MODELLING,

CONTROL AND
OPTIMISATION OF A
DUAL CIRCUIT
INDUCED DRAFT
COOLING WATER
SYSTEM

C.J. Muller
Sasol; University of Pretoria

Under supervision of:

Prof. I.K. Craig

University of Pretoria

February 2016

better together ... we deliver

Overview

• Introduction

• Process overview

• Modelling and validation

• Control and optimisation

• Case comparison

• Conclusion

better together ... we deliver 2

Introduction
• Process plants make extensive use of utilities (auxiliary process
variables) for example steam, electricity, compressed air, nitrogen and
cooling water.
• When it comes to optimisation, the focus is typically on the
consumption of the utility and not so much utility generation and/or
transportation/transmission
• Utilities account for a significant portion of fixed cost of a plant
• This study covers the modelling, control and optimisation of a dual
circuit induced draft cooling water system
• The purpose of the modelling is to provide a platform for simulation
and controller/optimiser design
• The control and optimisation objectives are to reduce energy
consumption/cost while honouring process and equipment
constraints

better together ... we deliver 3

Process overview
• Two Circuits: Tempered Water (TW) and Cooling Water (CW)

better together ... we deliver 4

Process overview
Dual circuit cooling water system with induced draft counter flow
cooling towers
Cooling Water
Return (CWR)
Tempered Water
Supply (TWS) CT-211 CT-212 CT-213 CT-214
TIC
101

E-101
PIC
201
Plant P-101

PIC
E-102 202
P-201
P-102
PIC
203
Tempered Water P-202
Return (TWR) E-103
P-103
PIC
204
P-203

E-104
P-104 PIC
205
P-204

Cooling Water
P-105 E-105 Supply (CWS) P-205

Process overview

better together ... we deliver 5

Process overview
• Two Circuits: Tempered Water (TW) and Cooling Water (CW)
• TW used in plant heat exchanger network where it collects heat
• TW transfers heat to CW though bank of heat exchangers

better together ... we deliver 6

Process overview
Dual circuit cooling water system with induced draft counter flow
cooling towers
Cooling Water
Return (CWR)
Tempered Water
Supply (TWS) CT-211 CT-212 CT-213 CT-214
TIC
101

E-101
PIC
201
Plant P-101

PIC
E-102 202
P-201
P-102
PIC
203
Tempered Water P-202
Return (TWR) E-103
P-103
PIC
204
P-203

E-104
P-104 PIC
205
P-204

Cooling Water
P-105 E-105 Supply (CWS) P-205

Process overview

better together ... we deliver 7

Process overview
• Two Circuits: Tempered Water (TW) and Cooling Water (CW)
• TW used in plant heat exchanger network
• TW transfers heat to CW though bank of heat exchangers
• Heat removed from the CW in the Cooling Towers (CTs) mainly by
means of partial evaporation

better together ... we deliver 8

Process overview
Dual circuit cooling water system with induced draft counter flow
cooling towers
Cooling Water
Return (CWR)
Tempered Water
Supply (TWS) CT-211 CT-212 CT-213 CT-214
TIC
101

E-101
PIC
201
Plant P-101

PIC
E-102 202
P-201
P-102
PIC
203
Tempered Water P-202
Return (TWR) E-103
P-103
PIC
204
P-203

E-104
P-104 PIC
205
P-204

Cooling Water
P-105 E-105 Supply (CWS) P-205

Process overview

better together ... we deliver 9

Process overview
• Two Circuits: Tempered Water (TW) and Cooling Water (CW)
• TW used in plant heat exchanger network
• TW transfers heat to CW though bank of heat exchangers
• Heat removed in Cooling Towers (CTs) mainly by means of partial
evaporation
• Each circuit is equipped with bank of pumps to provide flow

better together ... we deliver 10

Process overview
Dual circuit cooling water system with induced draft counter flow
cooling towers
Cooling Water
Return (CWR)
Tempered Water
Supply (TWS) CT-211 CT-212 CT-213 CT-214
TIC
101

E-101
PIC
201
Plant P-101

PIC
E-102 202
P-201
P-102
PIC
203
Tempered Water P-202
Return (TWR) E-103
P-103
PIC
204
P-203

E-104
P-104 PIC
205
P-204

Cooling Water
P-105 E-105 Supply (CWS) P-205

Process overview

better together ... we deliver 11

Process overview
• Two Circuits: Tempered Water (TW) and Cooling Water (CW)
• TW used in plant heat exchanger network
• TW transfers heat to CW though bank of heat exchangers
• Heat removed in Cooling Towers (CTs) mainly by means of partial
evaporation
• Each circuit is equipped with bank of pumps
• A temperature control valve is installed to bypass heat exchangers
on TW side to provide a handle for TW supply temperature control

better together ... we deliver 12

Process overview
Dual circuit cooling water system with induced draft counter flow
cooling towers
Cooling Water
Return (CWR)
Tempered Water
Supply (TWS) CT-211 CT-212 CT-213 CT-214
TIC
101

E-101
PIC
201
Plant P-101

PIC
E-102 202
P-201
P-102
PIC
203
Tempered Water P-202
Return (TWR) E-103
P-103
PIC
204
P-203

E-104
P-104 PIC
205
P-204

Cooling Water
P-105 E-105 Supply (CWS) P-205

Process overview

better together ... we deliver 13

Process overview
• Two Circuits: Tempered Water (TW) and Cooling Water (CW)
• TW used in plant heat exchanger network
• TW transfers heat to CW though bank of heat exchangers
• Heat removed in Cooling Towers (CTs) mainly by means of partial
evaporation
• Each circuit is equipped with bank of pumps
• A temperature control valve is installed to bypass heat exchangers on
TW side to provide a handle for TW supply temperature control
• Control valves exist on the discharges of the CW pumps, originally
used for pump overload protection

better together ... we deliver 14

Process overview
Dual circuit cooling water system with induced draft counter flow
cooling towers
Cooling Water
Return (CWR)
Tempered Water
Supply (TWS) CT-211 CT-212 CT-213 CT-214
TIC
101

E-101
PIC
201
Plant P-101

PIC
E-102 202
P-201
P-102
PIC
203
Tempered Water P-202
Return (TWR) E-103
P-103
PIC
204
P-203

E-104
P-104 PIC
205
P-204

Cooling Water
P-105 E-105 Supply (CWS) P-205

Process overview

better together ... we deliver 15

Process overview
• Two Circuits: Tempered Water (TW) and Cooling Water (CW)
• TW used in plant heat exchanger network
• TW transfers heat to CW though bank of heat exchangers
• Heat removed in Cooling Towers (CTs) mainly by means of partial
evaporation
• Each circuit is equipped with bank of pumps
• A temperature control valve is installed to bypass heat exchangers on
TW side to provide a handle for TW supply temperature control
• Control valves exist on the discharges of the CW pumps, originally used
for pump overload protection

• This is an example of a Hybrid system: contains both discrete and

continuous input variables

better together ... we deliver 16

Modelling and Validation
• Model derived mathematically:
• Pump calculations:
• Polynomial estimation from manufacturer’s pump curves
• Receives flow rate, produces discharge pressure
• Flow calculations:
• Mass balance, system flow coefficients, valve equations
• Duty/temperature calculations:
• Heat exchange equations, enthalpy change, energy balance, evaporative flow
• Energy consumption calculations:
• Rated power (for fans) and polynomial estimations of manufacturer’s curves
(pumps)
• Dynamics added to important variables to convert from steady-state to
dynamic model and derive state-space form
• Model verified against plant data for a period of 6 days (144 hours) during
which significant load changes occurred
• Genetic algorithm used in parameter estimation
to obtain better accuracy

better together ... we deliver 17

Modelling Results
• Correlation coefficient and least square error approaches applied to
gauge model quality
• Correlation between model and plant data:

• Adequate accuracy for the purposes of this simplified model

• Important to have correct directionality as verified by the step testing
results shown in the thesis

better together ... we deliver 18

Modelling Results (continued)
TW temperatures – simulated vs. plant data

Model response (solid line) vs. plant data (dotted line).

better together ... we deliver 19

Control and Optimisation
• Four cases were considered:
• Base case
• Advanced Regulatory Control (ARC)
• Hybrid Non-linear Model Predictive Control (HNMPC)
• Economic Hybrid Non-linear Model Predictive Control (EHNMPC)
• Two simulations for each case:
• Simulation 1: Artificial plant input data
• Simulation 2: Actual plant input data (same as that used for verification)

Simulation 1 Simulation 2

better together ... we deliver

Control and Optimisation (continued)
ARC Design:
• Aim is to make better use of base layer: use override selector control,
cascade control and rule-based switching logic to manipulate discrete
variables
• Overall objective is to minimise energy consumption by switching
equipment off when overcooling is provided
• No plant model required

better together ... we deliver 21

Control and Optimisation (continued)
ARC scheme illustration

FTW TTWS TTWI

TIC TIC Desired
FTV 103 102 Setpoint
TIC
101

MID
Plant
CT fan
FIC
101
switching

FTWI
TTWR FCW

TW pump CW pump
switching switching

better together ... we deliver 22

Control and Optimisation (continued)
ARC Design:
• Aim is to make better use of base layer: use override selector control,
cascade control and rule-based switching logic to manipulate discrete
variables
• Overall objective is to minimise energy consumption by switching
equipment off when overcooling is provided
• No plant model required
APC Design:
• Use the model of the system to develop a model predictive control
strategy
• Model is non-linear and hybrid which complicates controller design
• Genetic algorithm used as optimiser: capable of handling this type of
system directly
• Cost function mainly total energy consumption/cost
• Iteration time 30 minutes, prediction horizon 12, control horizon 4
• MVs: pumps, fans, flow controllers, temperature control valve
• CVs: TW supply and differential temperatures, power/cost
better together ... we deliver 23
Control and Optimisation (continued)
APC scheme illustration

FTW TTWS TTWI

FTV
APC

Plant
FIC
101

FTWI
FCW
TTWR

better together ... we deliver 24

Control and Optimisation Results
Base case (CVs) – Simulation 2

CVs

better together ... we deliver 25

Control and Optimisation Results (continued)
ARC case – Simulation 2
MVs CVs

better together ... we deliver 26

Control and Optimisation Results (continued)
HNMPC case – Simulation 2
MVs CVs

better together ... we deliver 27

Control and Optimisation Results (continued)
EHNMPC case – Simulation 2
MVs CVs

better together ... we deliver 28

Case Comparison
Energy/Power Consumption

better together ... we deliver 29

Case Comparison (continued)
Energy/Power Cost

better together ... we deliver 30

Case Comparison (continued)
Constraint Violations

better together ... we deliver 31

Conclusion
• Utility optimisation shows promising potential for optimisation
• By using ARC techniques, the bulk of the benefit may be realised at a
fraction of the cost and effort of APC
• APC allows for a marginal further optimisation though at the cost of
increased complexity and modelling requirements
• Hybrid systems complicate the control and optimisation design and
many utility systems are of a hybrid nature
• MINLP is still underdeveloped as an industrial option for control and
optimisation – GA proved to be an effective option for this study
• Always scope for further investigation and improvement – both utility
optimisation and hybrid systems are intriguing fields for further studies

better together ... we deliver 32

 Confidential Document

“The only true wisdom is

in knowing you know
nothing.”

Socrates

THANK YOU
FOR
YOUR TIME

better together ... we deliver