Sugarcane ash and limestone sludge yield stronger, greener concrete predicted by AI

Concrete may be the most widely used building material on Earth, but its key ingredient comes at a heavy environmental price. Cement manufacturing is estimated to account for roughly 5 to 8 percent of all anthropogenic carbon dioxide emissions, a figure that has pushed civil engineers to search for waste-derived substitutes that can partially replace Portland cement without sacrificing performance. A new open-access study now reports that two industrial and agricultural by-products, sugarcane bagasse ash and lime sludge ash, can be combined with polypropylene fibres to produce concrete that is measurably stronger and less permeable than conventional mixes, with an artificial neural network trained on the experimental data able to predict the resulting mechanical properties.

The research team tested concrete in which cement was partially replaced with sugarcane bagasse ash at 0, 10, 20 and 30 percent, alongside lime sludge ash at 0, 2.5, 5, 7.5 and 10 percent. Sugarcane bagasse ash is generated when bagasse, the fibrous residue left after sugar extraction, is burned as fuel in sugar mills; the ash used here came from Nirani Sugars Ltd. in Bagalkot, India. Lime sludge ash, a calcium-rich waste from the paper and pulp industry, was added because its calcium oxide content can accelerate hydration reactions and improve particle packing. The researchers report that the best performance across their tests came from a mix containing 20 percent sugarcane bagasse ash and 5 percent lime sludge ash.

With that optimum blend identified, the team carried out a second stage of testing in which 2 percent polypropylene fibres, by weight of cement, were added to the same ash-modified mix. The fibres used were 12 millimetres long with a diameter of roughly 50 micrometres, giving an aspect ratio of 240. Because polypropylene fibres tend to clump when dropped directly into dry concrete, the researchers first dispersed them in a liquid medium using a laboratory water-bath sonication process, in which ultrasonic cavitation separates fibre agglomerates before mixing. The team states that this dispersion step promoted even fibre distribution throughout the matrix, which is essential for effective crack-bridging.

The base concrete was an M20-grade nominal mix with a cement to fine aggregate to coarse aggregate ratio of 1:1.5:3 and a water-to-cement ratio of 0.4, designed according to Indian Standard IS 10262:2009. Cubes of 150 millimetres per side were cast for compressive strength, cylinders 100 millimetres in diameter and 200 millimetres tall for split tensile strength, and beams of 100 by 100 by 500 millimetres for flexural strength. Specimens were prepared in triplicate, demoulded after 24 hours and water-cured for 28 days before testing at 7, 14 and 28 days.

The strength results followed a consistent pattern for both ash types: performance rose with increasing replacement up to an optimum level and then declined. For lime sludge ash concrete, the control mix reached approximately 14, 20 and 27 megapascals at 7, 14 and 28 days, while the 5 percent mix reached roughly 18, 23 and 30 megapascals, gains of about 28.6, 15 and 11.1 percent at the respective ages. Beyond 5 percent, strength dropped slightly, which the authors attribute to dilution of cementitious compounds and reduced formation of calcium silicate hydrate gel. Sugarcane bagasse ash showed a similar optimum at 20 percent, where compressive strength improved by roughly 33.3, 20 and 14.8 percent at 7, 14 and 28 days compared with the control, before a marginal reduction at 30 percent replacement.

The combined system with fibres delivered the headline results. The concrete containing sugarcane bagasse ash, lime sludge ash and polypropylene fibres achieved a maximum compressive strength of 46.25 newtons per square millimetre at 28 days, an improvement of roughly 20.28 percent over the control mix, which reached 38.45 newtons per square millimetre. The abstract separately reports the optimum mix achieving 29.2 megapascals against a control of 27.45 megapascals, and split tensile and flexural strengths of 3.21 and 6.21 megapascals. The authors attribute the gains to pozzolanic reactions that refine the pore structure, secondary calcium silicate hydrate formation, and the fibre-bridging effect, in which fibres transfer tensile stresses across developing microcracks and shift failure behaviour from brittle toward pseudo-ductile.

Durability testing pointed in the same direction. Following the German standard DIN 1048-1, the team subjected 150-millimetre cubes to constant water pressure for 72 hours and then measured the maximum depth of water penetration after splitting the specimens longitudinally. The control mix allowed penetration to a depth of 14.5 millimetres, while the ash-modified concrete limited penetration to about 12 millimetres, an approximate 17 percent reduction in permeability. The researchers link this to improved particle packing from the fine ash particles and to secondary hydration reactions in which reactive silica from the bagasse ash consumes calcium hydroxide to form additional densifying hydration products, with the calcium oxide in the lime sludge ash supporting early hydration.

Microstructural evidence backed up the mechanical data. Scanning electron microscopy showed angular, porous ash particles acting as micro-fillers, dense layered and fibrous clusters characteristic of well-developed calcium silicate hydrate gel, calcium-rich crystalline phases attributed to the lime sludge ash, a more uniform particle arrangement with fewer voids, and hydration products adhering firmly to polypropylene fibre surfaces, indicating strong interfacial bonding. Energy dispersive X-ray spectroscopy detected strong silicon, oxygen, calcium and aluminium peaks, consistent with silica-rich pozzolanic phases and calcium-bearing hydration products, along with minor traces of magnesium, potassium, phosphorus, iron and titanium.

Statistical analysis supported the reliability of the findings. Coefficients of variation at 28 days ranged from about 3.85 to 4.04 percent for compressive strength, 10.42 to 11.63 percent for split tensile strength and 9.46 to 10.91 percent for flexural strength. In the fibre-content series, the mixture with 20 percent sugarcane bagasse ash and 5 percent lime sludge ash but no fibres reached 41.5 megapascals in compression, 3.8 megapascals in split tension and 5.5 megapascals in flexure at 28 days, while the 2 percent fibre mix reached 50.2, 5.5 and 7.8 megapascals respectively, improvements of approximately 20.96, 44.74 and 41.82 percent. One-way analysis of variance gave F-values of about 10.04, 5.20 and 5.45 with p-values of approximately 0.0006, 0.0091 and 0.0076, indicating statistically significant differences among fibre contents at the 5 percent level.

The machine-learning component used a feed-forward back-propagation neural network with a 3-10-3 architecture: three inputs, namely polypropylene fibre content, curing age and water-to-binder ratio, feeding ten hidden neurons with a TANSIG activation function, and three outputs for compressive, split tensile and flexural strength. The model was trained on 18 experimental observations, in which the ash contents were held at 20 and 5 percent while fibre dosage varied from 0 to 2.5 percent, curing age took the values 7, 14 and 28 days, and the water-binder ratio ranged from 0.40 to 0.50. Inputs and outputs were normalised with Min-Max scaling, training used the Adam optimiser with an initial learning rate of 0.001 for up to 150 epochs, and the data were split 70:15:15 into training, validation and testing subsets. The preliminary model achieved training R-squared values of 0.945, 0.907 and 0.849 for compressive, split tensile and flexural strength, and independent test R-squared values of 0.893 for both compressive and split tensile strength, with a comparatively lower 0.732 for flexural strength, which the authors link to the greater sensitivity of flexural performance to fibre distribution and crack-bridging mechanisms. The authors note that the study is limited to the investigated replacement levels and specimen scales, and that comparisons with other studies should be read as percentage improvements over each study’s own control rather than as absolute strengths, since binder compositions, curing conditions and testing procedures differ. Within those limits, they present the work as an integrated experimental and data-driven framework for designing sustainable concrete from agricultural and industrial wastes.

Subject of Research: Experimental and ANN-based evaluation of sustainable fibre-reinforced concrete using sugarcane bagasse ash, lime sludge ash and polypropylene fibres

Article Title: ANN-based prediction and experimental evaluation of sustainable fibre-reinforced concrete incorporating sugarcane bagasse ash and limestone slurry ash

Article References: K, G., V, S. G., & Jelila, Y. D. (2026). ANN-based prediction and experimental evaluation of sustainable fibre-reinforced concrete incorporating sugarcane bagasse ash and limestone slurry ash. Discover Mechanical Engineering, 5(1), Article 207. https://doi.org/10.1007/s44245-026-00380-y

Image Credits: AI Generated

DOI: 10.1007/s44245-026-00380-y

Keywords: sugarcane bagasse ash, lime sludge ash, polypropylene fibres, sustainable concrete, artificial neural network, compressive strength, split tensile strength, flexural strength, water permeability, supplementary cementitious materials, SEM analysis, machine learning