Home > The Lean Post> Heijunka: Mastering the Peaks and Valleys
The Lean Post
Sharing how the world is making things better through lean.

Heijunka: Mastering the Peaks and Valleys

by Jeff Smith
May 1, 2015

Heijunka: Mastering the Peaks and Valleys

by Jeff Smith
May 1, 2015 | Comments (3)

As a lean coach in manufacturing, I hear a lot of people express an inability to run different products down one line even though they would like to do so for cost savings purposes (i.e. people, equipment, machines, space, and so on). Most people have never heard of the lean principle of Heijunka, which is about using resources more effectively by sequencing production by sales ratio and then distributing the workload to people in order to achieve maximum cost-saving benefits as described above.

If you find yourself struggling with how to do it, here’s a scenario I hope you can use to get started. A hypothetical conversation between myself and fellow lean coach Danielle McGuinness:

Jeff: “Hey Danielle, let’s tackle the challenging subject of Heijunka for businesses with assembly operations.”

Danielle: “OK, what do you want to tackle about it?”

Jeff: “Let’s give an example of a multi-product line. We can show a sales breakdown of different types of models and their different build times. Want to show how to sequence those items so we can use the correct number of people and not create any bottlenecks (jobs over takt time)?

Danielle: “Right on. Let’s use stoves as a product. Three main types: 3 foot, 4 foot, and 5 foot (top of stove). Furthermore, we have a daily (from monthly) level demand of 45 units. And the customers order 50% 3 foot, 25% 4 foot, and 25% 5 foot.”

Jeff: “So the difference in the size refers to a larger stove with more goodies on it… like a griddle, more burners to cook on, possibly an additional oven, and so on? I take it these stoves take different times to build? And that we need to figure out if they can be made on the same line?”

Danielle: “Yes. So let’s say the 3 foot is 180 minutes of work, the 4 foot is 220 minutes, and the large 5 foot is 240 minutes total. Since this company works 450 minutes (7.5 hours/shift), the assembly line has a takt time of 10 minutes.”

Jeff: “Got it. So if we sum the weighted total cycle time for all models we can divide by takt time to find the allocated workload for a station/job, then go about figuring out a physical layout for how many people work on a unit per station and so on.”

Danielle: “Right. And we said the customer demand for a 3 foot stove is 50% and the total cycle time for this model is 180 minutes to build... The weighted cycle time for this 3 foot stove is then 90 minutes in our calculation… If we do the same for a 4 foot stove (220 minutes * 25%) we get 55 minutes total, and a 5 foot stove gives us (230 minutes * 25%) 57.5 minutes.”

Jeff: “So that means we can total those weighted sums of 90+55+57.5 minutes and divide by our takt time of 10 minutes. That means 202.5 minutes of weighted total cycle time / 10 minute takt time gives us 20.25 people to allocate work to for in a weighted average scenario, right? Probably means making it work with 20 people total?” 

Danielle: “Yes. We would accept that small challenge and allocate workload to 20 people. That could mean 2 people per station for a ten station line, or it could mean a 20 station line of 1 person per station. Also, it could be somewhere in-between.  Also, we can figure out how much work per person for each model, because we have the number of people already decided at 20.”

Jeff: “I think I’m following you… So a 180 minute 3 foot stove divided by 20 people is basically 20 jobs with a 9 minute workload per job. Same for a 4 foot stove (220 minutes/20 people) yielding 11 minutes a job, and a 5 foot stove would be (230 minutes/20 people) with 11.5 minutes job?”

Danielle: “Yes, and we can see how important Sequential Scheduling discipline would be to keep up with a ten minute takt time on a weighted average continuous flow assembly line making stoves one after another.”

Jeff: “I guess it means that if every other model is a 3 foot stove we gain a minute on the takt time to start working on the next product which is a 4 or five foot stove taking a minute to a minute and a half more to wash out the time over takt?”

Danielle: “To make it really wash out we could take that extra half minute on the longest model off line, or even eliminate some that amount of reach and walk waste out of the cycle time for a 5 foot to bring it down to a 4 foot time." 

Jeff: “This seems like a good plan to use the correct number of resources right out of the gate to build different cycle time stoves on one line. We could of course just sequentially supply big parts to the side of the line behind the operator to support and re-enforce their standardized work, as well as putting kits of small parts and fasteners down the line with the stoves to eliminate much walking (keeping people focused upon building quality in station).”

Danielle: “This would be a good starting baseline of Standardized Work in a weighted average or multi-product line environment. The managers and line associates would learn to master the peaks and valleys of work difference with the support of a good production scheduling sequence.”

Jeff: “Ok, sounds good. Let’s talk about Heijunka more in the future.”

Danielle: “Next time we can add some other challenges to the situation to learn from.”

The views expressed in this post do not necessarily represent the views or policies of The Lean Enterprise Institute.
Search Posts:
Breaking Through to Flow
By Ian Glenday
Creating Continuous Flow
By Mike Rother and Rick Harris
Was this post... Click all that apply
20 people say YES
20 people say YES
4 people say YES
7 people say YES
Related Posts
3 Comments | Post a Comment
Sabrina May 01, 2015

I'm not sure how this would change the scenario but Danielle originally said "the large 5 foot is 240 minutes total" so by my calculation (which could be incorrect), it would take 20.5 people. I can see at 20.25 people you would try to staff to 20.

What kinds of things would you consider if it was 20.5 people and you needed to make a decision for 20 versus 21 people keeping in mind you want to have the correct number of resources "right out of the gate" and you want to pay attention to sequential scheduling?

Reply »

Jeff smith May 01, 2015


Your correct on the staffing.

Good question about using 20 vs 21 team members.

In the article we were suggesting that 20 people would be used vs using 21. This means we would be accepting some overtime- until such time that the waste of walk and reach could be removed form the jobs, or if we moved work offline to a subassembly, or another tactical solution. 

The word "challenge" was used and may be more important than at first realized when reading the article.  Accepting the challenge to go with one less person than initially required creates the need to chase a target for the team. This would need to be a team decision based upon business needs and learning goals most likely.

Sequential supply of parts can help reduce the footprint to retrieve said parts, and may be in sinc with the steps of the team members Standardized Work (Working Sequence) if the parts are presented in that order. So "selecting parts" or choosing them can be minimized in terms of time because of presentation.

The challenge was to present the team with a next step after the line stabilizes at the 20 people with overtime required initially. The challenge is to get rid of the overtime, but first the line would need to stabilize. Yet maybe both stabilization and reduction or elimination of overtime could happen together.

I hope I have answered your question adequately.

Reply »

Rick Bohan May 06, 2015
1 Person AGREES with this comment

I was thinking, "Yes!  Finally an article that will really explain heijunka and what the heck it's good for!"  Instead, I get an imaginary discussion between two numbers geeks.  

As with pretty much every overview of heijunka I've read, this one assumes:

1.) Steady, stable demand from the customer, and

2.) No changeover time, and

3.) No other process variability.  

Which matches the conditions of pretty much none of my own clients or any company I've ever heard of.  (I will say that #1 is true of auto suppliers, and #2 is true of a consumer product maker I served, but all three at once?  Nope.)

My point isn't meant as a refutation of heijunka. It's just that I'm still looking for an explanation that will make sense to my clients.

Reply »

Search Posts:
Breaking Through to Flow
By Ian Glenday
Creating Continuous Flow
By Mike Rother and Rick Harris
Levelled Production: Why Do It?
"What Did I Transform Today?"
“You Gotta Kata,” Now What?
Please include links as plain text URLs only. Do not copy and paste directly from a web page or other document. Doing so may pick up additional HTML that will not function here.
URLs will be converted to functioning links when your comment is displayed on the site.
Here's an example:
See this article for more details: https://www.lean.org/whatslean