ken writes "I appreciate any sort of advice or guildlines here. I am using CFdesign to simulate the turning of a two bladed propeller. All experimental data is in NASA CR 196702. So basically this is a verification exercise.
I know in finite element space time decoupled formulation, the size of the time step affects both stability and accuracy. Here is the situation I have: In my first test case, I used this time step and get 90% matching in thrust and power. I consider that good enough. On the second case, initially using the same time step as test case 1, my converged solution only matches about 75% with experimental data. The difference between test case 1 & 2 is the case 2 has a higher propeller RPM, while the free stream Mach number and blade pitch setting is kept constant (To be more specific, M tip 1 = 0.77, M tip 2 = 0.89). The mesh of both cases are identical too. Therefore I decided to further reduce the time step and now i got 95% percent matching on thrust and power.
Here comes my question. How would I justisfy my approach of using different time step for the same type of analyses when I am dualing with real problems without known experimental data? Critics will shoot me on this "taking incosistent approach issue". So what should I reply them?"
|