[KVRC] Khamsin Virtual Racecar Challenge 2016

[RicME85]

Did anyone have massive problems when upgrading to the new one-click? I'm running 1.1-000

Hi, the lastest release is only running single core at the moment.

Ok, is there a way to parallel install? I have 8 cores that can run the current one-click in about 3hrs.

How easy is amazon to set up? How long do those simulations usually take?

AWS is easy to set up. Takes longer than you are used to, have a sim running now that so far has taken 5hrs. There is also the cost to consider.

[SR71]

Hi, the lastest release is only running single core at the moment.

Ok, is there a way to parallel install? I have 8 cores that can run the current one-click in about 3hrs.

How easy is amazon to set up? How long do those simulations usually take?

AWS is easy to set up. Takes longer than you are used to, have a sim running now that so far has taken 5hrs. There is also the cost to consider.

Thanks, that's insightful. I could see the benefit of being able to run multiple simulations at the same time. Is this possible with amazon? Can I have 4 running at a time for instance? That would be very attractive to me.

Also, does anyone know about parallel installing One-click? I'd like to keep 8 cores for the bodywork. use the single core for cooling/etc.

[RicME85]

Can have as many as you want. I have two on the go now.
It uses m3.large - Intel Xeon E5-2670 v2 (Ivy Bridge) - 2 vCPU -7.5 Mem (GiB) - $0.133 per Hour
I would be interested to know how long a simulation would take on m3.2xlarge which is 8 vCPU and 30 Men (GiB). That one costs around $0.5 per hour but could work out cheaper in the long term if it cuts down simulation time.

[RicME85]

Sorry, got that wrong, its m3.xlarge which is 4 vCPU, 15 Men (GiB) and costs $0.266 per hour

[etsmc]

just noticed a new version of OCCFD on the official site its version 1.1-009 what's new in this one??
I still have issues its now completing simulations but I get a can't find files in C:\DBD\pvs-x64\...\...\..\...
is this something to do with the wrong version of openmpi

[SR71]

Sorry, got that wrong, its m3.xlarge which is 4 vCPU, 15 Men (GiB) and costs $0.266 per hour

Nice, guess I'll be eating ramen soup to make sure my car is virtually fast!

I'm gonna try it out. I usually simulate at night when I'm sleeping, I'd rather wake up to 5 results than 1.

[SR71]

just noticed a new version of OCCFD on the official site its version 1.1-009 what's new in this one??
I still have issues its now completing simulations but I get a can't find files in C:\DBD\pvs-x64\...\...\..\...
is this something to do with the wrong version of openmpi

I'm nervous about the upgrade as well. Gonna give it a go.

BTW, when is the next race? Is the next race for points?

[etsmc]

April 14 is the next submission date for the first race which is for points

[CAEdevice]

Hi, I was thinking about a way to reduce the impact of the rule check and preprocessing.

My proposal is a rule check with no resubmission: a previuos submitted regular version car will be used or a penalty of 200N will be applied to downforce (the less favorable option would be chosen): only applied for the cars who take points.

[CAEdevice]

Another proposal is about the cooling devices: teams that can't run cfd have big problems (more than one team is going to have less than 50% of the engine poiwer available), so I would introduce a safety option, with no differential requirements, based on Opt1 but with much bigger inlets (100000mm2 projection on the frontal plane) and external templates (400mm extrusion). On the other side I would allow much more freedom (no minimum area inlets) for the teams that are compliant with the 2017 flows/differentials.

[etsmc]

The pressure at intake and exhaust are:
Engine intake, Area: 0.018m2 - Compliant
Surface integral of pressure: 4.35 Pa.m2
Engine exhaust, Area: 0.011m2 - Compliant
Surface integral of pressure: -4.59 Pa.m2
Cooling intake, Area: 0.401m2 - Compliant
Cooling exhaust, Area: 0.402m2 - Compliant
Differential of surface integral of pressure: -269.23 Pa.m2


A good start to the year now to get some more downforce.

can anyone explain the negative Differential of surface integral of pressure is this saying that air is actually being sucked out of my cooling intake and not reaching my heat exchanger?

[CAEdevice]

The pressure at intake and exhaust are:
Engine intake, Area: 0.018m2 - Compliant
Surface integral of pressure: 4.35 Pa.m2
Engine exhaust, Area: 0.011m2 - Compliant
Surface integral of pressure: -4.59 Pa.m2
Cooling intake, Area: 0.401m2 - Compliant
Cooling exhaust, Area: 0.402m2 - Compliant
Differential of surface integral of pressure: -269.23 Pa.m2


A good start to the year now to get some more downforce.

can anyone explain the negative Differential of surface integral of pressure is this saying that air is actually being sucked out of my cooling intake and not reaching my heat exchanger?

the presence of a negative differential is caused by the boundary condition (flow) applied to the inlet surface: it is more a mathematical effect than a physical fact. In the reality it means that the air flow would be (much) lower than the ideal (BC) one.

[cdsavage]

Hi everyone,

There was a big delay in finishing the legality checks - most of them have been sent out now, and the rest should be sent out tomorrow. The simulations have been completed already, since there was no resubmission for this round.

We are planning on pushing back the next round by a week, so that the submission deadline would be April 21. Is everyone OK with this?

Hi, I was thinking about a way to reduce the impact of the rule check and preprocessing.

My proposal is a rule check with no resubmission: a previuos submitted regular version car will be used or a penalty of 200N will be applied to downforce (the less favorable option would be chosen): only applied for the cars who take points.

What does everyone think about this? I would probably prefer using a time penalty rather than modifying the forces.

A different idea might be to keep the resubmission process, but impose a small penalty (something like 0.2% of the calculated lap time) for any entry that needs resubmission, as an added incentive to double-check the submission is legal before it is submitted. The full 1% penalty would apply if the resubmitted car is still illegal.

[LVDH]

Was there even a severe time problem on your side?
I always liked the process the way it was although you somehow always find something to nag about.

And when are we finally going to see who won the first race?
I cannot wait anymore.

[CAEdevice]

Hi everyone,

There was a big delay in finishing the legality checks - most of them have been sent out now, and the rest should be sent out tomorrow. The simulations have been completed already, since there was no resubmission for this round.

We are planning on pushing back the next round by a week, so that the submission deadline would be April 21. Is everyone OK with this?

Hi, I was thinking about a way to reduce the impact of the rule check and preprocessing.

My proposal is a rule check with no resubmission: a previuos submitted regular version car will be used or a penalty of 200N will be applied to downforce (the less favorable option would be chosen): only applied for the cars who take points.

What does everyone think about this? I would probably prefer using a time penalty rather than modifying the forces.

A different idea might be to keep the resubmission process, but impose a small penalty (something like 0.2% of the calculated lap time) for any entry that needs resubmission, as an added incentive to double-check the submission is legal before it is submitted. The full 1% penalty would apply if the resubmitted car is still illegal.

I would agree with the time penalty.

My proposal to re-use a previous submitted car, is to consider the case of a not legal detail that gives a big performance gain (more than 1%). Ex. A rear wing with a longitudinal section bigger than the allowed one, or a bigger diffuser or a slightly raked setup.