My background is high volume automotive production like Ford. I am now involved with static build production of HVAC units. How do I best apply stop the line methods with static builds? ie no line to stop.
Thanks for the question Graham. First of all, in the longer term we need to challenge the idea of what’s fixed and what’s stationary. Making work flow is the best way to make problems visible since there is a clear binary OK / NOK signal (flowing, not flowing). Boeing builds aircraft in a flow, Toyota Home builds houses in a flow, and companies like York make HVAC units in a flow. Where there is a will there is a way. Even so, volume, mix or cost factors may not make this practical for you in the near term so let’s address the issue of how to apply the TPS principle of stop and fix in a stationary process.
The term “stop and fix” is misleading. Many think this means that the product is flowing and when there is a problem it stops flowing. In fact this is not always the case. The product may continue to flow even when there is a problem until it reaches its fixed position, within a fixed position stop system. When a problem is found the worker stops working in order to alert a local team leader or staff member. So based on that understanding, you can have a stationary process so long as there is a system which enables the worker to stop and call for help.
The call signal itself should be as simple as possible. An audio-visual signal is best, something as simple as a flag or a light that switches on. Accompanying pleasant (or you may prefer annoying) music can help expedite the response by the supporting engineer or staff member. Line of sight visibility of the stop signal is very important, and this implies that the support personnel are within sight of the process. The support staff are located on the shop floor.
The team leader or supervisor responds to the call within an agreed time frame. In most cases this is within takt, or without delay, whichever is quicker. For stationary processes and low volume the takt may be very long and in these cases it is better to set the response time seconds or minutes. It is better at first that the agreed response time can be 100% met even if it is a bit longer than desired. Setting a fast response time and then having excuses for being late is not acceptable.
When responding to the call, problem containment is the first step. The responder needs to ascertain, “Is this problem a safety issue? Will this problem cause a defect? Will it delay shipment? Will it affect on-time for the downstream process?” and so forth. People and customers must be protected first with immediate containment plans, followed by longer term root cause countermeasures. The problem should be escalated as necessary if the problem cannot be addressed within an agreed time frame.
As pointed out in an earlier article, the success of an andon system or stop and fix system depends on quite a few things, most of them organizational. The same is true in a stationary environment. To that list we can add the need to show progress against time since the physical flow to show this is lacking. A clock that counts down lost time is one way to do this. A better way is to divide the work into packets and attach a card or other visual so that as each packet (say 30 min) of work is finished this card is placed on a board showing that work to be completed by 930AM has indeed been completed. So in summary the most important questions are, “How will we make abnormality visible?” and “Who will respond to each level of call?” and “How quickly will they respond?”
Hope that helps Graham, let me know how you get on.