Ask Gemba: Nuts and Bolts of the Andon System

By Jon Miller Published on April 12th, 2009

Bas Timmerman asked:

We are in the process of designing a synchronised flow process with an intermittent transportation system for a new assembly line. We have no buffer positions in the line, so it is critical that we organise support functions extremely well. An andon board is central to visualising line stops and signalling for support. Can you share your experience with ‘the nuts and bolts’ of such an andon-triggered support organisation? How does this work in practice in case of a line stop, who is triggered when by whom for what support? How are problems best escalated? How do you balance out immediate countermeasures and root cause corrective action?

Thanks Bas. This is a great question. I love the andon system because it’s so simple but to do it right there are so many things that management needs to put in place first. The andon system is really a great way to instill the right kind of mindset and behavior for lean operations to succeed. Each situation is different and you will need to work out the particulars of designing and running your andon system by trial and error. Experience has been my teacher, and here are a few things that will guarantee that your andon system fails. So don’t do these things:
Get rid of all buffer in the line. This is rather silly. Even at Toyota there is something called the fixed position stop system which allows the lined to travel a certain distance within a “buffer” space to allow the team leader to resolve the problem before the line stops. Several years ago one of the Toyota factories in Japan experimented with chopping down the ling lines into several smaller lines, in effect duplicating WIP and increasing buffer, but reducing the impact of line stops since any one assembly line stopping did not stop the other parallel lines from stopping. So think this through and use an appropriate level of buffer based on your reasonable ability to respond and resolve issues in real time. This should be not so much that you are hiding problems daily, and not so much that it keeps the support staff comfortably in their offices, but enough so that the system can operate.
Stop the line for every little problem. Same as above, we need clear rules on when a problem requires that the line keep running until a certain point in time, and when it must immediately stop (immediate safety issue or clear quality defect). The andon system is a “stop and call” system that causes the management and support staff to “stop and fix” if the team leader of the are cannot resolve the issue. The “fix” of “stop and fix” being on the line as much as possible for most problems, which should be small ambiguities or helping a worker catch up.
Set a lower limit on the size of a problem that requires workers to pull the cord. Some problems are too small to bother the team leader with, or intractable but negligible (minor near misses, minor flaws) and so we should set a threshold beneath with people should just let it go, and not pull the cord. Right? Wrong. At the Toyota factory in Georgetown, KY they reported 8,000 andon cord pulls during a typical day in full operation with 3,000 or so employees. If you are not prepared to thoroughly expose and address the most minor issues, even if this means an increased burden to the support staff int he beginning, you might as well remove the light bulb in the red part of the andon right now.
Keep the support resources comfortably in their offices. The andon system is more of a factory thing than a change to the way that engineers, managers and support staff work, right? Wrong. The team leader of the area is the first to respond, and they should be no more than a few steps away from their team, all day long. The group leader (supervisor) is the next level of elevation, and while they might have an office, it should be on the shop floor within their zone, ideally without walls, and with them rarely in it. The support functions of the area manager, engineer, quality, purchasing and so forth should be minutes away. Another way to think about this is that the design of any workplace should be based on minimizing the number of visual barriers (walls) and distance between the problem responders on-site and the points of problem occurrence (the line). How many doors, walls, and meters of walking between your head of site and the gemba?
Design the escalation system around the schedule of the support staff. These are busy people after all, with meetings to go to, chairs to sit in and webs to surf. Right? Wrong. A typical escalation for a line running at a 60 seconds might be seconds for the team leader, one minute (takt) for the group leader, 5 minutes for the area engineer or manager, 10 minutes for the plan manager and so forth. The length of time completely depends on the pace of work based on work volume and the cost of one second of down time. The entire management and staff exist to support the line which adds value and delivers to the customer, so the escalation system needs to be designed around the needs of the line. As a rule, containment should be immediate. Root cause countermeasures should be within 24 hours, but this depends to a large degree on the stability of the process (number of problems), the severity of the problems and the problem solving skills of the support staff. The assignment, location and schedule of support staff coverage is then synchronized with this rhythm.
Just pull the cord and expect the world to come running. It won’t happen unless you are already a high performance work team and the red lamps are the only missing bits to your problem response system. Each person should have specific and personal responsibility for responding to problems raised within a particular zone. Each person should have a clear person that they can go to when unable to resolve the issue, all the way up the chain to the head of site at a minimum. This requires thorough training, communication, practice and continuous improvement of the system.
Spend more time and money on the hardware and software than on the humanware. The andon system is all about lamps, digital display boards, automatic SMS alerts to the plant manager and other high tech toys, right? Wrong. The success environment of an andon system includes at a minimum:

A blame-free culture. It is completely safe to expose problems. Leaders ask “Why?” and not “Who?”

A visual (and audio) alert mechanism (andon, flag, etc.) with clear line of sight at both the line side and from the plant manager’s perch

A documented and agreed escalation system

Teams with reasonable span of control for team leaders so that they are able to respond quickly

Support staff positioned locally and able to respond rapidly

Training for all levels in roles and responsibilities of the andon system

Standardized problem solving process

Continuous improvement of the andon system

Hopefully this helps. A functioning andon systems is one of the best investments of time in any type of industry, process or environment. If you don’t have one, I recommend you experiment with the andon system. It will teach you a lot about what foundational elements of a lean organization you are missing.

Jon Miller

Jon has dedicated his 25+ year career to the field of kaizen, continuous improvement, and lean management. Jon spent the first eighteen years of his life in Japan, then graduated from McGill University with a bachelor’s in linguistics.