Please wait a minute...
浙江大学学报(工学版)  2021, Vol. 55 Issue (10): 1968-1977    DOI: 10.3785/j.issn.1008-973X.2021.10.019
机械与能源工程     
基于人工神经网络的超临界小火焰模型研究
高正伟1(),金台2,宋昌成1,罗坤1,樊建人1,*()
1. 浙江大学 能源清洁利用国家重点实验室,浙江 杭州 310027
2. 浙江大学 航空航天学院,浙江 杭州 310027
Application of artificial neural networks to supercritical flamelet model
Zheng-wei GAO1(),Tai JIN2,Chang-cheng SONG1,Kun LUO1,Jian-ren FAN1,*()
1. State Key Laboratory of Clean Energy Utilization, Zhejiang University, Hangzhou 310027, China
2. School of Aeronautics and Astronautics, Zhejiang University, Hangzhou 310027, China
 全文: PDF(1375 KB)   HTML
摘要:

为了解决超临界小火焰燃烧模型数据库过于庞大,导致计算机内存不足和取值性能下降的问题,提出使用人工神经网络(ANN)进行建库的超临界小火焰/过程变量模型FPV-ANN. 在先验性分析及在超临界水热火焰的大涡模拟计算中发现,FPV-ANN方法在温度、组分和其他目标变量的分布与传统FPV方法得到的结果吻合,说明FPV-ANN方法的准确性与传统FPV方法一致. 由于人工神经网络小火焰库大小只有传统库的1%,FPV-ANN方法在大规模并行计算中消耗更少的计算机内存. FPV-ANN方法的计算速度比传统FPV方法提升了30%. 可以看出,提出的FPV-ANN方法具有更好的计算性能.

关键词: 小火焰模型燃烧模拟人工神经网络(ANN)小火焰库建库方法计算性能    
Abstract:

Artificial neural networks (ANN) were utilized to build the library for the flamelet/progress variable (FPV) model and develop the FPV-ANN approach aiming at the problem that the enlarged lookup tables of the flamelet-based combustion model make the computer memory insufficient and slow down the interpolation process. Both the priori analysis and the large-eddy simulation of supercritical hydrothermal flames show that the distributions of temperature, species and other target variables obtained by FPV-ANN and classical FPV method achieve overall good agreement, verifying the accuracy of the FPV-ANN approach. Since the size of the ANN library is only 1% of the classical library, the use of FPV-ANN approach can produce a significant reduction in computer memory consumption during the large-scale parallel simulation. The computational speed of FPV-ANN approach is 30% faster than the classical FPV approach, which confirms that FPV-ANN approach has better computational performance.

Key words: flamelet model    combustion simulation    artificial neural network (ANN)    flamelet library construction method    computational performance
收稿日期: 2020-10-29 出版日期: 2021-10-27
CLC:  TK 16  
基金资助: 国家重点研发计划资助项目(2016YFB0600102)
通讯作者: 樊建人     E-mail: gaozw@zju.edu.cn;fanjr@zju.edu.cn
作者简介: 高正伟(1993—),男,博士生,从事超临界燃烧的数值模拟研究. orcid.org/0000-0003-1455-3493.E-mail: gaozw@zju.edu.cn
服务  
把本文推荐给朋友
加入引用管理器
E-mail Alert
作者相关文章  
高正伟
金台
宋昌成
罗坤
樊建人

引用本文:

高正伟,金台,宋昌成,罗坤,樊建人. 基于人工神经网络的超临界小火焰模型研究[J]. 浙江大学学报(工学版), 2021, 55(10): 1968-1977.

Zheng-wei GAO,Tai JIN,Chang-cheng SONG,Kun LUO,Jian-ren FAN. Application of artificial neural networks to supercritical flamelet model. Journal of ZheJiang University (Engineering Science), 2021, 55(10): 1968-1977.

链接本文:

https://www.zjujournals.com/eng/CN/10.3785/j.issn.1008-973X.2021.10.019        https://www.zjujournals.com/eng/CN/Y2021/V55/I10/1968

案例 各维度上的格子数( $ Z\times {Z}^{{'}{'}2}\times C $ 小火焰库大小/MB
1 150×10×50 60
2 300×25×300 600
3 600×50×600 4800
表 1  不同精度下多维结构化表格小火焰库的大小
图 1  人工神经网络的示意图
图 2  ANN训练结果:目标变量的回归分析图
图 3  水热燃烧器WCHB的结构和尺寸示意图
图 4  FPV-ANN方法目标变量在 $ Z $维度上的先验结果( $ C $= 0.117, $ {Z}^{{'}{'}2} $=0.15)
图 5  FPV-ANN方法的温度分布
图 6  目标变量在中轴线上的时均分布
图 7  温度和H2质量分数在Z-C空间内的分布( $ {Z}^{{'}{'}2} $=0)
图 8  FPV-ANN方法与FPV-ST方法的计算性能分析
1 NORBERT P Laminar diffusion flamelet models in non-premixed turbulent combustion[J]. Progress in Energy and Combustion Science, 1984, 10 (3): 319- 339
doi: 10.1016/0360-1285(84)90114-X
2 NORBERT P. Turbulent combustion [M]. Cambridge: Cambridge University Press, 2000.
3 PITSCH H Large-eddy simulation of turbulent combustion[J]. Annual Review of Fluid Mechanics, 2006, 38: 453- 482
doi: 10.1146/annurev.fluid.38.050304.092133
4 BENOIT F, VEYNANTE D, CANDEL S Modeling combustion chemistry in large eddy simulation of turbulent flames[J]. Flow, Turbulence and Combustion, 2015, 94 (1): 3- 42
doi: 10.1007/s10494-014-9579-8
5 STEPHEN B P Small scales, many species and the manifold challenges of turbulent combustion[J]. Proceedings of the Combustion Institute, 2013, 34 (1): 1- 31
doi: 10.1016/j.proci.2012.09.009
6 MATTHIAS I, SCHMITT C, PITSCH H Optimal artificial neural networks and tabulation methods for chemistry representation in LES of a bluff-body swirl-stabilized flame[J]. Proceedings of the Combustion Institute, 2009, 32 (1): 1527- 1535
doi: 10.1016/j.proci.2008.06.100
7 EMAMI M D, FARD A E Laminar flamelet modeling of a turbulent CH4/H2/N2 jet diffusion flame using artificial neural networks [J]. Applied Mathematical Modelling, 2012, 36 (5): 2082- 2093
doi: 10.1016/j.apm.2011.08.012
8 OWOYELE O, KUNDU P, AMEEN M M, et al Application of deep artificial neural networks to multi-dimensional flamelet libraries and spray flames[J]. International Journal of Engine Research, 2020, 21 (1): 151- 168
doi: 10.1177/1468087419837770
9 ZHANG Y, XU S, ZHONG S, et al Large eddy simulation of spray combustion using flamelet generated manifolds combined with artificial neural networks[J]. Energy and AI, 2020, 2: 100021
doi: 10.1016/j.egyai.2020.100021
10 WELLER H G, TABOR G, JASAK H, et al A tensorial approach to computational continuum mechanics using object-oriented techniques[J]. Computers in Physics, 1998, 12 (6): 620- 631
doi: 10.1063/1.168744
11 GAO Z, WANG H, LUO K, et al Evaluation of real-fluid flamelet/progress variable model for laminar hydrothermal flames[J]. The Journal of Supercritical Fluids, 2019, 143: 232- 241
doi: 10.1016/j.supflu.2018.08.014
12 BILGER R W, STÅRNER S H, KEE R J On reduced mechanisms for methane air combustion in non-premixed flames[J]. Combustion and Flame, 1990, 80 (2): 135- 149
doi: 10.1016/0010-2180(90)90122-8
13 REN M, WANG S, ZHANG J, et al Characteristics of methanol hydrothermal combustion: detailed chemical kinetics coupled with simple flow modeling study[J]. Industrial and Engineering Chemistry Research, 2017, 56 (18): 5469- 5478
doi: 10.1021/acs.iecr.7b00886
14 PENG D Y, ROBINSON D B A new two-constant equation of state[J]. Industrial and Engineering Chemistry Fundamentals, 1976, 15 (1): 59- 64
doi: 10.1021/i160057a011
15 CHUNG T H, AJLAN M, LEE L L, et al Generalized multiparameter correlation for nonpolar and polar fluid transport properties[J]. Industrial and Engineering Chemistry Research, 1988, 27 (4): 671- 679
doi: 10.1021/ie00076a024
16 PITSCH H. FlameMaster: A C++ computer program for 0D combustion and 1D laminar flame calculations[EB/OL]. [2020-09-20]. https://www.itv.rwth-aachen.de/downloads/flamemaster/.
17 PIERCE C D, MOIN P. Progress-variable approach for large-eddy simulation of turbulent combustion [D]. Stanford: Stanford University, 2001.
18 PIERCE C D, MOIN P Progress-variable approach for large-eddy simulation of non-premixed turbulent combustion[J]. Journal of Fluid Mechanics, 2004, 504: 73
doi: 10.1017/S0022112004008213
19 SMAGORINSKY J General circulation experiments with the primitive equations: I. the basic experiment[J]. Monthly Weather Review, 1963, 91 (3): 99- 164
doi: 10.1175/1520-0493(1963)091<0099:GCEWTP>2.3.CO;2
20 ABADI M, BARHAM P, CHEN J, et al. Tensorflow: a system for large-scale machine learning [C]// 12th Symposium on Operating Systems Design and Implementation. Savannah, GA: USENIX, 2016: 265-283.
21 PRÍKOPSKÝ K. Characterization of continuous diffusion flames in supercritical water [D]. Zurich: Swiss Federal Institute of Technology, 2007.
22 GAO Z, WANG H, SONG C, et al Large-eddy simulation of hydrothermal flames using extended flamelet/progress variable approach[J]. The Journal of Supercritical Fluids, 2020, 163: 104843
doi: 10.1016/j.supflu.2020.104843
23 WEN X, WANG H, LUO Y, et al Evaluation of flamelet/progress variable model for laminar pulverized coal combustion[J]. Physics of Fluids, 2017, 29 (8): 083607
doi: 10.1063/1.4999335
[1] 江衍铭, 张建全, 明焱. 集合神经网络的洪水预报[J]. 浙江大学学报(工学版), 2016, 50(8): 1471-1478.