The study, led by A. Hemanth Kumar and K. Karthikeyan with collaborator G. Murali, addresses a significant gap in concrete durability research. While lightweight aggregate concrete has become increasingly popular in coastal, offshore, and hydraulic engineering applications, comparative data on the abrasion resistance of different agricultural waste aggregates under high-energy underwater conditions have been scarce. The researchers set out to compare conventional concrete with lightweight aggregate concretes incorporating coconut shell, expanded clay aggregate, oil palm shell, and areca nut shell aggregates, each reinforced with natural fibers including hemp, jute, and coir.

What makes this research particularly compelling is its experimental design. Rather than testing materials under static conditions, the team subjected concrete specimens to underwater rotational abrasion at two distinct speeds: 1200 rpm and 1600 rpm. This approach allowed them to investigate the combined effects of rotational speed, abrasion time, aggregate properties, and the interfacial transition zone between aggregate and cement matrix — factors that collectively determine how well a concrete mixture resists mechanical erosion in real hydraulic environments.

The findings revealed a clear pattern: abrasion loss increased linearly with exposure time, confirming that underwater wear is a cumulative process rather than a sudden failure event. After six hours of testing at 1600 rpm, weight loss increased from 461.93 grams for conventional concrete to a range of 549.44 to 698.91 grams for the lightweight concretes. This means that while all lightweight mixtures experienced greater material loss than conventional concrete, the spread within the lightweight group was substantial, suggesting that aggregate choice and fiber reinforcement significantly influence abrasion behavior.

Wear rate data told an even more nuanced story. At 1200 rpm, wear rates ranged from 23.35 to 38.80 grams per hour across the tested mixtures. When the rotational speed was increased to 1600 rpm, wear rates jumped dramatically to between 55.42 and 116.49 grams per hour. The weight loss ratio varied between 1.78 and 2.60, indicating that some lightweight mixtures were considerably more sensitive to speed increases than others. This speed sensitivity is a critical parameter for engineers designing structures that will face varying hydraulic energy conditions throughout their service life.

Among all the lightweight mixtures tested, the palm oil shell-hemp fiber combination emerged as the standout performer. This mixture exhibited the lowest speed sensitivity and the most stable abrasion performance under severe underwater conditions. The finding suggests that the interaction between oil palm shell aggregates and hemp fibers creates a microstructure that better resists the escalating mechanical energy delivered at higher rotational speeds, potentially through improved crack bridging and enhanced bonding at the interfacial transition zone.

The implications of this research extend far beyond the laboratory. Agricultural waste products like coconut shells, oil palm shells, and areca nut shells are generated in enormous quantities across tropical regions, often ending up in landfills or open burning. Converting these byproducts into concrete aggregates offers a dual benefit: reducing waste disposal problems while decreasing reliance on natural stone aggregates, whose extraction carries significant environmental costs. Natural fibers such as hemp, jute, and coir further enhance the sustainability profile of these mixtures, as they are renewable, biodegradable, and require minimal processing energy compared to synthetic alternatives.

For the construction industry, the results provide practical guidance for material selection in hydraulic and marine applications. Engineers can now weigh the trade-offs between conventional concrete, which offers superior abrasion resistance but higher density and greater environmental footprint, and lightweight alternatives that reduce structural dead loads while incorporating recycled materials. The identification of palm oil shell-hemp fiber as the most stable performer under severe conditions gives designers a specific, evidence-based option for structures facing high-energy underwater environments.

The research also highlights the importance of understanding the interfacial transition zone — the thin region between aggregate particles and the surrounding cement paste — as a key determinant of abrasion behavior. The properties of this zone influence how cracks initiate and propagate under mechanical loading, and the natural fibers likely play a role in bridging microcracks that form during abrasion, slowing the rate of material loss. This microstructural perspective opens avenues for future research into optimizing fiber content, aggregate surface treatment, and mixture proportions to further enhance durability.

As climate change intensifies flooding events and coastal infrastructure faces growing hydraulic stresses, the need for durable, sustainable construction materials has never been more urgent. This study demonstrates that agricultural waste aggregates combined with natural fibers can produce lightweight concretes with measurable, quantifiable abrasion performance under controlled underwater conditions. While lightweight mixtures still trail conventional concrete in raw abrasion resistance, the palm oil shell-hemp fiber combination offers a promising balance of sustainability, reduced weight, and stable performance that could reshape material choices for the next generation of hydraulic and marine structures.

Subject of Research: Energy-dependent underwater abrasion behavior of sustainable lightweight concrete incorporating agro-waste aggregates and natural fibers under rotational hydraulic conditions

Article Title: Energy-dependent underwater abrasion behavior of sustainable lightweight concrete incorporating agro-waste aggregates and natural fibers under rotational hydraulic conditions

Article References: Kumar, A. H., Karthikeyan, K., & Murali, G. (2026). Energy-dependent underwater abrasion behavior of sustainable lightweight concrete incorporating agro-waste aggregates and natural fibers under rotational hydraulic conditions. Scientific Reports. https://doi.org/10.1038/s41598-026-72514-x

Image Credits: AI Generated

DOI: 10.1038/s41598-026-72514-x

Keywords: Energy-dependent, underwater, abrasion, behavior, sustainable, lightweight, concrete, incorporating, agro-waste, aggregates, natural, fibers

Silica fume is a byproduct of the silicon and ferrosilicon alloy industry, prized for its ultrafine particle size and high pozzolanic reactivity. When blended into cement, its amorphous silica reacts with calcium hydroxide (CH) generated during cement hydration to form additional calcium silicate hydrate (C-S-H), the glue-like phase that gives hardened cement its strength. This reaction refines pores and densifies the matrix, making silica fume a staple ingredient in high-strength concrete and ultra-high performance concrete (UHPC). But silica fume is expensive — often several times the price of ordinary Portland cement — because of limited supply, specialized baghouse filtration needed to capture its ultrafine particles, and growing demand from the high-performance concrete sector. Above roughly 20 weight percent replacement, it also makes fresh mixes noticeably less workable due to its enormous specific surface area, which drives up water demand and superplasticizer requirements.

Most practical mixes therefore keep silica fume below 5 to 10 weight percent, yet contents of 20 to 30 weight percent are common in UHPC, where the strength and durability gains are judged to outweigh the high unit price. This tension is exactly why the research team targeted a high silica fume window of 17.5 to 25 weight percent — the regime where the penalties are worst, and thus the most demanding test of whether a nano-additive can recover the lost performance. Their hypothesis was that carbon quantum dots could act as nanoscale organizers of the local hydration environment, adsorbing calcium ions and providing nucleation sites for C-S-H formation, thereby promoting more effective utilization of the silica fume already present in the system.

What sets carbon quantum dots apart from other carbon-based nanomaterials like carbon nanotubes (CNTs) and graphene oxide (GO) is their exceptional water dispersibility and surface chemistry. Structurally, CQDs consist of nanoscale carbon cores containing sp²-hybridized aromatic domains, with surfaces functionalized by hydroxyl and carboxylic acid groups. These oxygen-containing functionalities allow CQDs to disperse uniformly in water without sonication or surfactants — a major practical advantage over CNTs and GO, which typically require dispersing agents and are used at dosages of 0.05 to 0.5 weight percent, one to two orders of magnitude higher than the 0.02 to 0.10 weight percent employed in this study.

The experimental program was rigorous. The team used ordinary Portland cement Type I conforming to ASTM C150, silica fume with 98.36 weight percent SiO₂, and natural river sand. All mixtures maintained a constant water-to-binder ratio of 0.40, sand-to-binder ratio of 1.1, and superplasticizer dosage of 0.25 weight percent of binder. Three experimental groups were designed: one to investigate the influence of CQD dosage on composites containing 25 weight percent silica fume, one to assess whether CQDs could compensate for reduced silica fume contents of 20 and 17.5 weight percent, and one to evaluate the individual and combined effects of CQDs and silica fume using a factorial mixture design. Specimens were cast into 50 mm cube molds, cured at 20°C and above 95% relative humidity, and tested at 3, 7, and 28 days for compressive strength, workability, porosity, and water absorption.

The results were striking. In the 25 weight percent silica fume system, the optimal 0.04 weight percent CQD dosage increased 28-day compressive strength by 23.7% relative to the control, reaching 68.7 MPa. Workability improved simultaneously, with flow diameter increasing by 34.9% at the same dosage — a rare combination, since most nanocarbon additives improve strength at the expense of fresh-state performance. X-ray diffraction (XRD) and thermogravimetric analysis (TGA/DTG) confirmed that CQDs promoted C-S-H formation without altering the type of hydration products, while scanning electron microscopy (SEM) revealed progressively denser microstructures with fewer visible voids at the optimal dosage. The chemically bound water content increased from 8.41% in the control to 9.21% with 0.04 weight percent CQDs, indicating enhanced hydration-product development.

Even more remarkable was the performance at reduced silica fume contents. When the silica fume replacement level was lowered to 20 weight percent, the 0.04 weight percent CQD system achieved a 28-day compressive strength of 63.4 MPa — a 58.5% improvement over its CQD-free counterpart. At 17.5 weight percent silica fume, the enhancement reached 83.7%, with the modified composite achieving 59.1 MPa. These gains suggest that CQDs can partially compensate for the strength loss associated with lower silica fume contents by improving the efficiency of hydration-product formation from the available calcium ions, water molecules, and silicate species. Rather than acting as a direct replacement for silica fume as a SiO₂ source, CQDs function as local hydration-environment organizers that promote the precipitation and growth of hydration products, while the remaining silica fume contributes as a reactive silicate reservoir for secondary C-S-H formation.

The mechanistic investigation revealed how this works at the nanoscale. Zeta-potential measurements showed that CQDs in water carry a strongly negative surface charge of -36.26 mV, attributed to deprotonated carboxylate groups. In saturated calcium hydroxide solution, this charge shifted to -3.28 mV, indicating substantial adsorption of Ca²⁺ onto the CQD surface. Silica fume particles, which carry a negative charge of -27.86 mV in water, reversed to +24.53 mV in saturated calcium hydroxide solution due to Ca²⁺ adsorption onto negatively charged silanol surface sites. When CQDs were added to the ternary CQD-Ca²⁺-silica fume system, the zeta potential shifted to -28.27 mV and the hydrodynamic cluster size increased from 1,261 to 1,906 nanometers — evidence of Ca²⁺-mediated cation bridging between the negatively charged CQDs and silica fume surfaces. X-ray photoelectron spectroscopy (XPS) confirmed that carbon content on the silica fume surface increased from 35.59% to 47.73% after CQD modification, with enhanced C-O and HO-C=O functionalities indicating interfacial interactions between the two materials.

This interfacial organization brings Ca²⁺, reactive silicate species, and interfacial water into nanoscale proximity within the highly alkaline pore solution. Because nucleation constitutes an important kinetic barrier during C-S-H precipitation, this spatial organization provides a favorable chemical environment for the formation of incipient C-S-H domains, effectively lowering the energy barrier for nucleation and accelerating subsequent hydrate growth. The proposed mechanism suggests that CQDs do not merely act as passive nucleation substrates but function as interfacial organizers that facilitate the assembly of species required for C-S-H formation. At excessive dosages, increased CQD-CQD association or aggregation reduces the number of accessible and spatially separated functional sites, diminishing their effectiveness — which explains why 0.04 weight percent emerged as the optimum rather than higher loadings.

A two-way factorial analysis of variance (ANOVA) on 28-day compressive strength revealed that CQDs accounted for 60.2% of the total variation, compared with 25.3% for silica fume, while the CQD × silica fume interaction was not statistically significant (p = 0.211). This indicates that the performance enhancement arises primarily from the hydration-promoting effect of CQDs rather than from a strong synergistic interaction between the two additives. The combined incorporation of 0.04 weight percent CQDs and 25 weight percent silica fume achieved a 28-day compressive strength of approximately 69 MPa — a 36% improvement over the control — with the dual-additive system exceeding the sum of individual effects by roughly 10 percentage points, consistent with complementary rather than strongly synergistic mechanisms.

The economic analysis added a compelling dimension. Strength-normalized material cost (C_SN) — the total raw material cost per cubic meter divided by 28-day compressive strength — was reduced by 18.37% in the 25 weight percent silica fume system with 0.04 weight percent CQDs relative to the CQD-free control. The 20 weight percent and 17.5 weight percent silica fume systems with CQDs reduced C_SN by 8.80% and 12.33%, respectively, relative to the 25 weight percent silica fume control. A one-way sensitivity analysis showed that doubling the CQD price increased C_SN by only 0.96%, while the cost performance was governed primarily by the silica fume price due to its high replacement level. The addition of 0.04 weight percent CQDs transformed the 25 weight percent silica fume system from having a C_SN 21.86% higher than the non-additive cement composite to having a C_SN 0.53% lower, while still achieving a compressive strength of 69 MPa — a high-strength cementitious composite by any standard.

Compared with other carbonaceous additives reported in the literature, CQDs exhibit a prominent advantage. Carbon nanotube-modified systems generally achieve compressive strength enhancements of 20 to 36% at dosages of 0.02 to 0.40 weight percent, with one study reporting only a 6.15% improvement at 0.5 weight percent CNTs accompanied by a 42.1% reduction in workability. Graphene, graphene oxide, and graphene nanoplatelet systems exhibit strength enhancements ranging from 6 to 33%, with some requiring dosages as high as 1.0 to 1.5 weight percent. In comparison, the CQD-modified mixtures in this study achieved 83.7% and 58.5% strength enhancements at only 0.04 weight percent for the 17.5 and 20 weight percent silica fume systems, respectively, while also providing a 34.9% workability improvement in the 25 weight percent system — all without sonication or dispersing agents.

The implications for engineering applications are significant. High-silica fume binders are widely specified for ultra-high-performance concrete, marine structures, bridge decks, and repair mortars, but the high unit price of silica fume and its adverse effect on workability above 20 weight percent replacement have limited broader adoption. The results indicate that an ultralow CQD dosage of 0.04 weight percent addresses both constraints without requiring changes to conventional mixing or placement practice. Where the full 25 weight percent silica fume content is retained, the same dosage delivered a 34.9% increase in flow alongside a 23.7% strength gain, suggesting that superplasticizer demand may be reduced or a lower water-to-binder ratio adopted at equal flow. Because CQDs are water-dispersible and can be introduced through the mixing water using existing batching infrastructure, the approach is compatible with field conditions.

The researchers caution that the results were obtained on mortar specimens under controlled laboratory curing, and that validation at concrete scale, compatibility testing with commercial chemical admixtures, and durability assessment under chloride, sulfate, carbonation, and freeze-thaw exposure are required before the approach can be specified in practice. The economic analysis presented is a raw-material cost comparison and does not constitute a life-cycle assessment. Future work should extend CQD incorporation to concrete-scale mixtures containing coarse aggregates, evaluate long-term durability, and conduct comprehensive techno-economic and carbon-footprint assessments to determine whether the reduction in silica fume demand and improvement in strength-normalized binder efficiency can offset the cost and environmental impact of CQD production.

Nevertheless, the study demonstrates that carbon quantum dots function as local hydration organizers rather than solely as conventional fillers or passive nucleation sites. By modifying the local Ca²⁺-silicate hydration environment, CQDs can promote hydration-product development and improve the mechanical and economic efficiency of silica fume cement composites at very low dosages. The proposed hydration-organizing mechanism may not be restricted to silica fume — ordinary Portland cement inherently contains calcium silicate phases that release calcium and silicate species during hydration, suggesting that the local hydration-organization pathway could represent a more general route through which CQDs influence cement hydration. If validated at scale, this nanoscale strategy could reshape how the construction industry formulates high-performance binders, turning an expensive industrial byproduct into a more efficient and cost-effective ingredient for the sustainable concrete of tomorrow.

Subject of Research: Materials Science

Article Title: Carbon quantum dots enable efficient silica fume utilization by organizing the local hydration environment in cement composites

Article References: Charuchit, A., Nganglumpoon, R., Waiyaka, T., Jongrungrotbaworn, T., Wisuthseriwong, T., Jongvivatsakul, P., Kunawisarut, A., & Panpranot, J. (2026). Carbon quantum dots enable efficient silica fume utilization by organizing the local hydration environment in cement composites. Results in Engineering, 32, Article 113500. https://doi.org/10.1016/j.rineng.2026.113500

Image Credits: AI Generated

DOI: 10.1016/j.rineng.2026.113500

Keywords: carbon quantum dots, silica fume, cement, concrete, nanotechnology, construction materials, sustainability, Carbon, quantum, dots, enable, efficient

The technology, called Wax Foam 3D Printing (WF3DP), deliberately fuses additive and subtractive manufacturing rather than choosing between them. A robotic extrusion system deposits wax foam close to the final geometry, exploiting the material efficiency of 3D printing, while a five-axis CNC milling machine then cuts the printed preform to the precise contour and surface quality that concrete casting demands. The key innovation lies in the material itself: a wax-air composite in which roughly 20 percent of the volume is air, foamed mechanically without any chemical change. Because the wax can be remelted, filtered, and re-foamed indefinitely, the approach supports a closed-loop material cycle at the process level, not merely as an aspiration.

The hardware is substantial. The custom WF3DP unit integrates a heated reservoir holding molten wax at about 90 degrees Celsius, a mechanical foaming unit that whips filtered wax together with air and pre-cools the mixture into a homogeneous semi-solid foam, and a temperature-controlled delivery hose feeding a 12-millimetre extrusion nozzle. The print head is carried by an industrial six-axis KUKA robot with a working volume of 1.6 by 1.2 by 2 metres. Immediately before extrusion, the foam passes through a localized cooling stage regulated to within one degree Celsius, bringing it slightly below its freezing point so that a partially crystallized microstructure gives the extrudate dimensional stability and clean layer formation.

Throughput is impressive for a feasibility study. Stable deposition was achieved at rates of up to 28.5 kilograms per hour, and in the case study roughly 250 kilograms of wax foam were deposited in a continuous print lasting 8 hours and 45 minutes. The printed block measured about 1050 by 1550 by 500 millimetres, including a 20-millimetre offset for milling. Cooling from 65 to 25 degrees Celsius took 3 days and 21 hours, a dramatic improvement over the more than two weeks expected for a comparable solid wax block, whose enormous thermal mass must dissipate its heat slowly.

The material choice proved decisive. The team characterized a new formulation, Tudastar wax, against the ConFormWax (CFW) used in a decade of earlier wax formwork research at Braunschweig. CFW wax has a freezing range of only 62 to 66 degrees Celsius and begins softening well below it; in earlier trials with massive elements up to 1.5 tonnes, hydration heat from curing concrete had softened and partially melted the formwork surface, preventing reuse. Tudastar wax freezes at approximately 81 degrees Celsius, providing a much larger thermal margin. In compressive tests, solid Tudastar wax reached 4.69 newtons per square millimetre, and foaming with 20 percent air reduced its strength by only 6 percent, to 4.40, whereas foamed CFW wax lost 48 percent of its strength. Foamed Tudastar wax therefore carried 2.7 times the load of foamed CFW wax.

Temperature-dependent testing revealed how much of that strength evaporates as the wax warms. Printed Tudastar foam specimens softened from 3.59 newtons per square millimetre at room temperature to 3.03 at 35 degrees and 1.72 at 45 degrees, less than half the room-temperature value, with the failure mode shifting from partly brittle cracking to ductile deformation. Yet even at 45 degrees, the printed foam retained the room-temperature strength of foamed CFW wax, and the measured formwork temperatures during the case study, peaking at 40 to 47 degrees near the wax-concrete interface, stayed well within Tudastar’s thermal margin. Surface-energy measurements using the Owens-Wendt-Rabel-Kaelble method showed Tudastar wax to be almost entirely nonpolar, with a polar component of just 0.25 millinewtons per metre, suggesting favourable demoulding behaviour without release agents.

The full-scale case study put these properties to the test. A cooperating precast manufacturer commissioned a customized radial seating step, a two-step double-curved element with an inner radius of 115 centimetres and an outer radius of 155 centimetres, specified in fair-faced concrete class SB3, a demanding architectural surface standard. The conventional reference formwork had been built manually by a single carpenter in about 1.5 days from plywood and modelling foam, designed for up to 15 casting cycles but reused for only seven before being discarded as non-recyclable waste. The wax route instead printed a near-net-shape preform, scanned it in 3D, and finish-milled it in about 1.5 hours on a five-axis machine.

The print was not flawless. Edges warped upward by as much as 50 millimetres, producing a bowl-shaped deformation, and cracks developed between deposited strands where interlayer bonding proved insufficient. During machining, delamination occurred along layer interfaces, requiring local repair with molten wax, infrared reheating to densify the surface, and remachining. These defects matter because fine gaps between printed layers allowed cement paste and mixing water to penetrate by capillary action during casting, causing localized discoloration and an alternating rough and smooth texture on the demoulded concrete. The element reproduced its complex radial geometry accurately with sharply defined edges and no adhesion damage, but it met the SB3 surface requirements only in part and would need remedial treatment to comply fully.

Demoulding after roughly 23 hours of curing confirmed the expected release behaviour: no release agent was applied to the wax, and no concrete residue or adhesion was detected, though wooden wedges and mechanical force were needed to initiate separation, causing localized surface damage that prevented direct reuse of the formwork for a second casting. The team instead closed the material loop at process level: the used formwork was completely remelted in an 8-kilowatt tank, pumped through fine filtration units to remove concrete particles, re-foamed, and printed into new formwork elements for ongoing research. Earlier investigations had already shown CFW wax withstanding more than 200 remelting cycles without noticeable property changes, and comparable behaviour is expected for Tudastar wax provided oxidation is avoided.

The resource accounting is striking. Producing a formwork element of comparable size by the solid wax block route would have required roughly 1.13 tonnes of wax, most of it machined away without ever contributing to the casting surface. The near-net-shape printed route consumed about 250 kilograms, an approximately 80 percent reduction. The authors estimate that with structural infill patterns and dedicated slicing software, printing time could fall to about 2.5 hours and cooling to roughly one day, cutting the waiting time until the formwork is available to about one-tenth of the solid-block route. Labour input, currently high at around 17 hours including path planning and repair, is expected to become comparable to conventional routes once automation matures. The study identifies interlayer bonding, warpage control, geometric accuracy, surface quality, and damage-free demoulding as the principal hurdles between this prototype and an industrial-ready technology, but it establishes something concrete: a formwork that is printed, milled, cast, and melted down again, within a largely digital process chain powered by on-site solar energy.

Subject of Research: Materials science

Article Title: A novel formwork technology based on wax foam 3D printing and CNC milling

Article References: Jonischkies, S., Mainka, J., Kloft, H., Diesslin, R., & Hack, N. (2026). A novel formwork technology based on wax foam 3D printing and CNC milling. Frontiers in Built Environment, 12, Article 1934051. https://doi.org/10.3389/fbuil.2026.1934051

Image Credits: AI Generated

DOI: 10.3389/fbuil.2026.1934051

Keywords: 3D printing, concrete, formwork, recycling, CNC milling, circular economy, novel, technology, based, foam, printing, milling

Mill scale is generated at roughly 1 to 3 percent of the mass of steel rolled, equivalent to about 22 to 40 kilograms per tonne. Applied to global crude steel production of 1,882.6 million tonnes in 2024, that implies something on the order of 19 to 56 million tonnes per year, consistent with two independent published estimates of approximately 37 and 39 million tonnes. Because concrete is produced at an even larger scale, even marginal substitutions of natural sand with mill scale could be consequential — and the material has attracted interest on three distinct grounds: as a partial replacement for fine aggregate, as a dense additive for heavyweight and radiation-shielding concretes, and, more speculatively, as a source of limited chemical reactivity in alkaline pore solution.

One of the review’s most striking technical contributions is resolving a confusion that has plagued the construction-materials literature for years. Studies reporting mill scale as “93 to 99 percent Fe2O3” and studies reporting it as roughly half wüstite (FeO) and half magnetite (Fe3O4) are not in conflict at all — they are simply using different measurement conventions. X-ray fluorescence returns elemental concentrations that are conventionally normalised and reported as oxides, with all iron reported arithmetically as Fe2O3 regardless of its actual oxidation state. X-ray diffraction with quantitative phase analysis, by contrast, resolves the actual crystalline phases present. The single quantitative phase analysis of as-received mill scale the authors located reports FeO at 47.81 percent, Fe3O4 at 51.34 percent and metallic Fe at 0.85 percent — magnetite marginally exceeding wüstite, with hematite not present as a major phase.

The review also documents that mill scale is a genuinely variable material with identifiable causes of variability. Steel grade and alloy content govern the trace-metal inventory; the process route — basic oxygen furnace versus electric arc furnace, hot rolling versus wire drawing — sets the thermal and mechanical history; coiling temperature and cooling rate control the eutectoid decomposition of wüstite below approximately 570 degrees Celsius and therefore the final phase assemblage; descaling method and water contact determine particle size distribution and surface hydroxide formation; and handling, storage and oil contamination introduce rolling oils and greases that can impair the paste-particle bond. Stockpile weathering causes progressive oxidation and hydration, arguing for per-batch characterisation.

When the authors mapped what the primary literature actually reports about mechanical, durability and environmental performance, they found the reported optimum replacement level is not consistent across the corpus. Eighteen primary studies met the eligibility criteria, spanning mortars, conventional and high-performance concretes, self-compacting concrete, geopolymer concrete and precast paving blocks at replacement levels from 5 to 100 percent. Four independent studies converge on approximately 15 percent by mass or volume as a strength optimum in mortar and self-compacting concrete — the strongest single pattern in the corpus. But several conventional concrete studies report optima at 50 to 60 percent, and at least two report monotonic strength loss. The review treats this discrepancy as a genuine open question, most likely reflecting the interaction between the fineness of the mill scale and the gradation of the sand it replaces.

Workability decreases with increasing mill-scale content, consistently across studies — the most reliably reported effect. One study quantifies a 74 percent slump reduction at 60 percent replacement. The mechanism combines the flake morphology and rough surface texture of mill scale, which increase internal friction, with surface hydroxide layers and any entrained rolling oil, which alter water demand and the paste-particle interface. Because workability was managed differently across studies — some held water constant, some held slump constant, some added superplasticiser — the strength results across the corpus are not comparable without accounting for this, which the review treats as a first-order confounder.

Durability evidence is thin and unevenly distributed. Post-fire residual performance, thermal behaviour and acid resistance are reported as favourable by more than one group. Chloride evidence exists but rests substantially on the rapid chloride permeability test, whose interpretation is questionable for a matrix containing an electrically conductive, iron-bearing phase — the test measures total charge passed, which is a function of pore-solution conductivity as well as pore structure. Carbonation depth and freeze-thaw performance are unreported in the entire corpus. Most consequentially, no electrochemical study of reinforcement corrosion in mill-scale concrete was located anywhere — an iron-bearing admixture in reinforced concrete raises a corrosion question that has not been asked, let alone answered, and the review places it first among research priorities.

Quantitative pooling across the literature is currently obstructed by six specific and documentable deficiencies: heterogeneous replacement bases (mass versus volume; aggregate versus binder), inconsistent curing ages and outcome definitions, uncharacterised variation in mill-scale provenance, near-universal absence of dispersion statistics and replicate counts, confounding of mill scale with co-additives such as graphite nano/microplatelets or steel fibres, and probable non-independence among studies from single research groups. Of the eighteen studies, only two companion papers from one campaign report replicate counts, error bars and inferential statistics. A random-effects meta-analysis cannot be estimated from a corpus in which the variance components are unrecoverable.

To make future synthesis possible, the review specifies a pre-registered meta-analytic protocol — effect-size definition using Hedges’ g or raw mean difference, dose-response meta-regression, random-effects estimation with restricted maximum likelihood, heterogeneity quantification, pre-specified subgroup analysis, distributional diagnostics, sensitivity analysis and small-study-effect assessment — to be executed once the reporting deficiencies are remedied. It also proposes a minimum reporting standard for primary studies, requiring plant and process provenance, composition with the measurement technique named and the phase versus elemental basis stated explicitly, full particle size distribution, specific gravity, water absorption, chloride content, oil and grease content, complete mix proportions, and — for every outcome at every age — mean, standard deviation and number of specimens. Not mean alone. Not a bar chart without error bars.

What the evidence currently supports is cautious use of mill scale in non-structural and specialty applications. Four independent studies report shielding benefits attributable to mill scale’s density and iron content, including a linear attenuation coefficient of 0.085 per centimetre in a geopolymer composite and electromagnetic-interference shielding of 26.37 decibels at 11.82 gigahertz at 30 percent replacement in mortar. Notably, the shielding optimum lies at high replacement while the mechanical optimum in mortar lies near 15 percent — the two do not coincide, so a shielding mix must be designed against a strength requirement rather than assumed adequate. Radiation and EMI shielding, heavyweight and ballast concretes, mortars and renders at moderate replacement, and non-structural elements including paving blocks are the most defensible near-term applications. Structural reinforced concrete is not supported at any replacement level. The economics favour ready-mixed or precast operations near integrated steel plants, because the material is dense and transport cost scales with mass — the viable market is geographically bounded by its own density. The qualitative industrial-symbiosis case is strong; the quantitative case has not yet been made. But the reporting problem is solvable, and solving it is a precondition for the field to know what it collectively knows.

Subject of Research: Materials Science

Article Title: Steel mill scale in concrete and mortar: a critical review of the evidence base and a proposed framework for future quantitative synthesis

Article References: Loganathan, G., Senthil Pandian, M., & Deepa Nivethika, S. (2026). Steel mill scale in concrete and mortar: a critical review of the evidence base and a proposed framework for future quantitative synthesis. Frontiers in Built Environment, 12, Article 1932304. https://doi.org/10.3389/fbuil.2026.1932304

Image Credits: AI Generated

DOI: 10.3389/fbuil.2026.1932304

Keywords: steel mill scale, concrete, mortar, circular economy, construction materials, evidence synthesis, waste valorisation, durability, Steel, mill, scale, critical

The research team, led by Vishnu Vardhan Kasibhatla and Mohan Ganesh Gopal of the Department of Structural and Geotechnical Engineering, prepared three concrete mixes: one with no graphene oxide, one with 0.04 percent by weight of cement, and one with 0.08 percent. The graphene oxide itself was characterized before mixing, revealing wrinkled sheets roughly 500 nanometers across and about 5 nanometers thick, with an oxygen content near 21 percent. Those oxygen-bearing functional groups are the secret to the material’s effectiveness, because they give the sheets a strong chemical affinity for the calcium silicate hydrate gel—the glue-like compound that gives concrete its strength.

Getting nanoparticles to spread evenly through wet concrete is notoriously difficult, so the team took careful measures. The graphene oxide powder was weighed on a balance accurate to a thousandth of a gram, stirred into water, and then sonicated for twenty minutes in an ice bath to prevent overheating. A zeta potential measurement of −39.16 millivolts confirmed the resulting suspension was highly stable and well dispersed. Only then was the treated water gradually added to the dry ingredients—ordinary Portland cement, fine aggregate, and coarse aggregate in a mix designed for M25-grade concrete with a water-to-cement ratio of 0.46.

The mechanical results were striking. At 28 days, the concrete containing 0.04 percent graphene oxide reached a compressive strength of 40.6 megapascals, a 30.97 percent increase over the untreated control mix. Even more telling was the pace of that gain: at 7 days, the modified concrete was already 44 percent stronger than the control, suggesting the nanomaterial accelerates early hydration rather than merely strengthening the final product. The 0.08 percent mix also beat the control, but fell short of the lower dosage—a hint that more is not always better when it comes to nanoscale additives.

Scanning electron microscopy explained why. In the control mix at 7 days, the researchers saw loosely packed outer calcium silicate hydrate resembling cloud-like masses, visible gel pores, partially hydrated cement grains, and microcracks concentrated in porous zones. In the 0.04 percent mix, the picture changed dramatically: the calcium silicate hydrate gel appeared denser, with fibrillar crystalline networks that were more interconnected and better bridged across voids. The graphene oxide sheets, the authors conclude, act as nucleation sites—tiny platforms that give hydration products a head start, converting loose outer gel into tighter inner gel and reducing the pore connectivity that normally weakens the matrix.

At 14 and 28 days, the trend continued. The modified concrete showed reticular inner calcium silicate hydrate interwoven with thread-like crystalline fibres, which the researchers interpret as growth templates for further gel formation. Crack initiation within the gel was comparatively rare, indicating improved early-stage matrix integrity. But the 0.08 percent mix told a cautionary tale: localized agglomerations of clustered hydration products appeared, creating morphological irregularities that could concentrate micro-stresses and undermine the very uniformity the nanomaterial is meant to enhance. Excess graphene oxide, in other words, gets in its own way.

To connect these microscopic observations to macroscopic behavior, the team built a three-dimensional finite element model of a 150-millimeter concrete cube in COMSOL Multiphysics. The model treated each mix as a homogeneous, linearly elastic solid, with Young’s modulus derived from the experimentally measured compressive strength, a Poisson’s ratio of 0.2, and a density of 2,400 kilograms per cubic meter. The bottom surface was fully constrained while the top was pushed downward by a prescribed displacement of 0.165 millimeters—enough to keep the material within its elastic range. A mesh sensitivity analysis confirmed the results converged, with reaction forces varying by less than 0.01 percent across successive mesh refinements.

The simulation revealed a clear stiffness hierarchy. Under identical displacement, the control cube resisted with a reaction force of 697.88 kilonewtons, while the 0.04 percent graphene oxide cube pushed back with 798.65 kilonewtons—a 14.44 percent increase that reflects greater resistance to elastic deformation and improved load-bearing capacity. The 0.08 percent mix fell in between at 770.62 kilonewtons. Von Mises stress contours peaked at roughly 7.0 × 10⁷ newtons per square meter for the optimal mix, compared with about 6.0 × 10⁷ for the control, with the sharpest concentrations appearing near the constrained base edges—a boundary effect rather than a material flaw.

The authors are careful about what these numbers mean. Because the elastic moduli fed into the model were derived from the measured compressive strengths, the higher reaction force should be read as complementary numerical evidence of stiffness-dependent elastic behavior, not as an independent prediction of strength. The finite element analysis captures how a stiffer material responds under the same deformation, not how it fails. That distinction matters for engineers hoping to translate laboratory nanomodification into structural design codes, where the difference between stiffness and ultimate capacity is fundamental.

Still, the combined picture is compelling. A dosage of just 0.04 percent graphene oxide—four hundredths of one percent of the cement’s mass—produced a denser, more interconnected hydration network, faster early strength gain, and a stiffer elastic response, while doubling the dose reversed some of those benefits through agglomeration. As graphene oxide production scales up and costs fall, the study suggests that the future of concrete may lie not in pouring more material, but in engineering its microstructure one atomically thin sheet at a time.

Subject of Research: Effect of graphene oxide dosage on the microstructure and elastic mechanical response of concrete, studied by scanning electron microscopy and finite element modelling.

Article Title: Microstructural evolution and elastic response of graphene oxide–modified concrete: SEM and finite element modelling using COMSOL multiphysics

Article References: Kasibhatla, V. V., & Gopal, M. G. (2026). Microstructural evolution and elastic response of graphene oxide–modified concrete: SEM and finite element modelling using COMSOL multiphysics. Frontiers in Nanotechnology, 8, Article 1957798. https://doi.org/10.3389/fnano.2026.1957798

Image Credits: AI Generated

DOI: 10.3389/fnano.2026.1957798

Keywords: graphene oxide, concrete, cement, calcium silicate hydrate, compressive strength, microstructure, scanning electron microscopy, finite element analysis, COMSOL Multiphysics, nanomaterials, elastic modulus, construction materials

The core difficulty is class imbalance. In ordinary concrete CT images, voids — the pores and microcracks that govern how water, chlorides and gases migrate through a structure — occupy only about 3.2 percent of the pixels. Cement paste fills roughly 64.4 percent and aggregate the remaining 32.4 percent. When a three-class network is trained on such skewed data, the optimizer can satisfy itself by simply predicting the dominant classes everywhere. The study’s own controlled experiment makes this vivid: a single-stage Vision Transformer–U-Net hybrid, trained on exactly the same data split, preprocessing and loss functions as the proposed method, collapsed onto the majority aggregate class. It achieved just 31.81 percent overall accuracy and a mean Intersection over Union of 10.61 percent, with F1-scores of essentially zero for both void and paste. In other words, the model learned to call nearly everything aggregate and nothing else.

The proposed alternative decomposes the problem along the physical hierarchy of the material itself. Stage 1 is a binary task: separate void from non-void, exploiting the fact that air and cracks attenuate X-rays weakly and appear dark. Stage 2 then classifies the remaining pixels as paste or aggregate, with void pixels masked out of the loss entirely through an ignore index so they cannot contaminate the second stage’s learning. At inference, the two predictions are merged: anything Stage 1 calls void becomes void, and everything else receives the Stage 2 verdict. Each stage therefore faces a milder imbalance than the original three-class problem, and void no longer competes with paste–aggregate boundaries for the network’s attention during optimization.

Both stages share the same backbone: a Vision Transformer encoder (ViT-B/16, pretrained on ImageNet) paired with a convolutional U-Net-style decoder. The choice is deliberate. Self-attention lets the encoder aggregate long-range context across a 224-by-224-pixel patch, useful when distinguishing fine pores from large aggregate grains, while the upsampling decoder reconstructs dense, full-resolution masks. Because the Vision Transformer produces a flat sequence of patch tokens rather than multi-scale feature maps, the decoder omits encoder skip connections and instead applies four bilinear upsampling blocks that rebuild the image from a 14-by-14 feature grid. Each stage carries roughly 86.6 million trainable parameters, all fine-tuned end-to-end on concrete patches with no concrete-specific pretraining.

Training was equally careful about imbalance. Each stage optimizes a weighted blend of cross-entropy, focal, Dice and IoU losses, with coefficients of 0.5, 0.25, 0.15 and 0.10 respectively. Stage 1 up-weights void pixels fifteenfold; Stage 2 mildly favors aggregate. The optimizer is AdamW with a learning rate of 5×10⁻⁵ under cosine annealing, early stopping on validation loss, and mixed-precision training in PyTorch. Crucially, the single-stage baseline received the same recipe, including void-upweighted losses — showing that reweighting alone, a common remedy in the literature, could not restore minority-class performance under this extreme skew. The gains come from task structure, not hyperparameter tuning.

The results on 151 held-out test patches are striking. The two-stage pipeline reached 94.58 percent overall accuracy, 87.50 percent mean IoU and 93.31 percent mean Dice, with class-wise F1-scores of 0.92 for void, 0.96 for paste and 0.92 for aggregate. Void recall hit 96.4 percent — the number that matters most, since missed pores translate directly into underestimated porosity and unreliable durability predictions. Bootstrap resampling over the test patches showed tiny standard deviations, with overall accuracy at 94.90 percent plus or minus 0.07 percent, indicating the performance is stable rather than a lucky draw. An ablation confirmed that removing either stage causes substantial degradation: Stage 2 alone cannot find voids at all, and Stage 1 alone cannot separate aggregate from paste.

The study also stress-tested the framework beyond its algorithmic training labels. On a second corpus of 82 images with human-prepared masks, held out twelve at a time for testing, the two-stage model outperformed a traditional enhanced-mask baseline on both 8-bit inputs and percentile-normalized 16-bit inputs, with the two encodings performing nearly identically on overall metrics. A full-slice demonstration on an unseen CT image predicted phase fractions of 3.64 percent void, 59.38 percent paste and 36.98 percent aggregate — closely tracking the training distribution and suggesting the pipeline can plausibly map whole specimens when preprocessing matches training assumptions. The authors are careful, however, to frame this as transfer plausibility rather than certified generalization.

The error analysis is refreshingly honest about where the method still struggles. The largest residual confusion lies at paste–aggregate interfaces, where similar X-ray attenuation and partial-volume mixing produced more than 365,000 swapped pixels in the test confusion matrix. Thin voids and microcracks are sometimes absorbed into paste, and beam-hardening streaks near dense aggregate can trigger false void calls. Because the merge rule forces every Stage 1 void prediction to remain void, errors cascade: the 3.6 percent of void pixels missed by Stage 1 can never be recovered, and false voids bypass Stage 2’s judgment entirely. For engineering use, the authors note, bulk porosity and aggregate grading from well-resolved regions are far more trustworthy than interface statistics or sub-resolution pore counts without local audit.

The limitations define a clear research agenda. All primary data come from a single specimen, one scan and one mix design, with patch-level rather than slice-level splitting, so neighboring windows from the same CT slice can straddle train and test folds. The reference labels themselves are algorithm-generated pseudo-labels, verified against manual annotation on only a 57-patch subset (88.64 percent pixel agreement), and the manual-label external check rests on just twelve test images. Inference is strictly two-dimensional, with no inter-slice consistency, and the two forward passes roughly double the computational cost of a one-pass model — 27.9 versus 14.6 milliseconds per patch on an RTX 2060, acceptable for offline analysis but not real-time deployment.

Even within those bounds, the study delivers a lesson that resonates far beyond concrete. When one class is vanishingly rare, the fix may not be a cleverer loss or a bigger model but a restructuring of the task itself — teaching the network to find the rare thing first, and only then sort the rest. As infrastructure agencies worldwide push toward digital twins of aging bridges, tunnels and nuclear containment structures, the microscopic pores inside concrete are where durability is decided. A two-stage trick that finally makes those pores visible to machines could quietly become one of the more consequential pieces of AI in civil engineering.

Subject of Research: Two-stage deep learning segmentation of void, paste and aggregate phases in concrete X-ray CT images under severe class imbalance

Article Title: Two-stage deep learning for segmenting in concrete X-ray CT images

Article References: Wang, Y. (2026). Two-stage deep learning for segmenting in concrete X-ray CT images. Results in Engineering, 32, Article 113430. https://doi.org/10.1016/j.rineng.2026.113430

Image Credits: AI Generated

DOI: 10.1016/j.rineng.2026.113430

Keywords: concrete, X-ray computed tomography, deep learning, image segmentation, Vision Transformer, U-Net, class imbalance, porosity, cement paste, aggregate, civil engineering, digital twin

The research, led by R. Daniel Verdugo-Siqueiros with Luis Pellegrin and Jose Angel Gonzalez-Fraga, is explicitly data-centric rather than model-centric. Instead of inventing yet another neural architecture, the authors asked a different question: what if the real obstacle to automated crack inspection is not the detector but the scarcity of well-labeled examples spanning diverse materials and environmental conditions? Their answer was to exploit the zero-shot generalization abilities of foundational vision models, which have been trained on such vast image collections that they can perform tasks they were never explicitly taught, and to augment limited real training data with synthetic examples generated by artificial intelligence itself.

At the heart of the pipeline sits Segment Anything Model 2, or SAM2, Meta’s celebrated segmentation system capable of isolating objects in images and videos from simple prompts. SAM2 provides the framework’s precise, prompt-driven segmentation capability, tracing the often hairline, branching geometry of cracks that would defeat conventional thresholding methods. But SAM2 alone cannot know where to look. That role falls to Florence2, a unified vision foundation model developed at Microsoft Research, which supplies robust feature detection and localization, effectively telling the segmentation module which regions of a photograph deserve scrutiny. The pairing mimics a human inspector’s workflow: first scan the surface for suspicious marks, then examine each candidate closely.

The third component is perhaps the most surprising. Stable Diffusion XL, the high-resolution image synthesis model best known for generating photorealistic artwork, is repurposed here for two jobs: super-resolution enhancement of real inspection images and the generation of synthetic crack examples to flesh out sparse training sets. By upscaling imagery before analysis, the system preserves the fine detail that thin cracks demand. By manufacturing plausible synthetic cracks, it sidesteps the annotation drought that has plagued the field, where datasets covering concrete, asphalt, masonry and ceramics simultaneously are nearly nonexistent.

The numbers reported in the paper are striking for a system built largely from off-the-shelf components. Detection accuracy and precision both reach 92.3 percent, with recall of 89.7 percent and an F1-score of 91.0 percent, meaning the framework misses relatively few genuine cracks while raising few false alarms. The segmentation module, which must trace exact crack boundaries rather than simply flag their presence, starts at a mean Intersection over Union of 63 percent, a Dice coefficient of 77.2 percent, and a boundary F1-score of 68.1 percent. Those boundary metrics matter enormously in practice, because crack width is a key input to structural condition ratings, and sloppy edges translate directly into misjudged severity.

What elevates the work beyond a static benchmark is its use of active learning. Across three successive learning cycles, the segmentation performance climbs from an initial mIoU of 63 percent to 72 percent. In an active learning loop, the model flags the examples it is least confident about, human annotators label only those, and the model retrains on the enriched set. This targeted feedback means each hour of expert labeling yields maximum benefit, progressively refining the model exactly where it struggles, such as shadows, stains, and surface textures that masquerade as cracks. The authors describe this as a reflection of progressive refinement, and the trajectory suggests further gains with additional cycles.

The broader context explains why this matters. The paper’s own references trace a decade of effort, from classical image-processing techniques built on Otsu thresholding and differential images, through deep convolutional networks like DeepCrack and transformer-based systems such as Crackformer, to recent fine-tuning of vision foundation models for civil infrastructure. Each generation improved sensitivity, but each also inherited the same fundamental constraint: models are only as good as the labeled cracks they see. Surveys of crack detection technologies from 2022 and 2023 document the field’s rapid growth alongside persistent generalization failures, where a model trained on one pavement type stumbles on another. A framework that leans on zero-shot capability and synthetic augmentation is a direct strike at that weakness.

Computationally, the system is refreshingly lean. Processing high-resolution images in approximately 2.3 seconds on a standard graphics card means the pipeline could plausibly run on the kinds of hardware mounted on inspection drones, robots, or even field laptops, rather than being confined to cloud clusters. That opens the door to comprehensive monitoring programs in which bridges, tunnels, dams and highway networks are photographed routinely and analyzed automatically, with cracks identified, measured and tracked over time. The authors report robust performance across diverse materials and environmental conditions, the variability that has historically broken narrower systems. Data and models are available on request, and the team maintains a public GitHub repository, smart-infrastructure-inspector, dedicated to the work.

The stakes are not abstract. The collapse of the I-35W bridge in Minneapolis in 2007 and the 2021 closure of the I-40 bridge over the Mississippi River after a cracked tie girder was discovered put the cost of undetected deterioration on national display. Probability-of-detection studies for visual inspection of steel bridges have shown how inconsistent human inspection can be, and maintenance economics research repeatedly finds that early, inexpensive intervention prevents catastrophic, budget-breaking repairs. An automated inspector that never tires, never skips a hairline fracture in poor lighting, and improves with every labeled example could shift infrastructure management from reactive crisis response toward genuine predictive maintenance.

There are, of course, caveats. The reported metrics come from the authors’ own evaluation, and the paper does not yet demonstrate deployment at the scale of an entire highway network or a fleet of aging bridges. Synthetic data, however cleverly generated, risks embedding biases if the diffusion model’s imagined cracks diverge from real failure modes, and human annotators remain essential to the active learning loop. Still, the conceptual move is hard to dismiss: rather than waiting for a perfect crack dataset that may never exist, the framework manufactures its own training material and borrows perception from models trained on billions of images. As foundational vision models continue to mature, the authors’ integration strategy, pairing detection, segmentation and generation in a single data-centric loop, may well become the template for automated inspection across civil engineering, and a quiet but meaningful step toward infrastructure that tells us it is failing before it does.

Subject of Research: Automated crack detection and instance segmentation in construction materials using foundational vision models

Article Title: A data-centric framework for automated crack instance segmentation in construction materials via integration of foundational vision models

Article References: Verdugo-Siqueiros, R. D., Pellegrin, L., & Gonzalez-Fraga, J. A. (2026). A data-centric framework for automated crack instance segmentation in construction materials via integration of foundational vision models. Multimedia Tools and Applications, 85(10), Article 807. https://doi.org/10.1007/s11042-026-21964-6

Image Credits: AI Generated

DOI: 10.1007/s11042-026-21964-6

Keywords: crack detection, instance segmentation, foundation models, SAM2, Florence2, Stable Diffusion XL, infrastructure monitoring, computer vision, active learning, synthetic data, civil engineering, concrete

The research, led by M. Vignesh, S.S. Alan Prince, Christina Elangbam and Vasugi K, arrives at a moment when the concrete industry is undergoing a profound compositional transformation. To cut the enormous carbon footprint of cement production, manufacturers increasingly replace a portion of cement with supplementary cementitious materials, or SCMs, such as fly ash, silica fume and ground-granulated blast-furnace slag. These industrial by-products lower CO2 emissions while improving long-term strength and durability. Chemical admixtures, including high-range water reducers and accelerating agents, have expanded the design space even further, allowing engineers to push water-to-binder ratios lower and achieve strength faster without sacrificing workability. The trouble is that the classical empirical tools of the trade, from the Abrams water-cement law to the ACI 211.1 mix-design charts and the prescriptive provisions of EN 206, were calibrated for simpler binder systems. With ternary and even quaternary blends now common, mixes routinely fall outside the calibration range of the old equations.

Machine learning has long promised a way out, and neural networks, support vector regression and tree-based ensembles have all been applied to concrete strength prediction, with reported coefficients of determination on held-out test sets approaching or exceeding 0.90. But three persistent problems have kept these methods out of everyday engineering practice. Most published studies rely on small, single-laboratory datasets of uncertain provenance; validation typically rests on a single train-test split rather than rigorous cross-validation and residual diagnostics; and the most accurate models, gradient-boosted ensembles, are widely regarded as black boxes that offer no transparent account of why they predict what they predict. Before structural engineers and code committees can trust an algorithm’s output, they need physically justifiable relationships between inputs and outputs. The new study tackles all three shortcomings at once.

The foundation of the work is a dataset of 1,670 mix records drawn from the field ledger of the Guangxi Road and Bridge Group’s Road and Bridge Pavement Branch, deposited via Mendeley Data. Because the records come from normal construction practice rather than a controlled laboratory programme, they capture strength outcomes as they actually occur on site, including the effects of curing conditions, compaction and specimen handling. Each record contains nine measured variables: cement, fine aggregates, coarse aggregates, water, water-reducing admixture, fly ash, accelerating agent, silica fume and curing time. The target compressive strength spans a remarkable 4.3 to 76.3 megapascals, with a mean of 37.62 megapascals, covering everything from low-grade formulations to high-performance mixes. Curing ages cluster bimodally at 7 and 28 days, which allowed the model to learn age-dependent strength gain in a single training exercise. Crucially, no records were discarded and no outlier removal was performed, preserving the messiness of reality.

On top of the nine raw variables, the team engineered three physics-informed features drawn directly from classical concrete theory. The water-cement ratio encodes the Abrams law, sparing the algorithm from having to rediscover the single most famous relationship in concrete technology. The cement ratio, the binder’s share of the total aggregate skeleton, governs paste pressure and the density of the interfacial transition zone where strength-critical failure often begins. And the total cementitious content, the sum of cement, fly ash and silica fume, captures the full pool of hydration-active material available for long-term strength development. The final feature matrix thus held twelve predictors, and all were standardised using parameters fitted exclusively to the training data to prevent information leakage, with the strength target left unscaled so that error metrics remained directly interpretable in megapascals.

Six regression algorithms were then benchmarked on identical data partitions with a fixed random seed: linear regression, k-nearest neighbours, support vector regression with a radial-basis-function kernel, random forest, LightGBM and XGBoost. The hierarchy that emerged was textbook-clear. Linear regression managed only a coefficient of determination of 0.7755 with a root-mean-square error of 5.81 megapascals, its ordinary least-squares machinery simply unable to capture the nonlinear, interaction-dominated response surface. Support vector regression and k-nearest neighbours landed in the mid-0.83 range, while random forest crossed the 0.90 engineering-grade threshold. LightGBM pushed to 0.9209, and XGBoost led the untuned field at 0.9243 with a root-mean-square error of 3.37 megapascals. Ten-fold cross-validation confirmed the ranking, with XGBoost achieving a mean coefficient of determination of 0.9127 plus or minus 0.0318 across folds, evidence that the accuracy reflected genuine learned trends rather than a lucky test-set composition.

Hyperparameter optimisation via a five-fold grid search over 27 combinations, 135 model fits in total, nudged XGBoost to a final held-out performance of R-squared 0.9263, root-mean-square error 3.329 megapascals and mean absolute error 2.40 megapascals. The optimal configuration, 300 estimators, a maximum tree depth of 4 and a learning rate of 0.1, favoured shallower trees with faster learning than the baseline, consistent with the known behaviour of gradient boosting on moderately sized datasets. Residual diagnostics showed errors centred on zero across the entire 4-to-70-megapascal prediction range, with an essentially Gaussian residual distribution and no directional bias. Of the 334 test samples, 76 percent fell within 10 percent of the measured strength and 92.8 percent within 20 percent. Range-wise analysis added a note of caution: in the narrow 40-to-55-megapascal band the coefficient of determination dropped sharply, though absolute errors stayed low, a reminder that R-squared must always be read alongside error metrics and sample counts.

The real novelty lies in what came next: a three-layer explainability analysis. Global feature importance computed across four model families converged on the same ranking, with the water-cement ratio leading in both random forest and XGBoost, followed by total cementitious content, curing time and water. SHAP values, rooted in cooperative game theory, then quantified each feature’s contribution for every individual prediction. The water-cement ratio produced SHAP values spanning from minus 25 to plus 15 megapascals, the widest influence of any predictor: low ratios pushed predictions well above the 37.87-megapascal baseline while high ratios imposed penalties of 10 to 15 megapascals, a modern, multi-constituent confirmation of the century-old Abrams law. Curing time contributed negatively for 7-day specimens and positively for 28-day ones, and partial-dependence plots traced a near-linear strength gain from roughly 33.5 megapascals at 7 days to 40.5 megapascals at 28 days. The water partial-dependence curve revealed a strikingly nonlinear, monotonically declining response consistent with the classical gel-space ratio concept of paste porosity.

Individual waterfall decompositions show the practical payoff. For one weak mix that actually achieved 22.6 megapascals, the model predicted 26.7, and the explanation was legible in engineering terms: total cementitious content subtracted 4.16 megapascals from the baseline, the water-cement ratio another 4.00, short curing 3.28, and an unfavourable fly-ash dosage 1.27 more. The authors are careful to interpret such attributions correctly, noting that the sparse, zero-inflated field data on fly ash and silica fume, present in only 34.7 percent and 1.1 percent of records respectively, mean the model captures directional pozzolanic sensitivity rather than a validated dose-response curve. Independent microstructural studies, which found strength optima near 30 percent fly ash and 7.5 to 12 percent silica fume replacement, provide complementary mechanistic context. The team is equally candid about limitations: no new laboratory programme was conducted, so validation is statistical rather than experimental, and the web-based implementation is a proof-of-concept decision-support tool, not a replacement for standard strength testing. Still, by fusing transparent field data, physics-informed feature engineering, systematic benchmarking and multi-level explainability into one reproducible framework, the study sketches what trustworthy artificial intelligence for concrete design could look like, an algorithm that not only predicts strength within about three megapascals but can show its work, one mix at a time.

Subject of Research: Explainable machine learning prediction of compressive strength in SCM-enhanced concrete at the mix-design stage

Article Title: Explainable machine learning for mix-design-stage prediction of compressive strength in SCM-enhanced concrete

Article References: Vignesh, M., Alan prince, S. S., Elangbam, C., & K, V. (2026). Explainable machine learning for mix-design-stage prediction of compressive strength in SCM-enhanced concrete. Case Studies in Construction Materials, 25, Article e06595. https://doi.org/10.1016/j.cscm.2026.e06595

Image Credits: AI Generated

DOI: 10.1016/j.cscm.2026.e06595

Keywords: concrete, compressive strength, machine learning, XGBoost, SHAP, supplementary cementitious materials, fly ash, silica fume, water-cement ratio, explainable AI, mix design, construction materials

The target of the investigation was calcium silicate hydrate, or C-S-H, the binding phase that gives cement its strength. C-S-H is a layered material: stiff silicate chains form sheets that are stacked and held together by water molecules and calcium ions. That layered architecture is both a blessing and a curse. Within the sheets, the silicon-oxygen framework is extremely rigid, but between the sheets the material is comparatively soft and prone to sliding and separation. This anisotropy means that C-S-H — and therefore concrete — behaves very differently depending on the direction of the load, and it is precisely at these vulnerable interlayer regions that vibration damage tends to begin.

To probe how polymers could shore up these weak zones, the team built realistic atomistic models of C-S-H with a calcium-to-silicon ratio of 1.67, matching ordinary Portland cement, and inserted three representative polymers into the interlayer spaces: polyethylene (PE), polyvinyl alcohol (PVA), and polybutadiene (PB). These three were chosen deliberately because they differ in one crucial chemical respect. PE and PB are essentially nonpolar hydrocarbon chains, while PVA is decorated with abundant hydroxyl groups — the same -OH chemistry that makes water sticky. The polymer content was kept between roughly 1 and 5 percent by weight, consistent with what is used in real damping-modified concretes.

The simulations themselves were ambitious. Using the LAMMPS package with the ClayFF and CVFF force fields, the researchers first calibrated their model against known experimental and computational values, confirming that the tensile strength, elastic modulus, and shear modulus of their C-S-H matched earlier studies. They then subjected both pristine and polymer-modified C-S-H to two kinds of cyclic shear loading: constant strain-rate cycling and sinusoidal loading inspired by dynamic mechanical analysis, the standard laboratory technique for measuring viscoelastic damping. Each condition ran for 72 full loading cycles, with simulations lasting up to 46 nanoseconds — an enormous investment of computational time at the atomic scale.

The mechanical results were encouraging but nuanced. Adding polymers barely changed the tensile strength of C-S-H along its strong direction, and left shear strength along the stiff XY plane essentially untouched, because that direction is governed by the intrinsic silicate framework. But along the weak interlayer direction, the story changed dramatically. Polymer chains acted as molecular bridges spanning adjacent C-S-H sheets, raising shear strength in the XZ direction and — critically — providing residual strength after failure. When unmodified C-S-H was pulled apart along its weak axis, it fractured completely; when polymer was present, the broken pieces remained tethered together, delaying catastrophic failure and increasing the material’s ductility and toughness. Dosage mattered too: around 2.5 to 2.7 percent by weight gave the best tensile reinforcement, while 4.9 to 5.3 percent maximized shear enhancement.

The atomic-scale analysis explained why. Radial distribution functions showed that polymers sitting on the (010) surface of C-S-H readily form hydrogen bonds with interlayer water molecules at distances of roughly 2 to 3 angstroms, whereas polymers on the (001) surface form almost none. Hydrogen bond counting during deformation revealed that these bonds are the workhorses of interlayer cohesion: under tension along the weak direction, the hydrogen bond network collapses, but under shear it largely survives, allowing energy to be dissipated gradually through sliding and rearrangement rather than sudden breakage. Among the three polymers, PVA formed by far the most hydrogen bonds with C-S-H, thanks to its hydroxyl-rich chemistry, giving it the strongest interfacial grip.

When the team turned to damping — the heart of the study — the findings became genuinely striking. Under cyclic shear, plain C-S-H showed equivalent viscous damping of about 0.099 along the XY direction and 0.096 along XZ. After polymer modification, damping along the stiff XY direction rose only modestly, but along the weak XZ direction it jumped substantially: PE increased the loss factor by about 10.1 percent, PB by 11.3 percent, and PVA by 13.7 percent. Even more interesting, polymer modification flipped the anisotropy. Before modification, C-S-H dissipated more energy along its strong plane; after modification, the formerly weak interface became the better energy-dissipating interface. In effect, the polymers converted concrete’s structural liability into a functional asset.

The viscoelastic measurements reinforced this picture. Under sinusoidal loading, both the storage modulus, which measures elastic energy storage, and the loss modulus, which measures energy converted to heat, increased after polymer modification — with PVA again showing the largest gains. The non-zero phase shift between applied strain and resulting stress confirmed that C-S-H is genuinely viscoelastic, and the polymers widened that phase lag, particularly along the interlayer direction. The mechanism is elegant: hydrogen-bonded polymer bridges increase interlayer friction and enable controlled structural rearrangement during each cycle, so vibrational energy that would otherwise drive crack growth is instead bled away as heat, cycle after cycle.

The implications for infrastructure are considerable. A concrete that can dissipate vibration without sacrificing strength could extend the service life of bridges, tunnels, seismic-resistant buildings, and industrial facilities, while reducing maintenance costs and material consumption. The study is careful to acknowledge its limits: each simulation condition was run once, the nanosecond-scale loading frequencies far exceed real structural frequencies, and features like aggregates and heterogeneous hydration products are absent from the models. The authors frame their results as molecular-level guidance rather than engineering predictions, and call for experimental dynamic mechanical analysis and multiscale modeling to validate the mechanism at full scale. Still, the core insight stands: by choosing polymers with the right hydrogen-bonding chemistry — PVA chief among them — and orienting them within cement’s weakest layers, engineers may one day design concrete that doesn’t just resist vibration, but quietly swallows it.

Subject of Research: Molecular mechanisms of polymer-induced damping enhancement in calcium silicate hydrate

Article Title: Hydrogen-bond-mediated damping enhancement in polymer-modified calcium silicate hydrate

Article References: Liu, Y., Zaoui, A., & Sekkal, W. (2026). Hydrogen-bond-mediated damping enhancement in polymer-modified calcium silicate hydrate. Results in Engineering, 32, Article 113328. https://doi.org/10.1016/j.rineng.2026.113328

Image Credits: AI Generated

DOI: 10.1016/j.rineng.2026.113328

Keywords: calcium silicate hydrate, molecular dynamics, hydrogen bonds, polyvinyl alcohol, damping, cement, viscoelasticity, concrete, interlayer sliding, energy dissipation, C-S-H, infrastructure

A new study published in Case Studies in Construction Materials by Yuki Kobayashi, Taku Koyama, Keiichi Nakamura, and Mitsuo Ozawa tackles this problem with an elegant twist. Instead of combining steel fibers for strength with polypropylene fibers for fire protection, the team developed a UHPFRC reinforced exclusively with polyvinyl alcohol, or PVA, fibers. Their material achieved a compressive strength of at least 150 megapascals while carrying 1.7 percent by volume of PVA fibers, and their experiments reveal in quantitative detail how these hydrophilic synthetic fibers suppress explosive spalling through a two-stage mechanism that shifts as temperatures climb past the fibers’ melting point.

The physics of spalling in dense concrete is unforgiving. When a fire heats the surface, thermal stress builds at the heated face while water trapped in the pores near the surface turns to vapor. Because UHPFRC has an extremely low permeability, that vapor cannot escape quickly, and pressure builds rapidly in the near-surface pores. At the same time, the material’s high elastic modulus amplifies thermal stresses generated by steep internal temperature gradients. The combination can crack the surface layer and then blast it away. Once cover concrete is lost, embedded reinforcement heats up faster and the load-bearing cross-section shrinks, so preventing spalling is essential for structural fire safety.

The conventional defense has been to add low doses of polypropylene fibers, which melt at around 170 degrees Celsius and leave behind channels that vent vapor pressure. Steel fibers, meanwhile, boost tensile performance at ambient temperature but are considered ineffective against spalling on their own, and some studies suggest they may even increase spalling risk. PVA fibers occupy a unique middle ground: they are highly hydrophilic and form strong chemical bonds with the cement matrix, which improves crack dispersion and flexural toughness at room temperature while also contributing to fire-spalling mitigation. They are also immune to the corrosion that plagues steel fibers, opening applications in highly corrosive environments or structures requiring non-magnetic properties.

To test their material under conditions closer to real structures, the researchers used the ring-restraint heating test, a method standardized by the Japan Concrete Institute and recommended by RILEM. Rather than heating free-standing cylinders, they cast the UHPFRC into steel rings 300 millimeters in diameter that confine the specimen’s outer circumference, mimicking the restraint that surrounding structure imposes on a heated member. After preheating a small furnace to roughly 1050 degrees Celsius, they exposed the specimen’s surface to one-sided heating for 30 minutes, a deliberately severe thermal shock harsher than standard tunnel-fire curves such as ISO 834. Throughout each test they recorded internal temperature, internal water vapor pressure, and the restraint stress acting on the concrete at depths of 5, 10, 25, and 40 millimeters, using strain gauges on the steel ring corrected for thermal expansion and a silicone-oil-filled steel pipe calibrated to separate true vapor pressure from oil-expansion artifacts.

The experimental campaign covered four fiber volume fractions: zero, 0.5, 1.0, and 1.7 percent, with fibers replacing an equivalent volume of sand in a mix with a water-to-binder ratio of just 16 percent. Preliminary testing showed that anything above 1.7 percent destroyed workability, making that the practical ceiling. The PVA fibers themselves were slender, 0.2 millimeters in diameter and 12 millimeters long, with tensile strengths between 1092 and 1716 megapascals. Thermogravimetric analysis showed the fibers begin losing weight near 80 degrees Celsius, melt at 237 degrees Celsius, and thermally decompose between 270 and 460 degrees Celsius, a thermal fingerprint that proved central to interpreting everything that followed.

The results were striking. Spalling depth and spalling volume both decreased steadily as fiber content rose. The fiber-free control developed deep, mortar-shaped defects at the center of the heated surface, and rapid temperature surges repeatedly rippled inward as each spalling event peeled the surface away and shifted the heated front deeper. In the fiber-reinforced specimens, the damage took a different form: layered cracking parallel to the heated surface rather than explosive detachment. Because the surface concrete stayed in place, heat transfer into the interior was limited, and internal temperatures at depth remained markedly lower than in the control. Restraint stress also rose more gradually in fiber-containing specimens, reaching roughly 3 megapascals in the 1.7 percent mix within five minutes of heating, compared with 4.5 megapascals in the fiber-free specimen over the same interval.

One finding defied expectations. Peak internal vapor pressures were higher in the fiber-reinforced specimens, up to 5.7 megapascals at 0.5 percent fiber content, than in the fiber-free control, which peaked at 2.3 megapascals. The researchers attribute this counterintuitive trend to a synergy between the PVA fibers’ high melting point, their strong chemical bonding to the matrix, and the powerful crack-bridging effect of the high fiber dosage. Instead of venting early through cracks, as low-melting, weakly bonded polypropylene fibers allow, the vapor remained trapped because PVA fibers held crack surfaces together until temperatures approached the melting point. The pressure peaked in the range immediately before rapid fiber mass loss, then declined as the fibers melted and vanished, opening dissipation pathways through the voids they left behind.

Complementing the heating tests, residual strength experiments on small cylinders heated slowly to 200, 300, and 400 degrees Celsius mapped how the material degrades. Compressive strength fell to about 55 percent of its unheated value by 200 degrees Celsius, a sharper drop than seen in lower-strength PVA materials or in Eurocode 4 predictions for normal-weight concrete, likely because the dense, ultra-strong matrix generates larger thermal stresses that initiate fine map-like cracks early. The static modulus of elasticity proved even more sensitive, falling to just 18 percent of its initial value at 400 degrees Celsius, reflecting both the loss of fibers and the dehydration and decomposition of the cement hydrates that bind the matrix together.

Finally, the team applied a tensile strain failure model to connect the measurements to a coherent mechanism. When heated concrete expands but is restrained, compressive stress develops in the plane of the surface while tensile strain builds perpendicular to it through the Poisson effect; cracking occurs when that strain exceeds a critical value. Using experimentally derived residual modulus values and restraint stresses, the model reproduced initial spalling progression within about 10 millimeters of the heated surface when the tensile failure strain was set between 50 and 100 microstrain. Notably, for fiber contents of 1.0 percent and above, the model predicted earlier spalling than actually occurred, precisely because it cannot capture the fiber bridging and vapor dissipation that suppress spalling in reality. The authors propose that PVA fibers act in two stages: before melting, crack-bridging holds fractured fragments in place and prevents them scattering; after melting, the void network they leave behind dissipates accumulated vapor pressure. Direct microscopic confirmation of that void network remains future work, but the indirect evidence, from pressure curves that peak near the melting point and then fall, from enhanced air permeability reported in prior studies, and from notched-beam tests showing fibers intact at 200 degrees Celsius but gone at 300, makes a compelling case. If confirmed in full-scale beam tests under standard fire curves, this single-fiber strategy could simplify the design of fire-safe, corrosion-proof ultra-high-performance concrete for the infrastructure of the coming decades.

Subject of Research: Spalling suppression in ultra-high-performance fiber-reinforced concrete using polyvinyl alcohol fibers

Article Title: Fire-induced spalling suppression by polyvinyl alcohol fibers in ultra-high-performance fiber-reinforced concrete

Article References: Kobayashi, Y., Koyama, T., Nakamura, K., & Ozawa, M. (2026). Fire-induced spalling suppression by polyvinyl alcohol fibers in ultra-high-performance fiber-reinforced concrete. Case Studies in Construction Materials, 25, Article e06547. https://doi.org/10.1016/j.cscm.2026.e06547

Image Credits: AI Generated

DOI: 10.1016/j.cscm.2026.e06547

Keywords: UHPFRC, polyvinyl alcohol fibers, explosive spalling, fire safety, concrete, vapor pressure, thermal stress, ring-restraint heating test, construction materials, fiber reinforcement, residual strength, high-performance concrete