Research on heat transfer deterioration prediction of supercritical RP-3 based on optimized neural network
-
摘要:
目前针对超临界航空煤油RP-3流动传热过程中传热恶化的研究较少，现有研究手段多为经验关联式和数值模拟，对复杂流体传热时的非线性特征捕捉不够精确。为了精准预测超临界航煤RP-3在典型工况下的传热恶化特性，利用计算流体力学方法，在经实验数据对比验证后建立了涵盖管径为1～10 mm、压力为3.0～6.0 MPa工况的高保真数据库，设计并优化了多层神经网络结构，得到了泛化性能和预测准确性较高的调优神经网络预测模型。结果表明：随管径增加或压力降低，局部浮升力与热加速效应显著增强，导致传热恶化程度加剧，温度峰值位置沿流动方向后移。优化后的模型能精准捕捉这些非线性规律，预测误差在1.5%以内，为航空发动机热管理系统的高效设计提供了一种可靠的预测工具。
Abstract:Currently, research on heat transfer deterioration in the flow and heat transfer process of supercritical aviation kerosene RP-3 remains limited. Existing approaches mainly rely on empirical correlations and numerical simulations, which struggle to accurately capture the nonlinear characteristics associated with complex fluid heat transfer. To precisely predict the heat transfer deterioration behavior of supercritical RP-3 under typical operating conditions, computational fluid dynamics methods were employed; after experimental validation, a high-fidelity database covering pipe diameters from 1 to 10 mm and pressures ranging from 3.0 to 6.0 MPa was established. A multilayer neural network structure was then designed and optimized, resulting in a well-tuned model with strong generalization ability and high prediction accuracy. The results indicated that as pipe diameter increased or pressure decreased, local buoyancy and thermal acceleration effects were significantly intensified, exacerbating the degree of heat transfer deterioration and shifting the temperature peak downstream. The optimized model accurately captured these nonlinear trends, with prediction errors achieved within 1.5%, thus providing a reliable data-driven tool for efficient design of the thermal management systems in aero-engines.
-
图 2 不同湍流模型预测壁温的结果对比
Figure 2. Comparison of wall temperature predictions using different turbulence models
图 3 温度与表面传热系数随网格数量的变化
Figure 3. Variation of the temperature and surface heat transfer coefficient with grid number
图 4 不同热流密度下计算与实验壁面温度的对比分析
Figure 4. Comparison between the computed wall temperature and the experimental results under different heat flux densities
图 5 传热恶化条件下计算与实验壁面温度的对比分析
Figure 5. Comparison between the computed wall temperature and experimental results for heat transfer deterioration cases
图 6 多层前馈神经网络结构示意图
Figure 6. Schematic diagram of a multilayer feedforward neural network
图 9 验证集与平均参考集 R2 得分热力图
Figure 9. Heatmap of R2 scores on validation set and mean reference set
图 11 原数据集传热恶化峰值点标注图
Figure 11. Marked heat transfer deterioration peak point for original datasets
图 12 原始与缩减数据集预测温度分布对比图
Figure 12. Comparison of predicted temperature profiles under original and reduced datasets
图 13 缩减数据集传热恶化峰值点标注图
Figure 13. Marked heat transfer deterioration peak point for reduced datasets
图 14 ANN预测与经验关系预测对比图
Figure 14. Comparison between ANN prediction and empirical correlation prediction
表 1 物理模型工况参数表
Table 1. Operating conditions of the physical model
参数 数值 管径D/mm 1～10 系统压力p/MPa 3～6 进口温度Tin/K 423～573 壁面热流密度$q'' $/（kW/m2） 218～500 质量流量$\dot m $/（g/s） 0.55～55 
下载: 导出CSV
表 2 LS模型中常数和修正函数条件汇总
Table 2. Summary of constants and correction functions in LS model
类别 参数 表达式/数值 常数 Cμ 0.09 Cε1 1.44 Cε2 1.92 σk 1.0 σε 1.3 修正函数 f1 1.0 f2 1−0.3exp（−$Re^2_{\mathrm{t}} $） fμ exp{−3.4/[（1+Ret/50）2]}
下载: 导出CSV
表 3 异常值判断与清洗标准
Table 3. Abnormal value judgment and cleaning standards
层级 判定标准 判定阈值 统计层 Robust-Z分数法 Zrobust ≥ 3 物理层 温度、物性参数和能量守恒 T ≤ 1200 K，物性（密度、比热容、黏度等） > 0，能量守恒误差 |ΔQ/Q| ≤
1%（ Q为热流量；ΔQ为系统输入与输出热量的差值）空间层 努塞尔数空间二阶差分异常检查 相邻三截面Nu二阶差分值≥5倍平均值且符号相反
下载: 导出CSV
表 4 各网络结构R2得分对比表
Table 4. R2 scores for different neural network structures
网格层数 神经元数量 训练集 验证集 D5 D7 D9 平均值 训练时长/min 二层 1024 -5120.867 0.862 0.881 0.786 0.407 0.691 10.4 三层 256-128-64 0.946 0.945 0.897 0.876 0.885 0.886 3.2 三层 128-128-128 0.934 0.933 0.883 0.882 0.855 0.873 2.8 三层 512-256-128 0.943 0.943 0.916 0.899 0.857 0.891 6.5 四层 256-256-128-64 0.949 0.950 0.881 0.893 0.882 0.885 4.8 四层 512-256-256-128 0.944 0.944 0.892 0.864 0.846 0.867 8.5 四层 1024 -512-256-1280.997 0.996 0.993 0.991 0.993 0.992 13.4
下载: 导出CSV
表 5 输入参数对传热恶化峰值位置的影响程度
Table 5. Degree of influence of input parameters on the heat transfer deterioration peak position
排名 输入参数 归一化SHAP值 1 管径D 0.90 2 压力p 0.83 3 热流密度q′′ 0.62 4 质量流量$\dot m $ 0.38 5 入口温度Tin 0.07
下载: 导出CSV
表 6 原数据集预测误差
Table 6. Prediction errors for original dataest
D/mm xact/m xpre/m e/% 5 0.443 0.438 −1.084 7 0.502 0.498 −0.717 9 0.533 0.527 −1.126
下载: 导出CSV
表 7 缩减数据集预测误差
Table 7. Prediction errors for reduced datasets
D/mm xact/m xpre/m e/% 5 0.443 0.439 −0.813 7 0.502 0.506 0.717 9 0.533 0.536 0.450
下载: 导出CSV
表 8 经验判别预测结果
Table 8. Empirical prediction results
D/mm xpc/m δd/m xpre/m xact/m e/% 5 0.148 0.25 0.398 0.443 −10.16 7 0.207 0.26 0.467 0.502 −6.97 9 0.267 0.27 0.537 0.533 0.75
下载: 导出CSV
-
[1] 范学军, 俞刚. 大庆RP-3航空煤油热物性分析[J]. 推进技术, 2006, 27(2): 187-192. FAN Xuejun, YU Gang. Analysis of thermophysical properties of Daqing RP-3 aviation kerosene[J]. Journal of Propulsion Technology, 2006, 27(2): 187-192. (in Chinese doi: 10.3321/j.issn:1001-4055.2006.02.021FAN Xuejun, YU Gang. Analysis of thermophysical properties of Daqing RP-3 aviation kerosene[J]. Journal of Propulsion Technology, 2006, 27(2): 187-192. (in Chinese) doi: 10.3321/j.issn:1001-4055.2006.02.021 [2] HUANG Dan, WU Zan, SUNDEN B, et al. A brief review on convection heat transfer of fluids at supercritical pressures in tubes and the recent progress[J]. Applied Energy, 2016, 162: 494-505. doi: 10.1016/j.apenergy.2015.10.080 [3] 王彦红, 李素芬. 超临界压力下航空煤油传热恶化判别准则[J]. 推进技术, 2019, 40(11): 2528-2536. WANG Yanhong, LI Sufen. Criterion for heat transfer deterioration of aviation kerosene under supercritical pressures[J]. Journal of Propulsion Technology, 2019, 40(11): 2528-2536. (in ChineseWANG Yanhong, LI Sufen. Criterion for heat transfer deterioration of aviation kerosene under supercritical pressures[J]. Journal of Propulsion Technology, 2019, 40(11): 2528-2536. (in Chinese) [4] 王彦红, 孙文清, 李雨健, 等. 竖直内螺纹管中超临界RP-3航空煤油换热特性数值研究[J]. 工程热物理学报, 2024, 45(3): 826-835. WANG Yanhong, SUN Wenqing, LI Yujian, et al. Numerical research on heat transfer characteristics of supercritical RP-3 aviation kerosene in vertical internally-ribbed tubes[J]. Journal of Engineering Thermophysics, 2024, 45(3): 826-835. (in ChineseWANG Yanhong, SUN Wenqing, LI Yujian, et al. Numerical research on heat transfer characteristics of supercritical RP-3 aviation kerosene in vertical internally-ribbed tubes[J]. Journal of Engineering Thermophysics, 2024, 45(3): 826-835. (in Chinese) [5] 李中洲, 朱惠人. 超临界压力下航空煤油传热特性[J]. 推进技术, 2011, 32(2): 261-265. LI Zhongzhou, ZHU Huiren. Heat transfer characteristics of kerosene in micro-channel under supercritical pressure[J]. Journal of Propulsion Technology, 2011, 32(2): 261-265. (in ChineseLI Zhongzhou, ZHU Huiren. Heat transfer characteristics of kerosene in micro-channel under supercritical pressure[J]. Journal of Propulsion Technology, 2011, 32(2): 261-265. (in Chinese) [6] 陈玮玮. 管内超临界流动传热特性及应用研究[D]. 南京: 南京航空航天大学, 2016. CHEN Weiwei. Research on the heat transfer characteristics and application of in-tube supercritical flow[D]. Nanjing: Nanjing University of Aeronautics and Astronautics, 2016. (in ChineseCHEN Weiwei. Research on the heat transfer characteristics and application of in-tube supercritical flow[D]. Nanjing: Nanjing University of Aeronautics and Astronautics, 2016. (in Chinese) [7] 王彦红, 李素芬, 东明, 等. 超临界压力航空煤油热声振荡与传热恶化实验研究[J]. 推进技术, 2016, 37(3): 401-410. WANG Yanhong, LI Sufen, DONG Ming, et al. Experimental studies on thermoacoustic oscillation and heat transfer deterioration of aviation kerosene under supercritical pressure[J]. Journal of Propulsion Technology, 2016, 37(3): 401-410. (in ChineseWANG Yanhong, LI Sufen, DONG Ming, et al. Experimental studies on thermoacoustic oscillation and heat transfer deterioration of aviation kerosene under supercritical pressure[J]. Journal of Propulsion Technology, 2016, 37(3): 401-410. (in Chinese) [8] 王彦红. 超临界压力航空煤油热声振荡与传热恶化的机理及预测研究[D]. 大连: 大连理工大学, 2016. WANG Yanhong. Studies on mechanism and prediction of thermo-acoustic oscillation and heat transfer deterioration of aviation kerosene at supercritical pressures[D]. Dalian: Dalian University of Technology, 2016. (in ChineseWANG Yanhong. Studies on mechanism and prediction of thermo-acoustic oscillation and heat transfer deterioration of aviation kerosene at supercritical pressures[D]. Dalian: Dalian University of Technology, 2016. (in Chinese) [9] 王彦红, 李素芬, 赵星海. 超临界压力航空煤油不稳定流动实验[J]. 航空动力学报, 2018, 33(12): 2838-2844. WANG Yanhong, LI Sufen, ZHAO Xinghai. Experiment on flow instability of aviation kerosene under supercritical pressures[J]. Journal of Aerospace Power, 2018, 33(12): 2838-2844. (in ChineseWANG Yanhong, LI Sufen, ZHAO Xinghai. Experiment on flow instability of aviation kerosene under supercritical pressures[J]. Journal of Aerospace Power, 2018, 33(12): 2838-2844. (in Chinese) [10] 李海旺, 王渊, 朱剑琴, 等. 超临界碳氢燃料流动换热的一维模型[J]. 航空动力学报, 2017, 32(10): 2330-2337. LI Haiwang, WANG Yuan, ZHU Jianqin, et al. One-dimensional model of flow and heat transfer of supercritical hydrocarbon fuel[J]. Journal of Aerospace Power, 2017, 32(10): 2330-2337. (in ChineseLI Haiwang, WANG Yuan, ZHU Jianqin, et al. One-dimensional model of flow and heat transfer of supercritical hydrocarbon fuel[J]. Journal of Aerospace Power, 2017, 32(10): 2330-2337. (in Chinese) [11] PIZZARELLI M. The status of the research on the heat transfer deterioration in supercritical fluids: a review[J]. International Communications in Heat and Mass Transfer, 2018, 95: 132-138. doi: 10.1016/j.icheatmasstransfer.2018.04.006 [12] XIE Jingzhe, LIU Dechao, YAN Hongbin, et al. A review of heat transfer deterioration of supercritical carbon dioxide flowing in vertical tubes: Heat transfer behaviors, identification methods, critical heat fluxes, and heat transfer correlations[J]. International Journal of Heat and Mass Transfer, 2020, 149: 119233. doi: 10.1016/j.ijheatmasstransfer.2019.119233 [13] 王彦红, 东明, 李素芬, 等. 竖直下降圆管内超临界压力RP-3航空煤油传热恶化实验研究[J]. 推进技术, 2022, 43(6): 200635. WANG Yanhong, DONG Ming, LI Sufen, et al. Experimental study on heat transfer deterioration of supercritical-pressure RP-3 aviation kerosene in a vertical downward round tube[J]. Journal of Propulsion Technology, 2022, 43(6): 200635. (in ChineseWANG Yanhong, DONG Ming, LI Sufen, et al. Experimental study on heat transfer deterioration of supercritical-pressure RP-3 aviation kerosene in a vertical downward round tube[J]. Journal of Propulsion Technology, 2022, 43(6): 200635. (in Chinese) [14] 陶凯航, 朱剑琴, 程泽源. 基于人工神经网络模型的超临界RP-3热物性计算[J]. 航空动力学报, 2023, 38(4): 806-815. TAO Kaihang, ZHU Jianqin, CHENG Zeyuan. Calculation of thermophysical properties of supercritical RP-3 based on artificial neural network model[J]. Journal of Aerospace Power, 2023, 38(4): 806-815. (in ChineseTAO Kaihang, ZHU Jianqin, CHENG Zeyuan. Calculation of thermophysical properties of supercritical RP-3 based on artificial neural network model[J]. Journal of Aerospace Power, 2023, 38(4): 806-815. (in Chinese) [15] CHANG Wanli, CHU Xu, BINTE SHAIK FAREED A F, et al. Heat transfer prediction of supercritical water with artificial neural networks[J]. Applied Thermal Engineering, 2018, 131: 815-824. doi: 10.1016/j.applthermaleng.2017.12.063 [16] 周飞燕, 金林鹏, 董军. 卷积神经网络研究综述[J]. 计算机学报, 2017, 40(6): 1229-1251. ZHOU Feiyan, JIN Linpeng, DONG Jun. Review of convolutional neural network[J]. Chinese Journal of Computers, 2017, 40(6): 1229-1251. (in ChineseZHOU Feiyan, JIN Linpeng, DONG Jun. Review of convolutional neural network[J]. Chinese Journal of Computers, 2017, 40(6): 1229-1251. (in Chinese) [17] MOHSENI M, BAZARGAN M. The effect of the low Reynolds number k-e turbulence models on simulation of the enhanced and deteriorated convective heat transfer to the supercritical fluid flows[J]. Heat and Mass Transfer, 2011, 47(5): 609-619. doi: 10.1007/s00231-010-0753-9 [18] HE S, KIM W S, BAE J H. Assessment of performance of turbulence models in predicting supercritical pressure heat transfer in a vertical tube[J]. International Journal of Heat and Mass Transfer, 2008, 51(19/20): 4659-4675. [19] KIM W S, HE S, JACKSON J D. Assessment by comparison with DNS data of turbulence models used in simulations of mixed convection[J]. International Journal of Heat and Mass Transfer, 2008, 51(5/6): 1293-1312. [20] TAO Zhi, CHENG Zeyuan, ZHU Jianqin, et al. Effect of turbulence models on predicting convective heat transfer to hydrocarbon fuel at supercritical pressure[J]. Chinese Journal of Aeronautics, 2016, 29(5): 1247-1261. doi: 10.1016/j.cja.2016.08.007 [21] LAUNDER B E, SHARMA B I. Application of the energy-dissipation model of turbulence to the calculation of flow near a spinning disc[J]. Letters in Heat and Mass Transfer, 1974, 1(2): 131-137. doi: 10.1016/0094-4548(74)90150-7 [22] 张斌, 张春本, 邓宏武, 等. 超临界压力下碳氢燃料在竖直圆管内换热特性[J]. 航空动力学报, 2012, 27(3): 595-603. ZHANG Bin, ZHANG Chunben, DENG Hongwu, et al. Heat transfer characteristics of hydrocarbon fuel at supercritical pressure in vertical circular tubes[J]. Journal of Aerospace Power, 2012, 27(3): 595-603. (in ChineseZHANG Bin, ZHANG Chunben, DENG Hongwu, et al. Heat transfer characteristics of hydrocarbon fuel at supercritical pressure in vertical circular tubes[J]. Journal of Aerospace Power, 2012, 27(3): 595-603. (in Chinese) [23] DENG H W, ZHANG C B, XU G Q, et al. Density measurements of endothermic hydrocarbon fuel at sub- and supercritical conditions[J]. Journal of Chemical & Engineering Data, 2011, 56(6): 2980-2986. [24] LIU Bo, ZHU Yinhai, YAN Junjie, et al. Experimental investigation of convection heat transfer of n-decane at supercritical pressures in small vertical tubes[J]. International Journal of Heat and Mass Transfer, 2015, 91: 734-746. doi: 10.1016/j.ijheatmasstransfer.2015.07.006 [25] ZHANG C B. Investigation of flow and heat transfer characteristics of hydrocarbon fuel at supercritical pressures[D]. Beijing: Beihang University, 2011. [26] 杨观赐, 杨静, 李少波, 等. 基于Dopout与ADAM优化器的改进CNN算法[J]. 华中科技大学学报(自然科学版), 2018, 46(7): 122-127. YANG Guanci, YANG Jing, LI Shaobo, et al. Modified CNN algorithm based on dropout and ADAM optimizer[J]. Journal of Huazhong University of Science and Technology (Natural Science Edition), 2018, 46(7): 122-127. (in ChineseYANG Guanci, YANG Jing, LI Shaobo, et al. Modified CNN algorithm based on dropout and ADAM optimizer[J]. Journal of Huazhong University of Science and Technology (Natural Science Edition), 2018, 46(7): 122-127. (in Chinese) [27] 蒋昂波, 王维维. ReLU激活函数优化研究[J]. 传感器与微系统, 2018, 37(2): 50-52. JIANG Angbo, WANG Weiwei. Research on optimization of ReLU activation function[J]. Transducer and Microsystem Technologies, 2018, 37(2): 50-52. (in ChineseJIANG Angbo, WANG Weiwei. Research on optimization of ReLU activation function[J]. Transducer and Microsystem Technologies, 2018, 37(2): 50-52. (in Chinese) [28] 夏凡. 基于l2正则化的深度ReLU神经网络的泛化性能[D]. 武汉: 湖北大学, 2022. XIA Fan. The generalization performance of deep relu neural networks based on l2 regularization [D]. Wuhan: Hubei University, 2022. (in ChineseXIA Fan. The generalization performance of deep relu neural networks based on l2 regularization [D]. Wuhan: Hubei University, 2022. (in Chinese) [29] 周安众, 罗可. 一种卷积神经网络的稀疏性Dropout正则化方法[J]. 小型微型计算机系统, 2018, 39(8): 1674-1679. ZHOU Anzhong, LUO Ke. Sparse dropout regularization method for convolutional neural networks[J]. Journal of Chinese Computer Systems, 2018, 39(8): 1674-1679. (in ChineseZHOU Anzhong, LUO Ke. Sparse dropout regularization method for convolutional neural networks[J]. Journal of Chinese Computer Systems, 2018, 39(8): 1674-1679. (in Chinese) [30] 吕炜, 陈永刚, 沈晨. 带L2正则化项的神经网络逆向迭代算法收敛性分析[J]. 信息技术与信息化, 2015(6): 183-184, 186. LYU Wei, CHEN Yonggang, SHEN Chen. Convergence analysis of neural network inverse iterative algorithm with L2 regularization term[J]. Information Technology and Informatization, 2015(6): 183-184, 186. (in ChineseLYU Wei, CHEN Yonggang, SHEN Chen. Convergence analysis of neural network inverse iterative algorithm with L2 regularization term[J]. Information Technology and Informatization, 2015(6): 183-184, 186. (in Chinese) [31] 武妍, 张立明. 神经网络的泛化能力与结构优化算法研究[J]. 计算机应用研究, 2002, 19(6): 21-25, 84. WU Yan, ZHANG Liming. A survey of research work on neural network generalization and structure optimization algorithms[J]. Application Research of Computers, 2002, 19(6): 21-25, 84. (in ChineseWU Yan, ZHANG Liming. A survey of research work on neural network generalization and structure optimization algorithms[J]. Application Research of Computers, 2002, 19(6): 21-25, 84. (in Chinese) [32] 王彦红, 李素芬, 赵星海. 超临界压力下航空煤油传热恶化的分析与预测[J]. 化工学报, 2018, 69(12): 5056-5064. WANG Yanhong, LI Sufen, ZHAO Xinghai. Analysis and prediction of heat transfer deterioration of aviation kerosene under supercritical pressures[J]. CIESC Journal, 2018, 69(12): 5056-5064. (in ChineseWANG Yanhong, LI Sufen, ZHAO Xinghai. Analysis and prediction of heat transfer deterioration of aviation kerosene under supercritical pressures[J]. CIESC Journal, 2018, 69(12): 5056-5064. (in Chinese) -
点击查看大图
计量
- 文章访问数: 586
- HTML浏览量: 194
- PDF量: 25
- 被引次数: 0