Khamsin Virtual Racecar Challenge 2015

[CAEdevice]

Matteo time for a new machine? Get the i7 4770K, it is a nice CPU and I am running CFD on a 4.2 millions cells model in 9h45 with 3000 iterations I have now upgraded the cooling to water, so I could overclock the CPU and now I am running the same 3000 iterations in 7h15!

The i7 4770k is great! But I'm okay with the performance: in the office I have a "old" i7-3820 oveclocked @4.0gHz with two SSD (Raid0) and 32gb RAM + 32gb RAM disk (R:), anyway I was fond of the old machine (Q6600), it was my first "serious" workstation in 2007

Regarding the 2015 calendar, we are announcing that the season will consist of 6 rounds, spread over a longer time period compared to 2014. These are the submission dates for each round:

R1 - 2nd April (high downforce)
R2 - 14th May (high downforce)
R3 - 25th June (high downforce)

R4 - 6th August (medium/low downforce)
R5 - 17th September (medium/low downforce)
R6 - 5th November (medium/low downforce)

At the moment, the intention is for 2 out of the 6 rounds to be non-points scoring, 'practice' rounds. The actual tracks for each round are TBA, but we are aiming for the tracks within each of the 2 groups to all be close together in setup requirements.

I agree with MadMatt: the idea of spreading the races on a longer period is excellent.

About the "non-point scoring" ronds I have some doubts. Why don't consider e different (and increasing) distribution of the points? Ex. R1 and R4 half the points than R2, R3, R5, R6. So R1 and R4 would be good to practice less important but nobody could use (for example) a "dummy" car to not reveal too much.

Why high and low DF races are grouped? What about mixing the two kind of track?
I remeber that someone (maybe Chris, I'm not sure), during 2014, talked about this, but I can't find where: anyway, I wonder why there are only two track types? My classification would be:

- extreme high downforce: a track where drag has nearly zero influence (es. Monaco)
- medium dowforce: a track where df is much more important than drag, but drag can't be anyway high
- extreme low downforce: a trace where df is less important thatn drag reduction (ex. 2km long straight + slow 180° corner + 2km long straight + slow 180° corner)

In other words: I would group the tracks considering three and not two intervals for the slope of the track graph provided by Competition Car Engineering.

[MadMatt]

Shame for your computer Matteo. What did break down? Any way to fix it? The CPU on my laptop broke last year after 4 years of heavy usage, I purchased a second hand identical one on ebay for 8euros and it fixed the problem!

I agree with mixing the tracks, and I would also agree to add a medium DF track, but only 1, so the calendar would be:

R1 - Practice round

R2 - High DF
R3 - Low DF
R4 - Medium DF

R5 - Low DF
R6 - Medium DF
R7 - High DF

Have R1 as a non-counting round with no points, but Matteo's idea of half points is also ok, although I think 1 practice round is enough

[CAEdevice]

Have R1 as a non-counting round with no points, but Matteo's idea of half points is also ok, although I think 1 practice round is enough

I think the same, 1 practice race could be enough, but with a nearer date (ex. the end of February or the first week in March: no need to have the car completele refined, it would be just to test the CFD environment, the rules, the geometry check process).

@MadMatt
The problem was with one of the two 10000rpm Raptor disk (mounted in raid0): this kind of hd was the faster SATA at the time, but it produced a lot of heat and after so many working hours it failed. The CPU cooling system had problems too and while reparing it I think that an electrostatic charge dameged tu CPU... not worth to look for a replacement, the motherboard, a 775, is too old.

[cdsavage]

Excellent Chris, Why not using R1 and R4 as practice runs? That would be a good idea to gauge the cars and check that all parameters and details are working ok.

This is probably what we'll end up doing. I think it's important to have a practice round before the medium/low downforce rounds, since these rounds are more sensitive to getting the setup right. For the higher downforce rounds, there's less of a need to worry about undershooting or overshooting the optimum setup, but it would be nice to start things off with a non-championship round to give the first-time competitors a chance to get used to things.

[cdsavage]

I agree with MadMatt: the idea of spreading the races on a longer period is excellent.

About the "non-point scoring" ronds I have some doubts. Why don't consider e different (and increasing) distribution of the points? Ex. R1 and R4 half the points than R2, R3, R5, R6. So R1 and R4 would be good to practice less important but nobody could use (for example) a "dummy" car to not reveal too much.

Why high and low DF races are grouped? What about mixing the two kind of track?
I remeber that someone (maybe Chris, I'm not sure), during 2014, talked about this, but I can't find where: anyway, I wonder why there are only two track types? My classification would be:

- extreme high downforce: a track where drag has nearly zero influence (es. Monaco)
- medium dowforce: a track where df is much more important than drag, but drag can't be anyway high
- extreme low downforce: a trace where df is less important thatn drag reduction (ex. 2km long straight + slow 180° corner + 2km long straight + slow 180° corner)

In other words: I would group the tracks considering three and not two intervals for the slope of the track graph provided by Competition Car Engineering.

We originally intended to aim for 3 groups of tracks like you've said, but considering the overall number of rounds, we thought 2 was best. One of the goals this year is to make things a bit more competitive for those without big CFD resources, and part of this is reducing the need to constantly re-balance the car and find the best downforce level for every round. This way you have a high-downforce setup which you refine over a few rounds, and the same for the lower-downforce setup.

The practice round before the low-downforce rounds is mostly intended to allow everybody to get an idea of where the setup needs to be for the points-scoring rounds. We probably won't be releasing the laptime graphs which were supplied in 2013 and 2014, which makes this round more important.

If there is a concern that there are too few points-scoring rounds, then we could think about making the high-downforce practice round a points-scoring round instead.

[CAEdevice]

yes, 5 point scoring rounds would be better than 4 in my opinion.

[MadMatt]

Chris, it makes sense, and anything that can help reducing the advantage people running CFD simulations at home have is welcome! It would be nice to hear other people as well!

[CAEdevice]

Hi Julien,

I was using Paraview 4.1, so after unistalling it, I installed Paraview 3.98.
In that case the simulation stoppes after a few seconds: "pvbatch can not be found"

Where is ParaView 3.98 installed? In my case it is installed in (and it is found automatically):

Code: Select all

C:\Program Files (x86)\ParaView 3.98.0

Hi Julien,

I could not test the previous command because I had erased the result, but I think that Paraview could not be found beacuse it was the 3.98.1 release and the installation folder was:

Code: Select all

C:\Program Files (x86)\ParaView 3.98.1

.

Finally I installed Paraview 3.98.0 (installed in:

Code: Select all

C:\Program Files (x86)\ParaView 3.98.0

and I obtained a different problem during running: https://dl.dropboxusercontent.com/u/522 ... rgence.txt

The STL is the same of the previous test (the one completed without refinement).

[CAEdevice]

We originally intended to aim for 3 groups of tracks like you've said, but considering the overall number of rounds, we thought 2 was best. One of the goals this year is to make things a bit more competitive for those without big CFD resources, and part of this is reducing the need to constantly re-balance the car and find the best downforce level for every round. This way you have a high-downforce setup which you refine over a few rounds, and the same for the lower-downforce setup.

The practice round before the low-downforce rounds is mostly intended to allow everybody to get an idea of where the setup needs to be for the points-scoring rounds. We probably won't be releasing the laptime graphs which were supplied in 2013 and 2014, which makes this round more important.

If there is a concern that there are too few points-scoring rounds, then we could think about making the high-downforce practice round a points-scoring round instead.

I'm not convinced by the aim of make things more competitive for those without big CFD resources: last year it was different, but this year, once the "One click CFD" framework is completed, I think that everyone with a PC enough powerful to use a CAD as SolidWorks or any simulation game could be used to run a CFD test. The most interesting part of the KVRC is, in my point of view, the CFD challange (not the CAD modeling only).
Anyway: if the track charts will not be released, it could balance that point, so non problem for me.

I still think that two no scoring points races would be too much, I would prefer two "low" scoring races (maybe only one point to the winner) or one or two "free" run (including the HTML results) that each team could require during the season.

Another thing that could help, would be to include some automated measueres in the HTM report (es. floor DF, rear wing DF, ...) but I imagine that it would require some more components grouping in the STL file.

[RicME85]

Not everyone wants to have their average pc held up for the next 10 to 20 hours whilst it runs a simulation. I am at home pretty much all day every day due to my health so use my PC for gaming and media streaming. If I have a simulation running then that puts that out the window. I no longer have my Q6600 system I had been running the last few years of KVRC and because of budget restraints because of my lack of working I only have an i3 so not sure how well it will perform a simulation in comparison.
Im with Chris on this, the proposal is good, I dont want to be changing between configurations each week, I would prefer to work across races to perfect things then change to a different downforce level.

I dont understand the need for a point or 1 2 and 3 points for the podium places in the practice round, those couple of points arent going to make any real difference in the end.

[CAEdevice]

Not everyone wants to have their average pc held up for the next 10 to 20 hours whilst it runs a simulation. I am at home pretty much all day every day due to my health so use my PC for gaming and media streaming. If I have a simulation running then that puts that out the window. I no longer have my Q6600 system I had been running the last few years of KVRC and because of budget restraints because of my lack of working I only have an i3 so not sure how well it will perform a simulation in comparison.
Im with Chris on this, the proposal is good, I dont want to be changing between configurations each week, I would prefer to work across races to perfect things then change to a different downforce level.

I dont understand the need for a point or 1 2 and 3 points for the podium places in the practice round, those couple of points arent going to make any real difference in the end.

Sorry for your health. Anyway, indipendently from the "free practice" sessions, I think that the idea of allowing one (or two) free "private" run could help you and people who have not enough "CPU power". I i were not involved in the race if I were not involved in the race I would make available my workstation, but we could find a workaround (es. sponsorship for Amazon server: let me know, I just I do not want to be in conflict of interest )

[RicME85]

It all comes down to cost at the end of the day though doesnt it.
If there is a way for simulations to be run remotely that isnt going to cost the earth then that would be fantastic.

[CAEdevice]

It all comes down to cost at the end of the day though doesnt it.
If there is a way for simulations to be run remotely that isnt going to cost the earth then that would be fantastic.

I remeber that during 2014 or 2013 someone talked about the costs of remote server hiring, and they weren't too high.
If someone could help me to extimate that costs, I could think to a sponsorship limited to the possibility to give everyone access to one or twe "free" simulations during the season.

PS: You had a Q6600 too? My one just failed, anyway great processor!

[cdsavage]

Julien has made One-Click CFD compatible with Amazon web services, so you can run simulations remotely at a per-hour cost. There are some details on this page - http://www.khamsinvirtualracecarchallen ... -click-cfd. The prices are quite affordable I think, depending on how many simulations you intend to run. You should contact Julien if you are interested.

[RicME85]

PS: You had a Q6600 too? My one just failed, anyway great processor!

Yeah had a Q6600 G0. Bought it 7 or so years ago, was a fantastic processor, would still use it now if there were more up to date boards that had the 775 socket.