Stochastic Process for the Availability Assessment of Single-Feeder Industrial Energy System Sections

Abstract

Given the high interruption costs of industrial processes, dependent on the cause, duration and time, accurate availability assessment of the industrial energy systems supplying those processes is crucial in order to avoid unnecessary investments or indicate necessary improvements. A method is presented to sectionalize an industrial energy system based on its radial topology and recoverability options. Additionally, this paper introduces a novel non-renewal stochastic process able to compute the (un)availability of single-feeder sections with respect to cause, duration and time, taking into account competing risks, general transition rates and both minimal and perfect corrective maintenance. An accompanying solution strategy is developed which allows decoupling of the very diverse time constants. Finally, a numerical illustration is provided validating the proposed approach against the Markov and semi-Markov processes under their respective assumptions and illustrating the possible errors when modeling the problem using the aforementioned stochastic processes.

Redundancy Dependence in the Context of Competing Risks and Dynamic Degradation

“…1) competing risks, each with their respective maintainability and recoverability actions; 2) general failure rates, dependent on both calendar time and state sojourn time; and 3) dynamic degradation, where the duration of an abnormal loading event impact the resulting degradation. In prior work by the authors, a stochastic model was presented for the availability assessment of a component [19]. To allow for general transition rates and different restoration degrees, the state space is considered to be semicontinuous, including any state' sojourn time.…”

“…This section presents two models to account for redundancy dependence between components, incorporating the effect of abnormal loading duration. First, the concept of loading states is introduces in the context of the multistate semicontinuous state-space discussed in [19]. Next, a per-unit-degradation approach is presented to model the effects of multiple loading conditions on instantaneous failure rates.…”

“…A stochastic model is presented in [19] for the availability assessment of a single component. It assumes a semicontinuous state-space, meaning that the state description for any state x is extended with its sojourn time φ x resulting in (x, φ x ).…”

“…In [19], the system conditions are assumed to be known a priori, and therefore so are its failure rates. However, this assumption no longer holds if the loading of one component may become dependent on the state of others, affecting its reliability.…”

“…However, this assumption no longer holds if the loading of one component may become dependent on the state of others, affecting its reliability. Direct accounting for such dependencies using the stochastic process presented in [19] would require modeling all dependent components as a single system. As each component has its own normal operation sojourn time φ n , this would change the normal operation state (n, φ n ) into (n, φ n ).…”