No abstract

In an important class of systems, which arises in manufacturing, chemical process, and computer contexts, objects move sequentially from one work station to another, and rest between stations in buffers. In the manufacturing context, such systems are called transfer lines. The dynamic behavior of a buffered transfer line with unreliable work stations is modeled as a Markov chain. The system states consist of the operational conditions of the work stations and the levels of material in the buffers. The steady-state probabilities of these states are sought in order to establish relationships between system parameters and performance measures such as production rate (efficiency), forced-down times, and expected in-process inventory. The steady state probabilities are found by choosing a sum-of-products form solution for a class of states, and deriving the remaining expressions by using the transition equations. In this way, the order of the system of equations to be solved is drastically reduced. This algorithm suggests a general approach for solving large scale structured Markov chain problems.

No abstract

Although Energy Management and Control Systems (EMCS) have since the early 1970’s contributed significantly to the reduction (20-40 percent) of energy use in buildings without sacrificing occupants’ comfort, their full capabilities have not been completely realized. This is in part due to their inability to quickly detect and compensate for failures in the Heating, Ventilation and Air Conditioning (HVAC) system. In fact, no matter how good the control scheme for the HVAC system might be, the presence of undetected faults can completely offset any expected savings. This paper presents a methodology for detecting faults in an HVAC system using a nonlinear mathematical model and an extended Kalman filter. The technique was implemented in a computer program and successfully used to detect “planted” faults in simulations of the air handler unit of an HVAC system. Test results are presented to demonstrate the effectiveness of the methodology.

The paper deals with analytical modeling of transfer lines consisting of two machines decoupled by one finite buffer. In particular, the case in which a control policy (referred as "restart policy") aiming to reduce the blocking frequency of the first machine is addressed. Such a policy consists of forcing the first machine to remain idle (it cannot process parts) each time the buffer gets full until it empties again. This specific behavior can be found in a number of industrial production systems, especially when some machines are affected by outage costs when stops occur. The two-machine one-buffer line is here modeled as a discrete time markov process and the two machines are characterized by the same production rate. The analytical solution of the model is obtained and mathematical expressions of the most important performance measures are provided. Some significant remarks about the effect of the proposed restart policy on the behavior of the system are also pointed out