Because the Local Distribution Centers are only a short drive from each dealer, a “milk run” parts delivery vehicle can circulate from the LDC to every dealer every two hours, very much the way parts are sent from suppliers to lean assembly plants. And because the LDCs are large enough to stock a few of every active part, practically every car can be repaired the same day with no need for express freight from the Parts Distribution Center at the next level up the system. 6
When the customer first schedules service for a given day, a preliminary order for the necessary parts is prepared. Then, the day before the scheduled visit, when the dealer calls the customer to be sure that the repairs will be conducted the next day, firm orders for parts are placed with the LDC for delivery on the next milk run. Finally, on the morning of the service visit, the dealer technicians examine the car to see if any additional parts will be needed and place orders for these extra parts, to be supplied in two to four hours from the LDC.
While some features of this system may work only in regions with a very high population density—for example, Japan and many areas of Western Europe—the additional gain in parts system efficiency and level of service for the customer is striking, as shown in Table 4.1 .
T ABLE 4.1: P ARTS D ISTRIBUTION E FFICIENCY AND L EVEL OF S ERVICE , T OYOTA U.S.A. and JAPAN
Service parts warehouses are, of course, Type-One muda, being necessary to run service systems at the present time but not actually creating any value. However, as stock levels fall and replenishment orders grow smaller and more frequent, PDCs will look less and less like warehouses and more like cross-docking points. Many parts on their way to a dealer will simply be moved from the incoming container to a roll-cage containing the dealer’s order, without ever being binned. Instead of a series of deep lakes with little flow, the PDCs will gradually become wide spots in the channel where tributaries come together and parts are speeded to the required destination.
Perhaps at some completely lean point in the distant future it will be possible to use stereolithography and other emerging technologies to actually make parts at the dealership one by one as they are needed. However, the improvements instituted by Toyota in Japan and the United States in the past few years are available to any service business in any industry right now and constitute a remarkable leap compared with most current practice.
Is Chaos Real?
The introduction of pull in the Toyota service value stream, even to the partial extent achieved to date, raises profound questions going far beyond this particular value stream. Specifically, what happens to the “chaos” that observers have detected in many product markets when customers can pull value practically instantly from raw materials into reality? And what happens to the macroeconomy when lead times and inventories largely disappear?
Since James Gleick published his fascinating book Chaos 7 in 1987, it has become fashionable for business writers to talk about chaotic markets and the need for organizations to be able to instantly respond. Much of the writing on reconfigurable “virtual” corporations (whatever those are) and chaos management stems from this new perception of reality. Indeed, to apply to business MIT meteorologist Edward Lorenz’s original metaphor for a chaotic system—the world’s weather where the nonlinear nature of forces potentially makes it possible for a butterfly in Beijing to affect the weather a few days later in New York—managers today seem to be living in fear of butterflies.
In our view, this new way of thinking is appropriate for purely physical phenomena like the weather but miscomprehends the nature of customer-producer relations. Indeed, in looking at the great bulk of the world’s industrial economy, the most striking feature of this decade is the relative stagnation and predictability of most product markets. In activities ranging from motor vehicles to aircraft to industrial machinery to personal computers to home building, the trajectory of product technology is quite predictable. What’s more, the end-use demand of customers is inherently quite stable and largely for replacement. We believe that the volatility—the perceived marketplace chaos—in these industrial activities is in fact self-induced, the inevitable consequence of the long lead times and large inventories in the traditional world of batch-and-queue overlaid with relatively flat demand and promotional activities—like specials on auto service—which producers employ in response. 8
One solution—as recently proposed by Peter Senge 9 —is the creation of learning organizations which can reflect upon these phenomena and respond to them. One might think of a learning organization as a sort of intellectual MRP to take the kinks out of production and consumption.
We have a radically different proposal: Get rid of lead times and inventories so that demand is instantly reflected in new supply rather than the current situation of misjudged supply perennially searching for demand and creating chaos in the process. We are confident that the pattern of demand will suddenly be seen for what it is: remarkably stable except for a few new products—like multimedia—whose value and final form are being determined in real time.
Do We Really Need a Business Cycle?
If we can get rid of lead times and inventories to give people what they want when they want it, we believe that demand will stabilize for another reason: the damping effect on the traditional business cycle.
Conventional wisdom among economists is that about half of the down-swing of economic activity in business cycles is due to consumers and producers working off the inventories built up toward the top of the cycle. Similarly, about half the upswing is due to building up new inventories in expectation of higher upstream prices (“buy raw materials now to get a bargain before prices go up”) and to the expectation of greater downstream sales that require plenty of product in the distribution channel to supply, but which never quite materialize). 10 And no amount of government fine-tuning and countercyclical intervention has been able to damp the amplitude or frequency of the cycle during the fifty years since World War II. 11
Unfortunately, our hypothesis that largely eliminating inventories will greatly damp the cycle can’t be tested just yet, despite several decades of lean thinking in Japan and a decade of awareness about JIT in the United States and Europe. When one looks at the data on inventories, the amount associated with any given level of economic activity (normalizing for the business cycle) hasn’t budged in America, Europe, or Japan. The reason, we believe, is that most applications of JIT, even in Japan, have involved Just-in-Time supply, not Just-in-Time production, and batch sizes have not been reduced by much. Thus, nothing has happened except to push inventories of the same magnitude one step back up the value stream toward raw materials, and one of the great prizes of the lean leap is still waiting to be claimed .
Pulling Value in Pursuit of Perfection
We hope you can now see the need to precisely specify value and to identify every step in the value stream for specific products, then to introduce flow, and next to let the ultimate customer pull value from its source. However, much of the potential of lean thinking is lost unless you take the final principle to heart. We’ll end Part I of this book with some thoughts on perfection.
CHAPTER 5
Perfection
The Incremental Path
When Joe Day, the president of Freudenberg-NOK General Partnership (FNGP) of Plymouth, Michigan, began in 1992 to introduce lean thinking in the North American alliance between the world’s largest seal and gasket makers, 1 he noticed something very curious. No matter how many times his employees improved a given activity to make it leaner, they could always find more ways to remove muda by eliminating effort, time, space, and errors. What’s more, the activity became progressively more flexible and responsive to customer pull.
For example, when Freudenberg-NOK set out to reorganize the manufacture of vibration dampers in its Ligonier, Indiana, facility, an initial kaizen event achieved a 56 percent increase in labor productivity and a 13 percent reduction in the amount of factory space needed. However, in revisiting this activity in five additi
onal three-day kaizen events over the next three years, it was gradually possible to boost productivity by 991 percent while reducing the amount of space needed by 48 percent, as shown in Table 5.1 . What’s more, additional improvements are possible and planned for the future.
This seems to defy all logic. After all, there are diminishing returns to any type of effort, aren’t there? Kaizen activities are not free, and perfection — meaning the complete elimination of muda —is surely impossible. So, shouldn’t managers eventually stop efforts to improve the process and simply manage it in a steady state, avoiding variances from “normal” performance?
As we have reviewed data similar to those in Table 5.1 with senior managers in many firms around the world, we have found two prevalent reactions. One is that steady-state management—management of variances—really is the cost-effective approach once an activity has been “fixed.” The other was summarized by a senior manager of an English firm, which had done nothing to fix its product development, scheduling, and production systems but was planning to do something. “Why didn’t FNGP get the job done the first time! Why didn’t they conduct a thorough planning exercise to identify the perfect process at the outset so they wouldn’t waste three years before finally getting it ‘right’?”
T ABLE 5.1: R EPEAT K AIZENS ON S AME P ART N UMBER , FNGP L IGONIER , I NDIANA , F ACTORY , 1992–94
Both reactions show how traditional management fails to grasp the concept of perfection through endless steps, which is a fundamental principle of lean thinking. Because FNGP is one of the most relentless pursuers of perfection we have found, their approach makes an excellent illustration of what perfection means in practice and how to pursue it.
The Radical Path
There is an alternative, radical path to perfection, a total value stream kaikaku involving all the firms from start to finish. Glassmaking for the automotive industry provides an interesting example. Currently in North America, Japan, and Europe, manufacture of the fixed glass for cars and trucks (excluding the glass mounted in doors which moves up and down) involves very similar steps no matter which companies perform them. (These are shown in Figure 5.1 .)
The first step is the glass float, a vast device in which silica is melted and floated on a reservoir of liquid tin. Sheets of glass are pulled off the float, cut into rectangular shapes, and carefully cooled. Because of the size of the typical float and the problem of getting batch-to- batch consistency, large batches are produced and stored for considerable periods before shipment to the glass fabricator.
F IGURE 5.1: A UTO G LASS T ODAY
The glass fabricator cuts the glass to net shapes (discarding about 25 percent in the process). The net shapes are then heated to just below the melting point and positioned in dies of the desired shape, where they are “drooped” (without any pressure) or “pressed” (using an upper die to stamp them into shape) into the final geometry needed to precisely fit the frame of the car. Again, the complexity of changing the dies and the problems of achieving batch-to-batch consistency have caused glass fabricators to manufacture enormous batches of a given part number and to store them before shipment to the glass encapsulator.
The encapsulator takes the glass from its own incoming storage and inserts each piece in a molding machine which injects some form of rubber or plastic (most commonly polyvinyl chloride) into a channel around the perimeter of the glass to create a waterproof seal and an expansion joint for attaching the glass to the steel auto body.
After some additional storage at the encapsulator, the glass is shipped to the auto assembly plant, where it is installed in the car.
Clearly, there would be substantial gains from incrementally improving each step in this process. For example, pull systems like those described in the last chapter could be introduced for each replenishment loop and tool changes could be speeded up, particularly by the glass presser, to make smaller batches. However, there would still be enormous amounts of muda due to the distant location of the four plants involved and the large amounts of time-consuming, expensive transport. What’s more, quality problems causing high levels of scrap would still be difficult to address because of the long time lags between the pressing, encapsulation, and installation steps, where problems with the previous step are most likely to be discovered.
A radical leap toward perfection in this process would involve right-sizing the glass float for the amount of product needed by a specific customer, dramatically reducing batch sizes in the pressing step and conducting it at the end of the float to save the energy required to reheat the glass, then conducting the encapsulation step in continuous flow at the next workstation from the pressing step, and finally locating this whole activity across the road from the auto assembly plant so the pull of the plant could be answered instantly (as shown in Figure 5.2 ).
F IGURE 5.2: A UTO G LASS A FTER R ADICAL R EALIGNMENT
No one has pursued this approach because, like most truly radical rethinks of a value stream, a number of firms (four in this case) would need to cooperate in changing their methods by forming a lean enterprise for this product (which might best be defined as all of the fixed glass needed for a specific auto assembly plant). However, if a lean enterprise were formed to rethink the whole value stream, additional radical reconfigurations would no doubt follow as the enterprise asked: What is the real value here for the customer and how do we create it? At a minimum, it would be necessary to rethink the proper location for product design (the auto company, the glass presser, the glass encapsulator, or some alliance of all three?) and the flow of service and crash parts .
Continuous Radical and Incremental Improvement
In fact, every enterprise needs both approaches to pursue perfection. Every step in a value stream can be improved in isolation to good effect. And there is rarely any ground for concern about investing to improve an activity which will soon be replaced altogether. To repeat the lesson from Chapter 3 : If you are spending significant amounts of capital to improve specific activities, you are usually pursuing perfection the wrong way. Going further, most value streams can be radically improved as a whole if the right mechanisms for analysis can be put in place.
However, to effectively pursue both radical and incremental improvement, two final lean techniques are needed. First, in order to form a view in their minds of what perfection would be, value stream managers need to apply the four lean principles of value specification, value stream identification, flow, and pull. (Remember, you want to compete against perfection, not just your current competitors, so you need to be able to gauge the gap from current reality to perfection.) Then, value stream managers need to decide which forms of muda to attack first, by means of policy deployment (often called hoshin kanri in Japan, where these ideas originated).
The Picture of Perfection
At every step we’ve noted the need for managers to learn to see: to see the value stream, to see the flow of value, to see value being pulled by the customer. The final form of seeing is to bring perfection into clear view so the objective of improvement is visible and real to the whole enterprise.
We’ve just presented an example for glassmaking: a radical rethink of the whole value stream so that all value-creating steps are conducted immediately adjacent to the customer and exactly when needed. Toyota certainly had a picture of perfection—derived from its mastery of lean principles—when it set out in 1982 to rethink its Japanese service parts business, and then in 1989 when it began to apply the same concepts in North America. And Tesco needs a vision of perfection for the value and value stream of its beverage lines, as described in Chapter 2 .
Paradoxically, no picture of perfection can be perfect. If the value stream for automotive glass could be reconfigured as we suggest, it would then be time (immediately!) to imagine a new perfection which goes even further. Perfection is like infinity. Trying to envision it (and to get there) is actually impossible, but the effort to do so provides inspiration and direction essential to making progress along
the path. We’ll return to this theme in Part III .
One of the most important things to envision is the type of product designs and operating technologies needed to take the next steps along the path. As we have seen repeatedly in the preceding chapters, one of the greatest impediments to rapid progress is the inappropriateness of most existing processing technology—and many product designs as well—to the needs of the lean enterprise. A clear sense of direction—the knowledge that products must be manufactured more flexibly in smaller volumes in continuous flow—provides critical guidance to technologists in the functions developing generic designs and tools.
In addition to forming a picture of perfection with the appropriate technologies, managers need to set a stringent timetable for steps along the path. As we will see in the examples in Part II , the greatest difference between those organizations that have done a lot and those that have accomplished little or nothing is that the high achievers set specific timetables to accomplish seemingly impossible tasks and then routinely met or exceeded them. The low achievers, by contrast, asked what would be reasonable for their current organization and disconnected value streams to accomplish, and generally defeated themselves before they ever set out.
Lean Thinking Page 11