Because most of Bumper Works’s customers, as of 1992, were still ordering massive batches—one-month lots to be delivered by the last day of the month—Bumper Works decided to prepare for the future by creating its own daily schedule using a technique Toyota calls level scheduling. Shahid Khan’s production manager would take the orders for the next month, let’s say 8,000 of Bumper A, 6,000 of Bumper B, 4,000 of Bumper C, and 2,000 of Bumper D. She would add them up (to get 20,000) and divide by the number of working days in the month (say, twenty) to discover that Bumper Works would need each day to make 400 of Bumper A, 300 of Bumper B, 200 of Bumper C, and 100 of Bumper D (with a takt time of .96 minutes). This would require four changeovers of the blanking and stamping machines, totaling 88 minutes (9 percent of the 960 minutes of two-shift working time) at the maximum allowable changeover speed of 22 minutes.
The daily schedule was given to the welding booth to start the process. As the booth used up its reserve of inner and outer panels and brackets for Bumper A, the welders would slide the empty parts tub and its associated kanban, or signal card, down the short slide to the stamping machines. This provided the only signal needed to stamp more parts for Bumper A. Then, as the stamping press used up its blanks for Bumper A, the empty parts tub was sent back down the slide to the blanking machine, providing the only signal needed to make more blanks for Bumper A.
The in-plant MRP system that had been sending orders to every machine—but which never quite worked right so that expediting was always necessary to keep production going—was no longer needed. The new, simple system of pull and visual control always worked once the inevitable start-up problems were resolved. Bumper Works’s new operating doctrine could be summarized simply as “Don’t make anything until it is needed; then make it very quickly.”
But there was a problem right in the heart of the new system. The steel bumpers once welded required a coat of chrome before they could go to final assembly. This was a complex process conducted by specialist firms operating in batch mode. Shahid Khan’s chromer, Chrome Craft in Highland Park, Michigan (near Detroit), was the best supplier Bumper Works had found but was not in step with the new approach. Bumpers disappeared into Chrome Craft and didn’t reappear for weeks. What was more, getting a rapid turnaround on expedited orders was impossible.
Khan and the Toyota sensei were soon on their way to Chrome Craft, where president and owner Richard Barnett watched with some amazement as rapid changeovers were implemented on Barnett’s bumper polishing machines so that small lots could be pulled from the loading dock, taken through the necessary polishing process, and run through the long line of chroming tanks. (Chrome Craft was doing bumpers for other manufacturers and had dozens of bumper types going through its plant.)
By arranging for quick unloading and loading of the Bumper Works truck, it became possible to bring a load of bumpers in at 7:00 A.M. while picking up the load just completed, then to return at 3:00 P.M. to pick up the freshly chromed versions of the bumpers dropped off at 7:00 A.M. By 1995, a bumper’s time in the Chrome Craft plant had fallen from fifteen days, on average, to less than a day. What was more, at the end of every shift the entire output of Toyota bumpers was being trucked out of the plant, leaving zero in-process inventories. Chrome Craft’s inventory “turns” on Toyota bumpers had zoomed from about twenty to about five hundred per year.
Even this achievement is by no means the limit. In mid-1995, Chrome Craft helped install a right-sized chroming operation in a new Flex-N-Gate plant in Indiana making bumpers for the American Big Three. This brings time-in-process down from twenty-four hours (consisting of two eight-hour truck rides from Bumper Works to Chrome Craft and back, plus eight hours at Chrome Craft) to about eight hours.
As Bumper Works learned how to pull value through its system, it became capable of responding practically instantly to customer orders. Because of its quick changeover ability, Bumper Works could start welding a given type of bumper within about twenty minutes of receiving an order and it could easily vary its entire production as demand changed. All that was needed was to drop off a new set of order cards at the welding booth. Similarly, the time elapsed between the arrival of a flat sheet of steel on Bumper Works’s loading dock and the shipment of a finished bumper to the customer fell from an average of four weeks to forty-eight hours. Quality also zoomed, as it always does when flow and pull thinking are put in place together. As of mid-1995, Bumper Works hadn’t shipped a bad bumper to Toyota in five years.
The new system gave Bumper Works and Chrome Craft the ability to make small lots of bumpers at short notice—for example, a few replacement bumpers of the type Bob Scott needed—but Khan’s customers did not know how to take advantage of his new capabilities. Until very recently, even Toyota was still ordering large batches, then erratically changing its orders as shortages developed in the distribution system. Another step was needed to create a smoothly pulling value stream.
The Bad Old Days of Distribution
When Toyota introduced its Corona model in America in 1965, it suddenly began to sell large numbers of cars. These needed service parts, everything from new bumpers to replace those crumpled in accidents (like Bob Scott’s) to oil filters and spark plugs for periodic maintenance. Because of the long shipping time from Japan, Toyota needed large stocks of parts in North America and soon built a network of warehouses—called Parts Distribution Centers, or PDCs—stretching from Los Angeles to Boston.
In 1965, the Toyota Production System (TPS) was just being implemented in Toyota’s supplier plants in Toyota City. No one had given any thought to applying TPS principles to Toyota’s Japanese service parts warehouses, much less in faraway American warehouses. As a result, the eleven PDCs Toyota built in the United States were laid out like every warehouse in America. Each had vast bins stacked to a high ceiling, thousands and thousands of them, one for each type of part. The bins were lined up in long rows to create endless aisles in a massive square of a building.
The PDCs received parts from Japan in sealed containers, typically in large batches coming off massive container ships at weekly intervals. When the containers arrived at the PDC, they were opened in the receiving area and the parts were given to “stockers” with carts who walked up and down the aisles placing the parts in the proper bin. Because fifteen days were required back in Japan to assemble an order, another thirty-eight days were required for ocean shipping, and five days were needed at the PDC to bin the parts, the PDC needed to order parts at least fifty-eight days ahead of probable need to ensure uninterrupted supplies to Toyota dealers.
Toyota dealers, like Sloane Toyota, placed orders for parts once a week, by estimating likely increases or decreases in demand before the next weekly order. Because these forecasts were often wrong, they produced what Toyota calls “created demand”; that is, dramatic waves of orders traveling back up the value stream that are unrelated to actual demand expressed by real customers like Bob Scott. When the weekly orders were received at the PDC, a “picker” was dispatched to collect the appropriate parts from the appropriate bins in the appropriate aisles and take them to shipping. The parts were then delivered to the dealer by common carrier truck service the next day.
Because Toyota accepted the notion that large batches, expressed as “economic order quantities,” were efficient due to savings in shipping costs, and because overnight shipment of parts was expensive, it encouraged its dealers to order large amounts of each part whenever they replenished. To make this attractive, Toyota paid the freight for the large weekly batches and allowed dealers to send back up to 5 percent of the value of a weekly order for a credit if they ordered too many parts of a certain type—for example, for a special service promotion which failed to meet its target.
In the event that the dealer didn’t have a part in stock—for example, a bumper for Bob Scott’s pickup—a VOR or “vehicle off road” order system was able to locate and deliver the exact part needed to the dealership before noon the next day. This system
looked electronically into the inventory at the nearest PDC; then, at all PDCs; and finally, at Toyota’s national warehouse in Torrance, California, to find the part, print a shipping order, get the order to the pickers in the appropriate warehouse, and get it shipped. To cover the cost of this premium service, Toyota required the dealers or the customer, like Bob Scott, to pay the express freight charge for getting the parts there quickly. In this way dealers could keep large numbers of the most frequently used parts on hand while ordering special needs overnight.
At the PDC, the bins for each type of part were large and the shipping containers were larger. And the container ships were truly massive. Air freight to supply parts in the event of a shortage was very expensive, so it seemed like common sense to order large batches of a given part whenever stocks at the PDC began to run low. In addition, Toyota’s scheduling computer, reaching all the way back to the factories in Japan, was programmed to anticipate certain events—the onset of winter, when more bumpers are crumpled, or sales promotions, when a large number of oil filters and spark plugs are needed in a short time as dealers offer a “special” on routine service. Extra orders were added to ensure adequate supplies for these predictable surges in demand.
By the time Toyota’s warehouse network was fully in place in the early 1970s, the typical PDC had a six-month supply of the typical part. In addition, a special area of the national warehouse in Torrance housed very low-volume, rarely ordered parts, often for very old Toyotas. The months of supply on hand in this warehouse was difficult to calculate because some parts might never be ordered. Shortages still occurred, for reasons which always seemed mysterious, and some air freight across the Pacific was still necessary, but in general the system ran pretty well and permitted Toyota to achieve the highest “fill rate” (or percentage of parts available from the PDCs on demand) in the North American auto industry, at 98 percent. For fifteen years, it was “good enough.”
Lean Distribution for Pull
When Toyota began to assemble cars in the United States at the Fremont, California, joint venture with General Motors (NUMMI) in 1984, it began to develop a network of suppliers for bulky and “commodity” items—tires, batteries, and seats. Then, when Toyota opened its mammoth Georgetown, Kentucky, plant in 1988, it needed a comprehensive network of suppliers for a wide variety of parts.
These same parts were needed for routine service and body-shop work at Toyota dealers; so in 1986, Toyota had opened a receiving warehouse for American-made service parts in Toledo, Ohio. This Parts Redistribution Center, or PRC, was where Shahid Khan shipped his bumpers once he started producing for Toyota.
A major mission for this facility was to reduce shipping costs per part by consolidating the less than truckload shipments of parts received from suppliers into fully loaded trucks for onward shipment to each PDC. However, this focus on low freight cost per part created a classic batch-and-queue operation in which a month’s worth of parts were queued up at each supplier before shipment to the PRC. Upon arrival, the parts were queued again for quality inspection and then delayed one more time in a staging area awaiting a full truckload before shipment to each PDC.
As the yen strengthened at the end of the 1980s, and American competitors like Ford began to implement some aspects of the Toyota Production System, Toyota executives started to wonder how they could sustain their competitive advantage. In addition, Toyota’s four-year replacement cycle for every model, its steadily expanding range of models on offer in the United States, 3 and the tendency of Americans to drive their cars longer and longer, 4 meant rapid growth in “active” part numbers which Toyota needed to stock as replacement parts to keep its customers happy. This seemed to require larger and larger inventories of parts and ever-growing distribution costs.
As Toyota executives pondered this situation, it occurred to them that they had never applied any of Toyota’s lean thinking to their North American warehousing and distribution system. As they thought about this, it quickly became apparent that startling advantages could be gained if they did .
The Toyota warehouses at that time were run in the familiar batch-and-queue mode we described in the Introduction and again in Chapter 3 . Supervisors directed hourly workers to haul large carts or forklift loads of incoming parts from the receiving area down endless aisles for binning. The supervisors tried to ensure that the “stockers” were working hard when they were out of sight by giving each worker the same number of “lines” to stock during each shift. A “line” was a specific part number—for example, Bob Scott’s deluxe chrome bumper carries Toyota corporate part number 00228-35911-13—with a varying quantity of that part, perhaps only one but sometimes hundreds.
Each “line” could therefore involve a very different amount of work. Putting one hundred spark plugs on a low shelf was a lot easier and could be completed much faster than hoisting one heavy bumper into an upper bin, yet both counted as one stocking line. Because each supervisor gave each stocker the same number of lines to complete during the shift, there were endless claims of favoritism or punitive assignments. “You’re giving me all the heavy bumpers because I refused to go on the night shift when you were shorthanded,” etcetera. What was more, it was practically impossible for supervisors to determine the cause when stockers failed to complete their runs in the allotted time. Was it because bins were too full to hold more parts or because of a faulty forklift, or was it simply a matter of unsupervised workers relaxing on the job? Without accurate identification of causes it was hard to implement remedies and improve practices.
The same organization and logic regulated “picking” parts for the weekly shipments to the dealers. In addition, there was an expediting system in place for the “hot list” VOR parts which were needed the next day by dealers. Unfortunately, the VOR orders often caused chaos among the pickers and slowed down the routine picks for weekly dealer orders, and it’s easy to see why. A picker would be told at the last minute to run all the way across the warehouse to get a single part to meet the air freight pickup deadline. If this need had been anticipated, the pick could have been part of a complete circuit of the warehouse for many parts and would have been much more efficient.
But perhaps the worst features of the warehouse system in the late 1980s were the size of the bins, the inefficient use of storage space, and the size of the batches ordered as replenishments. Both the bins and reorder quantities were massive, involving hundreds or thousands of parts of a given type and number. This inevitably meant months of spare parts on hand and large facilities to hold all of them. Large facilities, in turn, were time-consuming for stockers and pickers to work their way around.
As Toyota executives thought about this situation, the solution to the stocking half of the problem became obvious: Toyota should dramatically shrink the size of the storage bins, and reduce the lot size for reorders. Instead of ordering from suppliers on a weekly or monthly basis, why not order daily and order just the amount shipped to the dealers that day? This was much more practical for domestic parts obtained from suppliers, like Bumper Works, who had mastered lean techniques and could respond to requests for small amounts. Fortunately, Toyota was rapidly transferring production of its parts from Japan to North America, and many suppliers were starting down the path pioneered by Bumper Works.
The other half of the problem, the picking, could be solved with an equally dramatic rethinking of relations with the dealers. Instead of asking dealers to order large batches weekly and then make special requests each night for missing parts, why not have the dealers order daily and order just the amount sold to customers that day?
Toyota knew that its dealers would strenuously object, unless the company offered to pay the freight for the daily shipments. Yet, a bit of analysis showed that if Toyota shipped parts from its eleven PDCs to the dealers in each of its eleven sales regions every night, the extra costs of the trucks would be offset by the simplification of the picking process, savings on inventory carrying costs, and the elimination
of express delivery charges. In addition, day-to-day consistency in orders, with no sudden waves, would allow consolidation of some truck routes.
There was one last problem to solve. This was the crisis at the dealer when a customer like Bob Scott came in with a request for a part not normally carried in the dealer’s parts inventory. Of course, the part could be supplied overnight by the new system, as it always had been, but the customer would be unhappy. Customers want their cars fixed right now!
Toyota realized that if dealers ordered every part daily to replace the exact number sold that day, dealer inventories of parts could be reduced dramatically. As dealers reduced their average stock of each part number, they could afford to increase the range of part numbers on hand. Instead of having hundreds of the most common parts requested and none of those requested less frequently, dealers could have a small number of each part across a very wide range. In this way they would be more likely to have a low-volume item like a bumper for an older vehicle when a customer like Bob Scott asked for one.
From Theory into Practice
The logic just described for introducing a pull system in warehousing that responds faithfully to actual customer demand was understood by Toyota’s North American executives by the late 1980s. Getting it fully in place, however, has required years, even in a supremely lean organization like Toyota, and the final steps needed are just now being taken. The translation of lean concepts into the warehouse has required considerable getting used to, for managers as well as employees, and Toyota has had to convince its employees that the new way of thinking would not cause anyone to lose his or her job.
Lean Thinking Page 9