Parametric Sweeps

$1.$ What are the parameters for the simulation?

const N$\_$TELLERS $=2$

const MEAN$\_$INTERARRIVAL$\_$TIME $=4\mathrm{.}0$

const MIN$\_$SERVICE$\_$TIME $=2\mathrm{.}0$

const MAX$\_$SERVICE$\_$TIME $=10\mathrm{.}0$

The fact that mean interarrival times are exponentially distributed is built into the code.

Interarrival times by their nature are exponentially distributed for independent arrivals. We

have no reason to believe the arrivals are not independent.

The fact that service times $($with a teller$)$ are uniformly distributed is built into the code.

We assumed a uniform distribution for the service times based on insufficient information to

determine the actual distribution of service times. An independent study to determine the best

distribution should be undertaken. But we will use a random distribution for this study.

$2.$ What are the viable ranges for the parameters?

Parameter Range

N$\_$TELLERS Integer $>=1$

MEAN$\_$INTERARRIVAL$\_$TIME Real $>0\mathrm{.}0$

MIN$\_$SERVICE$\_$TIME Real $>0\mathrm{.}0$

MAX$\_$SERVICE$\_$TIME Real $>=$ MIN$\_$SERVICE$\_$TIME

$3.$ What specific values do you want to change?

The four parameters in #$2.$ But, the minimum and maximum service times have been validated by

the bank and will not be in the parameter sweep.

$4.$ How will you change them?

All the parameters are numeric, and we will specify the values.

We will specify interarrival times as varying from $1\mathrm{.}0$ to $8\mathrm{.}0$ by $1\mathrm{.}0.$

The simulation is stochastic, and we must run at least $1,000$ runs to determine the distributions of

the output parameters. This means at least $5$ x $8$ x $1,000=40,000$ individual runs. If each individual

run takes $1$ second, this requires slightly over $11$ minutes to run.

Parameter Value Range

N$\_$TELLERS $2$ to $6$

MEAN$\_$INTERARRIVAL$\_$TIME $1\mathrm{.}0$ to $8\mathrm{.}0$ by $1\mathrm{.}0$