Henry Ford and Taiichi Ohno achieved flow for Ford and Toyota – Can we do the same?
Henry Ford
Henry Ford could be considered the father of the assembly line. In any case he recognised the importance of flow in production. In his autobiography “Today and Tomorrow” he boasted that the Ford Motor Company could take iron ore from their mine and deliver a car at the factory door in 81 hours (even today no manufacturer can match this sort of speed). Ford was so successful with his process that he was able to increase wages for his workers ($2.34 to $5.00 per hour) and reduce their workweek.
Ford developed his solution when manufacturers produced large numbers of identical products and is famous for saying, ”They can have any colour they want as long as it’s black”. Ford’s model T was the only car produced on the assembly line. The fact that his assembly line hat very limited space between workstations ensured flow, minimized work‐in‐process and eliminated local optima (no workstation could work faster than the slowest. Any problem immediately stopped the assembly line). Efficiency was important only at the limiting workstation.
Taiichi Ohno
After World War II Ohno and the Toyota Motor Company wanted to emulate the Ford production system. Ohno soon discovered that Toyota could not copy Ford. In Japan demand was for a variety of different automobiles – not the original single product concept of Ford. Ohno however wanted to keep the concept of flow. He needed a solution that would avoid massive amounts of inventory on the production floor – the inventory needed to produce different models using different parts.
History tells us that Ohno found his solution in an American supermarket. Supermarkets maintain small quantities of products on the store shelves and keep most of their stock in the back room (their warehouse). As consumers take from the shelves employees rapidly re‐stock them from the warehouse. Store shelves always look full, but inventory in the store (or production line) is minimal. This was the beginning of the Kanban (card) system and the Toyota production System (TPS).
The genius of Ohno’s Kanban system was that it told production workers when to produce and, at the same time, it prevented production when it was not needed. The Kanban system prevented unneeded production by, in effect, telling workers when NOT to produce. Even today many factories have processes in place to bring forward production (to utilize expensive equipment) even though there is no need for the products. In 2009 it is evident, by the huge amounts of unsold cars in parking lots around the globe, that many manufacturers did not heed the message of Kanban and Ohno!
The variety of cars that Toyota had to produce, coupled with Ohno’s Kanban system meant that production lines had to frequently switch from one product to another. Since set‐up times were long and since the common solution for long set‐ups was to produce (very) large batches Kanban and Ohno were n a collision course with traditional practice. Ohno being Ohno ignored tradition. Instead he found an ally in Shigeo Shingo the inventor of SMED – ‘Single Minute Exchange of Die’. There are examples of Toyota achieving changeovers in less than 10 minutes (single minute) when formerly it took many hours. Toyota still consumes probably more time than any other company for changeovers; however, this is what allows their production to flow like water down a river.
Here is an interesting fact. Toyota is still the most profitable and successful car company (and by a considerable distance). Why is this? Is it because Toyota focuses on flow and not cost reduction? It is true that Flow is Toyota’s focus (vs. cost). Nevertheless they probably have the best cost structure of any automobile company.
Is the Toyota Production System Universally Applicable?
Toyota has the luxury of relatively stable demand. Many other businesses are much more volatile with product life cycles much shorter than those of car manufacturers. Some businesses will have a combination of stable and unstable demand. Car manufacturers have the luxury of demand that lasts for a long time, that is stable over time week in and week out and work centre loads are also stable over time with only small changes as demand for products changes. Many production units do not have such stability.
Could it be that factories that don’t have the kind of stability of a car plant will not lend themselves to Lean production solutions1? At the very least the concept of flow needs to be re‐evaluated and probably must be implemented in a different way. Certainly individual work‐centres in many factories do experience significant overloads from time to time.
Ohno has shown that a less stable environment (than that of a single product production like the model T Ford) has even more to gain from flow. If that is true, then even less stable environments will likely also benefit from flow. The question is how to implement flow in the many less than stable environments.
Time Based Supply Chains
Little’s Law states that cycle time and work‐in‐process inventory are directly related. If cycle times are lowered, then ‘automatically’ work‐in‐process will decline too. Ohno developed his system by restricting them amount of stock in the system (with his Kanban cards). Instead of restricting inventory Goldratt’s Theory of Constraints restricts the amount of time available to produce an order. The order is released into production (much) later thus restricting the time and, at the same time, preventing production before it is truly needed. A Kanban system constrains production at every work centre which a time based system des not. Once released an order can flow through production without further restriction. The Kanban system mandates a (small) queue in front of each work centre from which it can pull inventory. In a timebased system most work centres will not have a queue. Instead the queue will collect in front of the slowest unit. The time‐based system needs less stock and protects the constraining work centre with more stock than a Kanban system does. It is more robust. The diagram below shows how work in process will collect in front of the slowest unit.
The CCR is the Capacity Constrained Resource – the slowest unit. The queue will always collect at this point. Queues elsewhere will be non‐existent or very short. All the production line needs is sufficient work in front of the CCR that it never runs out (unless there is insufficient market demand.
“Choke the Release” to Decrease “WIP” and Increase Flow (& Speed)
We cut production lead‐times by 50% by releasing work orders into the factory that much later. What will almost immediately happen is a reduction in work in process by about 50% ‐ much less material will be found in the factory. As long as client due dates are met or the constraining (slowest) resource is not starved of work, production will flow through the factory at a significantly higher speed. Cycle times will be cut in half. Flow has already been improved.
The choice of a 50% cut in cycle times should be quite safe considering that in most factories the actual touch time (the time material is actually being transformed) is less (often much less) than 10% current cycle times. Cutting cycle times in half will still give the factory a lot of queue time for every order and plenty of buffer for non‐instant availability of production resources. Our action should be quite safe!
Priorities
Meeting all client due dates is extremely important if we want to have a reputation as a reliable supplier. Reliability is often one of the important factors for choosing a certain supplier – so we had better be reliable despite the much shorter lead‐times. Work‐orders are quite varied – some contain a lot of work and have a relatively long cycle time, while others contain little work and have a correspondingly short cycle time. The factory needs to be able to prioritize among many different orders – long and short lead‐times. The chosen way is to use the consumption of the work‐orders cycle time relative to the other orders. An order that has consumed 90% of its cycle time and is still near the start of the process is in deep trouble and needs to be expedited. Any order somewhere in the middle of its process and with only 10% of its cycle time (or buffer) consumed is in good shape. There will also be orders with 90% of their buffer gone that are almost in the shipping area – these are also in good shape. Because we have reduced the number of orders in the system it is relatively easy for production to know which orders are in trouble and which are not. This is especially so if we colour code orders according to buffer (or cycle time) consumption into green (0‐33% consumption); yellow (33‐67%); red (67%‐100%) and black (late). Such a report could look like this:
A good rule of thumb is to have between 5 and 10% of orders in the red.
Batch Sizes
Production units often operate with relatively large batch sizes – especially if set‐up times are relatively long. The sales organisation also helps cause large batches by agreeing to lower prices if the client orders a big quantity (a large batch). Large batches are the cause for producing products that should not (because they are not really needed yet) be produced. If we cut times as discussed above large batches will make it difficult to meet client due dates. Waiting time of individual products is very long when operating with large batches. To flow like a river we want and need smaller batches. Lets look at the 2 orders below. Orders A and C are for different products so set‐ups for one cannot be used for the other. Clearly the 2 orders cannot be planned as shown here:
The best we can do if we maintain the integrity of batches is the following:
What often happens especially towards month end or if an order is at risk of being late is that production personnel will split batches. In the situation above the orange resource at the end can start work earlier – but only if partial batches are transported to it from the yellow operation. Thecycle time will be somewhat shorter. Splitting batches is actually quite common practice when an order is under time pressure or to provide a following resource work (since otherwise the resource would be idle).
If splitting batches are allowed we could have the following situation – the two orders can be completed much sooner. In fact the length of the longer green bar shows how much less time is needed if small transfer batches are used. (The shorter green bar shows the effect relative to splitting the last batches – as discussed above.)
Clearly the benefit can be quite large if there are sufficient resources available to move partial batches to the following stations. In the described case set‐up times are not an issue since the two products (A&C) are different and require different set‐ups. In the case discussed the two products require resources in the same sequence. In most production environments there is always a general direction of flow (or sequence of production steps). Variations will exist but a clear general sequence will almost always be evident. The example below shows a situation with 2 products that have a quite different sequence through the factory. Small (transfer) batches are still preferable but the benefit is less.
Again we have two different products. They fit quite well together as most of time there is no conflict for resources – except that the orange resources is required at the same time by both. A viable schedule looks like this. The black and white waiting time is the result of non‐instant availability of the orange resource.
Smaller transfer batches result in the following picture. You can see that the benefit of smaller (transfer) batches is much less simply because of the nature of the two products – their production sequence happens to be complementary.
The green bar shows the much smaller benefit. Nevertheless there is a benefit.
When should we use smaller transfer batches?
Obviously the current practice of using them when an order is in (time) trouble should continue. In other words using small transfer batches will help red orders complete sooner and lowers the risk of them becoming black. Larger batches, because of the long waiting time each individual product suffers, will always tend to be in the red. By splitting these batches and by using transfer batches such orders will complete much sooner and there will be fewer and fewer red (and black) orders. If smaller batches and transfer batches are used consistently, then product will flow more and more like a river through the factory.
Set-Up Times
Obviously smaller batches will cause more set‐ups and therefore more time spent on changeovers from one product to another. Set‐up times are ‘non value adding time’ so from that that point of view they are undesirable. So what do we do about that? Equipment in a factory rarely operates at capacity (sometimes equipment does due to big batches!) so at least from an equipment point of view set‐ups are usually not a problem – the machine would be idle anyway. However, some equipment is closer to its capacity than most of the others. Such equipment must not be turned into a bottleneck by many set‐ups. Set‐ups where there is plenty of capacity is not a problem (just work for people) and care must be taken to not turn a resource into a bottleneck. But what about that other resource – people? A rule of thumb in most operations is that there are also plenty of people resources. If this is true then performing more set‐ups or transporting material from one machine to the next is also not an issue – its just work.
People feel the pressure to at least look like they are working. They are very good in seeking work to be able to show that they are busy and essential to the operation. The fact that production units tend to fire employees when demand is low confirms to them that they must make themselves indispensable. When demand is lower the clever employees slow down their tempo and look just as busy as when demand is high. We don’t want employees producing things not needed yet, but we can give them the work of setting up machines more frequently. In this way they are busy and do not have to worry about what the boss might be thinking! Production flows more smoothly. Remember that workers in a factory represent only a very small portion of a business’ total workforce. They are also usually paid considerably less than many others within a business. Some extra expense to make production flow, should this expense become necessary, is probably well worth it – given the speed and reliability of the flow it creates. Speed and reliability helps create new customers, loyal customers and less pressure on price. The profitability of the business improves. BUT, we will be using a lot of resources to perform set‐ups and some set‐ups take a long time. There will be a limit to the number of set‐ups since they do consume capacity. Can we do something about these set‐ups? Of course we can – as Toyota has shown.
SMED – Single Minute Exchange of Die (part of the Toyota Production System)
Toyota was able to cut set‐up times from, in some cases, many hours to just a few minutes (less than 10). If they can do it, why can’t we? If set‐up times can be reduced to less than 10 minutes for all setups, then transfer and process batches can be reduced further and production will be able to flow even better. So, how is it done?
Give the workers a camera to film themselves performing set‐ups. The purpose of the filming is for the workers to watch how they do this job, and discover for themselves where time is lost. It is amazing how quickly they can improve their performance after seeing themselves on film. IMPORTANT: These films should be private to the set‐up team; management must not see them or use them for performance evaluation.
SMED Training: The set‐up team needs to understand the concepts that SMED is based on:
There are 2 important parts of set‐up – work that can be prepared before the set‐up starts and work that must be done on the machine when it is stopped.
It is important that all preparation steps are completed before the machine is stopped for a new set‐up. This is a ‘Full Kit’ application! If preparations are done well set‐ups will already be considerably shorter.
Changes made to the machine during a set‐up are analyzed in order to develop ways to do some of the work before the machine is stopped – this means moving work into the preparation phase.
Changes made to the machine during a set‐up are analyzed in order to develop ways to shorten the time it takes to do that work that must be done on the production unit or tool.
