DEVELOPMENT AND APPLICATION OF PREDICTIVE MARKOV-CHAIN CONDITION- BASED TRACTOR MAINTENANCE MANAGEMENT MODEL

Abstract

A repairable system (tractor) is subject to deterioration or repeated failure. At each inspection of failure status (partial, Combined and complete) a general repair (minor replacement of parts by preventive maintenance) or complete overhaul (major replacement of parts by corrective maintenance) is performed to restore the system to "as good as new ."state Considerable research has been conducted on the issue of periodic replacement times of failing systems. No doubt this is a reflection, at least in part, of the high capital cost, of many farming systems and on the importance of minimizing unnecessary failure costs. Despite relatively the large body of literature on this topic, analysis of dynamic maintenance schedule and their effect on performance of the system remains as an open problem. In response to these challenges this study is directed to develop a computerized recursive Markov-chain closed form analytical solution for prediction of tractor failure, analysis, control, scheduling of .maintenance activities and determining of tractor availability The developed algorithm of Markov chain for failure analysis looks at a sequence of events, defined as transition between states, and calculates the relative probability of encountering these events in short- run(partial),medium (combined) and long-run (complete). Hence, the algorithm is used to evaluate reliability and availability of tractor with .time-dependent transition rates using analytical matrix –based methods The developed procedure is written in Microsoft Excel (Spread sheet) operating environment. With the software the user will have the ability to manipulate and analyze the data directly using customized menus and .point and click mouse operation Typically, the failures time distribution are determined and applied, using three years tractor failure data collected from two different workshops at Sudanese Sugar Company(Sennar and Gunied) for three medium (72-120 .hp) tractors models To verify model accuracy, its basic functional relations (decision state matrix) are compared with Amri and McLanghim model (2004) and with .(WINQSB software using (t- test For purpose of model validation real data from the field is compared with that predicted by the model and results showed no significant difference .(between them (P =.05 Sensitivity analysis is used to utilize the model as experimental tool to explore the structure of various improvement scenarios using a two-step procedure and analysis of variance. Consequently, performance of the tractors with respect to six evaluation parameters and their ranks for the .various workshops is quantified