Professional Documents
Culture Documents
Applied Energy
journal homepage: www.elsevier.com/locate/apenergy
Process Systems Engineering Centre (PROSPECT), Research Institute of Sustainable Environment, Universiti Teknologi Malaysia, 81310 UTM Johor Bahru, Johor, Malaysia
Faculty of Chemical Engineering, Universiti Teknologi Malaysia, 81310 UTM Johor Bahru, Johor, Malaysia
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
a r t i c l e
i n f o
Article history:
Received 24 November 2014
Received in revised form 24 June 2015
Accepted 15 July 2015
Available online 28 August 2015
Keywords:
Heat exchanger network synthesis
Design target
DT min
Flexibility
Sensitivity
Operability
a b s t r a c t
This paper presents a systematic technique to select the optimal design target for the heat exchanger
network (HEN) synthesis by using a new trade-off plot which considers aspects of design, controllability
in terms of steady state exibility and sensitivity analysis, and cost. By selecting the HEN design target
according to this guideline, the designer is able to predict the design, operability, and controllability of
the designed HEN at the beginning of the synthesis stage. In this study, the HEN design target that needs
to be optimized is the value of the minimum temperature difference (DT min ). In traditional HEN synthesis,
designers only consider the trade-off between capital and operating costs in selecting the best DT min . As a
result, the HEN design at the selected DT min may not be optimum in terms of steady state controllability.
In addition to considering the capital and operating costs, the proposed new method provides additional
design insights in terms of energy recovery, operability, controllability (steady state) through the exibility and sensitivity. The proposed trade-off plot allows designers to choose the most suitable design target
either for the purpose of improving a networks energy recovery and/or controllability. A case study has
q
This article is based on a short proceedings paper in Energy Procedia Volume 161 (2014). It has been substantially modied and extended, and has been subject to the
normal peer review and revision process of the journal. This paper is included in the Special Issue of ICAE2014 edited by Prof. J. Yan, Prof. D.J. Lee, Prof. S.K. Chou, and Prof. U.
Desideri.
qq
The short version of the paper was presented at 6th International Conference on Applied Energy 2014 (ICAE2014) conference. This paper is the full paper with signicant
revision of the previously presented short version at the Conference.
Corresponding author at: Process Systems Engineering Centre (PROSPECT), Research Institute of Sustainable Environment, Universiti Teknologi Malaysia, 81310 UTM,
Johor Bahru, Johor, Malaysia. Tel.: +60 7 5535517; fax: +60 7 5536165.
E-mail address: kamaruddin@cheme.utm.my (Mohd. Kamaruddin Abd. Hamid).
http://dx.doi.org/10.1016/j.apenergy.2015.07.056
0306-2619/ 2015 Elsevier Ltd. All rights reserved.
1260
been applied to test the capability of the new proposed trade-off plot. The results show that DT min = 40 C
is the optimal design target to synthesize exible and operable HEN.
2015 Elsevier Ltd. All rights reserved.
1. Introduction
Heat exchanger network (HEN) is an important element for
enhancing energy efciency via heat recovery in the chemical
process industry. Extensive research has been done to improve
HENs design and synthesis over the past 40 years. Heat integration
based on the Pinch Analysis (PA) has been an established method
to enhance heat recovery in industries since its introduction by
Linnhoff and Flower [1]. PA aims to maximize energy recovery
and reduce external heating and cooling utilities.
Traditionally, a cost-effective HEN design aims to maximise heat
recovery by considering the trade-off between the capital and
operating costs for a given DT min . Sun et al. [2] investigated the
effect of DT min contribution on cost, by considering multiple utilities. Akbarnia et al. [3] applied PA to do supertargeting by considering piping cost, materials of construction and pressure rating.
Kravanja and Glavic [4] performed cost targeting on a heat exchanger network using a combination PA and a complex algorithmic
approach. The authors combined PA with a complex algorithmic
approach to target cost and to reduce feasibility region dened
over sets of discretized decision variables. Wang et al. [5] adapted
the PA technique to propose a graphical method to represent heat
duty versus time for batch streams.
After the HEN design is xed, the HEN controllability is
analyzed. Escobar and Trierweiler [6] simultaneously considered
control and optimization of an existing HEN design. Jschke and
Skogestad [7] have done self-optimizing control for stream split
of heat exchanger system. If the network is too complex, the
dynamic behavior becomes more sensitive and unstable. In such
cases, it may be necessary to apply advance process control in
order to stabilize the dynamic behavior. Iancu et al. [8] and
Giovanini and Balderud [9] proposed a proper MPC controller
approach for HEN retrot. To date, less research has focused on
HEN synthesis that considers the controllability criteria. Operational issues such as controllability have typically been neglected
at the early design stage because HEN design and control problems
have been solved separately as opposed to simultaneously. As a
result, it may not be possible to guarantee a robust performance
for such HEN designs, which may be exposed to controllability
and operability problems due to the xed and tight designs. Severe
propagation of disturbances through such process may render it to
be extremely difcult to operate or control, regardless of the
advanced control techniques used.
Research on integrating process design and process control
(IPDC) has been widely reported. Hamid et al. [10] presented the
model-based IPDC in order to solve reactionseparation system
with recycle by decomposing the problem into several stages.
Sharifzadeh and Thornhill [11] proposed an optimization framework for IPDC using the dynamic inversely controlled process
model to solve two heat-integrated series reactor. Although IPDC
research has been reported to solve chemical process system, there
are still less guidance to solve HEN that consider process design
and process control together.
There is a clear need to develop a new methodology for the
design and synthesis of a exible and operable HEN. The HEN
design can be further improved in terms of cost optimality as well
as controllability if these aspects are considered simultaneously
during the early design stage. Note that the DT min is a key parameter inuencing the HEN design and synthesis. In particular, how
max J
m X
n
X
wi;j Pi;j
i1 j1
Subjected to:
Process (dynamics and/or steady state) constraints:
dx=dt f u; x; y; d; h; Y; t
0 g 1 u; x; y h
1261
Fig. 1. Methodology for designing a exible and operable heat exchanger network.
0 h1 u; x; y
0 6 h2 u; x; y; d
CS y uY
1262
Ts
DT min
DT min
hT h hT t
2
2
Ts
DT min
DT min
hT c hT t
2
2
(
Ft
ft
/1 P1; R1 /1 P2 ; R2
Pin
F
14
Constraints
u = percentage of maximum manipulated variable
n = number of manipulated variable
Heat exchanger temperature crosses
Heat exchanger ft correction factors
/2 P1 ; R1 /2 P2 ; R2
DT M
q
10
Pin dyi
s
q h
i
1P1
R21 1 ln 1R
1 P 1
ft
p
1P 1 1R1 P 1 P 1 R21 1
p
R1 1 ln
2
1P 1 1R1 P 1 P 1
i1 u%
11
R1 1
i1 ddi
15
y = disturbances effect
d = disturbances
n = number of streams affected by disturbance
2.5. Stage 4: optimal selection and verication
P1
t2 t1
T 1 t1
12
R1
CB T 1 T 2
CA
t2 t1
13
1263
16
(see Fig. 2). The trade-off plot allows one to understand the
impacts of process design, process control, and process economics
during the early stage of HEN design.
Firstly, the range of the DT min was determined in Stage 1 based
on the practical operation of heat exchanger, which was set
between 5 C and 40 C [12]. The selection of the different DT min
provides important insights on the process design, process control
and the HEN cost. A smaller DT min (at the left side of Fig. 2) would
result in a HEN with higher energy saving and a lower operating
cost. However, the smaller DT min will result in a larger heat transfer
area and a higher capital cost. In terms of process controllability, a
smaller DT min will lead to a lower controllability performance,
leading to a network that is less exible and very sensitive to
disturbances. Choosing a larger DT min (at the right side of Fig. 2)
is expected to result in the effects opposite to those obtained with
a smaller DT min . A larger DT min will result in lower energy saving,
higher operating cost, and better process controllability.
2.7. Case study application
Index
Tmin
Small T min
Moderate T min
Large T min
Energy Recovery
Controllability
Capital Cost
Operaon cost
Fig. 2. The trade-off plot showing the four important aspects of a exible and
operable HEN.
In the illustrative case study, the operational window is considered based on Eqs. (7) and (8). Eqs. (17)(22) are the operational
window for streams H1, H2, H3, C1, C2, and C3. The operational
window design target (the optimal DT min is between 5 C and
40 C.
17
18
19
Table 1
Stream data for the HEN synthesis [17].
No
Stream
Descriptions
Enthalpy, DH (kW)
1
2
3
4
5
6
H1
H2
H3
C1
C2
C3
Reactor 1 Product
Reactor 2 Product
Distillate Product
Reactor 2 Feed 1
Reactor 2 Feed 2
Reactor 1 Feed 1
300
230
160
40
100
230
160
120
60
230
230
300
3
7
2
2
4
3
420
770
200
380
520
210
1264
20
21
22
Q FCpT in T out
23
1265
Fig. 3. Composite curves for HEN designs at (a) DT min = 5 C, (b) DT min = 25 C, and (c) DT min = 40 C.
It can be stated that the values of ft correction factors are quite low
at the small DT min for HE3, HE4, and HE6 and gradually increases as
the DT min increases. Therefore, the higher the DT min , the higher the
feasibility becomes, except for HE1. Fig. 6 shows that the curve for
HE1 is quite different because of the temperature supplies for hot
and cold streams of HE1 depend on the outlet of the previous heat
1266
Fig. 4. Grid diagrams of two different networks at: (a) DT min = 5 C, (b) DT min = 25 C, and (c) DT min = 40 C.
1267
DT min C
5
10
15
20
25
30
35
40
1080
1050
1020
990
960
930
900
870
340
400
460
520
580
640
700
760
5
5
5
2
2
exchanges. Cold feed stream for HE1 is the stream 16, which is also
the outlet stream of HE4. Therefore, the temperature of this stream
depends on the heat duty of HE4. Similarly, the hot feed stream for
HE1 which is stream 20 is also the outlet stream of HE2. The
temperature of this stream also depends on the heat duty of HE2.
These explain why the HE1 curve is quite different from other
curves in Fig. 6.
In addition, ft correction factors for all heat exchangers in the
HEN design at DT min = 30 C were above the minimum requirement. Therefore, the HEN design at DT min = 30 C was found as feasible and further tested together with HEN designs at DT min = 35 C
and 40 C. The other candidates were eliminated from the option.
Based on the information obtained in Stage 1 and Stage 2, the
options were narrowed down to several feasible HENs designs from
among the eight candidates. Only three candidates were
considered as a feasible design, which included HEN designs at
DT min = 30 C, 35 C, and 40 C. These candidates were tested
further in Stage 3 and Stage 4. In Stage 3, the feasible candidates
were veried with the exibility and sensitivity tests. The exibility test procedure is as follows:
(1) Increase the value of the manipulated variable which is the
feed owrate of stream H1 until the operation warning
occurs.
(2) Note down the maximum owrate that the network can tolerate, and then change back the owrate of stream H1 to the
initial value.
(3) Repeat steps 12 for the feed owrates of streams H2; C1
and C2.
(4) Calculate the average maximum of the increment owrate.
(5) Repeat steps 14 for the feed temperature.
The results of the exibility test are summarized in Table 3. It
can be seen that after the exibility test was done, the HEN design
at DT min = 30 C was less exible. Note that, even though the design
at DT min = 30 C was feasible, the network structure cannot tolerate
even the minimum feed owrate changes in the network involving
1 C change in temperature and 1 kg/h change in owrate. As a
result, this candidate was eliminated in the next analysis.
Therefore, the only remaining feasible candidates for the exibility
and sensitivity tests were designs at DT min = 35 C and 40 C.
Results for the exibility test shows that the candidate at
DT min = 40 C is more exible than the other design (see Table 3).
HEN candidates with the higher exibility can be the optimal
candidate for the process control, because they are more robust
and capable in maintaining the controlled variables with higher
owrate.
The next step is to do the sensitivity analysis. In this case, the
sensitivity test only involves the changes of feed stream temperatures since it is considered as the disturbance for the network.
Since only HEN design candidates at DT min = 35 C and 40 C are
Heat exchanger
Cooler
Heater
6
6
6
6
6
6
5
5
2
2
2
2
2
2
2
2
1
1
1
1
1
1
1
2
considered as feasible and exible designs, therefore, the sensitivity analysis were done for both designs. The sensitivity test consists of six steps:
(1) Increase the feed temperature for stream H1 to about 2%
from the initial.
(2) Note down the temperature changes in all streams.
(3) Calculate the derivative temperature effect of dy/dd for all
streams after the disturbance changes are applied.
(4) Change the temperature of stream H1 back to the initial
value.
(5) Repeat steps 14 for the temperature of streams H2; C1 and
C2.
(6) Calculate the average derivative value of the controlled
variables (temperatures in all streams except feed streams
temperature) with respect to disturbance dy/dd.
HEN design with higher sensitivity criteria is not a good design
since it will be too sensitive to the smaller changes in disturbances,
hence makes it very difcult to be controlled. The results of the
sensitivity analysis are summarized in Table 3. Note that the
design at DT min = 35 C is more sensitive than the other design.
3.4. Validation of the exible and operable HEN based on cost and
objective function
The remaining candidates were then further tested in Stage 4. In
this stage, the capital and operational costs for both HEN designs
were calculated. The heat exchanger surface area was calculated
by using Eqs. (24) and (25). The capital cost can be estimated from
the purchased cost diagram [18] by using the area obtained from
Eq. (10) and by assuming an ft value of 1. Calculations of the operating cost depend on the external utility obtained from Stage 1. The
calculated capital and operating costs are shown in Table 4. From
the results, it can be conrmed that the total heat exchanger area
for HEN designed at DT min = 35 C is less than the one designed at
DT min = 40 C. However, this observation contradicted with the
trade-off prediction. The reason is that, the number of units for
the HEN designed at DT min = 35 C was 1 unit less than the HEN
designed at DT min = 40 C (see Fig. 5d). Therefore, the total heat
exchanger area for the HEN designed at DT min = 40 C is slightly
larger than the one at DT min = 35 C.
Both candidates were then validated by calculating the
multi-objective function by using Eq. (16). However, before calculating the multi-objective function, the value of the objective
function, Px;x needed to be normalized. Since the range and unit
of each objective function value can be different, each objective
value was normalized with respect to its maximum value, as
shown in Table 5.
In this case, all the objective function terms are weighted
equally so that the decision-maker does not have any preference
1268
Fig. 5. HEN designs representation in the Aspen HYSYS process simulator: (a) DT min = 5, 10, 15 C, (b) DT min = 20, 25 C, (c) DT min = 30 C, (d) DT min = 35 C, and
(e) DT min = 40 C.
1269
Fig. 5 (continued)
Q UADT LMTD
Q = rate of heat transfer
U = mean overall heat transfer coefcient
A = heat transfer surface area
DTLMTD = Log mean temperature different
24
1270
1.05
1.05
0.95
correcon factors
0.95
0.85
0.75
0.65
0.85
0.75
0.65
0.55
0.55
35
0.45
0
10
15
20
25
30
35
40
Tmin
Tmin
HE1
HE3
HE4
HE6
HE5
Energy Recovery
Sensivity
Flexibility
Capital Cost
Operaon Cost
Fig. 7. Trade-off plot constructed for the two different HEN designs.
Table 3
Controllability analysis results for designed HENs at different DT min .
4. Conclusions
DTmin (C)
Feasibility
Flexibility (f), %
5
10
15
20
25
30
35
40
No
No
No
No
No
Yes
Yes
Yes
Not exible
24.720
42.183
1.580
1.065
Table 4
Capital and operating costs of the HEN designs at DT min = 35 C and 40 C.
DTmin
35
40
10,571
10,714
107,100
118,800
DT LMTD
Dt 1 Dt 2
lnDt 1 =Dt 2
25
Table 5
Multi-objective function calculations.
Energy recovery (design)
Flexibility (control)
Sensitivity (control)
P1;1
900.000
870.000
P2;1
24.720
44.183
P2;2
1.580
1.065
P3;1
10,571
10,714
P3;2
107,100
118,800
P1;1s
1.000
0.967
P2;1s
0.586
1.000
P2;2s
1.000
0.674
P3;1s
0.986
1.000
P3;2s
0.902
1.000
P1;1s
1.000
0.967
P2;1s
0.586
1.000
1=P2;2s
1.000
1.484
1=P3;1s
1.013
1.000
1=P3;2s
1.109
1.000
4.862
5.451
References
[1] Linnhoff B, Flower JR. Synthesis of heat exchanger networks: I. Systematic
generation of energy optimal networks. AIChE J 1978;24:63342.
[2] Sun KN, Wan Alwi SR, Manan ZA. Heat exchanger network cost optimization
considering multiple utilities and different types of heat exchangers. Comput
Chem Eng 2013;49:194204.
[3] Akbarnia M, Amidpour M, Shadaram A. A new approach in pinch technology
considering piping costs in total cost targeting for heat exchanger network.
Chem Eng Res Des 2009;87:35765.
[4] Kravanja Z, Glavic P. Cost targeting for HEN through simultaneous
optimization approach: a unied pinch technology and mathematical
programming design of large HEN. Comput Chem Eng 1997;21:83353.
[5] Wang Y, Wei Ying, Feng Xiao Chu, Khim Hoong. Synthesis of heat exchanger
networks featuring batch streams. Appl Energy 2014;114:3044.
[6] Escobar M, Trierweiler JO. Bypass design for control and optimization of heat
exchanger networks. In: Rita Maria de Brito Alves CAOdN, Evaristo Chalbaud B,
editors. Computer aided chemical engineering. Elsevier; 2009. p. 166570.
[7] Jschke J, Skogestad S. Optimal operation of heat exchanger networks with
stream split: only temperature measurements are required. Comput Chem Eng
2014;70:3549.
1271
[8] Iancu M, Cristea MV, Agachi PS. Retrot design of heat exchanger network of a
uid catalytic cracking plant and control based on MPC. Comput Chem Eng
2013;49:20516.
[9] Giovanini L, Balderud J. Control and operation of heat exchanger networks
using model predictive control. In: 2nd International workshop on networked
control systems. Rende, Italy; 2006. p. 516.
[10] Hamid MKA, Sin G, Gani R. A new model-based methodology for simultaneous
design and control of reaction-separation system with recycle. In: Jacek J, Jan
T, editors. Computer aided chemical engineering. Elsevier; 2009. p. 83945.
[11] Sharifzadeh M, Thornhill NF. Integrated design and control using a dynamic
inversely controlled process model. Comput Chem Eng 2013;48:12134.
[12] Shenoy UV. Heat exchanger network synthesis:: process optimization by
energy and resource analysis. Texas: Gulf Professional Publishing; 1995.
[13] Russel BM, Henriksen JP, Jrgensen SB, Gani R. Integration of design and
control through model analysis. Comput Chem Eng 2002;26:21325.
[14] Klemes J, Friedler F, Bulatov I, Varbanov P. Sustainability in the process
industry: integration and optimization. New York: Integration and
Optimization: McGraw Hill; 2010.
[15] Linnhoff B, Hindmarsh E. The pinch design method for heat exchanger
networks. Chem Eng Sci 1983;38:74563.
[16] Shah RK, Sekulic DP. Frontmatter. Fundamentals of heat exchanger design. San
Francisco: John Wiley & Sons, Inc.; 2007.
[17] Wan Alwi SR, Manan ZA. STEPa new graphical tool for simultaneous
targeting and design of a heat exchanger network. Chem Eng J
2010;162:10621.
[18] Peters MS, Timmerhaus KD, West RE. Plan design and economics for chemical
engineers. Boston: McGraw-Hill; 2004.