Seeking the Lean Machine
Intimate knowledge of your process is critical; don't be a "catalog engineer"
By George Koenigsaecker
Lean Investments, LLC
During the mid-1970s, I was working for Deere & Co., and spending a lot of time with a Japanese firm, Yanmar Diesel. After visiting dealers around Japan, touring factories and other facilities for several weeks, we had a meeting with senior management. During this meeting they reviewed their improvement gains for the prior three years.
They had increased their product range by four-fold to grow out of the "oil crisis." This type of growth usually messes up productivity, but they actually doubled enterprise productivity during the same period, and reduced average unit cost by 26%. I was astounded. As it turns out, for the prior three years they had weekend lean-conversion support from three members of Taiichi Ohno's Autonomous Study Group--the folks who "invented lean," which could be more accurately described as The Toyota Business System (TBS). From this time on, I read every book I could find on "Just In Time"--as we called it then--went to every seminar, and visited every plant where managers and workers looked as if they knew what they were doing.
During the early '80s I was with Rockwell International's Automotive Operations, and each quarter--for three years--I led teams to benchmark best manufacturing-enterprise practices worldwide. We quickly began focusing on Japanese firms, and also quickly found that the companies implementing the TBS were truly different. We benchmarked 144 firms--about 20% were part of the Toyota family--and they achieved, on average, four times the enterprise output per person as Japanese and US operations not based upon TBS. They also could make every product every day, had 90% fewer customer complaints, etc. We initiated some early experiments in applying these ideas within Rockwell, but we were just beginners.
In the mid-'80s I became President of Jacobs Vehicle Equipment Company (Jake Brake). Jake was in trouble, and after a decade of benchmarking and book learning, I was ready to try to implement the TBS. The real break came a few months after we started the "journey," when I found out that the guys who had led the conversion at Yanmar Diesel were retiring from Toyota. After a lot of effort, I convinced them to become my sensei and teach Jake how to apply lean. I was the student of these men for the next 15 years. From Jake Brake, I was moved to the role of Group President at the Automotive Group of Jake's parent, Danaher Corp. And very shortly thereafter, I became Group President of the Tool Group, the largest business unit of Danaher. Needless to say the lean enterprise conversion effort followed me around.
In the early '90s, I moved back to the small Iowa town where I grew up, and became president of the HON Company. In 2000, after eight years of getting lean institutionalized at HON, I formed Lean Investments LLC.
Conversion of the operations side of a business typically proceeds in three waves. The first "wave" occurs when you take the lean tools most folks have heard about, and apply them to a business that was designed around the concepts of craft/batch or mass production--but not around the concepts of lean. The vast majority of folks working to apply lean only know of this "wave," and it really amounts to the use of the tools and concepts of lean to improve a system that was designed around the "wrong" principles. This is usually the foundation step that allows an organization to move to other dimensions of lean.
The second wave is when the organization has achieved enough success in operations to apply the same tools and concepts to improve the administrative processes that were, of course, also built around the concepts inherent in craft/batch or mass. A few lean leaders are seriously into this wave, but just a few. And a few more are "testing the waters."
The third wave of operational lean is when you begin to reconfigure your production technology to be consistent with lean concepts and principles. To do so, you typically need several years of lean implementation experience so that these concepts, which are normally "opposite" the way we usually do things, begin to "feel OK." It's hard to invent a new production base until you actually believe in the concepts and principles of lean. In round numbers, there are really only a handful of North American firms that have made it to this stage.
The next stage is building a strong organization. One of the things that makes this stage difficult is that you have to do it yourself. My sensei would deride us for buying equipment--"catalog engineer" was one of the worst things they could say about you. But you find that to do this on your own you need to build up a group of skilled associates who can design and build machines--and more importantly, conceive of different approaches to designing such machines. This means that you may need to add skilled trades personnel, and some key engineering/design folks. These people look like "overhead" in traditional manufacturing thinking, and are usually the first persons fired when business slows.
The first steps to lean machine design come from modifying existing machines. Some of this work is done to incorporate setup reduction changes, some of it to add in poke yoke devices to ensure quality. One step that very few firms consider is to add "hanedashi" devices--simple auto-unload mechanisms--to each machine in the cells they build. Toyota found out long ago that it is hard to automate "load," but very easy and cheap to automate "unload." Therefore, after you build a cell with your inherited equipment, you should modify each machine so that at the end of the cycle it unloads the completed part into some sort of tray at the front of the machine. In "Toyota-Speak" this arrangement is called a chaku-chaku or load-load cell. It seems so simple that most folks don't do it.
Ohno's calculations indicated that this feature alone increases the productivity of a cell by 140%--i.e. output per person is 2.4 times that of the basic cell! If you think about the steps an operator takes to unload parts, the calculation makes sense. But we usually don't slow down enough to think of this waste. Not only does modification for setup reduction, poka yoke and hanedashi build quality and productivity, it also increases the skills of your future machine-design-and-build group. After a few years of improving your cells and building your skill base, you may be ready to take on the serious lean-machine effort.
In your inherited capital base, most of your equipment will not be "right-sized." And you will possess a number of monuments. These are the really big machines designed to be able to handle all the plant's volume. Now, no machine is inherently lean. It's the machine in its application--the cell structure, the volume of parts, etc.--that determines whether the machine is the lowest-cost way to provide flow. But design and construction of monuments is what machine builders do. The monument machine is their paradigm. Monuments are often paint systems, heat-treat systems, cleaning systems, or plating systems.
At Danaher we had gone down the path I describe above with our sensei. At HON, after about three years we started to build up our capability. By the eighth year of lean, we had five machine-design-and-build groups organized within the firm. But that's jumping ahead. Having been on the lean journey before, I was able to start our first machine-design-and-build group with a list of 13 key design practices for a "lean machine." They included things like:
- Design to takt time. A machine was not to be able to make a part in less than 1/2 of takt time. Like many of the easy-to-understand aspects of lean, this one is hard because it is 180º away from normal practice, which is to design a bigger, more capable, and faster machine than we need--"just in case."
- Design-in foolproofing mechanisms on all key characteristics. Pretty easy to understand.
- Design-in low to no setup time. The goal is a "one touch" setup--i.e., it takes one movement by the operator to change from one part to the next. It may take you half a dozen attempts to get to this point--but you need to realize that you can get there, if you keep trying.
- Design your machines to be no wider than a man's shoulders or no wider than 11/2 times the width of the basic part-holding tooling. This approach reduces walk time.
- Design machines with controls and start buttons on the left side of the machine, because a good lean cell runs "counterclockwise." You want to press the start button just as you leave the machine and head to the next one in the cell.
- Design machines so that all services are at the rear. You do not want to disrupt operator flow with machine servicing. And you certainly don't want chip conveyors dropping chips to the front and at the side--this extends walking distance AND causes flow disruption.
- Design machines so that you can see over the top of them. This requirement is part of visual control. Sometimes it's not possible, but it can almost always be done, if you focus on it.
- Design machines at the simplest and least-expensive level that will get THIS job done.
- And, of course, design each machine with hanedashi.
One of the things that is unusual about lean is that every time you reapply the tools and concepts to a given work area you will identify new levels of waste and make new improvements. You will not be substantially "lean" until you have restudied every process--both production and administrative--about six to eight times. At HON, even with someone who knew where we should be going and had a guideline for machine design, we were not able to "go directly there." It took several years of hard lean-conversion experience before folks thought that the stuff might really make sense. At that point, we started to do machine-modification work in addition to the setup reduction that we did early on. Our largest business involved a lot of sheetmetal work, and we had lots and lots of press brakes. So one of our first efforts sought to right-size a press brake.
Most of our brakes were 12 to 16 footers. Our earliest efforts at lean involved putting multiple sets of tooling in really long brakes as a kind of setup reduction. Of course, really long brakes are great barriers to flow. So we tasked ourselves to get "right-sized" press brakes.
Year One: We bought a commercially available six-foot press brake--an appropriate first step.
Year Two: Our newly formed machine-design-and-build group designed a three-foot press brake. A good step. And we were working to incorporate setup reduction concepts (robust design) and hanedashi. But we realized that we made many parts that did not require a machine of this size.
Year Three: We designed a two-foot-long press brake, and began serial production of them. Interesting. But as we looked at it, we realized we had employed expensive, more-complex variable-pressure hydraulics "just in case."
Year Four: We obtained simple hydraulics that only had enough power to make a part in 1/2 takt time, at the fastest.
Year Five: We finally "got it." On reflection, I realized that my guidance had been wrong. I told my design-and-build team that we no longer wanted to design right-sized press brakes. After four years of trying to do so, folks wondered where this effort was going. The new direction was to design self-actuating tooling that met the key design criteria. At that point our creativity exploded, and we found many ways to make tooling work--from air bladders, to hand-cranked presses for really small parts, etc.
Basically it took us eight years to understand the lean-machine concept well enough to consistently produce such equipment. So far it has not been possible to get machine builders to head down this path. Probably because it goes against several fundamental machine concept paradigms that they hold dear.
We traveled the same path on getting to right-sized paint systems. We started with very large systems, usually one per plant. We had several iterations that saw us develop smaller systems, before we really could "get it" and design a paint system to the needs of an individual cell. I have had operations that went from one monument washing system to small units in each cell, and from one monument heat-treat system to small units in each cell. With effort, we have been able to eliminate the need for every monument system that we encountered. I stress: with EFFORT.
Basically, any place you have to put a kanban system in place is a problem. In Toyota's world, kanban systems are a form of waste. They may be necessary waste today due to a monument system, but the goal should be to run without monuments.
Once you get to this point, there is a brainstorming tool that can lead you to breakthrough machine concepts. It's called 3P (for Production, Preparation, Process). Confusion can be caused by the fact that there are two versions of 3P. One is simply a checklist for pre-production planning. The almost-unknown one is a breakthrough approach to developing new machine-design concepts. By year eight, we were using 3P to come up with some breakthrough concepts for processes as well as just "right-sizing" existing concepts.
We were running 15 breakthrough machine-design projects each quarter. Our sensei had been trying to teach this approach at Boeing, but he found that they were not far enough along on the lean journey to accept the lessons. They would not complete a design effort. So he asked HON to invite a Boeing engineer to join each of our 15 machine design projects each quarter. Once they were teamed with six to eight HON folks who had traveled further along the lean journey, and were going to complete the project, the Boeing engineers would stick with it. And once they really tried it, they "got it" and went back to Boeing to form "moonshine" teams there.
It has taken Toyota more than 50 years of hard study to attain their present level of understanding of lean. It probably should not surprise anyone that it took us 10 years to learn these lessons! And from what I've seen, I suspect there are a lot more lessons that we can learn--if we stay on the journey long enough to do so!
This article was first published in the March 2004 edition of Manufacturing Engineering magazine.