International Journal of Applied Engineering Research ISSN 0973-4562 Volume 9, Number 23 (2014) pp. 21459-21471 © Research India Publications http://www.ripublication.com
A Constrained Genetic Algorithm Approach To Deliver A Quality Maintenance Global Service Toni Lupo* Università degli Studi di Palermo, 90132 Palermo (Italy) Dept. of Chemical, Management, Informatics and Mechanical Engineering * Corresponding author. Tel.: +399123861879 Email:
[email protected]
Abstract It is herein developed an approach to optimize the maintenance services policy related to a Global Service Contract. In particular, the latter requires the performing of corrective maintenance and replacements of the failed components on a set of equal vehicles of a logistic Company. The tackled problem concerns the determination of aneffective opportunistic maintenance policyon the basis of which when a fault occurs, it is replaced the failed component and, depending on the age of the otherscomponents, also replacements of others suitable components are performed, even if they are not yet broken, thus saving a substantial amount of system downtime. The problem is mathematically formulated by a constrained partition model aimed at the minimization of the global maintenance cost, which becomes difficult or very hard to solve by mathematical programming approach for large system as the one herein considered. For such reason, a suitable constrained genetic algorithm approach is employed to solve the considered problem.The performed optimization allows to point out components groups on which to perform maintenance actions when a system stop for failure occurs. In particular, a meaningful global maintenance cost reduction, up to 28%, can be obtained, thus demonstrating the effectiveness of the approach proposed. Keywords: opportunistic maintenance; global service contract; series system; corrective maintenance; maintenance services optimization; maintenance groups;
Introduction The Global Service (GS) is a particular typology of contract between two contracting parties: the Service Provider (SP) and the Outsourcer Company of maintenance
21460
Toni Lupo
services. It is strongly connoted by the exchanging of performance and, thus, it involves the assumption of duties andbonus for both the contracting parties. In particular, on one hand, the SP supplies maintenance services warranting fixed and measurable service performance levels and, on the other hand, the Outsourcer Company pays a fee for the obtained services (UNI 10146:2007). According to the UNI 10685:2007, “Criteria to prepare a maintenance Global Service”, the main contents of a GS contract are: Typology of requested services: they can be extremely heterogeneous since they can include not only the supplying of maintenance services but also of maintenance support servicessuch as, maintenance database system management, maintenance consulting etc..; Contract span (time duration): the particular connotation of a GS typically makes it a multi-year time duration contract. In this way, it is possible to evaluate with good reliability the achieved services performance levels; Payment modality: the way of payment can take into account the payment of a fee at a fixed periodicity or when a fixed value of a specific service indicator is achieved such as, for example, the performed hours of maintenance services or when a fixed number of maintenance actions are executed etc…; Performance indicators: according to the EN 15341:2007, they are defined by specific Service Level Agreements (SLAs) implying the definition of some Key Performance Indicators (KPIs). The latter are suitable and measurable indicators of service efficiency/effectiveness. The definition of the KPLs is based on the ideal service level that the Outsourcer Company takes into consideration for its good business. With regard to the definition of the KPLs, such parameters have to allow the impartial evaluation of the service levelsover the time and they have to be easily measured. As example, some KPIs usually considered in the GS field are reported below: Max value of system downtime; Max number of system failures during a fixed time horizon; Max number of contemporaneously failures system; Max time to restore a failed system. As mentioned before, this paper presents an optimization approach related to a GS contract between a SP and a logistic Company.The SP delivers maintenance services on a set of thirty equal vehicles of the outsourcer Company according to a GS contract of three-year duration. The maintenance services requested by the GS for each vehicle are: At least weekly, inspection and control of the critical parts as the brake system, the air compressor, the engine system, etc…; At least monthly, washing and lubrication of the transmission systems; The planned maintenance services with refer to the vehicle owner’s manual; Corrective maintenance activities with including the replacement of the fault component.
A Constrained Genetic Algorithm Approach
21461
For the latter activity, the GS individuates specific service performance levels. In particular, the KPIs specified by the contract are the following: a. for each vehicle, a total corrective maintenance time not greater than thirty days for year is permitted; b. for each vehicle, the duration of each corrective maintenance activity must not be greater than five days. In the light of the previous considerations, it is clear that the choice of a suitable maintenance policy matching the contract constraints (KPIs) is meaningful for the SP in order to obtain profitability. In fact, the latter parameter comes from the global maintenance cost reduction that the SP can obtain by choosing an effective maintenance policy.Based on these considerations, the present paper aims at the development of an effective maintenance policy for the considered GS contract. The remainder of the present paper is organized as follows: a brief literature review about maintenance models and optimization approaches is presented in the next Section, the mathematical model formulation of the tackled problem is given in Section 3. Section 4 describes the resolution approach based on a constrained genetic algorithm and Section 5 provides results of the case Company related to considered maintenance GS contract. Finally, conclusions, with a summary and directions for further future researches, close the work.
Literature review Over the years, many maintenance models and optimizations approaches have been developed in literature, including several review works.In 1965 McCall proposed the first survey on maintenance policy for systems subject to stochastic failures. In 1976 Pierskalla et al. proposed a survey on maintenance models for deteriorating systems citing about 250 references. Sheriff and Smith (1981) gave a survey on maintenance models for a single unit system subject to failure. Thomas (1986) proposed a survey on maintenance and replacement models of multi-item system. Valdez-Florez and Feldman (1989) developed a survey on preventive maintenance models between 1976 and 1988. Dekker et al (1997) provided an overview on multi-component maintenance optimization models published after 1991 with about 140 references. Wang (2002) made a survey on maintenance policies of deterioration systems that summarizes, classifies, and compares various existing maintenance policies for both single-unit and multi-unit systems. To our knowledge, Nicolai and Dekker (2006) made the most recent survey on maintenance policies of deterioration systems. In these surveys, a classification scheme of maintenance models is presented with the aim to support the decision maker in choosing the more suitable one with relation to the decision context in which she/he has to operate. In the light of the previous works, maintenance can be basically categorized into two major classes: preventive and corrective maintenance.Preventive maintenance (PM) is the maintenance that occurs when a system is in operating condition(Certa et al. 2012, 2013). According to MIL-STD-721B, PM means all actions performing in an attempt to retain an item in specified condition by providing systematic inspection, detection and prevention of incipient failure. On the contrary, corrective maintenance (CM) is the maintenance
21462
Toni Lupo
that is performed when a system failure occurs. Obviously, the CM is performed only on components that have insignificant failure consequences. According to MIL-STD721B, CM means that all maintenance actions are performed as result of a failure, to restore an item to a specified condition. With refer to the CM, many models have been introduced and extensively studied in the literature. The first study on CM optimization was developed by Radner and Jorgenson (1963). The Authors developed for a system with several monitored and stochastically failing parts an “opportunistic” replacement policy: when a system failure occurs it is replaced the failed part and, depending on the state (age) of the other parts, also others parts can be replaced, even if they are not yet broken. In such a way, a substantial amount of system downtime can be saved. Berg (1976) studied a system composed by two identical components subjected to CM. Under the considered maintenance policy, when a component failure occurs, the other component is also replaced by a new one if its age is greater than a pre-determined repair limit L. Zheng and Fard (1991) proposed an opportunistic maintenance policy based on failure rate tolerance for a system with K different unit types. The policy decision variables are L and u: the generic unit iis replaced (active replacement) when its hazard rate is within a pre-determined interval (Li-ui, Li) or at failure, and when an active replacement is performed, all the other units with their hazard rate within the related interval (L-u, L) are replaced at that time. Pham and Wang (2000) proposed two policies for the opportunistic maintenance optimization problem. The first policy works as follows: considering the decision variables and T with
