Software and Controls in the Age of AI and Decarbonization

The turbomachinery industry stands at a juncture. Driven by the urgent need for decarbonization and the transformative power of artificial intelligence (AI) and machine learning (ML), software and control systems are evolving and revolutionizing turbomachinery operations, maintenance, asset monitoring, and more, paving the way for a more sustainable and efficient future.

ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING

As AI steps onto the scene, experienced and knowledgeable workers are leaving, and AI is emerging as a crucial tool to address this experience gap and navigate the evolving landscape.

“One of the defining challenges today is the Great Crew Change, where people with decades of experience and expertise are retiring in mass—being replaced by fewer people with less operational experience,” said Shun Yoshida – Engineering Manager, HMI Software at Compressor Controls Corp. (CCC) by Honeywell. “This is while they are asked to manage energy efficiency, reliability of aging assets, and sustainability goals. They have a difficult job ahead in maintaining and even improving assets with less collective experience available in the team.”

Along this line, Yoshida named centralization and consolidation of data, analytics, and visualization as key trends “because many end users have increased their reliance on central SMEs to remotely support local operations teams where domain-specific expertise is hard to attain and retain,” he said. “And particularly for upstream offshore operations, there is a path toward unmanned platforms where connectivity is a foundational requirement.”

Ruben Alvarez, CEO of John Crane, sees growing pressure on manufacturers to capitalize on the potential of new technologies “to deliver measurable progress against sustainability targets while maintaining high performance and regulatory compliance.”

Integrating AI, ML, and even the Internet of Things (IoT) in turbomachinery applications has a lot of potential—from improving productivity to automating tasks, reducing the number of inspector tasks, and proactively adjusting controls.

According to Michael Domke, General Manager of Visual at Waygate Technologies, a Baker Hughes business, process automation, AI, and ML improve data quality and reduce inspection time. This is specifically seen in the close interaction between humans and the system: the better each side performs, the better the results.

“As with many other applications, the strength of ML and AI lies in the processing of substantial amounts of data and the automation of processes,” Domke said. “By moving some tasks away from the inspector and into the device, the inspector can focus on other tasks. For example, if we are inspecting a gas turbine, the AI-assisted software relieves the inspector of counting the turbine blades. The system also automatically recognizes a potential defect and notes the location. The inspector can then focus on the defects, check them, specify the details, and make the right recommendation regarding asset health to the owner.” The company’s video borescope, Everest Mentor Visual iQ+, does just that.

At EthosEnergy, AI and ML are now synonymous with its remote plant and asset-monitoring solutions, and it considers them essential technologies. “ML has come a long way since first introduced over 25 years ago, but at its core, modern software learns the normal behavior of assets based on the asset data received,” said Ashwin Kumar, VP of Marketing at EthosEnergy. “It automatically sets boundaries to proactively alert SMEs when there is a deviation from what was learned. This technology helps proactive engagement of an issue at the beginning stages of failure, avoiding costly downtime. With new advancements in AI, leveraging 25+ years of asset failure modes combined with ML, they aid in identifying the most likely culprits of the asset issue—vastly improving work scopes and urgency. The latest tech now blends ML and AI for forecasting failures, providing time perspective of what has been identified.”

Dinakar Deshmukh, Vice President of Data Science at GE Aerospace, said the company has been developing and deploying AI across its business for more than a decade. “During this time, we have more than doubled our investment in AI technology, positioning the company as one of the top AI patent holders within the aviation industry.”

Internally, GE Aerospace launched a companywide generative AI platform, AI Wingmate, in partnership with Microsoft to help improve employee productivity with daily work tasks and free up more time for employees to focus on problem-solving. The company leverages AI to optimize engine monitoring, part inspections, and the supply chain to improve efficiency, accuracy, and overall performance.

Credit: GE Vernova

“[We also launched] a new generative AI solution to enable faster searching of critical maintenance record assets for our airline and lessor customers, dubbed Gen AI Assistant,” Deshmukh said. “That was developed in close partnership with Microsoft and Accenture. [And we] equipped all our software developers with generative AI-based code assistants to boost efficiency and productivity.”

AI and ML also support decarbonization efforts and enable companies such as GE Vernova and CCC to explore their uses in asset performance management (APM) software and improve compressor reliability and availability.

Let’s dive deeper into decarbonization's impact on software and controls and where and how AI and ML fit in.

DECARBONIZATION

Decarbonizing global energy systems is driving a transformative shift in turbomachinery software and controls. OEMS are approaching this from different angles, including a controls platform that integrates carbon-capture plants and power plants; optimizing compressor operation to reduce energy consumption via controls and tests; and using APM to optimize assets, processes, and energy efficiency.

Carbon Capture

Carbon capture and storage (CCS) helps hard-to-abate industries decarbonize; however, operational flexibility is a common challenge.

GE Vernova developed a controls platform that enables tight integration between the power and carbon-capture plant to achieve greater than 95% CO₂ capture rates. In the short-term, the platform is providing additional control capabilities (beyond the typical power plant controls scope) for integration features specific to the carbon-capture plant, including:

• Controls related to exhaust gas recirculation (EGR), steam extraction, high backpressure, and CCS damper
• Control input/outputs for easier integration with the CCS controls system provided by the carbon-capture provider/EPC

The company also optimized the operations philosophy, such as early steam-extraction capabilities during start-up and minimizing emissions that negatively affect the CCS solvent life during steady-state operations. The EGR controls are developed to provide a concentrated flue gas stream with higher CO₂ concentration and lower oxygen concentration, resulting in lower CAPEX and OPEX for the CCS plant while protecting the gas turbine by controlling the amount of EGR.

Looking forward, GE Vernova is developing the capability to control both a full combined-cycle plant and a carbon-capture plant as an integrated system. “The key to this development has been the creation of a high-fidelity dynamic simulation that includes all aspects of both the combined-cycle power plant and the carbon-capture plant,” said Matthew Davidsaver, P.E. CCS Product Line Champion, GE Vernova. “Using this tool, control strategies are being developed to optimize overall plant operability and performance. For example, controlling steam extraction to the carbon-capture plant takes advantage of information originating in the gas turbine control system. This control and simulation platform is a continuation of tools developed over the past several decades and proven in developing various power plants, including combined cycle, integrated gasification combined cycle, aluminum smelting, LNG, and more.”

John Crane offers solutions like Sense Monitor, which allows companies to optimize infrastructure and operations by providing real-time data on equipment health. “This empowers operators to make informed decisions that reduce downtime, emissions, and leakages,” said Alvarez. “This enhances operational efficiency and supports industries like hydrogen and carbon capture in meeting their decarbonization goals by reducing their environmental impact.”

Turbocompressors

“Turbocompressors are often some of the largest energy consumers in a plant,” Yoshida said. “Eliminating suboptimal operations—such as manual control of anti-surge valves, reducing control margins, and improving control quality—represents sizable energy-saving opportunities that contribute positively to decarbonization.”

To solve upstream and midstream challenges, CCC has worked closely with its customers across the oil and gas value chain to create solutions. “For example, plugging and stiction of anti-surge valves can be a frequent issue upstream due to hydration formation and contaminants,” Yoshida said. “This can result in reduced effectiveness of compressor surge protection. CCC has introduced Valve Exerciser, which will systematically perform a valve test while monitoring its effects on process measurements. In contrast to valve self-diagnostics, this approach enables us to evaluate the overall health of the recycle line from the actuator, valve cage, plug, piping, scrubber, and strainer.”

As instrument manufacturers continue to enhance their products, control systems can also concurrently increase their tolerance of instrument failures using substitute measurements. There are compelling AI/ ML use cases for creating more graceful fallbacks. - Shun Yoshida, CCC by Honeywell

The company has also seen a buzz about its software’s ability to handle sensor drifts, freeze, and failures. “Further, our HMI software has been evolving to support seamless delivery of high-resolution data to remote experts with message queuing telemetry transport (MQTT) and web-based visualization,” he said. “We also continue to invest in cybersecurity at both organizational and product levels to help customers achieve digital transformation in a secure way.”

In critical compressor applications, CCC sees, in terms of IoT, end users moving toward centralized high-resolution time-series data and human-machine interface (HMI) screens. “On the other hand, we are still early in using AI/ML to take direct control actions,” Yoshida said. “We know from our customers that there is much to improve regarding compressor reliability and availability. For example, most turbomachinery controllers today are programmed with fallback modes that are triggered by instrumentation failures. These modes are designed to keep the compressors running without the risk of surging; however, this shift in operation modes can impact the process. As instrument manufacturers continue to enhance their products, control systems can also concurrently increase their tolerance of instrument failures using substitute measurements. There are compelling AI/ML use cases for creating more graceful fallbacks.”

Asset Performance Management

APM also plays a key role in decarbonization efforts “by helping to optimize assets, processes, and energy efficiency,” said Ryan Finger, Global Director of Software Product Marketing for GE Vernova’s Electrification Software Business.

It wears many hats and is designed to help teams improve equipment reliability, extend asset life, and reduce unplanned outages.

“Besides the traditional condition monitoring system, we see approaches like APM being adopted more,” Yoshida said. “APM gives end users a consolidated view of asset health and performance across different equipment types. Overall, we see high interest in computing and tracking the thermodynamic performance of turbomachines as an early indicator of equipment health degradation.”

APM supports company-driven sustainability efforts in industrial operations and boosts reliability and energy-efficient operations. Finger said it primarily impacts various maintenance aspects, including:

Renewable Energy Asset Maintenance/ Reduce Waste: APM uses predictive maintenance to ensure near-continuous operation of renewable energy assets, reducing reliance on fossil fuels during outages. Predictive maintenance also improves equipment precision and reduces product scrap and waste caused by poorly maintained equipment. It provides insights to help prevent asset failures by improving maintenance and reliability and reducing pollution risks from failures in critical infrastructure, such as pipelines.

Extended Asset Life: APM helps extend asset life, reducing the need for new equipment and supporting a circular economy.

Energy Efficiency: Proper maintenance enabled by APM helps equipment operate at peak efficiency, reducing energy consumption.

Work Planning: Operators/inspectors are often dispatched for distributed organizations. APM’s data insights help users better identify the work that needs to be done beforehand to validate the need for maintenance and only dispatch workers when necessary.

APM Strategy gives users access to APM Health for condition-based maintenance; APM Reliability for root cause and production loss analysis; SmartSignal for predictive and prescriptive diagnostics; and APM Mechanical Integrity for fixed asset inspections. This adds up to increased reliability with access to advanced monitoring so they can adjust their maintenance programs with near real-time data and deploy new strategies.

“APM is designed to focus on the ability to optimize organizations to get the most reliability/availability of their assets at the best cost and lowest risk,” Finger said. “For GE Vernova, the path to this starts with integrating data from all systems that store and use information that directly impacts reliability. Once correctly centralized, GE Vernova’s APM Strategy, designed to help organizations assign criticalities to assets, layer in ‘what-ifs’ and ultimately devise a scalable plan across identified assets.”

Its Autonomous Inspection leverages AI-powered computer vision in fixed cameras/ robotics to help remotely monitor assets and eliminate manual inspections. This reduces risks for monitoring volatile assets and O&M costs required for manual inspections. Data is then turned into a time series within APM to be leveraged along with additional collected information.

“As the rise of AI continues, GE Vernova’s APM is also running pilots with customers that include a built-in co-pilot for APM, the ability to ingest large documents with Generative AI, and more functionality for prescriptive analytics—all innovations that are designed to help keep a human-in-the-loop for more control and validation,” Finger said.

Use Cases

Some specific use cases of GE Vernova’s APM end-to-end value include the State Oil Company of the Azerbaijan Republic. It experienced:

• 20% reduction in maintenance costs
• 10% reduction in health, safety, and environmental incidents
• 10% increase in employee productivity

OQ, Oman’s global integrated energy group, saw a total savings of approximately $59 million. It avoided three turnaround events, saving $833K in maintenance costs and over $42 million in production losses. The company bypassed inspection costs for the removed damage mechanisms, saving an estimated $16 million, and optimized 1,320 assets by removing irrelevant damage mechanisms and adding relevant ones.

LOOKING AHEAD

“As we face the future of operations where fewer people are managing more assets, the responsibility of turbomachinery control systems should grow,” Yoshida said. “We can take early detection of faults as inputs and adjust control responses to achieve more uptime, reduce equipment damages, maintain process outputs, etc. For a given failure, some maintenance action is eventually necessary. However, the turbomachinery control system of the future can work to minimize the immediate impact of the failure and give some time for the already busy maintenance team to perform the right repair.”