You are on page 1of 22

Targeting Procedures for Energy Savings by Heat Integration across Plants

Hernan ´ Rodera and Miguel J. Bagajewicz
School of Chemical Engineering and Materials Science, University of Oklahoma, Norman, OK 73019

Heat integration across plants can be accomplished either directly using process streams or indirectly using intermediate fluids. By applying pinch analysis to a system of two plants, it is first shown that the heat transfer leading effecti®ely to energy sa®ings occurs at temperature le®els between the pinch points of both plants. In some cases, howe®er, heat transfer in other regions is required to attain maximum sa®ings. A systematic procedure to identify energy-sa®ing targets is discussed, as well as a strategy to determine the minimum number of intermediate-fluid circuits needed to achie®e maximum sa®ings. An MILP problem is proposed to determine the optimum location of the intermediate fluid circuits. The use of steam as intermediate fluid and extensions to a system of more than two plants are discussed.

Introduction
Since the onset of heat integration as a tool for process synthesis, energy-saving methods have been developed for the design of energy-efficient individual plants. Heat integration across plants Žthat is, involving streams from different plants in a complex. always has been considered impractical for various reasons. Among the arguments used is the fact that plants are physically apart from each other and, because of this separation, pumping and piping costs are high. However, an even more powerful argument against integration is the fact that different plants have different startup and shutdown schedules: if integration is done between two plants and one of the plants is put out of service, the other plant may have to resort to an alternative heat-exchanger network to reach its target temperatures. Plants may also operate at different production rates, departing from design conditions and needing additional exchangers to reach desired operating temperatures. All these discouraging aspects of the problem led practitioners and researchers to leave opportunities for heat integration between plants unexplored. Notwithstanding the aforementioned objections, in several practical instances, these savings opportunities are actually implemented either directly using process streams ŽSiirola, 1998; Zecchini, 1997., or indirectly through the use of the steam system in what has been called the steam belt ŽRobertCorrespondence concerning this article should be addressed to M. J. Bagajewicz.

son, 1998.. The first attempt to study the recovery of energy through integration between processes was made by Morton and Linnhoff Ž1984., who considered the overlap of grand composite curves to show the maximum possible heat recovery using steam. Later, Ahmad and Hui Ž1991. extended this concept to direct and indirect heat integration, also based on the overlapping of grand composite curves. In addition, they proposed a systematic approach to generate different heatrecovery schemes for interprocess integration. The concept of ‘‘total site’’ was introduced by Dhole and Linnhoff Ž1992. to describe a set of processes serviced by and linked through a central utility system. Using site-source and site-sink profiles Žbased on the combination of modified grand composite curves of the individual processes., they set targets for the generation and use of steam between processes. However, the elimination of process-to-process heat-exchange zones, also called ‘‘pockets,’’ from the grand composite curves of the individual processes reduces in certain cases the opportunities for energy recovery. This point is discussed in detail in the present article. One of the questions in total-site integration is whether a process fluid should be used to perform the heat transfer or an intermediate fluid should be used. In addition, the question of how to preserve energy efficiency when nonsimultaneous shutdowns take place needs to be addressed. The objective then is to have a dual design where both heat integration and 1721

AIChE Journal

August 1999 Vol. 45, No. 8

independent operation are achievable. In this first approach to the problem, energy targets are established. The nonsimultaneous operation of each of the plants is directly related to the ability of the heat-exchanger network in both plants to operate in either of the other modes. This multiobjective design procedure will be presented in detail in a follow-up article. In principle, direct transfer of heat from one plant to the other may involve many process streams, which results in many heat exchangers. The incentives to use intermediate fluids are the following. ⅷ Multiple pumps and compressors. Integration across plants may require the transferring of heat from a number of streams in one plant to a number of streams in the other. Thus, the cost of integration can be high because of the use of multiple pump and compressor units. ⅷ Pumping and compression costs. A fluid with a larger heat capacity than the process streams will result in a smaller flow rate of liquids to pump among plants. When process streams are gases, then the installation of compressors to cover large distances can be more expensive than the equivalent pumping of an intermediate fluid. ⅷ Safety. Process streams being pumped large distances may pose a hazard, should any spill occur. ⅷ Control. Piping process streams long distances also introduces long delays, which would eventually make process control more difficult. The use of intermediate fluids simplifies the problem. There are, however, some disadvantages worth mentioning. ⅷ The use of an intermediate fluid reduces the interval of effective heat transfer Žthat is, between pinches . by a multiple of the minimum temperature difference Ž ⌬Tmin .. Therefore, compared with the direct integration case, smaller savings can be obtained. ⅷ The number of heat exchangers involved in a setup that uses intermediate fluids can also be higher than using direct heat exchange. In this case, in the absence of other incentives, the trade-off is between the new number of heat exchangers and the pumping costs. In many cases, steam can be used as an intermediate fluid. This offers many possibilities, as the steam system is already in place. Recent work regarding the use of the utility system for the indirect integration of different processes ŽHui and Ahmad, 1994. focuses on the generation and use of steam to reduce utility costs. The present work introduces targets based on fixed steam pressures that are usually available in the plants. In this article, pinch analysis is used to establish target maximum energy savings for either direct or indirect integration of the case of two plants. The article is organized as follows: temperature intervals where heat transfer should take place are identified first, together with the identification of which plant should be the source. An LP problem is set up to determine these targets. The design of the intermediate fluid circuits is considered next. The possibility of using a single intermediate fluid circuit is evaluated. An MILP model is introduced to determine the location of the minimum number of circuits needed to achieve the target savings. Extensions to the problem of integration of a set of n plants are briefly discussed. To illustrate these concepts, examples using heat1722

Figure 1. Possible location of combined plant pinch point.

integration problems from the literature are solved. In addition, an example consisting of the integration between a crude unit and an FCC plant is solved.

Maximum Transferable Heat
Consider two plants and suppose that minimum utility targets are obtained independently for each of the plants using LP transportation or transshipment models ŽCerda et al., 1983; Papoulias and Grossmann, 1983.. When all the streams from both plants are included in the same set Žcombined plant., the total minimum heating and cooling utility targets are usually lower.

Location of the combined-plant pinch
Without loss of generality, assume that plant 2 has a pinch temperature that is higher than the pinch temperature for plant 1. Therefore, the pinch of the combined plant can fall in any of these three regions: Ža. above the pinch of plant 2; Žb. between the pinches; or Žc. below the pinch of plant 1 ŽFigure 1.. Consider first the case in which all intervals above the pinch of plant 2 in each of the individual plants are sinks of heat. If the corresponding intervals of both plants are added, the combined plant will also have sinks in these intervals. Therefore, the system considered has a combined pinch located no higher than the original pinch of plant 2. A similar analysis can be made for the region below the pinch of plant 1. When the preceding conditions are not met, the combined pinch can be located in any place. The implications of this will be investigated further. The first intuitive conclusion one can make is that heat should be transferred at temperatures between the pinch points of the original plants. Indeed, this is the only region where plant 2 is a heat source while plant 1 is a heat sink. This intuitive conclusion is in principle correct, but someAIChE Journal

August 1999 Vol. 45, No. 8

plant 2 does not need to use heat from the intervals between pinches to satisfy any demand below the pinch point of plant 1. Since any heat that is transferred to plant 1 at any temperature interval can be cascaded down to lower temperatures. Finally. no savings can be obtained by transferring heat in the regions above the higher pinch or below the lower pinch. Transfer of heat outside the region between both pinch points Transferring heat either above the higher pinch or below the lower pinch does not decrease the utility usage. Therefore. This Figure 2. say interval i . Figure 2 illustrates the effect of an amount of heat Q A transferred from plant 2 to plant 1 in the upper zone Žwithout loss of generality. Similarly. as some of the utility used by plant 1 is used to satisfy the heat demand of the first i y1 intervals. the amount that can be transferred is constrained by the original utility usage of plant I . If a case is unassisted. above or below both individual plant pinches is required to assist the realization of maximum savings. the combined 1723 Transfer of heat between both pinch points As illustrated in Figure 3. Similar upper limits for the transfer of heat are found if the lower zone is considered. Consider the case in which there are only sink intervals above the pinch point of plant 2 and only source intervals below the pinch point of plant 1. if the transfer is made in the first interval. Likewise. Figure 3. Therefore. Unassisted cases can also take place even though some intervals in plant 1 are sources of heat above the pinch of plant 2 or some intervals in plant 2 are sinks of heat below the pinch of plant 1.. In addition. A special case of unassisted heat transfer across plants is when plant 1 has only sink intervals above the pinch point of plant 2 and only source intervals below the pinch point of plant 1.times heat transfer in the opposite direction Žfrom plant 1 to plant 2. Only an equivalent amount of corresponding utility is shifted from one plant to the other. the amount of heat that can be transferred to plant 1 is not limited by such demand. In conclusion. This is explored next. Because all the intervals above the pinch point of plant 2 are sinks. the upper limit will be smaller. For example. An increase of the heating utility in the plant that releases the heat is followed by a reduction of the same utility in the plant that receives the heat. the real limitation on how much can be transferred is given by the ability of plant 2 to fulfill the demand at each interval. the whole demand of the heat in plant 1 is only satisfied by utility or by plant 2 from the intervals between pinches. the maximum amount that plant 2 can transfer is the resulting available heat it has between pinches.. Effect of transferring heat in the region between both pinch temperatures. The same effect on the cooling needs is observed if a certain amount of heat Q B is transferred from plant 2 to plant 1 in the lower zone ŽFigure 2. a certain amount of heat Q E is transferred from plant 2 to plant 1 between pinch points. 45. intervals are lumped to allow clarity of illustration . a maximum that can be transferred exists. This motivates the following definition: Definition. which is usually the cheapest. Assisted and unassisted heat transfer across plants We now investigate the upper limits in the amount of heat that plant 1 can accept and in the amount that plant 2 can deliver in the region between pinches. transfer has the effect of reducing the heating utility in plant 1 and cooling utility in plant 2. In addition. If the heat is transferred in an interval below the first 1. since all intervals below the pinch point of plant 1 are sources of heat. Smin one. Unassisted heat transfer across plants takes place when only heat transfer between pinches is needed to achieve maximum savings. to compute this upper limit one should subtract all the intervals that are heat sinks I Žnegative values. above the interval of transfer from Smin . it has no effect on the heating utility of plant 2 or the cooling utility of plant 1. AIChE Journal August 1999 Vol. However. Effect of transferring heat outside the region between both pinch temperatures. No. 8 . The maximum heat that plant 1 can receive is the actual sum of the demands it has in the intervals between pinches. transferring heat from plant 2 to plant 1 has the effect of reducing the lowest level of the heating utility demand on plant 1. transfer from plant 1 to plant 2 in either one of these regions is needed in some cases to facilitate the transfer of heat in the region between both pinch temperatures. as the temperature level at which the heat transfer in the upper zone takes place is lowered.

Consequently. Definition. this is an unassisted case. assume such heat sources are enough to produce a surplus that in the absence of integration across plants is effectively transferred in plant 1 through the location of the pinch point of plant 2. reducing the heating utility of plant 2. To prevent such limitation. only shifts utility load from one plant to the other. Furthermore. they do not explore further the significance of the assisted transfer in order to realize maximum savings. Dhole and Linnhoff Ž1992.. Assume now that some of the intervals in plant 1 above the location of plant 2 pinch are sources of heat. 1724 Figure 5. The heat transfer outside the region between pinches does not realize any savings. Indeed. one can transfer the surplus heat from plant 1 to plant 2.Table 1. Some of the surplus from plant 1 below its pinch can eventually be used to satisfy this demand. In order to compare the results obtained using the cascade diagram with methods that make use of grand composite Figure 4. In fact. 8 . Similarly. and allowing maximum heat transfer between pinches. and the actual value of these utilities for each of the plants as well as for the combined plant. The existence of assisted cases has been overlooked by Ahmad and Hui Ž1991. thus freeing the heat from plant 2 to be completely available to realize savings through transfer to plant 1 between pinches. Example of unassisted case Table 1 shows the interval balances. These savings are obtained by subtracting from the sum of the individual utilities the combined utility. the addition of all intervals and the fact that in general there is heat transfer across the location of the pinch of plant 2 indicates that the pinch does not lie above this temperature. the II . the total surplus that can be transferred is limited. Scale 140 120 100 80 60 40 Plant 1 qiI I ␥ iI Smin Plant 2 qiII II ␥ iII Smin Combined Plant qiCP ␥ iCP CP Smin y12 y12 y1 y13 y15 y28 y2 y30 2 y28 30 y19 y19 y1 y20 10 y10 5 1 y5 y4 20 y31 y31 y2 y33 y5 y38 3 y35 3 y32 38 I Wmin II Wmin CP Wmin 2 16 6 Max. A model to predict the exact amount of heat that needs to be transferred in each region is presented later. First. construct site source and site sink profiles based on the combination of modified grand composite curves of the individual processes. Assisted heat transfer across plants takes place when heat transfer between pinches needs to be assisted by heat transfer outside this region to attain maximum savings. These two cases motivate the following definition. Sink intervals are located above the pinch and source intervals are located below the pinch in either plant 1 or plant 2. In turn. if the surplus is larger than the heating utility of plant 2. in the presence of an assisted case. No. Pinch locations are shown with filled lines. the surplus of heat above the pinch of plant 2 needs to be used to satisfy the heat demand of plant 1 between pinches. AIChE Journal August 1999 Vol. In these modified curves process-to-process heat-integration zones or pockets are eliminated. In other words. An exact symmetric case happens below the pinch of plant 1. potential savings between pinches s 12 Max. opportunities for realizing maximum savings are lost and only limited savings between pinches can be pursued. and the surplus may become amount Q A is limited by Smin an effective limitation to realize all the potential for savings. the combined pinch is in between the original plant pinches. and therefore limit the maximum savings that can be obtained. As expected. if the heat demand of plant 2 in the corresponding intervals is not sufficiently large. Countercurrent composite curve profiles for Example 1. The same can be said for the region below the pinch point of plant 1. who only showed that sometimes more than one steam level is required for maximum indirect recovery between processes. Therefore. the heat cascade to determine the utilities. Example 1 Temp. possible savings s 12 pinch point of both plants lies in between pinches. 45. However. Figure 4 shows the cascade diagram solution after the integration is conducted. Cascade diagram solution for Example 1. some illustrative examples are shown. and only transfer between pinches is needed in order to obtain maximum savings. this may limit the amount that can be transferred from plant 2.

is employed. A source interval is located in the region above the pinch of plant 2 Žhigher pinch. the possibility of realizing maximum savings has been lost. to indirectly integrate the total site through the utility system. This is illustrated in Figure 7b. the limitation to obtain maximum potential savings cannot be totally removed. a transfer of the necessary amount from plant 1 to plant 2 above the higher pinch allows maximum potential savings to be realized. Scale 140 120 100 80 60 40 Plant 1 qiI I ␥ iI Smin Plant 2 qiII II ␥ iII Smin Combined Plant qiCP ␥ iCP CP Smin y18 y18 5 y13 20 y19 y19 y1 y20 20 y37 y37 4 y33 37 y5 y18 I Wmin y2 y20 2 y18 2 10 y10 II Wmin 5 y5 1 y4 16 5 y28 CP Wmin 3 y25 3 y22 15 Figure 6. However. in plant 1. Example 3 Temp. In the procedure introduced by Dhole and Linnhoff Ž1992. possible savings s 17 curves. Figure 5 shows the countercurrent profiles for the grand composite curves of the two plants. However. A source interval in plant 1 is found in the region above the pinch of plant 2. No. this is an assisted case. This is a result of the fact that all the limitation for transfer between pinches can be completely removed. Example 2 Temp. potential savings between pinches s 7 Max. since it makes possible the assisted transfer to plant 2 and allows full transfer of heat between pinches. possible savings s 3 August 1999 Vol. The pocket present in the composite curve of plant 1 has not been removed. potential savings between pinches s 17 Max. Therefore. Figure 6 shows the cascade diagram solution after the integration is conducted.Table 2. the approach of Ahmad and Hui Ž1991. AIChE Journal Max. Assisted cases below the two pinches are similar in nature and therefore examples are omitted. 45. 8 1725 . The grand composite curve of plant 1 has been inverted to be able to establish the maximum amount of direct heat transfer. the diagram of Figure 7a is obtained. Table 3 presents the data for Example 3. Figure 8 shows Table 3. When a comparison with the method that uses grand composite curves is performed. This source interval prevents plant 1 from receiving all the potential heat available to be transferred between pinches. as shown. Cascade diagram solution for Example 2. The extent of the maximum possible savings is reached whenever the profiles coincide in a point. a limit imposed by plant 2 arises in the heat that plant 1 can transfer above the higher pinch. Therefore. Thus. whenever the pockets are eliminated. Countercurrent composite curve profiles for Example 2. pockets are eliminated prior to the construction of the site-source and site-sink profiles. Example of assisted cases Table 2 presents the data corresponding to Example 2. The combined pinch lies between pinches. Figure 7. Scale 140 120 100 80 60 40 Plant 1 qiI y7 y5 I ␥ iI Smin Plant 2 qiII II ␥ iII Smin Combined Plant qiCP ␥ iCP CP Smin y7 y2 20 y10 y10 y10 y20 14 10 5 y6 4 9 20 y17 y17 y5 y22 y1 y23 7 y16 8 y18 23 y15 y17 y3 y20 3 y17 I Wmin II Wmin CP Wmin 3 29 15 Max. Unassisted cases are therefore readily tractable with the reported method.

The model that predicts the maximum possible energy savings that effectively occur between pinches Q E . the new heat transferred between intervals after integration between plants is implemented. Countercurrent composite curve profiles for Example 3. mI s mII s m. Figure 9. Cascade diagram solution for Example 3. A single direction of heat transfer is allowed: from plant 2 to plant 1 between pinches and from plant 1 to plant 2 outside this region. the temperature intervals are constructed using inlet and outlet temperatures of all streams from both plants Žthat is. This model predicts the amount of heat that needs to be transferred in each interval to achieve maximum savings. 8 . let qiE be the heat transferred between pinches in interval i from plant 2 to plant 1. the presence of a pocket in plant 1 allows the partial removal of the limitations in the transfer between pinches. Let ␦ and ␦ be the original minimum heating utility of plant 1 and plant 2. In this case. and qiA and qiB the heat transferred in the inverse direction above the higher pinch and below the lower pinch. The task is now to determine what amount is transferred at each interval to achieve maximum savings. an I II ˆ0 ˆ0 LP model is proposed. AIChE Journal August 1999 Vol.. respectively. and the eventual minimum amount of heat Q A and Q B to be transferred Maximum energy sa©ings Maximum savings that can be obtained by integration are computed subtracting the combined plant minimum heating utility from the summation of the individual plant’s minimum heating utilities. when no integration between plants is assumed. Figure 9a shows the composite curves for this example. In order to facilitate the computations. the results obtained by solving the LP transSmin portation or transshipment models for each of the plants sepI II ˆm ˆm arately.. let ␦ and ␦ be the original coolI II ing utilities ŽWmin and Wmin values.Figure 8. In the same way.. a model is constructed where heat can be transferred independently within each interval ŽFigure 10. Finally. Application to either direct or indirect integration is possible. Targeting Model for Heat Integration In this section. Also let ␦ iI and ␦ iII be 1726 Figure 10. The elimination of this pocket prevents the realization of the maximum possible savings as it is shown in Figure 9b. Splitting the heat transfer among intervals. 45. respectively. To obtain the amount of heat that has to be transferred in each interval. No. a model that allows the automatic determination of unassisted and assisted cases is presented. These values are I II and Smin . To do that. the cascade diagram solution after the integration is conducted.

a shift consisting of moving the hot scale of plant 2 Žand with it. p II ᭙ i s Ž p II q1 . Such shifting requires equipment and therefore represents additional investment without a benefit and should be avoided. the fluid circulates in the inverse direction than between pinches. Remark. in the region below the lower pinch. the cold scale too. . qiB G 0. . A shift of both hot and cold scales downward by ⌬Tmin degrees in plant 2 below its pinch. Therefore.. 8 . the cold scale. . namely flow rate and inlet and outlet temperatures. The design parameters for these circuits. has to be performed. This implies that independent reductions of these utilities are not possible. is a useful input for these models. will represent the total possible amount of heat to be transferred between pinches Q E . with it. These models. p I Ž1. This degeneracy is actually a flexibility that can be exploited later when a design is attempted. However. In this case. the cold scale. a shift of the hot scale of plant 1 Žand with it. 1727 August 1999 Vol. smaller savings than in the direct integration case may be achieved. or transferred down in plant 2 first. To reduce the utility in the amount transferred from plant 2 between pinches. qiE . a match between the cold scale of plant 1 and the hot scale of plant 2 is required in these two regions. The LP problem presented has degenerate solutions. . Finally. m Indirect Heat Integration The focus is now on the case of indirect heat integration by the use of intermediate fluid circuits. requiring the coincidence between the respective hot and cold scales. The temperature of the intermediate fluid leaving plant 1 Žregistered in its hot scale. The same result is obtained if one solves the problem minimizing the amount of cooling utility of plant 2. downward ⌬Tmin degrees in the region below its pinch is performed. ferred to plant 1 at a lower interval. should be equal to the starting temperature of the same fluid in plant 2. ␦p II II s 0 ␦p 5 5 5 ᭙ i s 1. ␦ iII . the heat can be transferred from plant 2 to plant 1 first and then transferred down. Therefore. the intermediate fluid acts as a hot stream. As a result of these temperature shifts. To accomplish this. the hot scale of plant 2 Žand with it. s. Therefore. Thus. new streams appear in each plant. . the excess consists of a simple shift of utility from plant 2 to plant 1. 45. andror compression costs. then this reduction in savings potential may or may not be compensated by the reduction in piping. for each unit of heat transferred between plants. . the scale shifts required are: 1. whereas in plant 2. . ..t. the knowledge of what are the intervals at which heat transfer from one plant to the other should take place Žin addition to the direction of such transfer . to transfer heat surplus from an interval in plant 2 to any interval in plant 1 between pinches. Minimizing the utility needed in plant 1 serves two purposes: 1. have to be calculated. I I ˆ0 ␦0 s␦ q Q A y QE II II ˆ0 ␦0 s␦ y QA A I ␦ iI s ␦ iI y 1 q qi y qi II A ␦ iII s ␦ iII y 1 q qi q qi I E ␦ iI s ␦ iI y 1 q qi q qi II E ␦ iII s ␦ iII y 1 q qi y qi B I ␦ iI s ␦ iI y 1 q qi y qi II B ␦ iII s ␦ iII y 1 q qi q qi I I ˆm ␦m s␦ y QB II II ˆm ␦m s␦ q QB y QE I I s 0. In the region above the higher pinch and below the lower pinch. and then transAIChE Journal I II Shift of scales When an intermediate fluid is used. many different paths are available. adding them to form the objective function of problem 1 is possible. Summarizing. If the use of intermediate fluids is not mandatory Ždue to safety or other considerations . its cold scale. the hot scale of plant 2 was already shifted by ⌬Tmin . The problem considers the conditions of minimum utility usage for both plants as the starting point. A simple balance around plant 1 proves that the summation of the solutions Žheat transferred amounts qiE . a shift of 2 ⌬Tmin degrees downward of the hot scale of plant 1 Žand. In plant 1. and it is assumed p I .in the regions outside the one between pinches is I II q ␦m Min Ž ␦ 0 . ␦ iI . . pumping.p II. In particular. the transfer to plant 1 is maximized. The results from the preceding models can now be used as target values for models that will determine the heatexchanger network needed to accomplish such savings. the temperature intervals are constructed using inlet and outlet temperatures of all streams of both plants. . To make sure that the amount of heat transferred from plant 1 to plant 2 is strictly the minimum needed. Consider the region between pinches first. . Hence. In this formulation p and p are the respective pinch point levels. it acts as a cold stream. The same situations occur in an inverse manner when the transfer takes place in any of the regions outside the region between pinches. both values are reduced by the same amount simultaneously. . No. Similarly. When a higher amount of heat than the minimum needed is transferred above pinches in the assisted case. The objective function used needs some explanation. as in the direct integration case. ᭙ i s Ž p I q1 . qiA . is shifted upward ⌬Tmin degrees in the zone above its pinch. 2. downward is needed. while the transfer from plant 1 to plant 2 below both pinches is kept at its minimum necessary to assist in the savings. which will be presented in a follow-up article will address the design of systems featuring the minimum number of exchanger units to accomplish a dual operation Žwith and without integration. . Indeed. Finally. . Now consider the possibility of assisted cases.

carrying heat from plant 1 to plant 2. The use of separate variables for the heat transferred to and from the intermediate fluid are defined as follows: qiFC for the heat received by the fluid in interval i . 8 . 3. The solution to problem 1 can be implemented in practice. AIChE Journal August 1999 Vol. A circuit covering m E intervals between pinches will receive heat from plant 2 and will deliver it to plant 1. as depicted in Figure 11. Ž2. this can be generalized for a set of m K intervals. In the assisted cases. No. problem 1 is solved. Next. 45. Feasibility of a single-fluid circuit Even though the target value for Q E can always be transferred between pinches and the eventual target amounts Q A and Q B can always be transferred outside this region.’’ while the one receiving that heat is considered the ‘‘heat-sink plant. Remark. Let us first explore the second law constraints regarding qiFC and qiFH. A generalization for any of the regions comes next. The plant that is providing the heat is considered the ‘‘heat-source plant. Indeed. where k F is the first interval where the transfer between plants takes place and m K represents the number of intervals covered by a circuit in any of the regions. Likewise. Figure 12a shows the unassisted case with m E intervals within the region between pinch temperatures Tp II and Tp I . However. to establish the regions in which heat transfer should be made to accomplish the overall target. one is interested in performing the transfer with the smallest amount of circuits possible. a circuit can be established for each interval that has a nonzero heat transfer qiE. k Fq mK k Fq mK QF s Ý is k F qiFC s Ý is k F qiFH . the question is whether these transfers can be achieved with a single circuit in each region. that is. and qiFH for the heat released by the fluid in interval i. or qiB. The total heat transferred Q F is already given by the targeting procedure. the heat released from plant 1 in the same interval does not have to be the same as the heat received by plant 2.’’ The following constraints prevent the temperature of the fluid from going 2. A shift of both hot and cold scales downward by 2 ⌬Tmin degrees in plant 1 below its pinch. The issue is investigated in the following sections. A shift of both hot and cold scales upward by ⌬Tmin degrees in plant 2 above its pinch. The assisted cases are shown in Figure 12b. Shift of scales to allow the use of intermediate fluids. In the region between pinches. Therefore. qiA. These shifts create gaps in the scales. Circuits of intermediate fluids in unassisted and assisted cases (gaps are omitted for simplicity). the heat released in each interval from plant 2 to the intermediate fluid does not have 1728 Figure 12. a circuit covering either m A or m B intervals will perform the inverse task. in the region above the higher pinch and the region below the lower pinch. Maximum energy sa©ings The maximum savings that can be obtained by integration can be computed by subtracting the combined plant minimum heating utility using the shifted scales from the summation of the individual plant’s minimum heating utilities.to be the same as the heat released by the fluid to plant 1 in the same interval. Temperature constraints Figure 11.

k Fq mK F Ž T0FH y Tk . The values of qiFC can be transformed to a new set q ˆiFC as follows: q ˆiFC s 0 ᭙ i s k F . qiFH s 0. and Q B are added and subtracted in the three defined zones of the cascade. . Ž8. s Q F Ž . a new solution with an upper temperature smaller than the one considered initially is possible. If this is the case. where q ˆiFC is another degenerate solution. Ý is k F qiFC q ˆiFC s qiFC ᭙ i s kqq2. . Let us now examine what values the initial temperatures of the intermediate fluid can take. Ž3. With these equalities. kq. . Ž k F q m K y1 . The target values Q E . Then by increasing the flow rate of the intermediate fluid and without limitations in the transfer of heat from the heat-source plant. . G Ý is k qiFC ᭙ k s Ž k F q1. . it should be noticed that the following has to be verified: F T0FC s TkFH qm K F T0FH s TkFC y1 . . . F Ž Tk F y 1 y Tk F q m K . G In the heat-sink plant. . . Ž5. Ž 13 . First. Ž 15. the heat-source plant will only be transferring heat to the intermediate fluid at temperatures lower than Tkq. Thus. . T0FC and T0FH are equal to Tky 1 and Tk . Q A. This value then can be replaced in Eqs. the set of Eqs. ᭙ i s k F. 1729 August 1999 Vol. G Ý is k F qiFH ᭙ k s k F . If any of these equations are not satisfied.G Ý is k qiFC ᭙ k s Ž k qm K .. . Adjusted heat-cascaded diagram (gaps are omitted for simplicity). . . Ž 16.higher than the interval temperature Tky 1 in the heat-source plant: k Fq mK source and sink plants are F T0FC s TkFH q m K s Tk F q m K Ž 12. . . F F T0FH y TkFH q m K s QF . . To find such a solution. . note that to guarantee feasibility of heat transfer. F Ž Tk y Tk F q m K . Ž k F q m K . Ž 14. This accounts for the supplies or demand each of the Ž9. No. . for some k Žthat is. m K . Ž7. they are confined to be end-interval temperatures . . intervals. The flow rate F can be calculated using Eq. This solution allows the circuit to be established between the intervals kqq1 and Ž k F q m K . . Now consider the case where the heat-source plant does not have any heat demand in the first set of Ž kqy k F q1. Ž 10. FŽ Tky 1yT0FC . Ž 11.. Ž6. . . . one can assume without loss of generality that the initial temperatures of the intermediate fluid in the heatAIChE Journal Figure 13. 45. that is. Ž k F q m K y1 . Ž k q1 . In order to assure a closed circuit. Equations 14 ᎐ 16 constitute a feasibility test for a single circuit transferring the heat Q F between plants.. 16. kq k FC FC q q ˆk q 1 s qkq q 1 q q Candidate heat-transfer sets The flexibility at hand for defining the general variables qiFH and qiFC is explored by an adjusted heat-cascaded diagram ŽFigure 13. intervals in the heat-source plant. . respectively. A similar argument can be made for the case where the last intervals in plant 2 do not transfer heat to the intermediate fluid. similar constraints are introduced to prevent the temperature of the fluid going lower than the interval temperature Tk : k Ý is k F qiFH ᭙ k s k F . FC F F Ž TkFC y 1 y T0 . . then a circuit between Tk F y 1 and Tk F q m K cannot transfer the maximum amount. F Ž Tk F y 1 y Tk . but perhaps some smaller value. 8 . . . 3 ᎐ 6 become k Ž4. 14 and 15. . . heat can be cascaded down from the first Ž kqy k F q1. . s Q F . F F F T0FH s TkFC y 1 s Tk F y 1 .

If in addition. . H In these equations. . and also the balance equation ŽEq. . ␴ k F s qk H H ˆ q ␦ k F y 1. . . but rather a calculation aid. In addition. The same set of equations can be written in terms of qiEH and qiEC: k I ˆpIII y QE q ␦k s␦ Ý Ž qiI q qiE H . .0 H H H Fqz F qk k q1 q ␭k F G 0 Ž 27 . is an acceptable candidate for a single circuit.zones will experience when the respective single-circuit candidates are considered. any nonnegative set of generalized values qiFH and that satisfies Eqs. the transfer will only occur in the subzone delimited by the real and induced pinches. or 23 ᎐ 24. . This solution is called the higher-circuit solution. the deficit of heat in the intervals below it need to be taken into acH at each interval is then decount. k F q m K y1. Ý Ž qiII q qiB H . To illustrate this. Ž 29. . p I . that is. A lower circuit solution is presented later. . . . G 0 i s p II q 1 Other sets can be found by making use of degeneracy. 1 satisfy the following relations: k I ˆpIII y QE q ␦k s␦ Thus. . The values obtained for the different intervals are not realistic heat-transfer amounts since some of them are negative. Ž 20 . Eqs. . 0 and ␴ kH s Max qk . 1730 Ž 24. m k II ˆpII y QE q ␦k s␦ ␴ kH 12 0 0 15 0 0 20 H H H qk q zk q 1 q ␭k H zk H ␻k Ž 23. that is. . . . k k F q1 F k q2 k F q3 k F q4 k F q5 k F q6 F 0 y1 y1 0 y15 y2 0 8 y5 y4 y2 y32 y4 20 0 y4 y3 y2 y17 y2 0 8 0 0 0 0 0 20 August 1999 Vol. . One candidate set is given by the solution of problem 1. Thus. the candidate set will be a feasible single-circuit solution in any of the zones. consider first the ˆkH F situation depicted in Table 4 for which ␦ y 1 s 0. Therefore. . . . .0 F F H H H F ᭙ k s k q1. G 0 i s p I q1 Table 4. . . G 0 i s p II q 1 ᭙ k s Ž p II q1 . . H F zk qm K s k II ␦k s ½ ½ ½ H H F ␭k q m K q ␴k Fq m K H H F qk q m K q ␭k Fq m K . k II ␦k s Ý Ž qiII y qiE . G 0 i s p II q 1 H zk s I 0 ᭙ k s Ž p q1 . at every interval. . G 0 ᭙ k s Ž p II q1 . . 21 ᎐ 22. . Determination of Heat Availability Interval qiH 12 y1 y1 15 y15 y2 20 H ␭k ᭙ k s Ž p I q1 . one can choose a set that prioritizes the heat transfer to the intermediate fluid over the heat transfer to the interval below in the source plant. p II k I ␦k s Ý Ž qiI y qiB C .0 H H H qk q zk q1 q ␭k G 0 0 H H H zk q1 q ␭k q ␴k Ž 26. . Ý i s p II q 1 Ž qiI q qiE . . G 0 is1 H H H H ␻k s Max qk q zk q1 q ␭k . p II Ž 21. qiFC qiFH s qiFC s qiF . let ␭ k F s 0. z k is an auxiliary variable that helps determine the amount of cumulative demand Žfrom the bottom. because the circuit starts and ends at the higher possible intervals. . k II II ␦k s ␦0 y QA q ˆ ÝŽ is1 qiII q qiA H . . No. . for the assisted cases the relations for the upper and lower zones are: k I ˆ0I q Q A y QE q ␦k s␦ 0 Ý Ž qiI y qiAC . m . Similarly. G0 Ž 22. ᭙ k s 1. . .0 H H F qk q m K q ␭k Fq m K G 0 H H H qk q zk q1 q ␭k . . Ž 28. the solutions of Eq. 14 ᎐ 16 are satisfied. . 8 AIChE Journal . . The maximization of the heat delivered to the intermediate fluid is the purpose of constructing a higher circuit solution. In particular. the availability ␻ k is given by ᭙ k s Ž p q1 . 2. 04 ᭙ k s k F. Ý Ž qiII y qiE C . . . . G 0 i s p II q 1 ᭙ k s Ž p I q1 . . II ᭙ k s k F q1. k Fq mK . 19 ᎐ 20. In this instance. . to establish the maximum amount of heat that each interval can provide to the intermediate fluid. Ž 18 . k q m k . ᭙ k s 1. p II I Ž 19. Moreover. Ž 25. 45. In the unassisted case. . The heat availability ␻ k H H 4 H 4 fined as follows: Let ␭ k s Min qk . . H Fs zk H H H F zk q 1 q ␭k F q ␴k F H H H Fq z F qk k q1 q ␭k F . because of these operations the adjusted heat-cascaded diagram of the heat-source plant may exhibit an induced pinch at each region where a circuit can be installed. . . p I Ž 17. . . p . A few generalized candidate sets are presented next. . .

y 1 y Tk F q m K Ý is k qiEC ᭙ k s k F q1. Transfer less heat from the intermediate fluid to the sink plant at any interval defined by the lower solution. Ž Tk y Tk Ž Tk F F qm K . Thus. . while ⌬Tmin for plant 2 is 10ЊC. FC H qk s Min ␻ k .0 Now let kq be the first interval with nonzero heat transferred from the intermediate fluid to the sink plant. only the minimum amount of heat should be transferred to the intermediate fluid. . Equation 30 states that at the first interval. H2 ŽHot. Q 4 qk k F Ž 30. . C2 ŽCold. On the source side. the higher-circuit solution is given by FC H F s Min ␻ F .0 150 2.08 116 10. 16 in Eqs. Problem 4sp1 C1 ŽCold.9 38 ⌬Tmin s 20ЊC 7. . Test Case ࠻2 from Linnhoff and Hindmarsh Ž1983. 160 93 260 138 270 82 762 589 876 1171 128 250 AIChE Journal August 1999 Vol. The interval between pinches goes from Ž 34. Table 5. S ŽSteam. C4 ŽCold. . by transferring heat to lower intervals in the source plant. . but covering a region that ends somewhere above it. This solution is called the lower-circuit solution because it starts and ends at the lowest intervals possible. that is. 0 . . . . . . FH C qk s Max y␪ k y ½ k y1 Ý is k F qiFH . 180. k q mK G Ž 32.5 20 8. 45. the higher solution is FH F sQ qk F FH qk s0 QF Ž Tk F y 1 y Tk . . QF y ½ k y1 Ý is k F qiFC 5 ᭙ k s k F q1.0 25 ᎏ 270 0. ᭙ k s kq. for the sink plant. Ž 33 . 5 ᭙ k s k F q1. a second test for the feasibility of a single circuit transferring the maximum savings Q F in any of the regions consists of constructing the higher or lower solution and checking if the following equations are satisfied: Ž 31. k G Ž Tk F y 1 y Tk F q m K . Ž k F q m K y1 . In such case.79 160 6. . At the other extreme. . S ŽStream. kq is such that qk s 0. . FH FH q / 0. Ž 39. F Ý is k F qiF H ᭙ k s k F .. As a result of the calculation of the cumulative heat demands. In turn. k F q m K . Table 7 shows the results of the pinch analysis. Data for Example 4 Ž 35. . no one-circuit solution can: 1. Ž kqy1. This minimum should be at least equal to qiFH to guarantee that temperature constraints have a chance of being satisfied. . From then on. Tt ŽЊC. Under such conditions. . C3 ŽCold. 2. k F q m K . a limit for the starting point of the unique circuit is established. qk all the surplus heat in the source plant can be transferred down until interval kq is reached. . all resulting heat flows cascaded down are positive. k F q m K . Ž 37 . Ž kqy1 . ᭙ k s Ž k F q m K .0 225. Example 4 In this example. . is plant 1 and problem 4sp1 is plant 2.62 60 8. the lower-circuit solution for the sink plant is given by FH C F s Max y ␪ F . minimizing the transfer to the intermediate fluid. Note that ⌬Tmin for plant 1 is 20ЊC. . . QF Ž 38 . 0 4 qk k These equations have been obtained by substituting Eq. ᭙ k s k F . Ž 36 . 60 125 60 100 270 82 Q ŽkW.55 249 ᎏ 270 5.0 107.0 90 3. Streams F ŽkW rЊC. Direct Integration Solution. the lower-circuit solution for the source plant is FC s0 qk FC FH qk s qk Test Case ࠻2 H1 ŽHot. . The adjusted cascaded heat values already account for this. H3 ŽHot. No. Then.5 240. CW ŽWater. one must make sure that the plant does not need the heat at the same interval. 14 and 15. In view of the preceding. It should be noted that the lower solution obtained by the preceding procedure is not always feasible. .5 40. Some other lower solutions might exist. then the higher-circuit solution will consist of this interval only. Start at a lower interval. all the heat available will be used provided it is lower than the overall maximum. . The data for the separate plants are shown in Table 5. Pinch temperatures and minimum utility consumption for each of the plants are shown in Table 6. . . we have a solution that maximizes the transfer of heat to the interval below in the heat-source plant. .0 262. . By construction. Indeed. Equation 31 states that at every interval. .68 38 ⌬Tmin s 10ЊC ᭙ k s k F . not necessarily covering the last intervals of the region between pinches. This means that if the sink plant can transfer all the maximum possible heat Q F in the first interval of transfer k F.either a higher circuit between pinches or a circuit above all pinches with an induced pinch. . 2. 8 1731 . the maximum that can be transferred is the surplus. . Ts ŽЊC. . Next. H4 ŽHot. . . the solutions for the sink and the source plants are somewhat related. k F q m K . Ž k F q1 . CW ŽWater.

8 353. 353.5 y67.0 297.5 258.0 145 y3.792 1. the maximum possible heat savings for the direct integration are the original minimum utility of plant 1. Table 10.1 y31.9 289. This solution is feasible. The intermediate circuits between the circuit spanning all intervals and the lowest possible circuit are proven feasible.5 y55.8 127. 225.2 87.6 30.0 y12.2 54. 40.3 W II y62. Note that Eqs.3 353.8 y127. in this case the shift does not have any effect in reducing the amount that can be transferred between pinches. This is also the result obtained by solving problem 1.1 y31.5 y107.3 60 y82.8 127.9 259. ŽkW.3 250. 0 107.2 39.0 93 y10.8 W C P min 149. ŽkW.8 y39. ŽkW. The test of feasibility for a single circuit is applied first to a circuit covering all the intervals between pinches.5 y67.5 kW.0 319.3 54.5 70 45. 45.4 324.5 120 y94.0 y17. Fewer intervals than in the case of direct integration are found.8 y127.0 0 qiC ŽkW.4 107. of Intervals 7 5 2 Tup ŽЊC.0 45 y12. II ␥ iII Smin qiC ŽkW. However. ŽkW.5 128 y4. In this sense. qk kW.5 W II 39.8 127. 90 90 150 F ŽkW rЊC. ␥ iI 0 I Smin ŽkW.5 126 y3.5 6.6 y60. the intermediate solutions between the circuit spanning all intervals and the highest possible circuit are feasible.1 y31.9 y33.5 y127.5 138 y6.5 0 0 0 0 0 0 122. Moreover.0 Table 8.4 min 122. 0 40. This is the consequence of the large availability of heat to transfer that plant 2 has in all the intervals between pinches.0 145 y8. ŽkW. 260 0 I ␥ iI Smin qiII ŽkW. 107.4 I Wmin 0 ŽkW. no solution will be able to start below the limit established by the lower solution.8 y127.8 278.5 28. Individual Plant Pinch Analysis for Example 4 Problem Test case ࠻2 4sp1 Pinch Temp.0 242..1 y31.0 250. The higher-circuit solution is shown in Table 10.6 274. ŽkW.4 259.5 y4.5 y55.8 232. ŽkW.5 27.8 250.0 y2. Indirect Integration Solutions.3 193.0 266.2 y82. qk kW.0 10.0 y16. ŽkW.0 Cooling Utility ŽkW.2 95.4 321.5 y127. 4sp1 Combined Plant CP ␥ iC Smin ŽkW.9 0 0 30.3 193. taking into account the resulting heat-exchanger network and the economic aspects. Some of the Indirect Solutions to Example 4 Solution All intervals Lower circuit Higher circuit No.9 4.5 90 31.8 221. 45.3 y82. 67.5 263.7 353.8 182.5 60 y82.5 116 y91. Table 8 shows the pinch analysis for the indirect integration. The position of the resulting circuit is shown in Figure 14.0 y14.0 y2.4 225. 38 and 39 are satisfied. ‘‘Higher Circuit’’ Solution to Example 4 Test Case ࠻2 4sp1 Q E s 107. the lower-circuit solution is presented in Table 11. ŽkW.5 120 y14.2 36. the problem has a large finite number of possible solutions that require further analysis.0 y91. Pinch Analysis for Direct Integration in Example 4 Test Case ࠻2 T ŽЊC. Therefore.5 Table 9.4 38. y175.0 122.5 rameters Žending temperatures and rate ᎐ heat-capacity product. ␻ k kW.0 0. y71.5 ŽkW.5 y58.2 min 83 103.5 28.5 y97.8 y127.4 ␥ iII 107.3 193.8 127. that is 107.5 56. which is equal to the maximum possible savings that can be obtained either with direct or indirect integration. Pinch Analysis for Indirect Integration in Example 4 Test Case ࠻2 T qiI ŽЊC.8 y39.0 0 45 y12.5 1.4 317.8 ŽkW.5 y76.0 y107.6 30.0 317.8 122.2 288.3 276.3 y33. 239 150 239 Tdown ŽЊC.2 36.8 222. and Table 9 shows the values obtained for the pa- 239 0 0 160 0 0 150 10. Table 8 shows that there is no demand in the upper intervals of plant 1.6 y22.3 y19.5 kW.7 182. and the shift does not appreciably decrease the large availability of heat in plant 2 in the region between pinches.721 1.0 31.7 ŽkW. An implementation of this indirect integration follows.9 7.5 122.0 W I 8.5 40.7 min 122. and its position is shown in Figure 15. 0. the resulting combined pinch is located at 249ЊC Župper limit of the interval between pinches . 90 249 Heating Utility ŽkW. No.0 0 297.2 250.5 kW I Ž EH Ž II Ž II Ž EC Ž Interval qk kW.6 167. 270 0 qiI 4sp1 qiII Combined Plant II Smin CP ␥ iC Smin ŽkW.5 122. ŽkW.208 Table 7.0 y62.5 56.4 28. ŽЊC. The combined pinch then results at the upper bound Ž239ЊC.5 90 90. ŽkW.5 200..3 y43. Therefore. qk kW.6 0 28.5 90ЊC to 249ЊC. Finally.0 y12.0 y64. p II q1 s 2 p II q2 s 3 p II q3 s 4 p II q4 s 5 p II q5 s 6 p II q6 s 7 p II q7 s 8 0 0 10 y8.5 1732 August 1999 Vol. After considering all streams in a single set.8 y127.1 y31. the interval between pinches is from 239ЊC to 90ЊC. The solution of problem 1 is Q F s 107.5 W C P min 149.2 ŽkW. After the hot temperature scale in plant 2 is shifted down 10ЊC.Table 6.0 107.9 y152..3 ŽkW.0 10. 67.8 250.2 95. since some of the extreme temperatures now coincide due to the shift.3 y19. 8 AIChE Journal .3 193. ŽkW. Other solutions can be found each time that a certain amount of heat could be cascaded and the one-circuit solutions that result are feasible.2 278.5 128.5 0 0 0 0 0 0 353.5 38.2 104.0 27.6 225.8 y127.4 181. This amount of heat is sufficient to supply the entire demand of plant 1.5 249 0 0 170 0 0 160 0 0 150 10.8 225.

␪ k kW. C4 ŽCold. are 38% lower than the savings from direct integration. an example taken from Trivedi Ž1988. The maximum possible savings for the direct integration are 104. Table 12. Pinch temperatures and minimum utility consumption for each of the plants are shown in Table 13.5 y4. qk kW. the resulting combined pinch is located at 200ЊC Župper limit of the interval between pinches .5 y107.4 0 0 0 0 2. is to be transferred in the zone above both pinches to attain the maximum possible savings. The interval between pinches is from 160ЊC to 180ЊC Žhot scale of plant 1. Lower single-circuit solution for Example 4. 404. Higher single-circuit solution for Example 4. Tt ŽЊC.0 353. Solving problem 1 gives the targeting values of the heat to be transferred in each of the zones.9 kW Ž Q A .7 kW Ž Q A . CW ŽWater.500 600 2. Direct Integration Solution. After considering all streams in a single set.4 kW.0 91. 8 1733 . In this case..0 585. qk kW. S ŽSteam.2 744. C5 ŽCold. Again. 351.2 kW. the combined pinch is at the upper bound of this interval Ž180ЊC. H2 ŽHot. Solving problem 1 gives the targeting values of the heat to be transferred in each of the zones. The interval between pinches goes from 160ЊC to 200ЊC.980. is plant 2. No.000 3.0 91.0 404.000 3. H6 ŽHot. Individual Plant Pinch Analysis for Example 5 Problem Trivedi Ciric and Floudas Pinch Temp.5 28.5 y2. S ŽSteam.0 y14. is plant 1 and example 1 from Ciric and Floudas Ž1991.144 96 7. 65.5 kW I Ž I Ž EH Ž II Ž EC Ž Interval qk kW.0 271 9. Figure 15.1 y31.2 95. 45.8 688.000 5.429. Ciric and Floudas H5 ŽHot.2 1. while above the pinch the same scale is shifted up 20ЊC. C2 ŽCold.Figure 14. F ŽkW rЊC.0 0 0 10 1.032 160 8.44 249 11. Table 14 shows the results of the pinch analysis.0 y91. Ts ŽЊC.0 200 100 270 190 180 300 90 1.7 875. ‘‘Lower Circuit’’ Solution to Example 4 Test Case ࠻2 4sp1 Q E s 107.8 1. C1 ŽCold. The data for the separate plants are shown in Table 12. and no integration is possible in this zone. p II q1 s 2 p II q2 s 3 p II q3 s 4 p II q4 s 5 p II q5 s 6 p II q6 s 7 p II q7 s 8 0 0 10 y8.43 70 ⌬Tmin s 20ЊC 10 300 120 200 15 70 25 70 50 70 ᎏ 300 105 70 ⌬Tmin s 20ЊC Table 11..8 600 Cooling Utility ŽkW. A minimum of 13. Data for Example 5 Streams Tri®edi H1 ŽHot. A gap of 40ЊC is then created in plant 2. 688.5 y16. has to be transferred in the zone above both pinches in order to attain the maximum possible savings. 110 138 106 146 160 217 250 300 90 Q ŽkW.3 y19.8 y39. The maximum possible savings from indirect integration.6 2. Indirect Integration Solutions. Table 15 shows the pinch analysis for the indirect integration.6 7.5 0 0 0 0 2. C3 ŽCold.100 Table 13. C6 ŽCold. Example 5 In this case. This represents 26% of the amount to be transferred in the direct case.6 30.000 12. 160 200 Heating Utility ŽkW. H3 ŽHot.296 115 18 140 ᎏ 300 34. CW ŽWater. qk kW. H4 ŽHot.5 14. the hot temperature scale in plant 2 below its pinch is shifted down 20ЊC.6 936. ŽЊC.100 AIChE Journal August 1999 Vol..816 227 7. A minimum of 52.5 14.

0 y390.2 195.0 120.5 283. 0 0 7 y224 y254.1 An implementation of this indirect integration follows. 0 0 7.4 y261.684.0 y519.5 2. Since an induced pinch appears in plant 1.4 I Wmin ŽkW.920.8 2.2 y404.620.2 y143.8 283. ␥ iI ŽkW.0 y119.3 0 0 Ciric and Floudas I Smin ŽkW.4 293.0 240.0 2.8 97.7 y148.191.0 1.4 2.0 600. No.0 qiC ŽkW.0 y3.0 y68.500.400.3 780.100.4 639.0 y900.1 47.8 qiII ŽkW.0 1.0 240.0 420.6 39.0 5.2 y61. II Wmin ŽkW.0 1.3 0 0 Ciric and Floudas I Smin ŽkW.8 689.0 y60.0 y90.4 113.6 2.4 113.0 600.0 ␥ iC ŽkW.0 y165. 0 0 7.9 86. 404.0 y329.0 y429. 100. the solution for the indirect integration in Example 5 is shown in Figure 16. In this case.0 y231.289. CP Wmin ŽkW.043.0 y30. ␥ iI ŽkW.0 570.903.7 y280. 100.0 y95. Once this amount is trans- ferred.0 y205. 0 0 7 y224 y254. 688.0 7.0 y900.07 y98.0 y336.3 kW.5 1.Table 14.8 1.020. 100. Combined Plant II Smin ŽkW.7 249.4 y143.0 y5.3 180.0 660.350.0 420.320.0 1.0 y105.0 2.0 780.0 y60.0 y195. Pinch Analysis for Direct Integration in Example 5 Trivedi T ŽЊC.3 1734 August 1999 Vol.7 y353.3 y620.0 90. 100.0 y100.5 283.9 326.2 y61. Then.0 123.0 2.8 283.8 qiII ŽkW.0 y68.0 y231.0 qiC ŽkW.7 y568.0 900.3 y339.6 33.0 y336.0 180.8 1.0 y35.0 y93. CP Wmin ŽkW.8 CP Smin ŽkW. 100.0 2.8 CP Smin ŽkW.7 y743.723.0 2.0 y29.100. II Wmin ŽkW.7 kW is required in the upper zone.3 y939. a single interval is left for the transfer of the amount Q A.0 2. 100.0 0.4 236.683.9 86.8 283.8 265. 8 AIChE Journal .8 y155.260.0 600.0 GAP 0.0 2.0 0.0 y600.0 180.0 y419.2 184.0 y600.1 47.443.0 1. a circuit is established in the interval between pinches.0 1.8 2.6 y261.0 y900.8 600.0 y5.0 y300.7 78.4 1.0 90.0 ␥ iC ŽkW.0 y95.0 1.0 y900. Use of Composite Cur®es.0 y90.4 y261.783.0 1.8 y155.7 249.0 1.2 y404. 300 290 271 270 249 237 227 210 200 180 160 146 138 135 116 110 106 100 qiI ŽkW.2 y404.0 y105.100.0 ␥ iII ŽkW.7 y69. For comparison.3 78.503.0 y50. 404. the method that uses countercurrent profiles for the grand composite curves is also applied to the direct integration of the plants ŽFigure 17a.2 754.8 7.8 283.8 1.0 y90.8 97.0 ␥ iII ŽkW.7 y323.0 750.0 y300.4 456. Only intervals above plant 1 pinch are shown since these intervals include the two zones of interest Župper and between pinches . Combined Plant II Smin ŽkW. 45.683.0 120.783.0 2.7 y623.220.400.0 1.0 570.0 591.0 y300.0 y195. This circuit will transfer 65.2 184. 688.500.0 y95.0 1.0 y300.2 456.8 7.0 y265. 100.6 y51..6 I Wmin ŽkW.4 236.0 y215.4 293. 100.0 y95. a coincidence in the higher and lower circuits is found. Table 16 shows the adjusted cascade heats.500.0 y85.1 167. 300 310 271 270 249 237 230 227 220 180 160 146 138 135 116 110 106 80 qiI ŽkW.336.0 600.0 939.1 167.0 5.6 13.2 865.0 y900.3 19.6 y196.0 123.0 420.8 600. A pocket in plant 1 allows the transfer of the required amount of heat above pinches that makes possible Table 15.0 2.2 754.9 326.0 y210.3 y320.7 y353.0 180. A transfer of 13.7 y104.6 y261.0 y105.736.6 2. Pinch Analysis for Indirect Integration in Example 5 Trivedi T ŽЊC..3 669.3 669.

3 2.86 2.5ЊC.6945 127. for both plants. Tt ŽЊC.8968 15. H6 ŽHot.0 y29.2 107.6ЊC Žequivalent to 10ЊF.78 89. C3 ŽCold. an amount of heat can be transferred from plant 1 to plant 2 inside the pocket to allow the realization of the maximum savings for the indirect integration.3 360..0252 348.93 12.96 0.7 353.64 77. H3 ŽHot.0 y210.6 29.73 7.4 0.0 y30.3 100. The data for the separate plants are shown in Table 17. 127.9 0.3 381.0 y105. 45.3 129.0 0. No.8 37. Direct Integration Solution. H5 ŽHot.5 232.0 y90.4 y143. due to the high tem- Table 17.73 1.98 9.7 y69.67 32.3 0.6ЊC 0. C2 ŽCold.2856 427. The vertical line in the profile of plant 2 represents the gap where no integration is possible.1 9. Countercurrent composite curve profiles for Example 5.86 35. The FCC unit needs a large amount of cold utility below its pinch.4 0.3 0.6ЊC.6ЊC 532.2 0.3 79.3 352.8 26.1 0.46 5.07 33.0334 326. Data for Example 6 Streams Crude Unit C1 ŽCold. and the downward shift of plant 2 during intermediate fluid integration is also 5.8 426.9990 15. H11 ŽHot.3 124.0 y35.0362 313.64 26.1503 190. Still.7 78.2 0.3 196. 8 1735 .6230 30. The ⌬Tmin in this case is 5. H12 ŽHot. 60.6 142. H2 ŽHot.1221 336. H7 ŽHot.3 239.7 5.32 Figure 16.18 9.2 0. while the FCC unit is plant 2. I ␪k ŽkW.7 Q ŽMW.2 0. Pinch temperatures and minimum utility consumption are shown in Table 18 and is the result of solving problem 1 for each of the plants.6 ⌬Tmin s 5.000 bblrd Ž64 MLrd.3 321.6 0 13. AIChE Journal August 1999 Vol.20 14.1ЊC to 143.5 0.3 491.1771 194. Considerable energy recovery is possible due to the large temperature difference between pinches Ž471.78 69.3 286.1439 261.7 ᎏ 600. Figure 17b shows how the same method can be used for the indirect transfer of heat between the plants. H1 ŽHot. Ts ŽЊC. II qk ŽkW. Figure 17.6 ⌬Tmin s 5.7 0.1262 538.0 GAP 600.7855 239. H8 ŽHot.5 0.59 5. and an FCC plant processing 40.0 y60.2744 163.0 y5.24 5. CW ŽWater.0831 471.85 22.0773 147.0 686.Table 16.07 2.0083 348.05 9. H15 ŽHot.3 0. H10 ŽHot.3 0.3 0. The crude unit is plant 1.7 37. 0 0 7.3 239. F ŽFuel. Assisted-circuit solution for Example 5.6 206.3 486.3 98. FCC Unit C4 ŽCold.3400 347.2 0.3 y13. F ŽFuel.3053 143.0078 243.6 13.2617 261.. Adjusted Cascaded Heats for Example 5 Trivedi Interval 1 2 3 4 5 6 7 8 9 10 I qk Ciric and Floudas ŽkW.2 21. CW ŽWater. H14 ŽHot.0655 127.0 the realization of maximum savings. H9 ŽHot. II Ž ␪k kW.42 11.7 26.2 537.000 bblrd Ž24 MLrd. Example 6 This example consists of a crude unit processing 150.9 115.9 37. H4 ŽHot.0 y231.8 268.8 26. H13 ŽHot.1 48.1 21. F ŽMW rЊC.0 y195.7 ᎏ y65.3 353.

1 y5.6 y6.7 y38. the crude unit needs a large amount of hot utility in order to heat up its streams during the fractionation process.1 y8.4 y21. 0.5 0.3 y3.8 9. ŽMW.9 y0.5 y42.3 y5. ŽMW.1 0.2 y42. 143.5 y55.4 5.9 y44.9 y3.0 163.1 0.2 y5.8 y55.1 y4.4 y40.1 y5.8 y13. and 1.1 3.5 y11.6 CP Wmin ŽkW.4 y57.0 y57. 163.6 0 y8.9 y42. 526.1 y0.6 2.0 CP Wmin 0.9 y1. ŽMW.2 MW smaller than the maximum possible heat found for the direct integration. II Wmin ŽkW.2 0.0 0.9 y2.3 0.5 y9.1 y0.1 y60.5 y57.3 0. ŽЊC.0 11.0 5.1 y13. 8 .4 y51.8 24.8 y2.1 y5.4 0.6 y3.8 2.9 0.9 y15.2 y5. This amount is 1. Table 20 shows the results of the pinch analysis for the indirect integration.1 y58.1 y5.3 0.4 244.8 26.6 y5.9 y8.8 ŽkW.8 y59. Pinch Analysis for Indirect Integration in Example 6 Crude Unit FCC Unit Combined Plant I II CP T qiI ␥ iI Smin qiII ␥ iII Smin qiC ␥ iC Smin ŽЊC.4 y7.8 69. After the hot temperature scale in plant 2 is shifted down 18ЊC.3 y4.8 y41. Higher and lower candidate single-circuit solutions for Example 6.9 y41.6ЊC to 143.8 y3. ŽMW.5 184.7 y0.2 268.8 y59.5 y67.7 y0.5 y3.5 3.7 y33.2 y13.7 0.7 206.8 2. ŽMW.1 y5. 143.3 y41.7 5. ŽMW. ŽMW. 532.6 8.8 y2.0 y4.8 y10.5 y2.6 y0.2 1.0 y13.1 y41.0 33.5 y9.7 268. ŽMW.9 3.9 243.3 y2.2 33.2 347.4 0.3ЊC.0 28.0 0.1 y5.2 Figure 18.7 9.7 y0.7 y16.9 y13.3 342.6 y2.0 0. The maximum possible savings are now 13.0 y57.5 y1.3 194.2 0.1 3.9 1.3 0.1 MW.0 y56.9 28.6 1.3 y33.0 y48.1 0.1 y41.8 69.2 y6.9 y18. On the other hand.9 y13.9 0.9 y2.8 y22.2 0 y8. The resulting combined pinch is located at 163.6 y67.6 y48.9 y3.1 0.8 y41.7 465.6 28.5 y38.1 y0.5 190.1 y53.2 y34.6 7.3 y1.7 y4. This Table 19.5 471.6 11. the heat that plant 2 has available in the rest of the intervals between pinches is not sufficient to supply the demand of the corresponding intervals of plant 1. Table 19 shows the results of the pinch analysis for the direct integration.1 0 0 0.Table 18.1 y5. the interval between pinches is from 465.2 0.4 y5.7 y38.1 y5. ŽMW.5 y5.9 239.1 y4. AIChE Journal 1736 August 1999 Vol.8 y3.8 y20. Without loss of generality.3 y54.7 y3.9 MW.1 ŽkW.2 y0.3 y3. ŽMW.6 y34.7 307. Note that 13.8 MW are transferred above the combined pinch. ŽMW. ŽMW.1 Heating Utility ŽMW. After this interval.4 4.3 326.3 15.9 y4.9 y38.3 MW below the combined pinch. and is the result of a compensation for the heat in the first interval of plant 1 by heat provided by plant 2.8 y4.6 y0.2 5.5 1. 45.3 ŽkW.5 0.7 y16.2 16.9 y6.1 I Wmin ŽkW.7 y4.9 y60.3 y0.7 y56. some of the intervals have been lumped to clarify the illustration in the upper part of the zone between pinches and below the pinch of plant 1.6 y21.4 y3.3 261.4 y3.0 y41.8 0.4 y4.2 y24.1 y5. 0.1 0 y8.1 1.8 y45.1 y3.9 y18.1 Cooling Utility ŽMW.0 y57.2 y69.1 60.7 0.0 y5.3 y1.1 10.5ЊC. 10.1 y2.1 348.8 y0.3 y26. Pinch Analysis for Direct Integration in Example 6 Crude Unit FCC Unit Combined Plant I II CP T qiI ␥ iI Smin qiII ␥ iII Smin qiC ␥ iC Smin ŽЊC.0 y5.7 3.1 10.8 232.3 226.3 Figure 18 shows the result of the higher-circuit solution.8 y51.3 y69.8 238.1 y8.6 y1.3 336.2 26.5 0.2 y68.6 347.3 326.2 206.4 3.1 59. 69.1 y8.7 336.1 0.8 0.1 y68.0 11.5 1. 2.1 y1.2 3. ŽMW. Indirect Integration.1 y3. ŽMW.8 y42.0 y4.1 y0.2 y1. ŽMW.3 194.1 peratures of the streams emanating from the reactor. The intervals below the lower pinch have been merged since heat recovery is not performed here.6 143.1 0 y8.1 y2.3 0.1 y59.8 0. Therefore. the maximum possible heat savings are 15.6 y41.2 y68.8 29. No.5 0. ŽMW. Individual Plant Pinch Analysis for Example 6 Plant Crude unit FCC unit Pinch Temp.2 33.5 142.9 y30.2 y0.9 1. ŽMW.9 239.3 Table 20.4 244.7 y4. ŽMW.8 II Wmin 12.3 2.3 261.0 y3.8 I Wmin 147. 13.4 y30.7 313.2 471.5 y44.

j s I.t. . j s I. Ž k q m E y1 . Ž k B q mB . ᭙ k s Ž k B q1 . II I I ˆm ␦m s␦ II II ˆm ␦m s␦ y QE I I s 0. . Ž k E q m E . Ž k A q mA . . The maximum amount of heat transferred by a single circuit. . . . . II s. qiEH . for which its location Žstarting and ending intervals. . . Consider the unassisted case first. . . I I ˆ0 ␦0 s␦ y QE II II ˆ0 ␦0 s␦ j j ␦ ij s ␦ iy 1 q qi I EH ␦ iI s ␦ iI y 1 q qi q qi II EC ␦ iII s ␦ iII y 1 q qi y qi j j ␦ ij s ␦ iy 1 q qi I BC ␦ iI s ␦ iI y 1 q qi q qi II BH qiII s ␦ iII y 1 q qi y qi j j ␦ ij s ␦ iy 1 q qi 5 ᭙ i s k B . . . is known. can be obtained by solving the following problem: I II q ␦m Min Ž ␦ 0 . . In this case. The lower-circuit solution is also shown in Figure 18. . m . . . . . . . Ž k B q m B y1 . Ý is k E k Eq mE k Eq mE FA qiEH Ý is k k Ý is k k Ž 43. . ␦ iI . For assisted cases. Ž k E q mE . II j s I. . one can still consider establishing a single circuit and realize only a portion of these total savings. II 5 ᭙ i s k A . Ž k A y1 . j s I. a single circuit will transfer a smaller amount of heat than the maximum Q E . . . . . ᭙ i s Ž k A q m A q1 . . qiEC G 0. Ž k B q m B . . . . 8 1737 . . The equalities that correspond to the heat balances in each interval have been split into two sets of equalities. . . In this last interval the value of the maximum possible heat Q E is reached. . . Ž 42. the test with Eqs. . . 45. ᭙ i s Ž k E q m E q1 . . The second set consists of the intervals in which the transfer is taking place. No. . . where a lower amount of heat is transferred. Ž k E y1 . . . ␦ iII . Ž k E y1 . ᭙ i s 1. . This is determined next. . . Then. . FE E Ý is k k q mE ⌬Ti G E Ý is k k q mE k Bq mB k Bq mB Ý is k Ý is k FE Ý is k E ⌬Ti s Ý is k E k Bq mB k Bq mB Ý is k ⌬Ti G ⌬Ti s Ý is k qiBH qiBH . . . Ž k B y1 . The first one considers only those intervals in which all the heat cascades down. . where ⌬Ti s Tiy 1 y Ti . . . . AIChE Journal k Bq mB k Bq mB Ý is k B Ý is k B Ž 44 . . k Aq mA k Aq mA FA Ý is k A Ý is k k Aq mA Ž 40. . Ž k A q m A . August 1999 Vol. FA qiEH ᭙ k s k . In addition to these equations. . . . . Ž 41. . Ž k A q m A y1 . II j j ␦ ij s ␦ iy 1 q qi ᭙ i s Ž k E q m E q1 . ᭙ i s 1. j s I. . . 38 and 39 fails for both candidate solutions. FB qiEC FB FB Ý is k B ⌬Ti G ⌬Ti s B Ý is k B qiBC qiBc B ᭙ k s k B . the following two new sets of constraints have to be added in order to account for the feasibility of a single circuit in each of the aforementioned regions: k k 5 ᭙ i s k E . j s I. . . ␦p II II s 0 ␦p FA Ý is k A ⌬Ti G ⌬Ti s Ý is k A qiAC qiAC A ᭙ k s k A . m . E E k Aq mA k k FE Ý is k E ⌬Ti G Ý is k k Aq mA ⌬Ti G ⌬Ti s A Ý is k k Aq mA qiA H qiA H A ᭙ k s Ž k A q1 . Only four intervals are required to transfer the maximum possible heat. II. . . . the set of equations for the balances in each interval in problem 40 are replaced by the following new sets: j j ␦ ij s ␦ iy 1 q qi I AC ␦ iI s ␦ iI y 1 q qi q qi II AH ␦ iII s ␦ iII y 1 q qi y qi j j ␦ ij s ␦ iy 1 q qi Maximum Amount Transferred by a Single Circuit In the case where a single circuit cannot realize the entire target savings.candidate solution covers all the intervals and in plant 2 transfers all the heat in each interval except in the last one. . . This problem is linear and offers no major difficulties. . FE Ý is k E k Eq mE ⌬Ti s Ý is k E k Eq mE FB qiEC ᭙ k s Ž k E q1 . . ᭙ i s Ž k B q m B q1 .

YpFH F II located at a low level. AIChE Journal 1738 August 1999 Vol. 5 ᭙ i s Ž p IIq1 . Ž p Iy1 . . Then II s 1 YpFC Ý i s p q1 II ZiE⌬Ti s Ý i s p q1 II Ž 49. . . p I ᭙ i s Ž p Iq1 . Z p q 1 s 0. . . qiEC . . II s 0 and the first interval of transfer will be otherwise. . Yi FH s Yi FC s ½ ½ 1 0 1 0 Interval Ž i q1 . . FE FE Ý ZiE⌬Ti G Ý qiEC is k is k pI pI ᭙ k s Ž p II q2 .Optimum Location of a Single Circuit Problem 40 is expressed for known fixed starting and ending intervals. . Let’s now consider that the third interval is the starting one. ZiE G 0 Yi EH . 1 4 . qiEH . . The case of a circuit between pinches is presented. . . I I ˆm ␦m s␦ II II ˆm ␦m s␦ y QE In addition. . . ␦ iII . . . . . Yi EC g 0. . . . For example. . the following variables are defined: ZiF s I I s0 ␦p II II s 0 ␦p ½ 1 0 Interval i is in the circuit Otherwise . An MILP formulation to determine the optimum points of insertion of such a circuit in any of the regions is presented next. p I qiEH y UZ iE F 0 qiEC y UZ iE F 0 E EH II Zp q 1 s Yp II EH EC ZiE s ZiE y 1 q Yiy 1 y Yiy 1 and F FH II Zp q 1 s Yp II s 1. . Consider the following general binary variables: Then II YpFC q 5 s 1 and FC F F FH II Zp q 6 s Z p II q 5 q Yp II q 5 y Yp II q 5 s 1q0y1 s 0. . . consider a circuit starting in the first interval below the pinch temperature of plant 2. . . Then ᭙ i s Ž p II q2 . YpFC must be zero in the interval in which YpFH is one in order for the circuit to span at least an interval. I I ˆ0 ␦0 s␦ y QE II II ˆ0 ␦0 s␦ j j ␦ ij s ␦ iy 1 q qi I EH ␦ iI s ␦ iI y 1 q qi q qi II EC ␦ iII s ␦ iII y 1 q qi y qi j j ␦ ij s ␦ iy 1 q qi To guarantee that only an interval is a startingrending one. ᭙ i s Ž p IIq1 . . there will not be a transfer in the sixth interval. II Ý i s p II pI Yi FH s 1 Ž 45 . An optimization problem based on the preceding binary variables to solve the unassisted case is then proposed. . k k These variables are related to Yi FH and Yi FC by the following equalities: F FH II Zp q 1 s Yp II FH FC ZiF s ZiF y 1 q Yiy 1 y Yiy 1 FE Ý i s p IIq1 pI ZiE⌬Ti G Ý i s p IIq1 pI qiEH ᭙ k s Ž p IIq1 . p II . . FE Ý i s p q1 pI II ZiE⌬Ti s pI Ý i s p q1 II qiEH ᭙ i s Ž p II q2 . . . . p I p I y1 Ý II YpFC q 2 s 1 and Yi EH s 1 II is p pI F F FH FC II Zp q 3 s Z p II q 2 q Yp II q 2 y Yp II q 2 s 0q1q0 s 1. m . II Ý i s p II q 1 Yi FC s 1. p I qiEC These equalities are needed to restrict the values of the heat transferred to and from the intermediate fluid to zero for intervals that are not in the circuit. . . j s I. That is. Ý i s p IIq1 Yi EC s 1 II II In any case. Therefore.t. p I ᭙ i s Ž p IIq1 . 8 . . . . . . the following inequalities are introduced: p I y1 ᭙ i s 1. ␦ iI . Finally. I II Min Ž ␦ 0 q ␦m . s. is the starting interval Otherwise Interval i is the ending interval Otherwise . Ž 47. . p I . . 45. consider that the circuit ends in the fifth interval. Ž 48 . . Ž 46 . No. . j s I.

. of Intervals 4 3 AIChE Journal August 1999 Vol. l y UZ i . Žl. The following constraints are introduced to eliminate this nonlinearity: j j ␦ ij s ␦ iy 1 q qi I I ˆm ␦m s␦ II II Žl. 51 is compared with the maximum heat possible to be transferred. p I E EH EC ZiE . y Ž 1y ZiE . 163. k k y Bi . l s Z iy 1 . l .In these equations. U is an upper bound of the total heat that can be transferred. . which represents 91% of the total possible savings predicted by problem 1. . FE EC Ý ZiE . p I ls1 n II ␦ iII s ␦ iII y 1 q qi y Ý ls1 C qiE . a minimum of two circuits is obtained. p I p y1 I Ý i s p II pI Yi EH . qiEH . . m . .. j s I. l y UZ i . 226. l G 0 EC Yi EH . a step-by-step increase of the number of circuits seems a logical procedure to reach the minimum required. .l y Bi . . l g 0.l ß Tup ŽЊC.293 The strategy to find the optimum number of circuits consists of a trial procedure.. . Assisted cases can also be solved by introducing similar constraints in the appropriate temperature intervals. . p II .l I ␦ iI s ␦ iI y 1 q qi q ¶ • ᭙ i s Ž p II q1 . . Yi . Z i . The two optimum solutions found are shown in Table 21 and Figure 19. l ᭙ i s Ž p II q1 . .6 MW. . Any of the circuits is capable of transferring 12.7 206.l Example 6 (continued) The linear formulation of problem 49 by introducing Eq. qi . . ␦ iII . . ˆ0 ␦0 q␦ y QŽ E II II ˆ0 ␦0 s␦ j j ␦ ij s ␦ iy 1 q qi ᭙ i s 1. Single Circuit Solutions to Example 6 Solution Optimum ࠻1 Optimum ࠻2 No. . 1739 Table 21. . This is a mixed-integer nonlinear problem having a single nonlinearity that consists of the product of a continuous variable times a binary variable. . If the difference is not zero. 49 is proposed in order to find the locaton of a number n of circuits: I II q ␦m Min Ž ␦ 0 . Ž p I y1 . . Example 6 (continued) If the formulation just presented is applied to Example 6. 0. Since a single circuit is not capable of transferring the maximum possible heat. FE Ý i s p q1 pI II ZiE . p I p Žl. . Žl. FE ŽF E ¶ H qiE . l ⌬ Ti s pI is k Ý i s p q1 II H qiE . . . The MILP model obtained is solved with the CPLEX solver. Then at each step the optimal location of an increasing number of circuits is to be found maximizing the overall heat transfer. . At each step. 45. 1 4 . l Optimum Location of Many Circuits When a single circuit is not capable of realizing the maximum target savings.3 F ŽMW rЊC. The following modification of Eq. l is k ᭙ k s Ž p II q2 . ᭙ i s Ž p II q2 . . . to problem 51 is shown in Figure 20. I I l. 50 is implemented in GAMS ŽBrooke et al. ¢ Bi y ZiE ⍀ F 0 Bi G 0 E Ž 51. II ° Bi s F EZiE FE ZiE ¶ •m~ Ž F G0 ß s Ž 0. l ⌬ Ti s Ý i s p II q 1 C qiE . .3 163. . . ⍀F0 Ž 50. 1 .t. Žl. p I ᭙ i s Ž p II q1 . l s1 ß EC E ␦ iI .3 Tdown ŽЊC. . As illustrated. l ⌬ Ti G Ý i s p II q 1 pI ZiE s Ž 0. . . l F 0 C E qiE . . this solution is the combination of one of the possible solutions obtained for a single circuit Žcovering four intervals. . The location of both circuits for one of the possible solutions Žfirst alternative . l s Yp II . II n H Ý qiE . . . . l y Yiy 1 . n E qiEH . . No. G 0 Ý i s p II q 1 p I ZiE . a new formulation is presented that achieves this target with the minimum number of circuits. l . l . l q Yiy 1 . j s I. . . FE I pI Ý i s p II q 1 ZiE . . 1996.199 0. . s. ᭙ k s Ž p II q1 . . the value obtained by solving Eq. ˆm ␦m s␦ y QE I I s0 ␦p II II s 0 ␦p ᭙ i s Ž p I q1 . then the value of l is increased to approach the target. . where ⍀ is a sufficiently large number. . 8 . l F 0 E EH II Zp q 1 .l •᭙ l s1. 1 . The minimum number of circuits resulting from this procedure will be less than or equal to the number of intervals between pinches. l s1 Ý i s p q1 II Yi FC . l ⌬ Ti G Ý qi . . This set of two circuits will transfer all the heat predicted by problem 1.

the use of latent heat of a single steam stream may reduce the opportunities of integration. they applied a procedure for multiple-utilities optimization ŽParker. An alternative is the use of the utility system to balance the steam supply and demand of source and sink plants. Consider now the sink plant and the zone between pinches. Due to degeneracy.. In any case.. Two alternative single-circuit solutions for Example 6. In this alternative. the two circuits overlap. 8 . Two-circuits solutions for Example 6 (first alternative).. No. Generalization for More than Two Plants Figure 20. is shown in Figure 21. 1989. assume first that a single steam-temperature level is specified between pinches. Indirect Integration Using Steam The use of steam for indirect integration imposes extra restrictions on the maximum amount of heat that can be transferred. and that is the only indirect fluid used ŽFigure 22. Consider that steam at fixed pressure levels is generated by the plant having an excess of heat Žhigher pinch temperature. 1994. Another possible solution Žsecond alternative . 45. In addition. For simplicity.Figure 19. Treating every single plant individually. By computing the cooling utility required for the source plant from its pinch down to the level of steam generation.. The two alternatives considered here were obtained using GAMS with CPLEX. the maximum load that the sink plant can accept is the deficit it presents between the steam level and the pinch. This plant will be able to use the steam coming from the source plant to reduce its heating utility demands only if this steam temperature is above its pinch temperature. Figure 21.’’ selling and buying utilities at fixed prices from the processes. there are a large number of possible solutions. This amount is the maximum heat this steam will be able to transfer to the sink plant. 1740 The concepts explored up to this point can be extended to the case in which a number n of plants is considered for AIChE Journal August 1999 Vol. and a circuit covering the last interval. Thus. Two-circuits solutions for Example 6 (second alternative). consider the utility as a ‘‘market. respectively ŽHui and Ahmad. the difficulties arise when more than one steam level is considered. Hui and Ahmad Ž1994. the heat load of this steam can be established.

The design of such networks will be attempted in a follow-up article. In turn. an issue that will be further investigated in follow-up articles. Other alternatives may consist of the use of a single circuit that splits in plant 2 and 3. the concepts and tools developed for two plants are a stepping stone for the generalized integration of a set of n plants. A downward shift is performed in the scale of the second and third plants to feasible transfer heat from the first plant. Among many other options. As a first-approach solution. The three possible ways of heat transfer are shown in Figure 23. Integration among three plants. accounting for the three possible combinations of two plants in the unassisted cases. were presented. new shifts of scales are required. Consequently. While all these studies determine the target savings. The increase in complexity is evident. In the case of indirect integration. 8 1741 . This must take place at a minimum investment cost. Models that account for maximum energy savings by direct and indirect heat integration. Therefore. No. and the highest and lowest pinches are identified. a strategy to capture these savings was developed. To predict maximum possible savings. 45. First. Conclusion There is a large incentive to perform heat integration across plants. all possible combinations of two plants have to be evaluated. including in this last case the location of the fluid circuits. AIChE Journal August 1999 Vol. accounting for each one of the combinations of two plants and the respective directions in which the heat will be transferred. problem 1 can be reformulated. the plants are ordered by increasing pinch temperatures. integration. since in principle. as well as when they are functioning separately. In turn. dual-use heat-exchanger networks featuring the minimum number of units accomplish this goal. Figure 22. consider the example of integration among three plants. three circuits can be established.the heat-exchanger network needed to accomplish the predicted savings. a second downward shift is needed in the third plant in order to transfer heat from the second plant. The results of this problem are useful targets for models that will determine Notation Fs product of heat capacity and flow rate ks auxiliary temperature intervals k As first transfer interval in the zone above both pinches k Bs first transfer interval in the zone below both pinches k Es first transfer interval in the zone between pinches p Is last interval above the pinch of plant 1 p II s last interval above the pinch of plant 2 Q F s total heat transferred in the generalized zone qs heat surplus or heat demand q Is heat surplus or heat demand in plant 1 q II s heat surplus or heat demand in plant 2 q Cs heat demand in the heat-sink plant q C P s heat surplus or heat demand in the combined plant q Hs heat surplus in the heat source plant Smin s minimum heating utility T0s initial temperature of the intermediate fluid Tss supply temperature Tts target temperature Tup s upper temperature of a fluid circuit Tdown s lower temperature of a fluid circuit Y F H s binary variable starting interval of a fluid circuit Y F C s binary variable starting interval of a fluid circuit Zs binary variable denoting an interval that belongs to a fluid circuit ˆs original minimum cascaded heat ␦ C ˆ s original minimum cascaded heat in the heat-sink plant ␦ ˆ H s original minimum cascaded heat in the heat-source plant ␦ ␥s variable used in the cascade of heat ␪s cumulative heat demands ␪ Is adjust cascaded heat in plant 1 Figure 23. there is still a need to determine a heat-exchanger network that can accomplish minimum energy consumption while the plants are integrated. For the unassisted cases. the zone delimited by these pinches is where all the integration can take place. Indirect integration using a single-steam temperature level. As a starting point. assisted cases will require transfer heat in the highest or lowest zones. picking up the required heat in each of these plants and performing a similar split when the heat is to be released.

and E. 745 Ž1983.... and Cooling. Parker. 18. V.. ‘‘Heat Exchanger Network Synthesis Without Decomposition.’’ Inst. E. 15. Dhole. DC Ž1988.␪ II s adjust cascaded heat in plant 2 ␭ Hs heat demand in the heat source plant Superscripts ACs cold fluid stream in the zone above both pinches AHs hot fluid stream in the zone above both pinches BCs cold fluid stream in the zone below both pinches BHs hot fluid stream in the zone below both pinches ECs cold fluid stream in the zone of effective transfer of heat EHs hot fluid stream in the zone of effective transfer of heat FCs cold fluid stream in the general case FHs hot fluid stream in the general case js chemical plant Subscripts js chemical plant rs hot stream ss cold stream Literature Cited Ahmad. 1998. Linnhoff. 45. Hindmarsh. 385 Ž1991.. R. Eng .’’ Chem. R. A User’s Guide. K. Univ. Cerda.’’ AIChE Meeting... 12. D. W. UK Ž1989.. Westerberg. No. Eng . Linnhoff. Chem.. Eng .. S. ‘‘Individual Process Improvements in the Context of Site-wide Interactions. Co-Generation. of Manchester Institute of Science and Technology.’’ Comput. Eng .. Manchester.. Eng . and re® ision recei® ed May 20. S.. Sci. B.. Linnhoff. S. and S.. Hui. ‘‘The Pinch Design Method for Heat Exchanger Networks.. 1742 August 1999 Vol. ‘‘Minimum Utility Usage in Heat Exchanger Network Synthesis.. UK Ž1984. ‘‘Total Site Heat Integration Using the Utility System.’’ Comput. Emissions. S101 Ž1992. 729 Ž1994. Washington. A. Mason. J.. 7.’’ Comput.. Sci. and I. 38. Bath.’’ Chem... Morton.. Washington.. Grossmann. 1999.’’ PhD Thesis. J. GAMS. Annual Research Meeting... A. Floudas. Chem. and A. ‘‘A Structural Optimization Approach in Process Synthesis ᎏ II: Heat Recovery Networks. Brooke. and B. ‘‘The Pinch Design Method for the Synthesis of Heat Exchanger Networks: The Constrained Case.. Ahmad. GAMS Development Corporation. ‘‘Total Site Targets for Fuel. Chem. R. C. and C. Eng . Ciric. J. 707 Ž1983. Chem. Eng . W.. 38.’’ Comput. Papoulias. 17. D. 8 AIChE Journal . Hui. W. and B. Meeraus. 17. A. ‘‘Heat Recovery Between Areas of Integrity. Eng .. Chem. 809 Ž1991. Manuscript recei® ed Aug. A. Trivedi. Linnhoff. DC Ž1996. K. Kendrick.. and B. ‘‘Supertargeting for Multiple Utilities.’’ Comput. 373 Ž1983. Chem.. and C.