Reliability of High Temperature Lightweight Valve Train Components in a Total Probabilistic Design Environment

Designing brittle material components to sustain transient structural loads without failing requires the use of a probabilistic design methodology. In the standard probabilistic treatment for ceramics, strength is assumed to be the sole stochastic parameter. However, variation in other parameters such as geometric dimensions, materials, and service load histories also exist and should be considered when evaluating the total probability of failure. Each of these parameter variations can cause changes in the stress state, which will in turn affect the probability of failure. The NASA CAKES/Life code has been updated to interface with the ANSYS finite element analysis (FEA) Probabilistic Design System (PDS) in order to assess the total probability of failure for a given system. Hence the objectives of this work are: 1) Perform time dependent reliability analysis for a silicon nitride ceramic valve subjected to an actual diesel engine cyclic thermomechanical load history, and 2) demonstrate the capability and value of the total probabilistic design approach in designing ceramic structures. Results indicate that the silicon nitride valves are extremely reliable even after very long service lives. Results of the PDS analysis illustrate that the probability of failure is not a single value, as defined by the traditional probabilistic analysis for ceramics, but a range spanning several orders of magnitude. In addition, the PDS sensitivity analysis is demonstrated to identify which parameters have the maximum effect on the valves' probability of failure (sensitivity plots) for efficient focus of resources.