Optimization of turbulent jet mixing

Abstract

We introduce an approach for controlling jet mixing that combines direct numerical simulation of an incompressible jet flow with stochastic optimization procedures. The jet is excited with helical and combined helical and axial actuations at the orifice. An objective function that measures the spreading of the jet evaluates the performance of the actuation parameters. The optimization procedure searches for the best actuation by automatically varying the parameters and calculating their objective function value. Solutions that lead to a pronounced spreading of the jet are found within reasonable time, although the evaluation of the objective function, the DNS of the jet, is expensive. For a jet flow at low Reynolds number the performance of different search algorithms (simulated annealing and evolution strategies) is evaluated. We compare various objective functions based on radial velocity and the concentration of a passive scalar, including functions that penalize actuation with high amplitudes. We find that a combined axial and helical actuation is much more efficient with respect to jet mixing than a helical actuation alone.