TR169: Decision Making for Schedule Optimization

This paper presents a novel formulation of scheduling and decision information which allows a concurrent optimization of both, the decisions leading to a specific schedule and the schedule itself. Our methodology allows a dynamic adaptation of the optimization criteria according to the quality measurement criteria of the involved decision making stakeholders. Major types of possible quality measurement criteria are project duration considerations, cost considerations, resource levelling considerations, safety considerations and some miscellaneous considerations like distances resources have to cover from one assignment to the next or time space conflicts of resources.
Decisions and their alternatives are represented in a Decision Breakdown Structure (DBS) (Kam, 2006). The DBS defines the search space for the optimization algorithm which is based on a Genetic Algorithm (GA) approach. The optimization algorithm uses the novel formulation of scheduling and decision information to find a Pareto optimal decision alternative combination which leads to a Pareto optimal schedule.

First tests of the decision and schedule optimization algorithm show that optimizations can be performed within one minute. This short latency suggests that the proposed concepts about decision optimization could, for instance, be utilized in meetings or in an Integrated Concurrent Engineering (ICE) environment where short latency is extremely important (Chachere, 2004) because stakeholders need to get a quick idea about good decisions and their predicted outcome.