Our goal is to model flows, investigate their physics, and optimize their behaviour.
We start from a physics-based model of a flow inside or around a device. If the model contains unknown parameters then we infer these from experiments using Bayesian Inference. This turns a qualitative model into a quantitative model.
We investigate the behaviour of the model and seek to understand the phenomena it predicts. If the model does not capture all the relevant physics, we revise the model.
We set targets and constraints and thereby derive the adjoint of the model. This shows, in a single calculation, how the targets are affected by all the model parameters.
We then use gradient-based algorithms to optimize the flow through or around the device.
Click on the links above to find out more.