JUST-IN-TIME, LEAN PRODUCTION, AND COMPLEMENTARY PARADIGMS:THREE PILLARS OF JUST-IN-TIME AND THE TOYOTA PRODUCTION SYSTEM

THREE PILLARS OF JUST-IN-TIME AND THE TOYOTA PRODUCTION SYSTEM

In broad terms, JIT is a management philosophy that seeks manufacturing excellence with an em- phasis on eliminating waste from all aspects of the production system. At its most basic level, a JIT system produces only what is used or sold, in the needed quantity and at the needed time. Accord- ingly, low levels of work-in-process and finished goods inventory are a prominent feature of JIT. Though it can be said that other production management approaches, such as material requirements planning (MRP), share this same overall goal, JIT differs in that it focuses on inventory reduction not only as an end but as a purposeful means to foster broader improvements in performance. Thus, JIT may be considered a dynamic system, always seeking to achieve still higher levels of performance. In contrast, conventional systems such as MRP or models such as economic order quantity (EOQ) are typically static in that they derive a solution for a given set of conditions (such as lead times or setup costs) and make no consideration for improvement of those conditions.

Another typical element of JIT is the use of a pull method of production coordination, such as the well-known kanban system. In a pull system, production is initiated only to replenish what has been actually used at the next stage of the production system (or sold to the customer). This is a reversal of the push concept, in which production is initiated in anticipation of future demand. Kanban functions as a pull system in that, as materials are used in a downstream stage of the production system, replenishment orders for component materials are relayed to the upstream stages in a pro- gressive cascade upwards. Because actual usage of materials downstream is the only trigger for making more of something upstream, production is initiated only when needed and automatically stopped when the demand ceases. Thus, kanban may be viewed as a decentralized, self-regulating system for controlling production and material flows, even in a complex manufacturing environment with thousands of discrete parts, such as an auto assembly plant. It should be noted, however, that kanban is most appropriate for high-volume, repetitive manufacturing environments and that it is only one means for implementing JIT.

The success of JIT, as well as kanban, is contingent on meeting several prerequisites: (1) smooth- ing of volume and variety; (2) development of a flexible, multiskilled workforce; and (3) implemen- tation of continuous improvement and autonomation. These prerequisites are discussed in the sections below. In addition, achievement of high quality levels is also essential to implement JIT. For this purpose, JIT and TQM efforts should be closely linked with each other, as will be discussed later.

Though some references for further information on JIT are provided in this chapter, no attempt is made to review the thousands of articles and books now available. For more exhaustive reviews of the early JIT literature, the reader is referred to Keller and Kazazi (1993), Golhar and Stamm (1991), and Sohal et al. (1989). Reviews of research focusing on the modeling / design of JIT and kanban systems are provided by Akturk and Erhun (1999) and Groenevelt (1993). General references on Toyota’s JIT system include the first publication in 1977 by Sugimori et al., as well as Ohno (1988), Shingo (1989), Monden (1998), and Fujimoto (1999).

Smoothing of Volume and Variety

Production leveling—in other words, smoothing of volume as well as variety to achieve uniform plant loading—is imperative for JIT implementation. What would happen if the production volume of an item were to fluctuate every day and we blindly pursued a pull policy of withdrawing the needed quantity of parts at the needed time from the preceding process? In that situation, the pre- ceding process must always maintain an inventory and workforce sufficient to meet the highest possible quantity demanded. Consequently, its production volume will exhibit fluctuations larger than those of the downstream process. Since a typical production system involves many sequential pro- cesses, there are many stages for these fluctuations to be transmitted from final assembly backward through to the suppliers. At each stage, the fluctuation is transmitted in amplified form to the previous stage, with the result that upstream processes and suppliers may incur vast inventory and waste. This is the so-called bullwhip effect known in the field of supply chain management.

In order to prevent these undesirable effects, an effort must be made to smooth out the production quantities of each item in the final assembly schedule and then keep that schedule fixed for at least some period of time. Smoothing the production schedule minimizes variation in the quantities of parts needed and enables the preceding processes to produce each part efficiently at a constant speed per hour or per day.

Figure 1 shows a simple illustration of production leveling on a daily basis. Based on demand forecasts and firm customer orders, a three-month production schedule is created and shared with relevant suppliers. Each month the production schedule is revised on a rolling basis to reflect the latest information. The first month of the schedule is then frozen and broken down into a daily production schedule as follows.

Suppose that the monthly production schedule for final assembly indicates demand for 2400 units of product A, 1200 units of product B, and 1200 units of product C. The level daily production schedule is determined by dividing these monthly demand quantities by 20 working days per month, that is, 120 A’s, 60 B’s, and 60 C’s per day, along with the corresponding numbers of parts. Minor adjustment of these quantities may occur on, say, a weekly basis, but alterations must be small (within ±10% empirically) so as to avoid introducing excess fluctuations into the system. Though suppliers use this leveled daily production schedule as a guideline for planning, the actual day-to-day produc- tion quantities are determined by actual demand from downstream stages and coordinated through the kanban system or similar approach. By initiating production only as demand occurs, kanban eliminates the possibility of waste due to excess production while absorbing minor, unavoidable fluctuations on the factory floor.

Next, the daily production schedule is smoothed for product variety through the generation of a production sequence. Rather than all A’s being made in one batch, then all B’s in a following batch, and so on, a mix of items should flow through the system. Because the demand for A’s, B’s, and C’s has a ratio of 2:1:1 in this case, an appropriate mix or sequence for their production would be a cycle of A B A C repeated through the day. For determining sequences in more complex situations, refer to algorithms in Monden (1998) and Vollmann et al. (1997).

Using this kind of mixed model sequencing, the line balance can be kept close to 100 percent even though the standard time or cycle time varies for each product. For example, suppose that the line’s takt time (cycle time to meet demand) is two minutes and the respective standard times of products A, B, and C are two minutes, three minutes, and one minute for a certain process. Then the cycle time for the A B A C sequence is balanced and equal to eight minutes. The individual time differences between products in the sequence can be accommodated by utilizing a space buffer between adjacent processes.

The ultimate goal is to have a production system so responsive that products can be made in a batch size of one and scheduled according to the market’s demand rate. This is also called one-piece flow production. However, to achieve one-piece flow, it is obvious that setup or changeover times must be short for all processes involved. The benchmark of reducing setup times to less than 10 minutes is well known. This is also known as single-minute exchange of dies (SMED) or rapid setup and has a number of associated techniques (e.g., Shingo 1985, 1989). Of these, the most important is to distinguish between internal setup tasks that must be performed while the machine is stopped and external setup tasks that can be done in parallel with continued operation. Once short setup times are realized, production lot size and lead time can be reduced, enabling upstream stages to work in response to the needs of the leveled final assembly schedule.

Development of Flexible, Multiskilled Workforce

Under one-piece flow production, different operations are potentially required for each succeeding item being produced. For it to be effective, workers must have the skills necessary for handling multiple operations properly. Consequently, a prerequisite for JIT is the development of multiskilled workers who can cope with the frequent changes in products and operations seen in a one-piece flow production environment.

In an automobile engine plant, for example, a multiskilled operator is typically responsible for several semiautomatic machine tools. This system is called multiprocess holding. Combining multi- process holding together with layout approaches such as U-shaped lines enables production efficiency to be maintained even when production volume varies. To illustrate, consider a fabrication line con- sisting of 12 machines and 6 operators and suppose that the production quantity is decreased from 100 to 80 per day due to reduced demand. Labor productivity still can be maintained if the operations can be rearranged to be shared by 4 multiskilled operators. This makes sense only with the existence of multiskilled operators capable of handling any of the 12 machines’ operations. U-shaped line layouts can facilitate flexible sharing of the operations by minimizing operator movement between the machines.

Figure 2(a) illustrates the type of changes seen in automobile final assembly lines. They have evolved from long straight lines into U-shaped lines and more recently have been rearranged into many short lines contrived so that each line consists of similar operations. The major purpose of this arrangement is to enable workers to learn multiple skills quickly and efficiently because they can easily learn the other operations within a line once they master one of its operations. After becoming a multioperation worker, he or she can move to another line to become a multifunction worker.

An extreme case of multiskilled workers is found in self-contained cell lines. Figure 2(b) illustrates a self-contained cell line at a Japanese microscope plant, where a single worker moves with his or her workstation from operation to operation. The experienced worker not only assembles and adjusts the entire product him- or herself but also takes responsibility for quality assurance and for production management issues such as scheduling. This facilitates production efficiency for high value-added, low-volume products and also provides workers with intrinsic motivation and satisfaction. At the same time, it is considered an effective means to maintain skills and technologies for a new generation of workers. The concern for preserving worker skills is also illustrated by a common industry practice regarding welding operations: by policy, many companies will automate up to 95% of weld points and reserve the remaining 5% for manual operation. Through the manual welding, skill and knowl-edge are preserved in the organization and can be applied, for example, to programming the robotic welders.

In today’s competitive environment, development of multiskilled operators is universally viewed as a critical issue. Well-organized training programs are seen at virtually every plant in Japan, irrespective of industry. As one means for motivating workers to develop further skills, many plants list workers’ names along with their acquired skills on a prominently placed bulletin board. Another important issue is for workers to acquire creative thinking and problem-solving skills and to feel themselves responsible for quality and productivity improvement. For this purpose, small group improvement activities such as QC circles and PM circles are organized. Workers receive training and develop kaizen capabilities through these activities. Another effective approach is the implementation of worker suggestion systems, wherein financial rewards are paid based on the value of the suggestion.

Not only do these various activities and systems contribute directly to improved performance, they help indirectly to maintain quality and productivity by increasing workforce morale.

Continuous Improvement and Autonomation

As previously mentioned, JIT is a production system that dynamically seeks ever-higher performance levels. Elimination of waste and defects is an important goal in this regard. Indeed, to achieve true just-in-time, all parts must be defect-free, as there is no excess inventory to draw upon for replace- ment. To achieve defect-free production, JIT production systems emphasize continuous improvement and utilize autonomation and other related techniques for assuring quality at the source.

The original meaning of autonomation in JIT / TPS is to stop a machine automatically when abnormal conditions are detected so as not to produce defective products. This is made possible by installing automatic detection and stopping devices in machines or equipment, thus making them capable of operating autonomously. The purpose of autonomation is to keep the machine working always for added value only and to prevent it from producing defects or waste. As discussed below, the idea of autonomation is extended in two ways: poka-yoke (mistake-proofing) and visual control.

Whereas autonomation is the application of detection and stopping devices to the operation of equipment, poka-yoke can be considered as the application of error detection and stopping devices to the activities of workers and the worker–machine interface. It is assumed that human errors are inevitable, so rather than rely on the vigilance of workers, various poka-yoke devices, such as jigs, sensors, and cognitive aids, are built into the work process (cf. Shingo 1986; Nikkan 1989). A simple example of poka-yoke is the use of color coding. Certain colors can be designated for the position and setting of a machine, so that the presence of any other color alerts the worker to an error and the need to take action. By reducing errors and the resulting defects, poka-yoke assumes an important role in assuring quality. In terms of job safety as well, poka-yoke is an indispensable means for protecting workers and preventing accidents in work operations.

Similarly, visual control systems (cf. Nikkan 1995) may also be viewed as an extension of the autonomation concept because they utilize various devices to share information and make abnor-

malities evident at a glance. A simple example is the designation of storage positions for tools, with an outline figure drawn for each tool. Any empty spot indicates the absence of a tool as well as its type. Perhaps the best-known example of visual control systems is the andon, a light board or stand consisting of red, yellow, and green lamps positioned so as to be visible from anywhere in the plant. A flashing red lamp indicates that a machine or workstation is experiencing some abnormal condition such as a breakdown. A green lamp stands for normal operation, and a flashing yellow lamp may indicate a setup operation or an upcoming tool exchange. The important role of the andon board is that it communicates the working status of the plant to everyone. In conjunction with andon, a common practice at many plants is to stop a production line when the red lamp flashes at any of the workstations so that the cause may be investigated and countermeasures taken.

Inventory level itself may also be considered as an important indicator for visual control. Not only is the inventory considered to be a non-value-adding expense, but high levels of inventory provide visible indication that waste and inefficiency may be hidden within the production system. To reveal the cause of the inefficiency, the inventory level is first decreased by removing some of the kanban containers from the system. The weakest part or bottleneck of the production system is then exposed because it will be the first affected by the removal of the inventory buffer. After reinforcing the weakest point, the inventory level is reduced further to expose the next weak point in an ongoing cycle. In this approach, the number of circulating kanbans is a visible measure of inventory level, and adjusting their number is a means of inventory reduction. More generally, this idea is employed as a continuous improvement cycle for improving production and material flows.