Response Surface Optimization of Structural Concrete Incorporating Two Gold-Mine Tailing Fractions

Abstract

Gold-mine tailings have attracted increasing interest as alternative constituents in cement-based materials, yet their use in structural concrete remains limited by the lack of multivariable approaches capable of capturing the interaction between tailing fractions with different functional roles. In this study, a tailing-derived fine aggregate and a fine tailing sludge from the Cisneros Project (Santo Domingo, Antioquia, Colombia) were jointly incorporated into structural concrete and evaluated through a response surface methodology based on a central composite design. The tailings were characterized by physical, chemical, mineralogical, and morphological analyses, while concrete mixtures proportioned according to ACI 211 were assessed in terms of 28-day compressive strength. The statistical model revealed a significant quadratic response and a strong interaction between both incorporation variables. The most favorable strength region, based solely on 28-day compressive strength, was identified at sludge contents below 20% and tailing aggregate replacement below 90%, with the latter interpreted as a preliminary mechanical threshold rather than as a practical recommendation for field application. Higher incorporation levels led to strength losses associated with the increasing fineness of the system and greater water demand. This study demonstrates that the performance of tailing-modified structural concrete depends on the coordinated dosage of fractions with distinct roles and provides preliminary mechanical incorporation limits based solely on 28-day compressive strength. Since durability and environmental safety tests, including heavy metal/cyanide leaching, permeability, shrinkage, and chemical resistance, were not conducted, these limits should not be interpreted as definitive recommendations for long-term structural application.

1. Introduction

Mining is essential to economic development, yet it also generates large volumes of solid residues that remain among the most persistent environmental liabilities of the extractive sector [1]. Mine tailings produced during crushing, grinding, concentration, and flotation processes are commonly stored in impoundments or containment areas, where inadequate management may lead to long-term impacts on soils, surface water, groundwater, and air quality [2]. Recent reviews have highlighted that mine tailings are generated worldwide on a massive scale and that the long-term stability of tailings storage facilities continues to raise environmental and geotechnical concerns, particularly in regions with intensive metal mining activity [3,4,5,6]. In gold-producing regions such as Antioquia, Colombia, tailings accumulation is not only a waste-management issue but also a territorial and environmental challenge associated with potentially toxic elements, geomorphological alteration, and increasing pressure on land use. Recent studies in Antioquia have reported environmental risk patterns in areas influenced by gold mining operations, reinforcing the need for strategies that move beyond disposal and toward the productive reuse of mining residues [7]. Within this context, the incorporation of mine tailings into cement-based materials has emerged as a promising route for reducing waste accumulation while partially replacing virgin raw materials used in construction.

The reuse of tailings in concrete is particularly attractive because these residues often contain silica-rich and aluminosilicate phases that may contribute either as fine aggregate substitutes or as mineral additions, depending on their mineralogical composition, morphology, and particle-size distribution [8]. Previous studies have shown that silicoaluminous residues can improve particle packing, alter hydration-related behavior, and, in some cases, enhance the physical performance of cementitious systems when their fineness and incorporation level are properly controlled [9,10,11,12]. More recent investigations on tailing-derived materials have confirmed that mine residues can be successfully incorporated into cement-based matrices, although their effect on strength development depends strongly on processing route, replacement ratio, and particle characteristics [13,14]. Despite this progress, most published studies have examined tailings as isolated constituents, typically evaluating a single fraction either as filler, mineral addition, cement replacement, or fine aggregate replacement [15,16]. This includes previous work by the authors [15], in which gold-mine tailings were mainly evaluated as a partial cement replacement and the discussion focused on identifying feasible replacement levels based on mechanical performance. However, that study did not address the simultaneous incorporation of two tailing fractions with differentiated functional roles within the same concrete mixture. Comparatively less attention has been given to the combined incorporation of fine tailing sludge and tailing-derived fine aggregate, even though these fractions may act through distinct and interacting mechanisms. Very fine sludge fractions may promote interstitial filling and local matrix densification, but excessive incorporation may also increase water demand and reduce mixture efficiency. In contrast, coarser tailing-derived fractions primarily affect the granular skeleton, packing density, and void structure. As a result, the behavior of tailing-modified concrete cannot be fully understood through one-factor replacement schemes alone, particularly in structural concrete, where mechanical performance must remain within practical and safe limits.

The tailings generated in the Cisneros Project, located in Santo Domingo, Antioquia, provide an appropriate case study for this type of assessment. These residues are associated with auriferous exploitation under specific geological and operational conditions and include quartz-rich and silicoaluminous fractions with potential for incorporation into cement-based materials. Previous characterization of the evaluated tailings identified quartz, feldspars, clay minerals, and an important amorphous component, suggesting that their engineering performance may depend on controlled dosage and on the functional role assigned to each fraction within the mixture [17,18].

From the perspective of construction materials engineering, the key question is therefore not only whether gold-mine tailings can be used in concrete, but under which combination of incorporation levels they remain compatible with structural performance. In this regard, Response Surface Methodology (RSM), particularly when coupled with Central Composite Design (CCD), provides a robust framework for simultaneously evaluating interacting variables and identifying nonlinear response regions that are difficult to capture through conventional trial-and-error approaches. This is especially relevant in multi-component systems where beneficial packing effects at moderate dosages may be offset by strength losses at higher incorporation levels due to excessive fineness and increased water demand [2,8,12]. This study investigates the combined incorporation of a tailing-derived fine aggregate and a fine tailing sludge from the Cisneros Project into structural concrete designed for a target compressive strength of 21 MPa. The materials were characterized by standardized physical tests, X-ray fluorescence, X-ray diffraction, and scanning electron microscopy, and their combined effects on 28-day compressive strength were evaluated through a Central Composite Design within a Response Surface Methodology framework. The present work evaluates a combined incorporation of two tailing fractions in which one fraction modifies the granular skeleton as a fine aggregate replacement, while the other alters the cementitious matrix as a partial cement replacement. Therefore, the substantive contribution of this study is not the use of RSM alone, but the identification of how both tailing fractions interact within the same structural concrete system. This approach enables the definition of practical incorporation limits based on the coordinated balance between particle packing, matrix refinement, water demand, and cement dilution.

2. Materials and Methods

2.1. Raw Materials

The materials used in this study were obtained from the Cisneros mining project operated by Antioquia Gold Ltd., located in the municipalities of Santo Domingo and Cisneros, Antioquia, Colombia. The geographical location of the study area is shown in Figure 1. This mining district is characterized by underground gold extraction and beneficiation processes that generate both flotation sludge and coarser tailing residues.

Two tailing fractions were selected for incorporation into concrete. The first was a fine tailing sludge generated during ore beneficiation, which was used as a partial replacement of cement by mass in the cementitious system. The second was a coarser tailing-derived material, which was used as a partial replacement for natural fine aggregate. Commercial Type I Portland cement complying with ASTM C150 was used as the binder. Natural river sand and crushed coarse aggregate were used as reference aggregates for the control mixture. The coarse aggregate had a nominal maximum size of 19 mm [19].

2.2. Material Characterization

The characterization program included physical, chemical, mineralogical, and morphological analyses of the conventional materials and tailing-derived fractions. Physical characterization was carried out in accordance with the applicable Colombian technical standards (NTC). For the cement–sludge system, density, normal consistency, and setting times were determined in order to evaluate the influence of the fine residue on binder-related behavior. For the aggregates, particle-size distribution, absorption, specific gravity, unit weight, and void content were determined for use in concrete proportioning. The tailing materials were additionally characterized by X-ray fluorescence (XRF), X-ray diffraction (XRD), and scanning electron microscopy coupled with energy-dispersive spectroscopy (SEM-EDS). XRD analysis was performed using a Panalytical X’Pert PRO MRD diffractometer (Malvern Panalytical, Almelo, The Netherlands) with Cu Kα radiation, operated at 40 kV and 30 mA. Diffraction patterns were collected over a 2θ range from 5.018° to 69.966°, with a step size of 0.026° 2θ and a counting time of 58.075 s per step. A fixed divergence slit of 1° and a receiving slit of 0.100 mm were used. Phase identification was conducted using X’Pert HighScore Plus software, version 3.0 (Malvern Panalytical, Almelo, The Netherlands), together with the corresponding diffraction database/library available in the software. The amorphous fraction was estimated by adding fluorite as an internal standard at 20 wt.% and applying Rietveld refinement to the identified crystalline phases. This approach allowed the crystalline and amorphous fractions to be indirectly estimated by comparing the refined intensity of the internal standard with the intensity of the crystalline phases present in the samples. XRF analysis was performed using a Zetium Mineral Edition 4 kW spectrometer (Malvern Panalytical, Almelo, The Netherlands) to determine the bulk chemical composition of the samples. To compare the total chemical composition obtained by XRF with the crystalline fraction quantified by XRD, an equivalent crystalline chemical composition was estimated from the mineralogical refinement and the stoichiometry of the phases included in the model. In this comparison, XRF represents the total composition of the original sample, XRD represents the contribution of the crystalline fraction estimated in the original sample excluding the internal standard, and the XRF–XRD difference corresponds to the chemical fraction not explained by the modeled crystalline phases. This difference was interpreted as a combination of the real amorphous fraction and possible deviations associated with the use of proxy phases in the refinement. SEM-EDS observations were performed using a JEOL JSM-5910LV scanning electron microscope (JEOL Ltd., Tokyo, Japan). The samples were analyzed as fragments coated with an 8 nm gold film to improve surface electrical conductivity. Localization micrographs were interpreted as backscattered electron images (BSE/BEC), where grayscale contrast was mainly associated with compositional differences and brighter domains tended to indicate a higher average atomic number. EDS spectra were treated as semi-quantitative point analyses and were interpreted together with morphology, grayscale contrast, and the exact position of each analyzed point, rather than as representative bulk compositions of the samples.

2.3. Concrete Mixture Design

A reference structural concrete mixture was proportioned according to ACI 211 for a target 28-day compressive strength of 21 MPa. This mixture was used as the control composition. The detailed mixture proportions used in the experimental program are presented in

Table 1. Mixture proportions of the control and tailing-modified concrete mixtures.

Mix	Water (kg/m3)	Cement (kg/m3)	Fine Tailing Sludge (kg/m3)	Natural Sand (kg/m3)	Tailing-Derived Fine Aggregate (kg/m3)	Coarse Aggregate (kg/m3)	w/c
Control	1.59	3.04	0	3.79	0	6.89	0.52
M1	1.59	2.43	0.61	0.76	3.04	6.89	0.52
M2	1.59	2.58	0.46	0.00	3.79	6.89	0.52
M3	1.59	2.74	0.30	0.19	3.60	6.89	0.52
M4	1.59	2.43	0.61	0.19	3.60	6.89	0.52
M5	1.59	2.74	0.30	0.76	3.04	6.89	0.52
M6	1.59	2.86	0.18	0.46	3.34	6.89	0.52
M7	1.59	2.43	0.61	0.19	3.60	6.89	0.52
M8	1.59	2.43	0.61	0.76	3.04	6.89	0.52
M9	1.59	2.74	0.30	0.19	3.60	6.89	0.52
M10	1.59	2.74	0.30	0.19	3.60	6.89	0.52
M11	1.59	2.43	0.61	0.19	3.60	6.89	0.52
M12	1.59	2.74	0.30	0.76	3.04	6.89	0.52
M13	1.59	2.74	0.30	0.76	3.04	6.89	0.52
M14	1.59	2.58	0.46	0.99	2.81	6.89	0.52
M15	1.59	2.31	0.73	0.46	3.34	6.89	0.52
M16	1.59	2.58	0.46	0.46	3.34	6.89	0.52
M17	1.59	2.43	0.61	0.76	3.04	6.89	0.52

. The nominal water-to-cement ratio was kept constant at 0.52 for all mixtures.

The experimental mixtures were obtained by modifying the control mixture through the incorporation of the two tailing fractions: partial replacement of natural sand with tailing-derived fine aggregate, and partial replacement of cement by mass with fine tailing sludge. The water-to-cement ratio was kept constant for all mixtures. The experimental domain was defined using two independent variables: the percentage of natural sand replacement by tailing-derived fine aggregate and the percentage of cement replacement by fine tailing sludge. These variables were selected to represent the two principal modes by which the mine tailings could affect the concrete system, namely modification of the granular skeleton and modification of the fine fraction of the cementitious matrix.

2.4. Preparation, Casting, and Curing of Specimens

Cylindrical specimens measuring 150 mm × 300 mm were cast and cured following NTC 550:2020. After an initial curing period of 24 ± 8 h, the specimens were demolded and stored under water at 23 ± 2 °C until the testing age. For each experimental mixture, three cylindrical specimens were produced and tested. Compressive strength was determined at 28 days according to NTC 673:2021, and the average value of the three specimens was used as the response variable for the statistical analysis.

2.5. Experimental Design and Statistical Analysis

The experimental program was structured using Response Surface Methodology (RSM) coupled with a Central Composite Design (CCD) in order to evaluate the combined influence of the selected factors on 28-day compressive strength.

Factor A corresponded to the replacement level of natural sand by tailing-derived fine aggregate, whereas Factor B corresponded to the replacement level of cement by fine tailing sludge. The CCD consisted of factorial points, axial points, and replicated center points distributed within the experimental region shown in Figure 2. The center point corresponded to 87.5% sand replacement and 15% cement replacement by sludge. In the adopted design, the axial domain extended approximately from 75% to 100% for sand replacement and from 5% to 25% for cement replacement by sludge, while the factorial region ranged from 80% to 95% sand replacement and from 10% to 20% cement replacement by sludge. In total, 17 experimental runs were produced according to the statistical design.

Based on the experimental results, a second-order polynomial model was fitted to the compressive-strength response, including linear, interaction, and quadratic terms. The statistical significance of the fitted model and its coefficients was assessed by analysis of variance (ANOVA). Response surface plots and contour plots were generated to identify the region of incorporation associated with the most favorable mechanical performance.

3. Results

3.1. Physical and Fresh-State Behavior of Tailing-Modified Systems

The preliminary physical characterization of the tailing-modified systems provided essential information for interpreting the subsequent mechanical response of structural concrete incorporating gold-mine tailings [16,20]. In particular, the evaluation of the cement–sludge system and the particle-size distribution of the tailing-derived fine aggregate made it possible to identify the principal mechanisms through which both residues affected fresh-state behavior, particle packing, and the overall efficiency of the cementitious matrix. Rather than acting as inert substitutions, the two tailing fractions modified the internal balance of the system through changes in density, water demand, setting behavior, and granular distribution, all of which are directly relevant to the compressive-strength trends discussed later [16,20]. As shown in Figure 3, the partial replacement of cement by fine tailing sludge produced a differentiated effect on density and normal consistency within the cementitious system. Density increased progressively with sludge incorporation up to 20%, reaching a maximum value of 2.96 g cm⁻³.

This trend suggests that, at moderate replacement levels, the very fine particles of the sludge promoted a denser solid arrangement within the binder system, most likely by occupying interstitial spaces between cement grains and improving particle accommodation. Such behavior is consistent with the well-known filler effect of finely divided mineral additions, which may enhance packing efficiency even when no direct cementitious reactivity is demonstrated [20,21]. However, the observed improvement was not indefinite, indicating that the contribution of the sludge was not governed exclusively by densification. Beyond a certain replacement level, the increasing amount of fine residue likely altered the balance between solid packing and water availability, thereby limiting the beneficial effect initially observed. The behavior of normal consistency further supports this interpretation. As sludge content increased, the amount of water required to achieve standard paste consistency also increased, reflecting the higher specific surface area introduced by the fine residue [21]. This response indicates that the tailing sludge did not merely occupy voids, but also increased the water demand of the cementitious system. In practical terms, this is a critical observation because the water-to-cement ratio was kept constant for all concrete mixtures; therefore, any increase in water demand induced by the sludge could directly reduce fresh-state efficiency and, ultimately, affect compressive strength development. The simultaneous increase in density and consistency requirement indicates that the sludge acted through two opposing mechanisms: on the one hand, it contributed to a more compact solid matrix through microfilling; on the other hand, it increased the surface area that had to be wetted, which may impair particle dispersion and effective hydration when replacement levels become excessive. This dual effect is central to understanding the nonlinear response observed later in the statistical analysis. The setting-time results shown in Figure 4 also revealed a systematic effect of the fine tailing sludge on binder behavior. Both initial and final setting times increased with sludge incorporation, indicating a progressive retardation of the setting process as cement replacement by mass increased.

This behavior is technically coherent with the partial replacement of cement by a residue that, although physically fine, does not necessarily contribute to early hydration to the same extent as Portland cement. In this context, the delay in setting can be interpreted as the result of a dilution effect combined with the greater water demand imposed by the sludge fraction. Because part of the cement was replaced by a non-clinker material, the concentration of hydraulically active phases in the paste decreased, which likely slowed the formation of the solid skeleton responsible for early stiffening. At the same time, the increase in fineness may have further modified the kinetics of the fresh system by affecting water distribution and particle interaction. These results are relevant because they confirm that the sludge affected not only packing but also the early temporal evolution of the binder phase.

From the standpoint of concrete proportioning, the observed behavior of the cement–sludge system suggests that the fine tailing sludge may provide a beneficial physical contribution only within a limited replacement range. At moderate dosages, the denser particle arrangement may favor matrix refinement and improved accommodation of solids. However, once the replacement level becomes too high, the increase in water demand and the slower setting kinetics may offset the packing-related benefit. This balance is particularly important in structural concrete, where fresh-state performance and hardened-state strength are strongly interconnected. In other words, the sludge cannot be interpreted simply as a beneficial fine addition; its effect depends on the point at which the positive contribution of microfilling is overtaken by the adverse consequences of excessive fineness and cement dilution. The grading behavior of the fine aggregate mixtures reinforced this interpretation from the perspective of the granular skeleton. As shown in Figure 5, the particle-size distribution curves of the natural sand and the tailing-derived fine aggregate combinations followed a continuous descending trend but progressively shifted toward finer distributions as the proportion of tailing-derived material increased.

The 100/0 mixture, composed entirely of conventional sand, exhibited the coarsest grading profile, whereas the 0/100 mixture, composed entirely of the tailing-derived material, showed the finest distribution, with 79.73% passing the 0.6 mm sieve. The intermediate combinations followed a gradual transition between these extremes, confirming that the replacement of natural sand with tailing-derived aggregate systematically modified the granular composition of the fine fraction toward finer particle assemblages. The fineness modulus varied from 2.01 to 1.57, indicating continuous gradations that remained technically admissible for cementitious mixtures, although clearly shifted toward greater fineness at higher replacement ratios [22].

This change in grading is highly relevant for the behavior of structural concrete because it directly affects both particle packing and workability. A somewhat finer granular skeleton may improve the redistribution of solids and reduce internal voids when used in controlled proportions. In this sense, the tailing-derived aggregate may contribute positively by refining the arrangement of the fine fraction and enhancing the contact between particles. Nevertheless, this same refinement also increases the total surface area of the aggregate system and therefore the amount of water required to coat the particles and maintain adequate mobility in the fresh state. Under the experimental conditions of this study, where the water-to-cement ratio was kept constant, this effect becomes particularly important. Any increase in surface area associated with finer aggregate replacement could reduce the amount of free water available for lubrication and hydration, thereby decreasing mixture efficiency and eventually limiting strength development. The grading shift observed in Figure 5 is therefore fully consistent with the trends identified in consistency and setting time. Although workability-related parameters were not included as independent response variables in the RSM model, the observed consistency of the mixes was consistent with the interpretation that higher tailing incorporation increased the water demand of the system. This behavior is consistent with the finer grading of the tailing-derived aggregate and the higher normal consistency required by the cement–sludge system, suggesting a reduction in effective water availability and fresh-state efficiency at higher replacement levels.

The combined interpretation of Figure 3, Figure 4 and Figure 5 indicates that both tailing fractions acted through distinct but convergent physical mechanisms. The fine sludge, used as partial cement replacement, influenced the cementitious matrix by modifying density, water demand, and setting kinetics. The coarser tailing-derived aggregate, in turn, affected the granular structure by shifting the fine aggregate system toward finer distributions. Although these two effects originated from different fractions, both increased the fineness of the overall concrete system. Consequently, the performance of the mixtures depended on the balance between improved particle accommodation at moderate incorporation levels and the adverse effects derived from excessive specific surface area at higher levels. This balance provides a physically coherent basis for the nonlinear and interaction effects later identified by the response surface model.

Additional physical properties of the conventional aggregates further support this interpretation by confirming that the reference materials provided an adequate and stable baseline for concrete production. The natural fine aggregate exhibited a bulk density of 1.18 g/cm³ and a void content of 56.42%, whereas the coarse aggregate showed a bulk density of 1.46 g/cm³ with 5.60% voids. Absorption values of 1.21% for the fine aggregate and 1.88% for the coarse aggregate were also recorded, together with densities of 2.72 g/cm³ and 1.58 g/cm³, respectively. These values indicate that the conventional constituents were technically suitable for structural concrete production and that the variations observed in fresh-state behavior were not attributable to deficiencies in the reference aggregates. Instead, the changes should be interpreted as a direct consequence of the intrinsic characteristics of the tailing fractions and their interaction with the conventional constituents. The preliminary physical results demonstrate that the incorporation of gold-mine tailings modified the concrete system through a coupled fineness-driven response rather than through isolated substitution effects. Moderate incorporation levels favored denser particle accommodation and potentially more efficient packing, while excessive replacement levels increased water demand, delayed setting, and reduced fresh-state efficiency. This interpretation is especially important because it anticipates the later strength response not as a monotonic function of either factor, but as the outcome of a balance between beneficial packing-related effects and the detrimental consequences of excessive fineness under constant water-to-cement conditions [23]. In this sense, the physical and fresh-state characterization does not merely describe the materials; it establishes the mechanistic framework required to understand why the compressive-strength response of tailing-modified structural concrete followed a nonlinear and interaction-controlled pattern.

3.2. Chemical, Mineralogical, and Morphological Characteristics of the Tailings

The chemical, mineralogical, and morphological characterization of the evaluated tailings provides the material basis for interpreting their subsequent behavior in structural concrete. Although both fractions were obtained from the same mining operation, they exhibited distinct textural and compositional features that support differentiated functional roles within the concrete system. In particular, the fine flotation tailing sludge (sample F) displayed characteristics consistent with a highly dispersed fine fraction capable of strongly affecting the cementitious matrix, whereas the coarser tailing-derived material (sample G) exhibited a more stable morphology compatible with its use as an alternative fine aggregate [10]. These differences are especially relevant because the performance of the evaluated mixtures depended not only on the presence of mine tailings, but on the coupled interaction between two fractions acting through different physical scales and mechanisms [24]. The SEM observations revealed clear morphological differences between both residues, as illustrated in Figure 6. Sample F showed a highly heterogeneous microstructure composed of very fine particles with angular edges, fractured surfaces, and abundant porous domains.

This morphology is consistent with the intensive grinding and flotation processes associated with ore beneficiation, which typically generate fragmented particles with irregular geometry and elevated specific surface area. From the standpoint of cement-based materials, such characteristics are important because they may favor interstitial filling and local matrix densification at controlled replacement levels. At the same time, however, the same features may increase water demand and reduce fresh-state efficiency when the proportion of ultrafine material becomes excessive. Therefore, the morphology of sample F supports the interpretation that this residue acts primarily through physical modification of the binder system rather than through the direct behavior expected from a conventional aggregate.

By contrast, sample G exhibited a more compact and comparatively uniform morphology, with angular but less rugged surfaces and a lower apparent degree of microstructural disruption. Although still irregular in shape, its particles appeared less porous and less dispersed than those of sample F, suggesting a more stable granular behavior. This distinction is important because it indicates that sample G is less likely to affect the cementitious matrix through high fineness or excessive surface-area effects and is more likely to contribute through modification of the granular skeleton. In practical terms, this supports its interpretation as a tailing-derived fine aggregate rather than as a highly reactive or strongly interactive powder. Consequently, the SEM observations confirm that the two evaluated tailing fractions should not be considered equivalent replacement materials, even if they originate from the same mining residue stream. Figure 6 clearly illustrates these morphological contrasts and highlights the more porous and disrupted structure of sample F relative to sample G.

The mineralogical characterization further reinforced this distinction. The XRD analysis showed that quartz was the dominant crystalline phase in both materials, with SiO₂ contents of 58.03% for sample F and 54.98% for sample G. Additional phases identified in both fractions included calcium carbonate, feldspars such as microcline, and clay-associated minerals, including kaolinite, biotite, and sericite [17]. This mineral assemblage is characteristic of silicoaluminous mining residues and is relevant for two complementary reasons [17]. First, the predominance of quartz indicates that a substantial fraction of the solids behaves as a stable mineral framework with limited intrinsic reactivity under ordinary cement hydration conditions. Second, the coexistence of feldspathic and clay-bearing phases suggests the presence of less ordered mineral domains that may influence the concrete system through filler-related effects and possible physicochemical interaction, particularly when present in very fine form. A particularly relevant result was the Rietveld-based estimation of the amorphous fraction using fluorite as an internal standard, which indicated an amorphous fraction of approximately 70%. This finding is important because amorphous silicoaluminous phases are generally more susceptible than highly crystalline phases to participate in physicochemical interactions within cementitious environments [25]. Nevertheless, within the scope of the present study, this result should be interpreted with caution. The presence of a substantial amorphous fraction is an indicator of potential activity, but it does not by itself demonstrate pozzolanic behavior or the formation of specific secondary hydration products. Since the present work did not include direct characterization of hydrated phases, such as thermogravimetric analysis, selective dissolution, or hydration-product-specific diffraction analysis after curing, the amorphous content should be discussed as a feature that may contribute to matrix behavior, but not as conclusive evidence of cementitious or pozzolanic reactivity. This distinction is important to preserve a rigorous interpretation of the material characterization and to avoid attributing mechanistic effects that were not directly verified experimentally [26].

The XRF analysis, summarized in Figure 7, further corroborated the silicoaluminous nature of both evaluated residues. In addition to the high silica content, the samples contained relevant proportions of Al₂O₃ (15.33–16.78%), CaO (5.68–6.86%), and Fe₂O₃ (6.10–8.24%), together with lower amounts of K, Na, Mg, and Ti oxides.

This compositional profile is consistent with the XRD results and confirms that both fractions belong to a chemically compatible mineral family for potential incorporation into cement-based materials [13,27]. The combined presence of silica, alumina, and calcium-bearing phases suggests that these tailings may influence concrete performance through more than one mechanism. On the one hand, the mineral fines can contribute physically by filling voids, modifying particle distribution, and improving local packing under controlled proportions. On the other hand, the amorphous silicoaluminous fraction may provide limited physicochemical interaction under hydration conditions. However, as noted above, the present evidence supports the discussion of compatibility and potential contribution, but not the direct confirmation of secondary hydrate formation.

An additional point of interest is that, despite the morphological differences revealed by SEM, the XRF results indicate that both samples remain broadly similar in elemental composition. This suggests that their different roles in concrete are governed less by major chemical dissimilarity than by differences in particle size, morphology, and physical dispersion state. In other words, the main distinction between both tailing fractions is not that one is chemically suitable and the other is not, but that each one interacts with the concrete system through a different structural scale. Sample F, because of its finer size, porous morphology, and elevated amorphous content, is better interpreted as a finely divided mineral fraction capable of modifying the cementitious matrix through microfilling and fineness-related effects. Sample G, in contrast, behaves more consistently as an alternative fine aggregate whose main contribution lies in reshaping the granular skeleton and altering the grading of the mixture.

Taken together, the SEM, XRD, and XRF results indicate that both tailing fractions are technically suitable for valorization in cement-based materials, although their roles within the concrete system are not equivalent. The fine tailing sludge should be understood primarily as a highly dispersed mineral fraction whose influence depends on the balance between microfilling, cement dilution, and the adverse consequences of excessive fineness [11]. The coarser tailing-derived material, in contrast, should be interpreted mainly as a fine aggregate substitute whose effect is governed by changes in particle-size distribution and granular arrangement. This distinction is essential for interpreting the subsequent compressive-strength response because it supports the view that the performance of the evaluated concretes was controlled not by the isolated action of each residue, but by the interaction between a matrix-modifying fine fraction and a granular tailing fraction acting simultaneously within the same mixture. Under this framework, the material characterization does not merely confirm the feasibility of incorporating mine tailings into concrete; it also explains why the resulting mechanical behavior followed a nonlinear pattern controlled by coordinated dosage rather than by independent replacement effects.

3.3. Response Surface Modeling of Compressive Strength

The mechanical performance of structural concrete incorporating gold-mine tailings was evaluated through a Central Composite Design (CCD), which enabled the simultaneous assessment of the effects of tailing-derived fine aggregate replacement, factor A, and cement replacement by fine tailing sludge, factor B, on 28-day compressive strength [10,13]. As described in Section 2, the experimental domain comprised 17 runs distributed according to the CCD, allowing the estimation of linear, interaction, and quadratic effects within the selected factor ranges. Based on the experimental results, the regression model was refitted by removing the non-significant higher-order terms A²B and AB² to reduce model complexity and avoid overfitting. The simplified second-order polynomial model used to describe the compressive-strength response is expressed in Equation (1). In Equation (1), the variables Sand and Sludge correspond to the actual replacement percentages used in the experimental design, not to coded or normalized variables. Sand represents the percentage of natural sand replaced by tailing-derived fine aggregate, while Sludge represents the percentage of cement replaced by fine tailing sludge.

$$R_c=-4845.74+413.59•\mathrm{Sand}+94.44•\mathrm{Sludge}-590.07•\mathrm{Sand}•\mathrm{Sludge}-120.22•{\mathrm{Sand}}^2-267.01•{\mathrm{Sludge}}^2$$

(1)

The statistical significance of the fitted model and its individual terms is summarized in

Table 2. ANOVA results of the response surface model for 28-day compressive strength.

Source	Sum of Square	p-Value
A-Sand	3.004 × 106	<0.0001
B-Sludge	1.781 × 105	0.0797
AB	4.186 × 106	<0.0001
A2	81,415	0.2184
B2	8.864 × 106	0.0012

.

The ANOVA results show that the interaction between sand replacement and cement replacement by sludge, AB, had a highly significant effect on compressive strength, with p < 0.0001. This indicates that the mechanical response of the concrete system cannot be explained by the isolated influence of each variable alone. Instead, the results suggest that compressive strength depended on the combined effect of the tailing-derived fine aggregate and the fine sludge fraction replacing cement.

The linear term associated with sand replacement, A, was statistically significant, with p < 0.0001, indicating that variation in the tailing-derived fine aggregate content had a direct influence on compressive strength within the evaluated domain. In contrast, the linear term associated with sludge replacement, B, was not statistically significant at the 95% confidence level, with p = 0.0797. However, the quadratic term B² was significant, with p = 0.0012, showing that the effect of cement replacement by sludge was mainly nonlinear. The quadratic term A² was not statistically significant, with p = 0.2184, suggesting that the curvature associated with sand replacement was less relevant than its linear effect and its interaction with sludge content.

Based on the refitted model, the final interpretation was focused on the significant linear contribution of A, the highly significant AB interaction, and the nonlinear contribution of B². The non-significant higher-order terms were removed from the final model to improve parsimony and reduce the risk of overfitting. The model-fit indicators were also included to assess model adequacy. The simplified model showed an R² of 0.9405 and an adjusted R² of 0.9135, indicating a strong agreement between the experimental and predicted compressive-strength values within the evaluated experimental domain. In addition, the comparison between experimental and predicted compressive-strength values showed an adequate agreement within the evaluated domain, supporting the reliability of the simplified model. No systematic deviations were observed in the residual trend, indicating that the model adequately represented the experimental response for the purpose of mechanical optimization. The fitted response surface, shown in Figure 8, illustrates the combined influence of both replacement variables on compressive strength. A region of favorable mechanical performance was identified at sludge replacement levels below approximately 20% and tailing-derived fine aggregate replacement levels below approximately 90%. This region was selected considering not only the predicted compressive-strength response, but also the criterion of maximizing the incorporation of both tailing fractions, particularly the highest feasible sludge content and tailing-derived fine aggregate replacement, while maintaining a favorable mechanical performance. Therefore, the selected region should be interpreted as a compromise solution between mechanical performance and maximum tailings valorization, rather than as a single numerical point of maximum predicted strength. This upper replacement level for the tailing-derived fine aggregate should be interpreted as a preliminary mechanical threshold obtained from the RSM model based on 28-day compressive strength, rather than as a definitive recommendation for concrete production or field-scale implementation.

Within this region, the mixtures reached the highest compressive-strength values, suggesting that the combined incorporation of both tailing fractions can be mechanically feasible when the dosage is controlled. Outside this range, the response progressively declined, indicating that excessive replacement levels may reduce mixture efficiency. This behavior is consistent with the physical characterization discussed in Section 3.1, where increasing tailing incorporation was associated with finer grading, greater water demand, and changes in the cement–sludge system.

A similar interpretation applies to the replacement of natural sand by the tailing-derived aggregate. As shown in Section 3.1, increasing the proportion of sample G progressively shifted the grading of the fine aggregate system toward a finer distribution. Under the nominally constant water-to-cement ratio adopted in this study, this effect becomes especially important because the fresh-state penalty associated with finer grading may reduce the effective water availability and cannot be compensated by additional mixing water. Consequently, the influence of the tailing-derived aggregate was not solely controlled by its replacement percentage, but also by its interaction with the sludge content [28].

The significance of the AB interaction term confirms that the acceptable dosage of one fraction depends on the level of the other. From a physical perspective, this interaction may be related to the simultaneous modification of the granular skeleton and the cementitious matrix. The tailing-derived fine aggregate progressively shifted the fine aggregate system toward a finer grading, which may improve particle packing at controlled replacement levels but also increases the surface area that must be wetted. At the same time, the fine tailing sludge partially replaced cement and introduced a highly dispersed fine fraction, increasing water demand and reducing the amount of hydraulically active binder. Therefore, under the nominally constant water–cement ratio, the effective water availability and binder efficiency may decrease when both replacement levels increase simultaneously. This provides a physical basis for interpreting why the AB interaction was statistically significant and why the compressive-strength response depended on the coordinated balance between aggregate grading, possible packing improvement, cement dilution, and water-demand effects.

Overall, the response surface analysis indicates that the use of gold-mine tailings in structural concrete is mechanically feasible within a bounded incorporation region. The simplified model did not support a purely linear interpretation of the system; instead, compressive strength was governed by the interaction between both replacement mechanisms and the nonlinear effect of sludge incorporation. This interpretation provides a preliminary mechanical basis for defining dosage limits for the valorization of gold-mine tailings in structural concrete.

3.4. Integrated Discussion of Tailings Performance in Structural Concrete

The combined interpretation of the physical, morphological, mineralogical, chemical, and statistical results provides a coherent explanation for the behavior of structural concrete incorporating gold-mine tailings. Rather than acting as equivalent replacement materials, the two evaluated tailing fractions fulfilled differentiated functions within the mixture. The fine tailing sludge primarily influenced the cementitious matrix through its high fineness, porous morphology, and elevated amorphous content, whereas the coarser tailing-derived fraction mainly modified the granular skeleton by altering fine-aggregate grading and particle arrangement. Consequently, the compressive-strength response depended on the balance established between matrix modification and granular stability rather than on the isolated effect of either residue. The SEM observations showed that sample F was composed of highly irregular, porous, and very fine particles, while the XRD and XRF analyses confirmed the predominance of silicoaluminous phases together with a substantial amorphous fraction. These features support the interpretation of sample F as a finely divided mineral fraction capable of enhancing interstitial filling and local matrix refinement at moderate levels of cement replacement [29]. However, the same characteristics also explain the strength reductions observed beyond the favorable region identified by the response surface. Once the amount of ultrafine material exceeds the capacity of the system to maintain adequate coating, dispersion, and hydration efficiency under a constant water-to-cement ratio, the beneficial filler-related effect is progressively offset by increased specific surface area, higher water demand, and cement dilution. The role of the sludge fraction is therefore intrinsically dual: favorable within a controlled replacement range, but detrimental when used excessively.

The contribution of the tailing-derived fine aggregate followed a complementary but distinct mechanism. Because this material behaved more consistently as an alternative fine aggregate, its principal effect lay in the reconfiguration of the granular skeleton [15]. At moderate replacement levels, its incorporation may improve particle accommodation and contribute to a more continuous solid framework. However, as the replacement level increased, the overall grading of the fine aggregate system shifted toward a finer condition, reducing the robustness of the granular structure and increasing the susceptibility of the mixture to workability-related penalties and inefficient water distribution. The reduction in compressive strength near the edges of the experimental domain is therefore better understood as the result of an imbalance between matrix refinement and aggregate efficiency than as the consequence of any single unfavorable variable acting alone.

From a mechanistic perspective, the favorable region defined by the response surface corresponds to a compositional window in which three conditions coexist: improved interstitial filling promoted by the fine sludge, acceptable fresh-state behavior despite the increase in fineness, and sufficient granular stability provided by the tailing-derived aggregate fraction. Within this window, the mixture benefits from enhanced particle distribution without incurring the excessive penalties associated with over-fineness and excessive cement replacement. This explains why the statistical interaction identified in the response surface model is not merely a mathematical result, but a direct expression of the physical interdependence between both residue fractions. In other words, the significance of the interaction term reflects the fact that the performance of the system emerges from coordinated dosage rather than from the independent maximization of each by-product.

This integrated interpretation also clarifies the practical relevance of the optimization strategy adopted in the study. The results indicate that the technical feasibility of incorporating gold-mine tailings into structural concrete depends on defining a controlled compositional range in which the positive effects of filler-assisted densification and granular restructuring remain dominant over the negative effects of excessive fineness, higher water demand, and cement dilution. This is a more meaningful approach than defining admissible replacement levels on the basis of a single residue or a one-variable analysis, since it reflects the actual multicomponent nature of the system [2]. Accordingly, the main contribution of the present work lies not only in showing that mine tailings can be incorporated into concrete, but in demonstrating that their successful use requires coordinated proportioning between fractions with different functional roles and different effects on mixture efficiency.

From a broader engineering perspective, these findings are also relevant in terms of resource efficiency and waste valorization. The proposed approach supports the performance-based reuse of mining residues in cement-based materials, reducing the demand for natural fine aggregates and promoting more efficient use of by-products from mineral processing [30]. At the same time, the results show that valorization should not be interpreted as unrestricted substitution, since the mechanical response remains strongly dependent on dosage balance and system compatibility [8]. In this sense, the study advances the discussion from simple waste incorporation toward optimization-based material valorization, in which the feasibility of reuse is demonstrated through measurable engineering performance rather than assumed on the basis of compositional similarity alone. The favorable response observed at moderate replacement levels results from the temporary predominance of improved particle accommodation and matrix continuity, whereas the decline in strength at high incorporation levels reflects the combined penalties of excessive fineness, increased water demand, and reduction of effective cementitious efficiency. Therefore, the incorporation of gold-mine tailings into structural concrete should be approached as a coordinated design problem, in which mixture performance depends on the controlled balance between physical refinement and material substitution.

4. Conclusions

This study evaluated the combined incorporation of two gold-mine tailing fractions into structural concrete through a Response Surface Methodology approach. The results demonstrated that these residues can be incorporated into structural concrete under controlled dosage conditions, although their effect is not governed by simple linear replacement behavior. Instead, the mechanical response depended on the interaction between a tailing-derived fine aggregate fraction and a fine sludge fraction used as partial cement replacement by mass. The fine tailing sludge primarily modified the cementitious matrix through fineness-related effects. At moderate replacement levels, it contributed to a denser solid arrangement within the binder system; however, it also increased normal consistency and delayed setting times, indicating greater water demand and a progressive reduction in early binder efficiency as cement replacement increased. The tailing-derived fine aggregate, in turn, altered the grading of the fine fraction by shifting the particle-size distribution toward finer conditions as its replacement level increased. This behavior suggests that the aggregate fraction may improve particle accommodation at moderate levels but may reduce mixture efficiency when the grading becomes excessively fine. The morphological, mineralogical, and chemical characterization confirmed that both tailing fractions fulfilled different functional roles within the concrete system. SEM observations showed that the sludge exhibited a highly heterogeneous and porous morphology with very fine particles, whereas the tailing-derived aggregate presented a more stable and compact granular structure. XRD and XRF analyses confirmed that both materials were predominantly silicoaluminous and chemically compatible with incorporation into cement-based materials. Although the presence of a substantial amorphous fraction suggests potential physicochemical contribution, the present study does not provide direct evidence of pozzolanic reactivity or secondary hydrate formation. The response surface model showed that the interaction between tailing-derived aggregate replacement and cement replacement by sludge was the most significant factor affecting 28-day compressive strength. This finding confirms that the performance of the system was controlled by coordinated dosage rather than by the isolated effect of each residue. The most favorable mechanical response, based solely on 28-day compressive strength, was identified at sludge replacement levels below approximately 20% and tailing-derived fine aggregate replacement levels below approximately 90%, where the mixtures maintained satisfactory compressive strength while benefiting from improved particle distribution. At higher replacement levels, compressive strength decreased due to the combined effects of cement dilution, excessive fineness, increased specific surface area, and higher water demand under constant water-to-cement ratio conditions. This study suggests that the reuse of gold-mine tailings in structural concrete may be approached as a multivariable design problem from a preliminary mechanical performance perspective, involving fractions with different physical roles rather than as an isolated substitution strategy. In this sense, the main contribution of the work lies in defining preliminary mechanical incorporation limits based solely on 28-day compressive strength, including tailing-derived fine aggregate replacement below approximately 90%, which should be interpreted as a preliminary mechanical threshold rather than as a practical recommendation for field application. However, durability and environmental safety indicators, such as heavy metal/cyanide leaching, permeability, shrinkage, and chemical resistance, were not evaluated in this study. Therefore, the proposed replacement ranges should be interpreted as preliminary mechanical feasibility limits rather than definitive recommendations for long-term structural use or field-scale application.