In this problem we will compare a M/M/1 queue with arrival rate, \(\lambda\)=120/hour and service rate, \(\mu\)=190/hour. We are interested in the following steady stae measures: time entities spent in system and expected number of entities processed in 100 hours.
W denotes the steady-state average time in the system and \(W=\frac { 1 }{ \mu -\lambda }\)
We will determine the number of expected entities processed by dividing the 100 hours by W and truncating to a round number.
Calculation will be done in R.
l <- 120
m <- 190
W <- 1/(m-l)
Entities <- floor(100*l)
W
## [1] 0.01428571
Entities
## [1] 12000
We will compare with result of simulation ran for 100 hours.
Result of simulation we have Average time in system = 0.0145 and number of entities processed = 11935 in 100 hours of simulations.
m/m/1 100 hours simulation run
Please note that when we ran model and capature statistic, we did not account for warm-up period.
Using the model previously set-up, we ran experiment of 100 repetitions. We set-up warm-up time to be 50 hours.
m/m/1 SMORE example
We will now build a simio model m/m/1 using processes instead of objects from standard library.
m/m/1 Model built using processes
Running the simulation for 50 repetitions of 100 hours.
m/m/1 Model built using processes Results
The results are similar than the one obtained by building the model using standar library objects.
m/m/1 Model Animation