0
Research Papers: Gas Turbines: Structures and Dynamics

Probabilistic Engine Performance Scatter and Deterioration Modeling

[+] Author and Article Information
Stefan Spieler

Institute of Aircraft Propulsion Systems, Stuttgart University, Pfaffenwaldring 6, 70569 Stuttgart, Germanystefan.spieler@ila.uni-stuttgart.de

Stephan Staudacher

Institute of Aircraft Propulsion Systems, Stuttgart University, Pfaffenwaldring 6, 70569 Stuttgart, Germany

Roland Fiola, Peter Sahm, Matthias Weißschuh

 Rolls-Royce Deutschland Ltd & Co KG, Eschenweg 11, 15827 Blankenfelde-Mahlow, Germany

Response surface models can be derived by monitoring specific changes (so-called experiments) of input variables. Often used experimental designs are Box–Behnken, Central Composite, or three-level full factorial designs, which differ in approximation quality and number of experiments (4).

Thereby, the square of Pearson’s correlation coefficient R2 was determined to only 0.945 with a deviance of HPC efficiency of 2.5%.

Term from production logistics: Components are produced simultaneously or promptly successively. This generally results in more homogeneous characteristics than when produced in single part production.

Such as Anderson–Darling or chisquare (10).

Hereby, different sources of uncertainty (such as calibration uncertainty, drift, hysteresis, etc.) as well as their interaction in the probe are analyzed to derive a distribution function (usually Gaussian) for their measuring error (10).

J. Eng. Gas Turbines Power 130(4), 042507 (Apr 29, 2008) (9 pages) doi:10.1115/1.2800351 History: Received May 24, 2007; Revised May 29, 2007; Published April 29, 2008

The change of performance parameters over time due to engine deterioration and production scatter plays an important role to ensure safe and economical engine operation. A tool has been developed which is able to model production scatter and engine deterioration on the basis of elementary changes of numerous construction features. In order to consider the characteristics of an engine fleet as well as random environmental influences, a probabilistic approach using Monte Carlo simulation (MCS) was chosen. To quantify the impact of feature deviations on performance relevant metrics, nonlinear sensitivity functions are used to obtain scalars and offsets on turbomachinery maps, which reflect module behavior during operation. Probability density functions (PDFs) of user-defined performance parameters of an engine fleet are then calculated by performing a MCS in a performance synthesis program. For the validation of the developed methodology pass-off test data, endurance engine test data, as well as data from engine maintenance, incoming tests have been used. For this purpose, measured engine fleet performance data have been corrected by statistically eliminating the influence of measuring errors. The validation process showed the model’s ability to predict more than 90% of the measured performance variance. Furthermore, predicted performance trends correspond well to performance data from engines in operation. Two model enhancements are presented, the first of which is intended for maintenance cost prediction. It is able to generate PDFs of failure times for different features. The second enhancement correlates feature change and operating conditions and thus connects airline operation and maintenance costs. Subsequently, it is shown that the model developed is a powerful tool to assist in aircraft engine design and production processes, thanks to its ability to identify and quantitatively assess main drivers for performance variance and trends.

FIGURES IN THIS ARTICLE
<>
Copyright © 2008 by American Society of Mechanical Engineers
Your Session has timed out. Please sign back in to continue.

References

Figures

Grahic Jump Location
Figure 1

Exemplary illustration of method of sensitivity parameter correction, impact of surface roughness increase on HPC efficiency (data removed due to confidentiality)

Grahic Jump Location
Figure 2

Schematic illustration of the modeling method

Grahic Jump Location
Figure 3

PDF of relative SFC change due to fan tip clearance scatter (data removed due to confidentiality)

Grahic Jump Location
Figure 4

Exemplary trend and scatter plots over time for HPT tip clearance (upper chart) and HPC chord length (lower chart) (data removed due to confidentiality)

Grahic Jump Location
Figure 5

Standardized empiric PDF of SFC scatter according to delivery test data with approximated Gaussian PDF (data removed due to confidentiality)

Grahic Jump Location
Figure 6

Comparison of model results with Real engine test data for the parameter SFC including production scatter PDF at T=0 (data removed due to confidentiality)

Grahic Jump Location
Figure 7

Methodology to generate the PDF of failure times out of the trend and scatter plot

Grahic Jump Location
Figure 8

Example of environmental data input, here: global sea salt and dust concentration of particles larger than 2μm(18)

Grahic Jump Location
Figure 9

Pareto bar chart and cumulative percentage line of the features with the most significant impact on New Engine SFC scatter (data removed due to confidentiality)

Grahic Jump Location
Figure 10

Pareto bar chart and cumulative percentage line of the features with the most significant impact on SFC trend (here, after 10,000h) (data removed due to confidentiality)

Tables

Errata

Discussions

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In