The Airplane game is a hands-on demonstration to experience the difference between “push” and “pull” in less than an hour. Participants form a line to manufacture a paper airplane by each performing one or more steps. In particular, participants see how how excessive work-in-process (WIP) hurts lead-time and quality, as well as how to solve these problems by using a Kanban-based “pull” system that limits the maximum amount of WIP in a line.
All of the materials you need are available for download. While we strongly recommend to actually play the game with your co-workers, this lesson is organized as a walk through that provide you with some intuition of what is happening.
The Airplane game models a manufacturing line with eight stations that turn a sheet of paper into a model airplane. These stations consist of the various folding operations, some involving a ruler to enforce accurate instructions, applying decals using various markers, stapling the fuselage together, and finally performing a quality inspection. Defects are introduced by exchanging a marker with one with a different color.
While it might be tempting to replace the paper airplane with a more complex object or even a real manufacturing product, using an object that can be completed in a minute or less enables you to experience problems that only arise at larger quantities in just a few minutes, while maintaining a birds-eye view of your process.
Find a table that allows eight people to comfortably sit and has enough space for instruction sheets, a work area at each station, and a stack of papers. Print the instructions, hand them to the participants and distribute the required tools. If you have less than eight participants, you can combine stations (not recommended). You will need:
- A stack of US Letter or DIN A4 paper
- Printed instruction sheets
- A stapler
- A red marker, a green marker, a blue marker, and two black/blue ball pens
- A ruler
- A stopwatch
The gallery below shows the instruction sheets:
Your first manufacturing line: push
The natural way to assemble something in a team is for everyone to perform their task, then push work-in-process (WIP) to the next station. By literally pushing it over, WIP is processed in the order it is received (first-in-first-out). Once the game starts, the majority of the players are idle, then gradually start working, and all working only when the first plane enters quality control. While this is not a problem in a manufacturing line that makes the same things every day, the resulting waiting time is indeed a problem in high-mix, low-volume environments.
We are now having a line that is fully loaded and in which WIP will build up in front of stations that are slower than others. As there are eight stations, you will need to start production of 30 planes to clearly see where the bottlenecks are.
Now mark an empty sheet with the current time and send it into the line. You can now observe when the marked sheet arrives at each station and when it finally leaves the quality control station. Check the time again and calculate the elapsed time. This is your lead time. You might have already observe that WIP negatively affects your lead time as your work piece needs to significantly wait at each bottleneck.
In order to simulate defects, replace the green marker with a blue marker. Tell the person at station eight (“quality control”) to be extra vigilant for this defect. Once the defect has been detected, change the blue marker back to a green marker, and count the number of planes with the wrong decal. You might notice that the number of broken planes again grows with the WIP in your line. Not only takes a systematic defect longer to detect, but the WIP actually acts as a buffer for broken planes, leading to more rework.
You now have experienced first-hand how WIP negatively affects lead time and quality. Imagine now you were making actual airplanes. This adds another dimension: space. Reducing WIP will help to substantially reduce storage space within your factory.
Your first “pull” system
We will now introduce a basic “pull” system that will limit the amount of WIP in your line. For this, delineate an area between each station that keeps only one piece of WIP. Discard all your other WIP, in particular all broken planes. Now instruct your team to only push WIP forward if the area is empty and wait otherwise.
Continue making airplanes and again introduce a time-stamped sheet as well as a “defect”. Again measure the lead time and the number of broken planes after the first defect is detected. You will find that your lead time and required rework (broken planes) is reduced by a factor of three or four. What you have just done is also known as a “single piece flow”
Notice that both systems are equivalent if your line is perfectly balanced, that is every step takes exactly the same time. Using
“pull”, you are therefore trading off lead time and quality with wait times at every station. In practice, a line is never perfectly balanced, however, much like a bunch of cars can never drive at the same speed on the highway without someone braking inducing a traffic jam. A pull system is therefore the equivalent of throttling traffic on the highway in order to prevent traffic jams. While individuals might be complaining about the seemingly unnecessary speed limit, they all will get to their destination faster. Also note that the rate at which airplanes are produced is exactly the same in both systems as both are limited by the station with the slowest cycle time. What looks like wasted wait time is therefore just overproduction.
It is rare that not using a pull system has a positive impact on customer value. This might be the case in high-mix, low-volume situations in which only a limited amount of items are being made. Here, individual lead-time is not as important as total makespan, i.e. the time it takes to make the entire lot, and overproduction will eventually find its way into a final product. Yet, depending on changeover time, it might make sense to change over each station in a rolling fashion and use “pull” to keep flow constant.
Pull using Kanban cards
Using a dedicated area to limit WIP (also known as a “Kanban square”) can also be achieved using so-called Kanban cards. In a Kanban system, workers only perform work when instructed to do so by a Kanban or “lot” card. In its most basic implementation, an Kanban card is attached to every product in a warehouse. If the product is sold, the Kanban card is removed and brought to the beginning of the line, triggering the next piece to be made.
Play the airplane game as in the push version, this time attaching sticky notes to each airplane. This is not any different and WIP will build up. Now stop issuing new sticky notes, but use sticky notes as they come out at the end of your line to trigger the next build. You will find that the last station is limiting the rate at which Kanban cards are released into the line, eventually leading to the WIP that has been built up during push diminishing.
So how many Kanban cards (sticky notes) should we use? If we use one per station (or eight total), we will approximate single piece flow. It turns out that using more cards (and you are welcome to try using the Airplane game) will gradually reduce lead time and quality, until the system is equivalent to a basic push system. The key in “pull” is that you introduce measures that put a hard upper bound on the WIP that can be in a system, which results in the know well understood advantages in terms of lead-time, quality, and space.
General advantages of “Pull”
You might have noticed that the airplane game is pushing some very specific notions of “lead time” and “quality”. If you don’t care about the actual time a specific item needs to get through your line or you have different ways of detecting defects early, the benefits of a pull system are not immediately obvious.
The intrinsic benefit of a pull system is that it limits your inventory. Using Pull will not only your WIP, but also your stock (if using a Pull system correctly there too and ordering at the same rate that you deplete items), and inventory of finished goods. This is advantageous for two reasons: first, inventory cost money to purchase (interest) and storage space. Second, every production process has an optimal inventory size. This becomes clear, if you – just briefly – imagine your inventory to be too small. Neither will you be able to meet customer demand, nor will you be able to make enough goods. Likewise, if your inventory (including stock, WIP, and finished goods) is too large you will incur unnecessary cost, storage space, and disorder. Using a Pull system will allow you to dial in a number that is right for your suppliers, process and customer demand, and will come with aforementioned advantages even if requiring constant adjustments.