Adaptive cycle counting for accurate fatigue damage prediction considering large and small cycles

点击翻译按钮获取中文摘要

Given the severe safety and financial consequences associated with any turbine failure, the ability to successfully predict the lifetime of a gas turbine is of critical importance to gas turbine owners and operators in the aerospace industry. Moreover, correctly identifying the lifetime of a turbine significantly reduces the cost of gas turbine fleet ownership, as owners are able to reduce maintenance costs by employing optimized overhaul and maintenance intervals as opposed to uncertain, and thus conservative estimates. The objective of the proposed research is to increase the accuracy and reliability of gas turbine prognostic systems, to further reduce the cost of turbine ownership. The current method for estimating damage accumulated during a fatigue cycle requires the strain range at fracture critical locations (FCL), which is calculated with FEA (finite element analysis) at the maximum and minimum conditions of the cycle. A Manson-Coffin type relationship is then used to relate strain range to fatigue damage. However, there is not necessarily a linear relationship between the measured operating parameter and the strain at the FCL; factors such as geometric stress concentration factor and material plastic behaviour may influence the strain range. Therefore, applying a constant pre-defined threshold could result in either missing significant cycles or counting unnecessary cycles. Improvements to the current method could take the form of an adaptive threshold, or other numerical techniques. This project will utilize non-linear FEA to analyze the accuracy of, and to make improvements to, the cycle counting method.