Comparison of discrete event simulation and agent based simulation for evaluating the performance of port container terminal

摘要

Event log obtained from Port Container Terminal (PCT) Surabaya is an asynchronous event log. This event log needs to be run in a simulation to reflect the real world performance which contains both time and cost. From the event log we gathered, we use forecast methods to predict the number of container for the following month. Several forecasting methods are evaluated; whereas discrete event simulation and agent based simulation are compared to handle asynchronous processes. The results of the experiments show that moving average have the lowest MSE compared to other forecast methods such as Simple Exponential Smoothing, Double Exponential Smoothing, and Linear Regression. Then, from the forecast results we successfully generate the event log for the following month and simulate it using agent based simulation and Discrete Event Simulation. The results of the simulation show that agent based simulation can handle the communication process which discrete event simulation cannot handle. Both the simulation results are depicted in Gantt charts.