Stochastic Model of Crankshaft Failure Material Endurance Strength & Alternating Stress as Random Variables Applied to Large-Bore Slow-Speed Integral Natural Gas Reciprocating Engines
The endurance limit of steel and the crankshaft bending stress from bearing misalignment and combustion gas load can be treated as random variables. The probability distribution for each is defined by a mean and standard deviation. Even if the steel endurance strength is significantly greater than the crankshaft running stress the statistical distributions can be shown to overlap. This overlap is the probability the crankshaft running stress is greater than the steel endurance limit thus contributing to an eventual high cycle fatigue failure. An average crankshaft probabilistic model predicts 0.0112 percent or about 1 in 9,000 cycles exceeds the rankshaft endurance limit. Every cycle with stress levels greater than the material endurance limits counts toward failure. The statistical model forecasts that one percent of crankshafts will have failed after 6 billion cycles or about 55 years of operation at 75% runtime. The Albuquerque Division of EPNG has experienced about a 1.5 percent total failure to date with large bore reciprocating engines in operation since roughly 1955. As a crankshaft nears fatigue failure its more likely to happen during a period of elevated stress levels. For example; detonation stress levels that would not have caused failure in a new or even middle aged (100s of millions of cycles) crankshaft can be high enough to cause failure in a crank with 6 to 8 billion cycles.