Retrofitting Industrial Heat Exchanger Network based on Pinch Analysis*

Transcription

1 6th International Syposiu on Advanced ontrol of Industrial Processes (AdONIP) May 28-31, aipei, aiwan Retrofitting Industrial eat Exchanger Network based on Pinch Analysis* Bao-ong Li and huei-in hang Abstract A systeatic buiple pinch based retrofit procedure is developed to lower the utility consuption levels of any given industrial eat Exchanger Network (EN) under new iniu teperature approach ahe cost of inor capital investent. his work is an extended and iproved version of our previous work (Li and hang, 2010). New visualized identification ethod for cross-pinch heat load is proposed to cover the case thahases change in the cross-pinch heat exchanger. Main characters of industrial ENs which are large scale, varied heat capacity flow rate and ultiple pinches are fully addressed. Specifically, each cross-pinch exchanger is identified and those exchanges with large loads are reoved first, and then, their heat loads on the hot and cold streas are both divided according to pinch teperatures. Next, at either side of or between the pinches, the divided heat loads on each strea are cobined to reove sall heat loads and atched according to a systeatic procedure derived fro siple pinch analysis. Note that odifications are focused on those cross-pinch atches and utility exchangers, which greatly siplifies the retrofitting procedure of an industrial EN. An industrial case of rude Oil Preheat rain (OP) is retrofitted to illustrate the effectiveness of the proposed procedure. I. INRODUION Depleting fossil and water resources, increasing environental concerns and energy prices provide the ipetus to iprove heat integration in existing process plants [1]. So, there is renewed interest in retrofitting the existing optial ENs which were designed under the outdated cost data [2]. he EN design ethod is a atured technology for energy integration in process industries, which has already been applied successfully in nuerous grass-root and revap projects for over two decades [3]. Differeno grass-root design of EN, the retrofit of EN is constrained by the existing exchangers and pipes should be reused to their axiu while the introduced odifications are iniized in order to iprove energy efficiency and reduce the required retrofit costs. Methods for retrofitting ENs can be broadly classified into three groups: (a) pinch analysis based ethods, (b) atheatical prograing based ethods, and (c) hybrid ethods. he erits and deerits of each group have been fully reviewed by Sreepathi and Rangaiah [4] recently and will not be repeated here for brevity. here are two possible ways to address the constraints of EN retrofit [5]: one is starting with a new grass-root design and evolving iowards the existing design *Research supported by National Natural Science Foundation of hina (Grant No ). Bao-ong Li is with the Departent of heical Engineering, Dalian Nationalities University, Dalian (corresponding author to provide phone: ; fax: ; e-ail: libh@ dlnu.edu.cn). huei-in hang is with Departent of heical Engineering, National heng Kung University,ainan, aiwan (e-ail: ctchang@ail.ncku.edu.tw). [6], the other is starting with the existing design and odifying io reduce heat exchange across the pinch [7], obviously, the later way is ore direct and effective. Our previous ethod did not address EN retrofit of industrial scale (for exaple, the OP). Main characters of industrial ENs are large scale, varied heat capacity flowrate [8] and ultiple pinches. In addition, ultiple iniu teperature approaches ( s) for different classes of process rather than one global are often required for an industrial EN to targehe utility consuption after retrofit. his work is an extended and iproved version of our previous work [7] in two key ways, the first one addresses larger industrial scale proble such as a crude oil preheat train which ay involve ultiple pinches and varying heat capacity flowrates, and the second one addresses cases involving phase change in cross pinch heat exchangers and ultiple s. For illustration clarity, the reinder of this paper is structured as follows: a otive exaple is presented first, the new visualized identification and partition ethods of the cross-pinch heat loads are then introduced in the next section, the pinch based retrofirocedure is listed in section IV. An industrial case of OP is provided in section V to illustrate the effectiveness of the proposed approach. onclusions are drawn ahe end of this paper. II. A MOIVE EXAMPLE A heat exchanger with heat duty of 23.9 is used as a 120-psig stea generator, the teperature of its hotrea reduces fro to while teperature of its cold strea rises fro 230 to 350 [9]. Note thahase change fro liquid to vapour happened in its cold strea which can be segented into two segents, one is water segent with the teperature scope of [230, 350] and enthalpy difference of 124 Btu/lb; the other is vaporization segent with the constaneperature of 350 and enthalpy difference is 871 Btu/lb. he pinch teperature for cold strea is accidently the sae as the outleeperature of the cold strea, and the is 35. Our previous version of identification of cross-pinch heat loads is based on the pinch teperature and the strea teperatures involved in the heat exchangers. he cross-pinch heat load in this exchanger sees to be 23.9 if our previous identification ethod is adopted and it is graphically represented in Fig. 1a. Notice thahe dashed line denotes the pinch-point location and it is associated with differeneperatures, i.e., p and t p, for the hot and cold streas respectively. he hatched area stands for the cross-pinch heat load (sae representation follows in subsequent figures). he result is not correct because the fact /17/$ IEEE 469

2 thahase change in the cold strea is neglected, this point becoes obvious if the eperature-enthalpy (-) diagra is adopted to represent heat flow in this exchanger as shown in Fig. 1b. Fro Fig. 1b, we can find that only the first segent of cold strea is situated below the pinch while the hotrea is above the pinch. he exact cross-pinch heat load can be calculated as 3.0 rather than 23.9, and detailed forula will be given in subsequenection = Figure (a) Representation of cross-pinch heat load for the otive exaple Based on this otive exaple, our previous identification ethod is only applicable to those exchangers which do not involve phase change, and both hot and cold streas assue the constant heat capacity flow rate [7]. III. IDENIFIAION OF ROSS-PIN EA LOAD BY - DIAGRAM A. No Phase hange is Involved in the eat Exchanger All possible scenarios for this category can be found in Fig. 2. In the exchangers described in Fig. 2, the cold-strea teperature rises fro t s to t t while the hot-strea teperature decreases fro s to t. It is obvious thahe entire heat duty of the exchanger in Fig. 2a is the cross-pinch heat load because thahe pinch line fully separates the hotrea and cold strea and no intersection with the. As for the other three scenarios, only part of the exchanger duty is transferred across the pinch and the exact cross-pinch heat load can be easily deterined according to the following forulas: Q F ( ) Q F ( ) for Fig. 2b cr s p p t Q F (t t ) F ( ) cr p s p t =F ( ) F (t t ) s p cr p s for Fig. 2c Q F (t t ) Q F (t t ) for Fig. 2d where Qcr and Q denote the cross-pinch heat load and the entire heat duty of the exchanger separately, F and F represenhe heat capacity flow rates of hot and cold streas respectively. Note thahere are two alternative forulas to deterine the cross-pinch heat load for scenarios (b-d) and the sae resulhould be obtained based on heat balance. Also, the cross-pinch heat load(s) can be considered as specific retrofiarget(s), i.e., the heat loads that can be reoved so as to achieve the desired utility consuption levels due to reduction = (b) 350 B. Phase changes in the heat exchanger When phase changes within a heat exchanger, three subcategories can be fored. he first one is that only the cold strea has phase change, the second one is that only the Figure 2. Four types of cross-pinch heat load for the case that no phase change is involved hotrea has phase change while the rest one is thahase changes in both cold and hotreas. Only the first one is discussed below because of siilarities aong these subcategories. Figure 3. ts (a) Four types of cross-pinch heat load for the case thahase changes in the cold strea All possible scenarios for this subcategory are presented in Fig. 3. he ain character of Fig. 3 is thahe heat capacity flow rate of cold strea changes with teperature because of phase change and should be segented. Siilar to Fig. 2, the entire heat duty of the exchanger in Fig. 3a is the cross-pinch heat load, eanwhile, the exact cross-pinch heat load in Fig. 3b can be deterined by the sae forula as Fig. 2b because only hotrea is involved in the calculation and it has no phase change. As for scenarios shown in Fig. 3(c-d), the following forula can be used to deterine the exact cross-pinch heat load: Q F (t t ) for t t cr L p s p t Where FL represents the heat capacity flow rate of cold strea before phase changes. For the otive exaple, the exchanger atches the scenario shown in Fig. 3c and F (c) (d) (a) (c) (b) (d) (b) L 470

3 3 can be easily deterined as Btu/(h ), and 3 Q ( ) 3.0. cr oparing with the previous identification ethod [7], the new visualized representation reveals phase change and varied heat capacity flow rate clearly, so the new ethod is ore coprehensive and ore accurate. IV. REROFI PROEDURE FOR INDUSRIAL EN argets for iniu energy consuption are based on the chosen value of. his paraeter reflects the trade-off between capital investent and energy cost [9]. With increasing energy costs, saller is preferred for EN retrofit. Note that ultiple s rather than only one global are required because of different classes of process exchangers, for exaple, exchangers of process strea against utilities could have a different [9]. After the (s) is or are chosen, the proposed retrofit ethod can be applied according to the following steps: Step 1: Deterine the hot and cold pinch teperatures and the corresponding energy targets according to specified (s), and ultiple pinches could be identified. Aspen Energy Analyser [10] and Pinch Analysis Spreadsheet developed by IheE are recoended for this step. Step 2: Identify and reove those ajor cross-pinch heat atches. Divide their heat loads on the process streas into two parts according to the corresponding pinch teperatures. Step 3: For each strea which has unatched load(s), a split load should be cobined into other unatched loads on the sae side of the pinch (i.e., above or between or below) if it is saller than a given value. he given value of heat load should been chosen based on the balance between avoiding those atches with very sall heat loads and keep existing heat loads distribution as uch as possible to reuse those existing heat exchangers. Step 4: At both sides of each pinch, atch the above cobined heat loads according to a odified version of the pinch design ethod [7]. Detailed explanation on this step is oitted here because of the liitation on paper length. Step 5: Break the heat load loops to reduce the total nuber of exchangers when necessary. Recalculate the heat duties, the inlet and outleeperatures of the involved atches. Step 6: Assign the new atches to available exchangers. Generally speaking, the industrial heat exchangers are alost always over designed by 15-30% in heat-transfer area. An existing exchanger can usually be adopted to realize a new atch between the sae hot and cold streas if its required heat-transfer area is close to the original design level (say 20%). If there are ore than one candidate atch copeting rude Feed fro ank, 65 F APA fro At ower, F F APA to At ower for the sae existing exchanger, the one with saller difference in heat-transfer area should be selected. E-1A/B 28.1 AOVD fro At ower, F F W E F F F Reflux to At ower Vacuu Resid fro Vacuu ower, F Furnace Fuel Oil to ank E Kerosene, E-17AD F F F Diesel F 90 F F o ank o ank ABPA fro At ower, F E-8A/B E-7A/B o ower F W F o Vacuu ower V. ASE SUDY AMPA fro At ower, F F E-3A/B 35.3 Desalter o Fluid atalytic racker(f) Unit his case is taken fro the work of Rossiter [9]. It is a 90,000 bbl/d rude Distillation Unit (DU) that includes both atospheric and vacuu tower. he OP is given in Fig. 4 and extracted strea data for pinch analysis is provided in able Ⅰ. eating and cooling utility and cost data are suarized in able Ⅱ. Stea generation (at 120 psig in this case) is considered as a cold utility and represented as a segented utility. Rule of thub values for that optiize the trade-off for different classes of processes, and between process streas and utilities are adopted, and selected value for process streas againsrocess streas is 70, for process streas againstea is 35 and process streas against cooling water is 30. Note thahere is data inconsistency in the original work of Rossiter [9], the inleeperature of Vaccu Resid to E-8A/B and the outleeperature of vacuu gas oil(vgo) fro E-6ABD in Fig. 4 have been highlighted and corrected to and respectively in order to keep the extracted strea data the sae as that adopted by Rossiter [9]to targehe utility usage. In addition, hotrea VGO leaving exchangers E-15 and E-16 are supposed to be ixed and teperature after ixing can be deterined as as shown in able Ⅰ by enthalpy balance, such treatent can reduce the nuber of process streas by 1. Furtherore, the W VGO fro Vacuu ower, F F F E-6ABD 93.4 E F E F F E-4A/B F o ower W E o ank Figure 4. he existing rude Oil Preheating rain AGO fro At ower, F F E

4 heat capacity flow rates in able Ⅰ are varied with their teperatures and should not be treated as a constant for ost streas; they are segented according to existing network configuration. ABLE I. SREAM DAA FOR PIN ANALYSIS OF OP eperature( ) eat duty eat capacity flow rate Strea Inlet Outlet () FP (MBtu/ -h) APA AOVD Kero e e-2 Diesel AMPA AGO e-2 VGO ABPA VR rude Oil1 rude Oil ABLE II. UILIY AND OS DAA Utility eperature( ) ost Inlet Outlet Btu/lb $/ per year Furnace N/A 49, ,500 Generation ooling water N/A N/A Basis: Furnace efficiency=85%. Working hours=8,400h/yr, Fuel cost=$5.00/mbtu(illion Btu). 120-psig stea cost=$4.50/mbtu. Aspen Energy Analyzer (v8.8) is used to deterine the pinch teperatures and potential for energy saving targets. he sae targets as those identified by Rossier [9] can be found and are suarized in able Ⅲ. wo pinches can be found for this case, the first one is process pinch with pinch teperature for hotrea is 481.8, and the second one is utility pinch with pinch teperature for hotrea is he heat integration opportunities in this case are best understood fro the suary inforation in able Ⅲ. he 120-psig stea is exported, so a negative scope iplies added value. Fro able Ⅲ, the energy saving target after retrofit is 33.8 for both hot and cold utilities, and cosaving is k$/yr. he cross-pinch heat loads are listed in able Ⅳ and those cross-pinch exchangers are highlighted in the grid diagra of Fig. 5. Fro able Ⅳ, we can find the total cross-pinch load againsrocess pinch is jushe sae as retrofiarget of 33.8 in able Ⅲ while that againshe utility pinch is 55.65, uch high than the target value. he reason is tharansferring heat across a utility pinch does not autoatically lead to an energy penalty and iiply substitutes high teperature utility for low teperatures [6]. ABLE III. Existing REROFI ARGES OF E ASE arget Scope Saving k$/yr Realized scope Fired , heater old total: 120-psig stea Gen ooling water otal 2611 ABLE IV. LIS OF ROSS-PIN LOADS ross-pinch duties, eat ot old process, 120-psig stea, exchanger strea strea E-3A/B Diesel Raw crude 14.9 E-4A/B AMPA Raw crude 0.97(0.9) 28.6 E-5 AGO Desalted crude E-6A-D VGO Desalted -1.35*(-1.3) crude E-7A/B ABPA Desalted 1.2 crude E-10 AMPA 120-psig stea gen Vacuu 120-psig resid stea gen otal Note: *A negative cross-pinch duty in a heat exchanger indicates thahe iniu teperature difference between the cold and the hotreas is less than the specified Miniu eperature Approach; those values reported by Rossier [9] are provided in brackets if different fro our results. In Fig. 5, the pinches appear as broken vertical lines, the hot and cold strea pinch teperatures are shown ahe top and the botto of the diagra, respectively. he process streas are shown as horizontal lines, with the hotrea running fro lefo right and the cold streas fro righo left. he teperature scale is norecise buhe start and targeeperatures of each strea are appropriately related to the pinches. It is then apparent which streas have segents above the process pinch teperature, between the two pinches, and below the utility pinch. eat exchangers are added to the diagra in Fig. 5. Process-to-process heat exchangers are shown as dubbells linking a hotrea to cold strea. Utility heat exchangers are shown as circles with a label identifying the type of used utility. If either the hot or cold portion of a heat exchanger extends across one or ore pinch boundaries, the appropriate circle is elongated to shown the teperature range relative to the pinches. he bar of a dubbell is drawn vertical when possible; however, it is noossible when the entire duty of a heat exchanger crosses a pinch(as is the case for exchanger E-4A/B crossing the utility pinch). able Ⅳ reveals thahe largest inefficiencies are in two atospheric iddle puparound (AMPA) heat exchangers, 472

5 E-4A/B (AMPA vs. raw crude, with 28.6 crossing the 120-psig stea pinch), and E-10 (AMPA stea generator, E-8A/B Vacuu Resid F ABPA F AMPA F E-7A/B E-10 E-4A/B F F S 23.9 AGO F Process pinch VGO F F F Desalted crude F F F with 23.9 crossing the process pinch). he next largest inefficiency (14.9 crossing the stea generation pinch) is in E-3A/B, the diesel vs. raw crude exchanger. All of the reaining cross-pinch loads are significantly saller (<10 ). So, the larger opportunities (in E-4A/B, E-10 and E-3A/B) are focused in this case, which account for ore than 73% of the heat crossing the process pinch and ore than 83% of the heat crossing the utility pinch. Now, those cross-pinch heat loads are split according to pinch teperatures as shown in Fig. 6. Specifically, the hotrea of Exchanger E-10 is copletely above the process pinch, its heat duty does not need to be split, buhould be unatched and renaed as E-10h, where the new added letter h in the new label stands for hotrea. As for Exchanger E-4A/B, its hotrea is crossing the process pinch and its heat duty should be split into two parts according to the pinch teperature, i.e., E-4ha of 0.98 above the process pinch and E-4h of 27.6 between the pinches, where the last letter in the new labels a and represent above the process pinch and between the pinches respectively. S 10.9 E-5 E-6A-D Diesel F Utility pinch F o ank F Figure 5. he existing prehearain as a grid diagra highlights those cross-pinch heat exchangers E-3A/B Kero F E-8A/B Vacuu Resid F ABPA F AMPA F APA F E-16 E-7A/B E-10h E-4ha F AGO F E AOVD F F Process pinch he cold strea of Exchanger E-4A/B is below the utility pinch, and does not need to be split, buhould be unatched and renaed as E-4c, where the last letter c in the new label stands for cold strea. Siilarly, the hot strea of Exchanger E-3A/B should be split into E-3h of 14.9 between the pinches and E-3hb of 20.4 below the utility pinch separately(where the last letter b in the new label represents below the utility pinch ), and the cold strea of Exchanger E-3A/B should be unatched and renaed as E-3c. In addition, light blue and green colours are also adopted to represenhose split heat loads of exchangers E-4A/B and E-3A/B respectively as shown in Fig. 6. An unatched load should be cobined if it is less than 10. here is only one unatched heat load, i.e., E-4ha, which needs to be cobined into heat load E-10h and the cobined load naed as E-10h+ as shown in Fig. 7. hree teperature intervals are fored because of two pinches existing in the case, and each interval should be considered as a separated subnetwork. Reaining procedure on retrofitting of this case is not provided here because of length liit on this paper. he final retrofit design is given in Fig. 7 which save in crude preheat and recover 18.6 for additional 120-psig stea generation as shown in able Ⅲ. VGO F E-4h F VGO product, VGOPA 310 E-18 E-17AD F F E-6A-D E-3h Diesel F 14.9 o ank Kero F F Desalted crude F F F S F F 10.9 E-1A/B E F F 21.6 E rude Feed 65 F Utility pinch APA E-4c E-3c F F F AOVD F E-18 E-17AD F F F E-1A/B E F F 21.6 rude Feed 65 F Figure 6. he prehearain as a grid diagra after splitting three larger cross-pinch heat exchangers F E-3hb 20.4 E-16 E F VGO product, VGOPA F

6 A new heat exchanger of E-new is added to fulfill hearansferring of process strea APA with a branch of rude Feed and parallel to E-3cf. oparing with the retrofit design obtained by Rossiter[9], the new design is ore accurate and reliable. Note that Rossiter s design fails to satisfy the following requireents: (1) the targeeperature of strea VGO to Vacuu ower is 310, this value is in his/her design. (2)he actual Miniu eperature Approach of new atch E-X1AB is as low as 11.1 which is uch lower than required 70 ; siilar case happens in Exchanger E-6A-D, its iniu teperature approach is 56.5 and lower than required value. In addition, the new design achieves both energy and onetary saving targets by 75%, which is uch higher than the reported value of 45% by Rossiter [9]. VI. ONLUSION New visualized tool of - diagra is developed in this paper to identify cross-pinch heat load to cover the case thahases change in the cross-pinch heat exchanger. A systeatic buiple pinch based retrofirocedure is developed to retrofit any existing industrial EN under new s. haracters of Large scale, varied heat capacity flow rate and ultiple pinches are fully addressed in the new ethod. he proposed ethod is applied to an industrial retrofit case of OP, the obtained design strictly satisfies those retrofit requireents and achieves both energy and onetary saving targets by 75%, which is uch higher than the reported value of 45% [9]. It is safely to conclude thahe new visual tool and proposed retrofit ethod is effective and siple to be adopted. NOMENLAURE AMPA atospheric iddle puparound Btu British theral unit DU rude Distillation Unit OP rude Oil Preheat rain F heat capacity flowrate EN eat Exchanger Network MBtu illion British theral unit Q heat duty t,eperature of cold and horocess strea respectively VGO vacuu gas oil he iniu teperature approach Subscripts cold strea L cold strea before phase changes cr cross hotrea p pinch F rude Feed fro ank, 65 F F E-1A/B 28.1 W AOVD fro At ower, F F F F Reflux to At ower s t E Kerosene, E-17AD F o ank Vacuu Resid fro Vacuu ower, F Furnace start target APA fro At ower, F F 6.5 E-new F F o ank ABPA fro At ower, F E-3hc E Diesel o ank F F W F F E-3cf 22.7 W AMPA fro At ower, F AKNOWLEDGMEN Financial supports fro National Natural Science Foundation of hina (Grant No ) and Fundaental Research Funds for the entral Universities (Grant No. D ) are gratefully acknowledged. REFERENES 310 F F E-4cf F o Vacuu ower Desalter AGO fro At ower, F F F F F E-8A/B E-7A/B E-10h+ E-6A-D E F o Fluid atalytic o ower VGO fro Vacuu racker(f) Unit ower, F F E-4h 27.6 o ower F Fuel Oil to ank Figure 7. he prehearain after retrofit [1] B.K. Sreepathi and G.P. Rangaiah, Iproved heat exchanger network retrofitting using exchanger reassignentrategies and ulti-objective optiization, Energy, 67, , 2014a. [2] B. Bakhtiari and S. Bedard, Retrofitting heat exchanger networks using a odified network pinch approach, Applied heral Engineering, 51, , [3] B. Linnhoff, Pinch analysis - a state-of-the-art overview, he.eng. Res. Des., 71, , [4] B.K. Sreepathi and G. P. Rangaiah, Review of heat exchanger network retrofitting ethodologies and their applications, Ind. Eng. he. Res., 53, , 2014b. [5] D.A. Jones, A. N. Yilaz and B. E. ilton, Synthesis techniques for retrofitting heat recovery systes, EP, July, 28-33, [6] I.. Kep, Pinch Analysis and Process Integration- A User Guide on Process Integration for the Efficient Use of Energy, 2nd, Elsevier Ltd. 2007, P59-60 and P [7] B.. Li and.. hang, Retrofitting heat exchanger networks based on siple pinch analysis, Ind. Eng. he. Res., 49, , [8] B.K. Sreepathi and G.P. Rangaiah, Retrofitting of heat exchanger networks involving streas with variable heat capacity: Application of single and ulti-objective optiization, Applied theral Engineering, 75, , [9] A. P. Rossiter, Iprove energy efficiency via heat integration, EP, Dec., [10] Aspen echnology, Aspen Energy Analyzer-user guide, Bedford, MA USA, 2015.