Browsing by Author "Kar, Tapas Ranjan"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Fuzzy logic-based fault diagnosis for mining equipment failures
    Kar, Tapas Ranjan (Virginia Tech, 1989-12-11)
    Equipment availability is the most significant factor in the productivity of many mines and processing plants. Machine breakdowns are not only expensive in terms of production losses but also important in meeting production schedules. In a complex piece of machinery like a shearer or a powered support system in a highly automated longwall face, such breakdowns can be due to one of the large number of possible faults. A large proportion, up to 80% of the down time is spent in locating the fault. For this reason, a need for an automated diagnostic method to assist the operator in the diagnosis process is felt. In this study, a diagnostic system is developed by modeling the partially known or imprecise relations and poorly defined variables found in a diagnostic environment. Logic of fuzzy sets and systems theory finds an interesting application in this area. This study presents a diagnostic algorithm, which relates the possible causes of failure to their respective symptoms through fuzzy logic paths. Applications of the diagnostic method are illustrated through examples of a compressor and a shearer.
  • Thumbnail Image
    Optimal stress screening for products sold under warranty
    Kar, Tapas Ranjan (Virginia Tech, 1993-12-05)
    In the face of increasing awareness among customers and today's competitive market, the warranty of a product has become an added feature in marketing strategy. A reliable product causes less warranty support cost. However, a more reliable product costs more to manufacture. Thus, a suitable trade-off between the cost and the benefit of a development and testing program is essential to optimize the performance measure, e. g., minimize total expected cost. Renewal theoretic models of sequences of failures over the burn-in and warranty periods and their costs are developed. Contrary to the usual asymptotic assumptions, transient behaviors of the renewal processes are considered. The expected costs associated with in-plant and field failures are balanced against the costs of implementing a burn-in program. A multi-component series system with different Weibull distributions for the components are considered. Burn-in is performed at the assembly level and the components are assumed to have different age accelerations under a common stress regimen. Models based on analyses both at the component and the system level are constructed. Two different burn-in policies are considered. These are "fixed duration" bum-in and "failure freell burn-in. A free replacement warranty for the components with policies of both fixed warranty period and renewed warranty period after each failure is considered in the models. The profit functions under different models are optimized with respect to burn-in period, stress parameters and warranty period. The models are extended to include reliability growth over the warranty period. Finally I solution procedures for optimizing the profit functions for all cases are given.