Methods for Approximating the Availability Functions
F. Michel Michel Dekking – Mentor (TU Delft - Applied Probability)
J.A.M. van der Weide – Mentor (TU Delft - Applied Probability)
Martina Kramer – Mentor (Shell Global Solutions B.V.)
Rommert Dekker – Graduation committee member ( Erasmus Universiteit Rotterdam)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Availability studies aim at modeling and quantifying the relation between system design and production effectiveness. Once this relation is modeled, different designs and operating strategies can be compared and ranked. Usually, the objective is to estimate and minimize losses that are due to equipment failures and planned shutdowns. Future performance of each unit/component may be modeled by the availability function that gives the probability that the unit is functioning at time t. Those functions are then combined together to assess future availability of a plant.
The main aim of this project was to improve existing methods for computing the availability functions on a component level. Formulation of the problem in a general framework allowed to obtain extra availability metrics that include the present condition of a component. Different methods for computing availability functions were tested and compared with respect to speed and accuracy. The best method was selected based on the performance on different problem families. For the purpose of the comparison the explicit formulas for the availability functions in case of Gamma alternating renewal process were derived. Additionally some new bounds on the availability functionswere obtained. The last part presents a new approach for modeling the availability of a component in situation when some failures are not visible (so called ”grace periods”).