Evaluating and Optimising the OpenFOAM Overset Mesh Solver for Offshore Renewable Applications
ARCHER2-eCSE07-07
PI: Prof Ling Qian, Manchester Metropolitan University
Co-Is: Dr Xiaohu Guo, STFC Daresbury Laboratory, Dr Zhihua Ma (Manchester Metropolitan University), Dr Wei Bai (Manchester Metropolitan University)
Technical staff: Dr Ranjodh Rai (Manchester Metropolitan University)
Subject Area:
The marine environment in which offshore renewable structures operate is naturally subject to harsh and potentially inconsistent wind and wave conditions over the long term, posing a significant challenge to the survivability of any renewable structure and its performance. Given this, there must be a thorough and robust design process, followed by extensive numerical and experimental testing. One of the pivotal tools required at a fundamental level is an accurate and efficient Computational Fluid Dynamics (CFD) model that deals with complex hydrodynamics and aerodynamics. The quality of any such model will also help ensure that the structure is as efficient as possible, both in a practical and financial sense.
The overset mesh method for CFD models has a major advantage over other techniques in that it allows for large amplitude motion of structures with complex geometries without changing its quality. This makes it particularly useful for aerodynamic analysis of fully rotating structures such as wind turbines—an application that many other methods cannot simulate. However, the primary disadvantage of the overset mesh approach is that it can be computationally expensive and inefficient.
Figure 1. Background mesh with the hole pushed back 5 layers of mesh cells from the turbine rotor boundaries in the yz (left) and xy (right) planes
Our eCSE work presents a new optimised version of the open-source ESI-OpenCFD OpenFOAM overset mesh solver that improves its computational efficiency. We tested the new version through two test cases. First, we considered regular wave interaction with a 2-D T-shaped floating body. It was found that the new method reduced computational time by 14% and 7% on 128 and 256 cores respectively — a modest amount but to be expected for a simple case with only small amplitude motion. Second, we considered the aerodynamics of an offshore wind-turbine rotor in constant wind. The new method reduced computational time by 64.6% on 1024 cores—approximately 3x faster whilst achieving practically identical results. The new method was also shown to have sufficiently good parallel efficiency. Indeed, the execution time on 4 nodes is less than half the time taken compared to the old method on 8 nodes, so the new method essentially cuts the required computational resources in half in addition to the reduction in computational time.
Publication
Rai, RS, Qian, L, Ma, Z, & Bai, W. “Evaluating and Optimising the Overset Solver in OpenFOAM for Complex Wave-Structure Interaction Problems.” Proceedings of the ASME 2024 43rd International Conference on Ocean, Offshore and Arctic Engineering. Volume 6: Polar and Arctic Sciences and Technology; CFD, FSI, and AI. Singapore, Singapore. June 9–14, 2024. V006T08A010. ASME.
Information about the code
OpenFOAM v2212 must be compiled in the user’s work directory rather than using the pre-installed Archer2 version. The cellCellStencil and inverseDistance directories in src/overset/cellCellStencil must then be replaced by the new versions. The stencils must then be compiled by executing ‘wclean’ and ‘wmake’ in src/overset. Errors may arise during compilation due to the other interpolation methods; comment out the offending lines accordingly.
Technical Report
The Technical Report is currently under embargo and will be published when available.