� Sony DRM gone astray | Main | Options - Hedging your bets �

November 17, 2005

Making of a trebuchet

Lately, most of my time has been consumed by coursework. As the semester winds down, I'm beginning to see the light at the end of the tunnel. One of the more interesting courses this semester is Systems Engineering. Our final team project is to build a trebuchet. It was a perfect sedgeway from my earlier post. The idea is to use system engineering tools and predictive analysis software (ie. FEA, Atreb V2.2, Minitab) to develop a robust trebuchet to satisfy customer requirements.

The course was divided into two phases:

Part I. Planning and predictive analysis
Part II.Validation, Fabrication and Testing.

Both Part I and Part II, make up what is called the System Engineering "V".
Left side is strictly design for robustness and prediction. The right side is the hardware validation and testing.
We were not allowed to build/test the treb until Part II.

The commercial software package Atreb V2.2, was tremendous, as it helped us understand the physics of a trebuchet. Although, our group is comprised of mechanical and electrical engineers. Nobody knew how the contraption worked, and I certainly wasn't born in the middle-ages. The treb operation was based upon first principles, but it clearly was non-trivial.

We used Minitab, a statistical tool, as a means to develop a sensitivity analysis, which would help validate our DOE.
We could then better understand the variability or the noise factors inherent to our design (ie. main arm pin friction, counterweight mass, main arm length).
After running Minitab, we studied the main-effects and then used this information to begin modeling the treb in CatiaV5.
Once we had a general idea how to model the package, we set out to improve robustness.

The Hypermesh FEA package was used to develop a contour plot to measure the fatigue stresses across the critical surface regions of the treb main arm. Constraining the main arm in space and applying a 50N force in -Z direction, helped us simulate the worst case force that our 30gram, lead shot filled projectile could induce on our treb arm. Yeah, I know the King's Men and the Roman Empire didn't have access to these tools, but they weren't trying to get an 'A' in graduate engineering course either. I suppose this is where we seize the advantage ;)

Each team built and exchanged CAD models and engineering drawings, to that the next team could fabricate their respective designs. To appropriately, satisfy the customer requirements (read the King), our device must consistently fire a projectile into an array of common party cups, nailed down to some plywood. The better the accuracy, the more points you score. There will probably be some environmental hazards (ie wind, rain, etc). If it's anything like our earlier battle-bot/stack block competition, we should be in for a real treat.

Our class has been divided into two teams and each team was responsible for building the others design. That is, we'd spent all our waking hours perfecting an excellent design only to have the other team build it.
Actually, this practice is very common in industry. One team does the strategy, the tosses it over the wall to the next team to execute the vision. We probably spent 80 man-hours collecting repeatability data, and resolving the design quirks of our opponents design. All modifications must be documented in a final paper.

If all goes well tonight, I will grab a few short clips of the teams in action. I will also post pics of the two competing trebs, and the targets. Time for bed..

Posted by AG at November 17, 2005 12:48 AM