IIOT Power

Predicting the Future? There's an App for That

Power plant operators are turning to predictive maintenance applications to monitor equipment and collect performance data. Asset management systems, using artificial intelligence and other methods, are part of the growing movement toward digitization in power generation.

The digital transformation of power generation has brought significant changes to the operation of power plants. Maintenance of equipment, including inspection, adjustments, repair, and even replacement, has become more streamlined, more efficient, and in turn safer, as plant operators use data collection to determine the condition of machinery and systems throughout a facility.

Predictive maintenance applications, designed to determine when equipment needs to be serviced, provide cost savings as plant operators can be proactive with fixes, rather than waiting until a part or machine breaks. This helps eliminate downtime, allowing for better scheduling of corrective measures, and can prevent unexpected equipment failure. Predictive maintenance differs from preventive maintenance because it relies on the actual condition of equipment, rather than average or expected service intervals, to predict when maintenance will be required—eliminating the cost of performing work that is not needed.

“Utilities are rapidly adopting technology to achieve digital transformation and keep pace with customer needs,” Matt Schnugg, vice president of Data Science and Analytics for GE Power Digital, told POWER. “Within the power industry, predictive maintenance is helping power and grid operators ensure energy remains reliable and efficient, especially as the grid of the future demands smarter, faster and more resilient power delivery. Asset performance management [APM] tools are giving operators insights in real time based on real-time performance and historical trends, which allow them to have greater oversight into asset lifecycle. With this intelligence, operators can ultimately lower operating costs and maximize production yields. Oil and gas, chemicals, and power generation companies using APM technology, like Marathon, Tasnee [a Saudi Arabian chemical company], and Exelon, are already achieving significant results.”

1. Today’s asset management systems include predictive maintenance tools that can give power plant workers information about equipment performance in real time, enabling proactive fixes before problems arise. Courtesy: Projetech

“Today, it would be more difficult to find somewhere where predictive maintenance is not being used in the power generation industry,” Steve Richmond, founder and CEO of Projetech Technology Solutions, a Cincinnati, Ohio-based provider of enterprise asset management systems (Figure 1), told POWER. “In all modern-day machinery, heat is the main killer that is wearing down machines, which must be closely monitored. But other places where predictive maintenance should be used include water quality, temperature [thermography], lubrication, or even electrical connections. This involves actively monitoring for electrical vibrations that need to be balanced to efficiently and accurately deliver results.”

Richmond said renewable energy resources also are contributing to more widespread use of predictive maintenance. “The biggest difference is that wind turbines and solar installations tend to be in very remote locations where there aren’t typically personnel onsite to manage and maintain the systems. Because of the age of wind turbines and solar arrays—most are usually younger, as opposed to fossil fuel generation, which are much older—the biggest difference is the newer systems are more adaptable, given they are smarter devices. If there’s something wrong with the actuator or the flow, they will communicate this—newer power generators are generally more sophisticated and communicative than older ones.”

Delivering Results for Operational Excellence

The executives who spoke with POWER said delivering results, be it in cost savings, reduced downtime, or improvements in operational efficiency, is key to successful predictive maintenance programs.

“The initiative is always to improve operational excellence,” Mike Brooks, senior director of Asset Performance Management consulting for Aspen Technology, told POWER. “What are the key performance indicators to manage any manufacturing facility?” Brooks said protecting high-value assets is always at the forefront of a predictive maintenance program. “Whatever the initiative is, it’s to support those [assets]. Asset performance management is where you need to get started.”

“The power industry is a commodity industry,” Richmond said. “Whether it’s the local utility or a co-op utility—anybody can put power on the grid. The reason for doing so is to make money. If your equipment is not operating at its full potential, and if it’s suffering from reliability issues, as a corporation you risk losing money. By investing in predictive maintenance, you are ultimately reducing your risks and therefore saving money.”

Many companies have developed predictive maintenance programs, and several of those are participating in POWER’s Connected Plant Conference, set for Feb. 25-27, 2020, in Atlanta, Georgia. The event focuses on the digital transformation and digitization for the power generation and chemical process industries, including digital monitoring and diagnostics for power plant operators.

Francois Laborie, president of Cognite North America, told POWER his company “established a program for an aluminum and renewable energy company in 2020 that will gradually help the company move from corrective and calendar-based maintenance to a predictive maintenance regime. We did so by first connecting data from different sources (control systems, work order management systems, spare parts inventory, production planning, etc.) and making it accessible to end-users.

“From there, we developed analytics tools that monitor alarms, the relative efficiency of turbines, and key operating parameters such as the duration of the subprocesses involved in turbine start/stop sequences. We used a number of different tools for the job, including our own products and third-party visualization tools.”

Laborie noted the capabilities of his company’s core product, Cognite Data Fusion, or CDF. “As the analytics tools continue to run, they will feed more and more data into Cognite Data Fusion,” he said. “Using machine learning, the company will then be able to make predictions based on equipment performance.” For example:

    ■ “The main inlet valve is taking more time than normal to open. We should inspect and potentially repair it in four weeks.”
    ■ “Turbine 1 is diverging from its predetermined optimal performance range. We should install new labyrinth seals in eight weeks.”
    ■ “There have been five alarms related to lubricating oil pump pressure in the last quarter. We should perform a live inspection next week.”

Laborie said current predictive maintenance applications “are supported by tailored models and simple GUIs [graphical user interfaces] to present users with model output and drill-downs. However, in the future there is a huge potential for this to be handled in a unified way in an application that covers the entire workflow, from scheduling maintenance to planning activities, to conducting and reporting on work done. Before we can talk about specific software applications, we need to solve the previous step: that power generation companies are set up to power these applications with the data they are already producing.”

Artificial Intelligence Enters the Picture

GlobalData in a recent report—“Thematic Research: Predictive Maintenance in Power”—listed some of what the group considers top providers of predictive maintenance solutions. The list includes International Business Machines Corp. (IBM), known for its Maximo Asset Performance Management system, which is offered by Projetech.

Richmond said his company’s platform is “integration agnostic. It will integrate any software product and collect data. Projetech products allow the input of numerous predictive systems to pool information that can later be analyzed by humans or AI [artificial intelligence] to determine a course of action and ultimately result in more accurate predictive maintenance.”

“We’re actually getting into the fringes of AI,” said Peter Stock, vice president-Utility Reliability Technical and Performance for Veolia, a global company providing operational, engineering, and technology solutions for industry, including power generators. “We’ve actually got a couple of IT people, data scientists, looking at the daily historian and stuff that we capture on our energy plants, to give us some idea of predictability on materials.

“What I’ve done, in terms of predictive technology, I used the RCM [reliability centered maintenance] tool so we can use predictive technology effectively,” Stock said. “A lot of this technology is expensive. It doesn’t make it cost-effective to use something like wireless vibration monitoring. But the cost is coming down… I think the price of a lot of those wireless [monitoring] systems [will] come down.”

In addition to IBM, other companies considered among the leaders in predictive maintenance programs include SAP, which offers its SAP Predictive Maintenance and Service solution; Microsoft, which with its Microsoft Azure provides a major public cloud platform for the Industrial Internet of Things (IIoT), including predictive maintenance; and Agder Energi, a Norway-based energy group that “uses Azure Digital Twins to determine ways to efficiently operate its electricity grid via distributed energy resources [DERs], device controls, along with predictive forecasting to avoid costly and tedious energy upgrades,” according to GlobalData.

Elsewhere in Europe, E.ON, the German-based utility, uses AI to warn of faults or other issues in the power grid. Schleswig-Holstein Netz, a German grid operator, is using the system on its medium-voltage grids. Siemens, another Germany-headquartered company, with major operations in the U.S. and worldwide, is notable for its cloud-based operating system MindSphere, a predictive learning platform that offers early notification of asset defects so maintenance can be performed, minimizing downtime.

2. The SKF Pulse is a Bluetooth sensor and mobile application that can be used to monitor rotating equipment. The sensor acts as a smart vibration tool, transmitting data wirelessly via the mobile app for real-time machine diagnostics. Courtesy: SKF USA

SKF USA recently introduced SKF Pulse (Figure 2), a portable Bluetooth sensor and mobile application for monitoring rotating equipment. The app provides operations and maintenance personnel with what could be called a “do-it-yourself” predictive maintenance program. The sensor acts as a smart vibration tool, transmitting data wirelessly via the mobile app for real-time machine diagnostics and insights. The vibration and temperature data can be stored and shared for further analysis. Josh Flemming, head of strategic marketing for SKF USA, in an email to POWER said, “The SKF Pulse app makes machine monitoring easier. Users can monitor machine health and receive timely, valuable insights for quick and strategic decision-making.”

Digital Twins

GE’s Predix platform is in widespread use. The program, according to GE, is “a scalable, asset-centric data foundation… the platform delivers shared capabilities that industrial applications require: asset connectivity, edge technologies, analytics and machine learning, big data processing, and asset-centric digital twins… [the] Predix Platform is optimized for high volume, low latency, and integration-intensive data management and analytics-driven outcomes.”

GE’s Schnugg told POWER, “Digital twins are software representations of physical assets that allow virtual monitoring, prediction and prevention of impending failure. Digital twins use historical and real-time data to give insights into asset lifecycle and maintenance needs, giving operators the ability to predict a failure and the potential consequences of it before it even happens. GE Digital has more than two decades of experience deploying digital twins as part of our APM suite. Oil and gas, chemicals, and power generation customers are able to leverage their existing data and synthesize it to optimize their processes and achieve efficiency and cost reduction. Companies that use GE digital twins typically see a 5–10% inventory cost improvement and a 2–6% increase in general availability. To date, we’ve been able to help customers save more than $1.5 billion in unforeseen maintenance or asset replacement costs by making recommendations that could save a company from devastating failures.”

Cognite’s Laborie agreed with Schnugg that “a key enabler for predictive maintenance is the digital twin,” noting it “enables data-driven engineering, production, and maintenance decisions. Historically, a digital twin has had a single dimension of contextualization to solve the use case it was specifically created to answer. However, it is possible for an organization to enhance the overall understanding of its operations by putting all OT and IT data through a contextualization pipeline to create an operational digital twin. This is the next frontier in the digital twin space.

“When a company creates an operational digital twin, the richness of the data describing the industrial reality allows the user to create many more correlations between data points. Power generation companies that deploy an operational digital twin will finally have true control over their data—the ability to understand where it comes from, how reliable it is, and how to enrich it over time. They will also be the first ones to scale successful predictive maintenance solutions on top of that data, which must be the priority of any digitalization initiative.”

The Bottom Line

Several energy companies and utilities have noted specific examples of cost savings from predictive maintenance. EDF Energy has reported saving more than $1 million through its use of Schneider Electric’s EcoStruxure Maintenance Advisor system. The company said it has used Emerson’s AMS Suite predictive maintenance software to optimize maintenance at a combined cycle gas turbine plant in the UK.

Duke Energy said it has used Schneider Electric’s Avantis PRiSM technology to save more than $7.5 million through the early detection of a crack in a turbine rotor, and said it also has used Genpact’s Lean Digital system to help with cost overrun predictability, and for asset optimization. Southern Company said it has used the Avantis PRiSM system for continuous monitoring at its natural gas-fired and biomass power plants, with savings of about $4.5 million.

Said Richmond: “There are many software applications available on the market. The software falls into some common categories for tracking—products for tracking thermal readings, products for tracking vibration, air quality, and lubrication. All the software essentially does the same thing. The key is that these products are compatible from a communications standpoint, and operate on the same platform to be able to exchange information freely.”

Brooks told POWER a successful predictive maintenance outcome is more than just reliability. “Criticality is what matters to us,” he said. “It’s the ability to plan. If I have equipment that needs to be available for 18 months, it needs to be able to be available for those 18 months. Availability to plan is the cornerstone, that’s the No. 1 thing. Uptime trumps everything.

“You need to get rid of the scourge of all condition monitoring, which is false alerts,” Brooks said. “Machine learning, it’s not about machine learning, it’s about what you do with machine learning. People say it’s a silver bullet, but it’s not. It will find correlations, but that’s not what you need. You need the data and information, going the right way, to find the right problems, to avoid those shut-down situations.” ■

Darrell Proctor is a POWER associate editor (@DarrellProctor1, @POWERmagazine).

SHARE this article