RavenCFD (computational fluid dynamics) is a full three-dimensional Navier-Stokes flow solver for compressible viscous flows on unstructured grids. RavenCFD is tailored for robust steady-state and time-accurate solutions for a variety of multi-dimensional flows.

 

Corvid has developed RavenCFD and exercised it extensively over the last ten years, using it to generate tens of thousands of large-scale, high-fidelity flow solutions for a variety of automotive and defense applications.

 

It has been validated through comparison to experimental data for Mach numbers from less than 0.1 to 15. RavenCFD has demonstrated linear to super-linear scalability on thousands of processors, providing an efficient framework for performing large scale calculations.

 

We employ it regularly as a predictive design tool to improve the aerodynamic characteristics of complex configurations. Since RavenCFD development is performed in-house at Corvid, new capabilities are implemented on an as-needed basis.

Features

 

• Numerical representation of the full three-dimensional conservation equations for fluid flow, including the Euler equations for inviscid flows and the Navier-Stokes equations for laminar or turbulent flows.

​​

• An arbitrary polyhedral formulation, meaning that the solver can accommodate unstructured grids with arbitrary cell shapes. This feature makes RavenCFD well-suited for aerodynamic predictions for complex geometries.

​​

• A variety of upwind flux-splitting schemes, limiters, and turbulence models to provide accurate, stable, and realizable solutions for a broad range of flow conditions.

​​

• Fully implicit time integration with Newton subiterations to recover time accuracy. This combination of features improves the robustness of the solver—providing good solutions even on grids with poor-quality cells—and permit simulation of unsteady flows without excessively small timesteps.

​

• Multi-species capability, including high-temperature thermodynamic properties for a variety of gases.

​

• Parallel scalability to permit solutions of large problems on large numbers of processors. While typical problem sizes are in the 50- to 300-million-cell range, we have performed calculations on grids as large as 500 million cells on thousands of processors. These large runs show the same performance as is observed on smaller calculations, and we have demonstrated linear to super-linear scalability on all of our problem sizes.

​

• Grid deformation, optimization, and adaptation (both h- and r-refinement), embedded into the flow solver itself.

​

Contact us for more details about our services or to request a quote.