Run Length and Statistical Error Estimation for Seakeeping Tests and Trials

A seaway and the resulting responses of models or full-scale craft are random processes. The accuracy of the final results depends heavily on the length of the test runs. Therefore in the planning, execution, and evaluation phases of a seakeeping test or trial, run length and its effect on the accuracy of the data must be accounted for. Two basic problem areas exist: 1. Run length estimation prior to a test or trial so that the mission can be carried out efficiently, and data gathered to a desired accuracy. 2. Error estimation for data taken in the past to determine what accuracy level was reached. The run length directly affects the level of accuracy or confidence in the processed results. This accuracy level is, in turn, a trade-off between the task priority and the statistical variability of the input medium (the seaway in this case) A higher priority mission will call for a higher data accuracy and longer run length while a changing sea condition will put constraints on this accuracy. In effect, items 1 and 2 must be considered in all missions to aid in pre-mission planning and to better understand the reliability of data gathered from a previous mission. This paper presents and describes methods to: 1. Determine the auto spectrum effective peak frequencies and half-power bandwidths needed to estimate run length and statistical error for given data samples. 2. Estimate random process mean values, standard deviations, variance, auto spectra and Response Amplitude Operators (RAO' s). 3. Estimate the statistical errors and run lengths associated with the mean value, standard deviation, variance, auto spectra and RAO's of a random process.