Visualizing and analyzing dynamic business process using Petri nets

Abstract

Business processes are modelled, visualized and analyzed using process maps, and implemented using productivity suites such as MS Visio, Business Process Extraction Language (BPEL), and some other productivity suites. However, the problem with such current process models is that it is static. Business process tasks modelled using non-interactive visualisation tool does not allow active engagement of the analyst or users developing the process maps with the process maps to determine what if scenarios. The constant changes to existing business processes are enforced due to several external and internal factors with an objective of improving process efficiency and effectiveness. So, we developed a Petri net process model to visualise business process tasks and allow interaction with the process model to allow the user or analyst to compute from an ‘as-is’ business process model to a ‘to be’ business process model that would optimise both the time it takes to complete the process (process duration) and the cost of executing the business process tasks (process costs). Petri nets whilst a form of visualisation of tasks also enables mathematical modelling of the process model. Methodology: We applied mathematical combination theory, combined with Petri nets formalism to create design options in generating alternative business process options with different process duration and cost factors to enable the process modeller or analyst make a better decision. We visualized and evaluated different dynamic ‘transitions’ sequence generated based on best duration and cost, modelled the ‘design options’ in the Platform Independent Petri Net editor (PIPE2) tool and performed further analysis of the Petri net properties of the modelled processes. Finally, we discussed how to visualize and analyze ‘design options’ and provided certain views on business processes. Results: We demonstrated through our research that Petri nets could be used in visualising different process tasks as and users can interact with the process model. This was implemented using a case study of a supermarket chain business process. This research helped in identifying six out of the eleven business process ‘design options’ generated are best optimised each with process duration of 28 hours and process cost of 35.57 dollars. However, further qualitative analysis of these optimised ‘design options’ effectively resulted in only one out of the eleven as the most optimal process design indicating the selected revised business process. The visualization is helpful in analyzing dynamic transitions for exploring and understanding complex behaviour in the business processes. The results suggest that the combination of mathematical approach in conjunction with the analytical properties of Petri Nets could be used to generate unbiased alternative business process designs that can be further evaluated by the business process re-engineer or analyst in order to revise an existing business process.