note that all subjects are seperated by a "TITLE : " string !
TITLE : Kanban Game
Date: Tue, 14 Sep 1999 09:31:50 -0400
From: Chris Lesadd <lesadd@mr.picker.com>
To: nwlean@egroups.com
Subject: NW Lean Mfg Net: Re: NW Lean Mfg. Net: Takt time
Try constructing a visual model to demonstrate what you want to achieve.
Richard Schonberger describes this in World Class Manufacturing. Schonberger
describes a "coffee cup simulator" where coffee cups represent kanban
containers and pieces of paper inside the cups identified the material, the
quantity, assy time, who uses it etc...
In my case I took a large drawing of the plant and cut pieces of red paper
representing the cells and overlaid them on the plant drawing. Each cell had
kanban squares for the parts it made and I cut business cards in half to
represent the parts. The "parts" also had data on them concerning assy time and
test time. The simulator looked and ran like a Monopoly game with one player
responsible for each cell.
When we ran the simulator we started with shipping, who pulled a system (bunch
of business cards). This left a signal for the final cell to test another unit.
The final test cell pulled parts (cards) from the assembly and sub-test cells
and so on. We went around the table with each player (cell) making moves to
satisfy demand signals
This exercise was very powerful. People could actually see what would happen in
a spacial/temporal sense. We also added the usual production SNAFUs like a bad
part, a customer change, a shortage to illustrate how these issues could be
resolved by the cell. The players wanted to win i.e. use their resources to
keep their kanbans filled. It really was astounding. Instead of a one hour
presentation where I tried to convince everyone that this was a good idea, 2
hours later the players were still hard at it trying to keep the squares
covered. I ran the simulator half a dozen times and every time the light went
on in a big way. People left the room thinking, "wow this could really work!"
Chris Lesadd.
"Conklin, Ed" wrote:
> I am just about to set up some flow lines in our manufacturing departments.
> Got the takt time and work stations defined. The problem is to convince
> Material Control to let us pull the material. I suggested starting with some
> end caps of a weeks worth of inventory followed by setting up kanbans based
> on vendors lead time on the line. NO GO. MC wants to kit material to the
> floor. They would like more info about KanBans. Is there a formula for
> setting up Kanbans? Any Advice?
> Eddie C
TITLE : Lego game
From: Scott Rezabek <scott_rezabek@yahoo.com>
Date: Mon, 19 Mar 2001 19:59:43 -0800 (PST)
Lego Simulation of the journey from Mass to Lean
Production has four three or four "rounds". The first round
illustrates Mass production with separate or satellite manufacturing
areas suppling a final assembly operation and performing to their own
schedules. Quality is inspected in after final assembly -- there is a
rework station -- and there is no 5S, Visual Factory or Error Proofing.
For round two, the participants do a "Kiazen" to apply some Lean
Manufacturing tools. The Measurements in round two -- First-Time
Through, Lead Time, Equipment Uptime, Build-to-Customer Demand, and
Cost typically improve by a factor of 50%. Round three is another
Kiazen that improves flow using a simple Pull System and introduces
Quick Changeover. In Round Four, flow in imporved again, cell
manufacturing is introduced to illustrate manufacturing flexiblity,
andQuick Changeover is applied. The typical round-to-round improvement
factor is 50%.
You can construct a "mock" manufacturing/assembly process to build a
Lego product (like airplanes or hyraulic valves. I have constructed
several of these, and they are alot of work to set up, and I question
whether you have enough time to do a good job before this Thursday.
TITLE : M&M Kanban Game
From: "Carl Lindenmeyer" <carlray@c-four.com>
Date: Thu, 18 Jan 2001 11:10:41 -0500
A neat (and fairly simple) way to demonstrate the KANBAN concept is to
obtain two small boxes (about the size of a small cigar box). Also,
purchase some (about 2 large bags) of peanut (my favorite anyway!) M&Ms.
Fill the boxes up equally (about 1/2 full or so -- and be sure to save
one-third of the supply of M&Ms). Have the class, one at a time, come up to
the front of the classroom and each student take out a number (calculate
this number based upon the number of students in the class) of the M&Ms from
one of the boxes (repeatedly) until there is no more. Inform them that they
are acting as an analogy to a production process which needs these "parts"
to perform the operations. Then switch the boxes, and after doing so, pour
the rest (the one-third you reserved) of the M&Ms into the empty (first)
box. Now inform them that your supplier has sent these replacement "parts"
based on an order you placed once the first box was emptied. Then, have
them again, repeatedly, remove M&Ms from the second box until empty, then
switch the boxes once more. At this point you are out of M&Ms to fill the,
now empty, box. Finally, state that you have ordered (or will order) a new
supply and will fill the empty box at the next class period -- if the
instructor will be so kind as to purchase the new supply! Hopefully, this
will inject a little humor to the demonstration!
Anyway, have fun and enjoy the M&Ms!
I have done this demonstration in my in-plant training for lean enterprise
development! It works, is fun, and the students remember it (which is the
most important outcome).
TITLE : Multitasking
From: JmBPotter@aol.com
Date: Sat, 27 May 2000 05:00:51 EDT
Subject: [cmsig] An excerpt from alt.projectmng about multitasking
To: "CM SIG List" <cmsig@lists.apics.org>
Danilo,
With open minded and thoughtful colleagues I have been taking the following
approach to demonstrating the multitasking effect:
1. Consider a collection of ten tasks. Upon completion, each task generates
revenue of $1000 (use more money per task if that improves the subject's
attention span). This is a fair (if somewhat abstract and simplified) model
of business activity. If tasks do not in some way (however indirect)
contribute to generating revenue, why do them?
2. All ten tasks are assigned to the same resource. No off-loading allowed.
This is also reasonable because it simulates a specialist with a work
backlog. It happens all the time.
3. Ask your associate to accept that each task requires a certain minimum
time to complete (say, eight hours for folks who prefer concrete examples).
Learning effects interacting with order of the work may change the total
time, but for sake of discussion accept that ...
- each task always completes in its minimum time
- if the resource multitasks, task switching wastes no time
Both these (normally, false) assumptions bend over backwards to deliver the
best possible performance under multitasking (reality will be MUCH worse).
Feel free to note this factoid, but do not make a big deal of it. With a
colleague who is just barely hanging on, do not even mention it, yet.
4. Now, ask your associate to consider the impact of "perfect" multitasking.
If the ten tasks will consume N days together, perform 1/N of the work on
each task each day. When will each task complete? What shape will the revenue
stream take? Compute the net present value (NPV) of the revenue stream under
"perfect multitasking." If you are doing a formal demonstration for a group,
you may want to assure that SOMEONE in the group "happens to have" a business
calculator or a suitable computer/software combination handy.
5. Now, repeat the same cash flow and NPV analysis under the assumption of no
multitasking (e.g., always finish a task without interruption before starting
the next task). How does the order of task starts influence the NPV result?
What order is best? Extra credit: What rule picks the order with the maximum
NPV?
6. Starting with ANY "no multitasking allowed" schedule, what happens to the
cash flow and NPV if you alter the schedule by adding one task switch? Why?
Consider the best "no multitasking allowed" schedule you discovered in "5,"
above. What happens to cash flow and NPV if you introduce even one task
switch?
7. If you "hook" a particularly inquiring mind, ask your associate to
consider the following ...
- How might task switching ACTUALLY influence the task completion times? Why?
- What about uncertainty? How might the uncertainty one normally encounters
in real life interact with the REAL task switching effects mentioned above? -
Under what circumstances can multitasking IMPROVE task performance for the
complete task collection? What happens when you carry that circumstance to
its logical conclusion?
- Recommend Goldratt's _Critical Chain_ or Robert Newbold's _Project
management in the Fast Lane_ to any associate who starts asking their own
questions and discuss the book with them as they read it (you get a refresher
course in any bargain that goes this far).
With a thoughtful colleague, I have ALWAYS reached the point where I receive
an acknowledgment that (at minimum, in theory) multitasking ALWAYS reduces
the NPV of a collection of tasks when the discount rate is positive. The
exception who proves this approach does not always work could walk up any
day, but (for the present) it should be worth a try.
TITLE : OPT - DBR Optimum Production Technology
From: Nicolas Hauduc <N.Hauduc@stg.co.uk>
Date: Fri, 2 Jun 2000 16:24:09 +0100
You can find a little simulator (that we call the OPT Executive Challenge)
at this place:
http://www.stg.co.uk/shockwave/optsolution/education/challenge/chalfs.htm
or here for HTML version only (if you don't have shockwave):
http://www.stg.co.uk/htmlonly/optsolution/opteducation/challenge/chalfs.htm
We use a similar tool in our training program to introduce people to the
OPT/TOC concepts in general and DBR in particular (this is at the core of
our scheduling softwares principles: OPT Advanced Planner and Scheduler and
OPT Scheduler - formerly ST-POINT).
If you manage to get good financial results from this little simulation
game, then it means you have understood some of the OPT/TOC concepts and you
managed to come up with a good Drum.
Have fun!
TITLE : Pull system & kanban demos.
I do a simple demo using a row of 5-6 people and have them fold paper
airplanes. I usually start with push system in batches of 5. Tell them
they are paid piecework. I intentionally build a real bottleneck and
also a secondary bottleneck by unbalancing the work content. The
secondary bottleneck is set up to disguise the real bottleneck. I run
for 5 minutes, count completions, WIP and calculate the cycle time for
the next unit coming into the line. In just five minutes of run time,
the queues usually gets up to several hours worth of work.
I then describe pull system and use anything available as a one-card
kanban. I run for another 5 minutes and the results are amazing.
Usually I get the same or more output with only 25 units in WIP and a
cycle time of minutes. The bottleneck is immediately apparent.
The only materials I need is a ream of photocopy paper.
TITLE : risk behaviour
From: Michel Baudin <Michel.Baudin@mmt-inst.com>
Date: Thu, 02 Nov 2000 10:27:44 -0800
If the company is doing well and lean manufacturing would be a way to
do even better, then the managers are likely not to be motivated. If it
is at risk of bankruptcy unless something is done, then they'll listen
to you. Next time you have an audience of five to ten, be it at work or
at a party, try the following game:
1.. Tell your audience to imagine they have just won a prize, and
hold two envelopes in front of your audience. In your left, say you
have $3,000; in your right, an 80% chance of having $4,000 and a 20%
chance of having nothing. Then ask which envelope they choose.
Almost everybody will choose the sure thing in the left hand.
2.. Tell your audience you are a judge about to impose a fine on them. In
your left hand, you hold a sentence for a $3,000 fine; in your right, an 80%
chance of a $4,000 fine and a 20% chance of no fine at all. Again, ask which
envelope they choose.
Surprise! Almost everybody chooses the right hand.
This was a psychology experiment carried out by Kahnemann and Tversky
and quoted by Peter Bernstein in "Against the Gods." I have tried it
multiple times and the outcome is always the same. What it says is that
we are risk-averse when it comes to maximizing gains, and risk-seekers
when it comes to avoiding losses.
If the business environment makes your management risk-averse, you can talk
until you are blue in the face and not get anywhere.
TITLE : Show Them the Money
Beschrieben in Soundview Aug 1999 Rodin, Hartman : Free, Perfect and Now
Take a product as example and show your people the price in terms of
physical coins. Make a pile of these coins.
Then explain the costing of this product. For each cost item you remove the
appropriate number of coins.
This should get them thinking.
TITLE : The Bead Game by Tony Rizzo
Beschrieben auch in Soundview Aug 1999 Rodin, Hartman : Free, Perfect and Now
The Bead Experiment
Tony Rizzo
To play the game, create teams of five people. The team performs two
projects, called the red project and the blue project. The projects
are identical, except for the fact that one has a red bowl and
plates, and the other has a blue bowl and plates.
(resource 1)
sortsortsortsort-|
|
| (resource 2)
|-processorangeprocessorange-|
| |
| (resource 2) |
|-processwhite---------------|
| |
| (resource 3) |
|-processblack---------------|
| (resource 1)
|-mixmixmixmixmix
The beads for each project are in the proportions 40 orange to 20 white to
20 black.
The first task requires that resource 1 sort the contents of a bowl.
The bowl contains glass beads of three colors: white, orange, and black.
the beads (actually called gems) are flat on one side and round on the
other side. The sorter uses a plastic spoon to move the beads from
the bowl to each of three plates. Each plate ends up with beads of
one color. Once all the beads are sorted and in their respective plates,
the sorting task is done, and the three successor tasks may be launched.
Also, since most organizations today use a phase and gate model, I have
the team complete the sorting task entirely, before any of the successor
tasks can be launched. Think of the sorting as the system engineering.
The successor might be the detailed component design tasks.
Once the sorting is done, resource no. 2 gets the plates with the orange
beads and the white beads. Only that resource can perform those two
tasks. Skills are not interchangeable. In addition, this resources
is a bottleneck, by design. Rizzo's policy is that the highest
level manager in the group gets this assignment, always.
Resource no. 3 gets the plate with the black beads.
Upon receiving their plate of beads, resources 2 and 3 begin the
processing steps. The processing steps are these:
1) spoon the beads onto the table (no pouring or flinging).
2) turn them all flat side downward, using your fingers.
3) turn them all again, flat side upward.
4) spoon them back into their respective plates, using only
the spoon.
Only the spoon may be used. Flinging and pouring of the beads are
not allowed. These are ISO-9000 violations.
Once the beads are all in their respective plates again, resource
no. 1 gets the plates and mixes the beads, by spooning them back
into the original bowl. Again, pouring and flinging are not allowed.
When the beads are in the original bowl, the project is done.
But, there's a catch. During the first experiment, the two projects
are staggered only 20 seconds apart. After 20 seconds, the sorter must
multitask and sort the beads of both projects. The sorter is required
to perform five operations on the red project and five operations
on the blue project, until both sorting tasks are done.
The resources who do the processing also must multitask, if they
have work before them from two projects simultaneously. So, resource
no. 2 must perform five operations on the orange beads of the
red project and five operations on the orange beads of the blue
project, whenever he/she has beads from both projects. When it's
time to process the white beads, resource no. 2 again must perform
five operations on the white beads of the red project and five
operations on the white beads of the blue project.
The same holds true for every task and every resource. If a
resource has only work from one project, then he/she need not
multitask. If the resource has work from both projects, then
he/she must multitask.
They won't want to multitask, because their common sense tells them
to not do it. Make them multitask. The purpose of the first
bead experiment is to model the performance of a multitasking
system. If they don't multitask during the first bead
experiment, then you won't be able to make a comparison between
the performance of a multitasking system and the performance
of a non multitasking system.
Before letting them perform this experiment, ask them if
multitasking is widespread in their organization. It's
important that they think about the degree to which multitasking
takes place, before they do the experiment. Most people
will acknowledge that multitasking is a way of life for them
and for their colleagues.
When performing the experiment, first give the team as much
time as needed, to learn all the steps. Then, have them
perform the multitasking experiment. As they work, have
a fourth person act as the project manager, whose
function is really to record the start time and the end time
of each project. A fifth person acts as the resource manager.
This person moves the work from resource to resource. He/she
also announces the start and the end of each project.
When the experiment is done, calculate the duration of each
project, in seconds, not in minutes and seconds.
Next, have the team perform the same projects, with the same
beads, and with the same resource assignments as before, but
without any multitasking. Give them instructions to work
each task to completion, before moving to another task.
This means that the projects are no longer staggered by a
mere 20 seconds. Now, they are staggered more naturally,
i.e., the sorting of the blue project begins only when
the sorting for the red project is completed. The same
holds true for all other tasks. NO MULTITASKING is the
rule for the second experiment.
When the non multitasking experiment is done, calculate the
duration of each project again, and compare these to the
times that you got from the multitasking experiment.
Then, let's have another discussion about multitasking.
Oh, you can order beads, bowls, plates, and spoons from the
following supplier, with which I have no financial ties:
Entre Design (973) 581-1995.
Ask for Joe Moran. Entre Design provides the kits that
I and others use to teach the TOC Multi-Project Management
Method to Lucent managers. Yes, he'll want to be paid.
His goal is to make money.
Tony Rizzo (908) 230-5348
tocguy@lucent.com
(R) Replication notice; Tony Rizzo, 1999, (908) 230-5348.
This article may be replicated freely, without written permission
from the author, but only if the article is replicated in its
entirety, including this notice. The article may be replicated in
any intelligible form. Such forms include, but are not restricted
to, hard copy publications that are either private, corporate,
public, or governmental in nature. Such forms also may include
electronic publications, such as Worldwide Web pages, electronic
newsletters, and electronic mail. The article also may be translated
into any language. Any such translations must include the complete
article as well as this notice. Translations may include attribution
for the translator(s). All translations must be made freely available
for further unrestricted replication by others.
TITLE : The Critical Chain Dice Game
Date: Sat, 08 Apr 2000 08:36:32 -0700
From: "D. Kissinger" <profits@7steps.com>
To: "CM SIG List" <cmsig@lists.apics.org>
Subject: [cmsig] Critical Chain Dice Game
While we're on the subject of dice games to illustrate how Theory of
Constraints (TOC) concepts work, I thought that I would share with you
what I call the Critical Chain Dice Game.
Critical Chain Dice Game
by Dohn Kissinger
Profit Solutions
profits@7steps.com
The objective of this game is to show that a project buffer
(contingency) of the proper size at the end of the project schedule
protects the project end date in most cases.
Take a simple schedule with 6 tasks. A review of these tasks by the
project team members indicates the following variability:
Task Optimistic Most Likely Pessimistic
(days) (days) (days)
1 3 5 10
2 3 5 10
3 3 5 10
4 3 5 10
5 3 5 10
6 3 5 10
The Critical Chain schedule is then, using the Goldratt approximation of
project buffer = 1/2 (Sum Pessimistic - Sum Most Likely):
Task: 1 2 3 4 5 6 Buffer Total
Duration: 5 5 5 5 5 5 15 45
days
One die (singular of dice) is then rolled to determine the actual
duration of the tasks. The number of rolls of a die until a 6 is rolled
is the actual duration for each task. The die is rolled until the task
duration for each task is determined.
One more thing: it is assumed that Parkinson's Law (the work expands to
fill the time available) is in effect for our simple project, so that
tasks can start on time or later than scheduled, but cannot start
earlier than scheduled.
It is good to do this with a group of at least 5 people, so that you
can see the variability in the schedules that result. I have done it
with about 100 people, but it is better to limit the group to about 20
people. Usually, about 80% - 90% of the schedules are completed within
the 45 day time frame.
TITLE : The Cup Game
Push/Batch vs Pull/Kanban Manufacturing Simulation
The Cup Game is a quick simulation to demonstrate the advantages of
Kanban/Pull manufacturing over traditional Push/Batch. The full
simulation with discussion can take less than an hour, and requires
only nine participants, although it is worthwhile if others are
observing. This Game was originally developed by Mike Studley for ACT
in England, and has been modified by HP and Lockheed, and now by Kevin
Meyer and John Vermillion at Abbott in Salt Lake.
1. Setup
Materials
? Small paper drinking cups (about 400)
? Green, red, and yellow circle stickers (about 500 each)
? 20 stickers different from above to use as "defects"
? Small pieces of paper, approximately 3"x3" (such as Post-It notes, about 400)
? Flipchart and markers
Set up a large conference room. Use flipchart paper to create stations
for the following:
? Warehouse station (does not need "In" box)
? Assembly Station #1
? Assembly Station #2
? Assembly Station #3
? Assembly Station #4
? Final Inspection station (does not need "Out" box)
Flipchart #1: Introduction
? Description of the Cup Game
? Agenda:
? - Introduction (10 minutes)
? - Push Simulation (10 minutes)
? - Push results discussion (5 minutes)
? - Pull/Kanban=6 simulation (10 minutes)
? - Discussion (5 minutes)
? - Pull/Kanban=1 simulation (10 minutes)
? - Discussion (10 minutes)
Flipchart #2: Roles
? Warehouse person
? Assembler #1
? Assembler #2
? Assembler #3
? Assembler #4
? Final Inspector
? Material Handler
? Engineer
? Observer
Flipcharts #'s 3,4,5: Simulation Instructions
? Instructions - see below
Create a "process instruction" by creating a final "product":
2. Introduction
Introduce the Cup Game and assign roles.
3. Push/Batch Simulation
In this simulation, the assemblers cannot talk to each other to
exchange quality information, and the Material Handler must be used to
transfer all product between stations in batches of 6. Run the
simulation for 10 minutes. The Material Handler can stage raw parts at
appropriate stations ahead of time (he will be very busy moving
product!). After the line is full and running, inject a "reject" unit
(a non-standard sticker) at the Assembler #2 station. Since no one can
talk about quality, this reject should make it all the way to Final
Inspection. Continue to sporadically inject defects into the line.
The line is unbalanced with Assembler #3 having to put three times as
many stickers on the paper as the others. Each operator to work as
fast as possible, and not to wait on the next station to finish, hence
the "In" area of slower stations will fill up rapidly. Follow
first-in-first-out principles with each batch.
Procedure:
? Warehouse person counts out 6 cups and 6 pieces of paper and calls
for Material Handler to move them to Assembler #1. Start working on
the next batch.
? Assembler #1 puts the paper into the cup and when done with 6
sub-assemblies, calls for the Material Handler to move the units to
Assembler #2. Start working on the next batch, if available.
? Assembler #2 puts the one green sticker on the paper and when done
with 6 sub-assemblies, calls for the Material Handler to move the units
to Assembler #3. Start working on the next batch, if available.
? Assembler #3 puts the three red stickers on the paper and when done
with 6 sub-assemblies, calls for the Material Handler to move the units
to Assembler #4. Start working on the next batch, if available.
? Assembler #4 puts the one yellow sticker on the paper and when done
with 6 assemblies, calls for the Material Handler to move the units to
Final Inspection. Start working on the next batch, if available.
? The Final Inspector inspects each assembly and puts them into a "Pass"
or "Reject" box.
? The Engineer calculates cycle time (time it takes for one product to
move from the warehouse through final inspection).
At the end of 10 minutes, stop the line. The Engineer then records on
the flipchart the cycle time, productivity (units shipped per person),
total WIP, and total shipped. The Final Inspector records on the
flipchart the yield. The Observer then leads the group in a discussion
of the simulation. Some questions:
? Where was the reject unit noticed?
? How much product is at risk due to potential risks?
? How much product is tied up in WIP?
? Where were the bottlenecks?
? How stressed out is the Material Handler?
4. Pull/Kanban=6 Simulation
In this simulation, the assemblers can talk to each other to exchange
quality information, and each assembler can pull subs/parts from the
previous station in groups of 6. Each assembler cannot make more parts
until the downstream assembler has pulled the group of 6 subs from
their "Out" area. Run the simulation for 10 minutes. The Material
Handler can stage raw parts at appropriate stations ahead of time, but
he will not have to do anything more except keep the stations stocked
with raw parts. After the line is full and running, inject a "reject"
unit (a non-standard sticker) at the Assembler #2 station. Since the
assemblers can talk about quality, this reject should be immediately
noticed at the following station, and removed from the line. Continue
to sporadically inject defects into the line. The line is now balanced
with each assembler putting on two stickers.
Procedure:
? Warehouse person counts out 6 cups and 6 pieces of paper and moves to
his "Out" area. Begin working on another batch when the "Out" area is
emptied.
? Assembler #1 pulls the batch of parts from the Warehouse "Out" area and
inserts the paper into the cup and when done with 6 sub-assemblies puts
them in his "Out" area. Begin working on another batch when the "Out"
area is emptied.
? Assembler #2 pulls the batch of parts from the Assembler #1 "Out" area
and puts the one green sticker and one red sticker on the paper and
when done with 6 sub-assemblies, moves them to his "Out" area. Begin
working on another batch when the "Out" area is emptied.
? Assembler #3 pulls the batch of parts from the Assembler #2 "Out" area
and puts the two red stickers on the paper and when done with 6
sub-assemblies, moves them to his "Out" area. Begin working on another
batch when the "Out" area is emptied.
? Assembler #4 pulls the batch of parts from the Assembler #3 "Out" area
and puts one yellow sticker on the paper and when done with 6
assemblies, moves them to his "Out" area. Begin working on another
batch when the "Out" area is emptied.
? The Final Inspector pulls the batch of assemblies from the Assembler #4
"Out" area and inspects each assembly and puts them into a "Pass" or
"Reject" box.
? The Engineer calculates cycle time (time it takes for one product to
move from the warehouse through final inspection).
At the end of 10 minutes, stop the line. The Engineer then records on
the flipchart the cycle time, productivity (units shipped per person),
total WIP, and total shipped. The Final Inspector records on the
flipchart the yield. The Observer then leads the group in a discussion
of the simulation. Some questions:
? Where was the reject unit noticed?
? How much product is at risk due to potential risks?
? How much product is tied up in WIP?
? Where were the bottlenecks?
? How stressed out is the Material Handler?
5. Pull/Kanban=1 Simulation
In this simulation, the assemblers can talk to each other to exchange
quality information, and each assembler can pull subs/parts from the
previous station in groups of 1. Each assembler cannot make more parts
until the downstream assembler has pulled the group of 1 sub from their
"Out" area. Run the simulation for 10 minutes. The Material Handler
can stage raw parts at appropriate stations ahead of time, but he will
not have to do anything more except keep the stations stocked with raw
parts. After the line is full and running, inject a "reject" unit (a
non-standard sticker) at the Assembler #2 station. Since the
assemblers can talk about quality, this reject should be immediately
noticed at the following station, and removed from the line. Continue
to sporadically inject defects into the line. The line is now balanced
with each assembler putting on two stickers.
Procedure:
? Warehouse person counts out cup and 1 piece of paper and moves to his
"Out" area. Begin working on another batch when the "Out" area is
emptied.
? Assembler #1 pulls the batch of parts from the Warehouse "Out" area
and inserts the paper into the cup and when done with 1 sub-assembly
puts them in his "Out" area. Begin working on another batch when the
"Out" area is emptied.
? Assembler #2 pulls the batch of parts from the Assembler #1 "Out"
area and puts the one green sticker and one red sticker on the paper
and when done with 1 sub-assembly, moves it to his "Out" area. Begin
working on another batch when the "Out" area is emptied.
? Assembler #3 pulls the batch of parts from the Assembler #2 "Out"
area and puts the two red stickers on the paper and when done with 1
sub-assembly, moves it to his "Out" area. Begin working on another
batch when the "Out" area is emptied.
? Assembler #4 pulls the batch of parts from the Assembler #3 "Out"
area and puts one yellow sticker on the paper and when done with 1
assembly, moves it to his "Out" area. Begin working on another batch
when the "Out" area is emptied.
? The Final Inspector pulls the assembly from the Assembler #4 "Out"
area and inspects it and puts it into a "Pass" or "Reject" box.
? The Engineer calculates cycle time (time it takes for one product to
move from the warehouse through final inspection).
At the end of 10 minutes, stop the line. The Engineer then records on
the flipchart the cycle time, productivity (units shipped per person),
total WIP, and total shipped. The Final Inspector records on the
flipchart the yield. The Observer then leads the group in a discussion
of the simulation. Some questions:
? Where was the reject unit noticed?
? How much product is at risk due to potential risks?
? How much product is tied up in WIP?
? Where were the bottlenecks?
How stressed out is the Material Handler?
6. Final Discussion
Discuss what happened. Specifically, notice the dramatic difference in
WIP, risk WIP, productivity, and cycle time between the three
simulations. With such an increase in cycle time and total shipped,
how many assemblers can be removed and still make the same amount of
product in the same amount of time?
How can this be applied to your current operation?
What is the next step?
---
RESULTS :
Below are the final stats for the Cup Game that all of us just went
through to demonstrate the advantages of "pull" one-piece-flow over our
traditional "push" batch process. We had modified the classic
HP/Lockheed Cup Game for our use, and will be writing up this modified
version if anyone would like to play the game with their employees.
A reminder that each simulation was for 10 minutes with 4 assembly
stations, 1 warehouse station, and 1 inspection station.
Push (batch=6) Pull (Kanban=6) Pull (Kanban=1 )
Cycle time (min) 8:40 5:05 0:32
Total WIP (units) 86 25 4
Output (shipped) 22 29 48
Productivity 3.2 4.9 8.0
Yield - final QA 96% 100% 100%
Yield - overall line 96% 97% 99%
Productivity is the output per operator, and cycle time is total time
from warehouse through inspection for each assembly. As you can tell
from the greatly reduced cycle time and increased productivity, you
could radically reduce the number of operators as you move toward
one-piece- flow. Since defects are caught immediately on the line,
final inspection becomes meaningless. The material handling function
was also reduced by 90% in the line-balanced kanban simulation. The
kanban simulations also pointed out several potential manufacturing
improvements, such as going to a hub/spoke assembly process instead of
straight-line-flow.
Now to figure out how to implement in the molding environment...
TITLE : The Dice Game
in analogy to the game explained in The Goal by Goldratt :
THE DICE PRODUCTION SIMULATION GAME
To explain the effects of dependent processes in a production enviornment, we
often use a series of simulations. Below are the instructions for one of the
best ones. Although we normally run the game with several teams, you can do it
with one team -- simply run the games several times to see the different
results. Here's the basics.
We set up two or three teams of 6-8 people per team.
Round 1:
2. We give each person four poker chips, legos, etc. The first person (raw
materials) has a box of inventory. Each person gets one die. The last person in
each team is the shipping department. They sit their parts out to the front of
the table, this is the shipped product.
3. We explain that the average roll of one die is 3.5 so each team should be
able to ship 70 chips over 20 days.
4. The exercise runs 20 days. Call out the rolling of the dice for each day and
have each player record their dice roll and how many they pass on. The hardest
part is explaining to people that they can only pass on what was in front of
them AT THE BEGINNING of the turn - everything received during the turn is
supply which arrived "after production". Sometimes, if they're low on
materials, they'll wait until they get chips passed to them and try to pass
them to the next person. Each person should always have something in front of
them at the end of each day.
Round 2:
5. After 10 days, count up the shipments. They normally fall short of 35 chips
they should have according to the schedule. Appoint a plant manager and give
them 5 overtime rolls for the next ten days. An overtime roll is an extra roll
the manager may have any team member take at the end of the day. However, since
overtime costs extra, tell them you will take one piece of shipped product as
the overtime premium.
Round 3:
6. You then play the next 5 days. Things probably aren't still looking much
better, so give the plant managers the option of moving people. They can have
someone roll with another team member, basically adding their rolls together.
The plant manager must decided who will be moved at the BEGINNING of each day,
before anyone rolls. Only one person may be moved each day per team. If a
person is moved, their station is basically shut down that day and zero chips
are shipped out from their station, althought parts can be shipped IN to their
station. If a plant manager wishes, they can have someone roll with someone
else, then come back and run overtime at their normal station.
7. We play the last 5 days. Even with all the moving and shuffling of people,
teams almost never get the 70 pieces out, and inventory will have gone up. OR,
if they do get out close to 70, they will have drained the plant and set
themselves up for disaster the next month.
This is why line balancing inevitably fails in plants -- the closer to balanced
they are, the worse the situation becomes.
We play a lot of different games during a seminar/project, but you can use this
one to at least get your managers thinking.
==========================================================
Date: Thu, 30 Mar 2000 09:58:03 -0500
From: Tom Gattiker <Tgattike@arches.uga.edu>
To: "CM SIG List" <cmsig@lists.apics.org>
Subject: [cmsig] RE: Dice Game Instructions
Dice Game
In the dice game each person represents a department. We sit in rows
of approximately 5 people. Each row could represent a production line,
a service activity, such as designing a building, or a distribution
system. The row begins at the left. The beginning of the row is raw
materials; the end of the row is completed units. Each poker chip is a
unit.
Each roll of the dice represents a day-s potential output at your
department. We will roll the dice simultaneously and move units
through the system. A low roll represents a day plagued with things
like absenteeism, equipment malfunctions, etc. A high roll represents
a good day.
No-lot-size version
Units move down the line, beginning at the left-most station and moving to the
right. Divide a piece of paper into half. This paper represents your
department. The left half represents your queue: units in line waiting to be
processed at your station. The right half represents a machine (or any type of
processing/transformation activity).
You will roll the die when the facilitator instructs you to. Your roll
represents the units that you process for the day. For example if you
roll a 3, take 3 units from your inbound queue, pass them over the
processing activity (the right half of your paper), and place the queue
of the station to the right of you This assumes you have 3 units in
your queue.
If you only have 2 units in your queue then you can only process 2
units. Instead of placing units in the next queue, the last person in
line moves his or her processed units off to the side. These units are
shipped units. Accounting: Next, record your actual roll, pieces
processed, "efficiency" (units processed divided by standard), and WIP
(units in your queue) on the score sheet. Note that if you need to save
time, do not calculate efficiency after every roll. Instead just
calculate it when you are done playing (column 3 total divided by the
standard). Next wait for the facilitator to call for the next roll of
the dice (the next day-s production) and begin again.
Lot size __________________
Standard __________________
Number of Days to Play __________________
Anticipated output system __________________
Lot size version (transfer batch)
Divide a piece of paper into thirds. This paper represents your
department. The first third is your inbound queue. The second third
represents a machine (or any type of processing/transformation
activity). The last third represents units that have been processed.
We will call this last third your holding area. Once a lot is
accumulated in your holding area, you can move it to the next station-s
inbound queue when the facilitator instructs. You will roll the die
when the facilitator instructs you to. Your roll represents the units
that you process for the day. For example if you roll a 3, take 3
units from your inbound queue and place them in your holding area.
This assumes you have 3 units in your queue. If you only have 2 units
in your queue then you can only move 2 units to your holding area.
At the end of the day, we transfer material to the next department
provided we have a lot-s worth of units in our holding area.
In other words, after you roll and have moved units from queue to
holding area, I will instruct you to move units from your holding area
to the next department-s queue if you have a lot-s worth in your
holding area.
You can only move material in multiples of the lot size. Each time you
receive a lot into your queue you incur a setup. The last person in
line moves units from his holding area off to the side. These units
are shipped units.
Accounting: Next, record your actual roll, pieces processed (moved from
queue to holding), efficiency (processed divided by standard), WIP
(units in queue + units in holding area) and setups (if you received a
lot into your queue, you had a setup) on the score sheet. Next wait for
the facilitator to call for the next roll of the dice (the next day-s
output) and begin again.
Lot size __________________
Standard __________________
Anticipated output system __________________
Low variability version
Same as no lot size version. But with the following modification to the rolls
If you roll then move
1, 2 or 3 3
4, 5 or 6 4
Unbalanced line version
Middle person in the line has the following modifications to his/her rolls.
If you roll then move
1, 2 1
2, 3 2
3, 4 3
Results: output should be about 2 with very high inventory.
Unbalanced line removes the effect of statistical fluctuations and
dependent events. The output is much closer to the theoretical output
(2 units per day)) than in the balanced lines.
Drum-Buffer-Rope version
Like unbalanced line version Except, we only put inventory into the
system at the rate the constraint can process it. First person only
passes 2 units to second station per turn. If he rolls a 1, he can
pass an additional unit next turn to compensate. Buffer the constraint
with 6 units. Do not buffer the stations downstream of the constraint.
(Or have el Nino hit and wipe out all inventory except at the
constraint).
Results- Dramatically less inventory than Unbalanced line version or other
versions.
Dice Game De-brief
Efficiency- We calculated eff as actual output for the department
divided by standard, which is reasonable. Several people noticed that
as lot sizes went down efficiency did not go down. This is a good
obsrevation. The reason for this is the artificiality of the game.
Notice that setups entailed no cost in volume produced. In reality,
each setup would have reduced the units produced because when a
department is setting up, output is zero. This reduction in units
produced would have harmed efficiency. Also note that smaller lots
resulted in less starvation of your downstream neighbor. This means
that your lower lotsize increases his efficiency, but it hurts your
efficiency. What I want you to understand about lot sizes is that they
generally hurt throughput. Our second run of the game was much more
successful in terms of units sold. Even though large lot sizes hurt
throughput, traditional thinking and its quest for efficiency and low
cost per piece encourages them.
To increase efficiency and decrease cost per part
1. Increase wip - 4 -
2. Increase lot sizes
To increase sales:
1. Increase WIP sometimes (buffers against variability, eg stat fluctuation and
variability in Quality)
2. Decrease lot sizes
3. Decrease variability
1. Note on WIP. With a straignt line making one product (Dice game),
increasing wip increases shipments; however, in a complex environment
with more products and more complicated routings, WIP will decrease
shipments. Furthermore, WIP has a negative effect on quality and on
the companies ability to remove process time variablility (JIT cause
and effect chain from class March 5).
2. If we can decrease lot sizes cost-free we will be better off. Goal
shows us that set-ups at nonconstraints really are cost free JIT shows
us that there are many ways to reduce the cost of setups (setup time
reduction, etc)
3. If we can decrease variability we will be better off. TOC accounts
for variablility using buffers (inventory). Note that it uses the
inventory differently than traditionally. JIT seeks to remove
variability through continuous improvement in processing times and
quality.
JIT shows us how to decrease lot sizes and decrease variability. These
increase shipments in themselves. Furthermore doing them actually
lets us drive down wip instead of increasing it. Furthermore,
producting in small lot sizes with high velocity allows us to produce
at a rate and mix that matches the market instead of producing in
batches.
TOC shows us that every system by definition has a constraint. TOC
shows us that you can never reduce variability enough to eliminate the
type of effect we saw in run 2 of the dice game. Build in extra
capacity except at the constraint. TOC suggests that efforts to reduce
vairability in quality and processing times everywhere is a folly. If
the constraint determines system output the focus improvement efforts
at the constraint. Roll Move from your queue to your neighbor to the
right-s queue Record
===========================================================================
Date: Mon, 10 Apr 2000 12:29:24 -0700 (PDT)
From: Tim Sullivan <sullytt@yahoo.com>
Subject: [cmsig] Re: Dice game and raw material costing
To: "CM SIG List" <cmsig@lists.apics.org>
I have done my own version of "the dice game" in my workshops for
years. I have incorporated valuing inventory at raw material cost into
the exercises.
To keep it as simple as possible, I say the end product sells for $110
and the raw material cost per unit is $10. We track the total work in
process inventory value and throughput at the end of each round of
rolls. In this way, WIP is stated in dollars not pieces, but it is
simply the number of pieces times 10. The throughput is also stated in
$, and is the number of pieces "shipped" times 100.
This has worked VERY well and has NOT seemed to detract from the smooth
flow of the exercises. I really like it because it gets ome of the
basic concepts of throughput accounting drilled in to the participants
in a very subtle, but effective way. (You see I teach in the workshop
that throughput has to be measured in $/time -- then if my exercise
focuses on pieces I am contributing to the confusion that is out
there.)
There is another thing I do that I think makes the exercise more
"realistic." I do not use 1 die. I use 2 dice. That way the output has
a central tendency like our real processes. The distribution is
pyramidal (centered at 7, range from 2 to 12) and not "normal", but it
is close enough that everyone accepts it...even the engineers in the
group!
---
Date: Wed, 28 Feb 2001 10:35:24 -0600
From: Tom Turton <tturton@ntx.waymark.net>
Someone mentioned finding the dice game on a TOC website; I found this one:
http://www.wvamc.com/dicegame.html
BUT, I am on my second read of The Goal...just got through Rogo's
dicegame with the Scouts, and can't recall if he "corrects" himself
later, but this dice game is flawed, or rather oversimplified.
It uses a "uniform" rather than "normal" distribution. Okay, okay, so
the point it demonstrates is valid, fluctuations mess up your production
line, but it also depends on how bad those fluc|uations are, and any
company that manufactures anything with a uniform distribution has got
more problems than they know where to begin!
Still, check out the above website...it adds a little more complexity to
the basic game by throwing in "overtime" and "moving workers on the line".
Date: Fri, 27 Apr 2001 17:17:06 -0400
From: Larry Hirschhorn <lhirschhorn@mail.cfar.com>
Phillip Bakker was kind enough to send me the URL for an article, by
Kelvyn Youngman on hospital waiting times. In it he describes the
following game
The Dice Simulator
"The behaviour of a closely coupled process can be simulated using a
dice and 4-5 boxes of matches. You will need from 2 to 6 players.
Establish 6 stations in sequence around a table. All but the 1st
station have 4 matches each to start with, the first station has all
the remaining matches. These represent patients in the system. We
have chosen 4 as about the average value that we can expect to throw
from a six-sided dice. Starting at the first station throw the dice
and put aside as many matches as possible, up to the value thrown,
from the group in front of you. Each station in turn does the same,
this represents one day of work. After the 6th station has done
this, each station passes their current days work to the station on
the station at the right to be consolidated with any that remain
unprocessed. Station 6 passes the work to an accumulating pile of
happily discharged patients. Repeat this sequence for 30-40 times.
If there are only two players, spilt the first 3 and last 3 stations
between you.
Mr. Youngman notes that he has an Excel version of this game, but
when I tried to email him the mail bounced back. So I went and made
up my on model, where I assigned 20 matches to the first station and
four each to the next five stations. I used a random number generator
(which varies betwee 0 and 1) instead of dice. In addition, I
simulated what would happen if DBR were applied, by assuming that in
a second and different game people processed matches/patients through
stations at a fixed rate equal to the constraint. So for example if
you processed 2 patients each period you would process all of them in
20 periods, if 3, then in 13 periods etc. I found to my surprise that
in the first game, in which processing times were determined
randomly, the curve of total patients processed by time period ,
never exceeded (rose higher than) that the curve of the second game
when processing time was equal to 3 patients per period. Another way
of saying this is that while the random game can "beat" the fixed
game, that is process patients faster, if the rate of processing in
the fixed game is <3, it never beat it when it is 3 or more?
Does this make sense to people. Assuming that I did the arithmetic
right, what principal am I stumbling upon here. Is it in some way
related to the fact that in a constrained system, you can't
compensate for delays with later accelerations?
TITLE : The TOC Exercise from The Race
From: JmBPotter@aol.com
Date: Thu, 2 Mar 2000 01:36:28 EST
Subject: [cmsig] Does TOC Have Credibility?
This is actually a judgment your firm can make for itself. The exercise below
may work well as an "off site retreat" activity. I recommend the following
procedure:
1. Purchase a copy of Goldratt's _The Race_ and
one quadrille pad for each member of your
executive team.
2. This (steps 2- 6) is the hard part. Get them
to each work ONE of the five exercises in
the back of the book on paper from the
quadrille pad. If you have fewer than five
executives (one per problem), have some
executives tackle more than one problem.
With more than five executives, assign some
exercises to more than one executive. Allow
about 5-10 minutes or until everyone
finishes (which ever comes first) for
working problems. Anyone taking more than
ten minutes on one of these exercises has
probably taken a wrong turn. Spare them
the agony by stopping them before they
waste more time. DO NOT cite this assertion
as a reason for stopping. Some utterance
along the lines of staying on schedule or
respecting the value of everyone's time
should play better.
3. Treat the book as a presentation. Read the
text (or hit the high points only; you have
an able audience) while the executives
follow the illustrations.
4. Have each person rework their original
exercise also on the quadrille pad without
consulting their original work. They will
need less time, now. Group people who did
the same exercise together for maximum
learning effect. If your executive team
includes fewer than ten people, you may
want to include enough other leaders to
reach ten to fifteen participants.
5. Form presentation teams of people who did
the same exercise. Each team should read
the "offical answer" following the exercises
in the book. Each presentation should
include a study of a "before" answer, an
"after" answer, and the "official" answer.
6. Each group presents their results to the
full executive team.
7. The executive team can now make a judgment
about ToC on their own. If they are unsure
but want a deeper investigation, they may
be willing to spring for $500 each ($300
each for companies which enrolled enough
students in the GSP) to see video tapes of
last year's GSP.
The above exercise with _The Race_ should take about three to five hours
depending upon how rapidly you do the presentation and how many questions you
must field. Few executive teams will be willing to make this kind of
commitment to informal self-training about something they do not understand.
Getting your executive team to make this commitment will be the hard part for
you.
You can save some time by assigning the first exercise as "prework." You may
substitute the video of _The Goal_ for presenting _The Race_. If you are
heavily pressed for time, the video might save half an hour to an hour. The
video may also spark fewer questions.
At minimum, an executive team who have been through this exercise, would be
well positioned to detect either a snow job or incompetence on the part of a
consultant evaluating ToC methods four you. A team without this introduction
might have no way to realize they have been sold a bum steer.
Especially for organizations interested in Six-Sigma or other improvement
processes, emphasize the "ToC focusing steps" as part of the presentation.
Simply applying the ToC focusing steps can aim a Six-Sigma program with such
accuracy that its value may improve by a thousand fold or more. The video of
_The Goal_ pounds on the focusing steps, but _The Race_ does not hit these
points strongly.
TITLE : Toss Them The Ball
Beschrieben in Soundview Aug 1999 Rodin, Hartman : Free, Perfect and Now
-Top Management als Teilnehmer incl 2te Ebene
-"This symbolizes a request regarding prices and delivery" and toss
somebody a yellow tennisball. Watch what he does with it and what comes out
of it.
TITLE : URL's for simulators
Date: Fri, 4 May 2001 11:33:40 -0400 (EDT)
From: Tony P <atperna@mail.com>
Subject: [cmsig] Constraint Simulation Games
Over the years I have come across some good simulation games, usually using
dice, to illustrate constraints and how to use them to improve flow and
reduce inventory. Does anybody know of a game that ties in the
financial/profit/money impact by applying TOC practices. I came across the
factory flow game at
www.factoryflowgame.com
which is OK. Someone mentioned an Excel simulation.
This topic has probably gone around before, sorry for the redundancy.
---
From: "Jean-Daniel Cusin" <jdcusin@cybernostic.com>
Subject: [cmsig] RE: Constraint Simulation Games.
Date: Fri, 4 May 2001 12:54:38 -0400
Productivity Inc. had a Kanban simulation game a few years ago. I have a
copy that I still use regularly. I can't find it on their site anymore
however.
Alternatively, there is the Buckingham Lean Game that you may want to look
at. Here is their URL:
http://www.axiom.co.uk/picsie/tbjgm.html
I have a Kanban Implementation Handbook that I have previously offered free
of charge to this list, which goes quite far into how to calculate the
number of kanbans one needs in ach kanban loop based on capacity
constraints, etc. Contact me directly if you would like a copy.
Bill DETTMER :
[Yes. The Management Interactive Case Study Simulator (MICSS), from MBE
Simulations, Ltd. (Israel) does that. Eli Schragenheim created the MICSS.
It's a computer simulation. It permits sensitivity analyses (changing
various policies from "traditional" to TOC, individually or in combination)
to evaluate/demonstrate the results. The program integrates production,
marketing/sales, purchasing, and finance functions. It shows the
interactive effects of changes in any one of these areas on the others.
A copy of the MICSS software comes with the purchase of the book
"Manufacturing at Warp Speed" (APICS/St. Lucie Constraint Management
Series -- try searching the On-Line Catalogue at www.apics.org). However,
this version is restricted to personal learning/use only. If you plan to
use it in an organizational environment, for "official" purposes, you'd
better contact MBE Simulations, Ltd. (www.mbe-simulations.com) to negotiate
terms.]
TITLE : Web pages for games
KANBAN Game : www.geocities.com/CapeCanaveral/2226/acad/systeng/jit.htm
From: "Brinton, Russ" <Russ.Brinton@PSS.Boeing.com>
To: "CM SIG List" <cmsig@lists.apics.org>
Subject: [cmsig] RE: Dice Game Instructions - Job Shop Game
Date: Mon, 17 Apr 2000 07:59:36 -0700
Go to:
http://www.vancouver.wsu.edu/fac/holt/em530/
And look under "Course Documents"
There are links for the
Job Shop Game Student Description
Job Shop Game Instructor Description
Job Shop Game Visual Aid
Job Shop Game Summary DBR vs. Traditional (36 Tasks)
TITLE : turnaround situation - standard costing
Company ABC corporation has 140 MioFF sales, the majority coming from
it's major product line P where they sell 10 Mio # p.a. at 10.-FF/#.
GrossEarning for product P is 30%, material and purchases account for
40% of sales. Installed line capacity for product line P is loaded at 75%
of max capacity.
The gross earnings of 30% according to the StdCOS modell are by chance
identical to GE=30% had ABC Inc. applied activity based costing.
Overall ABC Corp is making a loss of 3.6 MioFF p.a. The GenMgr has asked
his first line to come up with ideas to turn around the business or
they will have to close the plant.
The OpMgr committed his organisation to deliver 8% cost reductions
instead of the budgeted 4%. This does not bring them out of the sink
however because
8% of (100%-30%-40%)=30% = 2.4% cost reduction
2.4% of 140 MioFF is 3.36 MioFF cost reductions. This is not enough
as these reductions are bound to not materialize in the current fiscal
period.
The sales director turns up with a business opportunity to sell 2 Mio#
p.a. more of product P (which is well within the available capacity).
The downside is that the price would be 40% below the going price - in
other words 6.- FF/#. As it is a segmented market there is no fear that
the price in the main market will be effected.
What should ABC Inc do ?
Option A)
=========
Decline the additional business !
Our current rules forbid to accept a business at negative margin (30%
GE - 40% price decrease = -10% GE). Common business sense clearly says
that we have to decline that business as it will drive us deeper into
the losses.
Consequently ABC Corp will go bankrupt.
Option B)
=========
Accept the business !
This would give additional sales of 12 MioFF. There are no additional
operating expenses as we can produce on existing capacity. The only
cost arises from purchasing which is 2 Mio# * 40% of 10.-FF/# = 8
MioFF.
This business oportunity generates 4 MioFF net profit and allows ABC to
turn around, close with 2MioFF profit and survive although common
business rules would have killed it !
TITLE : P-Q Game
Raw Material Machine Assembly finished good
PP
5$/#
|
RM1 A C |
-------> ---> -- V --> 100 # p.week
20$/# 15min 10min \
\ D P
--> --------> -----> customers
/ 15min 90$/#
RM2 B C /
=======> ===> --
20$/# 15min 5min \ --> 50 # p.week
\ D Q
--> --------> -----> customers
/ 5min 100$/#
RM3 A B /
-------> ---> --
20$/# 10min 15min
Additional information :
- 8 hrs/day on 5 days/week open hours, no pauses
- no setup time required
- operational expenses of 6.000.- US$ per week
TITLE : Schragenheim on necessary conditions for games
Date: Wed, 11 Jul 2001 10:50:09 +0300
From: Eli Schragenheim <elyakim@netvision.net.il>
For a game to provide a learning experience, some necessary conditions
have to apply:
1. The game truly mimics (simulate) some aspects of reality and the
participants all agree, before the game starts, that it does represent
the more significant aspects of reality.
2. The participants agree they have some control on the results.
3. The participants fail in their first attempt.
4. The instructor is able to suggest the cause and effect behind the
failure in the first attempt.
5. The participant, with or without the help of the instructor, devise
an alternative approach that considers the cause and effect mentioned.
6. When applied the new rules lead to significantly superior results in
the subsequent attempts to play the game.
7. The participants, with or without the help of the instructor, are
able to bridge between the solution, that worked well in the game, and
their own reality.
And the game must be interesting to play, clearly non-trivial but not
too complex and the participants have a lot to do while playing.
TITLE : supply chain game
Date: Tue, 10 Jul 2001 21:05:06 -0400
From: Ed Walker <edwalker@gasou.edu>
Subject: [cmsig] Re: Supply chain game (was Braess)
Reply-To: cmsig@lists.apics.org
Rudy Burkhard asked:
I would like is a game that quickly demonstrates the fallacy of local
optimisation. The individuals in the game would do whatever is best for hem
locally. The result will be less than optimal for the game system.
and Brendan Fox replied:
There are several such games I can think of that demonstrate effects of local
optima that are manually played with tokens and such.
- The Production lego
- Production dice game
- The Beer game
I would add that I use a simple game in an undergraduate class to
demonstrate that better results can be achieved by working with your
suppliers and customers. (I do use my variant of the dice game for
production; but I find that the beer game, though excellent, takes more
time than I have available to do it effectively.)
The "supply chain game" is simple and straightforward. I have the class
work in groups of three (or so) members where each group represents a
company with one supplier and one customer.
You need a minimum of three companies because this is done in a circular
fashion. Call the three companies A, B and C. Company A supplies Company
B and is supplied by Company C. Company B supplies Company C (and is, of
course, supplied by Company A.) As a result, Company C supplies Company A
and is supplied by Company B.
Each Company has a set of cards in three different colors -- let's say
Blue, Green, and White -- and (at least) two envelopes.
The companies make or lose money based on the colors of the card they pass
to their customers and the card they receive from their supplier. The
table below shows the money made or lossed on the overall transaction.
Get
Give Blue Green White
Blue 20 -10 -10
Green -5 15 -5
White 15 -15 15
So, if you pass a Blue card to your customer and receive a white card in
return your company makes a negative profit of $10.
If we calculate the expected payback of giving each color (and assume that
there is an equal probability of receiving any of the three colors), we
find that the expected value of passing: a Blue card is $0; a Green card is
$1.67; and a White card is $5.
Instructions for every round:
The company is to decide what color card they want to pass to their
customer and their supplier and places that color into the envelope
(without allowing competing companies to see what color they choose.) All
companies then pass their card simultaneously to their customers and
suppliers. They then calculate the money made (or lost) on the selling
transaction and the purchase transaction.
Allow ten (10) buy and sell transactions and calculate the total profit or
loss.
Round one:
I tell the companies that they are not to converse with their suppliers or
customers and that there will be a reward for the company (or companies in
the event of a tie) that makes the most money.
It seems to always happen that the companies settle on passing the White
card as it is the safest (highest expected return, least probability of
loss, etc.), BUT at least one company will try to sabotage other companies
by passing a Green card every once in a while.
Round two:
Allow the companies to meet with their customers and suppliers for five
minutes, then repeat the process.
Astute companies will recognize that the way to make the most money is to
pass AND receive the Blue card.
****
This game simply illustrates the importance of working with your customers
and suppliers (and nothing more.) [I must give credit for the idea behind
this game to a couple of Organizational theorists named Pfieffer (sp?) and
Jones.]