Key success factors for adopting condition-based and predictive maintenance
The industrial sector is in the midst of the biggest disruption in decades, causing players to face major pressures to become more reliable and productive while reducing costs. With the rapid changes that are happening in the market, organizations can no longer afford to continue to operate in a reactive environment, it’s critical for organizations to reduce downtime, failure and production loss while extending equipment life and predicting expenditures – efficiencies that can transform operational costs. But how can that be achieved?
Hear from Covestro about how predictive analytics technologies are being put to work in industrial applications.
The adoption of efficient maintenance strategies is key to survival
Condition-Based Maintenance and Predictive Maintenance are in almost every discussion that involves improving performance and reducing costs.
In short, Condition-Based Maintenance (CBM), is maintenance when the need arises. Unlike traditional interval-based preventive maintenance, CBM is based on real-time data and only performed after one or more indicators show that equipment is going to fail or that performance is deteriorating. While proactive, and not time-based, condition-based maintenance is still a preventive maintenance paradigm as it does not give a full diagnostic on the best time to perform maintenance.
Predictive Maintenance (PdM), however, is a step up from CBM and is designed to help determine the condition of in-service equipment in order to predict when equipment is going to fail and, therefore, when maintenance should be performed. The indicators used can be vibration signatures, temperature changes, or even process parameters. Once the indicators are gathered, analyzed and the failure predicted, then the work management process can trigger a corrective activity. It involves the use of technologies like Artificial Intelligence (AI), Machine Learning (ML) and Advanced Pattern Recognition (APR)
These ways to leverage predictive analytics are just a tip of the iceberg. There is an enormous potential that big data analytics offers to improve the way organizations operate and ensure optimal utilization of resources.
Artificial Intelligence, Machine Learning and Advanced Pattern Recognition
However, it all seems too abstract, right? You must be asking, “What are these technologies all about and where can it be applied?”
Artificial intelligence (AI) is a term for simulated intelligence in machines, the ability to "think" like a human and mimic the way a person acts. To develop this kind of capability in a machine you would have to write millions of lines of codes with complex rules and decision-trees. But, in reality, artificial intelligence owes plenty to machine learning. It uses machine learning to train and build intelligent, human-like capabilities into machines. A key difference is that AI is much more proactive, interactive and “alive” in what it does.
The term Machine Learning (ML) was coined in 1959 by Arthur Samuel, who defined it as, “the ability to learn without being explicitly programmed.” It’s a field of artificial intelligence that explores the study and construction of algorithms that can learn from and make predictions on data using statistical techniques. ML is employed in a range of computing tasks where designing and programming explicit algorithms with good performance is difficult or not feasible (for example: email filtering, detection of network intruders, and computer vision).
The concepts are clear, but how are these technologies applied in the industrial scenario?
The answer is Advanced Pattern Recognition (APR).
APR is an engineering application of machine learning and consists of methods that a machine can use to recognize patterns in data to categorize them into classes and be able to predict outcomes. For example, when you are trying to predict if an asset is going to fail, this asset will follow a pattern of behavior. By recognizing and classifying that behavior, it’s possible to tell what the possible causes of the issue are and when the equipment is going to fail.
The path to create and sustain success with new maintenance strategies
When adopting a new maintenance strategy there are always many questions to ask and decisions to make. CBM, PdM or a combination of both? Which strategy is more cost-effective for the assets you want to maintain? Which technologies are more effective in finding the failure modes you’re looking for and how early will it be able to find them?
A good way to start is with a detailed asset-criticality assessment. Once it has been determined which assets are the most critical, the maintenance organization should then determine which strategies will be applied to each type of asset and seek to understand their failure modes. From this base level of knowledge, the team will then be able to identify the technologies that will find the failure modes early enough to allow the maintenance team to plan and make repair-or-replace decisions before a failure occurs.
Another question that should be asked is what kind of support and services are offered by the solution provider? Nowadays, many vendors are willing to partner with your company and set you up for success. You should look out for companies capable of delivering the full scope of the solution and helping you realize value.
Key Success Factors
Shifting from a reactive to a proactive or predictive maintenance approach is a big undertaking that creates many challenges. Often, stakeholders choose to ignore the complexity of the process completely, citing time as a limiting factor, which sets up many projects for failure.
Two things are mandatory before any initiative begins: a team of individuals willing to take on the challenge and a definition of expectations and goals. These goals should reflect the company’s core business principles and strategies as well as any industry-specific corporate and regulatory compliance standards.
Also, stakeholders need to be educated about the need for doing things differently. Maintenance and reliability workers also need to be supported and properly trained. Once a tool is in use, clear expectations must be in place about how the data and reports will be used to make better decisions, where will the data be stored and how will the data coming from maintenance tools integrate with the computerized maintenance management system (CMMS) and other monitoring systems.
A CBM or PdM program will also need to utilize root-cause analysis on any detected failure as well as structured reporting and documentation. If maintenance is to be perceived as a value rather than a cost, reporting and documenting findings are critical. All these actions together, performed on a regular basis, will help sustain the new maintenance program.
Making the transition from reactive maintenance to a more proactive or predictive strategy is a process. In many cases, the organization’s culture is the biggest obstacle to overcome and the company must change what it values most. In a reactive environment, “firefighting” maintenance is valued, but in a proactive environment, the strategy should focus on preventing the fires from ever sparking. When the organization is focused on preventing failures from happening, the culture begins to shift from reactive to proactive. There will always exist some reactive actions, but being able to minimize reactive work will allow for better planning and scheduling, increased equipment uptime, and an increase in overall equipment effectiveness (OEE).
To learn more, watch AVEVA and Covestro discuss how predictive analytics technologies are being put to work today in industrial applications.
Matheus is the Product Marketing Manager for Information products at AVEVA. With more than 7 years of experience in the industrial software industry, Matheus has extensive knowledge on automation systems, data management platforms, and the Industrial Internet of Things.