KEY CONCEPTS
• Industry 4.0 refers to the current revolution in the manufacturing and industrial sectors, characterized by automation and new digital technologies.
• While artificial intelligence (AI) is positioned to revolutionize the oil analysis industry, changes are happening incrementally. 
• Integrating Industry 4.0 technologies into an oil condition monitoring program can improve accuracy while increasing efficiency.

---

Meet the Presenter
This article is based on an Industry Insights presented by Jorge Alarcon on June 11, 2024, titled Artificial Intelligence - Integrating AI and Machine Learning with Oil Analysis Data. Industry Insights is a monthly virtual event series available for free to STLE members. These events provide a forum for members to collaborate with STLE industry peers and experts on a specific topic who give a brief presentation, then facilitate conversation by asking discussion questions and providing actionable takeaways. For more information, click here. 

Jorge Alarcon is global reliability manager for Pragma Reliability, where he specializes in condition monitoring, predictive maintenance and research. He holds degrees including bachelor of engineering industrial, master of science condition monitoring, master of science big data and business intelligence, master of science RAMS, master of business administration and is a doctoral candidate. Alarcon has worked in industries including power generation, manufacturing, wind energy and others. He has extensive knowledge of the predictive maintenance markets of Europe and the Americas and is focused on improving plant reliability through oil condition monitoring, lubrication best practices and digital transformation strategies. You can reach him at service@pragmatool.com.

Jorge Alarcon

---

The current technology environment is marked by unprecedented developments, including Industry 4.0, the Internet of Things (IoT), artificial intelligence (AI) and machine learning (ML). According to Jorge Alarcon, these developments have the lubrication industry positioned “at the cliff of transformation.” The merging of traditional oil analysis and Industry 4.0 technologies is set to redefine the industry, and Alarcon’s presentation explores the implications, opportunities and challenges presented as a result.

The interest in Industry 4.0 technologies is relatively recent. Although most of the concepts date back to the 1970s, Alarcon says, “After three years of dealing with the new concept of Industry 4.0, I was ready to present about AI and big data. It was 2015 at a conference related to reliability, and I only had three attendees at that time. One of those was my boss, and one was a tourist, so I basically had one attendee to my presentation. Nowadays, if you attend any AI presentation, you will have a lot of people because everyone’s paying attention to this new paradigm.”

What is Industry 4.0?
Industry 4.0 refers to the current revolution in the manufacturing and industrial sectors, characterized by automation and new digital technologies. The IoT, a network of devices and machines with embedded sensors and other software, is a key element of Industry 4.0. IoT allows the connected devices and machines to collect and share data automatically. 

According to Alarcon, the oil analysis industry is no exception to this new paradigm and is one of the latest innovators when it comes to AI. Industry 4.0 presents both opportunities and challenges to the oil analysis and lubrication industry. One of the first challenges is distinguishing true innovation from marketing. “We have been seeing a lot of these ‘new’ tools that are not really new: the same PDM tools, vibrational oil analysis, fluid analysis and thermal analysis. Marketing trends say that now we have an AI tool or a new tool based on AI, which, in some cases, is not true.”

Maintenance
When it comes to maintenance, Alarcon says, “We need to assume that we have an external demand, and we have to produce units of something to cover that external demand. We have our assets, and our assets are dependent on availability, reliability and maintainability.”

Alarcon describes four pillars of maintenance: predictive, preventative, corrective and prescriptive or proactive. He explains that oil analysis, as well as other lubrication tasks, can fall into any of these pillars, and that some industries operate at a very advanced level of maintenance while others have very poor practices. Reliability and maintainability are critical to a successful oil analysis program. Reliability means that the lubricant functions appropriately according to the application, while maintainability includes how easy it is to change the lubricant without leaving residue. Maintenance support goals also are critical, including proper storage, availability and cost effectiveness of a lubricant with the appropriate physical and chemical characteristics. “We want to design products—lubricants in this case—that can be easily maintained,” Alarcon says.

Applied to traditional oil analysis, following these principles can be expensive, both in time and money. According to Alarcon, AI offers a solution: “We can use virtual test benches and more volume of data.” He gives the example of analyzing a large volume of structured and unstructured data when selecting a gear oil. In a traditional method, the oil sample would be submitted to the lab, which would perform semi-automatic simulations and bench tests, then apply human knowledge to determine whether the oil is appropriate for the application. This might take months to complete. Using AI, virtual test benches allow the use of more data than what is collected during traditional testing. For gears, the standard test method is the FZG Test; some research and development labs are now using the traditional FZG Test alongside a virtual test. The conditions for speed, oil chemistry, load and temperature are the same in both tests.

AI sensors
Sensors and connectivity are definitive features of IoT and essential parts of Industry 4.0. Alarcon notes that the oil analysis industry has seen an increase in the use of sensors and maintenance software programs, particularly asset management software. Sensors currently on the market can be used for oil analysis to monitor particle count, viscosity, dielectric changes, chemical changes and water control. The sensors collect large amounts of data that can then be analyzed by AI programs much more quickly and with a higher level of sensitivity than traditional methods. For example, AI-powered anomaly detection algorithms can identify small variations in data that might go unnoticed by humans.¹ This can be critical in preventing a major failure and downtime. Alarcon says, “I’ve been working with many of these in the past years, and I can say that at some point sensors will probably replace the typical oil analysis in the lab.”

Alarcon shares an experience with working in a research and development (R&D) lab. The experience included an oil analysis lab wanting to launch a project that included a sensor. This was a bit unusual since implementing the sensor would potentially destroy part of the lab’s oil analysis business. However, the real results were different than expected; the sensor changed the way the lab performed oil analysis rather than making it obsolete. 

Changes in oil condition monitoring
While AI is positioned to revolutionize the oil analysis industry, changes are happening incrementally. Alarcon shares that he asks the same question across industries through his work: Are you planning to integrate an oil analysis sensor? In the past 10 years, he hasn’t seen much change in the very low number of responses that are planning to go fully digital. However, he has seen an increase in the number who are doing some pilot projects. Others still need to be sold on the value of sensors. “Is it a sensor that is going to help me to reduce shutdowns or unplanned maintenance or to know better the condition of my oil? People are still asking these questions,” shares Alarcon.

Oil failure mode detection
The most basic step to implementing an oil analysis program is knowing the oil failure mode, according to Alarcon. He says, “Whether you use an oil analysis lab or a sensor, if you don’t know the oil failure mode, you’re wasting time and money.” Reproducing oil failure mode in the lab is a long and expensive process, but AI applications can perform an assisted analysis, which still requires human intervention, to reduce this critical step. The assisted analysis involves identifying the chemical composition of the oil, plus the temperature, pressure and other conditions of the application. “The working conditions of the machine system must be part of any AI analysis. The oil is very important, but it’s also important to remember the rest of the things around the oil.” Then the specific neural network algorithm looks for the degradation markers that indicate early stages of failure. The role of human intervention is to redirect or correct the process as needed. This process depends on clustering, which is a key component of ML that involves sorting unlabeled, complex data according to patterns.² 

Conclusions
In the midst of the new digital revolution, Alarcon emphasizes the importance of adaptation. “Every single industry will need to jump into this. We need to be prepared.” While ML offers revolutionary efficiency benefits, it also introduces new challenges such as data quality, workforce upskilling and cybersecurity, data privacy and ownership.¹ Businesses will need to address these challenges to remain competitive in the evolving industry. Alarcon says, “In the midterm, AI will impact those businesses that have prepared to work with it.” Integrating Industry 4.0 technologies into an oil condition monitoring program can improve accuracy while increasing efficiency; Alarcon projects that AI can reduce hours spent on oil analysis by up to 90%. While a human still needs to review reports and make decisions, effective AI tools can significantly reduce the associated costs of industrial asset management once a program is in place.¹

REFERENCES
1. Alarcon, J. “Navigating the Digital Frontier: The Merging of Oil Analysis and Industry 4.0 in Shaping Tomorrow’s Industry Landscape.”
2. https://developers.google.com/machine-learning/clustering/overview

You can reach TLT editorial coordinator Nicole Gleeson at ngleeson@stle.org.