Comparison of Continual On-Board Inspections to a Single Mid-Life Inspection for Gas Turbine Engine Disks Brian D. Shook

1. Comparison of Continual On-Board Inspections to a Single Mid-Life Inspection for Gas Turbine Engine Disks Brian D. Shook, Harry R. Millwater Department of Mechanical EngineeringUniversity of Texas at San Antonio Steve J. Hudak, Michael P. Enright, and William L. Francis Southwest Research Institute AIAA/ASME/ASCE/AHS/ASC Structural Dynamics & Materials Conference

2. Introduction • Current maintenance requirements are that engine disks are removed after a certain number of usage hours. • On-board engine health monitoring will facilitate continual inspection of engine disks (once per flight). • On-board inspection could revolutionize the cost associated with gas turbine engine maintenance (retirement for cause).

3. Key Issue • Which has a Lower Probability-of-Fracture? • Continual monitoring (once per flight) with a low-precision inspection or • A single high-precision inspection at the mid-life of the disk • Is it Possible to Improve Safety and Decrease Maintenance Costs?

4. MeasuringProbability-of-Fracture (POF) • Monte Carlo Sampling Yes No Yes No

5. DARWIN® Software Package

6. Probability of Detection (POD) Curve • Gives the Probability of Detection as a Function of Crack size • Lognormal Distribution for Ultrasonic Sensors 1 0

7. Minimum Detectable Crack Size (MDCS) Failure Inspection Simulation 1 Sample Value 0

8. Sample Value Inspection Simulation Cont. POD(a(Ni)) Crack is detected for sensor realization j because POD(aj)<POD(a(Ni)) POD(a*j)

9. Differences in On-Board Inspection • Depot Inspection • Different Inspectors • Different Equipment • Human Error • On-Board Inspection • No Inspectors • Identical Equipment • No Human Error

10. DependenceaMDCS(j)  aMDCS(j+1)

11. Numerical Examples • 2 Mission Types with 8000 Flight Cycles • Air-to-Ground • Functional-Check-Flight • Stress from Flight-Data-Recorder RPM Data

12. Specific Cases Studied Depot Inspection Continual Inspections • 3 Median POD Values • 200 Mil • 400 Mil • 600 Mil • 32% Coefficient of Variation • 30 Mil POD • 32% Coefficient of Variation 5000 Monte Carlo Simulations for each Inspection Type

13. Computational Methodology • Large Numbers of Inspections Lead to Long Computational Times • 12 – 24 Hours • Condor Software Used to Pool Lab Resources • Distributes Input Files (Jobs) to Available Machines • Manages Jobs • Returns Results on Completion

14. Condor Network

15. Single Mid-Life Inspection No Inspection Continual Inspection Air-to-Ground Results

16. No Inspection Air-to-Ground Results Single Mid-Life Inspection Continual Inspection

17. No Inspection Air-to-Ground Results Continual Inspection Single Mid-Life Inspection

18. No Inspection Functional-Check-Flight Results Single Mid-Life Inspection Continual Inspection

19. No Inspection Functional-Check-Flight Results Single Mid-Life Inspection Continual Inspection

20. No Inspection Functional-Check-Flight Results Continual Inspection Single Mid-Life Inspection

21. Conclusions / Future Work • Conclusions • Continual Inspection is Significantly Better Than a Single Depot Inspection if the Sensor is Sufficiently Accurate. • Conservative Simulations can be Used to Assist Sensor Designers in Determining the Required Accuracy of an On-Board System. • Future Work • Investigate Other Mission Types • Instruments and Navigation • Live Fire • Target Tow • Etc. • Evaluate Anticipated POF when Experimental Data is Available.

22. Questions