Optimal Pressure Based Detection of Compressor Instabilities Using the Hurst Exponent

Paper #:
  • 2017-01-1040

  • 2017-03-28
  • 10.4271/2017-01-1040
Kerres, B., Mihaescu, M., Gancedo, M., and Gutmark, E., "Optimal Pressure Based Detection of Compressor Instabilities Using the Hurst Exponent," SAE Int. J. Engines 10(4):1917-1926, 2017, https://doi.org/10.4271/2017-01-1040.
The compressor surge line of automotive turbochargers can limit the low-end torque of an engine. In order to determine how close the compressor operates to its surge limit, the Hurst exponent of the pressure signal has recently been proposed as a criterion. The Hurst exponent quantifies the fractal properties of a time series and its long-term memory. This paper evaluates the outcome of applying Hurst exponent based criterion on time-resolved pressure signals, measured simultaneously at different locations in the compression system. Experiments were performed using a truck-sized turbocharger on a cold gas stand at the University of Cincinnati. The pressure sensors were flush-mounted at different circumferential positions at the inlet of the compressor, in the diffuser and volute, as well as downstream of the compressor. Results show that the previously identified threshold value distinguishing between surge and stable operation when the analysis was carried out for a different and smaller compressor can be used also for this much larger compressor. The investigation concerning the sensor locations reveals that pressure sensors at the outlet or shortly upstream the volute tongue give the clearest distinction between fully stable operation and operation close to the surge line. Further investigations show that as currently implemented, the criterion would need a minimum sampling duration of 500 ms and sampling frequency of 512 Hz. An extended algorithm based on distinguishing between a mono- and multifractal pressure signal is shown to have potential as an early warning indicator.
SAE MOBILUS Subscriber? You may already have access.
Members save up to 40% off list price.
HTML for Linking to Page
Page URL

Related Items

Technical Paper / Journal Article