基于代理模型的炼厂氢网络与脱硫系统同步优化

吴依凡; 夏志鹏; 吉旭; 周利

doi:10.14135/j.cnki.1006-3080.20220112002

基于代理模型的炼厂氢网络与脱硫系统同步优化

doi: 10.14135/j.cnki.1006-3080.20220112002

四川大学化学工程学院, 成都 610065

详细信息

作者简介:
吴依凡（1997— ），女，湖南岳阳人，硕士生，主要研究方向为过程系统工程。E-mail:wuyifanscu@163.com

通讯作者:
周利，E-mail：chezli@scu.edu.cn

中图分类号: TE624
计量
- 文章访问数: 20
- HTML全文浏览量: 14
- PDF下载量: 3
- 被引次数: 0
出版历程
- 收稿日期: 2022-01-12
- 网络出版日期: 2022-05-26

Surrogate-Assisted Refinery Hydrogen Network Optimization with Hydrogen Sulfide Removal

School of Chemical Engineering, Sichuan University, Chengdu 610065, China

摘要

摘要: 脱硫过程的准确建模是实现氢气分配网络和脱硫过程协同优化的基础。脱硫过程模型的简化和假设会导致结果次优，而基于过程热力学的严格过程模型会带来繁琐的计算工作量。为解决这一困境，本文提出了一种基于代理模型技术的炼油厂氢气网络和H₂S脱除过程耦合优化的新策略。基于严格脱硫过程模型开发其代理模型，再将代理模型集成到氢气网络优化的数学规划模型中，以实现高效准确求解。将该方法应用于国内某炼厂的氢网络集成，在有效控制系统中H₂S含量的同时实现了系统优化。相比于基于简化脱硫模型的文献方法，本文方法所得优化结果较优，证明了该方法的有效性。
- 炼油厂 /
- 氢气网络优化 /
- 硫化氢 /
- 代理模型技术
Abstract: In order to achieve the collaborative optimization of the hydrogen distribution network and the desulfurization process, it is necessary to accurately model the desulfurization process. The simplification of the desulfurization process model will affect the accuracy of the results, while a strict process model based on process thermodynamics is disadvantaged by its inherent high complicacy. To solve this dilemma, this paper proposes a new strategy for coupling optimization of the hydrogen distribution network and the hydrogen sulfide removal process in the refinery. Surrogate models are developed as approximations to the rigorous desulfurization processes. The surrogate model is then integrated into the mathematical programming model for hydrogen network optimization. The proposed method is applied to a case study taken from a refinery in Western China. The result indicates that the proposed model can effectively reduce the H₂S content in the system. Moreover, a comparison between the proposed method and the literary work based on simplified desulfurization models proves that the model presented in this paper is more competitive in obtaining practical optimization results.
- refineries /
- hydrogen network optimization /
- hydrogen sulfide /
- surrogate modeling techniques

HTML全文

图 1 氢气网络的超结构

Figure 1. State-space superstructure of the hydrogen network

下载: 全尺寸图片幻灯片

图 2 典型的MDEA脱硫工艺示意图

Figure 2. Diagram of a typical MDEA desulfurization process

下载: 全尺寸图片幻灯片

图 3 代理模型构建过程

Figure 3. The process of building the surrogate model

下载: 全尺寸图片幻灯片

图 4 案例中的氢气网络

Figure 4. The existing hydrogen network in the case study

下载: 全尺寸图片幻灯片

图 5 HP-DS单元的输入和输出变量

Figure 5. The selected input and output variables for surrogate model fitting of HP-DS units

下载: 全尺寸图片幻灯片

图 6 LP-DS单元的输入和输出变量

Figure 6. The selected input and output variables for surrogate model fitting of LP-DS units

下载: 全尺寸图片幻灯片

图 7 一阶、二阶、三阶、四阶代理模型间的准确性和复杂性比较

Figure 7. Surrogate model comparisons in terms of accuracy and complexity

下载: 全尺寸图片幻灯片

图 8 模型的优化结果

Figure 8. Optimization result of the case by the proposed method

下载: 全尺寸图片幻灯片

图 9 文献模型得到的优化结果

Figure 9. Optimal flowchart of the hydrogen network with the literature model^[25]

下载: 全尺寸图片幻灯片

表 1 现有氢气网络的详细流股信息

Table 1. Detailed hydrogen stream information of the existing hydrogen network

Item		Make-up						High pressure purge gas
Item		F/(mol·s⁻¹)	${\varphi}_{({\mathrm{H}}_{2}) }/{\text{%} }$	${\varphi}_{({\mathrm{H} }_{2}\mathrm{S})}/{\text{%}}$	p/MPa		F/(mol·s⁻¹)		${\varphi}_{ ({\mathrm{H} }_{2} )}/{\text{%} }$	${\varphi}_{ ({\mathrm{H} }_{2}\mathrm{S} )}/{\text{%} }$	${\varphi}_{(\mathrm{C})}/{\text{%} }$	p/MPa
KHT		22.1	93.8	0	2		61.3		87.83	0.9	11.27	5
DHT-2		193.1	93.8	0	2		797.5		86	1.05	12.95	5
GHT		86.8	93.8	0	2		725		86	1.05	12.95	5
DHT-1		252	98.2	0	2		752.6		86	0.95	13.05	5

Item	Low pressure purge gas								Inlet
Item	F/(mol·s⁻¹)	${\varphi}_{({\mathrm{H} }_{2})}/{\text{%} }$	${\varphi}_{({\mathrm{H} }_{2}\mathrm{S})}/{\text{%} }$	${\varphi}_{(\mathrm{C})}/{\text{%} }$	${\varphi}_{({\mathrm{N} }_{2})}/{\text{%} }$	${\varphi}_{(\mathrm{C}\mathrm{O})}/{\text{%} }$	p/MPa		F/mol·s⁻¹	${\varphi}_{({\mathrm{H} }_{2})}/{\text{%} }$	${\varphi}_{({\mathrm{H} }_{2}\mathrm{S})}/{\text{%} }$	p/MPa
KHT	5.6	64.4	1.8	31.04	2.76	0	1		83.4	89.4	0.66	7
DHT-2	36	42.3	1.9	43.82	4.15	7.83	1		990.6	87.5	0.85	7
GHT	19.8	31.32	1.9	59.88	5.41	1.49	1		811.8	86.8	0.94	7
DHT-1	37.2	48.89	1.8	40.19	7.27	1.85	1		1004.6	89.1	0.71	7

下载: 导出CSV

表 2 案例中各单元之间的管道距离

Table 2. Piping distances among the units in the base case

Item	Piping distances/m
Item	Reformer-2	Reformer-1	Hplant	KHT-1	KHT-2	DHT-2	GHT	DHT-1	PSA
KHT	1280	1300	1150	0	150	850	400	700	910
DHT 2	500	400	250	850	1000	0	890	180	480
GHT	1000	1400	1250	400	250	890	0	700	510
DHT 1	680	580	430	700	820	180	700	0	300

下载: 导出CSV

表 3 预留脱硫塔和氢阱之间的管道距离

Table 3. Piping distances among the reserved location for the desulfurization towers and the hydrogen sinks

Item	Piping distances/m
Item	HP tower	LP tower
KHT	550	910
DHT-2	550	480
GHT	340	510
DHT-1	365	300

下载: 导出CSV

表 4 氢阱入口流股的浓度约束

Table 4. Concentration constraints for the inlet streams of the hydrogen sinks

Item	$ {\varphi}_{{\mathrm{H}}_{2}}^{\mathrm{M}\mathrm{i}\mathrm{n}} $/%	$ {\varphi}_{{\mathrm{H}}_{2}\mathrm{S}}^{\mathrm{M}\mathrm{a}\mathrm{x}} $/%
KHT	89.4	0.1
DHT-2	87.5	0.1
GHT	86.8	0.1
DHT-1	89.1	0.1
PSA	—	0.2

下载: 导出CSV

表 5 HP-DS装置的输入变量范围

Table 5. Domain of the input variables for the HP-DS unit

Item	${F}_{ {\mathrm{H} }_{2} }^{\mathrm{h}\mathrm{i}\mathrm{n} }/({\mathrm{N}\mathrm{m} }^{3}\cdot {\mathrm{s} }^{-1})$	$ {F}_{{\mathrm{H}}_{2}\mathrm{S}}^{\mathrm{h}\mathrm{i}\mathrm{n}}/({\mathrm{N}\mathrm{m}}^{3}\cdot {\mathrm{s}}^{-1}) $	$ {F}_{\mathrm{C}}^{\mathrm{h}\mathrm{i}\mathrm{n}}/({\mathrm{N}\mathrm{m}}^{3}\cdot {\mathrm{s}}^{-1}) $	$ {F}_{\mathrm{m}\mathrm{d}\mathrm{e}\mathrm{a}}^{\mathrm{h}} /(\mathrm{k}\mathrm{g}\cdot {\mathrm{h}}^{-1}) $
Lower bound	9920	124	868	5000
Upper bound	22322	397	3100	80000

下载: 导出CSV

表 6 LP-DS装置的输入变量范围

Table 6. Domain of the input variables for the LP-DS unit

Item	$ {F}_{{\mathrm{H}}_{2}}^{\mathrm{l}\mathrm{i}\mathrm{n}} /( {\mathrm{N}\mathrm{m}}^{3}\cdot {\mathrm{s}}^{-1} )$	$ {F}_{{\mathrm{H}}_{2}}^{\mathrm{l}\mathrm{i}\mathrm{n}} /({\mathrm{N}\mathrm{m}}^{3}\cdot {\mathrm{s}}^{-1}) $	$ {F}_{{\mathrm{N}}_{2}}^{\mathrm{l}\mathrm{i}\mathrm{n}} /( {\mathrm{N}\mathrm{m}}^{3}\cdot {\mathrm{s}}^{-1}) $	$ {F}_{\mathrm{C}}^{\mathrm{l}\mathrm{i}\mathrm{n}} /( {\mathrm{N}\mathrm{m}}^{3}\cdot {\mathrm{s}}^{-1}) $	$ {F}_{\mathrm{C}\mathrm{O}}^{\mathrm{l}\mathrm{i}\mathrm{n}} /( {\mathrm{N}\mathrm{m}}^{3}\cdot {\mathrm{s}}^{-1}) $	$ {F}_{\mathrm{m}\mathrm{d}\mathrm{e}\mathrm{a}}^{\mathrm{l}} /( \mathrm{k}\mathrm{g}\cdot {\mathrm{h}}^{-1}) $
Lower bound	112	4	11	149	4	500
Upper bound	623	21	83	623	83	3500

下载: 导出CSV

表 9 文献模型和本文模型的年度成本的比较

Table 9. Comparison of the literature model and the proposed model in regard to the annual costs

Model	Annual cost/CNY
Model	HPlant	Desulfurization	Piping	Fuel	Electricity	Total
Literature [25]	1.565 ×10⁸	1.193 ×10⁶	1.228 ×10⁵	−6.288 ×10⁷	2.163 ×10⁷	1.166 ×10⁸
Proposed model	1.519 ×10⁸	5.099 ×10⁵	1.229 ×10⁵	−6.299 ×10⁷	2.157 ×10⁷	1.111 ×10⁸

下载: 导出CSV

表 10 本文模型和文献模型关于脱硫系统的成本比较

Table 10. Comparison of desulfurization cost between the literature model and the proposed model

Model	Cost/CNY
Model	HP-DS capital cost	HP-DS operation cost	LP-DS capital cost	LP-DS operation cost
Literature [25]	7.141×10⁴	1.029×10⁶	6.641 ×10⁴	4.570 ×10⁴
Proposed model	1.360 ×10⁵	2.120 ×10⁵	1.360 ×10⁵	2.559 ×10⁴

下载: 导出CSV

符号说明
a，b	成本费用系数
C	成本
CP	压缩机集合
CV	热值
D	管道直径
DS	脱硫单元集合（DS=1表示高分气脱硫单元，DS=2表示低分气脱硫单元）
ER	美元兑换人民币汇率
F	氢流股流量
I	氢源集合
J	氢阱集合
K	氢阱集合（同I）
L	管道长度
N	流股组分集合
T	塔板数
P	压力
PF	提纯装置集合
POW	压缩机功率
R	变压吸附装置氢气回收率
t	运行时间

下载: 导出CSV

TAC　　年度总费用
UP	单价
Y	氢流股组分体积分数
ZJ	年化系数
下角标：
i	氢源
j	氢阱
k	氢阱（同j）
cp	压缩机
pf	提纯装置
psa	变压吸附装置
comp	压缩机（同cp）
elec	电
heat	热量
n	氢流股组分
fuel	燃料
ds	脱硫装置
mdea	脱硫剂
pipe	管道
hm	HP-DS消耗的mdea
lm	LP-DS消耗的mdea
上角标：
Max	最大
Min	最小
HP	高分气
in	进口
LP	低分气
out	出口
P	产品流股
R	残余气流股

下载: 导出CSV

参考文献(37)

[1]	SIMPSON D M. Hydrogen management in a synthetic crude refinery[J]. International Journal of Hydrogen Energy, 1984, 9: 95-99. doi: 10.1016/0360-3199(84)90036-3
[2]	ELSHERIF M, MANAN Z A, KAMSAH M Z. State-of-the-art of hydrogen management in refinery and industrial process plants[J]. Journal of Natural Gas Science and Engineering, 2015, 24: 346-356. doi: 10.1016/j.jngse.2015.03.046
[3]	MARQUES J P, MATOS H A, OLIVEIRA N M, et al. State-of-the-art review of targeting and design methodologies for hydrogen network synthesis[J], International Journal of Hydrogen Energy , 2017, 42: 376-404.
[4]	ALVES J J, TOWLER G P. Analysis of refinery hydrogen distribution systems[J]. Industrial & Engineering Chemistry Research, 2002, 41: 5759-5769.
[5]	LIAO Z W, ROGN G, WANG J D, et al. Rigorous algorithmic targeting methods for hydrogen networks: part Ⅱ. Systems with one hydrogen purification unit[J]. Chemical Engineering Science, 2011, 66: 821-833. doi: 10.1016/j.ces.2010.10.019
[6]	HALLALE N, LIU F. Refinery hydrogen management for clean fuels production[J]. Advances in Environmental Research, 2001, 6: 81-98. doi: 10.1016/S1093-0191(01)00112-5
[7]	LIU F, ZHANG N. Strategy of purifier selection and integration in hydrogen networks[J]. Chemical Engineering Research & Design, 2004, 82: 1315-1330.
[8]	PERAMANU S, COX B G, PRUDEN B B. Economics of hydrogen recovery processes for the purification of hydroprocessor purge and off-gases[J]. International Journal of Hydrogen Energy, 1999, 24: 405-424. doi: 10.1016/S0360-3199(98)00105-0
[9]	LIAO Z W, RONG G, WANG J D, et al. Rigorous algorithmic targeting methods for hydrogen networks: Part Ⅰ: Systems with no hydrogen purification[J]. Chemical Engineering Science, 2011, 66: 813-820. doi: 10.1016/j.ces.2010.10.018
[10]	LIAO Z W, TU G, LOU J Y, et al. The influence of purifier models on hydrogen network optimization: Insights from a case study[J]. International Journal of Hydrogen Energy, 2016, 41: 5243-5249. doi: 10.1016/j.ijhydene.2016.01.104
[11]	GIRARDIM L, MARECHAL F, TROMEUR P. Methodology for the design of industrial hydrogen networks and the optimal placement of purification units using multi-objective optimisation techniques[J]. Computer Aided Chemical Engineering, 2006, 21: 1765-1770.
[12]	LIAO Z W, WANG J D, YANG Y R, et al. Integrating purifiers in refinery hydrogen networks: A retrofit case study[J]. Journal of Cleaner Production, 2010, 18: 233-241. doi: 10.1016/j.jclepro.2009.10.011
[13]	MA B Y, DENG C, CHEN H N, et al. Hybrid separation process of refinery off-gas toward near-zero hydrogen emission: Conceptual design and techno-economic analysis[J]. Industrial & Engineering Chemistry Research, 2020, 59: 8715-8727.
[14]	JIA N, ZHANG N. Multi-component optimisation for refinery hydrogen networks[J]. Energy, 2011, 36: 4663-4670. doi: 10.1016/j.energy.2011.03.040
[15]	UMANA B, SHOAIB A M, ZHANG N, et al. Integrating hydroprocessors in refinery hydrogen network optimisation[J]. Applied Energy, 2014, 133: 169-182. doi: 10.1016/j.apenergy.2014.06.080
[16]	UMANA B, ZHANG N, SMITH R. Development of vacuum residue hydrodesulphurization-hydrocracking models and their integration with refinery hydrogen networks[J]. Industrial & Engineering Chemistry Research, 2016, 55: 2391-2406.
[17]	WANG S H, ZHOU L, JI X, et al. A surrogate-assisted approach for the optimal synthesis of refinery hydrogen networks[J]. Industrial & Engineering Chemistry Research, 2019, 58: 16798-16812.
[18]	ZHANG J, ZHU X, TOWLER G P. A simultaneous optimization strategy for overall integration in refinery planning[J]. Industrial & Engineering Chemistry Research, 2001, 40: 2640-2653.
[19]	KANG L X, LIANG X Q, LIU Y Z. Optimal design of inter-plant hydrogen networks with intermediate headers of purity and pressure[J]. International Journal of Hydrogen Energy, 2018, 43: 16638-16651. doi: 10.1016/j.ijhydene.2018.07.044
[20]	WU L, LIANG X Q, KANG L X, et al. Saffron. Hydrogen network optimization by integrating impurity distributions of a fluid catalytic cracker and hydrogenation reaction kinetics[J]. Journal of Cleaner Production, 2018, 192: 542 - 552.
[21]	LI H R, LIAO Z W, SUN J Y, et al. Simultaneous design of hydrogen allocation networks and psa inside refineries[J]. Industrial & Engineering Chemistry Research, 2020, 59: 4712-4720.
[22]	LIU X P, LIU J, DENG C, et al. Synthesis of refinery hydrogen network integrated with hydrogen turbines for power recovery[J]. Energy, 2020, 201: 117623. doi: 10.1016/j.energy.2020.117623
[23]	LIU G L, TANG M Y, FENG X, et al. Evolutionary design methodology for resource allocation networks with multiple impurities[J]. Industrial & Engineering Chemistry Research, 2011, 50: 2960-2971.
[24]	ZHANG Q, LI J, FENG X. Thermodynamic principle based hydrogen network synthesis with hydrorefining feed oil sulfur content variation for total exergy minimization[J]. Journal of Cleaner Production, 2020, 256: 120230. doi: 10.1016/j.jclepro.2020.120230
[25]	ZHOU L, LIAO Z W, WANG D, et al. Hydrogen sulfide removal process embedded optimization of hydrogen network[J]. International Journal of Hydrogen Energy, 2012, 37: 18163-18174. doi: 10.1016/j.ijhydene.2012.08.151
[26]	YANG M B, FENG X. Simulation-based optimization and design of refinery hydrogen networks with hydrogen sulfide removal[J]. International Journal of Hydrogen Energy, 2019, 44: 23833-23845. doi: 10.1016/j.ijhydene.2019.07.108
[27]	PAN M, SIKORSKI J, SKROYD J, et al. Design technologies for eco-industrial parks: From unit operations to processes, plants and industrial networks[J]. Applied Energy, 2016, 175: 305-323. doi: 10.1016/j.apenergy.2016.05.019
[28]	GRACIANO E A, ROUX G A. Improvements in surrogate models for process synthesis application to water network system design[J]. Computers & Chemical Engineering, 2013, 59: 197-210.
[29]	KHALIGHI M, KARIMI I A, FAROOQ S. Comparing sicha and 4a zeolite for propylene/propane separation using a surrogate-based simulation/optimization approach[J]. Industrial & Engineering Chemistry Research, 2014, 53: 16973-16983.
[30]	KIDNAY A J, PARRISH W R, MCCARTNEY D G. Fundamentals of Natural Gas Processing[M]. [S. l.]: CRC press, 2019.
[31]	POSEY M L, ROCHELLE G T. A thermodynamic model of methyldiethanolamine-CO₂-H₂S-water[J]. Industrial & Engineering Chemistry Research, 1997, 36: 3944-3953.
[32]	GARUD S S, KARIMI I A, KRAFT M. Design of computer experiments: A review[J]. Computers & Chemical Engineering, 2017, 106: 71-95.
[33]	SOBOL I M. On the distribution of points in a cube and the approximate evaluation of integrals[J]. USSR Computational Mathematics and Mathematical Physics, 1967, 7: 86-112. doi: 10.1016/0041-5553(67)90144-9
[34]	MOROKOFF W J. Generating quasi-random paths for stochastic processes[J]. Society for Industrial and Applied Mathematics, 1998, 40: 765-788.
[35]	DIPPE A Z, WOLD E H. Antialiasing through stochastic sampling[C]// Proceedings of the 12st Annual Conference on Computer Graphics and Interactive Techniques, [S. l.]: [s. n.]: 1985.
[36]	JIN T W. Comparative studies of metamodelling techniques under multiple modelling criteria[J]. Structural and Multidisciplinary Optimization, 2001, 23: 1-13. doi: 10.1007/s00158-001-0160-4
[37]	BROOK A, KENDRICK D A A, Brook D. Gams: A user’s guide[J], ACM Signum Newsletter , 1988, 23: 10-11.