Free computational fluid dynamics software

[Winston]

I'm not a student or mentor, so I can't get the required Autodesk app for free, unfortunately. The latest Apogee newsletter about the CFD software:

https://www.apogeerockets.com/education/downloads/Newsletter438.pdf

 

[Cabernut]

If I was going for 50k feet or something then it might be worth it to cough up the $210 for a 1-year license. I wonder if you can get a discounted license for an older version like you can with NewTek Lightwave 3D?

 

[Winston]

If I was going for 50k feet or something then it might be worth it to cough up the $210 for a 1-year license. I wonder if you can get a discounted license for an older version like you can with NewTek Lightwave 3D?

There are some strange aerodynamic effects which I surmised an intuitive answer to in an experiment I described in another thread that I'd like to play with to find the actual reason, but it's not so essential that I'd be willing to pay anything to do it. As pointed out in the newsletter article, Tim Van Milligan is doing this for a TARC group he's mentoring, so he gets to use Autodesk for free.

 

[jmattingly13]

CFD is only useful if you know how to set up problems correctly and interpret results correctly. The solution changes quite dramatically depending on boundary conditions, grid generation (and to some extent domain size), and models used (viscous vs inviscid, which turbulence model, etc.). And then you have to figure out what to do with that information. If you're just going for the colorful pictures, it's probably not worth the investment. Even big aerospace companies fall into the trap of doing CFD for the sake of doing CFD. Just my two cents on the matter. (I for one am a big proponent of CFD.)

 

[aerostadt]

CFD is only useful if you know how to set up problems correctly and interpret results correctly. The solution changes quite dramatically depending on boundary conditions, grid generation (and to some extent domain size), and models used (viscous vs inviscid, which turbulence model, etc.). And then you have to figure out what to do with that information. If you're just going for the colorful pictures, it's probably not worth the investment. Even big aerospace companies fall into the trap of doing CFD for the sake of doing CFD. Just my two cents on the matter. (I for one am a big proponent of CFD.)

Well said Joe. When I retired 6 years ago I was running CFD using Fluent. Tim talks about importing the 3D CAD file, but does not mention how the CFD code sets up the grid. The grid can be set up with hexagons, bricks, etc. and is fine towards the boundary to capture the BL and gets coarse further away. When we ran Fluent, we set up the grid ourselves using Fluent or some other code. Just setting up the grid can be an art in itself. In general for setting up a problem if the vehicle is moving parallel to the flow, one may be able to find planes of symmetry to reduce the size of the 3-D problem. If the problem is truly axi-symmetric, then the problem can be run 2-D axi-symmetric. It is interesting to note that Tim said that he could take his model and put in an angle of attack. This immediately makes the problem into 3-D, which means that the code is definitely handling a 3-D problem, which is impressive both for gridding and CFD computation.

In all likelihood our model rocket flow problems are turbulent, although, technically there will be a short transition from laminar to turbulent flow at the front of the problem. Tim does not mention which turbulent model is used nor if the user in the trial version has a choice. There are at least a half-dozen popular turbulent models out there and probably many dozen altogether (heck, I even have a turbulent model that I have worked on for 40 years). The trial CFD version may not be telling what it is doing for turbulence modelling as well what it is doing for the gridding. (One could argue should we include the rocket exhaust in the model, but we are getting carried away.)

I am a TARC mentor, but I don't feel I have time to look into this, but it would be fun.

 

[Buckeye]

Well said Joe. When I retired 6 years ago I was running CFD using Fluent. Tim talks about importing the 3D CAD file, but does not mention how the CFD code sets up the grid. The grid can be set up with hexagons, bricks, etc. and is fine towards the boundary to capture the BL and gets coarse further away. When we ran Fluent, we set up the grid ourselves using Fluent or some other code. Just setting up the grid can be an art in itself. In general for setting up a problem if the vehicle is moving parallel to the flow, one may be able to find planes of symmetry to reduce the size of the 3-D problem. If the problem is truly axi-symmetric, then the problem can be run 2-D axi-symmetric. It is interesting to note that Tim said that he could take his model and put in an angle of attack. This immediately makes the problem into 3-D, which means that the code is definitely handling a 3-D problem, which is impressive both for gridding and CFD computation.

In all likelihood our model rocket flow problems are turbulent, although, technically there will be a short transition from laminar to turbulent flow at the front of the problem. Tim does not mention which turbulent model is used nor if the user in the trial version has a choice. There are at least a half-dozen popular turbulent models out there and probably many dozen altogether (heck, I even have a turbulent model that I have worked on for 40 years). The trial CFD version may not be telling what it is doing for turbulence modelling as well what it is doing for the gridding. (One could argue should we include the rocket exhaust in the model, but we are getting carried away.)

I am a TARC mentor, but I don't feel I have time to look into this, but it would be fun.

Well said. CFD is an art, and there are lots of parameters to deal with. Bachelors students in engineering are not fully equipped for CFD, let alone high school TARC students. That is why we hire mainly Masters grads. (BTW, Fluent meshing sucks for complex geometries (like automotive), but may be OK for aero shapes. There are more efficient meshers out there).

CAD programs like Autodesk like to include some simple CAE solvers, but they are limited and very black box in nature. When Tim said the simulations ran all of "5-10 min" on a PC, I knew this was a highly simplified CFD model. (My CFD jobs run for days on 100's of cores!) Also, I am not so sure about his interpretation of pressure and velocity contours, but whatever. The main benefit of CFD for hobby rocketry is better CD and CP prediction, neither of which you get from Autodesk, according to the article.

The only truly "free" CFD software is OpenFoam, and it is very capable.

 

[Cabernut]

I've tried simFlow 3.0, it uses OpenFOAM. Way over my head, but cool to tinker with.

Here's an example of a tutorial that I tried to follow, but not having an understanding of the underlying math, I didn't get expected results.

[youtube]wASYUxQ9LMc[/youtube]

 

[Winston]

CFD is only useful if you know how to set up problems correctly and interpret results correctly.

You shouldn't make blanket statements like that without knowing what someone needs to do. All I need to do is LOOK at the flow lines around the fins of a six-finned rocket at various angles of attack. Related thread:

https://www.rocketryforum.com/showt...R-1-2-altitude-of-Rocksim&p=701621#post701621

 

[Nytrunner]

Well said. CFD is an art, and there are lots of parameters to deal with. Bachelors students in engineering are not fully equipped for CFD, let alone high school TARC students. That is why we hire mainly Masters grads. (BTW, Fluent meshing sucks for complex geometries (like automotive), but may be OK for aero shapes. There are more efficient meshers out there).

CAD programs like Autodesk like to include some simple CAE solvers, but they are limited and very black box in nature. When Tim said the simulations ran all of "5-10 min" on a PC, I knew this was a highly simplified CFD model. (My CFD jobs run for days on 100's of cores!) Also, I am not so sure about his interpretation of pressure and velocity contours, but whatever. The main benefit of CFD for hobby rocketry is better CD and CP prediction, neither of which you get from Autodesk, according to the article.

The only truly "free" CFD software is OpenFoam, and it is very capable.

I think I've experience what you mean by "Black Box" when it comes to CAD flow tools. I use SolidWorks Flow Simulator for my CFD dabbling and it doesn't have a lot of the parameter adjustment capability as the OpenFoam based simFLow, but allows me to get some better resolution (Hmmmm.... better resolution of bad modeling may just result in shinier garbage. This requires more research.....). I'd be really stoked to try their legitimate CFD package, but I think that costs a couple grand per license.

5-10 minutes.....sure. A rough calculation of goals that short told me a 4"x53" Patriot's CP was 13" behind the nose tip......

I've tried simFlow 3.0, it uses OpenFOAM. Way over my head, but cool to tinker with.

My brief experimentation with simFLow was decidedly unromantic. I was trying to get some data for my old school's new IREC rocket, but the free version limited my computation space to 100,000 nodes. My uncle (longtime aero/propulsion CFD guy) saw that and pretty much said I was wasting time with such a rough resolution. From my experiments with SldWks FlowSim, I've been able to get closer to grid independence than I ever could with simFlow.

All I need to do is LOOK at the flow lines around the fins of a six-finned rocket at various angles of attack.

What sort of characteristics do you look for? Sharp deviations? Redirection trends?

If you're just going for the colorful pictures, it's probably not worth the investment. Even big aerospace companies fall into the trap of doing CFD for the sake of doing CFD.

My club's president is in the aero business and always refers to CFD as "Colors For Dollars" for that very reason.

 

[Winston]

All I need to do is LOOK at the flow lines around the fins of a six-finned rocket at various angles of attack.

What sort of characteristics do you look for? Sharp deviations? Redirection trends?

Anything that might explain the wobble I've seen near max-v at still very subsonic velocities for a rocket that should be highly stable, perhaps some kind of turbulent interaction between the unusual number of fins that decreases their effectiveness.

 

[Buckeye]

(Hmmmm.... better resolution of bad modeling may just result in shinier garbage.

+1

My club's president is in the aero business and always refers to CFD as "Colors For Dollars" for that very reason.

We like to say "Colors for Directors."

I get a lot of aero insight from CFD that is impossible to see or understand in wind tunnel tests. Complex geometry and interactions can only be efficiently visualized in CFD. Our lab at work has a force balance to measure forces and moments, but that is it. No visualization or monitoring equipment to speak of. So, when it spits out a bad Cd number, there is nothing else to understand "why." That is like your doctor declaring that "you're sick" without even taking your temperature.

Hook up a couple hundred batch CFD simulations into an optimization routine and you got yourself one hell of a tool that leaves hardware tests in the dust.