Khamsin Virtual Racecar Challenge 2015

[julien.decharentenay]

Is there any tutorial on how to run simulations with this package? I am quite savvy on computer stuff, but never tried Linux or this kind of stuff before, so could do with few guidelines!

@MadMatt The package is all based on Windows - no Linux command line. All the packages are Windows based - which is part of the challenge. Once you are past the third party installation issue(s), the GUI is pretty simple - as there is none (but neither is there any command line). The short story is: you open Windows File Explorer, right-click on the geometry file and select One-Click CFD>100 mph (local). It (in theory) starts, solves and outputs an html summary and the raw CFD results (in openFoam format)... The two (very) weaks points are: the geometry is to follow a number of conventions that are not described at the moment, and the installation of the third parties.

[julien.decharentenay]

Thank you, I tried with a different name and correct units: I receive a "geometry error" message.

Below is an extract of the first few lines of the STL file:

Code: Select all

solid "test"
	facet normal -1 1.87674541718276E-14 6.92113860124088E-15
		outer loop
			vertex 2.80331313717852E-15 -0.99998383905012 0.143374672109153
			vertex 2.83106871279415E-15 -0.999444623862472 0.145922790317555
			vertex 2.78943534937071E-15 -0.99976614522601 0.140779239809608
		endloop
	endfacet
	facet normal -1 1.26630998931759E-14 7.30271943886638E-15

Remove the characters " (the two of them) on the first line and it should work.

[julien.decharentenay]

The boundary conditions for the engine and cooling intake and exhaust will be based on nominal flow rates. This is to ensure that the problem is well define (this should result in better convergence than using pressure outlets at intakes, in my opinion).

Although I am unsure of what are realistics values, I am proposing the following:
- Engine intakes and exhausts (combined): 0.252 m3/s ( ~ 0.3 kg/s), corresponding to a face velocity of 100m/s at the exhaust;
- Cooling intakes and exhausts (combined): 1.5 m3/s (~ 1.8 kg/s), corresponding to a face velocity of 20m/s at the exhaust.

Let me know if you have any comment/suggestions. The value will be easy to change, but I have a bit of coding to do...

[CAEdevice]

hi Julien, just a curiosity: how can "one clik cfd" identify the inlet/outlet surfaces? Is it an automated process?

[CAEdevice]

Another question: the center of mass COM has been estasblished?

[CAEdevice]

About the flow rates: I agree with you about the convergence, but the have an opinion about the values, I'd need to know something about the engine (max power, turbocharged or not).

[MadMatt]

I just wanted to point out that with the current rule book there is the possibility to have the rear view mirror positioned vertically (150mm height and 75mm width).

[cdsavage]

I just wanted to point out that with the current rule book there is the possibility to have the rear view mirror positioned vertically (150mm height and 75mm width).

The rulebook states "150mm wide and 75m high", so I would interpret that as being illegal. I will add a clarification anyway, stating that the 150mm dimension is in the X-axis.

[cdsavage]

hi Julien, just a curiosity: how can "one clik cfd" identify the inlet/outlet surfaces? Is it an automated process?

I believe the inlet and outlet surfaces will be identified the same way that the wheels and wings were identified in 2014.

Another question: the center of mass COM has been estasblished?

This will be decided at our next meeting.

[CAEdevice]

Thank you, I tried with a different name and correct units: I receive a "geometry error" message.

Below is an extract of the first few lines of the STL file:

Code: Select all

solid "test"
	facet normal -1 1.87674541718276E-14 6.92113860124088E-15
		outer loop
			vertex 2.80331313717852E-15 -0.99998383905012 0.143374672109153
			vertex 2.83106871279415E-15 -0.999444623862472 0.145922790317555
			vertex 2.78943534937071E-15 -0.99976614522601 0.140779239809608
		endloop
	endfacet
	facet normal -1 1.26630998931759E-14 7.30271943886638E-15

Remove the characters " (the two of them) on the first line and it should work.

It's running!!! =D> =D> =D>

SnappyHexMesh is using 5Gb of RAM
SImpleFoam requires about 1Gb per core, it seems that only 3 cores of 4 are used: this is good for me (I'm using the PC for working in the meanwhile). I think that with 8Gb of RAM the simulation can be completed easily.

PS: I'm using my 2014 (F1) car, I tink that the 2015 car (LMP1) will require less mesh elements.

EDIT: After SimpleFoam finshed the simulation, I received another "geometry error"
https://dl.dropboxusercontent.com/u/522 ... log.03.txt

[MadMatt]

Thank you Julien, I have tried to re-install everything on windows 7 64bits, and tried to run a simulation using One-Click-CFD, but I get the following error:



I have seen your naming convention that I have applied to my STL files (do units have to be [m] as well, or is [mm] fine?), and have my Body.stl file that has a symmetry applied to it. Not sure how to combine all the files together for the simulation tho. I tried to look into the 2014 thread, but no success.

[MadMatt]

Regarding aero balance, how will this be taken in account for the dynamic simulation? Do we have to match the weight distribution of the car otherwise we lose performance? We haven't been given that weight distribution btw. Being a pure central engine powered LMP1 vehicle, I would imagine the weight distribution would be around 40/60 (FR/RR).

[cdsavage]

There will be a specific optimum COP, as was the case in 2013 and 2014. Virtual Stopwatch takes this value into account, and you will lose performance the further away you are from the optimum. The value has not been supplied yet but it will be soon.

[julien.decharentenay]

@CAEdevice I could not find the error. From the log file, it looks like the simulation ran to completion without any issue. In regards to number of processors, this is based on (a) number of processors available on your machine, and (b) size of the case.

@MadMatt The issue is with the installation of ParaView. If you installed ParaView 4.2, could you install ParaView 4.1 in C:\Program Files (x86)\ParaView 4.1.0 (I am assuming you are running a 64b windows)? You can download it from http://www.paraview.org/download/ - select the v4.1 in the Version drop-down menu.

I have got around writing a manual as a slide-presentation. You can view it on the webpage, or directly at http://www.slideshare.net/JuliendeChare ... sers-guide.

@MadMatt The PI idea has some merit - although I am not sure how practical/feasible it is. The issue with the KVRC Challenge is that the computational resource demand is high, but only when the Challenge is running (typically for about 1 week after the race submission). This demand pattern makes it well suited for Cloud computing...

[julien.decharentenay]

I have seen your naming convention that I have applied to my STL files (do units have to be [m] as well, or is [mm] fine?), and have my Body.stl file that has a symmetry applied to it. Not sure how to combine all the files together for the simulation tho. I tried to look into the 2014 thread, but no success.

To answer all your questions:
1) Use "meter" units.
2) Do not applied the symmetry in the file that you submit (otherwise, you will end up with a quarter of the geometry only);
3) I am uploading a new version (rc2) at http://www.khamsinvirtualracecarchallen ... .0-rc2.zip - it is not fully tested yet (let me know if something does not work), but it should solve all the issue that @CAEdevice has previously and should support multiple STL (binary or ASCII) files . So put one part in one file and make sure that all the files are in the same directory. Then when starting the analysis, select the directory where all the files are (not an STL file) then right-click and One-Click CFD>100 mph.