Modeling Time-Varied Arrival Rates: An Application Issue in Queuing Systems

This paper addresses the problem of modeling time-varied arrival rates in gicetcing systems. The approach is to break down a 24-hour period into small time durations such that an average arrival rate during the time duration can be reasonably viewed as constant. Discrete event simulation models were developed using SLAM II to evaluate the selection of the length of time durations. The simulation output data were then compared to real data collected from the Emergency Ambulance Service Center in Shanghai, China, a typical system with time-varied arrivals of service requests. Statistical data analyses on the system performance indicated that the 4-hour duration model with six different arrival rates in 24 hours closely matches the operation characteristics of the Shanghai system.