Taking Changeover Time into Account for Takt Time Calculation

How do you calculate takt time when there are regular changeover times within the production process? There are several ways to answer this question.
1. Take changeover time out of net available time. Let’s say we have a food packaging line that runs for 10 net available hours per shift and produces 1,000 packages. The takt time would be 10 hours x 60 minutes x 60 seconds divided by 1,000 packages, or 36,000 seconds / 1,000 package = 36 seconds per package. Each shift there is one hour of lost time due to changeovers. Unless this downtime is somehow accounted for, the production line will be short by one hour’s worth of packages each shift. So we can adjust down the available time by one hour to 9 hours making the calculation for net available time 9 hrs x 60 min x 60 sec or 32,400 seconds. The demand does not change so the new takt time is 32.4 seconds per package. The process either needs to be designed to be capable of producing one package every 32 seconds or the changeover time needs to be kaizened.
2. Add changeover time per piece to the cycle time. An approach that applies best to discrete manual operations and also for machine tool changes during a run is to add the changeover time to the cycle time per piece rather than to include it in the takt time calculation. The advantage of this approach is that in discrete manufacturing operations there may be some operations within a linked flow that have changeovers and others that do no, so simply subtracting changeover time as in the continuous processing example of 1 above gives a false picture of available time. In the case of a manual process such as a mixed model assembly or packaging line requiring changeovers in materials or tools by the operator, it is advisable to measure these changeover times, divide this number by the pieces run between changeovers, and add this changeover time per piece to the cycle time.
Tool changes are required for cutting tools for mills, drills, lathes and even grinding wheels. After a certain number of cuts these tools wear down and must be exchanged. While these changes are typically fairly quick, they do need to be accounted for and this lost time removed from the available capacity. Emerging online video star Ron Pereira of the Lean Six Sigma Academy explains the use of the table of production capacity by process (also called production capacity table) wherein time for tool changes is accounted for. Basically the time to change is divided by the number of pieces run until the change is needed. In the video example the changeover time is 600 seconds for 50 pieces so 12 seconds per piece is added on to cycle time. This can be for tool changes during a run or for changes from one model to another in cases where there is a set quantity patter production such that changeovers are done every 50 pieces, but it is traditionally used for the former.
3. Ignore the changeover time. In this “hard core” approach one neither subtracts time lost to changeovers from available time nor adds changeover time into the cycle time. By not taking changeover time into account, calculating takt time and running based on 100% operation within the net available time, losses due to changeovers will cause production shortfalls and demand attention immediately. Unless there is production capacity aplenty, this means applying SMED (single minute exchange of dies) techniques to achieving quick and smooth changeovers. In most cases companies that start with approach 3 will settle down to approach 1 at some point after making significant gains towards single minute changeovers and reaching a plateau for changeover reduction, at which point it becomes practical to take this lost time into account.
This approach is also advisable when there are irregular changeovers or unpredictable changeover times, making it more trouble than it is worth to determine accurate times to factor in. Of course this variability is in itself a problem and should be fixed, after which approach 1 or 2 above may be advisable.
Note that it is best not to take a similar approach to 1 or 2 above when considering OEE (Overall Equipment Effectiveness) and its effect on capacit or takt time. The calculation for OEE takes into account multiple losses such as changeover, minor stops, speed losses, yield, scrap and equipment breakdowns, many of which are variable and unpredictable. It is best to use method 3 to drive continuous improvement of OEE rather than factoring in OEE rate into takt time calculation.
However changeover time is taken into account for takt time calculation the method should be as simple as possible and linked to continuous improvement. The most important thing with any of these approaches is to not willfully increase the batch size or add resources in order to minimize the time lost to changeovers before doing everything possible to minimize changover time.

2 Comments

  1. Erik

    October 6, 2008 - 6:11 pm

    Jon,
    My apologies for the confusion but can you please explain #2 more clearly? Specifically, what is done with the cycle time number once we adjust it to include changeover time? Isn’t the rate of demand unaffected by cycle time? Wasn’t this essentially what Mike Gardner was talking about in his comment to the previous post? That we need to be careful to keep cycle time and takt separate and not confuse the two?
    I most likely am missing something here but I’ve gone over it several times looking for the connection and can’t seem to find it. My confusion comes from the relationship between cycle time and takt time. Expressed mathematically takt time (A) = Net Working Time (B) over Demand (C). However if we factor changeover time into Cycle Time this introduces another variable (D) that doesn’t appear in the above equation A = B/C. Therefore, I’m confused as to how adjusting D has an effect on A.
    I hope I have expressed my question clearly enough. Thank you as always for your insights.
    Erik
    PS – Since we’re on the subject of takt time again, what is the status of the Illustrated TPS book that was to be sent to the participants of the last takt time discussion?

  2. Jon Miller

    October 6, 2008 - 6:25 pm

    Hi Erik,
    The takt time calculation does not change for #2. Rather than taking changeover time out of available time, the time per piece to do the changeover is added to the cycle time. Essentially the same effect as #1, but more appropriate for discrete manufacturing. Sorry for the confusion.
    The TPS book is in traveling cyberspace between the printer and us. As soon as it stops traveling, we will be in a 30 day printing queue.