On a Simplified Delay Time Model of Reliability of Equipment Subject to Inspection Monitoring (Q3978741)

The investigated model is a single component system subject to periodic inspections. An observable defect can arrive at any moment \(t=y\) for a new unit but the system fails only after a delay interval \(h\). The detection of the defect at inspection time determines the replacement of the unit. The delay time is governed by the density function \(f(h)\) and the defect arising time by the density function \(g(y)\). The formula for the reliability \(R_ T(t)\) due to periodic inspection every \(T\) time units is derived by \textit{A. H. Christer} [ibid. 38, 329-334 (1987; Zbl 0608.90033)]. The formula is recurrent in the sense that the restriction \(r_ T^{(m)}(t)\) on the \(m\)-th interval of length \(T\) depends on the reliability \(r_ T^{(i)}(t)\) on every previous \((m-1)\) intervals. It also involves product convolutions on several intervals between \(g(y)\) and \(M(x)=\int^ \infty_ x f(h)dh\). Usually, one wishes to examine the changes in reliability when \(T\) or other parameters vary. As calculations are complex, the author's idea is to use linear or cubic fit on every inter-inspection interval and to restrict the calculations to \(kT\) points. The continuity: \(r_ T^{(m-1)}((m-1)T)=r_ T^{(m)}((m- 1)T)\), and the left and right derivatives on the first interval are sufficient to generate the linear and the cubic respective approximations. The author uses as main indicators of the appropriateness of the approximations the average reliability on the \(m\)-th inspection interval and the mid point reliability and the maximum discrepancy. The author presents the relative errors using linear and cubic fits for two cases (both \(f\), \(g\) exponential, \(f\) exponential and \(g\) uniform). In conclusion, knowing that usually \(f\) and \(g\) are also based on experimental data, altered by errors, the method simplifies the calculations considerably, while giving a reasonable good approximation. The paper is concise and very clear.