Hybrid Analytics and Modellling Applications in Fluid Flow: Case Study - Wind Energy, Drilling, Greenhouse and Urban City.

Hybrid Analysis and modelling (HAM) techniques aims to combine the interpretability, robust foundation and understanding of a physics-based approach with the accuracy, efficiency, and automatic pattern-identification capabilities of advanced data-driven machine learning and artificial intelligence algorithms for digital twin applications. These hybrid approaches can also be categorised into intrusive hybrid modelling and non-intrusive hybrid modelling. The intrusive hybrid approach involves techniques like the Corrective Source Term Approach (CoSTA) that involves augmenting the governing equation of a physics-based model with a corrective source term generated using a deep neural network or semi-intrusive reduced order models, while the non-intrusive hybrid methods involve techniques like non-intrusive Reduced Order Models (ROMs) developed based on data compression and deep learning techniques. Here, we present some of our HAM work applied to a drilling process, a wind turbine, a greenhouse and a urban landscape. Here, we use HAM to reconstruct temperature and velocity fields in the greenhouse and flow over a blade and flow over buildings using non-intrusive ROMS, and in drilling process, we use HAM to improve the accuracy of a 1D cutting hole transport model to predict accurate pressures during the flow of cuttings using the COSTA and continual learning approach.