kamilazdybal/ROM-zero-dimensional-reactor

Reduced-order model for a zero-dimensional reactor

This repository is licensed under:

Reduced-order model for a zero-dimensional reactor

Code and materials that accompany Chapter 8 from my dissertation: Reduced-order model for a zero-dimensional reactor.

Data

The data/ directory stores the training dataset corresponding to combustion of syngas in air. The dataset is generated using the zero-dimensional reactor model.

Code

The code/ directory stores Python code that can be used to reproduce results from my dissertation.

Results in Chapter 8 have been generated with the following Python package versions:

numpy version:		1.21.6
scipy version:		1.7.3
george version:		0.4.0
tensorflow version:	2.11.0
keras version:		2.11.0
PCAfold version:	1.6.0

Transport of PCA-derived manifold parameters: PC-transport.py

Running this script on your PC takes about 150 minutes. If you want much quicker results (of the order of 15 minutes), consider not running GPR (set run_GPR = False).

Transport of regression-aware AE-derived manifold parameters: AE-transport.py

Running this script on your PC takes about 10 hours. If you want much quicker results (of the order of 15 minutes), consider not running GPR (set run_GPR = False).

Visualizing the results

This Jupyter notebook can be used to upload and visualize the results of the PC-transport.py script.

This Jupyter notebook can be used to upload and visualize the results of the AE-transport.py script.

Results

The results/ directory stores figures and results in .csv files.

PC-transport

Test trajectory transported using PC-transport with the RBF closure model

Test trajectory transported using PC-transport with the GPR closure model

Test trajectory transported using PC-transport with the ANN closure model

Test trajectory transported using PC-transport with the kernel regression closure model

A posteriori predictions of the first and second projected source terms

Predicted thermo-chemistry

AE-transport

Test trajectory transported using AE-transport with the RBF closure model

Test trajectory transported using AE-transport with the GPR closure model

Test trajectory transported using AE-transport with the ANN closure model

Test trajectory transported using AE-transport with the kernel regression closure model

A posteriori predictions of the first and second projected source terms

Predicted thermo-chemistry