CIS461-DL                  Dr. J. Scher              GPSS Programming Assignments

PART I: Three 'Simple' GPSS Programs

READ: Chapters 1, 2 and 3 in text.

VIEW: Videotapes 16, 17 and 18

For your first GPSS/H programming assignment, I will ask you to write and run GPSS/H programs to solve three 'simple' problems given at the conclusion of Chapter 3 in the textbook. As previously mentioned, you may submit these any time up to the cutoff date, which will be near the end of April.

Use careful in-line documentation for each block in your GPSS/H source program. Include in your submission of each problem the .gps file and the .lis file, and an external page indicating your answers to the particular question (specify clearly how you arrived at your answer, highlighting specific components of your output file).

The three problems I will ask you to implement in GPSS/H are as follows:

1) Page 3-25, Problem 3, parts 'a' thru 'd' with the following modifications: assume that the time for conveyance between the first work station (where the 'inspection' takes place) and the second work station (where the 'setup' and 'assembly' takes place) is exactly 15 seconds, and that after the second work station, a conveyor takes the assembled parts (in precisely 27 seconds) to a third work station, staffed by one person, where a quality control check is performed which will take between 15 and 27 seconds.

i) Note that a 'resource,' as mentioned in the problem, is, in the GPSS sense, as defined on page 2-4 of the text, a "static entity in the system," as opposed to transactions which are, of course, the dynamic entities of the system.

ii) Rather than simulating for 500 parts, simulate the system for 60,000 parts.

2) Page 3-25 thru 3-26, problem 4 (again, repeat parts 'a' thru 'd' for your modified model).

3) Page 3-25, problem 7 (respond to both of the implicit parts of this problem).

i) Rather than simulate for 250 hours, simulate the system for 18,000 hours.

ii) The first part of the problem asks you to consider each operation individually, i.e., three separate "ADVANCE" blocks. The second part questions whether you can combine the three operations into a single ADVANCE block. (Mathematically speaking, you are being asked to empirically consider the issue of the convolution of three uniformly distributed random variables, and whether the result will also be a uniformly distributed random variable.) If your simulations show that there is a non-trivial difference in the results, either intuitively or mathematically explain why.

PART II: Capturing Waiting Time Performance Measures in GPSS

READ : Chapters 5 in text (presumes you have previously read Chapters 1, 2 and 3)

VIEW: Videotapes 19 and 20 (presumes you have also previously viewed GPSS videotapes 16 through 18)


                                 Simulation of a Simple Job Shop

On page 5-16 in the text, read the description of the simple job shop in problem 1. Note that according to the specifications of the problem, a single worker is required to spend processing time on each widget after it is placed in the jig. There is only one jig. The same worker is also required to spend processing time on the gadgets, but the gadgets do not require a jig. If a unit (widget or gadget) requires the worker, and the worker is occupied, then the unit cannot move any further, and must wait until the worker becomes available.. 

Assume that this job shop system is continuously processing, and there are no 'down-times' and the single worker is refreshed by a new worker (in zero time). 

SUBMIT AS FOLLOWS:

You will design a GPSS simulation study which will simulate this simplified Job Shop system for 500 days, to answer the systems analysis questions described in the textbook and below.

In addition to answering each of the questions a, b and c in the text, answer the following:

d) identify (in words) each point of congestion in this queueing system, where waiting may occur

e) what was the average waiting time and the maximum queue size at each point of congestion specified in (i) 

f) Suppose that the arrival rate of gadgets is subject to change, and we wish to study the behavior of this queueing system subject to these changes. 

A) Specifically, first suppose the average arrival rate of both types of arriving units increases by 15%, i.e., rather than an average  widget arrival rate of 2 per hour suppose it changes to 2.3 per hour, and similarly for gadgets, and also suppose that the 'spread' for widgets decreases by 15% (i.e., the 'spread'  will now vary between 0 and 25.5 minutes) Rewrite/rerun your GPSS model to study the behavior of this system in terms of this change in 'load,' and complete the answers to a,b,c,d and e. Do widgets have to wait longer with the increased arrival rate, and, if so, by how much on average?

B) Suppose next that the interarrival times INCREASE by 10%, i.e., the widgets arrive according to a uniform distribution with a mean of 33 minutes and a spread of 33 minutes, while gadgets have a constant mean interrarival time of 33 minutes. Again, modify your GPSS model and study the resultant behavior of the system, responding to each of the questions. Do widgets have to wait longer with the increased interarrival times, and, if so, by how much on average?

For your GPSS model(s), include in-line documentation on each line of your GPSS code. For answering the questions in each simulation study, you need to submit EXTERNAL documentation and to specify clearly which performance measure you used from your GPSS .lis file to answer the specific question (also, highlight in your GPSS output the performance measures you used, and specify which question they were used to answer).

You need to submit hardcopies of your GPSS source and output (.lis) files.