Takt time: can it be universally applied to all types of production?

This is the question Casey Ng asked in the TPS Only discussion group on LinkedIn. He elaborated as follow:

What are the essential conditions to implement takt time successfully? What are the cases when it fails and if you refer to the fundamental principle of takt and yet couldn’t find the solution? Then what are the exception areas and what alternate solution can be used?

To date, it has generated 43 comments, many with the high level of depth and the implementation examples that are characteristic to this group. It is a new LinkedIn group, with only 118 members — compared with 151,503 for Lean Six Sigma — but passionate and providing meaty technical content. I recommend it in general, and this discussion in particular. Here, I will be including only my own contributions on takt time, and its relevance to the following:


With monuments — the very large machines used for heat treatment, electroplating, painting, cutting of sheet metal, etc. — you usually have to load multiple parts simultaneously in order to meet demand, but these parts often do not have to be identical. I don’t recall anyone mentioning this in the earlier comments, but the parts that you load together can be a matching set rather than a batch. If your takt times are long enough, as, for example, in aircraft production, you can actually process one plane-set at a time, to takt time.

Gigantic products with long takt times

Gigantic products, like oil tankers, with takt times of six months or more, are built out of vertical hull slices made at much shorter takt times in a shop, and then welded together in the dry dock. Pioneered in the US with Liberty Ships during World War II, this is now standard shipyard practice, and enhances of the repetitiveness of the process, applying the concepts of takt time and one-piece flow.

Refueling outages in nuclear power plants

I am puzzled by Todd McCann‘s use of the takt time concept in the context of nuclear plant refueling outages, a problem I had the opportunity to work on 20 years ago in France, not in the US. I don’t know who the top performers are in this area today. At the time, it was a tie between the French and the Japanese, at about three weeks from shutdown to restart for one reactor, which is substantially longer than the 193 hours you were quoting for 2006, assuming that the work continues 24 hours/day, 7 days/week.

The French performance was accomplished by standardization. They had 55 reactors with only two designs, producing respectively 900MW and 1300MW, run by a single company. The procedures were the same everywhere, with any improvement quickly shared. The Japanese performance was based on using techniques from TPM. They had nine different reactor designs, run by different utility companies.

Even though they had the best performance in the world, I saw many opportunities for improvement, based on borrowing techniques from SMED, improved planning and scheduling for materials, tooling, and the 1000+ contractors involved, and operational details. Such a detail, for example, was security. Their procedures were effective at controlling access, but inefficient, with utility employees at all levels spending too much time getting contractors into and out of the facility.

The concept of takt, on the other hand, did not strike me as particularly relevant, given that a refueling outage is a yearly burst of intense activity for any given reactor, as opposed to a repetitive process.

Rate work versus response work

More generally about takt time, most businesses have both what my colleague Crispin Vincenti-Brown called Rate Work and Response Work. In manufacturing, if you do a Runner-Repeater-Stranger analysis of your products, Your Runners and Repeaters are rate work; your Strangers, response work.

Runners are products with enough volume to warrant a dedicated line. Repeaters are products that you group into families that, in aggregate, have enough volume for a line. Strangers are all the other products, including R&D prototypes, sample quantities of new products, spare parts for obsolete products, and any other special request. Even in aggregate, they account neither for a high volume nor for high revenue, but you still must produce them promptly. They require a small job-shop set up with your most flexible equipment, staffed with your most versatile operators, and its own operating policies.

While takt time is fundamental to line design for Runners and Repeaters, it isn’t much use for Strangers.

Non-repetitive operations

I have a hard time seeing the relevance of takt time in the absence of repetitiveness. In an assembly process, the takt time gives you an upper bound for the process time at each operation. As you broaden the mix of products you assemble on the same line, it becomes more difficult to balance the work among stations. Past a certain point, you are better off using approaches like bucket brigades, a.k.a. bump-back system, or even a yatai, which are not based on takt time.

Takt time and its calculation

Outside of mathematics, concepts are not reducible to formulas. Time/Demand is the way you calculate takt time, but it tells you neither the rules by which you are supposed to use that number nor how it maps to shop floor activity.

In mass production plants, managers use the inverse of this ratio: Demand/Time, which gives you the same information. Mathematically, working at a takt time of 1 minute and making 60 units/hour (uph) is equivalent. Yet, you and I know that, depending on whether the manager thinks the plant is producing at a takt time of 1 minute or making 60 uph, the shop floor will be radically different.

If you think in terms of uph, it doesn’t matter if nothing comes out for the first 59 minutes of each hour as long as all 60 come out in the end. If you think in terms of takt, 1 unit will come out like clockwork every minute.

What this says is that there is more to takt time than the formula. This is discussed extensively in Lean Assembly, with the following definition of takt time:

“Assuming we complete the product one unit at a time at a constant rate during the net available work time, the takt time is the amount of time that must elapse between two consecutive unit completions in order to meet the demand.”

As I recall, this is,more formally, the way Ohno described the concept in Toyota Production System.

Takt-driven production as the ideal state

The takt time allows you to define an ideal state, that John Shook and Pascal Dennis call True North, but that I prefer to call takt-driven production. In this state, you perform all operations one-piece at a time with process times that exactly match the takt time, and with instant transfer to the next operation at every beat. Of course, it is never perfectly realized, even on an assembly line. Real lines can only be approximations of it. The point is that it gives us a direction.

All deviations from takt-driven production translate to Ohno’s waste categories, overproduction, waiting, excess inventory, etc. Since any local project that move production in this direction eliminates waste, it can be undertaken with the confidence that it contributes to global improvement and is not sub-optimization.

Chronos, kairos, and takt time

Joachim Knuf : “… The ancient Greeks differentiated between two types of time: chronos (chronological, sequence of intervals, typically of equal extension) and kairos, best thought of as ‘the opportune moment.’ In this case, intervals begin and end under sets of conditions. This concept applies, for example, to healthcare, when next process steps are initiated by a preset configuration of values (patient’s blood pressure, glycemic index, bowel movement), not by elapsed time…”

I had never heard of Kairos, but, if the ancient Greeks made the distinction between Kronos and Kairos, why shouldn’t we? There is a rich toolbox associated with the pursuit of takt-driven production. Where the concept of takt does not apply, we can’t use these tools. As you said, extending usage of the word to Kairos-driven activities just adds confusion. These activities need different tools, and Casey pointed out some of them in his comments on Strangers. Let us keep different words to hang them on.

8 comments on “Takt time: can it be universally applied to all types of production?

  1. Comment in the TPS Only discussion group on LinkedIn:

    Takt time according to Taiichi Ohno in his book “Toyota Production System, beyond large scale production”
    “To prevent overproduction and make items as needed, one by one, we have to know when they are needed. Thus, the appropriate takt time becomes important.Takt is the length of time, in minutes and seconds, it takes to make one piece of the product. It must be calculated in reverse from the number of pieces to be produced. Takt is obtained by dividing the operable time per day by required number per day (pieces). Operable time is the length of time that production can be carried out per day. “
    This may be useful for us to understand the meaning and purpose of takt time when apply in the context of TPS, which have very clear difference of meaning and application from simple cycle time.

    • You made me reopen my copy of Ohno’s book. I had read it in Japanese 30 years ago and quickly found the section where he introduces Takt, which I had bookmarked and heavily annotated.
      The phrasing is ambiguous, because he says “how many minutes and seconds you need to make it.” (タクトとは、一個あるいは一台を何分何秒で作らなければならないという時間のことである。)
      This can be interpreted to mean how long it takes to make one unit, and this is apparently how it was translated into English. In the following sentence, however, Ohno introduces the formula, which makes it clear that he is talking about the required interval between units and not the process time of each unit.
      When I later read Jon Miller’s translation of Ohno’s Workplace Management, I was struck by how different the style was from Toyota Production System. Workplace Management feels like a conversation with Ohno around a beer, in which he says exactly what is on his mind. I have long suspected that Toyota Production System had been, if not ghostwritten, at least vetted and edited by Toyota’s PR department, which may have accidentally or deliberately introduced ambiguities.

      • Comment in the TPS Only discussion group on LinkedIn:

        Hi Michel: Just to clarify, an accurate translation of タクトとは、一個あるいは一台を何分何秒で作らなければならないという時間のことである。would be: “The takt is the time in minutes or seconds in which one piece or one unit must be made.” Note that Ohno is saying “…MUST be made.” So the English: “…you need to make it…” was probably not meant to mean: “…how much time it takes to make it…” It was probably meant to mean “…the time you need to make it in.” I haven’t read the context of the translation, but the choice the expression “…you need to make it…” is a candidate for editing. It’s a bit ambiguous. Sorry to be wordy, but I am a translator, so I can’t help it.

  2. Hello Michel… So your above statement from Mr. Ohno means takt time is not the time allotted to start and finish a task within a given moving assembly footprint… is this correct? I was under the impression that takt time was the time allotted to start and finish a task within the given cycle time of a single moving conveyor footprint. I’m curious now….

    • If you assemble on a moving conveyor at a takt time of 30 seconds, you can put marks on the belt so that one mark passes each station every 30 seconds, and there will be one unit of product between every two marks. The marks materialize the 30 seconds you have available to work on that unit of product at each station, which means that each assembly job at each station must be designed to fit within 30 seconds.

      The takt time is not how long it takes to do a task, but how much time you have to work on each unit at each station. These get confused all the time, even in Germany, where the word “Takt” comes from.

      The original “Taktsystem” was developed in the German aircraft industry in the 1920s and 30s. The planes were not assembled on a moving conveyor but in a sequence of stations, with movements between stations at fixed intervals, called “Takt.”

  3. Hello, I have a question. If I have different products of same family in the same line, but demand of one of them is significantly lower than the others, takt time should be different? I mean do I have to weight?
    At the end I want takt time to balance my line, but probably for high runners I’ll need 4 operators, but what about low runners? Do I need the same 4? Takt time should be different in order to optimize my resource? How can I group to define just one takt time? What should be the process?

    • A mixed-flow line running at a takt time of 1 minute completes one unit every minute, but consecutive units can be for different products. If one product is needed every two minutes, it occupies every other slot; if every ten minutes, every tenth slot.

      How many operators are needed to make one product is a function of the amount of labor needed to make it, not its takt time. It is difficult to run in the same line products with vastly different labor content. To equalize it, you can offload more or less work to feeder lines for different products, or apply more or less automation.

  4. Pingback: Driving Improvement Through Systems Thinking | Gregg Stocker | Michel Baudin's Blog

Leave a Reply

Your email address will not be published. Required fields are marked *