Practical Method for Estimating Number of Test Readings Required

Abstract

This paper provides a method for estimating the number of readings required in order to meet a desired level of accuracy in the results of a test. The uncertainty in the accuracy in this case is due to the scatter in the data having the largest influence on the results. This scatter is caused by variations in the system whose measurements are being recorded. The method is based on published statistical tables which can be used to estimate the standard deviation of readings in a given sample using the range, the difference between the largest and smallest readings in the sample. The paper deals with a pratical approach for arriving at the required number of readings and should supply information which can be very useful to the test engineer responsible for testing. This method was developed for steam turbine performance testing and is included in the newly proposed revision of the Power Test Code for Steam Turbines, PTC 6. For this application and as used in the examples presented, the desired level of accuracy due to scatter is 0.05 percent. The method, however, is suitable for any type of test for which a specific degree of accuracy is desired.