Experimental and numerical investigation of developing turbulent flow over a wavy wall in a horizontal channel

Turbulent flow over a wavy bottom wall in a horizontal channel is investigated by experimental and numerical methods. The ratio between wave length and wave amplitude is 10. This work assesses the predictive accuracy of the seven turbulence models and provides an experimental benchmark dataset suitable for numerical validation in both the developing and fully periodic regions of the wavy wall flow. The experiments are conducted using a particle image velocimetry system at a Reynolds number of 10700. The influences of the recirculation region and the shear layer region on the flow mean quantities are studied experimentally. Three Reynolds stresses (streamwise, wall-normal, and shear) are examined to assess the development of the turbulence quantities along the wavy channel. There are significant differences between the flow characteristics at the first wave compared to others. Flow periodicity is found at Wave 8. Computational fluid dynamics simulations are performed to predict the turbulent flow using four eddy viscosity turbulence models: standard k-epsilon, Realizable k-epsilon, standard k-omega, and SST; and three Second Moment Closure (SMC) turbulence models: LRR-IP, LPS, and SMC-omega. The standard k-epsilon, Realizable k-epsilon, and LPS models have the best overall agreement compared to the experiments. The eddy viscosity models predict similar results for the mean velocity and recirculation location. The LPS and SMC-omega models are also in good agreement with experimental data for the mean velocity. The wall treatment is shown to be critically important in the capability of the model to predict the flow separation. The results indicate no notable benefit of the SMC models compared to the eddy viscosity models in capability of predicting the mean flow and the separation and re-attachment locations.