Application Cases of Inverse Modelling with the PROPTI Framework - Data Set

<strong>Contents</strong> Set of simulation data, supplementary for a paper submitted to (published: 15 June 2019) the Fire Safety Journal, with the title "Application Cases of Inverse Modelling with the PROPTI Framework". See also our project at ResearchGate. This repository contains the complete input data for each IMP run of the mass loss calorimeter, shown in this paper. This comprises of the experimental data files, the templates for the simulation models and the input file for PROPTI. The data base files are provided. This includes the original ones created by PROPTI during the run, as well as the cleaned data base files, used to create the plots, and the extracted best parameter sets per generation. Plots, created during the IMP runs as means of monitoring the progress are also included. Furthermore, the repository contains a small collection of Jupyter notebooks which have been used to process the data base files and create the plots presented in this paper. The full factorial simulations were set up from within a Jupyter notebook. This notebook and the conducted simulations are also part of this repository. Data of the various TGA simulations are provided within a very similar repository, linked to a conference paper (ESFSS 2018, Nancy, France). Finally, the simulation input files, PROPTI input, as well as the custom script for file handling in concert with OpenFOAM, are provided. <strong>Technical Information</strong> Each ZIP archive represents a sub-directory of the original directory. For the analysis scripts, the Jupyter notebooks, to work properly out of the box it is necessary to keep this structure. Thus, simply extract all archives into the same directory. Note: Size on disc, after extraction, is about 4.1 GB. Version 2 adds about 5.1 GB. <strong>Version 2:</strong> Version 2 contains new IMP runs that address an error in determining the normalised residual mass, see Jupyter Notebook "RevisedTargetAssessment.ipynb", as well as input from the reviewers. The IMP runs are denoted by "08" after the optimisation algorithm label, e.g. "MLC_FSCABC_08_new_75kw_Ins".【数据集内容】本数据集为一篇已发表于2019年6月15日《火灾安全期刊》（Fire Safety Journal）、题为《PROPTI框架反演建模应用案例》的论文的配套仿真数据集，相关研究项目可参见我们在ResearchGate上的主页。本仓库包含本文中所展示的质量损失量热仪（mass loss calorimeter）每次IMP运行的完整输入数据，具体包括实验数据文件、仿真模型模板以及PROPTI的输入文件。本次提供的数据集文件包含两类：一是运行期间由PROPTI生成的原始文件，二是用于绘制论文图表的经清洗处理的数据集文件，以及每一代优化过程中提取得到的最优参数集。此外，本仓库还收录了用于监控IMP运行进度的各类绘图文件。同时，本仓库包含一批Jupyter Notebook（Jupyter笔记本），用于处理数据集文件并生成本文中展示的图表；全因子仿真的设置工作同样依托某份Jupyter Notebook完成，该笔记本及对应的仿真过程均包含在本仓库内。另有一个结构高度相似的仓库，收录了各类热重分析（TGA, Thermogravimetric Analysis）仿真数据，该仓库关联一篇发表于2018年法国南希ESFSS会议的论文。最后，本仓库还提供了仿真输入文件、PROPTI输入文件，以及用于配合OpenFOAM（OpenFOAM）进行文件处理的自定义脚本。【技术信息】每个ZIP压缩包对应原目录的一个子文件夹。为确保分析脚本与Jupyter Notebook可直接开箱即用，需保留原目录结构，请将所有压缩包解压至同一目录。注：解压后磁盘占用约为4.1 GB；版本2（Version 2）新增占用约5.1 GB。【版本2】版本2新增了若干IMP运行，用于修复归一化剩余质量计算的相关误差，相关细节可参见Jupyter Notebook"RevisedTargetAssessment.ipynb"，同时也纳入了审稿人提出的修改意见。本次新增的IMP运行在优化算法标识后以"08"作为后缀，例如"MLC_FSCABC_08_new_75kw_Ins"。