Troubleshooting Turbomachinery Problems Using A Statistical Analysis Of Failure Data.

Abstract

The use of statistical techniques to investigate machinery failure data is one of the most powerful troubleshooting tools available to the turbomachinery engineer. This analysis tool can provide a great deal of information and insight concerning causes of failures and the probability of future ones. Using a graphical technique known as a hazard analysis or Weibull analysis, failure data of the type commonly available in the field can be analyzed to determine failure modes, the mean time to failure, the effects of different failure modes and the probability of future failure. The graphical procedure described in this paper is very simple to use, and requires only elementary math. The technique uses such readily available parameters as running hr or installed time of the subject under investigation and can easily handle incomplete or censored data. The method used to construct and interpret a hazard, or Wei bull chart, using either special Wei bull paper or simple log-log graph paper is presented. A simple fictitious example is used to illustrate the method. Eight case histories are then presented to illustrate the procedure and show the many ways that the analysis can be applied to machinery problems.