I went to Virginia Tech for engineering after high school and after nearly failing out twice in my freshman year (I kid you not) I pulled it together, busted my rear, and finished school in four years. My senior year was dominated by my senior design project, Formula SAE. I can't be certain of this, but from all of the senior design projects I've seen over the years, nothing, and I mean NOTHING comes anywhere close the FSAE when it comes to preparing design engineers for the real world. I worked with a team of about 25 people to design, build, test, and race a small formula style car in less than 9 months. Designing something on your own is hard enough, but collaborating with 24+ other people, dealing with constant logistical issues and techincal problems, and coming together in the end to succeed at your goal is amazing. Especially for a young engineer. It was like a first job for me, and it taught me more about real world engineering than sitting in a classroom ever could.
My design area for FSAE was the engine intake manifold. I had some pretty lofty design goals before the year started...I remember drawing up all sorts of variable geometry intake manifolds similar to the designs used by BMW in their V8s at the time. Given the chance to do it all over again, I would have started simple and then simplified from there. The more complicated a design becomes, the more problems it brings. Add in the fact that 24+ other young "not quite engineers" were dealing with the same things, and the beauty of simplicity comes shining through.
I used CFD for my intake design, and it was anything but pretty. All universities have Fluent at their disposal, and after sitting down with a grad student over the course of several days, I learned how to open the program. If I told you Fluent was a bad tool, I'd be lying. I tend to overexaggerate things at times, but Fluent was really horrible for design engineering. The process went like this:
1. Create CAD model.
2. Export CAD model.
3. Import CAD model in Gambit.
4. Apply loads/mesh sizes.
5. Be frustrated.
6. Export mesh file.
7. Import mesh file into Fluent.
8. Hope everything is set up correctly.
9. Start simulation.
10. Wait for simulation to complete while watching numbers flash on the screen.
11. Attempt to interact with results after completion of the run.
It was borderline terrible. How can such a powerful program be so horrible to use? Maybe PhD level guys and gals enjoy pain and torture? Maybe they feel that there is no point in doing something if it isn't REALLY hard to do?
Fluent has it's place, it really does. It's just not on the desk of a mechanical engineer.
At some point near the end of the year, and aero engineering student approached me and showed me CFdesign. I was skeptical. The things he showed me were an order of magnitude more complicated than the stuff I had tested in Fluent, it took him 10 minutes to set it up, and he had results to look at right after hitting the go button. I could not believe it. This was well after my design phase when I saw the program for the first time. I was super anxious to throw my intake at CFdesign and see what it had to say about it.
This is where I apoligize to my fellow FSAE team members. I think we could have made a lot more power. CFdesign showed a few large issues with my design that Fluent didn't. I should say, CFdesign showed a few large issues with my design that I was unable to get Fluent to show me. As it turns out, I had used the wrong solvers, the wrong model setup, the wrong boundary conditions, the wrong turbulence mumbo jumbo...basically, everything I had done with Fluent was wrong. Now, this isn't to say that CFdesign couldn't have replicated the Fluent results, it's just that the combination of the default settings, super straightforward user interface, and general interactivity of CFdesign allowed me to set it all up correctly the first time. I was sold.
As luck would have it, my FSAE advisor, Dr. Kasarda, knew the small group of people who had created CFdesign. In fact, she went to school with them and was a close personal friend. After contacting the people at Blue Ridge, the makers of CFdesign, I drove up to Charlottesville from Blacksburg for an interview. It's the only job interview I've ever had.
My design area for FSAE was the engine intake manifold. I had some pretty lofty design goals before the year started...I remember drawing up all sorts of variable geometry intake manifolds similar to the designs used by BMW in their V8s at the time. Given the chance to do it all over again, I would have started simple and then simplified from there. The more complicated a design becomes, the more problems it brings. Add in the fact that 24+ other young "not quite engineers" were dealing with the same things, and the beauty of simplicity comes shining through.
I used CFD for my intake design, and it was anything but pretty. All universities have Fluent at their disposal, and after sitting down with a grad student over the course of several days, I learned how to open the program. If I told you Fluent was a bad tool, I'd be lying. I tend to overexaggerate things at times, but Fluent was really horrible for design engineering. The process went like this:
1. Create CAD model.
2. Export CAD model.
3. Import CAD model in Gambit.
4. Apply loads/mesh sizes.
5. Be frustrated.
6. Export mesh file.
7. Import mesh file into Fluent.
8. Hope everything is set up correctly.
9. Start simulation.
10. Wait for simulation to complete while watching numbers flash on the screen.
11. Attempt to interact with results after completion of the run.
It was borderline terrible. How can such a powerful program be so horrible to use? Maybe PhD level guys and gals enjoy pain and torture? Maybe they feel that there is no point in doing something if it isn't REALLY hard to do?
Fluent has it's place, it really does. It's just not on the desk of a mechanical engineer.
At some point near the end of the year, and aero engineering student approached me and showed me CFdesign. I was skeptical. The things he showed me were an order of magnitude more complicated than the stuff I had tested in Fluent, it took him 10 minutes to set it up, and he had results to look at right after hitting the go button. I could not believe it. This was well after my design phase when I saw the program for the first time. I was super anxious to throw my intake at CFdesign and see what it had to say about it.
This is where I apoligize to my fellow FSAE team members. I think we could have made a lot more power. CFdesign showed a few large issues with my design that Fluent didn't. I should say, CFdesign showed a few large issues with my design that I was unable to get Fluent to show me. As it turns out, I had used the wrong solvers, the wrong model setup, the wrong boundary conditions, the wrong turbulence mumbo jumbo...basically, everything I had done with Fluent was wrong. Now, this isn't to say that CFdesign couldn't have replicated the Fluent results, it's just that the combination of the default settings, super straightforward user interface, and general interactivity of CFdesign allowed me to set it all up correctly the first time. I was sold.
As luck would have it, my FSAE advisor, Dr. Kasarda, knew the small group of people who had created CFdesign. In fact, she went to school with them and was a close personal friend. After contacting the people at Blue Ridge, the makers of CFdesign, I drove up to Charlottesville from Blacksburg for an interview. It's the only job interview I've ever had.
No comments:
Post a Comment