CHEMICAL ENGINEERING TRANSACTIONS VOL. 70, 2018 A publication of The Italian Association of Chemical Engineering Online at www.aidic.it/cet Guest Editors: Timothy G. Walmsley, Petar S. Varbanov, Rongxin Su, Jiří J. Klemeš Copyright © 2018, AIDIC Servizi S.r.l. ISBN 978-88-95608-67-9; ISSN 2283-9216 Heat Exchanger Network Retrofit Under Fouling Effects with Cleaning Schedule Jaturaporn Yanyongsak, Kitipat Siemanond* The Petroleum and Petrochemical College, Chulalongkorn University, Soi Chulalong korn 12, Phayathai Rd., Pathumwan, Bangkok 10330, Thailand kitipat.s@chula.ac.th The energy conservation by heat exchanger network (HEN) is important in process design according to increase in energy costs and global environmental concerns. To minimize the energy consumption with positive net present value (NPV), the retrofitted HEN plays an important role in process energy systems. The HEN retrofit model is based on stage-wise superstructure by Yee and Grossmann (1990). In addition, fouling deposition on the surface area of heat exchangers causes extra energy consumption, production loss and maintenance costs. The new proposed model is retrofitting HEN with fouling effects. This method achieves HEN with the optimal trade-offs between energy savings, and investment over operating period. For cleaning schedule, retrofitted HEN shows better capable to recover heat and higher NPV than base-case considered from a lower number of cleaning requirement. In this study, the proposed model is combination of cleaning schedule strategy and HEN retrofitting with fouling effect to achieve greater profits. 1. Introduction Most HEN synthesis methods rely on sequential or step-wise procedures (Gundersen and Naess, 1988) which decompose design problem for synthesized network targets. After that, Dolan et al. (1987, 1989) and Yee and Grossmann (1990) proposed HEN model accounting for all types of costs simultaneously. Dolan et al. proposed the method of simulated annealing as a synthesis technique, while Yee and Grossmann formulated the model as mixed integer nonlinear programming (MINLP) model for synthesis and retrofit design. Both methods approach optimal operating and capital cost network. In addition, the main problem caused by fouling deposition has negative effects on thermal and hydraulic performance of heat exchangers. Fouling decreases overall heat transfer coefficient and thermal effectiveness of heat exchangers, resulting in extra hot and cold utilities consumption. In most of the cases, the cleaning schedule is applied for recovering heat exchanger efficiency as a systematic method to determine the optimal cleaning sequence in HENs under fouling. For predicting the fouling behaviours, the appropriate models are required. Ebert and Panchel (1995) was the first to give concept of fouling threshold. After that several modified models were proposed for improving the accuracy of crude oil fouling behaviour. In Polley’s model (2002a), wall temperature and Reynold number were used instead of film temperature and shear stress term Polly’s model is more accurate and easier to calculate comparing with Ebert and Panchel’s model. In addition, Rangfak et al. (2017) proposed HEN retrofit with fouling effects which help save utility for crude preheat train operation and achieve high NPV in long period. The combination of cleaning schedule strategy and HEN retrofitting with fouling effect will save more utilities and gain more profits. The purpose of this study is to retrofit HENs under fouling from oil refinery or petrochemical processes. The HENs with fouling effect model will be divided into sub-periods. The model of each period is formulated based on a stage-wise superstructure of Yee and Grossmann (1990). The HEN retrofitting under fouling effect model will be performed. And the cleaning schedule is applied to reduce energy consumption caused by fouling and get higher profit. DOI: 10.3303/CET1870259 Please cite this article as: Yanyongsak J., Siemanond K., 2018, Heat exchanger network retrofit under fouling effects with cleaning schedule , Chemical Engineering Transactions, 70, 1549-1554 DOI:10.3303/CET1870259 1549 2. The model of HEN retrofit In this study, the model of HEN retrofit is MINLP based on stage-wise approach as shown in Figure1. In order to modify former stage-wise model to HEN retrofit model, the constraints for existing exchanger matches have to be added to the synthesis model. The objective function of HEN retrofit model is maximizing profit as a function of utilities saving revenue and total investment cost from additional area and new heat exchanger units. In HEN retrofit part, the main assumptions are shown, as follows. • Constant heat capacities • Constant specific heat capacities • Counter current heat exchangers Figure 1: Stage-wise superstructure of HEN for two hot and two cold streams. (Yee and Grossmann, 1990) In order to formulate the MINLP model for the proposed superstructure described previously, the following definitions and equations are based on Yee and Grossmann (1990). And the modified model for retrofitting is proposed as follows. Maximize Profit = utilities saving revenue – total investment cost = + CCU×(Qcubase-∑ 𝑖 qcui) + CHU×(Qhubase-∑ 𝑗 qhuj) - cf× ∑ 𝑖,𝑗,𝑘 (zi,j,k-zbase,i,j,k) - cfcu× ∑ 𝑖 (zcui-zcubase,i) - cfhu× ∑ 𝑗 (zhuj-zhubase,j) - CA× ∑ (𝑖,𝑗,𝑘 ai,j,k - abase,i,j,k)B - CAC× ∑ 𝑖 (acui - acubase,i)B- CAH× ∑ (𝑗 ahuj - ahubase,j)B (1) 3. HEN retrofit under fouling effects strategy As mention above, the main problem in energy handling in industry is extra energy consumption caused by fouling deposition. In order to recondition thermal efficiency of HEN, there are many fouling mitigation strategies. Most common strategy used to operate HEN with fouling deposition is design of cleaning schedule but this strategy have to shut-down some exchangers or add spare exchangers. Thus, the production loss problem and extra investment cost may involve. In this study, the new proposed model composed of three main steps is shown in Figure. 2. For first step, the model is divided into twelve one-month periods for one year and then base-case HEN is simulated for twelve months with fouling accumulation by GAMS software. Without any periodic cleaning, HEN has to consume more utility due to decreasing heat recovery and overall heat transfer coefficient of network. The fouling deposition is based on fouling threshold model. In this study the fouling threshold models refer to Polly et a. (2002a) dRf/dt = αRe-0.8Pr-0.33EXP(-E/RTw)-γRe0.8 (2) The idea is to retrofit HEN during the shut-down period around the end of twelfth month. Thus, the HEN consumes lower energy consumption and gets better heat recovery by the increased area of each existing exchanger. For second step, base-case HEN at twelfth month under fouling condition is retrofitted by MINLP model using GAMS. For the third step, the retrofitted HEN from second step is operated under fouling effects for twelve months and utilities usage is calculated. The equations of fouling deposition and HEN retrofit are shown below: 1550 Rft = Rft-1 + Rf’t·∆t (3) 1/U = 1/hh + 1/hc + Rf (4) Objective = Minimize total utilities cost for twelve month = CCU× ∑ 𝑖,𝑡 qcui,t + CHU× ∑ 𝑗,𝑡 qhuj,t (5) Figure 2: Scheme of HEN retrofit model under fouling effects with cleaning schedule 4. HEN retrofit under fouling effects strategy with cleaning schedule In order to maximize profit, the cleaning schedule is applied. Wang et al. (2016) apply cleaning schedule for mitigating fouling and get the lower the cost comparing with practical fouling mitigation. The time of operation is divided into 2 types; operation and cleaning sub-periods, shown in Figure 3. The logical constraint, as shown in equation 6, defines the logic that if the effectiveness of heat exchanger ( 𝑄𝑡 𝑄𝑡0 ) is less than cleaning criteria (C), then the cleaning operation will be occurred. The cleaning status is indicated by binary variable Xcl,t. Where Xcl,t is one and zero referring to cleaning operation and non-cleaning, respectively. Equation 7 is used to indicate fouling resistance when cleaning operation is involved. -  ≤ (C - Qt/Qt0) – ( × Xcl,t) ≤ 0 (6) Rft = (Rft-1 + Rf’t) × (1 - Xcl,t) + (Rft0 × Xcl,t) (7) Figure 3: Time discretization for modelling cleaning in HEN. 5. Case study This crude preheat train case is used to illustrate the HEN retrofit model under fouling effects. The problem is accomplished in GAMS 24.2.1 solved by DICOPT as an MINLP solver on notebook computer (ASUS A45V Series (Intel® Core™ i7-3610QM CPU @ 2.30GHz, 8GB of RAM, Windows 10 (64-bit Operating system)). Project life (n) is five years with 20% of annual interest rate. The stream data is represented in Table 1. 1551 Crude preheat train HEN comprises of 10 hot and 3 cold process streams with 6 existing exchangers as presented in Figure 4a. At first, this base-case HEN requires hot and cold utility for 67,988 and 75,076 kW respectively. This base-case is improved to recover heat transfer efficiency using exchanger minimum approach temperature (EMAT) of 5 ˚C. When crude preheat train is operated for twelve months in first step, the result shows that HEN consumes more utilities due to decreasing heat recovery of HEN as shown in Figure 4b. Total hot and cold utility consumptions are 70,162 and 77,250 kW respectively. After twelve months, this HEN is modified by retrofit model. The HEN retrofit shows that there is one new exchanger needed as shown in Figure 4c. The area is increased from 3,913 to 8,424 m2. At the start of run, retrofitted HEN requires hot utility of 53,354 kW and cold utility of 60,442 kW. And retrofitted HEN operated for twelve months shows that hot and cold utilities are 55,851 kW and 62,939 kW, respectively as shown in Figure 4d. For base-case HEN, fouling accumulation rate is increased in existing exchangers, resulting in increasing heat load and decreasing overall heat transfer coefficient during all of the operating periods as shown in Figure 5. For the retrofitted HEN, the result shows that it saves total hot and cold utility along twelve months and gets positive NPV as shown in Table 2. Table 1: Stream data for real crude oil preheat train for base case Stream TIN (˚C) TOUT (˚C) FCP (kW.˚C-1) h (kW.m-2.˚C-1) H1 319.4 244.1 136.186 1.293 H2 73.24 30 6.842 5.063 H3 347.3 45 197.495 0.892 H4 263.5 180.2 123.06 1.361 H5 297.4 110 20.722 1.299 H6 248 50 63.166 1.344 H7 73.24 40 57.687 1.28 H8 231.8 120 48.526 1.396 H9 167.1 69.55 165.278 1.388 H10 146.7 73.24 253.551 0.505 C1 30 232.2 373.238 0.5165 C2 232.2 343.3 488.127 0.788 C3 226.2 231.8 392.55 3.328 Hot utility 120 ($/kW·y) 120 ($/kW·y) - 2 Cold utility 20 ($/kW·y) 20 ($/kW·y) 2 Heat exchanger cost = 26460 + 389×[area (m2)] 0.83 Cleaning cost = 500 $ Cleaning criteria (C) = 70% of thermal effectiveness Table2: Comparison between base-case and retrofit case Without cleaning With Cleaning Base-case for 12 months Retrofit case for 12 months Base-case for 12 months Retrofit case for 12 months NPV - $639,165 - $660,038 Hot utility saving - 20.40% 3.11% 22.23% Cold utility saving - 18.53% 2.82% 20.17% Utility cost $1,751,709 $1,388,799 $1,751,709 $1,381,134 Additional area cost - $419,755 - $419,755 Cleaning cost $3500 $2000 The last step, the cleaning schedule is applied by using cleaning criteria of 70 % of thermal effectiveness ( 𝑄𝑡 𝑄𝑡0 ) for comparing between a number of cleaning operations of base-case HEN and retrofitted HEN. The result shows that there are seven cleaning operations for base-case HEN while retrofitted HEN has four cleaning 1552 operations as shown in Figure 6. The comparison result of cleaning schedule is shown in Table 2. When cleaning schedule is applied with retrofitted case, HEN gets higher utility saving and NPV. (a) (b) (c) (d) Figure 4: (a) Existing HEN of crude preheat train (0-month). (b) Existing HEN of crude preheat train (12-month). (c) Retrofitted HEN of crude preheat train (0-month). (d) Retrofitted HEN of crude preheat train (12-month) Figure 5: Cumulative fouling rate of exchangers (a) Base-case and (b) Retrofitted case Figure 6: Optimal cleaning schedule of (a) Base-case and (b) Retrofitted case (a) 1 2 3 4 5 6 7 8 9 10 11 12 HX1 HX2 HX3 HX4 HX5 HX6 Month (b) 1 2 3 4 5 6 7 8 9 10 11 12 HX1 HX2 HX3 HX4 HX5 HX6 HX7 Month 1553 6. Conclusion In this study, the proposed HEN retrofit under fouling effects helps save total utility cost. The strategy is HEN retrofit model where network is designed involving additional area to recover more energy. Therefore the model achieve the best trade-offs between investment cost due to addition of area and exchanger and utility cost which is caused by fouling. Comparison between base-case HEN and retrofitted HEN, the retrofitted HEN with cleaning schedule overcomes the base-case one with lower number of cleaning operation. When the cleaning schedule is applied, the model shows that combination of HEN retrofit under fouling effects and cleaning schedule achieve lower energy consumption and higher NPV. Nomenclature Indices i hot process stream j cold process stream k index for stage 1 ... k t time interval Parameters  upper bound for heat exchanges hh film coefficient of hot stream (kw/°c.m2) hc film coefficient of cold stream (kw/°c.m2) CF fixed charge for exchangers ($) CHU per unit cost for hot utility ($/kW.year) CCU per unit cost for cold utility ($/kW.year) CA area cost coefficient ($/m2) B exponent for area cost α,β,γ dimensional parameters that vary for different substances (m2.°C/kW) C lower bound for thermal effectiveness Binary variables z existence of matching zcu cold utility exchanging zhu hot utility exchanging Xcl cleaning status Variables Q heat exchanged in heat exchanger (kW) Qhu heat exchanged in hot utility (kW) Qcu heat exchanged in cold utility (kW) a heat exchangers area (m2) U overall heat transfer coefficient (℃∙m2/kW) Rf fouling resistance (℃ ∙m2/kW) Rf’ fouling rate (℃ ∙m2/kW∙month) Re Reynolds number Pr Prandltr number Tw wall temperature of process stream (°C) R gas constant Acknowledgments On behalf of the authors would like to thank you to The Petroleum and Petrochemical College, Chulalongkorn University and Center of Excellence on Petrochemical and Materials Technology (PETROMAT) for funding support. References Polley G.T., Wilson D.I., Yeap B.L., Pugh S.J., 2002, Use of crude oil fouling threshold data in heat exchanger design, Applied Thermal Engineering, 22(7), 763-776. Rangfak S., Siemanond K., 2017, Heat Exchanger Network Retrofit with Fouling Effects. Computer Aided Chemical Engineering Volume 40, 775-780. Rodriguez C., Smith R., 2007, Optimization of Operating Conditions for Mitigating Fouling in Heat Exchanger Networks, Chemical Engineering Research and Design, 85, 6, 839-851. Tian J., Wang Y., Feng X., 2016, Simultaneous optimization of flow velocity and cleaning schedule for mitigating fouling in refinery heat exchanger networks, Energy 109 (2016), 1118-1129. Yee T.F., Grossmann I.E., 1990, Simultaneous optimization models for heat integration—II. Heat exchanger network synthesis. Computers & Chemical Engineering 14(10), 1165-1184. 1554