Data for: Accelerating and Stabilizing the Vapor-Liquid Equilibrium (VLE) Calculation in Compositional Simulation of Unconventional Reservoirs Using Deep Learning Based Flash Calculation

The folder contains the detailed models and the trained weights of a deep-learning based flash calculation module. The H5 files can be directly loaded into Keras. The weights file can be extracted for practical usage. classfication_model.h5 network structure of phase classification classfication_weights.h5 weights of phase classification concentration_model.h5 network structure of concentration determination concentration_weights.h5 weights of concentration determination Reference: Shihao Wang, Nicolas Sobecki, Didier Ding, Lingchen Zhu, Yu-Shu Wu Accelerating and Stabilizing the Vapor-Liquid Equilibrium (VLE) Calculation in Compositional Simulation of Unconventional Reservoirs Using Deep Learning Based Flash Calculation

本文件夹包含基于深度学习的闪蒸计算模块的详细模型架构与训练完成的权重参数。H5（Hierarchical Data Format 5）文件可直接加载至Keras框架中，权重文件可提取后用于实际工程应用。classification_model.h5为相分类任务的网络架构，classification_weights.h5为相分类任务的训练权重；concentration_model.h5为浓度测定任务的网络架构，concentration_weights.h5为浓度测定任务的训练权重。参考文献：Wang Shihao, Sobecki Nicolas, Ding Didier, Zhu Lingchen, Wu Yu-Shu. 基于深度学习闪蒸计算加速并稳定非常规油藏组分模拟中的气液平衡（Vapor-Liquid Equilibrium, VLE）计算。