ISO 9614-X standards enable to determine the sound power level from sound intensity measurements, providing that a physically stationary measurement surface can be defined and that the sound emitted by the source and by other significant extraneous sources are stationary in time. The standards provide the estimated highest values of the standard deviation of reproducibility for each one-third octave band (Type B uncertainty). The "true value" of the overall A-weighted sound power level is expected (at 95% confidence level) to be in the range ± 3 dB or ± 8 dB for the "Engineering grade" or "Survey grade" accuracy, respectively. Such wide ranges, even though acceptable for a standard that can be applied to any kind of source, are often meaningless when the result has to be used for validating numerical simulation or even to compare similar products. If a specific source is under investigation, for instance to detect the performance of noise reduction actions, repeated measurements should be preferred in order to determine Type A uncertainty. This paper presents the results of this approach applied to an external gear pump, showing the reduction of the uncertainty interval width as the number of repeated measurements increases. Acoustic measurements have been performed using the indication given in ISO 9614 part 2 (scanning method), which is probably the most applied methodology due to its simplicity. Results showed that 3 measures seem to be sufficient to have an expanded uncertainty value of repeatability lower than that indicated in ISO 9614-2. A greater number of repetitions would not significantly improve the uncertainty value.
