​Heat exchangers are home to phenomena ranging from fluid mechanics and thermodynamics to fouling—making the systems multi-scale by their very nature. Researchers at Liten, a CEA Tech institute, turned to parametric optimization to design the most “manufacturable” optimized shapes.

The researchers started by creating a metamodel using a method called “kriging” to rapidly change the design parameters. The resulting response surfaces were then explored so that the optimal set of parameters could be determined. For this step, the researchers used genetic algorithms (Pareto front) to change the parameters based on the results of previous parameters until they produced the solutions with the best compromises.

They implemented the technique on a heat-exchanger technology on local geometric parameters (elementary patterns) and broader parameters (distributor) to compare the performance of the initial and optimized designs. The technique could be of interest to companies across a range of industries. The case the researchers investigated focused on reducing diesel engine EGR valve fouling.