The skinny on lean maintenance

To what extent is predictive technology being used to create a lean maintenance department? That question formed the basis of a discussion among readers who participated in the latest Plant Services focus group in April. David Berger, contributing editor and maintenance management expert, moderated the discussion, which centered on participant’s widely varied efforts at reducing costs.

What we discovered is what we expected: that many manufacturers are focused on reducing costs, both in production and maintenance areas. What we didn’t expect was that, although predictive technology is playing a larger role in creating a lean environment, it’s the balance between reactive, preventive and predictive efforts that defines a successful lean maintenance program.

Please scroll down for an introduction of our participants. Here are some excerpts from that discussion:

What does lean thinking, lean manufacturing, lean management or lean maintenance denote to you?

Stan Gorka: Less cost.

Bill Bilbro: It seems to imply efficiency and best use of resources.

Joe Litza: It’s more of a preventive philosophy, using preventive maintenance to cut down on the cost of maintenance programs or time spent on maintenance.

John Trost: It’s basically doing more with less through cost cutting.

If we agree on the basic understanding of what “lean” means, have you investigated a formal maintenance program? Doing something to cut costs, get more for less?

Bill Capizzano: In our company, it’s not formally called lean, but that’s exactly what we do. We utilize computer maintenance management systems, procurement — and preventive-type methods to maximize our efficiencies and minimize our downtimes.

Gorka: We have a regular maintenance program called total commitment maintenance, a world-class maintenance program. There are a lot of different steps toward being lean. Are we there? I doubt it. But if someone opens up a way through constant cost monitoring, that’s part of it.

Ted Sorenson: We’ve done some case and time studies on maintenance issues, such as how to improve time and effort when we tear down a piece of equipment and get it back online in a timely fashion. Because we’re a growing company, we’ve got to study how we can work smarter.

Trost: We have a computerized maintenance management system (CMMS) and we do some predictive maintenance, oil analysis, etc. We still have a lot of room to grow. It’s just been in the last year that we’ve studied returns on it.

Litza: We’re just starting to get a formalized program together, where before everything was pretty much decentralized, and each area took care of its own issues. We are trying to get into more predictive maintenance, and ISO 9000 forces us to keep better logs and written information on our maintenance.

Bilbro: We’re such a small company that our mechanics are so busy that we cannot spare them for that type of activity. So when we get a new piece of equipment, we want to work as closely with the manufacturer as possible, to set up preventive maintenance (PM) schedules and such.

Tom Webb: We don’t have a formal program, but our equipment wears down after 3,000 running hours. We have timers on all the machines and the lasers have to be switched off every so many hours. So when the lasers go down, the technicians go in, and we go back and try to squeeze in the maintenance, such as lubrication, greasing, and try to coordinate it all together.

What percentage of your work is preventive versus predictive, versus reactive maintenance?

Webb: Preventive, we’re always looking for something to make our jobs easier. We’re taking out all our DC drive machines, including shields and filters and all this other stuff, and putting in AC drives machines, which have no filters. They last 15 to 20 years. The DC drives last only ten years. We don’t let anything go to the point where it breaks. It’s $500 an hour just for that machine, not to mention the time for labor. If something breaks, we’ll try to have it repaired. We just sent out a circuit board for repair and the shop told me two days — it’s been seven. We try to keep one of everything on the shelf, too.

Gorka: We do all PMs on a regular basis. We do a lot, maybe 30 to 40% reactive maintenance and mostly preventative maintenance. But it took about eight years to get there. Ten years ago when we started the program, some 90% was reactive and ten percent preventive, so it’s gotten a lot better. But there are still some times when it’s hectic. It subsides a little bit and then gets better. We do a lot of the predictive maintenance, such as changing ball screws or changing drives, but it’s not really where it should be. We’re very busy. It’s very tough to schedule stuff with the production department.

Capizzano: We do a lot of predictive maintenance and we’ve had our CMMS systems for about seven years. It’s really given us incredible rewards. We perform scoping and vibration analysis on all of our blowers and pumps. We also do vibration analysis, oil analysis and thermography on all of our electrical cabinets and roofs. On our air compressors and certain towers, we tear them down and rebuild. We combined the operations and the maintenance and engineering into one work order involving our CMMS. We came out with work order assistants called MORE, which stands for “maintenance operations or request engineering.” Everything goes into the CMMS, is prioritized and basically managed by the maintenance department; but operations, maintenance, engineering all are coordinated out of the same system. We have two different types of what we call PM programs; some that the maintenance department is responsible for, some the operators are responsible for. Since we’ve been recording the last three years, we have reached levels of 90 to 92% preventive versus reactive, which is a complete turn-around for us, considering eight years ago we were probably ten percent preventive and 90% reactive. It’s become a totally self-directed workforce. We used to have a planner, a clerk, three supervisors and a manager. Now we have a manager, an engineer and that’s it. The tracks are totally self-directed, so we’ve cut our workforce down, yet increased the amount of work and reduced our downtime by 60 to 70%.

Sorenson: Probably 70% reactive and 30% preventive maintenance; but we don’t have predictive maintenance. We’ve just now put in a new computerized maintenance management system. It will be helpful.

Trost: We have a CMMS system and just recently went from a DOS version to Windows. There was a little bit of politics involved with the rollout, so it’s been kind of a rocky one. Part of that had to do with centralizing our maintenance department. Mechanics worked for the production supervisors for a year, and then management decided to decentralize the departments again, which took about 15 months. The production supervisors learned from mechanics. The mechanics learned from production. So there was a plus side to it. But between decentralizing and using the DOS version, there was really no way to capture the data and make it useful as an asset management tool. So we opted for a Windows version that would work down at the mechanics’ level, to force them to do their own order entry, their own time card entry, etc. We’re at about 70% preventive scheduled maintenance, as opposed to 30% reactive. We do predictive maintenance, which is the vibration analysis and oil analysis.

Litza: We really don’t have a computer-based system, just a manual log-in system, and I don’t think we’ll be looking at one anytime soon. We’re a high production-type of outfit and we rely on a lot of CNC lathes and mills and on the manufacturers to do the work. We probably do 75% reactive, unfortunately. Like I said, we’re trying to keep a closer eye on more of our problems and try to prevent them. It always comes down to a “let’s try to get it going” approach and then a fever-pitched approach if we have to get the manufacturer involved.

Bilbro: I’m just hearing that scale, with all the maintenance efforts here, and we only have two guys. If I went to them and said, predictive maintenance, they’d say “what?” It’s 90% reactive and we have a manual work order preventive maintenance program to hit the high spots. But it’s two guys, six days a week, two shifts a day. I do a lot of my own laboratory instrument maintenance.

When we talk about lean, there’s a balance that occurs between preventive, predictive, reactive. One of the tricks is to determine the optimum. Some of you are at different stages and moving to a more planned environment, but all of you would agree that the desire is more for a planned environment versus 100% reactive emergency. How are you going to determine just what the percent breakdown should be? How will you make the determination as to what should be reactive and what should be predictive? How do you optimize that balance?

Litza: The biggest factor is the cost of downtime. We have to look at how much downtime a given piece of equipment is experiencing and see how much that affects us in each different area. We can’t say that we’re trying to wipe out all the reactive maintenance programs to get a better number on that.

Bilbro: You’d use past history information to start. And I’d use the manufacturer’s recommendations. I’d split the rest the best I could.
Trost: It would be nice to be able to use histories from a CMMS to get a good feel for it, and it would be nice to have a dollar attached to downtime to see how much it would cost when the line goes down. I guess it would depend on the technology that allows you to get into predictive maintenance, as well as the technical level and experience of your crew.

Gorka: In our plant we have a lot of CNC equipment and I believe that we if could bring that down to less than 100% a year we would be happy. However, in order to do more predictive maintenance, we have to first establish a period of time to implement the maintenance and start determining your history, current situation and where you want to go. The more you could reduce preventive and do more predictive would be helpful. How can you do preventive maintenance without building a history on every machine that you might have in the plant? Each has different problems.

Now consider the cost equation in the balance between reactive, preventive and predictive. If the principles of lean are to look at doing more with less and reducing costs, you have to find where it’s cost effective, and determine the cost of downtime versus the cost of implementing one of those three systems. You have to find that balance. That’s the principle of lean maintenance.

Capizzano: One of the things that a good organization has to do to move from a reactive to proactive environment is to constantly be dynamic. We are constantly evaluating predictive and preventive to see if they’re efficient. We constantly monitor equipment downtime and monitor the cost of resources. You know what your production costs are; therefore, you can evaluate pieces of equipment that give you problems and non-problems and look at how to make them more efficient. We break our line down into components. There are 300 or 400 components on each manufacturing line and each has its own asset number. Each one is monitored, but we also break it down into sections, where you can manage and see how, on any given month, these sections are functioning and what they are costing us as far as breakdowns, quality, efficiency or whatever. Then we evaluate. Did we go too far preventively or predictively? We monitor equipment breakdowns versus the frequency of our preventive repairs on it. Over time, we may go too far, and end up having a failure. So we evaluate: maybe we should go back to the way it was. That’s what we’re trying to do with every piece of equipment. Trying to find that happy medium takes a long time. It involves a lot of people and that’s where the advantage of the CMMS plays out well. You’re allowed to put the day-to-day operations down at the level of the people who evaluate the problem, determine what needs to be done and actually perform the work. The management of the overall picture — I’m talking about the manager, which is where it really needs to be because he’s the guy looking at it, setting frequencies and reviewing the costs such as labor expenses versus purchasing materials. He’s making constant evaluations on a day-to-day basis.

Webb: It’s still a learning experience. When I first started working here 18 years ago, we didn’t do any maintenance at all, we only brought equipment down once a year and changed the oil. I’ve been trying to tell [management] that it’s much harder to fix it after the problem occurs then to fix it before. So we started a database and log and we’ve been going at that for about four years now. We’ve got preventive maintenance. Every day we tear apart something because bearings have to be replaced and oil has to be replaced.

To what degree does instrumentation and control help measure success? How does instrumentation translate from the shop floor environment processes (shop floor control) to the management level?

Trost: We do predictive maintenance with spread spectrum and vibration analysis on our compressors, because our operations are heavily pneumatically controlled. We track that carefully.

Sorenson: We use metal detection devices and some of the scales on weighing things. We try to minimize our biggest problems there, and actually the line will go down, if it’s not working.

Capizzano: Some technology is very expensive and yet, the value is in return on investment. We have some very expensive components. Those have extremely long lead times, and are very costly to spare. Nobody wants to go out and buy a $200,000 piece of equipment just to stick it on a shelf. That’s the kind of equipment we focus on, especially with predictive maintenance, so when you start to see a deterioration of its efficiencies, it raises a red flag.

Bilbro: When we make a package, a food package with a modified atmosphere inside, we test residual oxygen as a measure of how well we gas flush. If there’s too much oxygen left, we haven’t flushed well enough. So the measurement of residual oxygen is a good measure of how well we achieved this. You take a sample from the package and put that through a filter, into a machine that reads the oxygen. If the filter becomes clogged, it can affect the reading and give you a false reading. But you also can feel it manually when we push in. The technicians know, once they start feeling that resistance that pushes in, to stop gassing. So the machine
itself is actually feeding back a signal that keeps itself calibrated.

What are some examples of reactive maintenance? Where do you get your return on investment? Is there something out there that will always be reactive?

Webb: Computerized equipment.

Litza: A lot of our machinery doesn’t really wear out by the time we have to get rid of it. We get rid of it mainly because technology is not up to par. The controls are changing so quickly they really don’t have time to wear one out. We get rid of it because we need to update our equipment.

You might try to predict when the board will no longer work. And that is changing because there are electronic tracers that are now entering the market.

Capizzano: Especially in operations, there’s always new operators that just don’t have the experience on that equipment. Probably a large portion of our routine maintenance is usually caused by operator error. Though we have an extensive training program; we also have high turnover. During the training process, operators make mistakes. A lot of our equipment breakdowns are caused by that.

Gorka: We have a union and they have the ability to post on different jobs, so we constantly have new people on the equipment and, of course, training is extremely important to prevent that.

Let’s move into preventive maintenance versus predictive. When does it become economical to move to predictive?

Sorenson: When you shut down a piece of equipment you’re losing — we have 96 confectionery pans. If we lose a pan, we lose a certain percentage of production. If you’re able to put that pan back online immediately, or by preventing it from happening, you’re saving dollars.

Gorka: When you’ve tested something. We have tested the feeds in the line and found a problem so we had to change it. We spent $30,000 for a new feed, but we prevented a shutdown of a couple days, so that cost could have been in the thousands. So that was a good decision.

When the cost of downtime is so exorbitant, clearly predictive maintenance makes a lot of sense—whether it’s vibration or thermography or whatever, it’s going to make money for you.

Capizzano: We have a tremendous amount of both AC and DC drives. The biggest thing that we focus on is thermography. And vibration analysis helps us study the cause of other equipment problems, because vibration can loosen terminals and loosen lugs. Five or six years ago, that was a big killer for us. That’s why some of our boards shorted out. You can make the investment in the equipment, then you’ve got to train maybe two people to use the equipment. We find that outsourcing a lot of the predictive maintenance is really the only way to go, because these people do it for a living. They’re experts at it. It probably takes us a few years for somebody to look at some of these charts and decipher exactly what he’s telling us. If somebody’s been doing it for five or six, ten years, you take their expertise into account.

Does anyone know what downtime costs?

Trost: They might know in accounting, but I don’t.

Gorka: We track maintenance-caused downtime.

Sorenson: Our extrusion line is about $3,000 an hour. A confectionery pan would be approximately $100 an hour.

Capizzano: Depending on which one of our facilities, it could be anywhere from $300 to $400 an hour — on our embossing line it could be between $400 and $500 an hour, versus our larger galvanizing line, which is probably around $5,000 to $6,000 an hour. So it really depends on each of the lines. We track it for every single line because we need to go back and determine what the downtime cost is.

Litza: We really don’t have any feel for it — we really don’t know how much it’s costing us.

Bilbro: I’m pretty sure they track it in a very fundamental way. Our production line makes changes. It can be difficult to set a standard because as soon as that project is over, there’ll be a new project with different calculation than the original.

What are some of the other ways your maintenance and operations area is focusing on getting lean? Is outsourced maintenance seen as a way to reduce costs?

Litza: On the machinery we’re working with, it’s constantly changing. We can’t always be the greatest people to do the maintenance on it. To be honest with you, we can’t always learn the newest systems that are out there. Who better to rely on than the machine builders or the distributors that do the maintenance of the machines.

Webb: We’ve hired a lot of people to come in to test the oil and the water. Electricians come in on constant basis to do visual work checks. We found we had a circuit breaker panel that had a three-phase motor hooked up to it. We found out that pulse power melted in half and that’s why we were only getting partial power. So it’s better for them to come in and do all that, because we get written warranties. With a contract, if we ever had an accident, we can always go back to them.

Gorka: We still selectively use outside maintenance help. A lot of it is for technical, CNC equipment. Some is for our compressor work.
Sorenson: We do a lot of equipment upgrades and bring in a lot of new equipment. Outsourcing is a cost known to us, and with a written estimate there aren’t any hidden surprises, so management doesn’t look at us like we’re idiots.

Capizzano: Absolutely, especially with predictive technologies and its analysis. You’re
constantly looking at the reward versus the expenditures. We can’t expect our craftsmen to be experts at everything. High tech includes constant technology changes and constant training, so for something you do once or twice a year, it isn’t cost- justifiable.

Litza: If they have the equipment to do it, and the training. When you only do something twice a decade, you’re not going to have a piece of equipment out there on your shelf that costs $10,000 to $20,000.

Bilbro: In some cases, they make it desirable to use an outside service. I’ve got a slicing machine that can be diagnosed over the telephone. All I have to do is pull the telephone wire out and plug it in. They read its condition over the phone and tell us what to fix, right then and there. We also depend upon outsourced training for our people.

What is upper management asking you to do, in terms of reducing your costs? If you’re at capacity and doing effective maintenance, getting product out the door, what is the major driver for that? Is management asking for an equipment reliability improvements, lower spare parts inventory, higher productivity?

ALL: All of the above.

Gorka: Really, the big thing is to make the month’s sales with the best productivity and reduced maintenance time. On-time delivery, and then cost, but if you can deliver faster than your competition, you can get more business.

Capizzano: Three years ago, we took on an economic value-added program. We selected specific drivers to drive that goal. Everybody throughout the organization has the same goals. It’s a good system because in the past, we had the normal goals and objectives that you see with decentralized operations, now everybody’s working cohesively, including the operators, to reach those economic drivers. It’s basically there just to reduce total costs, so that you become more efficient, more competitive in the market.

Sorenson: We did a study and there’s a certain point where we make money. Keeping the plant running at a level of 87 to 90% is our company, our management’s goal — without the maintenance being a factor.

Trost: I think it’s value-added. It comes from the top down. It’s the target that they have for all the divisions in the corporation. It’s the focus of value-added. If it goes out the door on time.