New tricks for old machines: don't run-to-fail, run-to-profit

How to make the transition from run-to-fail to run-to-profit

By Theresa Houck, associate editor

How many companies want to buy more maintenance? None, obviously. Management wants to invest in predictability, cost-effectiveness and on-time delivery. Yet most plants still operate in run-to-fail mode and spend too much on repairs. This, despite the availability and affordability of both basic and advanced technologies that can be part of an effective predictive maintenance (PdM) program (Figure 1), and despite the myriad proven financial benefits of equipment reliability.

So why have so few plants made the move to PdM?

Many top managers simply don’t understand how reliability affects the bottom line, says Tracy Strawn, director of international programs with Marshall Institute in Raleigh, N.C. “Predictive maintenance is a tool for increasing uptime and production and reducing maintenance costs,” he says. “Maintenance personnel have to continuously demonstrate that to management with metrics.”

“Predictive maintenance programs have to be taken in context of the overall asset management life cycle,” says Dick MacDonald, senior vice president of product management with SPL WorldGroup, San Francisco, Calif., and former president and CEO of Synergen, which recently merged with SPL. “Companies need to look at asset management as a core competency — look at how the various assets work together, define the critical elements of the overall system, then tie it into a condition assessment program.”

Some organizations avoid starting a PdM program, or implement their programs incorrectly, because of misperceptions. For example, Strawn says manufacturers often pin their hopes on PdM to solve all their problems. “Predictive maintenance doesn’t fix anything — it simply identifies potential failures early so that appropriate corrective action can be taken. Identifying problems early can ensure maximum reliability and uptime at the lowest costs, but you still have to follow through on the corrective actions to realize that,” he explains.

Another misperception is that the technology itself will improve equipment reliability. “When you look at an asset’s entire maintenance program, PdM only represents about 15%,” says Sandra DiMatteo, director of marketing with Burlington, Ontario-based Ivara. “Most of the maintenance program is based on visual or some other sensory inspection that relies on the tradesperson. Technology simply provides a way to monitor or capture that.”

Then there are the workload and cultural barriers. Setting up a PdM program takes extra effort that can create resistance in the maintenance organization. “Maintenance managers often feel burdened implementing a predictive program,” observes Rich Padula, president of Syclo in Hoffman Estates, Ill. “It’s extra work for them.”

Understand costs of failures
A new definition of failure is emerging that no longer means equipment has stopped running. Plants that are using reliability, rather than maintenance, as the measure of performance define failure as “equipment that isn’t performing at the level at which we need it to perform.”

“We need to look at predictive maintenance not as a maintenance tool, but as a plant optimization tool,” says Keith Mobley, principal with Life Cycle Engineering (LCE), Charleston, S.C. “Your program should be set up to improve the reliability of equipment, the cost of goods sold and the life-cycle cost of the assets. It’s a never-ending journey if you do it right.”

Steelmaker Dofasco in Hamilton, Ontario, is Canada’s second largest steel manufacturer. In the late 1980s, the company embarked on a strategic project to evaluate its maintenance operations. The process that evolved over the following years created a different way of thinking about maintenance that focuses on equipment reliability in the context of “manufacturing process reliability.” This provides predictable, stable operations that allow Dofasco to meet its business objectives for customers.

“The key is to make sure there’s a line of sight to maximizing shareholder value,” explains Ron Thomas, Dofasco’s senior equipment reliability consultant and project manager. “We don’t talk about maintenance anymore, we talk about equipment reliability.”

The company’s equipment maintenance programs for individual assets are the focus of the company’s maintenance function. “We develop technically based equipment maintenance programs that identify the activities we need to maintain the level of capability of our assets,” Thomas explains. “Predictive maintenance is integrated into the asset’s equipment maintenance program. We rely fairly heavily on condition monitoring.”

Dofasco partnered with Ivara to commercialize Dofasco’s business process and the in-house condition monitoring software that Dofasco developed (now marketed by Ivara under the name Ivara EXP). The strategy for equipment reliability has five primary components:

  • Adopt a business process for equipment reliability.
  • Incorporate best practices for performing maintenance.
  • Support that process with enabling technology, such as CMMS software.
  • Develop an implementation approach that respects the change management approach to the process.
  • Sustain it through proactive management.

“Maintenance programs come and go, but a business process defines the way you do business,” Thomas says. “Using the business process is a tremendous integrator of technology and best practices. So a business process includes equipment maintenance, and has different elements, such as work identification, planning, scheduling, execution, follow-up, etc. When we design the equipment maintenance program, we consider predictive technologies.”

Québec Cartier Mining (QCM), based in Port-Cartier, Quebec, also has had success using reliability-driven maintenance solutions. QCM is one of the leading producers of iron ore products in North America. Using Ivara EXP software, the company has added $7 million to the bottom line through increased equipment availability and decreased operating costs.

For example, the QCM plant gets more total performance from its wheel dozers with only half the fleet. The operations costs of large wheel loaders have dropped by 43.4%, while the lifespan of 190-ton off-highway trucks has been extended by more than 60%. In addition, worker efficiency is up by 5.1%, and spare parts inventory values have decreased by almost $10 million.

Industrial gas supplier American Air Liquide uses reliability as a business strategy to gain new business by minimizing costs and maintaining product availability. To assure high quality, the company has 17 national suppliers for maintenance services such as safety valve maintenance, compressor overhauls and electrical maintenance.

Partnering with Rockwell Automation for predictive maintenance services, Air Liquide centralized its PdM program at 107 U.S. sites. Data gathered from lube oil condition monitoring, infrared scanning and vibration analysis programs is stored in a central database accessible through the Web. The goal is to make the information available to both the field reps and management in a timely fashion.

“We developed a closed-loop work process that ensures that our field reps act upon the recommendations from Rockwell’s reports,” said Mark Lawrence, director of maintenance and reliability. “We really want to take action in a timely fashion so that we prevent a breakdown before it occurs. By assuring that each step of the process has clear roles and responsibilities and measuring compliance of each step, we assure ourselves that recommendations do not ‘fall between the cracks.’”

Accept investment
The first obstacle is the cost to get started. Because of the recent recession, many plants delayed spending the money on predictive technology and training, says DiMatteo. “However, the smart companies recognize that they need predictive maintenance even more in tough economic times, to focus on reliability. That will impact the bottom line.”

Energy provider Calpine Corp., San Jose, Calif., did just that. The utility’s business strategy is to expand generating capacity and optimize production capability to reduce the frequency and duration of forced outages, thereby increasing revenue and maximizing profits. In late 2000, they set out to reduce equipment failures and prevent downtime by implementing a PdM program that included condition-monitoring hardware and software from Rockwell Automation.

To date, 35 of Calpine’s 91 plants are enrolled in the program with about 21,000 points of data being collected from 2,100 pieces of equipment. In the first half of 2003, the program helped avoid approximately $1 million in repair costs.

The recession didn’t stop Ford Motor Company’s Dearborn Stamping Plant either. The plant produces the doors for the Ford F-150 series pickup truck. Management charged Process Engineer Jim Jackson with implementing infrared monitoring as part of the PdM program.

Jackson’s research led him to partner with Infrared Solutions, Plymouth, Minn. “Many preventive maintenance programs are paper-intensive, passing paper from one person to another,” he says. “I wanted to get the decision-making process about maintenance needs to the level of the guy doing the inspection. And I wanted more than just a camera.”

The program started in March 2004 and included training for the technicians and getting the infrared cameras, software and other devices in place. Now, the Dearborn Stamping Plant has the highest daily part output of Ford’s large plants, with 10,500 to 11,500 parts per day. The second highest producer is making 6,000 parts per day on similar equipment. Jackson says Dearborn’s high productivity is directly attributable to heightened awareness of the PdM program and use of infrared monitoring.

Quantify payback
Part of getting a return on investment (ROI) means buying predictive technologies only when there’s an actual need. MacDonald stresses the need to determine which assets will offer the most benefits from predictive technology investments. “You don’t just go out and buy predictive analysis software and expect to get huge returns,” says MacDonald, “because that might not be the area of significance to your particular facility. Step back and look at what can improve the longevity of the asset over its full life cycle.”

When it comes to investment in reliability improvement, DiMatteo says typical payback comes in less than one year for a reliability solution, with ROI in the range of five to 12 times within about three years. “Therefore, the most cost-effective way we have found is to first look at their process and their reliability program, and make sure they’re doing the right work at the right time,” she says.

Mobley says payback for total PdM can be even higher. “If you do predictive the way it was intended to be done — as an optimization tool — you can get 100:1 return on your investment, and that includes your recurring costs. I’ve seen that happening for my 40 years of experience.”

At Dofasco, performance analysis focuses not only on results, but also on the business process that produces those results, including work identification, planning, scheduling, execution and follow-up. “That’s set up so it feeds back into performance analysis, so you’re always looking at the performance of the maintained asset against required performance to support business goals,” Thomas says.

Personnel review two types of measurements. The first is results measures, which look at the measure of reliability, or equipment failure rates. “We’re measuring downtime as a contributor to asset availability, both planned downtime and unplanned,” he explains. “We’re also looking at maintenance costs as a contributor to total operating costs. Those are all results measures, often referred to as lagging indicators of performance.”

Dofasco also examines leading indicators, or business process metrics, which look at the individual elements of the business process. These reveal if they’re planning well, scheduling well and executing well. “You don’t manage results, you manage the process,” states Thomas.

At Calpine, one of the challenges Predictive Maintenance Engineering Manager Kevin Nordenstrom faced was getting each of the individual plants on board with the predictive maintenance program. In particular, he found it difficult to demonstrate the value of the PdM program because of difficulties in measuring and documenting the cost savings of a potential event that was prevented and ultimately never occurred.

“Cost avoidance is a difficult concept to communicate,” Nordenstrom says. “It’s a complex calculation, and since you have avoided an event from happening, its potential impact is difficult to measure with tangible numbers.”

Still, the plants have been able to record some impressive examples. In one instance, Nordenstrom and his team used PdM technology tools to identify an impending water pump failure. The problems were traced to a bad bearing on a 2,000-hp motor. Calpine opted to remove the motor and replace the bearing, resulting in minor downtime and minimal repair costs. By taking a proactive approach, Calpine avoided an expensive motor replacement and unplanned downtime costs that would have totaled an estimated $370,000.

Air Liquide has saved about $4.8 million since implementing the PdM program in 2002. Rockwell Automation has helped Air Liquide dramatically reduce travel time for predictive data collection — costs that previously consumed 75% of the program funding. More importantly, hundreds of interventions in the breakdown cycle have avoided downtime — at least 50 major cases have been documents in which PdM interventions saved large amounts of money for the company. In addition, since 2002, more than 750 correction action work orders have been written as a result of the PdM program.

Build the business case
Lack of management support and the resources that come with that support is one of the most common obstacles to implementing an effective PdM program. For plant personnel operating in reactive mode who already feel too busy to do anything else, it’s tough to transition from reactive to predictive mode.

To overcome these hurdles, the maintenance organization needs a PdM champion to develop a business case and present it to management. The business case presents the financial opportunities and the cost assessment for the benefits of the PdM program.

The type of metrics to include in the business case depends on the drivers important to each company’s success. However, every company has some measurements in common.

Sam Hess, director of software development and integration services at Revere, Birmingham, Ala., recommends including costs for an outside consultant to help set up the program. He says metrics to include are statistics on the current maintenance level, which includes lost production time; percentage of work spent on preventive maintenance; percentage of work spent on breakdown maintenance; inventory costs; and spare part costs.

Mobley recommends including front-end costs such as capital investments for predictive technologies and the cost for at least one week of training for technicians to use that technology. “Also, a typical analyst assigned to do predictive maintenance will get 5 to 15 days of training by the vendor as part of their front-end work,” he says. “Certification training for vibration and infrared vendors costs should be included, as well as five days of effective, repeatable data acquisition training for technicians.”

Mobley says other components to include are database development; an asset criticality assessment; continuing education for the technicians and analysts; and calibration, refurbishment and upgrades for new generations of equipment and software.

Plant managers might be startled by the initial start-up costs as well as long-term recurring costs, but if the business case was prepared properly, they should not be surprised. The key is to remember the ROI.
 
Assault resistance to change
Even armed with the knowledge of the financial benefits, many maintenance personnel resist PdM because they prefer the way they’ve always done their job.

“This is a huge obstacle at almost every customer,” reports Ralph DeLisio, business unit manager for integrated condition monitoring with Rockwell Automation, Milwaukee, Wis. “To overcome this, we have to prove we’ve done this before and show them that it’s going to make their jobs easier.”

At many plants, reactive maintenance is viewed as strength. “We did a facility analysis, and they were actually proud of how quickly they would react when something went wrong — yet they were running at 90% downtime,” says DeLisio. “Their company had a culture of being good at reacting to a crisis. They didn’t view that as a problem.”

Getting buy-in is a key to overcoming resistance to change. Ford’s Jackson knows this well. “One of the first obstacles we ran into is realizing that people have to buy into the need for preventive maintenance,” he says. “They have to understand that if you can keep the equipment ready to be used, you’re saving money because if I don’t incur any downtime when I’m trying to make product, I’m making money.”

“When we started our infrared program, we initially trained 12 people,” he says. “As time progressed, we ended up with three guys that were committed to the program. Some people didn’t quite understand what we were trying to do, so they decided to return to their original positions in the plant. It’s a different way of thinking. Many people struggled with the requirements and the level of technology being introduced.”

When it comes to software, technicians who may not be computer savvy tend to be overwhelmed by the many fields, tabs and functions of the technology. According to Marty Osborn, vice president of product strategy for Datastream Systems in Greenville, S.C., vendors are responding by simplifying the user interface, providing more information on one screen to make data entry and analysis easier and faster.
 
Improving morale may be a good way to convince maintenance technicians that PdM will help them. DiMatteo says many of her customers have talked about their improved quality of life as a result of changing the way they collect, consolidate and analyze predictive data to manage asset health.

“They’re not getting calls at 4 a.m. or on the weekends,” she explains. “In the past, people liked to work a lot of overtime, but today’s young engineers want to spend time with their families. So they are looking for innovative ways to be more proactive so that they can have a better quality of life outside of work.”

Think big, start small
When it’s time to implement a PdM program, it’s important to clearly understand your company’s performance and financial goals, equipment and maintenance needs, and long-term internal capabilities. Mobley recommends several key steps:

  • Perform an asset criticality analysis. “Focus your predictive needs on the critical assets that make you money — not on the auxiliary fans, pumps and fuse boxes. The criticality assessment determines what equipment should go with your predictive program,” says Mobley.
  • Perform a simplified failure modes and effects analysis on the critical assets so you understand how they’re going to fail.
  • Select the predictive technologies you need. Look for the specific failure modes of the critical assets.

“If you’re not sure about your goals, start small,” says DeLisio. “Run a pilot program in part of your facility and try a technology, such as infrared, as a baseline. Make sure you have a controlled area you can analyze so you can bring data back to management. Also, make sure any outside company you deal with shows you how they’re going to help you get ROI.”

Even if you do have clearly defined goals, start small. “It’s important to get started system-by-system so you start seeing value quickly,” DiMatteo says. “Build asset-by-asset until you’ve covered the entire plant. Then you don’t have a huge implementation when you go live.”

Padula has seen his share of customers who have tried to do too much too fast, and then became disappointed because their PdM program wasn’t as effective as they had hoped. ”Set up the program so that it’s not one big goal like, ‘We expect to have less downtime.’ Instead, work your way into the goal, like, ‘We want to increase downtime 10% in the next 3 months, 25% in the next 6 months, and 50% in the next 12 months.’ Setting internal expectations appropriately can really make the predictive maintenance program successful.”

The last piece of advice is to just do it. “Sometimes you have to be the maverick,” Ford’s Jackson says. “Now, people from other plants are coming here asking us, ‘What are you doing? How are you doing it? Why did you do it this way? Teach us.’ Good news travels fast.”