National Aeronautics and Space Administration
Washington, DC 20546–0001
And
U.S. Army Research Laboratory
Adelphi, Maryland 20783–1145
Simulation of Crack Propagation in Engine Rotating Components Under Variable
Amplitude Loading
The crack propagation life of tested specimens has been repeatedly shown to strongly depend on the loading history. Overloads and extended stress holds at temperature can either retard or accelerate the crack growth rate. Therefore, to accurately predict the crack
propagation life of an actual component, it is essential to approximate the true loading history.
In military rotorcraft engine applications, the loading profile (stress amplitudes, temperature, and number of excursions) can vary significantly depending on the type of mission flown.
To accurately assess the durability of a fleet of engines, the crack propagation life distribution of a specific component should account for the variability in the missions performed (proportion of missions flown and sequence). In this report, analytical and
experimental studies are described that calibrate/validate the crack propagation prediction capability for a disk alloy under variable amplitude loading. A crack closure based model was adopted to analytically predict the load interaction effects. Furthermore, a
methodology has been developed to realistically simulate the actual mission mix loading on a fleet of engines over their lifetime.
A sequence of missions is randomly selected and the number of repeats of each mission in the sequence is determined assuming a Poisson distributed random variable with a given mean occurrence rate. Multiple realizations of random mission histories are generated in this manner and are used to produce stress, temperature, and time points for fracture mechanics calculations.
The result is a cumulative distribution of crack propagation lives for a given, life limiting, component location. This information can be used to determine a safe
retirement life or inspection interval for the given location.
Several assumptions have been made to simplify the FASTRAN II simulation and to fill in for certain unknowns about the actual usage of the engine. First, it was assumed that the engines are ‘pooled’ at the depot. By this it is meant that an engine returned to the depot for maintenance will not necessarily return to the original ‘owner’ but will go to the first unit requiring an engine. This allows us to model the selection of each block of missions with an appropriately partitioned uniform random variable.
The probability of selecting a particular mission and ambient condition is directly linked to the fraction of time the engine is expected to spend performing that mission/ambient condition combination as described by the Army mission mix specification
Another, and perhaps oversimplifying, assumption is that, while stationed with a unit, the aircraft will perform one mission exclusively. For example, aircraft stationed at a training facility perform only the training mission. It was also assumed that the block length (number of missions between removal of the engines for overhaul at the depot) could be modeled as a Poisson process. The purpose of using a Poisson distributed random variable, as opposed to a fixed block length, was to more closely model the
actual engine usage (the T700 has no scheduled overhaul interval).
Summary and Conclusions
The FASTRAN II crack propagation analysis code, which is based on the crack closure model, accurately predicts crack propagation in surface cracked Kb bar specimens under variable amplitude loading.
A method of estimating crack propagation life distribution of helicopter engine rotating components subjected to thirty different missions of varying severity, length, and probability of occurrence, has been developed. This method utilizes several random variables to generate simulated engine histories that are then fed into the FASTRAN II crack propagation analysis code.
There is a direct correlation between the mean number of missions between removal for
inspection/overhaul and the variance in the predicted crack propagation lifetimes. Extra caution is therefore necessary when determining the retirement life or a safe inspection interval for critical rotating components. A more accurate accounting of the actual usage of the engines would also seem prudent.