PDA

View Full Version : How to generate random burrs on a surface?



paranormal_motion
08-23-2013, 11:56 AM
Hi,

This is my first post.

As a part of my thesis, I need to visualize burrs ( the protruding surface defects in manufacturing) on a surface. First of all, in deburring application, we remove burrs. I will try to do virtual equivalent of this system.

When the tool (deburring media) touches the burr, it removes it in a rate we call feed rate. During this operation, we may need the volume removed per second in a direction.

My questions are:
1- How can I randomly generate some protruding burrs on or along a surface?
2- Which method should I use, I mean, to draw them in 3D? Volume rendering, I mean marching cubes etc. ?

Due to the fact that I'm a mechanical engineer and have a very limited knowledge on opengl and computer graphics, I figured that it would be more easy if you guide me guys.

Thanks in advance,

kemal.

Carmine
08-23-2013, 01:22 PM
Show us a real picture of burrs and I bet you get a lot of responses.

paranormal_motion
08-23-2013, 01:46 PM
Show us a real picture of burrs and I bet you get a lot of responses.

Thank you Carmine,

Here is a basic burred surface:

(From top, left pictures show burrs, right pictures show deburred surfaces)

Look at the attachment!

Carmine
08-24-2013, 09:44 AM
Those are good pictures. Are you actually going to be working at the level of zoom shown in the pictures? As you get up close the 'burrs' might have to be modeled with polygons. That would be tough to do because they have such random, irregular shapes. If you're further away, it could probably be done with textures. Also, maybe someone who works with shaders could help you out.

paranormal_motion
08-24-2013, 02:36 PM
Those are good pictures. Are you actually going to be working at the level of zoom shown in the pictures? As you get up close the 'burrs' might have to be modeled with polygons. That would be tough to do because they have such random, irregular shapes. If you're further away, it could probably be done with textures. Also, maybe someone who works with shaders could help you out.

Yes, burrs are normally at an order of 10um on and between jet engine blades. I need especially volume information along with 3D representation. However, luckily, burrs can be simplifed as in the picture at the attachment of this post. I don't need to accurately represent 3D structure of burrs. I need an approximate representation and volume information.

First thing that dawned on me is "can I represent these burrs as made up of small boxes (very small)? If the answer to this question is 'yes', I could calculate volume information."

I tried to randomly generate tousands of cubes on a surface mimicking the burrs at structure level, but I failed because of the huge number of cubes and lacking of good data structure to store location of burrs. Also, I don't want to represent them as textures or some empty open surfaces which have noting to do with volume information...

I hope someone has answers for me! :-)

tksuoran
08-24-2013, 03:38 PM
Simulating the actual process that creates burr could possibly create best results, but might be quite a challege. Alternatively, if you could get some sort of surface scans of real instances, you could turn those into 3d grid and then to meshes and and splat them on a surface.

paranormal_motion
08-25-2013, 04:51 AM
Simulating the actual process that creates burr could possibly create best results, but might be quite a challege. Alternatively, if you could get some sort of surface scans of real instances, you could turn those into 3d grid and then to meshes and and splat them on a surface.

Burr formation simulations are basically done by finite element analysis... These methods are quite complicated. I can do it, but using for instance, ANSYS. But the process itself is very very slow, and cannot be handled in real-time.

Indeed, we will use conoscopic holography method to scan the surface. However, for the time being, I'm studying metal cutting principles for deburring operation in virtual environment. For this reason, I need to generate random burrs on a surface. (I started with some random terrain generation methods, I'm working on them now... I will use them to generate burrs instead of random terrain... However, these are generating some meaningless empty surfaces... :-) )

I read some papers which worked with some sort of small balls to represent burrs. When the tool contacts the surface of the burr, it disturbs or removes these balls. These papers were about scratching, though. I may need to do some tweak to make them suitable for deburring operation.

By the way, if I finish it, I will share it with everyone, here in OpenGL forum...

Roaoul
08-31-2013, 01:19 AM
Let me try to translate your physics into something closer to graphics. I'm a physicist, but only a graphics novice, so one of the experienced graphics folks will need to take it from here. This will be for approximation, and not high-fidelity finite-element simulation.

I think the burrs could be simulated by 2-D foliage (like a single, large curved leaf, with cuts in it, as opposed to a more space-filling conical or spherical tree crown). I think you do that with stencils? They should have variable height above the work surface (=width of foliage). They need to be attached to the work, at least until deburring; then body parts can fly. :) The mean width, tilt, curl, width variability, and cut-ness perhaps, should be tunable parameters.

After deburring, the foliage would shrink in the width and variability, and perhaps some texture gets added to the work surface to show abrasion.

He would need to be able to compute the area of the foliage, unless he is able to just use the mean width and cutness parameters.

He would have to assume a scenario-dependent value of the thickness of the foliage (like the thickness of a sheet of paper, or of a leaf), and multiply that by the area to get volume.

Please feel free to ask if any of this is unclear.