Uncertainty Analysis of Centrifugal Compressor Aero-Performance Test Data: Effects of Correlated Systematic Error
This paper presents the continuation of the work performed during the development of an uncertainty analysis method for estimating error levels in data gathered during factory aero-performance acceptance tests of centrifugal compressors. The previous work incorporated the effects of the variation and uncertainty levels associated with every parameter used in the calculation of centrifugal compressor aero-thermal performance. The work discussed herein focuses on the effects of the variation and uncertainty levels associated with the key measured variables, which are the parameters identified as having the greatest effect on the uncertainty of the performance measurements. Also included in this work is an evaluation of the effects of the correlated bias uncertainty components associated with said key variables, as well as comments on how these effects can be harnessed to reduce the uncertainty of the test data. The evaluation is performed via parametric studies, which present the test uncertainty levels achievable as a function of different correlation levels between the systematic uncertainty components of the measured data. Two different methods are used for the analysis of data measured for several machines. The first method is based on the direct use of the Monte Carlo simulation technique combined with real gas equations of state. The second method employs uncertainty propagation equations and the methodology included in the ASME PTC-19.1(1998) test code. Both approaches use the polytropic compression model and equations for performance evaluation included in the ASME PTC 10 (1997) Power Test Code. Data gathered during an on-site acceptance test of a centrifugal gas compression package are used to illustrate the effects of the uncertainty in the knowledge of the gas composition handled by the compressor over the uncertainty levels that can be obtained with this type of tests.