Business process improvement using multi-objective optimisation

• Published in: BT technology journal Vol. 24; no. 2; pp. 229 - 235
• Main Authors: Vergidis, K., Tiwari, A., Majeed, B.
• Format: Journal Article
• Language: English
• Published: Ipswich British Telecommunications PLC 01.04.2006
• Subjects: Algorithms; Automation; Business process reengineering; Competitive advantage; Design optimization; Information sources; Mathematical models; Objectives; Optimization algorithms; Quantitative analysis; Research methodology; Studies; Variables
• ISSN: 1358-3948; 1573-1995
• DOI: 10.1007/s10550-006-0065-2

Business process redesign and improvement has become an increasingly attractive subject in the wider area of business process intelligence. Although there have been many attempts to establish a business process redesign framework, there is little work on the actual optimisation of business processes with given objectives. Furthermore, most of the attempts to optimise a business process are manual and do not involve a formal automated methodology. This paper proposes a process improvement approach for automated multi-objective optimisation of business processes. The proposed framework uses a generic business process model that is formally defined. The formal definition of business processes is necessary to ensure that the optimisation will take place in a clearly defined, repeatable and verifiable way. Multi-objectivity is expressed in terms of process cost and duration as two key objectives for any business process. The business process model is programmed and incorporated into a software optimisation platform where a selection of multi-objective optimisation algorithms can be applied to a business process design. This paper outlines a case study of business process design that is optimised by the state-of-the-art multi-objective optimisation algorithm NSGA2. The results indicate that, although business process optimisation is a highly constrained problem with fragmented search space, a number of alternative optimised business processes that meet the optimisation criteria can be produced. The paper also provides directions for future research in this area. [PUBLICATION ABSTRACT]