Business Process Simulation with Differentiated Resources

Abstract

Business process simulation is an approach that allows us to perform the "what-if" analysis. With its help, we can analyse the current business process, manually find possible improvements, introduce them, and predict the impact of those changes by running a simulation. Simulation tools take business process models as input, accompanied by the additional details required for simulation, such as resource availability. Yet existing simulation tools support only basic process elements, among which are activities and decision points (gateways). In real life, however, a resource can perform activities not straight after enabling time but instead waiting for a group of activities to gather and then execute them in one go (batch processing), or a resource can prioritise one task over another when both of them are waiting for the execution (task prioritisation). Additionally, process simulation might benefit from introducing events to model various behaviour, for example, setting up a timer for 2 hours or interacting with external entities like calling a client or receiving a message from a client. We call these types of events intermediate events as they happen during a process. This thesis contributes to implementing those concepts above (batch processing, task prioritisation, intermediate events) based on the already implemented simulation engine with differentiated resources. Furthermore, the simulation engine we use as a basis, named Prosimos, does not have a web interface and can only be executed through the command line interface. This, in turn, has limited the adoption of Prosimos in practice. With this thesis, we also aim to diminish the knowledge requirement and allow people with no technical background, like business analysts, to utilise the tool. To achieve this, we implement a brand-new web application from ground up. During the development process, we write unit and integration tests, following the decision table testing approach, to verify the implementation continuously. For evaluation, we analyse the scalability of the newly introduced concepts. The results of this master’s thesis were already partially published as a demo paper.