Benchmarks Library ballred.gif (861 bytes) Multiproduct Scheduling Problem
line.jpg (1129 bytes)

Benchmarks

ballred.gif (861 bytes)  Multiproduct Scheduling Problem 

ballred.gif (861 bytes)  Benchmarks

Format of the data

The test instances are given in text files. The format of the data is the following:

States NUMBER_OF_STATES (PRODUCTS)

list of states names

Demands

list of demands

Units NUMBER_OF_UNITS

list of unit names

Tasks NUMBER_OF_TASKS

list of tasks names

Task_Unit_State_MinAmount_MinTime

list of tasks characteristics: task name, suitable unit, produced state, minimal amount, minimal time (Tmin)

Changeovers

list of changeover times

For each task, the production rate can be obtained as MinAmount / MinTime.

Series "Random"

The problem instances were generated randomly with the following parameters: DS ∈ [100,200], sij ∈ [1,10], ri ∈ [1,20]. For each task, the suitable unit was assigned at random. The correspondence of tasks and products was set in such a way that every product is produced by at least one task: tasks with the indices i < |S| were assigned to products with the same number s = i; for the other tasks the product was assigned at random. The instances can be downloaded as a zip archive rand.zip.

The following table shows the results obtained by the genetic algorithm.

Instance Cmax ch.time proc.time
rnd30 281 48 515
rnd40 291 57 526
rnd50 234 75 627
rnd60 295 92 793
rnd150 281 290 2527
rnd300 502 1026 3994

Series "Real"

The "Real" instances represent the set of subproblems arised when solving the real industrial problem using the hybrid decomposition method (see [1]). The instances can be downloaded as a zip archive real.zip. The following table shows the best Cmax values obtained by different algorithms [2].

Instance Greedy GA CPLEX
r55x10 222.86 224.167 224.17
r60x10 241.02 238.97 238.95
r78x10 456.57 395.3 414.8
r84x10 456.57 401.5 422.18
r109x10 514.24 447.86 468.44
r115x10 574.74 484.96 508.87
r120x10 574.74 480.73 536.13

References

[1] Borisovsky, P.A., Eremeev, A.V., and Kallrath, J. "Multi-Product Continuous Plant Scheduling: Combination of Decomposition, Genetic Algorithm, and Constructive Heuristic." Submitted to International Journal of Production Research.

[2] Dolgui, A., Eremeev, A.V., Kovalyov, M.Y., and Kuznetsov, P.M. 2010. "Multi-Product Lot Sizing and Scheduling on Unrelated Parallel Machines." IIE Transactions 42(7): 514-524.