PSAAP @ Stanford

The overarching problem for the PSAAP Center is the simulation of air-breathing hypersonic vehicles with a special focus on the prediction of off-design, transient conditions and their associated failure modes. Air-breathing hypersonic vehicles are envisioned as a means for reliable low-cost access to space. These vehicles are highly integrated systems whose performance depends on complex physics and the interactions between all of their components. Such performance-critical systems cannot be predicted with today's state-of-the-art simulation capabilities: a radically new integrated approach is required.

The PSAAP center implements a plan for the development of a validated and verified simulation environment for unsteady physical phenomena in the hypersonic regime. To validate the overall predictive capability, a suite of system-level data from previous and planned tests have been identified. World-class experimental facilities in Stanford's High Temperature Gasdynamics Laboratory (HTGL) will be used to conduct validation experiments for the key component physics and models. Verification methods will be developed and implemented as an integral part of our effort at both the component and the system levels. We will leverage advanced computer science methods developed at Stanford to ensure scalability, program correctness, and portability to future platforms with very large numbers of cores.

Failure-mode analysis of hypersonic vehicles simultaneously encompasses some of the most challenging problems of unsteady fluid and structural mechanics, plasma dynamics, chemical kinetics, radiation, numerical techniques, and multi-physics integration. These problems are of key importance to the NNSA Laboratories. The work represents a revolutionary advance in our predictive capabilities to mitigate a critical failure mode of hypersonic vehicles.

As driven by our simulation plan, the core of the work concerns the high-speed flow around the vehicle and through its supersonic combustion propulsion system (scramjet), as well as the prediction of unsteady thermal loads on the vehicle structure and fuel. Our simulations begin with a full-system calculation of the entire system leveraging the tools developed at Stanfords current ASC Center, including the Coupler for High-performance Integrated Multi-Physics Simulation (CHIMPS). The primary aim of the subsequent work is to add fidelity to the component models and software to improve prediction capability. Successful predictions will require accurate simulation of complex phenomena, including:

• Flow stability for supersonic mixing during combustion;
• Real gas effects, chemically-reacting and dissociated flows, and heat transfer;
• Turbulence, mixing, multi-phase flows, and associated interactions with oblique and intersecting shock waves,
• Moving shocks, unsteady shock/boundary layer interactions and separated flows,
• Transient thermal loads and conjugate heat transfer,
• Flow/structure interactions
• Laminar/turbulence transition
• Radiative energy transfer and plasma dynamics
  

Moreover, the environment associated with the unsteady failure modes of Mach 7+ flight makes the prediction of the vehicle performance quite sensitive to disturbances and uncertainties and one must ensure robustness to unknowns, such as those associated with modeling errors. A fundamental characteristic of our integrated simulation environment will be the ability to control the numerical error present in the highly integrated computations. This verification capability is considered a fundamental portion of the development of the software and not an add-on to existing approaches. This is particularly true of simulations where the discretization error, for example, can accumulate across interfaces. Together with modern methodologies for automatic differentiation and large-scale linear solutions, error estimation will be made an important part of the complex simulation environment proposed here.

Moving forward beyond the five-year duration of this program, next-generation supercomputers will be based on multi-core and streaming chip architectures for which MPI protocols may not be suitable. Accordingly, the Center will support a computer science research and educational effort on next generation programming environments and compiler developments for scientific computing.