Variance reduction sampling strategy-based stochastic reconstruction method

doi:10.11949/0438-1157.20231182

Abstract

Abstract:

In the sampling procedure of the stochastic reconstruction method, the number of samples for each structural attribute is unequal and varying. To use Latin hypercube sampling to reduce the variance of the random reconstruction model, based on the characteristics of the sampling process of the stochastic reconstruction method and the principle of Latin hypercube sampling, a new Latin hypercube sampling method suitable for the stochastic reconstruction method was proposed. In the novel Latin hypercube sampling, the sampling number for each structural attribute is determined by the values of the preorder structural attribute. A Latin hypercube sampling-based stochastic reconstruction model for a vacuum gas oil sample was developed. The determination of the sampling number for each structural attribute was introduced in detail with the building diagram. Multiple cases with different predefined molecular numbers were designed to investigate the effects of the novel Latin hypercube sampling on the variance and accuracy of the stochastic reconstruction model. The results showed that the novel Latin hypercube sampling method can significantly reduce the variance and objective function value of the model. As the molecular number ranging from 1000 to 50000, the standard deviations of the new model are 71.36%—74.53% lower than the traditional model, and the objective function values are 1.69%—13.82% lower than the traditional model. Based on the model accuracy and the computational cost of the simulation process, 4000—6000 was selected as the optimal molecule numbers for the new model. By comparing the values of objective function, bulk properties and mass fraction distributions in saturates and aromatics, it was found that the performance of the new model when the molecular number is 2000 is consistent with the performance of the traditional model when the molecular number is 10000, and the computing time of the new model is only 22.54% of that of the traditional model.

Key words: stochastic reconstruction method, novel Latin hypercube sampling, variance, simulation, Monte Carlo simulation

摘要：

在随机重构法的采样过程中，每个结构特征需要的采样数量是不相等且变化的。为了将拉丁超立方采样用于降低随机重构模型的方差，基于随机重构法采样过程的特征和拉丁超立方采样原理，提出了适用于随机重构法的新型拉丁超立方采样方法，探究了在多种分子数量设定情况下应用该方法对随机重构模型的方差和精度的影响。结果表明，应用该方法能够显著降低随机重构模型的方差，提高模型的精度，在分子数量为1000~50000范围内，新模型的标准差相较传统模型降低了71.36%~74.53%，目标函数值降低了1.69%~13.82%。综合模型精度和模拟过程的运算开销，选择4000~6000作为新模型最优的分子数量设定。

关键词: 随机重构法, 新型拉丁超立方采样, 方差, 模拟, Monte Carlo模拟

CLC Number:

TQ 028.8

Guangyao ZHAO, Minglei YANG, Feng QIAN. Variance reduction sampling strategy-based stochastic reconstruction method[J]. CIESC Journal, 2024, 75(5): 1939-1950.

赵光耀, 杨明磊, 钱锋. 基于降方差采样策略的随机重构法[J]. 化工学报, 2024, 75(5): 1939-1950.

Figures/Tables 14

Fig.1 Simple random sampling (a) and Latin hypercube sampling (b)

Fig.2 Diagram for generation process of values of structural attributes in pseduo mixtures with novel Latin hypercube sampling

Fig.3 Diagram of novel Latin hypercube sampling-based stochastic reconstruction method

Table 1 Properties of VGO sample

性　质	实验值	性　质	实验值	性　质	实验值
密度/(g/cm³)	0.9127	硫元素组成/10^-6		模拟蒸馏/K
碳元素/%(质量分数)	85.68	四氢苯并噻吩	566	初馏点	482
氢元素/%(质量分数)	12.38	苯并噻吩	7681	5%	605
硫元素/%(质量分数)	2.24	四氢二苯并噻吩	853	10%	637
氮元素/10^-6	456	二苯并噻吩	7141	20%	670
氢碳比(摩尔比)	1.72	四氢苯并萘噻吩	1269	30%	693
SARA组成/%(质量分数)		苯并萘噻吩	3393	40%	710
饱和烃	64.2	二苯并萘噻吩	1507	50%	725
芳烃	35.8	¹³C核磁/%(质量分数)		60%	739
同系物组成/%(质量分数)		脂肪烃 CH₃	14.5	70%	753
烷烃	22.1	脂肪烃 CH₂	52.5	80%	768
环烷烃	42.0	脂肪烃 CH	16.2	90%	785
单环芳烃	6.1	芳烃 CH	9.7	95%	797
双环芳烃	12.5	芳烃取代位 C	6.3	终馏点	827
三环芳烃	10.4	芳烃桥位 C	0.8
四环芳烃	6.2
五环芳烃	0.7

Table 2 Setting of structural attributes

序数	结构特征	缩写	分布类型	数值范围	参数	采样数量
1	分子类型	SA₁	直方图分布	1,2或3	x₁, x₂	n₁
2	链烷烃上是否存在侧链	SA₂	直方图分布	0或1	x₃	n₂
3	链烷烃上侧链的数量	SA₃	直方图分布	1,2或3	x₄, x₅	n₃
4	链烷烃的长度	SA₄	伽马分布	10~40	x₆, x₇	n₄
5	脂肪环数量	SA₅	伽马分布	1~6	x₈, x₉	n₅
6	苯环数量	SA₆	伽马分布	1~5	x₁₀, x₁₁	n₆
7	芳烃中脂肪环数量	SA₇	直方图分布	0,1或2	x₁₂, x₁₃	n₇
8	噻吩环数量	SA₈	直方图分布	0或1	x₁₄	n₈
9	吡咯环数量	SA₉	直方图分布	0或1	x₁₅	n₉
10	吡啶环数量	SA₁₀	直方图分布	0或1	x₁₆	n₁₀
11	环状结构上侧链的数量	SA₁₁	直方图分布	1,2,3或 4	x₁₇, x₁₈, x₁₉	n₁₁
12	环状结构上侧链的长度	SA₁₂	伽马分布	1~29	x₂₀, x₂₁	n₁₂

Table 3 Boundaries for parameters and constraints for parameters in histogram distributions

参数	下限	上限	递增约束	参数	下限	上限	递增约束
x₁, x₂	0	1	x₁< x₂	x₁₀	0	20
x₃, x₁₄, x₁₅, x₁₆	0	1		x₁₁	0	6
x₄, x₅	0	1	x₄< x₅	x₁₂, x₁₃	0	1	x₁₂< x₁₃
x₆	1	20		x₁₇, x₁₈, x₁₉	0	1	x₁₇< x₁₈< x₁₉
x₇	9	41		x₂₀	1	20
x₈	0	20		x₂₁	0	30
x₉	0	7

Fig.4 Building diagram of molecular construction

Table 4 Alignment of structural attribute values in paraffins

分子序数	SA₁	SA₂	SA₃	SA₄
1	s_1,1	—	—	—
2	s_2,1	s_1,2	s_1,3	s_1,4
3	s_3,1	s_2,2	—	s_2,4
4	s_4,1	—	—	—
5	s_5,1	s_3,2	s_2,3	s_3,4
6	s_6,1	s_4,2	—	s_4,4
︙	︙	︙	︙	︙
k₂	$s k 2, 1$	︙	$s n 3, 3$	︙
︙	︙	︙	—	︙
k₁	$s k 1, 1$	$s n 2, 1$	—	$s n 4, 4$
︙	︙	—	—	—
n	$s n 1, 1$	—	—	—

Table 4 Alignment of structural attribute values in paraffins

分子序数	SA₁	SA₂	SA₃	SA₄
1	s_1,1	—	—	—
2	s_2,1	s_1,2	s_1,3	s_1,4
3	s_3,1	s_2,2	—	s_2,4
4	s_4,1	—	—	—
5	s_5,1	s_3,2	s_2,3	s_3,4
6	s_6,1	s_4,2	—	s_4,4
︙	︙	︙	︙	︙
k₂	$s k 2, 1$	︙	$s n 3, 3$	︙
︙	︙	︙	—	︙
k₁	$s k 1, 1$	$s n 2, 1$	—	$s n 4, 4$
︙	︙	—	—	—
n	$s n 1, 1$	—	—	—

Table 5 Number of molecules in different cases

工况	分子数量	工况	分子数量	工况	分子数量
1	1000	6	6000	11	20000
2	2000	7	7000	12	30000
3	3000	8	8000	13	40000
4	4000	9	9000	14	50000
5	5000	10	10000

Fig. 5 Comparison of objective function values in two models

Fig.6 Comparison of standard deviations for parameter evaluation process in two models

Table 6 Comparison of bulk properties between estimation and experiment

性　质	实验值	文献^[48]	传统模型	本模型
密度/(g/cm³)	0.9127	0.9170	0.9269	0.9292
碳元素/%(质量分数)	85.68	85.40	85.54	85.51
氢元素/%(质量分数)	12.38	12.30	12.30	12.18
硫元素/%(质量分数)	2.24	2.24	2.12	2.26
氮元素/10^-6	456	608	458	448
氢碳比(摩尔比)	1.72	1.72	1.72	1.71
SARA组成/%(质量分数)
饱和烃	64.2	64.1	63.6	63.8
芳烃	35.8	35.9	36.4	36.2
同系物组成/%(质量分数)
烷烃	22.1	21.3	23.2	21.9
环烷烃	42.0	42.8	39.9	41.9
单环芳烃	6.1	6.2	6.2	5.4
双环芳烃	12.5	13.2	11.5	11.9
三环芳烃	10.4	9.8	10.2	10.3
四环芳烃	6.2	5.0	5.9	5.7
五环芳烃	0.7	1.8	3.0	3.0
硫元素组成/10^-6
四氢苯并噻吩	566	577	555	491
苯并噻吩	7681	7772	8194	8469
四氢二苯并噻吩	853	904	853	981
二苯并噻吩	7141	7061	5916	6628
四氢苯并萘噻吩	1269	1126	845	1023
苯并萘噻吩	3393	3391	3300	3478
二苯并萘噻吩	1507	1592	1538	1514
¹³C核磁/%(质量分数)
脂肪烃 CH₃	14.5	13.0	14.6	14.2
脂肪烃 CH₂	52.5	54.3	56.0	55.9
脂肪烃 CH	16.2	15.9	15.7	16.0
芳烃 CH	9.7	8.0	5.4	5.2
芳烃取代位 C	6.3	4.3	4.0	4.3
芳烃桥位 C	0.8	4.4	4.3	4.4

Fig.7 Comparison of distillation curve data between estimation and experiment

Fig.8 Predicted molecular composition in the traditional model and the new model

References 51

1	Ren Y, Liao Z, Sun J, et al. Molecular reconstruction: recent progress toward composition modeling of petroleum fractions[J]. Chemical Engineering Journal, 2019, 357: 761-775.
2	Chen Z, Wang Y, Li Y, et al. Explicit molecule-based reaction network simplification: theory and application on catalytic reforming[J]. Chemical Engineering Science, 2023, 277: 118833.
3	Zhang Y, Zhou Z, Zhang X, et al. Molecular characterization of heavy olefins in slurry-phase hydrocracking products using high-resolution mass spectrometry[J]. Energy & Fuels, 2023, 37(16): 11743-11753.
4	Chen Z, Wang G, Zhao S, et al. A molecular kinetic model for heavy gas oil catalytic pyrolysis to light olefins[J]. AIChE Journal, 2023, 69(8): e18116.
5	Guan D, Cai G, Zhang L. Dual-objective optimization for petroleum molecular reconstruction based on property and composition similarities[J]. AIChE Journal, 2023, 69(8): e18108.
6	Bi K, Qiu T. Novel naphtha molecular reconstruction process using a self-adaptive cloud model and hybrid genetic algorithm-particle swarm optimization algorithm[J]. Industrial & Engineering Chemistry Research, 2019, 58(36): 16753-16760.
7	Chen J, Fang Z, Qiu T. Molecular reconstruction model based on structure oriented lumping and group contribution methods[J]. Chinese Journal of Chemical Engineering, 2018, 26(8): 1677-1683.
8	Quann R J, Jaffe S B. Structure-oriented lumping: describing the chemistry of complex hydrocarbon mixtures[J]. Industrial & Engineering Chemistry Research, 1992, 31(11): 2483-2497.
9	Jaffe S B, Freund H, Olmstead W N. Extension of structure-oriented lumping to vacuum residua[J]. Industrial & Engineering Chemistry Research, 2005, 44(26): 9840-9852.
10	Quann R J, Jaffe S B. Additions and corrections: structure-oriented lumping: describing the chemistry of complex hydrocarbon mixtures[J]. Industrial & Engineering Chemistry Research, 1993, 32(8): 1800.
11	Ye L, Liu J, Xing B, et al. Molecular-level reaction network in delayed coking process based on structure-oriented lumping[J]. Chemical Engineering Science, 2021, 246: 116981.
12	Qin X, Ye L, Murad A, et al. Reaction network and molecular distribution of sulfides in gasoline and diesel of FCC process[J]. Fuel, 2022, 319: 123567.
13	Trauth D M, Stark S M, Petti T F, et al. Representation of the molecular structure of petroleum resid through characterization and Monte Carlo modeling[J]. Energy & Fuels, 1994, 8(3): 576-580.
14	Agarwal P, Sahasrabudhe M, Khandalkar S, et al. Molecular-level kinetic modeling of a real vacuum gas oil hydroprocessing refinery system[J]. Energy & Fuels, 2019, 33(10): 10143-10158.
15	Aye M M S, Zhang N. A novel methodology in transforming bulk properties of refining streams into molecular information[J]. Chemical Engineering Science, 2005, 60(23): 6702-6717.
16	Gomez Prado J, Zhang N, Theodoropoulos C. Characterisation of heavy petroleum fractions using modified molecular-type homologous series (MTHS) representation[J]. Energy, 2008, 33(6): 974-987.
17	Wu Y. Molecular management for refining operations[D]. Manchester: University of Manchester, 2010.
18	Liu L. Molecular characterisation and modelling for refining processes[D]. Manchester: University of Manchester, 2015.
19	Ren Y, Liao Z, Sun J, et al. Molecular reconstruction of naphtha via limited bulk properties: methods and comparisons[J]. Industrial & Engineering Chemistry Research, 2019, 58(40): 18742-18755.
20	Hudebine D, Verstraete J J. Molecular reconstruction of LCO gasoils from overall petroleum analyses[J]. Chemical Engineering Science, 2004, 59(22-23): 4755-4763.
21	Van Geem K M, Hudebine D, Reyniers M F, et al. Molecular reconstruction of naphtha steam cracking feedstocks based on commercial indices[J]. Computers & Chemical Engineering 2007, 31(9): 1020-1034.
22	Verstraete J J, Schnongs P, Dulot H, et al. Molecular reconstruction of heavy petroleum residue fractions[J]. Chemical Engineering Science, 2010, 65(1): 304-312.
23	Van Geem K M, Reyniers M F, Marin G B. Challenges of modeling steam cracking of heavy feedstocks[J]. Oil & Gas Science and Technology-Rev. IFP, 2008, 63(1): 79-94.
24	Klein M T, Neurock M, Nigam A, et al. In Monte Carlo modeling of complex reaction systems: an asphaltene example[M]//Sapre E A M. Chemical Reactions in Complex Mixtures. New York: Van Nostrand Reinhold, 1991: 126-142.
25	Neurock M, Nigam A, Trauth D, et al. Molecular representation of complex hydrocarbon feedstocks through efficient characterization and stochastic algorithms[J]. Chemical Engineering Science, 1994, 49(24): 4153-4177.
26	Petti T F, Trauth D M, Stark S M, et al. CPU issues in the representation of the molecular structure of petroleum resid through characterization, reaction, and monte carlo modeling[J]. Energy & Fuels, 1994, 8(3): 570-585.
27	Campbell D M, Klein M T. Construction of a molecular representation of a complex feedstock by Monte Carlo and quadrature methods[J]. Applied Catalysis A: General, 1997, 160(1): 41-54.
28	Campbell D M, Klein M T. Structural attribute reaction model for heavy hydrocarbons[J]. American Chemical Society, Division of Petroleum Chemistry, Preprints, 1997, 42(2): 300-302.
29	Horton S R, Zhang L, Hou Z, et al. Molecular-level kinetic modeling of resid pyrolysis[J]. Industrial & Engineering Chemistry Research, 2015, 54(16): 4226-4235.
30	Zhang L, Hou Z, Horton S R, et al. Molecular representation of petroleum vacuum resid[J]. Energy & Fuels, 2014, 28(3): 1736-1749.
31	Alvarez-Majmutov A, Chen J W, Gieleciak R. Molecular-level modeling and simulation of vacuum gas oil hydrocracking[J]. Energy & Fuels, 2016, 30(1): 138-148.
32	Deniz C U, Yasar M, Klein M T. A new extended structural parameter set for stochastic molecular reconstruction: application to asphaltenes[J]. Energy & Fuels, 2017, 31(8): 7919-7931.
33	Deniz C U, Yasar M, Klein M T. Stochastic reconstruction of complex heavy oil molecules using an artificial neural network[J]. Energy & Fuels, 2017, 31(11): 11932-11938.
34	Deniz C U, Yasar S H O, Yasar M, et al. Effect of boiling point and density prediction methods on stochastic reconstruction[J]. Energy & Fuels, 2018, 32(3): 3344-3355.
35	Gani R, Nielsen B, Fredenslund A. A group contribution approach to computer-aided molecular design[J]. AIChE Journal, 1991, 37(9): 1318-1332.
36	Marrero J, Gani R. Group-contribution based estimation of pure component properties[J]. Fluid Phase Equilibria, 2001, 183: 183-208.
37	Hukkerikar A S, Sarup B, Ten Kate A, et al. Group-contribution⁺ (GC⁺) based estimation of properties of pure components: improved property estimation and uncertainty analysis[J]. Fluid Phase Equilibria, 2012, 321: 25-43.
38	Yen L C, Woods S S. A generalized equation for computer calculation of liquid densities[J]. AIChE Journal, 1966, 12(1): 95-99.
39	Haktanır M, Karahan S, Yaşar M. Structurally explicit composition model of petroleum vacuum residue[J]. Fuel, 2021, 300: 120977.
40	Glazov N, Dik P, Zagoruiko A. Effect of experimental data accuracy on stochastic reconstruction of complex hydrocarbon mixture[J]. Catalysis Today, 2021, 378: 202-210.
41	Horton S R, Mohr R J, Zhang Y, et al. Molecular-level kinetic modeling of biomass gasification[J]. Energy & Fuels, 2016, 30(3): 1647-1661.
42	Horton S R, Klein M T. Reaction and catalyst families in the modeling of coal and biomass hydroprocessing kinetics[J]. Energy & Fuels, 2014, 28(1): 37-40.
43	Dufour R, Labeau P E, Henneaux P, et al. Stochastic optimization for reactive power planning problems[C]//2016 IEEE International Energy Conference (ENERGYCON). IEEE, 2016: 1-6.
44	Henderson S G, Nelson B L. Handbooks in Operations Research and Management Science: Simulation[M]. Amsterdam: Elsevier, 2006.
45	Amaran S, Sahinidis N, Sharda B, et al. Simulation optimization: a review of algorithms and applications[J]. Annals of Operations Research, 2016, 240: 351-380.
46	Rubinstein R Y, Kroese D P. Simulation and the Monte Carlo Method[M]. MYSE: John Wiley & Sons, 2016.
47	Loh W L. On Latin hypercube sampling[J]. The Annals of Statistics, 1996, 24(5): 2058-2080.
48	Alvarez Majmutov A, Gieleciak R, Chen J W. Deriving the molecular composition of vacuum distillates by integrating statistical modeling and detailed hydrocarbon characterization[J]. Energy & Fuels, 2015, 29(12): 7931-7940.
49	Das S, Suganthan P N. Differential evolution: a survey of the state-of-the-art[J]. IEEE Transactions on Evolutionary Computation, 2011, 15(1): 4-31.
50	Chen B, Zeng W, Lin Y, et al. An enhanced differential evolution based algorithm with simulated annealing for solving multiobjective optimization problems[J]. Journal of Applied Mathematics, 2014:1-13.
51	Ali M Z, Awad N H, Suganthan P N. Multi-population differential evolution with balanced ensemble of mutation strategies for large-scale global optimization[J]. Applied Soft Computing, 2015, 33: 304-327.

[1]	Jinshan WANG, Shixue WANG, Yu ZHU. Influence of cooling surface temperature difference on the high temperature proton-exchange membrane fuel cell performance [J]. CIESC Journal, 2024, 75(5): 2026-2035.
[2]	Yifei LI, Xinyu DONG, Weishu WANG, Lu LIU, Yifan ZHAO. Numerical study on heat transfer of dry ice sublimation spray cooling on the surface of micro-ribbed plate [J]. CIESC Journal, 2024, 75(5): 1830-1842.
[3]	Di WANG, Weiqian CHEN, Lingfang SUN, Yunlong ZHOU. Research of dynamic characteristics of photothermal coupled transcritical compressed carbon dioxide energy storage cycle [J]. CIESC Journal, 2024, 75(5): 2047-2059.
[4]	Fan LIU, Yuantong ZHANG, Cheng TAO, Chengyu HU, Xiaoping YANG, Jinjia WEI. Performance of manifold microchannel liquid cooling [J]. CIESC Journal, 2024, 75(5): 1777-1786.
[5]	Juan LI, Yaowen CAO, Zhangyu ZHU, Lei SHI, Jia LI. Numerical study and structural optimization of microchannel flow and heat transfer characteristics of bionic homocercal fin microchannels [J]. CIESC Journal, 2024, 75(5): 1802-1815.
[6]	Hansong QIN, Guoliang LI, Hao YAN, Xiang FENG, Yibin LIU, Xiaobo CHEN, Chaohe YANG. Theoretical study on the adsorption and diffusion behavior of methyl oleate catalytic cracking in hierarchical ZSM-5 zeolite [J]. CIESC Journal, 2024, 75(5): 1870-1881.
[7]	Jing LI, Fangfang ZHANG, Shuaishuai WANG, Jianhua XU, Pengyuan ZHANG. Effect of cavity structure on flammability limit of n-butane partially premixed flame [J]. CIESC Journal, 2024, 75(5): 2081-2090.
[8]	Lei XIE, Yongsheng XU, Mei LIN. Comparative study on single-phase flow and heat transfer of different cross-section rib-soft tail structures [J]. CIESC Journal, 2024, 75(5): 1787-1801.
[9]	Wenya WANG, Wei ZHANG, Xiaoling LOU, Ruofei ZHONG, Bingbing CHEN, Junxian YUN. Multi-microtubes formation and simulation of nanocellulose-embedded cryogel microspheres [J]. CIESC Journal, 2024, 75(5): 2060-2071.
[10]	Kang ZHOU, Jianxin WANG, Hai YU, Chaoliang WEI, Fengqi FAN, Xinhao CHE, Lei ZHANG. Foam rupture properties of mineral base oils based on molecular dynamics simulation [J]. CIESC Journal, 2024, 75(4): 1668-1678.
[11]	Dongfei LIU, Fan ZHANG, Zheng LIU, Diannan LU. A review of machine learning potentials and their applications to molecular simulation [J]. CIESC Journal, 2024, 75(4): 1241-1255.
[12]	Zheng ZHANG, Wuqiong WANG, Yajing ZHANG, Kangjun WANG, Yuanhui JI. Research progress in theoretical calculation of pharmaceutical formulation design [J]. CIESC Journal, 2024, 75(4): 1429-1438.
[13]	Xiaoying JI, Yuan ZHENG, Xiaopeng LI, Zhen YANG, Wei ZHANG, Shirui QIU, Qianying ZHANG, Canghai LUO, Dongpeng SUN, Dong CHEN, Dongliang LI. Controlled preparation of droplets, particles and capsules by microfluidics and their applications [J]. CIESC Journal, 2024, 75(4): 1455-1468.
[14]	Yiru WEN, Jia FU, Dahuan LIU. Advances in machine learning-based materials research for MOFs: energy gas adsorption separation [J]. CIESC Journal, 2024, 75(4): 1370-1381.
[15]	Yujiao ZENG, Xin XIAO, Gang YANG, Yibo ZHANG, Guangming ZHENG, Fang LI, Fengling WANG. Surrogate modeling and optimization of wet phosphoric acid production process based on mechanism and data hybrid driven [J]. CIESC Journal, 2024, 75(3): 936-944.