Simscript/Data Analysis Assignment
Due Thursday, March 24, 2005
In this assignment you will use Simscript models to answer some questions
about systems.
Telephone Problem (#1.29 pages 104-105)
The file phonesAB.sim contains a Simscript
program that models the telephone system described in problem 1.29 of
the text (and discussed in class). Copy this program to your directory
either by saving it from this page or using the cp command (the
complete pathname of the file is
~cpsc/public_html/Spring2005/CPSC420A/phonesAB.sim
Notice that the program is written with a loop in main so that it
can be run for several different numbers of phone lines between
cities A and B (the user enters a min and a max number).
- Compile and run the program using the system parameters given in
the problem description (note that the problem gives the interarrival
times for calls in seconds but the program asks for the
arrival rate per minute - so for the interarrival time of 10
seconds, the arrival rate would be 6 per minute). Run the program
for between 30 and 35 lines.
- You should note that the answers above do not meet the goal of no
more than 5% blocked calls. Try other numbers until you seem to get
numbers in the right range.
- Modify the program by adding an inner loop in main to do replications.
Also add code to write the results to a file - you should get a new
file for each number of lines (so the results for say 30 lines would
all go in one file, for 31 in another, etc.). A model to use is the
mmsexp.sim example we did in class (you should have a handout - this
simulated a system with one queue but multiple servers).
- Run the program for at least 5 different numbers of lines in
the range that you think contains the best solution (the smallest
number of lines that meets the 5% goal). For each number
of lines do at least 10 replications.
- Use Minitab to analyze the data from each number of lines.
- Use your Minitab analysis to determine the smallest number of lines
that meets the 5% goal. Give reasons for your answer.
A Variation on the Machine Problem (#1.22 on page 102)
In this variation, the manufacturing system contains two different types
of machines AND the machines have two different types of breakdowns (minor
and major). The differences between the two types of machines (Type I and
Type II) are as
follows:
- Type II machines run longer before breaking down. The runtimes
before breaking down for both types of machines are exponentially
distributed but the means are different. (The numbers will be
provided later - they do not affect the logic and should not be
hardcoded into the program!)
- Type II machines cost more per hour of downtime. (Numbers will be
provided later.)
Both types of machines have the same repair characteristics. Both
experience both ordinary breakdowns and major breakdowns. A major
breakdown requires two repair workers to be working on
it simultaneously ( you can request 2 units of a resource - be sure
to also relinquish 2). Fifteen
percent of the breakdowns are major. The repair times for ordinary
breakdowns are exponentially distributed (with a mean to be determined)
and the major ones are uniformly distributed (between two numbers).
The SIMSCRIPT program to model this system should be written from
the above information (no system parameters such as means or numbers
of machines of each type or numbers of repair workers hardcoded -
these numbers will be provided later). In the program, have a
MACHINE process with an attribute for the machine type. As in the
Quarry example, use arrays for the values that depend on the
type of machine. (NOTE: Unfortunately Simscript does not let you
tally or accumulate on an array variable. You will need to sum
your own downtime for each machine type.)
For experimental purposes,
the program should be written with a loop in main so replications
can be automatically performed and data written to a file for
easy analysis.
Be sure to document your program appropriately and have clear
prompts. The program should compute the following performance
measures:
- Cost per hour
- Utilization of each repair worker
- Average length of the machine queue
- Average delay in the queue
Goal of the Simulation: The current system has 5 Type I machines
and 2 Type II machines. The owner plans to purchase 3 new machines and
wants to know which types should be purchased (all type I, all type II,
or 2 of one and one of the other type) to minimize cost and whether or
not he/she needs to hire an additional repair worker (that could also
affect the minimum cost).
System Parameters: Click here
HAND IN: Copies of the programs and the data analysis files
for each problem. A written analysis for each with your conclusions
and justification for those conclusions.