Uncertainty estimation and Monte Carlo simulation method

Abstract It has been reported that the Monte Carlo Method has many advantages over conventional methods in the estimation of uncertainty, especially that of complex measurement systems' outputs. The method, superficially, is relatively simple to implement, and is slowly gaining industrial acceptance. Unfortunately, very little has been published on how the method works. To those who are uninitiated, this powerful approach remains a ‘black art’. This paper demonstrates that the Monte Carlo simulation method is fully compatible with the conventional uncertainty estimation methods for linear systems and systems that have small uncertainties. Monte Carlo simulation has the ability to take account of partial correlated measurement input uncertainties. It also examines the uncertainties of the results of some basic manipulations e.g. addition, multiplication and division, of two input measured variables which may or may not be correlated. For correlated input measurements, the probability distribution of the result could be biased or skewed. These properties cannot be revealed using conventional methods.