Improvement of Code Performance
Unsteady simulations with a few thousand or more panels can be very time-consuming. Thus, it has become necessary to improve the performance of the code. To achieve this, the following things have been implemented:
- The system of linear equations (ref. here) is solved by an efficient solver provided by the LAPACK software library.
- The OpenMP library is used and the computationally most intensive loops in the code have been parallelised.
To demonstrate the performance of the code various test case computations have been carried out. Below the results for the KCS Propeller (3,000 panels, non-fixed wake panels, 10 steady + 40 unsteady iterations) are presented.
Figure 1: Simulation Time for the Test Case with Varying Number of Processors.
Figure 2: Speed Up for the Test Case with Varying Number of Processors.