The problem of fractions. Something to think about before implementing Kanban.
If someone wants to pretend that they understood TPS, all you have to say is one by one is better than batch. I was not smart enough to say such a phrase and started the following discussion with my father, TPS coach.
Me: "I like batch production."
Me: "Ok, I love one-by-one. But think about this. Have you ever seen a "True" batch system outside Toyota's general-purpose machine area?"
And he finished the beer in front of him and ordered another one. He remained silent. Then the cold beer arrived, and he drunk like half of it.
Dad: "No. I have never seen a "True" batch outside Toyota."
Me: "So, Toyota has the best batch production."
So, what is this "True" batch production?
In the batch system, general-purpose equipment and methods are used to produce small quantities of output (goods or services) with specifications that vary greatly from one batch to the next.
In theory, the batch size is calculated based on the cost of setup. Cost accounting teaches this economic batch quantity calculation. The truth is that there is always a massive gap between this theory and the reality (shop floor).
The first gap is that this calculation did not take place. Many places have the batch size determined not based on calculations but based on experience or the opinion of somebody who had the loudest voice.
The second and more critical gap is the problem of "Fraction." When we go to the shop floor, we observe partially filled pallets or just a few materials inside a box. The problem of the fraction is these incomplete batches. And in most cases, these problems are neglected, although these are issues that are destroying the economic batch quantity.
If the management chooses to attack the fraction problem, many TPS philosophies are the only method to solve the issue.
1. Build-in quality
One of the common reasons why we have this fraction problem is because of defects. A general approach to this is to add allowance to defects. But this approach is strange. The moment you add the allowance, it is not the economic batch quantity anymore. Even if you said, let's say 1% for defect, sometimes the process makes 2% or .5% defect, then we will continue to have the same problem. The only way to produce in a batch is to create built-in quality.
One of the devasting moments in batch production is when you have to change over in the middle of a batch. The batch break happens due to rush orders. No, wrong statement. The batch break occurs because the order is not levelized. Taichi Ohno says that the assembly line needs to level so that the fluctuations of orders do not impact the upper stream of the supply chain, which typically does batch. Because batch production is less flexible, the need for leveled orders is higher than one-by-one production. Unfortunately, batch production creates false perception that it can handle any order because of the large quantity. Batch production is not flexible, and they are the ones that should be screaming for a more stable order.
Another issue that prevents accomplishing "true" batch production is machine issues. If there is a mechanical problem, the batch will stop. Similar to quality issues, many have tried to overcome them by adding allowances. The story is the same as quality, and this will not solve the fraction issues. The only way to solve this is to conduct excellent preventive maintenance.
The truth is that in most traditional batch production, it is allowed to break the economic batch quantity. When you break the economic batch quantity, technically speaking, we are not doing batch production. It is something else, but somehow it is accepted. Such rule-breaking is becoming the norm. Unfortunately, such habits are spreading to one-by-one production as well. Instead of taking action one by one, they send all the problems into a "parking lot" and work only on the "biggest" issues. The result will be the same disaster as wrong batch production. So the crucial difference between the two is not the quantity of production. It is the management taking action one-by-one or batch on problems. Without such effort by management, we can destroy any system.
I used to call the state that broken rules "chaos." The state following the regulations as a "system." Unfortunately, this method of distinguishment is becoming challenging to apply since many "chaos" is implementing information technology as a "system." Many information technologies collect data from every station and then re-calculate according to what they have made and adjust. But this adjustment is the best option under poor efficiency. Although the IT will make it look like the condition is in order, the need to problem solve one by one does not change.
Once I did get a chance to work at a factory doing "true" batch production.
Me: "what is your batch size, and why?"
Plant manager: "It's 6. And it is because the changeover time is two hours."
Me: "Ah, so if I Kaizen the changeover to 20 minutes, can we try one by one?"
Plant manager: "Yes, let's try."
After the first workshop, the changeover time was below one hour.
Plant manager: "Great. We should start three by three production from tomorrow."
After few more workshops, the changeover was hitting below 30 minutes. The plant manager pushed to implement one by one since he thought the organization needs to see that he is serious about this.
By the time the plant started one-by-one, my dad had come and asked if we shouldn't keep challenging the changeover to reduce the size of "1".
The plant manager agreed.
Later I had an opportunity to talk with the plant manager over a beer.
The plant manager praised TPS, but I asked the question, "why was his plant doing a "True" batch? The other sibling plants were not."
The plant manager laughed and said the plant used to work like the sibling plants. But when he was the production scheduling manager, he got tired of daily firefighting. He told the team that his group would not change the schedule due to problems. If they don't keep the batch, the cause of the broken batch reported as is. This started the plant to follow the batch production.
I told the plant manager that his anger might be similar to what Taichi Ohno felt when he started his activity.
Finally, please do not try to implement Kanban when you can't produce in batch.