yade-users team mailing list archive

["Feng, Chen" <fchen3@xxxxxxx>]

OK, I have got it, and I will spend time on it very soon, but today and 
tomorrow I will not have access to the FC3 computer in my lab, but I will 
reply email on Sunday.

>===== Original Message From Janek Kozicki <janek_listy@xxxxx> =====
>so I have prepared for you everything you need to code a collision of a
>sphere with a cyllinder. in the attachment you will find all the files.
>you can copy them to latest svn trunk (overwriting existing files), then
>compile and examine CylinderTest filegenerator.
>
>You have three problems to solve:
>
>
>
>1. yade-packages/yade-package-dem/src/PreProcessor/CylinderTest
>
>inside there is a function createCylinder(), apart from changing
>cylinder's position and orientation (which certainly you will do), you
>also should calculate correct mass and inertia of the cylinder.
>Currently there are formulas for sphere, which is of course totally
>wrong.
>
>
>
>
>2. 
yade-packages/yade-package-dem/src/Engine/EngineUnit/InteractingHollowCylinder
2AABB
>
>this function is crucial for correct detection of potential
>interactions. You must write a correct formula that will calculate AABB
>(Axis Aligned Bounding Box) from a hollow cylinder. In the graphical gui
>you can click checkbox 'drawBoundingVolume' to examine the results of this 
function.
>
>inside this file you have examples how it is done for a sphere and for a
>box. Based on that you should write it for a cylinder.
>
>currently it is calculating as if cylider was a sphere which is totally 
wrong.
>
>
>
>
>3. 
yade-packages/yade-package-dem/src/Engine/EngineUnit/InteractingHollowCylinder
2InteractingSphere4SpheresContactGeometry
>
>finally the most important part: calculate SpheresContactGeometry from a
>collision of a sphere with a cylinder. I've written a lenghty
>explamantion inside this file, so you should look there.
>
>
>
>
>
>Additional explanations:
>
>4. when you will load scene.xml generated by CylinderTest, you must
>click 'Fixed' in the 'Time Step' section. That is because the
>timestepper choses timestep=0 seconds. I'm not sure now why exactly he
>is doing that, we will find out later. To temporarily fix the problem
>you can also comment a line
>
>rootBody->engines.push_back(sdecTimeStepper);
>
>in CylinderTest.cpp, but finally that line should be uncommented, and
>this problem should be fixed. Otherwise the simulation will go too slow
>due to too small timestep, or will go wrong due to too big timestep.
>
>
>5. I have enabled 'useSpheresAsGround' because the cylinder will not
>interact with the box. It will just fall through, which is not what we
>want, at all :) You can disable it, and watch the results. If you want a
>collision between cylinder and a box, then you should write this class:
>
>yade-packages/yade-package-dem/src/Engine/EngineUnit/InteractingHollowCylinde
r2InteractingBox4SpheresContactGeometry
>
 ^^^^^^^^^^^^^^
>and appropriately register it inside CylinderTest.cpp, line 380.
>
>
>6. when you run the simulation you will see a cylinder that is
>overlapping with spheres. This is of course caused by the fact, that
>currently cylinder behaves like a sphere ;) This will be fixed when you
>solve problem 2 and problem 3 :)
>
>
>7. if you want a collision between two cylinders then you should write this 
class:
>
>yade-packages/yade-package-dem/src/Engine/EngineUnit/InteractingHollowCylinde
r2InteractingHollowCylinder4SpheresContactGeometry
>
 ^^^^^^^^^^^^^^^^^^^^^^^^^
>
>8. cylinder here is a dynamic object, so it is moving. probably for your
>triaxal test you want cylinder to be a non-moving thing. Or maybe you
>will want to move it with a TranslationEngine (DeuxExMachina). to have a
>non-moving cylinder you just assign isDynamic=false;
>
>9. if for example you will want to modify the radius of a cylinder (to
>compress spheres inside), you will have to write another DeuxExMachina,
>that will modify the radius. it will be similar to TranslationEngine,
>but instead of moving, it will modify the radius.
>
>
>good luck, and I'm waiting for interesting questions!
>
>--
>Janek Kozicki                                                         |