3.2 Statistics on Scheduling Problems and Performance Measures

3.2.1 Scheduling Problems

Table 3 shows the distribution of scheduling problems that are involved under this category. As for the general FMS schedul- ing studies, the parts dispatching was a popular item, which attracted the attention  of many  authors.

3.2.2 Performance Measures

Both flow-time and tardiness related measures were employed most frequently as  the  performance  measures.  This  is shown in  Table 4.

Table 3. Scheduling problem in multi-criteria scheduling    approaches.

Scheduling Number of Reference Period

problem publications number of

publications

Chryssolouris et al. [59] tested a modular system that  treated a production scheduling problem as a multi-criteria decision- making issue using a simulation model. The developed method attempted  to  model  the  decision-making  process,  at the work

  centre  level,  as  a  multi-criteria  decision  problem.  The simul-

Parts  dispatching  8 [59],  [83],  [85– 1985–1999

89], [91]

Machine  selection 1 [84] 1990

AGV  scheduling   3 [84],  [88],  [90]  1990–1996

Table 4. Performance measures used in multi-criteria    approaches.

ation model consisted of two different work centre configur- ations (i.e. one work centre, and five work centres) and com- pared the rule-based method with four dispatching rules, which were LCFS, FCFS, SPT, and LPT. The employed performance measures were average flow-time, WIP, number of jobs com- pleted, and mean tardiness. The results showed that  for one work centre, the SPT rule performed well, and in the case of

  five work centres, the rule outperformed all other dispatching

Performance Number of Reference Period measures publications number of

publications

Flow-time  related 5 [59],  [84],  [86], 1988–1996

[88], [90]

Tardiness or tardy  5 [59],  [84],  [86], 1988–1999

job related [87], [91]

Utilisation  related 3 [83],  [84],  [87]  1985–1994 (including system,

machine, station, etc.)

Cost related 2 [87], [91] 1994  and 1999

Inventory  related  1 [59] 1988

Others 4 [59],  [83],  [87], 1985–1996

[90]

3.3 Review of  Related Publications

Shanker and Tzen [83] found that random FMSs require bal- anced workloads to operate effectively. They studied a schedul- ing problem in an FMS that was considered to be a composite of two independent tasks: loading and sequencing. Formulations were  presented for  the  loading problem  with  two objectives: