Our initial objective in creating a value stream “map” identifying every action required to design, order, and make a specific product is to sort these actions into three categories: (1) those which actually create value as perceived by the customer; (2) those which create no value but are currently required by the product development, order filling, or production systems (Type One muda ) and so can’t be eliminated just yet; and (3) those actions which don’t create value as perceived by the customer (Type Two muda ) and so can be eliminated immediately. Once this third set has been removed, the way is clear to go to work on the remaining non-value-creating steps through use of the flow, pull, and perfection techniques described in the chapters ahead.
The Value Stream for a Carton 3 of Cola
The only way to make this method clear is to describe a typical value stream analysis. 4 We’ll use a product chosen more or less at random in the beverages aisle at Tesco, a cardboard carton of eight cans of cola. We should, however, tell you at the outset that what we will find is fairly horrific—a lengthy set of actions extending over three hundred days, most of which consume resources but create no value and are therefore muda. You should understand that looking at any of the thirty thousand other items in the typical Tesco store would produce very much the same result. The cola example is neither better nor worse than the norm.
You should also bear in mind that the firms arrayed along the cola value stream are all competently managed in terms of mass-production thinking. The problem is not the competence of managers operating the system in accord with an agreed logic. The problem is the logic itself.
Producing Cola
Even the mightiest river has modest headwaters. For cola one of these is literally water, supplied in the United Kingdom by the local Water Authorities. Other basic ingredients are the “essence” (in plain language, the taste) used in tiny amounts and supplied as a concentrate by the parent cola company, 5 beets for sugar, corn for caramels (to provide the “cola” color and additional taste), fir trees for cardboard to make the carton, and bauxite or recycled cans to create aluminum for the can. 6
Because the can rather than the actual beverage is by far the most complex aspect of a carton of cola 7 —and the one with the longest production lead time—we’ll initially focus our analysis on the flow of aluminum for the can, treating sugar, caramels, essences, and cartons as tributaries joining the stream farther down the valley.
As shown in the value stream map in Figure 2.1 , the first step is to mine bauxite in Australia. Although the ore could in principle be mined in small amounts and sent along to the next step within a few minutes of the receipt of an order, the mining machinery is truly massive and the actual process involves scooping out millions of tons of bauxite at a go in accord with a long-term production forecast. The mountain of ore is then transferred to massive trucks for shipment to a nearby chemical reduction mill where the bauxite is reduced to powdery alumina.
F IGURE 2.1: V ALUE S TREAM FOR C OLA C ANS
This process, which turns four tons of bauxite into two tons of alumina, requires about thirty minutes. When enough alumina is accumulated to fill an ultralarge ore carrier (over two weeks or so; about 500,000 tons or enough for 10 million cans), it is shipped by sea—a four-week trip—to Norway or Sweden, countries with cheap hydroelectric power, for smelting.
After about a two-month wait at the smelter, the application of an enormous amount of energy (twenty times that needed to melt down and recycle old cans) reduces two tons of alumina to one ton of aluminum in about two hours. Again, scale in smelting dictates that large amounts of aluminum be created in each batch, with the molten aluminum poured into dozens of ingots one meter on each side and ten meters long. These are then carefully cooled and stored for about two weeks before shipment by truck, boat, and truck to a hot rolling mill in Germany or Sweden.
After about two weeks of storage at the hot rolling mill, the ingot is heated to five hundred degrees centigrade and run through a set of heavy rollers three times to reduce the thickness from one meter to three millimeters. The actual rolling process takes about one minute, but the machinery is extremely complex and difficult to change from one specification of product to another, so management has found it best to wait until there are orders in hand for a large amount of material of a given specification and then to process these orders all at once. When this is done for the specification of aluminum needed for cola cans, the aluminum sheet emerging from the rolling mill is wound onto a ten-ton coil and taken to a storage area, where it sits for about four weeks.
When needed for the next step, the coil is taken from storage and shipped by truck to a cold rolling mill, either in Germany or Sweden, where it is stored for about another two weeks. Cold rolling (at 2100 feet of aluminum sheet per minute—about 25 miles an hour) squeezes the aluminum sheet from 3 millimeters to .3 millimeter, the thickness needed by can makers. Because the cold rolling equipment is also extremely expensive and difficult to change over to the next product, the managers of the cold rolling mills have also found it most economical to accumulate orders for products of a given specification and do them all at once. The thin sheet emerging from the cold roller is then slit into narrower widths, wound onto ten-ton coils, and stored for about a month on average.
When needed for can making, the aluminum coils are shipped by truck, by sea, and again by truck to the can maker in England, where the coils are unloaded and stored, again for about two weeks. When needed, the coils are taken from storage to the can making machinery and run through a blanking machine which punches circular discs out of the aluminum sheet at the rate of four thousand per minute. The discs are then fed automatically into “wall drawing” machines, which punch the disc three times in succession to create a can without a top, at the rate of three hundred cans per minute per machine. (Thirteen forming machines are downstream from each blanking machine.)
From the forming machines, the cans travel by conveyor through a washer, a dryer, and a paint booth applying a base coat and then a top coat consisting of the cola color scheme plus consumer information in different languages and varying promotional messages. The cans then travel through lacquering, necking and flanging (to prepare the cans to receive their tops after filling), bottom and inside spraying (to prevent discoloration and any aluminum taste from getting into the cola), and on to final inspection.
The can making machinery just described (really just one big interconnected machine) is a technical marvel capable of converting a sheet of aluminum into a finished, painted can—with no human intervention—in less than ten seconds of actual processing time. However, it is also extremely expensive to change over from one type of can to the next and one paint scheme to the next, so management tries to produce large lots of each type. From the can maker’s standpoint this is clearly the most economical approach, and it also meshes with the practice of the smelter, hot roller, and cold roller of processing specific types of aluminum in large batches.
After inspection, the cans proceed to an automated palletizing machine which loads the empty cans on pallets, eight thousand to each pallet, and sends them to a massive warehouse for storage until needed, usually four weeks. In the warehouse, they are stored by type of can because the bottling firm eventually filling the cola cans needs a variety of cans with different labels for beverages besides plain cola (for example, diet cola, caffeine-free cola, cherry cola). And even for plain cola, the bottler must support many different packaging configurations and promotional campaigns. Each package and many marketing campaigns require different information to be painted on the cans. 8
From the can maker’s warehouse, the cans are trucked to the bottler’s warehouse, where they are stored again, although this time only for about four days. They are then depalletized and loaded into massive can filling machines, where they are washed and filled. It is at this point that the major tributary streams converge in a massive tank adjacent to the filling machine.
In this step, water, caramels
, sugar, and essence are carefully mixed, and carbon dioxide (the fizz) is added to create cola. (Figure 2.2 shows the confluence of the tributaries.) The value streams for these items also require detailed analysis by Tesco, the bottler, and their suppliers, but the method for value stream analysis is best illustrated by sticking to the longest stream.
After the cola is poured into the cans (at the rate of fifteen hundred cans per minute), the cans are sealed with an aluminum can end containing the familiar “pop top,” supplied through a separate but very similar process by the can maker. The cans are then date stamped and packed into cartons of varying numbers of cans, eight in the present case. Each type of carton has its own paint scheme and promotional information.
The mixing and filling process, which brings all of the tributary value streams together, requires only one minute to proceed from washing to packing, but it is expensive and time-consuming to change over. In addition, putting cola in a few cans and then a clear soda in the next can requires purging the whole fill system, so the bottler has found it most economical to run large lots of each type of beverage through its complex equipment. 9
At the end of the filling/packing line, the cartons are palletized, stretch-wrapped (using equipment you will learn a bit more about in Chapter 6 ), and taken to the bottler’s central warehouse serving all customers in the U.K. Storage time for the pallets of cola is about five weeks .
F IGURE 2.2: C ONFLUENCE OF C OLA V ALUE S TREAMS
At the warehouse, the pallets are sorted and placed in designated areas by type. (A process called “stocking.”) They are then “picked” as needed and loaded onto one of the bottler’s trucks for conveying to one of Tesco’s regional distribution warehouses around the U.K.
Once at the Tesco warehouse things move much faster. Incoming pallets are stored for about three days before cases are taken from the pallets and placed in roll cages going overnight to each store. Once at the retail store, the roll cages are taken from the receiving dock to a storage area in the rear or directly to the shelves, and the cola is sold in about two days.
When the cola is taken home it is typically stored again, at least for a few days, perhaps in the basement if the shopper has bought a number of cartons to take advantage of a special promotional offer. Then it’s chilled and, finally, consumed. The last step probably requires about five minutes, after nearly a year along the stream.
A final important step, also shown in Figure 2.1 , is recycling the can to reintroduce it into the production process at the smelting stage. Currently, only 16 percent of aluminum cans in the U.K. are recycled (and shipped back to Norway), but the percentage is rising. If the percentage of cans recycled moved toward 100 percent, interesting possibilities would emerge for the whole value stream. Mini-smelters with integrated mini-rolling mills might be located near the can makers in England, eliminating in a flash most of the time, storage, and distances involved today in the steps above the can maker. (These activities would suddenly convert from type 1 in our typology—muda but unavoidable—to type 2—muda that can be completely eliminated right away.) The slow acceptance of recycling is surely due in part to the failure to analyze costs in the whole system rather than just for the recycling step in isolation.
When laid out this way, action by action, so it’s possible to see every step for a specific product, the value stream for physical production is highly thought-provoking. First, as shown in Table 2.1 , the amount of time when value is actually being created (3 hours) is infinitesimal in relation to the total time (319 days) from bauxite to recycling bin. More than 99 percent of the time the value stream is not flowing at all: the muda of waiting. Second, the can and the aluminum going into it are picked up and put down thirty times. From the customer’s standpoint none of this adds any value: the muda of transport. Similarly, the aluminum and cans are moved through fourteen storage lots and warehouses, many of them vast, and the cans are palletized and unpalletized four times: the muda of inventories and excess processing. Finally, fully 24 percent of the energy-intensive, expensive aluminum coming out of the smelter never makes it to the customer: the muda of defects (causing scrap).
T ABLE 2.1: T HE V ALUE S TREAM OF A C ARTON OF C OLA
The Root Cause of Muda
The simplest way to think about this situation is that a can of cola is very small and cola is consumed by the individual customer in small amounts, yet all of the apparatus used to make cola and get it to the customer is very large, very hard to change over, and designed to operate efficiently at very high speeds. The boats, warehouses, and processing machines we have been describing are truly massive and we can see that the primary objective of technologists in the beverage industry has been to scale up and speed up this equipment while removing direct labor, in a classic application of the ideas of mass production. 10
However, what appears to be efficient to individual companies along the stream—for example, purchase of one of the world’s fastest canning machines, operating at fifteen hundred cans per minute, to yield the world’s lowest fill cost per can—may be far from efficient when indirect labor (for technical support), upstream and downstream inventories, handling charges, and storage costs are included. Indeed, this machine may be much more expensive than a smaller, simpler, slower one able to make just what the next firm down the stream needs (Tesco in this case) and to produce it immediately upon receipt of the order rather than shipping from a large inventory.
For the moment, let’s just reemphasize the critical leap in embracing value stream thinking: Stop looking at aggregated activities and isolated machines—the smelter, the rolling mill, the warehouse, and the can filling machine. Start looking at all the specific actions required to produce specific products to see how they interact with each other. Then start to challenge those actions which singly and in combination don’t actually create or optimize value for the customer.
Ordering Cola
If it takes 319 days to bring a cola from bauxite to Tesco (and a similar amount of time to make most of the other items along Tesco’s aisles), there is a clear problem in ordering. Either orders must be completely uniform over time so the producers all along the stream can operate stable schedules with little inventory, or the upstream producers must maintain large inventories at every stage to deal with shifts in demand, or Tesco’s customers must learn to live with shortages. None of these is desirable because all create muda.
In fact, we encountered Tesco because this firm has made remarkable progress in recent years in streamlining its own ordering system to avoid these choices. It has dramatically reduced “stock-outs” (a situation of not having a product the customer wants) while also slashing its own in-store and warehouse inventories by more than half. Because Tesco was already one of the most efficient grocers in the world when it started this process, it appears that its current inventories are only half the U.K. average, a quarter the European average, and an eighth the North American average.
However, Tesco has recently realized that to move even further in reducing inventories, stock-outs, and costs on a total system basis (where more than 85 percent of the costs of a typical product like cola are outside Tesco’s corporate control), it will need to improve responsiveness and ordering accuracy all the way up its value stream, running across seven firms in this particular case. 11
To understand why Tesco reached this conclusion, let’s look at their current order-taking system, which is probably the most advanced in the world. Tesco installed a Point-of-Sale (POS) bar-code scanning system in the checkout lanes of all of its stores in the mid-1980s. This permitted each store to maintain a “perpetual inventory” of exactly how much of every item it had on hand and to make more accurate orders to suppliers. This was possible because every time a customer in the aisle took a carton of cola past the checkout, the system noted this fact along with the recent rate of sales and the number of cartons remaining. Replenishment orders could be automatically generated.
A few years later, Tesco transferred
decision making on what each store would purchase and when from the store manager, who had been ordering direct from each supplier, to a centralized system where Tesco placed orders combined from all stores to suppliers. At the same time, it opened a dozen Regional Distribution Centers (RDCs) in England so that suppliers for more than 95 percent of all sales volume (the exceptions being milk, sugar, and bread) would ship to the RDC rather than the store. Instead of sending a small truck, partially loaded, to each store, each supplier could send a large truck to each RDC and Tesco could send another large truck to the retail store each night.
In 1989, Tesco took a revolutionary step for the grocery industry by moving toward daily orders (rather than weekly or even monthly) for all fresh products and for many long-shelf-life items. Today, when each store takes inventory at the end of each day, the Tesco ordering system calculates the quantity needed to restore normal stocks plus any special demand likely to be caused by the day of the week, the time of year, the weather, or a sales promotion. After a quick review by the store manger, to check for glitches in the assumptions, this information is dispatched to Tesco’s central computer. There, the requirements from all stores in each region are accumulated and orders are dispatched electronically to each supplier during the night. 12 The suppliers are given a precise time (within fifteen minutes) on a precise day 13 to have the precise amount of goods delivered to a specific receiving dock in each RDC.
When the goods arrive at the RDC, they are taken to an area on the floor assigned to each store and consolidated as a load to be taken that night from the RDC to the store, arriving early in the morning. Thus, orders made by each Tesco store on Monday night result in replenishment goods from suppliers reaching each store before it opens on Wednesday morning, 14 effectively creating a twenty-four-hour continuous replenishment system. (The system is shown in Figure 2.3 .)
Lean Thinking Page 5