Reliability, Availability and Maintainability (RAM) Analysis of an Industrial Process

Abstract

The contribution made in this thesis is helpful to manufacturing industries for studying the performance of industrial process. The design of manufacturing process, its maintenance and satisfactory performance play vital role in order to produce reliable product. In order to analyse the performance of industry, a methodical scheme based on theory of Boolean algebra and Markov model has been developed in the present work. In addition, this thesis introduces a new numerical approach for evaluating transient computation of Markovian system of equations that are obtained by using supplementary variable technique. This approach incorporates different numerical methods including Finite Difference Methods, Lagrange’s interpolation method and Simpson’s one-third method. Further, an optimal maintenance-scheduling model has been developed by giving constraints on availability and cost of maintenance to find the optimal maintenance rates of industrial manufacturing machines or system when they underwent through preventive and corrective maintenance action. The proposed methodology is demonstrated on a fabric industry. From this application, it is concluded that mixer and slasher modules are sensitive towards the performance of industry. Further, systems of these modules have been analyzed by using the maintenance model in order to study maintenance rates of units of these systems.

This research work started with following three research objectives for studying the performance of the manufacturing process: 1. To conceptualize a RAM model for the selected industrial process/system 2. To transform the conceptualized RAM model into a quantitative and simulation ready model 3. To assess the influence of maintenance on the industrial process/system’s reliability and availability through simulation Keeping these objectives in view, the research work has been carried out. The whole research work is divided into seven chapters. The chapter wise summary of the thesis is as follows:

Chapter 1 Chapter 1 is introductory. In this chapter, apart from preliminary concepts to be used in the sequel, the literature on performance analysis of the systems along with a brief plan of the results presented in the subsequent chapters has been discussed.

Chapter 2 In this chapter, a methodical scheme is proposed to compute performance metric of the industrial process. This scheme utilizes the concepts of Boolean algebra and stochastic process, and is referred as Assiduity Progression Diagnosis (APD). The proposed scheme overcomes the limitation of stochastic models in which state space grows beyond certain limits while implementing on industrial process. Due to which number of equations in the system of differential equations, governing the performance of the industry, increases abruptly and it becomes a difficult task to compute performance parameters. This scheme helps the manufacturing industries to predict Reliability, Availability and Maintainability (RAM) of industrial process. The model thus developed helps to determine reliability, three kinds of availability (steady state, time dependent and inherent availability), maintainability, average production and other related metrics of the industrial process by using state probabilities efficiently. The proposed model has finally been implemented to the fabric industry for its performance analysis and has later been discussed in Chapter 5. The contents of this chapter meet our first objective.

Chapter 3 This chapter is in concerned with performance analysis of an industrial system where stochastic models are developed by using the supplementary variable technique (SVT) for a system having constant and variable transient rates. This technique can be applied to the RAM model (APD, as discussed in Chapter 2) when failure and repair rates of modules of an industry are variable. However, in this case the performance metrics such as reliability, time dependent availability and mean time to failure of process will not be computed from resulting mathematical model. In fact, the analytical methods such as Laplace Transform method, Lagrange’s method and separation of variable method have been used to solve the resultant system of simultaneous linear partial differential equations but analytical solution is so intricate that the industry persons will not be able to use them conveniently. In order to overcome this situation, a numerical technique is developed in this chapter to solve such a complex mathematical problem. The proposed numerical approach consists of three numerical methods (finite difference schemes, Simpson’s one-third method and Lagrange’s interpolation) which iteratively compute the transient solution of the Markovian model taken under consideration. This hybrid method is named as Lagrange Finite Difference Simpson Method (LFDSM). The contents of this chapter meet the second research objective.

Chapter 4 Three important criteria: (i) minimum replacement cost rate (ii) maximum availability and (iii) lower bound on the mission reliability are frequently used in literature for scheduling maintenance of an industrial system. By selecting anyone of these criteria, the other two criteria are generally ignored in the case of system maintenance. This ignorance may not provide flexibility for the decision makers to finalize the maintenance scheduling of the system. In this chapter, the problem of maintenance scheduling of the industrial system in terms of optimal failure and repair rates is considered. An optimization model is presented to maximize the total profit from the output of five units working system by imposing constraints on availability and maintenance cost of the system. The reliability criteria can also be used in this model under some conditions. This model is developed by taking into account the concept of stochastic process based on supplementary variable technique and optimization theory. A decision strategy, using the maintenance cost constraint, has next been proposed for a decision maker. The application of the proposed model is finally demonstrated on two industrial systems and has been discussed in Chapter 6. The contents of this chapter fulfil the third research objective.

Chapter 5 This chapter deals with the application of the APD model developed in Chapter 2 on a fabric industry to analyze its performance analysis. Certain parameters such as reliability, time dependent availability, steady state availability, maintainability, and average production metrics are used as performance measures for overall behavior of fabric industry to achieve desired production targets. The behavior analysis of APD model reveals that the mixer and slasher system of fabric industry are the most sensitive and affect more to the overall performance of the industry. A comparative study has also been presented in this Chapter between the APD model and the traditional approach for evaluating the performance of industry by using stochastic models.

Chapter 6 In this chapter, the optimal results for failure and maintenance rates for only mixer and slasher systems have been obtained by using the optimal maintenance scheduling methodology as discussed in Chapter 4. Further, in this chapter we have also discussed that how the steady state behavioral analysis can be used instead of transient analysis for studying optimal maintenance scheduling of these systems.

Chapter 7 In this concluding chapter, limitation and scope of the methodologies developed in this thesis to study RAM analysis of selected industrial process has been discussed. The industrial significance of the results as well as a scope for further work on this topic are also presented in the concluding chapter.