manchild wrote: ↑06 Mar 2018, 01:09
If you're going to do a CFD, make it as simpe as it gets.
Here it is.
I had to simplify it quite a bit, since no quarter of these mirrors is the same since air flow is far from perpendicular in this area of the car. To save computing resources, I decided to abandon my original idea to make the CFD model as close as possible in terms of geometry, which allowed me to test just a single quarter of this mirror.
The big unknown here is, of course, the internal geometry of mirrors and fairing, this is why it wouldn't be useful to make the model externally similar to actual thing.
There were 9 simulations in total, a combination of different slot gaps - 1, 2 and 3mm; and a combination of different air velocities - 30, 60 and 90 m/s, translating to around 100, 200 and 300 km/h. It is very hard to distinguish if there even is a slot around actual mirror, let alone the height of it. At first, I was looking at 1, 3 and 5mm slots, but realized 5mm is way too much.
Meshing, turbulent model and overall CFD set up is the same I used to correlate wind tunnel results with CFD for a much more complex model, so these results weren't dropped out of thin air. Let me show you the mesh, on the right you can see refined mesh made with adaptive meshing, automated by software. Blue cells pure fluid, green are partial cells.
So here are the results, side view in symmetry plane, pressure distribution with streamlines.
Our discussion was about whether or not there is choking inside these channels at different speeds and CFD results of this model show none of that. Pressure distribution is practically the same for the same slot height, while slightly different shapes of turbulent wake shouldn't be taken too much into consideration - it's turbulence after all and these were steady state simulations, not transient.
Numbers, in terms of drag coefficient multiplied by referent surface, tell the same tale. Differences are between 1-3% for the same slot height, which is within a tight margin of numerical error. Also, slots 2 and 3 offer almost 20% drag reduction compared to 1mm slot. Perhaps the optimum slot height is 2.5mm or something like that. Whatever it is, to make sure it is the same height all around the mirror with 2 or 3mm is not an easy thing with carbon fiber.
Other views of pressure distribution offer the same, let me show you a top view of 2mm slot at 60 m/s, hardly different from other 8 results.
The separation on outer surface here should be ignored, since this is just a single model without any optimisation. From this point of view, to achieve even better results the inside cone should be a bit more tapered in top view, as should the side fairing.
Just to add another thing into the mix, I first did a simulation of fairing with extended trailing edge:
Wow, a huge difference, isn't it?
It is, but not because of fairing, it's because of a decimal point in a turbulent factor value, which was 10 times bigger than it should be. This is why I always double check my first simulation, mistakes like that creep up all the time.
Having said that, here's hoping this discussion will not be resurrected.
Based on these simulations, there is no reason to believe there is much more to these mirrors other than drag and turbulent wake reduction.