Working with Model Data / Monte Carlo Methods

7.6

Simio Clinic Model

The most important elements to setting this simulation up, are the following elements:

  • A Data Table with Date Time property at 10 minute intervals between 8 AM and 12PM. An experiment was run from 8AM until 12:30.

  • A Source Arrival Mode referenced to the aforementioned data table’s DateTimeProperty, with an Arrival Time Deviation of Random.Triangular(-15,0,30), and a no-show probability (or second sink) of 10%.

Initial Model

Initial Model

We see the results: 10% missing, 24 total, and an 76.95% doctor utilization (3.46 total hours worked).

Now, if we were to run an experiment with 2-3 arrivals every 20 minutes, and with doctor willing to stay late, what are the results? Accomplishing this required:

  • an updated Work Schedule - extended to 1PM
  • a new Test Arrivals dateTime data table with 20 minute intervals
  • Random.Uniform(2,3) & Random.Triangular(-15,0,30) data stream configurations
  • A state statistic assigned to the Doctor server in order to tabulate the final TimeNow statistic upon exiting the server.
State stat

State stat

In a non-experimental approach, we see a randomly more efficient number of patients seen, 23, and an improvement in the P value. The doctor only stays .133 hours after noon, and the wait times are about 10 minutes. Experimentally, these values are similar, with (after 10 replications) 9.9 minute wait times.

Changed Model 1

Changed Model 1