Subject: Re: contact question Author: DENNIS NILSSON dennis.nilsson.656@ Date: 2001-05-12 08:43:00Lunyu, If you had followed the XANSYS netiquette you'd probably got your answers already!!!
If comparing the two methods by accuracy and efficiency the main differences are: For accuracy, the kinematic contact condition is always exactly enforced in the Lagrange multiplier method, thus making this method always the most accurate. The penalty method always introduces penetrations, which can be controlled by changing the penalty parameter. The accuracy of the penalty method therefor depends on the choice of the penalty parameter. Too small penalty parameter gives large penetrations, which may give unacceptable errors. Too big penalty parameter may produce numerical problems in the solution process.
For efficiency, the Lagrange multiplier method introduces new unknowns (the Lagrange multipliers), thus it always increases the dimension of the equation system to be solved, which may increase the time for solving the problem. The penalty method does not introduce new unknowns and therefor does not expand the equation system to be solved. However, extra contributions are added to the global stiffness matrix and these changes the profile of the stiffness matrix.
You have to choose the right method carefully depending of the problem you are running. If you are running and implicit solution you can choose both the Lagrange multiplier and the Penalty method. But for an explicit solution only the Penalty method is avaliable. However, this is not really true, the Lagrange method can be used in an explicit solution toghether with a time intregration method to obtain an explicit-implicit method.
It think you have to sit down and read some contact matrerial before you start running a contact analysis to make sure you understand the difference between the Lagrange multiplier method and the Penalty method. Notice that there are two more methods, the Augmented and the Pertubed method. Best regards
ANKER - ZEMER Design AB Dennis Nilsson MSc Mech.Eng, Application Engineer CAE dennis.nilsson@a...
----- Original Message -----
> I am also working on contact problem. But I am not familiar with > it. > Is there anyone who knows the penalty function and lagrange > multiplier?