Simulation of Agent-Based and Discrete Event for Analyzing Multi Organizational Performance

Abstract

Business process simulation is an interesting research for field of business process management and process mining. Through simulation, the parameters in the business process can be changed to determine the optimal simulation performance and evaluating existing business processes to improve performance for future business processes. Simulation in multiple organizations can be used to determine the performance of multiple organizations based on event log recorded on the organization's system. This study proposes simulations on multiple organizations using agent-based simulation and discrete event simulation. The multi-organizational example used in this research is port container terminal. In addition, this study also contributes with the establishment of event log in the future based on the results of forecasting the number of containers. This research proposes how to generate traces and event logs based on forecasting result of container number. This research resulted evaluation of business process performance in multi organization. The results show that the dwelling time in the existing event log and forecasted log exceeds the limit set by the government, which is a maximum of 3 days. Whereas in the event log the average dwelling time is 9.7 days based on agent-based simulations and 5.4 days of dwelling time using discrete-event simulation. Referring to the comparison of the two simulations, it is found that the difference in dwelling time in agent-based simulation and discrete-event simulation is influenced by the presence of asynchronous messages in agent-based simulation, resulting in longer dwelling time.