Home >    Community    > Forums
Topic Title: Cycle Time Calculation
Topic Summary:
Created On: 03/02/2011 09:03 AM
Linear : Threading
Send to a Friend Send to a Friend
Search Topic Search Topic
Topic Tools Topic Tools
View similar topics View similar topics
View topic in raw text format. Print this topic.
03/02/2011 10:53 AM
Print this message

Author Icon
pgph
Philip Peyrer - Heimstätt



I have a question regarding the calculation of cycle times in the following situation which we teach in our Lean course:

A process consists of two steps A & B. Each of these steps takes roughly 1 second, so the cycle time for that process should be 2 seconds.

However, in this example the takt time is 1,5 seconds, so I need to reduce the C/T. Since I cannot reduce it by eliminating NVA's or wastes, I will try to split this process in two by adding another resource.

The best way to do this would be to split the process into two separate steps: Step A with 1 sec C/T and Step B with 1 sec C/T. Now each of the processes is below T/T and the problem is solved.

However, another solution could be to duplicate the whole process, where the second resource also carries out both steps A & B. The number of resources is the same as in the other solution, but I know that this solution is sub-optimal because instead of having a smooth flow of materials (1 piece every second), I have fluctuations in output (2 pieces every 2 seconds). So when I explain this to my colleagues, I tell them that the first solution is the correct one.

However, where I am unsure is when it comes to calculating the C/T for the second solution. Is the correct C/T 2 seconds (because 1 piece still needs 2 seconds to leave the process), or is it 1 second (because every 2 seconds you get 2 pieces)?? If the answer is 1 second, then the next question would be "if the C/T is below T/T, what is wrong with duplicating the whole process?". My argument would be that it is better to level the output, so we get 1 piece every 1 sec instead of getting 2 pieces every 2 secs (which results in waiting times and therefore waste in the downstream processes).

What is the correct C/T, and what is the best explanation for the correct solution? Any inputs would be greatly appreciated!

Philip
03/02/2011 04:28 PM
Print this message

Author Icon
MarkRosenthal
Mark Rosenthal



To answer your question directly, the operator cycle time is the time it would take for one person to complete the operations required.

Your cycle time has not changed.

However, your question raises some other questions. When you are dealing with cycles and takt times shorter than about 10 seconds, you are getting into the domain where the work is not that suitable for people to do. Some would say even 10 seconds is too short.

To make the discussion easier to follow, I am going to convert your SECONDS into MINUTES.

Thus, your takt time is 90 seconds (1.5 minutes) and your total cycle time is 120 seconds (2 minutes).

The calculation to determine the number of people required is to take the total cycle time and divide by the takt time.

Thus:

120 / 90 = 1.5
As you pointed out, you have to round to two people.

Now, it is a natural solution to divide the work evenly between these two people. And sometimes, due to the nature of the work, you HAVE to do that.

But the preferable solution is to load as many people as possible to your operating takt time (or target cycle time, as Mike Rother calls it in Toyota Kata).

That would mean:
Person A: 90 seconds of work.
Person B: 30 seconds of work.

By consolidating all of the wait time onto one person you accomplish a couple of things.
Now the waste of waiting is more visible because it is consolidated onto a single person. What can you do to eliminate this waste?

You can improve flow by pulling another operation in to this one, and improve the work balance.

You can set a kaizen target to reduce the cycle time to the point where one person can do it.

You can move 30 seconds of work OUT of this flow over to someone who ALSO has 30 seconds of waiting, and take 3 people to 2 people total.

How you approach it depends on the context of the overall value stream.

Now, when you are dealing with fractions of a second, I would be looking for opportunities to combine other operations into this one. Can I get three people to two people?
03/03/2011 02:19 PM
Print this message

Author Icon
Sagala
Guntar Sagala



Dear all,
C/T = Total time to produce 1 pair (value add + non value add),
I want to share with yours the succesful of C/T(cycle Time), how & when we using C/T.
Many manufacture calculate C/T on production (when mass production) and some of manufacture using c/t for evaluation manpower, machinery,ect.
Power of C/T is :
1. Using C/T for calculate req of manpower,machinery ; it's mean C/t checking and count on development stage and some manufacture using bank data and revise when learning curve zone.
2. Reduce lead time ; Setting production target (T/T) , using c/t to improve your production output.

Can we reduce c/t (ofcourse YES), we can reduce by re-calculate motion on each activities and remove/avoid non value adding activities (doing kaizen).
To optimize and make sure productivity increase, we can analyze and standarize your c/t on learning curve zone and normal zone (should be difrent) the challenge is how to reduce lead time on learning curve zone. Example : shoe manufacture on stitching process learning curve 3-6 days; our challenge how to reduce until 1-2 days.
My sugest don't make operator think about Target /production result but make sure their only think achieve c/t with good quality product.
Cheers,

Sagala Gun
03/03/2011 02:19 PM
Print this message

Author Icon
leanwannabe
Brian Koenig



A couple items on top of Mark's nice overview on the advantages / disadvantages of putting in a second set of equipment.

Advantages-

A. Second set would allow the flow to keep going if one of the operations in the first set goes down. How reliable are the machines to maintain 1 second takt, day after day?

B. Depending on the operations and change-over, a second set could allow you to produce 2 different types of parts simultaneously while maintaining the same flow rate. (takt)

Disadvantages -

A. More space and possible higher operating cost.

B. Two lines would add another element of variation to the processing, which might impact furture processing or product performance.

A key Lean concept would be to keep the machines paired in the A - B pairs. Don't departmentalize and put the A's together and B's together.

Good Luck - Brian
03/03/2011 02:19 PM
Print this message

Author Icon
Boeing_Lean
Ken Hunt



Several years ago, I was on a team at Yamataki in Japan that was able to do just as Mark described. As I recall the takt time was something less than 10 seconds, and after moving some of the flow around, the number of operators went from 3 to 2.

One thing that did help is that we also took steps to reduce the distribution flow as well as the assembly flow.
03/03/2011 04:13 PM
Print this message

Author Icon
pgph
Philip Peyrer - Heimstätt



Mark,

Thank you very much for your detailed reply. My example was indeed taken from our training class where we simulate our production process on a small, simplified scale, so your point about seconds vs. minutes is of course correct. Also, we use participants of the class to act as machines, so again, in real life we would be talking about machines.

So, from what I understood from your first sentence is that the C/T does not change (i.e. continues to be 2 minutes) because each operator still takes 2 minutes to perform both tasks A & B, no matter how many people are working in parallel, correct? In that case I have another complication:

In the first run of our simulation, we have one operator performing both tasks with a C/T of 2 minutes. We calculate T/T and find that the C/T is above the T/T of 1.5 minutes. So in the second run, the participants decide to duplicate the process and have two operators perform both tasks (which we know to be a sub-optimal solution, because each person still takes roughly 2 minutes). All this I already explained in my previous post. But now when we measure the two person's individual C/T's, we of course find that they are not exactly the same - one takes, say, 1.8 minutes (because he is better at it) and the other takes 2.2 minutes.

So, given this situation, and reading your reply, am I correct to assume that the C/T for this process is still 2 minutes because that's the average of the two? There will always be variation even if the tasks were performed by two identical machines instead of humans - so is the correct solution to average them?

As for your other comments regarding loading one person to the T/T and thereby exposing the waiting waste in the other is interesting and I had not thought of that in this context.

Thanks again!
Philip
03/04/2011 01:00 PM
Print this message

Author Icon
MarkRosenthal
Mark Rosenthal



Assuming you are doing a true time observation study of multiple cycles, the cycle time you want to use is the lowest consistently repeatable time. This should give you a problem-free baseline.

Of course two people performing the same task are going to vary a bit. (This is one reason you don't use a "rabbit chase" in a multi-person work cell).

But it also offers an opportunity to understand why.
In your (admittedly hypothetical) example, you are looking at one person taking 108 seconds, and person #2 requiring 132 seconds. This is a HUGE variation, 20% of your nominal baseline.

Time to get back out to the shop floor and study the process some more.

The other thing to keep in mind is that in your standard work, the nominal cycle time is a TARGET CONDITION. It is what you are striving to achieve. You want to know when and why you don't achieve it. (This is another reason to load as many workers as possible right up to your operational takt (or target cycle time).... so you can see problems.

And that, I think, is a key point of the discussion. This is not so much about shaving hairs to calculate the cycle time exactly. It is about setting a baseline for what "should be happening" and structuring the work so that any difference between "should be" and "actually is" becomes immediately apparent so that you can respond, grasp the situation, seek understanding about why, and deal with something that you previously did not know.
03/04/2011 01:00 PM
Print this message

Author Icon
vincentloh
Vincent Loh



very good and detailed explanation! thks.
03/04/2011 01:00 PM
Print this message

Author Icon
3744
Ronald Turkett



If there is waste of motion cycle time can be reduced. For many machine cycles reduction cannot take place such as feeds and speeds fixed by material and tools.

Cycle time cannot be treated by itself to determine output unless the FTT and Operational Availability are 100%. Use Effective cyle time estimates by dividing the cycle time by the product of FTT and OA. This will adjust the cycle time by the loss due to quality rejects and downtime. Then compare ECT with Takt time.

Ron Turkett
03/07/2011 12:35 PM
Print this message

Author Icon
pgph
Philip Peyrer - Heimstätt



Thanks to all of you for your valuable inputs!

KR
Philip
03/14/2011 09:19 AM
Print this message

Author Icon
renganathan
Renganathan Iyengar



One of the Very useful and really Lean way of calculating time is to use the VA / NVA / RNVA classifications. Let me give you a step by step way for doing this:

1. Split the whole operation in to smaller activities say about 20 or 25 maximum. In case of many mfg jobs these activities would be typically like - load the job, clean the job, press the button, walk to the next machine etc.

2. Now split each activity into one of the following groups: a) Repetitive activities, b) Non-repetitive activities, c) Occasional activities and c) Random activities.

3. Now you also need to classify the activities as VA / NVA / RNVA.

4. Go to the operation spot and do a video of the operations -do for atleast ten cycles. If you do more than that, you will learn more about the operation.

5. Come back to your desk and play the video - go & measure the time for each of the operation category you have done in steps 1 to 3. You will ignore very low timings and very large timings - but do not delete such videos - they will contain vital clues about the Muda that is happening.

6. Take the average of each of the actitivity and sum them up - you will have the correct cycle time.

Now, First try to eliminate the NVA activities from the cycle. Then eliminate or reduce the RNVA - you will benefit more. Your 'real' benefical time would be from the VA elements only.

You will also find that usually it is very difficult to define what is VA / RNVA / NVA. No body fully agrees to any definitions !

Please note - what I had given is a mixture of old time study practice plus the current Lean method.

May be it would take a bit of time for you to understand this - but once you do, you will this kind of classifications help you to very systematically reduce the cycle time for any operations.

Rgds

Renga
03/13/2014 06:53 AM
Print this message

Author Icon
Giovanna
Giovanna Diaz



Hi,
Could you explain me one thing:. If our goal is to configure the production line to work at the pace of takt time, that is, achieve a CT that almost equals TT, then,how, after adding another resource, the CT remains invariable?
thanks
03/14/2014 08:24 AM
Print this message

Author Icon
285446
jeroen van deursen



If you duplicate the whole process, then each process only has to cope with 50% of the demand. So your TT doubles!

CT still is 2 seconds (for both), TT is 3 seconds. So each is below TT.

If you still realy want a part every 1,5 seconds (and not 2 parts every 3 seconds) then you have to do something with timing..

Jeroen
03/17/2014 07:04 AM
Print this message

Author Icon
Oldon
Donald Anderson



HI Giovanna:

The Cycle time does not stay invariable. Philip's theoretical example is intriguing for a number of reasons, which makes it a good training tool, not just for cycle time, but for "reality versus theoretical" implementation examples for trainees.

For example:
He stated you can't reduce NVA or Waste - but later have an example, as pointed out with a difference between two operators, which indicates waste does exist in the process. Reality vs. theory - waste always exists - the question is whether it is sufficient to bother removing through investment. Theory would say yes, reality maybe not; but meeting customer demand is necessary either way.

In adding another resource (another person) the assumption is that you double your throughput, a word not stated in anyone else's answer, because as lean practitioners we are focused on takt time versus cycle time (TT vs. CT) for process leveling. The implementation problem with that focus is that alternatives are not pursued that make business sense. Forced shutdown of equipment for maintenance (TPM), as just one example; application of people to other work cells, as another. You can't level processes perfectly in reality - there are always "trade offs" and that is an implementation lesson to learn.

The Toyota Production System focuses on throughput and removal of the partial worker from any defined work cell through improvement. People are the most flexible resource and a "half person" can be added to match CT to TT, which is the correct solution to the given example, not the "either / or" solutions presently discussed for the training. Mark and others point out this reality in their answers.

Adding people adds complexity as was noted in the timing differences and the actual NVA may go up, if they transfer product between each other for example, (or it could go down if it makes it more effective in the work cell layout) - the key is in the standard work design, which should vary for the process dependent on the number of workers in the cell. I actually developed an excel solver program for determining the optimal number of people in a cell, although having the people in the work cell themselves define it is the better solution as part of their training.

My final recommendation is to have all training take place on an actual process so that all of these reality issues can come out - even if it is a simulated process (or a filmed one as noted earlier) in the training room; otherwise they will come out when the training is applied to a real process and will take longer to resolve and even cause frustration in those doing the implementation. I hope this helps everyone in their focus on training and good luck!
Note: These forums are moderated by the Lean Enterprise Institute. All posts are reviewed prior to appearing on the site. Views expressed in these forums do not necessarily represent the views of the Lean Enterprise Institute.