Home  Contact Us  Search  Site Map  Downloads  About Unipass

Sample Set MPP in Monte Carlo Simulation

MPP was defined as the point with highest probability density in the failure domain. For the U-based MPP, it has the following properties:

1. It is the shortest distance point in the failure domain in the standard normal space. That is, the reliability index should be minimized.

2. It should be on the limit state surface, i.e., the g value should be zero. That is, the value of g(x) –g_init should be zero (where g_init is the initial value of g function defined in the Function Definition Window..

3. The gradient at this point should pass through the origin in the standard normal space. That is  , where  is the vector of directional cosine at the MPP u.

MPP can be effectively identified by the optimization algorithms. In the Monte Carlo simulation, we can define the Sample Set MPP as the highest density point among the sample points in the failure domain. If the sample set has enough sample points, this sample set MPP should be close to the U-based MPP.

At the end of simulation, a derivative limit state function is calculated at the sample set MPP to obtain the direction cosine . Then the associated first sensitivity measurements can be computed. These sensitivities are very useful in for upgrading the problem.  

Last Updated 02/08/10

Send mail to info@predictionprobe.com with questions or comments about this web site.
Copyright © 2001-2004 PredictionProbe, Inc.