Calculating Optimal Crew Size in Mixed Model Cells

By Ron Pereira Updated on January 13th, 2011

Calculating Crew SizeI have a question for you. How would you staff a cell that produces multiple products, all with varying cycle times?

In this article I will offer my thoughts and am excited to hear your comments as to how you would (or do) approach it in your organization.

And while I will be taking a manufacturing stance in this example, this topic definitely applies to other areas as well (e.g. the office).

Cycle Time Assumptions

Let’s assume, hypothetically speaking, we have developed a u-shaped cell were material flows counter-clockwise.

Let’s also assume, for the sake of this example, there are 5 similarly sized products produced in the cell.

The total manual cycle times (e.g. the time operators are needed) to produce each product are as follows.

– Product A: 5.1 minutes
– Product B: 6.8 minutes
– Product C: 4.9 minutes
– Product D: 7.9 minutes
– Product E: 3.9 minutes

We’ll also assume there is no auto or machine time required to assemble these units. Therefore, the average cycle time for the 5 products is 5.72 minutes.

Demand and Takt Time Assumptions

Now then, let’s also assume that on this particular day our “make to order” model cell is asked to produce the following.

– Product A: 23 units
– Product B: 26 units
– Product C: 3 units
– Product D: 94 units
– Product E: 14 units

In total, the cell must produce 160 similarly sized units. Assuming 8 net working hours (480 minutes) are available in the day our takt time comes to 3 minutes per unit (480 minutes / 160 units).

For the sake of this example, we will also assume no changeovers are required as we switch between products.

Calculating the Crew Size

Let’s get to the million dollar question. How should the team determine the optimal crew size for this particular day?

We know that the formula for optimal crew size is the sum of manual cycle time / takt time.

However, since we are not producing one product (like most of the lean books present us with) it’s not quite as straight forward.

If they use the average cycle time for the units our optimal crew size comes to 1.91 operators, which they round to 2 (5.72 minutes / 3 minute takt). Using this approach the team would staff the cell with 2 people, splitting the work up evenly.

Would this work out? I vote no.

Weighted Average Cycle Time

Since the team is not working with an even number of units averaging is the wrong thing to do, in my opinion. Asking these two people to succeed (without overtime) is the perfect example of muri.

Instead of taking the average, the team would be better served by taking a “weighted average” of the cycle times. This is easy to do in Excel using the “SUMPRODUCT” command. Here is how this particular example is being worked out in MS Excel.

When we take the weighted average, which accounts for the fact the team is asked to produce far more product D’s than anything else, we learn the weighted average cycle time is 6.91 minutes.

Weighted Average Crew Size

If we use the weighted average cycle time in the formula our optimal crew size comes to 2.3 people (6.91 minutes / 3 minutes takt).

This is to say that more than 2 people are needed in the cell on this particular day. Or, at a minimum, more than 2 people will be needed for parts of the day.

So, management should not expect 2 people to get the job done in the available time until some waste is eliminated and we are able to reduce the total cycle times.

What do you think?

What do you think? Do you (or would you) handle similar situations the same way? Or do you (or would you) approach it differently?

Subscribe to LSS Academy

If you enjoyed this article please consider subscribing to our full feed RSS. You can also subscribe by email and have new articles sent directly to your inbox.

  1. Tim Porter

    May 14, 2008 - 6:19 am

    In our business we have a lot more than 5 part numbers in each cell but I do think this method will help us as we also struggle with this manning issue. We have already met this morning here in the UK to form a pilot to test this approach. Thanks for the excellent advice.

  2. Ron Pereira

    May 14, 2008 - 8:00 am

    I’m glad it helped. Please do keep us informed as to how it works out for you. All the best, Tim.

  3. Mike Z

    May 14, 2008 - 10:10 am

    I would argue that you need to begin narrowing your scope by evaluating
    staffing on the hour level rather than the daily level. As you alluded in your
    article, you may require 3 people for a portion of the day. We’ve noticed a
    significant savings in the “flex time” employees would typically

    In addition, we have established “Expected Product Takt
    (EPT)” times for each product. This can be done in your example by dividing
    a product’s working minutes by the line’s takt time. SUM{Working Minutes *
    Product Units/day)/SUM(Product Takt * Product Units/day)

  4. Ron Pereira

    May 14, 2008 - 5:32 pm

    Good stuff, Mike. Thanks for the sharing your ideas.

  5. Paul Cary

    May 15, 2008 - 6:23 am

    Think of a short order cook that is cooking several different things at once. If the cell worker’s are familiar with the process they should be able to switch from one product to another without any lost time as long as the cell is set up to accomodate a mixed model demand.

  6. Mike Z

    May 15, 2008 - 8:25 am

    While I like Paul’s metaphor, I think it only applies to situations where the line is very flexible and the art of sequencing has been mastered. That is a bit of an ideal for most manufacturing environments.

    For example, we have 6 lines dedicated to product families. The crew of 12 staffs one or two lines at a time. Switching between the lines would cause a disaster. We allow a fair amount of sequencing on the individual lines to take place with the limitation being kit types.

    In Ron’s example it would likely work since the parts are assumed to be similar.

  7. Capt. Kaizen

    May 15, 2008 - 12:03 pm

    Hi Ron and others… Here is my thinking, as one would review the scheduling in advance for the next day or two, if you take the number of units times the cycle times of each unit needed, that will give you the total number of minutes needed in a given day. The art of manpower utilization is knowing what the requirements are going to be and adjusting your manpower accordingly. Thats assuming you have a fairly balanced process accross the entire chain and a highly cross trained workforce. With that increase there, most of the time that means that somewhere else there might be less people needed so that line might be able to borrow 1 person…just my thoughts

  8. Tajjammul Hussain

    May 15, 2008 - 1:02 pm


    I agree with the wieghted average method for determining the crew size , however there two more parameters that we need to consider..

    Expected Utilization of the Crew – While we know that a resource is expected to work for 8 hrs ( assuming that the time for lunch and coffee has been excluded), In reality the resource is productive for less than 100% of 8 hr shift.

    Depending on the organization, the expected productivity varies between 80% to 95% (The actual productivity may be a lot less )

    Which if applied, the total available minutes for the day becomes 432 minutes(7.2hrs) and the takt time becomes 2.7 minutes

    And hence the total crew size becomes 2.56 rounded to 3

    Now, the next parameter is the absenteesim-

    Coming from high volumes services background, it is common to have an absenteeism of around 10% and adding this buffer (3*1.1=3.3) makes it 3.3


  9. Sunil Yadav

    October 15, 2008 - 7:43 am

    What do think ,should optimum crew size be for a process with manual time 560 min & takt time 130 min with three models of different 420.540,720 min. respectively

Have something to say?

Leave your comment and let's talk!

Start your Lean & Six Sigma training today.