Turbulent adjoint optimisation framework

Context

For industrial thermal equipment, the performance gains left on the table are geometric — but exploring geometry by trial-and-error simulation is hopeless. Adjoint methods invert the problem: one extra solve yields the sensitivity of performance to every point on the surface at once.

My role

I was the fluid dynamics engineer developing the adjoint capability within a proprietary CFD solver and applying it to client equipment.

Approach

• Development of the 3-D adjoint solver for turbulent flow.
• Integration with the in-house proprietary CFD code.
• Surface-sensitivity-driven geometry improvement of heat exchangers and thermal equipment.
• Validation of optimised designs against baseline performance.

Outcome

An in-house optimisation capability that turned geometry improvement from guesswork into a directed, repeatable process.