Line Design is the step-by-step process for designing production line. At Leonardo Group Americas, we’re primarily interested in what’s called Mixed Model line design. That’s where we are building a family of products on the same line. There are many advantages to being able to do this, including:
- Shorter manufacturing cycle time
- Faster customer response
- Improved quality
- Reduced floor space
- One Line capable of building many products
- Increased flexibility
- Reduced waste (less walking)
- Lower inventory requirements
- Reduced working capital
There is a specific process involved in creating an accurate Mixed Model line design. Unfortunately, in our experience, most Lean practitioners don’t really know it. That’s largely because line design is far more technical than general Lean thinking or culture. You might find it helpful to think of Mixed Model Line Design not so much as a Lean philosophy, but rather as Lean Industrial Engineering.
(Mixed Model Line Design is one of our core courses. If you’re interested in diving deeper into the subject, you take the online course at the Lean Design Studio or join us for a 3-day live workshop hosted at Toyota Material Handling)
Let’s jump right into the most common mistakes when designing a production line.
MISTAKE #10: DESIGNING FROM THE CONFERENCE ROOM
When it comes to designing a production line, data is king. You will not get the most accurate data by spending too much time in the conference room. You have to go to the factory floor!
Beyond the numbers, there may be potential improvements that have not been documented. The truth is that your operators know the work better than anyone. When you’re designing, it’s important to have conversations with those operators. Ask respectful questions. Sometimes you have to dig a little deeper to discover what is really happening on the factory floor. When you’re putting together your design team, make sure to include operators on that team. The quality of your design will improve enormously.
Tip: Create Value Stream Maps and Process Flow Diagrams on the factory floor, NOT by memory.
If your line is complex, start at the end (shipping). We like to say, “Don’t put the donkey in front of the horses.” Plus, it’s easier to detect feeder lines from that direction.
MISTAKE #9: NOT UNDERSTANDING “PROCESS”
A process is work of the same type. This type of work is usually sequential. It should not be mixed.
Here’s the problem with building your Value Stream Maps based on processes:
- If it’s too high level, you will be technically correct, but the VSM will be useless
- If it’s too low level, again you will be correct, except this time the VSM will be overwhelming and confusing
Neither case makes for an effective use of your time.
Here are a few rules of thumb:
- Process is work of the same type. Don’t mix.
- Work is usually sequential, done one step after another.
- A capital machine is its own process. Don’t mix two major pieces of equipment.
- Work has a constant volume. From step to step, the volume does not change. Get one, make one. Not get one, make a dozen.
MISTAKE #8: MISCONNECTING UPSTREAM PROCESSES
In many factories, resources such as fabrication are shared with the primary processes. Responses from these other resources can be less predictable than the primary response times. Attempting to incorporate them directly into the main line can be disastrous.
What can you do to solve this problem?
The easiest solution is to connect them via a Kanban Supermarket. In this scenario, you would not include fabrication on your Process Flow Diagram. Instead, you should treat it seperately.
Over time, you will begin to reduce Kanban Supermarket quantities through strategies like Changeover reductions.
MISTAKE #7: UNDER-RESOURCING OPTIONAL PROCESSES
This one can be a huge mistake, and one that has afflicted me personally in the past, to be brutally honest.
Here’s the scenario: you have optional processes that are not required for every unit. Your line design is based on a certain assumption of mix and volume, and you calculate resources based on that assumption. You know, however, that that mix and volume will probably not ever actually happen.
When the demand for the option on any given day exceeds your design capacity, that option has now become a bottleneck. Let’s look at some of the potential solutions, to avoid this error:
- Sequence — If you never build that option back-to-back, you can overcome this potential constraint. This assumes you can control the build sequence.
- Over-Resource — If an option is required 60% of the time, on average, design for a higher percentage. Look at historic peaks in demand, instead of designing to the average.
- Put in place a FIFO lane — In other words, build ahead (if possible) a certain number of units, in order to handle more than one unit at a time.
- Combine these techniques — There’s no law against having a FIFO lane, plus some additional capacity.
- Don’t overstaff — Design so that operators can move in and out of this process as needed.
MISTAKE #6: EXCESSIVE OPERATOR PART SELECTION
This might not seem like a big deal on the surface, but here’s the rub: when an operator spends a high percentage of each cycle acquiring and selecting materials, over the course of a shift, that time adds up. Furthermore, efforts to reduce handling will result in additional work for operators.
The shorter the Takt time, the more critical this mistake becomes.
Luckily, in this case, there is a simple fix. Give your operators only what they need (within reason, of course). This approach represents a sea change in Lean thinking during the past few years.
Tip: To become even more efficient, design the workstations in a way that avoids the waste of motion. Especially avoid situations where the operator has to leave the station or turn around in order to select a part.
MISTAKE #5: JUMPING TO CAD TOO SOON
The temptation when designing a production line is go to the “layout” stage as soon as possible. But here’s the problem: the more complete a CAD drawing is, the harder it becomes to want to change it. This leads to sub-optimal designs being locked in early in the design process, without enough exploration taking place first.
So, when you feel the temptation to fire up CAD, take a deep breath and follow these steps:
- Start with a block layout.
- Ignore constraints. Focus on the flow, on signaling, and on material delivery.
- Experiment with multiple layout options, without using CAD.
Don’t worry, you’ll get to use CAD later. By following these steps first, your Mixed Model Line Design will improve by leaps and bounds.
MISTAKE #4: NOT USING IN-PROCESS KANBANS
In-Process Kanbans (IPKs) are spaces in a Mixed Model Line Design that are used to store partially completed units between workstations. In practice, an In-Process Kanban will act as a demand signal, as well as a buffer to assist with time variations between different models.
One of the biggest mistakes a Lean process designer can make is to classify IPKs as muda, or waste, and not use them for philosophical reasons.
Here’s an exercise: test it! Run a simulation model without In-Process Kanbans, then again with them. Your efficiency is almost sure to increase, sometimes dramatically.
There are a few things to keep in mind when designing lines with IPKs:
- There are many different techniques for using IPKs. We recommend learning them (say, for example, in an online course or a live workshop) and planning to use them if possible.
- The default number of IPKs is one between workstations. But, as I said before, it’s a good idea to test that with Simulation Modeling before implementation. In general, multiple IPKs tend to have diminishing returns.
- If you’re dealing with extremely large units or unusually long Takt time, it might not be possible to have IPKs.
MISTAKE #3: TOO MUCH LINE-SIDE MATERIAL
There is a common misconception that because material handling is waste, Lean specialists should focus on reducing line-side deliveries. It then follows that having more parts at the line increases flexibility.
In reality: too much line-side material lowers productivity and consumes floor space. We can strongly advice against it.
Instead, work with the material team to design delivery routes for a very modest amount of material at the workstations. You don’t want them to run out, but, contrary to popular opinion, having more parts line-side usually doesn’t help avoid that.
Depending on environment, target one day or less of material at workstations. Kanban supermarkets are a big part of this strategy by acting as a middle-man between the line and the warehouse.
MISTAKE #2: OVERSTAFFING
This one happens almost everywhere. One operator is assigned to one workstation. That operator never moves from that station. Even worse, that person isn’t cross-trained and doesn’t have the skills to work at other stations.
Why is this a mistake?
Under these conditions, any imbalance in the production line will result in lost production, either by being blocked or being starved.
Instead of assigning one operator to one station, try under-staffing the line. The truth is that you don’t usually need a person at every workstation. The proof is in the pudding. Just look at the output of your third shift. Often third shift is not staffed at 100%, and can produce as many units as first shift.
Want some more proof? Just look at the line and see how much of the time people are not actually working. This ties back to Mistake #10: Designing from the Conference Room. So I’ll say it again: You have to go to the factory floor!
Under-staffing the line is used in combination with IPK strategy (as a signal), and requires that Operators are cross-trained. They will also have to be trained in proper flexing technique.
MISTAKE #1: NOT TESTING MIX AND VOLUME
When you design for a given mix and volume, you have to somehow take into account the fact that, in the real world, mix and volume will change from day to day.
This might be a major mistake, but there is a simple solution for this: Simulation Modeling.
Simulation models are able to test a range of mixes and volumes that you cannot accurately account for with only a static model. And, if you have built your static model based on a comprehensive mixed model line design, you will already have the necessary data in order to create a simulation model.