jjn9128 wrote: ↑29 Apr 2020, 17:37
I think the main struggle for absolute downforce in this competition is the ride height (I'm sure a necessity of the number of entry's and turnaround to run all cases for a race), then the rotation around the front axle for rake instead of the splitter - so the whole of the ride height increases for even a small rake angle and you don't get the front wing much closer to the floor.
I also think the mesh doesn't work with any aggressive outwash. Where the flow should be wide it hits a join in mesh layers and just for sort of straightens. Again I suspect a mesh density issue. You've got to almost unlearn some things to work within the constraints. But it's not that easy.
Not complaining, just trying to work around the constraints.
I agree for ride height case. I think it is fair compromise for such competition, otherwise everyone would have to design own suspension models or have sets of pre-defined suspension members to choose from to keep all parts in position. To avoid what's happening now in the rear, where suspension doesn't match rear wheel. I think for 2019 the ride height variable was introduced for 1st time, wasn't it?
With current rake change, there are 2 (obviously) the most extreme ones: 0* and 1*, for which you need to have completely different approach, because as mentioned front wing position relatively to the ground is different, but also the diffuser volume changes significantly, plus rear wing goes much higher, than you may think, and that 1* + height difference can change rear wing behaviour completely. From theory 0* should give the less drag, and 1* should give the most downforce. However. It's not the case with the same geometry. I don't know if anyone has made the rake map of own car, but for mine it's neither 0, nor 1, nor 0,5
and that drives me bit crazy as trying to check boundary rake values, from time to time still gives me similar results compared to one which I use as the baseline.
I think having one parameter to play with is fair enough for this challenge. I think having both ride heights available to play with could be overkill, due to amount of data to process through and time needed to get it, and then develop it properly around certain amount of ride heights, to have some flexibility between the races.
So the thing with the rake needs bit more attention while developing the car, and once car is designed around one value, I think it won't be so easy to just switch it to different one to get for example directly less drag, because it also might not be the case
For the mesh thing - what you've mention: that could be the case, however (again) to make the competition available for most of the people the mesh thing is really huge trade-off. Compared to the LMP cars done in the previous year (whoever raced back then please confirm), current rules allow for really crazy detailed and refined elements which out of sudden drastically increase computation time. Worth to mention that then results can be skewed if some details are gone due to SnappyHexMesh way of meshing nad cells density, so it's important to keep parts quite simple, but fairly refined.
In my case from initial geometry car with ~5,5M cells, to more advanced one with ~6,5M cells (and it's not its final form!), the computation time increases from around 12h up to 15-16h. I guess to improve mesh density on the sides of the car some more restrictions could be made to the car parts itself to keep less high resoultion mesh elements (e.g. around wings).
