A combinatorial approach to a class of parallel-machine, continuous-time scheduling problems

Konstantin Kogan, Eugene Khmelnitsky, Eugene Levner

Research output: Contribution to journalArticlepeer-review


The paper analyzes a manufacturing system with N non-identical, parallel machines continuously producing one product type in response to its demand. Inventory and backlog costs are incurred when tracking the demand results in inventory surpluses and shortages respectively. In addition, the production cost of a machine is incurred when the machine is not idle. The objective is to determine machine production rates so that the inventory, backlog, and production costs are minimized. For problems with demand defined as an arbitrary function of lime, numerical methods are suggested to approximate an optimal solution. The complexity of the approximation methods is polynomial, while finding an exact optimal solution requires exponential time. In a case when production is to cope with a special form of a single-mode, >K-level piece-wise constant demand, we prove, with the aid of the maximum principle, that the exact optimal solution can be found as a combination of analytical and combinatorial tools in O(>KN 2(mnx{>K,2N})2) time.

Original languageEnglish
Pages (from-to)223-231
Number of pages9
JournalIIE Transactions (Institute of Industrial Engineers)
Issue number3
StatePublished - 2002
Externally publishedYes


Dive into the research topics of 'A combinatorial approach to a class of parallel-machine, continuous-time scheduling problems'. Together they form a unique fingerprint.

Cite this