Application of Hybrid-Temporal LES to Shock-Induced Boundary Layer Separation in Hyperloop Flows

The development of evacuated tube trains (ETT), generally referred to as Hyperloop, has gained significant attention in recent years, offering the potential for high-speed and energy-efficient transportation. Achieving sustainability in high-speed modes of transport requires aerodynamic drag optimisation in order to minimise energy consumption. One of the major challenges in ETTs is the choking of the flow around the pod, which occurs depending on operating conditions defined by the Mach number M∞ and blockage ratio BR, defined as the ratio between cross-sectional pod and tube area. As soon as the choking limit is surpassed, the resulting shock wave at the rear of the pod interacts with the boundary layer, producing a complex flow field behind the pod. The exact type of interaction and corresponding drag working mechanism is highly sensitive to the operating conditions (M∞, BR) with multiple distinct flow regimes recognised in the current literature. The complex nature of the shock-wave boundary layer interactions (SWBLI), the high sensitivity to operating conditions around the choking limit and a moving-wall boundary, leaves common Reynolds-averaged Navier-Stokes (RANS) CFD methods with significant challenges in accurately predicting flow characteristics. As much of the early research and development relies on CFD, both computationally affordable and sufficiently reliable results are essential. In this paper, we assess the feasibility of a hybrid RANS-LES approach in form of the Hybrid-Temporal large eddy simulation (HTLES) method, as implemented in STAR-CCM+ as Scale Resolving Hybrid (SRH), for drag and separation location prediction in the case of shock-induced boundary layer separation (SIBLS). We find good numerical stability and reasonable computational requirements for improved spatial and temporal resolution and subsequently improved spectral representation in configurations where SIBLS dominates the rear flow and wake of the pod. These hybrid RANS-LES methods are broadly applicable to a wide range of aeronautical applications, including transonic wing aerodynamics with physics similar to the presented case. Comparisons are conducted with respect to Menter's RANS k-ω SST turbulence model, with results based on previous research by the authors.