5 Hurdles to Flow Manufacturing

Welcome to Part 4 of a 5-part series on Lean Design. In Part 1 I made the case that the achievement of “flow” is a major goal in the design of a Perfect Value Stream, and in Part 2 I discussed different ways to measure flow and why this is important. In Part 3 I gave you a quick introduction to the design methodology itself. In this lesson I’ll continue that discussion with five hurdles that will need to be incorporated in your design to create the Perfect Value Stream.



You have designed your “dream home”, and you have a great set of drawings. If your architect is technologically advanced, he/she can show you a 3-D model of the house on a computer and take you on a virtual tour of the building. You can visualize yourself in it, walking around, and maybe come up with some adjustments to the design. That’s by far the cheapest time to make changes, before boards are cut and drywall is hung.

The methodology that has been presented in this series will get you very close to an ideal design, but it hasn’t been tested and proven yet. The concern is this: there are a lot of moving parts that interact. You have different products, different work content times, potentially different flow paths, different mix and volume of products that can change daily, normal human variability. These factors will have a negative impact on line performance. If you need 100 units a day, you may find that you can only get 83. You don’t want to wait until the line is physically running to find this out. You need to seriously consider adding a step to the process of creating the Perfect Value Stream, and it is called simulation modeling.

Using simulation software, you (or someone) can build a model of your proposed design, including process times and a daily production schedule. The model will run in simulation time (fast) and collect data on throughput and utilization of line resources. If the production levels and productivity of the model are not where they need to be, you can make changes to the model and your design to improve performance. Remember that the cheapest time to make changes is before you implement the design. The costlier the project, the more important it becomes to validate your design ahead of time.

If you have simulation expertise in-house, great. There are dozens of simulation software packages available, and thousands of free-lance people that can help you. The learning curve is steep, so I would advise against simply buying the software and trying to learn it, unless you are planning to build a career as a simulation specialist.


The term FIFO means First-In First-Out and following FIFO order makes a lot of sense in many cases, like inventory consumption. For scheduling a Mixed Model line, however, FIFO will probably not be the best choice. The order of production makes a huge difference in line performance, unless the variability from product to product is very low. For that reason, Lean production has developed the discipline of Heijunka, or level-loading. The goal is to create a production line-up or schedule that will optimize the work flow and avoid over-loading bottleneck processes when the work content varies. Companies will typically have a set of sequencing rules that Production Control will apply when developing the daily production plans. For example: never run two Model X’s back-to-back because they put a high demand on Feeder Line Y. For Heijunka to be possible you need to have at least some time to plan and mix products. If you have very little lead-time after receipt of the customer order (and there are many industries that need to respond quickly) then your ability to change the mix is much more limited.

You probably already have an idea of what your sequencing rules are, based on practical experience, but if you have a simulation model that’s a great way to run experiments to see which sequencing combinations will give you the best throughput. Formalize (document) these rules and be ready to use them as part of a daily production planning process.


We touched on the importance of buffers or IPKs in the Part 3. In a high-variability environment like a Mixed Model Value Stream, buffers should not be considered “un-Lean” or waste, but rather a necessary balancing tool to improve line performance. Guaranteed: a line without buffers will have lower throughput compared to the same line with them. Common sense needs to be applied to their location and number, but IPKs need to be used.

As with sequencing rules, the use of a simulation model is a powerful tool to test your proposed IPK strategy. It is easy to experiment with buffer locations and quantities to home in on your best throughput.


Having workers move to the work or flexing is one of the most under-appreciated strategies for boosting productivity, sometimes by 30% or more. If, due to imbalances in the flow which certainly occur, workers must wait then productivity goes down. If, however, they can move to a location in the line where they can continue to work instead of waiting, productivity will improve. The flexing strategy is to allow spaces in the line by under-staffing or over-designing the line. If you have fewer people than workstations, then workers will need to move to the open stations to maintain the flow. There is some loss of productivity created by the need to move, but this can be more than overcome by the improved ability to work at an open station.

Adding more workstations to a line than you currently need, or under-staffing a line, seems almost heretical. But the “proof is in the pudding”, and companies that have introduced a flexing strategy report large jumps in operator productivity. See the book The Basics of Self-Balancing Processes by Gordon Ghirann for more information.

I know you’re getting tired of this recommendation by now but having a simulation model will also allow you to test various staffing levels and the impact on utilization and productivity. This is one more reason why you will want to consider building a model.


This is a big one: everyone that will be working on or touched by the new Mixed Model line will need to be trained. The first line of defense in sustaining and improving your Value Stream performance will be with your line leaders and supervisors. These people are the closest management level to the work itself, and standard work for leaders includes the following responsibilities:

  1. Auditing the process for adherence to standard work, use of the material delivery system, flexing discipline.
  2. Training new operators, first in doing standard work at a primary workstation, and then expanding from there.
  3. Collecting and implementing improvement ideas gathered from the operators. This is sometimes called Quick and Easy Kaizen.
  4. Responding to problems, in a first responder role. The operators can “pull the cord” to stop the line and ask for help at any time.

Toyota spends a lot of time developing team leaders and supervisors, and you need to as well.

If you have followed this series, you can appreciate the reality that there is a lot to know and do, and that I have only been able to touch on the main topics. They all merit much more depth and detail. This knowledge and experience are out there in the world, so the need for life-long learning is not an exaggeration! In Part 5 I will be addressing some practical ways for you to advance your skills in designing and managing Mixed Model Value Streams.

Leave a Reply