Please use this identifier to cite or link to this item:
|Title:||RAM ANALYSIS OF INDUSTRIAL SYSTEMS USING PETRI NETS|
|Authors:||Sachdeva, Anish Kumar|
|Keywords:||MECHANICAL INDUSTRIAL ENGINEERING;RAM ANALYSIS;INDUSTRIAL SYSTEMS;PETRI NETS|
|Abstract:||Due to the continuous advancement in technology, Industrial systems are becoming complex and expensive to operate and maintain. Increasing attention needs to be given to reduce the cost during production, operation and maintenance of the repairable systems. These objectives can be achieved with the reliability based design of the systems and optimization of the maintenance and operational activities in the industrial systems to ensure their full utilization. This has made the job of reliability/ maintenance engineers more challenging as they attempt to study, characterize, measure and analyze the behavior and performance of such systems. The central focus of the present research is to analyze, plan and optimize the reliability., availability and maintainability (RAM) aspects in industrial systems. A comprehensive review of the literature was conducted to identify the gaps and relavent research issues in these areas. Several classic methods are reported in literature to perform RAM analysis in the real industrial systems. But, most of them are not flexible enough to take into account non-exponential laws in modeling the failure-to-repair process, partial production levels, process dependencies, resource constraints, etc. Hence, an approach based on Petri nets, which is capable of modeling these issues in order to perform RAM analysis, is proposed in the present research work. Petri nets are used to model the various systems of a process industry (paper production plant) with an effort to predict the system behavior more realistically. The reliability and availability analysis for measuring the performance of these systems of paper production plant has been performed using Markovian approach and Monte Carlo simulation. The application of Petri nets for identifying the state space of system through reachability graph is explained. The use of Petri nets as a modeling device enabled the system managers to critically analyze the iv effect of 'interdependencies between the components, maintenance policies and other system constraints on the performance of the system. Sensitivity analysis has also been performed to study the effect of subsystem conditions on system performance. It was observed that the Petri net based methodology is extremely useful for modeling the complex industrial systems and understanding the dynamism of the system in an effective manner. For effective implementation of RCM program in any organization, the problem of determining the preventive maintenance (PM) schedule plays a significant role. To this effect, a framework for determining Optimum preventive maintenance (PM) schedules based on multi criteria, viz., availability, maintenance cost, revenue earning and age replacement cost is proposed in the thesis. The Simulink toolbox of MATLAB is used to solve the formulated non linear equations for determining preventive maintenance schedules. The optimum results are obtained by integrating Simulink with Genetic Algorithm. The proposed framework for the optimization of PM schedules for components based on integration of Genetic Algorithm with Simulink did away the need for solving a plethora of mathematical equations. The effect of optimum PM schedules on system performance has been examined with Petri net modeling. The modeling of the PM schedules with Petri net model will help the maintenance managers to quantify improvements in 'availability', 'maintenance cost' and other performance indices. A new methodology based on fuzzy TOPSIS (technique for order preference by similarities to ideal solution) for evaluating the maintenance criticality index of failure causes is also proposed in the thesis, taking note of limitations of traditional FMECA approach into consideration. The proposed approach for determining the fuzzy maintenance criticality index (FMCI) is based on more number of criteria (chance of failure, chance of non detection, downtime length, spare parts criticality, and safety factor) than the number of criteria of RPN|
|Research Supervisor/ Guide:||Kumar, Dinesh|
|Appears in Collections:||DOCTORAL THESES (MIED)|
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.