Project Falcon

Reply
Active Member
Posts: 7
Registered: ‎11-05-2012
Message 31 of 50 (311 Views)

Re: This Must Be The Place (Native Melody)

11-20-2012 11:01 PM in reply to: Henrik14

When it came to the videos i cheated. I just used fraps to record. The videos Falcon makes is too low resolution. And buggy. Just like Inventor. As you can see once falcon gets over the hill its not a problem. Hence the recording in full HD and music. All recorded live.

 

 i was able to use all 8gb of ram on the previous version.

Does the grapic card matter a lot? i have a Geforce 680 with 2gb ram and watercooling if the need to push things comes.

 

I wanted to go finer mesh but the new baby interface will not allow finer than the 200% whatever that is. i now sit on 16gb more ram and i am not sure i need it.

 

But i haven`t really been trying the new one since i don`t like the new interface, the simplifications and since it crashes a lot.

I will try more now that i can direct the flow indicators.

 

It should be a save preferences button. So it becomes easier to trow in a new version/evolution on the piece you are testing. And a flip model function. Its a pain to make the model in inventor and discover that the model is on its side. Go back, flip it and export as STL? file and try again. opsy it was the wrong way. and.. one more time...

 

Picture from inventor stress simulation is attached. I think most would enjoy just forces from top or sides. XY and so on seems to be both. Its not lean. Who seems to be the thing in falcons development.

Board Manager
Posts: 1,084
Registered: ‎11-13-2006
Message 32 of 50 (303 Views)

Re: This Must Be The Place (Native Melody)

11-21-2012 07:13 AM in reply to: Henrik14

I have opened up a separate thread so we can specifically discuss system requirements.

 

http://forums.autodesk.com/t5/Project-Falcon/system-requirements-for-Project-Falcon/td-p/3707606



Scott Sheppard
Program Manager
Autodesk Labs
Autodesk, Inc.
Contributor
Posts: 22
Registered: ‎03-29-2012
Message 33 of 50 (299 Views)

Re: This Must Be The Place (Native Melody)

11-21-2012 03:39 PM in reply to: scott.sheppard
Unfortunately I got message that "t5" cannot be found for this new thread......... -- Regards from Croatia, the Homeland of one of greatest inventors from 19th and 20th Centuries, Engineer Nikola Tesla!
New Member
Posts: 2
Registered: ‎11-22-2012
Message 34 of 50 (292 Views)

Re: This Must Be The Place (Native Melody)

11-22-2012 04:22 AM in reply to: matso

Greetings to the people of autodesk
I give thanks for having made available to the public the Falcon software with which I carried out some tests.
The graphical interface is very intuitive and easy to handle, the object import section is fast and effective, at least in the type of file that I used 3ds and impressive visual results for such a small program.
I've been using it for short periods of time due to the limited capabilities of my team dualcore, 4GB of RAM and 512MB of graphics. The main problem is the intensive use of CPU (100% whether or not doing calculations) and excessive overheating. The 80% of the sessions my computer is turned off automatically.
The software is stable and has posted only trying to save images. I have not tried to save video given the limitations.
Sorry for the google translation.

Member
claud4auto
Posts: 4
Registered: ‎01-21-2012
Message 35 of 50 (265 Views)

Re: This Must Be The Place (Native Melody)

12-07-2012 02:52 PM in reply to: scott.sheppard

Scott

two issues:

1) I have reproduced a crash every time I change the voxel size, program crashes immediately. 

2) the drag calculation is overstimated by a factor of 2 (Cd=.48 instead of .25) in the same model under identical set up to another cfd.

 

I do not know if the voxel size can be changed any other way to get accuracy improvements. However, I do not believe that the default voxel will change the drag by the mentioned factor as noted.

Please let me know.

 

I use a Mac Pro with 28 gig ram.

 

Regards.

Contributor
Posts: 22
Registered: ‎03-29-2012
Message 36 of 50 (261 Views)

FALCON 2013

12-07-2012 03:58 PM in reply to: matso

Just today I have found out that some program wongly detect screen resolution of my computer which is 1280x1024 as 867x764 or similar. Maybe this is a problem? Did You try to run .stl file I sent you? Long time no answers are posted nor emailed to me. Could You please find some solution so I could use FALCON 2013? As I wrote, even reverting on EXPLORER 8 did  not helped, nor total deinstallation of previous version of FALCON, nor installing it again. I really do not know what to try next.........

Employee
hartogj
Posts: 152
Registered: ‎08-03-2011
Message 37 of 50 (254 Views)

Re: FALCON 2013

12-08-2012 06:34 AM in reply to: Henrik14

@claud4auto:

 

for pt 1, can you describe the exact steps that cause the crash when changing voxel size, ideally using one of the sample models so we can recreate it?

 

for pt 2, an important part of Project Falcon is validation against theory and physical test results. a brief validation doc is available here for your reference: http://labs.blogs.com/labs/ProjectFalcon2013Validation.pdf. you should be seeing closer results than what you reported. if not, please provide more detail in what you are seeing.

 

@henrik:

 

from your prior posts it's clear that you are having issues, but not clear exactly what system you are running and what the circumstances of the problems are. it sounds like an isolated issue and it's probably best to use another machine. however, if you need to run Falcon on this one machine that has been problematic, please email labs.alias.falcon@autodesk.com a succinct description of the OS, locale, graphics card, graphics driver, screen resolution, memory and exact steps for reproducing the issues you are having. we cannot guarantee that we can solve the problem on your system but can look into it.

New Member
Posts: 2
Registered: ‎11-22-2012
Message 38 of 50 (242 Views)

Re: This Must Be The Place (Native Melody)

12-10-2012 10:52 AM in reply to: scott.sheppard

Obviously I've had a cooling problem is already solved and I could do more intensive testing of the software.
At the moment I need to know what  computation time needed Falcon to obtain reliable results.

I left work a Falcon for periods of 10-15 minutes and the results are graphical computation time of 90 sec.
Using 3ds files around 2MB
My system is win7 32bit Cuore2duo, 4Gb ram using the half, 512MB graphics.
I tried to record video. Mov and have not succeeded. Save images as. Jpg hangs Falcon. I get images with screenshots.
Thanks

Member
claud4auto
Posts: 4
Registered: ‎01-21-2012
Message 39 of 50 (235 Views)

Re: FALCON 2013

12-10-2012 12:11 PM in reply to: hartogj

Ex:

1) I use the voxel size 'dial' to increase to 200% ...as soon as I do that ...crash app.

2) If I use the dial to decrease to 40% or less, then it run and became 'unstable' and calculations fail after awhile...i.e. giving negative drag values and finally crash. This one I understand it more, it has problems converging a very fine voxel size and since NO other controls exist to hand like time step.

 

Conclusion: as normally happens with grid sizes, voxel sizes matters on getting the solutions a) to run and b) converge.

 

The larger problem is that there is not much control if you want to increase the results accuracy.

 

3) I mentioned the results because I expected to have a much closer values. Maybe you should test on more elaborated shapes, not only on basic shapes where the errors may not be apparent. Fluids do not  behave the same at different regimes but also at different shapes, just to 'generalize'...

 

My suggestion is: test Falcon with more elaborated shapes, ex. wings.

I just did that, a full aircraft and got the 'issues'. Note: NOT very rare Reynolds, around 2-10 million.

 

Regards.

 

 

 

 

Employee
hartogj
Posts: 152
Registered: ‎08-03-2011
Message 40 of 50 (232 Views)

Re: This Must Be The Place (Native Melody)

12-10-2012 12:27 PM in reply to: matso

@mrgm05

 

It sounds like you are near the limit of what your system can handle, given what you mentioned about the overheating.

 

Falcon is a transient solver. It will capture unsteady flow phenomena more completely than a steady state solution which is more time-averaged. Given that is is fully transient, there is no final result. However, results should stabilize after enough time has elapsed. You can review the "time" output in the UI and see how much sim time has elapsed. Usually a few seconds are necessary. Additionally, you can review the drag plot and the average drag value to see when it levels off. We are looking at ways to automatically provide notice.

 

@claud

 

Thanks for the description. We'll take a look.

You are not logged in.

Log into access your profile, ask and answer questions, share ideas and more. Haven't signed up yet? Register

Announcements
Visit Autodesk Labs
Autodesk Labs is home for new prototypes, experiments, applications, and technologies.

Need installation help?

Start with some of our most frequented solutions to get help installing your software.

Ask the Community