Condition Monitoring Condition-based Maintenance Predictive Maintenance Reliability

Does 4.0-Based PdM Match The Hype?

EP Editorial Staff | May 14, 2020

The real magic of modern PdM solutions is in the software that interprets the raw sensor data using machine-learning (ML) algorithms that identify trends and estimate the remaining useful life of assets. Photo: Getty Images

Industry 4.0 technology provides a path to effective predictive and prescriptive maintenance practices.

By Justin Lesley, Motion Industries and Chantel Massie, Rexnord Corp.

Industrial maintenance and reliability practices have evolved since the beginning of the first industrial revolution in the late 1700s. Today, we’re well into the fourth industrial revolution (Industry 4.0) in which PdM (predictive maintenance) solution providers are promising a panacea of reliability.

Their vision for the “Factory of the Future” includes automatically generated, digital to-do lists that are created months in advance of required action. The expectation is that these lists will virtually eliminate the need for manual routes and time-based PMs by pointing reliability and maintenance professionals directly to assets that need attention and prescribing in detail which problems need to be addressed and when.

That’s not all. The software that generated these lists will also indicate the remaining useful life of each asset and even make recommendations about whether failing assets should be repaired or replaced, based on factors such as warranty, time in service, and repair history. The software may also automatically order parts so that they arrive at the factory just in time for service.

That is the vision, but what is the reality today and what will it take for solution providers to reach their goals?

Simple vs. Advanced

PdM solutions are in the midst of an evolutionary cycle of progress in the marketplace. Some PdM providers are just getting started on their development journey, which means focusing on the hardware used to generate machine health data. The most rudimentary solutions feature sensors that simply pipe raw data into a PLC near the machine using a hard-wired connection.

More-advanced solutions wirelessly transmit sensor data and incorporate some form of statistical analysis, such as machine learning, that translates raw sensor data into more useful insights.

The compute and storage functions associated with these advanced solutions could be executed locally using a network edge device (industrial computer) or at the cloud level, depending on system architecture and asset connectivity preferences.

Data collection is the foundation of any PdM solution and there is no shortage of opinions surrounding what data and how much is required to accurately assess an asset’s health or predict impending failure.

Reliability professionals at a potash mine in Saskatchewan, Canada, installed a Rexnord Smart Condition Monitoring System to monitor critical health of their Falk gear drive. The Rexnord Edge Device collects sensor data and computes health condition. The green Andon light indicates a healthy gear drive.

Data analysis

In any case, modern PdM solutions require continuous, consistent data collected directly from the monitored asset to record KPIs (key performance indicators) over time. But does sensor data need to be live streamed so that KPIs are tracked every second of every day, or can we poll sensors once an hour or only a few times daily? What about the type of data we collect? Is vibration all we really need to assess the performance of rotating equipment? Do we also need temperature, oil condition, electrical current, motor speed, and so forth?

These questions will be debated by solution providers and reliability practitioners for decades to come until successful track records are established for PdM accuracy across equipment categories.

Once sensors are integrated and data is flowing, the real magic of modern PdM solutions can materialize. That magic is in the software that interprets the raw sensor data using machine-learning (ML) algorithms that identify trends and link patterns to previously recorded outcomes. For instance, certain vibration signatures may be associated with known component failure points, e.g., bearing degradation versus shaft misalignment, through historical research. This is referred to as prescriptive maintenance (RxM) and is where established solution providers have an advantage over new PdM players who may only be able to detect change generically. In the same way, the remaining useful life of assets can be estimated by correlating data trends and patterns with historical outcomes.

Another strategy for interpreting machine health data is to use published ISO standards. Opponents of this method contend that ISO standards cannot possibly be relevant to the spectrum of PdM applications found throughout the industrial world. The reality lies in between. ISO standards are a good indicator of overall machine health but should be combined with asset manufacturer expertise to drive down specific repair recommendations.

IS Direct Data Enough?

So far, we have only discussed data collected directly from the asset in question. There are many other inputs that are relevant to machine performance. Consider the effects of environmental factors such as ambient air temperature, humidity, and atmospheric pressure. Additionally, the health and performance of auxiliary equipment associated with the asset in question will have an effect.

For instance, a misalignment at one end of a drive shaft could produce elevated vibration at the motor on the other end. As PdM solutions become more advanced, they will begin to incorporate these indirect data inputs into their predictive models, adding additional intelligence and accuracy.

The next frontier of advanced maintenance and reliability technology will involve the industrial supply chain. Repair or replace decisions are relatively simple to make after detecting a pending asset failure especially if the parts are readily available and/or conveniently stored on a spares shelf. Decisions change considerably when the parts have a multi-month lead time.

As PdM solution providers begin to work more closely with supply-chain partners, further inputs such as lead time, warranty coverage, time in service, and repair history will help reliability professionals make the decision to repair an asset or replace it with a new unit for maximum efficiency.

Start Small, Scale Fast

It can be overwhelming to develop a comprehensive PdM strategy for an entire facility. Solution providers understand this and many of them offer “starter kits” to help users test their solution without making a big commitment. These kits generally offer a few sensors and the additional hardware required to transmit and/or display data. Many starter kits use a local or “on prem” (premises) architecture with the option to upgrade to a connected or cloud-based architecture when scaling up. Once you are comfortable with a solution (users are trained, value is proven, and sensor installation is understood), it’s time to deploy more sensors and scale up.

Even if you don’t initially have a holistic PdM strategy, it’s a good idea to consider how many solutions it would take to monitor all of the critical assets in your facility and how the data they create will be used.

As you begin to chart the course ahead, you will want to build a list of all the maintainable assets in your facility and rank them by criticality. It would also be a good idea to note which assets are easy to replace versus those that were custom made, have long lead times, or are difficult to replace.

You also want to create a framework for ROI expectations or requirements for funding/approval purposes. ROI calculations for PdM solutions can be tricky as you will need to include “hard dollars” and “soft dollars.” End users should consider consulting an experienced third party when exploring predictive-maintenance solutions and the overall strategy for putting them to use in your organization.

Take Advantage Now

Despite the fact that PdM solutions aren’t perfect, they provide valuable machine-health insights and will only improve with time. Solution providers are documenting success stories every day and will document more and more as hardware and software improvements are made and PdM is embraced as mainstream.

Depending on the size of your company and its technology-adoption maturity, you may have vastly different resources available to leverage the benefits of today’s predictive-maintenance solutions. If your organization is not already putting these tools to use, it’s in your best interest to begin your PdM journey today. EP

Justin Lesley, Industry 4.0 Innovation Manager at Motion Industries., Birmingham, AL (motionindustries.com), directs IIoT strategy and partnerships related to the MRO industry. His career centers on operational efficiency supported by his Lean Manufacturing and Six Sigma certifications, combined with his engineering credentials.

Chantel Massie, Software Product Manager Digital Solutions at Rexnord Corp., Milwaukee (rexnord.com), manages the software design for Rexnord’s Smart Condition Monitoring Systems (SMCS) portal and mobile app. Massie began her career as a software engineer in the U.S. Air Force.

FEATURED VIDEO

CURRENT ISSUE

View Comments

Sign up for insights, trends, & developments in
  • Machinery Solutions
  • Maintenance & Reliability Solutions
  • Energy Efficiency
Return to top