Takt Time

One of the concepts used by companies that are moving toward leaner operations is takt time.  It is also a concept that is often misunderstood in companies that have had a little exposure to lean processes and have picked up the term.

Takt time is the beat at which we should make a product or component to be in synch with the needs of the customer. The customer may be an internal customer such as a downstream process, or it may be an external customer, a dealer or an end user. Takt is a German word for beat when the subject is music.  The baton used by a band or orchestra leader is called a taktstock.

Let us suppose that a customer of any type needs 900 units, widgets of some kind.  The factory works an 8-hour shift, but a half hour a day goes into breaks and other planned downtime, leaving 7.5 available hours.  Seven and a half hours times 60 minutes, times 60 seconds, equals 27,000 seconds of available time in a shift.  If we divide 27,000 seconds into 900 units, we find that our takt time is 30 seconds per unit.  To be perfectly synchronized with the customers requirements for a day, we need to produce one widget every 30 seconds and each operation in the process should take exactly 30 seconds.

This number is useful in a number of ways. It tells us exactly what we need to produce every 30 seconds to meet the customer’s requirement. We can now compare this with the actual cycle times of the workers who produce these widgets.  Let us suppose that there are six operators involved and the cycle times of each are as shown in the following bar chart.

 

This chart immediately shows us we have a problem.  Look at operator D.  We must produce a unit every 30 seconds but this operator is taking 34 seconds to do the required work for that operation. We are not going to successfully produce the 900 units that the customer requires. Meanwhile, the remaining five operators have some extra time on their hands.  Typically, they will over-produce, provided they have the material and the equipment to do the job.

The more common “solution” to this problem is to have everyone produce in batches and have a build up of material between one operation and the next, while everyone does they best they can to stay caught up. Operator D will probably have to put in some overtime now and then.  But suppose all of the operations were so close together that the work among operations can be redistributed.  Some of the work presently being done by operator D can be given to operator C, for example.  Not only that, we can figure out how to redistribute all of the work so that it now looks something like this.

Now we find that we only require five workers, four of them are working at or very close to takt time. We now have another problem. Operator E has a lot of time on her hands. She will use this time for mischief. It does not need to be (and in most cases will not be) malicious mischief. It will likely be with good intentions.  She may, for example stack the widgets, higher and higher, until the pile falls over and some widgets break and must be scrapped.  I have seen incidents of this type more than once.

The first chart data can be used mathematically to set improvement goals.  The sum of all the operator cycle times ads up to 150 seconds.  If we divide this number into the takt time, we see that the number of people needed to get this work done is five, the five shown in the second chart.

We can repeat the process on the second chart.  Now the sum of the cycle times is 124 seconds.  Divide that by 30 seconds and the number of people needed is 4.13.  However, the best thing to do in this case is to round down to four people.  There is no such thing as .13 of a person.  It is best to round down even from a number in which the digit to the right of the decimal is more than 5, such as 4.7 people.  To promote continuous improvement we always round down.

Practically, what does this mean?  Every operation has some waste in it. Perhaps the operator has to take time to reach for material, or it takes a long time to put the material in a fixture. Perhaps the material tends to stick to the fixture requiring extra work to release it. The solution can be in improving the method, the equipment, the location of the material or the material itself, or the lacations of the machines with respect to each other.

I worked with a colleague years ago, in which each of us was coaching different workshop exercises of the shop floor.  His involved making sub assemblies for automotive radiators.  I don’t have his exact data after all these years, but the five-day process started out with data similar to that in the first chart. The project team divided 150 seconds into 30 seconds, yielding five workers in the cell.  They went to work figuring out how to redistribute the work so that it could be done in a balanced way in five cells and wound up with a situation similar to the one in the second chart. Four workers were close to (but not over) takt time. One was way below.

They divided the new sum of the cycle times into the takt time. The consultant told them to figure out how to get the number of operators down to no more than four.  They worked at it, redid the calculation of sum of the cycle times divided by takt time, getting a result of something like 3.8 people needed for the work cell.  Once again they were challenged to round down to 3 people and figure out how to make it work.  In their next calculation they came up with soothing like 2.7 people.  They rounded down to 2 and went to work.  The managed to get it to 1.8 people needed.  At each step they worked with the operators to identify and eliminate waste.

That was as far as they got during our workshop, but a member of the improvement team asked, “If we get the number below one person, what do we do? Do we just do away with the operation?”  The answer was: if you get to the point where you do not need a full person to get the work done, you incorporate the operation into the line or cell to which you were previously delivering this product. There is no longer a need for a separate cell.

We will continue with this topic in the next post.