The technological obsolescence of a unit is characterized by the existence of challenger units displaying identical functionalities, but with higher performances. This paper aims to define and model in a realistic way, possible maintenance policies of a system including replacement strategies when one type of challenger unit is available. The comparison of these possible strategies is performed based on a Monte Carlo estimation of the costs they incur.
Most often, papers studying optimization of preventive or corrective maintenance policies rely on the assumption that failed or used pieces of equipment are replaced by identical items. Actually, the technological reality is often quite different. In practice, new equipments are regularly available on the market achieving the same missions, but with higher performances. These higher performances can be understood as smaller failure rates, lower energy consumption, a lower purchase cost, etc. At the same time, it can be more and more difficult or costly to find old-generation spares to replace degraded units. This situation is characteristic of technological obsolescence.
Managers then face important issues, such as, for instance: how to optimally schedule the replacement of old-type units by new-type ones? Is it economically more interesting to preventively replace all the old equipments, without benefiting from their residual lifetime, by their more performing challengers, or on the contrary is it preferable to replace gradually the old components in a corrective way, progressively with their normal outage, but at the risk of a larger number of failures? Such questions become even more crucial when spare parts are to be dealt with.
The aim of our work is therefore to define replacement policies of these obsolete equipments and to help the decision maker find an optimal strategy among them.
Previous works envisaged this problem in a simplified way.
In Ref. [1], the case of one single component subject to ageing, which can be either periodically maintained or replaced by a technologically more advanced unit was studied.
In Ref. [2], authors studied analytically the following case: A set of n identical and independent units can be either preventively or correctively replaced by new-type units. The replacements take a negligible time. The new-type units have a lower constant failure rate and a lower consumption rate. The so-called “K strategy” was introduced as follows [3]: first, new-type components are used only to replace failed old-type units; then, after K corrective actions of this kind, the n–K old-type remaining components are preventively replaced by new-type ones at the time of the Kth corrective intervention. The 0 strategy represents the preventive replacement of all old-type components at the initial moment. In Ref. [2], the authors reached the following conclusion: no matter which values are chosen for the data and the time horizon, only three strategies can be optimal: either all the components are replaced preventively (K=0), or one component is replaced correctively and the others preventively immediately after this first failure (K=1), or all the components are replaced correctively (K=n).
In Ref. [4], units subject to ageing and non-negligible stochastic replacement durations were envisaged and the same conclusions reached.
In the continuation of the works presented in Refs. [2] and [4], we introduce in this work more realistic maintenance actions, as extensions of the K strategy, and develop a complete model for the management of a set of identical units subject to obsolescence, in the presence of a maintenance policy and of challenger units with a limited number of maintenance teams.
This paper summarizes and extends the works presented in Refs. [5] and [6]. It is organized as follows: Section 2 describes the model proposed and the assumptions on which it is based. Section 3 illustrates, by numerical results based on Monte Carlo (MC) simulation, some aspects of the whole model and some of the strategies proposed. In particular, we compare in Section 3.1 our MC results with the analytical solution of the simplified problem from [2]. Section 3.2 treats how to deal with the spare part inventory and the time horizon on which the transition between technological generations takes place. In Section 3.3, we discuss on the basis of another set of data the ability to forecast a budget for the replacements, which is regularly distributed in time. Finally we conclude by some possible perspectives and extensions of the model.
