Reliable CFD analysis data is a key for each aerospace/turbomachinery project. With our professional team, OE is developing a unique CFD software for all range of subsonic and supersonic flows.
The compressible flow codes have been optimized to achieve better performance and flexibility by changing the data structure or even programming approach. In this process, all the methods that may be used in compressible flows have been investigated and the codes are really competitive with commercial codes in terms of memory and convergence rate.
The codes are using FV for discretization and simulation of different flow physics such as inviscid, laminar, turbulent, unsteady, real gas flows with moving boundary problems. Also, they benefit from several numerical techniques such as convergence accelerations by implicit and multigrid, high order accuracy, cell-center/cell-vertex and overlapped grids and more.
These codes have been deployed to run a project related to the laser cladding by powder in which the nozzle and the powder ejector were studied. Since the flow is incompressible, so many incompressible solvers were implemented such as MAC, Projection, Staggered and Co-located, SIMPLE and Artificial Compressibility with the first and second order accuracy by QUICK and TVD.
Regarding Heat Transfer, a software for simulation of unsteady heat transfer for deformed medias was developed in which FEM was deployed for Heat transfer and a torsional spring analogy for a 3D unstructured deformed mesh. Regarding an inadequate quality of the grid by the deforming technique when the solid deforms a lot, a new semi level set scheme was proposed and implemented to tackle this problem and it can manage ablation till disappearing whole solid. The software package is a GUI for these projects.
In RANS approach, Spalart-Allmaras and (k-ε) have been developed with standard wall functions for our CFD software. DNS was used to simulate the evolution of the vortex filament. The objective was to find a support for vortex filament equation (binormal) that has been solved analytically and numerically.
Some new schemes in moving boundary problems have been introduced and a new scheme for 2D general large amplitude boundaries motion has been implemented which provides a seamless high-quality domain while it resembles the overlapped grid approach. This method was extended and improved for 3D problems in which a simple robust scheme was proposed for the node deletion as a part of the algorithm. Since the parallel processing was utilized for the domain with moving mesh, some dynamic load balancing schemes and partitioning for it was deployed where another technique to improve the speed-up was proposed for these cases.
The overset grid scheme was implemented using kd-tree classification and a least squares interpolation.
A new scheme has been introduced for efficient scheduling of data transfers between processors. In the moving boundary part, since our scheme for moving mesh changes the topology and the grid size, the dynamic load balancing is a remedy where several partitioning schemes were studied to find most affordable one which makes minimal data migration with an acceptable edge-cut size.
SIMD approach was reimplemented in our CFD code for distributed memory. Also, the shared memory parallelism was deployed by OpenMP.
MRF (Moving Reference Frame) technique has been implemented to simulate aerodynamics of blades in helicopters. These tools were implemented for numerical analysis of an offshore floating wind turbine in which MRF technique and low Mach flows were added to the codes.
The computational grid is where the computational method are apply on, so having a robust, high quality grid generator is an essential. We have a 2D/3D structured/unstructured grid gnerator. For unstructured is a Delaunay quality grid and in strucutred is a PDE based tool with an efficient boundary orthogonality method.
The code benefits from a fast algorithm for a conservative interpolation of two planner meshes found in the interface of two attached blocks which is useful in the sliding mesh and unmatched multizone simulations.
Overset grids technique is a robust and an easy way for mesh preparation of complex geometries and quiet popular in moving boundary problems. The code has a 2D/3D overset grids preparation and a solver that can utilize this type of the grid and it has been used for the simulation of a wind turbine by full unsteady rotating blades.
Capturing whole flow details needs to have a very fine grid and it has a significant computational cost. A smart way to decrease this issue is to do simulation on a medium grid and then using some criterias find the complex feature of the flow then adapt those parts where we need more computational points called Adaptivity.
An intelligent smoothing and mesh refinement method have been developed which considers the desired criteria and the resulting solution in the movements of nodes. In this approach, the topology, memory and performance complexity will not change.
The codes (CFD and Heat Transfer) are second order in space and time. Also we have posibility to do very high order scheme (3rd, 4th and 5th order) using a new generation of DG (Discontinious Galerkin) for cartesian grids. Extension of this scheme to general unstrucutured grids is in progress now.