GUIDE · LINE DESIGN

What Is Mixed Model Line Design?

Mixed Model Line Design is a step-by-step engineering method for designing one production line that builds many different products, in any mix, one at a time, at a calculated pace. We call it Lean Industrial Engineering. It is the skill most Lean professionals were never taught, and it is where most of the benefits of Lean actually come from.

By Richard Rahn and Gerard Leone · 12 minute read

The short answer

Line Design is the discipline of arranging people, machines, work and material into a flow, so that each product moves from station to station without stopping, and every station has the right amount of work. It is not a common-sense rearrangement of the furniture. It is a method with inputs, calculations and checks, and it has been used thousands of times in every kind of industry.

The word mixed model tells you the line is not built for one product. It builds a family of products that share most of their processes, in whatever mix the customers order on a given day. That is how real factories work, and it is where the method earns its keep.

Why a line? Think of a river

In a flow line, products move through the processes without waiting, like water in a river. The Mississippi starts as a small stream in Minnesota, collects hundreds of tributaries, and never stops moving. Our metaphor for a factory is the same. The tributaries are your sub-processes, and they are all physically connected to the main flow.

When products are built one at a time, with no waiting between steps, the elapsed time through the plant approaches the actual work content time. Batch and queue production cannot do that, because most of a batch's life is spent waiting. Notice where the gain comes from. It does not come from working harder or faster. It comes from connecting the processes and removing the waits between them, and that is a design task, not an exhortation.

A useful yardstick. If the elapsed time through your value stream is within 300 to 400 percent of the work content on the critical path, you are in world-class territory. Most plants we visit are nowhere near that.

Why "mixed model"?

A dedicated line for every product is simple to design, and expensive to own. Each line needs its own people and floor space, and each product's sales cycle drags its line up and down. Building several products on one line has real advantages:

• Smoother demand. On any given day it is unlikely that every product is up. Add them together and the volatility washes out. Forecasts for a family are always better than forecasts for a single item.

• Better use of floor space and people. One well-used line beats three half-empty ones.

• Less exposure to sales cycles. No more hiring and firing operators to chase one product's peaks.

The price is that products do not take exactly the same time. Some have more work content than others. That is the central engineering problem of mixed model design, and the rest of this page is mostly about how we solve it.

Line Design and the rest of the Lean toolkit

Most Lean professionals know 5S, quick changeover, value stream mapping, standard work and kaizen. Those tools are valuable, and they all work better on a well-designed line. But improving a poorly designed line only produces a slightly better poorly designed line. Kaizen events find waste inside the process you have. Line Design decides what the process should be in the first place. That is why we say optimum process design is the source of most Lean benefits.

Line Design also needs the right setting. It works best in repetitive manufacturing, where the same families of products flow through the same processes. In an office, a job shop or a custom-engineered business the idea of takt begins to break down. That does not mean Lean does not work there. It means you dig a little deeper to get the same result.

The method, step by step

These are the steps we follow in our consulting and teach in our workshops. They are the same steps that make up the Lean Roadmap in our book, The Complete Guide to Mixed Model Line Design.

1. Assess and prepare

Know where you are before you start. Get management on board, because a purely bottom-up effort produces results that are easily lost. Train everybody, although not everybody needs the same training. Then pick a target area with logical, quantifiable criteria, and agree the benefits and metrics with the management team. A value stream map can help you get started, but do not make a career out of flow charting. Get out and get improvements going.

2. Understand the flow

List your products and the processes needed to build them. Draw a Process Flow Diagram for each product, showing which processes it needs and in which sequence. Then put them all in a Process Flow Matrix: products in the rows, processes in the columns, an X wherever a product uses a process. The matrix shows at a glance which processes everything shares, which are the main flow of work, and which are used by only a few products, because those are the ones that will give you trouble.

3. Define product families

Not every product belongs on the same line. We look for common processes, common work times and common material. As a rule of thumb we try to keep work content within about 30 percent across a family. Defining families is iterative. You will come back to it as you collect more data, and sometimes you will need more than one line to cope with differences in process time.

4. Calculate demand and takt time

Takt is the beat of the process. We calculate it with a simple formula:

Takt Time = Effective Work Minutes per Day ÷ Demand per Day

Two points trip up almost everyone. First, design for future volume, not for today's. Include growth, new products, product obsolescence and seasonality. Second, takt is calculated process by process, and different processes can have different takt times, because of scrap, rework, options and work schedules. A forecast of 20,160 drills a year, on a plant that works 240 days, is 84 drills a day, and every product on the line adds its own share to the volume of each process it touches. Effective minutes are not the hours on the time clock. A typical 510 minute day, less a lunch, two breaks, a short improvement meeting and start-up and clean-up time, leaves about 420 minutes of real work.

5. Document the work

The Standard Work Definition records, step by step, the work content of every process for every product, and separates value-added from non-value-added steps. We are not building a straitjacket. Standard Work is a baseline you can improve from, and it is the data source for the calculations that follow. You can use estimates to move faster, but the price is more adjustments after the line goes live.

6. Calculate the resources

Resources = Standard Time ÷ Takt Time

If a process needs 3.4 minutes of labor and the takt is 1.5 minutes, you need 2.26 people, so you plan three workstations. If a machine needs 12.7 minutes and the takt is 3.5, you need 3.63 machines. Batch machines deserve extra attention. A kiln with a 12 hour cycle and a 20 minute takt needs capacity for 36 units, and room for 36 more on each side of it. The numbers are simple. Reading them correctly is the skill.

7. Balance the work

Textbooks say "balance to takt". In a mixed model line that instruction breaks down, because the work time varies from product to product. Balance each product to a time target instead: the product's standard time divided by the number of workstations. Some products will run over takt and others under, and on average you meet the rate. If the mix shifts toward the heavy products, you will need more capacity, and the design should tell you how much.

When the first pass leaves stations over or under, we have six tools, and we use them roughly in this order:

• Eliminate waste, starting with the non-value-added steps

• Relocate work between stations

• Add resources

• Add inventory plus time

• Add in-process kanbans

• Sequence the mix

8. Link the line with pull signals

Processes must be connected by a formal signal that triggers work or movement based on need, not on a schedule. We use five tools: direct connect, in-process kanbans, FIFO lanes, kanban direct and kanban supermarkets. Direct connect, with no inventory between stations, looks the purest, but it makes the whole line move like a train and tempts operators to break the rules. We reserve it for large products and for takt times longer than about an hour. For most lines the in-process kanban, a single empty space between two stations, is the best compromise. It absorbs variation while still limiting inventory. The one non-negotiable rule: every material and every task gets a pull signal, without exception.

9. Lay it out and prove it

Build the conceptual layout as a team exercise, with every manufacturing function in the room, and then convert it to a CAD drawing that reflects columns, utilities, ergonomics and safety. Make it interactive, and have some fun with it. Allow criticism only during an "Open Season" meeting. When the season closes, everyone gets behind the design.

Then test it. A mixed model line is hard to judge on paper, because mix, volumes, work content and plain human variability all move at once. Simulation lets you run hundreds of scenarios before you pour a single anchor bolt, and the cost of a model is a fraction of the cost of redoing a physical line.

10. Go live and sustain it

Deploy with a plan, train the people who will run the line, and put audits and certification in place. We have seen too many good projects slip away after a change of management or a wave of turnover. Unless the right tools, culture and accountability are in place, your Lean gains are always at risk.

What you get

• The shortest possible cycle time. Work moves from step to step with almost no queue time.

• The best possible quality. Less clutter and less work-in-process mean problems are caught at the station where they happen.

• The highest productivity. In our experience we have rarely seen less than a 20 percent increase.

• Less inventory and less floor space.

• Flexible output. You raise or lower the rate by adding or removing people, not by rebuilding the line.

Mistakes we see most often

• Balancing every product to the average takt, which leaves the last station overloaded or idle.

• Designing for today's volume instead of the volume you will have.

• Freezing the schedule or building finished goods to hide variation, instead of designing for flexibility.

• Skipping the simulation, then learning how the line behaves the day it goes live.

• Having no plan to sustain the result.

Where to learn it

The fastest way is the hands-on way. We run a three-day Mixed Model Line Design workshop at Toyota Material Handling, with factory tours, and the full method is in The Complete Guide to Mixed Model Line Design. If you want to be recognized for the skill, our online certification program takes you through the course, a real design project and a final exam.

Material arriving at the right place is the other half of the picture. Read our companion guide, What Is Mixed Model Material Flow?

Learn it with us

Join a hands-on workshop at Toyota Material Handling, or book a 30-minute call to talk about your own line.