Henry A.J.,50 Durham Hall |
Nachlas J.A.,50 Durham Hall
Proceedings - Annual Reliability and Maintainability Symposium | Year: 2012
Over the recent past, the evolution in maintenance planning techniques has been encouraging. The transition from time and calendar based strategies under good as new assumptions to equipment status monitoring and non-stationary post maintenance time to failure models has been difficult but steady. In this paper, we extend the recently defined equivalent age model to condition-based maintenance (CBM) under non-stationary post-maintenance states. The use of an equivalent age model was recently proposed in the literature to account for the effect of the variations in intensity of use, ambient operating conditions and device technology (based on its vintage) on a device/system's aging rate. In certain applications, it is reasonable to foresee the existence of an interaction effect between usage profiles and operating conditions on a system's aging rate. This interaction effect was neglected in previous works where the effects of individual measures of usage intensity and operating conditions were assumed to be independent. In addition, there have been no attempts in the literature to investigate how equivalent age can be combined with a predictive maintenance model. In this paper, we relax the independence assumption seen in previous equivalent age literature and include the ability to model maintenance history. Moreover, we show practitioners how to develop a more complete maintenance strategy that allows for both corrective maintenance (CM) and CBM using a simple decision routine. We provide an analytical example that illustrates the model's sensitivity to the operating and ambient conditions. This leads to the construction of a more effective maintenance plan, which takes into account the probability of a future catastrophic equipment failure based on historical usage profiles, ambient conditions and vintage. Additionally, maintenance can be triggered by abnormal shifts in process conditions (which we assume to be associated with abnormal degradation or equipment wear). © 2012 IEEE.
Ellis K.P.,50 Durham Hall |
Sherali H.D.,50 Durham Hall |
Saunders L.,50 Durham Hall |
Shao S.,50 Durham Hall |
Crawford C.,50 Durham Hall
IIE Transactions (Institute of Industrial Engineers) | Year: 2014
An important strategic-level decision in the industrial gas industry involves the allocation of bulk gas tanks to customer sites. For a set of customers having specified demands, the bulk tank allocation problem determines the preferred size of bulk tanks to assign to customer sites in order to minimize tank investment costs and gas distribution costs for the industrial gas distributor. For a single gas product and multiple depots, the problem is modeled as a mixed-integer program and then solved using a decomposition approach. A heuristic for clustering customers and developing routes is proposed based on a sweep algorithm. These potential routes serve as input for the bulk tank allocation problem, which selects routes and assigns tanks to customer sites. The approach provides a valuable method for analyzing strategic-level decisions while incorporating operational-level characteristics. The approach is evaluated using data from an international industrial gas distributer to demonstrate the potential for improving the efficiency of gas distribution. © 2014 © 2014 Taylor & Francis.