[MVRC] Mantium Virtual Racecar Challenge 2016

[etsmc]

Wow 5th place is a surprise thought i had made a small improvement but not that much. my guess is i got near to 100% engine power.
I know i had a penalty but how much was it?

Not sure how much time i will have before the next round to fix things up and make any changes.

Everything can be found at the challenge page, calendar and results. I am too lazy to make a screenshot of the table so just head over and have a look. The penalty time was 107% of the best time as the average time was slower then that.
http://mantiumchallenge.com/mvrc-calender-results/

the table on the results page seems to be too wide...not all of it is showing

[CAEdevice]

A pity that cooling issues are still conditioning the results.

Cooling is part of the design
Look at KVRC data: our cars looked like spaceships with df/dr ratio around for the best ones... 2nd order scheme will be an important further improvement about realism.

Honestly you were right and JJR car was the best car in Sepang: he only choose the wrong (unbalanced) setup, but with a bit less rear wing he would have won.

[LVDH]

I always check the results with 100% engine performance and with the de-rated settings. The car that suffers the most is always Variante. Maybe this is an indicator of why the real life cars do not have such super crazy front wings. If you sort the data from this race by Cl/Cd, Variante would have been P5 which also also below what is expected for him. The design was simply not very good, with or without cooling.

BTW: Le Mans is available for testing in the virtual stop watch. For that race I have some highlights planned so I hope everyone is preparing the cars already.

[JJR]

A pity that cooling issues are still conditioning the results.

Honestly you were right and JJR car was the best car in Sepang: he only choose the wrong (unbalanced) setup, but with a bit less rear wing he would have won.

But I don ´t understand it because my own simulation was in 1,65 range. For high downforce tracks it was OK but for efficient tracks suddenly not. So I must find why.

[LVDH]

I was very surprised by your results as well as it was not your style.
The issue with your car was that the forces were not very flat, so probably some transient effects are going on with your car. To keep results comparable I cannot just run your car transient, so I let it run twice as long as usual and the forces are reasonably flat. If your own simulations showed better behavior maybe car is very close to the edge and small differences in mesh setup cause very different results. If you like you can send me your own simulation assuming you are using OF and I will have a closer look into this.

[variante]

Cooling is part of the design

I always check the results with 100% engine performance and with the de-rated settings. The car that suffers the most is always Variante. Maybe this is an indicator of why the real life cars do not have such super crazy front wings. If you sort the data from this race by Cl/Cd, Variante would have been P5 which also also below what is expected for him. The design was simply not very good, with or without cooling.

Yo, I wasn't complaining, you malicious. It was just an observation: half of the grid is suffering heavy cooling issues.
If you want me to complain, I could say that I've spent all my available simulations trying to figure out how WFlow/OF "thinks" (i.e. how it solves meshing on flat surfaces, how high is the porosity of the heat exchanger,...). I even thought that I did solve all the issues, as I took many precautions to prevent cooling problems (I was getting a very safe 8m3/s of airflow). Even the front wing LVDH always complains about was severely cut to increase airflow, and despite that I'm not getting better cooling compared to the first race where I was using the giant snow plow...

[CAEdevice]

Cooling is part of the design

I always check the results with 100% engine performance and with the de-rated settings. The car that suffers the most is always Variante. Maybe this is an indicator of why the real life cars do not have such super crazy front wings. If you sort the data from this race by Cl/Cd, Variante would have been P5 which also also below what is expected for him. The design was simply not very good, with or without cooling.

Yo, I wasn't complaining, you malicious. It was just an observation: half of the grid is suffering heavy cooling issues.
If you want me to complain, I could say that I've spent all my available simulations trying to figure out how WFlow/OF "thinks" (i.e. how it solves meshing on flat surfaces, how high is the porosity of the heat exchanger,...). I even thought that I did solve all the issues, as I took many precautions to prevent cooling problems (I was getting a very safe 8m3/s of airflow). Even the front wing LVDH always complains about was severely cut to increase airflow, and despite that I'm not getting better cooling compared to the first race where I was using the giant snow plow...

I know you were not complaining I was not malicious about you, but about a user that I discussed with and maybe he is reading now

If you had 8 m^3/s of air flow there must be something that doesn't work with your geometry/simulation. I had differences too between my solution and the official one, but very small (my solution: 3.1 m^3/s, the official one, slightly less than 3.0m^3/s) and acceptable.

[AratzH]

First congrats to CAEdevice for a well deserved win. It is also good to see that all cars are getting much closer, this makes everything much more interesting.
Regarding the results I also found divergences between the official and personal runs. In my case I think I am in the same situation as Matteo, I had 3,1 m3/s of cooling vs something lower that penalized me a little. Also it is interesting that even though my overall L/D matches in my tests and on the official results, my actual L and D numbers where lower.
I think this falls into the normal dispersion of CFD. I suspect also that using different number of cores gives slightly different results. I use 6 at day and 8 at night and I could clearly see a difference.

In any case a very interesting race and happy to see that my "radical" approach has potential. "Le Mans is going to be an interesting race...

[LVDH]

I suspect also that using different number of cores gives slightly different results. I use 6 at day and 8 at night and I could clearly see a difference.

This is interesting. It is a theoretical issue in CFD. Every time I get strange results in my work I check exactly that and it never proved to be the cause of my issues. I have noticed thanks to the logs of you guys that sometimes my results are not 100% what yours are. I thought it was a Windows / Linux thing.
I hope to soon get the online version done. Then we will all have the absolut exact setup.

[LVDH]

Hi,
I just noticed that there were many questions in the support forum which I did not answer. Somehow they did not show up properly. I only saw them today.
You can now find help if you still need it.
Also I have figured out what went wrong with Variante and his cooling flow. The problem is how monitoring surfaces are treated in OF. He made his surface slightly expand the limits of the radiator to make sure to cover every last piece of cooling flow. He also did not divide the two surfaces into a left and right side but kept both as one file. OF seems to guess values which are not in the domain which creates misleading results. I was not aware of this either.
The lesson is that the monitoring surface should match the geometrical extends it is supposed to measure and to divide it into a left and right hand side part. As we are only using the right hand side only using the one for the right hand side is enough. And on another note: The right hand side is the one on the right when sitting in the car (Y>0).

[CAEdevice]

Hi, I was thinking about two simple features to add to WFlow:

• I'd like to include the "reconstruct" commands directly into the launch file
• It would be great to have the possibility to know df and dr coefficients of single components (es. rear wing). I know I can do it in Paraview, but to have the data in a lg file would help

[Ft5fTL]

Hi, I was thinking about two simple features to add to WFlow:

• It would be great to have the possibility to know df and dr coefficients of single components (es. rear wing). I know I can do it in Paraview, but to have the data in a lg file would help

Thats a good idea. I have no clue about calculating df on paraview.

About cooling; is it normal that having negative values on engine exhaust parameter? I'm getting all the time after switching to wflow. My cooling numbers are this at the moment:

engine-exhaust, -0.376287[Pa*m^2]
engine-intake, 16.1303[Pa*m^2]
mSurf_mid-surface, 6.94291[m^3/s]

[CAEdevice]

Hi, I was thinking about two simple features to add to WFlow:

• It would be great to have the possibility to know df and dr coefficients of single components (es. rear wing). I know I can do it in Paraview, but to have the data in a lg file would help

Thats a good idea. I have no clue about calculating df on paraview.

About cooling; is it normal that having negative values on engine exhaust parameter? I'm getting all the time after switching to wflow. My cooling numbers are this at the moment:

engine-exhaust, -0.376287[Pa*m^2]
engine-intake, 16.1303[Pa*m^2]
mSurf_mid-surface, 6.94291[m^3/s]

Engine exhausts and intake are quite normal.
msurf flow is very high, but I am not sure if it is a issue (the of Variante), without knowing your geometry. Let's wait for LDVH.

[LVDH]

Yes the values that end with Pa*m^2 can be either positive or negative depending on the average pressure on that surface. For the engine intake you would want it to be positive and for the exhaust outlet negative. You should read the rule book it tells you the exact limits.
The monitoring surfaces where the values end with m^3/s are a bit confusing as the value depends on the orientation of you monitoring surface. As long as the flow has a good general direction, for example in the heat exchanger itself as it straightens the flow very much a "+" or "-" does not make much sense. At some point I think I changed the post-processing tool to only show the absolute value to remove this source of confusion. Some things are for me so normal that I do not think about them anymore, so your questions are all good to help me make the tools more usable for non-CFD people.

[Ft5fTL]

Thanks for the info. After submition for Sepang, i figured out that i need a more "direct" aproach to cooling so i changed everything about the cooling parts without waiting the results. The car looks like BMW v12 LM at the moment.