Afterbody longitudinal cavities for drag reduction and application of artificial neural network for optimization of groove geometry

The boattail model was found as an effective passive device for reducing the aerodynamic drag of axisymmetric models. For further decreasing aerodynamic drag, groove cavities made on the boattail region show a high potential technique. In this study, the geometry of longitudinal groove cavities was considered to understand its effect on the drag of the model. Then, an artificial neural network (ANN) was developed to predict the total drag to find the configuration with minimum drag. For that purpose, Reynolds-averaged Navier–Stokes with a k-ω shear stress transport model is used for the simulation. To generate data for the ANN, an automatic program was developed to create the model geometry, build the mesh, and conduct the calculation. The computational results were validated by experiments with the same flow conditions. The results show that for the baseline cases, the drag decreases with boattail angles to 14°, then, it increases again. However, by changing the geometry of longitudinal grooves, the aerodynamic drag decreases for boattail angles up to 20° with a maximum drag reduction of 34% in comparison to the blunt-based model. The mechanism of the drag reduction is due to the modification of the surface flow from a full separation for the baseline case to an attached flow with small local separation bubbles when the groove cavities were made. The pressure is found to significantly increase in the second-half region of boattail models with fully separated when the cavities were added. However, for boattail angles below 14°, the pressure distribution at the rear part of the boattail remains similar for models with grooves. The ANN results present that the drag of the model can be predicted well with an averaged uncertainty less than 2%. A groove configuration with the minimum drag and its aerodynamic characteristics are presented.