Jump to content

  • Log In with Google      Sign In   
  • Create Account

Resolving contact force to prevent penetration and LCP


Old topic!
Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.

  • You cannot reply to this topic
No replies to this topic

#1 Mr_Nick   Members   -  Reputation: 242

Like
0Likes
Like

Posted 02 August 2012 - 10:48 PM

Hello,
Im working through "Dynamic Simulations Of Multibody Systems" by Murilo Coutinho.
Ive got a ball rigid body (B1) and a plane. I've been trying to setup the LCP problem to calculate the force require to prevent pentration.
The problem is

a = AF + b
a >= 0
F >= 0
aF = 0

where
a = relative acceletaion at the contact point (to solve)
A = a constant (roughly the inverse mass at the contact point)
F = relative force at the contact point (to solve)
b = current relative acceleration at contact point + relative velocity x contactFrame time derivate.

The ball has a gravity force of -10 units, and mass = 1 so -10 acceleration too.
When there is no rotation, all works fine. But when there is rotation, the contact point on the ball gets a centrepetal acceleration so its accelertaion becomes -9.

So it calculates a force of +9 to apply at the contact. But the body still has an accn of -10, so applying this force won't prevent penetration. I could just calculate the force required to prevent penetraion but that would violate the aF = 0 constraint.

I've been googling all over the place the past 2 days and im nearly bald from tearing my hair out so Id be really really grateful for any help.
I know someone posted a similate comment (Bastian1978) but it doesnt help me maintain the constraints which I need so I can extend the system to sovling multiple simultaneous contacts.

Formule are:

b = B1.netForce/B1.mass + (B1.inertia^-1 * B1.Torque + B1.angualMomentum X B1.W) X r1
+ B1.W X (B1.W X r1);
+ (B1.V + B1.W X r1) * (coordFrame time derivate)

A = (I * B1.invMass) - (x1 * B1.inertia^-1 * x1);

where:
r1 = contactPoint - B1.centre of mass
x1 = r1.skewSymmetric
I = identity matrix

The final term of 'b' is zero as the contactFrame remains stationary.

Sponsor:



Old topic!
Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.



PARTNERS