You are on page 1of 6

AMMAR and WRIGHT

A Demonstration of a Push/Pull Assembly Line

workstation (WS) production process. The end


product is a carryout tray containing four cups with
lids, straws, and a colored dot (see Figure 1). The
worker in the first WS puts four cups in a tray. The
worker in the second WS places red dots on the cups.
The worker in the third WS puts lids on the cups.
Finally the worker in the fourth WS removes the
wrappers from straws and puts the straws in the holes
of the lids. The last workstation contains the task
with the longest time and is the expected bottleneck
of the assembly line. The process begins with an
inventory of four trays between each station. The
workers are told to produce continuously and to pass
the product down the production line (the push
system). Before long the trays of cup with lids will
begin to pile up in front of the last station. A marked
tray can be inserted into the process and the time
required to work its way to the end can be recorded.
After about twenty minutes of production the process
is stopped and a number of observations are made.
The most obvious is the build up of work-in-process
(WIP) at each station but most prominently in front
of the fourth station. The total WIP is one measure
of the cost of the operation. Another measure is the
time required for one item to work its way from the
first through the last station (turnover time.) The
space required (recorded in tables needed) to hold
the large WIP's is also a measure of cost. After a
discussion of the effectiveness of the process, it is
repeated to simulate a pull system. At each
workstation a WIP area is marked with tape that is
just large enough to hold four units. Each worker is
told to only produce when the WIP area is empty.
This simulates a KANBAN system in which
production takes place only as it is required. When
this system is executed the WIP between each station
is always less than four. The timepiece will work its
way through the system much more rapidly. Less
space is required. Hence the costs associated with
the production have been reduced while the
production rate has remained the same.

A Demonstration of Push/Pull
Assembly Line
Salwa Ammar and Ronald Wright
Le Moyne College
Syracuse, NY 13214
ammar@mail.lemoyne.edu
wright@mail.lemoyne.edu

Introduction
In the introduction to operations management course,
both at the MBA and undergraduate level, we use a
variety of demonstrations and games to illustrate key
concepts of production and operation (Ammar and
Wright, 1999). One particular topic, which is the
subject of this demonstration, is the difference
between push and pull production. Although it is
easy to describe the two systems of production, the
benefits of the pull system are hard to motivate
especially to undergraduate students with limited
exposure to assembly line manufacturing. In our
experience, an in-class demonstration of the two
systems allows for a better discussion and ultimately
a better understanding of the benefits. One very
effective demonstration is the cups game designed
by Peter Jackson (2). It illustrates clearly the
difference between push and pull production and
serves as a very stimulating vehicle for discussion
of the relevant concepts.

The demonstration is typically very effective. The


image of the WIP inventory piling up higher than
the workers in the push environment and the orderly
constant four-unit WIP for the pull system clearly
makes the desired point. The cups game however
requires a considerable supply of material. The

Figure 1: Cups Game Unit of Product

The cups game is a demonstration of a four

XX
INFORMS Transcations on Education 2:3 (XX-XX)

INFORMS

AMMAR and WRIGHT


A Demonstration of a Push/Pull Assembly Line

expense, availability and mere weight of the material


make setting up the demonstration a daunting task.
Also, the demonstration itself involves only four
participants leaving the rest of the students acting as
interested observers. This may not be a problem for
a class of motivated students, but we find it to be an
issue especially in our undergraduate classes. These
students particularly lack the experience and insight
and therefore can benefit most from participating in
the demonstration. Yet as observers, some of them
recognize that this can't possibly be on the exam and
find it easy to disconnect. Finally the effectiveness
of the demonstration depends on the instructor to be
a skillful moderator and to some extent an able
performer. This leaves some instructors, who
otherwise believe in their value, reluctant to conduct
such demonstrations.

observe. This demonstration although perhaps less


dramatic, still provides a very effective and easily
transferable tool to introduce the push-pull systems.
Assembly Line Demonstration
Objective
The objective of this demonstration is to illustrate
the difference between push and pull systems of
production. The push system requires each WS to
maximize its own utilization and produce as much
as possible while pushing WIP down the line. This
result in accumulation of WIP and the associated
added costs of production. The pull system requires
each workstation to produce as much as needed by
the next workstation. This is usually controlled using
the KANBAN system which keeps track of the WIP
of the following WS and when it needs to be
replenished. This of course controls the WIP
between workstations and eliminates the associated
additional cost. Comparing and contrasting these two
systems is the main objective of this demonstration.
This is achieved by first demonstrating the push
system, observing the accumulated WIP, discussing
the consequences and possible costs, and
investigating options to address these consequences.
The next step is to describe the KANBAN system
as a possible solution to controlling the large levels
of WIP. Then the pull demonstration with
KANBANS is conducted followed by a discussion
on the advantages and disadvantages of the two
production systems. Much like in the cups game,
the discussion is based on several observed measures
for each demonstration. These include WIP levels,
output rate, space needed for WIP, unit turnover time,
and worker utilization.

The cups game can be easily simulated in a computer


program or even a spreadsheet with either
deterministic or random task times. The program can
track the important statistics such as WIP, turnover
time, and required space and can even provide visual
representations in the form of pictures or graphs. This
will eliminate the need for the bulky material or the
time consuming set up. In our opinion, however, it
also reduces the effectiveness of the demonstration.
The physical staging of an assembly line and the
passing of products from one WS to the next provide
students with a production experience that they can
and do remember. Also rather than being told how
many units are in the WIP, seeing and counting the
accumulation reinforces the impact.
The demonstration described in this paper is a
compromise between the two approaches. It
maintains the physical staging of the assembly line
and the manipulation of the product down the line.
The product (in this case a penny) is pushed or pulled
down a four-station assembly line. At the same time
the work performed at each workstation is simulated
in a computer program using synchronized timers.
This eliminates the need for bulky material to
perform actual tasks. It contains the 'show' aspect
that is stimulating enough to maintain the interest
and focus of the participants. It also allows for
multiple staging of the assembly line and hence
allows more students to participate rather than just

Assembly Line
Each line requires four computers set up in a row.
Multiple lines can be run simultaneously in a
computer lab. Each computer represents a
workstation position on the line (one to four). The
space before the first computer contains raw
material. The spaces between first and second,
second and third, third and fourth, contain WIPS of
the second, third and fourth WS's. The space after
the fourth computer contains finished goods. A
XX

INFORMS Transcations on Education 2:3 (XX-XX)

INFORMS

AMMAR and WRIGHT


A Demonstration of a Push/Pull Assembly Line

penny that originates in the raw material inventory


and travels down the four WS's before it reaches the
finished goods inventory represents the unit of
product. The row of computers represents WS's
performing the following sequence of tasks: Press,
Carve, Wash and Polish, requiring 5, 6, 7, and 20
seconds respectively. The layout and task times are
designed to exaggerate the affect of the push system
on the resulting WIP. The bottleneck is placed at the
end of the line to illustrate the rapid accumulation of
WIP and facilitate the discussion toward the pull
system as a way to control WIP accumulation.

To start the demonstration each student at a computer


selects his or her respective position (WS #) from
the initial program screen (see Figure 2). Each WS
will present a different screen. The row of computer
will show the sequence of four screens corresponding
to the four WS's. Also, at the start, four pennies are
placed in each of the WIP's, a bunch of pennies
(depending on the duration of the demonstration)
are placed in the raw material space, and the finished
goods space is empty (see Figure 3).
The push and pull systems are designed in two
separate applications.

Figure 2: Push System Initial Form

Figure 3: Four-Station Assembly Line


XX
INFORMS Transcations on Education 2:3 (XX-XX)

INFORMS

AMMAR and WRIGHT


A Demonstration of a Push/Pull Assembly Line

Push System

stop button), which disables the program timers.

We run the push (Assembly-Push.exe) demonstration for five minutes. This usually requires about a
100 pennies in raw material and is ample time to
demonstrate the effect of the push system. Students
are asked to select their position and wait for instructions to start. We first describe their function
through out the demonstration. The individual WS
screens are controlled with timers. The timers track
the duration of the production process in the WS
and flash the word 'GO' along with the picture of
the processed penny, upon the elapse of the processing time. Students are asked to observe their screen
carefully. When the word 'GO' appears they are to
pass a penny from their WIP or raw material to the
WIP of the next WS or the finished goods inventory. The timers control the work rate at every WS.
For example the processed penny will appear every
five seconds in the first WS and every 20 seconds in
the fourth WS. All students will start (press the start
button and enable the respective timers) at the same
time when given the instruction to do so. At the end
of five minutes students are asked to stop (press the

Observations and Discussions


There are, as expected, 15 pennies in the finished
goods inventory. This helps clearly demonstrate that
the rate of output is dictated by the processing time
in the slowest WS, in this case every 20 seconds.
One observation is that there are 27 more pennies in
the WIP of the fourth WS, as well as some buildup
in the WIP of the second and third WS's. Students
quickly appreciate that if the demo went on for hours,
days, or weeks the WIP at every WS would get out
of control. The cost of this buildup is our next topic
of discussion. These include the holding, space,
defective rework, turnover costs etc.
We then spend some time working with a spreadsheet
(push-pull.xls) that is designed to calculate the
measures of performance as a function of the
duration of the demonstration (Run Time.) These
include total output, total WIP, space, turnover time
(or how long will take a new penny to get through
the system), and the 100% utilization of WS's associated with the push system (see figure 4.)

Figure 4: Push Spreadsheet

XX
INFORMS Transcations on Education 2:3 (XX-XX)

INFORMS

AMMAR and WRIGHT


A Demonstration of a Push/Pull Assembly Line

The inputs (in yellow) include the task time, the initial WIP, and the Run Time. The Run Time can be
controlled with a scroll bar and can be varied in small
or large increments. In return, the ending WIP, the
turnover time, the utilization are calculated (in aqua).
Also the spreadsheet indicates whether the allotted
WIP space is sufficient or not. This assumes that
the available space can hold a predefined number of
pennies (say 30). The task times can be varied but
are restricted in a way that prevents the starving of
any WS in order to exaggerate the affect of the push
system. Hence the times are required to be in ascending order in the WS sequence.

system requires that the production rate of each WS


be limited to the production rate of the slowest WS,
in this case WS 4. We can consider the initial WIP
of four units in each WS to be "FULL" WIP load. In
the KANBAN system a workstation is allowed to
pull a unit from its WIP and start its process only
when the KANBAN indicates that the WIP of the
following WS is "NOT FULL".
Therefore when the fourth WS pulls a unit from its
WIP and starts its process, the KANBAN of the third
WS indicates that the WIP of the fourth is "NOT
FULL". This in turn allows the third WS to pull a
unit from its WIP and start its process. When the
third WS completes its process it will add the unit to
the WIP of the fourth WS and the KANBAN will
again indicate that the WIP of the fourth WS is
"FULL". The third WS will not process any more
units until the fourth WS is ready to pull another
unit from its WIP and the KANBAN is again "NOT
FULL". This process carries through to the second
and first WS's and hence equalizes the rate of output
of all WS's.

As we look for solutions to the problem of WIP


accumulation, students, unlike in the cups game, are
less focused on balancing the line by grouping or
breaking down tasks. They realize that no line can
be perfectly balanced and that some tasks on the line
will take longer than others. In fact, the buildup in
the second WS of 10 more pennies over the course
of five minutes, is the result of time difference of
only one second. Although a more balanced line will
delay the buildup, it will not eliminate it. This can
be further illustrated by varying the Run Time on
the spreadsheet.

The screens of the pull application include a


KANBAN that indicates whether the WIP of the
following WS is "FULL" or "NOT FULL". If the
WIP is "NOT FULL" then the WS screen tracks the
processing of a unit much like in the push system. If
the WIP is "FULL" then the WS is idle. For example
in steady state, the third WS is idle 13 out of 20
seconds and is in process 7 out of 20 seconds. At the
end of five minutes, there are still fifteen pennies in
the finished goods inventory, indicating that the
overall productivity was not affected, and there is
no build up in the WIP's beyond the initial level of
4. The pull system addresses the issue of WIP
accumulation and the associated cost. Other
measures such as the space needed and the turnover
time are also improved with the pull system.

Eventually students will suggest attempting to


equalize the output rate of all WS's and begin to
describe a pull system. From there we investigate
options for implementing the pull system and later
describe the KANBAN process. Although this
demonstration is deterministic and the layout and
task times are designed to exaggerate the affect of
the push system, students can be made aware that
any WS could have the slowest output rate. In fact,
different WS's can be the slowest at different times
in the course of production. This variation can be
the result of variations in the task time or machine
maintenance and breakdown. The KANBAN system
is an easy and effective way to equalize the rate of
output of WS's even when the slowest station cannot
be easily identified or may vary in the course of
production.

Finally, we discuss the issue of idle time in WS's or


the reduced utilization. Some students suggest that
it can be reallocated in other aspects of the
production. Others suggest proportioning production
shifts as a way of dealing with the reduced utilization.
It is however easy to conclude that the KANBAN
system introduces no or little additional cost to the
production process. At the same time it reduces the

Pull System
We repeat the demonstration using the pull
application (Assembly-Pull.exe), which now
includes a KANBAN between the WS's. The pull
XX
INFORMS Transcations on Education 2:3 (XX-XX)

INFORMS

AMMAR and WRIGHT


A Demonstration of a Push/Pull Assembly Line

costs associated with the WIP accumulation.

References:

Conclusion

Ammar S. and R. Wright (1999), "Experiential


Learning Activities in Operations Management"
International Transactions in Operations Research,
Vol. 6, No. 2, pp. 183-97.

One of the major strengths of the cups game, that is


somewhat lacking in this demonstration, is that it
epitomizes the notion that a picture is worth a
thousand words. The unsightly appearance of the
trays of cups in the WIP of WS's is motivation
enough for the pull system. However using this
demonstration, the key issues can still be effectively
addressed with minimal cost and time investment.

Goldratt's Game: Reducing variation in a production


line, http://webserver.lemoyne.edu/~wright/
goldratt.htm)

The demonstration is simple and focused, which


makes it ideal as an undergraduate exercise. So far,
it has been used in a total of eight sections of
introductory OM undergraduate classes. Other
instructors who have used the demonstration describe
the exercise as informative and fun.
Other important assembly line concepts such as line
balancing, time variations, and WS starving are not
directly addressed in this demonstration. We believe
that the push/pull concept is important and somewhat
complex for an audience of undergraduate students.
We therefore choose to handle it separately. An
attempt to better balance the assembly line in this
demonstration will detract from the rapid WIP
accumulation. The same is true, we believe, for using
random task times and allowing for the possibility
of WS starving (i.e. no WIP to process.) Of course
these concepts are important in the general discussion
of the assembly line operation, and we address them
using other demonstrations, such as the Goldratt's
Game.

XX
INFORMS Transcations on Education 2:3 (XX-XX)

INFORMS

You might also like