Flexible Behaviour of Human Actors in Distributed Workflows

Abstract

Workflows often require participation of technical components as well as humans to achieve their objectives. In order to accurately model and simulate the flexibility of human and technical resources we have to consider diverse aspects such as (dynamic) (re)assignment, deadlines, escalation handling, faults and retrieval (backtracking). In order to meet these requirements we will represent configurations of a business process by graphs and operations changing these configurations by graph transformation rules. We will propose a domain-specific notation to visualise rules and configurations. Adding stochastic time, we can simulate process models to assess their performance. In a distributed setting, workflows can balance the load between different organisations, thus increasing flexibility, performance and reliability. We will use simulations to compare non-distributed and distributed versions of a system realising workflows in dynamic commercial environment. We will illustrate and validate the approach by means of a pharmacy case study.