Accounting for Uncertainty using Monte Carlo and Response Surface Methods
C-FER Technologies’ FEA simulations account for uncertainty in the model input parameters. We use Monte Carlo simulation techniques to consider a broad range of possible input parameters. These analyses require running the FEA simulation hundreds of times with different parameters to determine how the component behaviour changes. These results are used to estimate the likelihood, or probability, that a specific condition will occur.
If multiple uncertainties need to be considered in an analysis, the number of possible combinations of parameters could require an unmanageable number simulations. In these cases, a response surface methodology can be used to select critical parameter combinations. This can substantially reduce the number of FEA simulations that need to be performed to fully describe the effects of input variability on component behavior. The results of these simulations are expressed as a simplified, multi-variate relationship using a radial basis function, which uses the Euclidean distance between points to perform the interpolation.
The uncertainties that can be assessed with these techniques include:
- Manufacturing tolerances;
- Material properties;
- Operating loads, pressures and temperatures; and
- Wear, erosion and material degradation.