We are big fans of Simulation Modeling for lean designs, especially now that it is a lot easier to build models with tools like the Lean Design Simulator. We’re not alone…I heard from a simulation industry insider that Toyota has embraced modeling in a big way in the last few years.
Remember that your goal is to deliver a Value Stream design that is extraordinarily lean to begin with, and one that does not require a lot of Kaizen tweaking afterwards.
No matter how experienced you may be, it is virtually impossible to understand in your head how a Mixed Model workflow will perform in the face of changing volumes and mix. Trying to do so would be like an aeronautical engineer claiming that he doesn’t need to use a wind tunnel because he has a lot of aeronautical experience.
Don’t think that the use of simulation modeling tools is “unlean”!
CHALLENGES OF MIXED MODELING
When you attempt to mix different products in the same line or workflow, and those products have different work content times, different volumes, potentially different flows, different subassemblies, then one of four things can happen at a workstation.
- The operator at that station is able to do standard work and add value to the process.
- The operator is wanting to do work, but can’t move the previously completed unit out of his/her workstation because the downstream station is full. This is called being blocked.
- The operator is wanting to do work and his/her workstation is available but there is nothing to work on because the upstream workstation is taking longer. This is called Waiting Time.
- You could have a situation where there is legitimately no demand, i.e. nothing needed to be worked on. We call that Idle Time. The more differences you have in the work content, and variation in workflows, the more time an operator will spend being blocked or waiting, and the less time doing value-adding work. The impact to throughput and productivity can be huge. It gets even worse when the mix of products and volumes can change daily. You can even experience shifting bottlenecks depending on the mix, where the constraints to the flow are not always in the same place.
Another challenge that can bite you is the management of upstream processes. You may have a feeder process that is capacity-constrained, so that if the wrong mix of products is sequenced it gets overloaded and can’t keep up with its “internal customer”.
Knowing how to avoid this through proper planning is a part of a daily production discipline.
HOW IMPORTANT IS SEQUENCING?
By “sequencing” we mean putting the production plan in a specific order, to maximize throughput and productivity (and also deliver on time, of course). A Mixed Model manufacturer will develop a series of “sequencing rules” that are applied to the order backlog. This planning often takes place daily, or it may be needed less often if the volume and mix are more stable, or if the Takt Time is longer.
The sequence is the order in which products will be built, and maintaining the sequence is usually important. If the green doors arrive to be installed on the red car, that could be a problem!
Is Good Sequencing All You Need? No. Sequencing usually goes hand-in-hand with correctly sized In-Process Kanbans (buffers).
Think about this for a moment. One common sequencing strategy to even out a workflow is to alternate high and low work content products. The sum of the work content approximates the takt time, so when you build them together they “cancel each other out”. But…this can’t happen if you don’t allow an extra unit in-between.
If Station 1 is done sooner with the lower work content product, unless the product can be moved out so it can start another unit, Station 1 will be blocked. The buffer therefore helps to overcome those time differences (within limits). In summary, a good sequencing plan also includes a good buffering plan to go along with it.
SO HOW DO I CREATE SEQUENCING RULES?
You can develop sequencing rules based on your past experience or create them as you run the actual product line. Of course validating your sequencing strategies in the crucible of actual experience is essential.
Our recommendation is to create and test different sequences using simulation modeling as a part of the design process. Run many experiments in simulation time (fast) and build your strategy in that way first. Improve the rules later based on actual experience, but you should be close to a good set of rules on Day 1.
Your experiments may demonstrate sequencing options that you had not considered. For example, instead of trying to mix products evenly across the day you might find that building in small groups gives you better throughput and productivity. That is something you can easily test with the Lean Design Simulator and the AutoSequencer.
Here is a list of common sequencing scenarios that you can consider for testing:
- Mixing High and Low Work Content. This is a very common sequencing rule that attempts to cancel out time variations by alternating products. Your ability to actually do this is controlled by the size of your order backlog, and how long customers are willing to wait.
- Sequencing in Work Content Order. This is similar to Strategy 1, but the sequencing process involves starting with the highest work content product and working your way down the list (sorted by work content time). A rule to separate these products will allow you to space them out and not build them back-to-back, within the limits of the time slots available.
- Sequencing by Impact on Upstream Processes. You may know that certain upstream processes are capacity constrained so that you never want to build certain product close together in the sequence.
- Benefits of Building in Groups. Although it sounds a bit unlean to say so, there could be benefit in building in daily quantities or in small groups with the same product repeated. That’s because the station-to-station variation for the same product is much less than when you change products one after another. There will be a “bubble” of time difference when you change products, but that doesn’t happen after every unit.
- Sequencing by Product Parameters. If you have a paint process in your flow it is common to find a production plan sequenced by product color, going from light to dark colors. That’s because the changeover time between colors can be reduced if you are going to a similar but darker color.
- Sequencing by External Supplier Constraints. You may have external suppliers who are tightly linked to your production line and who are limited in their ability to supply certain components. A good sequencing plan will help you to manage those limitations.
As you can see, it gets complex!
That’s why the creation and use of a simulation model and an autosequencer can be so helpful in testing these many options.
Let’s take a look at an actual example of the Lean Design Simulator and the Autosequencer in action, to give you a better idea of how this tool works. Click on the video below to for a video introduction.
- Skillful sequencing of your demand can have a dramatic impact on your Value Stream performance. Throughput will go up, productivity and utilization will go up. All for the low price of doing a good job of planning.
- Your best option today is to build a simulation model of your proposed (or existing) production line or Value Stream and test its performance with a reasonable variety of sequencing scenarios. Computer simulation will let you test those scenarios quickly, and at a fraction of the cost of testing them in the physical world.
- The AutoSequencer is a tool to speed up the work of creating a sequence. You enter the desired parameters and the sequence is created in a few seconds.
- The Lean Design Simulator is a discrete-event simulation modeling tool that makes it easy to build models (in Excel), run them (in a few seconds), and analyze the results in detail.
Many thanks to Christoph Roser for his excellent series on Mixed Model Sequencing. Check it out at his website AllAboutLean.com.