mAOR: A heuristic-based reactive repair mechanism for job shop schedules

Abstract

Literature on job shop scheduling has primarily focused on the development of predictive schedules that generate an allocation sequence of jobs on machines. However, in practice, frequent deviations from a predictive schedule occur when the job shop experiences either external (e.g. unexpected arrival of urgent jobs) or internal disturbances (e.g. machine breakdowns) and renders the schedules inefficient. The reactive repair of the original schedule is a better alternative to total rescheduling, as the latter is not only time consuming but also leads to shop floor nervousness. Most of the existing schedule repair heuristics handle singular disruptions only. In this paper, the typical job shop disruptions are studied and their repair processes are decomposed into four generic repair steps, which are achieved using the proposed modified AOR (mAOR) heuristic. An extensive simulation study has also been conducted to evaluate the performance of the mAOR schedule repair heuristic, and the results indicate that the mAOR heuristic is effective in repairing job shop schedules when faced with disruptions.

Analysis of reactive deadlock-free scheduling in flexible job shops

“…Over the past decade, many approaches have been proposed to solve the job insertion problem. These included heuristics (Chiang and Hau 1996;Unal et al 1997;Subramaniam and Raheja 2003;Subramaniam et al 2005), scheduling rulesbased heuristics (Ourari and Bouzouia 2003;Hall and Potts 2004), disjunctive graphs Roubellat 2000, 2002), Genetic Algorithms (GA) (Jain and ElMaraghy 1997;Mesghouni and Rabenasolo 2002;Branke and Mattfeld 2005), and Tabu search (Liu et al 2005). As for total rescheduling, Raheja and Subramaniam (2002) stated that, in general total rescheduling can result in considerable system nervousness and take more time to implement than other repair approaches.…”

“…They modelled these measures as constraints in their proposed solution models, or as additional components in the objective function. In their comparison studies, Subramaniam and Raheja (2003) and Subramaniam et al (2005) and Branke and Mattfeld (2005) measured the average nervousness (AN) resulting in the system as the normalized sum of absolute deviations of staring times of operations of original jobs in the revised schedule from their starting times in the original schedule: AN ¼ P n i¼1 P K i k¼1 jðSR ij À SO ij Þj P n i¼1 K i where, n = number of original jobs in the system, K i = number of operations of job i, SR ij = starting time of operation j of job i in the revised schedule, SO ij = starting time of operation j of job i in the original schedule.…”

“…This is due to the occurrences of some unforeseen events, or disturbances, which usually require some degree of modification in the production schedule. Subramaniam and Raheja (2003) have listed 17 types of system disturbances. However, the focus in past literature has been mostly concentrated on arrival of new jobs in the system and machine breakdowns.…”

“…Furthermore, the variation in the originally set schedule affects other scheduled tasks like assignment of machine tools and delivery of raw material. In previous literature, this variation has been referred to as system nervousness, and has been usually estimated as the sum of absolute differences in the starting times of operations of the original jobs between the original and the revised schedules (Subramaniam and Raheja 2003).…”

Order Acceptance Decision in Make-to-Order System: Satisfaction Level Based Approach

“…Or it may be between the operations of new order and the forecasted order. To convert infeasible schedule into feasible one, right hand shift method of rescheduling is used as it is a popular method and is easy to implement (Subramanyam and Raheja, 2003). The schedule is said to be feasible if all the operations of orders can be released and finished within the above calculated time windows.…”