基于热力学计算与机器学习的增材制造镍基高温合金裂纹敏感性预测模型

doi:10.11900/0412.1961.2023.00050

金属学报

2023, Vol. 59

Issue (8): 1075-1086 DOI: 10.11900/0412.1961.2023.00050

研究论文

本期目录 | 过刊浏览

基于热力学计算与机器学习的增材制造镍基高温合金裂纹敏感性预测模型

穆亚航¹^,², 张雪¹^,², 陈梓名³, 孙晓峰¹(

), 梁静静¹(

), 李金国¹, 周亦胄¹

¹中国科学院金属研究所师昌绪先进材料创新中心沈阳 110016
²中国科学技术大学材料科学与工程学院沈阳 110016
³北京科技大学智能科学与技术学院北京 100083

Modeling of Crack Susceptibility of Ni-Based Superalloy for Additive Manufacturing via Thermodynamic Calculation and Machine Learning

MU Yahang¹^,², ZHANG Xue¹^,², CHEN Ziming³, SUN Xiaofeng¹(

), LIANG Jingjing¹(

), LI Jinguo¹, ZHOU Yizhou¹

¹Shi -changxu Innovation Center for Advanced Materials, Institute of Metal Research, Chinese Academy of Sciences, Shenyang 110016, China
²School of Materials Science and Engineering, University of Science and Technology of China, Shenyang 110016, China
³School of Intelligence Science and Technology, University of Science and Technology Beijing, Beijing 100083, China

引用本文:

穆亚航, 张雪, 陈梓名, 孙晓峰, 梁静静, 李金国, 周亦胄. 基于热力学计算与机器学习的增材制造镍基高温合金裂纹敏感性预测模型[J]. 金属学报, 2023, 59(8): 1075-1086.
Yahang MU, Xue ZHANG, Ziming CHEN, Xiaofeng SUN, Jingjing LIANG, Jinguo LI, Yizhou ZHOU. Modeling of Crack Susceptibility of Ni-Based Superalloy for Additive Manufacturing via Thermodynamic Calculation and Machine Learning[J]. Acta Metall Sin, 2023, 59(8): 1075-1086.

全文: PDF(3364 KB) HTML

摘要:

利用实验和热力学计算研究了镍基高温合金的增材制造裂纹敏感性，发现镍基高温合金增材制造裂纹以热裂纹为主，热裂纹敏感性系数(HSC)与实测裂纹面积分数相关性高。基于实验数据和热力学计算结果，建立高温合金裂纹敏感性的机器学习预测模型，该模型具有良好的预测和泛化能力，在训练集上和验证集上的相关性系数(R²)分别达到0.96和0.81，可以快速有效地计算出高温合金的热裂纹敏感性。采用SHapley Additive exPlanation (SHAP)方法对模型中的输入参数进行特征分析，获得了合金元素对裂纹敏感性的影响规律，并根据SHAP值对合金元素的裂纹敏感性影响进行了排序。结果表明，沉淀强化元素Ti、Al和微量元素C、B对镍基高温合金的裂纹敏感性的影响较大，其余合金元素对裂纹敏感性的综合影响排序为：Re > W > Cr > Mo > Ta > Co。

关键词 ：镍基高温合金, 裂纹敏感性, 增材制造, 机器学习, 热力学计算

Abstract：

The rapid development of aeroengines has led to high demand heat resistant blades. As a result, fabricating techniques and designing materials have taken center stage in producing aeroengines. Additive manufacturing (AM), which integrates design and manufacturing, has advantages in preparing blades with complex cavity structures. However, commercial Ni-based superalloys have poor additive manufacturability and are prone to defects such as cracks, severely hindering the development of the AM of superalloy blades. Therefore, finding a high-performance superalloy with excellent additive manufacturability is necessary. To alleviate this problem, many crack susceptibility criteria and test methods have recently been proposed to evaluate the crack susceptibility of alloys from a compositional and/or process point of view. However, the rapid prediction of the crack susceptibility of superalloys remains a challenge, hindering the widespread screening and designing of superalloys for AM. Nevertheless, using machine learning (ML) in conjunction with thermodynamic calculation may effectively predict the properties of alloys, and this combination is anticipated to grow as an important tool for designing alloys with low crack susceptibility for AM. Based on the aforementioned context, this study reports the development of an ML prediction model after combining experimental data and thermodynamic calculations to establish a Ni-based alloy crack susceptibility database. This ML model has an excellent prediction effect (R² = 0.96 on the training set and R² = 0.81 on the validation set) and enables accurate prediction of the crack susceptibility of the experimental alloys and published alloys. It is verified that a hot crack is the most typical type of crack in Ni-based superalloys during AM. The influence of elements on crack susceptibility is also analyzed using the SHapley Additive exPlanation method. Precipitation-strengthening (Al and Ti) and trace (C and B) elements greatly influence crack susceptibility. A small amount of Re can inhibit cracks, but excessive amounts produce a topologically close-packed phase, deteriorating the crack susceptibility and mechanical properties. The influence of other alloying elements on crack susceptibility is roughly ranked as follows: Re, W, Cr, Mo, Ta, and Co, which can provide a screening method for the composition design of subsequent AMed superalloys.

Key words： Ni-based superalloy crack susceptibility additive manufacturing machine learning thermodynamic calculation

收稿日期: 2023-02-10

ZTFLH:

TG146.1

基金资助:国家科技重大专项项目(Y2019-VII-0011-0151);国家科技重大专项项目(P2022-C-IV-002-001)

通讯作者: 孙晓峰，xfsun@imr.ac.cn，主要从事高温合金材料研制与构件制备的研究;梁静静，jjliang@imr.ac.cn，主要从事增材制造高温合金材料研发与工艺优化的研究

Corresponding author: SUN Xiaofeng, professor, Tel:(024)23971887, E-mail: xfsun@imr.ac.cn;LIANG Jingjing, professor, Tel:(024)23971787, E-mail: jjliang@imr.ac.cn

作者简介: 穆亚航，男，1997年生，博士生

	服务

	把本文推荐给朋友
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	穆亚航
	张雪
	陈梓名
	孙晓峰
	梁静静
	李金国
	周亦胄
44.8K

链接本文:

https://www.ams.org.cn/CN/10.11900/0412.1961.2023.00050 或 https://www.ams.org.cn/CN/Y2023/V59/I8/1075

图1 镍基合金裂纹敏感性预测模型建立流程

表1 增材制造镍基合金的成分设计范围 (mass fraction / %)

Type of crack	Crack susceptibility criteria	Formula	Ref.
Hot-tearing crack	FR	$F R = T l - T s$	[18]
	CSC	$C S C = t V t R = t f s = 0.99 - t f s = 0.90 t f s = 0.90 - t f s = 0.40$	[19]
	HSC	$H S C = T V T R = T f s = 0.99 - T f s = 0.90 T f s = 0.90 - T f s = 0.40$	[20]
	SCI	$S C I = d T d (f s 1 / 2) f s 1 / 2 → 1$	[4]
Solid-state crack	SAC	$S A C = d V f γ / d T, T ∈ [T γ *, T s]$	[5]
	M_SAC	$M S A C = [A l] + 0.5 [T i] + 0.3 [N b] + 0.15 [T a]$	[21]

表2 增材制造镍基高温合金的裂纹敏感性系数[4,5,18~21]

图2 沉积态下4#合金(Ni-8Co-10Cr-8W-4.5Al-1Ti-5Ta)试样中的微观组织SEM像，TCP相的BSE像、TEM像及EDS元素面分布图

图3 不同合金的裂纹面积分数及裂纹敏感性系数

图4 各裂纹敏感性系数与裂纹面积分数的关系图

图5 热裂纹敏感性(HSC)随单一元素含量的变化趋势

图6 镍基高温合金增材制造裂纹敏感性数据库数据散点图

图7 不同算法在训练集的拟合效果

图8 机器学习(ML)模型对不同裂纹敏感性系数在训练集中的拟合效果

图9 机器学习模型对不同裂纹敏感性系数在验证集中的预测效果

图10 机器学习裂纹敏感性预测模型特征参数的重要性评估

1	Lin X, Huang W D. High performance metal additive manufacturing technology applied in aviation field [J]. Mater. China, 2015, 34: 684
1	林鑫, 黄卫东. 应用于航空领域的金属高性能增材制造技术 [J]. 中国材料进展, 2015, 34: 684
2	Sun X F, Song W, Liang J J, et al. Research and development in materials and processes of superalloy fabricated by laser additive manufacturing [J]. Acta Metall. Sin., 2021, 57: 1471 doi: 10.11900/0412.1961.2021.00371
2	孙晓峰, 宋巍, 梁静静等. 激光增材制造高温合金材料与工艺研究进展 [J]. 金属学报, 2021, 57: 1471 doi: 10.11900/0412.1961.2021.00371
3	Rappaz M, Drezet J M, Gremaud M. A new hot-tearing criterion [J]. Metall. Mater. Trans., 1999, 30A: 449
4	Kou S. A criterion for cracking during solidification [J]. Acta Mater., 2015, 88: 366 doi: 10.1016/j.actamat.2015.01.034
5	Yu H, Liang J J, Bi Z N, et al. Computational design of novel Ni superalloys with low crack susceptibility for additive manufacturing [J]. Metall. Mater. Trans., 2022, 53A: 1945
6	Xu J H, Kontis P, Peng R L, et al. Modelling of additive manufacturability of nickel-based superalloys for laser powder bed fusion [J]. Acta Mater., 2022, 240: 118307 doi: 10.1016/j.actamat.2022.118307
7	Jain S. Benchmarking hot cracking behavior during localised melting using a new standard test methodolgy and thermodynamic predictors [D]. Ames: Iowa State University, 2021
8	Qin H, Yang G Y, Zheng X W, et al. Effect of Gd content on hot-tearing susceptibility of Mg-6Zn-xGd casting alloys [J]. China Foundry, 2022, 19: 131 doi: 10.1007/s41230-022-1117-z
9	Qian X, Yang R G. Machine learning for predicting thermal transport properties of solids [J]. Mater. Sci. Eng., 2021, R146: 100642
10	Hart G L W, Mueller T, Toher C, et al. Machine learning for alloys [J]. Nat. Rev. Mater., 2021, 6: 730 doi: 10.1038/s41578-021-00340-w
11	Johnson N S, Vulimiri P S, To A C, et al. Invited review: Machine learning for materials developments in metals additive manufacturing [J]. Addit. Manuf., 2020, 36: 101641
12	Zhu C P, Li C, Wu D, et al. A titanium alloys design method based on high-throughput experiments and machine learning [J]. J. Mater. Res. Technol., 2021, 11: 2336 doi: 10.1016/j.jmrt.2021.02.055
13	Menou E, Rame J, Desgranges C, et al. Computational design of a single crystal nickel-based superalloy with improved specific creep endurance at high temperature [J]. Comp. Mater. Sci., 2019, 170: 109194 doi: 10.1016/j.commatsci.2019.109194
14	Khatavkar N, Swetlana S, Singh A K. Accelerated prediction of Vickers hardness of Co- and Ni-based superalloys from microstructure and composition using advanced image processing techniques and machine learning [J]. Acta Mater., 2020, 196: 295 doi: 10.1016/j.actamat.2020.06.042
15	Wu J J, Li Y H, Zhao J B, et al. Prediction of residual stress induced by laser shock processing based on artificial neural networks for FGH4095 superalloy [J]. Mater. Lett., 2021, 286: 129269 doi: 10.1016/j.matlet.2020.129269
16	Zhu Y L, Duan F M, Yong W, et al. Creep rupture life prediction of nickel-based superalloys based on data fusion [J]. Comp. Mater. Sci., 2022, 211: 111560 doi: 10.1016/j.commatsci.2022.111560
17	Luo Y W, Zhang B, Feng X, et al. Pore-affected fatigue life scattering and prediction of additively manufactured Inconel 718: An investigation based on miniature specimen testing and machine learning approach [J]. Mater. Sci. Eng., 2021, A802: 140693
18	Singer A R E, Jennings P H. Hot-shortness of the aluminium-silicon alloys of commercial purity [J]. J. Inst. Met., 1946, 73: 197
19	Clyne T W, Davies G J. The influence of composition on solidification cracking susceptibility in binary alloy systems [J]. Br. Foundryman, 1981, 74: 65
20	Yu H N, Liu S M, Zhou L, et al. Study on solidification behavior and hot tearing susceptibility of Mg-2xY-xNi alloys [J]. Int. J. Metalcast., 2021, 15: 995 doi: 10.1007/s40962-020-00531-1
21	Tang Y T, Panwisawas C, Ghoussoub J N, et al. Alloys-by-design: Application to new superalloys for additive manufacturing [J]. Acta Mater., 2021, 202: 417 doi: 10.1016/j.actamat.2020.09.023
22	Xu B, Yin H Q, Jiang X, et al. Computational materials design: Composition optimization to develop novel Ni-based single crystal superalloys [J]. Comp. Mater. Sci., 2022, 202: 111021 doi: 10.1016/j.commatsci.2021.111021
23	Shi Z X, Dong J X, Zhang M C, et al. Solidification characteristics and hot tearing susceptibility of Ni-based superalloys for turbocharger turbine wheel [J]. Trans. Nonferrous Met. Soc. China, 2014, 24: 2737 doi: 10.1016/S1003-6326(14)63405-1
24	Zhao Y S, Zhang J, Song F Y, et al. Effect of trace boron on microstructural evolution and high temperature creep performance in Re-contianing single crystal superalloys [J]. Prog. Nat. Sci. Mater. Int., 2020, 30: 371 doi: 10.1016/j.pnsc.2020.05.003
25	Wang H W, Yang J X, Meng J, et al. Effects of B content on microstructure and high-temperature stress rupture properties of a high chromium polycrystalline nickel-based superalloy [J]. J. Alloys Compd., 2021, 860: 157929 doi: 10.1016/j.jallcom.2020.157929
26	Froeliger T, Després A, Toualbi L, et al. Interplay between solidification microsegregation and complex precipitation in a γ/γ' cobalt-based superalloy elaborated by directed energy deposition [J]. Mater. Charact., 2022, 194: 112376 doi: 10.1016/j.matchar.2022.112376
27	Xiong J, Shi S Q, Zhang T Y. Machine learning of phases and mechanical properties in complex concentrated alloys [J]. J. Mater. Sci. Technol., 2021, 87: 133 doi: 10.1016/j.jmst.2021.01.054
28	Sun X F, Jin T, Zhou Y Z, et al. Research progress of nickel-base single crystal superalloys [J]. Mater. China, 2012, 31(12): 1
28	孙晓峰, 金涛, 周亦胄等. 镍基单晶高温合金研究进展 [J]. 中国材料进展, 2012, 31(12): 1
29	Zhou Y Z, Volek A. Effect of carbon additions on hot tearing of a second generation nickel-base superalloy [J]. Mater. Sci. Eng., 2008, A479: 324
30	Zhou W Z, Tian Y S, Tan Q B, et al. Effect of carbon content on the microstructure, tensile properties and cracking susceptibility of IN738 superalloy processed by laser powder bed fusion [J]. Addit. Manuf., 2022, 58: 103016
31	Dong Y, Hao M S, Mu Y H, et al. Effect of carbon content on the microstructure and mechanical properties of GH3230 alloy formed by laser melting deposition [J]. Adv. Eng. Mater., 2023: 2201887
32	Hu Y, Yang X K, Kang W J, et al. Effect of Zr content on crack formation and mechanical properties of IN738LC processed by selective laser melting [J]. Trans. Nonferrous Met. Soc. China, 2021, 31: 1350 doi: 10.1016/S1003-6326(21)65582-6
33	Yu Q, Wang C S, Zhao Z S, et al. New Ni-based superalloys designed for laser additive manufacturing [J]. J. Alloys Compd., 2021, 861: 157979 doi: 10.1016/j.jallcom.2020.157979
34	Park J U, Jun S Y, Lee B H, et al. Alloy design of Ni-based superalloy with high γ' volume fraction suitable for additive manufacturing and its deformation behavior [J]. Addit. Manuf., 2022, 52: 102680

[1]	江河, 佴启亮, 徐超, 赵晓, 姚志浩, 董建新. 镍基高温合金疲劳裂纹急速扩展敏感温度及成因[J]. 金属学报, 2023, 59(9): 1190-1200.
[2]	卢楠楠, 郭以沫, 杨树林, 梁静静, 周亦胄, 孙晓峰, 李金国. 激光增材修复单晶高温合金的热裂纹形成机制[J]. 金属学报, 2023, 59(9): 1243-1252.
[3]	郑亮, 张强, 李周, 张国庆. 增/降氧过程对高温合金粉末表面特性和合金性能的影响：粉末存储到脱气处理[J]. 金属学报, 2023, 59(9): 1265-1278.
[4]	王磊, 刘梦雅, 刘杨, 宋秀, 孟凡强. 镍基高温合金表面冲击强化机制及应用研究进展[J]. 金属学报, 2023, 59(9): 1173-1189.
[5]	张禄, 余志伟, 张磊成, 江荣, 宋迎东. GH4169高温合金热机械疲劳循环损伤机理及数值模拟[J]. 金属学报, 2023, 59(7): 871-883.
[6]	袁江淮, 王振玉, 马冠水, 周广学, 程晓英, 汪爱英. Cr₂AlC涂层相结构演变对力学性能的影响[J]. 金属学报, 2023, 59(7): 961-968.
[7]	吴东江, 刘德华, 张子傲, 张逸伦, 牛方勇, 马广义. 电弧增材制造2024铝合金的微观组织与力学性能[J]. 金属学报, 2023, 59(6): 767-776.
[8]	李述军, 侯文韬, 郝玉琳, 杨锐. 3D打印医用钛合金多孔材料力学性能研究进展[J]. 金属学报, 2023, 59(4): 478-488.
[9]	刘来娣, 丁彪, 任维丽, 钟云波, 王晖, 王秋良. DZ445镍基高温合金高温长时间氧化形成的多层膜结构[J]. 金属学报, 2023, 59(3): 387-398.
[10]	冀秀梅, 侯美伶, 王龙, 刘玠, 高克伟. 基于机器学习的中厚板变形抗力模型建模与应用[J]. 金属学报, 2023, 59(3): 435-446.
[11]	唐伟能, 莫宁, 侯娟. 增材制造镁合金技术现状与研究进展[J]. 金属学报, 2023, 59(2): 205-225.
[12]	杨累, 赵帆, 姜磊, 谢建新. 机器学习辅助2000 MPa级弹簧钢成分和热处理工艺开发[J]. 金属学报, 2023, 59(11): 1499-1512.
[13]	于少霞, 王麒, 邓想涛, 王昭东. GH3600镍基高温合金极薄带的制备及尺寸效应[J]. 金属学报, 2023, 59(10): 1365-1375.
[14]	侯旭儒, 赵琳, 任淑彬, 彭云, 马成勇, 田志凌. 热输入对电弧增材制造船用高强钢组织与力学性能的影响[J]. 金属学报, 2023, 59(10): 1311-1323.
[15]	方远志, 戴国庆, 郭艳华, 孙中刚, 刘红兵, 袁秦峰. 激光摆动对激光熔化沉积钛合金微观组织及力学性能的影响[J]. 金属学报, 2023, 59(1): 136-146.

Viewed

Full text

Abstract

Cited

Shared

Discussed