Requirements of JIT/Lean

For a factory to operate as a just-in-time/Lean production facility, a number of steps must be taken. It is very important that JIT/Lean implementation be a part of a larger total qual­ity program; otherwise, many interdepartmental roadblocks will crop up as time passes. Like total quality, JIT/Lean re­quires an unwavering commitment from the top because production is more than just the manufacturing department. If these two elements (a total quality program and a commit­ment from the top) are in place, JIT/Lean implementation should be within reach. The following discussion touches on the issues that must be addressed as the implementation progresses.

1. Factory Organization

The JIT/Lean plant is laid out quite differently from that to which most people are accustomed. Most traditional facto­ries are set up according to the processes that are used. For example, there may be a welding shop, a machine shop, a cable assembly area, a printed circuit board assembly area, a soldering area, and so on. Each of these discrete processes may be set up in separate parts of the factory (all machin­ing operations done in the machine shop, all cable assembly done in the common cable and harness area, etc.), no mat­ter which of many products it might be for (refer to Figure 21.9). The JIT/Lean plant attempts to set up the factory by product rather than process. All the necessary processes for a given product should be located together in a single area and laid out in as compact a manner as possible.

The chart at the top of Figure 21.9 represents the old process-oriented traditional factory. Each of the processes has its own territory within the plant. Additionally, an area dedicated to warehousing is used for storage of production materials, work-in-process that is waiting for the next pro­cess, and perhaps finished goods awaiting orders. There is also an area set aside for shipping and receiving. Materials are received, inspected, processed, and sent to the warehouse area. Finished goods are taken from the warehouse or from final assembly, packed, and shipped. The upper illustration in Figure 21.9 maps the movement from the warehouse through the processes and finally to shipping in a traditional factory.

FIgure 21.9 Comparison of Factory Floor Layouts: Traditional Versus JIT/Lean.

The lower illustration in Figure 21.9 represents a JIT/ Lean factory that is set up to manufacture four different products. The warehousing area is gone. This cannot hap­pen overnight, but an objective of JIT/Lean is to eliminate all inventories. The second thing to notice is that the fac­tory is divided into discrete areas dedicated to the different products rather than to the different processes. Each product area is equipped with the processes required for that prod­uct. Parts bins are located right in the work area. These bins may have enough to last from a few hours to a few days or more, depending on the degree of maturity of the JIT/Lean
implementation and the nature of the product and its antici­pated production life.

Mapped out in the upper illustration of Figure 21.9 is a typical work-flow diagram for one product. Parts and ma­terials are pulled from several locations in the warehousing area and moved to a process A workstation. These materi­als may be in kit form (all the parts needed to make one lot of a product). The work instructions call for process A first, followed by process D. If process D is busy when the lot is finished by process A, the lot, now WIP, may be stacked up in a queue at process D or taken back to the warehouse for safekeeping. Eventually, process D will process the kit, and it will be sent to process E, perhaps waiting in queue or in the warehouse. This same sequence is repeated through process B, process C, and process F. From there, it goes to shipping. The diagram does not show any trips back to the warehouse between processes, but that could very well hap­pen after every step. The flow-diagram represents a best-case scenario. (This was done purposely to ensure clarity.)

Now observe the flow in the JIT/Lean factory of Figure 21.9. Product 1 is set up to follow exactly the same processing sequence (from parts bins to process A and then to process D, process E, process B, process C, process F, and shipping). In this case, the parts come straight from the bins located in the work cell, not from the warehouse and not in kit form, which is a waste of effort. The work cell is laid out in a U shape for compactness, to keep all the work cell members close to each other. The WIP flows directly from process to process without a lot of wasted movement. Moreover, because this is a JIT/Lean work cell, there will be small lot sizes, with work pulled through the process sequence by kanban. That means there will be no queue time on the floor or in the warehouse. Cycle time for this product in the JIT/Lean work cell can be ex­pected to be less than half of that for the same product in the factory at the top of Figure 21.9. An 80 to 90% reduc­tion would not be unusual.

Before one can lay out a JIT/Lean factory, the processes required for the product must be known. This is usually not a problem. Typically, the greatest difficulty comes in deter­mining how much of a process is needed. How many min­utes of a process does the product use? One would think that if the product had been built before in the traditional way, one should know how much process time is required at each step. This may be a starting point, but typically it is not very accurate. With all the wasted motion and waiting time in queues and in the warehouse, the real processing time becomes obscured. However, you can use the best informa­tion available and refine it over time. Now that the processes are put right into the product work cell, having just the right amount is important.

In the case of product 4 in Figure 21.9’s JIT/Lean fac­tory, it was determined that the product required more ca­pability in process A and process E than was available from single workstations, so they were doubled. Suppose that a product flow of 120 units per hour is anticipated. Each pro­cess has the following estimated capability for this product:

Because processes A and E are estimated to be capable of only about 60% of the anticipated demand, there is no point in trying to improve them. Rather, the process capability was doubled by putting in parallel equipment and workstations. This is a beginning. We now can watch for excess capacity that can be removed from the work cell or for bottlenecks that require other adjustment. Work cells are coarsely tuned at first, with fine-tuning taking place during the initial runs. Excess capacity should be removed, just as required added capacity must be brought into the work cell. Bottlenecks will be quickly discovered and corrected. From there on, it is a matter of continual improvement to increase efficiency forever.

2. Training, Teams, and Skills

Assuming an existing factory is converted to just-in-time/ Lean, one would assume that the people who had been operating it would be capable of doing it under JIT/Lean. Naturally, many of the skills and much of the training neces­sary for the traditional factory are required under JIT/Lean, but JIT/Lean does require additional training. First, the tran­sition from the traditional way of doing things in a factory to JIT/Lean involves profound changes. It will seem that ev­erything has been turned upside down for a while. People should not be exposed to that kind of change without prepa­ration. It is advisable to provide employees with training about why the change is being made, how JIT/Lean works, what to expect, and how JIT/Lean will affect them. Initial training should be aimed at orientation and familiariza­tion. Detailed training on subjects such as kanban, process improvement, and statistical tools should be provided when they are needed—a sort of just-in-time approach to training.

Most factory workers are accustomed to working indi­vidually. That will change under JIT/Lean, which is designed around teams. A JIT/Lean work cell forms a natural team. The team is responsible for the total product, from the first production process to the shipping dock. Perhaps for the first time the workers will be able to identify with a product, something that they create, and the processes they own. This doesn’t happen in a traditional factory. But with JIT/Lean, it is important to understand that workers must function as a team. Each will have his or her special tasks, but they work together, supporting each other, solving problems, checking work, helping out wherever they can. This may require some coaching and facilitating.

It was enough in the old way of production that work­ers had the skills for their individual processes. They did not need additional skills because they were locked into one pro­cess. This is not the case with JIT/Lean. Specialists are of far less value than generalists. Cross-training is required to de­velop new skills. As a minimum, work cell members should develop skills in all the processes required by their product. Naturally, there are limits to this. We do not propose that all the members of a work cell become electronics techni­cians if their cell employs one for testing the product, but the cross-training should broaden their skills as far as is rea­sonable. Even on the issue of technical skills, it is beneficial to move in that direction. For example, if an operator’s task is to assemble an electronic assembly that will be part of an end-item device, there is no reason that operators couldn’t test it when they complete the assembly. Go/no-go testers can be built to facilitate testing any electronic assembly, and they can be simple enough to operate that the assembler can easily perform the test. This frees the technician for the more complicated tests downstream and ensures that the assembly is working before it is passed on to the next higher level. It also gives operators a sense of ownership and accomplish­ment. Over time, they may even be able to troubleshoot an assembly that fails the test.

Requiring multiple skills in JIT/Lean teams is important for several reasons. First, when a team member is absent, the work cell can still function. Second, problem solving and continual improvement are enhanced by having more than one expert on whatever process is in question. New people will have fresh new ideas. Third, if one of the cell’s processes starts falling behind, another member can augment the pro­cess until it is back on track.

3. Establishing the Flow and Simplifying

Ideally, a new line could be set up as a test case to get the flow established, balance the flow, and generally work out initial problems. In the real world, this may not be feasi­ble. Normally, the new line is set up to produce deliverable goods. What typically happens is a line is set up and then operated with just a few pieces flowing through to verify the line’s parameters. It is very important to maintain strict dis­cipline on the line during pilot runs. Everyone must strictly adhere to procedures. Each operator must stay in his or her assigned work area, with no helping in another process. Only if pilot runs are conducted this way will the informa­tion gained be meaningful and valid. This will allow process times to be checked, wait times to be assessed, bottlenecks to be identified, and workers to become synchronized. It is not necessary to have a pull system in place for these preliminary runs because only a few pieces will be involved. In fact, until the flows have been established, kanban is not possible.

The second thing to look for in these pilot runs is how well the line accommodates the work. Are the workstations positioned for the least motion? Is there sufficient space but not too much? Can the operators communicate easily with each other? Is the setup logical and simple? Can any changes be made to make it better, simpler? Don’t overlook the pro­cesses themselves. Ultimately, that is where most of the sim­plification will occur.

4. Kanban Pull System

Having established the flow and simplified it to the extent possible, the company can now introduce the kanban pull system. As the work cell is being designed, the kanban scheme should be developed. For example, will a single or double kanban card system be used? Or, will kanban squares or bins be used? Or, will some combination or a different variation be used?

You may want to use an electronic kanban system, al­though it might be best to use one of the manual systems initially. After the kinks are worked out the electronic system will be easier to implement. Any kanban plan must be tai­lored to the application; there is no single, best, universally applicable kanban system.

Readers who are familiar with manufacturing may know that cards have been used in the manufacturing process as long as anyone can remember. They take the form of trav­eler tags, job orders, route sheets, and so on, but they are not at all the same as kanbans. These cards push materials and parts into a production process, such as PC-board stuffing. When the boards controlled by the card are all stuffed (the electronic components have been inserted into the boards), the entire batch is pushed to the next process—ready or not, here they come. The next process didn’t ask for them and may not be ready for them—in which case, they will stack up in front of the process or be removed from the production floor and stored with other waiting WIP. By contrast, in a JIT/Lean line, the succeeding process signals the preceding process by kanban that it needs its output. Be sure to under­stand the distinction; with kanban, the succeeding process pulls from the preceding (supplying) process. The kanban always tells the supplying process exactly what it wants and how many. The supplying process is not authorized to make more product until the kanban tells it to do so—nothing waiting, no stored WIP.

Ohno’s double card system uses two types of kanbans: the withdrawal kanban and the production kanban.

• The withdrawal kanban, also called the move kanban, is used to authorize the movement of materials or WIP from one process to another (see Figure 21.10). This kanban will contain information about the item it is au­thorizing for withdrawal, the quantity, the identity of the containers used, and the two processes involved supply­ing and receiving.
• The production kanban authorizes a process to produce an­other lot of one or more pieces as specified by the kanban (see Figure 21.11). This kanban also describes the piece(s) authorized, identifies the materials to be used, designates the producing process workstation, and tells the produc­ing process what to do with it when it is completed.

Consider the operation of two processes in a manufac­turing sequence to see how this works in practice. Figure 21.12 shows a preceding process that does grinding on metal parts. This is the supplier for the parts finishing workstation, the succeeding process. Figure 21.12 shows five segments, described in the following paragraphs:

Segment 1 reveals that the finishing workstation has containers at both its In and Out areas. The container at the In area carries a move kanban (MK) and has one part left to be used. The container at the Out area has five finished parts in it and is waiting for the sixth. Back at the grinding workstation, the Out container is filled with the six parts authorized by the production kanban (PK) attached. The container at the In area is empty, and work is stopped until another production kanban appears.

Segment 2 shows that the finishing workstation has completed work on the six parts, emptying the container at its In area. The empty container with its attached MK for six parts is taken back to the grinding workstation, which is ready to supply the parts.

Segment 3 shows that when the empty container is re­ceived at the grinding workstation, the move kanban is re­moved from the empty container and attached to the full container, which is sitting at the process’s Out area. This au­thorizes movement of the six parts to the finishing worksta­tion. At the same time, the production kanban is removed from the full container and attached to the empty one, which is placed at the grinding workstation’s Out area. This autho­rizes the grinding process to grind six more pieces.

Segment 4 shows that the finishing process has now pro­cessed two parts. The empty container at the In area of the grinding process has been taken back to the preceding process in order to obtain the parts it needs to grind six new pieces.

Segment 5 shows the finishing workstation halfway through its six pieces, with the grinding process started on its next six pieces. This cycle will repeat itself until there is no more demand pull from the right side (from the customers and the final processes).

The finishing workstation had its Out parts pulled by the next process in segment 2, triggering finishing’s pull de­mand on grinding in segment 3. That, in turn, resulted in grinding’s pull from its previous process in segment 4. The pulls flow from the right (customer side), all the way through the production processes to the left (supplier side). When demand stops at the customer side, pulling stops through­out the system and production ceases. Similarly, increase or decrease in demand at the customer side is reflected by auto­matically adjusted pulls throughout the system.

As suggested earlier, it is not always necessary to use ac­tual kanban cards. In many applications, it is necessary only to use kanban squares, kanban shelves, or kanban contain­ers. In Figure 21.12, for example, the two processes could have used any of these devices. The Out side of the grinding workstation could have the right side of its tabletop marked out in six kanban squares. One part ready for finishing would be placed on each square, like checkers on a checkerboard. The signal to grind six more parts would be the finishing workstation’s taking of the parts, leaving the kanban squares empty. In this case, the empty kanban square is the signal to produce more. Marked-off shelf areas, empty containers designated for so many parts, and various other devices can be used. Combinations are the rule.

Kanban is a shop floor control or management system. As such, it has some rules that must be observed: (1) never send forward a defective product, (2) withdraw only what is needed when needed, (3) produce only the exact quantity, (4) smooth production load, (5) adhere to kanban while fine tuning, and (6) stabilize and rationalize.¹⁵ These rules are ex­plained in the remainder of this section.

Instead, stop the process, find out why it was made defec­tive, and eliminate the cause. It will be much easier to find the cause immediately after it happened than it will be after time has elapsed and conditions have changed. Attention to the problem will escalate rapidly as subsequent processes come to a halt, forcing resolution. Only after the problem has been eliminated and the defective part replaced with a good one should the subsequent process be supplied. There can be no withdrawal without a kanban (of some sort). The number of items withdrawn must match the number authorized by the kanban. A kanban must ac­company each item.

Never produce more than authorized by the kanban. Produce in the sequence the kanbans are received (first in, first out).

Production flow should be such that subsequent pro­cesses withdraw from preceding processes in regular intervals and quantities. If production has not been equalized (smoothed), the preceding process will have to have excess capacity (equipment and people) to satisfy the subsequent process. The earlier in the production process, the greater the need for excess capacity. Because excess capacity is waste, it is undesirable. The alterna­tive would be for the processes to “build ahead” in an­ticipation of demand. This is not allowed by rule 3. Load smoothing will make or break the system because it is the only way to avoid these two intolerable alternatives.

In the previous section, we said that for a kanban system to work, the flow must first be established. Kanban cannot respond to major change, but it is a valuable tool for the fine-tuning process. All the pro­duction and transportation instructions dealing with when, how many, where, and so on are designated on the kanban. If the manufacturing process has not been smoothed, one cannot, for example, tell a pre­ceding process to do something early to compen­sate. Instructions on the kanban must be observed. Adhering to the kanban’s instructions while making small, fine-tuning adjustments will help bring about optimum load smoothing.

The processes need to be made capable and stable. Work instructions and methods must be simplified and stan­dardized. All confusion and unreasonableness must be removed from the manufacturing system, or subse­quent processes can never be assured of the availabil­ity of defect-free material when needed, in the quantity needed.

Observing the six rules of kanban all the time is diffi­cult, but it is necessary if the production flow in a JIT/Lean system is to mature and costs are to be reduced.

Kanban is often used by itself for shop floor control very effectively, but it can also be used in conjunction with automation, such as bar code and computer augmentation. Computer-based kanban systems exist that permit the fun­damental kanban system in a paperless environment. As with automation in general, such a computerized system must be designed, or tailored, to suit the application. Applying tech­nology simply for technology’s sake is never a good idea. Whatever you do, it is best to have the system working in its basic manual form before automating; otherwise, you are likely to automate your problems.

The demand pull system has proven itself far more ef­ficient than the traditional push system. If the advantages of just-in-time/Lean are wanted, there is no alternative but to use a pull system, and kanban, in one form or another, is what is needed.

5. Visibility and Visual Control

One of JIT/Lean’s great strengths is that it’s a visual sys­tem. It can be difficult to keep track of what is going on in a traditional factory, with people hustling to and fro storing excess WIP and bringing stored WIP back to the floor for the next stage of processing, caches of buffer WIP all over the place, and the many crisscrossing production routes. The JIT/Lean factory is set up in such a way that confusion is removed from the system. In a JIT/Lean factory, it is easy to tell whether a line is working normally or having a prob­lem. A quick visual scan reveals the presence of bottlenecks or excess capacity. In addition to the obvious signals, such as an idle workstation, JIT/Lean encourages the use of in­formation boards to keep all the workers informed of status, problems, quality, and so on.

Each product work cell or team should have one or more boards, perhaps on easels, perhaps on computer screens, on which they post information. For example, if the schedule anticipates the production of 300 subassemblies for the day, the workers will check off the appropriate number each time a succeeding process pulls subassemblies from its output. This keeps the team apprised of how it is doing and presents the information to managers, who only have to glance at the chart to gauge the work cell activity and its kan­bans to develop a clear picture of how well the line is doing. Another board charts statistical process control data as the samples are taken in the work cell. Anyone can spot develop­ing trends or confirm the well-being of the process with a quick look at the charts. Every time a problem beyond the control of the work cell or an issue with which the work cell needs help comes up, it is jotted down on a board. It stays there until resolved. If it repeats before it is resolved, annota­tions are made in the form of four marks and a slash for a count of five (see Figure 21.13). This keeps the concerns of the work cell in front of the managers and engineers who have the responsibility for resolution. The mark tally also establishes a priority for resolution. The longest mark “bar” gets the highest priority. Maintenance schedules for tools and machines are also posted in plain view, usually right at or on the machine, and normal maintenance activity, such as lubrication, cleaning, and cutter replacement, is assigned to the work cell.

Consider what happens when these charts are used. Information is immediately available to the work cell. The team is empowered to perform maintenance and solve all problems for which it has the capability. With the informa­tion presented to the team in real time, the team solves the problems at once and performs maintenance at appropriate times. This approach minimizes waste, keeps the machines in top shape, and produces a flow of ideas for improvement. The shop floor control loop is as tight as it can get. The op­erator detects and posts the information. The operator reacts to the information to solve problems or take action.

If a problem is beyond the work cell team’s capability, all the people who can bring skills or authority to bear are immediately brought in and presented with the data, and the problem gets solved—quickly. The control loop goes from information to action in one or two steps. In the tra­ditional factory, the operator may not even be aware of a quality problem. It is usually detected by a quality assurance inspector hours or even days after the defect was created. The inspector writes it up. The form may go to the manage­ment information system (MIS) department, where, after a period of time, the data are entered into the computer. Sometime later the computer prints a summary report in­cluding an analysis of quality defects. The report is sent to management through the company mail or via an intranet system. The report may rest in queue for a length of time before being examined. Managers in traditional plants are kept so busy with meetings and firefighting they hardly have time to read their mail, but eventually they will get around to looking at the report. They will see that the line is (or was) having a quality problem and pass the report to the floor supervisor for action. The floor supervisor will at­tempt to see whether the problem still exists. If it does not, case closed. It happened days or weeks ago, and the opera­tor, who up until now was unaware of the defect(s), can’t remember anything that would confirm the problem, let alone suggest a root cause. If the floor supervisor is lucky, the problem may still be there, and the cause may be found. But in the meantime, weeks of production may have been defective.

In this control loop, at least six functions are involved before the loop is closed. That is bad enough, but when the time delay factor is added, finding root causes of problems that come and go is unlikely. Process improvement is much more difficult in this kind of traditional production system. Having had personal experience with both, the authors can attest that the most expensive, most sophisticated computer- based defect analysis system, such as might be employed in the above example, is infinitely inferior to the simple one- or two-step, person-to-person, no-computers-involved control loop of JIT/Lean when it comes to presenting useful infor­mation on a timely basis for the purpose of problem solving and process improvement.

Before our plants changed over to JIT/Lean, a mainframe-based defect analysis system was used. The U.S. Navy designated it as a best practice in the industry. Other companies came to see it, and many of them used it as a model for their own new systems. It could analyze data and present it in many different forms. But it had one flaw: time delay. From the time a process produced a defective part until the loop was closed with the operator of the process, several days (at best) had passed. We are not suggesting that the system was unable to make improvements, because it did. But the real revelation came with implementing JIT/Lean and finding what could happen right inside the work cell when workers had the in­formation they needed while it was fresh and vital and were empowered to do something with it. Immediately, defects dropped dramatically, and they continued to drop as continual improvement was established. Before JIT/Lean, these plants were never able to achieve results remotely comparable, even with their megadollar computer-based system.

Every JIT/Lean line develops its own versions of infor­mation display techniques. But whatever the variation, ev­eryone has valuable, useful information available at all times. That kind of information is extremely difficult to find in a traditional line and most often comes to light in the periodic (weekly or monthly) computer analysis reports. By then, the trail to the root cause may have been obliterated by the passage of time, other problems, or events. In the JIT/Lean factory, real-time visibility lets people know of the problem right then and there, while the cause is obvious. Coupled with the JIT/Lean philosophy that says that the problem must be solved before going any further, this visibility be­comes a driver for elimination of problems and for process improvement.

6. Eliminating Bottlenecks

Richard Schonberger makes the interesting point that only the bottlenecks in a traditional factory forward work to the next process just-in-time.¹⁶ He explains that in a conven­tional manufacturing plant, the bottleneck process is one that goes as fast as it can all the time, barely keeping up with demand. If it breaks down, there is real trouble. To keep it running and to attempt to find ways to increase its output, the bottleneck receives attention out of proportion to the rest of the plant, monopolizing the efforts of engineering and management.

In a JIT/Lean plant, all processes are potential bottle­necks in the sense just discussed because there is little ex­cess capacity and there are no buffer stocks to fall back on when a process or machine shuts down. The upside of this is that all processes are constantly under scrutiny—none is ignored. As Schonberger also points out, the fact that all the processes must be watched carefully makes it imperative that the process operators play a major role in the care and moni­toring and improving of the processes because there cannot be enough engineers to go around when every process is a potential bottleneck.

For this discussion, though, the bottleneck is put into a slightly different frame of reference. We are talking primarily about the setup stage of a JIT/Lean operation when trying to establish a balanced, rational flow through the production system. In this early stage, it is not uncommon to have some real functional bottlenecks. For example, if the new JIT/Lean line is being established to produce as many as 1,000 parts per day, but the manual assembly process can turn out only 800, there is a bottleneck. One way or the other, the process must be brought up to 1,000 or more. If the process employs two people using hand tools, then the answer is simple: add a third person and the appropriate tools. Then the capacity for that process should be 1,200 per day. The extra capacity will have to be accepted until the process can be improved to bring the daily single-operator output up to 500 each, mak­ing it possible to go back to two operators.

Perhaps a machine can produce only 75% of the pro­jected demand. Here the options are a little different. This may be a very expensive machine, too expensive to replicate. Is it possible to put that machine to work somewhere else and put two lower-capacity, less expensive machines in the line, or maybe a single new, higher capacity machine? Can the old machine be modified to increase its output? If setup time is a part of the machine’s normal day, there is a poten­tial for improvement. Another possibility may be adding a second, smaller machine to augment the existing machine’s capacity, although two different machines on the same line making the same part or product is not a desirable solution.

Another kind of bottleneck can exist when a single physi­cal process is shared by two or more JIT/Lean lines. It is pref­erable to make each JIT/Lean product line independent and self-sufficient, but this is not always possible. An example might be a single-wave solder machine servicing two or more JIT/Lean lines. Because of the cost, size, and maintenance re­quirements of such a machine, it may not be feasible to put one in each JIT/Lean product line. Rather, all the JIT/Lean lines take their PC boards to a single-wave solder service cell for soldering. The JIT/Lean lines operate independently of each other. Therefore, it is difficult to predict when conflicts might develop. If they all need servicing at the same time, there is a bottleneck. If soldering delays cannot be accommodated, then one or more of the lines must have its own soldering capability.

Technology can often provide solutions to such prob­lems. For example, 20 high-quality drag soldering machines could be purchased for the price of one wave soldering ma­chine. Production rates of drag solder machines are much lower than those of wave machines, but in many applications, they are ideal for placement right in the JIT/Lean line, dedi­cated to the line’s product and controlled by the line. Such solutions are feasible with many other types of machines.

Whether your bottlenecks appear during the setup phase or during production, the best approach is to assign a cross-functional team to solve the problem. The team should have representation from engineering, manufactur­ing, finance, and any other relevant functional areas. Its job is to list all possibilities for eliminating the bottleneck. This can be done by brainstorming, setting aside those ideas that don’t make sense, and finding the most satisfactory solution in terms of quality, expense, efficiency, and timing.

Frequently, the solution to a bottleneck results in some degree of excess capacity in the process, as occurred earlier when the third operator was added. This is not always bad. Although JIT/Lean always works to achieve more and more efficiency—and, taken to the extreme, would have just ex­actly enough capacity to produce the demanded level and no more—in a practical sense, some excess capacity is desirable. If a line is running at top speed every day, the operators will have no time for problem solving or improvement activities. Some time should be set aside each week for those two items as well as for maintenance and housekeeping. For most ap­plications, 10 to 15% excess capacity is acceptable.

7. Small Lot Sizes and Reduced Setup Times

For a century, industrial engineers have been taught that the larger the production lot size, the greater the benefit from economy of scale. If one wanted to hold down cost of produc­tion, bigger lot sizes were the answer. This was the conven­tional thinking until the JIT/Lean manufacturing bombshell landed on our shores from Japan in the early 1980s. Under the leadership of Toyota and Taiichi Ohno, Japanese manu­facturers concluded that the ideal lot size is not the largest but the smallest. Is it possible that both the manufacturers and the universities could have been wrong all those years? Our conclusion is that the big lot was appropriate as long as mass production systems were used, although they cer­tainly had major problems even then. But once the Toyota Production System came into being, the big lot was not only out of step but impossible to justify.

It stands to reason that if a machine is used to produce different parts that are used in the subsequent processes of production and if the time it takes to change the machine over from one part type to another is six to eight hours, then once the machine is set up for a particular part type, one should make the most of it. It seems to make more sense to run the machine with the same setup for four days, setting up for the next part on the fifth day, than to run one day, spend the next on setup, and so on. The one-day runs result in about 50% utilization time for the machine, assuming a single shift for simplicity. The four-day run yields about 80% utilization.

So what is the problem? If there are four different parts to make on the machine, simply make 20 days’ supply in four days and then go to the next part. By the time production has used all the 20-day supply of the first part, the machine will have cycled back to make that part again. Perhaps a 30- day supply should be made, just in case the machine breaks down. Would a 40-day supply be better? Where does this stop? If we are willing to risk an occasional breakdown, the 20-day cycle is acceptable. A place to store a 20-day supply of not just one part type but four part types will be needed. Then the capability to inventory, retrieve, and transport these parts will also be needed. That represents land, facili­ties, and labor that would not otherwise be needed. None of it adds value to the product, so it is pure waste. It is likely that these costs add up to more than the supposed inefficiency of running the machine with a 50% utilization factor, but these costs are more acceptable to accountants. Land, buildings, and people in motion are not as apparent as examples of waste as machines that are not making product. Traditional thinking says, “Because the machine is busy, people are busy, floor space is full, it can’t be waste.” But it is.

In addition, suppose that a production flaw is found in one of the parts, caused by the machine. Every part made in that lot is suspect. Samples will be tested, and maybe the whole lot will have to be scrapped. This could be 20 days’ supply, representing significant cost. The line will be down until new parts can be made—a major disruption.

Suppose the engineering department corrects a design weakness in one of the parts. Is the entire inventory of parts already made scrapped, or do we use them up in production, knowing that they are not as good as the newly designed part? Either is a bad proposition.

Now assume that the one-day 50% utilization cycle on the machine was employed. The greatest loss we could take would be eight days’ inventory for any of these cases. The eight-day supply can be stored easier than a 20-day supply. This would reduce the cost of warehousing, control, and transportation. Any design changes can be cut-in in eight days. Everything seems positive except the 50% machine utilization.

Ideally, setup time might be reduced to 30 minutes, pro­ducing 1 day’s supply of each part every day. Utilization will be 75% and need for any warehousing may be eliminated. This may seem to be out of reach, but manufacturers using JIT/Lean have done far better, often taking setups from many hours to a few minutes. For example, by 1973 Toyota had reduced the setup time for a 1,000-ton press from four hours to three minutes. Over a five-year period, Yanmar Diesel reduced the setup time for a machining line from over nine hours to just nine minutes.¹⁷ These are not isolated examples.

The general rule seems to be that organizing properly for the setup, making sure the tools and parts that will be needed are in place, and having the right people there at the appointed time will yield an immediate 50% reduction. Then, by analyzing the setup process step by step, a company can usually streamline the process to cut time by half again. Ultimately, the machine itself may be modified to make setup faster and less difficult (e.g., by eliminating the need for adjustment). In any mature JIT/Lean factory, it would be a rare setup that took more than a few minutes, whereas the same setups were previously measured in hours.

The previously supposed advantage of manufactur­ing in big lots completely disappears when setup times are brought down to the kinds of times being discussed here. Machine utilization can be high to satisfy account­ing criteria, and lots can be small to prevent waste and to enable kanban pulling straight from the machine to the next process. Short setup times coupled with kanban have the advantage of flexibility of production. For example, Harley-Davidson used to run its motorcycle line in long production runs of the same model. If a dealer placed an order for a model that had just finished its run, it might have been several weeks before that model could be run again, allowing the order to be filled. Harley was one of the first North American companies to adopt the total quality methods—as a means of survival.

For many years now, Harley has been able to mix models on the production line. It no longer has to pro­duce its product in big lots because it was able to reduce setup times all along its line. Now when an order comes in, it is placed in the queue without regard for the model. Customers get their new bikes as they want them config­ured and far sooner.

Led by Nissan in the United States, auto production lines are beginning to be more flexible as well. Several manu­facturers have lines that accommodate two or more models of similar vehicles. The Nissan plant in Canton, Mississippi, which came on-line in 2003, has the capability to intermix five dissimilar models in lot sizes of one on the same line.¹⁸ Flexibility like that can happen only when model-to-model setup is eliminated or made insignificant. Who benefits? The customer gets more choice, higher quality, and lower cost, and the manufacturer becomes more competitive.

8. Total Productive Maintenance and Housekeeping

This is difficult to comprehend, but many manufacturers spend vast amounts on capital equipment and then ignore the machines until they self-destruct. By contrast, one can find relatively ancient machines in total quality Japanese factories that look like new and run even better. This must become the norm in the United States if U.S. companies are going to compete with the rest of the world. Because a JIT/Lean pro­duction line operates very close to capacity in every process, no tolerance exists for machine failure. When the machine is supposed to be running, it had better be, or the whole line will suffer. Companies that have adopted the Japanese philosophy of total productive maintenance have virtually eliminated machine breakdowns. Machines are cleaned and lubricated frequently, most of that work being done by the operators who run the machines. More technical preventive maintenance routines are performed by experts at frequent intervals. The machines are continually upgraded and modi­fied for closer tolerances, faster setup, and fewer adjustments. Not only do the machines last longer, but also during their entire life span they perform as well or better than when new.

The difficulty with TPM is finding the time in which to perform the maintenance, especially in factories in which three shifts are the norm. The third shift is rare in Japan and Europe, so companies there do not share this problem. Regardless of the workday schedule, it is imperative that maintenance time be provided. The operator-performed maintenance is done during the normal shift (one reason to have a bit more than just enough capacity—a half-hour to an hour a day of excess capacity should more than cover opera­tor maintenance needs).

An added benefit of turning some of the maintenance responsibility over to the operators is that the operators de­velop a sense of ownership for the machines they use and care for. They pay keen attention to the looks, sounds, vibra­tions, and smells of the machines to spot problems before they develop. For the first time, the operators are in a posi­tion to call for maintenance before breakdown occurs. TPM is a must for JIT/Lean production systems.

Housekeeping is another area that is different under JIT/Lean. It is not unusual for the operators themselves to take on the responsibilities formerly associated with janitors. In the better JIT/Lean plants, one will see planned downtime being taken up with cleaning chores—everything spotless, everything in its place. (Remember Five-S from Chapter 15.) It follows that better performance will result from a clean, tidy, and well-organized work area than from one that is dirty and cluttered with tools scattered all over. People like a clean, bright, rational place in which to work. Again, time will have to be made available for this activity.

9. Process Capability, Statistical Process Control, and Continual Improvement

Process capability, statistical process control (SPC), and continual improvement have already been discussed in detail in this book, but it is important to understand the dependence on them by just-in-time/Lean. Is JIT/Lean a necessary pre­requisite for process capability study and improvement, or for SPC, or for continual improvement? The answer is no. At least one of the three is being done in the majority of tra­ditional production plants. Still, there is a connection. The philosophy and discipline of just-in-time/Lean virtually de­mand that they be used in any JIT/Lean environment. While a traditional manufacturing operation may employ one or more of the three, the JIT/Lean manufacturing operation must, and it must be all three. The reason may be obvious to you by now. The JIT/Lean plant is fragile. Everything must work when it is supposed to, and it must work close to per­fection. There are no warehouses of buffer stock to come to the aid of a broken-down process. There is never much ex­cess capacity to help out in tight spots. All the processes with their machines and people must operate in top form all the time.

This is where process capability, SPC, and continual improvement come in. Even before the JIT/Lean line can be certified for full production, the line has to be balanced or rationalized, and a flow has to be established. Unless it is known what the processes are capable of doing in terms of quality and quantity, it will be difficult to achieve the even flow that is a necessary prerequisite of a kanban sys­tem. Without that, there is no JIT/Lean. In the traditional factory, not knowing the capability of the processes is not such a problem; normally, gross overcapacity exists, so parts are stored for the day things go wrong, and the bad parts are sorted out because there will still be good ones to use. In JIT/ Lean, no extra parts can be made, and all have to be good. Workers must have a handle on the processes.

Because one cannot afford (from the time or cost stand­point) to make defective parts, the processes must be in control at all times. The only way to ensure this is through statistical process control. This is not as necessary in a tra­ditional plant, but it is absolutely essential in JIT/Lean. Perfection is difficult to achieve in any circumstance, so it follows that in a complex manufacturing situation, perfec­tion is next to impossible. This is certainly true. We never quite get to the point where all the parts are perfect, but with solid, stable, in-control processes forming the basis of a re­lentless continual improvement program, we can come very, very close. (Some of the very best American plants target and achieve Six Sigma, 3.4 defects per million.) The best that can be achieved is the minimum that is acceptable for a JIT/ Lean factory. In the process of continual improvement, ways are found to do things better, faster, cheaper, and with con­stantly improving quality. The process never ends, and the diminishing-return syndrome doesn’t apply.

10. Suppliers

In the area of suppliers, JIT/Lean has different priorities from the traditional production system. The most obvious difference is the need for frequent, small-lot deliveries of parts, supplies, and materials, rather than the traditional in­frequent, huge-volume deliveries. We are finding more and more JIT/Lean plants in which the suppliers deliver mate­rials directly to the production cells, usually referred to as point-of-use. Several systems have been developed to cue the supplier that it is time to replenish materials. One is the dual-bin kanban system. Two parts bins are used. Bin capac­ity may range from a few hours’ to a couple of weeks’ supply, depending on value, size, usage rate, and intended frequency of replenishment. When the cell has withdrawn all the parts from one bin, the empty bin itself is the signal that it is time to replenish. The supplier routinely checks the bins on the factory floor, and whenever he or she finds a bin empty, it is refilled with the exact number and kind of part designated on the bin label, usually in bar code. The supplier’s bin checking must be scheduled frequently enough to ensure that the sec­ond bin is never exhausted before the first is replenished. In a variation on the dual-bin kanban scheme, the cell’s opera­tors signal the supplier that a bin is empty, either by bar code transmission or by automated electronic purchase order that is triggered by wanding the empty bin’s bar code.

Clearly, for this kind of point-of-use materials delivery system to work, the supplier must be 100% reliable, the ma­terials delivered must be of consistently high quality, and both the supplier and the manufacturing organization must be partners for the long haul. Consequently, choosing the suppli­ers for a JIT/Lean factory is a much more demanding job than it is for a traditional plant. Traditional factories are not so con­cerned with the delivery being on the dock at the precise date on the purchase order. It was going to be stored for a while anyway. Before that lot was used up, there would be another shipment in the warehouse. Neither do traditional factories concern themselves as much with quality from suppliers. The bad parts could always be sorted out, leaving enough good material to keep the line moving. The primary interest was price. Low price got the order. It quickly becomes apparent that this style of purchasing is incompatible with JIT/Lean.

The JIT/Lean plant must have its materials on the dock exactly on the day specified—in many cases at the hour and minute specified—or production may grind to a halt. Every part delivered must be a good part—there is no inventory cache from which to scrounge more parts to keep things moving. This means that the suppliers’ quality must be con­sistently at or above specified requirements. Delivery and quality performance requirements of JIT/Lean effectively rule out buying for price. There is an often used phrase in JIT/Lean and TQM purchasing: “cost versus price.” It sug­gests a holistic approach to the analysis of purchasing on the basis of total cost and value, not simply vendor price. How reliable is a particular vendor in terms of JIT/Lean deliver­ies? What kind of quality can be expected from the vendor? Does the vendor use JIT/Lean, SPC, and continual improve­ment? Are its processes stable and in control? A supplier that gives positive responses in these and other areas may not be the lowest price contender but may well be the lowest cost. Value is what the JIT/Lean purchasing manager must look for, not lowest price on a bid sheet, because in JIT/Lean that turns out not to be the whole story.

When a JIT/Lean factory finds a supplier that deliv­ers excellent materials on time, every time, there is every reason to want to continue to do business with it. More and more companies are turning to supplier partnerships to ce­ment these relationships. What this means is that the two companies agree to work together, not only as supplier and customer but also as unstructured partners. The JIT/Lean manufacturer may, for example, provide training and techni­cal assistance to the supplier to get it started in total quality, JIT/Lean, SPC, and other processes. The JIT/Lean firm may certify the supplier’s quality system to the extent that incom­ing inspections are eliminated, relying on the partner sup­plier to provide acceptable quality in all its deliveries.

The supplier partner may assign one or more employees to take up residence in the JIT/Lean manufacturer’s plant. Duties will include continually checking the kanban bins mentioned above, having them replenished appropriately, and coordinating on-time deliveries of materials, parts, and other supplies provided under the JIT/Lean partner­ship agreement between the manufacturer and supplier. In addition, the resident supplier employee is empowered to do whatever is necessary to solve supply problems before they can cause disruption in the JIT/Lean factory. (While this practice has been around for two decades in the United States, and much longer in Japan, it is now sometimes re­ferred to as JIT II. In the authors’ view, it is simply a logical variation of the materials element of just-in-time/Lean that can work very well in numerous situations.)

The supplier may also be called on to assist in the de­sign phase of a new product, bringing its unique expertise to the design team. Such relationships usually carry a multiyear agreement, so the supplier can count on the business as long as its performance remains high.

There may be preferential bidding treatment—say, an advantage of 10% or more over nonpartnership rivals. Effectively what happens is that the JIT/Lean manufacturer extends its factory right back into the supplier’s premises. They operate to each other’s requirements, and both are locked to each other. The results of this kind of arrangement have been excellent.

This kind of relationship is a far cry from the early ill- conceived attempts of some manufacturers to get into JIT/ Lean before developing a full understanding of the con­cept. In those days, some companies would determine that by using JIT delivery of parts and materials, money could be saved. That part had some merit, but the execution was flawed. The companies simply told their suppliers to deliver a week’s supply of materials once a week, rather than their customary 60 days’ supply every two months. The suppliers’ reaction is easy to imagine. They were being told, in effect, to store the materials in their own warehouses (the capacity for which they didn’t have) and to trickle the deliveries from the warehouses in small quantities weekly. This was simply a case of moving the storage facility from the manufacturer’s plant to the suppliers’. A GM or a Ford has the power to do that to a supplier, but the suppliers, being smaller and with less influence, couldn’t force the same back to their own sup­pliers, so they got caught in an intolerable situation. Only when the suppliers revolted and cried long and loud that this was not JIT—“and by the way, if you want me to store your goods for you, you’re going to pay the tab anyway”—did the would-be JIT/Lean manufacturers see the error of their ways.

The new approach is working well because both parties benefit enormously. If a company wants JIT/Lean, then it must have the best possible suppliers, and both must want to work together for the long haul.

Source: Goetsch David L., Davis Stanley B. (2016), Quality Management for organizational excellence introduction to total Quality, Pearson; 8th edition.