A Mode Based Reduced Order Model for Supersonic Store Separation

What is it about?

Whether they be experimental or numerical-based, the store separation modeling process is often both highly resource-intensive and time-consuming to comprehensively conduct. The expense associated with these modeling techniques has significantly limited the capability of engineers to both efficiently identify potential failure conditions of separation and avoid such failure conditions through the use of store design optimization and flight controller training. In pursuit of deriving a limited expense store trajectory prediction model, this work presents a CFD-based data-driven surrogate model that is capable of making high fidelity predictions for both surface pressure and shear stress distributions at a low computational cost. To construct this model, three supersonic Mach numbers were selected, $M$ = 1.2, 1.4, and 1.6, for fixed-wing CFD store separation simulations. Both pressure and shear stress distributions as computed through CFD simulation were analyzed with Proper Orthogonal Decomposition (POD) and reduced to a subspace of 64 modes. The Kriging interpolation method was utilized for interpolation of subspace such that future predictions of store load distributions were obtained. These predictions were then integrated and coupled with the six degrees of freedom equations of motion to predict store trajectories at Mach numbers of 1.3 and 1.5.

Featured Image

Photo by Darren Nunis on Unsplash

Photo by Darren Nunis on Unsplash

Why is it important?

Comparison between the POD reduced-order model (ROM) surrogate model and CFD simulation show that the resulting model was capable of producing highly fidelity trajectory predictions with a significant reduction in computational cost. Computational run times for a single trajectory prediction while using CFD simulation was 6 hours on 360 cores. When using the POD ROM surrogate model, the run time was reduced to 1 minute on a single core.

Perspectives