CHEMICAL ENGINEERING TRANSACTIONS VOL. 76, 2019 A publication of The Italian Association of Chemical Engineering Online at www.aidic.it/cet Guest Editors: Petar S. Varbanov, Timothy G. Walmsley, Jiří J. Klemeš, Panos Seferlis Copyright © 2019, AIDIC Servizi S.r.l. ISBN 978-88-95608-73-0; ISSN 2283-9216 Integrating Hydrogen Turbines into Refinery Hydrogen Network for Power Recovery Xuepeng Liua, Jian Liua, Chun Denga,*, Jui-Yuan Leeb, Raymond Tanc aState Key Laboratory of Heavy Oil Processing, College of Chemical Engineering and Environment, China University of Petroleum, Beijing, 102249, China bDepartment of Chemical Engineering and Biotechnology, National Taipei University of Technology, Taipei 10608, Taiwan cDepartment of Chemical Engineering, De La Salle University, 2401 Taft Avenue, 1004 Manila, Philippines chundeng@cup.edu.cn The operation pressure levels of refinery hydrogenation units (e.g., hydrocrackers and hydro-treaters) and hydrogen supply units (i.e., hydrogen plant and continuous catalyst reforming unit) are normally different. The hydrogen compressors (i.e., make-up hydrogen and recycle hydrogen compressors) are widely utilized to raise the pressure of the hydrogen-rich streams. The valves are also used to reduce the pressure of hydrogen-rich streams, and this leads to the waste of power. Hydrogen turbines have been successfully applied in practical industrial plants to recover power. This paper presents a novel superstructure of refinery hydrogen networks integrated with hydrogen turbines to recovery the power. The superstructure consists of a hydrogen production plant, a continuous catalyst reforming unit, other hydrogen sources and sinks, hydrogen compressors and turbines. The mathematical model including mass balance and logical constraints is proposed, with the objectives of minimizing flowrate of hydrogen utility and minimizing the difference between consumed power in compressors and recovered power using turbines. A simplified industrial case study is solved to illustrate the proposed methodology. Results show that 1 MW of power can be recovered via the integrated turbines. 1. Introduction The purchase and processing amount of inferior crude oil has been increasing yearly in modern refineries. At the same time, stricter environmental regulations on sulfide content in the product oil (i.e., the European Standard (EN 228+A1, 2017) and Chinese national standard (GB 17930-2016, 2016)) require that the upper bound of sulfur content in unleaded petrol is 0.001 %. Thus, the proportion of hydrotreating and hydrocracking processes continues to improve, which consume a large amount of hydrogen. Hydrogen effective management has become a critical issue in refineries. On the other hand, compression work which is generated by using compressors to increase pressure raises the operating cost in hydrotreating and hydrocracking processes. It is desirable to conserve energy while optimizing the hydrogen system to save fresh hydrogen in refinery. The methodologies for the optimal synthesis of hydrogen networks include insight-based pinch techniques and mathematical programming approaches. Alves and Towler (2002) firstly proposed the hydrogen surplus diagram to determine the minimum flowrate of hydrogen utility before detailed hydrogen network design. Then, many other insight-based Pinch techniques have been developed to target the minimum flowrate for hydrogen network, such as the material recovery pinch diagram (El-Halwagi et al., 2003), the source composite curve (Bandyopadhyay, 2006), limiting composite curve (Agrawal and Shenoy, 2006) and the improved problem table (Deng, 2015a). However, these approaches do not consider pressure constraints for hydrogen network. In general, hydro-cracking and hydro-treating processes operate at particular pressure levels. The streams (i.e. make-up hydrogen streams and recycle streams) flowing from low pressure level to higher pressure sinks requires compression work. Ding et al. (2011) proposed the average pressure profiles of hydrogen sources and sinks to determine whether the compressors should be installed or not. Bandyopadhyay et al. (2014) presented a rigorous targeting methodology to minimize compression work in hydrogen allocation network. Their method considered to break hydrogen allocation network with multiple pressure levels into various subproblems having two pressure levels only and minimize DOI: 10.3303/CET1976203 Paper Received: 18/03/2019; Revised: 17/06/2019; Accepted: 11/07/2019 Please cite this article as: Liu X., Liu J., Deng C., Lee J.-Y., Tan R.R., 2019, Integrating Hydrogen Turbines into Refinery Hydrogen Network for Power Recovery, Chemical Engineering Transactions, 76, 1213-1218 DOI:10.3303/CET1976203 1213 cross-flows between these pressure levels. In recent years, various mathematical models based on superstructure have been developed to optimize hydrogen network with pressure constraints. Hallale and Liu (2001) proposed a mathematical model based on superstructure which aims to maximize hydrogen recovery in refineries. Their method considers pressure constraints, existing equipment and capital costs. Kumar et al. (2010) developed modified model considering variable inlet and outlet pressures as well as recycle rate of compressor. Jagannath and Almansoori (2017) presented a mathematical programming model for the optimization of hydrogen networks considering minimum compression cost. Deng et al. (2018) presented an improved nearest neighbors’ algorithm to design hydrogen networks with minimum compression and it is validated using mathematical models. It may be noted that no methodology has been presented to recover power when streams flow from higher pressure to lower pressure. In this paper, a novel superstructure of hydrogen network is integrated with hydrogen compressors and turbines. The superstructure consists of hydrogen sources and sinks, hydrogen compressors and turbines and their interconnections. The mathematical formulation mainly includes mass balance equations and logic constraints, with the objective function considering total power consumption. One literature case is analyzed to illustrate the application of proposed approach. 2. Process Description The problem can be stated as follows. A set of hydrogen sources ( s NSR ) can be allocated to hydrogen sinks or discharged to the fuel system. Each hydrogen source has its outlet flowrate ( iSR F ), impurity composition ( iSR y ) and pressure ( iSR P ). A set of hydrogen sinks ( k NSK ) accept hydrogen sources via reuse/recycle. Each hydrogen sink is characterized by its required inlet flowrate ( jSK F ), upper bound of impurity composition ( max jSK y ) and specified pressure ( jSK P ). The hydrogen utility is treated as supplementary hydrogen source. Compressors are placed to increase the pressure of hydrogen streams to satisfy the inlet pressure of hydrogen sinks and turbines also can be installed to recover power. It is assumed that the total power consumption of hydrogen network is difference between consumed power in compressors and recovered power by turbines. The aim of this study is designing an optimal hydrogen network with minimum flowrate of hydrogen utility and total power consumption. HT Purge HC Purge CCR HT HC Fuel SK S M M M S S S Compressors Turbines H2 PlantHU SR Figure 1: Superstructure of refinery hydrogen network integrated with hydrogen compressors and turbines 3. Mathematical Model The superstructure of the synthesis of hydrogen network is shown in Figure 1. As shown, HU denotes the hydrogen utility. SR and SK denote the process hydrogen sources and sinks. Fuel system denotes the fuel gas system. S and M denote the splitter and mixer. When the pressure of the hydrogen source is lower than that of the hydrogen sink and the supply relationship exists, the compressor is implied to increase the pressure of hydrogen source. When the pressure of the hydrogen source is higher than that of the hydrogen sink and the supply relationship exists, the turbine is applied to recover power. We present the model formulations for the 1214 synthesis of hydrogen network to optimize the flowrate of hydrogen utility and work. A list of indices, sets, parameters and variables are given in the Notation, where all the parameters are denoted with upper-case symbols and all the variables are denoted with lower-case symbols. 3.1 Constraints The model formulations are extended from our previous work (Deng et.al. 2018). In the previous literature, the formulation of compress work in the objective function is a nonlinear equation, and it is modified to be a linear equation in this paper. Eq.(A14) is the objective function to minimize the compression work. The item comp , , ( ) u k u k k u fuk z  − denotes compression work between U and K, which contains continuous variables and binary variables, resulting in the nonlinearity of the objective function. Compression work comp ,a b W is , ( ) a b b a fuk   − when comp ,a b z is one; compression work comp ,a b W is zero when comp ,a b z is zero. The modified compression work calculation is formulated as Eqs(1-7). max comp , , comp a b comp a b W M z  (1) comp max , , , ( ) (1 ) comp a b a b b a a b comp W fab z M −  −  −  (2) comp max , , , ( ) ( 1) comp a b a b b a a b comp W fab z M −  −  −  (3)  comp comp, , ,, comp a b u k s k W W W (4)  comp comp comp, , ,,a b u k s kz z z (5)  , , ,,a b u k s kfab fuk fsk (6)  , ,a u k s    (7) where a and b denote the pressure of the hydrogen supplier and hydrogen receiver. , comp a b W denotes the compression work between hydrogen supplier a and hydrogen receiver b. When the hydrogen flow rate is not zero and the pressure difference between the hydrogen supplier a and receiver b is positive, turbines can be considered to set to recover power. Eqs(8-9) can be represented as judging the setting of turbines. , , , , 1 and 0 1 and 1 P a b a b a t t b z z z z  = = → = = (8)  , ,k ,k,a b u sz zuk zsk (9) The logical constraint (i.e. Equation (8)) is formulated as Eq.(10): turbine , , , 0 P a b a b a b z z z  − + +  (10)  turbine turbine turbine turbine, , , ,, ,a b u k s k s fz z z z (11) where turbine ,a b z is binary variable. When ,a b z is one and , P a b z  is zero, the value of turbine ,a b z must be one, or else the value of turbine ,a b z can be one or zero. When turbine ,a b z is one, the work turbine ,a b W recovered by turbine is , ( )a b a bfuk   − ; if turbine ,a b z is zero, the work turbine ,a b W is zero also. The logical calculations can be expressed by Eqs(12-18). max turbine , , turbine a b turbine a b W M z  (12) max , , , ( ) (1 ) turbine turbine a b a b a b a b turbine W fab z M −  −  −  (13) 1215 max , , , ( ) ( 1) turbine turbine a b a b a b a b turbine W fab z M −  −  −  (14)  , , , ,, , turbine turbine turbine turbine a b u k s k s f W W W W (15)  , , , ,, , turbine turbine turbine turbine a b u k s k s f z z z z (16)  , , , ,, ,a b u k s k s ffab fuk fsk fsf (17)  , , ,a u k s f     (18) where , turbine a b W denotes the work recovered by the hydrogen supplier a and hydrogen receiver b. 3.2 Objective functions The flow rate of hydrogen utility can be minimized to reduce the capacity of hydrogen plant and its operating cost. The objective function is given by Eq(A13) in Appendix A Mathematical Model of literature (Deng et al., 2018). The energy takes great part of operating cost of hydrogen network and the objective function is reformulated to solve the minimum total power consumption, which can be formulated as Eq(19). , , , , , min comp comp total u k s k u NHU k NSK s NSR k NSK turbine turbine turbine u k s k s f u NHU k NSK s NSR k NSK s NSR f NSF W W W W W W           = + − − −           (19) 3.3 Model summary Two models (P1 and P2) are considered in sequence for the synthesis of hydrogen network with minimum total work. P1 – minimum flowrate of hydrogen utility fhu as objective function (Deng et al., 2018), P2 – minimum as objective function min given in Eq(19) s.t. mass balance constraints (A1)(A2)(A3)(A4) and (A6) composition constraint (A5) pressure index calculation 0 0 ln P P P    =     for isothermal process and 1 0 0 1 P P P     −    =    −    for adiabatic/polytrophic process; logic constraints: (A7)(A9)(A10) and Eqs(10-11) connection constraints: (A11) number of compressors: (A12) compression work: Eq(1-7) turbine work: Eqs(12-18) All the models are coded in GAMS 24.2.1 on a PC with an Intel(R) CoreTM i7-4790 CPU 3.60 GHz, 8 GB RAM, running Window 7, 64-bit operating system. The LP and MILP problems are solved using CPLEX. The absolute optimality tolerance for all solvers is set as 10-6. 4. Case Study The Limiting data shown in Table 1 for the refinery hydrogen network is extracted from the literature (Hallale and Liu, 2001). The outlet hydrogen streams of naphtha hydro-treating (NHT), jet fuel hydrotreating (JHT), cracked naphtha hydrotreater (CNHT), diesel hydrotreater (DHT), hydrocracker (HC) and catalytic reformer (CCR) are taken as process hydrogen sources (denoted as S1, S2, S3, S4, S5, S6, respectively). The inlets of NHT, JHT, CNHT, DHT, HC and isomerization (IS4) are considered as process hydrogen sinks (denoted as D1, D2, D3, D4, D5, D6, respectively). The hydrogen utility (denoted as S0) is supplied via hydrogen production plant. The surplus hydrogen source discharged to the fuel system is denoted as D0. 1216 Table 1: Limiting data for hydrogen network Type Impurity composition (mole fraction) Flow rate (Sm3·s-1) Pressure (kPa) μadiabatic (kJ·Sm-3) Sources H2 plant (S0) 0.08 14.598 2,068 839.55 NHT (S1) 0.4 3.324 1,379 747.72 JHT (S2) 0.35 2.596 2,413 877.36 CNHT (S3) 0.25 13.183 2,413 877.36 DHT (S4) 0.3 3.334 2,758 911.48 HC (S5) 0.25 31.791 8,274 1,247.55 CCR (S6) 0.25 3.359 2,068 839.55 Sinks NHT (D1) 0.312 5.136 2,068 839.55 JHT (D2) 0.279 4.015 3,447 971.48 CNHT (D3) 0.229 14.737 3,447 971.48 DHT (D4) 0.248 4.219 4,137 1,023.43 HC (D5) 0.197 40.802 13,790 1,443.61 IS4 (D6) 0.25 0.013 2,068 839.55 Fuel system (D0) 0.4 3.263 1,379 747.72 The minimum flowrate of hydrogen utility supplied by hydrogen production plant and the flowrate of waste hydrogen to be discharged to the fuel system are determined to be 14.598 Sm3·s-1 and 3.263 Sm3·s-1 via solving mathematical model P1. For the results comparison with our previous work (Deng et al., 2018), the upper bound of the connections is set to 17, and the upper bound of compression work is set to 15.09 MW. Solving the mathematical model P2, the minimum total power consumption of the hydrogen network is 14.09 MW. The total compression power is still identical with 15.09 MW, while the recovered power via turbines is 1 MW. Thus, the result reduced by 1 MW compared with that reported in our previous work (Deng et al., 2018) and the literature (Bandyopadhyay et al., 2014). Based on optimization results of mathematical model P2, the optimal hydrogen network can be plotted shown as Figure 2. CCR H2 plant IS4 NHT JHT HC DHT CNHT 12.721 1.686 1.498 1.827 12.723 3.71 0.981 2.596 0.061 3.263 Fuel system Flowrate:Sm 3 ·s -1 0.460 28.081 0.161 0.050 0.013 0.161 2.329 Figure 2: One optimum hydrogen network and the placement of compressors and turbines 5. Conclusions In this paper, a novel superstructure of refinery hydrogen networks integrated with hydrogen turbines is proposed. Two sequential mathematical models (i.e. P1 and P2) are proposed to determine the minimum flowrate of hydrogen utility and total power consumption. One literature case is analyzed to show the applicability 1217 and feasibility of the proposed model. Results show that 1 MW of power can be recovered via the placement of hydrogen turbines. Future work will consider the hydrogen network with purification reuse/recycle, the cost of equipment and the optimization of operating costs. Acknowledgements The research is supported by National Natural Science Foundation of China (No. 21878328) and Science Foundation of China University of Petroleum, Beijing (No. 2462018BJC003). References Agrawal V., Shenoy U. V., 2006, Unified conceptual approach to targeting and design of water and hydrogen networks, AIChE Journal, 52(3), 1071-1082. Alves J. J., Towler G. P., 2002, Analysis of Refinery Hydrogen Distribution Systems, Industrial & Engineering Chemistry Research, 41(23), 5759-5769. Bandyopadhyay B., 2006, Source Composite Curve for Waste Reduction, Chemical Engineering Journal, 125(2), 99-110. Bandyopadhyay S., Chaturvedi N.D., Desai A., 2014, Targeting Compression Work for Hydrogen Allocation Networks, Industrial & Engineering Chemistry Research, 53(48), 18539-18548. Deng C., Zhou Y., Chen C-L., Feng X., 2015, Systematic approach for targeting interplant hydrogen networks, Energy, 90, 68-88. Deng C., Zhu M., Zhou Y., Feng X., 2018, Novel Conceptual Methodology for Hydrogen Network Design with Minimum Compression Work, Energy, 159, 203-215. Ding Y, Feng X, Chu K.H., 2011, Optimization of hydrogen distribution systems with pressure constraints, Journal of Cleaner Production, 19(2-3), 204-211. El-Halwagi M. M., Gabriel F., Harell D., 2003, Rigorous Graphical Targeting for Resource Conservation via Material Recycle/Reuse Networks, Industrial & Engineering Chemistry Research, 42(19), 4319-4328. EN 228+A1, 2017, Automotive fuels - unleaded petrol - requirements and test methods, European Standard , accessed 17.06.2019. GB 17930-2016, 2016, Gasoline for motor vehicles, Beijing Standards Press of China , accessed 17.06.2019. Hallale N., Liu F., 2001, Refinery hydrogen management for clean fuels production, Advances in Environmental Research, 6(1), 81-98. Jagannath A., Almansoori A., 2017, A mathematical model for optimal compression costs in the hydrogen networks for the petroleum refineries, AIChE Journal, 63(9), 3925-3943. Kumar A., Gautami G., Khanam S., 2010, Hydrogen distribution in the refinery using mathematical modelling, Energy, 35(9), 3763-3772. 1218