Research Fellows

Pavlos Alexias

Early Stage Researcher 4 at Engys IT

• Research
• Outreach
• Biography

• Overview
• Publications
• Mesh Optimization
• Sensitivity Smoothing
• Automated Remeshing
• Drag Minimization

In the context of gradient-based optimization, the adjoint method can be considered the most computational effective as the gradient calculation of the objective function is independent of the number of the design variables. That allows the creation of a very rich design space which leads to better optimum solutions.

My research focuses on the integration of the adjoint sensitivities (the gradients of the objective function w.r.t. every surface node of the computational domain) inside a CAD-free morphing framework. This can be done in two different ways : 

1. Create control boxes that control the surface movement through parametrized control points and then translate the adjoint sensitivities into gradients w.r.t. to the control point movement.
2. Smooth the sensitivities in the boundary surface and directly deform the surface nodes based on the magnitude of the sensitivity derivative. This requires the existence of a mesh displacement algorithm to guarantee us that the computational domain will be valid and suitable to solve the Navier-Stokes equations.