Uncertainty Analysis A practical guide

During the DELPHI project a first attempt was made to standard-ise the way uncertainty analysis was carried out. However, some of the documents that were produced contain errors w.r.t. the treatment of systematic errors and could easily lead to a misinterpretation of the underlying theory. Therefore Philips Research and CQM decided to produce a document that deals with this subject in a more precise manner. This document was originally aimed at the PROFIT partners, but it may also be of interest to others and is therefore also available as a Nat.Lab. technical note. The main message of this note is that uncertainty analysis consists of two parts: the analysis of systematic and random errors. Both should be treated in their own way. Identified systematic errors should be quantified as much as possible and then used to correct measured values. Any uncertainties in these corrections are reported together with the other random errors to give a confidence interval for the measured result. A series of examples are given to illustrate the method.