Expected demand for service parts when failure times have a Weibull distribution (Q799232)

The time between two successive failures being a random variable z with a Weibull probability density function: \[ f(z)=(\alpha\lambda )\quad (\lambda z)^{\alpha-1}\exp [-(\lambda z)^{\alpha}]. \] The function m(N,\(\tau)\) defined as the expected demand during a period time \(\tau\) of a N-parts system satisfies \[ m(1,T)=\sum^{\infty}_{j=1}c_ j(\alpha)\quad P(j,\lambda^{\alpha}T^{\alpha}) \] where P is the incomplete gamma function and \(c_ j(\alpha)\) are weighting factors. The number of service parts growing by B batches of size N/B, the expected total demand in (0,T) is: \[ M=(N/B)\quad\sum^{B}_{i=1}\quad m(1,T- t_ i), \] with ith batch entering at time \(t_ i=T_ 0(i-1)/(B-1) (i=1,...,B)\). An approximation M \(\hat A\) for M is proposed: \(\tilde M=N\lambda^{\alpha}(T-T_ 0/2)^{\alpha}\). Numerical results ''prove'' the quality of this approximation, especially for certain values of the parameters \(\lambda\), \(\alpha\).