Increasingly, the sophistication of rotorcraft flow simulation has become more complex, involving aerodynamically interacting environments such as a helicopter landing on a ship deck. These problems involve solving the Navier-Stokes equations on various grid types (structured, unstructured, or hybrid). As the complexity increases, the range of length and time scales may vary significantly within the problem domain, making simulations more demanding in terms of computational resources. To make RotCFD an affordable tool at the early stages of the design cycle, RotCFD's paradigm is to provide engineering solutions to complex problems in one computational environment. The key to obtaining an engineering solution is to employ one or more of the following techniques: (a) higher order schemes, (b) complex algorithms that exploit the physics intelligently, (c) error reduction techniques such as multigrid, (d) adaptation automation, and (e) parallel processing. This paper focuses on the development of parallel processing in RotCFD using Graphics Processing Units (GPUs) and the associated technical details, which show considerable promise for computational efficiency and design.
展开▼