High Resolution Modeling for Evaluation of Measurement Uncertainty

  1. French College of Metrology
  1. Marco Wolf,
  2. Martin Müller,
  3. Dr. Matthias Rösslein and
  4. Prof. Walter Gander Eth

Published Online: 3 FEB 2010

DOI: 10.1002/9780470611371.ch54

Transverse Disciplines in Metrology

Transverse Disciplines in Metrology

How to Cite

French College of Metrology (2009) High Resolution Modeling for Evaluation of Measurement Uncertainty, in Transverse Disciplines in Metrology, ISTE, London, UK. doi: 10.1002/9780470611371.ch54

Author Information

  1. Zürich, Switzerland; Empa St. Gallen, Switzerland

Publication History

  1. Published Online: 3 FEB 2010
  2. Published Print: 1 JAN 2009

ISBN Information

Print ISBN: 9781848210486

Online ISBN: 9780470611371

SEARCH

Keywords:

  • high resolution modeling;
  • gauge block calibration;
  • standard deviation;
  • chemical process;
  • software package

Summary

The determination of the measurement uncertainty is a rather tedious task. The Guide to the Expression of Uncertainty in Measurement introduced an evaluation method for the uncertainty. Usually we choose a model of a perfect world that behaves preferably in a linear way and includes as few dependencies between the influences as possible to keep the effort for the calculation small. Using the Monte Carlo method that will be propagated in the first supplement of the GUM we allow with our simulation software a more detailed view of the uncertainty calculation model. We try to be as close to the real world as possible.

Usually we try to use a simple model to keep the effort for the uncertainty calculation as small as possible. The five or six most important influences are chosen, they are quantified using normal distributions and put into a linearized model equation to evaluate an estimation for the measurement uncertainty. In some cases this maybe an accurate choice to keep time expenses and costs as small as possible. Using existing software packages for the GUM framework simplifies the work once more.

However, we think a more realistic modeling can enhance the quality of measurement results a lot more. Highly detailed, more realistic models can lead definitely to a better inside to the mechanisms of configuration parameters of measurements and therefore help to increase the reliability of results. MUSE supports the user of the software to model the measurement process and automates the simulation runs and analyzing data. Dividing the model equation into different parts that use different sets of variables and parameters allows a very well arranged modeling process. We showed first in an example of GUM and GS1 that MUSE arrives at the same results and allows a logical structuring of the model equation. In a rather small example from chemistry we then explained how our method fits the needs of the user. Different scenarios can be evaluated quite easily and a direct comparison of results using various parameter sets is possible.

It is hard to decide in advance which influences are the most important to a measurement scenario due to the measurement uncertainty. We support the user of our software in finding the most important influences to the measurement, as we make it possible to test different settings with very small effort, for example in using different kinds of loops for repeated simulations with different parameter sets. Whole parts of the measurement can easily be replaced, edited or deleted.

All in all we hope to support and automate the user in enhancing the measurement uncertainty evaluation task with our software package MUSE.