Home > The Lean Post> Ask Art: Why Do I Need to Switch From Batch to Flow?
The Lean Post
Sharing how the world is making things better through lean.

Ask Art: Why Do I Need to Switch From Batch to Flow?

by Art Byrne
April 12, 2018

Ask Art: Why Do I Need to Switch From Batch to Flow?

by Art Byrne
April 12, 2018 | Comments (16)

No company can become lean unless they move from batch manufacturing or, in the case of service type companies, from functional specialties to flow. This is one of the most important principles when it comes to becoming lean. Functional specialties in, say, a hospital or insurance company create long lead times, excess processing, lots of waiting, excess staffing and unacceptable levels of quality. For manufacturers, we can add excess inventory and floor space to this list. Flow, on the other hand, exposes and eliminates these problems without the need for capital spending.

Let’s start with the batch approach, which is the most common method for traditional manufacturing companies—in fact just about all companies. Maybe it is just human nature that people believe batching is more efficient. I don’t really know. But this belief has been around for many decades. As manufacturing companies moved from early craft production to specialization, and then as things became more machine-based, to a system of one-man-one-machine; somewhere along the line we also decided that similar machines are happiest when they are placed next to other machines just like them. So, we grouped similar machines into functional departments: all punch presses together, all screw machines together all drilling machines together, etc. This was done to keep the machines happy and to try and get efficiencies by having one man say, run two machines. We did a similar thing in hospitals, where we created a large number of independent fiefdoms based on narrow specialties, such that a patient has to travel through 10-12 different fiefdoms to get treated.

Once these functional manufacturing departments are established the fun really begins. Products being made may have to travel through eight different functional departments. Simply figuring how these parts move through the various departments adds an enormous amount of unnecessary complication. People need to create routings and move tickets. You will also need labor tickets to keep track of the cost that is being added at each step. This requires sophisticated MRP (materials resource planning) systems to even have a chance of keeping track of everything.

With this approach the parts travel long distances from department to department to get made. Sometimes miles worth. They are small. They get very tired. We want to take good care of our parts so we build them a nice place to rest in-between their travels. This is the WIP inventory area or as I call it “The Parts Hotel”—nothing more that a bunch of sleeping money that is totally unnecessary.

As you might expect, getting everything through this maze takes a lot of time. Lead times are in the 6 to 8-week range and are locked in by the MRP System that we paid a lot of money to install.  The same thing happens in a bank where a loan application may take 3 weeks to get processed when the actual touch time labor is only seven minutes. A schematic of what this would look like is shown below.

Typical Batch in Manufactiuring

This shows a process for product A that has to travel through eight different functional departments, with rest stops at the parts hotel, in order to get made. A pretty typical situation. But the fun is just beginning. The machines in each of the various departments, and sometimes within a given department, all run at different speeds. This means we need to adjust for this in our planning so that we won’t make too many bases and not enough covers, for example.

Everything goes through each department in batches. The batch size is determined by those clever boys in the finance department based on the length of time it takes to changeover the various machines. The changeover times are taken as a given, “not much we can do about that” and the batch sizes err on the side of being too big rather than too small.

You will find a similar situation in a life insurance company that takes 44-48 days to underwrite a new policy (i.e. give a quote) when the actual time for an underwriter to do the work once all the information that is needed is available is about 15 minutes on average. Think of the waste that exists here. This occurs as the work gets passed back and forth between the underwriters and the case managers just like the functional departments in manufacturing.

A primary focus of any company is to “make-the-month” so we skimp a bit on machine maintenance. The result is that machines break down all the time and always at the worst time. For service companies the computer system goes down. But the finance boys still want to make the absorption hours in order to “make-the-month”. So if one of the eight machines in this example is broken today, finance will insist on running the other seven to make the absorption hours even if we cannot make one single product today that we can sell to a customer. If the machine that makes bases is broken down we still will blast out covers and wind up with a big imbalance in parts. The inventory piles up. We have to move and store it and move it again. We lose it, we spend time looking for it. We damage it in the moving but hey, so what, we made the absorption hours and the month looks pretty good. We have plenty of inventory to cover up the problems and six weeks (the lead time) to get the broken machine fixed and catch up. What could go wrong?

And, oh yes, I forgot to mention the quality issues that this batch approach can cause. How, for example, do we fix the problem if we don’t find out we have that problem until six weeks later at final assembly? How can we answer any the following questions: What department caused the defect? What machine in that department was the defective part run on? Who was the operator, we work on two shifts? Whose raw material was it, we buy from three different vendors? Oh, and we have 6 weeks of defective parts in the system; what do we do with them?  The same is true for the underwriting department, whenever the number of cases stretch beyond 48 days because they somehow got lost in the shuffle. Do we even bother to complete them? After all it is only a quote and our hit rate on quotes is well below 50%.

I think you get the picture. The waste that exists in this functional approach is enormous but we can’t see it. “But we’ve always done it this way.” But what about flow? What are the benefits and how would you do it?

To switch to flow we start with understanding the takt time. What is the rate of customer demand that we have to meet? Let’s say it is 60 seconds required per product A in this case. Based on that we can observe the labor needed in each department to make the part. What we find is that it is very uneven. Almost every operator requires less than the determined takt time to do the work. We don’t see this in the batch state because the work is done in 8 functional departments. In fact, by lining up the machines in a one piece flow cell as shown below we discover that if we load each operator to takt time we can go from having 8 operators to 3 and get the same output. 

One Piece Flow Cell

Of course we have to move a number of machines, and the operators will have to learn to run more than one machine to do this. That is a bit of a challenge, but going from 8 to 3 operators gives us more than a 60% productivity gain, and significant quality gains as well. These come from the fact that now if there is a quality problem you see it right away, not 6 weeks later; and you know what machine, what operator and whose material caused the problem. This gives you a high likelihood of not only solving the problem but of getting a permanent solution. In fact, in my experience, moving to a one-piece-flow can very quickly result in a tenfold improvement in quality. And you never wind up with 6 weeks of bad inventory as you only have 8 pieces in process when you find the problem.

One-piece-flow also forces you to establish a preventative maintenance program. Now, when one machine goes down they all stop. You have an emergency, and so you have to fix the problem now. This may cause some problems early on but they will be solved both quickly and permanently. Also, having things in a one-piece-flow eliminates the need to push product through the system. Now you can pull it. You don’t need a complicated MRP system to schedule the shop floor anymore. Also you no longer need to build to a forecast, which most of the time is wrong anyway. The arrival of a simple kanban card will initiate the order for product A. Inventory goes down, travel distance is reduced, space is freed up, safety is much improved, quality is way better and you have a big productivity gain. Life is good or at least getting better.

But wait, I forgot to mention the biggest gain of all, the strategic benefit that results from moving to a one-piece-flow. Can you guess what that is? Well, in the batch state with functional departments and push scheduling you had a 6-week lead time from when you started till a finished product came out of the maze. Now, from the arrival of a kanban card and the start of production it takes only three minutes, 3 workers times one minute each, to go from raw material to in-the-box. This is a significant strategic win for you as you can now leverage this in the market place by offering a 1-2 day lead time while your competitors are still stuck at 6 weeks. You will gain market share, most of it at full book price, because of your shortened lead times. You will grow and be more profitable as you don’t need to give away your productivity gains if you can compete on speed.

So, in answer to your question, flow production takes something that is inherently very complicated, functional departments and batch production, and greatly simplifies it to your great advantage. So what are you waiting for? It’s clear that flow is the way to go!

The views expressed in this post do not necessarily represent the views or policies of The Lean Enterprise Institute.
Was this post... Click all that apply
26 people say YES
22 people say YES
18 people say YES
16 people say YES
Related Posts
16 Comments | Post a Comment
Andrew Bishop April 12, 2018
1 Person AGREES with this comment


Thanks for the well laid-out arguments.

I think this is one of those cases where a clear direction for change (from the top? from a sensei?) may be necessary vs. getting to flow through the organic, incremental process we rely on for improvement so much of the time. Moving from batch to flow is a radical shift, for all the reasons you've laid out.

I've seen workgroups "discover" one-piece flow in a simulation once or twice but, in actual work on the shop floor, we only ever "got to flow" as a result of a challenge from outside the work group.

What are your thoughts and experiences in driving this change?


Reply »

art byrne April 12, 2018
2 People AGREE with this reply

Andrew, thanks for your comments. My thoughts are that you are correct. Changing from traditional batch to flow requires some outside stimulous. It's best if this comes from the CEO but a good sensei can also get it done if supported by the CEO. Trying to do it without a driver like this is ususually a wasted effort. There will be way too many people throwing up every excuse you can imagine as to why things can't change. Regards, Art.

Reply »

Shahrukh A Irani April 13, 2018

Hi Art:

What are the obstacles to converting a factory from batch production to flow?  I have experienced serious challenges when I work with clients that are not like a Toyota making the same product over and over again on well-defined assembly lines?  Assuming that Lantech, or wherever else you now consult at, has machine shops, fabrication shops, etc. that feed parts to your assembly lines (or bays), I would expect that the nice and clean batch-to-one-piece flow concept would need to be modified/adapted.  How do you do that?

An admirer of all all that you have achieved,

Dr. Shahrukh Irani

Reply »

art byrne April 13, 2018

Dr. Irani, thanks for your question. You are correct that every business is slightly different and as a result what needs to be done to get to flow is never the same. The principles however are the same and so is the basic approach. In your example you would probably start with final assembly and make it flow then work on creating flow in each of the areas supplying component parts. A lot of this would probably invole a lot of focus on set up reduction in the component areas. Once they were in a flow configuration you could start attaching them to the final assembly line to get more of the total process linked up. None of this is hard. The difficulty comes in overcoming all of the objections that the traditional managers will throw up to avoid making any changes. 

eric February 02, 2021
1 Person AGREES with this reply

hello Dr

Nicole April 14, 2018

How do you facilitate one piece flow with processes such as batch passivation or tumbling for micro part fabrication?  These processes are essential to part integrity but also force batch processing.

Reply »

art byrne April 14, 2018

Nicole, thanks for your comments I'm sure this is a question that many people would ask. It is also a good example of the type of objection that can stop a move to one piece flow and lean in its tracks. Some processes are inherently batch by nature. The size of the batch on the other hand is another matter. Batch sizes can be dramatically reduced by cutting set up times or going to smaller more dedicated equipment. You can institute a "behind the curtain" type of approach whereby a small quantity [ lets say 36 parts] are accumulated at the end of step 5 in a tray. When the quantity of part A in the tray reaches 36 then that tray is sent to tumbling. Another tray of 36 parts that has come back from tumbling will then start to feed the succeding steps in the process, in this case step 6. This could be a tray of part A or it could be a tray of part B or C or D etc.

A number of years ago I was on the board of a small jewlery manufacturer making rings with gold and precious stones. It took them 8 weeks to make a ring. One of the issues was exactly like you describe. The rings had to go to a cleaning operation at the far end of the plant twice during the process. This caused long delays. The rings were washed/cleaned in large batches in a very big room. Instead we got a couple of low cost sonic cleaners like the ones used in laboratories and installed them in the line. The lead time to make a ring dropped from 8 weeks to 2 days. Anything is possible if you don't let traditional thinking get in the way.

Ovidiu Contras April 17, 2018

Art, thank you for this important reminder. Lost in the mountain of available Lean literature, this point gets overlooked, with no business benefits as a consequence ... 

What I found interesting in dealing with manufacturing complex products, the trades become batches - sheet metal workers do their magic, then the electrical people come and do their magic, and so on ..... -  with very long lead times . Just to add to your point about functional specialities, this is widespread practice in developing new complex products, like airplanes ...

Reply »

art byrne April 17, 2018
1 Person AGREES with this reply

Ovi, thanks for your comments. I think it is a wide spread approach for many industries, not just building airplanes. I am always amazed at how many companies seem to think they can implement lean while still keeping a funtional organizational structure and continuing to batch.

Reply »

Aki April 18, 2018

Toyota uses functional structure or matrix, right? But no batching... How do they coordinate the flow of the value stream? Comparing diffefent literatures, what is the difference begween a value stream manager and a process owner? Both responsible of flow performance... 

Reply »

art byrne April 18, 2018

Aki, I am not an expert on Toyota's management approach so I don't want to comment on that. I beleive that they control the flow through extensive use of kanban [which are by now probably more electronic than their prior system of physical cards]. As for the difference between a value stream manager and a process owner this is har to answer as different companies have different definitions. I woulld say a value stream manager has cotrol of all the things needed to make a product family complete from raw material to in the box and in the warehouse. A process owner has a much narrower scope like being in charge of a functional department like say stamping.

Sam Beaird April 20, 2018


I thought your article is very good at laying out the benefits to moving to one-piece-flow from batch processing.  However, I have a case which is hard for me to envision going to one-piece-flow due to the processes involved.  Contiuous flow, yes.  One-piece-flow, not so easy.  How do you create one-piece-flow in the chemical industry where the product is made in a batch (mixture of chemicals) and progresses to very small unit of product?  I have faced this in the pharmaceutical industry where the product is made in a batch and then ends up in thousands of capsules and/or pills.  Also, think of the perfume industry where the product is also mixed in batches and progresses to perfume in a very small bottle.  How do you get to one-piece-flow in these type of processes? 





Reply »

art byrne April 21, 2018

Sam, good question. I'm sure others will have the same thoughts. There are many industries/ processes like the one you describe. Obviously it would be very difficult to make an economic batch size of one pill or one perfume bottle. You can make a one piece flow line to bottle the perfume or make the pills but the raw material will probably always be made in a batch of some size. Making the batch size smaller and smaller by focusing on set up reduction then becomes the key approach and you will realize lots of gains by taking this approach, lower inventory, faster customer service, better quality, lower cost are just a few. Product will flow much faster from raw material to final product. Maybe you can just make a one day batch such that everything becomes a finished product the same day. I think you get the point. 

Reply »

Samuel Beaird April 23, 2018

Good thoughts!  I particularly like the suggestion of reducing the batch size.

Reply »

Abhay April 11, 2020

Thanks for such detailed arguments.

I always wonder why we chose batch manufacturing over one-piece flow in the beginning?

What is the history of batch flow?

Reply »

art byrne April 13, 2020

Abhay, I am no historian but I think you can trace batch manufacturing from how the origional craft production evolved. In the ever ending search for productivity people felt that instead of having one craftsman produce a product it would be more efficient to develop specialists to do parts of the process on a repetitive basis as each one could produce more parts per hour than the craftsman and as they were less skilled they were cheaper. Eventually machines were developed to speed each of these production steps. This started with one man one machine but then for productivity sake similar machines were grouped together into functional departments so that you had the opportunity to have one man run more than one machine. Organizing this way of course creates lots of inefficiencies. But the wizards in the finance department were so commited to the idea that cost is reduced the more pieces you can make per hour that lots of pressure was put on production speed even though the different departments and equipment produced at different rates. This of course just piled up unbalanced amounts of inventory which took up massive floor space and added extra costs to move around and count etc. With the drive for speed no one cared about how long it took the machines to changeover just creating even worse batches. I guess you can think your way logically into this situation but once you create it it is a mess. More importantly it is very difficult for those who grew up in this type of organization to even think about how to do it a different way.

Reply »

Avoid the Costly Work of Rework
Beware the Comforts of Waste
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