Random Variation in a Model


Nonhomogeneous Poisson Process

Subsections:

There are many systems in which the arrival rate of entities depends strongly on time (for example, arrivals of patients at an emergency room and the arrival of calls at a customer service center). In Simulation Studio, a nonhomogeneous Poisson process (NHPP) based on either count or rate data can be used to generate a time-dependent arrival process on the time interval $(0,S]$.

For either the count-based or rate-based case, a Numeric Source block with the NHPP option specified under the Data Driven option can be connected to the InterArrival time port of an Entity Generator block. In the Entity Generator block, the option At First Interarrival Time for First Entity Creation should be selected to ensure that the first arrival time is an actual event time for the NHPP. If either the At Start Time or After Signal Arrival option is selected, the first entity is generated at the start time specified in the Entity Generator block or the signal time and is not an arrival (event) time generated from the specified NHPP. However, all subsequent arrival times are generated from the NHPP.

Also, the start time of the NHPP might not correspond to the start time of the simulation. For example, suppose an NHPP is defined on the time interval from noon to 2:00 p.m., but the simulation start time is 8:00 a.m. In this case, if the time unit is in hours, an extra subinterval from 8:00 a.m. to noon with count (or rate) 0 could be added so that the NHPP is defined on the time interval $(0,6]$. No events are generated for the time interval $(0,4]$ and the process effectively begins at simulation time 4 (which corresponds to noon).

After the simulation clock reaches time S, the entity generator block associated with the NHPP shuts down and no more entities are generated. If necessary, a signal can be sent to the Entity Generator block to generate an entity at a specific time after S.