Trebuchet simulator questions.

[jim frank]

I'm a total newb to Blender and Bullet, so forgive any dumb questions.

I want to model a floating arm trebuchet. It has fairly simple elements; a weight that falls straight down, a lever hinged to the weight, and an axle in the middle of the lever that translates along a horizontal axis as the weight falls.

What I've not found in the few searches I've done so far is a mention of how to translate the rotation point of a long thin object.

Modelling the sling and projectile is a project for a far later date.

My ultimate goal is to allow the user to modify most of the variables; the weight of the projectile and the counterweight, the length of fall, the length of the beam, sling, and the location of the axle on the beam.

Any tips, hints, or pointers to tutorials would be greatly appreciated.

What a cool program this is!

Jim


[edit] Right after I posted this, I Googled a page with pretty much exactly what I was looking for. Oh well, thanks for any replies anyway.

[jim frank]

Well. after investigating the model I found, I realized that it doesn't really use the physics engine at all. It just maps a path for the arm of a floating arm trebuchet.

That's not useful for what I want to do.

So I guess I'll just keep tweaking and learning.

Thanks.

Jim

[jim frank]

OK, I've been tweaking and playing, and I have many more questions than answers.

Is there a detailed explanation on how the "Rigid Body Joint" constraints actually work? They seem to act as a variety of either rigid rods, rubber bands, or bizarre hyperkinetic links that accelerate the constrained object into a whirling helicopterlike mess.

I just want to hinge a long narrow beam at one end to a falling mass, and lock the center of the beam so it can rotate but not translate.

The manual really sucks. I quote:

Description


This constraint is mainly useful when using the game engine.
Options
Joint Types
foo
toObject
foo
ShowPivot
foo
Pivot X, Y, Z
foo
Ax X, Y, Z
foo
Example

Not exactly illuminating.


I looked at a semi promising ragdoll tutorial, but it refers to things that don't work on my version of Blender (2.48a). When I try to approximate, I get rigid links instead of hinges.

Where can I go from here?

[jim frank]

OK, I'm finally making some progress.

Here's the model so far. I still need to figure out how to constrain the beam from side to side. On my real life model, the weight buckets serve that function. Once I get the lateral motion stabilized, I need to come up with a sling and bucket of some sort. I was thinking a mesh sphere with half removed might work for the bucket, and I've made a compound pendulum simulation that might work for a rope if I extend the number of nodes out a ways. Comments and hints are welcomed.

The model's parts still needs to be re-labeled for clarity.

http://www.dolcevoce.com/trebuchet.blend

[jim frank]

It's a sharp learning curve for this program. I finally have a workable trebuchet model, with projectile and target. The sling and basket are rough but they work.

Next goal-textures, colors, etc.

I've given up on user interface adjustments of parameters. Until I have time to learn Python, that is, or until some more detailed documentation arises.

It's been fun so far.

http://www.dolcevoce.com/trebuchet2.blend

[movrev]

Wow, cool stuff. Do you have any idea why after you run the simulation a few times the ball trajectory changes, though? Keep it up.

[jim frank]

It changes shot-to-shot because the simulation is not very exact. The hinges are loose and springy, the baskets and other collision surfaces are semi-permeable, and it's like making a model out of celery with rubberbands holding it together.

But, it does agree with my observed results well enough that I can make some general observations about the performance of the floating arm trebuchet.

For instance, on the average, making the 'sling' reach from the beam end just to the roller on the beam gives the best range. Also, a ratio of about two to one for length of beam from roller to tip vs. roller to falling weight, gives a broad peak in performance.

I'm working on an Excel spreadsheet to do some number crunching for the design, to see if I can get a little more rigor.

[jim frank]

Well, Excel is a PITA as far as any real calculation goes, so I switched to VB6. I can calculate the beam tip path and velocity so far, and now it's time to start working with the actual forces.

[lastcoolnameleft]

Did you ever find a solution which worked for you? I'm looking to do the exact same thing. We build an 8' floating arm last year for a contest, but looking to go with a more precise 12' one this year.

Also, the links at http://www.dolcevoce.com/trebuchet.blend don't work.

Any help would be greatly appreciated.

Cheers,
-Tommy.