Author Topic: AutoDesk Flow Design / CFD Simulation  (Read 11775 times)

0 Members and 2 Guests are viewing this topic.

Offline bubruins

  • Full Member
  • ***
  • Posts: 106
AutoDesk Flow Design / CFD Simulation
« on: December 01, 2014, 02:44:14 PM »
Has anyone here tried using AutoDesk Flow Design for improving the shape of their vehicle?

It looks like it can show the flow dynamics and Cd at different air speeds. I am helping a team that is early in the planning phase for a car. I have limited 3D modeling experience (Sketchup) and am learning Solidworks.  The game plan is to create a really good model of the car we're building and a few alternative scenarios (hood scoop styles, ground clearance, etc...), then export to Autodesk Flow Design as a .DWG file, and learn as much as we can before building a model for 1:18 scale wind tunnel testing.

Is there anything I should be cautious of when using this software? I'm going to guess that the Cd will not necessarily be true to real world figures that would be gathered in a 1:1 wind tunnel. We are planning on bench-marking performance based on relative gains instead of whatever Cd number it spits out.

If you don't use Flow Design but use a similar program, please chime in and explain what you use and why you like it.
« Last Edit: December 01, 2014, 02:51:44 PM by bubruins »

Offline WOODY@DDLLC

  • Hero Member
  • *****
  • Posts: 1803
  • ECTA made it to AR-Kansas!
    • Design Dreams, LLC
Re: AutoDesk Flow Design / CFD Simulation
« Reply #1 on: December 01, 2014, 03:50:41 PM »
Check my webpage - I have a head start on you!  :-o :-D
All models are wrong, but some are useful! G.E. Box (1967) www.designdreams.biz

Offline bubruins

  • Full Member
  • ***
  • Posts: 106
Re: AutoDesk Flow Design / CFD Simulation
« Reply #2 on: December 01, 2014, 04:12:17 PM »
Woody, no doubt you have a head start. I have checked out every page on your site. Can you elaborate on what software you use without giving away any trade secrets? Unfortunately I'm just a guy sitting at home that plays engineer on nights and weekends.

Offline WOODY@DDLLC

  • Hero Member
  • *****
  • Posts: 1803
  • ECTA made it to AR-Kansas!
    • Design Dreams, LLC
Re: AutoDesk Flow Design / CFD Simulation
« Reply #3 on: December 01, 2014, 09:13:31 PM »
All CFD software allows you to build a digital wind tunnel to test your tater! I use the FlowSimulation add-in for SolidWorks. The AutoDesk Flow Design used to be called CF Design before AutoDesk acquired them. Methods and interface - same thing only different! It took me about 3 months to get my first results that I thought were useful but then I had to say, "Compared to what?" Most of that 3 months was reviewing all my fluid mechanics lessons and books so I could attempt to understand what I was really looking at. I've been at it for 8+ years now and I'm still learning something every time I run another lump! Like many at first I only looked at drag and lift which was a huge mistake! Drag is the easy part! What degree of aero stability will you have? Where is the center of pressure? What does that really mean? What are the pitch, yaw and roll moments? Resulting aero reactions on the tires? Trust me it's complicated and there are no single good references to point you to! But you are on the right track with CFD. CFD helps you understand the why not just the what! There are some programs out there that are cheaper than either package but each has it's own limitations. Get some quotes first and price a workstation! $$,$$$.$$ Then you'll spend a whole lot of hours pondering your results and trying out every anecdotal B'ville factoid you have ever heard!  :-o BTW: I would not spend time or money on 1:18 wind tunnel testing. CFD is full scale testing and depending on the model can be very close to wind tunnel results. I have done a couple of wind tunnel follow-ups and were within 4%~10% on results. [The big guys are 1%~2%!] Use CFD and REAL wind tunnel time. CFD is best for concepts / comparisons and can tell you what NOT to take to the wind tunnel or track! Remember computer models are perfect but real world vehicles have lots of rivets, warts and wrinkles! Hope this helps - have fun!  :cheers: [Or do you want a quote? :-D]
All models are wrong, but some are useful! G.E. Box (1967) www.designdreams.biz

Offline bubruins

  • Full Member
  • ***
  • Posts: 106
Re: AutoDesk Flow Design / CFD Simulation
« Reply #4 on: December 01, 2014, 09:56:32 PM »
Awesome info. That's exactly the kind of experience I was hoping to hear! I wish I had access to SW Flow Simulation, but I am not in a position to fork out the $$,$$$ for it. AC Flow Design is $210/year and looks promising. I'll be building my first model and testing the software out in the coming months.

Back to that foolish rookie mistake of only examining Cd and lift at first... Flow Design has a Cd, but cannot provide lift. We are planning on optimizing the form the best we can given the drivetrain / general layout constraints we have and find a slippery shape. Then we want to do small scale tests in a 1:18 tunnel to get a relative measure of lift/downforce and go back and forth between a CFD program and 1:18 tunnel to optimize the shape the best we could before building a 1:1 form for the 1:1 $2000/try wind tunnel. Would that be a bad way to attack this on a budget? If not, how would you measure downforce/lift before going to a 1:1 tunnel?

We have a good idea of where the center of pressure and center of weight will be, but are still early in the planning process. I have not even began to start thinking about pitch, yaw, roll, and how we can prepare a good car for these factors.

Thanks for the information provided so far!

Offline WOODY@DDLLC

  • Hero Member
  • *****
  • Posts: 1803
  • ECTA made it to AR-Kansas!
    • Design Dreams, LLC
Re: AutoDesk Flow Design / CFD Simulation
« Reply #5 on: December 02, 2014, 09:18:01 AM »
Need to clarify something. What you are looking at is not the former CF Design as I first thought. Flow Design was formerly known as Project Falcon and this is the commercial version. I actually took part in the beta testing of this and spoke with the developers. I asked where's the lift and the ground plane? Looks like there may now be a ground plane and they have cleaned up the interface a lot. They have added some useful visual functionality. The solver is actually very powerful and visualizing the yaw effects even without all the data is still useful. You will see how much drag changes in yaw [or in a spin]. All the other torques and forces will be changing also! For conceptual and a first look it's worth a try! The other great thing about CFD is it will get you thinking in ways you would have never imagined before!  [After looking at the specs again, for $210/yr I will likely sign up, too!]

The biggest problem with scale model wind tunnels is similarity. At 1:18 for a real car @ 100 mph your wind tunnel speed needs to be 1800 mph!  :-o Supersonic not good!  :cry: If your wind tunnel speed is 100 mph your test speed is 100/18 = 5.56 mph.  :cry: The big guys use 40% or 50% scale models for a reason. Models have to be super accurate as well. Can still be an informative experience though! Does the tunnel exist or are you building it? This is what I would build if I were going the scale model route: http://www.f1technical.net/forum/viewtopic.php?f=1&t=9961
All models are wrong, but some are useful! G.E. Box (1967) www.designdreams.biz

Offline Peter Jack

  • Hero Member
  • *****
  • Posts: 3776
Re: AutoDesk Flow Design / CFD Simulation
« Reply #6 on: December 02, 2014, 11:08:41 AM »
It looks like the videos have been taken down Woody. That's really too bad.

Pete

Offline bubruins

  • Full Member
  • ***
  • Posts: 106
Re: AutoDesk Flow Design / CFD Simulation
« Reply #7 on: December 02, 2014, 11:32:21 AM »
I also could not access the videos.

Version 1 of the tunnel exists, but we don't have a formal fan set up or anything. Right now we're just using an air compressor hose and/or a small high velocity industrial fan. We are not measuring air speed at this point. Even with just the fan we are able to produce measurable amounts of downforce. With the air compressor hose it is difficult to lift the model off the surface because of the amount of downforce it creates (for the most radical design that has been built). At this point we are just looking for gains in downforce and reductions in Cd. Is that a poor place to start? What other kinds of information should we be seeking out at this early stage? What are the next subjects that should be studied to build a safe, fast vehicle (roll, yaw, pitch, etc...)?

Offline bubruins

  • Full Member
  • ***
  • Posts: 106
Re: AutoDesk Flow Design / CFD Simulation
« Reply #8 on: December 02, 2014, 11:45:32 AM »
The other interesting part about using 1:18 scale is that we can build models by hand easily (wood) or we can print models (plastic) to put in the 1:18 tunnel. Of course, all models are sanded and painted the same.

Offline WOODY@DDLLC

  • Hero Member
  • *****
  • Posts: 1803
  • ECTA made it to AR-Kansas!
    • Design Dreams, LLC
Re: AutoDesk Flow Design / CFD Simulation
« Reply #9 on: December 02, 2014, 10:23:42 PM »
Sorry I did not check the video links - my bad!  :-( This one is very similar but not quite as sophisticated: http://www.f1technical.net/forum/viewtopic.php?f=14&t=14495
You have to have the same test conditions [air speed & density] or you cannot compare results from one test to the next! Otherwise you are just blowing air! You will need to look at L/D not just one or the other. The two most common comments heard in the wind tunnel are, "Look at that!" and "I have no idea!"  :-o Have fun with it but don't take your results too seriously!  :cheers:
All models are wrong, but some are useful! G.E. Box (1967) www.designdreams.biz

Offline bubruins

  • Full Member
  • ***
  • Posts: 106
Re: AutoDesk Flow Design / CFD Simulation
« Reply #10 on: February 13, 2015, 10:00:12 AM »
I want to update this thread with some progress in the hopes that it will solicit some feedback and will inform others of what is possible using relatively easy to use, easy to get software.

I built several models of a car body in Google Sketchup 8. This was possible using the standard tool kit in Sketchup 8 and an plug-in called soapscreen bubble (used for creating a complex "skin" based on lines). All the information I used to build the model is based on YouTube tutorials and lots of time testing and building in Google Sketchup. I personally like Google Sketchup because it's free and there is a ton of information out there on how to use it.

Once we had a detailed model that looked good and other team members liked, we imported the file into AutoCAD 2015. Then we saved the file as a 2010 .DWG file. The conversion from a .SKP file to .DWG looked great! From there, we opened the file using Autodesk Flow Design and began running scenarios. We used a freshly built desktop that has 8gb solid state ram and a 4.4ghz processor, but it still took an extremely long amount of time to get 3D results processed. It appeared that the higher the wind speed the longer it took to get the results and a realistic average CD. 2D results are instantaneous, but only show the air over one plane of the car at a time.

At this point we have not done nearly enough testing to decide on a shape, but have proven that it is possible to build a model in Sketchup 8 and run it in Flow Design. I'll admit, there were several "look at that" and "I have no idea" comments in the first session that we ran it, but we can only learn more from here. Thus far, it has been very fun to build and play with. Now comes the hard work and trying to interpret our results in the most scientific way that we can.

Next steps for us are to test standard shapes (cubes, spheres, airfoils, etc...) and check the Flow Design CD values against industry standards. This will mostly be a sanity check, but if we see a trend we may conduct a correction value analysis. After that we will construct 2 more detailed car models. With the total of 3 detailed models we will make some changes to the shape if we see any obvious areas for improvement. Simultaneously the guys that are contributing the $ to this project are ironing out the details for the final drivetrain dimensions. Then it will be time to conduct analyses with the 3 detailed models we have to include push bars, parachute packs, etc.
« Last Edit: February 13, 2015, 10:10:34 AM by bubruins »

Offline Graham

  • Jr. Member
  • **
  • Posts: 35
    • Fluids Laboratory for Interdisciplinary Projects - FLIP
Re: AutoDesk Flow Design / CFD Simulation
« Reply #11 on: February 14, 2015, 07:14:36 PM »
I tried introducing a CFD component to a 2nd year fluid mechanics course using one of these CAD-based flow packages (in Solidworks). Was pretty dreadful, we actually found it worse than throwing the students in a deeper end with ANSYS in their 3rd year, as it led to a lot of false expectations and assumptions.

The packages you're describing are probably pretty good for assessing a few different configurations and picking a preferable one to pursue - comparisons between one dubious model and another can be quite useful if you use the same settings. However I wouldn't use it for detailed work or any design where you're on any kind of limit or margin - not reliable.

With a full-feature package, yes they have a steeper learning curve, much higher cost, etc. but the user and their experience/expertise has a huge influence over the result. With the modules in CAD packages (and I won't pretend I've tried others, so maybe I'm off the mark), you are generally limited by the program making lots of choices for you and serving up a nice result from a black box process - e.g. I could not get a package like that to give me an accurate result and I've been doing CFD for a living for 10 years.

So I would say, early design tool (or if it's the only available tool!) - great
Detailed design tool or replacement for physical testing, particularly when trying to assess tricky balance/stability problems - not great, proceed with caution
Graham Doig
Fluids Laboratory for Interdisciplinary Projects - www.thinkflip.net
Aerospace Engineering Department
California Polytechnic State University

Offline bubruins

  • Full Member
  • ***
  • Posts: 106
Re: AutoDesk Flow Design / CFD Simulation
« Reply #12 on: February 15, 2015, 07:46:56 PM »
Thanks  for the feedback! Physical testing in a 1:1 scale tunnel is on the list to do before building the final body design. The struggle I'm having is just how much detail we can test for in the tunnel. For example, I'm trying to figure out ways to allow air under the car in different settings.

Thinking about this process has helped me to understand why so many of the cars look so similar. Conservatism seems to pay off here because a radical idea appears to be difficult to efficiently make minor adjustments to in 1:1 testing.