt price of some $15/g. Current prices can be found e.g. at www.kitco.com.
 204 Process System Innovation by Design

In the scenario a number of breakthroughs that result in stepwise decremented

PEM platinum loading. Continuous improvement gradually reduces required Pt-
loading. Similarly, an increase in system economic lifespan is modelled as a number
of steps in combination with incremental improvement. In the scenario,
technological progress yields a 70% reduction in PEM platinum-loading and an
extension the economic lifespan of fuel cell stacks to five years. Since PEMFC
vehicles will incorporate battery systems, regenerative breaking and advanced
transmission, the rated fuel cell power output is 30 kW for passenger vehicles.
Heavy duty PEMFC vehicles average 160 kW of system power output, which
together with peak battery power yields a performance characteristics similar to
conventional diesel engine powered vehicles.
Stack development
6

5 Tech.
Lifetime
Oper.
4 [Years]
[g/kW], [Years]

2 Platinum
Loading
[g/kW]
1

0
2005 2010 2015 2020 2025 2030
Year

Figure 6-6: Stack development scenario: Platinum loading and technical lifespan for
automotive PEMFC systems.
Market development expressed as the number of systems sold determines the order-of-
magnitude of the platinum and methanol demand, respectively. Today, worldwide
20 million passenger vehicles and 600,000 heavy-duty vehicles (trucks, buses) are
sold each year (Anonymous 2003a; van den Hoed 2004). In the conservative
'Superior PEM vehicles' scenario, 10% of these annual sales are fuel cell powered by
2020.

New systems per year Running fleet

(No. of systems in operation)
25 300
Total
250
20

200 Auxiliary
15
Million

Million

150
10 Heavy-
100 Duty*100

5
50 Light-Duty

0 0
2005 2010 2015 2020 2025 2030 2005 2010 2015 2020 2025 2030
Year Year

Figure 6-7: Market development scenario for automotive PEMFC systems.

Fuel Cell Vehicles and Industry Development 205

Auxiliary power systems. Developments are underway to change the voltage of

passenger vehicles on-board electric DC-power. In cars the electric power
consumption of auxiliary systems continues to increase. Amongst others, fuel cells
are expected to be integrated into automobile electric systems. In the scenario, all
combustion engine passenger vehicles sold are expected to have a 42V PEM fuel
cell auxiliary system on board.
Fuel diet. The share of methanol in the fuel diet is expected to drop from an initial
100% to only 50% in all these markets. We expect that methanol will be the only
fuel used initially because we believe methanol will be more readily accepted, and
on-board methanol storage and reforming using engine waste-heat does not present
serious problems. We do expect that hydrogen technology picks up after a few years
for auxiliary systems. This gradual change is incorporated in the scenario. In PEM
fuel cell equipped cars, we do expect competition between methanol fuelled and
flexible fuel reformer (FFR) equipped cars that may use LPG, gasoline or diesel. In
the scenario, we have assumed that the introduction of these FFR-PEM systems will
limit the final methanol share of the fuel cell car market to 50%. In heavy-duty
applications we expect a similar situation to develop. Notably, the calculated
methanol estimates do not include the possible development of on-site manufacture
of hydrogen from methanol in the fuel distribution infrastructure.
Mass production development has been modelled to reflect realism and above given
market penetration. All three vehicle markets' forecasts are characterised by a very
low volume up till 2005, followed by rapid growth characterised by two steps in
feasible levels of mass production per facility: tens of thousands of systems per
facility from 2005-2010; hundreds of thousands of systems from 2010 to 2015, and
sufficient growth to justify mass production plants for 500,000 systems each.
Demand stabilises after 2019 to obtain some idea of the 'steady-state' of the
scenario.

6.5.2 IMPACTS ON PLATINUM

In Figure 6-8, the relation between the world's platinum market, the methanol
market and the various PEMFC markets is visualised: the four major markets for
PEMFC clearly will represent magnificent drivers for both methanol and the
platinum related industry.
Mass production of PEM fuel cells for passenger vehicles appears to have the
potential to dramatically impact the world platinum market because it can introduce
a substantial imbalance in worldwide supply and demand. Using the above
characteristics, Arthur D. Little calculated that an annual production of 500,000 cars
requires 52 metric tons of Pt at the current level of technology. This represents
some 30% of current worldwide production capacity from mining operations, which
amounts to some 160 metric tons (Johnson Matthey, Platinum Today).
Platinum requirements for PEM automotive fuel cell systems represent a completely
new type of demand, which comes on top of demand for other platinum
applications (Figure 6-8, page 206). Successful introduction of PEM technology may
therefore lead to platinum scarcity and soaring prices. When one considers the
above estimates for PEM-related platinum demand, one may conclude only a small
percentage the worlds 200-300 Million vehicles can be replaced by PEMFC vehicles.
206 Process System Innovation by Design

This would imply that some or all of these markets may never blossom. In other
words, the limited size of the world platinum market appears to be a serious
constraint for PEMFC vehicle market development.

Stationary - Small
(1-5kW)
PEM FC
Passenger Vehicles
Production (30 kW)
Trucks, Buses
(300 kW)
Vehicles Auxiliary
(0.2 kW)
Natural Supply MeOH Distribution
Winning
Gas Infrastructure production Infra
MTBE>Gasoline
additive
Platinum
Pt- Platinum Catalysts for oil refining MMA>PMMA windows
Mining Markets /
Ore Extraction
Distribution Catalyst for chemical plants Other: (plastic)
materials
Computer harddisks

LCD manufacture Gasoline: Pt-Loadign

Jewellery Down
Exhaust catalysts
Diesel: Pt-Loading
Investors Industrial Up

Figure 6-8: Overview of related platinum, methanol and PEM fuel cell markets.
Therefore, we calculated a forecast for a period of 30 years. The results are
visualised in Figure 6-9. The combined developments - the number of systems sold
increases while the amount of platinum per system decreases - result in a peak of
total Pt-demand of 160 tons per annum. Due to the lifespan of PEMFC power
systems, Pt-supply from PEM Pt-recovery operations trails behind some 5 to 7
years. In the scenario, however, the Pt recycled from one early vehicle can be used in
multiple new vehicles that use more advanced technology. As a consequence, in the
scenario virgin-Pt demand peaks at some 100 ton per annum in 2012 and decreases
rapidly to extinction around 2020 when markets and technology have matured.
Finally, the amount of platinum from recycling will match or exceed net Pt demand
for new systems, in this scenario after some 30 years. Thus, a net inflow of Pt from
the PEMFC fleet changes to a net outflow of Pt from the PEMFC fleet. This
implies that Pt-production capacity must be carefully planned and that capacity
required for Pt recycling is fairly predictable and reaches some steady-state level.
Obviously total PEM related platinum demand and virgin demand levels will go
down if technology advancement with respect to Pt-loading is more successful. The
shape of the curve, however, remains the same: total Pt-demand will rise until
PEMFC markets have matured (in our scenario: around 2022, i.e. in 20 years time),
and then drop to a quasi steady-state level that is a function of technology status.
In Figure 6-9 the total amount is plotted of Pt stored in the PEMFC automotive fleet
as well: the first 20 years, it grows dramatically, then levels off to a steady-state of
some 800 tons (!). Thus, the long-term dynamics of the worldwide Pt material cycle
have been elucidated.
An important question is: can the systems in the worldwide platinum material cycle
respond adequately to the dramatic dynamics resulting from only moderately
Fuel Cell Vehicles and Industry Development 207

successful PEM technology introduction? Can production capacity of virgin Pt and

the availability of Pt from spent PEM fuel cell recycling be adequately managed?
The model predicts a net increased demand for an already scarce and precious
material. Indeed, according to Johnson Matthey (2000) the steep increase of Pt-
prices in early 2000 was due to the announcement of a number of consortia between
Ballard Power Systems, the leading fuel cell manufacturer, automobile companies
and oil companies. Johnson Matthey analysts, however, attribute the same price
effects to the erratic issue of Pt export quota's by Russia's prime minister Putin! By
mid-2001 supply-demand balance had largely been restored, and prices have
dropped to 'normal' levels again of some $450/troy oz.

125

100 Pt - demand

75
[Ton/Year]

50 Pt - recovered

25

0 Pt - virgin
2005 2010 2015 2020 2025 2030
-25
Year

Figure 6-9: Model forecast for world Pt-demand, 'Superior PEM vehicles' scenario.
From statistics compiled by Johnson & Matthey (www.johnsonmatthey.com), it can
easily be seen that Pt supply has doubled since 1975, from a net supply of 81
ton/year to the current level of 174 ton/year. This development has occurred largely
in response to continuously increasing demand. In their most recent yearly outlook
(2001) Johnson Matthey quote that for example South Africa have huge yet
undeveloped Pt resources, and that both prospecting activity worldwide has
increased as well as expansion of mining and refining planned. In addition, where
(e.g. Hampden-Smith 2000) states that Pt markets are extremely volatile and its price
unpredictable, Johnson Matthey indicates that the relatively large jewellery market
share effectively results in a price-cap at some $650/troy oz. where jewellery
consumers will simply stop buying the material, which would allow a shift from e.g.
jewellery to fuel cell manufacture by Pt. For example, in case jewellery consumption
would go back to 1991 levels, some 44 tons of Pt would become available on the
market. Supply and demand data also show that with the introduction of auto-
exhaust catalysts and the development of a new market outlet for Pt, a few years
later also Pt becomes available from recycling of said catalysts for, amongst others,
the recovery of Pt. The historic data suggest that a net decline in Pt-demand for
automobile exhaust catalyst is foreseeable. Together with the continuous growth in
supply, there is thus no reason to believe that Pt availability is a constraint for the
development of a fuel cell vehicle fleet!
208 Process System Innovation by Design

6.5.3 METHANOL FORECASTS

Where platinum is required to manufacture the required hardware, methanol is one
of the possible fuels for PEMFC powered transport. Using the model, a forecast
has been calculated of the methanol capacity required in the Superior PEM vehicles
scenario (Cadre 6-2, p.203). In Figure 6-10, the total forecast for PEM fuel cell
related methanol demand is given, as the total supply required, and the change of
supply per year. By inspection of these graphs it becomes clear that total methanol-
demand increases continuously to a level of at some 70 MTA. The required
methanol capacity addition is 0.5 MTA/year in 2008, grows to some 1.5 MTA/year in
the from 2009 to 2013 to explode to some 5 MTA/year from 2014-2018. In this
conservative scenario, methanol demand growth would require a single
MegaMethanol53 plant to be built every two years initially. Subsequently, a single
MegaMethanol plant is required. At peak growth, three of four such plants must be
built in a single year! In the scenario, after 2018 demand growth slows down because
of technological progress and changing fuel diet.

80

60 MeOH demand
[MTA], [MTA/Yr]*10

40

20

MeOH new
0 capacity*10
2005 2010 2015 2020 2025 2030
-20
Year

Figure 6-10: Total automotive methanol demand and yearly capacity growth.

An increase in PEM fuel cell system efficiency will only imply a somewhat slower
growth in methanol demand after 2010, when the more efficient systems are put on
the road. The effects are minimal in relation to the uncertainties in market
development and mileage covered per vehicle. The shape of the curves remains the
same: methanol -demand will rise until PEM fuel cell markets have matured (in our
scenario: around 2022, i.e. in 20 years time), and then level or drop slightly to a quasi
steady-state level that is a function of technology status and fuel diet.

53 MegaMethanolTM is the tradename of the state-of-the-art methanol plant design developed by

Lurgi. The development of this process system concept represented a step-change in feasible single-
train methanol capacity. The resulting economy-of-scale has dramatically driven down methanol
production cost.
Fuel Cell Vehicles and Industry Development 209

There are three underlying questions to ask whether such development of methanol
demand represents a constraint to PEMFC vehicles market development:
1. What are the long term dynamics of the worldwide methanol market?
Can this system adequately respond to the forecast - temporary - dramatic
growth in methanol demand for PEM fuel cell related use?
2. Does automotive transport through PEM offer a “sure” future methanol outlet?
Although investments in PEM fuel cell research and development have been
large and an increasing number of companies are entering the PEM fuel cell
community, the current status of the technology is still not good enough to
make large-scale introduction a commercial success (see Appendix A. 9). This
would imply that the 'PEM outcompeted' scenario or 'PEM - the eternal
promise' become reality Figure 6-5, p. 204).
3. What are the costs and risks of a methanol distribution infrastructure?
Can a feasible strategy be developed to bring-into-being a methanol distribution
infrastructure that supports the operation of a PEMFC vehicle fleet?
(1) Long term dynamics. In the Superior PEM vehicles scenario, the methanol required
for PEM automotive fuel cell systems (Figure 6-10, p. 208) represents a completely
new type of demand. Ultimately a demand of 180 MTA methanol is created, which
is about 5 times the current methanol output levels and between 2015-2030 100 new
world scale plants would have to be realised. The effects on world oxygenate
production for the gasoline is considered to be minimal.
The present worldwide nameplate capacity for methanol is some 35 MTA, and
present level of the demand54 is some 30 MTA (Tazelaar 2001). This is a normal
balance between capacity and demand at an utilisation of some 85%. Supply and
demand data also show that with the introduction of oxygenates-enriched fuel,
methanol demand in existing fuel has steeply increased in the past two decades. In
the period 1994-2000 for example worldwide capacity increased from 26 MTA to 35
MTA, or some 1.5 MTA per year.
(2) Uncertainties At present, some older methanol capacity is being phased out, whilst
a few new large scale plants are under construction and various studies for
“Megascale” methanol have been initiated (Tazelaar 2001). There is a concern,
however, that a possible ban on MTBE in the US will lead to a situation of serious
oversupply. As a consequence, world methanol prices may become depressed just at
a time when demand from PEM fuel cell cars may be growing. Today, a new
methanol plant can be erected and operational in some 3-4 years time. The question
really is: can new capacity added match the demand growth-rate .

54 At present, methanol is used in a variety of chemical and fuel applications. In Western Europe,
only the volumes used for formaldehyde manufacture exceed those for gasoline additive production.
MTBE is the most important oxygenate that replaces lead-additives in gasoline to ensure proper
combustion (anti-knock) of the fuel. Other important -chemical- applications include the production
of acetic acid and methyl methacrylate. The chemical markets can be considered mature. The MTBE
market faces some uncertainty because of possibly forthcoming legislation banning its use in gasoline
in the U.S. Substitute oxygenates are being introduced, but some of these also require methanol.
210 Process System Innovation by Design

The forecast calculated for automotive methanol demand alone implies a definite
change of the character of methanol from chemical to fuel. PEMFC methanol
demand may have dramatic impact on the structure of the worldwide industry. The
trend towards megascale methanol plants will be accelerated and older plants that
have operating capacities well below the megascale plants may be phased out.
As in platinum, the evolution of worldwide methanol production capacity must be
carefully considered before adding new methanol capacity. Contrary to Pt, in such
an analysis, the effects of existing methanol plants being outcompeted and
subsequently being phased out must be taken into account, as well as the possible
ban on MTBE, the world's largest current methanol outlet.
(3) Distribution infrastructure. The development of methanol distribution to fuel service
stations for the consumer market has the characteristics of a catch-22; because the
number of methanol-fuelled cars is limited, it is not attractive to invest in
distribution infrastructure; because of lacking distribution infrastructure, consumers
do not invest in methanol-fuelled cars.
In recent years by introducing Shell PuraTM this company has shown that in a
relatively short time period an additional fuel outlet can created. The World
Methanol Institute (1999) has calculated the installation costs for methanol outlet
per station to some 80,000 euro55 (Stikkelman 2001). In the Netherlands the track
record of change of fuel stations is very good. Companies, for example, have
invested significantly in environmental precautions induced by legislation. The
government funded part of the changeover costs. There is no reason why a similar
scheme could not be developed for the expansion of fuel stations to include
methanol outlets.
The Netherlands is home to a total of 2600 fuel service stations (CBS). The total
cost of a maximum fine-grid of methanol distribution in the Netherlands therefore
would amount to 2,600 * 100,000 = 260 Million Euro, which is about half the
investment for a single world scale methanol plant. Together with the present
growth in natural gas supply, there is thus reason to believe that
1. The development of a methanol distribution system is no less feasible than the
development of any other fuel distribution system, and thus with careful
business development is not expected to present a problem.
2. Ample methanol supply capacity can be developed in time, and thus methanol
shortage or high prices do not by themselves constrain the development of a
fuel cell vehicle fleet.
3. The status of the technology and the pace of technological progress largely cause
uncertainty with respect to the development of methanol PEM fuel cell markets.
PEMFC vehicle development may result in methanol becoming a new 'logistic fuel',
which opens the way for dramatic investment in methanol production capacity and
distribution facilities.

55This quote includes all the necessary work and hardware per station. As a matter of course, other
costs will be incurred to create a methanol-to-consumer distribution system, e.g. for the set up or
adaptation of administrative and other information systems.
Fuel Cell Vehicles and Industry Development 211

6.6 Process system innovations for industry development

In the previous sections we have presented a functional model of 'automotive
transport' and categorized innovations that may enable its transition. The superior
PEMFC vehicles scenario considerably impacts the worldwide platinum material
cycle and the production and distribution of methanol. The dynamics involved not
only have to be catered for, however, they present tremendous opportunities for
networked process system innovations. When the industry developments lands in a
region such as the Port of Rotterdam (PoR), innovative chemical complexes may be
created.
As 'proof' of the methodology for process system innovation by design, in this
section innovations are specified for industry, which thereby may benefit from and
shape the transition to 'automotive transport' by PEM fuel cell vehicles. The results
have been obtained by combining the functional model developed (§ 6.3), the
scenario background (§ 6.4), the model results (§ 6.5) and our knowledge on metal
cycles (Verhoef et al. 2004a) and the petrochemical industry and chemical process
systems (this thesis and references cited).
Finally, the process of bringing into being specific process system innovations is
discussed.

6.6.1 PLATINUM-RELATED OPPORTUNITIES

When the functional element 'Provide Automotive Transport' (Figure 6-3, p. 193) is
realized by PEMFC vehicles, platinum is a key resource and fuel cell stacks need to
be produced, replaced and recycled. These activities are part of a global platinum
material cycle management.
Amongst others the PoR appears to be a suitable location for platinum trading,
material handling and logistics, fuel cell assembly, fuel cell recycling and (precious)
metal recovery.
While the supply of virgin-Pt is largely a concern for the countries where Pt-ore is
found, mined and processed, there is no reason whatsoever why a large-scale
production facility for PEM fuel cells or parts-thereof cannot be located in the PoR.
On the contrary, the PoR has all the logistics facilities to bring in the variety of
materials needed and the skill-level in the Netherlands is high. The PoR
management could approach the industry association FME in this respect to enquire
after developments in the metals and electronics sector.
With the successful development of Autorecycling Nederland as an example, the
presence of the London Metal Exchange in Rotterdam, and the recent
developments in metal recycling activity, the development of superior PEM vehicles
is an interesting option for Rotterdam to become a centre for platinum material
cycle management. When in Europe 500,000 PEMFC vehicles are sold, initially at
the midlife of these cars their fuel cell stacks need to be replaced and recycled. Total
vehicle recycling is at the end-of life is mandatory (10-15 years). This represents a
huge potential volume of stack recycling and metal recovery operations. Via low-
urgency shipping scrapped FC stacks and/or systems can be shipped to Rotterdam
from around the world for processing. Thus, superior PEMFC vehicles could
superbly position the PoR for a major role in the recycling of all platinum and other
metals contained in industrial and consumer goods. The associated activities
212 Process System Innovation by Design

includes collection, disassembly, transportation and storage of systems, components

and/or materials, and finally extraction to obtain the Pt and other valuable metals.
Such a role in material cycle management and the associated spectrum of activities
would largely contribute to sustainable metal use. Careful planning and management
is required, however, to effectively deal with market developments. A focus on
recycling activities, however, appears to bear limited risk. The supply forecast for
virgin Pt is that additional capacity is only required for a decade or so. This implies
all else remaining equal that plants will become obsolete. In stack recycling and
metal recovery, however, capacity must be built and is there to stay: the platinum
and metals cycle through the system at limited or no consumption of virgin material.
In addition, the capacity for Pt recovery and recycling can be realised gradually as
the market expands. It may be expected, though, that at the time when net demand
for virgin Pt collapses, the prices for Pt from recovery operations will also collapse
because it is a single market. A certificate system could prevent some of this,
labelling the Pt from virgin and recovery, e.g. to distinguish between adverse
environmental effects.
In Figure 6-11, p. 213, an overview is given of innovations in the worldwide
platinum material cycle and of associated opportunities for the Port of Rotterdam

6.6.2 METHANOL-RELATED INNOVATIONS

In the superior PEMFC vehicles scenario it is assumed that FE 1 'Convert to
suitable power source' is realized initially as methanol production. The projected
market development would imply a dramatic growth in global methanol capacity. In
addition, a distribution infrastructure (FE 2) must be realized.
In the Netherlands, a single methanol plant in Delfzijl has been in operation since
1973. The Port of Rotterdam, however, is the methanol seaport with the largest
volume of methanol landed in Western Europe and the associated jetty, storage and
facilities for distribution inland by ship, rail and road. Since the PoR offers off-
loading and storage for large ocean-going vessels, it has benefited from the shift of
worldwide methanol production towards remote locations where isolated natural gas
field are exploited successfully. Rotterdam-Rijnmond, however, also appears to be a
perfect location for a large-scale methanol production facility. Such a plant would
match the existing (petro) chemical cluster; all kinds of utility and logistics facilities
are available, which include liquid-bulk storage and a connection to the Hi-Cal
natural gas grid. Personnel of appropriate skill-level are available from the greater
Rotterdam region. Notably, natural gas pipeline capacity is expected to be available
for a world scale methanol plant because a large ammonia plant recently has been
shutdown. The only reason why a methanol plant would not be erected in
Rotterdam is the competition from low-cost methanol manufactured in remote
locations. With demand growing, due to fuel cell developments, however, capacity in
Rotterdam may become required in the not too distant future.

6.6.3 HYDROGEN INFRASTRUCTURE OPTIONS

In the superior PEMFC vehicles scenario it is assumed that FE 1 'Convert to
suitable power source' is realized ultimately in part by hydrogen or on-board
conversion of logistic fuels.
Fuel Cell Vehicles and Industry Development 213

Figure 6-11: The opportunities for the PoR in the world platinum material cycle.
214 Process System Innovation by Design

Today, the industrial markets for hydrogen represent a large-volume market56, where
regional, national and international pipeline grids for hydrogen transport have been
established. In the PoR area, for example, Air Products exploits a regional hydrogen
infrastructure, and Air Liquide exploits a 'basin' hydrogen infrastructure that spans
Rotterdam, Antwerp up to Lille in France.
The network, however, largely has the function of a backup facility for local
hydrogen production facilities, which indicates that hydrogen transport over large
distances by pipeline is still uneconomic.
Worldwide pure hydrogen is produced predominantly by steam reforming of natural
gas, which has favourable economics compared to the use of other fossil resources.
The syngas produced is subject to a number of CO-shift and purification processes.
In crude oil refineries hydrogen make-up (95% hydrogen) is presently also produced
via residue gasification followed by treatment of the product gas. Due to the nature
of refinery processes, a significant level of (CO) content is allowed, which makes
this route economic. Depending on the CO-tolerance of PEMFCs, both hydrogen
production from natural gas or oil products may become economic in the future.
Alternative sources of (relatively pure) hydrogen are a number of petrochemical
processes that include dehydrogenation, notably styrene.
The existence of hydrogen pipeline infrastructure in the PoR offers the possibility to
tie in both 'virgin' and 'waste' hydrogen production, as has been done with the Air
Liquide network, which takes in purified waste hydrogen from the Dow Terneuzen
styrene plants and the General Electric chlorine plants in Bergen op Zoom.
The possibilities with respect to hydrogen in relation to fuel cells have been
indicated in Figure 6-11 and Figure 6-12. Hydrogen is part of the precursor of
methanol, synthesis gas that is a mixture of CO, CO2 and H2. As the demand for
hydrogen increases, possible synergies may emerge in the production of synthesis
gas for methanol and hydrogen production respectively. In both cases, synthesis gas
can be produced out of a variety of resources, the synergy being that for each mix of
resources available an optimal mix of syngas qualities can be produced: for hydrogen
production the syngas H2 :CO ratio needs to be as high as possible, whilst methanol
synthesis requires an optimal ratio of 2:1, and allows the presence of some CO2.
Thus, natural gas and biomass can be combined to yield an optimal syngas
composition.
In the PoR and in the region the hydrogen can be distributed via pipeline or bottles.
A facility for small containers can readily be added to the existing facilities. In other
studies, the use of hydrogen as an additive to the Dutch natural gas grid for
households has been advocated. Hydrogen capacity thus can be built to first cater
for this market, and later to cater (partly) for the fuel cell market, thus providing a
smooth growth and/or transition. The combination of a hydrogen pipeline
infrastructure and a growing market for hydrogen may provide synergy to other
industries that generate hydrogen as a co-product that in other locations must be
used for (furnace)-fuel only.

56 The small-volume market of hydrogen is mainly catered for by bottled hydrogen in industrial
cylinders. These are transported by road-cargo. An intermediate-volume market also exists, where
hydrogen is generated at a customer-site.
Fuel Cell Vehicles and Industry Development 215

6.6.4 A 'GREEN' INDUSTRIAL COMPLEX

Using the building blocks discussed in the previous sections, a transition can be
realized towards a green methanol production, distribution and supply system. In
the green complex design (Figure 6-12, p.217) a number of 'supply-chains' is
integrated in an innovative single networked process system or green complex,
which consists of (1) natural gas to methanol, (2) natural gas to hydrogen, (3) crude
oil to oil products and (4) biomass to agro- and food products. The activities in the
'left' part of these supply-chains often are based in 'remote' locations, which range
from 'just outside the PoR' area, to 'the other end of the world'. The major
opportunity for the PoR is the provision and integration of syngas production as a
system link between resources from remote locations, the industry already located in
the Rotterdam port area and the distribution of automotive fuels and other products
to consumers.57.
The elements located in Rotterdam that are relatively unaffected by the green
complex are drawn as plain boxes in Figure 6-12. An example is the agro-and food
industry. Elements that do exist today, but which can be expanded considerably in
the development of a green complex, are indicated by a grey-fill of half the shape.
An example is the conventional production of hydrogen. Finally, elements of the
green complex that are not present in the PoR today, such as a methanol plant, have
been given a complete fill.
The production of methanol, hydrogen and oil products can be located on many
sites around the world. In practice, however, sites are selected on the basis of the
economics expected. The complex configuration presented in Figure 6-12 offers a
lot of synergies, which translate into favourable economics and ecology.
At the input-side, in Rotterdam a connection with the Dutch Hi-Cal high-pressure
natural gas grid is available. This allows the supply of Dutch, Norwegian and even
Russian gas. Other remote locations may be connected via LNG-facilities, for which
a terminal is available in the PoR. Both for hydrogen gas and liquid methanol
transfer facilities are available, as well as a distribution infrastructure and storage.
This implies that for hydrogen and methanol plants, the outside battery limit
investments can be reduced.
Kuipers argued that the industrial cluster in the PoR is characterised by limited
flexibility (Kuipers 1999). The creation of a synthesis gas complex, however, opens
the way to a more flexible cluster.
Syngas is a mixture of CO, hydrogen and CO2. It is in intermediate product, of
which each application requires a particular composition. At the production side, each
combination of feedstock (residue, natural gas, coal, biomass, waste) and technology
(reforming, partial oxidation, pyrolysis) yields a particular syngas composition, and a
varying quantity and spectrum of impurities. Thus, whenever a range of feedstock
and technologies is available at a single location, as well as a spectrum of syngas

57At the beginning of the twentieth century a similar opportunity in Rotterdam was seized by the
Bataafsche Petroleum Maatschappij. The construction of its refinery in Pernis initiated the
development of the Port of Rotterdam into one of the world's major transport hubs and conversion
centres for crude-oil and oil products. Shell Refining Company still operates what today is one of the
four worldscale oil refineries located in the Port of Rotterdam area.
216 Process System Innovation by Design

consumers, optimum flexibility can be achieved, albeit at the possibility of under

utilisation of some production units. When purification is added to the system, the
syngas product quality increases (no or few impurities, CO2 can be removed), which
results in yet more flexibility of the system. Adding purification units would allow
the production of 'clean' synthesis gas from a variety of sources. The pure hydrogen
manufactured would be suitable for chemical processes that require clean synthesis
gas or hydrogen. The syngas purification could include the removal of CO2 to
achieve the correct composition for methanol synthesis or any other required
CO/H2 ratio suitable for other chemical processes. The syngas produced could be
used to match synthesis gas produced from natural gas and for the co-production of
pure CO2.
In the PoR refineries, large-scale gasification of heavy residue from refinery
operations is already practiced. The hydrogen produced is largely used within these
refineries to 'whiten-the-barrel' and to remove the sulphur present in crude oil and
its straight-run products. These fractions from crude distilling are subjected to a
series of conversion and upgrading operations to obtain a maximum of 'white
products', viz. gasoline, kerosene and diesel out of each barrel of oil processed.
Chemically bound sulphur is converted to hydrogen sulphide in the process. In the
various refinery processes various qualities of hydrogen can be used, which can open
the way for hydrogen obtained from other sources at acceptable cost. Combining
residue gasification, which is basically a waste conversion process, with the
gasification of other hydrocarbon-containing waste streams58 could provide some
additional economy-of-scale.
These networked process system innovations would result in more optimal
realisations of fossil resource use. Eventually, however, a true sustainable industry
and energy supply must be based on renewable energy sources (solar, wind, tidal,
water, biomass) and renewable or recyclable materials (biomass, metals). Biomass,
however, today is primarily cultivated for our world food-supply, a host of agro-
industrial crops, timber and wood-supply to the pulp and paper industry. The use of
biomass for heating and electricity supply often is from natural biomass stock
(forests), and cannot be considered sustainable.
Biomass-to-energy presents little added value when the final products are heat
(space-heating) and electric power. The conversion of biomass-to-fuel, notably
methanol, however, may be economically and ecologically attractive (Stikkelman
2001). The integration of biomass conversion and hydrocarbon processing an
optimal utilisation of resources can be achieved. In a green complex, valuable
organic components may be extracted prior to the bulk conversion of biomass to
liquid fuel. Both the extraction and conversion can be by chemical process
technology or via bioprocessing.
Location of a large-scale facility in PoR would have the major advantage that
shipping of biomass waste from all-around the world can moderate regional
seasonal variations. The associated CO2 -credits (if any!) would have to split
somehow between the supplying country and the Netherlands.

58 Mixed plastic waste, paper, wood, tyres, textiles etc.

Fuel Cell Vehicles and Industry Development 217

Figure 6-12: Opportunities in the industrial infrastructure around syngas.

218 Process System Innovation by Design

Notably biomass waste presents an attractive source for final conversion to

synthesis gas because it is considered waste, and generally cannot be used for food
or other purposes anymore; it originates from the large agro-industry and can be
considered at least CO2 neutral.
At the product end the realisation of large-scale hydrogen production from
methanol is an alternative that allows the use of methanol as a long-distance
hydrogen-carrier. Location in the PoR is attractive because industrial waste-heat can
be used to run such an installation. The co-produced pure CO2 could be fed to a
local pipeline infrastructure and used for supercritical extraction, welding, urea
manufacture or to promote plant growth in nearby greenhouses.
In (Dijkema and Kuipers 2001) it was advised to the port authority to concentrate
and coordinate efforts to realise an olefins production facility, notably a naphtha
cracker in the region. The ethylene, propylene and butylenes from such a facility
would greatly enhance the strength of the existing chemical cluster, and ensure
adequate supply of these key chemicals in the region. Methanol developments to
cater for Superior PEM vehicles related demand may very well be combined with
methanol development to provide feedstock for a methanol-to-olefins (MTO)
facility in PoR (Kvisle et al. 2001). Such a plant produces largely ethylene and
propylene by pyrolysis of methanol. Only low quantities of aromatics are produced,
which is an advantage because of the glut of aromatics that have to be taken out of
the gasoline-pool. The size of the methanol market forecasted would allow the
development of a global-sized site for methanol and olefins that would cater
competitively for both methanol fuel and feedstock requirements of the region as
well as the demand for olefins. Thus, either harbour facilities, storage and other
logistics facilities may be erected in Rotterdam initially to cater for such a plant. A
local methanol plant, however, would provide a perfect physical and economic
backup or balancing facility. In the design of such a facility, there must be additional
synergies that can be exploited, e.g. the reuse and recycle of hydrogen and/or
methane produced in the MTO plant.
Thus, a number of degrees-of-freedom are introduced. These allow selection of
system element configuration and capacity that is optimal with respect to product
flexibility, economy and ecology.

6.7 Discussion and Conclusion

In this chapter we have illustrated the use of functional modelling for analysis and
innovation of sectors in relationship with their environment. Fuel cell development
for automotive transport was analyzed in a system context; functional models,
scenarios and a dynamic model were presented. We have shown that market
development, technology breakthrough, economy and ecology may not only
dramatically impact part of the petrochemical industry, they also may present
tremendous opportunities for networked process system innovation. In the previous
section, a number of such innovations have been synthesized. What are the odds
that these are actually brought into being?
Fuel Cell Vehicles and Industry Development 219

Setting the stage for MeOH-to-consumer distribution

The situation described with respect to the distribution infrastructure for methanol
is a typical catch-22 that must be breached to support the market development of
PEMFC powered vehicles.
What is the current and future role of oil companies in this? Large oil companies
typically are vertically integrated to control the entire value-chain from well to wheel.
All logistic fuels (LPG, gasoline, diesel) originate from crude oil and major oil
companies have large stakes in exploration, winning and production, refining,
wholesale and retail. This vertical integration together with the oligopolistic
character of oil production, refining and distribution presents a formidable barrier to
market entry. Exceptions noted, in many countries the majors have dominated the
logistic fuel markets for many years. Thus, introduction of an entirely new logistic
fuel, which is not derived from crude oil, appears to be difficult and risky because it
requires the realisation of an entire new production and value-chain under fierce
competition from vested players with considerable market power. Indeed,
competition between alternative fuel infrastructures is virtually non-existent.
Major oil companies, however, increasingly have become energy companies that also
have a large stake in the worldwide natural gas business, and companies like Statoil
have built their own methanol facilities, albeit primarily for captive-use of the
methanol produced in oxygenates. One may argue, however, that new style oil
companies in the near future may want to extract additional value themselves out of
the gas-to-consumer supply-chain. The supply of methanol as a logistic fuel thus can
well become part of their strategy.
Other options to bring methanol to the consumer that may be considered are forced
'market defragmentation' through legislative and regulatory reform. Some Third-
Party-Access regime to fuel service stations may provide a key element to develop
methanol distribution. Under such a regime, vested fuel service station owners or
franchisers would be obliged to allow a Third Party access to their site to sell a
different fuel than LPG, gasoline or diesel. Such a regime-change could be arranged
during or in the aftermath of auctioning concessions to exploit A-locations along the
highways etc. Thus, a possible market niche and viable entry-point is created for all
players in the natural gas, hydrogen and methanol market.

Methanol and platinum - opportunities or constraint?

We have argued that 'automotive transport' realised through methanol and PEM fuel
cell technology would not only open up a complete new 'fuel channel', but may also
enable a true transition of 'automotive transport' towards the use of non-oil derived
fuels. In addition, if successful it will have a dramatic impact on the world methanol
industry through the sheer volume of methanol required to fuel an expanding fleet
of methanol PEM vehicles. Simply put, if the technology and infrastructure
materialises and a substantial number of methanol-fuelled mileage is realised, the
associated methanol demand cannot be met by existing production capacity. New
plants will thus be built. Investors will employ the most competitive technology
available (e.g. Lurgi MegaMethanolTM), which may give the prospective owners of
these plants possibly some leverage to out compete vested methanol producers in
the petrochemical industry. In refining, depending on the fuel-of-choice for PEM
220 Process System Innovation by Design

vehicles, the fuel mix to be produced will change. Possibly, aromatics will be phased
out in logistic fuels for PEM altogether, as these require more sophisticated on-
board reforming, if at all feasible. Similarly, MTBE, which is required for anti-knock
properties (RON), may not be required anymore as a gasoline constituent.
A second important sector that will be affected is the world's platinum material cycle
(and Ruthenium) because PEM fuel cells do contain significant amounts of these
precious metals. The key parameter is the total amount of Pt per fuel cell system.
Today, the major uncertainty is the speed of technology advancement required to
bring down this amount. Backed by a scenario study, we conclude, however, that
stack lifetime and stack Pt-content do not pose a serious problem with respect to
bringing-into-being a worldwide PEM fuel cell system.
We conclude that the only real barrier to all these developments is the technological
advancement of PEM fuel cell technology and consequently the final breakthrough
of the technology to the marketplace. Not the platinum content, not the fuel
consumption, nor the emissions, but simply the total anticipated lifecycle cost to a
PEM vehicle owner is crucial to final demand for these systems. The present barrier
may very well be overcome by the already initiated introduction of small-scale
systems for vehicle auxiliary power generation. PoR must then be prepared to
facilitate the launch of the development of said synthesis gas cluster and further
develop its position in the world's platinum markets by developing the PEM Pt-
recycling infrastructure.
The only reason technology advancement is mandatory is to bring down system cost
significantly, and improve user characteristics. In our analysis, we have concluded
that, with adequate planning and management of the world's platinum material cycle,
availability and these precious metals need not be a problem.

Networked process system innovations

The main business opportunity for the chemical industry is the supply of methanol.
The scenario results in a final annual demand of 70 MTA methanol phenomenal
demand growth, with a true explosion of required added methanol capacity between
2015 and 2020.
Anticipating that some of the required methanol capacity could be realised in PoR,
new objective-defined functions of the petrochemical industry and possible system
concepts for implementation in the Rotterdam industrial cluster were explored. The
existing industrial cluster and harbour facilities would allow the realisation of unique
integrated industrial complexes around the flexible manufacture and use of synthesis
gas. At the feedstock side, a connection with the European natural gas network
already exists, but oil residue, coal, biomass and biomass waste can also be
upgraded/converted to synthesis gas. Although methanol would consume the major
share of the synthesis gas, hydrogen manufacture, metals processing, olefins and
other chemical processes could be part of the such complexes. Synergy is achieved
because all feedstock mentioned yield different qualities of synthesis gas, and a
spectrum of applications would allow a perfect match. In case of methanol,
transport, distribution and storage facilities already exist. The presence of the agro-
food industry in PoR combined with a syngas cluster could make PoR the port in
the Western hemisphere for the processing and upgrading of biomass and agro-food
Fuel Cell Vehicles and Industry Development 221

waste. The possible 'course-of-events' sketched illustrates that this could very well
provide a route towards real greening of the Port's chemical industry.

Realising networked process system innovations in Rotterdam

The development of PEM fuel cell market for a long time hinged on lack-of-funding
and lack-of-involvement of vested industry. With the forging of joint ventures
between notably Ballard Power systems, automobile manufacturers, oil companies
and other technology suppliers, the latter barrier has been removed. With these joint
ventures, also industry funding of PEM fuel cell development has increased
dramatically the past 5 years. The scenario's results indicate that ample opportunities
exist in development of the platinum material cycles, both with respect to platinum
winning and primary production, as well as platinum recovery and recycling.
In the previous section, we have outlined a number of networked process system
innovations that appear attractive for Port of Rotterdam in view of the conclusions
of a recent study for PoR (Dijkema and Kuipers 2001), were we among others
concluded that
- whilst the effects of technological developments in the petrochemical industry in
general span decades, two major developments appear to have taken-off and
impact the industry in Rotterdam at large: (1) biotechnology and (2) the
conversion of methanol to olefins.
- With respect to business developments, although there is significant capacity-
build-up in the regions that have large reserves of oil and natural gas, the
refining and petrochemical industry is to stay in Western-Europe for a period
measuring decades.
- With respect to policy developments, the environmental pressure on vested
industries from external stakeholders is a fact and it will increase. Companies are
responding adequately by increasingly effective communication with their
stakeholders.
- Regulations with respect to construction and environmental permits are
considered to be clear, stringent but just. The associated procedures can be
executed swiftly in cooperation with the authorities when regulations are met,
and thus the meeting of milestones may be anticipated in project planning.

Interpreting these developments in combination with the existing Port facilities and
Industrial Cluster in the PoR area, the timely launch of a PEMFC-AGV project can
provide a nucleus and a driver for a methanol or synthesis gas based industrial
infrastructure or complex in the Port of Rotterdam.

Functional modelling, scenario-building and dynamic modelling

In this chapter, we have used functional modelling to explore and analyse possible
alternative realisations of automotive transport. The qualitative analysis was
augmented with a quantitative assessment based on system dynamics modelling and
scenario analysis. Thus, the impact and constraints in potential transition scenarios
could be analysed and system concepts for the transition were formulated.
222 Process System Innovation by Design

Through abstraction from currently employed system concepts, we were able to

specify novel system content. Opportunities and barriers for realisation of said
options have been discussed. Thus, we have combined novel and proven technology
or system concepts in networked system designs to yield industrial complexes of
optimal structure with respect to economic and ecological performance.
The networked system innovations developed are 'proof' of the methodology for
process system innovation by design. Combined with scenario-building and dynamic
modelling, a powerful set of methods results to anticipate and respond to changes in
the world outside the industry. By realisation of the innovations developed the
chemical industry may reap the benefits of a transition in automotive transport.
 7 Conclusions and Recommendations

7.1 Introduction
Process system innovations are defined as changes in the system structure or system
design of an industry, its industrial complexes, or individual plants that can be
enabled by technological inventions or vice versa. The central theme of this thesis is
the specification of process system innovation content for a sustainable
petrochemical industry. The main result is a methodology for said specification that
is based on system representation using the input-output paradigm and functional
modelling.
In the subsequent sections, the innovations specified are reviewed. The
contributions to (engineering) science are discussed and the overall research
approach is reflected upon. Finally, recommendations are given.

7.2 Process system innovations for sustainability

A number of innovation options have been specified for olefins production. In the
steam cracker, separate dehydrogenation and cracking may increase olefin yield.
Aromatics extraction from the cracker feed could reduce methane by-product,
increase efficiency and allow direct aromatics use. Together with an aromatics
converter, the aromatics banned from logistic fuels could be effectively utilised.
The integration of olefins and syngas production from aliphatic hydrocarbons would
represent a true process system innovation, which exploits the conditions favourable
for syngas production from C5-C8 and steam cracking of the C2-C3-C4 co-products.
Trigeneration systems for the production of chemicals, electricity and heat represent
a class of process system innovations for improved resource utilisation. A
trigeneration system was developed and patented for a new type of fuel cell.
Promising candidates for trigenerate fuel cell reactors were explored using early
economic evaluation. Process-integrated applications of fuel cells in chemical plants
are feasible when creating a leverage effect on chemical plant capacity and
performance. Trigenerate petrochemical complexes significantly reduce total CO2
emissions.
The methanol capacity required for fuel cell vehicles would provide a favourable
climate for process system innovation and transition of existing petrochemical
clusters. The associated platinum use requires adequate material cycle management
for continued platinum availability. In two or three decades, clusters around
methanol and platinum may emerge that serve as the industry's infrastructure for
sustainable transport and sustainable packaging.
Together, the case study results illustrate that the methodology developed fosters
'out-of-the-box' thinking for the use of unit operations, single plants, industrial
sectors and complexes alike. It allows process system innovation content
categorisation and specification. Thus, it may help to develop a roadmap for
transition towards a sustainable petrochemical industry.
224 Process System Innovation by Design

7.3 (Engineering) Science contributions

We conjectured the process system engineering community has been largely focused
on the development of “new flow sheets for existing problems.” Sustainable
development, however, requires innovation at all system levels, from reaction path to
process system, from chemical plant to petrochemical complex or industry. In short,
sustainability requires a systems approach that addresses multiple system aggregate
levels in concert. This is the ambition of the emerging discipline of industrial
ecology, which to date, however, suffers from 'lack-of-content'. The petrochemical
industry is a mature industry where the erection of any new processing plant
represents a huge capital project that requires adequate risk management. The
industry is conservative in the adoption of novel technology or systems and often
prefers 'proven designs'. To be accepted, any innovation suggested must have sound
technological and system content. We believe that the methodology presented in this
thesis helps to meet this criterion because it is firmly embedded in chemical process
system engineering. As such, it is a system engineering foundation for industrial
ecology.
A method has been developed to translate the need for sustainable networked
industrial systems into primitive problem formulations suitable for exploration and
detailing under the General Design Paradigm. Starting from mathematical system
theory using the input-output paradigm, a coherent system representation was
formulated that links system function, behaviour and transformation to system
performance characterisation and valuation. This representation is the foundation of
a practical approach for system assessment by the combination of stream-valuation
and material/energy balancing. The combination provides a proxy to the more
complicated approach of Second Law based analysis or exergy analysis.
While system modelling and decomposition is not a trivial task, the decomposition
strategy is hardly addressed explicitly, not in studies reported in the literature and not
in textbooks. Functional modelling is a means to arrive at a coherent system
decomposition using mutual contingency as a stop-criterion when formulating
appropriate objective-defined functions of existing systems. Identification of hidden
or new, emergent objective-defined functions required for sustainable process
systems involves abstraction of the current method of 'fulfilment' or realisation of
the functions of industry or parts thereof. Innovations reported in the literature can
easily be categorised and related. The case study on olefins illustrates that when
using said abstractions an innovation space that spans multiple system levels can be
explored for novel functions and alternative ways of function 'fulfilments'. In the
specification of trigeneration systems, chemical engineering, power system
engineering and cost engineering are combined. The availability of fuel cell
technology, an innovative energy transformation, was translated into coherent
innovations involving the transport sector, petrochemicals, hydrogen and metals
production.
We developed a modelling-decomposition-synthesis strategy for primitive problem
formulation, specification and exploration of the design-space available to solve
wicked environmental problems by synthesis of sustainable systems.
Conclusions and Recommendations 225

7.4 Research approach

We conjectured that a transition towards sustainability not only requires adequately
engineered decision processes59, but also specification of process system innovation
content. Since it appeared there is a 'white-spot' in R&D with respect to this type of
innovations, our central research question was 'is it possible and worthwhile to
devise some procedure to structure the search of the process system innovation
space and to foster the specification of innovation content'.
The methods and procedure developed have been specified and refined in the
course of a number of case studies that cover a range of system aggregate levels. At
the level of single petrochemical plants, exploratory system research into the
potential use of fuel cells in the chemical industry has led to a number of patents.
This work also led to the conceptualisation of trigeneration options and a framework
for trigeneration analysis. Application of the methodology to olefins provided a
concise system description, a convenient classification of published inventions and
R&D and a number process system innovation concepts. Methanol-fuelled fuel cell
vehicles and the associated infrastructure represent a radical system innovation
initiated outside the petrochemical industry. By using the specification procedure, it
was demonstrated that these developments present a tremendous window of
innovation opportunities for the creation of a sustainable industry.
The methods and procedures developed can be labelled successful if their use results
in many innovations that foster sustainability. The case study results serve as an
illustration of the usefulness albeit not a formal proof or validation of the methods.
This is a common problem in system-oriented studies, which it shares with, for
example, macro-economics where measures are proposed to shape the economy at
large. In contrast with the natural sciences, it is not possible to conduct experiments
in an environment where all but the parameters of interest are varied. These
considerations have been coined in a meta-model of the research, which in
retrospect is a roadmap to arrive at the innovation procedure developed.
Another question is whether the innovation concepts specified could not have been
developed otherwise. The answer to this question is: yes, but the extensive literature
research conducted and reported did not reveal similar structured methods that link
creative thinking with assessment of performance, structure and technology of
existing complex systems. Functional modelling provides a structured method for
abstraction that allows one to freely consider novel realisations of existing functions
of the chemical industry at every system level of interest. It allows one to 'unfreeze'
from the 'common, established, proven, accepted' ways 'things are done'. In
combination with prioritisation of weak elements, it represents one method out of
many to address the open problem of process system innovation. Only after new
concepts have been developed, the hard work can begin to achieve their realisation
and adoption through R&D and assessment of their economic and ecological
performance.

59A premises of the M.Sc. in Systems Engineering, Policy Analysis and Management curriculum of
TU Delft is that multi-stakeholder decision processes can be engineered to address effectively a
variety if problems and opportunities that for example exhibit a varyiety of scale, complexity and
impact over a number of timescales,
226 Process System Innovation by Design

7.5 Recommendations
Focus on sustainable “supply chain level” or “networked industrial systems.” Sustainability
cannot be achieved in petrochemical plants or by promising technologies alone.
These must leverage system sustainability of supply chains or networked industrial
systems that adequately link with multiple interconnected material cycles.
Integrate or relate process system engineering and industrial ecology. The methodology
developed and the theoretical formulation provide a bridge between industrial
ecology concepts and the body-of-knowledge of process system engineering. Only
one example is the conceptual specification of a coal-based industrial complex for
base metal, hydrogen and synthesis gas production that is currently contemplated by
Port of Rotterdam and the industrial stakeholders.
Explore the implications of biotechnology for the petrochemical industry. In the long run, fossil
resources must be replaced by renewable bio-resources. If the slate of high-volume
petrochemicals remains and define the sector, the nature of the conversion
technologies must change from a chemical conversion paradigm to a biological
paradigm. Or some hybrid conversion paradigm may be adopted. The related
innovation space around biotechnology must be explored.
Efficient fossil resource use by trigeneration. Trigeneration would contribute significantly to
CO2 reduction. The steam cracker is one of many candidate processes. The
feasibility of trigeneration in liberalised and re-regulated energy markets must be
evaluated, the incentives of CO2 emission-trading elucidated whilst taking into
account the possibilities of modern furnace technology and design.
Build a coherent set of models and tools for innovation. Qualitative tests involving experts
should be developed to further underpin the approach. Quantitative evaluation
should be extended and incorporate dynamics and economics of system innovation
concepts. Thus, a portfolio of generic tools to foster innovation would result.
Transfer and use of research results in industry and society at large
The methods can be used to address 'wicked' environmental problems of industrial
sectors and society at large. They will foster the specification of innovation content
and will particularly be transferable to the sectors characterised procedurally
cohesive as the 'process industry': refining, base metal, pulp and paper, agro- and
food industry. Their use need not be limited to these sectors or sustainability!
Link to policy analysis and strategy Functional modelling must be linked to
(inter)national public policy development and regulation and corporate strategy
development. While the conceptual process system innovations require further
detailing, we trust the examples presented stimulate you as a stakeholder in any
process industry characterised by maturation to help reinvent your industry.
Incorporate the methods in transition management
As one of 'the industries of industry', a petrochemical industry is essential to achieve
sustainability. Its relation with the public at large is indirect and characterised by
lack-of-interest, discomfort and lack-of-acceptance. Use of the methods, however, is
expected to forge a link between societal needs, business strategy and process system
innovation as these can facilitate interaction in process development teams and
between stakeholders involved in a transition towards a sustainable society.
References
Abernathy, W.J. and J.M. Utterback (1978). "Patterns of Industrial Innovation." Technology
Review 80: 40-7.
Albright, L.F., B.L. Crynes and S. Nowak (1992). Novel Production Methods for Ethylene,
Light Hydrocarbons, and Aromatics. Dekker, New York.
Allenby, R.B. and J.B. Richards, Eds. (1994). The Greening of Industrial Ecosystems. National
Acadamy Press, Washington.
Alqahtany, H., D. Eng and M. Stoukides (1993). "Methane Steam Reforming Over Fe
Electrodes in a Solid Electrolyte Cell." Energy & Fuels 7(4): 495-504.
Anderson, J.L., Ed. (1996). Process Synthesis. Academic Press, New York.
Anonymous (1990). Chem-facts Netherlands. Chemical Intelligence Services, London.
Anonymous (1991). Chem-facts Ammonia. Chemical Intelligence Services, London.
Anonymous (1995a). Opportunities in Refining and Feedstocks beyond 2000. Report for Rotterdam
Municipal Port Management, ChemSystems. London.
Anonymous (1995b). Ullman's encyclopedia of Industrial Chemistry.
Anonymous (2000). Ullmann's Encyclopedia of Industrial Chemistry.
Anonymous (2001a). BP Statistical Review.
Anonymous (2001b). "Special Issue South East Asia." European Chemical News.
Anonymous (2003a). Scania Annual Report 2003. english.annualreport2003.scania.com.
Anonymous (2003b). Survey. European Environmental Agency.
Anonymous (2004). Statistical Review of World Energy. BP. www.bp.com.
Arnbak, J. and J.H. Weber (1999). Transmission systems engineering. lecture notes et4-036;
TU Delft, Faculty of Information Technology and Systems, Department of Electrical
Engineering.
Arthur D. Little (2000a). Cost Analysis of Fuel Cell System for Transportation - Baseline
Cost Estimate. Report to Dept. of Energy.
Arthur D. Little (2000b). "Distributed Generation: Understanding the Economics."
www.arthurdlittle.com.
Asbjørnsen, O.A. (1992). System Engineering Principles. Skarpodd. Houston.
Asbjørnsen, O.A. (1995). System Engineering at the Norwegian Institute of Technology.
Annual Report. Trondheim.
Asselbergs, A. (1998). "De chemische industrie ontmengt." NPT-Procestechnologie
6(November/December).
Aström, K.J. and B. Wittenmark (1984). Computer Controlled Systems. Prentice-Hall,
Englewood Cliffs, N.J.
Avramenko, Y., L. Nystrom and A. Kraslawski (2002). Selection of Internals for Reactive
Distillation Column - Case-based Reasoning Approach. European Symposium on Computer
Aided Process Engineering 12 (Escape-12), 157-162. Elsevier Science, The Hague.
Ayres, R.U. and L.W. Ayres (1996). Industrial Ecology; Towards Closing the Materials Cycle. Elgar,
Cheltenham.
Ayres, R.U. and L.W. Ayres, Eds. (2002). A Handbook of Industrial Ecology. Elgar, Cheltenham.
Azapagic, A. (1999). "Life cycle assessment and its application to process selection, design
and optimisation." Chemical Engineering Journal 73(1): 1-21.
228 Process System Innovation by Design

Bakshi, B.R. (2002). "A thermodynamic framework for ecologically conscious process
systems engineering." Computers & Chemical Engineering 26(2): 269-282.
Bansal, V., J.D. Perkins and E.N. Pistikopoulos (2002). "A case study in simultaneous design
and control using rigorous, mixed-integer dynamic optimization models." Industrial &
Engineering Chemistry Research 41(4): 760-778.
Bansal, V., J.D. Perkins, E.N. Pistikopoulos, R. Ross and J.M.G. van Schijndel (2000).
"Simultaneous design and control optimisation under uncertainty." Computers & Chemical
Engineering 24(2-7): 261-266.
Baylin, E.N. (1990). Functional Modeling of Systems. Gordon and Breach Science Publishers,
New York.
Bekkum, H.v. (2001). TU Delft. Personal Communication.
Biegler, L.T., I.E. Grossmann and A.W. Westerberg (1997). Systematic Methods of Chemical
Process Design. Prentice Hall PTR, Upper Saddle River, NJ.
Blair, R.N. and C.W. Whitston (1971). Elements of Industrial Systems Engineering. Prentice Hall,
Englewood Cliffs, NJ.
Blanchard, B.S. and W.J. Fabrycky (1998). Systems Engineering and Analysis. Prentice-Hall,
Upper Saddle River.
Bogle, D. and T. Perris (1999). "CAPE.NET - Responding to the business challenges."
Computers & Chemical Engineering 23: S779-S782.
Boot, A.H. and F.G.H. van Wees (1982). Industrial Process Heat in Relation to the
Temperature Level. ECN. Petten.
Boulding, K.E. (1956). "General Systems Theory - the Skeleton of Science." Management
Science 2 (3): 197-208.
Bouma, J.M., A.G. Cramwinckel and P.C. Klein (1996). De produktie van epichloorhydrine
volgens het Showa-Denko proces. Fabrieksvoorontwerp FVO Nr. 3176, TU Delft.
Delft.
Boustead, I. and G.F. Hancock (1979). Handbook of Industrial Energy Analysis. Horwood,
Chichester.
Bras-Klapwijk, R.M. (2001). Adjusting Life Cycle Assessment Methodology for Use in
Public Policy Discourse. Dissertation, TU Delft. Delft.
Brennan, D. (1998). Process Industry Economics: an International Perspective. IChemE,
Rugby, UK.
Bretz, R. and P. Fankhauser (1997). "Life-cycle assessment of chemical production
processes: A tool for ecological optimization." Chimia 51(5): 213-217.
Burgess, A.A. and D.J. Brennan (2001). "Application of life cycle assessment to chemical
processes." Chemical Engineering Science 56(8): 2589-2604.
Buxton, A., A.G. Livingston and E.N. Pistikopoulos (1997). "Reaction path synthesis for
environmental impact minimization." Computers & Chemical Engineering 21(1): S959-S964.
Carnot, S. (1824). Re'flexions sur la Puissance Motrice du Feu, et sur les Machines Propres a
De'velopper cette Puissance. Sadi Carnot
Carson, R.L. and L. Darling (1962). Silent Spring. Riverside Press, Cambridge, Mass.
Caserza, G., T. Bozzoni, G. Porcini and A. Pasquinucci (1996). Hydrogen Fuel Cells in
Chemical Industry: The Assemini Project. 1996 Fuel Cell Seminar, 747-749. Courtesy
Associates, Orlando, Florida, US.
CBS (1992). Energiestatistiek Nederland 1990. Voorburg.
CBS (2002). Energiestatistieken Nederland. www.cbs.nl
References 229

Charpentier, J.-C. (2002). "The triplet "molecular processes-product-process" engineering:

the future of chemical engineering ?" Chemical Engineering Science 57(22-23): 4667-4690.
Chauvel, A. (1981). Manual of Economic Analysis of Chemical Processes. McGraw-Hill,
New York.
Chauvel, A. and G. Lefebvre (1989a). Petrochemical Processes. Technical and Economic
Characteristics. 1. Synthesis-Gas Derivatives and Major Hydrocarbons. Gulf Publishing
Company, Houston, London, Paris, Tokyo.
Chauvel, A. and G. Lefebvre (1989b). Petrochemical Processes. Technical and Economic
Characteristics. 2: Major Oxygenated, Chlorinated and Nitrated Derivatives. Gulf
Publishing Company, Houston, London, Paris, Tokyo.
Chenery, H.B. and P.G. Clark (1959). Interindustry economics. John Wiley & Sons Inc., New
York.
CMAI (2000a). "CMAI completes 2000 World Light Olefins Analysis." CMAI News.
CMAI (2000b). "CMAI's World Ethylene Cost Study Examines Energy's Impact on Global
Competition." CMAI News.
CMAI (2001a). "CMAI completes World Propylene Supply Study." CMAI News.
CMAI (2001b). "CMAI Outlook for Polyolefins." CMAI News.
Cornelissen, R.J. (1997). Thermodynamics and Sustainable Development. The use of exergy
analysis and the reduction of irreversibility. Dissertation, Twente University. Enschede.
Corrigan, R.E. and R.A. Kaufman (1965). Why System Engineering. Fearon Publishers, Palo
Alto, Calif.
Daichendt, M.M. and I.E. Grossmann (1997). "Integration of hierarchical decomposition
and mathematical programming for the synthesis of process flowsheets." Computers &
Chemical Engineering 22(1-2): 147-175.
Delissen, W.k. (1997). Nuttige Toepassing of Vernietiging? Een kwantitatief criterium voor
dit onderscheid bij co-verbranding van afvalstoffen in de cementindustrie. M.Sc. thesis,
TU Delft. Delft.
Denbigh, K.G. (1956). "The second-law efficiency of chemical processes." Chemical
Engineering Science 6(1): 1-9.
Dijkema, G.P.J. (1996). Process Design for Combined Generation of Power, Heat and
Chemicals. 5th World Congress on Chemical Engineering, 373-379. AIChE, San Diego,
California.
Dijkema, G.P.J. (1998). MCFC with separate CO2 channel - Early identification of applications. Delft
University of Technology, Faculty of Technology, Policy and Management. Delft.
Dijkema, G.P.J. (1999). Fuel Cells in the Chemical Industry. Joint International Meeting.
196th Meeting of the Electrochemical Society. 1999 Fall Meeting of the Electrochemical
Society of Japan, 1013. The Electrochemical Society, Inc., Hawaii.
Dijkema, G.P.J., Ed. (2000). Mergers and Acquisitions; the truths; the tactics; the trends. Proc. 21st
Annual European AIChE Colloquium. American Institute of Chemical Engineers -
Netherlands/Belgium Section, Scheveningen, The Netherlands.
Dijkema, G.P.J., Ed. (2001a). The Chemical Industry in the New Economy. e-Business: Strategy or
Artifice. Proc. 22nd Annual European AIChE Colloquium. American Institute of
Chemical Engineers - Netherlands/Belgium Section, Scheveningen, The Netherlands.
Dijkema, G.P.J. (2001b). The Development of Trigeneration System Concepts. 6th World
Congress of Chemical Engineering, Melbourne, Australia
230 Process System Innovation by Design

Dijkema, G.P.J. and J. Grievink (2000). Technological change by system design - the
industrial production of aromatics. European Symposium On Computer Aided Process
Engineering, Escape-10, 631-636. Elsevier Science, Florence, Italy.
Dijkema, G.P.J., J. Grievink and M.P.C. Weijnen (2001). Process System Innovation for a
Sustainable Petrochemical Industry 6th World Congress of Chemical Engineering, Melbourne,
Australia.
Dijkema, G.P.J., J. Grievink and M.P.C. Weijnen (2003). "Functional Modelling for a
Sustainable Petrochemical Industry." Transactions of the Institution of Chemical Engineers Part
B: Process Safety and Environmental Protection 81(5): 331-340.
Dijkema, G.P.J. and B. Kuipers (2001). De chemie van Mainport Rotterdam - Middellange
termijn ontwikkeling van de chemie in de Rotterdamse haven: analyse, beleid en
stimulering. TU Delft, Faculteit Techniek, Bestuur en Management. Delft.
Dijkema, G.P.J., C.P. Luteijn and M.P.C. Weijnen (1998). "Design of trigeneration systems -
Process integrated applications of energy conversion devices in chemical plants." Chemical
Engineering Communications 168: 111-125.
Dijkema, G.P.J. and I.S. Mayer (2001a). Corporate Strategy and Product-Oriented
Environmental Policy. 6th World Congress of Chemical Engineering, Melbourne, Australia.
Dijkema, G.P.J. and I.S. Mayer (2001b). Information Exchange on Sustainable Product
Chains in the Dutch Chemical Industry? 3rd European Congress on Chemical Engineering,
Nuernberg, Germany.
Dijkema, G.P.J. and M.A. Reuter (1999). "Dealing with complexity in material cycle
simulation and design." Computers & Chemical Engineering 23: S795-S798.
Dijkema, G.P.J., M.A. Reuter and E.V. Verhoef (2000). "A new paradigm for waste
management." Waste Management 20(8): 633-638.
Dijkema, G.P.J., J.N. Schinkel and M.P.C. Weijnen (1994a). "Doelmatig gebruik van
grondstoffen in de procesindustrie." NPT-Procestechnologie 2(5).
Dijkema, G.P.J., J.N. Schinkel and M.P.C. Weijnen (1995). Synthese, Technologische
mogelijkheden voor een doelmatig beheer van primaire en secundaire grond stoffen, dl.
1. Interduct. Delft.
Dijkema, G.P.J., E.M. Steenkamp and P.J.T. Verheijen (1997). "Implementation of novel
processes in existing process networks: Evaluation of the industrial production of C-4
chemicals by optimization." Computers & Chemical Engineering 21: S493-S498.
Dijkema, G.P.J. and R.M. Stikkelman (1999). A Decision-Support System for Eco-Industrial
Parks - Development of an Agro-Industrial Complex -. Int. Conf. on Industrial Ecology,
Troyes, France.
Dijkema, G.P.J. and L. Stougie (1994). Environmental and economic analysis of the Veba
Oel option for the processing of mixed plastic waste. Interduct. Delft.
Dijkema, G.P.J., A.F. Veenstra, M. Vriesendorp and H.C. de Lathouder (1994b). Design of a
MCFC-upgrade kit for a steam-reforming methanol plant. 1994 International Fuel Cell
Seminar, 299-302. Courtesy Associates Inc., San Diego.
Dijkema, G.P.J., J. Vervoort, R.J.E. Daniëls and C.P. Luteijn (1996). Ammonia synthesis and
er-mcfc-technology - a profitable combination? 1996 Fuel Cell Seminar, 303-307. Courtesy
Associates Inc., Orlando, Florida.
Douglas, J.M. (1988). Conceptual Design of Chemical Processes. McGraw-Hill, New York.
Dow Chemical Benelux (1994). Schone cumeenproductie door procesontwikkeling. DSM
Milieutechnologie prijsuitreiking. TROS Aktua in Bedrijf.
References 231

Drenckhahn, W., K. Hassmann and A. Lezuo (1995). "Brennstoffzellen - eine attraktive

option fur Anwendungen in der Industrie." Brennstoff-Warme-Kraft: 386-389.
Drinkenburg, A.A.H. (1994). Personal Communication.
Drinkenburg, A.A.H. (1999). Research themes Process Engineering. www.tue.nl/research.
Eastman, A.D., J.H. Kolts and J.B. Kimble (1990). "Catalytic Production of Ethylene at
Phillips .2. Ethane Oxidative Dehydrogenation." Abstracts of Papers of the American Chemical
Society 199: 72-IAEC.
Edgar, T.F., D.A. Dixon and G.V. Reklaitis (1999). Vision 2020: Computational Needs of
the Chemical Industry. Impact of Advance in Computing and Communication Technologies on
Chemical Science and Technology. National Research Council, National Academic Press,
Washington D.C.
Edgar, T.F., D.M. Himmelblau and L.S. Lasdon (2001). Optimization of chemical processes.
McGraw-Hill, Boston.
Emig, G., H.-J. Krämer and J. Weitkamp, Eds. (2001). "Creating Value from Light Olefins -
Production and Conversion" Proceedings of the DGMK-Conference. DGMK
Tagungsbericht, Hamburg, Germany.
Fahy (1990). Methanol: fuel or chemical. Financial Times Business Publications.
Farooque, M. (1988). Apparatus and method for methanol production using a fuel cell to
regulate the gas composition entering the methanol synthesizer. US Pat. 19860892246
19860731. Energy Research Corporation. US.
Fellows, R., E.W. Sloetjes and R. Ottervanger (1998). "Stack Networking for System
Optimisation: an Engineering Approach." J. Power Sources 71(1-2): 138-143.
Ferguson, P.R. and G.J. Ferguson (1988). Industrial Economics. Issues and Perspectives. The
Macmillan Press Ltd., 2nd London.
Floudas, C.A. (1995). Nonlinear and Mixed-Integer Optimization; Fundamentals and
Applications. Oxford University Press, New York.
Fonyo, S. and P. Mizsey (1990). A global approach to the synthesis and preliminary design
of integrated total flowsheets. Annual AIChE Meeting, Chicago.
Franck, H.G. and J.W. Stadelhofer (1988). Industrial aromatic chemistry; raw materials,
processes, products. Springer, Berlin.
French, M.J. (1999). Conceptual Design for Engineers. Springer, 3rd London.
Freshwater, D. (1997). People, pipes and processes; a short history of chemical engineering
and the Institution of Chemical Engineers. Institute of Chemical Engineers, Rugby.
Frosch, R.A. and N.E. Gallopoulos (1989). "Strategies for Manufacturing." Scientific American
261(3): 144-152.
Frydenlund, F.A. (1993). Method of Preparing Ammonia. EP 0423177. Haldor Topsoe,
Denmark.
Geissler, K., E. Newson, F. Vogel, T.B. Truong, P. Hottinger and A. Wokaun (2001).
"Autothermal methanol reforming for hydrogen production in fuel cell applications."
Physical Chemistry Chemical Physics 3(3): 289-293.
Georgescu-Roegen, N. (1991). Thermodynamics and We, the Humans. AES vol. 25/HTD-
Vol. 191 Second Law Analysis - Industrial and Environmental Applications, 115-125.
ASME.
Gielen, D. (1999). Materialising Dematerialisation. Integrated Energy and Materials Systems
Engineering for Greenhouse Gas Mitigation. Dissertation, TU Delft. Delft.
232 Process System Innovation by Design

Glasser, D. and D. Hildebrandt (1997). "Reactor and process synthesis." Computers &
Chemical Engineering 21: S775-S783.
Goossens, J.C.J.M. (1995). Conceptual Design of a Fuel Cell System for an Electrothermal Phosphorus
Process. Faculty of Chemical Engineering and Material Science. Delft.
Govaerts, F. (1999). Largest air-separator in the World. Presentation. AIChE-NLB company
visit to Air Products, Rotterdam-Botlek,
Graedel, T.E. (2003). Cross-scale Analysis of the Contemporary Copper and Zinc Cycles.
ISIE '03: International Society for Industrial Ecology Second International Conference, 1.
International Society for Industrial Ecology, Ann Arbor, Michigan MI, USA.
Graedel, T.E. and R.B. Allenby (1995). Industrial Ecology. Prentice Hall, Englewood Cliffs,
New Jersey.
Greer, D., R. Houdek, R. Pittman and J. Woodcock (2001). Maximizing Propylene Yields in
Fluid Catalytic Cracking and Related Technologies - Solutions for Increasing Propylene
Production. DGMK, Hamburg, Germany.
Grievink, J. (1994). Ontwerpen voor Procesvoering, Schakel tussen Molekuul en Markt. TU
Delft.
Grievink, J. (1998). Discussion on Symbolic System Representation.
Grievink, J. (2003). Symbolic System Representation.
Grievink, J. and J.v. Schijndel, Eds. (2002). European Symposium on Computer Aided Process
Engineering 12. Elsevier, Amsterdam.
Grigg, J.B., Ed. (1974). The Agricultural Systems of the World - an Evolutionary Approach.
Cambridge University Press.
Grossmann, I.E., J.A. Caballero and H. Yeomans (2000). "Advances in mathematical
programming for the synthesis of process systems." Latin American Applied Research 30(4):
263-284.
Grossmann, I.E. and A.W. Westerberg (2000). "Research Challenges in Process System
Engineering." AIChE Journal 46(9): 1700-1703.
Grubbard (1992). Energie & Milieuspectrum.
Grunert, K.G., H. Harmsen, M. Meulenberg, E. Kuiper, T. Ottowitz, F. Declerck, B. Traill
and G. Goransson (1997). A Framework for Analysing Innovation in the Food Sector.
Product and Process Innovation in the Food Industry. B. Traill and K. G. Grunert, Eds.: 1-33.
Blackie Academic and Professional, London.
Hamel, G. and C.K. Pralahad (1994). Competing for the future. Harvard Business School Press,
Boston.
Hampden-Smith, M. (2000). The Design & Performance of Ultra Low Pt Electrocatalysts
produced by spray based routes for fuel cell applications,. paper presented by Superior
MicroPowders, LLC, Albuquerque, 2000.
Hartmann, K. and K. Kaplick (1990). Analysis and Synthesis of Chemical Process Systems. Elsevier,
Amsterdam.
Heijnen, P.W. and Z. Verwater-Lukszo (2003). Batch Plant Integration in Need of more
Support. Foundations of Computer Aided Process Operation FOCAPO-2003, Florida.
Hemmes, K. and G.P.J. Dijkema (1998). Method of operating a molten carbonate fuel cell, a
fuel cell, a fuel cell stack and an apparatus provided therewith. TU Delft, NL-2628 BL
Delft.
Hemmes, K., W.H.A. Peelen and G.P.J. Dijkema (1998). New MCFC with separate CO2
supply, Proof of Principle. 1998 Fuel Cell Seminar.
References 233

Herder, P.M. (1999). Process Design in a Changing Environment: Identification of Quality

Demands Governing the Design Process. Delft University Press, Delft.
Herder, P.M. and M.P.C. Weijnen (1999). "Assessment of the quality of the design process
and the design of chemical plants with Piquar." Computers & Industrial Engineering 37(1-2):
125-128.
Hirschofer, J.H., D.B. Stauffer and R.R. Engleman (1994). Fuel Cells, a Handbook.
Morgantown.
Hohmann, E.C. (1971). Optimum Networks for Heat Exchange. Dissertation, University of
Southern California.
Homburg, E., A.S. Travis and H.G. Schröter, Eds. (1998). The Chemical Industry in
Europe, 1850-1914: Industrial Growth, Pollution, and Professionalization. Kluwer
Academic, Dordrecht.
Hounshell and M.J. Smith (1989). R&D at DuPont.
Hutcheson, P., A.W. Pearson and D.F. Ball (1995). "Innovation in Process Plant: A Case
Study of Ethylene." Journal of Product Innovation Management 12(5): 415-430.
IChemE (2003). What is Chemical Engineering? www.icheme.org.
INCOSE (2004). A Consensus of the INCOSE Fellows.
http://www.incose.org/practice/fellowsconsensus.aspx.
IPCC (2001). IPCC Third Assessment Report - Climate Change 2001. www.ipcc.ch.
Ismail, S.R., P. Proios and E.N. Pistikopoulos (2001). "Modular synthesis framework for
combined separation/reaction systems." AIChE Journal 47: 629-649.
Jennings (1991). Ammonia.
Jungerhans, R.J. (1984). Process for Ammonia Syngas Manufacture.
Kaibel, G. and H. Schoenmakers (2002). Process Synthesis and Design in Industrial Practice.
European Symposium on Computer Aided Process Engineering 12 (Escape-12), 9-22. Elsevier
Science, The Hague.
Kalitventzeff, B., F. Marechal and H. Closon (2001). "Better solutions for process
sustainability through better insight in process energy integration." Applied Thermal
Engineering 21(13-14): 1349-1368.
Ketelaar, J.A.A. (1968). Ökonomische Betrachtungen zur Wahl der Brennstoffe fur
Brennstoffelemente. 289-298. Akademie Verlag, Berlin.
Khan, F.I., B.R. Natrajan and P. Revathi (2001). "GreenPro: a new methodology for cleaner
and greener process design." Journal of Loss Prevention in the Process Industries 14(4): 307-328.
Kniel, L., L. Winter and K. Stork (1980). Ethylene, Keystone to the Petrochemical Industry, . Marcel
Dekker Inc., New York.
Kondepudi, D. and I. Prigogine (1998). Modern Thermodynamics: from Heat Engines To
Dissipative Structures. Wiley, Chichester.
Kondili, E., C.C. Pantelides and R.W.H. Sargent (1993). "A general algorithm for short-term
scheduling of batch operations-I: MILP formulation." Computers & Chemical Engineering
17(2): 211-227.
Konijn, P. (1994). The make and use of commodities by industries: on the compilation of
input-output data from the national accounts. Dissertation, Twente University.
Kordesch, K. and G. Simader (1996). Fuel Cells and their Applications. VCH, Weinheim, New
York.
Kotas, T.J. (1985). The Exergy Method of Thermal Plant Analysis. Butterworths, London.
234 Process System Innovation by Design

Kotler, P. (1991). Marketing Management - Analysis, Planning, Implementation, and

Control. Prentice-Hall, 7th London.
Kuipers, B. (1999). Flexibiliteit in de Rotterdamse Havenregio. Dissertation, TU Delft. Delft.
Kussi, J., R. Perne and A. Schuppert (2002). Process Software in the Chemical Industry -
The Challenge of Complexity. European Symposium on Computer Aided Process Engineering 12
(Escape-12), 23-30. Elsevier Science, The Hague.
Kvisle, S., H. Reier Nilsen, T. Fuglerud, A. Grønvold, B.V. Vora, P.R. Pujadó, P.T. Barger
and J.M. Andersen (2001). Methanol to Olefins (MTO): State of the Art and
Perspectives. DGMK, Hamburg, Germany.
Lance, J.R. (1984). "The Economics of a Fuel Cell Cogeneration System Using a By-Product
Hydrogen Stream." Int. J. Hydrogen Energy 8: 219-224.
Li, X.N. and A. Kraslawski (2004). "Conceptual process synthesis: past and current trends."
Chemical Engineering and Processing 43(5): 583-594.
Linke, P. and A.C. Kokossis (2004). "Advanced process systems design technology for
pollution prevention and waste treatment." Advances in Environmental Research 8(2): 229-
245.
Linnhoff, B. (1993). "Pinch Analysis - a State-of-the-Art Overview." Chemical Engineering
Research & Design 71(A5): 503-522.
Linse, K. (2002). Performance Excellence - An Overview. 23rd Annual European AIChE
Colloquium, Kurhaus, Scheveningen.
Lipsey, R.G. and P.O. Steiner (1991). Macro Economics.
Loffel, H. and M. Schulz (1984). "Gas-Turbine for Energy Supply-Systems of an Industrial-
Plant." Brennstoff-Warme-Kraft 36(6): 243-248.
Malhotra, S. and R. Dattra (1996). "Feasibility Studies of a Fuel Cell for Cogeneration of
Homogeneously Catalyzed Acetaldehyde and Electricity from Ethanol." J. Electrochem.
Soc. 143(10): 3058-3065.
Marchetti, C. and N. Nakicenovic (1979). The Dynamics of Energy Systems and the Logistic
Substitution Model. IIASA. Laxenburg, Austria.
Marcoulaki, E.C. and A.C. Kokossis (2000). "On the development of novel chemicals using
a systematic synthesis approach. Part I. Optimisation framework." Chemical Engineering
Science 55(13): 2529-2546.
Maslow, A.H. (1954). Motivation and Personality. New York/London.
Mayer, I.S. and G.P.J. Dijkema (1999). Naar een infrastructuur voor de uitwisseling van
productgerelateerde milieugegevens : een strategische analyse van de
gegevensuitwisseling in acht bedrijfssectoren en een agenda voor vernieuwende
initiatieven. TU Delft, Faculteit TBM. Delft.
Mazanec, T.J. (1993). "Apparatus for Hydrocarbon Fuel Processing Electropox: BP's Novel
Oxidation Technology " Act. Dioxygen Homogeneous Catal. Oxid.: 85-96.
Meadows, D.H., D.L. Meadows and J. Randers (1972). The Limits to Growth; a report for
the Club of Rome's project on the predicament of mankind New American Library, New
York.
Meadows, D.H., D.L. Meadows and J. Randers (1992). Beyond the Limits to Growth. Confronting
Global Collapse, Envisioning a Sustainable Future. Chelsea Green Press, Durham, New
Hampshire.
Mellor, W., E. Wright, R. Clift, A. Azapagic and G. Stevens (2002). "A mathematical model
and decision-support framework for material recovery, recycling and cascaded use."
Chemical Engineering Science 57(22-23): 4697-4713.
References 235

Melman, A.G., H. Boot and G. Gerritse (1991). Energiebesparingspotentiëlen 2015. TNO.

Apeldoorn.
Mol, A.P.J. (1995). The Refinement of Production, Ecological Modernization Theory and
the Chemical Industry. van Arkel, Utrecht.
Molle, W. and E. Wever (1984). Oil Refineries and Petrochemical Industries in Western
Europe, Buoyant Past, Uncertain Future. Gower Publishing Company Ltd., England.
Montfoort, A.G. (1992). Over ontwerpen gesproken.... . Afscheidsrede TU Delft. Delft.
Montfoort, A.G. (1993). Personal Communication. Delft.
Montfoort, A.G., F.A. Meijer and A.v.d. Ham (1989). De Chemische Fabriek. Dl. 1A. Flowsheet-
theorie en Ontwerp; . TU Delft, Faculteit der Scheikundige Technologie en der
Materiaalkunde, Vakgroep Chemische Technologie. Delft.
Mulderink, J., W. van Heugten, L. Jansen, C. Bakker, H. Bouwmeester, T. Kievid, G. van
Grootveld and P. Vergragt, Eds. (1997). DTO Sleutel Chemie. Zon en biomassa: bronnen van de
toekomst. Interdepartementaal onderzoeksprogramma Duurzame Technologische
Ontwikkeling, Den Haag.
Nauman, E.B. (1988). Introductory Systems Analysis for Process Engineers. Butterworth-
Heinemann, Stoneham.
Nelson, J.H. (1972). Methodology for the Systems Engineering Process. Vol. 1. System
Functional Activities. NTIS. Springfield.
NRC (1988). Frontiers in Chemical Engineering. Research Needs and Opportunities.
National Academy Press, Washington D.C.
OECD (1999). World Energy Outlook. www.oecd.org.
Overwater, K. (2001). Technip; Personal Communication.
Papalexandri, K.P. and E.N. Pistikopoulos (1996). "Generalized modular representation
framework for process synthesis." AIChE Journal 42: 1010-1032.
Pavitt, K. (1984). "Sectoral Patterns of Technical Change: Towards a Taxonomy and a
Theory." Research Policy 13: 343-373.
Peters, M.S. and K.D. Timmerhaus (1991). Plant Design and Economics for Chemical Engineers.
McGraw-Hill, New York.
Piduhn, M. (1999). Metallocen-katalysierte Polymerisation von Propen und Ethen aus der
Gasphase. Mensch-und-Buch, Berlin.
Pierucci, S., Ed. (2000). European Symposium on Computer Aided Process Engineering 10.
Computer-Aided Chemical Engineering. Elsevier, Amsterdam.
Pinto (1977). Methanol. US4065483. ICI, United Kingdom.
Pisano, G. (1997). the Development Factory; Unlocking the Potential of Process
Innovation. Harvard Business School Press, Boston
Planck, M. (1926). Treatise on Thermodynamics. New York, Dover.
Prigogine, I. and I. Stengers (1984). Order out of Chaos: Man's New Dialogue with Nature. Bantam
Books, Toronto.
Quakernaat, J. (1981). Chemie en Kolen. Een overzicht van mogelijkheden in Nederland.
ECN, The Netherlands. Petten.
Reuter, M.A. (1998). "The simulation of industrial ecosystems." Minerals Engineering 11(10):
891-918.
Reuter, M.A. and G.P.J. Dijkema (2003). Addressing LCA. PoliWaste Proposal. BGRM, Ed.
Rothaemel, M. and H.-D. Holtmann (2001). MTP, Methanol to Propylene - Lurgi's Way.
DGMK, Hamburg, Germany.
236 Process System Innovation by Design

Roussel, P.A., K.N. Saad and T.J. Erickson (1991). Third generation R and D; managing the link
to corporate strategy. Harvard Business School Press, Boston.
Rowe, D.A., J.D. Perkins and S.P. Walsh (1997). "Integrated design of responsive chemical
manufacturing facilities." Computers & Chemical Engineering 21: S101-S106.
Rudd, D.F., S. Fathi-Afshar, A.A. Trevino and M.A. Stadtherr (1981). Petrochemical Technology
Assessment. Wiley, New York.
Rudd, D.F., G.J. Powers and J.J. Siirola (1973). Process Synthesis. Prentice-Hall, Englewood
Cliffs.
Rudd, D.F. and C.C. Watson (1968). Strategy of Process Engineering. John Wiley & Sons. Inc.,
New York.
Ryu, J.H., V. Dua and E.N. Pistikopoulos (2004). "A bilevel programming framework for
enterprise-wide process networks under uncertainty." Computers & Chemical Engineering
28(6-7): 1121-1129.
Sargent, R.W.H. (1998). "A functional approach to process synthesis and its application to
distillation systems." Computers & Chemical Engineering 22(1-2): 31-45.
Sas, H., J.P. van Soest, M.J. Zegwaard and G.P.J. Dijkema (1994a). Disposal of plastic household
waste: analysis of environmental impacts and costs. Centrum voor Energiebesparing en Schone
Technologie. Delft, The Netherlands.
Sas, H., J.P. van Soest, M.J. Zegwaard and G.P.J. Dijkema (1994b). Verwijdering van
huishoudelijk kunststofafval: analyse van milieu-effecten en kosten, hoofdrapporten en
bijlagen. CE. Delft.
Satterfield, C.N. (1991). Heterogeneous catalysis in industrial practice. McGraw-Hill, 2nd New
York.
Schinkel, J.N., G.P.J. Dijkema and M.P.C. Weijnen (1994). Novel opportunities for fuel cell
commercialization: process integrated applications,. International Conference 'Bringing Fuel
Cells down to Earth', 239-249. South Coast Air Quality Management District, Los Angeles.
Scholes, G. (1999). Integrating Sustainable Development into the Shell Chemicals Business.
Environmental Performance Indicators in Process Design and Operation. M. P. C. Weijnen and P.
M. Herder, Eds. EFCE Event 616: 69-82. Delft University Press, Delft.
Schumpeter, J. (1942). Capitalism, Socialism and Democracy. Harper and Row, New York.
Seader, J.D. (1982). Thermodynamic Efficiency of Chemical Processes. Massachusetts Institute of
Technology,
Sederquist, R.A., J.C. Trocciola, J.F. Farris and M.J. Smith (1993). Method of Generating High-
purity Nitrogen Gas. US5330857. Int. Fuel Cells Corporation (US).
Seider, W.D., J.D. Seader and D.R. Lewin (1999). Process Design Principles: Synthesis, Analysis,
and Evaluation. Wiley, New York.
Setac (1994). LCA manual oid.
Sharratt, P. (1999). "Environmental criteria in design." Computers & Chemical Engineering
23(10): 1469-1475.
Sheldon, R.A. (1993). Selectiviteit in de Organische Synthese: Schoonheid in de Chemie. TU
Delft, Delft.
Sherwood, T.K. (1963). A Course in Process Design. The MIT Press, Cambridge, MA.
Siirola, J.J. (1996). Industrial Applications of Chemical Process Synthesis. Process Synthesis. J.
L. Anderson, Ed. 23: 2-61. Academic Press, New York.
Simon, H.A. (1969). Science of the Artificial. MIT Press, Cambridge, MA.
Sinclair, J.M., Ed. (1995). Collins Cobuild English Dictionary. HarperCollins Publ., London.
References 237

Smith, J.M., H.C. Van Ness and M.M. Abbott (2001). Introduction to chemical engineering
thermodynamics. McGraw-Hill, 6th New York.
Smith, R. (1995). Chemical Process Design. McGraw-Hill, New York.
Smits, D.F. and G.P.J. Dijkema (1994). Techniekbeschrijvingen, opzet en inhoud. Interduct. Delft :
Interduct.
Spillman, R.W., R.M. Sotnitz and J.T. Lundquist (1984). "Why not make chemicals in fuel
cells." ChemTech 15: 226.
Spiro, T.G. and W.M. Stigliani (1996). Chemistry of the Environment. Prentice-Hall, Upper
Saddle River, NJ.
SRI (1989). Mass Conversion Factors of Industrial Chemical Processes. Stanford Research
Institute, Calif.
Stadtherr, M.A. and D.F. Rudd (1976). "Systems study of the petrochemical industry."
Chemical Engineering Science 31(11): 1019-1028.
Stadtherr, M.A. and D.F. Rudd (1978). "Resource use by the petrochemical industry."
Chemical Engineering Science 33(7): 923-933.
Stanley, S.J. (2001). Recent Advances in the Manufacture of Light Olefins by Steamcracking.
DGMK, Hamburg, Germany.
Steffens, M.A., E.S. Fraga and I.D.L. Bogle (1999). "Multicriteria process synthesis for
generating sustainable and economic bioprocesses." Computers & Chemical Engineering
23(10): 1455-1467.
Stephanopoulos, G. (1984). Chemical Process Control. Prentice-Hall, Englewood Cliffs.
Stikkelman, R.M. (2001). Groene Methanol uit Groningen. Interduct rapport. Delft.
Stobaugh (1988). Innovation and Competition. The global management of petrochemical
products. Harvard Business School Press, Boston.
Stull, D.R., E.F. Westrum and G.C. Sinke (1969). The Chemical Thermodynamics of Organic
Compounds. Wiley, New York.
Südholter, S., M.A. Reuter and J. Kruger (1996). "Eco-techno-economic synthesis of process
routes for the production of zinc using combinatorial optimization." Metallurgical and
Materials Transactions B-Process Metallurgy and Materials Processing Science 27(6): 1031-1044.
Tazelaar, F. (2001). Nieuwe methanolcapaciteit 2001-2006. Memorandum RMPM.
Thompson, R.G., J.A. Colloway, L. Nawalanic and J. Becker (1978). The Cost of Energy and a
Clean Environment Gulf, Houston.
Truxal, J.G. (1972). Introductory System Engineering. McGraw-Hill, New York.
Twiss, B.C. (1992). Managing Technological Innovation. 4th edition, Pitman Publishing, London.
van Breda, L.M. and G.P.J. Dijkema (1998). "Eia's contribution to environmental decision-
making on large chemical plants." Environmental Impact Assessment Review 18(4): 391-410.
van den Hoed, R. (2004). Driving Fuel Cell vehicles. How Established Industries React to
Radical Technologies. Dissertation, Delft University of Technology. Delft.
van der Graaf, M.H.K. (1980). Psychologische Aspekten van de Organisatie. Samsom, Alphen a/d
Rijn, Brussel.
van Geem, P. (1994). "Methanol to Olefins." NPT Procestechnologie.
van Klaveren, R. (2003). BDO and derivatives business Europe. Presentation, AIChE-NLB
company visit to Lyondell Rotterdam-Botlek, Spijkenisse
van Schaik, A. and M.A. Reuter (2004). "The time-varying factors influencing the recycling
rate of products." Resources Conservation and Recycling 40(4): 301-328.
238 Process System Innovation by Design

van Zanten, D.J. and G.P.J. Dijkema (2002). Public roles and private interests in
petrochemical clusters - Model-based decision support of a Regional Development
Agency. Internal Research Paper, TU Delft, Fac. TBM. Delft.
Vanderborght, B. and G.P.J. Dijkema (1996). Perspectives for the use of combustible waste in the
Belgian and French cement industries. School of Systems Engineering, Policy Analysis and
Management / Interduct, Delft University of Technology.
Vayenas, C.G. (1988). "Catalytic and electrocatalytic reactions in solid oxide fuel cells." Solid
State Ionics 28-30: 1521-1539.
Vayenas, C.G., S. Bebelis, I.V. Yentekakis and H.G. Lintz (1992). "Non-Faradaic
Electrochemical Modification of Catalytic Activity - a Status-Report." Catalysis Today
11(3): R5-438.
Vayenas, C.G., S.I. Bebelis and C.C. Kyrizais (1991a). ChemTech 21: 50.
Vayenas, C.G., S.I. Bebelis and C.C. Kyrizais (1991b). "Cogeneration - Electricity +
Chemicals .1. Solid Electrolytes." ChemTech 21(7): 422-428.
Verhoef, E.V., G.P.J. Dijkema and M.A. Reuter (2001). Managing the Dynamics of Waste
Management. Proc. 3rd Annual Symposium Delft Interfacultary Research Center Design and
Management of Infrastructures 125-152. Delft University Press.
Verhoef, E.V., G.P.J. Dijkema and M.A. Reuter (2002a). Waste co-incineration in the
European Community - is the sky the limit? Int. Conf. on Resources, Recovery and Recycling
RRR2000, 1028-1036. Toronto, Canada.
Verhoef, E.V., G.P.J. Dijkema and M.A. Reuter (2004a). "Knowledge, waste management
and metal ecology." Journal of Industrial Ecology 8(1-2).
Verhoef, E.V., J.A. van Houwelingen, G.P.J. Dijkema and M.A. Reuter (2002b). Industrial
Ecology and Flexible Waste Infrastructure. 5th Int. Conf. on Technology, Policy and Innovation,
The Hague, Delft2001 Critical Infrastructures, 238 and CD-ROM. Lemma, Utrecht, The
Netherlands.
Verhoef, E.V., J.A. van Houwelingen, G.P.J. Dijkema and M.A. Reuter (2004b). "Industrial
Ecology for Waste Infrastructure Development. A roadmap for the Dutch Waste
Management System." Technological Forecasting and Social Change In press.
Villermaux, J. (1996). Opening Address. 5th World Congress of Chemical Engineering, AiChE, San
Diego, California.
VNCI (1989). Integrated Substance Management. VNCI. Leidschendam.
VNCI (1998). Jaarverslag. Leidschendam.
Vogel, F. (2001).
von Bertalanffy, L. (1951). "Problems of General Systems Theory." Human Biology.
von Weiszacker, E.U. (1996). Factor Four: Doubling Wealth, Halving Resource Use - A
Report to the Club of Rome. Earthscan, London.
VROM (1972). UrgentieNota Milieuhygiene. Tweede Kamer 1971-1972, 11906, Den Haag
Weber, A. (1909). Über den Standort der Industrien DII. Tübingen.
Wei, J., T.W.F. Russell and M.W. Swartzlander (1979). The Structure of the Chemical Processing
Industries, Function and Economics. McGraw-Hill, Inc., New York.
Weijnen, M.P.C. (1991). Naar een doelmatiger beheer van primaire en secundaire
grondstoffen. Projectvoorstel, Interduct. Delft.
Weijnen, M.P.C. (1999). The Greening of Natural Gas. Research Proposal, Faculty of
Technology, Policy and Management. Delft.
References 239

Weijnen, M.P.C. and A.A.H. Drinkenburg, Eds. (1993). Precision Process Technology, Perspectives
for pollution prevention. Kluwer Academic Publishers, Dordrecht.
Weissermel, K. and H.-J. Arpe (1994). Industrielle organische Chemie; bedeutende Vor- und
Zwischenprodukte. VCH, 4. Aufl. Weinheim.
Welford, R. (1997). Hijacking Environmentalism: Corporate Responses to Sustainable
Development. Earthscan, London.
Westerberg, A.W. (1989). "Synthesis in Engineering Design." Computers & Chemical
Engineering 13(4-5): 365-376.
Westerberg, A.W. (2004). Comment on the design course thought by Prof. T.K. Sherwood.
Westerberg, A.W. and O.M. Wahnschafft (1996). Synthesis of Distillation-Based Separation
Systems. Process Synthesis. J. L. Anderson, Ed. 23: 64-170. Academic Press, New York.
Wetenschappelijke Raad voor het Regeringsbeleid (2003). Naar nieuwe wegen in het milieubeleid.
Sdu Uitgevers, Den Haag.
Whitehead (2000). "Looking from the outside in: Achieving value growth in asset-intensive
industries." Chimica Oggi 18(5).
Witlox, F.J.A. (1998). Modelling Site Selection : a Relation Matching Approach Based on
Fuzzy Decision Tables. Dissertation, TU Eindhoven, Faculteit Bouwkunde. Eindhoven.
Woltjer, G. (2002). De economische manier van denken. 2e druk, Coutinho, Bussum.
Wood, D., Ed. (2001). Proceedings of the 6th World Congress on Chemical Engineering,
Melbourne, Australia.
World Commission on Economic Development (1987). Our Common Future. Oxford
University Press, Oxford, UK.
Worrell, E. (1994). Potentials for improved use of industrial energy and materials.
Dissertation, Utrecht University. Utrecht.
WRI (2003). 2004: Decisions for the Earth: Balance, voice and power. World Resources 2002-2004.
World Resources Institute. governance.wri.org.
Yang, Y. and L. Shi (2000). "Integrating environmental impact minimization into conceptual
chemical process design -- a process systems engineering review." Computers & Chemical
Engineering 24(2-7): 1409-1419.
Yee, T.F. and I.E. Grossmann (1990). "Simultaneous-Optimization Models for Heat
Integration .2. Heat-Exchanger Network Synthesis." Computers & Chemical Engineering
14(10): 1165-1184.
Yee, T.F., I.E. Grossmann and Z. Kravanja (1990). "Simultaneous-Optimization Models for
Heat Integration .1. Area and Energy Targeting and Modeling of Multi-Stream
Exchangers." Computers & Chemical Engineering 14(10): 1151-1164.
Yentekakis, I.V. and C.G. Vayenas (1989). "Chemical Cogeneration in Solid Electrolyte Cells
- the Oxidation of H2s to So2." Journal of the Electrochemical Society 136(4): 996-1002.
Zeppenfeld, R. and R. Walzl (2001). Steamcracker Revamp Projects: Challenges and
Technologies. DGMK, Hamburg, Germany.
Zevnik, F.C. and R.L. Buchanan (1963). Chem. Eng. Progr. 59(2): 70.
Symbols

Chapter 1
I total plant Investment
C plant capacity
p parameter

Suffix denotes:
0 standard or reference condition

Chapter 3
A Availability [J]
B By-product output ( Stream label )
B behaviour of a system
D Domain
∆ Difference or change (∆G, ∆H, ∆U, ∆S)
E Emission output ( Stream label )
Ex Exergy [J]
ηa Reference efficiency
ηb Product efficiency
ηc Emission efficiency
ηd Total emission efficiency
H Enthalpy [ J/mole ]
I Inputs ( Stream label )
Λ system boundary
Σ sum
Σ system
σ set of streams
σ Entropy created [ J/K ]
S Scarce resources input ( Stream label )
S Entropy [ J/K ]
Su set of inputs
Sy set of outputs
S1 set of objective-defined functions
S2 set of objective-defined functions
T Technology
T Ubiquities input ( Stream label )
T Temperature [K]
t time [ time unit ]
κu,y(t) vectors of conversion factors [ unit/flow unit ]
κ• net accumulation of chemical element, mass or energy [ unit ]
La Reference Loss [ unit ]
Lb Product Loss [ unit ]
Lc Emission Loss [ unit ]
Ld Total Emission Loss [ unit ]
Le Balance Loss [ unit ]
M Main Product output ( Stream label )
242 Process System Innovation by Design

O Outputs ( Stream label )

Ω performance of a system
P value function
P Pressure [ N/m2 ]
Φ function of a system
φ flow [ unit / time ]
Ψ transformations within a system
q heat [J]
U Internal Energy [J]
u(t) vector of input flows [ flow unit/time ]
V Volume [m3]
W Waste inputs ( Stream label )
w Work [J]
X general variable
Y general variable
y(t) vector of output flows [ flow unit/time ]
Z general variable

Suffix denotes:
0 ‘at standard conditions’
a allowed ( Stream label )
c co- (feed or product) ( Stream label )
d desired ( Stream label )
f primary ( Stream label )
i incidental ( Stream label )
irr. irreversible
l limited ( Stream label )
m material
n accepted ( Stream label )
p process-related ( Stream label )
r reference ( Stream label )
r reaction
u utility-related ( Stream label )
w wasted ( Stream label )

Chapter 5
Ac matrix of input-output coefficients
α electric power per ton unit capacity [ kW/(ton/day) ]
β conversion factor 1,16 *103 [ J/Cal]•[g/ton]•[kJ/kWh ]
c vector of capacities
d vector of demands
f vector of some objective function’s coefficients
∆G0 Gibb’s free energy change at standard conditions [ kCal/mole ]
∆Gr Gibb’s free energy change of reaction [ kcal/mol ]
ε electrical efficiency [ 0..1 ]
γ fraction of fuel cell system investment [ γ ≤ 1 ]
I Unity matrix
I Investment, expressed as dollars per unit capacity [ $/(ton/day) ]
Ie Investment, expressed as dollars per unit power generated [ $/kW ]
λth ratio of the electricity revenue and value added of the chemical
Symbols 243

M molecular weight of product [ kg/kmole ]

p vector of products
Rh revenue flux for heat produced [ $/day ]
Re revenue flux electricity [ $/day ]
Rr,h,(H2) flux of substitute fuel value of hydrogen [ $/day ]
r vector of feedstocks
re revenue for electricity [ $/kWh ]
rf revenue for feedstock [ $/ton ]
rh revenue per unit heat [ $/kWh ]
rp revenue for product [ $/ton ]
s vector of supply
θ the technical lifespan of a facility [ days ]
x0 vector of design variables
V profit difference between fuel cell or conventional reactor [$]

Suffix denotes:
c ‘conventional reactor’ or ‘conventional plant’
e ‘expression in electric dimensions’ (e.g. Ie)
f ‘fuel cell reactor’ or ‘fuel cell use in plant’
s ‘stand-alone power generation’
t ‘trigeneration’:
Abbreviations
ABS Acrylonitrile Butadiene Styrene copolymer
AC Alternating Current
AIChE American Institute of Chemical Engineers
Akzo Akzo Nobel (company, NL)
Al Aluminium
AVR Afvalverwerking Rijnmond (company, NL)
B Benzene
BASF Badische Anilin und Soda Fabrik (company, Germany)
BDO Butanediol
BMW Bayerische Motorwerke AG (company, Germany)
BTU British Thermal Unit
BTX Benzene, Toluene and Xylenes
BTXCon Consumption system BTX
BTXExtr. BTX Extraction process
BTXProd Production system BTX
BTXSep BTX separation
BTXTot Total system BTX
C Carbon
C1 Hydrocarbons with only one C-atom
C2 Hydrocarbons with two C-atoms (e.g. ethane, ethylene)
C3 Hydrocarbons with three C-atoms (e.g. propane, propylene)
C4+ Hydrocarbons with four or more C-atoms
C5+ Hydrocarbons with five or more C-atoms
CaCO3 Lime or Calcium Carbonate
CAPE Computer Aided Process Engineering
CBR Case-Based Reasoning
CBS Centraal Bureau voor de Statistiek (organisation, NL)
CE Centrum voor Energiebesparing (company, NL)
CEO Chief Executive Officer
CFC Chlorinated Fluoro Carbons
CH4 Methane (natural gas)
Cl2 Chlorine
CMAI Chemical Marketing Associates Inc. (company, US)
CO Carbon monoxide
CO2 Carbon dioxide
CPC Computer and Process Control
Cu Copper
DC Direct Current
DHV The Dutch company DHV
DME Dimethylether
DMT Dimethyl Terephtalate
DSM The Dutch company DSM
DTO Duurzame Technologie Ontwikkeling
EB Ethylbenzene
ECN Energie Centrum Nederland (research centre, NL)
EDC Ethylene Dichloride
EOP Electrochemical Oxygen Pumping
EPDM Ethylene Propylene Dimonomer
246 Process System Innovation by Design

Escape European Symposium on Computer Aided Process Engineering

EU European Union
FCC Fluid Catalytic Cracking
FE Functional Element
FFR Flexible Fuel Reformer
FME Federatie Metaal Elektro (association, NL)
FOCAPD Foundations of Computer Aided Design
FOCAPO Foundations of Computer Aided Operations
FOI Faciliterende Organisatie Industrie (organisation, NL)
FUM Functional Unit Method
g gramme
GHR Gas Heated Reformer
GNP Gross National Product
GT Gas turbine
H2 Hydrogen
H2O Water
H2S Hydrogen disulfide
HDA Hydro Dealkylation process
HDPE High Density Polyethylene
Hi-Cal High Calorific
HT High Temperature
HTU Hydrothermal Upgrading
I/O Input/Output
IChemE The Institution of Chemical Engineers (association, UK)
ICI Imperial Chemical Industries (company, UK)
ICT Information, Communication and Telecom
IIASA International Institute for Applied Systems Analysis (research centre)
iMCFC Improved Molten Carbonate Fuel Cell
INCOSE International Council on Systems Engineering (organisation)
IPCC Intergovernmental Panel on Climate Change (organisation, UN)
kg kilogramme
kW kilowatt
kWe kilowatt electric
kWh kilowatt-hour
LCA Life Cycle Analysis
LCI Life Cycle Inventory
LDPE Low Density Polyethylene
LHV Lower Heating Value
LLDPE Linear Low Density Polyethylene
LNG Liquefied Natural Gas
LP Linear Programming
LPG Liquefied Petroleum Gas
LT Low Temperature
MCFC Molten Carbonate Fuel Cell
MEA Membrane electrode assembly
MeOH Methanol
Mg Magnesium
MIBK Methyl isobutyl ketone
MINLP Mixed Integer Non-Linear Programming
MMA Methylmethacrylate
Mn Mangan
MTA Million Ton per Annum
Abbreviations 247

MTBE Methyl-tert-butyl-ether
MW Megawatt
MWh Megawatt hour
mX meta-Xylene
NaCl Sodium Chloride (rock salt)
NGL Natural Gas Liquids
NGO Non Governmental Organisation
NH3 Ammonia
NIMBY Not In My Backyard
NL The Netherlands
NOx Nitrogen oxides
NRC National Research Council (organisation, US)
ODF Objective-defined Function
OECD Organisation for Economic Co-operation and Development
OPEC Organisation of the Oil Exporting Countries
oX ortho-Xylene
oXylSep ortho-Xylene separation
PAFC Phosphoric Acid Fuel Cell
pc Personal Communication
pCO2 Partial CO2 pressure
PdVC Polydivinylchloride
PE Polyethylene
PEM Polymer Electrolyte Membrane
PEMFC Polymer Electrolyte Membrane fuel cell
PET Poly Ethylene Terephtalate
PG Propylene Glycol
Phenol1 Phenol process 1
Phenol2 Phenol process 2
PJ Pèta Joule (1015 Joule)
PMMA Polymethylmethacrylate
Pmt Product-market-technology combination
Pmt-R Product-market-technology-Resource combination
PO Propylene Oxide
PoR Port of Rotterdam
POX Partial Oxidation
PP Polypropylene
PS Polystyrene
PSA Pressure Swing Adsorption
PSE Process System Engineering
Pt Platinum
PtcAnh Phtalic Anhydride
PTFE Poly tetra fluoro ethylene
PUR Polyurethane resin
PVC Poly vinylchloride
pX para-Xylene
PyGsl Pyrolysis Gasoline production
R&D Research and Development
R/P Resource to Production ratio
RC Responsible CareTM
RefGsl Refinery Gasoline production
RIVM Rijksinstituut voor Volksgezondheid en Milieu (research centre, NL)
ROACE Return on Asset Capital Employed
248 Process System Innovation by Design

ROI Return on Investment

RON Research Octane Number
Sabic Saudi Aramco Basic Industries (company, Saudi Arabia)
SAN Styrene Acrylonitrile copolymer
Sasol Sasol (company, South Africa )
SBR Styrene Butadiene Rubber
SEP Samenwerkende Elektriciteits Producenten (association, NL)
SETAC Society of Environmental Toxicology and Chemistry
Si Silicium
SIC Standard Industrial Classification
SM Styrene monomer
SMDS Shell Middle Distillates Synthesis
SO2 Sulphur dioxide
SOFC Solid Oxide Fuel Cell
SRI Stanford Research Institute (company, US)
Styren1 Styrene process 1
Styren2 Styrene process 2
T Toluene
TBA Tert-butyl-alcohol
TCOU Total Cost of Ownership and Use
TiO2 Titanium dioxide
TNO Toegepast-Natuurwetenschappelijk Onderzoek (research centre, NL)
TOE Ton Oil Equivalent
TPA Terephtalic Acid
UK United Kingdom
UN United Nations
US United States of America
VCM Vinylchloride Monomer
VNCI Vereniging Nederlandse Chemische Industrie (association, NL)
VROM Volksgezondheid, Ruimtelijke Ordening en Milieu (Ministerie)
VOC Volatile Organic Compounds
WRI World Resource Institute (organisation)
X Xylene
Zn Zinc
ZnO Zinc Oxide
 Appendices
─────────────
Appendices 251

A. 1 Overview of petrochemical products, Pmt-Rs and supply chains.

Basic inorganic chemicals such as sulphuric acid, ammonia, phosphoric acid, sodium
hydroxide, chlorine and nitrogen fertilizers have long dominated listings of
chemicals’ production volumes. This reflects the importance of the high-volume
industry of fertilizer production that supported the massive increase in agricultural
productivity world-wide. Ethylene, the major organic base chemical has moved up
these ranks, however, at least in the US (Brennan 1998), which reflects the dramatic
growth of the petrochemical industry in the ‘60-‘70s driven by increasing demand
and production of polymers such as polyethylene, polyvinyl chloride (PVC) and
polystyrene.
The top-30 chemicals production listing, USA, 1996.
Chemical USA Annual Chemical USA Annual
prod. cap. Growth prod. cap. Growth
‘85-’95’ ‘85-’95’
(MTA) (%) (MTA) (%)
1 Sulphuric Acid 43.3 0 16 Vinyl 6.8 5
Chloride
2 Ethylene 21.3 6 17 PVC+copol. 5.6 6
3 Ammonia 16.2 0 18 Styrene 5.2 4
4 Diammonium 14.3 4 19 Methanol 5.1 8
phosphate
5 Phosphoric 11.9 2 20 Polypropylene 4.9
Acid
6 Sodium 11.9 2 21 Formaldehyde 3.7 4
hydroxide
7 Propylene 11.7 6 22 DMT 3.6 2
8 Polyethylene 11.7 5 23 Ethylene 3.5 3
(LD+HD) oxide
9 Chlorine 11.4 2 24 Hydrochloric 3.3. 3
acid
10 Nitric acid 7.8 1 25 Cumene 2.6 5
11 MTBE 8.0 25 26 Polystyrene + 2.6 3
copolymers
12 Ethylene 7.8 4 27 Ethylene 2.4 2
Dichloride glycol
13 Ammonium 7.3 2 28 Ammonium 2.4 2
Nitrate sulphate
14 Urea 7.1 2 29 Mono- 2.4 5
Ammonium
Phosphate
15 Ethylbenzene 6.2 6 30 Acetic Acid 2.1 5
Note 1: Data taken from (Brennan 1998: 22) (based on Chem. Eng. News, 24.06.’96).
Note 2: Font used indicates product class: Petrochemicals, petroleum-derived end-
products, chemicals from natural gas. Inorganics
252 Process System Innovation by Design

A. 2 Applied thermodynamics
State properties
In thermodynamics, the concept of state properties has proven to be very useful. A
state property of a particular substance, mixture or stream only depends on the
particular state (pressure, temperature, composition), and not on how that state has
been realised. The value of a state property is always quoted relative to some
reference state60.
Important state properties are internal energy U, enthalpy H, free energy G (all
[J/mol] and entropy S [J/mol/K]. These have been defined in chapter 3:
∆U = q + w (Eq. 3.2)
The internal energy of a system changes in case heat q or work w flow in or out of
the system.
Properties correctly derived from combinations of state properties also are state
properties. Enthalpy is a state property defined as
∆H = ∆U + ∆(PV). (Eq. 3.3.)
It follows from eqn. 3.2 and 3.3 that in case only PV work is allowed on a system
(i.e. no shaft Work, w=0) that ∆H = qp.

Entropy, or disorder, is defined as

∆S = q / T. (Eq. 3.4)
According to the Second Law of Thermodynamics the sum of the entropy changes
in the system and its surroundings can only be positive or zero.
In chemical engineering thermodynamics Gibb's free energy is used. The Gibb's free
energy of a system is defined as:

∆G = ∆H - T∆S (Eq. 3.6)

In a chemical reaction, thus the change of Gibb's free energy thus is a function of
the enthalpy of reaction, the temperature, and the change in entropy between the
final and entry states. Since G is a state property, ∆G is independent of the course
of the transformation process in the system
It can be shown that the maximum amount of Work that can be extracted from a
chemical reaction equals the ∆G of that reaction. Alternatively, for a reaction to
occur, the minimum amount of Work to be supplied also equals ∆G of that reaction.
A reaction where ∆G is sufficiently negative will complete spontaneously once
initiated. An example are combustion and explosion reactions, where both ∆H is

60 For thermodynamic properties such as enthalpy, this is the state of the elements and standard
conditions (P=1 atm, T=298 K). For industrial properties such as the Lower Heating Value and
exergy, one prefers to use the lowest energy state found on Earth (i.e. CO2 + water). The definition
of a reference state for other elements is somewhat arbitrary, as their natural appearance may vary.
As a consequence, discussions have been ongoing in the scientific community to arrive at a consensus
that can be used for consistent energy and exergy efficiency calaculations (e.g. Kotas 1985).
Appendices 253

negative, and ∆S is positive (entropy creation), which results in a strongly negative

∆G. (Stull et al. 1969).
A chemical reaction where the number of reactants is the same as the number of
products, which occurs at infinitesimal slow speed, and which is considered an
isolated system, represents a reversible system where dS = 0. As a consequence, ∆G
= ∆H.
To enable calculations around energy transformations that involve chemical
reactions, the state property Enthalpy of formation, Hfo [kJ/mol] is used: in case the
Enthalpy of formation of products and reactants is known, the enthalpy change
between products and reactants equals the enthalpy of reaction ∆Hr.
In energy statistics, the Lower Heating Value or LHV [MJ/kg], is a common
measure of energy content of fuels. It is a measure of the amount of heat q released
when the fuel is combusted. The LHV is an expression of the amount of chemical
energy stored in a material; the LHV is equal to the reaction enthalpy of combustion
of the fuel with oxygen under standard conditions to gaseous products. Thus, this
excludes the amount of energy that is released when part of these products, the
water vapour, is condensed to liquid at standard conditions . In the Higher Heating
Value (HHV) of a material this amount of energy is included.
Since internal energy, enthalpy, LHV and HHV are a thermodynamic properties of
state, in the calculation of system energy balances the energy content of streams can
be expressed as ∆H0, LHV or HHV or any other energy measure, but the same
expression must be used for all streams in the final calculation.

Open, closed and isolated systems

In thermodynamics, a distinction is made between isolated, closed, and open
systems. A totally isolated system can neither exchange matter nor energy with its
surroundings. A thermally isolated system can only exchange Work with its
environment. A closed system can only exchange energy with its surroundings.
Finally, an open system can exchange both matter and energy with its environment
(Seader 1982; Georgescu-Roegen 1991).
A chemical reactor that operates perfectly adiabatically is a thermally isolated system.
A heat-engine based on the famous Carnot-cycle is a closed system, as the specific
system requirement is that the 'working medium' is brought back to its original state
in the process of converting heat to work. The Earth can be considered a closed
system, for all practical purposes, in case the input of meteorites and the slow loss of
Helium to the Universe are neglected. A chemical plant and the petrochemical
industry are open systems because these exchange at least feedstock and product
streams with their surroundings.

Manifestations of energy
Energy has many forms or manifestations. The best known manifestation of energy
is heat, in the form of radiation, convection, or direct heat transfer. Other energy
forms are movement (kinetic energy), gravitational potential energy, and of course
electrical potential energy.
254 Process System Innovation by Design

Energy can be stored in materials as chemical energy or chemical potential. Some of

these materials are our commonly used fuels Energy can be converted from one
form into another. An overview is presented in Figure A.2-1.

Oil Heat Mechanical

Crude Oil Electricity Propulsion
Products Thigh Energy
1 2 5 6

Powering
Coal Equipment

Natural Gas Lighting

Powering
Equipment
7
Propulsion
8
Heat Industrial
Tmedium Processing
3 9
Heat Space-
Tlow heating
4 10
Energy sources
or
stocks Forms of energy or energy-carriers Energy use

Figure A.2-1: Energy conversions in the utilisation of fossil fuels.

Legenda for number labels Figure A.2-1
No. Conversion Conversion
1 Conversion / separation (refinery) 6 Shaft power-to-electricity (generator)
2 Combustion (power plants) 7 Shaft-power to equipment (machinery,
electronics)
3 Combustion (industrial furnace) 8 Shaft-power to propulsion (transport)
4 Combustion (Central Heating) 9 Heat transfer (industrial heat-exchangers)
5 Heat-to-Power (e.g. turbomachinery) 10 Heat transfer (radiators Central Heating)

The First Law of Thermodynamics

According to the First Law of Thermodynamics, energy cannot be lost not created.
This implies that the energy of the Universe is constant. Thus, the internal energy of
a system can only increase when simultaneously the energy of the system
surroundings decreases. In energy analysis confusion results when the First Law of
thermodynamics is not properly understood by engineers, scientists, or policy-
makers and consultants who have had only limited or no formal training in
thermodynamics.
Particularly the use of the misnomen 'energy consumption' is widespread. Instead of
'energy consumption', however, we rather speak of energy use. Energy is often
reported as Lower Heating Value. This LHV requires consistent calculation, as its
value depends on the reference state selected. In the promotion material for boilers
Appendices 255

to be used in central heating systems, for example, misuse of LHV leads to common
'deceit'. Boiler efficiencies greater than 100% are proclaimed for systems that utilise
the energy released during condensing the water vapour present in the flue gas. The
energy content of the natural gas consumed is listed as the Lower Heating Value, a
measure which assumes that water vapour is not condensed. Thus, both exergy and
LHV are thermodynamically sound properties once the 'dead' reference state has
been defined, which must include the phase state of the components included.

The Second-Law of Thermodynamics

Where the First Law states that energy cannot be lost nor created, the Second Law
implies that the total energy quality of the Universe must deteriorate whenever any
real process completes. The Second Law of thermodynamics states that the result of
any process that completes in a finite time-span is that the total entropy S of the
Universe increases . It implies that in a closed system heat cannot be converted
completely into Work: there will always be some heat that is rejected to the
environment. The maximum conversion of heat-to-Work is obtained in a Carnot-
cycle (Carnot 1824; cited in Kondepudi and Prigogine 1998: 69). The associated
Carnot-efficiency depends only on the temperature of the heat reservoir and the
temperature of the heat sink between which part of the heat is converted into Work.
The potential to perform useful Work can be determined for any form of energy. It
is a measure of the quality of energy and has been coined exergy (e.g. Kotas 1985). In
case we select a reference temperature To for the heat sink in the Carnot-cycle, the
exergy of a heat reservoir or stream q at temperature T is given by the familiar
Carnot factor:
Ex = A = q (1 – To/T) (Eq. 3.9)
This equation coins the familiar result that when the temperature T of a heat
reservoir q or heat flow q• approaches T0 its Exergy decreases to zero.
In case the transformation of (chemical) material occurs at T0,
∆Ex = ∆H - To∆S = ∆G0

The work of Denbigh (1956) provides a link between energy systems and chemical
reaction systems as he derived the equation
wt = To σ + ∆A
This equation states that the total Work wt required to let a process progress at a
finite speed equals the wasted work To σ or irreversible entropy production plus the
change in of availability the process materials ∆B. In case the transformation occurs
and T0, this equals ∆G0
256 Process System Innovation by Design

A. 3 Overview of cohesion types

In this appendix, an overview of cohesion types is given. Each is illustrated by an
explanatory example. Subsequently, the applicability of various cohesion types in the
petrochemical industry is briefly addressed.
• coincidental: ‘no common function, data, procedure’, i.e. not serving some purpose
It may be seen that coincidental cohesion in systems decomposition and modelling is
to be avoided. A decomposition of a nations industry by companies, for example,
may result in 'coincidental cohesion' unsuitable for any modelling or analysis. The
very establishment of the fact that the only cohesion in a business organisation is
perceived as coincidental may explain a lot of business reorganisations, as the very
conclusion is this case unavoidably is that the organisation does not serve a purpose
or mission.
• logical: the elements ‘perform certain logical classes of functions’
Logical cohesion involves identification of subsystems by grouping them according
to some conceptual classification scheme. In a logical system decomposition the
elements ‘perform certain logical classes of functions’. The best example in the
industrial economy is the SBI-code used in economic statistics (CBS 1992).
• temporal: modules are related logically and by time of execution, but not necessary
by procedure
Temporal cohesion is important in the daily operation of systems, when sound
timing of activities must be achieved, and in weekly or yearly planning of systems.
• sequential: no other relation than that output of one element provides input to the
next
Other than in temporal cohesion, in this case there is either a physical or
information link between system elements. It relates, for example, to the traditional
method of project execution in the chemical industry. The subsequent project
phases (e.g. 'Strategic idea', 'Conceptual Design', 'Basic Design', 'Detailed Design',
'Plant Construction' (van Breda and Dijkema 1998) are executed sequentially based
on the information packages generated in the previous phase. It may be seen that in
actual practice, when further decomposition of system elements is required,
procedural cohesion can be applied, or temporal cohesion when timing is important.
• procedural: elements are all part of a procedure
Procedural cohesion can be achieved when grouping of subsystem is done according
to a common procedural context, based on similarity on how specific activities are
performed. A procedural decomposition of the chemical industry, for example, is the
distinction between continuous processing, and batch processing. This is widely used
in chemical engineering literature, and even in the formulation of research
programmes and proposals.
Procedural cohesion is useful to understand existing production systems: realising
production objective-functions must be done in an economic context. At any given
time because of the set of available technology, (market) regulations (arrangement),

61 Only if the initial and final (P,T) of the materials processed is the same, and equals (P,T) of the

medium at which the process exchanges heat with the environment.

Appendices 257

control mechanisms, notably information technology, often a particular type of

procedure for operation is economically the most advantageous. This largely explains
the procedural cohesion in the chemical industry, where on the one hand large-scale
operations prevail for high-volume products, on the other hand batch-wise
operations for small-scale, flexible production for fine-chemicals and
pharmaceuticals. Exceptions often can be explained by lack-of-suitable technology
(e.g. batch wise polymerisation for PVC, a product produced in bulk-quantities). As
the systems environment changes, the procedural cohesion may no longer be valid
and can disappear, which e.g. is occurring in the power industry which is ‘in
transition’ through the dramatic changes in regulation and to promote economic
restructuring.
• communicational: the elements all operate on the same set of data (object-
orientation)
Communicational cohesion, for example, relates to emerging concept of concurrent
engineering, where the prime cohesion in activity is indeed that they relate to the
same object (a chemical plant) that must be designed, engineered, and built.
Communicational cohesion can also be labelled relational cohesion. It is then
provides a means for cohesive modelling of a system’s environment, and it may
provide a powerful means to explain the impact of such concepts as ‘substance-
chain management’, Product Stewardship, supply-chain management etc., which
relate a chain of operation to a common object or substance. Another area of
relevance is that of infrastructures. In an increasing number of infrastructure sectors
new roles are defined, e.g. for network and market regulation. The actors involved all
act on and relate to one infrastructure.
• functional cohesion (see Chapter 3, §3.3 p. 93

Cohesion types and the petrochemical industry

When attempting to achieve procedural cohesion the focus is on the types of activity
of a particular industry, sector or plant. Fossil resources are being used or converted
for the greater part in power generation and the process industry. The latter term is a
collocation of those industries that employ large-scale continuously operated
processes (see Figure A.3.1). In a largely a logical classification one may distinguish
'winning and treatment of mineral ores', 'crude-oil refining', 'petrochemical industry',
'chemical industry', 'base metals industry', 'fertilizer industry', 'food industry' and
'polymer industry'. The use of large-scale continuous processes is the common
procedural denominator of the activities in these sectors.
258 Process System Innovation by Design

Base Metal

Coke
Inorganic Chemicals Food
factories
industry

Fertilizers
Large-scale process industry
Other Organic
Chemicals

Power Plants

Polymers

Petroleum Refineries
Petrochemicals

Figure A.3.1: Industry sectors that employ large-scale, continuous processing.

A further suitable procedural decomposition of continuous large-scale plants in all
the industries mentioned is not trivial.
In the petrochemical industry, each chemical plant constitutes a specific network of
unit operations. In the design of a chemical plant, sequential cohesion between its unit
operations must be achieved. In chemical engineering, the classes of unit operations,
reactors, pumps, compressors, separators etc. represent a procedural decomposition
at this aggregate level. The equivalent of these operations again are found in most of
the process industry sectors illustrated (Figure A.3.1). A further procedural
decomposition in, for example, reactor engineering is the distinction between gas,
liquid, solids, homogeneous, heterogeneous, gas/liquid, gas/solid etc.
Another decomposition of the chemical industry can be completed by considering
the procedural denominators in parts of the production chains or process routes. At
the beginning of the production chains, a common denominator is that multi-
purpose or mixed feedstock is converted to pure, single- or limited-purpose
feedstock for individual chemical plants. The primary example in the petrochemical
industry is the naphtha cracker, gasification to synthesis gas and the MTO-process62.
Further down chemical production chains consist of chemical plants that can only
convert a limited span of feedstock to a limited number of products. Finally, at the
end of many production chains products are made in very small quantities in plants
that are truly ‘multi-purpose’: they can process a variety of feedstock into a variety of
final products.

62 Methanol To Olefins
63 Shell Middle Distillate Synthesis
Appendices 259

A. 4 Systems representation with the input-output paradigm64

Characterisation of dynamics systems according to the lines of thought in
mathematical systems theory (MST), using the input-output paradigm.
We abstain from the more general concept of a dynamic system in which the
distinction between inputs and outputs is dropped and the interaction between
system and environment is represented by interaction variables, which can act both in
an input and an output fashion, changing over in time.
We develop the input-output description using the state-space concept initially.
Using this concept with the associated time evolution of the states presumes
knowledge about the internal workings of the system. Then we relax on this
knowledge of internal workings and drop the state concept, taking the system to be
of the black-box Input-Output type with time evolution. Finally, the time evolution is
discarded.
It is assumed that the function of the system, Σ, is to generate a behaviour B to
transform a set of inputs (Su) into a set of outputs (Sy), which relate the system to the
external world. The performance of the system (Ω) is assessed by means of value
functions (P).
The specification of the system , Σ, is built up in two steps, viz. (1) defining the
behaviour (2) defining the performance.
Starting with (1) the behaviour, B , first three different spaces are defined for the
inputs u ∈SU, the outputs y ∈ SY and the internal states x ∈ SX. To account for the
evolution of the system in time, the scalar time variable t is introduced in space T.
Each space is given a properties, like a metric in order to be able to define size of an
element and distance between objects in the space. Next, one defines the mappings:
* from input space to state space,
G: SU => SX ,
* the evolution from state space and time to state space and time,
F: SX * T => SX * T,
* and from state space to output space,
H: SX => SY .
The mappings satisfy certain general smoothness properties.
For mapping F special conditions apply, relating to the transitivity aspects of time
operations:
F(x0, t0 => x2, t2 ) = F(x0, t0 => x1, t1 ) * F(x1, t1 => x2, t2 )
The precise nature of the mappings and the specific properties define the specific
behaviour B of the system Σ:
Bstate={G o F o H}.
Note that the causality of the system is rooted in the specification of the flow of
information from inputs to states to outputs in combination with the transitivity
property of F.

64 This appendix is largely based on internal discussions and research note (Grievink 2003).
260 Process System Innovation by Design

Now that the behaviour of the system has been defined, the next item is the measures
for the performance of the system.
The performance will be measured by so-called values ω ∈ Ω. These values are
elements of the performance space Ω. Again this space is given an internal structure
to make sure that the elements have a size and that a distance measure is in place.
The performance values are obtained from a performance functions (P) , which is a
mapping from the inputs, outputs and, possibly, the states over a certain time
interval of evolution to the performance measures:
Pstate: SU * SX * SY * T => Ω
The mapping P to the performance space is called the value map.
The values can be written as a functional over time interval [t0,t1]:
ω = P(∫t0t1 p(u(t),y(t),x(t),t) dt
Some examples of performance values are: controllability, controllability and
economic profit, technological indicators, like exergy losses. Hence, the system Σstate,
is now fully characterised by:
Σstate = { SU , SX , SY , T, Ω, Bstate, Pstate }

An example of such a system description will now specified for linear dynamic
systems.
The behaviour, Blin, can be characterised by a triplet of matrices:
Blin = {B, A, C}.
The corresponding behaviour is given by:
d x(t) /dt = A.x(t) + B.u(t) and y(t) = C. x(t)
The performance of such linear systems is often specified in a quadratic fashion:
ω(t1 ,t0 ) = x(t0)T.W0.x(t0) + x(t1)T.W1.x(t1)
+ ∫tot1[u(t)T.Wu.u(t) + y(t)T.Wy.y(t)] dt

Hence, the performance functional can now be characterised by the matrix quartet:
Pquadratic = {W0, W1, Wu, Wy }

For a black-box type of input – output description, where the internal state is not
considered, the behavioural description will reduce to:
BI-O: SU * T => SY * T

The performance model reduces to:

PI-O: SU * SY * T => Ω

Hence, the system ΣI-O, is now fully characterised by:

ΣI-O = { SU , SY , T, Ω, BI-O, PI-O }
Appendices 261

When evolution in time is not considered (e.g. representing the steady state of a
system) the description further simplifies to:
BI-O,ss: SU => SY ,
PI-O,ss: SU * SY => Ω
ΣI-O, ss = { SU , SY , Ω, BI-O,ss, PI-O,ss }.

In conclusions, any system is characterised by the specification of its:

o Input space (SU)
o Output space (SY)
o Performance space(Ω)
o Behaviour map (B)
o Value map (P)
262 Process System Innovation by Design

A. 5 Algorithm to check for mutual contingency

The criterion for mutual contingency of a set objective-defined functions { Ф } (Ch. 3,
§3.3.3 p. 97) can be written as a (pseudo) algorithm, which returns the Boolean S
where S: { Ф }= Mutual Contingent Set.
{the system has m possible instances, and n objectives Ф 1…n}
{an instance equates to a possible system state expressed as a 'value' of Ф 1…n }
{Start - init}
S=TRUE
{Loop over all objectives Ф (1…n)}
FOR i=1…n
FOR j=1…n
{per pair of objectives Loop over all instances of the system (1..m)}
{init}
k=0
P = FALSE
M = FALSE
A = FALSE
While k < m
k = k+1
{parallel?}
Pk = (Фi,k AND Фj,k)
{mutually exclusive?}
Mk = ((Фi,k OR Фj,k) AND NOT (Фi,k AND Фj,k))
{achievement?}
Ak = OR (Фi,k AND Фj,k+1)
{Thus it is implicitly determined when there is no relation between Фi and
Фj in any of the instances because Mk contains Фi or Фj, never both
{add result of this instance to result for pair i,j}
{latch P when parallel achievement in some instance of the system}
P= P OR Pk
{latch M}
M= M OR Mk
{latch A}
A = A OR Ak
END {while}
{test whether in the set pairs of instances exist that are both parallel and mutually
exclusive; if this is the case, then mutual contingency has not been achieved}
Si,j = (P OR M OR A) AND (NOT (P AND M))
S = S AND Sij
END {FOR j}
END {FOR i}
Appendices 263

A. 6 Assessment of the Dutch industrial olefins system

Production facilities (1990) included in the assessment

In the tables below, an overview is given of the Dutch production facilities that
together formed the industrial olefins systems in the Netherlands as of 1990. To
date, both Shell and Dow have completed major rejuvenation/revamp/expansion
projects on their steam cracker complexes. DSM has spun off its petrochemicals, PE
an PP activities to Sabic Petrochemicals.

Table A.6.1: Production capacities for ethylene and propylene

Process company Capacity [ ton/year] Remarks
Steam cracker (LPG, Dow Benelux ethylene: 1.000.000 expansion:
naphtha, gas oil) propylene: 400.000 ethylene: 100.000
Steam cracker DSM ethylene: 925.000 expansion:
(naphtha, gas oil) propylene: 470.000 ethylene: 100.000
Steam cracker (LPG, Shell Moerdijk ethylene: 625.000
naphtha, gas oil) propylene: 350.000

As indicated in Figure 4-2: Industrial production of ethylene derivatives and example

applications and in Figure 4-3: Industrial production of propylene derivatives and
example applications (p. 124-125) are relatively 'loosely' connected, while they share
a common source, the steam cracker. Therefore, production capacities of ethylene
and propylene derivatives are listed separately below (Tables A.6.2 and A.6.3)
Table A.6.2: Capacities of ethylene derivatives in the Dutch industrial olefins system
From / to ... Capacity Company Remarks
[ton/year]
Ethylene 2.550.000 Dow, DSM, Shell expansion: 200.000
Polyethylene
LDPE 580.000 Dow, DSM
LLDPE <380.000 Dow, DSM DSM: one plant for both
HDPE <110.000 DSM LLDPE and HDPE
UHMWPE 5.000 DSM “Dyneema” fibre
EPDM rubber 65.000 DSM expansion: 10.000
PE/VAc copolymer 20.000 Vinamul expansion: 24.000
Vinylchloride and derivatives
Ethylene dichloride 600.000 Rovin
Vinyl chloride 480.000 Rovin
Ethylene diamine 49.000 Dow, Delamine estimate
Polyvinylchloride 410.000 Rovin, Limburgse expansion: 80.000
Vinyl Unie
Glycols
Ethylene oxide 285.000 Dow, Shell
Ethyleneglycol 345.000 Dow, Shell
Glycol ethers 74.000 Dow, Shell
Various
Vinylacetaat 60.000 diversen estimate
264 Process System Innovation by Design

Table A.6.3: Capacities of propylene derivatives in the Dutch industrial olefins system
Product Capacity [ton/year] Company Remarks
Propylene 1.220.000 Dow, DSM, Shell
Polypropylene 480.000 DSM, Basf, Shell Expansion: 190.000
Propylene oxide 370.000 Shell, Arco
Propylene glycol 60.000 Arco
Propylene glycol ether 100.000 Arco
Polyols 180.000 Dow, Du Pont, Estimate
ICI, Shell
Acrylonitrile 175.000 DSM
acrylamide 15.000 Cytec Estimate
polyacrylamide 15.000 Cytec Estimate
isopropylalcohol 290.000 Shell
diisopropylether ? Shell
Acetone 150.000 Shell
PMMA 9.500 ICI
Methylisobutyl ketone 60.000 Shell Estimate
Methylisobutyl carbinol
Allylchloride 80.000 Shell
Epichloro hydrin 80.000 Shell
Bisphenol A 90.000 Shell
Epoxy resins 55.000 Shell

Modelling and assessment

The assessment has been executed using the methods described in chapter 3. These
were implemented in a database system and a spreadsheet programme.
The olefins system was modelled as a single industrial complex. The set of system
elements consists of the industrial processes in commercial use. The subsystems
'olefins production' and 'olefins consumption' were included by appropriate selection
of system elements.
Subsystems and system elements are connected by internal flows, the inputs and
outputs of single process plants (as depicted in Figure 3.2). The net inputs and
outputs of the subsystems and the entire complex were calculated in a linear model
on the basis of actual capacity-installed (100% utilisation), maximum internal
deliveries, and the input/output characteristic of each individual plant viz.
subsystem.
Input-output data on individual plants used was largely taken from open-literature
data on 'typical plants', such as published in (Chauvel and Lefebvre 1989a). These
were augmented and corrected by using information and data supplied by various
companies65. Although some datasets used may have been outdated or incomplete,
these were thought to suffice to for the objective of the analysis, which was to
illustrate the possibilities of networked petrochemical system assessment to obtain
entry-points to look for improvements, and not the assessment of individual plants.

65 Data and information were supplied with the understanding that these would be treated as
confidential.. Therefore, analysis results are only presented in aggregate form. Details on the data set
used and the reconciliation process applied are available from the author.
Appendices 265

The data sets obtained were converted prior to their use in the assessment. The data
reconciliation involved a check for data-consistency and completeness per data set
on an individual plant. Intricate process knowledge and mass and energy balancing
were used to correct or complete data sets. To be able to complete the process the
following assumptions had to be made:
- with respect to additions for completion of plant's mass balances, is was
assumed that a deficiency (i.e. more mass going into the process than going out)
is caused by the generation of light gases that are emitted as CO2 and H2O. This
assumption can be considered valid, since most process schemes include an off-
gas treatment section that remains otherwise unspecified, whilst 'hydrocarbons'
can always be utilised as fuel after proper treatment. The carbon and hydrogen
thus lost will eventually end up as CO2 and H2O.
- the specific energy content of streams not reported in a particular data set can be
estimated. However, since often nothing is known about the energy content or
the application of these streams, these are assumed to be zero in the analysis.
Mass loss La, Lb and mass efficiency ηa, ηb do not require complete datasets, they
immediately give an impression of the performance of the process. Calculation of
these indicators does not require an assumption with respect to the fate of the
external mass loss!
Computation of Industrial System Performance indicators
The energy content and exergy content of the input and output streams can
conveniently be estimated from thermodynamic data reported in the literature (Stull
et al. 1969). The computation of Lower Heating Value and exergy require adherence
to a suitable thermodynamic reference state (P0, T0, standard composition of the
environment). This is acceptable for the analysis performed, which is intended to
obtain an overall picture in a reasonable amount of time, which is reasonably
accurate and which can serve to make a first selection of starting points for an
assessment of the (technological) potential for improvement. Because most
production plants take-in feedstock at ambient temperature. In specific cases,
however, the actual losses may differ substantially from the ones calculated in the
analysis.
As with mass balances, however, hardly ever the data reported on industrial
processes result in a closed energy balance because streams are missing. Energy
leaving an industrial process, for example, as flue gas, cooling water, or condensate
are hardly ever listed. In addition, some losses occur by radiation. Properly defined,
energy is a thermodynamic property of state. Energy content, however, also
represents a valuation of streams. In chapter 3, it was argued it is a prime candidate
for use in process system assessment for sustainability.
266 Process System Innovation by Design

Table A.6.4: Assessment result olefins system, Mass Indicators

Mass
Efficiency Loss
0…100% [kton/year]
a b e a b e
System Subsystem Process name Capacity
Production Ethylene Steam cracker 73 73 44 2550 350 350 1211
Production Propylene Steam cracker 73 73 44 1220 350 350 1211
Production C4/C5/PyGsl/Steam cracker 0 0 0 1043.1 0 0 0
Consumption Ethylene EDC+VCM 85 85 47 480 88 88 549
Consumption Ethylene Ethylene diamine 40 40 23 49 74 74 164
Consumption Ethylene Ethylene oxide 54 54 11 285 242 242 2255
Consumption Ethylene Ethylene glycol 81 92 92 345 79 32 32
Consumption Ethylene Glycol ethers 81 81 94 74 17 5 5
Consumption Propylene Propylene oxide 91 94 90 370 151 93 166
Consumption Propylene Propylene glycol 81 90 90 60 14 7 7
Consumption Propylene Propylene glycolether 81 90 90 60 24 13 13
Consumption Propylene Polyols 96 97 88 180 8 6 25
Consumption Propylene Acrylonitril 64 72 20 175 97 77 760
Consumption Propylene Acrylamide 75 75 39 15 5 5 23
Consumption Propylene Isopropylalcohol 89 89 57 290 37 37 216
Consumption Propylene DIPA 0 0 0 0 0 0 0
Consumption Propylene Acetone 90 90 64 150 16 16 85
Consumption Propylene MEK 75 75 75 60 10 10 10
Consumption Propylene Allylchloride 49 93 93 80 84 12 12
Consumption Propylene Epichloro hydrin 38 38 2 80 131 131 3400
Consumption Propylene Bisphenol A 86 86 83 90 15 15 18
Consumption Propylene VinylAcetate 0 0 0 60 0 0 0
Polymerisation Ethylene LDPE 97 97 52 580 18 18 536
Polymerisation Ethylene LLDPE 98 98 90 380 9 9 35
Polymerisation Ethylene HDPE 97 97 91 110 3 3 11
Polymerisation Ethylene UHMWPE 97 97 64 5 0 0 3
Polymerisation Ethylene EPDM Rubber 98 98 90 65 1 1 7
Polymerisation Ethylene PE/Vac Copol 98 98 56 20 1 1 19
Polymerisation Ethylene PVC 99 99 27 410 4 4 1099
Polymerisation Propylene Polypropene 96 97 86 480 21 16 77
Polymerisation Propylene PMMA 100 100 8 9.5 0 0 116
Polymerisation Propylene Epoxy resins 68 68 26 55 26 26 159
Appendices 267

TableA.6.5: Assessment results, olefins system, Energy Indicators

Energy
Efficiency Loss
0…100% [PJ/Yr]
a b e a b e
System Subsystem Process name Capacity
Production Ethylene Steam cracker 65 65 65 2550 23.7 23.7 23.7
Production Propylene Steam cracker 65 65 65 1220 23.7 23.7 23.7
Production C4/C5/PyGsl/Steam cracker 0 0 0 0 0.0 0.0 0.0
Consumption Ethylene EDC+VCM 64 64 64 480 5.2 5.2 5.2
Consumption Ethylene Ethylene diamine 39 39 39 49 2.3 2.3 2.3
Consumption Ethylene Ethylene oxide 65 65 64 285 4.2 4.2 4.4
Consumption Ethylene Ethylene glycol 52 61 61 345 5.4 4.4 4.4
Consumption Ethylene Glycol ethers 65 75 75 74 1.0 0.7 0.7
Consumption Propylene Propylene oxide 76 80 80 370 15.5 13.1 13.1
Consumption Propylene Propylene glycol 61 70 70 60 0.8 0.6 0.6
Consumption Propylene Propylene glycolether 69 78 78 60 1.2 0.9 0.9
Consumption Propylene Polyols 78 79 79 180 1.3 1.2 1.2
Consumption Propylene Acrylonitril 51 56 56 175 5.3 4.9 4.8
Consumption Propylene Acrylamide 71 71 68 15 0.1 0.1 0.2
Consumption Propylene Isopropylalcohol 63 63 63 290 5.3 5.3 5.3
Consumption Propylene DIPA 0 0 0 0 0.0 0.0 0.0
Consumption Propylene Acetone 50 50 50 150 3.3 3.3 3.3
Consumption Propylene MEK 97 97 97 60 0.0 0.0 0.0
Consumption Propylene Allylchloride 58 68 68 80 1.4 1.0 1.0
Consumption Propylene Epichloro hydrin 54 54 54 80 1.2 1.2 1.2
Consumption Propylene Bisphenol A 73 73 71 90 1.1 1.1 1.2
Consumption Propylene VinylAcetate 0 0 0 60 0.0 0.0 0.0
Polymerisation Ethylene LDPE 83 83 83 580 5.1 5.1 5.1
Polymerisation Ethylene LLDPE 86 86 84 380 2.3 2.3 2.7
Polymerisation Ethylene HDPE 86 86 84 110 0.8 0.8 0.9
Polymerisation Ethylene UHMWPE 83 83 83 5 0.0 0.0 0.0
Polymerisation Ethylene EPDM Rubber 78 78 78 65 0.8 0.8 0.8
Polymerisation Ethylene PE/Vac Copol 93 93 88 20 0.0 0.0 0.1
Polymerisation Ethylene PVC 76 76 75 410 2.3 2.3 2.4
Polymerisation Propylene Polypropene 82 83 83 480 4.7 4.5 4.5
Polymerisation Propylene PMMA 87 87 86 9.5 0.0 0.0 0.0
Polymerisation Propylene Epoxy resins 88 82 81 55 0.3 0.3 0.3
268 Process System Innovation by Design

Once the composition of a stream is known, the LHV and the exergy value can
easily be computed from thermodynamic basic data and the reference state. In oil
and petrochemicals, however, often only an approximate composition is known and
thermodynamic basic data may be lacking or inaccessible. Therefore a pragmatic
approach has been used: for hydrocarbon mixtures such as 'naphtha', exergy has
been equated to LHV reported in reputable sources66.
In the computation of Industrial System Performance indicators as suggested in
Chapter 3, energy is calculated as Lower Heating Value (LHV) with respect to a
standard reference environment, the so-called dead state (see e.g. Kotas 1985)). The
LHV and exergy of the dead state are zero, i.e. energy is at a minimum, and entropy
is at a maximum. As a result, LHV and exergy of most hydrocarbons greater than
zero, which allows us to define exergetic efficiencies.

The assessment results are given in Tables A.6.4 and A.6.5on the previous pages.

Calculation example: The Showa-Denko process for epichloro hydrin

The definitions and the use of these indicators are illustrated below for the
production of epichloro hydrin (ECH).
As per the method developed in Chapter 3, the ECH process is considered as a
black box characterised by its inputs and outputs of feedstock, (spent) utilities,
(spent) processing agents, products, by-products and wastes. The process data used
in this example were taken from the preliminary process design by three chemical
engineering students of Delft. Their design was based on the Showa-Denko process
for the production of epichloro hydrin, ECH, which is one of the constituents of
epoxy resins.
The main steps in this process are
1 allyl acetate production from propylene, acetic acid and oxygen
2 conversion of allyl acetate with water to allyl alcohol, whereby acetic acid is
recovered
3 reaction of allyl alcohol and chlorine to dichloro hydrin, where HCl is a catalyst
4 conversion of dichlorohydrin with calcium hydroxide to ECH and calcium
chloride
The overall mass balance calculated in a student process design (Bouma et al., 1996)
is listed in Table A.6.6. These data suffice to get a first impression of the eco-
efficiency of the process design. The first step is to label the process input streams as
either feedstock, processing agents, or utilities. Processing agents, such as solvents
and catalysts, typically are not consumed, but can be used over and over again in a
continuous production process. Energy utilities, such as electricity, and steam
typically do not end up in the process’ product, but only supply the energy required
to operate the process. At the output side, we must distinguish between wanted

66 Reconciliation was done per individual plant. Details of thermodynamic calculations can be

obtained from the author.

Appendices 269

products, i.e. the main product and some valuable by-products, and the unwanted
products, such as low-quality by-products, emissions, and wastes.
From the summary of main process steps of the Showa-Denko process, we
conclude that propylene, oxygen, chlorine, and calcium hydroxide are feedstock,
whereas water, steam, hydrochloric acid, and acetic acid are processing agents. Water
also is a reactant of the first process step, but the overall process reaction does not
include water.
propylene + ½ oxygen + chlorine + ½ calcium hydroxide -->
ECH + ½ calcium chloride
The main product is ECH, there are no wanted by-products. Propane, waste gas,
waste water, and organic chlorines are all labelled as unwanted by-products.
Table A.6.6: Overall mass balance for a Showa-Denko preliminary process design (from
Bouma et al., 1996)
In [kg/s] Out [kg/s]
Oxygen 1,06 ECH 3,44
Propylene 2,18 Propane 0,69
Water 1,06 Waste gas 1,15
Acetic acid 0,004 Waste water 38,16
(CaCl2) (2,2)
Hydrochloric 0,13 Organic 0,39
acid chlorines
Chlorine 2,91
Aqueous lime 32,1
(Ca(OH)2 (1,6)
Steam 4,37
Total 43,83 Total 43,83

The direct efficiency expresses how fit for purpose the process evaluated is. The
Showa-Denko process design has an ηa of only 0,44. The product efficiency of the
process ηb is equal to ηa, as there are no wanted by-products. Both indicators ηa and
ηb can also be expressed as the quotient of products and all inputs, i.e. feedstock,
and processing agents combined. Only ηc provides additional information, however.
In case of the Showa-Denko process design the total efficiency is only 0,078, which
indicates that the process uses a lot of additional resources that are degraded to
waste. Most of these unwanted by-products can be recovered, however, at the
expense of additional process equipment and energy input.
270 Process System Innovation by Design

A. 7 PEMFC - the important markets projected beyond 2005

The focus of the fuel cell community has long been the development of PEMFC
systems suitable for the passenger-vehicle market, for buses and for heavy-duty
vehicles. Automobile companies have joined forces and invested in these fuel cell
companies to integrate PEMFC systems in to existing and novel car concepts. Oil
companies have largely focused on the possibilities of on-board conversion of
'logistic fuels', a label for the commonly distributed motor fuels, gasoline, diesel and
to a lesser extent LPG. A survey of internet resources of reputable character reveals
that at present PEM fuel cells and PEMFC systems are being developed for 6 classes
of applications:
I. Stationary applications:
1. small-scale (1-5 kWe) for household micro-cogeneration systems
2. medium scale (500 kW - 2 MWe) for utility building applications
II. Automotive applications
3. light duty vehicles (50 kWe): passenger vehicles
4. heavy duty vehicles (205 kWe): trucks and buses
5. heavy duty vehicles (300 kWe): AGVs
6. auxiliary power (1 kW): 42 V on board - next generation passenger vehicles
III. Portable applications
7. a variety of (+/- 200 We) 'large battery' applications
IV.Other applications
8. motive power for ships and submarines (0.5 - 100 MW)
9. motive power for (cargo) trains (0.5 - 100 MW)
Various companies have been producing stacks viz. systems and have demonstrated
working systems for category I - III.
Above application represent an equal number of markets for fuel cell use. At
present, it appears that a number of the PEM fuel cell markets that are currently
targeted may become -at least partly- methanol-powered fuel cell markets.
Automotive applications today appear to be the most likely candidates for methanol FC
development. In all classes mentioned under category II, currently methanol
PEMFC systems are in an advanced state of development. In these markets there
will be competition from Flexible Fuel Reformer (FFR) based systems that tie in to
the existing infrastructures for logistics fuels. A major disadvantage of these systems,
however, is their high operating temperature (900-1100 oC) of the flexible fuel
processor that would be able to convert LPG, gasoline, diesel and methanol. In
addition, FFRs are more complex, less efficient, and more costly than methanol
reformers, and their current track record appears to be less advanced than that of
MeOH-based systems(ADL, 2000a). According to Vogel (2001) the 'well-to-wheel'
efficiency of FFR on logistic fuels and methanol systems are similar.
Additional competition must be expected from hydrogen powered light-duty
vehicles, which according to ADL (2000a) represent the fastest route to the market
today, with projected on-board 300 bar hydrogen storage, or the use of certain metal
Appendices 271

hydrides. Notably BMW appears to be investing heavily in this system concept.

Health and safety issues around hydrogen and their perception, however, are
expected to remain a risk to general public acceptance.
Auxiliary power systems face competition from hydrogen based systems: as the fuel
requirement of these systems is limited, probably cost effective and safe hydrogen
storage can be incorporated in a next generation conventional cars. Methanol
powered auxiliary PEMFC systems, however, would allow the use of engine waste-
heat for methanol reforming, thereby increasing the overall 'well-to-wheel' efficiency.
Ballard Power (2001) have recently launched their NexaTM system concept.
In Stationary Power application we foresee some use of methanol based systems in
small-scale applications: those households that do not have a connection to a natural
gas grid. As in automotive, in that case the 'well-to-kW' efficiency and cost of a
methanol based fuel cell system can be expected to be the same or better than that
of a logistics fuel FFR based system. In medium-scale applications, especially those
employed in Uninterrupted Power Supply service (UPS), there may also be a
substantial share of methanol, the alternative again being logistics fuel (UPS) or
natural gas (other).
The fuel of choice for Portable applications is bottled hydrogen: only a fuel cell stack is
needed for power generation.
With respect to the other applications mentioned: it may well be the case that other
types of fuel cells will be employed in these applications, notably Solid Oxide Fuel
Cells.
Finally, in case the Direct Methanol Fuel Cell development is successful, reformer based
FC systems are expected to be phased out entirely in automotive and stationary,
non-natural gas grid connected micro-cogeneration. The required breakthroughs in
technology, however, are dramatic. Realistically, commercial systems cannot be
expected the coming 10-15 years.
272 Process System Innovation by Design

A. 8 Fuel cell, Pt and MeOH scenario model equations

To analyse the possible effects of PEMFC fuel cell developments on platinum and
methanol markets, a scenario model has been developed. A scenario for systems that
incorporate fuel cell technology is translated into total associated platinum and
methanol demand.

Let i = 1….n denote scenario year 1…n

Let j = 1….n denote year of system production 1…n
Let k = 1….m denote market segment 1…m
Let p = 1…q denote parameter 1..q

Let Π be a two-dimensional matrix π[p,k]of parameters for typical 'market' systems

characteristics. Π(1,k) is economic lifespan [year], π(2,k) is rated system output [kW],
π(3,k) is typical system fuel input [ton/kW].
Let D be a two-dimensional matrix of elements d(i, k) denoting market demand per
market segment [no. of systems/year].
Let Τ denote a matrix of scenario parameters, T(j, k) denoting average Technical
Lifespan [year] per year cohort.
Let θ denote a matrix of scenario parameters θ(p,k) determining system phase-out
characteristics per market segment [fraction].
Let R denote a matrix of fractions r(i,j,k) of systems phased out per market segment
[fraction/year]

R = f(T, θ, Π). (Eq. A.8.1)

Let s = (j-i)/ min (π[1,k], T(j,k)) (Eq. A.8.2)

Then the modelling of R as a function T and θ is such that

- ∇ k | ∇ i ≤ j | r(i, j, k) = 0 (Eq. A.8.3)

- ∇k|i=j| r(i, j, k) = θ(1, k) (Eq. A.8.4)

- ∇k|i>j| r(i, j, k) = θ(2, k) (Eq. A.8.5)

- ∇ k | 0.6<s<1.2|r(i, j, k) = f(θ(3..5, k)) (Eq. A.8.6)

Thereby a proxy normal distribution of cohort phase out is modelled around the
average lifespan, which standard deviation increases with average lifespan.
Appendices 273

Let A denote a three-dimensional matrix of elements a(i, j, k) denoting number of

systems in the running fleet [no. of systems].

∇k|i≤j

a(i, j, k) = 0 ; (Eq. A.8.7)

∇k|i>j

a(i, j, k) = a(i-1, j-1, k)*(1-r(i-1, j-1, k)) + d(i-1, k) (Eq. A.8.8)

Let PL be a matrix of scenario parameters, PL(i, k) denoting platinum loading of

systems produced [g/kW].
Let Pt be a matrix of elements Pt (p, i, k) indicating platinum total demand Pt (1, i,
k) , recycling demand/supply Pt (2, i, k) and net platinum virgin demand Pt(3, i, k),
all in [g/year]
Pt (1,i,k) = d(i,k)* π(2,k)*PL(i,k) (Eq. A.8.9)

Pt (2,i,k) = ∑j=1,i {a(i-1, j-1, k)*r(i-1, j-1, k))*π(2,k)*PL(i-1,k)} (Eq.A.8.10)

Pt (3,i,k) = Pt (1, i, k) - Pt(3, i, k) (Eq.A.8.11)

Pt_stored = ∑i ∑k Pt (3,i,k) (Eq.A.8.12)

Let MF be a matrix of scenario parameters, MF(i, k) denoting fraction of systems
fuelled by methanol[fraction].
Let MeOH be a matrix of elements MeOH(i, k) indicating methanol fuel demand
[ton/year]
MeOH (i) = ∑k ∑j=1,i {a(i, j, k))*π(3,k) (Eq.A.8.13)
274 Process System Innovation by Design

A. 9 Key issues with respect to PEMFC success

The success or breakthrough of PEMFC is driven viz. depends on the economy,
ecology and technology of PEMFC systems for the realisation of 'automotive
transport'.
Ecology leads to an ever increasing 'sense-of-urgency' with respect to the necessity of
fuel cell technology to become successful for the benefit of mankind. On a global
scale, there is the threat of global warming by human CO2 and other hydrocarbon
emissions, as well as the concern for depletion of fossil resources. On a macro-scale,
notably in the US there is a growing concern of the country's dependence of
(foreign) oil.
On a meso- or micro-scale there is the issue of air quality, that is seriously affected
by soot-emissions (diesel engines) and CO and unconverted hydrocarbons (gasoline)
that result in heavy, toxic smog in cities such as Athens, Mexico City and Greater
Los Angeles. Smog-related problems have been the driving factor behind the series
of Policy and Legislative packages issued by the State of California over the last 20
years.
On a meso- and micro-scale, noise pollution is an issue that increasingly puts a
constraint on expansion of transport-related activities.
Economy of fuel cell systems appears to become less and less a critical issue, as more
and more niche markets are developed, mass production is foreseen to begin in
2005, and regulatory constraints on the automotive and logistics fuel systems
become increasingly tighter.
Technology development today, however, is still largely focused on (Chalk, 2000)
(1) bringing down 'System and Component' costs
The fuel cell requires 60% of current system costs calculated at $300/kW for
a 50 kW system (ADL, 2000a), the fuel processor some 20%, system
assembly and 'Balance-of-Plant' accounting for the remainder 10%. Notably,
the cost of the so-called Membrane Electrode Assembly (MEA) must go
down. According to ADL, the cost of the stack in today's 50kW systems still
accounts for almost 50% of system cost (76% of fuel cell cost). At least 30%
of FC system cost are material related, and platinum alone accounts for 20-
25% of system cost, or some $60/kW!
(2) increasing durability to extend the operating lifespan of key components
Throughout the life of a fuel cell, its performance (% fuel converted to
usable power) decays slowly. Today, some 40,000 hrs of continued feasible
PEMFC operation is possible (ADL, 2000a). Especially for stationary
systems, this figure (5 yrs of continued service) is still to low. In automotive
systems, due to the nature of drive-cycles to be expected, stack life is
expected to be much less. Indeed, according to Chalk, durability of MEAs
and fuel processor catalysts up to 5,000 hrs. has not been demonstrated yet,
and according to ADL, system's durability beyond 2000 hrs of service with
power degradation (< 5%) must still be demonstrated.
Appendices 275

(3) improving the dynamic performance and reliability of fuel processors

The CO content of hydrogen-rich gas sent to the FC stack presently must be
lower than 100 ppm in order not to poison the catalytic MEA and thereby
avoid drastic reduction of its lifespan. Catalyst development and MEA
development is under way for higher 'CO-tolerance'. Meanwhile, system
start-up times of 20-30 seconds must be realised, compared to the current 6
to 20 minutes.
(4) Improvement of air management
Both air delivery (sub-component: air compression) and utilisation (sub-
component: the fuel cell) must be improved to improve overall system
efficiency, volume and weight.
(5) health and safety
Health and safety issues must be addressed. Specifically, the high operating
temperature of gasoline-based systems is seen as problematic (ADL 2000a).
The perceived risks of hydrogen are high, whilst in actual practice huge
hydrogen distribution networks have operated without problems for decades.

The current 'distance-to-target is summarized for a number of critical performance

characteristics for Passenger Vehicles operating on Gasoline is indicated in Figure
A.9.1. This serves to illustrate why MeOH-fuelled PEMFC powered AGVs present
such an interesting option:
• Specific Power, Power Density and Start-up to Full Power are non-critical issues
for AGVs. A large weight is an advantage.
• Transient response is less of an issue, as driving cycles for AGVs are considered
less demanding than for passenger vehicles
• In relation to noise and other emissions, cost is not so much of an issue. Energy
efficiency must move from 34 to some 50% of system efficiency to achieve
acceptable 'well-to-wheel' efficiency
• System durability is for methanol powered vehicles is probably a much less
problematic issue than in gasoline powered FC-cars. Extensive tests are
underway with hydrogen and methanol fuelled cars and buses to prove
durability. Apparently, there is still a durability issue in gasoline reformer
catalysts that must operate around 900-1100 oC. In methanol reforming, a
different catalyst can be used, and operation is around 350 oC.
276 Process System Innovation by Design

Figure A.9.1: Status of 50 kW gasoline operated PEMFC development (from ADL,

2000a:3).

Finally, with respect to health and safety methanol does not have the disadvantage of
high-temperatures and it avoids storage of hydrogen. Compared to gasoline, health
issues appear to be rather similar (Stikkelman, 2001).
Summary
In the petrochemical industry large-scale chemical processes are employed to
convert petroleum-derived feedstock into base chemicals. Huge industrial complexes
of many interconnected processes are the blue-print image of this industry, which is
linked to virtually all important economic sectors such as agriculture, automobiles,
construction, electronics, housing, pharma, textiles, consumer durables and
packaging. The core business of this industry of industries is to transform finite
resources of fossil origin into building blocks of polymers - plastics, resins, fibres.
Petrochemical operations also generate by-products that can be detrimental to our
health, safety and environment if emitted or wasted. In the 1960-70's, the state-of-
the-art industrial practice led to the persistent label ‘Dirty, Damaging, and
Dangerous’ of 'chemical industry'. Over time, however, the control and prevention
of harmful emission and waste management have improved dramatically. Carbon
dioxide, however, continues to be emitted because CO2 formation is linked with
petrochemical process energy supply, feedstock, product and polymer degradation.
To accommodate stable or modest growth in petrochemicals demand, the industry
must timely respond to the prospects of fossil resource depletion and a cap on CO2
emission induced by global-warming concerns. A petrochemical industry is required
that is characterised by appropriate resource selection, effective resource utilisation,
the avoidance of waste and emissions and a product slate that fosters a sustainable
society.
How can these objectives be met? In addressing this question, one must realize that
the petrochemical industry has the structure of an interconnected network. The
nodes represent the processing plants, the vertices are the exchange flows of
intermediate products and energy.
The networked petrochemical industry has an impressive record of innovation of its
industrial plants by invention and improvement of system components - e.g.
catalysts, reactors, separation technology- and products enabled by fundamental and
applied science. Industry maturation, however, has reduced the scope for process
innovation by “inventing new flow sheets for existing problems.” Changes of its
structure caused by R&D breakthroughs or by the use of extra-sector innovations
have become incidental.
The system structure of the industry generally has been considered fixed and the
effect of most R&D on industry network structure is not planned. A literature
review revealed that conceptual design of the industry structure and the specification
of R&D themes for multiscale systemic industry change is a 'white spot' on the
science and engineering map. The discipline of engineering science that comes
closest to dealing with these issues is process system engineering. PSE has evolved
as a sub discipline of chemical engineering that “is concerned with the understanding
and development of systematic procedures for the design and operation of chemical
process systems, ranging from microsystems to industrial scale, continuous and
batch processes.” In PSE multiscale process synthesis approaches that “span a
chemical supply chain -from mastering the molecules to chemical enterprises-” only
have been suggested, however.
280 Process System Innovation by Design

Starting from a higher level of aggregation , economy and policy studies could offer
options, but these suffer from technological lack-of-content. We conjectured
therefore a sustainable petrochemical industry requires process system innovation,
which is defined as “change in the system structure or system design of the
petrochemical industry, its industrial complexes, or individual plants. These can be
enabled by technological inventions or vice versa.” In technology and systems
development and in the adoption of extra-sector innovations the industry structure
explicitly must be considered a degree-of-freedom. Therefore, the central theme of
this thesis is the specification of system innovation content for a sustainable
petrochemical industry.
Systemic innovations do require a long time for development and implementation.
Early identification of promising concepts thus is preferred to facilitate sustainable
development at affordable and acceptable costs and to maximise the effect of R&D.
Therefore, the research questions addressed in this dissertation were (1) what is a
suitable method to assess the resource utilisation and the scope for improvement of
the production systems in the petrochemical industry, (2) can the search of the
innovation space be structured to foster the specification of process system
innovation content and (3) what is the usefulness and the scope of application of the
methods thus developed?
These research questions have been addressed using system theory,
thermodynamics, PSE and computer systems modelling. The concept of and the
methodology for 'process system innovation by design' was developed, validated and
tested in case studies that each span multiple system aggregate levels.
In a technology-free systems representation using the input/output paradigm, any
system is considered a ‘black-box’, which upon opening may consist of a structured
collection of system elements or sub-systems. At each system aggregate level it is
assumed that the function of a system is to generate a behaviour to transform a set
of inputs into a set of outputs in order to satisfy an objective. Together, these relate
the system to its external world. Its performance is assessed by means of criteria or
value functions. The petrochemical industry is a system embedded and linked to the
global material cycles. In a layered networked system view, its system elements are
petrochemical complexes and individual plants. In chemical engineering, these plants
are modelled as structured collections of unit operations connected by flows.
To define indicators for sustainability, mass and energy balancing and stream
valuation on the basis of input scarcity, output fate, economy and ecology were
combined in criteria for system resource utilisation. In quantitative analysis these
indicators yield an image of performance. Together with known plant capacities,
such quantitative system assessment results in a ranking of petrochemical plants with
respect to performance and scope for improvement. This ranking is a starting point
for systematic search for R&D options and themes.
An early revelation of our research was that inclusion of fuel cells in chemical plants
would have a dramatic effect on plant performance by exploiting hitherto
unrecognised fuel cell functionality, which allows efficient interconversion of
chemical, electric and thermal energy. The functional description of the fuel cell as a
unit operation and the subsequent recognition petrochemical plants throughout the
Summary 281

industry as candidates for fuel cell integration, however, requires some suitable
method for system decomposition.
Functional modelling for computer system design, was related to and adapted for the
technology-free petrochemical system representation using the input-output
paradigm. It enables cohesive system decomposition using 'mutual contingency' as a
stop-criterion, which indicates the system decomposition exhibits loose coupling and
sufficient relation between the system elements. The criterion is formalised in a
logical algorithm. In the multiscale modelling exercise subjective functions describe
the system and parts thereof. Upon completion of a mutual contingent
decomposition the functional model is built.
The modelling of system functions rather than system technology provides a
mechanism for abstraction from current design and physical realisation of industrial
systems. Stream valuation and functional modelling were the missing links between
resource utilisation assessment and the systemic exploration of the multiscale
innovation space for chemical conversion and energy interconversion. In system
synthesis, these links enable the specification of process system innovation content -
or the formulation of primitive problems under the General Design Paradigm. The
main result is a methodology for said specification that is based on system
representation using the input-output paradigm and functional modelling.
The use of fuel cells in the petrochemical industry leads to trigeneration systems that
produce useful chemical products, electricity, and heat. Functional modelling helps
to specify new trigenerate system concepts at the aggregate level of reactors,
petrochemical plants and complexes, one of which was patented. Addressing
innovation for the industrial olefins system showed that ongoing R&D and
innovations reported can be categorised and augmented. Quantitative assessment
using stream valuation allows one to focus on 'the weakest link'. Functional
modelling enabled specification of a true process system innovation, the integration
of olefins and syngas production from aliphatic hydrocarbons. To exploit a radical
extra-sector innovation, the advent of methanol-fuelled fuel cell vehicles, process
system innovations were specified around the Rotterdam petrochemical cluster,
whereby the Rotterdam process industry becomes a key industrial infrastructure in
methanol, olefins and platinum.
The results of these case studies illustrate that the methodology fosters 'out-of-the-
box' thinking at the aggregate level of unit operations, single plants, industrial
complexes and clusters alike. Since quantitative evaluation of each innovation
specified requires extensive work, some proxy methods for evaluation have been
adopted, such as economic assessment of fuel cell reactor concepts and dynamic
simulation of fuel cell vehicle adoption and its consequences for methanol and
platinum. The innovation results and proxy evaluations do not constitute formal
proof or validation of the methodology. It remains possible that the concepts
specified could have been developed otherwise. Extensive literature research,
however, did not reveal similar structured methods that link creative thinking with
assessment of performance, structure and technology of existing complex systems.
The methodology and concepts presented in this dissertation can be improved and
extended. Qualitative tests that involve experts can be developed to further underpin
the innovations specified. Thereby process system innovation can be linked to
282 Process System Innovation by Design

(inter)national public policy development, regulation and corporate strategy. In the

design process and PSE the relation between innovation, systems analysis, synthesis
and conceptual design can be elaborated. The quantitative assessment should be
extended dynamic modelling of global material cycles. A transition from chemical
conversion to bio-chemical conversion would enable a global carbon-cycle that
includes fossil resources, renewable bio-resources, petrochemicals, polymer waste
and control of global atmospheric CO2. Thereby a bridge between industrial ecology
concepts and the body-of-knowledge of process system engineering would be
forged. Thus the methodology can be expanded to a portfolio of methods,
procedures and tools, which allows dealing with 'wicked' environmental problems by
fostering innovation for sustainability in many industries - refining, base metal, pulp
and paper, agro- and food industry.
The role of the petrochemical industry as 'an industry of industries' is critical to
achieve sustainability. The methodology described in this dissertation can help this
industry, its technology suppliers, engineering service providers and academia to
specify process system innovations. In addition, the methodology can facilitate
interaction between all stakeholders involved in a transition towards a sustainable
society by forging a link between societal needs, business strategy and process
system innovation.
Samenvatting
In de petrochemische industrie worden grootschalige chemische processen gebruikt
voor de omzetting van aardoliegrondstoffen in basischemicaliën. Grote industriële
complexen die bestaan uit veel onderling verbonden processen bepalen het beeld
van deze industrie. De petrochemie heeft een relatie met bijna alle economisch
belangrijke sectoren, te weten de landbouw, de automobiel industrie, de weg- en
waterbouw, elektronica, woning- en utiliteitsbouw, de farmaceutische industrie, de
textielindustrie, duurzame consumptiegoederen en verpakkingen. De kernactiviteit
van deze industrie van industrieën is de omzetting van eindige grondstoffen van
fossiele oorsprong in bouwstenen van polymeren - plastics, harsen en vezels.
Petrochemische fabrieken genereren ook bijproducten die schadelijk kunnen zijn
voor onze gezondheid en veiligheid en het leefmilieu wanneer ze worden
uitgestoten of gestort als afval. Door de toenmalige stand der techniek en industriële
praktijk verkreeg de chemische industrie in de jaren '60 en '70 het etiket 'Vies,
Schadelijk en Gevaarlijk'. Dit etiket is erg hardnekkig gebleken, ondanks het sterk
terugdringen van schadelijke emissies en het ontstaan van gevaarlijk afval. Echter, de
emissie van koolstofdioxide, CO2, gaat onverminderd door omdat de vorming van
CO2 direct gekoppeld is aan het grondstofgebruik en de energievoorziening van
petrochemische installaties en een onvermijdelijk gevolg is van de uiteindelijke
afbraak van organische verbindingen zoals basischemicaliën en polymeren.
Om tegemoet te kunnen blijven komen aan de naar verwachting blijvende en
stijgende vraag naar petrochemische producten moet de industrie tijdig maatregelen
nemen. Ze zal adequaat moeten anticiperen op de uitputting van fossiele
grondstoffen én op de instelling van een plafond aan de toegestane CO2 uitstoot in
verband met wereldwijde zorg om klimaatverandering. Er is kortom een
petrochemische industrie nodig die wordt gekenmerkt door de keuze van geschikte
grondstoffen, door effectief gebruik van deze grondstoffen, door het vermijden van
afval en emissies, en door een productspectrum dat helpt een duurzame samenleving
te realiseren.
Hoe zijn deze doelen te realiseren? Bij de beantwoording van deze vraag is het goed
ons te realiseren dat de petrochemische industrie een netwerkstructuur heeft. De
knopen in het netwerk zijn de petrochemische fabrieken, langs de lijnen worden
tussenproducten en energie uitgewisseld. Deze genetwerkte petrochemische
industrie heeft een indrukwekkende staat van dienst wat betreft procesinnovatie en
productontwikkeling ondersteund door fundamenteel- en toegepast-
wetenschappelijke kennis. Voor haar fabrieken worden voortdurend
systeemcomponenten verbeterd en nieuwe componenten ontwikkeld zoals
katalysatoren, reactoren en scheidingstechnologieën. Omdat de volwassen chemische
industrie nog slechts matig groeit zijn er slechts beperkte financiële prikkels voor
procesinnovatie - het vinden van nieuwe processchema's voor bestaande problemen.
Fundamentele wijzigingen van haar structuur onder invloed van wetenschappelijk
doorbraken of door de toepassing van innovaties van buiten de sector zijn
incidenteel. De structuur van de industrie ligt in hoofdzaak vast en effecten van
R&D op deze netwerkstructuur zijn meestal niet gepland.
284 Process System Innovation by Design

Een literatuurstudie heeft laten zien dat conceptueel ontwerp van de

industriestructuur en de specificatie van R&D die een verandering van de industrie
op meerder schaalniveaus bewerkstelligt een witte vlek is op de onderzoekskaart. De
technische discipline die in aanzet dit probleem aanpakt is chemisch fabrieksontwerp
of 'process systems engineering' (PSE). Dit vakgebied heeft zich ontwikkeld als een
subdiscipline van de scheikundige technologie en is gericht op 'het begrijpen van en
de ontwikkeling van systematische procedures voor het ontwerp en de operatie van
chemische voortbrengingsprocessen, van het microschaalniveau van moleculen tot
de microschaal van de industrie, van continue en van ladingsgewijs bedreven
processen". In PSE zijn echter tot dusver ontwerpmethoden die "een gehele
chemische voortbrengingsketen omspannen - van het beheersen van moleculen tot
meerdere chemiebedrijven" slechts gesuggereerd. Economisch of beleidsmatig
georiënteerde studies zouden geschikte aanknopingspunten kunnen bieden omdat ze
juist van dit hogere macroschaalniveau uitgaan, maar dit type studies lijdt over het
algemeen aan een gebrek aan technologische inhoud. Wij postuleerden daarom dat
voor een duurzame petrochemische industrie proces systeeminnovatie nodig is. Dit
is gedefinieerd als "veranderingen in de systeemstructuur of het systeemontwerp van
de petrochemische industrie, haar industriële complexen, of individuele fabrieken.
Deze worden mogelijk gemaakt door technologische inventies en omgekeerd". In de
ontwikkeling van technologie en systemen en bij het gebruik van innovaties van
buiten de sector moet de industriestructuur expliciet beschouwd worden als een
vrijheidsgraad. Daarom is het centrale thema van dit proefschrift de specificatie van
de inhoud van systeeminnovatie voor een duurzame petrochemische industrie.
Systeeminnovaties vergen een lange periode voor ontwikkeling en implementatie.
Vroege identificatie heeft daarom de voorkeur om duurzame ontwikkeling mogelijk
te maken tegen betaalbare en acceptabele kosten. Doel is het te verwachten effect
van R&D inspanningen te maximaliseren. Daarom zijn in dit proefschrift de
volgende onderzoeksvragen uitgewerkt: (1) wat is een geschikte methode om de
benutting van grondstoffen en energiedragers te beoordelen en het
verbeteringspotentieel vast te stellen? (2) Kan het verkennen van de innovatieruimte
worden gestructureerd ter bevordering van de specificatie van systeeminnovaties? (3)
wat is de gebruikswaarde en de reikwijdte van de aldus ontwikkelde methodes?
Deze onderzoeksvragen zijn uitgewerkt onder gebruikmaking van systeemtheorie,
thermodynamica, PSE en modellering van computer- en informatiesystemen. Het
concept en de methode voor 'systeeminnovatie op recept' is ontwikkeld, gevalideerd
en bewezen in een drietal case studies die elk meerdere schaalniveaus bestrijken.
In een technologievrije representatie van een systeem gebruikmakend van het
input/output paradigma kan elk systeem beschouwd worden als een zwarte doos, die
als ze geopend wordt bestaat uit een gestructureerde verzameling systeemelementen
of subsystemen. Op elk aggregatieniveau wordt aangenomen dat de systeemfunctie is
'het genereren van gedrag om een set inputs te transformeren in een set outputs en
daarmee een doelstelling te bewerkstelligen'. Gezamenlijk relateren deze het systeem
aan de wereld buiten het systeem. De systeemprestatie wordt beoordeeld met behulp
van criteria of waardefuncties.
De petrochemische industrie is een systeem dat onlosmakelijk verbonden is met
wereldwijde materiaalcycli. In een gelaagde en genetwerkte systeemvisie zijn haar
Samenvatting 285

systeemelementen industriecomplexen of chemische fabrieken. In de scheikundige

technologie worden deze gemodelleerd als gestructureerde verzamelingen van
eenheidsbewerkingen (zgn. 'unit operations') die met elkaar zijn verbonden door
massa- en energiestromen.
In dit proefschrift zijn indicatoren voor duurzaamheid van systemen geformuleerd
door het opstellen van massa- en energiebalansen te combineren met de waardering
van stromen op basis van de schaarste van grondstoffen en energiedragers, de
bestemming van producten en economische en ecologische aspecten. Deze
combinatie levert criteria voor grondstofbenutting van systemen waarmee in een
kwantitatieve analyse de prestatie van individuele fabrieken kan worden beoordeeld.
Door de resultaten te combineren met de capaciteit van fabrieken kan een rangorde
naar prestatie of verbeteringspotentieel worden verkregen van alle petrochemische
installaties in de industrie. De zwakste schakels vormen een startpunt voor het
gestructureerd zoeken naar R&D onderwerpen en thema's.
Een vroege constatering op basis van ons onderzoek was dat de integratie van
brandstofcellen in chemische fabrieken een dramatische verbetering van de prestatie
van de laatste kan bewerkstelligen. Dat komt omdat voorheen niet onderkende
functionaliteit van de brandstofcel in een chemische fabriek kan worden benut om
meer efficiënte omzetting van chemische, elektrische en thermische energie te
realiseren. De functionele beschrijving van de brandstofcel als een 'unit operation'
die onderdeel kan uitmaken van een chemisch procesontwerp, de integratie van de
brandstofcel in het procesontwerp en de zoektocht naar kandidaat-industriële
processen vroeg echter om een goede methode voor het onderverdelen van
systemen in subsystemen en systeemelementen - systeemdecompositie.
Het modelleren met behulp van objecten en functies - hierna functioneel modelleren
- zoals ontwikkeld voor het ontwerpen van computer- en informatiesystemen is
gemodificeerd en gerelateerd aan de technologievrije representatie van chemische
processen onder gebruikmaking van het input-output paradigma. De zo ontwikkelde
methode maakt functioneel coherente systeemdecompositie mogelijk. Daarbij wordt
als stopcriterium het bereiken van 'mutual contingency' gehanteerd. Dit betekent
zoveel als 'een verdeling die wordt gekarakteriseerd door een balans tussen
wederkerige afhankelijkheid en voldoend onderscheidende ontkoppeling'. Dit
criterium is geformaliseerd in een logisch algoritme. Tijdens het functioneel
modelleren worden - subjectief - op meerder systeemniveaus functies geformuleerd
die betrekking hebben op het systeem of delen daarvan. Als aan het criterium wordt
voldaan, is de decompositie functioneel coherent en het model voltooid.
Het modelleren van functies in plaats van de technologie van systemen biedt een
mechanisme voor abstractie van bekende ontwerpen en bestaande fysieke realisatie
van industriële systemen. Het waarderen van stromen en functioneel modelleren
vormen de ontbrekende schakels tussen het beoordelen van
grondstofbenutting/energiegebruik en de het op meerdere schaalniveaus tegelijk
verkennen van de innovatieruimte voor chemische conversie en energieomzetting.
In de synthese van systemen zijn het deze schakels die specificatie van de inhoud
nodig voor 'systeeminnovatie op recept' mogelijk maken. Met andere woorden, de
formulering van primitieve problemen in het Algemeen Ontwerp Paradigma wordt
hiermee gefaciliteerd.
286 Process System Innovation by Design

Het hoofdresultaat beschreven in dit proefschrift is een methodologie voor

'systeeminnovatie op recept' gebaseerd op de representatie van systemen onder het
input/output paradigma en gebruik makend van functioneel modelleren.
Het gebruik van brandstofcellen in de petrochemische industrie leidt tot
trigeneratiesystemen die tegelijkertijd bruikbare chemische producten, elektriciteit en
warmte produceren. Functioneel modelleren maakt het mogelijk om nieuwe
systeemconcepten voor trigeneratie te specificeren op het systeemniveau van
reactoren, fabrieken en industriecomplexen. Op een dergelijk concept is octrooi
verleend. De case studie naar innovatie rond olefines heeft laten zien dat functioneel
modelleren te gebruiken is voor het ordenen van lopend onderzoek en in de
literatuur gerapporteerde innovaties. De kwantitatieve analyse leidde inderdaad tot de
zwakste schakels in het systeem. Een echte systeeminnovatie is op recept
ontwikkeld, namelijk de combinatie van synthese gas- en olefine-productie uit
lineaire koolwaterstoffen. Om de chemie een radicale innovatie van buiten de sector
te laten benutten - de komst van brandstofcelauto's op methanol- zijn genetwerkte
systeeminnovaties ontwikkeld voor het Rotterdams petrochemisch cluster. Indien
deze systeeminnovaties werkelijk worden gerealiseerd, dan wordt de Rotterdamse
procesindustrie een belangrijke schakel in de industriële infrastructuur voor
methanol, olefines en platina.
De resulaten van de case studies laten zien dat methodologie voor 'systeeminnovatie
op recept' het denken buiten bestaande kaders stimuleert cq. faciliteert, of het nu het
schaal niveau van unit operations, fabrieken, industriecomplexen of
industriesectoren betreft. Omdat verdere uitwerking en kwantitatieve analyse van
elke innovatie op recept zeer veel werk met zich mee zou brengen, zijn een aantal
methoden ontwikkeld voor evaluatie op hoofdlijnen, te weten economische analyse
van brandstofcelreactoren en een dynamisch model dat markt- en
technologiescenario's voor brandstofcelvoertuigen simuleert. Daarmee kunnen in
een vroeg stadium de consequenties voor methanol en platina worden geïllustreerd.
De gespecificeerde systeeminnovaties en de evaluaties vormen geen formeel bewijs
of sluitende validatie voor de ontwikkelde methodologie. Immers, de gespecificeerde
concepten zouden ook op een andere wijze kunnen worden ontwikkeld. Uitgebreide
literatuurstudie heeft echter geen vergelijkbare gestructureerde methoden opgeleverd
die creativiteit verbinden met prestatieanalyse en de structuur en technologie van
bestaande complexe systemen.
De methodologie en de concepten gepresenteerd in dit proefschrift kunnen op een
aantal punten worden uitgewerkt. Kwalitatieve testprocedures dienen te worden
ontwikkeld waarin experts worden geraadpleegd. Daarmee kunnen de
gespecificeerde concepten verder worden onderbouwd en uitgewerkt. Ook kan zo
'systeeminnovatie op recept' gerelateerd worden aan (inter)nationale
beleidsontwikkeling, regulering en de strategie van bedrijven. De relatie tussen
innovatie, systeemanalyse en -ontwerp voor duurzaamheid kan verder worden
uitgewerkt door de kwantitatieve analyse uit te breiden naar de wereldwijde
materiaalcycli en de analyse van dynamische systemen. In het ontwerpproces en in
PSE dient de relatie tussen de probleemformulering, systeemanalyse, synthese van
het primitief ontwerpen en het conceptueel systeemontwerp verder uitgewerkt te
worden. Een transitie van chemische conversie naar biochemische conversie zou een
Samenvatting 287

wereldwijde koolstofcyclus mogelijk maken waarin optimaal gebruik wordt gemaakt

van fossiele bronnen, hernieuwbare biogrondstoffen, petrochemische producten,
plastic afval terwijl tegelijkertijd het atmosferisch CO2-gehalte wordt beheerst.
Daardoor wordt een brug geslagen worden tussen de concepten zoals geformuleerd
in industriële ecologie en de opgebouwde kennis in 'process systems engineering'.
De methodologie voor 'systeeminnovatie op recept' kan zo worden uitgebouwd tot
een palet van methodes, procedures en hulpmiddelen waarmee een oplossing voor
de 'taaie' milieuproblemen van deze tijd dichterbij komt. Gebruik van dit palet zal
helpen innovatie voor duurzaamheid tot stand te brengen in vele sectoren van de
economie - olieraffinage, basismetaal, de pulp- en papierindustrie en de agro- en
voedings- en genotmiddelenindustrie.
De rol van de petrochemische industrie als een industrie van industrieën is cruciaal
voor de verduurzaming van onze samenleving. De methodologie beschreven in dit
proefschrift kan de spelers in deze industrie, haar technologie- en diensten
leveranciers en universitaire onderzoekers helpen om systeeminnovaties te
specificeren. Door het tot stand brengen van interactie tussen alle spelers in de
kenniseconomie en door afstemming van systeeminnovaties op de strategie van
bedrijven en de behoeften van de samenleving, tenslotte, zal het gebruik van de
methodologie de transitie naar een meer duurzame samenleving bevorderen.
 Curriculum Vitae
Gerhard Pieter Jan Dijkema was born in Veenhuizen, municipality Norg. He
completed the Gymnasium-β in Assen at the “Dr. Nassau College” (1980) and
graduated as a chemical engineer (hons.) from Twente University of Technology
(1986). After his graduation he joined Shell International in the Hague. Among
others, he spent 2 years on the start-up team of the NAM sour-gas-desulphurisation
plant in Emmen. In 1990 he married Elly van der Gugten. Together they have two
children, Jenske (1990) and Roeland (1992). In 1991 he joined Interduct, the Delft
University Clean Technology Institute, where his research focused on the large-scale
process industry and infrastructures for power generation and waste management.
Since mid-1996 he is affiliated with the faculty of Technology, Policy and
Management, section Energy and Industry. The theme of most of his education and
research is the interaction between industry, government and management, with an
emphasis on innovation for improved sustainability through system thinking and
sound technology. A second focus is decision support for the development and
transition of large-scale systems that span industry and infrastructure sectors. Since
1999 he is Academia Liaison for the local section of the American Institute of
Chemical Engineers.