Improving maintenance operations with data analytics

Increase reliability by planning and integrating analytics using Big Data that’s already being collected during maintenance operations. See five steps to implementing data analytics in maintenance operations.

By Aaron Merkin March 7, 2022
Courtesy: Fluke Reliability Solutions

 

Learning Objectives

  • Data analytics is the analysis of raw data to make informed decisions.
  • Advanced data analytics, which takes advantage of Big Data, can help manufacturers make better decisions and improve operations.
  • An asset criticality analysis can help a company find out where to start improving analytics and build a program from there.

Industrial organizations have used data analysis, trending, graphing and other visualization techniques since people started recording readings from machinery. While data analytics remains a constant, not many maintenance managers fully appreciate what it is, how it impacts operations or how it will shape Industry 4.0.

What is data analytics in maintenance?

Data analytics is the analysis of raw data to make informed decisions. That’s it. Current technology has been doing this for decades. What’s changing now is the amount of data collected and who, or “what” in this case, is doing the analysis.

Historically, technicians have collected data while managers or experts analyze those inputs and draw conclusions. These two actions get more digitized than before with each iteration of advanced technology and software.
Data analytics isn’t just important to maintenance today; it’s also key to the future. The next evolution in maintenance strategy will use prescriptive analytics, where software collects and analyzes data and also gives recommendations for maintenance tasks that might not otherwise be performed. In this world of “prescriptive maintenance,” advanced artificial intelligence (AI) and machine learning (ML) software will help decide what actions to take and when.

Few have a true AI maintenance software for practical purposes. Many companies are racing to get there.

However, it’s still a pipe dream on most industrial shop floors. Operations often run on manual data readings while the advanced few use wireless sensors and look toward a more automated future.

Figure 1: Maintenance strategies are progressing toward prescriptive analytics, where software will not only collect and analyze data, but also offer recommendations. Courtesy: Fluke Reliability Solutions

Figure 1: Maintenance strategies are progressing toward prescriptive analytics, where software will not only collect and analyze data, but also offer recommendations. Courtesy: Fluke Reliability Solutions

Manual vs. automated data analytics

Manual data analytics is a lot of playing with data and looking at squiggly lines. Plenty of industrial operations offer data analytics services, alleviating managers of the hassle data sifting, source verification and determining which data needs expert analysis.

Whether you have the in-house expertise for manual data analysis, or prefer to use sophisticated automated analysis, the basis of it all is in the large amounts of data generated from condition monitoring sensors and controls.

Big Data and data analytics for manufacturing

Big Data is any large or complex data set. In the maintenance world, it includes industrial measurements, operational data and wireless sensor readings. However, gathering and storing all this data isn’t the point. The information must be extracted and leveraged to be useful.

Data analytics is the key to unlocking information from Big Data. Expert analysts can extract meaning from a seemingly incomprehensible series of values and codes. As Industry 4.0 continues to revolutionize maintenance and repair operations, this analysis will transform into intelligent software capabilities the beginnings of which are already here today.

While AI data analysis is still in the future for many, current maintenance software systems are leveraging more and more data to assist maintenance teams and augment easily automated tasks.

Industrial technology and data analytics

Industrial data sources include operations control data such as supervisory control and data acquisition (SCADA), programmable logic controller (PLC) systems, building management systems, integrated or third-party sensors, technicians with connected tools, and more. With the growing adoption of IIoT sensors on assets, Big Data comes from more sources than ever. Thermographic tools also can be used to take readings from multiple assets. Vibration sensors conduct continuous condition monitoring and can sense problems like a misaligned motor shaft. Data from a technician spot-check using a handheld tool can be sent to the cloud immediately. Software can make inferences from fusing data sources into a comprehensive picture.

Figure 2: Today’s wireless sensors are helping pave the way for the future, where condition monitoring data will feed into AI-powered software. Courtesy: Fluke Reliability Solutions

Figure 2: Today’s wireless sensors are helping pave the way for the future, where condition monitoring data will feed into AI-powered software. Courtesy: Fluke Reliability Solutions

Analyzing industrial data

A commercial cheese manufacturer recently celebrated completing a multi-million-dollar expansion, increasing capacity by 25%. A lot of new equipment was coming in and management knew they’d need to monitor assets for uptime.

The company’s manufacturing utility process engineer said the team knew they needed to maintain a flow of product through the facility. They also wanted to know right away about any problems with the equipment.

They used wireless vibration sensors to upload constant readings to the cloud and analysis software to conduct vibration monitoring for the most common faults. That data provides steady insights to the maintenance team such as asset condition status, event information, warnings and more.

Five steps to implementing data analytics in maintenance operations

The path to the data analytics-enhanced future isn’t the same for everyone. Some already have reliability-centered maintenance (RCM) ingrained into operations. Others are starting their reliability journey and need to get the basics down before digging into Industry 4.0. However, everyone can benefit from these steps regardless of where they are on the path.

Step 1: Complete an asset criticality analysis

This manual bit of analysis guides teams in prioritizing asset health and maintenance on a hierarchy of importance. Teams grade each asset by its use within the organization, not necessarily its standardized use in the industry, as well as the business impact when it fails.

The asset criticality analysis also informs teams on which assets are prime candidates for condition monitoring and screening, providing analytics sources.

Step 2: Plan a pilot program

Like with most technology deployments or process changes, it’s good to start with a small set of assets to glean insights from. During asset criticality analysis, equipment crucial to day-to-day operations will be identified. Start condition monitoring on these more critical assets, generating the required component for data analytics (manual or automated).

Step 3: Launch program

Launching a program isn’t one and done; the plan will be refined during deployment to make sure it fits maintenance and operational needs. If a process or automation isn’t working correctly, refine and gather more data. Don’t get discouraged, either. Too many organizations drop a pilot program because it isn’t giving them what they thought they wanted. Instead, think like the Marine Corps – Improvise, adapt and overcome.

Step 4: Review results with leadership

The pilot program’s launch isn’t the end. With the data in hand, users can prove the point for further program expansion to leadership. They also may have suggestions based on years of business management and process change experience. Prove to them the program is sound and expandable with data.

Step 5: Grow the data analytics program

Once leadership has blessed the project, return to the asset criticality analysis to determine where to widen the condition monitoring program. Growth can be in the same facility, between facilities, or even between different nations.

It also can be helpful to refine existing setups to gain better data.

Growing the data analytics program also means testing new sources for industrial data. Sensors, handheld tools, equipment-integrated SCADA and PLC systems and other resources can be fused, improving analytics in the process. While vibration monitoring is a great starting point for new programs, thermal imaging, oil analysis and other condition-based maintenance (CBM) resources also are useful.

Figure 3: Data analysis and subsequent decisions making has usually fallen to managers or experts. New technology is digitizing that process more than ever before. Courtesy: Fluke Reliability Solutions

Figure 3: Data analysis and subsequent decisions making has usually fallen to managers or experts. New technology is digitizing that process more than ever before. Courtesy: Fluke Reliability Solutions

Data analytics and the future of industrial AI

The principles above lead to quality data analytics and are a foundation for future technology and software. They’re also part of a RCM program. Solutions should be integrated with the cloud, bringing Industry 4.0 into operations.

Now is the time to lay the data analytics groundwork for the coming AI/ML age. Those who accomplish their data analytics goals today will have everything they need to integrate emerging technology when it’s available.

Aaron Merkin is chief technology officer (CTO) of Fluke Reliability Solutions. Edited by Chris Vavra, web content manager, Control Engineering, CFE Media and Technology, cvavra@cfemedia.com.

MORE ANSWERS

Keywords: Data analytics, Big Data

CONSIDER THIS

How can data analytics and Big Data improve your facility?


Aaron Merkin
Author Bio: Aaron Merkin is chief technology officer (CTO) of Fluke Reliability Solutions. His responsibilities include developing and executing IIoT strategy and leading the technology team in the continued creation of innovative solutions for customers. Merkin brings more than two decades of experience developing enterprise software across a variety of industries and markets, including roles at IBM, Dell, ABB, Aclara (now Hubbell), and Honeywell. This includes positions as the CTO of ABB Enterprise Software, CTO of Aclara, and most recently, CTO of Honeywell Connected Industrial. Merkin holds a Master’s degree in Computer Science and a Bachelor’s degree in Mathematics.