AIChE Journal - 2020 - Zhao - Waste High Density Polyethylene Recycling Process Systems For Mitigating Plastic Pollution

15475905, 2021, 4, Downloaded from https://aiche.onlinelibrary.wiley.com/doi/10.1002/aic.17127 by Cornell University, Wiley Online Library on [07/11/2022].
See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Received: 21 May 2020 Revised: 28 October 2020 Accepted: 24 November 2020
DOI: 10.1002/aic.17127
PROCESS SYSTEMS ENGINEERING
Waste high-density polyethylene recycling process systems

for mitigating plastic pollution through a sustainable design
and synthesis paradigm
Xiang Zhao1 | Fengqi You1,2
1
Systems Engineering, Cornell University,
Ithaca, New York Abstract
2
Robert Frederick Smith School of Chemical This article addresses the sustainable design and synthesis of open-loop recycling
and Biomolecular Engineering, Cornell
process of waste high-density polyethylene (HDPE) under both environmental and
University, Ithaca, New York
economic criteria. We develop by far the most comprehensive superstructure for
Correspondence
producing monomers, aromatic mixtures, and fuels from waste HDPE. The super-
Fengqi You, Systems Engineering, Cornell
University, Ithaca, NY 14853, USA. structure optimization problem is then formulated as a multi-objective mixed-integer
Email: [email protected]
nonlinear fractional programming (MINFP) problem to simultaneously optimize the
Funding information unit net present value (NPV) and unit life cycle environmental impacts. A tailored
Division of Chemical, Bioengineering,
global optimization algorithm integrating the inexact parametric algorithm with the
Environmental, and Transport Systems, Grant/
Award Number: 1643244 branch-and-refine algorithm is applied to efficiently solve the resulting nonconvex
MINFP problem. Results show that the optimal unit NPV ranges from $107.2 to
$151.3 per ton of HDPE treated. Moreover, the unit life cycle greenhouse gas emis-
sions of the most environmentally friendly HDPE recycling process are 0.40 ton
CO2-eq per ton of HDPE treated, which is 63% of that of the most economically
competitive process design.
KEYWORDS
life cycle assessment, superstructure optimization, tailored global optimization algorithm,
techno-economic analysis, waste HDPE recycling process systems
1 | I N T RO DU CT I O N to recycle waste plastics.8 The closed-loop process produces recov-

ered plastics with high cost, while the open-loop process manufac-
Plastic is a common material used in households and factories around tures various products that can be applied rather than producing
the U.S.1 With uninterrupted development in plastic production, the original plastics.8 Therefore, implementing the open-loop recycling
whole country has long been self-sufficient in plastics. A large amount process provides an opportunity to enhance the economic viability
of waste plastics are generated by households, yet only 8% of the and environmental sustainability of waste plastic processing.
plastics are recycled.2 Comparatively, 76% of waste plastics go to Chemical recycling methods used in open-loop waste plastics
landfill, which contribute to growing environmental concerns.2 Land- recycling process, including pyrolysis,9-11 gasification,12 as well as fluid
filled plastics deteriorate the ecosystem via breaking down into toxic catalytic cracking (FCC),13 have drawn growing attention in the con-
3
substances that are ingested by plants, or through watering into text of circular economy. Among these methods, pyrolysis is often
microplastics which can be accumulated in the food web.4-6 To reduce used for cracking plastics that are hard to depolymerize, such as high-
the share of detrimental landfill processes in plastic waste treatment, density polyethylene (HDPE),14 and eliminates the pollution associ-
additional large-scale facilities that are dedicated to processing plastic ated with traditional landfill process. Furthermore, this process can
waste must be designed and established in the short term.7 Closed- handle highly contaminated plastic waste, such as the automotive
loop and open-loop recycling processes are two main methodologies shredder residue contaminated by chloride or heavy metal, to produce
AIChE J. 2021;67:e17127. wileyonlinelibrary.com/journal/aic © 2020 American Institute of Chemical Engineers 1 of 25

https://doi.org/10.1002/aic.17127
15475905, 2021, 4, Downloaded from https://aiche.onlinelibrary.wiley.com/doi/10.1002/aic.17127 by Cornell University, Wiley Online Library on [07/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2 of 25 ZHAO AND YOU
economically profitable pyrolysis oil.15,16 Specifically, fast pyrolysis involved in light component separation, heavy chemical component
process has a high yield of monomers and pyrolysis oil.17,18 Given the separation, and aromatic extraction sections to reduce utility usage.
relatively high profitability and large demand for these products from Based on the superstructure, we apply a cradle-to-gate LCA, and the
HDPE, several economically viable fast pyrolysis processes have been functional unit is defined as one ton of waste HDPE treated in the
designed.19,20 Differences between these processes lie in types of recycling process systems. The system boundary consists of five life
reactors and reaction conditions. Besides, various studies have been cycle stages, namely HDPE collection, separation, and transportation,
performed to emphasize on the environmental impacts derived from natural gas production and transportation, waste HDPE processing,
the fast pyrolysis process accounting for the global warming potential external utilities and electricity generation, and wastewater treatment.
(GWP) indicator or ReCiPe end-point score indicator.21 Recent works The LCA is further integrated with TEA to provide environmental and
combine these two viewpoints to address environmental sustainability economic parameters to the corresponding superstructure optimiza-
through fixed technologies when conducting techno-economic analy- tion problem, which aims to optimize the economic performance and
sis (TEA) and life cycle assessment (LCA).22 However, the lucrative life cycle environmental impacts simultaneously. We apply fractional
pathway for the open-loop recycling process of waste HDPE evalu- objective functions to calculate the net present value (NPV) and
ated depends on the market condition.23 To address this concern, it is greenhouse gas (GHG) emissions or ReCiPe end-point score per ton
important to investigate the sustainable design of waste HDPE of HDPE treated, so as to avoid optimizing linear objective functions
recycling process systems in consideration of the economic perfor- that result in an unreasonably maximum or minimum amount of HDPE
mance and environmental impacts. Various studies systematically treatment. Optimizing fractional objective functions can lead to a bal-
addressed the design and synthesis of sustainable process systems to anced treatment amount, so that the sustainable design of and syn-
24,25
produce biofuels, or to process shale gas. However, although vari- thesis of waste HDPE recycling process systems can be systematically
ous researchers have attached importance to environmental sustain- addressed.33 The superstructure optimization problem is then formu-
26,27
ability in the recycling process of waste HDPE, the issue on the lated as a multi-objective mixed-integer fractional programming
sustainable design of HDPE recycling process systems has not been (MINFP) problem. We simultaneously optimize the economic and
systematically addressed. environmental objective functions in this problem. The economic
Several research challenges need to be tackled to systematically objective function is to maximize the NPV within the project lifetime
address the sustainable design and synthesis of waste HDPE recycling per ton of HDPE treated (unit NPV), and the environmental objective
process systems. The first challenge is to generate a novel and compre- function is to minimize the life cycle GHG emissions or life cycle ReC-
hensive superstructure of waste HDPE recycling process systems that iPe end-point score per ton of HDPE treated, respectively. However,
can simultaneously manufacture high-purity products and save energy. the combinatorial nature and the pseudo-convexity of fractional
High-quality products and minimum energy usage can improve eco- objective functions make this MINFP problem computationally chal-
nomic competitiveness and environmental sustainability for HDPE lenging for general-purpose global optimizers.34 Therefore, we apply a
recycling process systems. The second challenge is associated with the tailored global optimization algorithm integrating the inexact paramet-
acquisition of high-fidelity process parameters for the corresponding ric algorithm with the branch-and-refine algorithm to tackle this com-
superstructure optimization problem. Key parameters, such as split frac- putational challenge. The proposed framework is applied to an HDPE
tions and capital costs of the HDPE recycling process, are unavailable in recycling process design with the upper bound of treatment capacity
the existing literature. Therefore, it is essential to perform rigorous pro- around 295,000 ton HDPE/year in the U.S., corresponding to the esti-
cess simulation to extract these process and economic parameters for mated annual landfill amount of waste HDPE in New York State.35,36
major technology alternatives in the superstructure of waste HDPE The major novelties of this work are summarized below:
recycling process systems.28 The third challenge is to tackle the non-
convexity involved in multiple objective functions when solving the • The most comprehensive superstructure of waste HDPE recycling
corresponding superstructure optimization problem.29 process systems that involves 1,728 processing routes and nine
In this work, we develop a comprehensive superstructure of sections;
waste HDPE recycling process systems that integrates multiple tech- • A tailored global optimization algorithm that integrates the inexact
nologies for HDPE pyrolysis, as well as technologies for separating parametric algorithm with the branch-and-refine algorithm is
and processing downstream products. The superstructure is further applied to efficiently solve the superstructure optimization prob-
improved by incorporating multiple technology alternatives for CO2 lem for addressing the sustainable design and synthesis of waste
separation to reduce direct emissions. The proposed superstructure HDPE recycling process systems;
consists of 1,728 processing routes and nine sections, namely HDPE • A sustainable process design of waste HDPE recycling process sys-
preprocessing, HDPE pyrolysis, light component separation, heavy tems that has a unit NPV of $142.9/ton, and a reduction by more
chemical component separation, aromatic extraction, hydrogen pro- than 90% in unit life cycle GHG emissions compared to the landfill
duction, hydrotreating, inner heat and electricity generation, and CO2 process.
separation. We leverage experimental data from various literature to
determine the reaction conditions for HDPE pyrolysis.30-32 In addi- The rest of the article is organized as follows. We describe each
tion, heat integration is implemented into each technology alternative section of the superstructure of waste HDPE recycling process systems
ZHAO AND YOU 3 of 25
in the next section in detail. Based on this superstructure, necessary separation, absorption and desorption unit processes for the aromatic
information and assumptions, as well as parameters that formulate the extraction section. The technologies and unit processes are
corresponding superstructure optimization problem are given in the corresponding to the technology options shown in Figure 1. Only one
Problem Statement section. The formulation of the resulting MINFP technology and unit process shown in Figure 2 can be selected in each
problem and the introduction of a tailored global optimization algorithm section of the flow sheet for integration into one processing pathway.
are provided in the Model Formulation and Solution Algorithm section. Detailed function and description for each section, and
To address the sustainable design and synthesis of waste HDPE corresponding flow sheets are given in following subsections.
recycling process systems, a case study accounting for different envi-
ronmental criteria is presented before the conclusion.
2.1 | HDPE preprocessing and pyrolysis
2 | PROPOSED SUPERSTRUCTURE OF The proposed superstructure starts with the HDPE preprocessing sec-
WA ST E HD P E R E CY CL I N G tion, where the waste HDPE is ground into small particles to be trans-
ported to the downstream pyrolizer. Four technology options with
In this work, we develop by far the most comprehensive superstructure pressure swing-adsorption (PSA) are considered: “FBR, SiO2, 675 C,
of waste HDPE recycling process systems, as given in Figure 1. The with PSA,” “FBR, SiO2, 650 C, with PSA,” “FBR, HZSM-5, 500 C, with
superstructure includes 1,728 processing pathways and nine sections, PSA,” and “FBR, HZSM-5, 600 C, with PSA.”30-32 The “FBR” denotes
namely HDPE preprocessing, HDPE pyrolysis, light component separa- the fluidized bed reactor. The sand, which is approximated as SiO2, is
tion, heavy chemical component separation, aromatic extraction, hydro- circulated to maintain high-temperature in the first two technology
gen production, heavy component hydrotreating, inner heat and options, while the HZSM-5 catalyst used in the last two technology
electricity generation, and CO2 separation. The corresponding flow options improves the yield of monomeric molecules.37 The
sheet is given in Figure 2, which consists of various technologies and unconverted HDPE is mixed with the catalyst in the last two technol-
unit processes for waste HDPE pyrolysis, as well as downstream prod- ogy options, and the catalyst is regenerated in a furnace by igniting
uct separation and processing. We consider distillation processes for the unconverted HDPE. All gaseous products are separated, cooled,
sections of light component separation, heavy chemical component and split into two streams with light and heavy chemical components.
F I G U R E 1 Overview of the proposed superstructure of waste high-density polyethylene (HDPE) recycling process systems [Color figure can
be viewed at wileyonlinelibrary.com]
F I G U R E 2 Proposed superstructure of waste high-density polyethylene (HDPE) recycling process systems, which includes HDPE
preprocessing, HDPE pyrolysis, light component separation, heavy chemical component separation, aromatic extraction, hydrogen production,
heavy component hydrotreating, inner heat and electricity generation, and CO2 separation sections [Color figure can be viewed at
wileyonlinelibrary.com]
The light stream is then sent to the PSA unit to separate the inert flu- 2.2 | Light component separation
idized gas. The treated light stream and heavy stream are fed into the
light component separation section and heavy chemical component The main goal of this section is to produce ethylene, propane, and
separation section, respectively. Detailed flow sheets of this propylene from waste HDPE. Similar to the light olefin separation pro-
section are given in Figure 3. cess used in shale gas processing,21,26 the technology options
F I G U R E 3 Process flow sheets of the high-density polyethylene (HDPE) preprocessing and HDPE pyrolysis using SiO2 or HZSM-5 catalyst
[Color figure can be viewed at wileyonlinelibrary.com]
F I G U R E 4 Process flow sheets of the ultimate depropylenizer in the light component separation section to separate products from
pyrolyzing high-density polyethylene (HDPE) (a) using SiO2 (sand), (b) using HZSM-5 catalyst [Color figure can be viewed at
employed in this section, namely ultimate depropylenizer and ultimate separator (around 105 C, 20 bar) as the overhead liquid, while the
heavy component separator, differentiate themselves in terms of the propylene product is split in the C3 separator (around 56 C, 22 bar).
sequence of component separation. Heat integration is applied to Detailed flow sheets for the ultimate depropylenizer and ultimate
save energy for all configurations. The refrigeration cycles are inte- heavy component separator are given in Figures 4 and 5, respectively.
grated to maintain the cryogenic conditions, as well as reduce cooling Detailed operating condition of each equipment unit is given in
utilities usage.38 Tables S3 and S4.
In all process configurations, the light stream from the upstream
HDPE pyrolysis section is fed into the demethanizer column
(−33–6 C, 20 bar) to remove methane. The raffinate is sent to the 2.3 | Heavy chemical component separation
deethylenizer column (27–57 C, 20 bar) to be split into C2 stream
and liquids with heavier components. The C2 stream is further split This section produces n-butane, i-butane, butene mixture, and C5 mix-

into ethane and ethylene stream in the C2 separator (−26 C, 20 bar). ture from waste HDPE. n-butane and i-butane are important ingredi-
In the ultimate depropylenizer configuration, the bottom product from ents in petroleum refinery, butene mixture is used as copolymers,39 and
deethylenizer is sent to the heavy component separator to split heavy C5 mixture is a high performance solvent, and important component in
components (average molar weight > C3) as the bottom product. The naphtha.40 Three types of technology alternatives are included in this

overhead liquid is fed into the C3 separator (67 C, 20 bar) to produce section, namely second step C4 separation, third step C4 separation,
propane and propylene. In the ultimate heavy component separator and C4 and C5 co-separation. The major difference between the
configuration, propane is produced from the heavy component technology alternatives lies in the separation sequence of the products.
F I G U R E 5 Process flow sheets of the ultimate heavy component separator in the light component separation section to separate products
from pyrolyzing high-density polyethylene (HDPE) (a) using SiO2 (sand), (b) using HZSM-5 catalyst [Color figure can be viewed at
In the second step C4 separation configuration, the stream of has a high aromatic loading capacity.42 This solvent also shows a high
heavy components from the light component separation section is solubility in water to reduce its emissions from the overhead gas dur-
split into a C4 stream and stream of heavier components (>C4) in the ing the washing process. In addition, TEG is miscible in water with a

C4 splitter (around 69 C, 8 bar). The C4 stream is used for producing high selectivity to aromatics.43
butene and butane in the second distillation column, namely C4 sepa- In all technology options, the heavy liquid produced from the
rator (around 36 C, 4 bar), while the heavier components is split into heavy chemical component separation section is treated with solvent
C5 mixture and the heavy liquid in the C5 splitter (around 120 C, in the extractive distillation column (around 170 C, 2.53 or 3.53 bar).
8 bar). In the third step C4 separation configuration, the heavy liquid The raffinate from the distillation column is washed by water in the
is produced as a bottom product in the heavy component splitter washing tower (83–101.4 C, 1.01 bar) to remove heavy organic com-
(103 C, 8 bar). The overhead liquid from the heavy component splitter ponent. The stream of heavy organic component, which is the over-
is sent to the C5 splitter (around 67 C, 8 bar) to produce C5 mixture, head liquids of the washing tower, is then sent to the downstream
and the overhead liquid from the C5 splitter is fed into the C4 splitter liquid hydrotreating section. The bottom product from the washing
(around 36 C, 4 bar), which is the third distillation column, to separate tower is split into the aromatic-rich stream and the solvent stream in
butene and butane. In the C4 and C5 co-separation configuration, the the solvent stripper (83–101.4 C, 1.01 bar). The aromatic-rich stream
heavy liquid is split in the heavy component splitter (around 102 C, is split into the aromatic mixture product and water stream in the
8 bar), and the overhead liquid is used for producing butene in the decanter. Ultimately, the solvent and water streams are then mixed
butene splitter (around 37 C, 4 bar). Butane product and C5 mixture with their relevant makeup streams and reused in the next cycle.
product are separated simultaneously in the C5 splitter (around 83 C, Detailed flow sheets for the sulfolane extraction, TEG extraction,
8 bar). The heavy liquid from all configurations is sent to the aromatic and sulfolane-TEG are given in Figures 9–11, respectively. Detailed
extraction section. Detailed flow sheets for the second step C4 sepa- operating condition of each equipment unit is given in Table S7.
ration, third step C4 separation, as well as the C4 and C5 co-
separation are given in Figures 6–8, respectively. Detailed operating
condition of each equipment unit is given in Tables S5 and S6. 2.5 | Hydrogen production
To satisfy the need of hydrogen in the heavy component hydrotreating

2.4 | Aromatic extraction section, three typical technology alternatives are considered for produc-
ing hydrogen, namely electrolysis, photocatalysis, and steam methane
This section aims to produce the aromatic mixture, which is sold as an reforming, as given in Figure 12. In the electrolysis process, water is
important ingredient for benzene and toluene production. Typical aro- sent to the electrolysis cell, and then split into oxygen and hydrogen.44
matic extraction processes are comprised of liquid–liquid extractive In the photocatalysis process, the oxygen and hydrogen are produced
distillation and solvent recycling process.29,30,41 Three technology from pressurized water in a separate bed to avoid direct contact, so as
options are considered in this section, namely sulfolane extraction, tri- to reduce the risk of explosion. For the steam methane reforming, the
ethylene glycol (TEG) extraction, and sulfolane-TEG extraction. Nota- methane flue gas from the demethanizer is mixed with the purchased
bly, sulfolane is a typical solvent for UOP aromatic extraction, and it steam and then heated to trigger steam methane reforming reaction, so
F I G U R E 6 Process flow sheets of the second C4 separation in the heavy chemical component separation to separate products from
F I G U R E 7 Process flow sheets of the third C4 separation in the heavy chemical component separation to separate products from pyrolyzing
high-density polyethylene (HDPE) (a) using SiO2 (sand), (b) using HZSM-5 catalyst [Color figure can be viewed at wileyonlinelibrary.com]
as to produce hydrogen, carbon monoxide, and unconverted methane. hydrotreating section, two typical alternatives for treating upstream
To improve the hydrogen production, a water-gas shifting process is raffinate are considered, namely direct hydrogenation and liquid
performed afterward.45,46 PSA unit is then adopted to separate hydro- hydrocracking, as given in Figure 13. In the direct hydrogenation pro-
gen from the mixture of carbon monoxide, carbon dioxide, and cess, the raffinate is treated by the preheated hydrogen and NiMo
unconverted methane. catalyst in the hydrogenator (150 C, 1.01 bar) to produce gasoline
(assume to be C6–C7 components), diesel (mostly C8–C16 compo-
nents), and wax (>C16 components). Notably, the properties of the
2.6 | Heavy component hydrotreating upstream raffinate resemble to Fischer–Tropsch (F–T) liquids, so it is
reasonable to apply the F–T liquid hydrocracking method in the liquid
The integration of this section into the superstructure aims to produce hydrocracking section.46 Preheated hydrogen stream and precooled
gasoline, diesel, wax to be sold as fuels and chemical products. In this upstream raffinate stream are fed into the hydrocracker (50 C,
F I G U R E 8 Process flow sheets of the C4 and C5 co-separation in the heavy chemical component separation to separate products from
F I G U R E 9 Process flow sheets of the sulfolane extraction in the aromatic extraction section. (a) Sulfolane extraction with post-heater,
(b) Sulfolane extraction with post-cooler [Color figure can be viewed at wileyonlinelibrary.com]
22.89 bar) to get the fuel gas, gasoline, and diesel. The fuel gas is then HDPE recycling process systems and reduce the cost of purchasing elec-
sent to the downstream inner heat and electricity generation tricity. The energy obtained from the ignition of overhead gas streams is
section to generate electricity and heat. converted into heat and electricity in this section,45 as given in Figure 14.
All overhead gas streams from the methane reformer, demethanizer, and
hydrocracker are mixed and combusted in the furnace. The flue gas
2.7 | Inner heat and electricity generation stream outputted from the furnace is then sent to all sections of the
superstructure to exchange heat. The heat in the remaining gas stream is
A steam turbine is implemented in the inner heat and electricity genera- used for evaporating the precooled water, so that the vapor can push the
tion section to provide high-temperature heat for all sections in waste steam turbine to do the work. The work is then converted into electricity
F I G U R E 1 0 Process flow sheets of the triethylene glycol (TEG) extraction in the aromatic extraction section. (a) TEG extraction with
post-heater, (b) TEG extraction with post-cooler [Color figure can be viewed at wileyonlinelibrary.com]
F I G U R E 1 1 Process flow sheets of the sulfolane-triethylene glycol (TEG) extraction in the aromatic extraction section. (a) Sulfolane-TEG
extraction with post-heater, (b) Sulfolane-TEG extraction with post-cooler [Color figure can be viewed at wileyonlinelibrary.com]
in a turbine generator (not given in the flow sheet). The mixture of vapor absorption, membrane capture process, and PSA process, as given in Fig-
and water coming out of the steam turbine is pumped into the condenser ures 15 and 16. MEA shows a high loading capacity to acid gas and high
to be precooled into liquid water again. solubility in water, contributing to a high selectivity in the CO2 capturing
process.47 However, solvent regeneration requires much energy. For the
membrane capture process, the polyvinyl amine (PVAm) membrane is
2.8 | CO2 separation used to selectively sieve out the CO2 from the flue gas.48 In the PSA pro-
cess, the flue gas is treated by the PSA unit to split out the CO2.49 Like
To reduce the life cycle environmental impacts of waste HDPE recycling the MEA absorption, the CO2 compressor chain is used to gather CO2 in
process systems, three technology alternatives are considered for the other two technology alternatives. Ultimately, the cleaned flue gas is
removing CO2 from the flue gas, namely monoethanolamine (MEA) emitted to air, while the gathered CO2 is liquified for storage.
F I G U R E 1 2 Process flow sheets of the photocatalysis, electrolysis, and steam methane reforming in the hydrogen production section [Color
figure can be viewed at wileyonlinelibrary.com]
F I G U R E 1 3 Process flow sheets of the direct hydrogenation, liquid hydrocracking in the heavy component hydrotreating section [Color
figure can be viewed at wileyonlinelibrary.com]
3 | PROBLEM STATEMENT superstructure optimization problem.45,51 In summary, LCA and TEA give
access to evaluate and compare the environmental and economic perfor-
In this section, we formally state the LCA and TEA approaches, rele- mance of each process alternative in the superstructure when executing
vant assumptions, as well as parameters to formulate the superstruc- the corresponding superstructure optimization, which systematically
ture optimization problem of waste HDPE recycling process systems. In addresses the sustainable design and synthesis of waste HDPE recycling
this work, a process-based LCA is adopted to quantitatively evaluate the process systems.
environmental footprint of each activity or process within waste HDPE
recycling process systems, and provide environmental parameters to the
corresponding superstructure optimization problem.29 Typically, a 3.1 | Goal and scope definition
process-based LCA contains four phases, namely goal and scope defini-
tion, inventory analysis, impact assessment, and interpretation.50 In addi- In the phase of goal and scope definition, we define main features of
tion, we apply TEA to evaluate and determine the economic parameters, waste HDPE recycling process systems through setting the LCA goal,
such as the capital and operating costs, for the corresponding system boundary, the functional unit, and key assumptions.
F I G U R E 1 4 Process flow sheet of

the inner heat and electricity generation
[Color figure can be viewed at
F I G U R E 1 5 Process flow sheet of the monoethanolamine (MEA) absorption in the CO2 separation section [Color figure can be viewed at
The goal of this work is to investigate the process design of collection, separation, and transportation, natural gas production and
HDPE recycling process systems with optimal economic and environ- transportation, waste HDPE processing, external utilities and electric-
mental performance. Due to the absence of end-of-life phases for ity generation, and wastewater treatment, as given in Figure 17.
final products (e.g., butene) manufactured from HDPE recycling pro- A functional unit is then defined as the unit amount of waste
cess systems, this study considers a cradle-to-gate system bound- HDPE treated, or unit amount of final products manufactured from
ary.52 The system boundary consists of five life cycle stages to the HDPE recycling process systems.53,54 Notably, the amount of the
account for the material flows, as well as emissions from the HDPE output product is proportional to the amount of HDPE in a fixed
F I G U R E 1 6 Process flow sheets of the membrane capture and pressure swing-adsorption (PSA) in the CO2 separation section [Color figure
can be viewed at wileyonlinelibrary.com]
F I G U R E 1 7 System boundary of a cradle-to-gate life cycle assessment (LCA) for waste high-density polyethylene (HDPE) recycling process
systems is confined by a dotted red box, including five life cycle stages: waste HDPE processing (denoted as a big grass-green block), HDPE
collection, separation, and transportation (denoted as an orange block), natural gas production and transportation (denoted as a shallow orange
block), external utilities and electricity production (denoted as a shallow yellow block), and wastewater treatment (sour orange block). Direct and
indirect environmental impacts in the producing stage are denoted as orange and dark green tick, respectively. Major processes and material
flows are represented by gray boxes and ticks, respectively [Color figure can be viewed at wileyonlinelibrary.com]
process design. Since the type and composition of the final products 3.2 | Inventory analysis
vary when selecting different processing technology options, it is rea-
sonable to employ a functional unit defined as one ton of waste The second phase is the inventory analysis, where life cycle invento-
24,29
HDPE treated in the recycling process systems. Accordingly, we ries (LCIs) are set up for all activities and processes within the system
evaluate the economic performance and life cycle environmental boundary.55 Specifically, the mass and energy balances for the offsite
impacts based on this functional unit. production of utilities and natural gas, transportation of HDPE and
natural gas, as well as each technology alternative within the waste of the ReCiPe mid-point scores can provide guidance on operating
HDPE recycling process systems are quantified by the predefined the sustainable waste HDPE recycling process systems.
29
functional unit. The data of these LCIs are extracted from Ecoinvent As for both methodologies, we specify the characterization fac-
V3.6 database and process simulations developed in Aspen Plus.56,57 tors that are used for systematically evaluating the life cycle environ-
Harnessing the obtained data, the mass and energy balance relation- mental impacts of each processing route included in waste HDPE
ships can be formulated as mathematical constraints to aggregate the recycling process systems.
relevant LCIs, so that the LCI of various design alternatives in the
superstructure can be automatically evaluated.28
3.4 | Interpretation
3.3 | Impact assessment In this work, the interpretation phase is directly integrated with super-
structure optimization to systematically address the sustainable
In the phase of impact assessment, we apply the GWP indicator and design and synthesis of waste HDPE recycling process systems. The
ReCiPe end-point score indicator to make a functional-unit-based evaluated LCA and TEA data are fed into the corresponding super-
quantification of the life cycle environmental impacts of waste HDPE structure optimization problem so that the economic and environmen-
recycling process systems. The quantification parameters are obtained tal performance for each technology alternative of recycling waste
from the Ecoinvent V3.6 database.56,57 HDPE can be systematically evaluated and automatically optimized. A
We focus on the GWP indicator due to its wide application in set of optimal solutions are obtained and plotted as two Pareto-
calculating GHG emissions when quantitively assessing the environ- optimal curves to reveal the trade-offs between the economic perfor-
mental impacts of plastic processing systems.58,59 Moreover, GHG mance and the life cycle environmental impacts (in terms of unit life
emissions are always addressed in regulations and legislation to cycle GHG emissions or unit life cycle ReCiPe end-point score).
restrict the flue gas emissions of a chemical process. We use the The interpretation of economic performance follows the typical
GWP in a time horizon of 100-year to represent the relative green- methodology of TEA,25 which accounts for the capital expenditure
60
house impact compared to CO2. For example, emitting 1 kg of (CAPEX) and operating expenditure (OPEX) of waste HDPE recycling
methane is equivalent to emitting 28 kg of CO2 over the course of process systems. CAPEX consists of the direct and working capital of all
100 years, so the GWP of methane is 28. Hence, the GWP indicator equipment units. OPEX includes the operation cost of HDPE collection,
can be used for quantifying greenhouse impacts of GHG emissions. separation, and transportation, feedstock cost, utility cost, as well as the
The GHG emissions are divided into two categories within the sys- cost for operation and maintenance (O&M), property tax and insurance,
tem boundary, namely direct and indirect emissions, as given in sale expense, and income tax. We use the linear depreciation method
Figure 17. when performing the NPV calculation. The NPV is directly calculated by
Since the GWP indicator only accounts for one life cycle envi- the summation of discounted annualized cash flow subtracting CAPEX in
ronmental impact category, it is necessary to investigate the envi- each year.
ronmental sustainability of a process design of waste HDPE Based on the proposed superstructure and the LCA and TEA
recycling process systems from a wide spectrum of impact catego- approaches, necessary assumptions used for formulating the super-
ries. ReCiPe hierarchical end-point score provides access to system- structure optimization problem of waste HDPE recycling process sys-
atically evaluate the life cycle environmental impacts of waste tems are provided below:
HDPE recycling process systems in terms of 18 mid-point impact
categories, namely climate change, ozone depletion, terrestrial acidi- • Linear relationships between mass flow rates and energy consump-
fication, freshwater eutrophication, marine eutrophication, human tions, mass flow rates and environmental impacts are postulated.63
toxicity, photochemical oxidant formation, particulate matter forma- • All reactants in reactor, pyrolizer, and combustor, unless specifi-
tion, terrestrial ecotoxicity, freshwater ecotoxicity, marine cally noted, are fully converted.
ecotoxicity, ionizing radiation, agricultural land occupation, urban • The equipment unit procurement cost of each technology alterna-
land occupation, natural land transformation, water depletion, min- tive is scaled with capacity through power function.34
eral resource depletion, and fossil fuel depletion. 61
These mid-point • The fixed capital investment, working capital investment and main-
indicators are finally aggregated and classified into three ReCiPe tenance cost are estimated by a percentage of total capital cost.55
end-point impact categories, which account for the damage to
human health, the damage to ecosystem, and the damage to Based on these assumptions, the relevant parameters used in the
resource availability.61 We focus on the same system boundary and corresponding superstructure optimization problem are provided
life cycle stages as those used to assess GHG emissions to evaluate below:
the ReCiPe end-point score. ReCiPe mid-point scores are then cal-
culated to demonstrate the degree of impacts in 18 different impact • Process parameters for each technology alternative, including the
categories for each life cycle stage within the system boundary of chemical compositions of feedstocks and products, temperatures
HDPE recycling systems.62 A systematic analysis of the distribution for inlet and outlet streams of each equipment unit.
• Parameters related to mass balance relationships in waste HDPE • Techno-economic analysis variables that consider the NPV,
recycling process systems, including the upper and lower bounds CAPEX, and OPEX to operate the waste HDPE recycling process,
for processing capacity, and split fraction of each equipment unit. as well as the breakdowns of CAPEX and OPEX, and the break-
• Parameters related to energy balance relationships in waste HDPE downs of the annualized cost.
recycling process systems, including the unit flow rate-based usage • Environmental impacts assessment variables that consider the
rate of cooling, heating utilities, and electricity of each equipment GHG emissions, ReCiPe end-point score, and ReCiPe mid-point
unit, and unit flow rate-based heat generation rate in the furnace. score of the waste HDPE recycling process, as well as their break-
• The annual operating hours and lifespan of the whole plant. downs for the direct emissions, HDPE collection, separation, and
• A set of environmental impact characterization factors, including transportation, natural gas production and transportation, waste
those of HDPE transportation, components in off-gas emissions, HDPE processing, external electricity generation, and cooling
feedstocks, productions and transportations of offsite electricity, water production.
heating, and cooling utilities, using the GWP and ReCiPe end-point
score indicators.
• Economic parameters, including direct capital costs, coefficients 3.5 | Model formulations for the superstructure
for calculating O&M, market prices for feedstocks and products, optimization problem
utility costs, and discount rate for evaluating annualized
cash flow. The superstructure optimization problem is formulated as a multi-
objective model. The general formulation of this superstructure
The unit NPV is calculated through dividing the NPV by the optimization problem is provided below. The functional-unit-based
mass of waste HDPE treated within the project lifespan. The unit economic objective function, and environmental objective functions
life cycle environmental impacts are calculated via dividing the life are named as OBJeconomic, GWP, RciPe, respectively. For compactness,
cycle environmental impacts by the mass of waste HDPE treated the detailed model formulation is provided in Section S4.
annually. Hence, all objective functions are in fractional forms. If
optimizing the linear objective function, the treatment amount of NPV
max OBJeconomic = in Eq: ðS56Þ
HDPEyr
waste HDPE can be directly pushed to its maximum or minimum
value that might not be reasonable.29 Comparatively, the optimiza- 8
> ðEECST + EEH + EEE + EEW + EED + EENÞGWP
tion of these fractional objectives can lead to a balanced optimal >
<GWP = in Eq: ðS36Þ
HDPE
HDPE treatment amount, so that each technology alternative can min
>
:RciPe = ðEECST + EEH + EEE + EEW + EED + EENÞRECIPE in Eq: ðS37Þ
>
be identified and compared. Thus, the optimal design of waste HDPE
HDPE recycling process systems are systematically addressed. The
decision variables are superstructure configuration variables and s.t. Mass balance constraints (S1)–(S24)
mass balance variables, which are used for directly determining the Energy balance constraints (S25)–(S29)
configuration and mass flow rates of the waste HDPE recycling Environmental impacts assessment constraints (S30)–(S35)
process systems. Detailed descriptions are provided below: Techno-economic evaluation constraints (S38)–(S55)
Superstructure network configuration constraints (S57)–(S67)
• Superstructure network configuration variables that consider the
selection of technology alternatives and network topology in the As stated in the Problem Statement section, the unit NPV is cal-
waste HDPE recycling process. culated through dividing the NPV by the mass of waste HDPE treated
• Mass balance variables that consider the capacity of each equip- within the project lifespan. The unit life cycle environmental impacts
ment unit, the capacity for power generation in the waste HDPE are calculated via dividing the life cycle environmental impacts by the
recycling process, and flow rates of streams for each technology mass of waste HDPE treated annually. The life cycle environmental
alternative. impacts are the summation of those of HDPE collection, separation,
and transportation (EECST), external heat utilities production (EEH),
Other variables, such as the energy balance variables, as well as external electricity production (EEE), cooling water production (EEW),
variables corresponding to the techno-economic analysis and environ- direct emission (EED), as well as the natural gas production and trans-
mental impacts assessment, are dependent on the values of the afore- portation (EEN). The unit NPV, unit life cycle GHG emissions, and unit
mentioned decision variables. Detailed description of these variables life cycle ReCiPe end-point score are denoted by OBJecomomic, GWP,
are listed below: and RciPe, respectively. The economic and environmental objective
functions are optimized subject to the constraints corresponding to
• Energy balance variables that consider the amount of heating the mass balance, energy balance, environmental impacts assessment,
energy and electricity generated by the steam turbine in the waste techno-economic evaluation, and superstructure network configura-
HDPE recycling process. tion. We formulate economic and environmental objective functions
as fractional terms to address the functional-unit-based life cycle per- considered to be around three times of Ashley process's capacity
formances.33 Notably, nonlinear terms are introduced in capital cost for meeting New York State's supply, although the optimal capac-
calculation in the form of separable concave terms and in the frac- ity determined by optimization could be much smaller than the
tional objective functions, while all other constraints have linear rela- upper bound. Notably, the Foreland Refinery Corporation in the
tionships with continuous and binary variables. Therefore, the U.S. can accept crude oil, gasoline, and diesel from multiple wells
superstructure optimization problem can be formulated as a noncon- and is still exploring other opportunities to maximize potential of
vex MINFP problem. production.68,69 On this regard, we propose to establish the waste
HDPE recycling system near the Foreland Refinery Corporation, so
the gasoline and diesel products can be sent to this refinery to
3.6 | Solution algorithm improve its production potential. We perform a case study to
address the optimal HDPE recycling process design with maximum
The superstructure optimization problem of waste HDPE recycling unit NPV and minimum unit life cycle GHG emissions or unit ReC-
process systems is reformulated as a nonconvex MINFP problem. iPe end-point score. The interest rate is 10%, and the project
Due to the combinatorial nature and pseudo-convexity of frac- lifespan is 20 years when calculating the NPV.25
tional objectives, as well as separable concave terms for capital All computational experiments are performed on a DELL
costs estimation, the solution of this MINFP problem is computa- OPTIPLEX 7040 desktop with Intel(R) Core (TM) i7-6700 CPU @
tionally challenging for general-purpose global optimizers.34 Thus, 3.40 GHz and 32 GB RAM. The product distributions for HDPE
we apply a tailored global optimization algorithm that integrates pyrolysis are leveraged from relevant literature.30-32 The process
64
inexact parametric algorithm with the branch-and-refine algo- simulation models for each technology alternative in HDPE pyroly-
rithm65 to tackle this computational challenge.34 A pseudo-code of sis, light component separation, heavy chemical component sepa-
this algorithm is given in Figure 18. ration sections are developed in Aspen Plus V10.0.70 The process
simulation models for the aromatic extraction section are devel-
oped in Aspen HYSYS V10.0 to use the glycol property package.56
4 | RESULTS AND DISCUSSIONS The capital cost for each equipment unit is calculated using Aspen
Capital Cost Estimator V10.071 and references to data from other
The proposed framework is applied to an HDPE recycling process literature.44,47-49,72,73 All relevant data to solve the corresponding
design with the upper bound of treatment capacity of around superstructure optimization problem are given in Tables S8–S19.
295,000 ton HDPE/year, corresponding to the estimated landfill The mathematical formulation of the corresponding superstructure
amount of waste HDPE in New York State.35,36 We note that there optimization problem and related solution process are coded in
is a real-world plastic pyrolysis plant located in Indiana, which is a GAMS 24.8.3,74 using CPLEX 12.7 as the MILP optimizer, and
part of a project called Ashley, with a treatment capacity of BARON 17.1,75 and SCIP 3.2 as MINLP global optimizers.76 The
67
100,000 ton/year of waste plastic pyrolysis. Therefore, the tolerances of relative optimality for the inexact parametric and the
upper bound of HDPE treatment capacity in this case study is branch-and-refine algorithm are set as 10−6.77
F I G U R E 1 8 The pseudo-
code of the tailored global
optimization algorithm that
combines the inexact parametric
algorithm and the branch-and-
refine algorithm
4.1 | Case study: Sustainable design of HDPE determined by the New York State's annual landfill amount of waste
recycling process systems accounting for optimizing HDPE, the mass flow rates of gasoline and diesel from the superstruc-
unit NPV and unit life cycle GHG emissions or ReCiPe ture can reach to their maximum values, which are 432.67 and
end-point score 448.41 kg/hr, respectively. The values are extracted from different
optimal solutions. If we set the upper bounds of the gasoline and die-
4.1.1 | GHG emissions breakdowns and the sel production as their corresponding maximum production amounts
optimal HDPE recycling process systems and solve the superstructure optimization problem again, the Pareto-
optimal curve and the good-choice solution B remain unchanged.
In the first subsection, we consider the superstructure of waste HDPE The maximum unit NPV given on the curve is $151.38/ton
recycling process systems given in Figure 2 and solve the superstructure (corresponding NPV: $3.4 billion) at point C, which is higher than that
optimization problem to simultaneously maximize the unit NPV and mini- given in the related literature by about 25%.78 This improvement is
mize the unit life cycle environmental impacts. To reveal the benefit of mainly contributed by the multiple products manufactured from the
using fractional functions when solving the superstructure optimization ultimate heavy component separator, and C4 and C5 co-separation.
problem, we also optimize two linear objective functions, namely total Those multiple products not only include the ethylene and propylene
GHG emissions and NPV. The optimal results can be plotted on a Pareto- given in the literature related to design an HDPE refinery,22 but also
optimal curve to reveal the trade-off between economic and environmen- contain butene, butane, and the C5 mixture. Besides, the optimal
tal performance. To compare with existing LCA results in other literature, operating condition of ultimate heavy component separation, C4 and
we focus on minimizing the unit life cycle GHG emissions. C5 co-separation, and sulfolane extraction greatly improves the purity
Each point on the Pareto-optimal curve corresponding to an opti- of hydrocarbon products and their manufacturing amount. Such
mal solution of the superstructure optimization problem, as shown in improvements lead to a more economically competitive HDPE
Figure 19. Notably, the operating parameters input to the superstruc- recycling process for the optimal solution C. Moreover, the improve-
ture optimization problem are extracted from the high-fidelity process ment in economic performance can be partially attributed to the heat
simulation and remain constant. Moreover, the operating conditions integration and the on-site production of high-temperature energy
of the optimal design are validated and are close to those from the and electricity, which helps to reduce the cost of electricity and natu-
process simulation for each technology option. On these regards, the ral gas procurement.
process topology and design implication corresponding to an optimal However, compared to the HDPE recycling process for the opti-
solution is invariant. mal solution A (the most environmentally friendly solution), the draw-
In the superstructure, the mass flow rate of heavier components back of the HDPE recycling process for solution C in terms of GHG
is determined by the nonzero split fraction extracted from Aspen-Plus emissions is also evident. The unit life cycle GHG emissions of solu-
process simulation and product distribution of the upstream pyrolysis tion A is 398.51 kg CO2-eq/ton, which is reduced by 43% in compari-
process. Thus, the heavier components, of which the mass flow rate son with the result at point C (700 kg CO2-eq/ton). Nevertheless, the
could not be reduced to zero, are sent to the direct hydrogenation unit NPV at point A (corresponding NPV: $2.4 billion) is 70% of that
and liquid hydrocracking sections to produce gasoline, diesel, or wax at point C, and values of unit NPVs are $107.15/ton, and $151.39/
to avoid their direct emissions. On this regard, it is impossible to avoid ton, respectively. This result effectively illustrates the trade-off
the production of gasoline or diesel. between economic and environmental performance in the sustainable
Notably, if the treatment capacity is fixed to the upper bound of design and synthesis of HDPE recycling process systems with maxi-
the HDPE treatment, which is 33.67 ton HDPE per hour and mum unit NPV and minimum unit GHG emissions.
F I G U R E 1 9 Pareto-optimal
curve reveals the trade-offs
between the unit NPV and unit
life cycle GHG emissions. The
figure presents the selected
technologies for high-density
polyethylene (HDPE) recycling
process for the good-choice
solution, and the emission
breakdowns of the HDPE
recycling processes of good-
choice, the most environmentally
friendly, and the most
economically competitive
solutions [Color figure can be
viewed at wileyonlinelibrary.com]
T A B L E 1 Calculated product distribution for three optimal solutions A, B, and C are given in pie charts near the Pareto-optimal
solutions A, B, and C curve. In addition, the integration of technology selection for point B
Product components Solution A Solution B Solution C is given above the Pareto-optimal curve. For the emission break-
Ethylene (kg/hr) 512.33 512.33 512.33 downs, the direct emissions include the environmental impacts of the
off-gas, while the indirect emissions are of HDPE collection, separa-
Propylene (kg/hr) 3,336.50 3,336.50 3,336.50
tion, and transportation, natural gas production and transportation,
Propane (kg/hr) 990.99 990.99 990.99
external electricity production, heat utilities production, and cooling
1-Butene (kg/hr) 3,474.53 3,474.53 3,474.53
water production. The optimal HDPE recycling process for point B
I-Butene (kg/hr) 316.01 316.01 316.01
selects MEA absorption as the CO2 separation process. In this amine-
Butane mixture (kg/hr) 743.04 743.04 743.04
based process, the flue gas from the combustor (not specifically pres-
C5 mixture (kg/hr) 2,456.64 2,456.64 2,456.64
ented), which connects with the steam turbine, is directly treated by
Aromatic mixture (kg/hr) 91.41 91.41 91.41 amine to remove CO2 effectively in the absorber. The absorbed gas is
Gasoline (kg/hr) 76.66 76.66 174.78 then desorbed from the rich solution under high-temperature and
Diesel (kg/hr) 181.14 181.14 91.76 CO2 is liquefied and stored after being cooled and pressurized. More
Paraffin (kg/hr) 24.46 than 85% of GHGs are removed from the flue gas in this process,
which greatly decreases the direct emissions compared to the most
economically competitive design of solution C.
The HDPE recycling process for point B (the good-choice solu- To decipher the advantage of optimizing fractional functions
tion) shows a pronounced economic performance and minimum unit when solving the superstructure optimization problem, we need to
life cycle GHG emissions. We use the elbow method to select the get the optimal solutions by optimizing two linear objective functions,
good-choice solution B on the Pareto-optimal curve to reveal the namely total GHG emissions and NPV, as shown in Figure 20. The unit
HDPE recycling process that has a pronounced economic perfor- NPVs of the good-choice optimal solution B and the most economi-
mance and minimum unit life cycle GHG emissions. Notably, the cally competitive optimal solution C are $142.9/ton and $151.39/ton,
good-choice solution B locates on the elbow of the curve. If we respectively, which are identical to results from solving fractional
choose the solution that has more unit NPV than solution B, such as objective functions. However, the unit GHG emissions of solution D
solution C, the unit life cycle GHG emissions would enhance by more are 410 kg CO2-eq/ton with a treatment capacity of 13.6 ton/hr
than 50%, so that the economic benefit is less worth than additional (maximum capacity), which is less environmentally friendly than solu-
life cycle environmental impacts. If we choose the solution that has tion C (398.51 kg CO2-eq/ton). The technology options of CO2 sepa-
less life cycle environmental impacts, such as solution A, the unit NPV ration are different from optimal solutions C and D. The optimal
would be reduced by more than 30%. Thus, the benefit of reducing solution C selects the PSA unit, while the solution D chooses the
life cycle environmental impacts is less worth than the NPV loss. membrane unit. Notably, the capital cost of the PSA unit ($4.8 million)
Moreover, as shown in Figure 19, for the point adjacent to point B is much higher than the membrane unit ($1.3 million) with the same
with lower unit NPV, the corresponding process configuration inte- CO2 separation capacity (such as 4.92 ton CO2/hr). Hence, when opti-
grates the ultimate depropylenizer (the process configuration is not mizing the linear objective functions, the process design of optimal
shown on the graph; the good-choice solution B incorporates the ulti- solution D is inclined to adopt the technology option with lower capi-
mate heavy component separator in the superstructure). For the point tal cost to improve the total NPV. Moreover, the total NPV would be
adjacent to point B with higher unit NPV, the superstructure configu- enhanced if the treatment capacity increases. Therefore, the optimal
ration integrates the third step C4 separation, while the good-choice solution would tend to select the most economical option if optimize
solution B incorporates the C4, C5 mixed separation. Notably, the the total NPV and total GHG emissions simultaneously, as illustrated
value of unit GHG emissions is in one-to-one correspondence to the in the literature.79 On this regard, when optimizing linear objective
process configuration, or an optimal solution. Hence, for the area that functions, the most environmentally friendly solution A, which has a
is confined by these three points on the Pareto-optimal curve shown low NPV, is substituted by optimal solution D.
in Figure 19, there is no more optimal solution within this area. On Notably, the treatment capacity of the optimal solution E is 12.8
these regards, we choose solution B as the good-choice solution to ton/hr, which is lower than the maximum capacity. If the treatment
reflect the trade-off between the unit NPV and unit life cycle GHG capacity of the corresponding processing pathway reaches the maxi-
emissions. The unit NPV at point B ($142.9/ton, corresponding NPV: mum treatment capacity, the total GHG emissions would be higher
$3.2 billion) is 94.4% of that at point C. The optimality is evidently than that of optimal solution B, so the solution E can be replaced by
revealed when the unit life cycle GHG emissions at point B is just the optimal solution B. Hence, the trade-off between the total NPV
10% higher than that at point A, while the unit NPV at point B is and total GHG emissions prevents the capacity from reaching its max-
improved for more than 30%. imum amount. Thus, the benefit of optimizing the fractional objective
To find out the factors for maintaining the minimum unit life cycle function is from a more pronounced trade-off between NPV and
greenhouse emissions in this sustainable design of waste HDPE GHG emissions, so the sustainable design and synthesis of waste
recycling process systems, the emissions breakdowns for optimal HDPE recycling processing systems can be addressed.
Selection of life cycle inventory models and corresponding data- The improvement of environmental sustainability is evident when
bases, as well as the source of economic parameter in the literature comparing the unit NPV and unit GHG emissions at point B (the
can influence the results of techno-economic analysis and life cycle good-choice solution) with existing results in relevant literature. Nota-
assessment. Economic parameters, such as the CAPEX for equipment bly, the GHG emissions extracted from Gracida-Alvarez et al22 are cal-
22
units analyzed in Gracida-Alvarez et al, are extracted from govern- culated based on IPCC 2013, which is the same database as we use in
ment reports and relevant books instead of the Aspen Plus Capital this work. Compared with the GHG emissions from the pyrolysis pro-
Cost Estimator used in this study, are extracted from relevant litera- cess in Gracida-Alvarez et al22 (0.99 ton CO2-eq/ton), the optimal
80,81
ture and relevant books. However, we extract these parameters HDPE recycling process with MEA absorption for point B can reduce
from the Aspen Plus Capital Cost Estimator and government reports. the unit life cycle GHG emissions by 55.6% (0.44 ton CO2-eq/ton).
Moreover, the modified accelerated cost recovery system is used for Thus, the combination of the open-loop recycling and the CO2 separa-
22
calculating NPV in Gracida-Alvarez et al, while we use linear depre- tion processes in the sustainable design of waste HDPE recycling
ciation method to calculate depreciation cost and thus the NPV.46 process systems has a far-reaching environmental significance.
Notably, various relevant literature prefers to extract the economic
parameters directly from Aspen Plus due to the high accuracy in cal-
culating CAPEX for equipment units, such as heat exchangers and 4.1.2 | Cost and ReCiPe total score breakdowns
compressors.46 On this regard, the economic performance calculated
in this work is more accurate. It is worth noting that various literature To provide a detailed analysis and a pervasive standard to investigate
corresponding to waste plastic recycling applied IPCC 2013 as the the sustainable design of HDPE recycling process systems, we focus
59
database of life cycle inventories. Hence, the results of LCA for this on ReCiPe end-point score in the second subsection to directly inter-
work is more typical. pret the impacts of life cycle inventories for each life cycle stage. The
F I G U R E 2 0 Pareto-optimal curve
reveals the trade-offs between the total
NPV and total life cycle GHG emissions.
The figure presents the selected
technologies for high-density
polyethylene (HDPE) recycling process for
the most environmentally friendly and
good-choice optimal solutions [Color
figure can be viewed at
F I G U R E 2 1 Pareto-optimal
curve shows the trade-offs
between the unit NPV and the
unit life cycle ReCiPe end-point
score of the high-density
polyethylene (HDPE) recycling
processes of the good-choice, the
most environmentally friendly,
and the most economically
competitive solutions. The figure
presents relevant CAPEX and
OPEX breakdowns, and ReCiPe
end-point score breakdowns
environmental objective is formulated as minimizing the unit life cycle The unit NPV and unit life cycle ReCiPe end-point score of each opti-
ReCiPe end-point score per ton of HDPE treated, while the economic mal solution is given on the resulting Pareto-optimal curve in Figure 21.
objective is maximizing NPV per ton of HDPE treated within the pro- The optimal process flow sheets of the most environmentally friendly,
ject lifespan. The optimal results can be plotted in a Pareto-optimal good-choice, and the most economically competitive solutions are dis-
curve to reveal the trade-off between economic and environmental played in Figure 22, and they are related to points A, B, and C, respec-
performance. The good-choice solution from the resulting Pareto- tively. The HDPE recycling process for point B (the good-choice solution)
optimal curve systematically addresses a sustainable HDPE recycling integrates ultimate heavy component separation, C4 and C5 co-separa-
process, which weighs minimum life cycle environmental impacts in tion, sulfolane extraction with post-heater, methane steam reforming,
terms of human health, ecosystem quality, and resource scarcity.61 direct hydrogenation, steam turbine, and MEA absorption.
F I G U R E 2 2 Process flow sheets of the most environmentally friendly solution A, good-choice solution B, and the most economically
competitive solution C, respectively [Color figure can be viewed at wileyonlinelibrary.com]
To better illustrate the economic competitiveness for each opti- and transportation, external electricity production, heat utilities
mal HDPE recycling process, the breakdowns of CAPEX and OPEX for production, and cooling water production. For optimal HDPE
each optimal solution are given as a series of pie charts below the recycling processes for solutions A, B, and C, the heat utilities pro-
Pareto-optimal curve in Figure 21. Due to the high cost for procure- duction shares more than 60% of the total ReCiPe score. The life
ment and installation for various separation equipment units and reac- cycle stages of the production and transportation of natural gas,
tors for downstream processing, the annualized capital cost shares and the HDPE collection, separation, and transportation also pose
around 25% of the total expense for optimal HDPE recycling pro- impacts on the environment, which share 6.8 and 3.2% of the total
cesses of the three optimal solutions. The O&M, which is used for ReCiPe scores for solution A, 6.6 and 4% for solution B, and 5.4
maintaining the normal plant operations, also shares a large portion and 3.2% for solution C. The ratios of ReCiPe end-point score of
(25–27%) in the total expense. Values of the O&M, annualized total direct emissions to the total score are 0.0482, 0.0495, and 0.2542
capital investment, utility cost, and feedstock cost for each waste for solution A–C, respectively. Thus, the result of ReCiPe total
HDPE recycling process for solutions A, B, and C are given in score breakdowns effectively demonstrate the environmental sus-
Figure 23. The optimal HDPE recycling process for solution A (the tainability for solution B.
most environmentally friendly solution) reduces the indirect emissions
by avoiding using heat utilities in the PSA unit in the CO2 separation
section. However, it has the highest annualized capital investment 4.1.3 | ReCiPe mid-point and end-point score
($26.65 million/year) and O&M ($27.07 million/year) due to the high breakdowns
costs of purchasing and installing this unit. The HDPE recycling pro-
cess for point C (the most economically competitive solution) with the In this section, we present and analyze the comprehensive environ-
minimum annualized capital investment ($16.88 million/year) and mental impact breakdowns of different Pareto-optimal solutions. As
O&M ($18.09 million/year) does not include the CO2 capture process, mentioned in the problem statement section, the ReCiPe approach for
in order to further reduce the capital cost. LCIA includes 18 mid-point impact categories and three end-point
Notably, the income tax is evaluated as a percentage of the income. impact categories.61 As given in Figures 24, S1, and S2, the impact
If a fixed amount of waste HDPE, for example 1 ton, is fed to the waste components for optimal HDPE recycling processes for solutions A, B,
HDPE recycling process systems, the feedstock cost is the same for all and C include external heat utilities production, external electricity
optimal HDPE recycling processes. Therefore, a higher ratio of the income production, cooling water production, HDPE collection, separation,
tax to the feedstock HDPE cost indicates a higher profit per year and and transportation, direct emissions, and natural gas production and
finally a higher unit NPV. The ratio of the income tax to the feedstock transportation. The dominant impact component of the good-choice
cost for point B is 7.84, while the ratio for point C is 8.21. Therefore, the solution B is the external heat utilities production, which contributes
optimal HDPE recycling process for solution B (the good-choice solution) more than 60% of the mid-point score in each mid-point impact cate-
has a comparably economic competitiveness to that of solution C. gory. For solutions A and C, the life cycle environmental impacts of
ReCiPe total score breakdowns given in Figure 21 denote the producing external heat utilities are also pronounced. The massive
share of environmental impacts for each life cycle stage. The direct usage of heat utilities, especially low-pressure steam (LP), is indispens-
emissions include the environmental impacts of the off-gas, while the able to maintain the temperature for operating distillation columns in
end-point scores related to indirect emissions account for HDPE the light component separation, heavy chemical component separa-
collection, separation, and transportation, natural gas production tion, and aromatic extraction sections.
F I G U R E 2 3 Values of the
O&M, annualized total capital
investment, utility cost, and
feedstock cost for each optimal
high-density polyethylene (HDPE)
recycling process for optimal
solutions A, B, and C in Figure 21
that show the economic
competitiveness [Color figure can
be viewed at
F I G U R E 2 4 Breakdowns of the ReCiPe mid-point scores in percentage that show detailed environmental impacts of high-density
polyethylene (HDPE) recycling process for point B (the good-choice solution) in Figure 21 [Color figure can be viewed at wileyonlinelibrary.com]
For solutions A and B, the impacts of the external electricity pro- Therefore, we perform a sensitivity analysis on the market price and
duction are also pronounced in 10 out of the 18 mid-point impact cat- NPV to show how economic performance changes with price vari-
egories, namely agriculture land occupation, freshwater ecotoxicity, ances. We obtained the sensitivity analysis results via using the prices
freshwater eutrophication, human toxicity, ionizing radiation, marine of products or utilities prices to resolve the superstructure optimiza-
eutrophication, metal depletion, natural land transformation, urban tion problem. The result is given in Section S3. The most influential
land occupation, and water depletion. The HDPE collection, separa- economic parameter is the price of butene mixture products.
tion, and transportation has visible environmental impacts posed on
fossil depletion, human toxicity, metal depletion, natural land transfor-
mation, ozone depletion, particulate matter formation, photochemical 4.3 | Computational performance
oxidant formation, terrestrial acidification, and terrestrial ecotoxicity.
For solution C, the direct emissions from flue gas dominate the envi- The superstructure optimization problem of waste HDPE recycling
ronmental impacts on climate change (Figure 24). process systems is reformulated as a multi-objective MINFP prob-
ReCiPe end-point score breakdowns account for three end-point lem, which consists of 104 integer variables, 63,906 continuous
impact categories are provided in Figures 25, S3, and S4 which are used variables, and 67,743 constraints. To effectively solve this noncon-
to clearly show the dominant impact category for each life cycle stage. vex problem, a tailored global optimization algorithm is applied.
For solutions A, B, and C, the external heat utilities production, which has The advantages are evident when comparing the computational
substantial environmental impacts, poses the life cycle environmental performance of the tailored global optimization algorithm with
impacts mainly on ecosystem quality and resource availability. All other general-purpose NLP global optimizers, such as BARON 17.1 and
life cycle stages except direct emissions have same dominant impact cate- SCIP 3.2. In Table 2, the computational performances of solving
gories. For direct emissions, they can affect human health and ecosystem three MINLP problems corresponding to solutions A, B, and C are
quality, and their life cycle environmental impacts on human health are summarized. The economic performance of Instance 1 to Instance
more significant, especially for the HDPE recycling process for solution 3 is the value of unit NPV at points A, B, and C. For all cases,
C. Thus, it is reasonable to choose an HDPE recycling process for the BARON 17.1 starts to find solutions in the local search phase in a
good-choice solution B that uses MEA absorption process to effectively few seconds with no global optimal solution returned in 44,000 s.
reduce the direct emissions. The other global optimization solver, SCIP 3.2, also fails to find a
global optimal solution due to the combinatorial nature and non-
convexity shown in fractional objective functions. The resulting
4.2 | Sensitivity analysis upper and lower bounds of the objective value solved by BARON
17.1 and the objective value solved by SCIP 3.2 are given in
Notably, the choice of economic parameters that are directly related Table 2. By contrast, the tailored global optimization algorithm
to markets is also crucial for technology selection and application. successfully solves all three instances within a few CPU seconds.
F I G U R E 2 5 Breakdowns of the
ReCiPe end-point score that show
detailed environmental impacts of the
high-density polyethylene (HDPE)
recycling process for point B (the good-
choice solution) in Figure 21 [Color figure
can be viewed at wileyonlinelibrary.com]
T A B L E 2 Computational results of solving three instances on the Pareto-optimal curve for HDPE recycling process systems using BARON
17.1, SCIP 3.2, and the tailored global optimization algorithm
BARON 17.1 SCIP 3.2 Global optimization algorithm
No. Unit NPV ($/ton) CPU time (s) Unit NPV ($/ton) CPU time (s) Unit NPV ($/ton) CPU time (s)
a b
1 (−∞,113.3) 44,000 N/A 31,260 107.2 9
2 (100.5,179.2)a 44,000 N/Ab 31,260 142.9 5
3 (−∞,189.5) a
44,000 −160.6 c
168 151.3 4
Abbreviation: HDPE, high-density polyethylene.

a
The optimizer terminates automatically and returns the upper and lower bounds for the objective function.
b
The optimizer does not give primal bounds, but gives dual bound of 0.00346 for Instance 1, and 0.0863 for Instance 2.
c
The solution is suboptimal.
F I G U R E 2 6 Performance of solving
Instance 3 when applying the global
optimization algorithm to solve the
superstructure optimization problem
The detailed computational performance of solving Instance 3 is 5 | CONC LU SION

given in Figure 26. The upper and lower bounds for each inner itera-
tion are displayed simultaneously to show the relative optimality In this work, we addressed the sustainable design and synthesis of
gap. Only two outer iterations with eight total iterations are needed waste HDPE recycling process systems under economic and environ-
to reduce the relative optimality gap to zero. The upper bound is mental criteria. We developed by far the most comprehensive super-
gradually decreasing, while the lower bound slowly increases. More structure to manufacture products from waste HDPE and mitigate
feasible solutions are added as partition points to improve the accu- plastic pollution. The superstructure included 1,728 processing path-
racy of the piecewise approximation successively. Both the upper ways and nine processing sections, namely HDPE preprocessing,
and lower bounds approach zero at the end of the second outer iter- HDPE pyrolysis, light component separation, heavy chemical compo-
ation, representing the convergence of the global optimization nent separation, aromatic extraction, hydrogen production, hydro-
algorithm. treating, inner electricity production, and CO2 separation. Based on
this superstructure, a superstructure optimization problem was formu- RE FE RE NCE S

lated following the life cycle optimization approach to simultaneously 1. Sharuddin SDA, Abnisa F, Daud WMAW, Aroua MK. Energy recovery
maximize the unit NPV and minimize unit life cycle environmental from pyrolysis of plastic waste: study on non-recycled plastics (NRP)
data as the real measure of plastic waste. Energ Conver Manage. 2017;
impacts. The LCA was corresponding to a system boundary from cra-
148:925-934.
dle to gate with a functional unit of one ton of waste HDPE treated to 2. Plastics: Material-Specific Data. https://www.epa.gov/facts-and-figures-
associate the feedstock and products. TEA corresponding to the about-materials-waste-and-recycling/plastics-material-specific-data
superstructure was also performed to deliver economic parameters. A 3. Chae Y, An Y-J. Current research trends on plastic pollution and eco-
logical impacts on the soil ecosystem: a review. Environ Pollut. 2018;
multi-objective MINFP problem was formulated to systematically
240:387-395.
address the sustainable design and synthesis of waste HDPE recycling 4. Galloway TS, Cole M, Lewis C. Interactions of microplastic debris
process systems by maximizing unit NPV and minimizing unit life cycle throughout the marine ecosystem. Nat Ecol Evol. 2017;1(5):116.
environmental impacts. To tackle the computational challenge caused 5. Haward M. Plastic pollution of the world's seas and oceans as a con-
temporary challenge in ocean governance. Nat Commun. 2018;9(1):667.
by the nonconvexities in objective functions, a tailored global optimi-
6. Ivleva NP, Wiesheu AC, Niessner R. Microplastic in aquatic ecosys-
zation algorithm that integrated the inexact parametric algorithm with
tems. Angew Chem Int Ed. 2017;56(7):1720-1739.
the branch-and-refine algorithm based on successive piecewise linear- 7. Joshi C, Browning S, Seay J. Combating plastic waste via trash to tank.
ization was applied to solve this optimization problem. Nat Rev Earth Environ. 2020;1(3):142-142.
This bicriteria MINFP problem was solved to obtain the Pareto- 8. Ragaert K, Delva L, Van Geem K. Mechanical and chemical recycling
of solid plastic waste. Waste Manag. 2017;69:24-58.
optimal curve in consideration of optimizing unit NPV and unit life
9. Anuar Sharuddin SD, Abnisa F, Wan Daud WMA, Aroua MK. A review
cycle environmental impacts. From the perspective of unit life cycle on pyrolysis of plastic wastes. Energ Conver Manage. 2016;115:
GHG emissions, the good-choice solution and the relevant HDPE 308-326.
recycling process were identified from results from the Pareto-optimal 10. DeNeve D, Joshi C, Samdani A, Higgins J, Seay J. Optimization of an
appropriate technology based process for converting waste plastic in to
curve. This design integrated HDPE preprocessing, HDPE pyrolysis
liquid fuel via thermal decomposition. J Sustain Dev. 2017;10(2):116-124.
using FBR and HZSM-5 catalysts in 500 C, ultimate heavy component 11. Joshi CA, Seay JR. Total generation and combustion emissions of
separator, C4 and C5 co-separation, sulfolane extraction with post- plastic derived fuels: a trash to tank approach. Environ Prog Sustain
heater, steam methane reforming, direct hydrogenation, steam Energy. 2019;39(5):1-10.
12. Lopez G, Artetxe M, Amutio M, Alvarez J, Bilbao J, Olazar M. Recent
turbine, and MEA absorption. The unit NPV for this process was
advances in the gasification of waste plastics. A critical overview.
$142.9/ton, which was closed to the value of the most economically Renew Sustain Energy Rev. 2018;82:576-596.
competitive design ($151.3/ton) and 23% higher than that reported in 13. Buekens AG, Huang H. Catalytic plastics cracking for recovery of
the literature.78 Moreover, the unit life cycle GHG emissions of this gasoline-range hydrocarbons from municipal plastic wastes. Resour
Conserv Recycl. 1998;23(3):163-181.
waste HDPE recycling process were reduced to 44.4% of that from
14. Scheirs J, Kaminsky W. Feedstock Recycling and Pyrolysis of Waste
the pyrolysis process in other literature.22 From the perspective of the Plastics. Chichester, England: John Wiley & Sons; 2006.
unit life cycle ReCiPe end-point score, the same good-choice solution 15. Perugini F, Mastellone ML, Arena U. A life cycle assessment of
was obtained from the Pareto-optimal curve, and the corresponding mechanical and feedstock recycling options for management of plas-
tic packaging wastes. Environ Prog. 2005;24(2):137-154.
HDPE recycling process reduced the harmful impact on human health
16. Vermeulen I, Van Caneghem J, Block C, Baeyens J, Vandecasteele C.
by implementing the MEA absorption process. A sensitivity analysis
Automotive shredder residue (ASR): reviewing its production from
was performed to reveal that the most influential economic parame- end-of-life vehicles (ELVs) and its recycling, energy or chemicals'
ters was the price of butene mixture. The computational results of valorisation. J Hazard Mater. 2011;190(1–3):8-27.
solving the MINFP problems showed that the tailored global optimiza- 17. Williams EA, Williams PT. Analysis of products derived from the fast
pyrolysis of plastic waste. J Anal Appl Pyrolysis. 1997;40-41:
tion algorithm greatly outperformed the general-purpose global
347-363.
optimization solvers. 18. Scott D, Czernik S, Piskorz J, Radlein DSA. Fast pyrolysis of plastic
wastes. Energy Fuel. 1990;4(4):407-411.
ACKNOWLEDGMENT 19. Westerhout R, Van Koningsbruggen M, Van Der Ham AG, Kuipers J,
Van Swaaij WPM. Techno-economic evaluation of high temperature
The authors acknowledge the financial support from National Science
pyrolysis processes for mixed plastic waste. Chem Eng Res Des. 1998;
Foundation (NSF) CAREER Award (CBET-1643244). 76(3):427-439.
20. Fivga A, Dimitriou I. Pyrolysis of plastic waste for production of heavy
AUTHOR CONTRIBUTIONS fuel substitute: a techno-economic assessment. Energy. 2018;149:
865-874.
Xiang Zhao: Data curation; methodology; visualization; writing-
21. Alston SM, Arnold JC. Environmental impact of pyrolysis of mixed
original draft; writing-review and editing. Fengqi You: Conceptualiza- WEEE plastics part 2: life cycle assessment. Environ Sci Technol. 2011;
tion; funding acquisition; investigation; project administration; 45(21):9386-9392.
resources; software; supervision; writing-review and editing. 22. Gracida-Alvarez UR, Winjobi O, Sacramento-Rivero JC, Shonnard DR.
System analyses of high-value chemicals and fuels from a waste high-
density polyethylene refinery. Part 2: carbon footprint analysis and
ORCID regional electricity effects. ACS Sustain Chem Eng. 2019;7(22):18267-
Fengqi You https://orcid.org/0000-0001-9609-4299 18278.
23. Gao J, Ning C, You F. Data-driven distributionally robust optimization 44. Saur G. Wind-to-Hydrogen Project: Electrolyzer Capital Cost Study.
of shale gas supply chains under uncertainty. AIChE J. 2019;65(3): Golden, CO: National Renewable Energy Laboratory (NREL);
947-963. 2008.
24. Garcia DJ, You F. Network-based life cycle optimization of the net 45. Wang B, Gebreslassie BH, You F. Sustainable design and synthesis of
atmospheric CO2-eq ratio (NACR) of fuels and chemicals production hydrocarbon biorefinery via gasification pathway: integrated life cycle
from biomass. ACS Sustain Chem Eng. 2015;3(8):1732-1744. assessment and technoeconomic analysis with multiobjective super-
25. Gong J, You F. A new superstructure optimization paradigm for pro- structure optimization. Comput Chem Eng. 2013;52:55-76.
cess synthesis with product distribution optimization: application to 46. Swanson RM, Platon A, Satrio J, Brown R, Hsu DD. Techno-Economic
an integrated shale gas processing and chemical manufacturing pro- Analysis of Biofuels Production Based on Gasification. Golden, CO:
cess. AIChE J. 2018;64(1):123-143. National Renewable Energy Laboratory (NREL); 2010.
26. Almustapha MN, Andrésen JM. Recovery of valuable chemicals from 47. Li K, Leigh W, Feron P, Yu H, Tade M. Systematic study of aqueous
high density polyethylene (HDPE) polymer: a catalytic approach for monoethanolamine (MEA)-based CO2 capture process: techno-
plastic waste recycling. Int J Environ Sci Develop. 2012;3(3):263-267. economic assessment of the MEA process and its improvements. Appl
27. Achilias DS, Roupakias C, Megalokonomos P, Lappas AA, Energy. 2016;165:648-659.
Antonakou ΕV. Chemical recycling of plastic wastes made from poly- 48. Grainger D, Hägg M-B. Techno-economic evaluation of a PVAm CO2-
ethylene (LDPE and HDPE) and polypropylene (PP). J Hazard Mater. selective membrane in an IGCC power plant with CO2 capture. Fuel.
2007;149(3):536-542. 2008;87(1):14-24.
28. Tian X, Meyer T, Lee H, You F. Sustainable design of geothermal 49. Ho MT, Allinson GW, Wiley DE. Reducing the cost of CO2 capture
energy systems for electric power generation using life cycle optimi- from flue gases using pressure swing adsorption. Ind Eng Chem Res.
zation. AIChE J. 2020;66(4):e16898. 2008;47(14):4883-4890.
29. Yue D, Kim MA, You F. Design of sustainable product systems and 50. Guinée JB, Lindeijer E. Handbook on Life Cycle Assessment: Operational
supply chains with life cycle optimization based on functional unit: Guide to the ISO Standards. Vol 7. Dordrecht, Netherlands: Springer
general modeling framework, mixed-integer nonlinear programming Science & Business Media; 2002.
algorithms and case study on hydrocarbon biofuels. ACS Sustain Chem 51. Gong J, You F. Sustainable design and synthesis of energy systems.
Eng. 2013;1(8):1003-1014. Curr Opin Chem Eng. 2015;10:77-86.
30. del Remedio Hernández M, García AN, Marcilla A. Catalytic flash 52. Gong J, You F. Value-added chemicals from microalgae: greener,
pyrolysis of HDPE in a fluidized bed reactor for recovery of fuel-like more economical, or both? ACS Sustain Chem Eng. 2015;3(1):82-96.
hydrocarbons. J Anal Appl Pyrolysis. 2007;78(2):272-281. 53. Zhao N, Lehmann J, You F. Poultry waste valorization via pyrolysis
31. del Remedio Hernández M, Gómez A, García AN, Agulló J, Marcilla A. technologies: economic and environmental life cycle optimization for
Effect of the temperature in the nature and extension of the primary sustainable bioenergy systems. ACS Sustain Chem Eng. 2020;8(11):
and secondary reactions in the thermal and HZSM-5 catalytic pyroly- 4633-4646.
sis of HDPE. Appl Catal Gen. 2007;317(2):183-194. 54. Gao J, You F. Economic and environmental life cycle optimization of
32. Gracida-Alvarez UR, Mitchell MK, Sacramento-Rivero JC, noncooperative supply chains and product systems: modeling frame-
Shonnard DR. Effect of temperature and vapor residence time on the work, mixed-integer Bilevel fractional programming algorithm, and
micropyrolysis products of waste high density polyethylene. Ind Eng shale gas application. ACS Sustain Chem Eng. 2017;5(4):3362-3381.
Chem Res. 2018;57(6):1912-1923. 55. Gong J, You F. Consequential life cycle optimization: general concep-
33. Gao J, You F. Integrated hybrid life cycle assessment and optimization tual framework and application to algal renewable diesel production.
of shale gas. ACS Sustain Chem Eng. 2018;6(2):1803-1824. ACS Sustain Chem Eng. 2017;5(7):5887-5911.
34. Gong J, You F. Global optimization for sustainable design and syn- 56. Aspen HYSYS. Aspen HYSYS Customization Guide. Burlington, MA:
thesis of algae processing network for CO2 mitigation and biofuel Aspen Technology; 2010.
production using life cycle optimization. AIChE J. 2014;60(9): 57. ecoinvent V3.6. https://v36.ecoquery.ecoinvent.org/
3195-3210. 58. Astrup T, Fruergaard T, Christensen TH. Recycling of plastic: account-
35. DiNapoli TP. Local Governments and the Municipal Solid Waste Landfill ing of greenhouse gases and global warming contributions. Waste
Business. Albany, NY: Office of the New York State; 2018. Manag Res. 2009;27(8):763-772.
36. Advancing Sustainable Materials Management: 2016 and 2017 Tables 59. Khoo HH. LCA of plastic waste recovery into recycled materials, energy
and Figures. Washington, D.C.: United States Environmental Protec- and fuels in Singapore. Resour Conserv Recycl. 2019;145:67-77.
tion Agency; 2019. 60. Hartmann D, Tank A, Rusticucci M. IPCC Fifth Assessment Report, Cli-
37. Rahimi N, Karimzadeh R. Catalytic cracking of hydrocarbons over mate Change 2013: The Physical Science Basis Ipcc Ar5. Vol 5. Cam-
modified ZSM-5 zeolites to produce light olefins: a review. Appl Catal bridge University Press, Cambridge, United Kingdom, and New York,
Gen. 2011;398(1):1-17. NY, USA; 2013:31-39.
38. Yang M, Tian X, You F. Manufacturing ethylene from wet shale gas 61. Goedkoop M, Heijungs R, Huijbregts M, De Schryver A, Struijs J, Van
and biomass: comparative technoeconomic analysis and environmen- Zelm R. A Life Cycle Impact Assessment Method Which Comprises
tal life cycle assessment. Ind Eng Chem Res. 2018;57(17):5980-5998. Harmonised Category Indicators at the Midpoint and the Endpoint Level.
39. Kelly M, Goyal SK, McEachern JJ. Short chain branching control on The Hague, the Netherlands: Ministry of VROM. ReCiPe; 2009.
ethylene-butene copolymers. US patent 10,221,266. Inventors: U.S. 62. Goedkoop M, Heijungs R, Huijbregts M, De Schryver A, Struijs J, Van
Patents, assignee; 2019. Zelm R. ReCiPe 2008. A Life Cycle Impact Assessment Method which
40. Buekens AG, Froment GF. Thermal cracking of propane. Kinetics and Comprises Harmonised Category Indicators at the Midpoint and the End-
product distributions. Ind Eng Chem Process Des Dev. 1968;7(3):435-447. point Level. Vol 1. Netherlands; 2009:1-126.
41. Meyers RA. Handbook of petroleum refining processes. 548. McGraw- 63. Chen Y, Adams TA, Barton PI. Optimal design and operation of static
Hill New York; 2004. energy polygeneration systems. Ind Eng Chem Res. 2011;50(9):5099-
42. Meyers RA. Handbook of Petroleum Refining Processes. Vol 548. 5113.
New York, NY: McGraw-Hill; 2004. 64. Zhong Z, You F. Globally convergent exact and inexact parametric
43. Wang W, Gou Z, Zhu S. Liquid-liquid equilibria for aromatics extrac- algorithms for solving large-scale mixed-integer fractional programs
tion systems with tetraethylene glycol. J Chem Eng Data. 1998;43(1): and applications in process systems engineering. Comput Chem Eng.
81-83. 2014;61:90-101.
65. You F, Grossmann IE. Stochastic inventory management for tactical 76. Gamrath G, Fischer T, Gally T, et al. The Scip Optimization Suite. Vol 3.
process planning under uncertainties: MINLP models and algorithms. Berlin, Germany: Zuse Institute; 2016:2.
AIChE J. 2011;57(5):1250-1277. 77. Gong J, You F. Resilient design and operations of process systems:
66. Glover F. Improved linear integer programming formulations of nonlinear adaptive robust optimization model and algorithm for resil-
nonlinear integer problems. Manag Sci. 1975;22(4):455-460. ience analysis and enhancement. Comput Chem Eng. 2018;116:
67. Goldsberry C. Brightmark Energy Narrows Site Search for Next 231-252.
Advanced Plastic Recycling Plant. 2020; https://www.plasticstoday. 78. Gracida-Alvarez UR, Winjobi O, Sacramento-Rivero JC, Shonnard DR.
com/recycling/brightmark-energy-narrows-site-search-next- System analyses of high-value chemicals and fuels from a waste high-
advanced-plastic-recycling-plant/35082194462817. Accessed April density polyethylene refinery. Part 1: conceptual design and techno-
13, 2020. economic assessment. ACS Sustain Chem Eng. 2019;7(22):18254-
68. AFPM United States Refining and Storage Capacity Report. 2019; 18266.
https://www.afpm.org/system/files/attachments/AFPM-Capacity- 79. Zhu H, Huang G. SLFP: a stochastic linear fractional programming
Report-2019.pdf. Accessed August 2019. approach for sustainable waste management. Waste Manag. 2011;31
69. American Liberty Petroleum Corporation. Announces Foreland Refin- (12):2612-2619.
ing's Crude Oil Purchase Contract For All Production from Gabbs Oil 80. Peters MS, Timmerhaus KD, West RE. Equipment Costs - Plant Design
Field. https://www.businesswire.com/news/home/20120424006140/ and Economics for Chemical Engineers. New York, NY: Mc Graw Hill;
en/American-Liberty-Petroleum-Corp.-Announces-Foreland-Refinings- 2002.
Crude-Oil-Purchase-Contract-For-All-Production-from-Gabbs-Oil. 81. Winjobi O, Shonnard DR, Zhou W. Production of hydrocarbon fuel
Accessed April 2012. using two-step torrefaction and fast pyrolysis of pine. Part 1: techno-
70. Aspen Plus A. V10.0. Burlington, MA: Aspen Technology; 2017. economic analysis. ACS Sustain Chem Eng. 2017;5(6):4529-4540.
71. Aspentech AP. V10. Aspen Properties V10; Aspen Process Economic
Analyzer V10. 2017.
72. Jones SB, Meyer PA, Snowden-Swan LJ, et al. Process Design and Eco-
SUPPORTING INF ORMATION
nomics for the Conversion of Lignocellulosic Biomass to Hydrocarbon
Additional supporting information may be found online in the
Fuels: Fast Pyrolysis and Hydrotreating Bio-Oil Pathway. Richland, WA:
Pacific Northwest National Laboratory (PNNL); 2013. Supporting Information section at the end of this article.
73. Pinaud BA, Benck JD, Seitz LC, et al. Technical and economic feasibil-
ity of centralized facilities for solar hydrogen production via photo-
catalysis and photoelectrochemistry. Energ Environ Sci. 2013;6(7): How to cite this article: Zhao X, You F. Waste high-density
1983-2002. polyethylene recycling process systems for mitigating plastic
74. Rosenthal R. GAMS - A User's Guide. Washington, D.C.: GAMS Devel-
pollution through a sustainable design and synthesis paradigm.
opment Corporation; 2016.
75. Tawarmalani M, Sahinidis NV. A polyhedral branch-and-cut approach AIChE J. 2021;67:e17127. https://doi.org/10.1002/aic.17127
to global optimization. Math Program. 2005;103(2):225-249.

AIChE Journal - 2020 - Zhao - Waste High Density Polyethylene Recycling Process Systems For Mitigating Plastic Pollution

Uploaded by

Copyright:

Available Formats

AIChE Journal - 2020 - Zhao - Waste High Density Polyethylene Recycling Process Systems For Mitigating Plastic Pollution

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

AIChE Journal - 2020 - Zhao - Waste High Density Polyethylene Recycling Process Systems For Mitigating Plastic Pollution

Uploaded by

Copyright:

Available Formats

15475905, 2021, 4, Downloaded from https://aiche.onlinelibrary.wiley.com/doi/10.1002/aic.17127 by Cornell University, Wiley Online Library on [07/11/2022].

PROCESS SYSTEMS ENGINEERING

Waste high-density polyethylene recycling process systems

Xiang Zhao1 | Fengqi You1,2

1 | I N T RO DU CT I O N to recycle waste plastics.8 The closed-loop process produces recov-

AIChE J. 2021;67:e17127. wileyonlinelibrary.com/journal/aic © 2020 American Institute of Chemical Engineers 1 of 25

To satisfy the need of hydrogen in the heavy component hydrotreating

F I G U R E 1 4 Process flow sheet of

BARON 17.1 SCIP 3.2 Global optimization algorithm

Abbreviation: HDPE, high-density polyethylene.

The detailed computational performance of solving Instance 3 is 5 | CONC LU SION

this superstructure, a superstructure optimization problem was formu- RE FE RE NCE S

You might also like