New Fatigue Reliability and Random Vibration Methods for Linear and Nonlinear Systems

Principal Investigators: Zissimos P. Mourelatos, Oakland University, mourelat@oakland.edu
Dorin Drignei, Oakland University, drignei@oakland.edu
Student: Vasiliki Tsianika, Oakland University
Government: Monica Majcher, Igor Baseski, Mark Brudnak, U.S. Army TARDEC
Industry: Kurt Munson, HBM Prenscia (nCode & ReliaSoft)
John Skarakis, BETA

Project begins 2017.

Fatigue life estimation, reliability and durability are of paramount importance in acquisition, maintenance and operation of vehicle systems. Assessment of fatigue damage and durability in a laboratory environment using compressed or Accelerated Life Testing (ALT) methodologies is also very important. However, the problem is very difficult both theoretically and practically. Fatigue damage tests have showed that damage accumulation is very sensitive to non-Gaussian excitation and is also affected by elasto-plastic deformation in a multiaxial, non-proportional stress environment under non-stationary and non-Gaussian stress signals. The research will address these issues. In automotive applications, it is common to experience non- Gaussian input excitation such as road irregularities (e.g., bumps and potholes). The traditional Gaussian random test signals do not represent these road irregularities accurately. Higher field failure rates can be observed if the right type of test signal is not used during vibration testing.

We will improve the current fatigue life reliability estimation (not fatigue prediction) and random vibration methods using multiaxial fatigue damage models for linear and nonlinear dynamic systems under non-Gaussian and non-stationary loads. We will also develop an ALT methodology using non-Gaussian excitation to avoid applying unrealistically high loads. The ALT method will be implemented in TARDEC’s Physical Simulation Lab. To support the proposed research in fatigue reliability and ALT, new random vibration methods for nonlinear dynamic systems will be also developed using Singular Value Decomposition (SVD) and Polynomial Chaos Expansion (PCE). We will remove the stationary, Gaussian and narrow-band random process assumptions, and will use the Miner’s linear model under random loading as well as multiaxial stress-strain fatigue models to account for damage due to both elastic and plastic deformation. Cycles will be counted using a rainflow algorithm in the time domain avoiding the restrictive and commonly used narrow-band assumption. All developments will improve current practices in durability assessment using short-duration tests and will be institutionalized at TARDEC’s Physical Simulation Lab.

Publications from closely related prior work:

  • Z.P. Mourelatos, M. Majcher, V. Pandey and I. Baseski, “Time-dependent Reliability Analysis Using the Total Probability Theorem,” ASME Journal of Mechanical Design, 137(3), 031405 (8 pages), 2015.