Feedback on PyFluent with other Python ecosystem

[PySRG]

Hello,

I have been exploring PyFluent with other Python ecosystem such as Numpy, Pandas, scikit-learn, Seaborn, Plotly,... I am able to create an example which performs DOE, fits ML models and predicts the output for unseen data. Overall experience is very positive!

Following are key highlights:

• All packages are installed under a separate conda environment.
• Jupyter Notebook is the only IDE used so far. It worked as expected. Auto-Complete worked nicely for PyFluent.
• PyFluent usage was limited to a few APIs. Case setup (Mixing Elbow) was done a priori in a standard Fluent.
• DOE: Defined manually using Numpy arrays for Cold and Hot Velocity Inlet. Average temperature at outlet was extracted as an output using TUI API, stored in a 2D Numpy array.
• Response Surface: Plotted using Plotly. Faced some error initially but was resolved after installing jupyter, ipykernel and nbformat in this separate conda env.
• Data Pre-Processing: Using Pandas and scikit-learn.
• ML Model Fitting and Prediction: Linear Regression with Polynomial features and Random Forest Regressor of scikit-learn. XGBoost Regressor installed as a separate library.
• Parity Plot: Seaborn and Matplotlib.

Regards,
Shitanshu.

[dnwillia-work]

@PySRG Awesome! Please let us know if anything you require is missing or not working.

[donff2j]

This is fantastic. I would like to see a demo and overview.

[PySRG]

@donff2j @dnwillia-work

Please see a short demo here:

FluentDOE_ML_Demo.mp4

Regards,
Shitanshu.