Toyota Kata : Managing People for Improvement, Adaptiveness and Superior Results

by Rother, Mike

  Ideally, every production process would have a target condition. Leaders would be able to check and mentor improvement activity by going from process to process daily, observing, and asking the five questions at each stop. Certainly no process in a production facility should operate without a defined standard that it is striving to achieve. However, it would be overwhelming and infeasible to begin by applying the improvement kata at many processes simultaneously.

  One common answer to the question of where to start is at the loop in the value stream with the greatest potential for improvement. In the simplified value stream map in Figure A1-1, clearly the stamping loop, with its eight days of lead time, has greater improvement potential than the assembly loop, which generates only a half day of lead time. Many of us would logically begin in the stamping loop.

  Figure A1-1. A value stream with two loops

  As part of the research leading to this book, I studied how Toyota works with its suppliers. One thing I observed is that after Toyota supplier support personnel walk a value stream for some time—in order to gain a broad understanding of the overall situation—they usually began by focusing on the assembly loop of a value stream, even if it had far less inventory and lead time than the upstream loops. In the value stream depicted in Figure A1-1, Toyota would most likely begin in the assembly loop. Why?

  In Toyota’s way of thinking, the first place in the value stream to establish and drive toward a target condition is at the “pacemaker process,” rather than at upstream “fabrication” processes. The pacemaker process, or loop, in a value stream is the set of downstream steps that are dedicated to a family of products, and where that family of products is finished for the external customer. The external-customer takt time applies to this process. Often this is an assembly process and its associated scheduling process (Figure A1-2).

  Note that a pacemaker process means something different than a bottleneck process, although they could by coincidence be the same process.

  Toyota tends to begin at the pacemaker loop because it occupies a critical position in a value stream and is worthy of special attention. Fluctuation and instability in the pacemaker loop can quickly affect the just-downstream external customer, and simultaneously cause hard-to-follow, amplifying demand fluctuations for the upstream processes.

  Figure A1-2. Pacemaker and fabrication processes

  I first came across this effect when I visited a plant and was told that the biggest problem was the upstream machining area. The assembly process could often not fulfill its production schedules because it was frequently running out of machined parts. Yet when we got to the machining area, a few calculations revealed excess capacity there. The machining supervisor cleared things up when he said, “Yes, we have enough capacity here, but no one can expect us to keep up with assembly the way they constantly change their production schedule.” At that point we went back to assembly—in the pacemaker loop—and started taking a closer look there.

  Many problems in the upstream processes of a value stream actually have their origin in a poorly operated pacemaker loop. If the pacemaker is operating in a unstable or unleveled manner, it becomes difficult to discern where problems in the value stream are actually coming from. Problem solving and improvement are difficult. Toyota’s improvement strategy here is to strive to develop a stable, leveled pacemaker process first, and then see what problems remain in the upstream processes and migrate there as needed.

  Sometimes, of course, you cannot start at the pacemaker loop because there is a show stopper problem at an upstream process. What Toyota often does in this situation is to fix this upstream problem quickly, within a few weeks at most—even by temporarily increasing inventory there—and then get back to concentrating on the pacemaker.

  A special focus on pacemaker processes, particularly at the start, may take some practice and extra effort to bring it into an organization. At one company I know, the vice president of manufacturing regularly visits the manufacturing facilities; a common practice for manufacturing VPs. Despite having been instructed on the pacemaker focus, plant managers would invariably want to walk the visiting VP through the factory to show “all the improvements we have made”; that is, to show scattered point improvements made at many places in the factory. To change this and get people more focused, it took the VP saying, “For the near future when I visit your plant, I will go to your pacemaker processes first, where I will be asking the five questions.”

  As you continue to focus on the pacemaker process and strive to achieve successively tighter target conditions there, the causes of obstacles will increasingly lie up- or downstream of the pacemaker or even elsewhere in the organization. When conditions in other processes and areas become the obstacles preventing you from achieving the next target condition at the pacemaker, you can migrate to them (Figure A1-3). This is an elegant way to expand into the value stream—following where the problems lead you—because then you are always working on what you need to work on, and individual improvement efforts tie together. Eventually you will be striving for target conditions at all processes, but in a connected and concerted way. And as you move into other processes, the value stream mapping tool will prove helpful for understanding and planning how you would like the flow to tie together next.

  Figure A1-3. Migrating into the value stream and other areas as needed

  Appendix 2 Process Analysis

  The purpose of this appendix is to show you a procedure for analyzing the current condition of a production process. This is done to help obtain the facts and data you need in order to define an appropriate process target condition.

  I have used this process analysis on a wide variety of production processes; some more automated and some less automated. In some cases adjustments will be necessary in order to fit the analysis to the characteristics of a particular type of process, but the basic concept as presented here is usually about the same.

  The purpose of the process analysis is not to uncover problems or potential improvements, but to grasp the current process condition (Figure A2-1) and obtain the facts and data you need for establishing an appropriate next process target condition. This is an important point. This is not a hunt for waste in the process. Going through the steps of this process analysis is intended to force you to look into and confront the details of a process, so you can define how the process should be operating. Once you have a target condition, then you can strive to move toward it, ask the five questions, and identify what you need to work on.

  The process analysis and establishing a target condition take some time, but once a target condition has been established, the coaching cycles can be frequent and short. Try practicing the steps of this process analysis, establishing a target condition and applying the rest of the improvement kata. Once you understand the thinking and pattern behind this process analysis, you may well decide to modify it to better suit your environment.

  Figure A2-1. Process analysis helps you grasp the current condition

  Start with the Value Stream

  Improvement happens at the process level, but conducting at least a “value stream scan” is a prerequisite before conducting a process analysis and establishing a first target condition. Such a scan helps you understand the overall flow from dock to dock and to identify the segments or “loops” of a value stream.1

  A value stream scan often does not take too much time, typically one day or less. Do not try to get all the details, just a basic overview of the value stream by asking the questions below. You can add detail to this value stream map later, as you begin to gain a deeper understanding of the pacemaker process.

  Questions for a Value Stream Scan

  1. Which value stream (product family) have you selected?

  2. What are the processing steps? (Figure A2-2)

  Figure A2-2.

  3. Is the process dedicated (D) or shared (S)? (Figure A2-3)

  Figure A2-3.

  4. At what points along the value stream is inventory kept?
(Figure A2-4)

  Figure A2-4.

  5. How does each process know what to produce (information flow)? (Figure A2-5)

  Figure A2-5.

  6. At what processes are changeovers needed? (Figure A2-6)

  What is the changeover time, current lot size, current number of changeovers per day, and the estimated EPEI at those processes? (Every-Product-Every-Interval: this is the interval of time over which a process produces every high-volume product it makes.)

  Figure A2-6.

  7. What are the “loops” in this value stream? (Figure A2-7)

  Which loop is the pacemaker loop? (See Appendix 1 for an explanation of the pacemaker process or loop.)

  Figure A2-7.

  8. With a one- to two-year time horizon in mind, where:

  Do you think 1×1 flow should be possible?

  Do you think inventory should be replaced with a Pull or FIFO system?

  Now Focus On One Process in the Value Stream

  You are now dropping down from the value stream level to the process level, to conduct the process analysis. Start at the pacemaker loop and stay focused on it. Often this means you will be analyzing an assembly or similar process (Figure A2-8).

  There is a logic behind the order of these steps. However, the effort quickly becomes iterative. As you move through the analysis, you will often have to go back and review or recalculate an earlier step based on what you are learning as you move forward. This is normal. You are trying to get a deep understanding of the current condition.

  Figure A2-8. Start at the pacemaker loop

  Figure A2-9. Steps of process analysis

  The only equipment you need to conduct a process analysis is:

  A stopwatch that measures seconds

  Graph paper

  Pencil

  Eraser

  Calculator

  Do not forget shop-floor courtesy:

  Approach the process via the team leader or supervisor

  Introduce yourself

  Explain what you are doing

  Do not interrupt the operators while they are working

  Explain that you are watching the work, not the operator. (People will not believe you when you say this, but if it is what is in your heart, eventually they will.)

  Show any notes you’ve taken.

  Say “Thank you” before you leave.

  Perhaps keep your hands out of your pockets on the shop floor. People are working hard here, and hands-in-pockets sends a too casual message. A better message is: “We are all working hard for the customer.”

  Assess Customer Demand and Determine Line Pace

  Here are two numbers you should know (Figure A2-10).

  Figure A2-10. Takt time and planned cycle time

  Takt time (TT). This is the rate of customer demand for the group of products produced by a process. Takt time is calculated by dividing the effective operating time of a process by the quantity of items customers require from the process in that operating time. You can see the formula in Figure A2-11, and an example in Figure A2-12.) “Effective operating time” is the available time minus planned downtimes such as lunches, breaks, team meetings, cleanup, and planned maintenance. Unplanned downtime and changeover times are not subtracted, because they are variables we want to reduce.

  Figure A2-11. The takt time calculation

  Example:

  Figure A2-12. Example takt time calculation

  Interpretation of the example: The customer is, on average, currently buying one unit every 58 seconds. (Of course, customer demand rates change over time. For example, Toyota recalculates takt time every 30 days and reviews it every 10 days.)

  Planned cycle time (Pc/t). Once you have calculated takt, then also subtract changeover time and, perhaps, other losses, such as unplanned downtime and scrap and rework rates, from the operating time to arrive at the planned cycle time (Pc/t). This is the actual speed at which the line should be running.

  a.Changeover time. In your first Pc/t calculation you can simply use the number of changeovers currently done per day, and the total time that currently takes. You can also calculate with other patterns of changeovers and changeover times, in order to explore different scenarios.

  b.Downtime. There are two kinds of downtime: short stoppages throughout the day that add up, and rarer but longer-lasting catastrophic failures. In calculating Pc/t, we are concerned with only the small stoppages. You cannot cover for occasional catastrophe with a faster Pc/t.

  Toyota subtracts changeover time in calculating Pc/t, but not unplanned downtime. This is because Toyota factories maintain a time gap after each shift, which is used to make up for small stoppages that occurred during the shift. If you do not currently have that option, then you will probably have to accommodate for some unplanned downtime in calculating the Pc/t.

  One tactic is to strive for a Pc/t that is only 15 or 20 percent faster than takt, and prescribe that changeover time and other losses should be controlled to fit within that 15 or 20 percent gap.

  The following simple capacity analysis using the L-shaped stack chart is an exceptionally useful tool for calculating planned cycle time, which you should master.

  In the stack, show each category of losses individually, rather than, for example, combining them in an OEE figure (Overall Equipment Effectiveness). This way you can better understand the issues.

  Start with a one-day interval to make the Pc/t calculation.

  If you are seeking Pc/t, calculate down. If the Pc/t is fixed, say because of an unchangeable machine cycle, then calculate up.

  Use the optimal changeover sequence to minimize total changeover loss.

  Always put changeover time at the top of the stack.

  Figure A2-13. Capacity analysis

  Figure A2-13 includes an example of using capacity analysis to determine Pc/t.

  First Impressions of the Process

  What do you see?

  Get to know the process by trying to sketch a block diagram of it. Draw a straight-line sketch of the workstations in the process. Do not draw to scale or worry about the shape—the layout—of the line. Simply make each box about the same size as shown in Figure A2-14. Each box equals one workstation or machine. This sketch can get messy as you see deeper and deeper into the process. That’s ok.

  Now observe the process and try to answer the following three questions. Write down your observations. You can ask questions, but do not interview people. Learn to see and understand for yourself.

  Figure A2-14. A block diagram sketch of a process

  Is there a 1×1 flow?

  Do parts move directly from one value-adding step to the next?

  Are each operator’s work steps the same from cycle to cycle?

  Is output consistent at the end of the process?

  Use a stopwatch to time 20 successive cycles at the output end of the process. Select a point and time how often a part comes by this point. Chart the individual times as shown in Figure A2-24. Do not calculate or use averages.

  Check Machine Capacity

  What is meant here by “machines” is automatic equipment that runs even if an operator walks away. A drill press that is operated by a person, for example, is not automatic. A drilling machine that drills by itself after a person unloads and loads it is automatic.

  The questions we are trying to answer with this step of the process analysis are:

  Can the automatic equipment in this process meet the planned cycle time?

  What is the fastest planned cycle time that the automatic equipment can currently support? (This is current process capacity.)

  Theoretically, an automatic machine’s cycle time has to be right at or faster than the planned cycle time. For example, if the planned cycle time for a process is 20 seconds, then the automated machines in the process would need to go through their full cycle in 20 seconds or less. In practice, however, this is not quite correct.

  Every machine has a certain small fluctuation from cycle to cycle. Some
times the time to unload and load the machine varies slightly, or the machine cycle itself varies a small amount. Due to this “personality” of machines, a close-coupled 1×1 flow will not be sustainable if any of the automated machines in it require the full Pc/t interval to complete their cycle. In a 1×1 flow, if one machine goes over the planned cycle time, then this variation can telegraph up- and downstream and disrupt the 1×1 flow.

  For this reason, automatic machines should finish their cycle a little before the planned cycle time is up, at the latest. A guideline—only a guideline—is that the total machine cycle time for any automated equipment in a 1×1 flow should be no longer than about 90 percent of the planned cycle time. This guideline applies only to machines, not operators. Operator work should ideally be filled up to the planned cycle time. Looked at another way, the fastest planned cycle time with which a line is able to consistently run a 1×1 flow is depicted in Figure A2-15. This quotient represents the current capacity limit of a 1×1 flow process.