Hi Mohammad,

The variance between the results of queuing models and simulation ones are normal especially for one replication. There are some points that may explain the variance source:

1. The used distribution: for example in M/M/1 the arrival and service follow Poisson and exponential distribution. So, you should use the same distributions in the simulation model.

2. Even if you used the same distributions, the simulation model results are based on random generations of the time between arrivals and service time. On the other hand, queuing models depend on average only. So, for one replication results will be definitely different.

3. The number of replications plays a crucial rule in this issue. The larger the number of replications the closer the results will be. Just try to run 10, 20, 30 replicates and I think the results will converge.

Hope my answer may help..

Best,

Zak

Hi sir,

Simulation enables the study of, and experimentation with, the interactions of a complex system (or a subsystem thereof). Informational, organisational, and environmental changes can be simulated and the changes to the model’ s behaviour can be observed. Knowledge gained by studying the model can be of great value , and may suggest improvement to the system under investigation. Changing simulation inputs can help understanding how certain components of the system interact.

http://www.macs.hw.ac.uk/~jphillips/DAS/DataAnalysisTopic8.pdf