毕业设计&课设-在Python中运行MATLAB油藏模拟工具箱（MRST）的包装器.zip

# pyMRST Python wrappers to run simple reservoir simulations from MATLAB Reservoir Simulation Toolbox (MRST). Still in development. <div> <img src="https://user-images.githubusercontent.com/51282928/109408566-7d209800-79bd-11eb-89b3-294343680217.png" width="500"/> </div> MATLAB Reservoir Simulation Toolbox (MRST) is a free open-source software for reservoir modelling and simulation, developed primarily by the Computational Geosciences group in the Department of Mathematics and Cybernetics at SINTEF Digital. While pyMRST is a wrapper that allows users to run simulations from MRST in a Python environment. Currently, it has single-phase reservoir simulations for water, oil, gas, polymer, and thermal effect; and two-phase oil-water. |Simulations|Python notebook| |:--:|:--:| |1-phase water|[Notebook](https://github.com/yohanesnuwara/pyMRST/blob/main/notebooks/pyMRST_water_incomp.ipynb)| |1-phase oil *)|[Notebook](https://github.com/yohanesnuwara/pyMRST/blob/main/notebooks/pyMRST_oil_slightly_comp.ipynb)| |1-phase gas *)|[Notebook](https://github.com/yohanesnuwara/pyMRST/blob/main/notebooks/pyMRST_gas_comp.ipynb)| |1-phase polymer *)|Notebook| |1-phase compressible with thermal effect|Notebook| |2-phase oil-water|[Notebook](https://github.com/yohanesnuwara/pyMRST/blob/main/notebooks/pyMRST_oilwaterflooding.ipynb)| *) No flow boundary condition and constant BHP well assumption <!-- ## Single-phase Fluid ### Water Example inputs: * mu = 1 cp * rho = 1000 kg/m3 Formula: * mu, rho = constant ``` FLUID1 water 1,1000 ``` ### Oil Example inputs: * mu = 1 cp * rho_r = 850 kg/m3 (Reference rho @ reference pressure) * pr = 200 bar (Reference pressure) * c = 1e-3 1/bar (Fluid compressibility) Formula: * mu = constant * rho(p) = rho_r * exp(c * (p - p_r)) ``` FLUID1 oil 1,850,200,0.001 ``` Example inputs for gas: * mu0 = 5 cp (Viscosity at zero) * rho_r = 850 kg/m3 * pr = 200 bar * c = 1e-3 1/bar * c_mu = 2e-3 1/bar (Viscosity coefficient) Formula: * mu(p) = mu0* (1 + c_mu * (p - p_r)) * rho(p) = rho_r * exp(c * (p - p_r)) ``` FLUID1 gas 5,850,200,0.001,0.002 ``` <!-- Experimentations of MATLAB Reservoir Simulation Toolbox in Google Colab to port it with Python and utilize free GPUs for faster computation <p align="center"> <img src="https://user-images.githubusercontent.com/51282928/100498951-68ebb580-3198-11eb-95c7-87ed7c1e6e9c.png" width="700" /> </p> <!-- Experimentations of MATLAB Reservoir Simulation Toolbox in Google Colab to port it with Python and utilize free GPUs for faster computation ### Successful: Phase 1. Full codes and documentations are preserved in [Zenodo]() * Use Google Colab to run a MATLAB (Octave) script of MRST simulation of five-spot waterflooding in SPE10 model. * Optimizing the well placement by coupling the simulation with a Python optimizer, such as Bayesian optimization. ### On experimentation: * Modifying the MRST script to use `gpuArray` so that the simulation can use the free GPU in Colab and speed up simulation. * Using `Tensorflow Probability` to improve optimizations, powered by free GPU in Colab. * Experimenting with various optimizers (`Scipy`, `Optuna`, `Platypus`, etc.) and multi-objective optimizers (`Pymoo`). ### Future: * Well placement optimization for different scenarios (CO2, surfactant, and polymer injection)