A Reimagining of the DPLR Code

The NASA Ames Data Parallel Line Relaxation (DPLR) Computational Fluid Dynamics (CFD) code has been a cornerstone of the NASA aerothermodynamics community since its original development. The code offers excellent computational efficiency for computing supersonic and hypersonic flows in chemical and thermal nonequilibrium. It has been used for the design and analysis of Thermal Protection Systems (TPS) for the Space Shuttle Orbiter, the Orion capsule, and planetary entry probes. It is also used extensively to model both hypersonic flight tests and ground test facilities.

Recently, the NASA Glenn Research Center (GRC) utilized the DPLR code to model laminar-to-turbulent transition in the NASA Ames Panel Test Facility (PTF) to better understand a series of test conditions to which high-temperature seals were exposed. High-temperature seals are porous structures that admit some flow through their depth when subjected to a pressure gradient. The amount of enthalpy ingested into high-temperature seals is dependent upon the enthalpy profile through the boundary layer which is a function of whether the boundary layer is laminar or turbulent. A series of laminar-to-turbulent transition models were implemented by GRC in the DPLR code and were subsequently used to analyze the PTF flow environment to determine the effects of a potential transition on the performance of high-temperature seals.

The application of DPLR to the PTF environment necessitated several changes to the DPLR code that ultimately resulted in a complete rewrite of the DPLR code. The first change was to unify the separate two- and three-dimensional solvers distributed in the original version of the DPLR package into a single solver that handles both dimensionalities without a loss of computational efficiency. The second change was necessitated by a grid convergence study in which numerical instabilities manifested on O-H grids as the mesh spacing was refined. This was resolved through the application of one-sided second-order numerical stencils in the vicinity of domain corners. Other changes described include the handling of block splits in the DPLR line direction, a new turbulence subsystem implementation, and a new implementation of the block tridiagonal Lower-Upper (LU) triangular factoring scheme.

Finally, a series of validation cases are presented to demonstrate that the updated DPLR code produces identical solutions to the original NASA Ames DPLR code. Three validation cases are presented and include a two-dimensional planar case, a two-dimensional axisymmetric case, and a three-dimensional case.