Understanding uncertainty is an important part of any scientific measurement process and the ability to evaluate and understand uncertainty is a requirement of the International Standards for quality control. The basic uncertainties relating to the measurement of airborne sound insulation in the field can be assessed using the methods in BS5725. However, identifying the components that contribute to the total variability is beyond the scope of the standard and more detailed information requires a more advanced approach. Recent developments in the “Guide to the expression of uncertainty in measurement” (GUM) suggest an approach can be used where identification of the input variables and their likely contribution will result in a solution that can be modelled providing enough information is available. However, recent research on uncertainty in sound insulation using GUM has identified problems involving the correlation between frequency bands, which leads to an overestimate of uncertainty. An empirical approach is used in this thesis, which incorporates advanced analysis of variance (ANOVA) and a specific model called Gauge Repeatability and Reproducibility (GRR). It enables the components of variance in the measurement system to be partitioned and provides an estimate of their contribution. In addition, ANOVA highlights any interaction between factors. In the GRR, carried out on a lightweight timber floor and a heavyweight concrete floor, significant interaction was detected between the operator and part. Good agreement is obtained in the repeatability and reproducibility calculated for each construction and the samples are combined with measurements of test elements that provide a wider range of sound insulation performance. The uncertainty associated with the instrumentation, operator, interaction and part are calculated in each case. It is shown that the interaction component is important and should be contained in any approach evaluating uncertainty. Further evidence reveals that the total uncertainty in the measurement process is dependent on the construction being measured.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:592871 |
| Date | January 2013 |
| Creators | Whitfield, William |
| Contributors | Gibbs, Barry G.; Seiffert, Gary |
| Publisher | University of Liverpool |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://livrepository.liverpool.ac.uk/14517/ |
Page generated in 0.002 seconds