Acs Iecr 0c03829

Subscriber access provided by CORNELL UNIVERSITY LIBRARY
Process Systems Engineering

Optimal Retrofit of Multi-Plant HEN with Fair Benefit Allocation Plan
Fang-Yi Lo, Yu-Xuan Zhu, and Chuei-Tin Chang
Ind. Eng. Chem. Res., Just Accepted Manuscript • DOI: 10.1021/acs.iecr.0c03829 • Publication Date (Web): 28 Aug 2020
Downloaded from pubs.acs.org on September 3, 2020
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,

Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the
course of their duties.
Page 1 of 58 Industrial & Engineering Chemistry Research
1
2
3
4
5
6
7
8
9
10
11
12
13
14 Optimal Retrofit of Multi-Plant HEN with Fair Benefit Allocation
15
16 Plan
17
18
19
20
21
22
23
24
25 Fang-Yi Lo, Yu-Xuan Zhu, Chuei-Tin Chang*
26
27
28
29
30 Department of Chemical Engineering, National Cheng Kung University, Tainan, Taiwan
31
32 70101, ROC
33
34
35
36
37
38
39
40
41
42
43 August 31, 2020
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59 *Corresponding author’s email: [email protected]
60
1
ACS Paragon Plus Environment
Industrial & Engineering Chemistry Research Page 2 of 58
1
2
3
4
5
6 Abstract
7
8
9 A model-based procedure has recently been developed to address the benefit
10
11 allocation issue among members of an interplant heat integration scheme in the spirit of
12
13 cooperative game.1 Although satisfactory results were reported, their approach is only
14
15
16 applicable to grass-root designs. In practical applications, the existing plants on an industrial
17
18 park were usually built to meet the targeted market demands which arose during different
19
20 periods in the past and each must have already been equipped with a heat exchanger network
21
22
23
(HEN) by the time when its construction was completed. Therefore, the aforementioned
24
25 allocation problem is more likely to take place when a multi-plant HEN retrofit project is
26
27 called for to facilitate further reduction in utility consumption. This paper presents three
28
29
viable strategies to solve the more realistic revamp problems. Depending upon other
30
31
32 requirements in practical applications, e.g., safety concerns, spatial limits and operability, etc.,
33
34 these strategies are devised by introducing different levels of restrictions on the new and
35
36 original matches, on re-piping and reuse of existing units in the multi-plant HEN, and on the
37
38
39 installation of purchased new heat exchangers, coolers and heaters. The actual financial
40
41 benefits allocated to the participating members of the interplant heat integration scheme are
42
43 then determined according to the corresponding Shapley values. A simple example is
44
45
46
presented in this paper to illustrate the aforementioned HEN retrofit and saving distribution
47
48 procedures.
49
50
51
52
53
54
55
56
57 Keywords: Retrofit design; Multi-plant heat-exchanger network; Shapley value; benefit
58
59 allocation; Cooperative game.
60
2
1
2
3
4
5 1. Introduction
6
7 Heat exchanger network (HEN) is traditionally used for maximum heat recovery in a
8
9
10
single chemical plant, while the multi-plant counterparts have been studied primarily for the
11
12 purpose of reaping additional energy savings. Since the early works on the latter issue
13
14 focused only upon minimization of the total energy cost of the entire site, the resulting
15
16 arrangements were often infeasible due to the fact that the individual savings may not be
17
18
19 reasonably distributed and therefore not always acceptable to all participating parties.
20
21 Although various methods have already been proposed to address this issue based upon game
22
23 theory, a common weakness of the available approaches is due to the assumption of grass-
24
25
26 root designs.1 In most cases in the real world, the existing plants on an industrial park were
27
28 built to meet the targeted market demands which arose during different periods in the past
29
30 and each should have already been equipped with a HEN by the time of its completion. In
31
32
33
other words, the above benefit allocation problem occurs mainly when a multi-plant HEN
34
35 retrofit project is initiated for the purpose of gaining extra energy savings.
36
37 As mentioned above, it is often possible to significantly reduce the total utility cost of
38
39 two or more standalone chemical plants via interplant heat integration, e.g., see Bagajewicz
40
41
42 and Rodera,2 Kralj3 and Liew et al.4,5 Although there are other effective techniques available
43
44 in literature, the model-based method is adopted in the present work as the primary HEN
45
46 synthesis tool since it is in general believed to be more rigorous and thus better equipped for
47
48
49 locating the true optimum. Zhang et al.6 proposed to use a superstructure for building a
50
51 mixed-integer nonlinear programming (MINLP) model to synthesize the multi-plant HEN
52
53 designs. Chang et al.7 presented an optimization methodology for interplant heat integration
54
55
56
using the intermediate fluid circle(s). On the other hand, it has also been well-established that
57
58 the model-based single-plant HEN synthesis strategies may be classified into two types, i.e.,
59
60 the simultaneous and sequential approaches. In the former case, by constructing a
3
1
2
3 superstructure and the corresponding MINLP model, the HEN design with the lowest total
4
5
6 annual cost (TAC) can be produced accordingly in a single step.8 In the latter case, the HEN
7
8 is generated in three consecutive steps. A linear transshipment model is formulated for
9
10 determining the minimum total utility cost in the first step, while a MILP model with the
11
12
13
embedded constraint of minimum utility cost is adopted in the second step to identify the
14
15 minimum number of matches and their heat duties.9 In the final step of the sequential
16
17 approach, the NLP model developed by Floudas et al.10 can be used to synthesize the optimal
18
19
HEN that minimizes the total capital investment. A modified version of the aforementioned
20
21
22 simultaneous HEN synthesis strategy has been formulated in the present study. This practice
23
24 is due to the need to achieve a better tradeoff between capital and energy costs and, also, due
25
26 to the easiness in model reformulation so as to facilitate identification of revamp
27
28
29 opportunities for interplant heat integration.
30
31 As indicated in Cheng et al.,11 many existing interplant heat integration arrangements
32
33 are often not implementable in practice due to the fact that the profit margin might be
34
35
36
unacceptable to one or more participating party. This drawback can be primarily attributed to
37
38 the conventional HEN design objective, i.e., minimization of overall energy cost. Thus, the
39
40 key to a successful total site heat integration (TSHI) scheme should be to allow every plant
41
42
obtain reasonable extra benefit while striving for the largest overall saving at the same time.
43
44
45 To address this benefit distribution issue, the above authors developed a non-cooperative-
46
47 game based sequential optimization strategy to generate the “fair” interplant integration
48
49 schemes via direct heat exchanges between the hot and cold process streams across plant
50
51
52 boundaries. To further improve the practical feasibility of TSHI projects, Chang et al.12
53
54 modified this sequential optimization approach by replacing the direct interplant heat-transfer
55
56 options with indirect ones. However, it should be noted that all aforementioned strategies are
57
58
59
weakened by two obvious drawbacks. First of all, the HEN design produced with the
60
4
1
2
3 sequential optimization method cannot always reach a true optimum. More importantly, for
4
5
6 total-site heat integration, the assumption of non-cooperative game may not be valid.
7
8 Instead of the non-cooperative game, Hiete et al.13 first treated the above benefit
9
10 distribution issue as a cooperative game and planned the required interplant heat exchanges
11
12
13
based on heuristic judgments. Tan et al.14 presented a LP model based on cooperative game
14
15 theory for optimal distribution of cost savings in eco-industial park. On the other hand, Jin et
16
17 al.1 developed a rigorous MINLP-based procedure to handle this allocation problem in the
18
19
spirit of a cooperative game. Their approach is implemented basically in two stages. The
20
21
22 minimum total annual cost (TAC) of each and every potential coalition was first determined
23
24 with a modified version of the conventional MINLP model,8 while the benefit allocation issue
25
26 is addressed in the second stage on the basis of the risk-based Shapley values. An effective
27
28
29 cost sharing scheme is constructed in this second stage according to the core solution of a
30
31 cooperative game15,16 and the risk-based Shapley values of all players.17,18 The former
32
33 ensures coalition stability, while the latter yields a reasonable cost distribution plan.
34
35
36
Although the above methods have been successfully applied to resolve the benefit
37
38 allocation issue for total site heat integration, all of them focused on the grass-root designs
39
40 only. As indicated in the beginning of this section, the benefit allocation problem occurs
41
42
primarily when a multi-plant HEN retrofit project is contemplated for enhancing overall
43
44
45 energy efficiency of an existing industrial park. Ciric et al.19,20 developed a model-based
46
47 method for determining the optimum retrofit design of existing HEN configuration in a single
48
49 plant. Yee and Grossmann21 proposed a superstructure-based MINLP model to handle the
50
51
52 retrofit designs of the single-plant HENs. Sorsak and Kravanja22 described a MINLP model
53
54 for the retrofit of single-plant HENs that comprise different exchanger types. Ponce-Ortega et
55
56 al.23 also presented an optimization model based on superstructure to produce the redesigned
57
58
59
60
5
1
2
3 HEN that considers the plant layout. Smith et al.24 applied the network pinch approach in
4
5
6 HEN retrofit design with temperature-dependent thermal properties.
7
8 It can be concluded from the above discussions that, although various reliable single-
9
10 plant HEN retrofit methods have already been proposed, the model-based simultaneous
11
12
13
multi-plant HEN retrofit design method and the corresponding benefit allocation scheme
14
15 have never been considered before. Thus, the research objective of present study is to first
16
17 formulate and solve a rigorous MINLP model for revamping and merging the existing HENs
18
19
into a multi-plant design via the conventional simultaneous optimization strategy.8 Since an
20
21
22 existing HEN is assumed to be present in each plant, a modified objective function, i.e., the
23
24 extra TAC saving (TACS), is utilized in the model formulation of the multi-plant counterpart.
25
26 The subsequent allocation approach taken in the present study is basically the same as that
27
28
29 adopted in Jin et al.1 In particular, the task of devising the benefit distribution scheme is
30
31 considered to be analogous to that of engaging in a cooperative game. More specifically, an
32
33 effective benefit allocation plan is stipulated according to the core of a cooperative game and
34
35
36
the Shapley values of all players. Three alternative revamp strategies are devised and
37
38 compared in a simple example in this paper to demonstrate the feasibility of the proposed
39
40 design and allocation methods.
41
42
43 2. Superstructure of Multi-Plant Heat Exchanger Networks
44
45 An example of the multi-plant counterpart of traditional single-plant superstructure8 is
46
47 shown below in Figure 1. This superstructure has been adopted in the present study to address
48
49
50
the needs to incorporate design options for placing interplant matches and for consuming
51
52 external utilities across plant boundaries. For illustration convenience, Figure 1 shows the
53
54 structure used for two fictitious plants (say P1 and P2). Plant P1 is equipped with two hot
55
56
utilities (HPP1 and HOP1), one cold utility (CWP1), one hot stream (H1P1) and one cold stream
57
58
59 (C1P1), while P2 has two hot utilities (HPP2 and HOP2), one cold utility (CWP2), one hot
60
6
1
2
3 stream (H1P2) and one cold stream (C1P2). Notice that the interplant and inner-plant matches
4
5
6 in this superstructure are represented with grey-coloured and uncoloured circles, respectively.
7
8 Since there are a total of 2 hot streams and 2 cold streams in the multi-plant HEN, the number
9
10 of stages (denoted as NOK) of this superstructure is set to be 2. Notice also that, in order to
11
12
13 introduce revamp flexibility, an extra bypass is placed on each process stream in every stage
14
15 and also at the end of stream.
16
17 The main body of model formulations can be produced on the basis of this
18
19
20 superstructure, while additional constraints are imposed to facilitate realization of specific
21
22 revamp strategies. The former set of equations and inequalities are included in Part A of the
23
24 Supporting Information for the sake of illustration brevity.
25
26
27
28
29 Figure 1. An example of superstructure for two-plant HEN synthesis
30
31
32
33
34
3. Revamp Strategies
35
36 Three revamp strategies have been devised in this work to facilitate interplant heat
37
38 integration. Basically, each differs from the others mainly in the reclaimed energy and in the
39
40
capital investments and re-piping costs of the resulting multi-plant HEN structure. The
41
42
43 design guidelines adopted in these three strategies are described in the sequel.
44
45
46
47  Strategy I: Only new interplant matches can be introduced into the revamp design. The
48
49
50 existing exchanger matches located within each plant must be kept unchanged, while the
51
52 existing cooler and heater matches may adopt utilities from any plant in the multi-plant
53
54 HEN. Existing coolers and heaters may not be utilized in the revamp design if the
55
56
57
corresponding cooling and heating duties are not required. The interplant matches
58
59 should be housed in new heat exchangers, coolers or heaters purchased externally. Every
60
7
1
2
3 inner-plant match should be housed in its original unit and, if a larger heat-transfer area
4
5
6 is called for, this unit can be connected with an extra new one in series to fulfil the
7
8 required heat duty (see Figure 2).
9
10  Strategy II: Only new interplant matches are allowed to be introduced into the revamp
11
12
13
design. The existing exchanger matches located within each plant must be kept
14
15 unchanged, while the existing cooler and heater matches may adopt utilities from any
16
17 plant in the multi-plant HEN. Existing coolers and heaters may not be utilized in the
18
19
revamp design if the corresponding cooling and heating duties are not required. The
20
21
22 interplant matches should be housed in new heat exchangers, coolers or heaters
23
24 purchased externally. Every inner-plant match can be housed in either its original heat
25
26 exchanger or another existing one of the same type within the same plant and, if a larger
27
28
29 heat-transfer area is called for, this unit can be connected with an extra new one in series
30
31 to fulfil the required heat duty (see Figure 2).
32
33  Strategy III: Both new inner-plant and new interplant matches can be introduced into
34
35
36
the revamp design, while some of the existing matches may not be utilized after
37
38 retrofitting. Every existing heat exchanger should be kept within the plant where it is
39
40 originally located. Existing coolers and heaters may not be utilized in the revamp design
41
42
if the corresponding cooling and heating duties are not required. Any match in revamp
43
44
45 design can be housed either in a purchased heat exchanger or an existing one and, if a
46
47 larger heat-transfer area is called for in the latter case, this unit can be connected with
48
49 another new heat exchanger in series to fulfil the required heat duty (see Figure 2).
50
51
52
53
54 Figure 2. Connections between an existing heat exchanger and a new one to enlarge the
55
56 heat-transfer area
57
58
59
60
8
1
2
3 To facilitate clearer illustration of the above strategies, their unique features are summarized
4
5
6 in Table S1, which is placed in Part B of the Supporting Information. Since a greater financial
7
8 gain can usually be identified by solving a MINLP with more relaxed constraints, it can be
9
10 expected that Strategy III extracts the most benefit, Strategy I the least and Strategy II yields
11
12
13
a cost saving that lies between the above two. However, it should also be noted that selecting
14
15 an appropriate strategy actually depends upon additional practical issues, e.g., safety concerns,
16
17 spatial limits and operability, etc.
18
19
20
21
22 4. Additional Constraints to Generate Retrofit Designs
23
24 Other than the constraints described in Part A of the Supporting Information, it is
25
26
27
necessary to incorporate additional ones in the programming models for realization of the
28
29 aforementioned revamp strategies. These constraints are outlined below:
30
31 4.1 Extra constraints needed for implementing Strategy I
32
33
In addition to the sets, parameters and variables defined in Part A of the Supporting
34
35
36 Information, more are introduced below to facilitate illustration of the proposed model
37
38 formulation.
39
40
41
42
43 Sets:
44
45 𝐶𝑝𝐻𝑇 Set of cold streams that consume hot utilities in the original HEN of plant 𝑝.
46 𝐶𝐿𝑝 Set of matches between hot streams that consume cold utilities in the original HEN
47 of plant 𝑝 and identical cold utilities from any other plant.
48
49
𝐶𝐿𝑀𝑝 Set of inner-plant matches between hot streams and cold utilities in the original HEN
50 of plant 𝑝.
51 𝐻𝑝𝐶𝐿 Set of hot streams that consume cold utilities in the original HEN of plant 𝑝.
52 𝐻𝑇𝑝 Set of matches between cold streams that consume hot utilities in the original HEN
53 of plant 𝑝 and identical hot utilities from any other plant.
54 𝐻𝑇𝑀𝑝 Set of inner-plant matches between hot utilities and cold streams in the original HEN
55
56 of plant 𝑝.
57 𝑌𝑝 Set of existing inner-plant matches in the original HEN of plant 𝑝.
58
59
60 Parameters:
9
1
2
3 𝐴𝐸𝑋 Heat transfer area of existing heat exchanger 𝑦𝑖𝑝,𝑗𝑝 (𝑦𝑖𝑝,𝑗𝑝 = 1,2,⋯,𝑁𝑖𝑝,𝑗𝑝) used for
𝑦𝑖𝑝,𝑗𝑝
4
5 housing match (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝 in the original HEN of plant 𝑝.
6 𝐴𝐸𝑋
𝑖𝑝,𝑛𝑝 Heat transfer area of existing cooler used for housing match (𝑖𝑝,𝑛𝑝) ∈ 𝐶𝐿𝑀𝑝 in the
7 original HEN of plant 𝑝.
8
𝐴𝐸𝑋
𝑚𝑝,𝑗𝑝 Heat transfer area of existing heater used for housing match (𝑚𝑝,𝑗𝑝) ∈ 𝐻𝑇𝑀𝑝 in
9
10 the original HEN of plant 𝑝.
11 𝐿𝑅𝑅𝐶𝑗𝑝,𝑘 Lower bound for bypass flow fraction of cold stream 𝑗𝑝 ∈ 𝐶𝑝 at stage 𝑘 ∈ 𝑆𝑇.
12 𝐿𝑅𝑅𝐶𝑈𝑗𝑝 Lower bound for bypass flow fraction of cold stream 𝑗𝑝 ∈ 𝐶𝑝 at the end of stream.
13 𝐿𝑅𝑅𝐻𝑖𝑝,𝑘 Lower bound for bypass flow fraction of hot stream 𝑖𝑝 ∈ 𝐻𝑝 at stage 𝑘 ∈ 𝑆𝑇.
14
15 𝐿𝑅𝑅𝐻𝑈𝑖𝑝 Lower bound for bypass flow fraction of hot stream 𝑖𝑝 ∈ 𝐻𝑝 at the end of stream.
16 𝐿𝑋𝑖𝑝,𝑗𝑝 Lower bound for heat-transfer area of the augmented unit for housing match
17 (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝 in the multi-plant HEN.
18 𝐿𝑋𝑖𝑝,𝑛𝑞′ Lower bound for heat-transfer area of the augmented unit for housing cooler
19
20 match (𝑖𝑝,𝑛𝑞′) ∈ 𝐶𝐿𝑝 in the multi-plant HEN.
21 𝐿𝑋𝑚𝑞,𝑗𝑝 Lower bound for heat-transfer area of the augmented unit for housing heater
22 match (𝑚𝑞,𝑗𝑝) ∈ 𝐻𝑇𝑝 in the multi-plant HEN.
23 𝑁𝑖𝑝,𝑗𝑝 Total number of existing heat exchangers used for housing match (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝 in
24
25 the original HEN of plant 𝑝.
26 Λ𝑖𝑝,𝑗𝑝 A large enough constant which is not smaller than the largest heat-transfer area of
27 the augmented unit for housing match (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝 in the multi-plant HEN.
28 Λ𝑖𝑝,𝑛𝑞′ A large enough constant which is not smaller than the largest heat-transfer area of
29 the augmented unit for housing cooler match (𝑖𝑝,𝑛𝑞′) ∈ 𝐶𝐿𝑝 in the multi-plant
30
31 HEN.
32 𝛬𝑚𝑞,𝑗𝑝 A large enough constant which is not smaller than the largest heat-transfer area of
33 the augmented unit for housing heater match (𝑚𝑞,𝑗𝑝) ∈ 𝐻𝑇𝑝 in the multi-plant
34 HEN.
35
36
37
38 Variables:
39
40 𝐴𝑖𝑝,𝑗𝑝,𝑘 Heat-transfer area of match (𝑖𝑝,𝑗𝑝) at stage 𝑘 in the multi-plant HEN.
41 𝐴𝑖𝑝,𝑛𝑞′ Heat-transfer area of cooler match (𝑖𝑝,𝑛𝑞′) in the multi-plant HEN.
42 𝐴𝑚𝑞,𝑗𝑝
43
Heat-transfer area of heater match (𝑚𝑞,𝑗𝑝) in the multi-plant HEN.
𝑦𝑖
44 ,𝑗
𝑒𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 Binary variable used for determining whether or not the existing heat exchanger 𝑦𝑖𝑝,𝑗𝑝
45 (𝑦𝑖𝑝,𝑗𝑝 = 1,2,⋯,𝑁𝑖𝑝,𝑗𝑝) adopted for housing match (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝 in the original HEN of
46
47
plant 𝑝 can be used to house the same match at stage k of the multi-plant HEN.
48 𝑟𝑧𝑖𝑝,𝑘 Binary variable used for determining whether or not hot stream 𝑖𝑝 ∈ 𝐻𝑝 requires a
49 bypass stream at stage k of the multi-plant HEN.
50 𝑟𝑧𝑗𝑝,𝑘 Binary variable used for determining whether or not cold stream 𝑗𝑝 ∈ 𝐶𝑝 requires a
51
bypass stream at stage k of the multi-plant HEN.
52
𝑟𝑧𝑐𝑢𝑗𝑝 Binary variable used for determining whether or not cold stream 𝑗𝑝 ∈ 𝐶𝑝 requires a
53
54 bypass stream for its heater in the multi-plant HEN.
55 𝑟𝑧ℎ𝑢𝑖𝑝 Binary variable used for determining whether or not hot stream 𝑖𝑝 ∈ 𝐻𝑝 requires a
56 bypass stream for its cooler in the multi-plant HEN.
57 𝑦𝑖
58
,𝑗
𝑋𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 Heat-transfer area of the augmented unit of the existing heat exchanger 𝑦𝑖𝑝,𝑗𝑝 (𝑦𝑖𝑝,𝑗𝑝
59 = 1,2,⋯,𝑁𝑖𝑝,𝑗𝑝) adopted for housing match (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝 at stage k of the multi-plant
60 HEN.
10
1
2
3 𝑋𝑖𝑝,𝑛𝑞′ Heat-transfer area of the augmented unit of the existing cooler for housing match
4
5
(𝑖𝑝,𝑛𝑞′) ∈ 𝐶𝐿𝑝 in the multi-plant HEN.
6 𝑋𝑚𝑞,𝑗𝑝 Heat-transfer area of the augmented unit of the existing cooler for housing match
7 (𝑚𝑞,𝑗𝑝) ∈ 𝐻𝑇𝑝 in the multi-plant HEN.
8 𝑦𝑖 ,𝑗 Binary variable used for determining whether or not the existing heat exchanger 𝑦𝑖𝑝,𝑗𝑝
9
𝜎𝑖𝑝,𝑗𝑝 𝑝,𝑘𝑝
10 (𝑦𝑖𝑝,𝑗𝑝 = 1,2,⋯,𝑁𝑖𝑝,𝑗𝑝) adopted for housing match (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝 in the original HEN of
11 plant 𝑝 can be used to house the same match at stage k of the multi-plant HEN by
12 enlarging its heat-transfer area according to Figure 2.
13 𝜎𝑖𝑝,𝑛𝑞′ Binary variable used for determining whether or not the existing cooler for housing
14
match (𝑖𝑝,𝑛𝑞′) ∈ 𝐶𝐿𝑝 in the original HEN of plant 𝑝 can be used to house the same
15
16 match in the multi-plant HEN by enlarging its heat-transfer area according to Figure
17 2.
18 𝜎𝑚𝑞,𝑗𝑝 Binary variable used for determining whether or not the existing heater for housing
19 match (𝑚𝑞,𝑗𝑝) ∈ 𝐻𝑇𝑝 in the original HEN of plant 𝑝 can be used to house the same
20
match in the multi-plant HEN by enlarging its heat transfer area according to Figure
21
22 2.
23
24
25 Other than all the constraints mentioned in Part A of the Supporting Information, extra
26
27 ones should be included for implementation of Strategy I. These model constraints are given
28
29
30 in the sequel:
31
32 Firstly, to house the existing matches in their original heat exchangers, the following
33
34 constraints can be imposed:
35
36
37
38
∑𝜉
𝑘 ∈ 𝑆𝑇
𝑖𝑝,𝑗𝑝,𝑘 ≤ 𝑁𝑖𝑝,𝑗𝑝 ; 𝑖𝑝 ∈ 𝐻𝑝; 𝑗𝑝 ∈ 𝐶𝑝; (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝. (1)
39
40 𝑃
41
42
𝜉𝐶𝑈 𝑝
𝑖𝑝,𝑛𝑞′ ≤ 1; 𝑖𝑝 ∈ 𝐻 ; 𝑛𝑞′ ∈ ⋃ 𝐶𝑈 ; (𝑖 ,𝑛 ) ∈ 𝐶𝐿 .
𝑞′ = 1
𝑞′
𝑝 𝑞′ 𝑝 (2)
43
44
𝑃
45
46
47
𝜉𝐻𝑈
𝑚𝑞,𝑗𝑝 ≤ 1; 𝑚𝑞 ∈ ⋃𝐻𝑈 ; 𝑗
𝑞=1
𝑞
𝑝 ∈ 𝐶𝑝; (𝑚𝑞,𝑗𝑝) ∈ 𝐻𝑇𝑝. (3)
48
49
50
where, 𝑝 = 1,2,⋯,𝑃. Note also that the binary variables 𝜉𝑖𝑝,𝑗𝑝,𝑘, 𝜉𝐶𝑈 𝐻𝑈
𝑖𝑝,𝑛𝑝 and 𝜉𝑚𝑝,𝑗𝑝 should be set
51
52 to zero if they are not associated with the existing matches, i.e.,
53
54 𝜉𝑖𝑝,𝑗𝑝,𝑘 = 0; 𝑖𝑝 ∈ 𝐻𝑝; 𝑗𝑝 ∈ 𝐶𝑝; (𝑖𝑝,𝑗𝑝) ∉ 𝑌𝑝. (4)
55
56
57 𝜉𝐶𝑈 𝑝 𝑝
𝑖𝑝,𝑛𝑝 = 0; 𝑖𝑝 ∈ 𝐻 ; 𝑛𝑝 ∈ 𝐶𝑈 ; (𝑖𝑝,𝑛𝑝) ∉ 𝐶𝐿𝑝. (5)
58
59
60
11
1
2
3 𝜉𝐻𝑈 𝑝 𝑝
𝑚𝑝,𝑗𝑝 = 0; 𝑚𝑝 ∈ 𝐻𝑈 ; 𝑗𝑝 ∈ 𝐶 ; (𝑚𝑝,𝑗𝑝) ∉ 𝐻𝑇𝑝. (6)
4
5
6 Exactly one existing heat exchanger (which is originally used to house an existing match in
7
8 the single-plant HEN) can be used to house the same match in a distinct stage in the multi-
9
10
plant HEN, i.e.
11
12 𝑁𝑖𝑝,𝑗𝑝
13
14
15
∑
𝑦𝑖𝑝,𝑗𝑝 = 1
𝑦𝑖
𝑒𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 ― 𝜉𝑖𝑝,𝑗𝑝,𝑘 = 0; 𝑖𝑝 ∈ 𝐻𝑝; 𝑗𝑝 ∈ 𝐶𝑝; (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝; 𝑘 ∈ 𝑆𝑇.
,𝑗
(7)
16
17
18
19
∑𝑒
𝑘 ∈ 𝑆𝑇
𝑖𝑝,𝑗𝑝,𝑘 = 1; 𝑖𝑝 ∈ 𝐻𝑝; 𝑗𝑝 ∈ 𝐶𝑝; (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝; 𝑦𝑖𝑝,𝑗𝑝 = 1,2,⋯,𝑁𝑖𝑝,𝑗𝑝. (8)
20
21
22 Also, the heat-transfer areas of the units in the multi-plant HEN should be constrained as
23
24 follows:
25
26 𝑦𝑖
𝐴𝑖𝑝,𝑗𝑝,𝑘 ― 𝐴𝐸𝑋
,𝑗
𝑦𝑖𝑝,𝑗𝑝𝑒𝑖𝑝,𝑗𝑝,𝑘 ≥ 0;
𝑝 𝑝
27
28 (9)
29 𝑝 𝑝
30 𝑖𝑝 ∈ 𝐻 ; 𝑗𝑝 ∈ 𝐶 ; (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝; 𝑘 ∈ 𝑆𝑇; 𝑦𝑖𝑝,𝑗𝑝 = 1,2,3,…,𝑁𝑖𝑝,𝑗𝑝.
31
32 𝑃
33
34
𝐴𝑖𝑝,𝑛𝑞′ ― 𝐴𝐸𝑋 𝐶𝑈 𝑝
𝑖𝑝,𝑛𝑝𝜉𝑖𝑝,𝑛𝑞′ ≥ 0; 𝑖𝑝 ∈ 𝐻 ; 𝑛𝑞′ ∈ ⋃ 𝐶𝑈 ; (𝑖 ,𝑛 ) ∈ 𝐶𝐿 .
𝑞′ = 1
𝑞′
𝑝 𝑞′ 𝑝 (10)
35
36
𝑃
37
38
39
𝐴𝑚𝑞,𝑗𝑝 ― 𝐴𝐸𝑋 𝐻𝑈
𝑚𝑝,𝑗𝑝𝜉𝑚𝑞,𝑗𝑝 ≥ 0; 𝑚𝑞 ∈ ⋃𝐻𝑈 ; 𝑗
𝑞=1
𝑞
𝑝 ∈ 𝐶𝑝; (𝑚𝑞,𝑗𝑝) ∈ 𝐻𝑇𝑝. (11)
40
41
42
Since an inner-plant match in the multi-plant HEN is housed in an existing unit which is used
43
44 to house the same match in the single-plant HEN and a new heat exchanger may or may not
45
46 𝑦𝑖 ,𝑗
be added according to Figure 2, the heat-transfer area of this augmented unit (𝑋𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘) can be
47
48
49 determined as follows
50
51 𝑦𝑖 𝑦𝑖
𝑋𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 = (𝐴𝑖𝑝,𝑗𝑝,𝑘 ― 𝐴𝐸𝑋
𝑦𝑖𝑝,𝑗𝑝)𝑒𝑖𝑝,𝑗𝑝,𝑘;
,𝑗 𝑝 𝑝 ,𝑗
52
53 (12)
54 𝑝 𝑝
55 𝑖𝑝 ∈ 𝐻 ; 𝑗𝑝 ∈ 𝐶 ; (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝; 𝑘 ∈ 𝑆𝑇; 𝑦𝑖𝑝,𝑗𝑝 = 1,2,3,…,𝑁𝑖𝑝,𝑗𝑝.
56
57
58
59
60
12
1
2
3 On the other hand, the inner-plant coolers and heaters in the multi-plant HEN are always not
4
5
6 smaller than the existing ones due to equations (10) and (11). Their augmented heat-transfer
7
8 areas (𝑋𝑖𝑝,𝑛𝑞′ and 𝑋𝑚𝑞,𝑗𝑝) can be expressed as
9
10 𝑃
(13)
11
12 𝑋𝑖𝑝,𝑛𝑞′ = (𝐴𝑖𝑝,𝑛𝑞′ ― 𝐴𝐸𝑋
𝑖𝑝,𝑛𝑝)𝜉𝑖𝑝,𝑛𝑞′; 𝑖𝑝 ∈ 𝐻 ; 𝑛𝑞′ ∈
𝐶𝑈 𝑝
⋃ 𝐶𝑈𝑞′; (𝑖𝑝,𝑛𝑞′) ∈ 𝐶𝐿𝑝.
13 𝑞′ = 1
14
𝑃
15 (14)
16
17
𝑋𝑚𝑞,𝑗𝑝 = (𝐴𝑚𝑞,𝑗𝑝 ― 𝐴𝐸𝑋
𝑚𝑝,𝑗𝑝) 𝜉𝐻𝑈
𝑚𝑞,𝑗𝑝; 𝑚𝑞 ∈ ⋃𝐻𝑈 ; 𝑗
𝑞=1
𝑞
𝑝
𝑝
∈ 𝐶 ; (𝑚𝑞,𝑗𝑝) ∈ 𝐻𝑇𝑝.
18
19
20 To facilitate calculation of the capital cost of augmented unit for each exsiting match in the
21
22 multi-plant HEN, the following logic constraints must be imposed:
23
24 𝑦𝑖 ,𝑗 𝑦𝑖 ,𝑗
(15)
25
𝑋𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 ― Λ 𝜎 𝑝𝑝 ≤ 0;
𝑖𝑝,𝑗𝑝 𝑖𝑝,𝑗𝑝,𝑘
26
27
𝑖𝑝 ∈ 𝐻𝑝; 𝑗𝑝 ∈ 𝐶𝑝; (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝; 𝑘 ∈ 𝑆𝑇; 𝑦𝑖𝑝,𝑗𝑝 = 1,2,3,…,𝑁𝑖𝑝,𝑗𝑝.
28
29
𝑃
30 (16)
31
32
𝑋𝑖𝑝,𝑛𝑞′ ― 𝛬𝑖𝑝,𝑛𝑞′𝜎𝑖𝑝,𝑛𝑞′ ≤ 0; 𝑖𝑝 ∈ 𝐻 ; 𝑛𝑞′ ∈ 𝑝
⋃
𝑞′ = 1
𝐶𝑈𝑞′; (𝑖𝑝,𝑛𝑞′) ∈ 𝐶𝐿𝑝.
33
34 𝑃
(17)
35
36 𝑋𝑚𝑞,𝑗𝑝 ― 𝛬𝑚𝑞,𝑗𝑝𝜎𝑚𝑞,𝑗𝑝 ≤ 0; 𝑚𝑞 ∈ ⋃ 𝐻𝑈𝑞; 𝑗𝑝 ∈ 𝐶𝑝; (𝑚𝑞,𝑗𝑝) ∈ 𝐻𝑇𝑝.
37 𝑞=1
38
39 Similarly, to facilitate calculation of the capital cost of bypasses in the multi-plant HEN, the
40
41 following logic constraints must also be imposed:
42
43
44 𝑟𝑟ℎ𝑖𝑝,𝑘 ― 𝑟𝑧𝑖𝑝,𝑘 ≤ 0; 𝑖𝑝 ∈ 𝐻𝑝; 𝑘 ∈ 𝑆𝑇. (18)
45
46 𝑟𝑟𝑐𝑗𝑝,𝑘 ― 𝑟𝑧𝑗𝑝,𝑘 ≤ 0; 𝑗𝑝 ∈ 𝐶𝑝; 𝑘 ∈ 𝑆𝑇. (19)
47
48
49 𝑟𝑟ℎ𝑢𝑖𝑝 ― 𝑟𝑧ℎ𝑢𝑖𝑝 ≤ 0; 𝑖𝑝 ∈ 𝐻𝑝. (20)
50
51 𝑟𝑟𝑐𝑢𝑗𝑝 ― 𝑟𝑧𝑐𝑢𝑗𝑝 ≤ 0; 𝑗𝑝 ∈ 𝐶𝑝. (21)
52
53
54 In addition, to avoid impractically small heat-transfer areas of the augmented units, the
55
56 following constraints should be incorporated:
57
59 𝑋𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 ― 𝐿𝑋𝑖𝑝,𝑗𝑝𝜎𝑖𝑝,𝑗𝑝 𝑝,𝑘𝑝 ≥ 0; (22)
60
13
1
2
3 𝑖𝑝 ∈ 𝐻𝑝; 𝑗𝑝 ∈ 𝐶𝑝; (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝; 𝑘 ∈ 𝑆𝑇; 𝑦𝑖𝑝,𝑗𝑝 = 1,2,3,…,𝑁𝑖𝑝,𝑗𝑝.
4
5
𝑃
6
7
8
𝑝
𝑋𝑖𝑝,𝑛𝑞′ ― 𝐿𝑋𝑖𝑝,𝑛𝑞′𝜎𝑖𝑝,𝑛𝑞′ ≥ 0; 𝑖𝑝 ∈ 𝐻 ; 𝑛𝑞′ ∈ ⋃ 𝐶𝑈 ; (𝑖 ,𝑛 ) ∈ 𝐶𝐿 .
𝑞′ = 1
𝑞′
𝑝 𝑞′ 𝑝 (23)
9
10 𝑃
11
12 𝑋𝑚𝑞,𝑗𝑝 ― 𝐿𝑋𝑚𝑞,𝑗𝑝𝜎𝑚𝑞,𝑗𝑝 ≥ 0; 𝑚𝑞 ∈ ⋃𝐻𝑈 ; 𝑗 𝑞
𝑝 ∈ 𝐶𝑝; (𝑚𝑞,𝑗𝑝) ∈ 𝐻𝑇𝑝. (24)
13 𝑞=1
14
15 The impractically small bypass flow fraction of every bypass should also be prohibited as
16
17 follows:
18
19
20 𝑟𝑟ℎ𝑖𝑝,𝑘 ― 𝐿𝑅𝑅𝐻𝑖𝑝,𝑘𝑟𝑧𝑖𝑝,𝑘 ≥ 0; 𝑖𝑝 ∈ 𝐻𝑝; 𝑘 ∈ 𝑆𝑇. (25)
21
22 𝑟𝑟𝑐𝑗𝑝,𝑘 ― 𝐿𝑅𝑅𝐶𝑗𝑝,𝑘𝑟𝑧𝑗𝑝,𝑘 ≥ 0; 𝑗𝑝 ∈ 𝐶𝑝; 𝑘 ∈ 𝑆𝑇. (26)
23
24
25 𝑟𝑟ℎ𝑢𝑖𝑝 ― 𝐿𝑅𝑅𝐻𝑈𝑖𝑝𝑟𝑧ℎ𝑢𝑖𝑝 ≥ 0; 𝑖𝑝 ∈ 𝐻𝑝. (27)
26
27 𝑟𝑟𝑐𝑢𝑗𝑝 ― 𝐿𝑅𝑅𝐶𝑈𝑗𝑝𝑟𝑧𝑐𝑢𝑗𝑝 ≥ 0; 𝑗𝑝 ∈ 𝐶𝑝. (28)
28
29
30
31
32 4.2 Extra constraints needed for implementing Strategy II
33
34 In addition to the sets, parameters and variables defined in Part A of the Supporting
35
36
37
Information and in subsection 4.1, more should be introduced below to facilitate illustration
38
39 of the proposed model formulation for implementing Strategy II.
40
41
42
43
Sets:
44
45
46 𝐶𝑝𝐻𝑇𝑙𝑝 Set of cold streams that consume hot utilities of type 𝑙𝑝 in the original HEN of plant
47 𝑝.
48 𝐶𝐿𝑤 𝑝 Set of existing cooler matches (𝑖𝑝,𝑛𝑝) of type 𝑤𝑝 in plant 𝑝.
49 𝑝
50 𝐶𝑈𝑤𝑝 Set of cold utilities with type 𝑤𝑝 in plant 𝑝.
51 𝐻𝑝𝐶𝐿𝑤𝑝 Set of hot streams that consume cold utilities of type 𝑤𝑝 in the original HEN of plant
52 𝑝.
53 𝑙𝑝
54
𝐻𝑇𝑝 Set of existing heater matches (𝑚𝑝,𝑗𝑝) of type 𝑙𝑝 in plant 𝑝.
55 𝐻𝑈𝑙𝑝 Set of hot utilities with type 𝑙𝑝 in plant 𝑝.
56
57
58
59
60
14
1
2
3 Parameters:
4
5
6
𝐿𝑝 Total number of heater types in plant 𝑝.
7 𝑁𝑝 Total number of existing heat exchangers in the original HEN of plant 𝑝, i.e., 𝑁𝑝 =
8 ∑(𝑖 ,𝑗 ) ∈ 𝑌𝑝𝑁𝑖𝑝,𝑗𝑝.
9 𝑝 𝑝
10 𝑁𝑤𝑝
𝐶𝐿
Total number of existing coolers of type 𝑤𝑝 in the original HEN of plant 𝑝
11
12 𝑁𝑙𝐻𝑇
𝑝 Total number of existing heaters of type 𝑙𝑝 in the original HEN of plant 𝑝
13
14 𝑊𝑝 Total number of cooler types in plant 𝑝.
15
16
17
18
19 Variables:
20
21 𝑦𝑖 ,𝑗
𝑒𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 Binary variable used for determining whether or not the existing heat exchanger
22 𝑦𝑖𝑝,𝑗𝑝 (𝑦𝑖𝑝,𝑗𝑝 = 1,2,⋯,𝑁𝑖𝑝,𝑗𝑝) adopted for housing match (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝 in the original
23
24 HEN of plant 𝑝 can be used to house another match (𝑖𝑝,𝑗𝑝 ) ∈ 𝑌𝑝 at stage 𝑘 of the
25 multi-plant HEN.
26 𝑒𝑖𝑖𝑝𝑝,𝑛𝑝,𝑤𝑝
,𝑛𝑞
Binary variable used for determining whether or not an existing cooler of type 𝑤𝑝,
27 ′
i.e., (𝑖𝑝,𝑛𝑝) ∈ 𝐶𝐿𝑤 𝑝 ,
𝑝
in the original HEN of plant 𝑝 can be used to house another
28
29 cooler match (𝑖𝑝,𝑛𝑞′) in the multi-plant HEN.
30 𝑒𝑚 𝑝,𝑗𝑝,𝑙𝑝 Binary variable used for determining whether or not an existing heater of type 𝑙𝑝,
𝑚𝑞,𝑗𝑝
31
32
i.e., (𝑚𝑝,𝑗𝑝) ∈ 𝐻𝑇𝑙𝑝𝑝, in the original HEN of plant 𝑝 can be used to house another
33 heater match (𝑚𝑞,𝑗𝑝) in the multi-plant HEN.
34 𝑦𝑖 ,𝑗 Heat-transfer area of the augmented unit of the existing heat exchanger 𝑦𝑖𝑝,𝑗𝑝 (𝑦𝑖𝑝,𝑗𝑝
35 𝑋𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘
36 = 1,2,⋯,𝑁𝑖𝑝,𝑗𝑝) of match (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝 in the original single-plant HEN for housing
37 another match (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝 at stage 𝑘 of the multi-plant HEN.
38
𝑋𝑖𝑖𝑝𝑝,𝑛𝑝,𝑤𝑝 Heat-transfer area of the augmented unit of the existing cooler of type 𝑤𝑝, i.e.,
39 ,𝑛𝑞 ′
40 (𝑖𝑝,𝑛𝑝) ∈ 𝐶𝐿𝑤𝑝 𝑝, in the original HEN of plant 𝑝 for housing another cooler match (𝑖𝑝,
41 𝑛𝑞′) in the multi-plant HEN.
42
43
𝑋𝑚 𝑝,𝑗𝑝,𝑙𝑝
𝑚𝑞,𝑗𝑝
Heat-transfer area of the augmented unit of the existing heater of type 𝑙𝑝, i.e.,
44 (𝑚𝑝,𝑗𝑝) ∈ 𝐻𝑇𝑙𝑝𝑝, in the original HEN of plant 𝑝 for housing another heater match (
45 𝑚𝑞,𝑗𝑝) in the multi-plant HEN.
46 𝑦𝑖
47
,𝑗
𝜎𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 Binary variable used for determining whether or not the existing heat exchanger
49 HEN of plant 𝑝 can be used to house another match (𝑖𝑝,𝑗𝑝 ) ∈ 𝑌𝑝 at stage k of the
50 multi-plant HEN by enlarging its heat transfer area according to Figure 2.
51
52 𝜎𝑖𝑖𝑝𝑝,𝑛𝑝,𝑤𝑝
,𝑛𝑞
Binary variable used for determining whether or not the existing cooler of type 𝑤𝑝,
′
53 i.e., (𝑖𝑝,𝑛𝑝) ∈ 𝐶𝐿𝑤 𝑝
𝑝
, in the original HEN of plant 𝑝 can be used to house another
54 cooler match (𝑖𝑝,𝑛𝑞′) in the multi-plant HEN by enlarging its heat transfer area
55
56 according to Figure 2.
57 𝜎𝑚 𝑝,𝑗𝑝,𝑙𝑝
𝑚𝑞,𝑗𝑝
Binary variable used for determining whether or not the existing heater of type 𝑙𝑝,
58 i.e., (𝑚𝑝,𝑗𝑝) ∈ 𝐻𝑇𝑙𝑝𝑝, in the original HEN of plant 𝑝 can be used to house another
59
60 heater match (𝑚𝑞,𝑗𝑝) in the multi-plant HEN by enlarging its heat transfer area
15
1
2
4
5
6
7 Together with all the constraints mentioned in Part A of the Supporting Information,
8
9 additional ones should be included for implementing Strategy II. The needed constraints are
10
11
described below:
12
13
14 First of all, for housing the existing exchanger matches of the multi-plant HEN with
15
16 the available heat exchangers, the following inequality should be imposed:
17
18
19 ∑ ∑𝜉 𝑖𝑝,𝑗𝑝,𝑘 ≤ 𝑁𝑝 (29)
20 (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝𝑘 ∈ 𝑆𝑇
21
22 where, 𝑝 = 1,2,⋯,𝑃. On the other hand, for housing the existing cooler and heater matches of
23
24
25 multi-plant HEN with the available coolers and heaters of the same types, the following
26
27 constraints should be used:
28
29 𝑃
30
31
∑ ∑ ∑
𝑞′ = 1𝑖𝑝 ∈ 𝐻𝑝 𝑤𝑝𝑛 ∈ 𝐶𝑈𝑤𝑞′
𝜉𝐶𝑈 𝑤𝑝
𝑖𝑝,𝑛𝑞′ ≤ 𝑁𝐶𝐿𝑝; 𝑤𝑝 = 𝑤𝑞′ = 1,2,⋯,𝑊𝑝. (30)
32 𝐶𝐿 𝑞′
33
34 𝑃
35
36 ∑ ∑ ∑
𝑞 = 1𝑚 ∈ 𝐻𝑈𝑙𝑞𝑗𝑝 ∈ 𝐶𝑝 𝑙𝑝
𝜉𝐻𝑈 𝑙𝑝
𝑚𝑞,𝑗𝑝 ≤ 𝑁𝐻𝑇𝑝; 𝑙𝑝 = 𝑙𝑞 = 1,2,⋯,𝐿𝑝. (31)
37 𝑞 𝐻𝑇
38
39 In addition, equations (4) – (6) are still valid in the present case. Since every existing match
40
41
42 can be housed in either its original heat exchanger or another existing one of the same type
43
44 within the same plant, the corresponding constraints should be written as
45
47
48 ∑ ∑ 𝑦𝑖
𝑒𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 ― 𝜉𝑖𝑝,𝑗𝑝,𝑘 = 0; 𝑖𝑝 ∈ 𝐻𝑝; 𝑗𝑝 ∈ 𝐶𝑝; (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝; 𝑘 ∈ 𝑆𝑇.
,𝑗
(32)
49 (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝𝑦𝑖𝑝,𝑗𝑝 =1
50
51
52 ∑ 𝑒𝑖𝑖𝑝𝑝,𝑛𝑝,𝑤𝑝
,𝑛𝑞
′
― 𝜉𝐶𝑈
𝑖𝑝,𝑛𝑞′ = 0;
53 𝑤
(𝑖𝑝,𝑛𝑝) ∈ 𝐶𝐿𝑝 𝑝
54 (33)
55
56 𝑖𝑝 ∈ 𝐻𝑝𝐶𝐿𝑤𝑝; 𝑛𝑞′ ∈ 𝐶𝑈𝑤𝑞′; 𝑤𝑞′ = 𝑤𝑝 = 1,2,…,𝑊𝑝; 𝑞′ = 1,2,…,𝑃.
57
58
59
60
16
1
2
3
4 ∑ 𝑒𝑚 𝑝,𝑗𝑝,𝑙𝑝 𝐻𝑈
𝑚𝑞,𝑗𝑝 ― 𝜉𝑚𝑞,𝑗𝑝 = 0;
5 (𝑚𝑝,𝑗𝑝) ∈ 𝐻𝑇𝑝𝑝
𝑙
6 (34)
7
8 𝑚𝑞 ∈ 𝐻𝑈𝑙𝑞; 𝑗𝑝 ∈ 𝐶𝑝𝐻𝑇𝑙𝑝; 𝑙𝑞 = 𝑙𝑝 = 1,2,⋯,𝐿𝑝; 𝑞 = 1,2,…,𝑃.
9
10
Furthermore, each existing heat exchanger should be used to house exactly one match in the
11
12
13 multi-plant HEN, i.e.
14
15
16 ∑ ∑𝑒 𝑦𝑖𝑝,𝑗𝑝
𝑖𝑝,𝑗𝑝,𝑘 = 1; 𝑖𝑝 ∈ 𝐻𝑝; 𝑗𝑝 ∈ 𝐶𝑝; (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝; 𝑦𝑖𝑝,𝑗𝑝 = 1,2,⋯,𝑁𝑖𝑝,𝑗𝑝. (35)
17 (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝𝑘 ∈ 𝑆𝑇
18
19 Also, if the existing units can be adopted in the multi-plant HEN, the corresponding heat-
20
21
transfer areas should be constrained as follows:
22
23 𝑦𝑖
,𝑗
24 𝑦𝑖𝑝,𝑗𝑝𝑒𝑖𝑝,𝑗𝑝,𝑘 ≥ 0;
𝑝 𝑝
25 (36)
26
27 𝑖𝑝,𝑖𝑝 ∈ 𝐻 ; 𝑗𝑝,𝑗𝑝 ∈ 𝐶 ; (𝑖𝑝,𝑗𝑝),(𝑖𝑝,𝑗𝑝) ∈ 𝑌 ; 𝑘 ∈ 𝑆𝑇; 𝑦𝑖𝑝,𝑗𝑝 = 1,2,⋯,𝑁𝑖𝑝,𝑗𝑝.
𝑝 𝑝 𝑝
28
29 𝐴𝑖𝑝,𝑛𝑞′ ― 𝐴𝐸𝑋 𝑖𝑝,𝑛𝑝,𝑤𝑝
≥ 0; 𝑖𝑝 ∈ 𝐻𝑝; 𝑛𝑝 ∈ 𝐶𝑈𝑝; (𝑖𝑝,𝑛𝑝) ∈ 𝐶𝐿𝑤
𝑖𝑝,𝑛𝑝𝑒𝑖𝑝,𝑛𝑞 𝑝 ;
𝑝
30 ′
31 (37)
32 𝑤𝑞′
𝑖𝑝 ∈ 𝐻𝑝𝐶𝐿𝑤𝑝; 𝑛𝑞′ ∈ 𝐶𝑈 ; 𝑤𝑞′ = 𝑤𝑝 = 1,2,…,𝑊𝑝; 𝑞′ = 1,2,…,𝑃.
33
34 𝑚𝑝,𝑗𝑝,𝑙𝑝 𝑙𝑝
35 𝐴𝑚𝑞,𝑗𝑝 ― 𝐴𝐸𝑋 𝑝 𝑝
𝑚𝑝,𝑗𝑝𝑒𝑚𝑞,𝑗𝑝 ≥ 0; 𝑚𝑝 ∈ 𝐻𝑈 ; 𝑗𝑝 ∈ 𝐶 ; (𝑚𝑝,𝑗𝑝) ∈ 𝐻𝑇𝑝 ;
36 (38)
37 𝑙𝑞
38 𝑚𝑞 ∈ 𝐻𝑈 ; 𝑗𝑝 ∈ 𝐶𝑝𝐻𝑇𝑙𝑝; 𝑙𝑞 = 𝑙𝑝 = 1,2,⋯,𝐿𝑝; 𝑞 = 1,2,…,𝑃.
39
40 Since an inner-plant match in the multi-plant HEN may be housed in any existing unit and a
41
42
43
new heat exchanger may or may not be added according to Figure 2, the heat-transfer area of
44
𝑦𝑖
45 the augmented units (𝑋𝑖 ,𝑗𝑝 𝑝,𝑘, 𝑋𝑖𝑖𝑝,𝑛𝑝,𝑤𝑝
and 𝑋𝑚 𝑝,𝑗𝑝,𝑙𝑝
,𝑗
,𝑛𝑞 𝑚𝑞,𝑗
) can be determined as following equations:
46 𝑝 𝑝 𝑝 ′ 𝑝
47
𝑦𝑖 𝑦𝑖
𝑦𝑖𝑝,𝑗𝑝)𝑒𝑖𝑝,𝑗𝑝,𝑘;
,𝑗 ,𝑗
48 𝑝 𝑝
49
(39)
50
51 𝑖𝑝,𝑖𝑝 ∈ 𝐻 ; 𝑗𝑝,𝑗𝑝 ∈ 𝐶 ; (𝑖𝑝,𝑗𝑝),(𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝; 𝑘 ∈ 𝑆𝑇; 𝑦𝑖𝑝,𝑗𝑝 = 1,2,⋯,𝑁𝑖𝑝,𝑗𝑝.
𝑝 𝑝
52
53
𝑋𝑖𝑖𝑝𝑝,𝑛𝑝,𝑤𝑝
= (𝐴𝑖𝑝,𝑛𝑞′ ― 𝐴𝐸𝑋
𝑖𝑝,𝑛𝑝)𝑒𝑖𝑝,𝑛𝑞 ; 𝑖𝑝 ∈ 𝐻 ; 𝑛𝑝 ∈ 𝐶𝑈 ; (𝑖𝑝,𝑛𝑝) ∈ 𝐶𝐿𝑝 ;
𝑖𝑝,𝑛𝑝,𝑤𝑝 𝑝 𝑝 𝑤𝑝
54 ,𝑛𝑞
′ ′
55 (40)
56 𝑤𝑞′
𝑖𝑝 ∈ 𝐻𝑝𝐶𝐿𝑤𝑝; 𝑛𝑞′ ∈ 𝐶𝑈 ; 𝑤𝑞′ = 𝑤𝑝 = 1,2,…,𝑊𝑝; 𝑞′ = 1,2,…,𝑃.
57
58
𝑚𝑞,𝑗𝑝 = (𝐴𝑚𝑞,𝑗𝑝 ― 𝐴𝑚𝑝,𝑗𝑝)𝑒𝑚𝑞,𝑗𝑝 ; 𝑚𝑝 ∈ 𝐻𝑈 ; 𝑗𝑝 ∈ 𝐶 ; 𝑚𝑝,𝑗𝑝
𝑋𝑚 𝑝,𝑗𝑝,𝑙𝑝 𝐸𝑋 𝑚𝑝,𝑗𝑝,𝑙𝑝 𝑝 𝑝 ( (41)
59 )
60
17
1
2
3
∈ 𝐻𝑇𝑙𝑝𝑝; 𝑚𝑞 ∈ 𝐻𝑈𝑙𝑞; 𝑗𝑝 ∈ 𝐶𝑝𝐻𝑇𝑙𝑝; 𝑙𝑞 = 𝑙𝑝 = 1,2,⋯,𝐿𝑝; 𝑞 = 1,2,…,𝑃.
4
5
6 To facilitate calculation of the capital cost of augmented unit for each exsiting match in the
7
8 multi-plant HEN, the following logic constraints must be imposed:
9
11 𝑋𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 ― Λ 𝜎𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 ≤ 0;
𝑖𝑝,𝑗𝑝
12 (42)
13
14 𝑖𝑝,𝑖𝑝 ∈ 𝐻𝑝; 𝑗𝑝,𝑗𝑝 ∈ 𝐶𝑝; (𝑖𝑝,𝑗𝑝),(𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝;𝑘 ∈ 𝑆𝑇; 𝑦𝑖𝑝,𝑗𝑝 = 1,2,⋯,𝑁𝑖𝑝,𝑗𝑝.
15
16
― 𝛬𝑖𝑝,𝑛𝑞′𝜎𝑖𝑖𝑝𝑝,𝑛𝑝,𝑤𝑝
≤ 0; 𝑖𝑝 ∈ 𝐻𝑝; 𝑛𝑝 ∈ 𝐶𝑈𝑝; (𝑖𝑝,𝑛𝑝) ∈ 𝐶𝐿𝑤 𝑝
;
17 ,𝑛𝑞′
,𝑛 𝑞′ 𝑝
18 (43)
19 𝑖𝑝 ∈ 𝐻𝑝𝐶𝐿𝑤𝑝; 𝑤𝑞′
𝑛𝑞′ ∈ 𝐶𝑈 ; 𝑤𝑞′ = 𝑤𝑝 = 1,2,…,𝑊𝑝; 𝑞′ = 1,2,…,𝑃.
20
21
22 𝑋𝑚 𝑝,𝑗𝑝,𝑙𝑝 𝑚𝑝,𝑗𝑝,𝑙𝑝 𝑝 𝑝 ( ) 𝑙𝑝
𝑚𝑞,𝑗𝑝 ― 𝛬𝑚𝑞,𝑗𝑝𝜎𝑚𝑞,𝑗𝑝 ≤ 0; 𝑚𝑝 ∈ 𝐻𝑈 ; 𝑗𝑝 ∈ 𝐶 ; 𝑚𝑝,𝑗𝑝 ∈ 𝐻𝑇𝑝 ;
23
24
(44)
𝑙𝑞
26
27 Furthermore, to avoid using impractically small heat-transfer areas of the augmented units,
28
29
30 the following constraints should be incorporated:
31
𝑦𝑖 𝑦𝑖
32 ,𝑗
𝑋𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 ― 𝐿𝑋𝑖𝑝,𝑗𝑝𝜎𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 ≥ 0;
,𝑗
33
34
(45)
35 𝑖𝑝,𝑖𝑝 ∈ 𝐻 ; 𝑗𝑝,𝑗𝑝 ∈ 𝐶 ; (𝑖𝑝,𝑗𝑝),(𝑖𝑝,𝑗𝑝) ∈ 𝑌 ; 𝑘 ∈ 𝑆𝑇; 𝑦𝑖𝑝,𝑗𝑝 = 1,2,⋯,𝑁𝑖𝑝,𝑗𝑝.
𝑝 𝑝 𝑝
36
37
38
,𝑛𝑞 ― 𝐿𝑋𝑖𝑝,𝑛𝑞′𝜎𝑖𝑖𝑝𝑝,𝑛𝑝,𝑤𝑝
,𝑛𝑞 ≥ 0; 𝑖𝑝 ∈ 𝐻𝑝; 𝑛𝑝 ∈ 𝐶𝑈𝑝; (𝑖𝑝,𝑛𝑝) ∈ 𝐶𝐿𝑤
𝑝 ;
𝑝
′ ′
39 (46)
40 𝑖𝑝 ∈ 𝐻𝑝𝐶𝐿𝑤𝑝; 𝑤𝑞′
𝑛𝑞′ ∈ 𝐶𝑈 ; 𝑤𝑞′ = 𝑤𝑝 = 1,2,…,𝑊𝑝; 𝑞′ = 1,2,…,𝑃.
41
42
𝑚𝑞,𝑗𝑝 ― 𝐿𝑋𝑚𝑞,𝑗𝑝𝜎𝑚𝑞,𝑗𝑝 ≥ 0; 𝑚𝑝 ∈ 𝐻𝑈 ; 𝑗𝑝 ∈ 𝐶 ; 𝑚𝑝,𝑗𝑝 ∈ 𝐻𝑇𝑝 ;
44
45
(47)
𝑙𝑞
47
48 Finally, notice that equations (18) - (21) and (25) - (28) should also be included in the present
49
50
51 application.
52
53 4.3 Extra constraints needed for implementing Strategy III
54
55 In addition to the sets, parameters and variables defined in Part A of the Supporting
56
57
58
Information and in subsections 4.1 and 4.2, extra definitions are introduced below to facilitate
59
60 model formulation for implementing Strategy III.
18
1
2
3 Sets:
4
5
6
𝑍𝑝 Set of all possible inner-plant matches between hot stream 𝑖𝑝 ∈ 𝐻𝑝 and cold stream
7 𝑗𝑝 ∈ 𝐶𝑝 in plant 𝑝.
8 𝑍𝐶𝐿𝑤 𝑝 Set of all possible cooler matches between hot stream 𝑖𝑝 ∈ 𝐻𝑝 and cold utility 𝑛𝑝 ∈
9 𝑝
10 𝐶𝑈𝑝 which can be housed in coolers of type 𝑤𝑝.
11 𝑍𝐻𝑇𝑙𝑝𝑝 Set of all possible heater matches between hot utility 𝑚𝑝 ∈ 𝐻𝑝 and cold stream 𝑗𝑝 ∈
12 𝐶𝑝 which can be housed in heaters of type 𝑙𝑝.
13
14
15
16 Variables:
17
𝑦𝑖
18 ,𝑗
𝑒𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 Binary variable used for determining whether or not the existing heat exchanger
20
21
HEN of plant 𝑝 can be used to house another match (𝑖𝑝,𝑗𝑝 ) ∈ 𝑍𝑝 at stage 𝑘 of the
22 multi-plant HEN.
23 𝑒𝑖𝑖𝑝𝑝,𝑛𝑝,𝑤𝑝
,𝑛𝑞
Binary variable used for determining whether or not an existing cooler of type 𝑤𝑝,
24 ′
i.e., (𝑖𝑝,𝑛𝑝) ∈ 𝐶𝐿𝑤 in the original HEN of plant 𝑝 can be used to house another
𝑝 ,
𝑝
25
26 cooler match (𝑖𝑝,𝑛𝑞′) in the multi-plant HEN.
27 𝑒𝑚 𝑝,𝑗𝑝,𝑙𝑝
𝑚𝑞,𝑗𝑝
Binary variable used for determining whether or not an existing heater of type 𝑙𝑝,
28
i.e., (𝑚𝑝,𝑗𝑝) ∈ 𝐻𝑇𝑙𝑝𝑝, in the original HEN of plant 𝑝 can be used to house another
29
30 heater match (𝑚𝑞,𝑗𝑝) in the multi-plant HEN.
31 𝑢𝑖𝑝,𝑗𝑝,𝑘 Binary variable used for determining whether or not a new heat exchanger should
32 be purchased to house match (𝑖𝑝,𝑗𝑝 ) ∈ 𝑍𝑝 at stage 𝑘 of the multi-plant HEN.
33
34 𝑢𝐶𝑈,𝑤
𝑖𝑝,𝑛𝑝
𝑝 Binary variable used for determining whether or not a new cooler of type 𝑤𝑝
35 should be purchased to house cooler match (𝑖𝑝,𝑛𝑝 ) ∈ 𝑍𝐶𝐿𝑤 𝑝
𝑝
in the multi-plant
36
37
HEN.
𝐻𝑈,𝑙𝑝
38 𝑢𝑚 Binary variable used for determining whether or not a new heater of type 𝑙𝑝
𝑝,𝑗𝑝
39 should be purchased to house heater match (𝑚𝑝,𝑗𝑝 ) ∈ 𝑍𝐻𝑇𝑙𝑝𝑝 in the multi-plant
40
41
HEN.
42 𝑣𝐶𝑈,𝑤
𝑖𝑝,𝑛𝑞
𝑝 Binary variable used for determining whether or not a new cooler of type 𝑤𝑝
′
43 should be purchased to house inter plant cooler match (𝑖𝑝,𝑛𝑞′) in the multi-plant
44 HEN.
45
𝑣𝐻𝑈,𝑙𝑝 Binary variable used for determining whether or not a new heater of type 𝑙𝑝
46 𝑚𝑞,𝑗𝑝
47 should be purchased to house inter plant heater match (𝑚𝑞,𝑗𝑝) in the multi-plant
48 HEN.
49 𝑦𝑖 ,𝑗
𝑋𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 Heat-transfer area of the augmented unit for the existing heat exchanger 𝑦𝑖𝑝,𝑗𝑝
50
(𝑦𝑖𝑝,𝑗𝑝 = 1,2,⋯,𝑁𝑖𝑝,𝑗𝑝) of match (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝 in the original single-plant HEN used
51
52 for housing another match (𝑖𝑝,𝑗𝑝) ∈ 𝑍𝑝 at stage 𝑘 in the multi-plant HEN.
53 𝑋𝑖𝑖𝑝𝑝,𝑛𝑝,𝑤𝑝 Heat-transfer area of the augmented unit of the existing cooler of type 𝑤𝑝, i.e.,
,𝑛𝑞
54 ′
55
(𝑖𝑝,𝑛𝑝) ∈ 𝐶𝐿𝑤𝑝 𝑝, in the original HEN of plant 𝑝 for housing another cooler match (
56 𝑖𝑝,𝑛𝑞′) in the multi-plant HEN.
57 𝑚𝑝,𝑗𝑝,𝑙𝑝
𝑋𝑚 Heat-transfer area of the augmented unit of the existing heater of type 𝑙𝑝, i.e.,
58 𝑞,𝑗𝑝
59 (𝑚𝑝,𝑗𝑝) ∈ 𝐻𝑇𝑙𝑝𝑝, in the original HEN of plant 𝑝 for housing another heater match (
60 𝑚𝑞,𝑗𝑝) in the multi-plant HEN.
19
1
2
3 𝑦𝑖 ,𝑗 Binary variable used for determining whether or not the existing heat exchanger
4 𝜎𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘
5
𝑦𝑖𝑝,𝑗𝑝 (𝑦𝑖𝑝,𝑗𝑝 = 1,2,⋯,𝑁𝑖𝑝,𝑗𝑝) originally adopted for housing match (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝 in the
6 single-plant HEN of plant 𝑝 can be used to house another match (𝑖𝑝,𝑗𝑝 ) ∈ 𝑍𝑝 at
7 stage 𝑘 of the multi-plant HEN by enlarging its heat transfer area according to
8 Figure 2.
9
10 𝜎𝑖𝑖𝑝𝑝,𝑛𝑝,𝑤𝑝
,𝑛𝑞
Binary variable used for determining whether or not the existing cooler of type
𝑤𝑝, i.e., (𝑖𝑝,𝑛𝑝) ∈ 𝐶𝐿𝑤
′
11 𝑝
𝑝
, in the original HEN of plant 𝑝 can be used to house
12 another cooler match (𝑖𝑝,𝑛𝑞′) in the multi-plant HEN by enlarging its heat
13
14
transfer area according to Figure 2.
15 𝜎𝑚 𝑝,𝑗𝑝,𝑙𝑝
𝑚𝑞,𝑗𝑝
Binary variable used for determining whether or not the existing heater of type 𝑙𝑝,
16 i.e., (𝑚𝑝,𝑗𝑝) ∈ 𝐻𝑇𝑙𝑝𝑝, in the original HEN of plant 𝑝 can be used to house another
17
heater match (𝑚𝑞,𝑗𝑝) in the multi-plant HEN by enlarging its heat transfer area
18
20
21
22 Other than all the constraints mentioned in Part A of the Supporting Information,
23
24
25
additional ones should be included for implementation of Strategy III. Such constraints are
26
27 given below:
28
29 For housing the exchanger matches of the multi-plant HEN with either available or
30
31
new heat exchangers, the following inequality should be imposed:
32
34
35
36
∑ ∑ 𝑦𝑖 ,𝑗
𝑒𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 + 𝑢𝑖𝑝,𝑗𝑝,𝑘 ― 𝜉𝑖𝑝,𝑗𝑝,𝑘 = 0;
(𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝𝑦𝑖𝑝,𝑗𝑝 = 1 (48)
37
38
39 𝑖𝑝 ∈ 𝐻𝑝; 𝑗𝑝 ∈ 𝐶𝑝; (𝑖𝑝,𝑗𝑝) ∈ 𝑍𝑝; 𝑘 ∈ 𝑆𝑇.
40
41 On the other hand, for housing the cooler and heater matches in multi-plant HEN with either
42
43
44 available or new coolers and heaters of the same types, the following constraints should be
45
46 used:
47
48
,𝑛𝑝 + 𝑢𝐶𝑈,𝑤 𝑝 𝐶𝑈
𝑖𝑝,𝑛𝑝 ― 𝜉𝑖𝑝,𝑛𝑝 = 0;
50 𝑤
51 (49)
52
53 𝑖𝑝 ∈ 𝐻𝑝; 𝑛𝑝 ∈ 𝐶𝑈𝑝; (𝑖𝑝,𝑛𝑝) ∈ 𝑍𝐶𝐿𝑤
𝑝 ; 𝑤𝑝 = 1,2,⋯, 𝑊𝑝.
𝑝
54
55
,𝑛𝑞′
+ 𝑣𝐶𝑈,𝑤
′
𝑝 𝐶𝑈
𝑖𝑝,𝑛𝑞 ― 𝜉𝑖𝑝,𝑛𝑞′ = 0; (50)
57 𝑤
58
59
60
20
1
2
3
𝑖𝑝 ∈ 𝐻𝑝; 𝑛𝑞′ ∈ 𝐶𝑈𝑤𝑞′; 𝑤𝑝 = 𝑤𝑞′ = 1,2,⋯, 𝑊𝑝; 𝑞′ ≠ 𝑝.
4
5
6
7
∑ 𝑙
𝑒𝑚 𝑝,𝑗𝑝,𝑙𝑝 𝐻𝑈,𝑙𝑝 𝐻𝑈
𝑚𝑝,𝑗𝑝 + 𝑢𝑚𝑝,𝑗𝑝 ― 𝜉𝑚𝑝,𝑗𝑝 = 0;
(51)
9
10
𝑚𝑝 ∈ 𝐻𝑈𝑝; 𝑗𝑝 ∈ 𝐶𝑝; (𝑚𝑝,𝑗𝑝) ∈ 𝑍𝐻𝑇𝑙𝑝𝑝; 𝑙𝑝 = 1,2,⋯, 𝐿𝑝.
11
12
13
14 ∑ 𝑙
𝑒𝑚 𝑝,𝑗𝑝,𝑙𝑝 𝐻𝑈 𝐻𝑈
𝑚𝑞,𝑗𝑝 + 𝑣𝑚𝑞,𝑗𝑝 ― 𝜉𝑚𝑞,𝑗𝑝 = 0;
(52)
16
17
18 𝑚𝑞 ∈ 𝐻𝑈𝑙𝑞; 𝑗𝑝 ∈ 𝐶𝑝; 𝑙𝑝 = 𝑙𝑞 = 1,2,⋯,𝐿𝑝; 𝑞 ≠ 𝑝.
19
20 In the multi-plant HEN, the total numbers of heat exchangers, coolers and heaters that are
21
22 housed in existing units should be bounded, i.e.
23
25
26
27
∑ ∑ ∑ ∑𝑒 𝑦𝑖𝑝,𝑗𝑝
𝑖𝑝,𝑗𝑝,𝑘 ≤ 𝑁𝑝. (53)
(𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝𝑦𝑖𝑝,𝑗𝑝 = 1(𝑖𝑝,𝑗𝑝) ∈ 𝑍𝑝𝑘 ∈ 𝑆𝑇
28
29
𝑃
30
31
32
∑ ∑ ∑ ∑
𝑞′ = 1(𝑖 ,𝑛 ) ∈ 𝐶𝐿𝑤𝑝𝑖𝑝 ∈ 𝐻𝑝𝑛 ∈ 𝐶𝑈𝑤𝑞′
𝑒𝑖𝑖𝑝𝑝,𝑛
′
𝑝,𝑤𝑝
,𝑛𝑞 ≤ 𝑁𝑤
𝐶𝐿; 𝑤𝑝 = 𝑤𝑞′ = 1,2,⋯,𝑊𝑝.
𝑝
(54)
𝑝 𝑝 𝑞′
33 𝑝
34
𝑃
35
36
37
∑ ∑ ∑ ∑𝑒
𝑞 = 1(𝑚 ,𝑗 ) ∈ 𝐻𝑇𝑙𝑝𝑚 ∈ 𝐻𝑈𝑙𝑞𝑗𝑝 ∈ 𝐶𝑝
𝑚𝑝,𝑗𝑝,𝑙𝑝
𝑚𝑞,𝑗𝑝 ≤ 𝑁𝑙𝐻𝑇
𝑝
; 𝑙𝑝 = 𝑙𝑞 = 1,2,⋯,𝐿𝑝. (55)
38 𝑝 𝑝 𝑝 𝑞
39
40 Every existing heat exchanger should of course be used to house exactly one exchanger
41
42 match in the multi-plant HEN, i.e.
43
44
45
46
∑ ∑𝑒 𝑦𝑖𝑝,𝑗𝑝
𝑖𝑝,𝑗𝑝,𝑘 = 1; (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝; 𝑦𝑖𝑝,𝑗𝑝 = 1,2,⋯, 𝑁𝑖𝑝,𝑗𝑝. (56)
(𝑖𝑝,𝑗𝑝) ∈ 𝑍𝑝𝑘 ∈ 𝑆𝑇
47
48
49 If the existing units can be adopted in the multi-plant HEN, the corresponding heat-transfer
50
51 areas should be constrained as follows:
52
53 𝑦𝑖
,𝑗 𝑝 𝑝
𝑦𝑖𝑝,𝑗𝑝𝑒𝑖𝑝,𝑗𝑝,𝑘 ≥ 0; 𝑖𝑝,𝑖𝑝 ∈ 𝐻 ; 𝑗𝑝,𝑗𝑝 ∈ 𝐶 ;
𝑝 𝑝
54
55 (57)
56
57
(𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝; (𝑖𝑝,𝑗𝑝) ∈ 𝑍𝑝; 𝑘 ∈ 𝑆𝑇; 𝑦𝑖𝑝,𝑗𝑝 = 1,2,⋯,𝑁𝑖𝑝,𝑗𝑝.
58
59
60
21
1
2
3 𝑖𝑝,𝑛𝑝,𝑤𝑝
𝐴𝑖𝑝,𝑛𝑞′ ― 𝐴𝐸𝑋
𝑖𝑝,𝑛𝑝𝑒𝑖𝑝,𝑛𝑞 ≥ 0; 𝑖𝑝,𝑖𝑝 ∈ 𝐻𝑝; 𝑛𝑝 ∈ 𝐶𝑈𝑝; (𝑖𝑝,𝑛𝑝) ∈ 𝐶𝐿𝑤
𝑝 ;
𝑝
4 ′
5 (58)
6 𝑤𝑞′
7
𝑛𝑞′ ∈ 𝐶𝑈 ; 𝑤𝑞′ = 𝑤𝑝 = 1,2,…,𝑊𝑝; 𝑞′ = 1,2,…,𝑃.
8
𝑚𝑝,𝑗𝑝,𝑙𝑝 𝑙𝑝
9 𝐴𝑚𝑞,𝑗𝑝 ― 𝐴𝐸𝑋 𝑝 𝑝
𝑚𝑝,𝑗𝑝𝑒𝑚𝑞,𝑗𝑝 ≥ 0; 𝑚𝑝 ∈ 𝐻𝑈 ; 𝑗𝑝,𝑗𝑝 ∈ 𝐶 ; (𝑚𝑝,𝑗𝑝) ∈ 𝐻𝑇𝑝 ;
10
(59)
11 𝑙𝑞 𝑝
12 𝑚𝑞 ∈ 𝐻𝑈 ; 𝑗𝑝 ∈ 𝐶 ; 𝑙𝑞 = 𝑙𝑝 = 1,2,⋯,𝐿𝑝; 𝑞 = 1,2,…,𝑃.
13
14 The heat-transfer areas of the augmented units can be expressed as
15
16 𝑦𝑖 𝑦𝑖
𝑦𝑖𝑝,𝑗𝑝)𝑒𝑖𝑝,𝑗𝑝,𝑘; 𝑖𝑝,𝑖𝑝 ∈ 𝐻 ; 𝑗𝑝,𝑗𝑝 ∈ 𝐶 ;
,𝑗 𝑝 𝑝 𝑝 ,𝑗 𝑝
17
18 (60)
19
20
(𝑖𝑝,𝑗𝑝) ∈ 𝑌 ; (𝑖𝑝,𝑗𝑝) ∈ 𝑍 ; 𝑘 ∈ 𝑆𝑇; 𝑦𝑖𝑝,𝑗𝑝 = 1,2,⋯,𝑁𝑖𝑝,𝑗𝑝.
𝑝 𝑝
21
22 𝑋𝑖𝑖𝑝𝑝,𝑛𝑝,𝑤𝑝
,𝑛𝑞 𝑖𝑝,𝑛𝑝)𝑒𝑖𝑝,𝑛𝑞 ; 𝑖𝑝,𝑖𝑝 ∈ 𝐻 ; 𝑛𝑝 ∈ 𝐶𝑈 ;
= (𝐴𝑖𝑝,𝑛𝑞′ ― 𝐴𝐸𝑋 𝑖𝑝,𝑛𝑝,𝑤𝑝 𝑝 𝑝
23 ′ ′
24 (61)
25 (𝑖𝑝,𝑛𝑝) ∈ 𝐶𝐿𝑤
𝑝 ;
𝑝 𝑤𝑞′
𝑛𝑞′ ∈ 𝐶𝑈 ; 𝑤𝑞′ = 𝑤𝑝 = 1,2,…,𝑊𝑝; 𝑞′ = 1,2,…,𝑃.
26
27
28
𝑚𝑝,𝑗𝑝,𝑙𝑝
𝑋𝑚 𝑞,𝑗𝑝
= (𝐴𝑚𝑞,𝑗𝑝 ― 𝐴𝐸𝑋
𝑚𝑝,𝑗𝑝)𝑒𝑚𝑞,𝑗𝑝 ; 𝑚𝑝 ∈ 𝐻𝑈 ; 𝑗𝑝,𝑗𝑝 ∈ 𝐶 ;
𝑚𝑝,𝑗𝑝,𝑙𝑝 𝑝 𝑝
29 (62)
30 (𝑚𝑝,𝑗𝑝) ∈ 𝐻𝑇𝑙𝑝𝑝; 𝑚𝑞 ∈ 𝐻𝑈𝑙𝑞; 𝑙𝑞 = 𝑙𝑝 = 1,2,⋯,𝐿𝑝; 𝑞 = 1,2,…,𝑃.
31
32
33 The presence (or absence) of each augment unit can be determined with the binary variables
34
35 in the following logic constraints:
36
37 𝑦𝑖 ,𝑗 𝑦𝑖
𝜎𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 ≤ 0; 𝑖𝑝,𝑖𝑝 ∈ 𝐻𝑝; 𝑗𝑝,𝑗𝑝 ∈ 𝐶𝑝;
,𝑗
38 𝑋𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 ― Λ
𝑖𝑝,𝑗𝑝
39 (63)
40
41 (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝; (𝑖𝑝,𝑗𝑝) ∈ 𝑍𝑝; 𝑘 ∈ 𝑆𝑇; 𝑦𝑖𝑝,𝑗𝑝 = 1,2,⋯,𝑁𝑖𝑝,𝑗𝑝.
42
43
― 𝛬𝑖𝑝,𝑛𝑞′𝜎𝑖𝑖𝑝𝑝,𝑛𝑝,𝑤𝑝
≤ 0; 𝑖𝑝,𝑖𝑝 ∈ 𝐻𝑝; 𝑛𝑝 ∈ 𝐶𝑈𝑝; (𝑖𝑝,𝑛𝑝) ∈ 𝐶𝐿𝑤
𝑝 ;
𝑝
44 ,𝑛𝑞
′
,𝑛𝑞 ′
45 (64)
46 𝑤𝑞′
𝑛𝑞′ ∈ 𝐶𝑈 ; 𝑤𝑞′ = 𝑤𝑝 = 1,2,…,𝑊𝑝; 𝑞′ = 1,2,…,𝑃.
47
48
𝑚𝑞,𝑗𝑝 ― 𝛬𝑚𝑞,𝑗𝑝𝜎𝑚𝑞,𝑗𝑝 ≤ 0; 𝑚𝑝 ∈ 𝐻𝑈 ; 𝑗𝑝,𝑗𝑝 ∈ 𝐶 ; 𝑚𝑝,𝑗𝑝 ∈ 𝐻𝑇𝑝 ;
50 (65)
51 𝑙𝑞 𝑝
52
𝑚𝑞 ∈ 𝐻𝑈 ; 𝑗𝑝 ∈ 𝐶 ; 𝑙𝑞 = 𝑙𝑝 = 1,2,⋯,𝐿𝑝; 𝑞 = 1,2,…,𝑃.
53
54 To avoid using impractically small heat-transfer areas of the augmented units, the following
55
56 constraints should be incorporated:
57
58
59
60
22
1
2
3 𝑦𝑖 𝑦𝑖
𝑋𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 ― 𝐿𝑋𝑖𝑝,𝑗𝑝𝜎𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 ≥ 0; 𝑖𝑝,𝑖𝑝 ∈ 𝐻𝑝; 𝑗𝑝,𝑗𝑝 ∈ 𝐶𝑝;
,𝑗 ,𝑗
4
5 (66)
6
7 (𝑖𝑝,𝑗𝑝) ∈ 𝑌 ; (𝑖𝑝,𝑗𝑝) ∈ 𝑍 ; 𝑘 ∈ 𝑆𝑇; 𝑦𝑖𝑝,𝑗𝑝 = 1,2,⋯,𝑁𝑖𝑝,𝑗𝑝.
𝑝 𝑝
8
9 𝑋𝑖𝑖𝑝𝑝,𝑛𝑝,𝑤𝑝
― 𝐿𝑋𝑖𝑝,𝑛𝑞′𝜎𝑖𝑖𝑝𝑝,𝑛𝑝,𝑤𝑝
≥ 0; 𝑖𝑝,𝑖𝑝 ∈ 𝐻𝑝; 𝑛𝑝 ∈ 𝐶𝑈𝑝; (𝑖𝑝,𝑛𝑝) ∈ 𝐶𝐿𝑤
𝑝 ;
𝑝
,𝑛𝑞 ,𝑛𝑞
10 ′ ′
11 (67)
12 𝑤𝑞′
𝑛𝑞′ ∈ 𝐶𝑈 ; 𝑤𝑞′ = 𝑤𝑝 = 1,2,…,𝑊𝑝; 𝑞′ = 1,2,…,𝑃.
13
14
𝑚𝑞,𝑗𝑝 ― 𝐿𝑋𝑚𝑞,𝑗𝑝𝜎𝑚𝑞,𝑗𝑝 ≥ 0; 𝑚𝑝 ∈ 𝐻𝑈 ; 𝑗𝑝,𝑗𝑝 ∈ 𝐶 ; 𝑚𝑝,𝑗𝑝 ∈ 𝐻𝑇𝑝 ;
16 (68)
17 𝑙𝑞 𝑝
18 𝑚𝑞 ∈ 𝐻𝑈 ; 𝑗𝑝 ∈ 𝐶 ; 𝑙𝑞 = 𝑙𝑝 = 1,2,⋯,𝐿𝑝; 𝑞 = 1,2,…,𝑃.
19
20 Finally, the presence (or absence) of bypasses can also be determined with the binary
21
22 variables in equations (18) - (21), while the impractically small bypass flow fractions should
23
24
25 be prohibited in equations (25) - (28).
26
27 5. Objective Functions
28
29 In this study, the overall saving achieved by retrofitting and building the multi-plant
30
31
32 HEN is used as the objective function to be maximized. This function (𝑇𝐴𝐶𝑆) can be
33
34 expressed as follows
35
36 𝑇𝐴𝐶𝑆 = 𝑇𝑈𝐶 ― 𝑇𝑈𝐶′ ― 𝑎𝑓(𝐴𝑇𝐶𝐶 + 𝑁𝑇𝐶𝐶1 + 𝑁𝑇𝐶𝐶2 + 𝑁𝑇𝐶𝐶3 + 𝑁𝑇𝐶𝐶4 + 𝑇𝑈𝑃𝐶) (69)
37
38
39 where, 𝑇𝑈𝐶 denotes the sum of utility costs of all single-plant HENs which should be
40
41 regarded as a given constant in the corresponding MINLP models; 𝑇𝑈𝐶′ denotes the total
42
43 utility cost of the multi-plant HEN after retrofit; 𝑎𝑓 is the annualization factor which is
44
45
46 another given constant; 𝐴𝑇𝐶𝐶 is the total capital cost of all augmented units; 𝑁𝑇𝐶𝐶1 denotes
47
48 the total capital cost of all new units purchased for interplant matches; 𝑁𝑇𝐶𝐶2 denotes the
49
50
total capital cost of all bypasses; 𝑁𝑇𝐶𝐶3 denotes the unit reassignment cost; 𝑁𝑇𝐶𝐶4 denotes
51
52
53 the total capital cost of all new units purchased for inner-plant matches; 𝑇𝑈𝑃𝐶 is the total
54
55 capital cost of pipes if the existing coolers and heaters adopt utilities from other plant in the
56
57
58
multi-plant HEN. Other than the aforementioned two constants (i.e., 𝑇𝑈𝐶 and 𝑎𝑓), the
59
60 detailed expressions of the remaining cost models are listed in the subsequent subsections and,
23
1
2
3 for the sake of brevity, all embedded model parameters (or cost coefficients) are first defined
4
5
6 below:
7
8
9
10 Parameters:
11
12 𝐵𝑌𝑖𝑝
13
Cost of a single bypass on hot stream 𝑖𝑝.
14 𝐵𝑌𝑗𝑝 Cost of a single bypass on cold stream 𝑗𝑝.
15 𝐶𝐴𝑖𝑞,𝑗𝑞′ Variable cost coefficient in the cost model of heat exchanger between hot stream
16 𝑖𝑞 and cold stream 𝑗𝑞′ (𝑞,𝑞′ = 1,2,⋯,𝑃) .
17 𝐶𝐴𝑖𝑝,𝑛𝑞′
18
Variable cost coefficient in the cost model of cooler between hot stream 𝑖𝑞 and
19 cold utility 𝑛𝑞′ (𝑞′ = 1,2,⋯,𝑃) .
20 𝐶𝐴𝑚𝑞,𝑗𝑝 Variable cost coefficient in the cost model of heater between hot utility 𝑚𝑞 (𝑞
21
= 1,2,⋯,𝑃) and cold stream 𝑗𝑝.
22
23
𝐶𝐹𝑖𝑞,𝑗𝑞′ Fixed cost in the cost model of heat exchanger between hot stream 𝑖𝑞 and cold
24 stream 𝑗𝑞′ (𝑞,𝑞′ = 1,2,⋯,𝑃)
25 𝐶𝐹𝑖𝑝,𝑛𝑞′ Fixed cost in the cost model of cooler between hot stream 𝑖𝑞 and cold utility 𝑛𝑞′
26
27
(𝑞′ = 1,2,⋯,𝑃) .
28
𝐶𝐹𝑚𝑞,𝑗𝑝 Fixed cost in the cost model of heater between hot utility 𝑚𝑞 (𝑞 = 1,2,⋯,𝑃) and
29 cold stream 𝑗𝑝.
30 𝐶𝑀𝑖𝑖𝑝𝑝,𝑗,𝑗𝑝𝑝 Reassignment cost for existing heat exchanger which houses match (𝑖𝑝,𝑗𝑝) in the
31
32
original HEN of plant 𝑝 and houses different match (𝑖𝑝,𝑗𝑝) in the multi-plant
33 HEN.
34 𝐶𝑀𝑖𝑖𝑝𝑝,𝑛𝑝,𝑤𝑝
,𝑛𝑞
Reassignment cost for existing cooler of type 𝑤𝑝 which houses match (𝑖𝑝,𝑛𝑝) in
35 ′
the original HEN of plant 𝑝 and is utilized by different hot process stream 𝑖𝑝 in
36
37 the multi-plant HEN
38 𝐶𝑀𝑚 𝑝,𝑗𝑝,𝑙𝑝
𝑚𝑞,𝑗𝑝
Reassignment cost for existing heater of type 𝑙𝑝 which houses match (𝑚𝑝,𝑗𝑝) in
39 the original HEN of plant 𝑝 and is utilized by different cold process stream 𝑗𝑝 in
40 the multi-plant HEN
41
42
𝐶𝑃𝑖𝑝,𝑛𝑞′ Fixed cost of pipe for existing cooler which can house interplant match (𝑖𝑝,𝑛𝑞′) in
43 the multi-plant HEN by adopting utilities from other plant
44 𝐶𝑃𝑚𝑞,𝑗𝑝 Fixed cost of pipe for existing heater which can house interplant match (𝑚𝑞,𝑗𝑝) in
45 the multi-plant HEN by adopting utilities from other plant
46 𝐶𝑄𝑖𝑝,𝑛𝑞′ Unit cost of cold utility 𝑛𝑞′ (𝑞′ = 1,2,⋯,𝑃) for cooling hot stream 𝑖𝑝.
47
𝐶𝑄𝑚𝑞,𝑗𝑝 Unit cost of hot utility 𝑚𝑞 (𝑞 = 1,2,⋯,𝑃) for heating cold stream 𝑗𝑝.
48
49 𝛽 Exponent of heater transfer areas in variable cost terms in the cost models of heat
50 exchanger, cooler and heater.
51
52
53
54
5.1 Cost models utilized for implementation of Strategy I
55
56 For Strategy I, the cost items embedded in equation (69) are presented in detailed as
57
58 follows
59
60
24
1
2
3 𝑃 𝑃
4
5
𝑇𝑈𝐶′ = ∑∑ ∑ ∑
𝑝 = 1𝑞′ = 1𝑖𝑝 ∈ 𝐻𝑝𝑛𝑞 ∈ 𝐶𝑈𝑞′
𝐶𝑄𝑖𝑝,𝑛𝑞′𝑞𝑖𝑝,𝑛𝑞′
6 ′
7 (70)
8 𝑃 𝑃
9
10 + ∑ ∑ ∑ ∑ 𝐶𝑄
𝑞 = 1𝑝 = 1𝑚𝑞 ∈ 𝐻𝑈𝑞𝑗𝑝 ∈ 𝐶𝑝
𝑚𝑞,𝑗𝑝𝑞𝑚𝑞,𝑗𝑝
11
12
13 𝑃 𝑁𝑖𝑝,𝑗𝑝
∑ ∑ ∑ ∑[ )]
14 𝛽
15 𝐴𝑇𝐶𝐶 =
𝑦𝑖 ,𝑗
𝐶𝐹𝑖𝑝,𝑗𝑝𝜎𝑖𝑝,𝑗𝑝 𝑝,𝑘𝑝 +𝐶𝐴𝑖𝑝,𝑗𝑝 𝑋𝑖𝑝,𝑗𝑝 𝑝,𝑘𝑝 ( 𝑦𝑖 ,𝑗
16 𝑝 = 1(𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝𝑦𝑖𝑝,𝑗𝑝 = 1𝑘 ∈ 𝑆𝑇
17
18 𝑃 𝑃
19
20
+ ∑∑ ∑ ∑ ∑
𝑝 = 1𝑞′ = 1𝑖𝑝 ∈ 𝐻𝑝𝐶𝐿𝑛𝑞 ∈ 𝐶𝑈𝑞′(𝑖𝑝,𝑛𝑞′) ∈ 𝐶𝐿𝑝
[𝐶𝐹𝑖 ,𝑛 𝜎𝑖 ,𝑛
𝑝 𝑞′ 𝑝 𝑞′
+ 𝐶𝐴𝑖𝑝,𝑛𝑞′ (𝑋𝑖𝑝,𝑛𝑞′)𝛽] (71)
21 ′
22
23 𝑃 𝑃
24
25
+ ∑∑ ∑ ∑ ∑
𝑞 = 1𝑝 = 1𝑚𝑞 ∈ 𝐻𝑈𝑞𝑗𝑝 ∈ 𝐶𝑝𝐻𝑇(𝑚𝑞,𝑗𝑝) ∈ 𝐻𝑇𝑝
[𝐶𝐹𝑚 ,𝑗 𝜎𝑚 ,𝑗 𝑞 𝑝 𝑞 𝑝
+ 𝐶𝐴𝑚𝑞,𝑗𝑝(𝑋𝑚𝑞,𝑗𝑝)𝛽]
26
27
28 𝑃 𝑃
29
30
𝑁𝑇𝐶𝐶1 = ∑ ∑ ∑ ∑ ∑ [𝐶𝐹𝐹
𝑝 = 1𝑞′ = 1𝑖𝑝 ∈ 𝐻𝑝𝑗𝑞 ∈ 𝐶𝑞′𝑘 ∈ 𝑆𝑇
𝑖𝑝,𝑗𝑞′𝜉𝑖𝑝,𝑗𝑞′,𝑘 +𝐶𝐴𝑖𝑝,𝑗𝑞′(𝐴𝑖𝑝,𝑗𝑞′)𝛽]
31 𝑞′ ≠ 𝑝 ′
32
33 𝑃 𝑃
34
35
+ ∑ ∑ ∑ ∑
𝑝=1 𝑞′ = 1𝑖𝑝 ∉ 𝐻𝑝𝐶𝐿𝑛𝑞 ∈ 𝐶𝑈𝑞′
[𝐶𝐹𝐹𝑖 ,𝑛 𝜉𝑖 ,𝑛
𝑝 𝑞′ 𝑝 𝑞′
+ 𝐶𝐴𝑖𝑝,𝑛𝑞′(𝐴𝑖𝑝,𝑛𝑞′)𝛽] (72)
36 ′
𝑞′ ≠ 𝑝
37
38
𝑃 𝑃
39
40
41
+ ∑ ∑ ∑ ∑ [𝐶𝐹𝐹
𝑞 = 1𝑝 = 1𝑚𝑞 ∈ 𝐻𝑈𝑞𝑗𝑝 ∉ 𝐶𝑝𝐻𝑇
𝑚𝑞,𝑗𝑝𝜉𝑚𝑞,𝑗𝑝 + 𝐶𝐴𝑚𝑞,𝑗𝑝(𝐴𝑚𝑞,𝑗𝑝)𝛽]
42 𝑞≠𝑝
43
44 𝑃
45
46
𝑁𝑇𝐶𝐶2 = ∑ ∑ [𝐵𝑌 (𝑟𝑧ℎ𝑢 + ∑ 𝑟𝑧
𝑝 = 1𝑖𝑝 ∈ 𝐻𝑝
𝑖𝑝 𝑖𝑝
𝑘 ∈ 𝑆𝑇
)]
𝑖𝑝,𝑘
47
48 (73)
𝑃
49
50
51
+ ∑ ∑ [𝐵𝑌 (𝑟𝑧𝑐𝑢 + ∑ 𝑟𝑧
𝑝 = 1𝑗𝑝 ∈ 𝐶𝑝
𝑗𝑝 𝑗𝑝
𝑘 ∈ 𝑆𝑇
𝑗𝑝,𝑘)]
52
53
54 𝑁𝑇𝐶𝐶3 = 0 (74)
55
56 𝑁𝑇𝐶𝐶4 = 0 (75)
57
58
59
60
25
1
2
3 𝑃 𝑃
4
5
𝑇𝑈𝑃𝐶 = ∑ ∑ ∑ ∑
𝑝=1 𝑞′ = 1𝑖𝑝 ∈ 𝐻𝑝𝐶𝐿𝑛𝑞 ∈ 𝐶𝑈𝑞′
(𝐶𝑃𝑖𝑝,𝑛𝑞′𝜉𝑖𝑝,𝑛𝑞′)
6 𝑞′ ≠ 𝑝 ′
7
8 (76)
𝑃 𝑃
9
10
11
+ ∑ ∑ ∑ ∑ (𝐶𝑃
𝑞 = 1𝑝 = 1𝑚𝑞 ∈ 𝐻𝑈𝑞𝑗𝑝 ∈ 𝐶𝑝𝐻𝑇
𝑚𝑞,𝑗𝑝𝜉𝑚𝑞,𝑗𝑝 )
12 𝑞≠𝑝
13
14 5.2 Cost models utilized for implementation of Strategy II
15
16
17 The above cost models are all applicable in the present scenario except 𝐴𝑇𝐶𝐶. In
18
19 particular, equation (71) and (74) should be replaced by the following formula:
20
∑ ∑ ∑ ∑ ∑[ )]
22 𝛽
23 𝐴𝑇𝐶𝐶 =
𝑦𝑖 ,𝑗
( 𝑦𝑖
𝐶𝐹𝑖𝑝,𝑗𝑝𝜎𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘 +𝐶𝐴𝑖𝑝,𝑗𝑝 𝑋𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘
,𝑗
24 𝑝 = 1(𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝𝑦𝑖𝑝,𝑗𝑝 = 1(𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝𝑘 ∈ 𝑆𝑇

25
26 𝑃 𝑃 𝑊𝑝
27
28
+ ∑∑ ∑ ∑
𝑝 = 1𝑞′ = 1𝑤𝑝 = 𝑤𝑞′ = 1(𝑖 ,𝑛 ) ∈ 𝐶𝐿𝑤𝑝𝑖𝑝 ∈ 𝐻𝑝 𝑤𝑝𝑛 ∈ 𝐶𝑈𝑤𝑞′
∑ ∑ [𝐶𝐹 𝑖𝑝,𝑛𝑝,𝑤𝑝
𝑖𝑝,𝑛𝑞′𝜎𝑖𝑝,𝑛𝑞
′
29 𝑝 𝑝 𝑝 𝐶𝐿 𝑞′
30
31
(77)
32 + 𝐶𝐴𝑖𝑝,𝑛𝑞′ 𝑋𝑖𝑖𝑝𝑝,𝑛(
𝑝,𝑤𝑝
,𝑛𝑞 )𝛽]
33 ′
34
35 𝑃 𝑃 𝐿𝑝
36
37
+ ∑∑ ∑ ∑
𝑞 = 1𝑝 = 1𝑙𝑝 = 𝑙𝑞 = 1(𝑚 ,𝑗 ) ∈ 𝐻𝑇𝑙𝑝𝑚 ∈ 𝐻𝑈𝑙𝑞𝑗𝑝 ∈ 𝐶𝑝 𝑙𝑝
∑ ∑ [𝐶𝐹 𝑚𝑝,𝑗𝑝,𝑙𝑝
𝑚𝑞,𝑗𝑝𝜎𝑚𝑞,𝑗𝑝
38 𝑝 𝑝 𝑝 𝑞 𝐻𝑇
39
40
41 + 𝐶𝐴𝑚𝑞,𝑗𝑝 𝑋𝑚 (
𝑝,𝑗𝑝,𝑙𝑝
𝑚𝑞,𝑗𝑝 )𝛽]
42
43
45
46
𝑁𝑇𝐶𝐶3 = ∑ ∑ ∑ ∑ ∑ 𝐶𝑀 𝑖𝑝,𝑗𝑝 𝑖𝑝,𝑗𝑝,𝑦𝑖𝑝,𝑗𝑝
𝑖𝑝,𝑗𝑝𝑒𝑖𝑝,𝑗𝑝,𝑘
47 𝑝 = 1(𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝𝑦𝑖𝑝,𝑗𝑝 = 1 (𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝 𝑘 ∈ 𝑆𝑇
48 𝑖𝑝 ≠ 𝑖𝑝 ∨ 𝑗𝑝 ≠ 𝑗𝑝
49 (78)
51
52 + ∑∑ ∑ ∑ ∑ ∑ 𝐶𝑀𝑖𝑖𝑝𝑝,𝑛𝑝,𝑤𝑝 𝑖𝑝,𝑛𝑝,𝑤𝑝
,𝑛𝑞 𝑒𝑖𝑝,𝑛𝑞
′ ′
53 𝑝 = 1𝑞′ = 1𝑤𝑝 = 𝑤𝑞′ = 1(𝑖 ,𝑛 ) ∈ 𝐶𝐿𝑤𝑝𝑖𝑝 ∈ 𝐻𝑝 𝑤𝑝𝑛 ∈ 𝐶𝑈𝑤𝑞′
𝑝 𝑝 𝑝 𝐶𝐿 𝑞′
54 𝑖𝑝 ≠ 𝑖𝑝
55
56
57
58
59
60
26
1
2
4
5 + ∑∑ ∑ ∑ ∑ ∑ 𝑚𝑝,𝑗𝑝,𝑙𝑝 𝑚𝑝,𝑗𝑝,𝑙𝑝
𝐶𝑀𝑚 𝑞,𝑗𝑝
𝑒𝑚𝑞,𝑗𝑝
6 𝑞 = 1𝑝 = 1𝑙𝑝 = 𝑙𝑞 = 1(𝑚 ,𝑗 ) ∈ 𝐻𝑇𝑙𝑝𝑚 ∈ 𝐻𝑈𝑙𝑞𝑗𝑝 ∈ 𝐶𝑝 𝑙𝑝
𝑝 𝑝 𝑝 𝑞 𝐻𝑇
7 𝑗𝑝 ≠ 𝑗𝑝
8
9 5.3 Cost models utilized for implementation of Strategy III
10
11
12 In this case, the total annual utility cost of the multi-plant HEN (𝑇𝑈𝐶′) and the total
13
14 capital cost of all bypasses (𝑁𝑇𝐶𝐶2) can be determined according to equations (70) and (73)
15
16
respectively. The other cost models are presented below:
17
𝑃
19
∑ ∑ ∑ ∑ ∑ [𝐶𝐹 )]
𝛽
20
21
𝐴𝑇𝐶𝐶 =
𝑖𝑝,𝑗𝑝𝜎𝑖𝑝,𝑗𝑝,𝑘 ( 𝑦𝑖
+𝐶𝐴𝑖𝑝,𝑗𝑝 𝑋𝑖𝑝,𝑗𝑝 𝑝𝑝,𝑘
,𝑗
𝑝 = 1(𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝𝑦𝑖𝑝,𝑗𝑝 = 1(𝑖𝑝,𝑗𝑝) ∈ 𝑍𝑝𝑘 ∈ 𝑆𝑇

22
23
𝑃 𝑃 𝑊𝑝
24
25
26
+ ∑∑ ∑ ∑ ∑ ∑ [𝐶𝐹
𝑝 = 1𝑞′ = 1𝑤𝑝 = 𝑤𝑞′ = 1(𝑖 ,𝑛 ) ∈ 𝐶𝐿𝑤𝑝𝑖𝑝 ∈ 𝐻𝑝𝑛 ∈ 𝐶𝑈𝑤𝑞′
𝑖𝑝,𝑛𝑝,𝑤𝑝
𝑖𝑝,𝑛𝑞′𝜎𝑖𝑝,𝑛𝑞
′
27 𝑝 𝑝 𝑝 𝑞′
28
29 (79)
30 (
+ 𝐶𝐴𝑖𝑝,𝑛𝑞′ 𝑋𝑖𝑖𝑝𝑝,𝑛𝑝,𝑤𝑝
,𝑛𝑞 ′
)𝛽]
31
32 𝐿𝑝
𝑃 𝑃
33
34
35
+ ∑∑ ∑ ∑
𝑞 = 1𝑝 = 1𝑙𝑝 = 𝑙𝑞 = 1(𝑚 ,𝑗 ) ∈ 𝐻𝑇𝑙𝑝𝑚 ∈ 𝐻𝑈𝑙𝑞𝑗𝑝 ∈ 𝐶𝑝
∑ ∑ [𝐶𝐹 𝑚𝑝,𝑗𝑝,𝑙𝑝
𝑚𝑞,𝑗𝑝𝜎𝑚𝑞,𝑗𝑝
𝑝 𝑝 𝑝 𝑞
36
37
38
39
(
+ 𝐶𝐴𝑚𝑞,𝑗𝑝 𝑋𝑚 𝑝,𝑗𝑝,𝑙𝑝
𝑚𝑞,𝑗𝑝 )𝛽]
40
41 𝑃 𝑃
42
43
𝑁𝑇𝐶𝐶1 = ∑ ∑ ∑ ∑ ∑ [𝐶𝐹𝐹
𝑝 = 1𝑞′ = 1𝑖𝑝 ∈ 𝐻𝑝𝑗𝑞 ∈ 𝐶𝑞′𝑘 ∈ 𝑆𝑇
𝑖𝑝,𝑗𝑞′𝜉𝑖𝑝,𝑗𝑞′,𝑘 +𝐶𝐴𝑖𝑝,𝑗𝑞′(𝐴𝑖𝑝,𝑗𝑞′)𝛽]
44 𝑞′ ≠ 𝑝 ′
45
46
𝑃 𝑃
47
48
49
+ ∑∑ ∑ ∑
𝑝 = 1𝑞′ = 1𝑖𝑝 ∈ 𝐻𝑝𝑛𝑞 ∈ 𝐶𝑈𝑞′
[𝐶𝐹𝐹𝑖 ,𝑛 𝑝 𝑞′
+ 𝐶𝐴𝑖𝑝,𝑛𝑞′(𝐴𝑖𝑝,𝑛𝑞′)𝛽]𝑣𝐶𝑈
𝑖𝑝,𝑛𝑞′ (80)
50 𝑞′ ≠ 𝑝 ′
51
52 𝑃 𝑃
53
54 + ∑ ∑ ∑ ∑ [𝐶𝐹𝐹 𝑚𝑞,𝑗𝑝 + 𝐶𝐴𝑚𝑞,𝑗𝑝(𝐴𝑚𝑞,𝑗𝑝)𝛽]𝑣𝐻𝑈
𝑚𝑞,𝑗𝑝
55 𝑞 = 1𝑞 = 1𝑚𝑞 ∈ 𝐻𝑈𝑞𝑗𝑝 ∈ 𝐶𝑝
𝑞≠𝑝
56
57
58
59
60
27
1
2
4
5 𝑁𝑇𝐶𝐶3 = ∑ ∑ ∑ ∑ ∑ 𝐶𝑀 𝑖𝑝,𝑗𝑝 𝑖𝑝,𝑗𝑝,𝑦𝑖𝑝,𝑗𝑝
𝑖𝑝,𝑗𝑝𝑒𝑖𝑝,𝑗𝑝,𝑘
6 𝑝 = 1(𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝𝑦𝑖𝑝,𝑗𝑝 = 1 (𝑖𝑝,𝑗𝑝) ∈ 𝑍𝑝 𝑘 ∈ 𝑆𝑇
7 𝑖𝑝 ≠ 𝑖𝑝 ∨ 𝑗𝑝 ≠ 𝑗𝑝
8
10
11
+ ∑∑ ∑ ∑ ∑ ∑
𝐶𝑀𝑖𝑖𝑝𝑝,𝑛𝑝,𝑤𝑝 𝑖𝑝,𝑛𝑝,𝑤𝑝
,𝑛𝑞 𝑒𝑖𝑝,𝑛𝑞
′ ′ (81)
12 𝑝 𝑝 𝑝 𝑞′
13 𝑖𝑝 ≠ 𝑖𝑝
14
16
17 + ∑∑ ∑ ∑ ∑ ∑ 𝐶𝑀 𝑚𝑝,𝑗𝑝,𝑙𝑝 𝑚𝑝,𝑗𝑝,𝑙𝑝
𝑚𝑞,𝑗𝑝 𝑒𝑚𝑞,𝑗𝑝
18 𝑞 = 1𝑝 = 1𝑙𝑝 = 𝑙𝑞 = 1(𝑚 ,𝑗 ) ∈ 𝐻𝑇𝑙𝑝𝑚 ∈ 𝐻𝑈𝑙𝑞𝑗𝑝 ∈ 𝐶𝑝
𝑝 𝑝 𝑝 𝑞
19 𝑗𝑝 ≠ 𝑗𝑝
20
21 𝑃
22
23
𝑁𝑇𝐶𝐶4 = ∑ ∑ ∑ [𝐶𝐹
𝑝 = 1(𝑖𝑝,𝑗𝑝) ∈ 𝑍𝑝𝑘 ∈ 𝑆𝑇
𝑖𝑝,𝑗𝑝 + 𝐶𝐴𝑖𝑝,𝑗𝑝(𝐴𝑖𝑝,𝑗𝑝,𝑘)𝛽]𝑢𝑖𝑝,𝑗𝑝,𝑘
24
25
26 𝑃 𝑊𝑝
27
28
+ ∑∑ ∑
𝑝 = 1𝑤𝑝 = 1(𝑖 ,𝑛 ) ∈ 𝑍𝐶𝐿𝑤𝑝
[𝐶𝐹𝑖 ,𝑛𝑝 𝑝
+ 𝐶𝐴𝑖𝑝,𝑛𝑝(𝐴𝑖𝑝,𝑛𝑝)𝛽]𝑢𝐶𝑈,𝑤
𝑖𝑝,𝑛𝑝
𝑝
(82)
29 𝑝 𝑝 𝑝
30
31 𝑃 𝐿𝑝
32
33
34
+ ∑∑ ∑
𝑝 = 1𝑙𝑝 = 1(𝑚 ,𝑗 ) ∈ 𝑍𝐻𝑇𝑙𝑝
[𝐶𝐹𝑚 ,𝑗 𝑝 𝑝
+ 𝐶𝐴𝑚𝑝,𝑗𝑝(𝐴𝑚𝑝,𝑗𝑝)𝛽]𝑢𝐻𝑈,𝑙𝑝
𝑚𝑝,𝑗𝑝
𝑝 𝑝 𝑝
35
36
𝑃 𝑃 𝑊𝑝
37
38
39
𝑇𝑈𝑃𝐶 = ∑∑ ∑ ∑ ∑ ∑ [𝐶𝑃
𝑖𝑝,𝑛𝑝,𝑤𝑝
𝑖𝑝,𝑛𝑞′𝑒𝑖𝑝,𝑛𝑞
′
]
40 𝑞′ ≠ 𝑝 𝑝 𝑝 𝑝 𝑞′
41 (83)
43
44 + ∑∑ ∑ ∑ ∑ ∑ [𝐶𝑃 𝑚𝑝,𝑗𝑝,𝑙𝑝
𝑚𝑞,𝑗𝑝𝑒𝑚𝑞,𝑗𝑝 ]
45 𝑞 = 1𝑝 = 1𝑙𝑝 = 𝑙𝑞 = 1(𝑚 ,𝑗 ) ∈ 𝐻𝑇𝑙𝑝𝑚 ∈ 𝐻𝑈𝑙𝑞𝑗𝑝 ∈ 𝐶𝑝
𝑞≠𝑝 𝑝 𝑝 𝑝 𝑞
46
47
48 6. Case Studies – Maximum Saving Designs
49
50
51 As an illustrative example, let us consider three chemical plants (P1, P2 and P3) and
52
53 their stream data and utility data given in Tables 1 and 2. Let us further assume that the
54
55 existing single-plant HENs were synthesized according to the conventional simultaneous
56
57
58
optimization strategy8 and they are presented in Figure 3. The minimum TACs of these HENs
59
60 were found to be 316,565 USD/yr (P1), 56,294 USD/yr (P2) and 287,769 USD/yr (P3)
28
1
2
3 respectively. The sum of utility costs of all three single-plant HENs, i.e., 𝑇𝑈𝐶, were
4
5
6 determined to be 630,350 USD/yr.
7
8
9
10 Table 1. Stream data of illustrative example
11
12
13 Table 2. Utility data of illustrative example
14
15 Figure 3. Single-plant HENs of (a) 𝐏𝟏, (b) 𝐏𝟐 and (c) 𝐏𝟑
16
17
18
19
20
The parameters used in both the above and also the subsequent calculations are listed as
21
22 follows:
23
24  Lower bounds of heat duties (𝐿𝑄𝑖𝑝,𝑗𝑝, 𝐿𝑄𝑖𝑝,𝑗𝑞′, 𝐿𝑄𝑖𝑞,𝑗𝑝, 𝐿𝑄𝑖𝑝,𝑛𝑝, 𝐿𝑄𝑖𝑝,𝑛𝑞′, 𝐿𝑄𝑚𝑝,𝑗𝑝 and 𝐿𝑄𝑚𝑞,𝑗𝑝)
25
26 are all set to be: 30 kW.
27
28  Lower bounds of match flow fractions (𝐿𝑅𝐻𝑖𝑝,𝑗𝑝,𝑘, 𝐿𝑅𝐻𝑖𝑝,𝑗𝑞′,𝑘, 𝐿𝑅𝐶𝑖𝑝,𝑗𝑝,𝑘 and 𝐿𝑅𝐶𝑖𝑞,𝑗𝑝,𝑘) are
29
30 all set to be: 0.1.
31
 Lower bounds of bypass flow fraction (𝐿𝑅𝑅𝐶𝑗𝑝,𝑘, 𝐿𝑅𝑅𝐶𝑈𝑗𝑝, 𝐿𝑅𝑅𝐻𝑖𝑝,𝑘 and 𝐿𝑅𝑅𝐻𝑈𝑖𝑝) are
32
33
all set to be: 0.1.
34
35  Lower bounds of heat-transfer areas for the augmented units (𝐿𝑋𝑖𝑝,𝑗𝑝, 𝐿𝑋𝑖𝑝,𝑛𝑞′ and 𝐿𝑋𝑚𝑞,𝑗𝑝)
36
37 are all set to be: 1 m2.
38
39  Based on a life length of 10 years and an arbitrarily-chosen yearly interest rate of
40
41 5.85 %, an annualization factor (𝑎𝑓) of 0.1349 is adopted in this study.
42
43
 The exponent of area in the variable cost term (𝛽) is 0.83.
44
45
46  The variable cost coefficients of the heat exchangers (𝐶𝐴𝑖𝑝,𝑗𝑝, 𝐶𝐴𝑖𝑝,𝑗𝑞′ and 𝐶𝐴𝑖𝑞,𝑗𝑝) and
47
48 those of coolers (𝐶𝐴𝑖𝑝,𝑛𝑝 and 𝐶𝐴𝑖𝑝,𝑛𝑞′) are all set to be: 380 $/m1.66; The variable cost
49
50
51 coefficients of the heaters (𝐶𝐴𝑚𝑝,𝑗𝑝 and 𝐶𝐴𝑚𝑞,𝑗𝑝) are all set to be: 700 USD/m1.66.
52
53  The fixed costs of interplant heat exchangers, coolers and heaters (𝐶𝐹𝑖𝑝,𝑗𝑞′, 𝐶𝐹𝑖𝑞,𝑗𝑝, 𝐶𝐹𝑖𝑝,𝑛𝑞′
54
55
56 and 𝐶𝐹𝑚𝑞,𝑗𝑝) are all set to be: 30000 USD; The fixed costs of all inner-plant units (𝐶𝐹𝑖𝑝,𝑗𝑝,
57
58 𝐶𝐹𝑖𝑝,𝑛𝑝 and 𝐶𝐴𝑚𝑝,𝑗𝑝) are all assumed to be: 10000 USD.
59
60
29
1
2
3  The re-piping cost of every bypass (𝐵𝑌𝑖𝑝 and 𝐵𝑌𝑗𝑝) is 500 USD.
4
5
6  The reassignment cost for existing units (𝐶𝑀𝑖𝑖𝑝,𝑗,𝑗𝑝, 𝐶𝑀𝑖𝑖𝑝,𝑛𝑝,𝑤𝑝
and 𝐶𝑀𝑚 𝑝,𝑗𝑝,𝑙𝑝
) are all set to
𝑝 𝑝 𝑝,𝑛𝑞 ′
𝑚𝑞,𝑗𝑝
7
8 be: 2000 $.
9
10  The fixed costs of pipes for adopting utilities from other plants (𝐶𝑃𝑖𝑝,𝑛𝑞′ and 𝐶𝑃𝑚𝑞,𝑗𝑝) are
11
12 all set to be: 5000 $.
13
14
Finally, it should be noted that the optimal solutions presented in the sequel were all obtained
15
16 with solver BARON in GAMS 27.3 on a personal computer (Intel Core i7 6700; 16G).
17
18 6.1 Optimal solution obtained with Strategy I
19
20
An MINLP model can be constructed according to the formulations described in Part
21
22
23 A of Supporting Information, subsection 4.1 and subsection 5.1. The actual numbers of real
24
25 and integer variables in the GAMS code were 1503 and 466 respectively, while that of
26
27 constraints was 3167. After solving this model, an optimal revamped design of the three-plant
28
29
30 HEN can be generated (see Figure 4). The computation time in this case was around 31000 s.
31
32 Notice that every interplant match in Figure 4 is represented with vertically connected circles
33
34 filled with gray colour and each should be housed in a purchased new heat exchanger. All
35
36
37
inner-plant matches are represented with vertically connected circles without colour and they
38
39 are housed (at least partially) in the existing heat exchangers. The inner-plant matches that
40
41 require augmented units are indicated with double circles, while the others are marked with
42
43
single circles. Table 3 shows the arrangements of new units which house interplant matches.
44
45
46 The net saving of this design, i.e., 𝑇𝐴𝐶𝑆, was determined to be 239,218 USD/yr, while the
47
48 sum of utility costs of all single-plant HENs, i.e., 𝑇𝑈𝐶, and the sum of the yearly utility cost
49
50 and annualized total capital investment of the retrofit design, i.e., 𝑇𝑈𝐶′ +𝑎𝑓
51
52
53 (𝐴𝑇𝐶𝐶 + 𝑁𝑇𝐶𝐶1 + 𝑁𝑇𝐶𝐶2 + 𝑁𝑇𝐶𝐶3 + 𝑁𝑇𝐶𝐶4 + 𝑇𝑈𝑃𝐶), were 630,350 USD/yr (a constant
54
55 parameter) and 391,132 USD/yr, respectively. Finally, by comparing Figure 3 and Figure 4, it
56
57
can be observed that a large amount of utility saving can be realized via retrofit, i.e.
58
59
60
30
1
2
3  The hot utilities used on the two cold streams in plant P1 can be reduced
4
5
6 significantly, i.e., the heat duty on 𝐶1𝑃1 is decreased from 97 kW to 0 kW and
7
8 the heat duty on 𝐶2𝑃1 from 640 kW to 237.5 kW;
9
10
11
 The cold utility used on hot stream 𝐻1𝑃2 in plant P2 can be reduced from 110.5
12
13 kW to 0 kW;
14
15  The cold utility used on hot stream 𝐻2𝑃3 in plant P3 can be reduced from 629.5
16
17
18 kW to 243.2 kW.
19
20
21 Figure 4. Interplant heat integration scheme obtained by revamping the single-plant
22
23
24
HENs of P1, P2 and P3 with Strategy I
25
26
27
28 Table 3. Assignments of new units using Strategy I
29
30
6.2 Optimal solution obtained with Strategy II
31
32
33 Another MINLP model was constructed according to the formulations described in
34
35 Part A of Supporting Information, subsection 4.2 and subsection 5.2. The actual numbers of
36
37 real and integer variables in the GAMS code were 1514 and 558 respectively, while that of
38
39
41
42 HEN can be generated and this design is presented in Figure 5. The computation time in this
43
44 case was around 134000 s. The symbols used in Figure 5 follow exactly the same
45
46
47
conventions described in subsection 6.1. Since Strategy II allows assignment of every
48
49 existing heat exchanger, cooler or heater to any existing match of the same type within the
50
51 same plant, additional information can be extracted from the optimal solution. Table 4 shows
52
53
the placement scheme of the existing units in the retrofit design, while Table 5 shows the
54
55
56 arrangements of new units. Notice from Table 3 that the two existing heat exchangers used to
57
58 house two matches (𝐻1𝑃2,𝐶1𝑃2) and (𝐻1𝑃2,𝐶2𝑃2), respectively, in the original single-plant
59
60
31
1
2
3 design of P2 are switched in the revamp design. Notice also from Table 4 that the match
4
5
6 (𝐻1𝑃1,𝐶𝑊𝑃1) is not housed in its original cooler in the retrofit design and, instead, the hot
7
8 stream 𝐻1𝑃1 is matched with the cheapest cooling water from plant P3. Similarly, the
9
10
11
matches (𝐿𝑃𝑃1,𝐶2𝑃1) and (𝐿𝑃𝑃2,𝐶2𝑃2) are also not housed in their original heaters in the
12
13 retrofit design and, instead, the cold streams 𝐶2𝑃1 and 𝐶2𝑃2 are matched with the cheapest
14
15
low-pressure steam from plant P3. Furthermore, the existing cooler for original match
16
17
18 (𝐻1𝑃2,𝐶𝑊𝑃2) and existing heater for original match (𝐿𝑃𝑃1,𝐶1𝑃1) are not needed in the
19
20 retrofit design. Finally, the net saving of this second design, i.e., 𝑇𝐴𝐶𝑆, was determined to be
21
22
239,399 USD/yr, while the sum of utility costs of all single-plant HENs, i.e., 𝑇𝑈𝐶, and the
23
24
25 sum of the yearly utility cost and annualized total capital investment of the retrofit design, i.e.,
26
27 𝑇𝑈𝐶′ +𝑎𝑓(𝐴𝑇𝐶𝐶 + 𝑁𝑇𝐶𝐶1 + 𝑁𝑇𝐶𝐶2 + 𝑁𝑇𝐶𝐶3 + 𝑁𝑇𝐶𝐶4 + 𝑇𝑈𝑃𝐶), were 630,350 USD/yr
28
29
30
(a constant parameter) and 390,951 USD/yr, respectively. Finally, by comparing Figure 4 and
31
32 Figure 5, it can be observed that
33
34  The HEN structures obtained with Strategies I and II are essentially the same,
35
36
while the total utility consumption rates in both cases are almost equal;
37
38
39  Although it is necessary to include the extra capital cost for the aforementioned
40
41 reassignment of exchanger units in plant P2 so as to implement Strategy II, the
42
43 total capital cost of augmented units and new interplant units of plant P1 is
44
45
46 higher. Since the extra capital cost of the former is lower than that of the latter,
47
48 Strategy II achieved a slightly larger TAC saving when compared with Strategy I.
49
50
51
52
53
Figure 5. Multi-plant HEN design obtained by revamping the single-plant HENs of P1,
54
55 P2 and P3 with Strategy II
56
57
58
59
Table 4. Assignments of existing units using Strategy II
60
32
1
2
3
4
5 Table 5. Assignments of new units using Strategy II
6
7
8
9
10 6.3 Optimal solution obtained with Strategy III
11
12 A third MINLP model was constructed according to the formulations described in
13
14 Part A of Supporting Information, subsection 4.3 and subsection 5.3. The actual numbers of
15
16
17 real and integer variables in the GAMS code were 1532 and 694 respectively, while that of
18
20
21 HEN can be generated according to Strategy III and this design is presented in Figure 6. The
22
23
computation time in this case was around 165000 s. The symbols used in Figure 6 follow
24
25
26 exactly the same conventions described in subsection 6.1, while the new inner-plant matches
27
28 housed in the new purchased units are represented with vertically connected circles filled
29
30 with green colour. Since Strategy III allows assignment of every existing heat exchanger,
31
32
33 cooler or heater to any match of the same type within the same plant, additional information
34
35 can be extracted from the optimal solution. Table 6 shows the assignments of the existing
36
37 units in the retrofit design, while Table 7 shows the arrangements of new units. Notice from
38
39
40 Table 6 that the two existing heat exchangers used to house two matches (𝐻1𝑃2,𝐶1𝑃2) and
41
42 (𝐻1𝑃2,𝐶2𝑃2), respectively, in the original single-plant design of P2 are switched in the
43
44
revamp design. Notice also from Table 7 that inner-plant match (𝐻2𝑃3,𝐶1𝑃3) is housed in
45
46
47 new purchased heat exchanger. On the other hand, the existing cooler for original match
48
49 (𝐻1𝑃2,𝐶𝑊𝑃2) and the existing heater for original match (𝐿𝑃𝑃1,𝐶1𝑃1) are not needed in the
50
51
52 retrofit design, while hot stream 𝐻1𝑃1 and cold streams 𝐶2𝑃1 and 𝐶2𝑃2 are matched with the
53
54 cheapest utilities from plant P3 (see Table 6). Finally, the net TAC saving of the last design,
55
56 i.e., 𝑇𝐴𝐶𝑆, was found to be the highest among all three strategies considered so far (i.e.,
57
58
59 241,947 USD/yr), while the sum of utility costs of all single-plant HENs, i.e., 𝑇𝑈𝐶, was
60
33
1
2
3 630,350 USD/yr (a constant parameter) and the sum of the yearly utility cost and annualized
4
5
6 total capital investment of the retrofit design, i.e., 𝑇𝑈𝐶′ +𝑎𝑓
7
8 (𝐴𝑇𝐶𝐶 + 𝑁𝑇𝐶𝐶1 + 𝑁𝑇𝐶𝐶2 + 𝑁𝑇𝐶𝐶3 + 𝑁𝑇𝐶𝐶4 + 𝑇𝑈𝑃𝐶), was the lowest at 388,403 USD/yr.
9
10
Finally, by comparing Figure 5 and Figure 6, it can be observed that
11
12
13  The HEN structures obtained with Strategies II and III are similar but not
14
15 identical, while the total utility consumption rates in both cases are almost equal;
16
17  When compared with the HEN design generated with Strategy II, the design
18
19
20 produced with Strategy III needs one extra purchased unit for an inner-plant
21
22 exchanger match in plant P3.
23
24  Although it is necessary to buy a new unit for the above purpose, the capital
25
26
27
investment of augmented units for Strategy III is significantly less than that for
28
29 Strategy II.
30
31
32
33
Figure 6. Multi-plant HEN design obtained by revamping the single-plant HENs of P1,
34
35
36 P2 and P3 with Strategy III
37
38
39
40 Table 6. Assignments of existing units using Strategy III
41
42
43
44 Table 7. Assignments of new units using Strategy III
45
46
47
48
49 7. Core and Shapley Values
50
51 The retrofit design of the above multi-plant HEN actually lacks one critical component.
52
53
54 In particular, only the total saving of the entire system, i.e., 𝑇𝐴𝐶𝑆, is determined by solving
55
56 the corresponding MINLP model, while the practical issues of benefit allocation are not
57
58 addressed at all. In the present work, this allocation problem is viewed as a cooperative game
59
60
34
1
2
3 and all players of the game form a so-called “coalition.” Typically, the core and Shapley
4
5
6 values are used to characterize the reasonable and fair solution(s) for distributing the financial
7
8 benefit within the coalition. Although extensive discussions on their evaluation procedures
9
10 have already been published, e.g., see Branzei et al.17, a brief summary is still given in the
11
12
13
sequel for the sake of illustration clarity.
14
15 “Core” is the solution set of a cooperative game. Each solution in the set depicts a
16
17 realizable plan for every member of the coalition to receive a reasonable portion of total cost
18
19
saving after retrofitiing. To facilitate illustration, let us use 𝛹 = {1,2,⋯,𝑛} to represent the set
20
21
22 of all players in a game and Σ ⊆ 𝛹 denotes a coalition. Then all possible coalitions should
23
24 form the power set of 𝛹 (denoted as 2𝛹) and a characteristic function 𝜈( ∙ ) can be defined
25
26
accordingly as the mapping 𝜈: 2𝛹→𝑅. The function value 𝜈(Σ), where Σ ∈ 2𝛹, is the total cost
27
28
29 saving (𝑇𝐴𝐶𝑆) realized by coalition Σ as a whole. To ensure function consistency, it is also
30
31 required that ν(∅) = 0. Let us further denote the annual cost saving allocated to plant 𝑖 ∈ 𝛹 in
32
33 coalition Σ ⊆ 𝛹 as 𝑥Σ,𝑖. Thus, ν(Σ) = ∑𝑖 ∈ Σ𝑥Σ,𝑖 and 𝐱Σ = [𝑥Σ,1,𝑥Σ,2,⋯] is referred to as the pre-
34
35
36 imputation vector of coalition Σ. The pre-imputation vector of grand coalition 𝛹, i.e., 𝐱𝛹, in
37
38
the core 𝐶(ν) should possess the following properties
39
40
41
42 1. Individual rationality: The cost saving allocated to player 𝑖 in the grand coalition 𝛹
43
44 should be larger than or equal to that achieved by a single player individually, i.e.
45
46 𝑥Ψ,𝑖 ≥ 𝜈(𝑖), ∀𝑖 ∈ 𝛹 (84)
47
48
49 2. Group rationality: The total cost saving realized by the grand coalition should be
50
51 entirely distributed to all its members, i.e.
52
53
54 ∑𝑥 𝛹,𝑖 = ν(𝛹) (85)
55 𝑖∈𝛹
56
57 3. Coalition rationality: The total cost saving realized by a subcoalition should not be
58
59 greater than the sum of cost savings allocated to the members of this subcoalition by the
60
35
1
2
3 grand coalition, i.e.
4
5
6
7
∑𝑥
𝑖∈Σ
𝛹,𝑖 ≥ ν(Σ), ∀Σ ⊆ 𝛹 (86)
8
9
10
4. No cross subsidization: The cost saving allocated to player 𝑖 by coalition 𝛹 should be
11
12 samller than or equal to the marginal contribution of player 𝑖 to the total cost saving of
13
14 coalition 𝛹, i.e.
15
16 𝑥𝛹,𝑖 ≤ ν(𝛹) ― 𝜈(𝛹\𝑖), ∀𝑖 ∈ 𝛹. (87)
17
18
19 Notice that 𝜈(𝛹\𝑖) above denotes the total cost saving achieved by the subcoalition of the
20
21 grand coalition 𝛹 and this subcoalition is formed by excluding player 𝑖 from 𝛹. In other
22
23
words, equation (87) is needed because, if otherwise, the members in 𝛹\𝑖 do not have the
24
25
26 incentive to accept player 𝑖.
27
28
29 It is clear that the core 𝐶(𝑣) only represents a feasible region. A one-point solution
30
31 can be obtained by computing the Shapley values. This allocation approach essentially calls
32
33
34
for dividing and distributing the 𝑇𝐴𝐶𝑆 of a coalition according to the contribution level of
35
36 each participating member. Before evaluating these so-called Shapley values for benefit
37
38 allocation, it is necessary to calculate the cost savings achieved by all possible subcoalitions
39
40
41
Σ ⊆ 𝛹. To enumerate all scenarios exhaustively, let us first consider the 𝑛! permutations of
42
43 the 𝑛 players in 𝛹 and collect the corresponding sequences in set Π(𝛹). Let us further
44
45 express an element in Π(𝛹) as 𝛑𝜎 (where, σ = 1,2,⋯,𝑛!), i.e., Π(Ψ) = {𝛑1,𝛑2,⋯,𝛑𝑛!}, while a
46
47
48
particular sequence σ′ in Π(𝛹) may be written explicitly as 𝛑𝜎′ = (𝜋𝜎′(1),𝜋𝜎′(2),⋯,𝜋𝜎′(𝑛)). A
49
50 sequence of marginal contributions of the total cost saving (denoted as 𝐦𝜎) can then be
51
52 computed for every sequence 𝛑𝜎 in Π(𝛹), i.e.
53
54
55
56 𝐦𝜎 = {𝑚𝜎𝜋𝜎(1),𝑚𝜎𝜋𝜎(2),⋯,𝑚𝜎𝜋𝜎(𝑘),⋯,𝑚𝜎𝜋𝜎(𝑛)} (88)
57
58
59 where,
60
36
1
2
3 𝑚𝜎𝜋𝜎(1) = ν(𝜋𝜎(1)) ― 𝜈(∅) (89)
4
5
6
7 𝑚𝜎𝜋𝜎(𝑘) = ν(𝜋𝜎(1),⋯,𝜋𝜎(𝑘)) ― 𝜈(𝜋𝜎(1),⋯,𝜋𝜎(𝑘 ― 1)) (90)
8
9
10 and 𝑘 = 2,3,⋯,𝑛. It should be noted that the precedence order of the elements in sequence 𝐦σ
11
12 corresponds to that in sequence 𝛑𝜎. These elements can be rearranged according the original
13
14
15 order in 𝛹 and then placed in another column vector 𝐨𝜎. After obtaining 𝐨𝜎 that stores the
16
17 rearranged marginal contributions for every sequence 𝛑𝜎 in Π(𝛹), one can then compute the
18
19
20
corresponding arithmetic averages and store them in the Shapley-value vector 𝛗𝑁 as follows
21
22
1
23
24
𝛗𝛹 =
𝑛! ∑
𝛑𝜎 ∈ 𝛑(𝛹)
𝐨𝜎 (91)
25
26
27
28 where, 𝛗𝛹 = [𝜑𝛹,1 𝜑𝛹,2 ⋯ 𝜑𝛹,𝑛]𝑇 and 𝜑𝛹,𝑖 (where, 𝑖 = 1,2,⋯,𝑛) denotes the average
29
30
31 benefit (cost saving) allocated to player 𝑖 by coalition 𝛹. It should also be noted that the
32
33 above notation on the Shapley values can be extended to any subset of the grand coalition,
34
35 i.e., Σ ⊆ 𝛹. If the players in Σ form a coalition, then the Shapley value of player 𝑖 (∀𝑖 ∈ Σ) can
36
37
38
be written as 𝜑Σ,𝑖.
39
40
41 To summarize, the required procedure for computing the Shapley values in the present
42
43 application is outlined below:
44
45 1. Determine the total cost savings (𝑇𝐴𝐶𝑆s) of coalitions formed by all possible
46
47
48 combinations of plants according to the proposed MINLP models;
49
50 2. Determine the marginal benefits of each plant according to the results obtained
51
52 in step 1 and equations (89) and (90) for all possible precedence orders of this
53
54
55 plant joining the coalition.
56
57 3. Determine the Shapley value of each plant by taking the arithmetic average of
58
59 the marginal benefits obtained in step 2 according to equation (91).
60
37
1
2
3
4
8. Case Studies – Benefit Allocation
5
6 To illustrate the aforementioned computation procedure, let us revisit the retrofit
7
8 problem discussed in section 6. As mentioned before in section 7, the function value 𝜈(Σ),
9
10
where Σ ∈ 2𝛹 and Σ ⊆ 𝛹, should be viewed as the total cost saving (𝑇𝐴𝐶𝑆) realized by the
11
12
13 corresponding coalition. In the present case, 𝛹 = {𝑃1,𝑃2,𝑃2} and let us further define the
14
15 following subsets of 𝛹:
16
17
 Σ11 = {𝑃1}, Σ12 = {𝑃2}, Σ13 = {𝑃3};
18
19
20  Σ21 = {𝑃1,𝑃2}, Σ22 = {𝑃2,𝑃3}, Σ23 = {𝑃3,𝑃1}.
21
22 Thus, the Shapley-value vector can be written as
23
24 2𝜈(𝛹) + 𝜈(Σ23) + 𝜈(Σ21) + 2𝜈(Σ11) ― 2𝜈(Σ22) ― 𝜈(Σ12) ― 𝜈(Σ13)
[ ]
25
6
[ ]
26 𝜑𝛹,𝑃1
2𝜈(𝛹) + 𝜈(Σ21) + 𝜈( Σ22) + 2𝜈(Σ12) ― 2𝜈(Σ23) ― 𝜈(Σ13) ― 𝜈(Σ11)
27 𝛗𝛹 = 𝜑𝛹,𝑃2 = 6 (92)
28 𝜑𝛹,𝑃3 2𝜈(𝛹) + 𝜈(Σ22) + 𝜈(Σ23) + 2𝜈(Σ13) ― 2𝜈(Σ21) ― 𝜈(Σ11) ― 𝜈(Σ12)
29
6
30
31
32
The allocation schemes of cost savings achieved with different retrofit strategies can all be
33
34
35 devised according to equation (92).
36
37 8.1 Allocation of cost saving achieved with Strategy I
38
39 By repeated solving the MINLP models described in Appendix, subsection 4.1 and
40
41
42 subsection 5.1 for coalitions 𝚺𝟐𝟏, 𝚺𝟐𝟐, 𝚺𝟐𝟑 and 𝜳, one can obtain the following function values.
43
44
45  ν(∅) = ν(Σ11) = ν(Σ12) = ν(Σ13) = 0;
46
47
48
 ν(Σ21) = 106801, ν(Σ22) = 28713, ν(Σ23) = 162366;
49
50  ν(𝛹) = 239218.
51
52
53 Therefore, the cost savings allocated to P1, P2 and P3 can be computed according to equation
54
55
56 (92), i.e., 𝜑𝛹,𝑃1 = 115,029 USD/yr (48.1%), 𝜑𝛹,𝑃2 = 48203 USD/yr (20.1%) and 𝜑𝛹,𝑃3
57
58 = 75,985 USD/yr (31.8 %).
59
60
38
1
2
3 8.2 Allocation of cost saving achieved with Strategy II
4
5
7
9
10
11  ν(∅) = ν(Σ11) = ν(Σ12) = ν(Σ13) = 0;
12
13
14  ν(Σ21) = 108904, ν(Σ22) = 28713, ν(Σ23) = 162366;
15
16
 ν(𝛹) = 239399.
17
18
19
21
22 (92), i.e., 𝜑𝛹,𝑃1 = 115,440 USD/yr (48.2%), 𝜑𝛹,𝑃2 = 48,614 USD/yr (20.3%) and 𝜑𝛹,𝑃3
23
24
= 75345 USD/yr (31.5 %).
25
26
27
28
8.3 Allocation of cost saving achieved with Strategy III
29
31
33
34
35
 ν(∅) = ν(Σ11) = ν(Σ12) = ν(Σ13) = 0;
36
37
38  ν(Σ21) = 108904, ν(Σ22) = 28713, ν(Σ23) = 168878;
39
40
41  ν(𝛹) = 241947.
42
43
45
46 (92), i.e., 𝜑𝛹,𝑃1 = 117,375 USD/yr (48.5%), 𝜑𝛹,𝑃2 = 47,293 USD/yr (19.6%) and 𝜑𝛹,𝑃3
47
48
49 = 77,279 USD/yr (31.9 %).
50
51
52 8.4 Implications of allocation results
53
54
55 It can be observed from the aformentioned Shapley values that, although the
56
57 percentages of cost savings allocated to P1 and P3 rise as as the model restriction gradually
58
59
relaxes from Strategy I to III, the share of P2 decreases from Strategy II to III, which means
60
39
1
2
3 that the average marginal contribution of P2 declines from Strategy II to III. On the other
4
5
6 hand, despite the fact that the cost savings achieved by the three-plant HENs for ν(𝛹)
7
8 increase in the corresponding three cases, Σ22 = {𝑃2,𝑃3} still remains unaffected by the
9
10
different revamp strategies under consideration. Thus, the above observations seem to
11
12
13 suggest that the contribution levels to overall cost saving of the three-plant heat integration
14
15 scheme can be ranked as (1) P1, (2) P3 and (3) P2 in the present example.
16
17
18
19
20
21
22
9. Conclusions
23
24 A comprehensive design procedure is proposed in this paper to revamp the multi-plant
25
26 HENs for lowering the overall utility consumption level and to divide and allocate the
27
28 resulting benefit (cost saving) fairly to all members of the interplant heat integration scheme.
29
30
31 Three retrofit strategies are devised to satisfy practical requirements, such as safety issues,
32
33 space limitations and/or operability problems, etc. The corresponding MINLP models can be
34
35 constructed by augmenting the superstructure-based formulation presented in Appendix with
36
37
38
different versions of additional constraints imposed on the new and original matches, on re-
39
40 piping and reuse of existing units in the multi-plant HEN, and on placement of purchased
41
42 heat exchangers, coolers and heaters. As expected, it can be observed from the illustrative
43
44
example that a greater financial gain can always be realized with a less-constrained MINLP
45
46
47 model. The allocation plan of overall cost saving is drawn up according to the well-
48
49 established Shapley values in this study. From the allocation results in the same example, the
50
51 contribution levels of the players in the corresponding cooperative game can also be easily
52
53
54 identified. In summary, the proposed revamp design is indeed cost effective and the
55
56 associated allocation plan is fair enough if factors other than marginal benefits, e.g., risk of
57
58 coalition collapse,1 are not important. Therefore, under this condition, the proposed allocation
59
60
40
1
2
3 plan should be acceptable to all parties participating in the multi-plant HEN retrofit project.
4
5
6 Furthermore, the proposed MINLP models and the corresponding calculation procedure for
7
8 stipulating the benefit allocation plans should be applicable to any practical retrofit design for
9
10 total-site heat integration.
11
12
13
14
15 Supporting Information
16
17 Part A: Superstructure-Based Model Constraints; Part B: Summary of Revamp Strategies;
18
19 This information is available free of charge via the Internet at http://pubs.acs.org/
20
21
22
23
24 Acknowledgement - This work is supported by the Ministry of Science and Technology of
25 the ROC government under grant MOST 109-2221-E-006-085.
26
27
28
29
30
31
References
32
33
34 1. Jin, Y.; Chang, C. T.; Li, S.; Jiang, D. On the Use of Risk-Based Shapley Values for Cost
35
Sharing in Interplant Heat Integration Programs. Applied Energy 2018, 211, 904–920.
36
37 2. Bagajewicz, M. J.; Rodera, H. Multiple Plant Heat Integration in a Total Site. AIChE
38
39 Journal 2002, 48, 2255–2270.
40
41 3. Kralj A. K. Heat Integration between Two Biodiesel Processes Using a Simple Method.
42 Energy & Fuels 2008, 22, 1972–1979.
43
44 4. Liew, P. Y.; Alwi, S. R. W.; Klemes, J. J.; Varbanov, P. S.; Manan, Z. A. Algorithmic
45
46 Targeting for Total Site Heat Integration with Variable Energy Supply/Demand. Appl.
47
48
Therm. Eng. 2014, 70, 1073–1083.
49 5. Liew, P. Y.; Theo, W. L.; Alwi, S. R. W.; Lim, J. S.; Manan, Z. A.; Klemes, J. J.;
50
51 Varbanov, P. S. Total Site Heat Integration Planning and Design for Industrial, Urban and
52
53 Renewable Systems. Renew. Sust. Energ. Rev. 2017, 68, 964–985.
54
6. Zhang, B. J.; Li, J.; Zhang, Z. L.; Wang, K.; Chen, Q. L. Simultaneous Design of Heat
55
56 Exchanger Network for Heat Integration Using Hot Direct Discharges/Feeds between
57
58 Process Plants. Energy 2016, 109, 400–411.
59
60
41
1
2
3 7. Chang, C. L.; Chen, X. L.; Wang, Y. F.; Feng, X. Simultaneous Optimization of Multi-
4
5 Plant Heat Integration using Intermediate Fluid Circles. Energy 2017, 121, 306–317.
6
7 8. Yee, T. F.; Grossmann, I. E. Simultaneous Optimization Models for Heat Integration—II.
8
9
Heat Exchanger Network Synthesis. Computers & Chemical Engineering 1990, 14(10),
10 1165–1184.
11
12 9. Papoulias, S. A.; Grossmann, I. E. A Structural Optimization Approach in Process
13
14 Synthesis—II. Heat Recovery Networks. Computers & Chemical Engineering 1983, 7,
15
707–721.
16
17 10. Floudas, C. A.; Ciric, A. R.; Grossmann, I. E. Automatic Synthesis of Optimum Heat
18
19 Exchanger Network Configurations. AIChE Journal 1986, 32, 276–290.
20
21 11. Cheng, S. L.; Chang, C. T.; Jiang, D. A Game-Theory Based Optimization Strategy to
22 Configure Interplant Heat Integration Schemes. Chemical Engineering Science 2014, 118,
23
24 60–73.
25
26 12. Chang, H. H.; Chang, C. T.; Li, B. H. Game-Theory Based Optimization Strategies for
27
28
Stepwise Development of Indirect Interplant Heat Integration Plans. Energy 2018, 148,
29 90–111.
30
31 13. Hiete, M.; Ludwig, J.; Schultmann, F. Intercompany Energy Integration. J. Ind. Ecol.
32
33 2012, 16(5), 689–698.
34
14. Tan, R. R.; Andiappan, V.; Wan, Y. K.; Ng, R. T. L.; Ng, D. K. S. An Optimization-
35
36 Based Cooperative Game Approach for Systematic Allocation of Costs and Benefits in
37
38 Interplant Process Integration. Chem. Eng. Res. Des. 2016, 106, 43-58.
39
40 15. Fernandez, F. R.; Hinojosa, M. A.; Puerto, J. Core Solutions in Vector-Valued Games. J.
41 Optim. Theory Appl. 2002, 112, 331–360.
42
43 16. Grabisch, M.; Xie, L. J. A New Approach to the Core and Weber Set of Multichoice
44
45 Games. Math. Method Oper. Res. 2007, 66(3), 491–512.
46
17. Branzei, R.; Dimitrov, D.; Tijs, S. Models in Cooperative Game Theory (second edition)
47
48 2008, Springer, Berlin.
49
50 18. Frisk, M.; Gothe-Lundgren, M.; Jörnsten, K.; Ronnqvist, M. Cost Allocation in
51
52 Collaborative Forest Transportation. Eur. J. Oper. Res. 2010, 205(2), 448–458.
53 19. Ciric, A. R.; Floudas, C. A. A Retrofit Approach for Heat Exchanger Networks.
54
55 Computers & Chemical Engineering 1989, 13(6), 703-715.
56
57 20. Ciric, A. R.; Floudas, C. A. A Comprehensive Optimization Model of the Heat Exchanger
58
59
Network Retrofit Problem. Heat Recovery Systems and CHP 1990, 10(4), 407-422.
60
42
1
2
3 21. Yee, T. F.; Grossmann, I. E. A Screening and Optimization Approach for the Retrofit of
4
5 Heat-Exchanger Networks. Ind. Eng. Chem. Res. 1991, 30(1), 146-162.
6
7 22. Sorsak, A.; Kravanja, Z. MINLP Retrofit of Heat Exchanger Networks Comprising
8
9
Different Exchanger Types. Computers & Chemical Engineering 2004, 28(1-2), 235-251.
10 23. Ponce-Ortega, J. M.; Jiménez-Gutiérrez, A.; Grossmann, I. E. Simultaneous Retrofit and
11
12 Heat Integration of Chemical Processes. Ind. Eng. Chem. Res. 2008, 47(15), 5512-5528.
13
14 24. Smith, R.; Jobson, M.; Chen, L. Recent Development in the Retrofit of Heat Exchanger
15
Networks. Applied Thermal Engineering 2010, 30(16), 2281-2289.
16
17 25. Chen, J. Comments on Improvements on a Replacement for the Logarithmic Mean.
18
19 Chemical Engineering Science 1987, 42(10), 2488-2489.
20
21
22
23
24 For Table of Contents Only
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
43
1
2
3
4
5
6 Figure Legends:
7
8
9 Figure 1. An example of superstructure for two-plant HEN synthesis
10
11
12 Figure 2. Connections between an existing heat exchanger and a new one to enlarge the heat-transfer area
13
14
15 Figure 3. Single-plant HENs of (a) P1, (b) P2 and (c) P3
16
17
18 Figure 4. Interplant heat integration scheme obtained by revamping the single-plant HENs of P1, P2 and P3 with Strategy I.
19
20
21 Figure 5. Multi-plant HEN design obtained by revamping the single-plant HENs of P1, P2 and P3 with Strategy II
22
23
24 Figure 6. Multi-plant HEN design obtained by revamping the single-plant HENs of P1, P2 and P3 with Strategy III
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43 ACS Paragon Plus Environment
44
45
46
1
2
3
4
5
6
Stage Stage
7
C1P1- k=1 k=2
8
HPP1
9
10
11 C1P1-
12 HOP1 H1P1-
13 H1P1 H1P1- H1P1-
14 CWP1
C1P1 C1P1 C1P1
15 C1P1-
16 HPP2
17 H1P1-
18 CWP2
H1P1- H1P1-
19 C1P1- C1P2 C1P2
20 HOP2
21
22
23 C1P2-
HPP1 H1P2- H1P2-
24 H1P2-
C1P1 C1P1
25 CWP1
26
C1P2-
27
28 HOP1
H1P2-
29
H1P2 H1P2- H1P2-
CWP2
30 C1P2 C1P2 C1P2
C1P2-
31
HPP2
32
33
34 C1P2-
35 HOP2
36
37 Temperature location Temperature location Temperature location
38 k=1 k=2 k=3 Figure 1
39
40
41
42
44
45
46
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23 Figure 2
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
44
45
46
1
2
3
4
5 Stage 1 Stage 2 Stage 1 Stage 2
6 (EX2-2)
(EX1-1) (EX2-1)
7 149.1 m2 11.1 m2 91.9 m2 10 m2 3.2 m2
8 P1-H1 P2-H1
200 141 90.1
150 H1P1- 61 61 H1P1- H1P2- H1P2- H1P2-
9 C1P1 CWP1 40 C2P2 C1P2 CWP2 70
10 147 kW 110.5 kW
11
12 1.9 m2
P1-C1 P2-C1
13 140
LPP1- 129.2 H1P1- 60 60
110
110 110 H1P2- 30
C1P1 C1P1 C1P2
14
15 97 kW 623 kW 280 kW
16 30 m2 5.8 m2
P1-C2 P2-C2
17 LPP1- 110 110 110 LPP2- 183.3 H1P2- 140 140
18 190 C2P1
190 C2P2 C2P2
19 640 kW 50.5 kW 324.5 kW
20
(a) (b)
21
Stage 1 Stage 2
22
23 (EX3-1) (EX3-2)
63.5 m2 23.8 m2
24 P3-H1
370 H1P3- 200 H1P3- 150
25 C1P3 C1P3 150
26
27 (EX3-3)
41.9 m2 21.6 m2
28 P3-H2
29 200 200 H2P3-
C1P3
154.5 H2P3-
CWP3
40
30
629.5 kW
31
32 1.9 m2 199 H1P3- 0.37
P3-C1 C1P3
33 360
C1P3- 312.3 H1P3-
C1P3
199 110
HOP3 150 kW
34 H2P3-
510 kW C1P3
35 214.5 kW 199 0.63
36 250.5 kW
(c)
37
Figure 3
38
39
40
41
42
44
45
46
1
2
3
4
5 Stage 1 Stage 2 Stage 3 Stage 4 Stage 5
6 (EX1-1) (149.1 m2)

610.4 m2 11.1 m2
7 P1-H1
150 150 150 H1P1- 61 61 61 H1P1-
40
8 C1P1 CWP3
9 147 kW
10
147.9 H1P1- 0.79
P1-C1 C1P1
140 140 140 60 60 60
11
2
140 623 kW 20.5 m
H2P3-
12 0.2
152.9 m2 110.8 C1P1 0.21
13 P1-C2 30 m 2 153.5 H1P2-

C2P1
0.78
14 190 197.3
LPP3-
C2P1
160.3
0.8
160.3 48.7 m2 110 110 110 110
15 237.5 kW 184.1
H2P3-
C2P1 0.22
16
17 (EX2-1) (EX2-2) (10 m2)
36.7 m2
91.9 m2
18 P2-H1
200 H1P2- 160.2 H1P2- 111 H1P2- 70 70 70 70
19 C2P2 C2P1 C1P2
20 270.5 kW
21 P2-C1 109.9 H1P2-

C1P2
0.81
110 110 110 30 30 30

22 110 225.5 kW
8.7 m2
23
H2P3-
110.4 C1P2 0.19
24
0.34
P2-C2 5.8 m2 5.8 m2
25 190
LPP3-
C2P2
183.3
198.1
H1P2-
C2P2
154.1
0.66
H1P3-
C2P2
140 140 140 140
26 50.5 kW 219 kW
27
(EX3-1)
28 P3-H1
63.5 m2 0.84 H1P3- 156.5
29
C2P2 (EX3-2)
370 H1P3- 198.2 150 150 150 150
23.8 m2 150
C1P3 105.5 kW
30 0.16
H1P3-
C1P3 114.9
31 0.27 H2P3- 111 0.72 H2P3- 87.1
0.19
21.6 m2
32 P3-H2
200 200
C2P1 (EX3-3) (41.9 m2)
2 111.8
C1P1
84.2 84.2 84.2 H2P3-
249.5 m 40
33 132 kW
H2P3-
97 kW
H2P3-
0.81 CWP3 29.7
34 (1.9 m2) 0.64

0.73 C1P3 112 0.28 C1P2 77 243.2 kW
35 H1P3- 54.5 kW
P3-C1 2.9 m2 197.2 0.1
C1P3
HOP3- 311.7 H1P3- 197.2 110 110 110 110
36 360 444.3 C1P3 0.36 C1P3 39.3 kW
H2P3-
37 217.2 kW 515.3 kW 197.2 C1P3 0.9
38 353.3 kW
Figure 4
39
40
41
42
44
45
46
1
2
3
4
6 (EX1-1) (149.1 m2)

574.7 m2 11.1 m2
7 P1-H1
150 150 150 H1P1- 61 61 61 H1P1-
8 C1P1 CWP3
40
9 147 kW
10 P1-C1
147.6 H1P1-
C1P1
0.79
140 140 140 60 60 60

11 140 623 kW 20.1 m2
12
H2P3-
145 m2 111.3 C1P1 0.21
0.19
13 P1-C2 30 m2 153.5 H1P2- 0.79
C2P1
14 190 197.2
LPP3- 159.6
C2P1 0.81
159.6 42.7 m2 110 110 110 110
15 243 kW 182.5
H2P3-
C2P1 0.21
16
17 (EX2-2) (10 m2) 0.42 (EX2-1)
30.1 m2 91.9 m2
18 P2-H1
200 H1P2- 161 H1P2- 111 H1P2- 70 70 70
19 C2P2 C2P1 0.58 C1P2 40.3
70
20 275 kW
21 P2-C1 110 H1P2-

C1P2
0.81
110 110 110 30 30 30

22 110 225.5 kW
7.2 m2
23
H2P3-
110 C1P2 0.19
24 P2-C2 5.8 m2 6 m2
25 190
LPP3-
C2P2
183.3 H1P2-
C2P2
154.7 H1P3-
C2P2
140 140 140 140
26 50.5 kW 214.6 kW
27
28 (EX3-1)
63.5 m2 0.84 H1P3- 156.5
P3-H1
29
C2P2 (EX3-2)
370 H1P3- 200 150 150 150 150
23.8 m2 150
C1P3 110 kW
30 0.16
H1P3-
C1P3 115.1
31 0.25 H2P3- 111 0.73 H2P3- 87.9
0.19
21.6 m2
32 P3-H2
200 200
C2P1 (EX3-3) (41.9 m2)
112.3
C1P1
84.7 84.7 84.7 H2P3-
354 m2 40
33 122 kW
H2P3-
97 kW
H2P3-
0.81 CWP3 29.8
34 0.75 C1P3 112.7 0.27 C1P2 76.3 246 kW
1.9 m2 54.5 kW
35 P3-C1 199 H1P3- 0.1
C1P3
HOP3- 312.3 H1P3- 199 110 110 110 110
36 360 C1P3 C1P3 40.1 kW
H2P3-
37 214.5 kW 510 kW 199 C1P3 0.9
38 360.5 kW Figure 5
39
40
41
42
44
45
46
1
2
3
4
6 (EX1-1) (149.1 m ) 2
627.7 m2 11.1 m2
7 P1-H1
150 150 150 150 H1P1- 61 61 H1P1-
40
8 C1P1 CWP3
9 147 kW
10 P1-C1
148 H1P1-
C1P1
0.79
140 140 140 140 60 60

11 140 623 kW 20.6 m2
12
H2P3-
159.2 m2 110.5 C1P1 0.21
0.19
13 P1-C2 30 m2 154.6 H1P2-
C2P1
0.77
14 190 197.2
LPP3- 160.1
C2P1 0.81
160.1 160.1 50.6 m2 110 110 110
15 238.9 kW 178.9
H2P3-
C2P1 0.23
16
17 (EX2-2) (10 m2) 0.42 (EX2-1)
30.1 m2 91.9 m2
18 P2-H1
200 H1P2- 161 161 H1P2- 111 H1P2- 70 70
19 C2P2 C2P1 0.58 C1P2 40.3
70
20 275 kW
21 P2-C1 110 H1P2-

C1P2
0.81
110 110 110 110 30 30

22 110 225.5 kW
8.3 m2
23
H2P3-
110 C1P2 0.19
24 P2-C2 5.8 m2 6 m2
25 190
LPP3-
C2P2
183.3 H1P2-
C2P2
154.7 154.7 H1P3-
C2P2
140 140 140
26 50.5 kW 214.5 kW
27
28 (EX3-1)
63.5 m2 0.84 H1P3- 156.5
P3-H1
29 370 H1P3- 200 200
C2P2 (EX3-2)
23.8 m2
150 150 150
150
C1P3 110 kW
30 H1P3-
115.1
0.16 C1P3
31 (EX3-3) 0.19
41.9 m2 0.28 H2P3- 111.1 0.72 H2P3- 87 21.6 m2
32 P3-H2
200 200 H2P3- 192.2
C2P1
2 111.6
C1P1
84 84 H2P3-
204.2 m 40
33 C1P3 126.1 kW
H2P3-
97 kW
H2P3-
0.81 CWP3 29.7
34 0.72 C1P3 111.7 0.28 C1P2 76.4 242 kW
35 1.9 m2 H1P3- 54.5 kW

P3-C1 199 0.1
C1P3
HOP3- 312.3 H1P3- 199 H2P3- 189.4 110 110 110
36 360 C1P3 C1P3 C1P3 40.1 kW
H2P3-
37 214.5 kW 510 kW 43 kW 188.4 C1P3 0.9
38 317.4 kW Figure 6
39
40
41
42
44
45
46
1
2
3
4 Table Legends:
5
6
8
9
11
12
14
15
16 Table 4. Assignments of existing units using Strategy II
17
18
20
21
23
24
25 Table 7. Assignments of new units using Strategy III
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

1
2
3
5
6
7 Plant Stream 𝑇𝑇𝑇 (℃) 𝑇𝑇𝑇𝑇 (℃) 𝐹 (kW/℃) ℎ (kW⁄m2 ℃)
8 H1 150 40 7.0 1.2
9
10 P1 C1 60 140 9.0 1.6
11
12 C2 110 190 8.0 1.0
13
14 H1 200 70 5.5 1.5
15 P2 C1 30 110 3.5 1.1
16
17 C2 140 190 7.5 1.2
18
19 H1 370 150 3.0 1.4
20
P3 H2 200 40 5.5 1.1
21
22 C1 110 360 4.5 1.3
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

1
2
3
5
6
7 Unit Cost
Plant Utility 𝑇𝑇𝑇 (℃) 𝑇𝑇𝑇𝑇 (℃) ℎ (kW⁄m2 ∙ ℃)
8 (USD/kW ∙ yr)
9
10 Cooling water 25 30 1.2 200
11
12 LP steam 200 200 1.5 375
13
P1 MP steam 250 250 1.8 575
14
15 HP steam 300 300 2.3 775
16
17 Hot oil 500 475 1.8 900
18
19 Cooling water 25 30 1.2 250
20
LP steam 200 200 1.5 400
21
22 P2 MP steam 250 250 1.8 600
23
24 HP steam 300 300 2.3 800
25
26 Hot oil 500 475 1.8 1000
27
Cooling water 25 30 1.2 150
28
29 LP steam 200 200 1.5 350
30
31 P3 MP steam 250 250 1.8 550
32
33 HP steam 300 300 2.3 750
34 Hot oil 500 475 1.8 850
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

1
2
3
5
6
7 Interplant matches using new heat exchangers
Areas of new heat exchanger (m2 )
8 in retrofit design (𝑖𝑝 , 𝑗𝑞′ ) or (𝑖𝑞 , 𝑗𝑝 )
9
10 (𝐻1𝑃2 , 𝐶2𝑃1 ) 152.9
11
12 (𝐻1𝑃3 , 𝐶2𝑃2 ) 5.8
13
(𝐻2𝑃3 , 𝐶1𝑃1 ) 20.5
14
15 (𝐻2𝑃3 , 𝐶2𝑃1 ) 48.7
16
17 (𝐻2𝑃3 , 𝐶1𝑃2 ) 8.7
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

1
2
3
4 Table 4. Assignments of existing units using Strategy II
5
6
7 Existing Assignments of
Exchanger order Areas of existing heat
8 matches/exchangers exchanger matches in
9 (𝑦𝑖𝑝 ,𝑗𝑝 ) exchangers (𝑚2 )
10 before retrofit (𝑖𝑝 , 𝑗𝑝 ) retrofit design �𝚤̃𝑝 , 𝚥̃𝑝 �
11
12 (𝐻1𝑃1 , 𝐶1𝑃1 ) (1) (𝐻1𝑃1 , 𝐶1𝑃1 ) 149.1
13
(𝐻1𝑃2 , 𝐶1𝑃2 ) (1) (𝐻1𝑃2 , 𝐶2𝑃2 ) 91.9
14
15 (𝐻1𝑃2 , 𝐶2𝑃2 ) (1) (𝐻1𝑃2 , 𝐶1𝑃2 ) 10
16
17 (𝐻1𝑃3 , 𝐶1𝑃3 ) (1) (𝐻1𝑃3 , 𝐶1𝑃3 ) 63.5
18
19 (𝐻1𝑃3 , 𝐶1𝑃3 ) (2) (𝐻1𝑃3 , 𝐶1𝑃3 ) 23.8
20
(𝐻2𝑃3 , 𝐶1𝑃3 ) (1) (𝐻2𝑃3 , 𝐶1𝑃3 ) 41.9
21
22 Existing Assignments of cooler
23 Cooler type Areas of existing
24 matches/coolers matches in retrofit
25 (𝑤𝑝 ) coolers (𝑚2 )
26 before retrofit (𝑖𝑝 , 𝑛𝑝 ) design �𝚤̃𝑝 , 𝑛�𝑞′ �
27
(𝐻1𝑃1 , 𝐶𝐶𝑃1 ) (1) (𝐻1𝑃1 , 𝐶𝐶𝑃3 ) 11.1
28
29 (𝐻1𝑃2 , 𝐶𝐶𝑃2 ) (1) / 3.2
30
31 (𝐻2𝑃3 , 𝐶𝐶𝑃3 ) (1) (𝐻2𝑃3 , 𝐶𝐶𝑃3 ) 21.6
32
33 Existing Assignments of heater
Heater type Areas of existing
34 matches/heaters matches in retrofit
35 (𝑙𝑝 ) heaters (𝑚2 )
36 before retrofit (𝑚𝑝 , 𝑗𝑝 ) design �𝑚
� 𝑞 , 𝚥̃𝑝 �
37
38 (𝐿𝐿𝑃1 , 𝐶1𝑃1 ) (1) / 1.9
39
40 (𝐿𝐿𝑃1 , 𝐶2𝑃1 ) (1) (𝐿𝐿𝑃3 , 𝐶2𝑃1 ) 30
41 (𝐿𝐿𝑃2 , 𝐶2𝑃2 ) (1) (𝐿𝐿𝑃3 , 𝐶2𝑃2 ) 5.8
42
43 (𝐻𝐻𝑃3 , 𝐶1𝑃3 ) (2) (𝐻𝐻𝑃3 , 𝐶1𝑃3 ) 1.9
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

1
2
3
5
6
7 Interplant matches using new heat exchangers
Areas of new heat exchanger (m2 )
8 in retrofit design (𝑖𝑝 , 𝑗𝑞′ ) or (𝑖𝑞 , 𝑗𝑝 )
9
10 (𝐻1𝑃2 , 𝐶2𝑃1 ) 145
11
12 (𝐻1𝑃3 , 𝐶2𝑃2 ) 6
13
(𝐻2𝑃3 , 𝐶1𝑃1 ) 20.1
14
15 (𝐻2𝑃3 , 𝐶2𝑃1 ) 42.7
16
17 (𝐻2𝑃3 , 𝐶1𝑃2 ) 7.2
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

1
2
3
5
6
7 Existing Assignments of
Exchanger order Areas of existing heat
8 matches/exchangers exchanger matches in
9 (𝑦𝑖𝑝 ,𝑗𝑝 ) exchangers (𝑚2 )
10 before retrofit (𝑖𝑝 , 𝑗𝑝 ) retrofit design �𝚤̃𝑝 , 𝚥̃𝑝 �
11
12 (𝐻1𝑃1 , 𝐶1𝑃1 ) (1) (𝐻1𝑃1 , 𝐶1𝑃1 ) 149.1
13
(𝐻1𝑃2 , 𝐶1𝑃2 ) (1) (𝐻1𝑃2 , 𝐶2𝑃2 ) 91.9
14
15 (𝐻1𝑃2 , 𝐶2𝑃2 ) (1) (𝐻1𝑃2 , 𝐶1𝑃2 ) 10
16
17 (𝐻1𝑃3 , 𝐶1𝑃3 ) (1) (𝐻1𝑃3 , 𝐶1𝑃3 ) 63.5
18
19 (𝐻1𝑃3 , 𝐶1𝑃3 ) (2) (𝐻1𝑃3 , 𝐶1𝑃3 ) 23.8
20
(𝐻2𝑃3 , 𝐶1𝑃3 ) (1) (𝐻2𝑃3 , 𝐶1𝑃3 ) 41.9
21
22 Existing Assignments of cooler
23 Cooler type Areas of existing
24 matches/coolers matches in retrofit
25 (𝑤𝑝 ) coolers (𝑚2 )
26 before retrofit (𝑖𝑝 , 𝑛𝑝 ) design �𝚤̃𝑝 , 𝑛�𝑞′ �
27
(𝐻1𝑃1 , 𝐶𝐶𝑃1 ) (1) (𝐻1𝑃1 , 𝐶𝐶𝑃3 ) 11.1
28
29 (𝐻1𝑃2 , 𝐶𝐶𝑃2 ) (1) / 3.2
30
31 (𝐻2𝑃3 , 𝐶𝐶𝑃3 ) (1) (𝐻2𝑃3 , 𝐶𝐶𝑃3 ) 21.6
32
33 Existing Assignments of heater
Heater type Areas of existing
34 matches/heaters matches in retrofit
35 (𝑙𝑝 ) heaters (𝑚2 )
36 before retrofit (𝑚𝑝 , 𝑗𝑝 ) design �𝑚
� 𝑞 , 𝚥̃𝑝 �
37
38 (𝐿𝐿𝑃1 , 𝐶1𝑃1 ) (1) / 1.9
39
40 (𝐿𝐿𝑃1 , 𝐶2𝑃1 ) (1) (𝐿𝐿𝑃3 , 𝐶2𝑃1 ) 30
41 (𝐿𝐿𝑃2 , 𝐶2𝑃2 ) (1) (𝐿𝐿𝑃3 , 𝐶2𝑃2 ) 5.8
42
43 (𝐻𝐻𝑃3 , 𝐶1𝑃3 ) (2) (𝐻𝐻𝑃3 , 𝐶1𝑃3 ) 1.9
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

1
2
3
4 Table 7. Assignments of new units using Strategy III.
5
6
7 Matches using new heat exchangers in retrofit
Areas of new heat exchangers (m2 )
8 design �𝚤̃𝑝 , 𝚥̃𝑝 �, (𝑖𝑝 , 𝑗𝑞′ ) or (𝑖𝑞 , 𝑗𝑝 )
9
10 (𝐻2𝑃3 , 𝐶1𝑃3 ) 204.2
11
12 (𝐻1𝑃2 , 𝐶2𝑃1 ) 159.2
13
(𝐻1𝑃3 , 𝐶2𝑃2 ) 6
14
15 (𝐻2𝑃3 , 𝐶1𝑃1 ) 20.6
16
17 (𝐻2𝑃3 , 𝐶1𝑃2 ) 8.3
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Acs Iecr 0c03829

Uploaded by

Copyright:

Available Formats

Acs Iecr 0c03829

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

Acs Iecr 0c03829

Uploaded by

Copyright:

Available Formats

Subscriber access provided by CORNELL UNIVERSITY LIBRARY

Process Systems Engineering

is published by the American Chemical Society. 1155 Sixteenth Street N.W.,

24 𝑝 = 1(𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝𝑦𝑖𝑝,𝑗𝑝 = 1(𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝𝑘 ∈ 𝑆𝑇

𝑝 = 1(𝑖𝑝,𝑗𝑝) ∈ 𝑌𝑝𝑦𝑖𝑝,𝑗𝑝 = 1(𝑖𝑝,𝑗𝑝) ∈ 𝑍𝑝𝑘 ∈ 𝑆𝑇

6 (EX1-1) (149.1 m2)

13 P1-C2 30 m 2 153.5 H1P2-

21 P2-C1 109.9 H1P2-

110 110 110 30 30 30

34 (1.9 m2) 0.64

6 (EX1-1) (149.1 m2)

140 140 140 60 60 60

21 P2-C1 110 H1P2-

110 110 110 30 30 30

34 0.75 C1P3 112.7 0.27 C1P2 76.3 246 kW

140 140 140 140 60 60

21 P2-C1 110 H1P2-

110 110 110 110 30 30

34 0.72 C1P3 111.7 0.28 C1P2 76.4 242 kW

35 1.9 m2 H1P3- 54.5 kW

ACS Paragon Plus Environment

ACS Paragon Plus Environment

ACS Paragon Plus Environment

ACS Paragon Plus Environment

ACS Paragon Plus Environment

ACS Paragon Plus Environment

ACS Paragon Plus Environment

ACS Paragon Plus Environment

You might also like