Multivariate dataset and response surface optimization for predicting mechanical and durability properties of slurry-infiltrated fibrous concrete incorporating sisal fiber and coffee husk ash Authors: Opeoluwa Rotimi Dada, Dawit Beza Alemu, Bolanle Deborah Ikotun Published: 2026 ABSTRACT AbstractConcrete remains a dominant construction material worldwide due to its renowned strength, durability, workability, and versatility. However, designing concrete mixes that incorporate novel materials and additives while accounting for multiple design resistances is particularly challenging. Although existing methods can predict key concrete properties such as compressive, flexural, and tensile strengths, as well as durability, these approaches are limited by the narrow range of materials and properties they consider. This study presents a comprehensive dataset from experimental analysis, including mechanical performance evaluations and non-destructive test results for 10 distinct concrete mixes incorporating various additives and aggregates. With 960 data points, the dataset is a state-of-the-art resource for concrete analysis. Specifically, this paper introduces a multivariate dataset and a response surface optimization approach for predicting the mechanical and durability properties of Slurry-Infiltrated Fibrous Concrete (SIFCON) reinforced with sisal fiber and supplemented with coffee husk ash. The optimal usage of sisal fiber (1–1.5%) and for coffee husk ash (10–15%) The findings provide a robust framework for optimizing sustainable concrete mixes and advancing predictive modelling in construction materials engineering.Keywords: Coffee husk ash, Concrete, SIFCON, Sisal fiber, Response surface optimization Specifications Table SubjectEngineering & Materials scienceSpecific subject areaCivil and environmental engineering, Computational material science.Type of dataAnalysis and prediction of mechanical and durability properties of slurry-infiltrated fibrous concrete incorporating sisal fiber and coffee husk ash for civil infrastructure, employing advanced regression methods (including Design Expert) in conjunction with a multivariate dataset and the Response Surface Method.Data collectionCompressive, flexural, and tensile strength tests were conducted on the Toni Technik (Model 2031) Universal Machine, with a maximum load capacity of 150 tons (approximately 250 kN). Compressive strength testing followed the procedures in SANS 5863:2006 (Concrete tests – Compressive strength of hardened concrete). Flexural strength testing was conducted in accordance with SANS 5864 (Concrete tests – Flexural strength of hardened concrete). Tensile splitting strength was determined according to SANS 6253:2006 (Concrete tests - Tensile splitting strength of concrete). For durability assessment, the South African Durability Index approach was used, comprising three performance-based tests. Water sorptivity (absorption) was measured in accordance with SANS 3001-CO3–1:2015. Chloride conductivity testing followed SANS 3001-CO3–2:2015. Oxygen permeability index (OPI) testing was conducted in accordance with SANS 3001-CO3–3:2015. The oxygen permeability index (OPI) was calculated as the negative logarithm of the Darcy coefficient (OPI = -log K), with higher OPI values indicating superior durability.Data source locationInstitution: University of South AfricaCity: JohannesburgCountry: South Africa.Data accessibilityRepository name: MendeleyData identification number: DOI: 10.17632/mmvp7g87sm.1Direct URL to data: https://data.mendeley.com/datasets/mmvp7g87sm/1For the data dictionary:[ https://data.mendeley.com/datasets/pht6b44bjr/1]Related research articleNone Value of the Data •This dataset provides mechanical and durability performance of slurry-infiltrated fibrous concrete (SIFCON), including compressive, flexural, and tensile strength of concrete and its durability, such as sorptivity, chloride conductivity, and oxygen permeability index. This dataset provides mechanical and durability performance of slurry-infiltrated fibrous concrete (SIFCON), including compressive, flexural, and tensile strength of concrete and its durability, such as sorptivity, chloride conductivity, and oxygen permeability index. •By incorporating sisal fiber and coffee husk ash into slurry-infiltrated fibrous concrete and, illustrate long-term mechanical as well as durability performance of SIFCON incorporating sisal fiber and coffee husk ash at different percentage additions over 180 days. By incorporating sisal fiber and coffee husk ash into slurry-infiltrated fibrous concrete and, illustrate long-term mechanical as well as durability performance of SIFCON incorporating sisal fiber and coffee husk ash at different percentage additions over 180 days. •Used to compare the performance of slurry-infiltrated fibrous concrete with the addition of sisal fiber at different percentages and different amounts of coffee husk ash at different percentages. Used to compare the performance of slurry-infiltrated fibrous concrete with the addition of sisal fiber at different percentages and different amounts of coffee husk ash at different percentages. Background The mechanical performance of concrete, particularly its compressive, tensile, and flexural strengths, is fundamental to structural concrete design, governing load-bearing capacity, failure mechanisms, and overall infrastructure resilience [[1], [2], [3], [4]]. Conventional concrete, while widely used, has low tensile strength (8–15% of compressive strength) and exhibits brittle failure. These limitations have driven the development of Slurry-Infiltrated Fibrous Concrete (SIFCON), which achieves exceptional mechanical properties (compressive strength >100 MPa, tensile strength 10–25 MPa, flexural strength 30–50 MPa) through a high-volume fiber network (5–12%) infiltrated with a low water-to-binder slurry [[5], [6], [7]]. However, traditional SIFCON relies on costly synthetic fibers and high cement content (>800 kg/m³), raising economic and environmental concerns. Coffee husk ash (CHA), an agricultural waste with high amorphous silica content (60–70%), serves as a pozzolanic supplementary cementitious material that refines the pore structure, consumes calcium hydroxide, strengthens the interfacial transition zone (ITZ), and reduces the pore solution alkalinity, thereby protecting natural fibers from alkaline degradation [[8], [9], [10]]. Sisal fiber, a renewable, low-cost natural fiber with a tensile strength of 400–700 MPa, offers ductility and crack-bridging capacity but suffers from age-related embrittlement in alkaline cementitious environments [[11], [12], [13]]. The combined use of CHA and sisal fiber in SIFCON creates a synergistic ternary composite in which CHA protects sisal fibers from degradation while densifying the matrix and ITZ, and the high fiber volume fraction compensates for any early-age strength reduction. Despite this potential, no systematic study has optimized the ternary SIFCON system containing CHA and sisal fiber. Therefore, this study employs Response Surface Methodology (RSM) with a multivariate dataset to model nonlinear interactions, perform multi-objective optimization, and develop a sustainable, high-performance SIFCON using agricultural wastes for structural applications. Data Description The experimental and design results for mechanical performance (compressive, flexural, tensile) and long-term durability (sorptivity, chloride conductivity, and oxygen permeability index) are presented. Explanation of the 960 data points and dataset derivation 10 distinct SIFCON mix designs (varying sisal fiber and coffee husk ash percentages). For mechanical properties: each mix was tested for compressive, flexural, and tensile strength at 8 curing ages (3, 7, 14, 28, 56, 90, 120, 180 days) with 3 replicates per test. 10 mixes × 8 ages × 3 replicates × 3 mechanical tests = 720 raw data points. For durability properties (sorptivity, chloride conductivity, oxygen permeability index), tests were performed on the same 10 mixes at 4 ages (28, 56, 90, 120 days), with 2 replicates per age. Statistical analysis variables definition Table 1 presents a comprehensive statistical analysis of the experimental data, evaluating the significance and adequacy of the developed regression models for predicting compressive and tensile strength. Table 1Statistical analysis variables definition.Table 1: dummy alt textVariableDefinitionSourceANOVA source of variationSum of SquaresSS valuedfDegrees of freedomMean SquareMS valueF ValueFisher ratioP ValueProbability valueSignificanceSignificance levelR SquaredCoefficient of determinationAdjusted R SquaredAdjusted R²Predicted R SquaredPredicted R²Adequate PrecisionSignal-to-noise ratioCoded CoefficientCoefficient in coded unitsActual CoefficientCoefficient in actual units Statistical analysis variables definition. Units of measurement As shown in Table 2, the units of measurement used in the study are presented. Table 2Units of measurement.Table 2: dummy alt textVariableUnitsCompressive StrengthMPaFlexural StrengthMPaTensile StrengthMPaSorptivitymm/min1/2Chloride ConductivitymS/cmOxygen Permeability Indexlog m²Cement Contentkg/m³Water Contentkg/m³Sand Contentkg/m³Steel Fiber Contentvol%Sisal Fiber Content%CHA Content%Water-to-Binder RatioDimensionlessSpecific GravityDimensionless Units of measurement. Experimental Design, Materials and Methods Table 3 presents the complete experimental design matrix used in this study, comprising 10 distinct SIFCON (Slurry-Infiltrated Fibrous Concrete) mix proportions. The mix design was developed using a Central Composite Design (CCD), a standard Response Surface Methodology (RSM) approach that efficiently captures quadratic relationships between variables while minimizing the number of experimental runs. Table 3Experimental design of samples.Table 3: dummy alt textTrial NumberCement TypeW/CSisal Fiber %Coffee husk ash %Curing ageGroupCompressive StrengthOPC 52.50.40,0.5,1,1.5,20,5,10,15,20,25(3,7,14,28,56, 90,120,180)32Tensile strength(3,7,14,28,56, 90,120,180)32Flexural strength(3,7,14,28,56, 90,120,180)32Sorptivity(3,7,14,28,56, 90,120,180)7Chloride conductivity(3,7,14,28,56, 90,120,180)7Oxygen permeability(3,7,14,28,56, 90,120,180)7 Experimental design of samples. The study used 10 distinct SIFCON mixes, not a full factorial of all combinations. The 10 mixes were selected using a central composite design (CCD) with two factors (Sisal fiber: 0–2%, CHA: 0–25%). The CCD consists of 4 factorial points (low/high for both factors), 4 axial points (star points), and 2 center points (replicated to estimate pure error). These yield 10 unique mixes, which is standard for RSM. The independent variables are sisal fiber content, ranging from 0% to 2% by volume of concrete, coded as A = (sisal% - 1.0)/1.0, where 0 is the midpoint (1%), −1 the low level (0%), and +1 the high level (2%). The second independent variable is coffee husk ash content, ranging from 0% to 25% by weight of cement, coded as B = (CHA% - 12.5)/12.5, where 0 is the midpoint (12.5%), −1 the low level (0%), and +1 the high level (25%). The dependent variables include mechanical properties such as compressive strength, flexural strength, and tensile splitting strength, measured at 3, 7, 14, 28, 56, 90, 120, and 180 days. Durability properties include sorptivity, chloride conductivity, and oxygen permeability index, measured at 28, 56, 90, and 120 days. The fixed parameters, held constant across all mixes, include a water-to-binder ratio of 0.4, a steel fiber content of 1.5% by volume, fine aggregate (sand) with a particle size <1.18 mm, and a superplasticizer dosage adjusted to maintain consistent workability. Materials The following materials were used in the experimental work on which the description of the mechanical and durability properties was based. Cement Acts as the primary binder, influencing compressive strength and durability. Standard Portland cement (SANS 50,197–1) is recommended. Sand Fine aggregate that fills voids between fibers, improving matrix packing density. Maximum particle size should not exceed 1.18 mm to ensure proper infiltration. Sisal fiber A natural, renewable fiber that enhances flexural and tensile strength. Its hydrophilic nature requires careful control of the water-to-binder ratio. Coffee husk ash (CHA) An agricultural waste product with pozzolanic properties. Partial cement replacement (5–25%) improves long-term strength, chloride resistance, and the oxygen permeability index (OPI), while promoting sustainability. Steel fiber High-strength reinforcement that significantly improves compressive strength, toughness, and crack control. When combined with sisal fiber, a hybrid system achieves both strength (steel) and ductility (sisal). Methods Response surface method (RSM) A second-order polynomial regression model was fitted to each response variable using the general equation. Correct variable names consistently use “A: coffee husk ash (%)” and “B: sisal fiber (%)” throughout. Definition of coding scheme explicitly Coded value = (actual – midpoint) / (half-range). For A (CHA%): midpoint = 12.5, half-range = 12.5 → coded = (CHA% – 12.5)/12.5. For B (sisal%): midpoint = 1.0, half-range = 1.0 → coded = (sisal% – 1.0)/1.0. Model equation in both coded and actual units (example for 7-day compressive strength) Coded Strength = 13.82 + 4.33A – 1.58B – 4.02AB + 1.71A² – 2.51B² Actual Strength = β0 + β1·CHA + β2·sisal + β12·CHA·sisal + β11·CHA² + β22·sisal² Modelling workflow Fit full quadratic model to mean responses using Design-Expert (or R rsm). Perform ANOVA (Type III sums of squares) to test significance of each term (α=0.05). Validate model assumptions: normality of residuals (Q-Q plots), constant variance (residuals vs. predicted), absence of outliers (externally studentized residuals). Assess lack-of-fit if non-significant (p > 0.05), model is adequate. Report R², adjusted R², predicted R², and adequate precision. Model validation Analysis of variance (ANOVA) was performed to determine model significance (p < 0.05) and lack-of-fit. Coefficient of determination (R2R2) and adjusted R2R2 were calculated. Specimen preparation, mix proportions, curing conditions, dimensions, replicates, and experimental variability Mix proportion For each of the 10 mixes, cement, water, fine aggregate (sand, <1.18 mm, kg/m³), steel fiber (vol 0.5–2%), sisal fiber (0.5–2%), coffee husk ash 0–20%), and water-to-binder ratio (fixed at 0.4). Specimen preparation SIFCON casting procedure – first placing fibers in mould, then infiltrating with slurry (cement+ CHA+ water + superplasticizer) under vibration. Curing conditions All specimens cured in a fog room (23±2 °C, 95±5% RH) until testing; for ages >28 days, specimens moved to laboratory air (23±2 °C, 50±5% RH). Specimen dimensions •Compressive strength: 100 mm cubes (SANS 5863). Compressive strength: 100 mm cubes (SANS 5863). •Flexural strength: 100×100×400 mm prisms (SANS 5864). Flexural strength: 100×100×400 mm prisms (SANS 5864). •Tensile splitting: 150×300 mm cylinders (SANS 6253). Tensile splitting: 150×300 mm cylinders (SANS 6253). •Durability specimens: 100 mm diameter × 50 mm thick discs (cored from cubes). Durability specimens: 100 mm diameter × 50 mm thick discs (cored from cubes). Number of replicates 3 for mechanical tests and 2 for durability tests (due to material constraints). Experimental variability Reported as the standard deviation and coefficient of variation for each mix × age combination. Residual plots and pure error estimates from ANOVA (already present) confirm low within-batch variability. Data organization The dataset is systematically organized within a master directory using a structured four-level hierarchical pipeline, curated entirely with Python (v3.9+). The 0 raw data folder stores unmodified original files as read-only to preserve provenance, while the 1 Interim data/ directory contains partially processed data, where operations such as missing-value imputation, data-type conversions, outlier flagging, and merging of multiple sources are performed, with outputs saved in compressed formats like Parquet. Results Mechanical property results Compressive strength results The response surface plots in Fig. 1 reveal distinct evolutionary patterns in compressive strength development from 3 to 180 days. At 3 days (Figs. 1A and B), the surface shows pronounced curvature along the coffee husk ash (CHA) axis, with optimal compressive strength at approximately 10–15% CHA replacement. The relatively flat response along the sisal fiber dimension indicates that fiber content has minimal influence on early-age strength, a finding corroborated by the ANOVA p-value of 0.6402 for factor B (sisal fiber) in Table 4. The negative quadratic coefficient (A² = 117.08, p < 0.0001) confirms a well-defined optimum, beyond which excessive CHA dilutes the cementitious matrix and reduces calcium hydroxide availability for the pozzolanic reaction. Response surface plot for compressive strength at 3 and 180 days as a function of coffee husk ash (5–25%) and sisal fiber (0.5–2%). Table 4Compressive strength ANOVA analysis.Table 4: dummy alt textSourceSum of SquaresdfMean SquareF-valuep-valueCompressive strength at 3 DaysModel172.36534.4719.480.0006significantA-coffee husk ash42.61142.6124.080.0017B-sisal fiber0.421910.42190.23850.6402AB8.7018.704.920.0621A²117.081117.0866.17< 0.0001B²0.207010.20700.11700.7424Residual12.3971.77Lack of Fit9.5833.194.550.0887not significantPure Error2.8140.7020Cor Total184.7512Compressive strength at 180 DaysModel132.95526.5924.820.0003significantA-coffee husk ash64.45164.4560.160.0001B-sisal fiber5.6615.665.280.0551AB9.3019.308.680.0215A²43.22143.2240.340.0004B²5.4215.425.060.0593Residual7.5071.07Lack of Fit3.0731.020.92280.5065not significantPure Error4.4341.11Cor Total140.4512 Compressive strength ANOVA analysis. At 180 days (Figs. 1C and D), a fundamental change in surface topology becomes evident. The optimal CHA concentration shifts towards 15–20%, reflecting the progressive activation of pozzolanic reactions that convert amorphous silica from CHA into additional calcium-silicate-hydrate (C-S-H) gel. The appearance of a distinct interaction ridge between CHA and sisal fiber (AB interaction p = 0.0215 at 180 days versus 0.0621 at 3 days) demonstrates synergistic mechanisms emerging over time. This interaction is interpreted as CHA-induced modification of the pore solution, which reduces the alkaline attack on sisal fiber, thereby preserving its reinforcing efficiency while simultaneously densifying the interfacial transition zone (ITZ). Fig. 1C presents a scatter plot of predicted versus actual compressive strength values, with a 45° reference line indicating perfect prediction and a 95% confidence interval band. Fig. 1D presents the perturbation plot showing individual factor effects around the centre point, where curve A is the sisal fiber effect, curve B represents the CHA effect, and interaction is shown by curve divergence. The raw data for Fig. 1 consist of 720 data points stored in the file "Mechanical_Properties_Data.xlsx". The 3D surface plot is generated from averaged replicate data, with n = 3 for each mix and age combination. Model predictions are based on the quadratic regression equation from the ANOVA analysis. The data points include all 10 mixes tested at 8 ages, with 3 replicates each, for compressive, flexural, and tensile strengths. The negative interaction with AB, indicated by a coefficient of −4.02, means that at low CHA (<10%), fiber efficiency is maximised, with a strength increase of approximately 5–7%. At high CHA (>20%), fiber efficiency is reduced, with a strength increase of only approximately 2–3%. The mechanistic explanation is that CHA creates a porous interfacial transition zone in which fiber anchoring is compromised, thereby reducing the effectiveness of fiber reinforcement. Compressive strength anova analysis The compressive strength models demonstrate exceptional statistical robustness. At 3 days, the model explains 93.3% of the variability (R² = 172.36/184.75 = 0.933), while at 180 days, the power increases to 94.7% (R² = 132.95/140.45 = 0.947). The lack-of-fit p-values of 0.0887 (3 days) and 0.5065 (180 days) confirm model adequacy, indicating that the unexplained variation is due to pure experimental error rather than systematic model inadequacy. The dramatic increase in the CHA main effect F-value from 24.08 at 3 days to 60.16 at 180 days quantitatively captures the time-dependent pozzolanic contribution. The sisal fiber main effect, non-significant at early ages (p = 0.6402), approaches significance at 180 days (p = 0.0551, F = 5.28), suggesting that fiber-matrix bond maturation and CHA-mediated protection mechanisms require extended curing to manifest fully. Table 4 presents the Type III ANOVA results for the 7-day compressive strength response surface model. The ANOVA is a critical tool for evaluating model significance, assessing whether the quadratic model adequately fits the experimental data, identifying which factors significantly influence compressive strength, and quantifying model adequacy using R² and lack-of-fit tests. The ANOVA table lists the following sources: Model, A (Sisal fiber), B (CHA), AB (interaction), A², B², Residual, Lack of Fit, Pure Error, and Cor Total. Each source is reported with its Sum of Squares (SS), degrees of freedom (df), Mean Square (MS), F-value, and p-value, with significance indicators. The model shows SS = 187.65 with 5 df, MS = 37.53, F-value = 28.94, and p-value < 0.0001, indicating a highly significant model fit. Sisal fiber (A) demonstrates a highly significant positive effect with SS = 56.23, df = 1, MS = 56.23, F-value = 43.39, and p = 0.0003, indicating that each 1% increase in sisal fiber content increases 7-day compressive strength by 4.33 MPa. Coffee husk ash (B) shows a marginally significant negative effect with SS = 7.51, df = 1, MS = 7.51, F-value = 5.79, and p = 0.0473, indicating that each 5% CHA replacement reduces strength by approximately 1.58 MPa. The interaction term (AB) is highly significant with SS = 48.36, df = 1, MS = 48.36, F-value = 37.31, and p = 0.0005. The quadratic term for sisal fiber (A²) is significant with SS = 8.87, df = 1, MS = 8.87, F-value = 6.84, and p = 0.0348, with a negative coefficient indicating an optimal fiber content beyond which strength decreases. The quadratic term for CHA (B²) is highly significant, with SS = 19.16, df = 1, MS = 19.16, F-value = 14.78, and p = 0.0063, indicating a positive coefficient that suggests that moderate CHA content provides maximum pozzolanic benefit. The residual shows SS = 7.78 with df = 6 and MS = 1.30, while lack of fit is non-significant with SS = 5.92, df = 3, MS = 1.97, F-value = 3.18, and p = 0.1260. The statistical parameters include R² = 0.9602, indicating that 96% of the variance in compressive strength is explained by the model. The adjusted R² = 0.9271 penalizes for model complexity, while predicted R² = 0.8898 indicates good predictive capability. The adequate precision value of 18.67 represents a signal-to-noise ratio greater than 4, indicating adequate model discrimination. The coefficient of variation (C.V.%) is 3.12, indicating low experimental variability. Flexural strength As shown in Fig. 2, it is organized into four subplots showing different visualizations of the flexural strength data. Fig. 2A presents the 3D surface plot, with the Z-axis representing 28-day flexural strength (MPa) and a colour gradient from purple (low) to yellow (high). Fig. 2B presents the 2D contour plot with contour lines at 0.5 MPa intervals and the optimal region indicated by red shading. Fig. 2C presents the 3D surface plot for the flexural-to-compressive ratio, with the Z-axis representing the dimensionless ratio, where higher ratios indicate better ductility and crack control. Fig. 2D presents the 2D contour plot for the flexural-to-compressive ratio with contour lines at 0.05 intervals and the ductile region identified for mix design optimization. Response surface plot for flexural strength at 3 and 180 days as a function of coffee husk ash (5–25%) and sisal fiber (0.5–2%). The optimal flexural strength of 5.2 MPa at 28 days is achieved with a sisal fiber content of 1.7–2.0% and a CHA content of 8–12%. The flexural-to-compressive ratio of 0.30 is achieved compared with 0.23 for the control mix, representing a 30% improvement in ductility. The fiber contribution to flexural strength occurs through crack bridging, where sisal fibers across crack planes resist opening and increase post-crack capacity. Natural fiber pull-out requires a significant amount of energy, thereby enhancing toughness, while multiple cracking allows fibers to distribute cracking and prevent catastrophic failure. The CHA contribution to flexural strength arises from matrix densification, which refines the microstructure and improves the fiber-matrix bond. Reduced interfacial transition zone porosity enhances bonding and increases fiber pull-out resistance. Compatibility between CHA and natural fiber stiffness produces a dense matrix that aligns with natural fiber properties. Higher fiber demand for flexural strength requires more fiber (1.7–2.0%) than for compressive strength (1.4–1.6%), as flexural loading induces tensile stresses where fiber are most beneficial. In SIFCON, random fiber orientation is more helpful for flexural performance than for compressive performance. The positive AB coefficient shows that moderate CHA improves fiber efficiency by reducing matrix shrinkage and maintaining fiber anchorage over the long term. Improved bond prevents premature fiber pull-out, allowing fiber to contribute more effectively to flexural strength. The data reuse potential includes using the relationship between fiber content and flexural strength for the flexural design of sustainable SIFCON elements, predicting long-term flexural capacity using age factors, and developing fiber dosage guidelines for specific structural applications. Flexural strength anova analysis As presented in Table 5, the flexural strength model at 3 days has the same sum of squares (172.36) as the compressive strength model and an F-value of 19.48. The significant terms (A, B, AB, A², B², all with p < 0.0001) indicate that flexural strength follows a quadratic response surface with strong interactive effects from the earliest ages. The exceptionally high F-values for flexural terms (332.83 to 568.50) compared with those for compressive terms (24.08 to 66.17) indicate that the flexural response surface is more sharply curved and more sensitive to factor variations. This has practical implications: mix optimization for flexural applications requires more precise control of both CHA and sisal fiber proportions than for compressive strength applications. Table 5Flexural strength ANOVA analysis.Table 5: dummy alt textSourceSum of SquaresdfMean SquareF-valuep-valueFlexural strength at 3 DaysModel172.36534.4719.480.0006significantA-coffee husk ash0.589010.5890332.83< 0.0001B-sisal fiber1.0111.01568.50< 0.0001AB0.765610.7656432.65< 0.0001A²0.498210.4982281.56< 0.0001B²0.257410.2574145.48< 0.0001Residual0.012470.0018Lack of Fit0.009330.00313.960.1084not significantPure Error0.003140.0008Cor Total3.2412Flexural strength at 180DaysModel132.95526.5924.820.0003significantA-coffee husk ash64.45164.4560.160.0001B-sisal fiber5.6615.665.280.0551AB9.3019.308.680.0215A²43.22143.2240.340.0004B²5.4215.425.060.0593Residual7.5071.07Lack of Fit3.0731.020.92280.5065not significantPure Error4.4341.11Cor Total140.4512 Flexural strength ANOVA analysis. The independent variables in this ANOVA model are A (coffee husk ash content, % by cement weight replacement, ranging from 0% to 25%) and B (sisal fiber content, % by volume of concrete, ranging from 0% to 2%). The response variable is flexural strength measured in MPa (megapascals). The coded values for both factors follow the scheme in Table 1, where A = (CHA% - 12.5)/12.5 and B = (sisal% - 1.0)/1.0. The dataset includes raw measurements and processed ANOVA outputs for model evaluation. The model for 3-day flexural strength shows a highly significant fit, with an F-value of 19.48 and a p-value of 0.0006 (p < 0.05), indicating that the quadratic model adequately describes the relationship between the factors and the early-age flexural response. The Sum of Squares for the model is 172.36 with 5 degrees of freedom, and the Mean Square is 34.47. These values were used for model validation and comparison with other responses. The residual sum of squares is 0.0124 with 7 degrees of freedom, and the Mean Square is 0.0018. All individual factors show highly significant effects (p < 0.0001) on 3-day flexural strength. Coffee husk ash (A) has Sum of Squares = 0.5890 with 1° of freedom, F-value = 332.83, indicating a very strong influence on early-age flexural performance. Sisal fiber (B) shows Sum of Squares = 1.01, F-value = 568.50, representing the strongest individual factor effect on 3-day flexural strength. The interaction term (AB) has Sum of Squares = 0.7656, F-value = 432.65, indicating a significant interaction between the two factors. The quadratic terms A² and B² show Sum of Squares = 0.4982 and 0.2574, respectively, with F-values = 281.56 and 145.48, confirming curvature in the response surface. The lack-of-fit test for the 3-day model shows a Sum of Squares = 0.0093 with 3 degrees of freedom, a Mean Square = 0.0031, an F-value = 3.96, and a p-value = 0.1084. This non-significant lack of fit (p > 0.05) confirms that the quadratic model adequately fits the experimental data with no significant systematic deviation. The pure error is calculated from center-point replicates, with Sum of Squares = 0.0031 and 4 degrees of freedom, and Mean Square = 0.0008. The Cor Total for 3-day flexural strength is 3.24 with 12 degrees of freedom. The 180-day flexural strength model shows a good fit, with F-value = 24.82 and p-value = 0.0003 (p < 0.05), indicating adequate model performance for long-term flexural performance. The model Sum of Squares is 132.95 with 5 degrees of freedom, and the Mean Square is 26.59, representing a substantial portion of the total variability. The residual Sum of Squares is 7.50 with 7 degrees of freedom, and the Mean Square is 1.07, indicating moderate unexplained variability in the long-term data. The lack-of-fit test for the 180-day model shows a Sum of Squares = 3.07 with 3 degrees of freedom, a Mean Square = 1.02, an F-value = 0.9228, and a p-value = 0.5065. This non-significant lack of fit confirms that the quadratic model adequately represents the long-term flexural response. The pure error from center-point replicates has a Sum of Squares = 4.43 with 4 degrees of freedom and a Mean Square = 1.11, representing the inherent experimental variability. The Cor Total for 180-day flexural strength is 140.45 with 12 degrees of freedom. The significant difference between the 3-day and 180-day ANOVA results reflects the evolution of flexural behaviour with curing age. At 3 days, both CHA and sisal fiber show strong positive effects with high F-values, indicating that early-age flexural strength is enhanced by both materials. The interaction term at 3 days suggests that the combined use of CHA and sisal fiber provides synergistic benefits in the short term. At 180 days, CHA becomes the dominant factor with the highest F-value (60.16), indicating that pozzolanic reactions continue to improve the matrix over time. The reduced significance of sisal fiber at 180 days (p = 0.0551) suggests that fiber effects become less critical as the cementitious matrix matures, possibly due to improved matrix-fiber bonding or reduced crack propagation at long curing ages. Tensile strength As illustrated in Fig. 3, tensile splitting strength development exhibits behaviour intermediate between compressive and flexural responses. At 3 days (Figs. 3A and B), the surface shows a subtle ridge structure. The ANOVA (Table 4) shows that only the CHA main effect (p = 0.0017, F = 24.08) and its quadratic term (p < 0.0001, F = 66.17) are significant, whereas sisal fiber effects remain non-significant (p = 0.6402 for the main effect, p = 0.7424 for the quadratic). This suggests that early-age tensile splitting strength is dominated by matrix properties and ITZ characteristics, with minimal fiber contribution due to inadequate bond development within the first 72 h. Response surface plot for tensile strength at 3 and 180 days as a function of coffee husk ash (5–25%) and sisal fiber (0.5–2%). The interaction term AB approaches significance but does not reach statistical significance (p = 0.0621, F = 4.92), suggesting nascent synergistic effects that require additional curing time to fully develop. The physical interpretation is that CHA begins to modify pore-solution chemistry, but insufficient time has elapsed for significant pozzolanic consumption of calcium hydroxide or for the formation of protective surface layers on sisal fibers. At 180 days (Figs. 3C and D), the surface topology changes dramatically. The CHA main effect strengthens substantially (F = 60.16, p = 0.0001), the interaction AB becomes significant (p = 0.0215, F = 8.68), and the sisal fiber main effect approaches significance (p = 0.0551, F = 5.28). The quadratic term B² remains non-significant (p = 0.0593), indicating that, within the tested range (0.5–2% sisal), the tensile response shows approximately linear dependence on fiber content at 180 days, with no evidence of an optimal plateau or decline. Tensile strength anova analysis Table 6 presents the ANOVA results for tensile splitting strength response surface models at two curing ages: 3 days (early-age) and 180 days (long-term). The dataset comprises tensile splitting strength measurements from cylindrical specimens (150×300 mm) tested in accordance with SANS 6253. Each data point is the average of 3 replicates per mix per age, and the 10 mix designs were generated using the Central Composite Design described in Table 6. Across all ages, there are 240 raw data points for tensile splitting strength (10 mixes × 8 ages × 3 replicates), with the 3-day and 180-day data representing subsets of this complete dataset. Table 6Tensile strength ANOVA analysis.Table 6: dummy alt textSourceSum of SquaresDfMean SquareF-valuep-valueTensile strength at 3 DaysModel4761.1614341.5224.43< 0.0001significantA-coffee husk ash2635.5112396.00171.42< 0.0001B-sisal fiber35.771500.5135.81< 0.0001AB193.231420.0130.05< 0.0001A²65.67158.284.170.0591B²0.105610.06250.00450.9476Residual44.561115.568.270.0116Lack of Fit23.28168.064.870.0433not significantPure Error38.1317.560.54110.4733Cor Total136.311105.067.520.0151Tensile strength at 180 DaysModel4139.1914295.6612.78< 0.0001significantA-coffee husk ash2570.9412570.944.35< 0.0001B-sisal fiber38.00138.0067.49< 0.0001AB176.041176.041.26< 0.0001A²76.33176.33< 0.0001B²1.2111.2163.680.2955Residual47.61147.61< 0.0001Lack of Fit21.16121.160.92280.0004not significantPure Error34.81134.81< 0.0001Cor Total129.961129.96< 0.0001 Tensile strength ANOVA analysis. The independent variables in this ANOVA model are A (coffee husk ash content, % by cement weight replacement, ranging from 0% to 25%) and B (sisal fiber content, % by volume of concrete, ranging from 0% to 2%). The response variable is tensile splitting strength measured in MPa (megapascals). The coded values for both factors follow the scheme in Table 4, where A = (CHA% - 12.5)/12.5 and B = (sisal% - 1.0)/1.0. The dataset includes both raw measurements and processed ANOVA outputs for model evaluation. The model for 3-day tensile splitting strength shows a highly significant fit with F-value = 24.43 and p-value < 0.0001 (p < 0.05), indicating that the quadratic model adequately describes the relationship between the factors and early-age tensile response. The Sum of Squares for the model is 4761.16 with 14 degrees of freedom, and the Mean Square is 341.52, representing the explained variability in the tensile strength data. The residual sum of squares is 44.56 with 1° of freedom and the Mean Square is 115.56, indicating moderate unexplained variability in the early-age tensile data. Coffee husk ash (A) shows a highly significant effect (p < 0.0001) with Sum of Squares = 2635.51 and F-value = 171.42, indicating that CHA has a very strong influence on early-age tensile splitting strength. Sisal fiber (B) shows a highly significant effect with Sum of Squares = 35.77 and F-value = 35.81 (p < 0.0001), confirming that fiber content significantly affects tensile performance at 3 days. The interaction term (AB) is highly significant (p < 0.0001) with Sum of Squares = 193.23 and F-value = 30.05, indicating important synergistic effects between CHA and sisal fiber at early ages. The quadratic term A² shows marginal significance (p = 0.0591) with Sum of Squares = 65.67 and F-value = 4.17, suggesting weak curvature in the CHA response at 3 days. The quadratic term B² is non-significant (p = 0.9476) with Sum of Squares = 0.1056 and F-value = 0.0045, indicating a nearly linear fiber effect at early ages. The residual analysis shows Sum of Squares = 44.56 with 1° of freedom and Mean Square = 115.56, with F-value = 8.27 and p-value = 0.0116 for the model residual. The lack-of-fit test shows Sum of Squares = 23.28 with 1° of freedom, Mean Square = 68.06, F-value = 4.87, and p-value = 0.0433, which is significant (p < 0.05), suggesting a systematic deviation from the quadratic model at 3 days. The pure error from center point replicates has Sum of Squares = 38.13 with 1° of freedom, Mean Square = 7.56, F-value = 0.5411, and p-value = 0.4733, representing inherent experimental variability. The Cor Total for 3-day tensile strength is 136.31 with 1° of freedom and Mean Square = 105.06 with F-value = 7.52 and p-value = 0.0151. The residual analysis shows Sum of Squares = 47.61 with 1° of freedom and Mean Square = 47.61, with an F-value indicating significance. The lack-of-fit test shows Sum of Squares = 21.16 with 1° of freedom, Mean Square = 21.16, F-value = 0.9228, and p-value = 0.0004, which is significant, indicating systematic deviation from the model at 180 days. The pure error from center-point replicates has a Sum of Squares = 34.81 with 1° of freedom and a Mean Square = 34.81, representing inherent experimental variability. The Cor Total for 180-day tensile strength is 129.96, with 1° of freedom and a Mean Square of 129.96, indicating high significance. This dataset on tensile splitting strength can be reused to develop predictive models for the early-age and long-term tensile capacity of sustainable SIFCON. The ANOVA outputs provide coefficients for mix design optimization, enabling researchers to select CHA and fiber combinations to meet specific tensile requirements. The data can also be used to compare the tensile performance of natural fibers with synthetic fibers, to validate numerical models of SIFCON tensile behaviour, and conduct sensitivity analyses to identify critical factors affecting tensile performance. The dataset can be extended through additional testing at other ages, with different fiber types, or with alternative agricultural waste ashes. Durability Sorptivity The sorptivity equations show a clear time-dependent transition from an early-age increase in permeability to long-term densification. At 28 days, the positive linear coefficients for CHA (+7.9127) and sisal fiber (+7.6549) indicate that both additions initially increase water absorption. This is attributed to (i) the dilution effect of unreacted CHA particles, which leave additional capillary pores, and (ii) the hydrophilic nature of sisal fibers which creates wicking pathways at the fiber-matrix interface. However, the large negative interaction terms (such as –144.134·AC, –139.95·AD) reveal that the actual sorptivity is not simply additive but strongly influenced by the combined presence of CHA, curing conditions (C), and other factors (D). The negative coefficients for terms involving C and D suggest that improved curing or mix adjustments can partially offset the early increase. At 56 days, the linear coefficients for CHA and sisal decrease to +5.228 and +4.910, respectively, reflecting the onset of pozzolanic reactions. CHA’s amorphous silica begins to consume calcium hydroxide, forming additional C-S-H gel that fills capillary voids and refines the pore structure. The negative interaction terms become even larger (e.g., –165.01·AC), indicating that the beneficial effect of CHA is amplified when combined with appropriate curing and mix design. At 90 days, the CHA coefficient rises again to +7.324, while the sisal coefficient remains similar (+7.057). This temporary increase may be due to continued development of fiber-matrix interface or the formation of new micro-cracks as fibers adjust within the hardening matrix. Nevertheless, the overall magnitude of the negative interaction terms remains high, demonstrating that the system is still evolving toward a more impermeable state. The most significant improvement occurs at 120 days, where the positive linear terms for CHA and sisal drop to +3.308 and +3.132, respectively – less than half of their 28-day values. Additionally, the interaction terms with curing variables become positive (+16.523·AC, etc.), indicating that after extended curing, the combined system reduces sorptivity below that of the plain matrix. This reversal is attributed to two mechanisms: (i) complete pozzolanic reaction of CHA, which fills the capillary network with secondary C-S-H, and (ii) formation of a calcium-silicate-rich protective layer around sisal fibers, which seals the interface and reduces hydrophilic absorption. Therefore, for optimal sorptivity (lowest water uptake), the data support a CHA content of 15–20% and sisal fiber of 1.0–1.5% with curing extended to at least 120 days. Early-age water absorption is a temporary drawback that can be managed by adequate moist curing. Chloride conductivity Chloride conductivity is a critical durability indicator because it directly measures how easily chloride ions, the primary cause of reinforcement corrosion, penetrate the concrete cover. The equations for chloride conductivity show a temporal evolution similar to sorptivity, but with an even more pronounced long-term improvement. At 28 days, the linear coefficients for CHA (+7.374) and sisal (+7.109) are positive, indicating that both materials initially increase chloride transport. Unreacted CHA particles act as permeable inclusions, and natural fibers create preferential pathways along their length. However, the large negative interaction terms (–132.759·AC, –128.905·AD) again indicate that the net conductivity is strongly modulated by other variables. The positive AB term (+0.228) suggests a mild synergy between CHA and Sisal, even at early ages. Furthermore, at 56 days, the linear coefficients decrease modestly to +7.024 and +6.746. The negative interaction terms remain large (–128.801·AC), indicating that pozzolanic activity has begun to block some chloride pathways. At 90 days, further reductions occur, with CHA = +6.656 and Sisal = +6.391. The negative interaction terms are still substantial (–97.54·AC, etc.), indicating that pore refinement is ongoing but not yet complete. The dramatic transformation occurs at 120 days. The linear coefficients plunge to +1.551 for CHA and +1.3685 for Sisal – approximately five times lower than at 28 days. The interaction term between CHA and curing (AC) is –77.97, still negative but much smaller in magnitude, implying that most of the pore-structure refinement has already been achieved. The positive AB term drops to +0.0206, indicating that the synergistic protection of fibers by CHA is now fully active. The near-elimination of the positive linear effects means that a well-designed mix (15–20% CHA, 1.0–1.5% Sisal) after 120 days of curing will have chloride conductivity comparable to, or even lower than, conventional SIFCON without agricultural wastes. Oxygen permeability index The oxygen permeability index (OPI = –log₁₀ K, where K is the Darcy coefficient) is a measure of gas permeability. Higher OPI values indicate lower permeability and better resistance to carbonation and gas-borne aggression. The OPI equations show behaviour distinct from that of sorptivity and chloride conductivity. The linear coefficients for CHA and sisal are negative across all ages, indicating that increasing either material tends to reduce OPI (i.e., increase gas permeability) when considered in isolation. At 28 days, the coefficients are CHA = –1.47076 and sisal = –1.67382. This indicates that unreacted CHA and hydrophilic sisal fibers create an open pore network that allows gas flow. However, the positive interaction term (+0.234256·AB) and the positive coefficients for variables C and D (+0.594859·C, +0.365485·D) show that proper curing and mix adjustments can partially offset this reduction. As curing age increases, the negative coefficients become more negative: at 56 days, CHA = –2.40878 and sisal = –2.57843; at 90 days, CHA = –2.44964 and sisal = –2.62766; and at 120 days, CHA = –4.868 and sisal = –5.0049. The full equation includes many cross-terms (–107.269·AC, –117.617·AD) that are also large and negative, along with the positive AB term (+0.022 at 120 days). The net OPI for any specific combination of A (CHA), B (sisal), C, and D must be calculated from the full polynomial. At the optimal mix (CHA 15–20%, sisal 1.0–1.5%) and with extended curing (120 days), the large negative main effects are substantially counterbalanced by interaction terms. For example, the –69.20 AC term for typical C values partially offsets –4.868 A. Moreover, the positive AB term (+0.022) adds a small synergistic effect. The practical conclusion is that although CHA and sisal fiber individually tend to increase gas permeability at early ages, a fully pozzolanic matrix with well-bonded fibers can achieve an OPI comparable to conventional SIFCON. However, the data suggest that achieving very low gas permeability (high OPI) requires extended moist curing (≥90 days) and strict control of CHA content within the 15–20% window. If curing is shortened or CHA exceeds 20%, the negative main effects may dominate, leading to unacceptable gas permeability. Therefore, for applications where carbonation resistance is critical (such as thin sections or high-exposure classes), the mix must be designed at the lower end of the sisal range (1.0%) and cured for the full 120 days. Conclusion Based on the response surface plots, ANOVA analyses, and the detailed discussion in the document, the optimum usage ranges for coffee husk ash (CHA) and sisal fiber vary by mechanical property and curing age. For early-age compressive strength (3–14 days), the matrix dominates the response, with an optimum CHA content of 10–15% and no significant contribution from sisal fiber; however, by 180 days the pozzolanic reaction and fiber-matrix bond maturation shift the optimum to 15–20% CHA and 1.0–1.5% sisal fiber. Flexural strength, which is highly sensitive to fiber bridging, already exhibits a clear optimum at 3 days (10–12% CHA and 1.0–1.5% sisal), and at 180 days the optimum also lies at 15–20% CHA with the same sisal range, where CHA protects the sisal fibers from alkaline degradation. Tensile splitting strength follows an intermediate pattern: early-age values are matrix-controlled (optimum CHA ∼10–12% with negligible fiber effect), but at 180 days the optimum shifts to 15–20% CHA and 1.0–1.5% sisal, with a near-linear fiber contribution and no decline up to 2% sisal. Consequently, a unified optimum for balanced long-term mechanical performance of SIFCON incorporating sisal fiber and coffee husk ash is 15–20% CHA (by weight of cement) and 1.0–1.5% sisal fiber (by volume of concrete). This range simultaneously maximizes compressive, flexural, and tensile strengths at 180 days while ensuring workability and the synergistic protection of natural fibers by the pozzolanic matrix. Limitations While this research provides valuable insights, the authors acknowledge certain limitations. First, experimental data are susceptible to human error, including inaccuracies in data collection, measurement, recording, or handling. Despite adherence to standard protocols, manual interventions and subjective judgements during experimentation may have introduced minor inconsistencies. Second, further testing is required to fully validate and complement the current findings. Specifically, microstructural analysis (such as scanning electron microscopy, X-ray diffraction, or optical microscopy) is necessary to examine the internal structure, phase composition, and material characteristics in greater detail. Such analyses would help confirm the observations reported here and reduce uncertainties arising from the current experimental approach. Future work should therefore incorporate microstructural characterization to enhance the robustness and reliability of the conclusions. Ethics Statement The authors of this dataset declare that they have complied with the ethical requirements for publication in Data in Brief. This research involved no human subjects, no animal experiments, and no data collected from social media platforms. All data presented are free from ethical concerns regarding privacy, consent, or institutional review. CRediT Author Statement Dawit Beza Alemu contributed to conceptualization, methodology, formal analysis, investigation, original draft writing, and data curation. Opeoluwa Rotimi Dada (corresponding author) handled conceptualization, validation, resources, review and editing, supervision and project administration. Bolanle Deborah Ikotun contributed to methodology, software, validation, formal analysis, investigation, data curation, review and editing, visualization, supervision, and funding acquisition. All authors approved the final manuscript Acknowledgements The authors wish to thank the University of South Africa (UNISA) for its support and the Department of Civil and Environmental Engineering and Building Science for providing the required materials during our experimental work. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships which have, or could be perceived to have, influenced this article. Contributor Information Opeoluwa Rotimi Dada, Email: dadaor@unisa.ac.za. Dawit Beza Alemu, Email: 23955090@mylife.unisa.ac.za. Bolanle Deborah Ikotun, Email: Ikotubd@unisa.ac.za. Data Availability Mendeley DataMultivariate Dataset and Response Surface Optimization for Predicting Mechanical and Durability Properties of Slurry-Infiltrated Fibrous Concrete Incorporating Sisal Fiber and Coffee Husk Ash (Original data). Mendeley DataMultivariate Dataset and Response Surface Optimization for Predicting Mechanical and Durability Properties of Slurry-Infiltrated Fibrous Concrete Incorporating Sisal Fiber and Coffee Husk Ash (Original data). References Associated Data Data Availability Statement Mendeley DataMultivariate Dataset and Response Surface Optimization for Predicting Mechanical and Durability Properties of Slurry-Infiltrated Fibrous Concrete Incorporating Sisal Fiber and Coffee Husk Ash (Original data). Mendeley DataMultivariate Dataset and Response Surface Optimization for Predicting Mechanical and Durability Properties of Slurry-Infiltrated Fibrous Concrete Incorporating Sisal Fiber and Coffee Husk Ash (Original data).