This thread is for the Beer Game. This will be an essay question on the midterm related to this experience.

Here are some links for the Beer Game:

http://web.mit.edu/jsterman/www/SDG/beergame.html

http://www.forio.com/nearbeer.htm


The comments from last year's MC506 class are available below, please add your comments for 2005. Worth +1 point on the Midterm.

There's NO BEER in the beer game? So much for Metrologists playing.... :biglaugh: :bonk:

Hershal

I'm confused. I understand all about "rubber banding" or "bull whip" in supply chains. I am unsure about the mechanics of this game. The player instructions seem to imply the consumer demand is "random" (within limits) as represented by a "deck of cards." The discussion, however, states the consumer demand is level, starting at 4 cases/week, rising to 8 cases/week. Did I miss something in the information presented or are there "secret instructions" (i.e. a rigged deck of cards?)

This looks like a lot of fun (at least as much as "red beads.") Thanks for bringing it to our intention.

I'm confused. I understand all about "rubber banding" or "bull whip" in supply chains. I am unsure about the mechanics of this game. The player instructions seem to imply the consumer demand is "random" (within limits) as represented by a "deck of cards." The discussion, however, states the consumer demand is level, starting at 4 cases/week, rising to 8 cases/week. Did I miss something in the information presented or are there "secret instructions" (i.e. a rigged deck of cards?)

This looks like a lot of fun (at least as much as "red beads.") Thanks for bringing it to our intention.

Actually it is "rigged". You tell the participants that "anything can happen" to demand, but all that actually happens is the demand goes from a steady state of 4 per week to a steady state of 8 per week. And yes, the results are very predictable, both groups of students experienced the bull whip. I will be charting and posting the groups' results in excel here.

This is an interesting problem. I'm sure other minds have contemplated this challenge, but it got into my head and couldn't quite let it go. As I think about it, I have concluded that a "bull-whip" is 1) inevitable and 2) the best solution to the problem as stated (e.g. known delays up the supply chain, no marginal cost for sudden changes in production).

I think I have a pretty logical and efficient algorithm for placing orders and maintaining inventory (even when the orders at the retail outlet are not simply 8 after the first day), but then I realized this is also a class project and I don't want to be doing the students' homework for them ;). So maybe I'll give my answer later to see if it can match the students' scores.

Other observations:
1) This seems like the funnel problem - continually reacting to changes just makes things worse.
2) This also seems like the problem of controlling a furnace's temperature when doors are opened, loads change, etc. There are well developed algorithms for such situations that could be applicable here.



Tim F

Here's a "game site" which seems to come close to replicating the beer game. Try it a little and see if you agree

http://www.lemonadegame.com/

I think I have a pretty logical and efficient algorithm for placing orders and maintaining inventory (even when the orders at the retail outlet are not simply 8 after the first day), but then I realized this is also a class project and I don't want to be doing the students' homework for them ;). So maybe I'll give my answer later to see if it can match the students' scores.
Tim F

If you "google" on The Beer Game, you will find some sites that do propose computer algorithms for this problem. The primary issue is keeping track of what you have already ordered in the past that is yet to come to you. It is sort of like trying to judge the 2nd derivative (acceleration) of the problem. I am reminded of the problem of "hovering" manually in a submarine. I was never good at judging the amount of water to bring on or pump out. You really had to have a good judge for acceleration and once the depth bottomed out, immediately start flooding water in, rather than waiting to get to the ordered depth before starting to bring on water, or the submarine will build up more and more oscillations about the ordered depth.

Steve,

I know you are teaching Quantitative Methods which involve creating forecasting algorithms but have you thought about how a series of control charts could prevent the problems described?

Tim commented on the beer game similiarity to the funnel problem which is the reason for the question.

Bill

Steve,

I know you are teaching Quantitative Methods which involve creating forecasting algorithms but have you thought about how a series of control charts could prevent the problems described?

Tim commented on the beer game similiarity to the funnel problem which is the reason for the question.

Bill

Forecasting is the last topic covered, and unfortunately the book gives very canned (not data you'd see in the real world) examples to support moving averages and seasonal corrections. That will be a very interesting topic to teach, but at least I do have my Red Bead results to show the fallacies of those techniques on most real world data.

The odd thing about the beer game is even if you at your station realizes what is happening, it is hard to prevent the whip saw effect if everyone else falls for it. You still must meet your incoming orders. I did have one odd session where for some reason the retailer refused to increase her order rate, and by under reacting, the losses in backlogs were incredibly high.

If the order rate at the retailer were indeed "random", then definitely a control chart of customer orders, with sharing of that information all the way to the factory could help keep the system from whipsawing. But in the canned scenario of the Beer Game, you need to measure (maybe on a control chart) some combination of your current inventory, current outflows, and expected inflows from orders currently placed.

Steve,

If you continue to teach the course and play the game, it would be interesting to control chart some of the results. Maybe if you collected the data each time the game is played, you could develop some control charts to prevent the tampering.

When I was learning forecasting techniques from an operations management consultant we used, I felt there was an opportunity to use control charts. The consultant did use control charts of the variation from forecast but it appeared that all he was doing was demonstrating that they were tampering since he used prior data to forecast the next periods demand when all that was really happening was normal variation.

Maybe there is an opportunity to tie this game to your systems thinking article and control charting. Keep us informed of the game and the students' essays on their results.

Bill

Here are charts of the orders and the inventories for one of the two City University teams. This team was "Drink Fast".

Hi Fellow MC506 Students
Wasn't that an interesting experience playing the beer game in class last night? I can't wait to see the results of our manic beer making! The effects of non-communication in a manufacturing environment were so apparent...communication really is the key in so many situations. Had we been allowed to discuss orders, demand, manufacturing levels etc. my guess is we would have been able to keep the inventory and backlog to a minimum.
Lori b

I cannot find the Wholesaler sheet for the "Pale Ales", but here are the charts for the Hiccups.

What an interesting "game" ! I am not quite sure how the backlogs begin, however, being the factory we tried to constantly second quess the distributor, ( who I suppose was trying to second guess the retailer etc...). If there is only one lesson learned from this game it is again interactions. Without the interactions of the 4 parts the whole was skewed! To be successful they (as everyone) needed to work together and communicate. Communication is an important part of making the parts equal the whole.

The backlogs arise due to the "bow wave" or "whip saw" effect in the orders. You can see the first cycle in the ordering chart for the Hiccups.

The customer increased its order to 8 per week at week 4. It took a few weeks for this to deplete inventories and start the panic cycle in orders. Notice that as the order sizes increased, it created even bigger backlogs "upstream" later in time.

The Retailer made a peak order of 20 at week 15.

Three weeks after that, the Wholesaler peaked at 40

The odd thing is the Distributor peaked early at 25 in week 16. The Distributor spread the peak between week 16 and week 21.

Finally, in week 20 the Factory peaked at 50.

Notice the delay in time, and the growth in the order size. Then orders shut down in week 22.

Another odd thing I haven't seen before was the runup by the Distributor and then the Factory in the last few weeks. It looks like the Wholesaler started a small runup in week 25 that peaked at week 27 with an order of 20. The Distributor then peaked at week 31 with an order size of 100, and the Factory peaked at week 33, also with an order size of 100. Meanwhile, the Retailer had completely stopped ordering. I estimate there were about 300 cases of beer in the supply chain at that point, at least 25 weeks of supply.

A study in psychology more than anything, I thought. It was very difficult for a "control-freak" like myself to realize that I, essentially, had no control over outcomes. I found myself feeling very frustrated with my "assistant" when I realized that she/I had not kept track of what was owed to us (the distributor). We knew the backlog for our customer, but not from our supplier. We were lost from that point on. I was the one making the final decision for orders; but, I tried to play psychologist in determining trends: ie. did we control the production cycle, and reduce our backlog by ordering less beer, and trying to play "catch-up" with our backlog? Or, would we replenish more quickly if we ordered one big batch (of 100)? No rationale seemed to connect to outcomes (later found out about the bull-whip effect, which we experienced in spades!). For me, this was an ongoing frustrating attempt to discover a cause and effect from trends, which did not exist. However, was great fun and learning exercise! Anita

This thread is for the Beer Game. This will be an essay question on the midterm related to this experience.

Here are some links for the Beer Game:

http://web.mit.edu/jsterman/www/SDG/beergame.html

http://www.forio.com/nearbeer.htm


The comments from last year's MC506 class are available below, please add your comments for 2005. Worth +1 point on the Midterm.

The beer game provides a provocative and informative lesson in the importance of managing your supply chain. I find it amazing that customer demand at the retailer remained constant beginning at four cases and then moving only to eight cases, yet the supply chain from the factory to the wholesaler tried to anticipate demand and escalated the supply far beyond the system needs. Yes, the bull whip effect is real as demonstrated by the exercise. It is also very interesting to me that when the data from various groups has been plotted on graphs the bull whip remains fairly constant from group to group. Businesses in the real world must strive to improve the information flow in their supply chain, improve forecasting, and improve ordering processes to alleviate this problem. In the game the results were entertaining, but in the real world similar results would be devastating for a business.

It was unfortunate that I missed the Beer Game. From the looks of the spreadsheets, everyone was very busy!!

I was reading one of the articles suggested about supply chains:

"Another classic in Mr Lee's repertoire is the “bullwhip effect”, named after the way the amplitude of a whip increases down its length—just as variations in orders tend to get amplified along the supply chain. Why is it, for instance, that Procter & Gamble has to deal with widely fluctuating orders for its nappies, when babies' consumption is generally quite steady? The reason is that each retailer bases his orders on his own, slightly exaggerated, forecast, thus increasingly distorting the information about real consumer demand. This is one of the most important causes of inefficiency in a supply chain. "

It really struck home with me, as my husband worked for 25 years for a very popular consumer supply company. They have recently closed down several of their plants and have the sales running through another company. I think that the overproduction of goods at several different times was for exactly the reasons stated above, as well as no communication with the individuals actually selling the product. The "bullwhip effect" probably became one of the major reasons for their closure.

I agree with my class and teammates about the frustration of not being able to communicate; that would have been essential to finding out what was going on with the orders (or lack of). As the retailer, once I figured out that I was going to get slammed with more beer than I could sell, stopped ordering and think that by the end of the 50 weeks it would balance out. I'm still trying to figure out that if I had put in a huge order in the beginning, then just stuck with the 8 per time, if it would have worked out.

But....an excellent learning experience - much easier to remember than just reading about the process. I may test it out this weekend with real product.:)
tammy

Loved the Beer Game. The setup was simple but effective. Plant the seed that demand is variable, make it cost more to be short than have inventory, and not allow discussion among the supply chain. Like everyone, I initially feel for the idea to have some excess inventory would be a good thing. I also did my part to support the whip saw effect. After about 20 turns, I realized what was happening. Even though I had my orders written down, I did not keep track of what I needed to send out. Therefore, it was difficult to determine how to get back into balance. Without being able to discuss this with the team, it is impossible to get the backlog flowing thru the supply chain without inducing the next level up the chain to begin ordering more. It would seem that with all of our tools and software that the whip saw effect would cease to exist, yet it is obvious that the lack of communication and "assumptions" that are made exist exactly the same way in the real world.