Sequential Bayesian Inference of Finite-strain Visco-elastic Visco-plastic model parameters of 22-month aged PA12 bulk material printed along different directions
收藏资源简介：
These are the data related to aged PA12 (22 months) following the methodology described in the following publication in which non-aged PA12 has been tested: title = "Bayesian inference of high-dimensional finite-strain visco-elastic-visco-plastic model parameters for additive manufactured polymers and neural network based material parameters generator.",journal = "International Journal of Solids and Structures",year = "2023",volume = "283",pages = "112470",doi = "10.1016/j.ijsolstr.2023.112470",author = "Wu, Ling and Anglade, Cyrielle and Cobian, Lucia and Monclus, Miguel and Segurado, Javier and Karayagiz, Fatma and Freitas, Ubiratan and Noels Ludovic" Contrarily to the non-aged material, since high-strain-rate tests are not available, only 5 Maxwell's branches are considered herein. Description BI code and results of the inference of a pressure-dependent visco-elastic visco-plastic model developed in [NGU16] with a umat implementation in https://gitlab.uliege.be/moammm/moammmPublic/code/-/tree/main/MaterialModels/FiniteStrain/Finite_VEVP The sequential BI is described in [WU23] The experimental protocol is reported in [COB22,COB22b] but is herein applied on aged PA12 To run the BI you need the open source code https://gitlab.onelab.info/cm3/cm3Libraries If you use these data or model, we would be grateful if you could cite the related papers: [WU23] L. Wu, C. Anglade, L. Cobian, M. Monclus, J. Segurado, F. Karayagiz, U. Santos Freitas, L. Noels, Bayesian inference of high-dimensional finite-strain visco-elastic-visco-plastic model parameters for additive manufactured polymers and neural network based material parameters generator, International Journal of Solids and Structures (2023) 112470: https://doi.org/10.1016/j.ijsolstr.2023.112470 [COB22] L. Cobian, M. Rueda-Ruiz, J.P. Fernandez-Blazquez, V. Martinez, F. Galvez, F. Karayagiz, T. Lück, J. Segurado, M.A. Monclus, Micromechanical characterization of the material response in a PA12-SLS fabricated lattice structure and its correlation with bulk behaviour, Polymer Testing 110 (2022) 107556: https://doi.org/10.1016/j.polymertesting.2022.107556 (in Open access) [COB22b] Data of “. Cobian, M. Rueda-Ruiz, J.P. Fernandez-Blazquez, V. Martinez, F. Galvez, F. Karayagiz, T. Lück, J. Segurado, M.A. Monclus, Micromechanical characterization of the material response in a PA12-SLS fabricated lattice structure and its correlation with bulk behaviour, Polymer Testing 110 (2022) 107556” http://dx.doi.org/10.5281/zenodo.6136935 (in Open access) [NGU16] V. D. Nguyen, F. Lani, T. Pardoen, X. Morelle, L. Noels, A large strain hyperelastic viscoelastic-viscoplastic-damage constitutive model based on a multi-mechanism non-local damage continuum for amorphous glassy polymers. International Journal of Solids and Structures 96 (2016): 192-216; https://dx.doi.org/10.1016/j.ijsolstr.2016.06.008, Open access: https://orbi.uliege.be/handle/2268/197898 Directories All the codes and experimental results are in three directories: Experiment_PA12_AGED: experimental data of aged material, see the README.txt in each subdirectory for details BayesianVE: BI of the visco-elastic parameters2.1. PlotExperimentalCurves: to vizualize the experimental curves and prepare the observations for the BI in the VE range2.1.1. Loadcase_H.py and Loadcase_V.py read experimental results and create Load_ExpVE_H.dat and Load_ExpVE_V.dat, which keep the experimental observations and loading conditions to perform the BI.2.1.2. PrintDir_H & PrintDir_V subdirectories with the functions called by Loadcase_H.py and Loadcase_V.py2.1.3. Load_ExpVE_H.dat and Load_ExpVE_V.dat created files with the observations and loading conditions to perform the BI2.2. VE_V and VE_H: BI for viscoelastic properties of "V" specimen (VE_V) and "H" specimen (VE_H)2.2.1. BI_allpos_sequence.py runs the BI using Predict_VETest.py and creates the MCMC_VE_....dat2.2.2. WarmStart = True is used to restart an inference2.2.3. MCMC_VE_....dat in the VE_V and VE_H directories are the BI results2.3. CheckBayRes: to visualize predictions of a BI sample and experimental curves 2.3.1. MCMCRes.py is used to check the numerical predictions of a BI parameter sample (read last sample by default, V or H direction can be selected at line 7)2.3.2. ResKGEmu.py plots the evolution of elastic properties with time2.3.3. uses as input VE_V/MCMC_VE_....dat or VE_H/MCMC_VE_....dat2.3.4. uses local ViscoElasticTest.py, line.geo, line. msh as interface with https://gitlab.onelab.info/cm3/cm3Libraries code2.3.5. uses local functions plotExp.py2.4. ViscoElasticTest.py, line.geo, line.msh: interface with https://gitlab.onelab.info/cm3/cm3Libraries code used by VE_V and VE_H to call the VEVP model BayesianVEVP: BI of the visco-elastic and visco-plastic parameters3.1. PlotExperimentalCurves: to vizualize the experimental curves and prepare the observations for the BI in the VE-VP ranges3.1.1. Loadcase_H.py and Loadcase_V.py read experimental results and create Load_ExpVEVP_H.dat and Load_ExpVEVP_V.dat, which keep the experimental observations and loading conditions to perform BI at the viscoplastic stage.3.1.2. PrintDir_H & PrintDir_V subdirectories with the functions called by Loadcase_H.py and Loadcase_V.py3.1.3. Load_ExpVEVP_H.dat and Load_ExpVEVP_V.dat created files with the observations and loading conditions to perform the BI3.2. VP_V and VP_H: BI for viscoelastic-viscoplastic properties of "V" specimen (VP_V) and "H" specimen (VP_H)3.2.1. BI_allpos_sequence.py runs the BI using Predict_VETest.py and creates the MCMC_VP_....dat3.2.2. WarmStart = True is used to restart an inference 3.2.3. It starts from the VE prosterior as prior, see point 2, and generates a MCMC_VP_?1Step.dat (? being H or V)3.3. CheckBayRes: visualize predictions of a BI sample and experimental curves3.3.1. MCMCRes.py is used to check the numerical predictions with 3 BI parameter samples (inclusing MAP, V or H direction can be selected at line 12) using the samples of BayesianVEVP/VP?/MCMC_VP_?1Step.dat (? being H or V)3.3.2. plot_hist.py is used to plot histograms of all the inferred parameters using the samples of BayesianVEVP/VP?/MCMC_VP_?1Step.dat (? being H or V)3.3.3. Plot_Prop.py plots joints histograms of the inferred parameters using the samples of BayesianVEVP/VP?/MCMC_VP_?_1Step.dat (? being H or V) 3.3.4. ResKGEmu.py plots the evolution of elastic properties with time3.4. VEVPTest.py: interface with https://gitlab.onelab.info/cm3/cm3Libraries code used by VP_V2Step and VP_H2Step to call the VEVP model Figures (reference to the number in [WU23] but for aged PA12) Fig. 5 (Selected observations): From directory BayesianVE/PlotExperimentalCurves/PrintDir_? (? being H or V), run python3 plotExp_T.py or plotExp_C.py Fig. 7: BayesianVEVP/CheckBayRes/Plot_Prop.py with direct = "V" and then with direct = "H" and with Var = [0,1,14,18,22,23,24,25] Fig. 8 (Predictions of 3 inference realisations): BayesianVEVP/CheckBayRes/MCMCRes.py with direct = "V" (requires https://gitlab.onelab.info/cm3/cm3Libraries code) Fig. 9 (Predictions of 3 inference realisations): BayesianVEVP/CheckBayRes/MCMCRes.py with direct = "H" (requires https://gitlab.onelab.info/cm3/cm3Libraries code) Fig. 14A: From directory BayesianVE/PlotExperimentalCurves/PrintDir_? (? being H or V), run python3 plotExp_T.py or plotExp_C.py Fig. 15B: BayesianVEVP/CheckBayRes/Plot_Prop.py with direct = "V", Var = [2,3,8,9,10,11,12,13] and [14,15,16,17,18,19,20,21] Fig. 16B: BayesainVEVP/CheckBayRes/plot_hist.py with direct = "V" Fig. 17B: BayesianVEVP/CheckBayRes/plot_hist.py with direct = "V" Fig. 18B: BayesianVEVP/CheckBayRes/Plot_Prop.py with direct = "H", Var = [2,3,8,9,10,11,12,13] and [14,15,16,17,18,19,20,21] Fig. 19B: BayesianVEVP/CheckBayRes/plot_hist.py with direct = "H" Fig. 20B: BayesianVEVP/CheckBayRes/plot_hist.py with direct = "H" This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 862015.
本数据集针对老化22个月的聚酰胺12（PA12），其研究方法源自下述已对未老化PA12开展测试的学术成果： 标题：《面向增材制造聚合物的高维有限应变黏弹性黏塑性模型参数贝叶斯推断及基于神经网络的材料参数生成器》 期刊：《固体与结构国际期刊》 发表年份：2023年 卷号：283 页码：112470 DOI：10.1016/j.ijsolstr.2023.112470 作者：Wu, Ling、Anglade, Cyrielle、Cobian, Lucia、Monclus, Miguel、Segurado, Javier、Karayagiz, Fatma、Freitas, Ubiratan、Noels Ludovic 与未老化材料不同，因无法获取高应变率测试数据，本文仅考虑5组麦克斯韦（Maxwell）分支单元。 ## 数据集说明 本数据集包含基于[NGU16]提出的压力相关黏弹性黏塑性模型的贝叶斯推断（Bayesian Inference, BI）代码与结果，该模型的用户自定义材料子程序（umat）实现代码位于：https://gitlab.uliege.be/moammm/moammmPublic/code/-/tree/main/MaterialModels/FiniteStrain/Finite_VEVP。 序贯贝叶斯推断方法详见[WU23]，实验流程已在[COB22, COB22b]中公开，本数据集将其应用于老化PA12材料。 运行贝叶斯推断需依赖开源代码库：https://gitlab.onelab.info/cm3/cm3Libraries。 若您使用本数据集或相关模型，恳请引用下述学术文献： [WU23] L. Wu, C. Anglade, L. Cobian, M. Monclus, J. Segurado, F. Karayagiz, U. Santos Freitas, L. Noels, 面向增材制造聚合物的高维有限应变黏弹性黏塑性模型参数贝叶斯推断及基于神经网络的材料参数生成器, 《固体与结构国际期刊》(2023) 112470: https://doi.org/10.1016/j.ijsolstr.2023.112470 [COB22] L. Cobian, M. Rueda-Ruiz, J.P. Fernandez-Blazquez, V. Martinez, F. Galvez, F. Karayagiz, T. Lück, J. Segurado, M.A. Monclus, PA12-SLS制备点阵结构的材料响应微观力学表征及其与块体行为的相关性, 《聚合物测试》110 (2022) 107556: https://doi.org/10.1016/j.polymertesting.2022.107556（开放获取） [COB22b] 文献来源：Cobian, M. Rueda-Ruiz, J.P. Fernandez-Blazquez, V. Martinez, F. Galvez, F. Karayagiz, T. Lück, J. Segurado, M.A. Monclus, PA12-SLS制备点阵结构的材料响应微观力学表征及其与块体行为的相关性, 《聚合物测试》110 (2022) 107556: http://dx.doi.org/10.5281/zenodo.6136935（开放获取） [NGU16] V. D. Nguyen, F. Lani, T. Pardoen, X. Morelle, L. Noels, 基于多机制非局部损伤连续体的无定形玻璃态聚合物大应变超弹性黏弹性黏塑性损伤本构模型, 《固体与结构国际期刊》96 (2016): 192-216; https://dx.doi.org/10.1016/j.ijsolstr.2016.06.008, 开放获取链接：https://orbi.uliege.be/handle/2268/197898 ## 目录结构 所有代码与实验结果均存储于三个主目录下： 1. Experiment_PA12_AGED：老化材料的实验数据，详细说明请参见各子目录下的README.txt文件。 2. BayesianVE：黏弹性参数的贝叶斯推断 2.1 PlotExperimentalCurves：用于可视化实验曲线，并为黏弹性范围内的贝叶斯推断准备观测数据 2.1.1 Loadcase_H.py与Loadcase_V.py：读取实验结果并生成Load_ExpVE_H.dat与Load_ExpVE_V.dat，二者存储了用于贝叶斯推断的实验观测数据与加载条件。 2.1.2 PrintDir_H与PrintDir_V子目录：包含Loadcase_H.py与Loadcase_V.py调用的功能函数。 2.1.3 Load_ExpVE_H.dat与Load_ExpVE_V.dat：存储用于贝叶斯推断的实验观测数据与加载条件的生成文件。 2.2 VE_V与VE_H：分别针对“V”型试样（VE_V）与“H”型试样（VE_H）的黏弹性参数贝叶斯推断 2.2.1 BI_allpos_sequence.py：通过调用Predict_VETest.py执行贝叶斯推断，并生成MCMC_VE_....dat文件。 2.2.2 设置WarmStart = True可重启贝叶斯推断任务。 2.2.3 VE_V与VE_H目录下的MCMC_VE_....dat文件为贝叶斯推断结果，其中MCMC为马尔可夫链蒙特卡洛（Markov Chain Monte Carlo）缩写。 2.3 CheckBayRes：用于可视化贝叶斯推断样本的预测结果与实验曲线 2.3.1 MCMCRes.py：用于验证贝叶斯推断参数样本的数值预测结果（默认读取最后一组样本，可在第7行选择V或H方向）。 2.3.2 ResKGEmu.py：用于绘制弹性属性随时间的演化曲线。 2.3.3 输入文件为VE_V/MCMC_VE_....dat或VE_H/MCMC_VE_....dat。 2.3.4 通过本地文件ViscoElasticTest.py、line.geo、line.msh与https://gitlab.onelab.info/cm3/cm3Libraries代码库实现接口对接。 2.3.5 调用本地功能函数plotExp.py。 2.4 ViscoElasticTest.py、line.geo、line.msh：与https://gitlab.onelab.info/cm3/cm3Libraries代码库的接口文件，供VE_V与VE_H目录调用黏弹性黏塑性（VEVP）模型。 3. BayesianVEVP：黏弹性与黏塑性参数的贝叶斯推断 3.1 PlotExperimentalCurves：用于可视化实验曲线，并为黏弹性-黏塑性范围内的贝叶斯推断准备观测数据 3.1.1 Loadcase_H.py与Loadcase_V.py：读取实验结果并生成Load_ExpVEVP_H.dat与Load_ExpVEVP_V.dat，二者存储了黏塑性阶段贝叶斯推断所需的实验观测数据与加载条件。 3.1.2 PrintDir_H与PrintDir_V子目录：包含Loadcase_H.py与Loadcase_V.py调用的功能函数。 3.1.3 Load_ExpVEVP_H.dat与Load_ExpVEVP_V.dat：存储用于贝叶斯推断的实验观测数据与加载条件的生成文件。 3.2 VP_V与VP_H：分别针对“V”型试样（VP_V）与“H”型试样（VP_H）的黏弹性-黏塑性参数贝叶斯推断 3.2.1 BI_allpos_sequence.py：通过调用Predict_VETest.py执行贝叶斯推断，并生成MCMC_VP_....dat文件。 3.2.2 设置WarmStart = True可重启贝叶斯推断任务。 3.2.3 以黏弹性参数的后验分布作为先验分布（详见第2点），并生成MCMC_VP_?1Step.dat文件（?代表H或V方向）。 3.3 CheckBayRes：用于可视化贝叶斯推断样本的预测结果与实验曲线 3.3.1 MCMCRes.py：用于验证3组贝叶斯推断参数样本（包括最大后验估计Maximum A Posteriori, MAP）的数值预测结果，可在第12行选择V或H方向，输入文件为BayesianVEVP/VP?/MCMC_VP_?1Step.dat（?代表H或V）。 3.3.2 plot_hist.py：基于BayesianVEVP/VP?/MCMC_VP_?1Step.dat文件（?代表H或V）的样本，绘制所有推断参数的直方图。 3.3.3 Plot_Prop.py：基于BayesianVEVP/VP?/MCMC_VP_?_1Step.dat文件（?代表H或V）的样本，绘制推断参数的联合直方图。 3.3.4 ResKGEmu.py：用于绘制弹性属性随时间的演化曲线。 3.4 VEVPTest.py：与https://gitlab.onelab.info/cm3/cm3Libraries代码库的接口文件，供VP_V2Step与VP_H2Step目录调用黏弹性黏塑性（VEVP）模型。 ## 图表说明 以下图表对应[WU23]中的编号，但针对老化PA12材料： - 图5（选定观测数据）：进入目录BayesianVE/PlotExperimentalCurves/PrintDir_?（?取H或V），执行python3 plotExp_T.py或plotExp_C.py命令。 - 图7：执行BayesianVEVP/CheckBayRes/Plot_Prop.py命令，分别设置direct = "V"与direct = "H"，并设置Var = [0,1,14,18,22,23,24,25]。 - 图8（3组推断结果的预测曲线）：进入目录BayesianVEVP/CheckBayRes，执行python3 MCMCRes.py命令，设置direct = "V"（需依赖https://gitlab.onelab.info/cm3/cm3Libraries代码库）。 - 图9（3组推断结果的预测曲线）：进入目录BayesianVEVP/CheckBayRes，执行python3 MCMCRes.py命令，设置direct = "H"（需依赖https://gitlab.onelab.info/cm3/cm3Libraries代码库）。 - 图14A：进入目录BayesianVE/PlotExperimentalCurves/PrintDir_?（?取H或V），执行python3 plotExp_T.py或plotExp_C.py命令。 - 图15B：执行BayesianVEVP/CheckBayRes/Plot_Prop.py命令，设置direct = "V"，Var = [2,3,8,9,10,11,12,13]与Var = [14,15,16,17,18,19,20,21]。 - 图16B：执行BayesianVEVP/CheckBayRes/plot_hist.py命令，设置direct = "V"。 - 图17B：执行BayesianVEVP/CheckBayRes/plot_hist.py命令，设置direct = "V"。 - 图18B：执行BayesianVEVP/CheckBayRes/Plot_Prop.py命令，设置direct = "H"，Var = [2,3,8,9,10,11,12,13]与Var = [14,15,16,17,18,19,20,21]。 - 图19B：执行BayesianVEVP/CheckBayRes/plot_hist.py命令，设置direct = "H"。 - 图20B：执行BayesianVEVP/CheckBayRes/plot_hist.py命令，设置direct = "H"。 本项目获欧盟地平线2020研究与创新计划资助，资助协议编号：862015。
