Turbine Engine Diagnostics Using a Parallel Signal Processor

Abstract : For the past several years, the Measurement and Computing Systems Laboratory has been working in close cooperation with the United States Air Force at Arnold Engineering Development Center (AEDC), Arnold AFB, to develop techniques for large scale instrumentation systems. In depth, online analysis of test data from turbine engine testing is critical to ensuring an accurate, timely evaluation and diagnosis of engine performance. Given the complexity of the analysis algorithms and the quantity of data, the computations overrun the capability of the fastest supercomputers. This paper describes the development of Computer Assisted Dynamic Data Monitoring and Acquisition System (CADDMAS). The CADDMAS is a 48 channel, 50 KHz full time analysis system, capable of flexible analysis of signals in the time and frequency domains. Data is presented on real time displays, showing, for example, spectrums, Campbell Diagrams, engine order tracking. The system (both hardware and software) is synthesized using a novel model based technique. The approach has been used to generate several systems used for online military and commercial turbine engine data analysis at Arnold AFB and for analysis of the SSME for NASA. On-line analysis has had a significant impact on turbine engine testing, reducing the time necessary to meet testing objectives and improving the quality of testing results. Substantial savings have been demonstrated by allowing immediate access to reduced data.