Scheduling Back-End Operations in Semiconductor Manufacturing

Improving performance of dispatch rules for daily scheduling of assembly and test operations

“…Referring to Deng et al (2010), GRASP_opt was designed to uncover a diversity of good feasible solutions by randomly selecting the machine setups at the upper level in accordance with an adaptive greedy measure and then solving the resultant lower level problem to obtain the optimal lot assignments. Since ASAP is somewhat limited in how it can be customized, Rule_GRASP_asap only adopts the GRASP_opt logic for solving the upper level problem and uses a combination of filters already available to determine the lot assignments.…”

“…The first attempt to use optimization technology to achieve these goals was undertaken by Deng, Bard, Chacon, and Stuber (2010) who formulated the scheduling problem as a mixed-integer program (MIP) with the following four objectives given in order of priority: (1) minimize the shortage of critical devices, (2) maximize the weighted throughput of the remaining lots, (3) minimize the number of machines used, and (4) minimize the makespan. Solutions were obtained with a reactive greedy randomized adaptive search procedure (GRASP) designed to examine a diversity of machine-tooling combinations and lot assignments over many iterations [the literature on GRASP is extensive; e.g., see Festa and Resende (2009) for an annotated bibliography of algorithms, and Feo, Venkatraman, and Bard (1991) and Monkman, Morrice, and Bard (2008) for manufacturing applications].…”

Learning to Dispatch Operations with Intentional Delay for Re-Entrant Multiple-Chip Product Assembly Lines

“…In particular, simulation-based studies have attempted to understand the characteristics of lot flows by performing tasks virtually in advance [6,13,[15][16][17][18]. Case base reasoning Utilization [31] Artificial neural network Flow time [32] Utilization [33,34] There is another line of research that aims to control lot flows by using dispatching rules [4,[19][20][21][22][23][24]26]. Dispatching rules have been widely used in various areas because they have the advantages of short computation time and ease of implementation [35,36].…”

Integrating optimisation and simulation approaches for daily scheduling of assembly and test operations

“…When scheduling AT operations, the goals are to achieve low cycle times, high throughput and high utilisation without violating agreed upon delivery dates. The first attempt to use optimisation technology to achieve these goals was undertaken by Deng et al (2010), who formulated the scheduling problem as a mixed integer programme (MIP) with the following four objectives given in order of priority: (1) minimise the shortage of critical devices, (2) maximise the weighted throughput of the remaining lots, (3) minimise the number of machines used and (4) minimise the makespan. Solutions were obtained with a reactive greedy randomised adaptive search procedure (GRASP) designed to examine a diversity of machine-tooling combinations and lot assignments over many iterations (the literature on GRASP is extensive; e.g.…”

“…To decide on the best machine-tooling set-ups and how to assign lots to machines, a three-phase heuristic was used. In phase I, a single-pass algorithm derived from the GRASP in Deng et al (2010) is run to obtain a tentative solution. Initial machine configurations are examined and the completion times of the lots in process at time zero are determined.…”