In a ship model resistance test, the towing force measurements usually contain high amplitude and low frequency noise. The length of the towing tank is finite, which induces an uncertainty in the estimated mean value of the towing force. In practical work, a time-window is manually selected from the complete measurement time history to compute the mean value to be used in the further analysis. Due to the combination of high-amplitude, low frequency noise and limited length of time series, the selection of the time window is found to matter for the resulting mean value. The idea pursued in this paper is that the uncertainty in the estimated mean value of resistance can be improved by considering multiple time windows rather than just one.
First, the paper deals with the estimation of mean value of resistance from the given towing force data using 1) Single Time Window (STW) technique 2) Multiple Time Windows (MTW) technique. By applying these techniques to a given time series with a known mean, we can compare reliability of these estimators. Finally, the carriage and ship model are mathematically modeled as moving spring-mass-damper system. Comparison of the simulated towing force data with the experimental data allows us to understand the major sources of noise in towing force data.