Rolling process variation estimation using a Monte-Carlo method

No technical process is totally certain, but subjected to uncertainties. They may originate in the process itself or in the input materials and determine the precision of the product. Two questions are here especially of interest: 1) How do variations in the input workpiece evolve within the process? 2) Which process steps are crucial to influence this behavior? Answers to these questions can be obtained by analyzing production data or by numerical methods. The usage of Monte-Carlo-methods for estimation of variations and tolerances is a well proven approach in some fields, but was first applied by the authors to rolling processes. The inputs are all varied at once by drawing random samples from given distributions, so cross-dependencies are included in the analysis. The method has the favor of general applicability, i.e. the simulation procedure can be regarded as black box. So the method is generally agnostic to the used simulation core, but needs a large number of simulation evaluations, so fast simulation models are favorable.