Applying Impulse-Based Force Reconstruction to a Pitching Wing in a Transverse Gust

Abstract: Developers of CFD codes are often faced with tradeoffs in the frequency of data saves from high-fidelity unsteady analysis due to file and count.  This can limit the quality of subsequent data analysis techniques such as modal analysis.  Implementing specialized output functions into codes adds to the complexity of development and maintainability.  Considerable research has yielded techniques to access information directly from the memory of the running solver code, known variously as ‘in situ’ or ‘co-processing’.  We present a new co-processing approach for enhancing CFD workflows that provides services for visualization, data science, job monitoring, provenance capture and logging along with computational steering of the solver code. The software, called Kombyne® accomplishes these functions with a very low code footprint, few to no external dependencies and direct support for ‘in transit’ operation where the workload and memory requirements are delegated to a separate process, working in tandem with the solver code.  The instrumentation API is open-source, and the Kombyne® source code is available to solver code developers at no-cost. 

Abstract: Fluid-induced loads can be estimated from flow-field data using momentum- or impulse-based force reconstruction methods. Momentum approaches are less prone to random error, but impulse approaches allow for the decomposition of force components from each source of vorticity in the flow. 

Here, we applied impulse-based force reconstruction to a pitching wing in a transverse gust using experimental PIV and numerical simulations. Aside from experimental errors and finite wing effects, there were two main challenges to overcome: (1) accounting for vorticity that leaves the limited FOV and (2) masking of external gust vorticity. Out-of-frame vorticity was modeled by a point vortex for each time-step to ensure all vorticity generated at the wing surface satisfies Kelvin’s Circulation Theorem. Masking of gust vorticity was done carefully by hand but still required correction for misattributed or entrained vorticity. Wing-generated vorticity within the gust shear layer regions was modified to match the expected impulse and quantify misattributed vorticity. Each source of error was systematically isolated with various 2D URANS test-cases.

Results show accurate force reconstruction for no-gust cases, without external vorticity to mask, even with a limited FOV, which validates the out-of-frame modeling approach. From our entrainment modification we observed an 8-12% error in misattributed vorticity. The biggest challenge still standing is accurate masking of external vorticity.