Takt Time Concept Still Misunderstood | LinkedIn Discussion

Mark DeLuzio

“What does TAKT Time mean to you and how have you used it to better your business?”



Sourced through LinkedIn

Michel Baudin‘s comments: It’s 2017, and this question should be unnecessary, but the responses reveal that confusion about this concept is still widespread. As I belong to The Takt Times Group, I felt compelled to participate; at the same time, I didn’t want to repeat everything else I have written on the topic.

About what Takt Time is, there was a litany of the usual oversimplifications about “the heartbeat of the customer,” “the rate of customer demand,” “consumer demand for capacity allocation,” all of which confuse using takt time with making-to-order and forget about the effect of net available work time. Before presenting my own perspective, however, I would like to quote a few posts I agreed with.

Dan Micheau got it right when he wrote:

“Takt time is calculated by available time divided by required output. The units are seconds, minutes or even hours. Anything longer and I’d suggest introducing the concept of “pitch”. If we can set up our system to perform each takt time (stability!) then we can start to improve that system.

Every takt time is a chance to run a scientific experiment to improve. It exposes abnormalities (we welcome them!) and it’s crucial for pacing and synchronizing other parts of our system. I don’t understand how operations can call themselves Lean when people don’t know what the takt time is (or what it means). I always think of takt time as the interval of time that the customer’s giant hand reaches into our factory to grab the next product.”

Mark DeLuzio himself had a good description of the use of takt time in on-going operations that I am used to, with yamazumi charts in assembly or work-combination charts in machine-centered operations:

“You never balance operators to TAKT. You front load all operators and bring them as close to TAKT as possible, and all of the wait time is pushed down to the last operator, who is called the “least man.” So, if TAKT time is 60 seconds and total operator cycle time is 150 seconds, the first two operators are loaded to takt time (60 seconds) and the last operator is only loaded to 30 seconds.

As you continue to do Kaizen and reduce operator cycle time, let’s say by 15 seconds in total from any of the operators , you front load again bringing the first two operators back to takt (60 seconds) and reduce the last operator to 15 seconds. As you continue to Kaizen and reduce operator cycle time, you will eventually be able to eliminate the 3rd operator and run the cell with two people.

However, if you balanced the operators they would all have 50 seconds of load, and you would produce faster than TAKT time therefore overproducing. You also will forever be stuck with three operators.”

And Greg McFalls chimed in:
“We used this same idea in a factory that had over a billion possible variants down one production line. You front load the line. The last operators will be under takt time. Think of takt time as the price is right. You want to be as close as possible without going over. You start with operator 1 and work your way down. Technology goes a long way here and this can be done very quickly by computers. Balancing is a misnomer. You don’t balance the operators to each other – you balance to takt time.” 
My own contributions to the discussion went as follows:
Definition: Assuming we complete the product one unit at a time within the net available work time, the take time is the time that must elapse between two successive unit completions in order to meet the demand. (Lean Assembly, p.43) Sorry to disagree with Shahrukh, but it’s applicable in machining or fabrication as well as assembly.
For details, see:

Takt-driven production is the ideal situation in which each step of the process takes exactly one takt time to complete and transfer to the next step is instantaneous. It only exists in approximations but, if realized, would be free of any of Ohno’s wastes.

To Shahrukh Irani‘s assertion that takt time was only relevant in assembly, I responded that it’s the demand pattern, not the process technology, that determines whether the concept applies. I have been personally involved in cell projects based on takt time in:

  • Making cores in automotive foundries
  • Machining engine blocks in automotive
  • Secondary machining of aircraft components
  • Gear making
  • Machining surgical implants
  • Welding agricultural machinery components
  • Machining surgical implants
  • Welding agricultural machinery components

Usually, takt time is relevant directly to runners, and, with heijunka, to repeaters. It does not apply to strangers. Often the complexities in machine operations are due to mistakes in equipment selection, favoring a small number of monuments over a large number of smaller, less flexible machines that can be dedicated to a runner or a family of repeaters. In a machine shop, this means using a machining center with 60 tool pockets to do repetitive work that you could do more cheaply and with the same quality in a cell of lathes, milling machines, drill presses, etc.

The consideration of takt time should drive both equipment selection and shop floor layout, to avoid all the specific problems you raise. It is of limited use if all you can act on is scheduling. You may nibble percentages in performance, but not achieve orders of magnitude improvements.

Dimensionally, the takt time is not a pace, just like a wavelength is not the same as a frequency. Mathematically, a takt time of 1 minute and a pace of 60 units/hour means the same. Practically, designing a production line to put out one unit every minute gives you a different result from designing it for 60 units/hour. That’s why it matters. When I walk into a plant and ask the plant manager what the takt time is, if he or she says “59 seconds” without hesitation, I have an idea of what to expect on the shop floor. If the answer is “What’s a takt time?” or “We make 61 units/hour,” it gives me a different idea of what to expect inside, like batch processing and WIP floating around.

According to another responder, the idea of takt time as a rate is found in the iSixSigma dictionary, along with the statement that “Takt is the German word for the baton that an orchestra conductor uses.” It is actually “Taktstock.”  I have heard Takt used in German for a bar in sheet music, the time between trains on a line where they run at fixed intervals, and strokes in a car engine. For example, a 4-stroke engine is a “Viertaktmotor.” Unfortunately, nothing named “Six Sigma” is a reliable source on TPS concepts. The difference between a rate and a time is exactly the kind of nuance the Six Sigma community misses.

Also, the net available production time influences the takt time as much as the demand. If you go from 1-shift operation to 2, you double the takt time without any change in the demand. You calculate takt time as a ratio and, in a ratio, the numerator matters as much as the denominator.

The concept of organizing production around a takt time originated in the 1920s in aircraft manufacturing at Junkers in Germany, was taught by German engineers at Mitsubishi Aircraft in Nagoya in 1942, and migrated to Toyota along with aircraft engineers after the war, where it was adapted and refined for car manufacturing. For details, see:

#takttime, #lean, #TPS, #SixSigma

11 comments on “Takt Time Concept Still Misunderstood | LinkedIn Discussion

  1. I think we should see Takt Time as the customer accurately defined purchase rate, not sales forecast. It takes the section of the voice of the customer that defines their volume requirements into their supplier’s workplace. It also addresses what Ohno saw as the greatest waste; “After trying to identify and eliminate every kind of waste, we concluded that the real waste was making products that don’t sell.” Taiichi Ohno.
    Takt Time is also a central part of Standard Work. Having heard the customer’s voice, SW then defines; the work sequence, number of operators, machinery and equipment and the standard WIP required to produce at that rate.
    We talk about QA (quality assurance); but standard work also gives us CA (cost assurance) & DA (delivery assurance – Tact Time). All these costs & values are now controlled at the point of occurrence, not after the event in the accounts or other departments. I think this is a point often missed about Takt Time and Standard Work.

  2. Takt time is easy to use when production line is a conveyor belt where speed is given by the belt. However, in many companies (i.e. automotive suppliers), production is performed in cells (mostly flexible and U shape). In that case, Takt time is not easy to calculate when there are SKUs with different “time per unit” in a cell.
    Because Takt time, which is calculated by dividing available time by required output can be different per SKU. What works well in these situations is to calculate customer orders as minutes and see the needed production minutes to satisfy the customer. Decide on the production strategy by setting the shift pattern, open days, nr. of operators, etc. on the cell.
    Once production strategy is clear and production minutes are fixed, distribute these minutes to SKUs with an intelligent algorithm to stay in the limits of the production minutes. And all these can be done by the help of an advanced planner.

    • You are describing Advanced Planning and Scheduling (APS) as the first resort when it is the last. It’s what you do when all else failed. What is “all else”?

      First, when you design cells for repeaters, you have choices to make in grouping the repeaters in families to which you can dedicate a cell. It’s the point where group technology has a role. You use feature and process similarity to mitigate the variation in resource requirements among family members.

      If that’s not enough, you may offload some operations for some products to a feeder line, so that the work remaining inside the cell is closer for all family members.

      If that’s not enough, you can apply different levels of automation to the different products in the family at selected operations within the cell.

      If that’s not enough, you define a fixed sequence in which to process the products in the cell, so that it’s the sequence that becomes “the product” rather than the individual family members. This requires frequent adjustments with changes in mix.

      If that’s not enough, instead of takt time, you apply the bucket-brigade concept to organize the work in the cell.

      If it’s still not enough, then you may consider using an advanced scheduling system. At that point, the cell concept itself may not be a good fit.

  3. Great post. I would like to add that when talking about takt time, most people mean only the customer takt time. However, this is only one side of the coin, and there is also the takt time of your production system (regardless of assembly, making cores in automotive foundries, or machining surgical implants).

    This takt time of the line/process/production system is equally important, since it has to match the customer takt, ideally the system takt being a notch faster than the customer takt. Yet, while everybody knows customer takt (albeit often incorrectly as you stated above), I usually receive blank stares when I talk about line takt.

    • I can see plenty of practical uses for what you call “line takt” to guide both the design and operation of production lines. It is often possible to make goods one at a time at fixed intervals during production time to match the demand.

      On the other hand, I don’t see what practical use I could make of a “customer takt,” given that manufactured goods are not commonly shipped from factories to customers one unit at a time at fixed intervals.

      Customer demand comes in many forms, ranging from individual consumer orders placed online to 10-year defense contracts for delivery of a fixed number of units every month.

      Usually, Production does not respond to the raw demand. It doesn’t make to orders in the sequence they arrive in. In most plants, Production’s customer is actually the Sales department, which decides what it needs in order to serve the market. Even at Toyota, it includes units that have not been sold.

  4. Comment on Linkedin:

    For some people, a manufacturing system simply means one or more production lines each of which makes somewhat similar products one after another. They may even make a convenient assumption that the somewhat similar products on each line have steady demand or their cumulative demand is steady over time. Lean methods work wonderfully for such systems.

    However, I am yet to see a TPS consultant who has a good grasp of “real” production issues of complex job shops.

    • I am sure such shops must exist, but I have yet to see a machine shop that has exclusively sporadic demand, making exclusively one-of-a-kind or short runs of many items. The closest I have seen to this is the department making fixtures and jigs inside a machine shop. They rarely make more than one unit of a fixture, but they still go through a repetitive sequence of phases, if not operations.

      All the shops I have seen have a mix that includes runners, repeaters, and strangers, and the problem that I have seen is the opposite of what you are describing: they treat all items as if they were strangers.

      Until you show them the data, some managers are not even aware that they have runners and repeaters, and that they are losing productivity by not taking advantage of the repetitiveness hidden in the flow of orders.

      It doesn’t mean that you should forcefit strangers into lines meant for other product categories. The strangers deserve their own job shop, operating next to flexible lines for families of repeaters and dedicated lines for runners.

      The perception of a chaotic, unstructured demand pattern is more common than its reality, and sometimes caused by customers’ purchasing practices. A commercial aircraft maker with firm orders for one plane every three days for the next 18 months will use an ERP system that turns this steady rate into alternations of feast and famine for suppliers. A supplier who understands this can use this information in production control.

      Often, you have to dig into the data to make the repetitiveness visible. Make-to-order versus make-to-stock is not the issue. The runner I was alluding to earlier was a single order from a single customer for three years of work. Our client made the mistake of allocating its most modern, most flexible machining center to this work, when they should instead have used a cell of simpler, single-purpose lathes, milling machines, drill presses, grinders, etc., while keeping the machining center available for strangers.

  5. Comment on LinkedIn:

    Can I read up on your method to find the Runners, Repeaters and Strangers method that you use in any of your books? I saw a Heijunka calculation where setup time between batches was assumed to be zero. Really! In a job shop?

    • The basic P-Q analysis explained in Chapter 2 of Lean Assembly applies to machine shops too. Back when I wrote it, I used “A,” “B,” and “C” instead of Runners, Repeaters, and Strangers. See Chapter 15 of Lean Logistics on Production scheduling for Lines with Setups between Products. In the same plant, a runner was a part for a semiconductor production machines with 6,000 units to be made in 3 years at a steady pace, and a stranger was an assembly fixture for a car company. In another plant, the runners were fittings for aircraft engines.

  6. Comment on LinkedIn:

    Based on my experience with HMLV production in many job shops, I would like to explain my perception of demand pattern for those shops. A HMLV job shop keeps receiving a stream of low-volume orders with unpredictable receiving times, volumes and process requirements.

    Some of those orders come with customer-specified due dates. For many orders, the shop does not keep final goods inventory, that is, it makes the required products only after receiving the orders. In terms of the structure of its production system, it may not be easy and economical to transform the system into a set of independent production lines (based on product families) such that similar products pass through each production line one after another and each product family has stable demand. I want to see a strong evidence that Toyota methods are really effective for production control in this kind of job shops.

    In my opinion, make-to-order versus make-to-stock is a very important issue for job shops. There may be hundreds of thousands job shops all over the world with major part of production being of MTO type. They may fall below the radar of Lean consultants. I look for direct experience with complex job shops from Lean consultants.

    • Make-to-order and make-to-stock are different business processes, particularly in Sales and Supply Chain Management. By the time you get to production, it no longer matters. Whether it is to go into a finished goods stock or to be shipped to a customer, you still have to make it. What matters is how much in how long. A runner can be made to order and a stranger to stock.

      It’s easier to make money from a mix where runners and repeaters account for most of the volume than if it is all strangers, and I think it is the reason this pattern is so common. Being a supplier to a fickle industry is not a recipe for long-term survival, particularly when you consider that the ups and downs are amplified as you go up the supply chain.

      The semiconductor industry has cycles; the semiconductor equipment industry, more violent cycles; machine shops that supply components to the semiconductor equipment industry, even more violent ups and downs. Ditto for oil and gas.

      To survive, you try to supplement the business by supplying industries that may pay less but more reliably, like defense, aeronautics, automotive, construction machinery, etc.

Leave a Reply

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