[KVRC] Khamsin Virtual Racecar Challenge 2016

[cdsavage]

its just that my car is such a large file in sketchup. it takes ages sometimes to make simple changes. sketchup has an annoying limitation of only utilising one core... i'm currently waiting on a face intersection process while posting this...

Most of CAD tools use very few multicore features... but SketchUp uses the only core badly. It is not a suitable tool for mechanical/aerospace design.

If you don't like parametric CAD (freeCAD), my advice is to learn Blender (I think Mantium uses it, with excellent results!).

I would agree that Blender is a better option than Sketchup. However it can be a lot of work to get to grips with a new piece of (complex) software.

[AratzH]

[quote="CAEdevice"]Just submitted an updated version of the car: I will probably obtain a bit more than 9000N (with balancing issues even if not as much as the pre season race) but facing some risks about cooling.[/quote

This is going to be a tight race! My predictions (not OCCFD based) are close to that figure too.

Now some pictures of the baby:

[Alonso Fan]

sorry chris, I clicked on the submit button twice, was in a hurry, finally done

[CAEdevice]

Just submitted an updated version of the car: I will probably obtain a bit more than 9000N (with balancing issues even if not as much as the pre season race) but facing some risks about cooling.[/quote

This is going to be a tight race! My predictions (not OCCFD based) are close to that figure too.

Now some pictures of the baby:

http://postimg.org/image/eeidxm9n5/

http://postimg.org/image/y6305m0k1/

http://postimg.org/image/r4vftw4ep/

I have always liked your cars, but this year your project is even more realistic. I am sure that you have podium perspectives in the next race and you will one of the favorites for the first place in the last two races.

[CAEdevice]

Hi, after the submission for Race 1, I finally had some time to dedicate to OCCFD.

I have to thank Julien and Team HP (who suggested to check the version number of mpi_x64 and noticed the dll previously installed is not erased automatically: now multicore works perfectly)!
It also quite fast: only 10% slower than linux, with similar hardware.

The only two issues are:

1) (Very minor): the "test" option of OCCFD does not work
2) It sometimes happens that the post processing stops (I think when generating the images). It is quite strange, because it happens only half of the times. It is not a big issue, since I use Paraview to post process the simulation, but it would be useful to have an automated computation of pressure differential and COP. Would it be possible to include them in the log? I think that (for the KVRC purposes) a simple text report with forces, moments, COP, pressure differential would be enough, since Paraview is very easy to use for graphical post processing. This would make unnecessary some of the utilities that sometimes generate problems.

[RicME85]

Test works on AWS btw so its only a local issue

[Alonso Fan]

@CAEdevice

Where do I find the .dll to erase it? And which is the version of mpi_x64 you have installed?

[CAEdevice]

@CAEdevice

Where do I find the .dll to erase it? And which is the version of mpi_x64 you have installed?

I am not at the pc now, but you can find it in the previous pages. You should look for a message by HP team.

[CAEdevice]

Uninstall any version of MSMPI you have on your computer and make sure that these files are deleted:

- Mpiexec.exe
- Smpd.exe
- Msmpi.dll

the first 2 are found in the installation directory of MSMPI and the last is in C:/Windows/System32

On my PC the uninstall routine didn't delete Msmpi.dll for some reason...

Then you just need to install this MSMPI into it's standard installation directory:

https://www.microsoft.com/en-us/downloa ... x?id=36045

To check your installation you can look at the file versions of the 3 mentioned files. They should all be 4.0.3906.0


Alex

[Alonso Fan]

Thanks, will do this as soon as I get home

[HP-Racing]

Hi, after the submission for Race 1, I finally had some time to dedicate to OCCFD.

I have to thank Julien and Team HP (who suggested to check the version number of mpi_x64 and noticed the dll previously installed is not erased automatically: now multicore works perfectly)!
It also quite fast: only 10% slower than linux, with similar hardware.

The only two issues are:

1) (Very minor): the "test" option of OCCFD does not work
2) It sometimes happens that the post processing stops (I think when generating the images). It is quite strange, because it happens only half of the times. It is not a big issue, since I use Paraview to post process the simulation, but it would be useful to have an automated computation of pressure differential and COP. Would it be possible to include them in the log? I think that (for the KVRC purposes) a simple text report with forces, moments, COP, pressure differential would be enough, since Paraview is very easy to use for graphical post processing. This would make unnecessary some of the utilities that sometimes generate problems.

No Problem, we're happy to help.

We also have the same post processing issue, but only with the "light" option.

[CAEdevice]

I noticed that things go better with Paraview working in single core mode (only Paraview), but I am not sure it solves everything.

[Alonso Fan]

@CAEdevice, how long did occfd in parallel take for you compared to single core?

and what are your pc specs?

[CAEdevice]

@CAEdevice, how long did occfd in parallel take for you compared to single core?

and what are your pc specs?

Single core: 9h
Multicore: 5h

Hardware: i7 3820 + 32gb RAM (total 64gb, but I use 32gb as a ramdisk for other tasks, different from occfd).

[Alonso Fan]

Thanks

Jeez man that must have taken a large chunk of your wallet

Jokes aside, so if I get it working i should be able to half the time roughly. So about 14 hours instead of 24...