Real-World Diagnostics and Prognostics for Grid-Connected Battery Energy Storage Systems

This is a sponsored article brought to you by The University of Sheffield.

Across global electricity networks, the shift to renewable energy has fundamentally changed the behavior of power systems. Decades of engineering assumptions, predictable inertia, dispatchable baseload generation, and slow, well-characterized system dynamics, are now eroding as wind and solar become dominant sources of electricity. Grid operators face increasingly steep ramp events, larger frequency excursions, faster transients, and prolonged periods where fossil generation is minimal or absent.

In this environment, battery energy storage systems (BESS) have emerged as essential tools for maintaining stability. They can respond in milliseconds, deliver precise power control, and operate flexibly across a range of services. But unlike conventional generation, batteries are sensitive to operational history, thermal environment, state of charge window, system architecture, and degradation mechanisms. Their long-term behavior cannot be described by a single model or simple efficiency curve, it is the product of complex electrochemical, thermal, and control interactions.

Most laboratory tests and simulations attempt to capture these effects, but they rarely reproduce the operational irregularities of the grid. Batteries in real markets are exposed to rapid fluctuations in power demand, partial state of charge cycling, fast recovery intervals, high-rate events, and unpredictable disturbances. As Professor Dan Gladwin, who leads Sheffield’s research into grid-connected energy storage, puts it, “you only understand how storage behaves when you expose it to the conditions it actually sees on the grid.”

This disconnect creates a fundamental challenge for the industry: How can we trust degradation models, lifetime predictions, and operational strategies if they have never been validated against genuine grid behavior?

Few research institutions have access to the infrastructure needed to answer that question. The University of Sheffield is one of them.

Sheffield’s Centre for Research into Electrical Energy Storage and Applications (CREESA) operates one of the UK’s only research-led, grid-connected, multi-megawatt battery energy storage testbeds. The University of Sheffield

Sheffield’s unique facility

The Centre for Research into Electrical Energy Storage and Applications (CREESA) operates one of the UK’s only research-led, grid-connected, multi-megawatt battery energy storage testbeds. This environment enables researchers to test storage technologies not just in simulation or controlled cycling rigs, but under full-scale, live grid conditions. As Professor Gladwin notes, “we aim to bridge the gap between controlled laboratory research and the demands of real grid operation.”

At the heart of the facility is an 11 kV, 4 MW network connection that provides the electrical and operational realism required for advanced diagnostics, fault studies, control algorithm development, techno-economic analysis, and lifetime modeling. Unlike microgrid scale demonstrators or isolated laboratory benches, Sheffield’s environment allows energy storage assets to interact with the same disturbances, market signals, and grid dynamics they would experience in commercial deployment.

“The ability to test at scale, under real operational conditions, is what gives us insights that simulation alone cannot provide.” —Professor Dan Gladwin, The University of Sheffield

The facility includes:

• A 2 MW / 1 MWh lithium titanate system, among the first independent grid-connected BESS of its kind in the UK
• A 100 kW second-life EV battery platform, enabling research into reuse, repurposing, and circular-economy models
• Support for flywheel systems, supercapacitors, hybrid architectures, and fuel-cell technologies
• More than 150 laboratory cell-testing channels, environmental chambers, and impedance spectroscopy equipment
• High-speed data acquisition and integrated control systems for parameter estimation, thermal analysis, and fault response measurement

The infrastructure allows Sheffield to operate storage assets directly on the live grid, where they respond to real market signals, deliver contracted power services, and experience genuine frequency deviations, voltage events, and operational disturbances. When controlled experiments are required, the same platform can replay historical grid and market signals, enabling repeatable full power testing under conditions that faithfully reflect commercial operation. This combination provides empirical data of a quality and realism rarely available outside utility-scale deployments, allowing researchers to analyse system behavior at millisecond timescales and gather data at a granularity rarely achievable in conventional laboratory environments.

According to Professor Gladwin, “the ability to test at scale, under real operational conditions, is what gives us insights that simulation alone cannot provide.”

Dan Gladwin, Professor of Electrical and Control Systems Engineering, leads Sheffield’s research into grid-connected energy storage.The University of Sheffield

Setting the benchmark with grid scale demonstration

One of Sheffield’s earliest breakthroughs came with the installation of a 2 MW / 1 MWh lithium titanate demonstrator, a first-of-a-kind system installed at a time when the UK had no established standards for BESS connection, safety, or control. Professor Gladwin led the engineering, design, installation, and commissioning of the system, establishing one of the country’s first independent megawatt scale storage platforms.

The project provided deep insight into how high-power battery chemistries behave under grid stressors. Researchers observed sub-second response times and measured the system’s capability to deliver synthetic inertia-like behavior. As Gladwin reflects, “that project showed us just how fast and capable storage could be when properly integrated into the grid.”

But the demonstrator’s long-term value has been its continued operation. Over nearly a decade of research, it has served as a platform for:

• Hybridization studies, including battery-flywheel control architectures
• Response time optimization for new grid services
• Operator training and market integration, exposing control rooms and traders to a live asset
• Algorithm development, including dispatch controllers, forecasting tools, and prognostic and health management systems
• Comparative benchmarking, such as evaluation of different lithium-ion chemistries, lead-acid systems, and second-life batteries

A recurring finding is that behavior observed on the live grid often differs significantly from what laboratory tests predict. Subtle electrical, thermal, and balance-of-plant interactions that barely register in controlled experiments can become important at megawatt-scale, especially when systems are exposed to rapid cycling, fluctuating set-points, or tightly coupled control actions. Variations in efficiency, cooling system response, and auxiliary power demand can also amplify these effects under real operating stress. As Professor Gladwin notes, “phenomena that never appear in a lab can dominate behavior at megawatt scale.”

These real-world insights feed directly into improved system design. By understanding how efficiency losses, thermal behavior, auxiliary systems, and control interactions emerge at scale, researchers can refine both the assumptions and architecture of future deployments. This closes the loop between application and design, ensuring that new storage systems can be engineered for the operational conditions they will genuinely encounter rather than idealized laboratory expectations.

Ensuring longevity with advanced diagnostics

Sheffield’s Centre for Research into Electrical Energy Storage and Applications (CREESA) enables researchers to test storage technologies not just in simulation or controlled cycling rigs, but under full-scale, live grid conditions.The University of Sheffield

Ensuring the long-term reliability of storage requires understanding how systems age under the conditions they actually face. Sheffield’s research combines high-resolution laboratory testing with empirical data from full-scale grid-connected assets, building a comprehensive approach to diagnostics and prognostics. In Gladwin’s words, “A model is only as good as the data and conditions that shape it. To predict lifetime with confidence, we need laboratory measurements, full-scale testing, and validation under real-world operating conditions working together.”

A major focus is accurate state estimation during highly dynamic operation. Using advanced observers, Kalman filtering, and hybrid physics-ML approaches, the team has developed methods that deliver reliable SOC, SOH and SOP estimates during rapid power swings, irregular cycling, and noisy conditions where traditional methods break down.

Another key contribution is understanding cell-to-cell divergence in large strings. Sheffield’s data shows how imbalance accelerates near SOC extremes, how thermal gradients drive uneven ageing, and how current distribution causes long-term drift. These insights inform balancing strategies that improve usable capacity and safety.

Sheffield has also strengthened lifetime and degradation modeling by incorporating real grid behavior directly into the framework. By analyzing actual market signals, frequency deviations, and dispatch patterns, the team uncovers ageing mechanisms that do not appear during controlled laboratory cycling and would otherwise remain hidden.

These contributions fall into four core areas:

State Estimation and Parameter Identification

• Robust SOC/SOH estimation
• Online parameter identification for equivalent circuit models
• Power capability prediction using transient excitation
• Data selection strategies under noise and variability

Degradation and Lifetime Modelling

• Degradation models built on real frequency and market data
• Analysis of micro cycling and asymmetric duty cycles
• Hybrid physics-ML forecasting models

Thermal and Imbalance Behavior

• Characterizing thermal gradients in containerized systems
• Understanding cell imbalance in large-scale systems
• Mitigation strategies at the cell and module level
• Coupled thermal-electrical behavior under fast cycling

Hybrid Systems and Multi-Technology Optimization

• Battery-flywheel coordination strategies
• Techno-economic modeling for hybrid assets
• Dispatch optimization using evolutionary algorithms
• Control schemes that extend lifetime and enhance service performance

Beyond grid-connected systems, Sheffield’s diagnostic methods have also proved valuable in off-grid environments. A key example is the collaboration with MOPO, a company deploying pay-per-swap lithium-ion battery packs in low-income communities across Sub-Saharan Africa. These batteries face deep cycling, variable user behavior, and sustained high temperatures, all without active cooling or controlled environments. The team’s techniques in cell characterization, parameter estimation, and in-situ health tracking have helped extend the usable life of MOPO’s battery packs. “By applying our know-how, we can make these battery-swap packs clean, safe, and significantly more affordable than petrol and diesel generators for the communities that rely on them,” says Professor Gladwin.

Beyond grid-connected systems, Sheffield’s diagnostic methods have also proved valuable in off-grid environments. A key example is the collaboration with MOPO, a company deploying pay-per-swap lithium-ion battery packs in low-income communities across Sub-Saharan Africa. MOPO

Collaboration and the global future

A defining strength of Sheffield’s approach is its close integration with industry, system operators, technology developers, and service providers. Over the past decade, its grid-connected testbed has enabled organizations to trial control algorithms, commission their first battery assets, test market participation strategies, and validate performance under real operational constraints.

These partnerships have produced practical engineering outcomes, including improved dispatch strategies, refined control architectures, validated installation and commissioning methods, and a clearer understanding of degradation under real-world market operation. According to Gladwin, “It is a two-way relationship, we bring the analytical and research tools, industry brings the operational context and scale.”

One of Sheffield’s earliest breakthroughs came with the installation of a 2 MW / 1 MWh lithium titanate demonstrator. Professor Gladwin led the engineering, design, installation, and commissioning of the system, establishing one of UK’s first independent megawatt scale storage platforms.The University of Sheffield

This two-way exchange, combining academic insight with operational experience, ensures that Sheffield’s research remains directly relevant to modern power systems. It continues to shape best practice in lifetime modelling, hybrid system control, diagnostics, and operational optimization.

As electricity systems worldwide move toward net zero, the need for validated models, proven control algorithms, and empirical understanding will only grow. Sheffield’s combination of full-scale infrastructure, long-term datasets, and collaborative research culture ensures it will remain at the forefront of developing storage technologies that perform reliably in the environments that matter most, the real world.