Optimal placement of plasma actuators on trucks : A drag reduction study using adjoint methods

The fuel consumption of vehicles plays an important environmental role and ways of improving the fuel economy is beneficial for a sustainable future. Active flow control is one method that can potentially reduce the aerodynamic drag significantly and therefore the fuel consumption. A plasma actuator introduces momentum in its vicinity and can be used for active control of the flow. This thesis aimed to find the optimal placement of plasma actuators on the A-pillars of a truck, using Reynolds-Averaged Navier-Stokes (RANS) simulations performed with STAR-CCM+. The adjoint solver was used to obtain a sensitivity map of the drag with respect to the variation of momentum. This was then used when the accuracy of this adjoint solution was evaluated by performing parametric studies where the actuation was placed at different locations around the point of flow separation. A simple case consisting in a half-submerged cylinder was first studied as both wind tunnel experiments and Large Eddy Simulations (LES) have been previously performed on this flow case at KTH. When placing actuation on top of the cylinder, the drag reduction obtained with RANS was 4.3% which was comparable to the previous LES work, where the reduction was 4.65%. With the flow separating at 98 degrees in a local cylindrical coordinate system, which starts at the leading edge of the cylinder, the adjoint solution showed that the optimal placement was located at 105 degrees. The actuation was placed at several locations between 90 and 118 degrees and the minimum was found to be located at 106 degrees. The Ground Transportation System (GTS) model representing a generic tractor-trailer combi-nation, both in two and three dimensions, was then used with the same solution procedure. In the two-dimensional case, the flow was found to separate at 63 degrees. The optimal placement predicted by the adjoint solver was 69 degrees, while the parametric study showed the optimal location to be at 75 degrees. The added complexity of an extra dimension resulted in the wind speed having to be lowered in order to produce similar results for the actuation. However, agreement on the optimal location was observed between the two- and three-dimensional cases. In general, the adjoint solver showed varying levels of accuracy between simulations, but still gave a qualitative indication of the optimal placement. A correlation between the point of separation of the flow and the adjoint solution was observed and along with the results from the parametric studies, the optimal placement was concluded to be slightly downstream of the separation point. In order to obtain the exact optimal position, a parametric study is needed for each individual case