Heuristic algorithms for scheduling an automated wet-etch station

Using simulation and hybrid sequencing optimization for makespan reduction at a wet tool

“…Conceptually, a wet tool can be considered as a series of tanks (T1-T6) filled with either etchant chemicals, used to etch away the exposed photo-resist from the wafer layers, or deionized water, used to terminate the etchant action. The succession of chemical and water tanks is not necessarily alternating as reported in most research studies (Geiger, Kempf, and Uzsoy 1997;Bhushan and Karimi 2004;Zeballos, Castro, and Méndez 2011); in real plants, consecutive chemical tanks can be observed. For some chemical tanks, overexposure to the etchant will damage the wafers exposed; as a consequence, a Zero-Wait (ZW) constraint is applied; this suggests that the batch must be moved out of the "unsafe" tank when the prescribed exposure time is reached.…”

“…The equations used are not reported in this paper as their illustration goes beyond the analysis purpose. The trial and error algorithm developed can be considered a conceptual generalization of the JAT algorithm proposed by Bhushan and Karimi (2004). The two algorithms take the ZW and NIS constraints into account while generating a batch schedule; they also consider the robot availability constraint before proceeding with the calculation of the starting time for a batch at the next tank.…”

“…In order to evaluate the efficiency of the scheduling logic implemented in the model, the production scenario used in Bhushan and Karimi (2004) to illustrate the impact of rescheduling delayed tasks was simulated. The results obtained show that the scheduling algorithm developed here is able to generate initial solutions that might prove better than those reported in Bhushan and Karimi (2004); in particular, scheduling the 25 jobs reported in one of the examples required 519.6 time units using the Bhushan's algorithm against the 501.5 units estimated by the scheduling algorithm illustrated in this study, thus obtaining 3.5% makespan reduction before the implementation of sequencing optimization approaches. The scheduling algorithm has also shown good performance in terms of computational times as fully detailed batch schedules are generated in about 0.01 seconds.…”

“…Less rigid assumptions, especially those regarding the material handling system, are also considered in more recent models (Aguirre, Méndez, and Castro 2011;Lee, Lee, and Lee 2007). Tabu search is used by Geiger, Kempf, and Uzsoy (1997); whereas Bhushan and Karimi (2004) develop a simulated annealing based approach to address larger instances of the scheduling and sequencing problem at wet stations with one robot. Bhushan and Karimi (2003) also adopt a Mixed Integer Linear Programming (MILP) approach to generate optimal schedules at a wet station; based on the MILP approach, a two-step heuristic approach is developed.…”

Applying Milp/Heuristic Algorithms to Automated Job-Shop Scheduling Problems in Aircraft-Part Manufacturing

“…In the past, simple heuristic procedures were used to provide a primary solution, far to the optimal one, for this kind of problems, when full-space methods had become untreatable for solving industrial examples, due to the high number of decisions involved in the model. In the other hand, simple heuristic methods, like two-stage approaches (BHUSHAN; KARIMI, 2004), are difficult to implement when sequencing decisions of both stages are strongly linked. Thus, any changed in one stage's decisions could turn the problem infeasible if other decisions are not carefully revised.…”