CFD Workflow Best Practices
Whether you are using one of the Top500 supercomputers or a modest cluster for CFD, post-processing can be a big bottleneck. Since this is where decisions are made and your goal is to make the best decisions possible, why not optimize your post-processing workflow? Intelligent Light can help in three important ways:
1. Automating post-processing to perform repetitive tasks has great leverage in streamlining a workflow. Our SnapShot solution delivers wizard-driven comparison and reporting for CFD solutions at all scales. So, if you are looking at variations on a geometry, use a standardized naming convention across all solutions (like ‘wheel’ or ‘impeller-suction-side’) – the same worklow can be used repeatedly across all solutions via a loop, even if the solutions come from different solver codes. Standardized reports have never been easier!
2. Data management: most organizations simply cannot move or even store all of the results files from solvers. This is emerging as a significant and costly CFD optimization challenge. Intelligent Light introduced ‘Extract Post-processing’ into the commercial FieldView post-processor in 2005, providing data reduction and organization through the integrated eXtract database or XDB mechanism. Post processing objects such as streamlines, cut planes, iso-surfaces, etc. can be saved and restored via XDBs, with one hundred-fold reduction in file size over the CFD result files. XDBs can form the basis for archives and sharing in your CFD workflow with speed and ease. You can produce the extracts on your HPC system or the cloud and then post-process locally with files that are 100 to 1000 times smaller than the volume files.
3. In situ and in transit processing is evolving as the must have technique for the applications needing the largest CFD cases or many smaller cases. Flows dominated by unsteady flow physics, such as aeroacoustics, combustion, rotorcraft, wind turbines and motion and those that require very large meshes such as Formula 1, high-lift aircraft configurations and digital twin analysis that requires modeling ‘as built’ geometry. Data extraction functions are integrated with the CFD code to pull out surfaces and sub-volumes of interest directly from the solver memory, rather than having to write huge volume files and then read them in. We have helped customers implement these workflows for the DoD CREATE program, solver developers, turbomachinery manufacturers and aerospace agencies.
Our domain experts can work alongside your methods team to implement any of these state-of-the-art workflows for you!