How to model business processes with GPN

Abstract

Organizations today face increasing pressure to reduce time to market, i.e. to improve the design and the operations of business processes in terms of lead time and meeting due dates. Formal analysis using a mathematical graph-based approach can help to achieve this kind of improvement. We will apply business graphs to scheduling workflows in terms of time-based optimization. We will concentrate on performance measures like completion time, flow time and tardiness. From a business process network we derive two types of directed graphs, one representing the task net (task graph) and the other one representing the resource net (resource graph). In the task graph a node is representing a task and its duration and arcs are representing different kinds of precedence constraints between tasks. The resource graph is similar to a Petri net and represents resource constraints and flows of jobs. In order to compute optimal or near-optimal workflow schedules the algorithms have to relate to the structure of the business graphs. We will show that a variety of data structures commonly assumed in modern scheduling theory can be represented within the framework of business graphs. Based on these data structures specific scheduling algorithms to optimize time-based performance measures can be applied with the objective to reduce time to market.