Predictive Maintenance

Guidebook How does PdM generate

operational efficiencies? / p.2

8 Financial Reasons to
Deploy PdM Technology / p.3
Your handbook on how to
build an effective PdM The Predictive
program, from business Maintenance Quiz / p.16

case to technology Is Online Monitoring

innovations. Right for You? / p. 24

Watch Your Assets

from the Cloud / p.35

Special Report © 2015 Azima DLI | All Rights Reserved.

 PREVIOUS PAGE NEXT PAGE

PLAY VIDEO
http://youtu.be/3mZaPS8P830

How does PdM generate operational efficiencies?

Listen in on this exclusive conversation between Thomas Wilk, chief editor
of Plant Services, and Burt Hurlock, CEO of Azima DLI.

Predictive Maintenance Guidebook | 2

8 Financial Reasons to Deploy Predictive
Maintenance Technologies
Leave time-based maintenance and reactive maintenance
for the non-critical equipment

Maintaining equipment on a set schedule has long been an accepted practice for industrial
facilities. If a plant can “fix it before it breaks,” precious uptime is preserved.

Is uptime ALL that matters? Is overhauling a machine just because the schedule says so the
best use of your maintenance dollars and resources?

Today’s reliability experts say “no” to both of these questions.

A growing number of facilities are improving uptime and productivity while simultaneously
reducing maintenance costs. They are doing this by moving away from the concept of
preventative (time-based) maintenance in favor of predictive (just-in-time) maintenance.

By employing non-invasive technologies such as vibration analysis, oil analysis, and IR

surveys, plants are able to determine what specific maintenance is needed and when it is needed.
Pinpointing a specific worn part and replacing it before it fails or triggers broader damage is
far less costly than overhauling an entire machine with no knowledge of what – if anything –
is wrong with it.

Few maintenance professionals will dispute the benefits of predictive maintenance. After all,
who wouldn’t want to know about a fault before it becomes critical, and who wouldn’t want
to plan maintenance around what actually needs repair?

Moving to a predictive maintenance program, however, is not always easy to do. Getting
management to approve the budget needed for routine, condition-based monitoring can be a
tough sell.

8 Financial Reasons to Deploy Predictive Maintenance Technologies | 3

 Here are eight financial reasons your maintenance department can use to justify the move to
management.

• Preventative Maintenance Creates Unnecessary Work

• Unnecessary Maintenance Can Introduce Equipment Failures
• Condition Monitoring Enables Better Maintenance Planning
• Predictive Maintenance Helps Satisfy Agency Regulations
• Labor is a Finite Resource
• Knowing Which Equipment is Critical Drives Down Maintenance Costs
• Monitoring & Analysis Improves Planned Outage Efficiency
• Detecting Problems While Equipment is Under Warranty Saves Money

Reason #1:
Preventative Maintenance Creates Unnecessary Work

Preventative maintenance may very well fix a problem before it happens, but it comes at a
higher cost.

Plants performing time-based maintenance typically spend more in annual maintenance

costs than those using predictive maintenance. Blindly overhauling equipment on a schedule
means you are likely paying for unnecessary repairs on a healthy machine.

Heather DeJesus, director of technical services at Azima DLI (www.azimadli.com) and

a certified vibration analyst, says the most common unnecessary repair is changing out
components after so many hours of service.

“Plants will change out bearings that are in absolutely fine condition simply because they
have been in use for a set period of time,” she says. “That is a waste of maintenance dollars
and resources.”

DeJesus said she’s encountered customers that stop equipment and swap them with their
spares just because the schedule says it is time to do so.

8 Financial Reasons to Deploy Predictive Maintenance Technologies | 4

 “It’s not necessarily a good idea to do this without justification,” she says. “You are tying
up maintenance personnel and usually impacting production. Plus, an electric motor going
through starts and stops induces a lot of electrical failures. Why run that risk without
knowing whether the equipment actually needs to be taken out for maintenance?”

Time-based programs also see their costs rise when equipment problems arise in between
scheduled overhauls.

“That’s when your time-based maintenance program turns into a reactive maintenance
program,” says DeJesus. “Now you’re looking at even higher costs. You may need to pull
maintenance personnel from another area, you may need to pay overtime or a premium for a
vendor to make an emergency call, and you may need to expedite shipping on a part. There are
countless ways the costs grow in this situation.”

DeJesus says the bottom line is that condition monitoring – whether outsourced or a hybrid
– saves money.

“By avoiding the time-based PMs and doing work when PdM analysis tells them to, plants
typically save 16 to 20 times the cost of the monitoring program itself,” she says.

Reason #2:
Unnecessary Maintenance Can Introduce Equipment Failures

DeJesus, who spent time testing Naval helicopters and worked in the power and automotive
industries prior to joining Azima DLI a decade ago, says it is common for failures to be
Case Study introduced simply by doing maintenance.

Read about managing critical “Any time you work on equipment you run the risk of introducing problems,” DeJesus says. “So,
roll-bearing deterioration with a time-based program you may be sending a healthy motor out to be rebuilt just because the
http://info.azimadli.com/managing-
at Orchids Paper schedule says to do it. Then, it comes back with issues it didn’t have before you sent it. All kinds of
critical-roll-bearing-deterioration things can happen. Just the vibration from it being transported can cause problems.”
CLICK HERE

8 Financial Reasons to Deploy Predictive Maintenance Technologies | 5

 This means your plant is not only paying for unnecessary work on a healthy machine, but is
also paying to repair problems resulting from that work.

“Why risk introducing problems if you don’t really need to be working on the equipment in
the first place?” DeJesus asks. “There are technologies that can confirm when and where work
is needed. Why not use them?”

In-house maintenance can also introduce problems, says Dan Hogan, a vibration and
oil analyst at Azima DLI. Hogan’s background is primarily with mechanical drives and
gearboxes, as well as electrohydromechanicals, or mechatronics.

Hogan says over lubrication, use of the wrong oil/grease, and improper belt tension are the
most common mistakes that result in equipment issues.

“Over greasing bearings can have a detrimental effect on bearing life,” he says. “The wrong
lube oil or grease can have similar outcomes. Loose or, more often, too much belt tension can
cause rapid belt failure and chronic bearing failures.”

Reason #3:
Condition Monitoring Enables Better Maintenance Planning

You still have PMs to do, but, by monitoring the condition of your equipment, you can
schedule those PMs with more accuracy.

DeJesus points out that being armed with information about exactly what repairs are
needed enables you to plan the work in the most cost effective manner. Plus, she says, if you
also perform a criticality assessment of your equipment, you can ensure the machines that
impact production the most are worked on first.
 “The bottom line is it saves money and a whole lot of stress and aggravation,” she says.

DeJesus cites one client with a large gearbox from a manufacturer in France. Getting parts
for it from overseas could take as long as five to six weeks.

“We were picking up what looked like gear-tooth chatter in the vibration readings,”
she explains. “Oil analysis verified the vibration findings based on particulate in the oil
sample. We then looked back to our vibration data to pinpoint the exact gear that needed
replacement.”

Because of this information, the replacement was able to be ordered early enough to have it
on hand for the plant’s scheduled outage.

Bill Ruiz, another Azima DLI analyst, is a certified oil analyst and has 17 years of experience
in infrared thermography. He says oil analysis is often used to confirm vibration findings, but
can also stand on its own as a predictive technology.

“We know the importance of oil changes in our own automobiles, and we use mileage
as the oil-change marker,” he explains. “Industry relies on oil analysis to determine the
protection quality and remaining life of the oil. Particle count analysis tells us if the machine is
clean. Metals in oil samples are indicators of component wear. With this information in hand,
maintenance schedulers can plan ahead of future problems for repair or for corrective action
before costly failures.”

Another technology, thermography, when done routinely can eliminate the guesswork with
electrical repairs, Ruiz says.
Case Study
“With routine inspections, a thermographer can quickly detect issues with most any
Read about how IR electrical, mechanical, process equipment, piping and vessels, or any other problems,” says
thermography prevented Ruiz. “It’s fast, noninvasive and safe. Most of the faults that are found in electrical inspections
http://www.plantservices.com/as-
devastating damage are loose wire connections, stress and tension on wire and connection, improperly sized
sets/Media/1403/Azima-DLI-Total-
at Total Petrochemicals equipment or components, and improperly installed equipment. IR will eliminate guesswork
Petrochemicals-IR.pdf
CLICK HERE and accurately pinpoint problems, which can then generate work orders for repair.”
Reason #4:
Predictive Maintenance Helps Satisfy Agency Regulations

Besides improving uptime and the bottom line, predictive maintenance programs are
proving their worth across verticals for other reasons.

Many industries are shifting to condition-based monitoring to comply with the rules and
regulations of outside agencies.

“I see it being used for safety compliance in the oil and gas sector,” says DeJesus. “With new
demands from the American Petroleum Institute and OSHA to prove system integrity, oil and
gas companies are relying on PdM results to properly and effectively maintain their systems to
meet these requirements.”

She says pharmaceutical manufacturers are also starting to embrace the predictive mindset
relative to FDA requirements.

“Pharmaceutical manufacturers are finding that condition-based maintenance satisfies FDA

requirements with respect to the integrity of their process equipment. They are able to meet
these requirements by showing that they are addressing problems before they impact the
process, instead of just doing time-based maintenance,” DeJesus says.

Reason #5:
Labor is a Finite Resource

The truth is, most facilities are lucky to get their PMs done at all.

“The main bottleneck in most maintenance programs, besides the financial budget or
bottom line, is the available labor time,” says Hogan. “Available labor is a finite resource.
You can help to direct your labor most effectively by using multiple technologies, such as IR,
vibration, and oil, to identify the faults, assess the severity, and predict the progression.”

Doing this, he says, arms the maintenance staff with the data they need to go after the assets
that have the most impact on production.
 DeJesus says some of her customers close out preventative maintenance tickets based on the
reports that come from the predictive technologies. This, she says, can help a heavily taxed
maintenance staff keep focused on the areas in which they are really needed.

“One power plant had reduced its maintenance staff from 15 to three, so they only had
time for necessary PMs,” she explains. “From a CMMS standpoint, it looked bad for them to
have a backlog of time-based maintenance work that they just couldn’t get to. By deploying
vibration analysis they proved they didn’t need to do the time-based PMs.”

Hogan says predictive technologies are also effective as an alternative to intrusive PMs, such
as overhauls and rebuilds. For example, he says, oil analysis may allow you to increase oil
change intervals by using data to flag the appropriate end of life. Vibration and IR allow you
to monitor a rotating asset and, in a similar method, trend and track the change in mechanical
condition to identify the best time for overhaul or rebuild.

DeJesus agrees, noting that often maintenance work requires quite a bit of labor, such as
in the case of changing a bearing. Being able to extend the frequency of the work goes a long
way in managing your resources.

“The labor to disassemble a motor and change out the bearing could be half a shift with
multiple people or even a full shift,” says DeJesus. “If you rely on vibration and oil analysis
for bearings, you can postpone changing out a bearing for a month or even a year, depending
on how well the machine is cared for and the quality and care of the bearing. The same goes
for motor current testing. Often the motor is pulled out and sent to a motor shop, and that’s
not necessary. That’s even more expensive.”
Case Study
Hogan says many companies are outsourcing their condition-based programs as fewer and
Read about Conoco Phillips
fewer have dedicated in-house PdM employees. “Outsourcing this type of work is attractive
http://www.plantservices.
Ferndale’s condition-
given the lower upfront costs versus a full in-house program and staff,” he explains. “One
http://info.azimadli.com/conoco-
com/assets/Media/1402/
monitoring success
of the biggest advantages a contracted program has is dynamic scale and controlled costs. A
phillips-ferndale
Azima-DLI-Paper-Mill.pdf contracted, outsourced program has the flexibility to dynamically add and shrink the size or
CLICK HERE
cost of programs, which can help keep costs within budget
 However, he says the most successful PdM programs recognize the value of condition-based
monitoring and ensure their program provides “a full picture of their whole plant.”

DeJesus agrees that consistent, routine monitoring delivers the best results.

“A lot of customers want to know if they can get away with doing vibration twice a year on
a machine that runs all the time,” she says. “Monthly monitoring is predominantly what we
shoot for. We build a baseline to determine what’s normal for the machine. Based on what we
see, we can pick up seasonal changes for outside machines. The more mature your program is,
the more accurate it is.”

She recalls a paper mill that had an induction fan that constantly came out of balance due
to dust buildup.

“A big unbalanced fan can be catastrophic,” DeJesus says. “From the routine vibration
readings, it seemed to happen on a three-month cycle, so they now have a PM that says,
‘Verify vibration readings and clean fan blades.’ ”

Reason #6:
Knowing Which Equipment Is Critical Drives Down Costs

Knowing the impact of each machine is the key to developing a PdM program that keeps your
facility on-line.

A criticality analysis can help a facility prioritize its maintenance work and focus predictive
technologies where they have the most impact. You will spend maintenance dollars more wisely
and simultaneously increase the efficiency of your facility-- all of which drives down overall costs.

“You don’t paint every asset with the same paintbrush,” explains DeJesus. “You look at
safety, production, and compliance. If you deem a machine very important to any of those
areas, then it’s automatically a candidate for predictive maintenance.”

8 Financial Reasons to Deploy Predictive Maintenance Technologies | 10

 She said looking at the purpose of
each machine is the key. The Composite Risk Index
and Why You Should Know
“Maybe a 25-hp machine doesn’t need What They Can’t Tell You
By Burt Hurlock, Azima DLI CEO
to be monitored,” she says. “But a 25-
hp machine running a giant compressor
does. You really have to look beyond the The Composite Risk Index (CRI Report)
machine itself and consider the domino was developed to help our customers
effect from a failure of the machine.” (with bigger problems than cost) see risk.
The report uses machine health data to
DeJesus says she often performs generate a score of site operating risk
criticality assessments to help her and to rank sites by that score. The CRI
customers determine whether a machine Report is automatically available to all
should be on a time-based or condition- Azima DLI multi-site customers, and by
based maintenance program. no coincidence it is the industry leaders
among them that are most likely to use
She says a lot is taken into it. CRI Report users share some common
consideration to make the traits. Leadership and line management
determination. For example, in some tend to be aligned through clear command
cases equipment uptime is tied to chains and communications channels.
contractual obligations in which there There’s a premium on high-quality
are stiff penalties for failing to deliver on information and a shared interest in the
time. Others have strict environmental rapid adoption of best practices. Perhaps
regulations that must be met to avoid most interestingly we see organizational
being shut down and/or fined. humility, an acknowledgment by line
managers that they can’t always
Hogan points out that redundancy makes see what management needs to know, the
a machine less critical. This, too, is a factor same way battalion commanders can see
to be considered in the assessment. the battlefield but not the front. The CRI
Report is the first in a family of enterprise
“The power generation industry performance metrics Azima DLI will
is the standard in this regard,” says deliver to improve visibility of production
Hogan. “Most plants double up or have site performance and expose more best
redundant support-equipment assets practices sooner across the fleet.
on every function to ensure there’s aa

8 Financial Reasons to Deploy Predictive Maintenance Technologies | 11

always one available asset in case of
an unknown or unplanned failure. Why customers have the best ideas
Identification of critical assets requires and the genesis of risk reporting
evaluation and examination of the whole When customers spend money we assume
production process or the experience and they need to justify the investment, and the
knowledge of a site engineer to pinpoint inclination is to present measures of return-
these critical and important machines.” on-investment (ROI). The problem with ROIs is
that they can be subjective and are too easily
All of this is taken into consideration critiqued for being self-serving. They also defy
when a Compound Risk Index (CRI) is standardization because some companies
determined for each machine, DeJesus says. place more value on some variables than
others, which means the same formula can
produce very different pictures of return. Even
Reason #7: audiences within the same customer differ on
the relevance of performance measures.
Monitoring & Analysis Improves
Planned Outage Efficiency When a long-standing customer told us
we didn’t charge enough to merit an ROI
When you have a scheduled outage calculation and he had larger problems than
approaching, information from your cost, he had our full attention. “Risk,” he said,
predictive technologies can help you “is my problem. I know it’s out there and I can’t
determine which machines to pull for see it. Can you show me risk?” It was a seminal
repair. moment in the evolution of our organization
because (1) the request was modest and
Repairing a machine that truly needs easily accommodated and (2) the relationship
it at a time when there is no impact on acquired a new dimension, becoming strategic,
production avoids the high cost of it as well as economic. By focusing exclusively
failing when the plant is back online. on the economic equation, we overlooked
what the customer wanted and underestimated
“It’s a common practice to use your the value of available information he needed.
PdM program to queue and schedule Customers always have the best ideas because
work based on the findings of vibration, they frame the vendor’s role in the context of
oil, and IR prior to an outage,” says their needs faster than can the vendor. And
Hogan. “I just had a power customer what matters to one customer often matters to
request an expedited review of data another, as we have found.
collected to provide this ability for the

8 Financial Reasons to Deploy Predictive Maintenance Technologies | 12

 program. An outage or shutdown is the
best time to do the bulk of preventive The disruptive power of visibility
work or work impossible to perform To speak of risk and presume to manage it
when a site is in production.” feels, well, risky. While risk can be expressed
numerically, it remains an abstraction, and
Faults or problems identified over risk scores rarely capture the full spectrum
multiple testing intervals and trending of risk factors. But, unlike ROIs, risk can be
toward an increasing severity are the depicted, both in an absolute sense, subject
prime candidates for repair during an to the limitations of the risk model, and, more
outage, advises Hogan. Conditions importantly, in a relative sense. The power of
severe enough to warrant immediate the latter cannot be underestimated because
inspection or suspected conditions with it creates the context for performance relative
unknown failure rates are also good to potential. If we agree on the risk factors
candidates for outage repairs and work. and accept they can be measured, then it
follows that Site A is more risky than Site B
One thing to note, however, is that when more risk factors are present; and this
predictive technologies work best is what our customer knew. Risk factors were
when employed over time. If you have present, and they needed to be quantified.
never monitored a machine and then The highest risk sites never want to
collect vibration from it once, you can see their scores, especially relative to
diagnose current faults. However, not the balance of the fleet. The information
having a history of data to trend means is threatening even though it simplifies
you lack the ability to identify fault identifying and mitigating risks. Ironically, the
patterns and progression leading to managers who need it most tend to reject
early warnings of faults. it while the ones who need it less embrace
it, further reducing their risks. Visualizing
“Regular data collection on the relative risk and sharing it across the fleet
Case Study needed interval is the best prescription will expose cultural strengths and weaknesses
for identifying machinery problems and will polarize the operating community.
Read about how Kimberly and fault progression,” says Hogan. Continuous improvers will use it to improve
http://www.plantservices.
Clark Paper Mill avoided “Annual screening or once-in-a-while performance and advance, while laggards
$5com/assets/Media/1402/
million in lost production monitoring isn’t the most effective use will find it meddlesome. The disparity
in performance and riskiness between
Azima-DLI-Paper-Mill.pdf of the PdM toolbox.”
CLICK HERE adopters and laggards will widen.

8 Financial Reasons to Deploy Predictive Maintenance Technologies | 13

 Knowing the outage schedule is also important.

“A nuclear plant, for example, will shut down every 18 months,” explains DeJesus. “When
they have a planned outage, they’ll ask us what maintenance they should do during that time.
We’ll use vibration as our heavy hitter, plus infrared, motor current analysis, and oil analysis.
Sometimes problems manifest themselves in other technologies before we see them in vibration.
You can see something in a motor current analysis long before you see it in a vibration reading.
We might use oil analysis to verify what we’re picking up with vibration analysis.”

Reason #8:
Detecting Problems While Equipment Is Under Warranty Saves Money

If something is broken, why not get it repaired while it is still under warranty?

It sounds simple, but in practice it is not always easy.

After an outage, an installer generally warranties his work for 30 to 60 days. With time-
based maintenance, you likely would not notice a problem until it escalates. The time that
takes could very well put you outside of the warranty.

With predictive technologies, you can quickly check equipment post-installation to check for
signs of problems.

“We often perform testing right after an outage, machine overhaul, or the commissioning of
a new asset,” DeJesus says. “When you put something back together you have the opportunity
to make mistakes. Things being aligned or bolted down properly is very important. When
you’re trying to get through an outage quickly, you might put a bearing in wrong, and it will
show up immediately in a vibration reading.”

If the work was done by a third party and a problem is caught within 30 to 60 days, it is
usually under warranty, she says. If the work was done in-house, you may need to review your
installation processes and procedures to make sure they are not the cause of the problem.

8 Financial Reasons to Deploy Predictive Maintenance Technologies | 14

 Predictive technologies provide a clear picture of the asset before and after maintenance
work or a repair.

“This documents the effectiveness, or lack of in some cases, and helps to identify any areas
of improvement,” says Hogan. “It’s a common practice for most of my customers to want to
document the effectiveness of a repair and confirm the work.”

Varied installation practices and protocols can introduce all sorts of headaches, much like a
driver with a heavy foot or a knack for hitting potholes impacts a car, says Hogan.

“All these technologies can help identify faults with newly installed assets. A factory-tested
machine may have passed its vibration compliance testing in the factory only to fail after it
was installed improperly on the ground at a customer’s site,” he says. “The identification of
these faults is important to the longevity and reliability of these machines. Vibration and IR
are two excellent tools for this work.”

One of Hogan’s customers had an emissions control fan refurbishment during an outage.
The fan was rebuilt with new bearings, coupling inserts, and a standard alignment and
balance.

“During our routine vibration analysis a month after the work, the data showed the fan
end had severe indications of bearing looseness and wear,” explains Hogan. “Our report was
expedited due to the high severity of the condition and the customer was in disbelief, given
they had just spent a large sum to fix these fans.”

The contractor was contacted and confirmed the condition. An operations problem was
identified as the cause of the condition and corrected. The fan had to be rebuilt, balanced, and
aligned again before going back into service.

“Our routine monitoring helped identify a failing fan, identify the faulty operations
procedures, and put corrective action in place to avoid failures in the future,” says Hogan.

8 Financial Reasons to Deploy Predictive Maintenance Technologies | 15

The predictive maintenance quiz
How does your program stack up?

H ave you ever wondered how your predictive maintenance (PdM) program stacks up to
others? Would you like to know how to improve? If you don’t already use PdM, are you
curious about how it could benefit you? If you answered yes to any of these questions, then
you’re in the right place. Take our quick quiz, tally your results, and then read on to learn
ways to help ensure your PdM program meets best practices.

Step 1: Take the quiz

1. RELIANCE OR COMPLIANCE?
− Do you rely on your PdM program to improve uptime and drive down costs? (Yes = +5)
− Is the ONLY reason for your PdM program to comply with a mandate? (Yes = +2)
2. PRETENDING YOU’RE TRENDING?
− Does your program trend machine conditions over time to maximize your maintenance
effectiveness and efficiency? (Yes = +5)
− Is your primary goal to identify machines with serious or critical issues? (Yes = +2)
3. IS YOUR SKILL FOR REAL?
− Does your staff have the technical field expertise to detect issues on ALL of your critical
equipment? (Yes =+5)
− Is your team trained on how to combine the results of vibration, infrared, oil,
ultrasound, and motor testing to get to the root of a problem? (Yes = +5)
4. ABORTING REPORTING?
− Are regular reports being generated on equipment health, collection compliance,
program performance metrics, and maintenance performed in response to
recommendations? (Yes = +5)
− Are reports generated only for machines identified with serious or critical problems?
(Yes = +2)
5. IS YOUR COST LOST?
− Do you know the cost of your PdM program within 10% of total spend? (Yes = +5)
− Do you have the necessary information to accurately calculate your PdM program’s
ROI? (Yes = +5)

The Predictive Maintenance Quiz – How does your program stack up? | 16
Step 2: Tally your score
25+ Points
Congratulations! You have a solid PdM program in place. Check the topics below for ideas on
how to fine-tune and sustain your program.

18-24 Points
You are on the path to a very successful PdM program. Read the topics below to find out
where you can make further improvements.

10-17 Points
You have the potential to get more from your PdM program than you are currently realizing.
Review the topics below to identify areas where you can make improvements.

2-9 Points
Congratulations on having a PdM program in place – that’s an important first step, but it
doesn’t end there. To be truly effective, it needs to meet more than the minimum standards.
Use the information below to identify where your program can be strengthened.

Step 3: Step up your game

Reliance or Compliance?
“Both mandated and chosen PdM programs can be equally successful or unsuccessful,”
says Tim Kelley, Director of Watchman Services at Azima DLI (www.azimadli.com). “One
approach is not necessarily better than the other.”
Programs that are internally mandated may experience slower adoption or internal
resistance compared to those brought in by choice, and they may require more internal selling
and education up front. On the other hand, having executive support increases its odds of
sustainability, and tying plant managers' incentives to performance can enhance its degree of
success.
Chosen programs benefit from champions who understand the value of PdM and are
willing to put forth the effort to implement it. However, if your champion or other PdM talent
retires or moves on, the program risks failure, particularly if the duties are handed off to
someone lacking the time or interest to ensure continued success.

The Predictive Maintenance Quiz – How does your program stack up? | 17
 “The perfect mesh is when a champion is invested in implementing a PdM program and the
concept is sold to corporate. That’s the best of both worlds,” adds Kelley.
Regardless of how it originates, a program’s effectiveness depends on how it is implemented.
Too often, companies collect condition data but fail to analyze or use it properly.
Consequently, maintenance remains largely reactive and subject to firefighting, thus limiting
the PdM program’s financial benefits. A best-practice PdM program predicts when it’s time
to do maintenance so outages can be efficiently and proactively scheduled, and steps can be
taken to ensure troubled machines keep running until the planned shutdown.

Pretending You’re Trending?

If you are not trending your data, you are not alone. However, by focusing only on serious
or critical issues and not trending your condition data, you miss opportunities to optimize
maintenance performance.
Visibility into the early progression of a fault allows you to properly plan ahead.
“In the past, people had more familiarity with their machines. They were tuned in to how
the equipment should sound and operate, and had a better grasp on where to direct their
attention,” explains Michael DeMaria, Director of Product Management at Azima DLI. “Now,
personnel turnover and changing technology has increased the need for PdM.”
Capturing condition information regularly and tracking it over time allows work orders to
be created and completed in time to avoid failure. As machine issues decline and firefighting
wanes, there is more time to focus on lower priority opportunities, for instance design and
operational performance.
Very often, companies lack the means or capabilities to trend data. Those responsible have
additional duties and too little time to spare, or the data may not be centrally accessible. Asset
management systems can be part of the problem.
“The goal of a CMMS or EAM system is to be a maintenance repository but their
usefulness depends on the initial setups and often times data and asset information is
incomplete,” says Kelley.
Another obstacle to trending is the choice of technology. For example, management may
want a vibration program, but the cheapest technology – broadband meters – does not support
trending. What is needed are narrow-band tools that enable you to see subtle nuances in
machines as they start to develop or deteriorate. By finding such conditions early, you can
plan downtime that caters to the repair, and only order the parts necessary. Once the repair
is complete, you can assess its effectiveness and ensure no new problems were introduced
because of it by comparing the machine’s post-repair state to a baseline.

The Predictive Maintenance Quiz – How does your program stack up? | 18
 Because some PdM is better than none, companies should tackle the easy wins first and
quantify the savings over time. Tracking return on investment (ROI) information makes it
easier to justify expanding the program to additional bad actors and machine types.

Is Your Skill for Real?

The ability to run an effective PdM program with your current staff depends on workload
and availability, equipment expertise, and PdM skills. Getting up to speed requires a
significant, albeit worthy, investment.
“The most successful programs always originate the hard way – following an expensive
problem,” says DeMaria.
In companies that attempt in-house programs, PdM is usually a collateral duty rather than
a dedicated role. If your staff is already spread too thin and struggling to complete preventive
maintenance tasks, then PdM success is unlikely. There is also the risk of employee turnover.
Latest trends show that every three to five years, personnel tend to change or leave jobs. If the
expertise is lost, the program will have to start from scratch.
When PdM teams favor the machines most familiar to them and lack real-world experience
with all of the critical machines in their plant, PdM success is constrained. It takes years to
develop expertise on the various machines and component types.
Furthermore, the best programs leverage multiple PdM technologies such as vibration,
infrared, oil, ultrasound, and motor testing. There is an art to gathering and aggregating
the results from multiple sources and analyzing the data effectively. Vibration, for example,
is not an exact science. There are subtle nuances in vibration readings, and brand new
machines will respond differently than older equipment. Combining technologies can settle
uncertainties. If a vibration technician suspects tooth chatter on a gearbox it can be verified
by pulling an oil sample.

Effective PdM teams have:

• Time available to perform PdM activities
• Expertise with the various PdM technologies
• Knowledge of multiple equipment and machine types
• Continuity of knowledge despite personnel turnover
• Funding to maintain adequate skill sets and certifications

The Predictive Maintenance Quiz – How does your program stack up? | 19
 Developing the needed skills can be hit-or-miss. For instance, vibration classes use clean
data to show what an imbalance looks like, but real-world scenarios may look very different.
Or, you may hire a college graduate freshly certified in PdM, but he or she lacks hands-on
machine experience, making it difficult for them to diagnose complex problems.
Outsourcing can be a cost-effective solution to the skills gap.
“If I were a plant manager starting a vibration analysis program, I wouldn’t want my staff,
fresh from training, to test their knowledge on my most critical machines,” says DeMaria.
“Does your analyst have enough experience to make a $1 million judgment call? Even the
best analysts sometimes need a second opinion.”
Azima DLI developed software that performs PdM analysis and automatically trends each
unique fault condition, using healthy machines as the baseline. A pump requires a different
specific frequency range than an electric motor, for example, and certain frequencies come
out of a machine naturally. Therefore proper database setup is crucial. Understanding how
to capture and account for faults and naturally occurring frequencies can be tricky. External
support from a PdM expert or an OEM can be helpful.
In the best outsourcing programs, plants retain the data collection routine while a third
party performs analysis. PdM partners can also recommend upgrades to the overall program,
for instance changing out certain bearing types, tweaking setups and alarms, or optimizing
the EAM/CMMS software. By partnering together, the plant personnel can be the eyes and
ears for the program locally and they also have the historical maintenance background of the
machines. The PdM services provider can add value with their vibration expertise and also
compare similar applications to help identify any maintenance, design or installation faults.
While PdM services from OEMs and local motor shops tend to be side work, Azima DLI’s
core business model, capability, and investment centers on PdM options and continuous
improvement. “We have many analysts who are well-qualified and experienced on all aspects
of PdM and the various equipment being monitored, who can be called upon at any time for
help,” says Clint Goodin, a Vibration Analyst at Azima DLI.

Aborting Reporting?
Assess the reports that your PdM program generates. The time invested in reporting can
have a major impact on your overall uptime. If they clearly reveal the faults and how they are
progressing, you can plan your actions in a timely, cost-effective manner.
If reports show serious or critical machine problems, but lack overall machine health
trends, developing issues, repair recommendations or actions taken, then PdM efforts will be

The Predictive Maintenance Quiz – How does your program stack up? | 20
ineffective. If your staff doesn’t have time to create the reports or they are not disseminated
to the right people, then your PdM investment is wasted. If your reports don’t relay
operational factors such as efficiency, uptime, and ROI, then you aren’t able to measure and
communicate the PdM program’s performance.
Poorly designed reports are another concern.
“If a report says ‘bad bearing,’ which bearing is it? If it reports ‘looseness,’ is that at the fan
or motor? Pinpoint the problem and give an actionable recommendation, and your PdM will
be more efficient and effective,” recommends DeJesus.
For instance, if a trend of fan or pump misalignments are noted, consider retraining the
technician on how to align machines. If similar machines all have the same problem, it could
be a design problem, a bad bearing, the wrong bearing, or it’s installed incorrectly.
“Show the most critical issues at the front of the report so the users can avoid wasting
time weeding through dozens of pages,” suggests Dan Hogan, a Vibration and Oil Analyst
at Azima DLI. “Include pictures or mechanical schematics showing problem areas for less
experienced field technicians. Talk with your report users about the contents and flow, and
consider leveraging a PdM service provider to apply the requested improvements if in-house
availability is limited. They know their audiences and can tailor a message without jargon.”
Azima DLI’s reports include a History Table that shows all monitored assets by name
and their color-coded severity over a full year, giving management insight into reported
condition trends and progress on the faults. They can determine how long it took for a fault
to progress, determine whether an unreliable machine or process caused the problem, and
judge whether the technicians are not acting quickly enough on corrections. If a catastrophic
failure occurs, they can assess whether the condition was not found or the equipment was
not monitored.
“It’s difficult to get people to read reports. Our clients can take our report and create work
orders from it,” says Goodin.
Delivery format flexibility is also important. Azima DLI’s Watchman Reliability Portal
provides the data online in a central repository, but it can also be emailed in a PDF file or
physical copies can be printed. Azima DLI’s reports automatically tie in to the customer’s
internal planning systems, without a separate import step; the data is visible in both systems
concurrently.

Is Your Cost Lost?

You should always know, within 10% of your budget, what your PdM program is costing
you. Likewise, it is essential to document and communicate how much ROI your program

The Predictive Maintenance Quiz – How does your program stack up? | 21
is realizing and measure its performance against other plants and business units across the
globe. A solid PdM program will enable you to track these figures easily and effectively.
Cost controls are a necessity. An in-house PdM program includes fully loaded costs for
each team member plus the costs of the equipment, software, and perhaps some outside
consulting services. Data collectors alone can cost as much as $40,000 each, and they need
to be refreshed every three to five years with new, supported technology. Annual hardware
and software support costs further inflate the budget, as do skills training, certification, and
recertification costs.
Some or all of these costs can be assumed by a PdM service provider. Azima DLI’s full-
service PdM, offered on a monthly subscription basis, is the most predictable for budgeting
purposes. No capital investment is required and the agreements are fixed for three- to five-
year terms. Companies with a larger resource pool and budget may instead purchase the
hardware and software, perform the data collection, and leverage Azima DLI for analysis
and to fill in when internal resources are unavailable. Be aware that some other providers
will require change orders for out-of-scope services and charge for time and materials.
Any money invested in an effective PdM program will quickly be offset by maintenance
cost savings. Your maintenance budget will decline by replacing scheduled PMs and run-to-
failure conditions with work based on actual asset conditions.

PdM ROI potential is substantial:

• Replace excessive maintenance with condition-based maintenance
• Avoid unscheduled downtime by better planning work
• Avoid production losses by preventing emergency shutdowns
• Reduce spare parts inventory by reducing emergency repairs
• Reduce labor and overtime costs by avoiding late-night breakdowns
• Recover lost productivity by avoiding the distraction of emergencies
• Reduce lost business and late-delivery penalties by improving on-time shipments
• Avoid unfinished product disposal by preventing breakdowns mid-batch run
• Avoid hazardous material incineration costs by preventing breakdowns mid-batch run

The Predictive Maintenance Quiz – How does your program stack up? | 22
 Senior management responds to ROI data. Did you pay $50,000 for a PdM technology
that prevented a production shutdown that might have cost $200,000 per day? Did you
replace bearings annually but now you can wait until signs of failure appear?
“If you do a good job at minimizing equipment failures and downtime, driving efficiencies
and reducing costs, but fail to take adequate credit, management could easily defund the
program or reduce the service frequency, assuming it’s not important anymore,” explains
Kelley. “Within six months, all of the reliability gains will have been lost.”

Conclusion
We hope that taking this quiz, learning your score, and identifying opportunities for
improvement helps you to move closer toward best-practice PdM. Whether you repurpose
your staff and processes, or bring in PdM partners for assistance, you can increase efficiency,
lower your overall maintenance costs, and do more with what you have.
Remember, there is no need to outsource your PdM entirely – a hybrid approach where
your staff’s strengths are maximized and supplemented with a strong partner is equally
effective. Your staff is smart, they work hard, and you need them. Get them the help they
need so that they can focus where you need them most.

The Predictive Maintenance Quiz – How does your program stack up? | 23
Is an online monitoring
system right for you?
What you should know
by Sheila Kennedy, contributing editor, Plant Services magazine

T he decision to implement an online condition monitoring system should not be made

lightly. Both the pros and cons must be weighed, as well as the ultimate payoff potential.
Which machines are the best candidates? What advantages will be gained? Is it worth the costs?
What is the best configuration? Understanding these considerations is essential in order to
design the most effective and economical solution for your predictive maintenance (PdM) goals.
Your ideal solution could end up being a stand-alone online system — maybe wireless,
maybe automated — or you might prefer a hybrid online and walk-around monitoring
approach. The data and analysis could be managed in-house, or third-party services might be
preferred. You might even choose to forgo online monitoring altogether until any red tape and
budget constraints are overcome. Either way, it’s best to begin the journey with open eyes and
a clear PdM vision.

What’s driving online condition monitoring?

The motivation for online monitoring is rooted in core business challenges: safety,
accessibility, equipment complexity, labor constraints, and insurance requirements.
Increasingly sophisticated online monitoring solutions are able to solve problems that walk-
around approaches could not.

Safety: With online systems, hazardous equipment and locations can be monitored without
risks to personnel or the environment. “Walk-around inspections in unsafe conditions require
special access and lockout/tagout procedures, which are labor-intensive, costly, and still
potentially risky,” says Laurent La Porte, director of technical services at Azima DLI (www.
azimadli.com). An alternative is to install sensors permanently and run the wires out to a safe
area and use a walk-around meter to access those sensors.

Is an online monitoring system right for you? | 24

 Equipment accessibility: Machines in distant or hard-to-reach locations are prime
candidates for online monitoring. “If you can’t spare the time or cost of a person to drive
200 miles out to a pumping station to collect data every month, then it makes sense to hook
up a system that’s automated and will report in on schedule,” suggests Joe Van Dyke, VP of
operations at Azima DLI. Wind industry turbines are another example. They are commonly
installed on remote land, rough terrain, challenging climates, or at sea, and normally no one is
allowed up in a tower while it is operating.

Machine complexity: As machines become more complex and efficient, they require
more predictive maintenance attention. “Forty years ago, the machines and subassembly

“Online monitoring is becoming more

prevalent due to the complex nature
and running state of the machines.”

Is an online monitoring system right for you? | 25

components driving process manufacturing were basic, but that is no longer the case. In
chemical manufacturing, for example, online monitoring is becoming more prevalent due to
the complex nature and running state of the machines,” says La Porte. “Acquiring more and
better data, and having it properly assessed, is becoming a necessity.”

Labor pool: If there are not enough personnel at a particular site to absorb the testing tasks,
an online strategy can fill the gap. “When your desire for machinery condition awareness
exceeds what you can reasonably achieve from walk-around data collections, then an online
system will be of value to you,” explains Van Dyke. “If you want to test twice a day but don’t
have the resources available, then you need an automated system.” The industry trend is
toward fewer personnel and more remote work, particularly in the United States, where labor
costs are high. Online systems don’t require salaries or benefits, and they don’t take time off
for coffee breaks, holidays, or vacations.

Insurance requirements: Insurance companies may request that certain machine classes be
covered by an automated surveillance system, without necessarily defining the specific kind
of system or how responsive it should be. Online vibration monitoring systems tend to satisfy

Online monitoring design considerations

• Selectable input types: DC, 4-20mA, ICP
• Selectable data types: Scalar or process, waveform, FFT (spectra), demodulation,
advanced bearing detection algorithms
• Automation: Data collection, diagnostics, and resulting repair recommendations
• Simple and complex logical parameters: Simple = if X, then Y; complex = if X and/or
Y, then Z
• Ability to set multiple alarms, thresholds, baselines, averages
• Data buffering and/or database storage
• Wireless or wired
• Modular and expandable (hardware and software)
• Fully customizable and instantly re-configurable settings
• Ability to interface with third-party software

Is an online monitoring system right for you? | 26

this need. “One major pharmaceutical company permanently installed sensors and a dedicated
monitoring system on one of its gas turbine generators just because a mutual insurance
company required an online, automated condition monitoring system for this critical unit,”
says Van Dyke.

Benefits of online versus walk-around monitoring

Taking an online approach under the right conditions yields compelling process and cost
advantages, such as greater operational efficiency, downtime avoidance, extended asset life,
simplified training, and system scalability.

Process efficiencies: Data collection can account for 80% of the labor involved in a walk-
around vibration monitoring program. With an online, automated program, the labor burden
is near zero because users can remotely command tests, modify test frequencies, and run
special tests when needed. “It’s also easier to get online condition system data into other
enterprise systems and in front of more people,” says Dave Geswein, director of engineering
development at Azima DLI.

Early fault detection: Online systems can detect mechanical problems much sooner
than walk-around systems. Delivering notifications and expert diagnostics within seconds
allows the equipment to be taken off production and repaired before greater harm is done.
“We monitored an air separation unit at a commercial air product company for several
years and noticed the compressor pinion’s bearing clearances open up to the point that the
OEM had to become involved. The OEM did an in-field, low-cost bearing replacement,
avoiding a catastrophic failure of the machine,” says La Porte. “If the unit had gone down
catastrophically, it would have cost in the $1 million range for repairs and downtime, or
possibly multi-millions for replacement.”

Lifecycle extension: Some machinery operators have equipment with known, common flaws,
but they still need to run them. Installing an online system can allow continued use of the
equipment while covering them against a fault. If the equipment exhibits signs of the failure
mode, it can be addressed right away. “A government agency had a certain class of marine
reduction gears with a design flaw operating in four locations. They implemented an online
system to look at that gearbox and check for signs of the fault. If detected, they could turn the
gearbox down or off, or use a companion unit. This allowed them to maintain operation of the
units until a redesign could be instituted a couple years later,” explains Van Dyke.

Is an online monitoring system right for you? | 27

Interference
Industrial wireless condition sensor
Node
technologies fall into two general
Node
technical categories: mesh technology
or a direct-node-to-access-points (APs) Node
Node
arrangement like regular Wi-Fi 802.11. In
either case, they typically use ISM band
Node
frequencies such as 2.4GHz typical.
Mesh devices form their own self- Node
Node Node
healing network and can theoretically
survive the temporary introduction of an
electromagnetic interference source or
sink that interrupts a communications
path between elements of the mesh. Node

This model is indeed robust, but not

Mesh Technology
failure-proof, since a node that has only Node
one connection can get disconnected,
and interference can affect more than
one pathway.
Node-to-AP-based arrangements Controller
Node
rely on a robust array of receivers and/ LAN
or repeaters for reliability. While the
set of APs can be designed to provide
redundancy, many wireless devices will
lock onto a preferred wireless access
Monitoring
point and, even when it is weakened or System
unreliable, will continue to use it rather
than switch to a temporarily stronger

Is an online monitoring system right for you? | 28

more constant access point signal. This
can introduce reliability issues.
Sources of interference or wireless Node Node Node
signal problems can be as simple as
an object or structure placed between
a transmitter and a receiver, or a
Access
strong magnetic field like having a Point

transmitter or receiver too close to

an energized motor or other wireless Node Node Node

device. Sometimes having the devices

themselves too close to each other
can cause problems. To remedy this, LAN
there are Wi-Fi surveyor companies
and tools that are available on the
market. Use of these services or tools Node Node Node Node Repeater Node

can help identify persistent issues,

but don’t always capture intermittent
interference, which can be the most
Access Access
common type of interference. Point Point

When sensors, surveyed assets, or

access points are subject to intermittent
strong EMF, a wireless system that
transmits only small amounts of data
Monitoring
will be more robust. A system that System
must transmit long sets of readings will
have more trouble because the longer
time required is more likely to span an Access Point / Repeater technology
interference event.

Is an online monitoring system right for you? | 29

Support Models

Remote/distance support models cover the spectrum and differ widely but can be divided roughly into three definitions.
1. Data and systems stay in the plant and with remote access allowed via secure connection. This is more common for plants
with restrictive corporate data policies.
2. Data stored/hosted off-site with access to the hosted data for analysis via Web apps, websites, or dashboards. This is
becoming a very common model, with the availability of highly secure hosted data and data services. In this scenario,
systems for acquisition of data communicate their information off-site for secure storage and management and analysis. The
plant personnel access the data via applications running on workstations or corporate servers or as Web applications.
3. Mixed, with remote access to components in place at the plant, and hosted cloud components with some data on-site and
some off-site. This is generally the most flexible solution, offering the best plant systems integration options while still allowing
for good distance support.

Everything onsite and Data and analysis systems

Optional services outsourced hosted offsite

CBM Monitoring, Optional Expert CBM Monitoring,

OPC, SCADA Analysis, and Assitance via OPC, SCADA Analysis, and
Historian reporting system remote-desktop Historian reporting system
and services or equivalent and services

Plant IT Plant IT
Infrastructure Infrastructure

CBM Devices CBM Data CBM Devices CBM Data

Is an online monitoring system right for you? | 30

 Training savings: Walk-around data collection can be taught in a couple of hours, but
becoming expert in data analysis can take months or years of ongoing training. On the
other hand, automated acquisition tools eliminate data-collection training altogether, and
outsourcing data analysis to a third-party service provider eliminates analytical training and
turnover concerns.

Scalability advantage: Online condition monitoring systems should be modular and

expandable in order to accommodate additional sensors and greater data volume as
new requirements or opportunities are identified. “A maintenance team used a condition
monitoring system for walk-around inspections and later added an online system to one of
their gas turbine generators. Because the existing monitoring system was scalable, they were
able to collect the online data in the same database where the walk-around data resided, and
no new training or software systems were required,” says Van Dyke.

Understanding the cost factors

A hybrid approach of both online and walk-around monitoring is often the best and most
economical course of action. “It’s usually not practical to put online systems on all machines,”
says Geswein. “I normally recommend putting online systems on the most important
equipment and continuing manual monitoring on the rest of the plant.” Over time, additional
classes of machines may be added to the online system if the walk-around program reveals a
need for better coverage.
Implementing an online condition monitoring system is usually expensive compared to a
walk-around system, and the price can be a hurdle for some organizations. Typically, about
40% of the cost is sensors and cable, 30% is installation labor, and the remaining 30%
is the monitoring system itself. Additional costs may include the infrastructure (software,
network), training, outsourced monitoring, extended warranties, and future hardware and
software upgrades.
Getting permission to install the system carries costs, as well. The approval process can be
particularly difficult if significant modifications are required to an existing plant. In industries
such as nuclear power, the red tape can be considerable, even if the business justification is solid.

Is an online monitoring system right for you? | 31

 Another often unexpected cost pertains to the role of the information technology (IT)
department. “Many companies don’t include IT people in the decisions until the day of
the install,” explains Geswein. “The data won’t get from Point A to Point B if a firewall
is blocking communications; it’ll have to be configured to allow the data through. At one
nuclear facility, getting the firewall modified took two or three months to complete because no
procedure existed for doing so. They had never done that type of change before.”
Wired systems have higher installation costs. “I would budget about an hour per sensor to
install it and run conduit. That’s the biggest cost of field installation,” says La Porte. Using
wireless monitoring devices can reduce installation costs by reducing
the amount of cabling and red tape involved, although the associated
batteries introduce maintenance and reliability costs and can limit the
data quantity and/or quality.
An online system can be up and running within 30-45 days,
depending on the number of machines involved. After that, they
generally require little maintenance. Sensors and wires can be damaged,
acquisition devices have a finite lifetime, wireless components can
be less than robust, and anything that needs batteries requires
maintenance. Calibration requirements can also add to the cost. “If
you have 100 sensors, you could average a half hour every two months
troubleshooting them remotely. For in-plant troubleshooting, I’d
estimate about an hour every two months,” adds La Porte.

Calculating the value

The true value of an online monitoring system is recognized over the
remaining life of the facility. Avoiding machinery failures, eliminating
data collection labor time, and preventing safety or environmental
incidents can easily and very quickly produce bottom-line benefits.
“Given enough time, the system’s benefits will outweigh its costs.
For instance, if a power plant receives early notification of a turbine
problem and takes quick action to prevent further damage, it can be
“The heightened awareness of worth millions of dollars,” says Geswein.
a machine’s status gives you “The heightened awareness of a machine’s status gives you the
the opportunity to make more opportunity to make more saves and avoid more failures and unplanned
saves and avoid more failures downtime,” explains Van Dyke. “Timeliness is paramount. A wind
and unplanned downtime.” turbine operator detected signs of a gearbox bearing failure while the

Is an online monitoring system right for you? | 32

gearbox rebuild was still under warranty, so the rework was conducted for free. Had the fault
been discovered outside of warranty, a $20,000-$30,000 bearing replacement would still be
far more economical than a $200,000 gearbox replacement.”

Standardization and compliance advantages

Automated online monitoring programs facilitate data collection, testing, reporting, insurance,
and warranty compliance. Online monitoring systems allow for standardization of how
condition data is collected, stored, and presented. “I’ve seen fingers point to vibration analysis
as being ineffective after a critical equipment failed, when it was actually a lack of compliance
in data collection that
hindered the analysis
process,” remarks Van Dyke. Increasingly sophisticated online
“When you remove people monitoring solutions are able to
completely from the process
and let computers do the
solve problems that walk-around
work, you get consistent approaches could not.
results every time,” explains
Geswein. “In vibration analysis, all you are really looking for is a change. If you have
consistency from the ground up and a change is detected, then you know it’s real and not
some outside influence.”
A corporation that standardizes its predictive maintenance and risk reporting practices
and systems will benefit from a larger knowledge base, better history for compliance, and
centralized decision making. Other companies prefer their plants to be autonomous and make
decisions that are best for them. “Larger organizations like power or paper companies usually
let their individual sites determine which technology resources to use for vibration monitoring
and analysis,” says Van Dyke.

Technology and tool considerations

“A lot of people get caught up in the specifications of the online systems,” cautions Geswein.
“To me, what’s more important is whether the software and system that is receiving the data is
able to handle the volume that will be collected. People who go from all-manual data collection
to even just a few online systems are suddenly swimming in more data than they’ve ever had
before. You’ll need the system to screen the data and flag the machines that require attention.”

Is an online monitoring system right for you? | 33

 Maintaining data security is essential. A company may not want its data to go up to someone’s
cloud-based server for fear it falling into the hands of a competitor or hacker. In that case,
containment of the data and isolation of the systems come into play. Government agencies with
tight restrictions on IT fall into this category, and steel plants sometimes do, as well.
Security concerns also drive
some companies to purchase
the online system, install it
on their own or with some “Acquiring more and better data,
assistance, and then they run and having it properly assessed,
and analyze the data internally is becoming a necessity.”
or put it in the hands of the
consultant in charge of the
facility. “We call this an ‘inside the fence’ solution when they want to keep all of their data
inside. It’s a fairly common model,” says Van Dyke.
The adoption of wireless sensors in vibration monitoring is growing but not particularly
widespread. While less expensive to install, wireless technology usually comes with batteries,
which require maintenance and can therefore cause the sensors themselves to be less reliable.
Additionally, because wireless systems usually transmit the data over radio waves, they try
to reduce the data down to a few single parameters rather than the high-definition data that
an analyst may prefer to have on every measurement. “Rather than complete high-resolution
spectra and waveforms, the analysts may only receive overall measurements. If they detect
any changes in an overall measurement, they’ll need to reconfigure the system to collect the
spectral data if possible,” adds Van Dyke.

Conclusion
Weighing the pros and cons of online monitoring will help you decide if it’s right for you.
Each company’s business case is unique, but the flourishing predictive nature of online
monitoring systems is simplifying their justification. Automated solutions, enhanced data
capture, rapid analytics, and early warning of potential failures are among the compelling
features designed to solve operational and budget challenges. Organizations that incorporate
online monitoring with or without a walk-around program are empowered to operate more
efficiently, reliably, and safely, with fewer mechanical failures and costly unplanned outages.

Is an online monitoring system right for you? | 34

Monitoring Magic
Watch your assets from the cloud
by Sheila Kennedy, contributing editor, Plant Services magazine

T he introduction of cloud computing and a new design paradigm for predictive

maintenance (PdM) instrumentation are adding versatility to the implementation of PdM
programs. Automated online condition monitoring, supported by the cloud, is shifting the
burden from front-line maintenance personnel to remote specialists, whether in-house, third-
party service providers, or both.

Compared to the early days when screwdrivers were pressed to the ear to hear machinery
problems, this new approach works like magic. It is far more practical, comprehensive, safe,
and cost-effective.

“Since 2000, the instruments have gotten faster. They can measure and store larger quantities
of data, the graphics have improved, and the data exchange process is much faster,” says Ken
Piety, vice president of technology at Azima DLI (www.azimadli.com). In addition, systems now
exist that allow unmanned and more frequent data collection, and the introduction of cloud
computing makes the data and analysis more widely accessible to authorized personnel.

The reliability gains that result from online data collection and remote diagnostic analysis
are game changing. “Our communications infrastructure and mobile devices have shrunk the
world, and radically changed both our personal and business lives,” adds Piety.

Why change was needed

Improvements to PdM over the years have enabled more timely information, sophisticated
analysis, effective data presentation, and impressive return on investment (ROI). As a result,
the number and scope of PdM programs and the size of the audience are expanding within
and beyond the four walls of the plant. This is causing the machines, reliability professionals,
and other stakeholders to become more geographically dispersed, and increasing the demand
for the flexibility of a distributed PdM model.

Monitoring magic | 35
 Traditional monitoring models brought in-house or external analysts to the machines.
Internally driven programs were generally informal and limited in scope because the startup
and maintenance costs and resource demands were high. Often, interest in the program would
wane if the champion moved on. Programs that were outsourced created a dependency on the
supplier for the equipment, data collection, and analytical services. Timeliness was sometimes
lacking and any changes or increases in the machines monitored came at a cost.

The Distributed Model

Data Collection Execute Corrective

Actions

Program and Plant

Management

Data Analysis Plan and Schedule

Maintenance

The distributed approach brings PdM data and

information to the people who need it, when they
need it, wherever they are located at the time.

Monitoring magic | 36
 The more modern option today is the distributed approach, where PdM data and
information is brought to the people who need it, when they need it, wherever they are
located. “With a distributed model, companies can elect to operate entirely with an in-
house PdM staff, or they can outsource all manpower, use a blended staff routinely, or seek
assistance during specific periods or for special problems or second opinions,” explains Piety.

Interdependent Stakeholders

Corporate Management

Plant Management

Maintenance Planning and Performance

Data Analysis and Data Collection

Machine Health
Reporting

With the information stored in the cloud,

authorized personnel at any location can track
machinery performance, troubleshooting data,
corrective actions, and program results.

Monitoring magic | 37
 Remote monitoring access can be provided through an Internet connection. Companies with
strict corporate policies may elect to maintain the data and systems at the plant and enable
secure remote access. Others prefer offsite hosting and storage of the data, with analysis
performed over the Internet.

“All of that information is sent to a central repository and analyzed by trained analysts
24/7, so maintenance personnel fix only the equipment that needs attention at that moment,”
explains Joe Van Dyke, vice president of operations at Azima DLI. “The benefits can be very
significant, resulting in advanced notice on machinery problems that could otherwise have
devastating financial and safety ramifications.”

A third, more versatile and increasingly popular approach leverages a combination of

plant-based and cloud-based components. “The mixed approach provides remote access to
components in place at the plant, as well as hosted cloud components with some data on-site
and some off-site,” says Van Dyke. “This is generally the most flexible solution, offering the
best plant systems integration options while still allowing for good distance support.”

Program benefits
The primary functional benefits of a distributed PdM program include automation of data
collection, access to dynamic information, and the elimination of communication barriers.

Monitoring automation: Automated monitoring systems enable more frequent and more
predictable machine condition monitoring without the aid of data collection personnel. They
can gather readings on many machines on a near-continuous basis, making them more cost-
effective than the labor-intensive walk-around programs. “Near-real-time test frequencies are
possible when testing high-speed equipment or gathering overall vibration or scalar data like
temperatures and pressures,” says Van Dyke. As a result, plant personnel can focus on higher-
value activities while the system sends its updates to the cloud.

Monitoring magic | 38
 Dynamic awareness: Automated or online systems can observe dynamic changes in machine
fault conditions arising from certain load, speed or operating condition changes, but at a
lower cost, according to Van Dyke. Compared to walk-around monitoring programs, online
systems are always on duty so the sought-after events won’t be missed. When integrated into
SCADA systems, the online system can be designed to collect data only under specified loading
and operations conditions, which avoids collecting data when machines are idling, unloaded,
or turned off.

Reliability Portal Dashboard

Automated monitoring systems enable

more frequent and more predictable machine
condition monitoring on a near-continuous basis,
making them more cost-effective than labor-
intensive walk-around programs.

Monitoring magic | 39
 Open communication: Cloud-based condition monitoring breaks down communication
barriers within the plant, across the corporation, and among trusted reliability service
providers. Plant and corporate operations, maintenance operations, diagnostic center
operations, and the field data collection and data analysis teams all have a stake in the
program and its outcome. With the information stored in the cloud, authorized personnel at
any location can track machinery performance, troubleshooting data, corrective actions, and
program results.

Strategic benefits
Best practices in condition monitoring are supported by this model, and it addresses strategic
concerns about costs, expertise, safety, and corporate performance.

Reliability cost efficiencies: Remote data collection and diagnostic analysis ensures more
consistent, near-real-time insight into equipment conditions, enabling proactive steps to
improve performance and avoid failures. It generates ROI through accurate and timely
machine performance assessments, including efficiency, throughput, etc., and relating that data
to the machine condition.

Sustainable expertise: Mitigating the loss of critical talent is a universal goal in the
industrial space. “Machinery analysts tend to turn over every three to five years, and it
requires a minimum of two years to train a new analyst,” says Piety. “In addition, very few
analysts are experts in analyzing all types of machines and all types of faults.” The distributed
model allows companies to assign program functions to those who can execute the tasks at
the lowest cost and with the greatest effectiveness, whether it’s internal personnel on site or at
a sister plant, or an external service provider.

Improved safety: Online monitoring systems help to keep personnel away from unsafe
environments and equipment. “It allows for collecting data in sensitive, dangerous, or
hazardous areas without risking harm to your personnel, processes, tools, or the assets you
are monitoring,” says Van Dyke. “Even when the equipment or the environment is considered
safe, safety ratings should be expected to improve because automated data acquisition
eliminates the need for personnel to be physically entering machinery spaces or carrying
instrumentation up and down stairs and ladders, and it all but eliminates the potential of
contacting energized equipment.”

Monitoring magic | 40
 Corporate standard: The distributed PdM model encourages the adoption of industry best
practices across all sites. “It allows the sharing of information with corporate counterparts for
the purposes of consultation, troubleshooting, collaboration, and reporting of KPIs and other
performance-related issues,” says Van Dyke.

Outsourcing and the cloud

Those seeking outside assistance might consider a diagnostic service provider and storing
the PdM data in a commercial virtual data center. Virtual diagnostic centers are generally
staffed by large teams of experts with diverse expertise who use state-of-the-art, cloud-
based communication methods. Their high-bandwidth physical data centers are accessible
globally over the Internet, and their hardware and software excels at communicating data
and information. Closed-loop communication methods connect all team members, and the
operational infrastructure is designed to secure data from loss.

Piety summarizes the business benefits of this approach as follows:

• Centralization: PdM data is centrally stored for access by multiple users
• Collaboration: Multiple analysts can collaborate, thus increasing diagnostic accuracy
• Integration: PdM is accessible to other business systems, which opens the door to data
exchange between systems
• Transparency: The more people who see PdM information, the more valuable the program
becomes
• Communication: Email alerts or text messages can notify specified individuals when specific
events have occurred, such as health state change notifications, due and overdue reminders,
and the delivery of program KPIs
• Participation: Others can gather data, contribute knowledge, and record observations
• Optimization: The right people can perform each function at the lowest cost

Special considerations
A rigorous IT data security policy is needed for condition monitoring systems that push data
to the cloud or allow access in from remote users. Firewalls must be constructed and managed
to restrict all unnecessary and unauthorized access, and suitable data encryption methods
must be employed.

Monitoring magic | 41
 Users of data centers and cloud-based systems must also address data ownership and data
protection. Although data is generally considered the property of the asset owner, Van Dyke
recommends that explicit, contractual provisions be made to ensure that the data is protected,
not copied, and returned if and when the contract is canceled.

Solution: The Distributed Model

Sustainable Increased Controlled

expertise effectiveness costs through
through through efficiencies
virtualization collaboration

A distributed, automated condition monitoring

system leveraging cloud computing and today’s
PdM instrumentation is far more flexible and
functional than prior options.

Monitoring magic | 42
 Continuity of services from an IT perspective is another consideration. If a critical link is
severed because someone at the data center accidentally changes a password on a server used
by a remote monitoring system, it can cause cessation in monitoring. Another example is if a
monitoring system data feed stops due to a permissions issue, it could generate an erroneous
problem report, leading to wasted time chasing the problem at the plant.

Conclusion
A distributed condition monitoring system leveraging cloud computing and today’s PdM
instrumentation is far more flexible and functional than prior options. It heightens the breadth
and depth of expertise accessible to an organization; supports transparent communications
and collaboration; improves skills application; and helps to control costs and mitigate safety
concerns. It pulls the curtain back to expose the true value of predictive and condition
monitoring.

Not only will it ease the rollout of PdM where such programs do not already exist, but it
is a logical upgrade and enhancement for plants running traditional, in-house programs. Any
number of predictable or unpredictable events can upend a condition monitoring initiative;
analysts may leave, funding may be cut, or a program expansion or new tools beyond
current capabilities may be desired. No other approach can ease these transitions as well as
distributed PdM.

Monitoring magic | 43
< Previous

Sponsored by

Azima DLI World Headquarters

300 TradeCenter, Suite 4610 
Woburn, MA 01801 USA
Toll-free U.S. 800-482-2290
International (+1) 781-938-0707
www.AzimaDLI.com

Produced by

555 W. Pierce Road, Suite 301, Itasca, IL 60143

630-467-1300 | www.PlantServices.com