Salgótarján is a city of roughly 35,000 people whose economic life was built on brown coal. Mining began in the mid-nineteenth century, and over at least 170 years the city developed a heavy industrial portfolio that included a coal-fired power plant, iron and steel smelting, glassworks, and mining machinery production. When the communist system collapsed in the early 1990s, the mines and heavy industries declined, but their waste did not disappear. Coal mine dumps, coal ash heaps, and smelter slag deposits remained scattered across the urban landscape. In several cases, these industrial leftovers were repurposed as filling material during post-industrial redevelopment, meaning that playgrounds, parks, and kindergartens may literally sit on top of the by-products of a vanished industrial era.

The research team focused on strontium because of an unusual geochemical property. Strontium has four naturally occurring stable isotopes, and the proportion of strontium-87 increases over geological time through the radioactive decay of rubidium-87, an element with a half-life of 48.8 billion years. As a result, ancient rocks and the soils derived from them tend to carry elevated ratios of strontium-87 to strontium-86, while industrial by-products such as coal ash typically carry lower ratios. Strontium also substitutes readily for calcium in minerals, which means it concentrates in the limestone additives and carbonate-rich materials used during coal combustion. When coal burns, strontium becomes enriched in the resulting ash: the Salgótarján brown coal contained 52.8 milligrams of strontium per kilogram, while a coal ash sample collected from a cone just half a kilometer from the former power plant contained 113.6 milligrams per kilogram.

To capture these signals, the researchers analyzed fifteen urban soil samples collected from kindergartens, playgrounds, parks, and roadsides in the northeastern part of the city, at depths of five to fifteen centimeters, following the Euro-Geo-Surveys sampling protocol. Samples were digested using microwave-assisted acid treatment, purified through extraction chromatography resin, and measured with thermal ionization mass spectrometry at a laboratory in Fukushima, Japan, using the multi-dynamic technique and the NIST-SRM-987 standard for quality control. The results revealed considerable variation: strontium concentrations ranged from 17.5 to 51.3 milligrams per kilogram, and the strontium-87 to strontium-86 ratios spanned 0.710303 to 0.719559, a spread wide enough to encode multiple distinct sources.

The spatial pattern was striking. More than half of the samples showed a strong positive correlation, with a coefficient of determination of 0.86, between the isotopic ratio and distance from the former coal-fired power plant. Soils closest to the plant, such as a roadside sample 0.6 kilometers away and a park sample 0.9 kilometers away, carried the lowest ratios, consistent with heavy coal ash input, since the local coal ash itself registered a ratio of 0.711652. At intermediate distances the ratios climbed, and beyond three kilometers they approached values typical of the loess deposits that blanket the Pannonian basin, indicating that coal ash influence had faded and local soil dominated. The local coal ash ratio is itself explained by the marine carbonates within the coal seams, which carry the globally uniform seawater signature of roughly 0.7080 to 0.7090, slightly modified by the limestone sorbents added during combustion.

Not every sample fit this tidy distance gradient, and the deviations proved just as informative. A playground sample located near the former iron and steel works recorded the lowest ratio of the entire dataset, 0.710303, even lower than the coal ash itself, despite not being the closest site to the power plant. Two other samples, from a kindergarten and a roadside location, showed similarly depressed values. The researchers attribute these anomalies to smelter slag, a known industrial source in the city that incorporates strontium from carbonate and phosphate minerals with inherently low isotopic ratios, and possibly to unknown technogenic soils transported by humans during land cover changes. Microscopic examination of these samples revealed reddish oxidized layers and slag grains, physical corroboration of the isotopic evidence.

At the opposite extreme, three samples exceeded even the local soil end-member, with the highest ratio reaching 0.719559 at a roadside site. These elevated values point to artificial infills, quartz-rich materials commonly introduced for road construction and soil stabilization during urbanization. Such materials derive from parent rocks rich in rubidium, like granites and metamorphic rocks, in which radiogenic strontium-87 has accumulated over geological timescales. In other words, even the apparently clean end of the isotopic spectrum in Salgótarján tells a story of human landscape engineering rather than untouched nature.

To move from qualitative fingerprints to quantitative estimates, the team applied a two-end-member mixing model, plotting the isotopic ratio against the inverse of strontium concentration. Around eighty percent of the samples fell along a strong linear trend, with a coefficient of determination of 0.89, confirming that local soil and coal ash are the two dominant contributors. The model estimated that coal ash accounts for roughly 77 percent of the strontium in a roadside sample, 71 percent in a park sample, and 67 percent in another park located 4.4 kilometers from the plant, where ash was likely introduced as fill rather than by atmospheric deposition. Conversely, samples from kindergartens showed local soil contributions of 71 to 97 percent, suggesting minimal industrial contamination at those particular sites. The authors emphasize that these figures are semi-quantitative, intended to indicate spatial trends rather than precise apportionment, since the urban environment likely contains additional sources that a two-component model cannot fully resolve.

The physicochemical context reinforced the isotopic story. The soils were predominantly weakly alkaline, with a mean pH of 7.28, and samples rich in coal ash showed elevated pH reflecting the calcium and magnesium content of combustion residues, while samples near the smelter slag hill were more acidic, consistent with the secondary minerals like gypsum and iron oxy-hydroxides that form as slag weathers over decades. Phosphorus and sulfur contents peaked at sites influenced by industrial by-products, and total organic carbon and total nitrogen were highest in vegetated parks, where organic matter appears to help stabilize strontium in the soil matrix. Notably, a previous study by part of the same team had already documented reduced microbial activity at several of these sites, linked to arsenic, lead, mercury, and cadmium from coal ash and slag, underscoring that the strontium signal tracks co-contaminants of direct health concern.

The broader significance of the study lies in its demonstration that strontium isotope geochemistry can serve as a robust, relatively inexpensive tracer for legacy industrial contamination in post-industrial cities worldwide. Comparable isotopic signatures have been reported in urban soils from Kyrgyzstan and in contaminated sediments from Chinese rivers, suggesting the approach travels well across different geological and industrial settings. For city planners and environmental agencies, the message is practical: before repurposing industrial waste as fill, and before siting playgrounds and kindergartens on former industrial land, the isotopic composition of the soil can reveal exactly what lies beneath. As former industrial regions across Central and Eastern Europe continue to be redeveloped, the humble strontium atom may become one of the most valuable witnesses to what the twentieth century left in the ground.

Subject of Research: Strontium isotope tracing of industrial contamination sources in urban soils of a former Hungarian coal and steel city

Article Title: Strontium isotopic signatures and source apportionment in urban soils of Salgótarján: unraveling the impact of industrial emissions

Article References: Maghsoudlou, M., Tserendorj, D., Abbaszade, G., Kavasi, N., Völgyesi, P., Tóth-Bodrogi, E., Sahoo, S. K., Štrok, M., Inoue, K., Kovács, T., & Szabó, C. (2026). Strontium isotopic signatures and source apportionment in urban soils of Salgótarján: unraveling the impact of industrial emissions. Environmental Science and Pollution Research, 33(30), 15829-15848. https://doi.org/10.1007/s11356-026-38225-6

Image Credits: AI Generated

DOI: 10.1007/s11356-026-38225-6

Keywords: strontium isotopes, urban soils, coal ash, coal-fired power plant, smelter slag, soil contamination, source apportionment, Salgótarján, heavy metals, geochemistry, post-industrial cities, environmental forensics

The research, led by Yufei Wang, Nan Li, Hui Shi, Gang Xie, and Xiaoyin Nie of the Shanxi Key Laboratory of Advanced Control and Industrial Intelligence at Taiyuan University of Science and Technology, addresses a challenge that has grown more urgent as renewable energy floods modern power grids. When wind and solar output fluctuates, coal-fired units are increasingly asked to ramp their load up and down, cycling through operating conditions far more varied than the steady baseload duty for which they were originally designed. That flexibility demand places new stress on the dynamic models used to predict and control how the plant responds, and inaccurate predictions can translate directly into safety risks and efficiency losses.

At the heart of the difficulty is the sheer complexity of a boiler-turbine system. It is a strongly nonlinear, multivariable machine in which fuel flow, feedwater, turbine valve positions, and air supply interact in tangled feedback loops. Worse still, different variables respond on different time scales: main steam pressure may react to a disturbance within seconds, while main steam temperature drifts slowly over minutes because of the thermal inertia of massive metal components. A single, one-size-fits-all neural network architecture struggles to capture this variable-specific, multi-scale character, and conventional data-driven models that do capture it often hide their internal logic behind millions of opaque parameters.

The team’s answer is a framework they call the Multi-dimensional Dynamic Attention Network, or MDANet. Rather than treating all inputs and all time scales identically, the architecture is built to mirror the physical structure of the plant it models. The first pillar of the design is adaptive receptive-field temporal feature extraction, which allows the network to learn, for each measured variable, how far back in time it needs to look. Fast-moving signals get short memory windows; sluggish, high-inertia signals get long ones. This directly encodes the multi-scale dynamic behavior that engineers know characterizes boiler-turbine systems, and it does so without hard-coding specific time constants, letting the data reveal the appropriate scales.

The second pillar, adaptive feature re-weighting, tackles the problem of shifting operating conditions. A coal-fired unit swinging between low-load and high-load operation does not obey a single fixed set of relationships; the relative importance of one input variable to an output can change as the plant moves through its operating envelope. MDANet’s re-weighting mechanism dynamically adjusts how much attention each variable receives, so the model remains faithful to the plant’s behavior across a wide range of conditions rather than overfitting to one narrow regime. This is the kind of condition-dependent sensitivity that classical linear models cannot represent and that naive deep networks learn only implicitly, buried where no engineer can inspect it.

The third pillar addresses the fact that a boiler-turbine unit must be modeled as a coupled multi-output system. Unit load, main steam pressure, and main steam temperature are not independent quantities; they are linked through shared energy and mass flows, and predicting them in isolation discards valuable information. To capture this coupling, the researchers employ a temporal context-aware multi-gate mixture-of-experts structure. In this design, multiple specialized expert sub-networks process the input, and learned gating networks decide, moment by moment and task by task, which experts should contribute to each output prediction. The result is a form of structured knowledge sharing: related outputs draw on overlapping expertise while still retaining the flexibility to specialize where their dynamics diverge.

Validation was carried out on real distributed control system data from a 600 MW coal-fired boiler-turbine unit, giving the evaluation an industrial realism that laboratory benchmarks often lack. The authors report that comparative experiments, ablation studies, and out-of-sample robustness analysis all demonstrate that MDANet achieves accurate and stable predictions for the three key outputs: unit load, main steam pressure, and main steam temperature. The ablation studies are particularly telling, because they show that removing any one of the three architectural pillars degrades performance, confirming that each component earns its place rather than adding decorative complexity.

What sets the work apart, however, is its insistence on interpretability as a first-class requirement rather than an afterthought. The researchers used attention-based visualization to expose which input variables and which time windows the network relied on when forming its predictions, and they supplemented these visual maps with quantitative interpretability analysis. The crucial finding is that the learned relevance patterns inside the model are consistent with the expected dynamic behavior of the boiler-turbine system. In other words, when the network predicts a change in main steam temperature, the attention patterns show it watching the inputs that thermodynamics says should matter, over the time horizons that the plant’s thermal inertia implies. The model’s internal logic and the engineer’s physical understanding converge.

This alignment between learned representations and physical expectation is precisely what the term physics-aligned is meant to convey. The framework does not embed explicit physical equations as hard constraints, a strategy used in some physics-informed neural networks. Instead, it shapes the architecture so that the network can naturally discover physically meaningful structure from data, and then verifies that the discovered structure matches reality. For plant engineers, this means a prediction can be audited: if the model’s attention points to a plausible causal chain, confidence rises; if it points somewhere nonsensical, the anomaly itself becomes a diagnostic signal. That auditability is the difference between a research curiosity and a tool that can be trusted in safety-critical operation.

The broader implications extend well beyond a single power station. As grids worldwide absorb ever-higher shares of variable renewable generation, the flexible operation of remaining thermal plants has become a linchpin of energy security, and transparent dynamic models are a prerequisite for advanced monitoring, fault detection, and control in that context. The Taiyuan team’s results suggest a practical middle path between the accuracy of black-box deep learning and the transparency of first-principles modeling: architectures whose inductive biases guide them toward physically sensible solutions, paired with interpretability tools that make the learned behavior visible. If such physics-aligned approaches generalize to other industrial processes, from chemical reactors to cement kilns, they could help dissolve the trust barrier that has long separated powerful machine learning from the engineers who operate the machines. The study, published open access with funding support from the National Natural Science Foundation of China and several Shanxi provincial programs, offers a concrete demonstration that in industrial artificial intelligence, seeing what the model sees may matter as much as how well it predicts.

Subject of Research: Physics-aligned deep learning for interpretable dynamic modeling of coal-fired boiler-turbine systems

Article Title: Interpretable dynamic modeling of coal-fired boiler–turbine systems: a physics-aligned deep learning approach

Article References: Wang, Y., Li, N., Shi, H., Xie, G., & Nie, X. (2026). Interpretable dynamic modeling of coal-fired boiler–turbine systems: a physics-aligned deep learning approach. Complex & Intelligent Systems. https://doi.org/10.1007/s40747-026-02479-x

Image Credits: AI Generated

DOI: 10.1007/s40747-026-02479-x

Keywords: boiler-turbine system, dynamic modeling, deep learning, attention mechanism, interpretability, multi-task learning, mixture-of-experts, coal-fired power plant, renewable energy integration, industrial AI, time series prediction, control systems