Mustafa zuhair

My research mainly focuses on approaching fluid flow problems from a data point of view. I am looking for the hidden features of turbulent flows behind big data that can be generated from both experiments and direct numerical simulations.

My interest in fluid and heat transfer-related phenomena started during my undergraduate study, I was passionate about the development of the boundary layer, the instability of the flow during the transition process to turbulence, and the effect of heat transfer on it. In my master's study, I worked on developing new empirical correlations of transitional natural convection boundary layer flows in inclined cylinders and annular passages. I used both the experimental and CFD data to obtain and validate the results. During my Ph.D. study, I decided to dive into turbulence-related problems using the power of machine learning and (CFD-PIV data). I have worked on developing reduced-order models for massive separation flows, turbulent inflow generation of fully developed and developing turbulent boundary layers, super-resolution reconstruction of turbulent flows, and flow
control using deep reinforcement learning. In my research, I showed that using physics-guided-deep learning models can remarkably improve the generalization of the models that can be used to predict flow fields at Reynolds numbers that are outside the training range. Furthermore, I showed that using transfer learning can help reduce the training data and the computational time without affecting the performance of the models.

• Reduced order modeling of turbulent flows
• Prediction of turbulent flows
• Turbulent inflow generation for simulations of wall-bounded turbulent flows
• Reconstruction of turbulent flows
• Active flow control
• Convection heat transfer
• Machine learning