Early in my consulting career, working with Kei Abe, I was surprised to hear him make seemingly contradictory recommendations about the organization of maintenance in a small auto parts plant and in a large car assembly plant. In both, the managers were thinking of splitting the maintenance group into smaller teams, each dedicated to a production line. In the parts plant, Kei Abe talked them out of it; in car assembly, he supported it. When I asked him why he just said: “the parts plant is too small.”
This is a translation of Bodo Wiegand’s latest newsletter, about Lean in Germany, followed by my comments:
This week I was with a company that is on its way to implement industry 4.0. All machines were networked. The manager could see from his desk which machines were running and which were not. All data were collected centrally and also shown locally to the machine operator. The trend was easy to see. One third of the machines had a malfunction. With an average OEE of 62%, the machines do not always run.
“As long as we buy new machines, we have to live with this,” was his answer to my question.
But, it was not only the newest, but also the older machines that don’t need to be smeared with oil and dirty, even even while generating chips. Provided on request, the Fire-Fighting-factor reported to us by the maintenance technicians was above 75%. The chief knew exactly: 76.6%. An OEE of 62% and 76.6% Firefighting means in plain language: In this business, there is no stable processes.
But what drives intelligent managers then to link his whole company, only to find that the processes are unstable? With some thought they could have discovered this without networking and invested first in stabilizing the processes. Introducing Industry 4.0 For industry on unstable processes will fail. The crucial question: how I manage to stabilize the processes and avoid unplanned shutdowns?
This is a translation of the bulk of Bodo Wiegand’s latest newsletter, about Lean in Germany, followed by my comments:
At the beginning of this year I was at a company with a high level of Lean in Manufacturing and went into a discussion with the Board about how to go further the realize the full potential .
They did not want to get into the administrative areas, since there the world bosses were allowed to have their say — even though there was real potential there . But life in a matrix organization, as has been frequently noted , is very pleasant. Before doubly protected kingdoms can be torn down, it takes usually a crisis or a new boss. You know my motto: “Give a slave of two masters and he is a free man . ”
Well then, what? We talked about opportunities in Manufacturing and, on our tour of the facilities, spent a bit more time on Maintenance. They were quite proud of the TPM plans they showed me, with a regular preventive maintenance plan, and involvement of production operators in routine maintenance. The whole range of tools set up and implemented was classic. Their pride was a new conveyor system, to which a maintenance technician was dedicated for inspections, routine maintenance, and troubleshooting . It sounded to me like: “With such investments, we must be able to afford this, to avoid the risk of failure .”
In another area , there were identical machines; in the next hall several different presses. With the exception of heat treatment (3 shifts ) all areas were still working in 2 shifts. Of course, Maintenance is an area where you can see what happens inside just by looking from the outside. But my gut was telling me not to scream “wonderful” about the perfect organization. Instead, alarm bells went off and immediately came the question “What does the value stream look like?” Proudly, the manager led me to the team leader room. There hung the map. And I immediately saw the date on which it was drawn. It was three years ago. Well, I expressed my concern: ” Is the new system taken into account ? ”
This is a mistake we encounter often. Value streams change with the actions we perform and should be revised especially after new investments or major changes. Bottlenecks migrate and thereby change the production system. Back in the office, we discussed again his question of why the value stream is important to Maintenance. I told him about Lean Maintenance. He asked “Should Maintenance be organized according to the value stream? – Why? ”
” In the value stream,” I answered, “bottlenecks are detected, critical facilities are identified from a customer perspective and process stability is visible. Priorities given to equipment are the basis for maintenance and spare parts stocking strategies.”
That was too high-level for him.
So – I tried again. Equipment that is the bottleneck or is in close proximity to customers is prioritized because it is important for delivery, and the bottleneck caps the production volume. If the bottleneck stops, so does the whole production system. If the last machine stops, which is important for delivery, the safety stock increases .
Maintenance and stocks ? – The Board id not understand. ” What does Maintenance have to do with working capital ? ”
“Well, safety stocks are usually based on the worst-case interruption time for repairs and mostly with people-related impact to it. ” In this case, it was three weeks.
It is usually two to three weeks – no one knows why.
“Can the maintenance strategy reduce working capital? ”
“Sure,” I answered. ” By prioritizing the facilities you identify the ones that are important for delivery . There you focus your maintenance activities and develop your spare part strategy. This is the only place where it is important whether this system fails. Failure analysis identifies the components that may be responsible. Then individual maintenance strategies must be developed for these components .
This starts with wear-dependent important components that are not predictable with sensor monitoring, and goes as far as the maintenance strategy of “creating redundancy.” The aim is to increase the process stability and to allow no loss. This reduces the need for safety stock . From that we get a feel for what would be the biggest shutdown and can estimate this time .
The next step is to optimize the maintenance time, ie to reduce the repair times to a minimum . If it is possible to organize the maintenance response to a quasi Formula 1 – standard, and you also develop a maintenance strategy adapted to it , you can make the maintenance times as short as possible. The safety stocks can then be lowered furthe . Gut feel no longer prevails. Instead, you have clear maintenance strategies based on numbers , facts, and figures. ”
” But isn’t that more effort? ”
“Perhaps on the facilities with high priorities. But why do you inspect machines that you take out of production for entire shifts? You have many working only 2 shifts. If a machine fails, it can be replaced by others. And why are you dedicating one person to your new conveyor system , which is certainly not a bottleneck? Why do you thoroughly inspect your presses and have not considered how the failure of one could be compensated by the use of another. On such equipment “farms,” you do not need preventive maintenance in the classical sense, only a maintenance strategy that is appropriate for this case. ”
It is important to deliver and therefore you need a stable process. For this, you should evaluate the maintenance person, and not by cost. With a Lean Maintenance approach you will go from failure-driven maintenance to largely planned and predictable maintenance, requiring less effort, providing higher process stability and reducing costs for emergency response .
The result: we have reduced the worst-case repair time from 2.5 days to 8 hours, and safety stocks to two to three days, while reducing the costs of external maintenance services by 80%.
The necessary investments in the sensors, redundancy or spare parts have been more than covered by the reduction in working capital. The annual reductions amount to a low seven-figure sum . The greatest gain was that the production and the maintenance staff are now working towards a common goal and are understood as a team . It culminated in this statement of the initially reluctant maintenance manager : “We want to be measured by the manufacturing productivity and working capital. “
What I read in Wiegand’s words is the focus of improvement in Maintenance should not be on structures and tools but on purpose. We maintain production facilities not to comply with a mandate or fulfill formal requirements but because it allows us to deliver goods to customers without large safety stocks. You might add that, if your products are custom, or even if you just have high variety, there is no way you can hold stocks large enough to deliver promptly.
In most companies, “Lean Maintenance” is taken to mean TPM and, within TPM, the only component that is implemented in the most basic, autonomous maintenance. The headings for the higher levels of TPM include equipment improvement, quality maintenance, and maintenance prevention but, even in Japan, you often hear managers say “We looked into implementing these, but decided they were not worth the cost.”
When you stick with autonomous maintenance, you have an approach to how the work is done but not what it is. This is a whole other topic. Wiegand states as the goal of maintenance to make interruptions of service less frequent and shorter. This is exactly what United Airlines focused on in the late 1960s when the Boeing 747 was introduced, and they called in “Reliability-Centered Maintenance” (RCM).
As part of this effort, they discovered that the “bathtub curve” of failure rates — that staple of reliability textbooks — only applies to about 4% of the aircraft components. In particular, many exhibited no tendency to fail more when aging, which made policies of periodic replacement pointless. They also developed the technique of Failure-Mode-Effect-Analysis (FMEA), on the basis of which they set policies for systematic replacement and spare parts stocks, and selected some items for targeted redundancies.
RCM was later adopted in nuclear power and process industries, and some RCM thinking has found its way into machine-shops, for example in the form of redundant tools in machining center pockets.
The criticism of RCM that I have heard is that it is a workaround to the limitations of the equipment rather than an improvement of it. It is better to have a cutting tool that lasts twice as long than to put a redundant tool on standby in the machine but then, you have to find such a tool.
Wiegand also seems to think that failures are not a problem when you have multiple, interchangeable machines with overcapacity. Technically, that’s unquestionable, but it is another story from the human point of view. It won’t be a problem next week, but what happens over time when overcapacity in an area allows you to have 25% of your equipment down? Your performance will eventually settle at a point where you actually have one machine in four down at any time. Why bother keeping all of them up all the time when they are not needed? Settling for this low availability, however, turns this process into a bottleneck.
See on Scoop.it – lean manufacturing
Mark Graban on LeanBlog: “From my experience, you have to be cautious when somebody says either, “Lean says you should….” or “Toyota would tell you to…” because those statements, even if stated authoritatively, can be wrong.
At a recent speaking engagement (I won’t disclose where), a professor (one who teaches about Lean) made a curious comment that I’d put in the Lean As Misguidedly Explained (orL.A.M.E.) category.
The professor made a point that, when working in healthcare, we have to be careful about applying all methods and tools from Toyota. I agreed with that part of his statement. We’re not literally hanging “andon cords” or putting tape around every piece of equipment just because a factory does it. We have to be solving hospital problems and not just copying tools. I get that.
His example, though, was a bit off base.
The prof talked about “takt time” (or the rate of customer demand) and how we balance the service or production time to match up with takt. Again, that’s correct.
In his hypothetical, he said let’s assume that a doctor’s office is supposed to be seeing a patient every 20 minutes. What if the patient has been in the room for 19:59 already.
The prof said, ‘Toyota would tell you to kick the patient out of the room at 19:59 because you have to keep on takt time.'”
The example is about the hypothetical application of the concept of takt time to a doctor’s office. Mark’s post rebuts a statement that it would imply kicking out a patient at the end of the allotted time regardless of whether the patient’s problem was solved.
This is actually what psychiatrists do when they tell patients “Your 50 minutes are up.” But that is because these patients would otherwise linger on indefinitely. At the opposite end of the spectrum, a general practitioner I know who is an excellent diagnostician once explained that she knew what was wrong with 90% of her patients almost the second they walked into her office, and could confirm it within minutes but stayed longer with each patient just so that they would feel cared for and would trust her diagnosis.
In general, however, I don’t see the concept of takt time as applicable in situations where the work content of transactions varies in a way that cannot be anticipated. An MD can’t know how much time a patient will need. Likewise, a maintenance technician cannot know how a long a work order will require, even in preventive maintenance, because you can’t know exactly what you will find when you open a machine.
Incidentally, takt time is not “the rate of customer demand,” first because it is a time and not a rate, and second, because it is not only a function of demand but also of work time available. It is the time that must elapse between two consecutive unit completions at every operation in order to meet demand within the net available work time. It takes 26 words rather than 5, but the definition really cannot be simplified further.
It is an extremely useful concept to plan repetitive sequences of operations done by different people and machines during a shift. But I don’t see much value in applying it, for example, to people who are on call 24×7 to respond to emergencies, particularly when they do it individually. An MD in an office treats a patient end-to-end; it is in not similar to an assembly line, even if patients sometimes feel that way.
There are other approaches to managing such situations. For example, a takt-based approach to computer networks called “token ring” had its day 30 years ago. A token was passed around between computers in a loop at fixed intervals, and only the computer that had the token was allowed to speak while the others listened.
This takt-based approach was abandoned and superseded by Ethernet, in which computers essentially “grab the mike” whenever they have something to say, with a protocol to resolve confilcts when two or more speak at once. It was a better fit to the way computers communicated and is still the basis for your local office or home network today.
See on www.leanblog.org
Superman, Wonder Woman, or Spider-Man action figures may have captured your attention when you were a kid, but did you ever see action figures for Maintenance Man or Maintenance Woman?
I didn’t think so.
Rarely cast as heroes, members of the maintenance department come to the rescue when a machine is down and it takes superpowers to get it back online. They respond to a crisis with their super-strengths, which include inventiveness, know how, and a wide range of technical skills. Then, like Clark Kent, the maintenance men and women go back to their roles as ordinary characters.
See on www.manufacturingpulse.com
This article describes a method involving initial testing and extensive training used by an Alabama steel mill to increase Maintenance’s share of its work force to almost 30%.
Jim Peck drew my attention to it on NWLEAN through a post in which he questioned their approach to recruitment as training people who didn’t need it or turning down people with the right skills. This kind of information, of course, is not in the article.
The article points out the growing of share of Maintenance in the work people do in a manufacturing operation as it evolves. Based on the numbers in the article, close to one in four employees of the mill works in Maintenance today, and they are trying to increase this ratio. Steel is an industry that has had enormous productivity increases in the past decades. As they point out in the article, they went from 45,000 employees in the 1940s to 2,100 today, who produce as much.
In today’s labor-intensive manufacturing activities, maintenance’s share of the labor force is on the order of 5%, and I believe we can expect that number to rise. For example, an auto plant that employs 5,000 today may produce the same amount with the same depth of manufacturing with 500 people 25 years from now — if cars are still around in 2037… And, out of these 500 people, 150 to 200 will be in Maintenance, the rest being primarily programmers of automatic machines.
Whether testing is appropriate or not depends on the relevance of what people are tested on. An organization has the right to decide what “qualified” means for its own needs. On the other hand, I find testing inappropriate if there is a hidden agenda.
Many Silicon Valley software companies, for example, subject applicants to “coding interviews,” in which they are tested on such topics as the details of sorting algorithms. A computer science student learns this in college but rarely uses it as a professional programmer, because 90% of the time you need to sort records, you just invoke a sort function without worrying about what is under the hood. As a consequence, this kind of test is an effective way to bias the interviews in favor of recent college graduates and filter the experienced programmers.