The Future of IIoT Predictive

Maintenance
A Combined Study by Emory University and Presenso
Introduction
Summary of Findings
1.0 The Current State of Predictive Maintenance
2.0 The Outlook for Industry 4.0 Maintenance Technologies
3.0 Perspectives on IIoT Predictive Maintenance
4.0 Implementation of IIoT for Preventive Maintenance
5.0 Impact of IIoT Predictive Maintenance
6.0 Recommendations
Appendix A – Additional Research Data
Introduction
In April 2013, at the Hanover Messe conference in Germany,
the guiding principles of Industrie 4.0 or Industry 4.0 were
released. Over the past five years, Industry 4.0 has moved from
German government policy to executive-level strategy across
the globe. Today we are in the third wave: active
implementation.
Industry 4.0 is considered the fourth Industrial Revolution, and
industry analysts have forecasted a significant and broad
economic impact. The application of the Industrial Internet of
Things (IIoT), Artificial Intelligence and Machine Learning to
industrial maintenance or Predictive Maintenance 4.0 is a core
element of Industry 4.0.
The Emory University Future of IIoT Predicti ve Maintenance
research study was designed to identify the gaps between the
high-level strategic and business drivers of change and the
reality of implementation. For this purpose, we interviewed
Maintenance and Reliability professionals responsible for
Predictive Maintenance in their organizations.
This study’s goal is to provide a field perspective on the
following topics:
 The current state of Predictive Maintenance in industrial plants
 The level of satisfaction with current Predictive Maintenance systems
 IIoT Maintenance systems most likely to be adopted within the next
five years
 The extent to which the Digital Twin is likely to be deployed
 The disconnect between executives and O&M professionals
responsible for implementation
 Reasons for delays in investments in new IIoT Predictive Maintenance
solutions
 Factors blocking the implementation of IIoT Predictive Maintenance
solutions
 The likely impact of IIoT Predictive Maintenance on current O&M
practices
For this study, 103 O&M professionals were surveyed ac ross
Europe, North America, and Asia Pacific. A combination of
quantitative research (online survey) and in -depth interviews
were used. In addition, feedback was solicited in public forums
in Asset Maintenance LinkedIn groups.
Six Emory University students participated in the research and
writing of this report: Arnav Jalan and Nathan Brooks (project
co-leads) and Dilsher Dhupia, Ian Goldstein, Hannah Laifer and
Sabiha Officewala.
Summary of Findings
In 2017 and 2018 alone, significant advances in cognitiv e
analytics have been applied to the discipline of Predictive
Maintenance. In parallel, Industry 4.0 has been embraced by
the senior management of worldwide industrial facilities.
Our research indicates a growing chasm between the potential
for PdM4.0 and the reality in today’s industrial plants. We found
no urgency to upgrade legacy Maintenance and Reliability
practices from the 1970’s and 1980’s. Microsoft Excel is still the
default analytics tool.
Concerns that are raised about PdM4.0 and Maintenance 4.0
stem from practical considerations regarding the feasibility of
deployment and the lack of resources. O&M professionals view
PdM4.0 positively but expect an incremental change in the form
of improvements to existing systems and processes.
This report analyzes the following topics.
1. Current State of Predictive Maintenance: IIoT for Predictive
Maintenance is still in its infancy. Despite the promise of PdM4.0, there
is little discontent with current Predictive Maintenance systems.
Traditional Predictive Maint enance, including vibration monitoring, oil
residue analysis, and thermal imaging, still dominates, and manual
statistical modeling such as Excel has not been replaced by more
advanced technologies.
2. Outlook for Industry 4.0 Maintenance Technologies: O&M
professionals expect that Automated Failure Reporting and Automated
Repair Scheduling are most likely to be widely adopted over the next
five years. There are limited expectations for the deployment of
Robotics Assisted Repair and Drone/Robotics Assisted Ins pection. The
 Digital Twin concept is not widely known by O&M professionals and is
not forecast to play a major role in industrial plants within the next five
years.
3. Perspectives on IIoT Predictive Maintenance: O&M professionals
are less enthusiastic about IIoT for Predictive Maintenance than is
senior management. Part of this is attributed to the “hype” that
resonates less with the Maintenance and Reliability workers who are
responsible for implementation. Almost 40% of respondents in the
online survey cite a lack of IIoT strategy as a reason for delays in
adoption. In the long term, there is an expectation that the perceived
ROI from IIoT Predictive Analytics will justify the expenditures.
4. Implementation of IIoT for Preventive Maintenance: The most
significant inhibitor of IIoT for Predictive Maintenance deployment is a
skill shortage of Big Data Scientists and a lack of understanding of
Industry 4.0. The complexity of software and access to sensor data are
considered less significant factors affecting stall ed deployment.
5. Impact of IIoT Predictive Maintenance: Overall, O&M professionals
have a positive view ofIoT Predictive Maintenance. Improvements to
Operational Equipment Efficiency (OEE) are widely expected.
Furthermore, most survey respondents believe tha t utilizing and
analyzing the data in real-time will allow for better decision making.
From an organizational perspective, there are only limited concerns
that the roles and responsibilities of O&M professionals will change. In
general, there was not much support for the outlook that IIoT
Predictive Maintenance will force the convergence of Information
Technology and Operational Technology organizations.

1.0 The Current State of Predictive Maintenance

IIoT for Predictive Maintenance is still in its early stage and
O&M professionals see little impetus for adoption. The promise
of PdM4.0 is recognized but little discontent exists with current
Predictive Maintenance systems.
Traditional PdM including vibration monitoring, oil residue
analysis and thermal imaging has not been displaced. Microsoft
Excel is considered the default for predictive modeling.
1.1 Assessment of Current Predictive Maintenance Practices
Maintenance 4.0 may be a core element of an executive’s
strategic vision. It could be included in an i ndustrial plant’s
technology roadmap. However, from a practical perspective, it
is not widely deployed.
Today, the most common processes for Predictive Maintenance
are traditional methods such as vibration monitoring, oil residue
analysis, and thermal imaging. Industrial plants still rely on
manual statistical modelings, such as Microsoft Excel for
Predictive Analytics (44% of respondents), far more than they
do on advanced statistical modeling (23%) and Machine
Learning (12%).

Why is Excel so prevalent? The reason is basic. Reliability &

Maintenance professionals already use Microsoft Excel as
opposed to other tools, such as MATLAB. Familiarity with Excel
creates momentum; people start with a small table that grows
exponentially and becomes the default du e to a lack of viable
alternatives.
O&M Practitioner Insight
Q. Manual statistical modeling is the most common method of
Predictive Analytics (e.g., Excel). Do you think this is likely to
change in the next few years? Why or why not?
Jack R. Nicholas, Jr., P.E., CMRP, CRL, IAMC
Statistical modeling is rarely used in predictive analytics,
particularly at the local level. The most common method
is Regression Analysis in various forms.
The second most common method is Pattern Recognition (e.g.,
for Infrared thermography and off -line and online motor circuit
analysis) or visual analysis/recognition (of faults as indicated in
images or graphical presentation of data).
The next most common method is Relative Comparison (from one
machine to another or a group of machines).
After that comes Tests Against Limits or Ranges. This is related
to regression analysis in that, while a trend may be developed
and a sudden jump in parametric data occurs to a level above
an alert or alarm presents, action will be based on the latter
and not on the trend.
Statistical Process Control (SPC) is practiced in many programs
and is common in SCADA systems. This is a visual/graphical
analysis method that can be practiced by mill deck personnel
and requires only simple mathematical and plotting skills if
done manually (e.g., number of readings over time in an
adverse direction towards upper and/or lower control limits).
The most powerful method is called Correlation Analysis. This
involves using analysis results from two different technologies
taken at the same time or under the same conditions of
operation to confirm a fault or from the same family of
technologies (e.g., Vibration Analysis and Shock Pulse
Analysis) sequentially in time to track the same fault.
When advanced analytics are used (which is currently most
effectively done using IIoT/Cloud Computing Technology), five
(5) basic methods are prevalent at the present time. These are
Visual Analysis (e.g., for outliers), Clustering Analysis (e.g.,
Density-Based Spatial, Hierarchical or other common methods
such as k-Means), Data Mining (e.g., for Differentiating
Analysis or Association Rule-based), Time Series Analysis and
Statistical Analysis (which includes Regression, such as Partial
Least Square and Multiple Linear Analysis).
Usman Mustafa Syed
Manual statistical modeling will gradually be replaced by
Machine Learning & AI-based Predictive Analytics. With the
current rate of technology advancement in the field of Machine
Learning, AI, and Sensors & Connectivity, we will be entering
an era of off-the-shelf Predictive Analytics tools that could be
deployed economically and rapidly.

Reflecting an acceptance of the status quo, most survey

respondents are relatively satisfied with the current options for
Predictive Maintenance systems. Although only 2% were Very
Satisfied with current systems, a larger percentage (46%) were
Somewhat Satisfied and an additional 28% were Neutral. A
minority of only 8% were Very Dissatisfied.
With respect to rules-based SCADA systems, 30% of survey
respondents recognized that limitations are widely recognized,
whereas 38% were neutral.
1.1.2 Review of Analyst & Industry Research
The following reports were considered as part of our evaluation:

Analyst Title Date

Plant Engineering Plant Engineering 2018 Maintenance Study 2018

Survey of technology use for PdM indicated the use of

spreadsheets/schedules (55%), Computerized Maintenance
Management Systems (CMMS, 53%) and paper records of
maintenance reports (44%).

Analyst Title Date

PwC Predictive Maintenance 4.0: Predict the Unpredictable 2017

Survey of technology use for PdM indicated MS Excel/Access:

67%, WIFI: 34%, data warehouse: 18%, statistical software:
18%, conditioning monitoring software: 40%, cloud: 13%, data
software 33%, mobile networks: 20%, IoT: 14%.
1.2 Metrics to Evaluate Maintenance Solutions
Unsurprisingly, O&M professionals view operational
efficiencies/labor cost savings as the primary metric for
evaluating the impact of Predictive Maintenance solutions.
However, there is also a recognition of the financial benefits of
a proactive approach to maintenance. As depicted in Chart 3, in
terms of the most important metric for a predictive maintenance
solution, savings from lower downtime were rated almost as
high as were operational efficiencies (7.3/9.0 versus 7.4/9.0).
Furthermore, revenue from increased uptime (6.9/9.0) was
ranked even higher than improvements in spare parts and other
logistics management (6.3/9.0).
Even if O&M professionals do not maintain the same
enthusiasm for PdM4.0 that senior management does, they
understand the strategic value of increased uptime, higher
production yield rates, and lower asset downtime.

2.0 The Outlook for Industry 4.0 Maintenance

Technologies
The primary change expected from Industry 4.0 maintenance
technologies is improvements to automated workflows and
processes. In contrast, there is a lesser likelihood for the
deployment of robotics or drone-assisted inspection.
The Digital Twin concept has not attracted much attention. Most
O&M professionals do not expect to see significant deployment
within the next five years.
2.1 Adoption Rates of Industry 4.0 Technologies
The primary change forecast is in the incremental adoption of
the automation of processes and workflows. This is not
surprising given that Computerized Maintenance Management
System (or Software) or CMMS has been prevalent since the
early 1990’s.
A drill-down into the expected adoption of Automated Failure
Reporting (Chart 4.1) indicates that 45% of respondents expect
full deployment while an additional 22% expect a scenario of
mostly deployment.

With respect to the adoption of Robotics Assisted Repair, 35%

do not expect adoption while an additional 17% expect only
somewhat adoption. (See Chart 4.2.)
O&M Practitioner Insight
Q. Do you think there will be a migration to IIoT Predictive
Maintenance in the next few years? Why or why not?
Fred Schenkelberg
No. Gathering more data does not solve anything. We already don’t
use the data we have nor know what to do to analyze and use the data
today… until we learn how to gather the right data to help solve real
problems (which may or may not include PdM), then IIoT is not useful.
We must get the basics right first.
Jørgen Grythe
Yes. Most organizations have at least one skilled maintenance
person who can step onto the factory floor and sense, through
sight, sound, smell, vibration and temperature, the conditions in
the factory.
These human condition monitoring experts are retiring. There
are not enough people to replace them. So, I think we are
increasingly seeing a centralization of monitoring, either by
remote sensing (not completely automatic, but you don’t ha ve
to be on-site to monitor the plant) or/and by automated
condition monitoring with machine learning algorithms.
2.1.1 Review of Analyst & Industry Research
The following report was considered as part of our evaluation:

Analyst Title Date

McKinsey & Company Where Machines Could Replace Humans – And Where They Can’t
(Yet)
2016

Following is a summary of the research:

 Automation will eliminate very few occupations in the next
decade. However, it will affect portions of almost all jobs based
on the type of work they entail.
 Predictable physical activities account for 1/3 of workers’ overall time
in the manufacturing sector. Based on technical considerations alone,
manufacturing is the second most readily automatable sector (after
services).
 Ninety percent of the tasks of welders, cutters, solderers, and brazers
has the technical potential for automation.
 Potentially59% of all manufacturing tasks are susceptible to
automation.
2.2 Adoption Rates of the Digital Twin
In contrast to some analyst reports, the deployment outlook for
the Digital Twin is relatively modest. Sixty-one percent of O&M
professionals claim that they are not even familiar with the
Digital Twin concept. An additional 20% of respondents do not
expect that the Digital Twin will be deployed within five years.
Only 4% of respondents expect complete deployment. (See
Chart 5.)

O&M Practitioner Insight

Q. The Digital Twin is not yet widely used. Why is this?
Georgi Kirilov
Even if there were support from the management, to implement
such modeling, it would require local (facility) knowledge. Here,
IT and consultants are not enough. The main drivers for this
change would be reliability engineers, but incorporating their
knowledge into software would threaten their jobs.
Usman Mustafa Syed
The Digital Twin requires a lot of expertise, which is scarce in
the market; this is a big hurdle.
The bigger hurdle is the fact that GE has so aggressively
associated itself with the name and concept of Digital Twin that
it has somehow become a “pet brand” for it. This leads other
competitors to avoid it. Also, the end users of non -GE assets
tend not to consider it.
Jack R. Nicholas, Jr., P.E., CMRP, CRL, IAMC
Developing an accurate digital twin is expensive and time-consuming.
This is because even in fairly new plants, configuration control is rarely
practiced to the extent it should be. Drawings are usually out of date
(even for newly commissioned assets) and specifications for
performance are imprecisely documented over time and are, after a
few operating cycles, ignored [because] assets are driven beyond the
original design limits. To construct an accurate digital twin requires
extensive research into current performance and condition
requirements, not just referral to original plant specs.
Keeping a digital twin accurate over a life cycle is also
expensive and requires at least periodic attention [from] data
scientists, modeling specialists and business needs analysts –
skills scarce in most organizations and usually available at high
cost only from large service providers such as IBM, Accenture,
and GE. Any plant modification affecting performance or
required reliability conditions requires a change to the digital
twin model and related algorithms.
Added to this is the fact that only a small fraction of available
data is analyzed, and the data is rarely linked to functional
failure modes (even if they are known by personnel who should
be aware of them).
Added to that is the fact that brownfield plants haven’t provided
connectivity from critical assets even with wireless technology
to analysis centers in-plant or via the IIoT because the work to
identify return on investment is seldom done and justified in a
comprehensive plan containing all steps that must be
accomplished to make cost-effective use of both analysis at the
edge (near the machine), in the plant (before any firewalls) and
in the Cloud.
In today’s cybersecurity climate, the safest approach is to have
information and control functions on separate networks, the first
allowing connectivity to the Cloud and the latter totally isolated
from any external digital hacking. This, too, adds to cost.
2.2.1 Review of Analyst & Industry Research
The following reports were consider ed as part of our evaluation:

Analyst Title Date

 Deloitte Industry 4.0 and the Digital Twin 2017

Gartner Prepare for the Impact of Digital Twins 2017

Following is a summary of the research:

 Half of large industrial companies will use Digital Twins by 2021. The
result is an expected 10% improvement in “effectiveness.”
 ROI is expected to vary widely and will be based on monetization
models that drive them.
 “CIOs will need to work with business leaders to develop economic and
business models that consider the benef its in light of the development
costs, as well as ongoing digital twin maintenance
requirements.” Alfonso Velosa, VP of Research, Gartner

3.0 Perspectives on IIoT Predictive Maintenance

There is a disconnect between the views of O&M professionals
and senior executives. Executives recognize the potential
represented by IIoT for Predictive Maintenance but the people
responsible for implementation are less sanguine.
A lack of IIoT strategy is viewed as a significant factor in the
delay of PdM4.0 adoption.
3.1 Attitudes Toward IIoT Predictive Maintenance
Senior executives are more likely to recognize the potential of
IIOT Predictive Analytics than are facility staff.
Many O&M professionals still view PdM4.0 as over-hyped. In
response to the statement “The hype of industry 4.0 and IIoT is
exaggerated,” over 40% are Neutral. Twenty -three percent of
online survey participants Agree with the notion and an
additional 9% Strongly Agree. In contrast, only 6% Strongly
Disagree and 18% Disagree. (See Chart 6.4, Appendix A.)
Almost 50% of O&M professionals surveyed say they Strongly
Disagree or Disagree with the notion that facility staff
recognizes the potential. An additional 15% are Neutral.
This contrasts with O&M professionals’ perception of senior
executive. Fifty percent of online survey respondents Strongly
Agee or Agree with the notion that senior executives recognize
the potential of IIoT Predictive Analytics. Only 14% of
respondents Strongly Disagree. (See Charts 6.1 and 6.5,
Appendix A.)
Thirty-eight percent of online survey respondents indicate that
they Strongly Agree or Agree with the statement that there are
delays in new investment until a decision is reached related to
an IOT strategy. Only 11% of respondents Strongly Disagree
with this statement. (See Chart 6.2, Appendix A.)

O&M Practitioner Insight

Q. What is the reason for the disconnect between O&M
professionals and senior management?
David Naus
There is a huge disconnect with upper management. I have talked
with many decision makers who don’t care what happens on the
maintenance side of their business, which is a huge asset to
them and cost.
They will never take the time to learn how they could maintain
their machines better and cut costs in doing it at the same time.
They will always pass on to an employee in maintenance who is
not qualified, and they won’t take the time to even discuss it.
Howard W Penrose, Ph.D., CMRP
I’m seeing something a little different. Many companies and
maintenance organizations are championing IIoT devices, and
some that you may not realize are IIoT, such as power
monitoring, protective devices, and monitoring networks. When
the device applications are presented at conferences, there is
normally a significant turnout. However, [for discussions of] things
like proper setup or, worse, security related to connected devices, the
attendance is lacking. People see these issues as “someone else’s
problem.”
Carlos De Leon
As Sundeep Sanghavi said in his article “Cognitive Predictive
Maintenance Managed Data Flood”: The IIoT has the potential to
transform this industry in a dramatic way, but this will be possible only
if the volume of data is effectively leveraged by enhancing sensor
connectivity.
3.1.1 Review of Analyst & Industry Research
The following report was considered as part of our evaluation:
 Analyst Title Date

MIT Sloan/Boston Reshaping Business with Artificial Intelligence:

Consulting Group 2017
Closing the Gap Between Ambition and Action

Following is a summary of the research:

 Seventy-five percent of executives expect Artificial Intelligence to
enable their organizations to move into new businesses.
 Nearly 85% think that AI will help them obtain or sustain a competitive
advantage. However, only one in five has incorporated AI in some
offerings or processes.

4.0 Implementation of IIoT for Preventive

Maintenance
A skill shortage of Big Data Scientists and a lack of
understanding of Industry 4.0. are the two most significant
factors negatively affecting PdM4.0 deployment. O&M
professionals are significantly less concerned about the
complexity of software and access to sensor data.
4.1 Factors Impacting the Implementation of PdM4.0
For O&M professionals, the biggest factor preventing the
deployment of IIoT Predictive Analytics is a shortage of data
scientists (4.2/5.0). Forty-four percent of online survey
respondents say this skill shortage will have a Strong Impact.
An additional 37% indicate a Moderate Impact. (See Chart 7.9,
Appendix A.)
The second-highest ranked negative impact on deployment is a
lack of understanding of Industry 4.0 (3.8/ 5.0). Thirty-three
percent of online survey respondents view this factor as
Moderate, while an additional 30% indicate a Strong Impact.
(See Chart 7.8 in Appendix A.)
Interestingly, the two factors that are least likely to impact
deployment of IIoT Predictive Analytics are the complexity of
the software and the inability to access sensor data. Both of
these are ranked 3.2/5.0.
From a resource perspective, online survey participants are
more likely to believe that their organization has the core
infrastructure needed for IIoT Predictive Analytics (4.8/9.0) than
that they have sufficient data scientists to deploy IIoT
Predictive Analytics (3.3/9.0).

O&M Practitioner Insight

Q. Are the specific accelerators or blockers of IIoT Predictive
Maintenance in your industry?
Fred Schenkelberg
Not just IIoT PdM.
PdM, in general, requires paying attention to the equipment in a
systematic manner, identifying and modeling indicators, and
having funding/support to implement.
Furthermore, this approach doesn’t address every cause of
maintenance actions – for industries with a high percentage of
failures due to predicable failure mechanisms, using IIoT or not
first needs to prove to the organization that the approac h works
(adds value).
IIoT is a buzzword and may attract some in the market, yet unless they
do the basics first, it will be an expensive experiment with little chance
of success.
Jørgen Grythe
I cannot speak of the Predictive Maintenance industry as a
whole since this is huge. We, however, focus on acoustic
beamforming array and are using beamformed sound from
machinery to classify a healthy or error state of a machine.
The biggest blocker for us right now is that every project is
different, and there doesn’t seem to exist one universal black
box solution to cover a wide variety of customers. So, if I were
to generalize, I would say that it is difficult to really expand into
a large-scale system that works for everyone. Kind of like if the
iPhone had to be customized for each individual user; you can
imagine the cost that would impose on the product and the lack
of scalability.
Jack R. Nicholas, Jr., P.E., CMRP, CRL, IAMC
Current literature on Big Data, Advanced Analytics, Cloud
Computing Technology and related subjects spells out the need
for the merger of OT and IT, tying the mill deck to the
boardroom.
While this is desirable in many ways, it often requires a
significant change in culture in organizations with large if not
dominant IT influence on digital transition. Many C -level
executives (if not all by this time) are scared to death of a
career-ending data breach.
Cybersecurity in most companies is assigned to IT personnel as
a collateral duty and may be based on just a week or two of
training – totally inadequate in today’s cyberspace climate.
While methodologies such as reliability-centered maintenance
and total productive maintenance have been around for
decades, they still haven’t been adopted to the degree needed
for personnel in plants to have firm knowledge of failure modes
being experienced and [to] link them to data needed to provide
advanced warning of onset at stages where economic mitigation
can be planned and executed in an orderly fashion. Thus, there
is no way for them to justify adoption of IIoT predictive
maintenance.
Usman Mustafa Syed
In my opinion, many IIoT Service Providers (except for the big
brands, mostly OEMs) come with a background in IT, Software
and Business Analytics, with a bit of a disconnect from the
Industrial Engineering fraternity.
They don’t speak the same language as the engineers
managing the assets within different industries. This must
change. Furthermore, there are concerns regarding cost, ROI,
data security and leadership commitments in general.
Another important fact is that this wave of IIoT advancement
came up during generally uncertain times. Global economies
are sluggish right now, which has also impacted its success.
 But with the ongoing advancements and gradually improving
economic conditions, things will improve.
Anonymous survey respondent
I was at a seminar last week with some very influential
European organizations present. Some very telling facts were
presented about the IOT in manufacturing.
As the two days progressed, there was a distinct split in the
expectations of the data/tech guys and the “grease monkeys”
responsible for day-to-day production. It seems that I4.0 is a dream of
the tech guys that, at this team, the hardware cannot support.
Two key statistics stood out for me – there are 92% or 64
million machines in the world without network connectivity. The
other statistic was the average of the machines in Italy is 13
years. This would lead to a gulf in desire versus ability.
It is my opinion that the large companies (BMW, Siemens) will
have the resources to trial the new systems and machines and
assess the ability to adopt across production facilities.
Small companies will have the flexibility to adopt and adapt.
This would then leave the medium size to invest in lower cost solutions
and therefore experience [fewer] benefits from I4.0.
It’s clear that we will all be affected by IoT but in my opinion,
this is likely to be driven by a consumer “need” versus a
manufacturing readiness. It certainly will take longer than the
predicted five to eight years to adopt.
4.1.1 Review of Analyst & Industry Research
The following report was considered as part of our evaluation:

Analyst Title Date

PwC Industry 4.0: Building the DigitalEnterprise 2016

Following is a summary of the research:

The biggest challenges or inhibitors for building digital
operations capabilities were:
 Lack of a clear digital operations vision and support/leadership from
top management – 40%
 Unclear economic benefit and digital investment–38%
 High financial investment requirements – 36%
 Unresolved questions about data security and data privacy in
connection with the use of external data – 25%
 Insufficient talent – 25%

5.0 Impact of IIoT Predictive Maintenance

Overall, O&M professionals have a positive view of IoT
Predictive Maintenance.
In terms of change, most survey respondents do not expect that
employment levels will change due to PdM4.0 but they do
expect thatjob roles will change.
Improvements to Operational Equip ment Efficiency (OEE) are
widely expected. Furthermore, most survey respondents believe
that utilizing and analyzing the data in realtime will allow for
better decision making. From an organizational perspective,
there are only limited concerns that the ro les and
responsibilities of O&M professionals will change. In general,
there was not much support for the outlook that IIoT Predictive
Maintenance will force the convergence of Information
Technology and Operational Technology organizations.
5.1 Impact of IIoT Predictive Maintenance on Financial Results
Regarding the financial impact of PdM4.0, the strongest belief
that O&M professionals hold is that the perceived ROI from a
Predictive Analytics solution will justify the costs. Forty -eight
percent Agree with the statement that perceived ROI will justify
the cost, while an additional 11% Strongly Agree. Only 4% of
respondents Strongly Disagree with this statement.

5.2 Impact of IIoT Predictive Maintenance on Employment

Most O&M professionals do not think that employment levels
will change dramatically with the adoption of Industry 4.0
practices such as PdM4. Twenty-four percent of survey
respondents believe that employment levels will fall and only
6% believe they will rise. By far, the prevalent view is that
employment levels will stay the same but job roles and
functions will change (57% of respondents).
5.3 Impact of IIoT Predictive Maintenance on O&M Practices
Almost all O&M professionals expect that standard O&M
practices will change due to PdM4.0, although opinions are split
as to the extent of this change.
More than half (54%) of online survey respondents expect
moderate changes to standard O&M practices, while an
additional 5% expect no changes to practices. A minority of
respondents (41%) expect major changes to standard O&M
practices.

5.4 Overall Impact of IIoT Predictive Maintenance on Plant/Industry

Most O&M professionals expect that PdM4.0 will result in
positive change. There is a significant consensus that Overall
Equipment Efficiency (OEE) will increase. In fact, 54% of
respondents Strongly Agree with this statement, while an
additional 36% Agree. Similarly, regarding the statement that
 utilizing and analyzing data in real time will allow for better
decision making, 38% of respondents Agre e and 54% Strongly
Agree.
There is less support for the notion that PdM4.0 will force the
convergence of Information Technology and Operational
Technology.

O&M Practitioner Insight

Q. What changes do you see happening in the Reliability &
Maintenance discipline over the next five years?
Fred Schenkelberg
I see an increased emphasis on establishing consistent
processes, gathering and using data in a meaningful manner
(manual first, then, where useful, automating it).
I also expect atenfold increase in the collection of data that goes
under-utilized. Currently, many factories collect tons of data and there
is little or limited use of it. Adding more sensors adds more data, not
information.

6.0 Recommendations
6.1 Recommendations to Accelerate PdM4.0 and Industry 4.0
Deployment
The following high-level recommendations for accelerating the
deployment of IIoT Predictive Maintenance are provided based
on research in this report.
 Include O&M professionals in developing the organization’s PdM4.0
strategy. Not only does this help secure organization -wide buy-in, but
their inclusion can provide practical insight into adoption.
 Implement Proof of Concepts with multiple vendors and evaluate
results based on pre-defined criteria.
 Include O&M professionals in the process of creating functional and
specification documents, Requests for Proposals (RFPs) and Proof of
Concepts.
 Create Centers of Excellence for the Application of Machine Learning.
 Develop and nurture an Industry 4.0 vendor and solutions ecosystem
so that complementary technologies can be rapidly evaluated and
adopted when needed.
 Design a company-wide IIoT Technology Roadmap that is vendor -
neutral.
 Form a working group representing Operational Technology and
Information Technology to explore technological solutions.
 Develop operational metrics to measure the adoption of PdM4.0. These
should not only include deployment (number of assets covered), but
also relate to the enabling infrastructure (people, process,
technology,etc.).
 Include elements of Artificial Intelligence and Machine Learning in
O&M Skills Development programs.
6.1.1 Review of Analyst & Industry Research
The following reports were considered as part of our evaluation:

Analyst Title Date

McKinsey & Company Notes from the AI Frontier: Insights from Hundreds of Use Cases 2018

Following is an abstract of recommendations included in the

report:
 There are steps to be taken before embarking on pilots or PoC’s.
McKinsey suggests a “holistic approach”: creating a prioritized portfolio
of initiatives across the enterprise, including both AI and the wider
analytic/digital techniques.

Analyst Title Date

PwC Netherlands Predictive Maintenance 4.0: Predict the Unpredictable 2017

Following is an abstract of recommendations included in the

report:
 Set up an IIoT infrastructure including “choosing the right protocols for
wireless connectivity, data encryption and security.”
 Install feedback loops. The ML solution may provide insights into the
cause of machine asset failure. “Perhaps the PdM 4.0 business case
for a particular asset type needs to be re-evaluated: it may be more
expensive or yield worse returns than initially thought. Or the criticality
of assets may change over time and warrant new feasibility st udies.”

Appendix A – Additional Research Data

This section contains additional graphics not included in the
main report.
1.0 Attitudes Towards IIoT Predictive Maintenance
2.0 Factors Impacting the Deployment of IIoT Predictive Maintenance
3.0 Impact that IIoT Predictive Maintenance Will Have on Plant and/or
Industry