Testing Concrete for the Construction of Winemaking Tanks

Featured Application

The present work can find direct application in the process of producing concrete winemaking tanks.

Abstract

This work focuses on the design of concrete for the construction of winemaking tanks, as well as coating behaviour and stability of the systems in wine immersion. More specifically, alternative laboratory concrete mixtures were investigated by replacing cement with natural pozzolan and using silicate aggregates and quartz sand as filler in order to obtain self-compacting concrete of strength class C 20/25. The optimal mixture was selected and further tests were carried out on the mechanical properties of permeability, durability and thermal conductivity. Three coatings and plain concrete were tested for their leachability of heavy metals in wine. The results show that the selected composition with 20% cement replacement by natural pozzolan has the desired workability and strength and is comparable to a reference concrete without natural pozzolan. The leachability tests show that heavy metals do not leach out upon contact with wine, but only calcium and potassium oxide, which can be easily addressed by coating or treating the surface of the concrete. Also, the optimum coating did not influence the pH of the wine.

1. Introduction

The wine fermentation stage is one of the most important stages of the winemaking process, whether it is white, red or reddish (rosé) winemaking. This phase can last from a few months to a few years, resulting in the wine’s taste changing from wild and astringent to soft and velvety in combination with the rich aroma that will have developed. It is usually carried out in stainless steel tanks, oak tanks or oak barrels.

Regarding stainless steel tanks, while being highly durable, the thermal conductivity and impermeable nature of the material necessitate external cooling and micro-oxygenation systems [1]. Wine fermented in stainless steel tanks is characterized by a more astringent character than oak tanks. Regarding oak tanks and barrels, their main advantage is the fact that they allow the inflow of a small amount of oxygen (2.5 mg/L per month), which is necessary for the fermentation process [2]. They also exhibit better behaviour than stainless steel tanks in terms of their thermal conductivity [3]. At the same time, however, they have a limited lifespan (usually around five uses). Winemaking in oak barrels usually produces a wine with a more blunt character, while the wood itself contributes aromas to the final product, something that is usually sought in red wines. Beyond these cases, in recent years the value of winemaking in concrete tanks has begun to be recognized internationally. In recent decades, their production and use have grown rapidly in traditional wine-producing countries (France, the USA), as these tanks seem to combine the advantages of the other two options [3]. Research has been ongoing for both fermentation [4] and maturation [5] processes.

The main advantage of concrete is its great durability over time. Also, the porosity of concrete allows for natural micro-oxygenation inside the tank [1]. At the same time, the nature of the material in combination with the thickness of the tank allows for slow (not abrupt) temperature changes inside, in relation to the environment, ensuring the maturation process [6]. These characteristics, according to the experience so far in wine production in concrete tanks, contribute to the production of wines that combine the characteristics of oak and stainless-steel tanks, i.e., full and balanced body due to the long and balanced maturation, while maintaining all the aromas of the fruit intact, as concrete does not contribute its own aromas.

Another important advantage of concrete is the possibility of constructing tanks of particular shape that aid the winemaking process. These shapes usually refer either to amphora tanks, based on the corresponding clay containers used in antiquity, or to a more modern proposal of an oval-shaped tank, allowing the wine to be in constant motion, giving a smooth and uniform fermentation [4,7].

Taking into account the interests and prospects that exist internationally in this field, the present work concerns the study of concrete for the construction of oval-shaped winemaking tanks. Thus, it was chosen to design test compositions of self-compacting concrete (SCC) of strength class C 20/25, at least. The choice of SCC was made in order to achieve the correct placement of the concrete in the curved geometry of the shape [8] and to achieve the required adhesion with the concrete reinforcement [9]. The design of the test compositions and the individual materials selected for the compositions follow the requirements of the relevant regulations for the manufacture of SCC, such as the Greek National Technical Specification ELOT ΤC 1501-01-01-06-00 [10] and the relevant EFNARC regulation [11].

2. Materials and Methods

The selection of materials and the design of the test compositions were carried out based on the requirements and specificities of the application. In the following subsections, a reference is made to these parameters.

2.1. Raw Materials

The type of cement chosen for the application is Portland cement CEM I 42.5 R (according to EN 197-1) [12], manufactured by TITAN S.A., Athens, Greece, as it does not contain any other additives besides clinker and gypsum, a basic requirement for absolute control of each additional additive that will be selected to be included in the mix. Also, this cement is characterized by high final strength, but also high early strength (R index), which is considered necessary as the tanks will not be cast on site at their point of use.

Part of the cement quantity was decided to be replaced with natural pozzolan from Milos island, Greece, with the aim of improving, on the one hand, the environmental profile of the compositions and, on the other hand, to contribute to the consumption of portlandite. Consequently, these features will improve ultimately the early age properties of the tanks [13]. The following table (

Table 1. Chemical composition of OPC and pozzolans that were used to produce test samples, percentages (%) of metal oxides and loss of ignition (L.O.I.).

Binder	Na2O	K2O	CaO	MgO	Fe2O3	Al2O3	SiO2	SO3	L.O.I.
OPC	<0.01	1.10	61.7	1.20	3.07	3.14	15.1	3.34	4.12
Pozzolan < 20 μm	1.27	2.27	3.08	0.34	1.46	13.78	71.09	<0.01	4.91
Pozzolan < 45 μm	1.47	2.45	0.56	0.27	1.08	12.67	74.89	<0.01	4.61

) gives the chemical composition of the ordinary Portland cement (OPC) and pozzolan (in two different fineness gradations, <20 μm and <45 μm) used for the test compositions. The determination of metal oxides was conducted with X-Ray Fluorescence (XRF), using a S8 Tigger instrument of Bruker, Karlsruhe, Germany. The loss of ignition was determined from room temperature up to 1000 °C in the oven.

The XRF results (Table 1) show that pozzolanic materials conform with ASTM 618-03 [14], according to which SiO₂ + Al₂O₃ + Fe₂O₃ content should be above 70% w/w.

In the context of this research, the use of river siliceous aggregates was chosen for three main reasons:

• Siliceous aggregates, based on experience, exhibit lower water absorption, an important parameter for self-compacting concrete, so that the amount of free water in the mixture is not affected by the aggregates. It is typically reported that the average water absorption of limestone sand is 1.5% [15], while as mentioned below the absorption of the silica sand used in the project was measured at 1.24%.
• Due to their geometry (rounded edges), river aggregates contribute more positively to the fluidity of concrete than crushed quarry aggregates, which are characterized as more angular.
• Their use reduces the calcium content in the concrete, the reduction of which is considered positive for the wine fermentation environment.

2.2. Design of the Wine Tank

The design of the wine tanks was based on a number of objectives, both technical and operational. In terms of functionality, the tanks should have a useful volume of at least 1.2 cubic metres of liquid to ensure sufficient productivity during vinification. At the same time, a certain maximum hydrostatic pressure must be maintained on the walls of the tank over its entire height to ensure the best possible vinification. With this in mind, the internal walls of the tank were designed in an egg-shaped layout, which meets all the necessary requirements while ensuring a smooth flow—internal circulation—that the liquid grape must have during its transformation into wine. This shape is consistent with the good structural behaviour of the tank.

For the best structural capacity of the tanks, their centre of gravity should be low and at the same time, the weight of the tank should be balanced towards the ground. It was therefore decided to support the tanks on three legs, always forming a plane, balancing thus the weight on uneven surfaces. These legs have specific height dimensions to ensure that the structure will not overturn in the event of local seismic activity, while at the same time allowing the liquid discharge pipe to be placed at the bottom of the tank. The interface between the floor and the ground is designed to prevent the floor from developing higher than allowable stresses that could lead to failure of the floor. At the same time, the feet also ensure operational objectives as they can be used to tie down the tanks for transport.

The self-weight of the tank had to be as low as possible so that it could be easily transported. At the same time, the tank had to accommodate two manholes of about half a meter in diameter, one on the side walls at a sufficient height and one at the top. The walls would therefore have to meet all the above objectives, but be thin enough to ensure low weight, accommodate the necessary reinforcements with appropriate coatings, and the tank would have to perform elastically under all the necessary load combinations so that the tank would not crack and leak liquid.

To ensure the structural integrity of the tanks, an analysis was carried out using the finite element method. This analysis was performed using the Abaqus v2022 finite element program. Concrete was simulated using C3D8R 8-node linear brick elements with reduced integration and hourglass control, with a mesh size of 40 mm. Reinforcement was simulated using truss elements. The concrete damage plasticity model (CDP) was used to simulate the concrete, with mechanical properties defined for the C20/25 class. The reinforcing steel was simulated using a plasticity model with certain steel grade B500C values. Three load combinations were tested to fully represent the operational, transport and roll-over-drop conditions of the tank

2.3. Composition of Concrete Specimens

All the above-mentioned parameters regarding the geometry of the tanks in combination with the existence of the required reinforcement impose the use of aggregates with a maximum grain size of 16 mm. After carrying out test granulometric compositions and preparing test compositions of SCC, the final composition of the aggregates was decided, the granulometric curve of which is given in Figure 1. To improve the workability of the concrete, quartz sand of gradation 0–0.3 mm was added as filler. Quartz sand, in comparison to other aggregate fillers (e.g., limestone), exhibits reduced water absorption, which is necessary for the SCC mixture so that the amount of free water in the mixture would not be affected by fine aggregates.

As for the chemical admixtures, a third-generation superplasticizer was used, based on polycarboxylate polymer technology, as well as a viscosity modifier, based on a modified starch complex. The concrete mix design was carried out through successive optimization of each of the parameters that were necessary for the desired final mix. These parameters are summarized in the following points:

• Total amount of binder;
• Water/Binder ratio;
• Cement replacement percentage and fineness of pozzolan;
• Amount of filler (kg/m³) in the mixture;
• Amount of superplasticizer (% w/w of fines) in the mixture;
• Amount of viscosity modifier (% w/w of fines) in the mixture.

The main criterion for the study of the compositions is the properties of the fresh concrete, which were measured after each composition in order to decide the next steps. Two of these tests are standardized, while all of them are described by the corresponding National Technical Specification [10] and EFNARC [11].

Slump-flow test: The slump-flow test is used to assess the free horizontal flow of fresh concrete without obstruction and is based on the corresponding concrete slump test (EN 12350-2 [16]). Along with the diameter of mixture spreading on a flat surface (recommended values ≥ 650 ± 50 mm) the time required for the diameter of the concrete circle to reach 500 mm (T500) is also measured. Comparatively short times indicate high flowability.

L-box test: The test is used to assess the flowability of the SCC, as well as the degree to which the SCC clogs the reinforcement. The test is also described in the EN 12350-10 [17] regulation. The device is essentially an L-shaped duct of rectangular cross-section, where the vertical section is separated from the horizontal section by a removable door in front of which vertical reinforcement bars are placed. The vertical box is filled with SCC and then the door is lifted, allowing the mixture to flow into the horizontal section. When the flow is complete, the ratio H2/H1 of the surface level of the mixture in the horizontal and vertical sections is calculated. The closer this ratio is to the value 1 (ideal water flow condition), the better the flowability of the concrete.

Segregation resistance test: The test is used to assess the resistance to segregation (stability) of the SCC, by pouring a quantity of the mixture (10 L) through a sieve with a 5 mm opening. The result is expressed as a percentage of the mass passing to the initial mass of the material placed on the sieve. For values above 15% and especially above 30% there is a significant possibility of segregation of the SCC.

A decisive factor for the development of the test compositions is also the mechanical behaviour of concrete. For this reason, 15 cm cube specimens were produced, following the requirements of EN 12390-3 [18], for the determination of compressive strength.

Table 2. Preliminary compositions and parameters investigated.

Acronym	Binder Quantity (kg/m3)	Water/Binder Ratio	Cement Replacement Rate (%)	Pozzolan Fineness (μm)	Filler Quantity (kg/m3)	Superplasticizer (% w/w of cem)	Viscosity Modifier (% w/w of fines)	Flowability-Time for 50 cm Spread (s)	Final Spread Diameter (cm)	L-Box (L2/L1)	Segregation (%)	Average 7-Day Compression Strength (MPa)	Average 28-Day Compression Strength (MPa)
C1	500	0.43	0	-	80	1.9	0.30	2.5	61	1.00	1	42.3	49.9
CP1	500	0.43	30	20	80	2.0	0.35	2.7	56	0.55	0	36.0	44.0
C2	460	0.48	0	-	120	1.8	0.40	5.9	65	0.85	7	36.7	43.9
C3	460	0.45	0	-	120	1.8	0.40	2.7	62	0.80	4	40.3	45.8
CP2	460	0.45	30	20	120	2.0	0.40	1.5	56	0.55	1	33.1	42.0
CP3	460	0.45	30	20	160	2.0	0.35	-	43	-	0	31.8	40.1
CP4	460	0.45	30	40	160	2.0	0.20	2.0	61	0.64	4	28.8	35.6
CP5	460	0.45	30	40	160	2.0	0	6.3	60	0.70	1	32.3	42.6
CP6	460	0.45	25	20	120	2.0	0.40	-	46	0.36	0	36.9	42.7
CP7	460	0.45	25	40	120	2.0	0.40	4.3	52	0.55	0	32.9	39.4
CP8	460	0.45	20	20	150	2.0	0.40	3.3	53	0.56	0	35.8	42.5
CP9	460	0.45	20	40	150	2.0	0.20	4.3	55	0.70	0	37.5	46.7
CP10	460	0.45	20	40	150	1.8	0.10	2.7	69	0.85	2	35.3	42.7
CP11	460	0.45	20	40	150	1.5	0	3.5	62	0.64	0	32.3	41.2
CP12	460	0.45	20	40	150	1.8	0	2.7	67	0.85	3	41.5	47.2
CP13	460	0.45	20	40	150	2.0	0	3.2	-	0.80	3	39.2	48.9
CP14	460	0.42	20	40	150	2.0	0	3.7	70	0.80	5	37.2	48.5

summarizes the preliminary compositions carried out, with all of the aforementioned parameters that were investigated, showing how the optimization of the mixture evolved.

All the studied mixtures have shown satisfactory strength and segregation resistance. Workability-flowability performance (according to above combined flowability and L-box test results) of mixtures CP10, CP12 and CP14 seem to be optimal of all mixtures containing pozzolan as cement replacement. Among these three mixtures, CP12 and CP14 gave indications of a tendency to segregation, after visual inspection during production. Mixture CP10 contains a small percentage of viscosity modifier (0.1% w/w of fines), which resulted in a more coherent mixture, without any visible segregation tendency. Taking into account all the above factors (strength, flowability, coherency, environmental profile), mixture CP10 was chosen as optimal for the construction of the tanks.

Along with the CP10 mixture that contains cement and pozzolan as binders, C3 mixture was selected for the study as the optimal reference composition (with comparable strength and flowability performance) with cement as the sole binder in the mixture. The constituents of the two mixtures are shown in

Table 3. The optimal composition of C3 and CP10 mixtures.

Materials	Specific Gravity	C3 (kg/m3)	CP10 (kg/m3)
Cement I 42.5	3.14	460	368
Pozzolan, natural (<45 μm)	2.39	-	92
Water	1.00	207	207
Fine aggregate, siliceous (0–4 mm)	2.68	760.70	749.8
Coarse aggregate, siliceous (4–8 mm)	2.68	456.40	449.9
Coarse aggregate, siliceous (8–16 mm)	2.68	304.30	299.9
Fine quartz 0–0.3 mm (filler)	2.71	120	150
Superplasticizer	-	8.28	8.28
Viscosity modifier	-	2.35	0.62

.

2.4. Testing of Concrete Specimens

For these two mixtures, further testing of physical and mechanical properties, durability tests, as well as leachability tests were carried out. More specifically, the flexural and splitting tensile strengths were determined according to EN 12390-5 [19] and EN 12390-6 [20], respectively. The open porosity was measured according to EN 12390-7 [21] and RILEM CPC11.3 [22] (in water under vacuum) regulations and the static modulus of elasticity was according to BS 1881-121 [23].

Durability testing after freeze-thaw cycles is of particular importance for tanks as they may be placed in low-temperature areas for wine preservation after being exposed to higher ambient temperatures, with a potential risk of deterioration or cracking. After subjecting the specimens to freeze-thaw cycles, tests are carried out on mass loss (weighing), concrete structure (by determining the dynamic modulus of elasticity) and finally on the compressive strength of the specimen. All these properties are compared with the corresponding properties of the specimens before they were subjected to cycles, in order to determine the potential deterioration of concrete. The test follows the ASTM C666 [24] standard. Each freeze-thaw cycle lasts 4 h and covers the temperature range of −18 °C to 4 °C.

One of the main benefits of concrete wine tanks is the increased thermal resistance of concrete (and therefore low thermal conductivity), which leads to more subtle temperature changes inside the tank, despite greater external temperature changes, as compared to, for example, metal tanks. Thermal conductivity testing was carried out on concrete samples measuring 200 × 200 × 25 mm (tiles). The apparatus used for the tests was a THERMTEST MODEL HFM 100 thermal conductivity measuring device for building materials (λ), which follows the EN 12667 [25] standard. The calculation of thermal conductivity was carried out using the method of permanent heat flow “Heat Flow Meter Technique” (heat flow measurement method). The instrument provides results at two different temperatures 10 °C and 20 °C, and since these are possible working temperatures, both records are given in the Results section.

The permeability of concrete to water vapor was also measured as an important parameter of the material of a winemaking-maturation tank, for the micro-oxygenation of wine. The test was carried out following EN 1015-19 [26]. The tiles were maintained for 2 days in a chamber with constant temperature and humidity conditions (20 °C and 95% RH) and then for 26 days in a chamber with 21 °C and 50% relative humidity, according to the standard, with the aim of stabilizing their hygroscopic properties.

2.5. Leaching Test

The two selected compositions were part-immersed in white and red wine for 30 days with the aim of testing the leachability of metals from the concrete into the wine and the stability of the concrete structure under the influence of wine, the latter being an acidic solution. The concrete was characterized before and after part-immersion. As an alternative for the construction of the interiors of the tanks, three coatings were also tested for leachability in wine. Two of them were commercially available epoxy resin coatings for edible goods’ tanks, coded D and S. Another coating, coded A, was produced by mixing waterglass with soil from the area of the winery, to introduce aspects of the locality (terroir) of the crop to the wine through the maturation process in the tank.

The leaching test consisted of the following steps:

• Large containers (60 × 30 × 30) cm were filled with red and white wine.
• The pH value of the wines and the pH value of the concrete samples were determined prior to immersion.
• The pH value of the wines and the pH value of the concrete samples were determined prior to immersion.
• Coated (three different coatings) and uncoated concrete specimens (150 × 150 × 150) mm, at 28 days of age, were partially immersed in red and white wine.
• After 30 days, 50 mL of wine quantity was removed from every container and the pH value of the wine was determined.
• To continue, 1 mL of 1 M HNO₃ acid was added to the vials to preserve the wines until the determination of total metals. The samples were heated in a sand bath at 95 °C for 1 h. At the end of this period, 5 mL of concentrated nitric acid was added again and left for a further hour at the same temperature. The pre-treated sample is then removed from the sand bath and allowed to come to room temperature. The procedure is completed by adding it to a 100 mL volumetric flask and filling it up with deionized water. The pre-treated samples were filtered with a vacuum pump before the determination of the metals.
• Total metal oxides determination was conducted with flame atomic absorption spectroscopy (FAAS) for metals reported at the mg/L level and graphite (GFAAS) for metals reported at the μg/L level. The determination of mercury was carried out with a cold vapor atomic absorption apparatus (CVAAS).

2.6. Evaluation of Chemical Properties Before and After Leaching Test

The crystalline phases of the plain concrete samples were determined by X-ray diffraction analysis. Diffractograms were obtained using a Bruker D2 Phaser diffractometer (Bruker, Billerica, MA, USA) with a 30 kV/10 mA X-ray generator, Cu(Ka) radiation, step size 0.02 from theta 5° to 70°. The samples were dried at 80 °C and were then ground. The powdered samples were sieved and the fine powder (<75 μm fraction) was collected for the measurement.

Attenuated Total Reflectance (ATR) contributed to the identification of the IR fingerprint of the concrete samples, before and after immersion in wine. A Cary 630 FTIR instrument (Cary, Addison, IL, USA) with ATR probe of Agilent Technologies (Santa Clara, CA, USA) was used for the determination of absorption patterns from 375 cm⁻¹ to 4000 cm⁻¹ wavenumbers with a spectral resolution of 2 cm⁻¹ and 32 scans per measurement. The background of all spectra was collected on the ambient atmosphere before each measurement. The sampling was made from the surface of the specimens and the samples were ground to produce powder.

Thermograms were obtained with a Netzsch STA 449 F5 Jupiter Simultaneous TG-DSC instrument (Netzsch, Selb, Germany), in an N₂ atmosphere (50 mL per minute) from 50 °C to 1000 °C and 20 °C per minute step, at 28 days. The calculated calcium hydroxide and carbonated species of the pastes were made by utilizing the mass loss between 350 °C and 550 °C and 600 °C and 800 °C, respectively.

3. Results and Discussion

3.1. Numerical Modeling and Optimization

The tank responds elastically in all three loading cases. This is important because, if the concrete were in the plastic stage, microcracks would form and liquid would escape. The next picture shows the worst results of all the finite element analysis combinations, as well as the low stress concentration of the reinforcement, which indicates the structure’s elastic response. Using finite element and parametric analyses, the amount of concrete required has been reduced, thus reducing the weight and making construction easier by using a reinforcement grid.

Following parametric finite element analysis, it was decided that the specific wall thickness should be in the range of 8–15 cm, with the greatest thickness at the lowest points of the structure. Figure 2 shows a visualization of the results of the finite element analysis.

3.2. Mechanical and Physical Properties

As shown in

Table 4. Results of physical and mechanical properties.

Property	C3	CP10
Open porosity (%)	6.79	10.55
7-day compressive strength (MPa)	40.3	35.3
28-day compressive strength (MPa)	45.8	42.7
Static modulus of elasticity (GPa)	37.2	28.0
Flexural strength (MPa)	7.97	7.81
Splitting tensile strength (MPa)	1.94	2.33

, a small increase in open porosity was observed in CP10 composition, which is a common feature in concretes/mortars that contain a quantity of pozzolan. This is connected to the lower specific gravity of the pozzolan compared to cement powder [27]. Regarding compressive strength, it seems in principle that there was no significant difference between the two compositions. In C3 composition, it seems that the development of strength was faster in the first 7 days compared to CP10, which is also expected, as the presence of pozzolan leads to the development of strength at a slower rate. In addition, the aim of pozzolan addition was the capturing of portlandite into the structure, rather than contributing to excessive compression. Overall, the strength level achieved was considerably higher than the required for C20/25 concrete, which means that there was room for cement content reduction in the concrete mixture. However, the authors decided to keep the cement content high and have a higher safety margin for the actual production of the concrete. After some real scale trial mixtures, cement content could be reduced to improve the environmental profile of the concrete tanks. Additionally, the static modulus of elasticity for both mixtures was in the expected range of values for concrete, with a small decrease observed for CP10, in agreement to other studies in which the dynamic modulus is slightly reduced due to pozzolan addition [28]. For tensile strength (flexural and splitting) there was no significant difference after the addition of pozzolan, on the contrary, a small increase in splitting tensile strength was observed. Table 4 shows the mechanical properties, as well as the open porosity of the two studied mixtures.

Table 5. Results of experimental durability tests before and after the freeze-thaw cycling.

Composition	Weight (g)		Dynamic Elastic Modulus (GPa)		Compressive Strength (MPa)
	Before	After	Before	After	Before	After
C3	7.98	7.98	46.00	44.09	45.8	46.00
CP10	7.81	7.81	44.75	44.47	42.7	46.25

gives a comparative analysis of the results for compositions C3 and CP10, before and after the freeze-thaw cycles. The results from the thermal conductivity test on two tiles of each composition are given in

Table 6. Thermal conductivity coefficient results.

Composition	λ [W/(m × K)]		R [(m2 × K)/W]
	10 °C	20 °C	10 °C	20 °C
C3	1.2847	1.3753	0.0188	0.0173
CP10	1.3972	1.5010	0.0211	0.0198

.

Table 7. Calcium hydroxide content, calcium carbonate and pH value in concrete samples of compositions C3 and CP10, at 28 days of age.

Composition	Ca(OH)2 (% w/w)	CaCO3 (% w/w)	pH
C3	9.24	8.22	13.08
CP10	4.85	8.22	12.96

presents the results of the water vapor permeability resistance coefficient for the two compositions C3 and CP 10.

Regarding the durability test (Table 5), no significant modification is observed in the weight change, structure and strength of the concrete after freeze-thaw cycles.

The same similar behaviour between the two compositions is also observed in the thermal conductivity/resistance test (Table 6). For both compositions, the thermal conductivity of the concrete is relatively low, compared for example to the standard values of the Greek Regulation on the Energy Performance of Buildings, for lightly reinforced concrete (1.65–2.0 W/(m × K)) [29], therefore the composition is considered successful in terms of thermal behaviour, for the construction of the tanks. The water vapor permeability test showed an increased value of the resistance coefficient for the composition with pozzolan (55.8 m) compared to the reference composition (31.8 m). Despite the greater open porosity of the CP10 composition, the lower permeability could be attributed to the pore distribution modifications, as the addition of pozzolanic materials refines the pore structure [30]. Both compositions had relatively low water vapor permeability compared to the literature [15].

3.3. Chemical Properties

The results of the determination of the calcium hydroxide (Ca(OH)₂) and calcium carbonate (CaCO₃) contents are given in Table 7, as a percentage by weight (% w/w). The pH value of the concrete sample (measured on a dissolved in water sample with a glass electrode) is also given in Table 7. The measurements were conducted before the immersion of the specimens in wine. TG results show that the addition of pozzolan reduced the calcium hydroxide content (portlandite), while it bound calcium into calcium-silicate formations [29,31]. Consequently, the addition of pozzolan succeeded in reducing portlandite content before 28 days and, thus, potentially reduces the calcium leaching or the interaction between portlandite and wine. The pH values of the samples were similar at the age of 28 days, showing that pozzolan addition did not modify the alkalinity of cement.

The concrete specimens were removed from wine 30 days after part-immersion and the compressive strength of the specimens, the pH value and the calcium hydroxide content were determined. Additionally, ATR spectra were obtained from the surface material of the specimens to identify the clarity of the concrete structure. In Figure 3, the IR assignments of surface samples of immersed concrete specimens in red and white wine are displayed.

The spectra show that wine did not interfere with the concrete structure, regardless of the presence of pozzolan in the concrete composition, or the type of wine that the specimens were immersed in. The concrete compositions’ carbonates (1412 cm⁻¹, 871 cm⁻¹, 776 cm⁻¹) and the CSH compounds (963 cm⁻¹) did not present any modifications. The CSH peak intensity of CP10 and C3 samples in red wine was slightly lower compared to white wine (CP10L and C3L in Figure 3), although the difference is insignificant.

The decrease in pH values after the immersion in wine (either white or red wine) is probably due to the wine quantity absorbed and/or remaining on the surface of the concrete (

Table 8. Values of pH and calcium hydroxide content on the surface of concrete samples before and after immersion in wine.

Composition	Immersion in Wine	pH	Ca(OH)2 % w/w
C3	-	13.08	9.24
CP10	-	12.96	4.85
C3_K	Red	11.76	11.71
CP10_K	Red	9.80	8.30
C3_L	White	10.92	11.63
CP10_L	White	9.14	7.79

). It is also shown that the pH of CP10 is less resistant in wine immersion. The portlandite content of CP10 series was lower, compared to C3 series, due to the consumption of Ca(OH)₂ by the pozzolanic reaction. In

Table 9. Determination of metals and pH value of white wine, after 30 days immersion of concrete in white wine.

	Units	White Wine	Limits 1	DL_C3	DL_CP10	SL_C3	SL_CP10	AL_C3	AL_CP10	RL_C3	RL_CP10
pH	-	4.01		3.91	3.92	6.6	6.58	5.73	6.41	6.32	6.31
Ca	mg/L	21.4		16.9	18.3	124.9	149.8	111	79.1	169.2	161.5
Cr	mg/L	0.31		0.36	0.25	0.5	0.46	0.49	0.41	0.46	0.46
Cu	mg/L	0.18	<1.0	0.25	0.23	0.56	0.55	0.42	0.4	0.45	0.43
Fe	mg/L	1.9		2.22	2.6	10.6	29.2	17.36	14.84	15.7	18.6
K	mg/L	314.4	<100	308.6	302.2	367.2	370	341.4	369	362.4	353.6
Mg	mg/L	115.2		114.2	111	142.8	144.4	139.8	140.6	142.8	141.6
Mn	mg/L	1.81		1.82	1.85	2.05	2.24	3.62	3.83	2	1.96
Na	mg/L	11.2		14	12.2	173.8	175.6	140.8	392	99.2	105.4
Zn	mg/L	0.93	<50	0.82	0.89	0.72	0.75	0.8	0.62	0.71	0.77
As	μg/L	12.2		0.2	ND 2	32.4	25.2	25.6	19.6	28.4	30.6
Cd	μg/L	0.6		0.4	ND	ND	ND	ND	ND	ND	ND
Co	μg/L	3	<1000	3	3	28.4	29.8	15.8	42.2	22.6	19.2
Hg	μg/L	ND		ND	ND	ND	ND	ND	ND	ND	ND
Ni	μg/L	38.4	<100	43	26.6	236.8	256	189.2	199.2	193.6	201.6
Pb	μg/L	ND	<5000	ND	ND	ND	ND	ND	ND	ND	ND
Se	μg/L	0.4		10	4.4	5.8	4.6	4.4	5.8	4	5.2

¹ As proposed by IOC OENO 18/2003 [32]. ² ND: not detected.

, the Ca(OH)₂ % w/w content follows this trend. Lower pH values and lower portlandite contents are attributed to the same influence factor as with the coefficient of resistance in water vapor permeability results, which is the interaction of portlandite with pozzolan. Furthermore, portlandite content increased after the immersion of the concrete cubes in wine. As the concrete cubes remained in wine for 30 days the cement hydration evolution led to the formation of additional portlandite. The latter is a piece of evidence that immersion in wine did not hinder the cement hydration in time. Accordingly, the decrease in pH value cannot be connected to the deterioration of the concrete samples but could only be connected to the decrease of calcium hydroxide content.

The mineralogical composition of the samples before and after immersion did not present significant differences, which supports the stability of the C3 and CP10 in wine (Figure 4).

The identified crystalline phases were common to all samples. The dominant mineralogical phase of the samples was quartz, originating from silica sand. Crystalline phases are found in lower proportions: albite, albite, microcline, phlogopite, portlandite and calcite. Albite, anorthite and orthoclase are sodium and potassium aluminosilicate phases due to the presence of silica sand. Portlandite found before immersion was derived from the hydration of clinker. Chlorite and magnesium silicate peaks in the C3_K sample are impurities of the silica sand. Table 9 and

Table 10. Determination of metals and pH value of red wine, after 30 days immersion of concrete in red wine.

	Units	White Wine	Limits 1	DK_C3	DK_CP10	SK_C3	SK_CP10	AK_C3	AK_CP10	RK_C3	RK_CP10
pH	-	4.01		4.3	4.33	6.6	6.51	6.32	7.09	6.35	6.3
Ca	mg/L	21.4		22.3	23.2	85.6	81.2	83	59.3	87.7	89.3
Cr	mg/L	0.31		0.31	0.36	0.40	0.40	0.36	0.34	0.39	0.4
Cu	mg/L	0.18	<1.0	0.14	0.14	0.42	0.34	0.33	0.51	0.38	0.35
Fe	mg/L	1.9		9.0	8.68	7.4	7.38	7.74	8.37	5.54	6.39
K	mg/L	314.4	<100	688	718.4	754.8	744	736	800	754	772
Mg	mg/L	115.2		119.4	117.8	144.2	145.4	143	135.8	146.8	145.8
Mn	mg/L	1.81		1.62	1.79	1.46	1.54	3.56	2.7	1.56	1.52
Na	mg/L	11.2		11.3	10	176.2	147	174.8	648	88	75.4
Zn	mg/L	0.93	<50	0.89	1	0.86	0.68	0.65	0.9	0.81	0.69
As	μg/L	12.2		2.4	2.8	26.4	27.8	19.6	19.8	24.8	24.4
Cd	μg/L	0.6		ND 2	ND	ND	ND	ND	ND	ND	ND
Co	μg/L	3	<1000	5.2	6.6	15.8	22.4	26.6	58.8	16.4	15.6
Hg	μg/L	ND		ND	ND	ND	ND	ND	ND	ND	ND
Ni	μg/L	38.4	<100	55.2	77	216.8	190	127.6	243.2	162.8	153.6
Pb	μg/L	ND	<5000	ND	ND	ND	ND	ND	ND	ND	ND
Se	μg/L	0.4		7.2	7.4	8.6	8.8	8	6.4	7	8.2

¹ As proposed by IOC OENO 18/2003 [32]. ² ND: not detected.

display the concentration of metals in mg/L or μg/L in the samples of white and red wine, after concrete immersion for 30 days. The determination of metals in the wine revealed the presence and the absence of specific metals in mg/L and μg/L. Mercury and lead were completely absent. Selenium and cadmium, along with chromium and copper, were found in very small amounts in white wine. Potassium and magnesium were found in significant amounts, compared to the other metals. The values of the plain white wine were considered as the values to compare with the affected wine, after the immersion of coating of coated concrete specimens and uncoated concrete.

The three tested coatings and coating-free samples behaved differently in the wine environment, within 30 days of influence. Coating D presented the optimum results, compared to the other systems. More precisely, the pH values and the content of the metals in the white wine after the immersion of concrete with D coating remained approximately stable. Wine is an acidic solution for concrete. As a result, this coating protected concrete from wine influence, as well as it protected the white wine from concrete leaching. Coatings A and S presented similar behaviour. The metals that had the greatest leaching from coating to white wine were calcium, sodium and nickel. Ferron, arsenite and cobalt were found to increase after immersion and very small modifications of the concentration were detected for chromium, copper, potassium, magnesium and manganese.

According to Cortiella et al. [4], wine acidity is a characteristic that is affected by the type of vessel or the container of the fermentation. In the present study, the pH value of wines is modified and seems to vary depending on the different coatings of the concrete tank, as depicted in Table 9 and Table 10. Coating D offered the least modification of wine pH and the least leaching of metals in the wine bulk.

The influence of red wine on coated and plain concrete was slightly different compared to white wine. The leaching of potassium in red wine was found to be significant. Potassium concentration after the immersion of concrete samples was found to be more than two times greater. Also, sodium was leached to a higher degree in S and A coatings, compared to white wine influence.

The increase in Ca and K concentration in wine indicates the leachability of calcium hydroxide and potassium oxide. At the same time, this phenomenon leads to an increase in the pH of the wine due to the dilution of hydroxyl ions (OH⁻), for both the white and red wine effect on the samples, except for coating D. In contrast, the decrease in magnesium and manganese concentration in the wine following concrete immersion indicates the potential absorption of these metals in the pores of the concrete. The metals identified in trace amounts, at concentrations of the order of magnitude of μg/L, do not show significant variations, except for chromium (Cr) and nickel (Ni), which are absent from the wine and were found in the wine after contact with the concrete for 30 days. Nickel concentration meets the standards of water intended for human consumption, following the European Directive 2020/2184 [33]. According to the European Regulation 2023/915 [34], for maximum permitted levels for food contaminants, the metals of interest are lead (Pb), cadmium (Cd), mercury (Hg), tin (Sn) and arsenic (As). After the leachability test, Pb, Cd and Hg concentrations were below the limit of detection in μg/L, against the regulation that limits to 0.02 to 0.05 mg/kg. Additionally, the measurements meet the standards of the International Oenological Codex (IOC, Table 9 and Table 10), below 1 mg/kg for Cr and below 5 mg/kg for Ni.

The main interest of the leaching test of concrete is focused on the calcium leaching [35], coming from the dissolution of portlandite [36] because this component is essential for the micro-structure, the strength development and the durability of the concrete. In the present study, the total calcium hydroxide content (Table 9) is much greater than the leached quantity in mg/L found in Table 10 for calcium. R series (plain uncoated concrete) presented a lower leachability in red wine than in white wine. Adding to that, the presence of pozzolan did not influence the leachability of calcium and potassium, as both CP10 and C3 uncoated (R) had similar leaching amounts of these cations. Consequently, the present study cannot find correlations between pozzolan addition and calcium leaching reduction. The latter might be connected to the lower open porosity and greater density of C3 concrete, compared to CP10. But also, it should be taken into account that the utilization of pozzolan in the concrete mixture contributed to the properties before the leaching test, as the calcium hydroxide was reduced (Table 7) prior to durability tests.

The modification of pH value of wine is attributed to the dissolution of hydroxyl anions into the wine, as previously explained by Kamali et al. [37]. Then, the pH of the wine increases. This phenomenon explains the modification of pH value on the uncoated samples (Table 8). Three of the four tested coatings presented similar behaviour, as their utilization did not prevent the pH decrease. Epoxy-based coating (D) eliminated the ion exchange and the dissolution of metals, as well as hydroxylic anions in the wine.

The present effort focused on studying and comparing the mechanical properties of concrete compositions, as well as the durability and the influence of different coatings, to clarify and propose the optimum options for their use in concrete tanks for wine fermentation. The water to binder ratio of the concrete matrix, the addition of pozzolan and the acidity of the solution are key factors for the leaching behaviour of the concrete, or the coated concrete [38]. Consequently, the selection of a durable and resistant coating and the stability of this dual system play an important role in the performance of the winemaking tank. The tested coatings had a different impact on the leaching of metals, but also the composition of the wine seems to have a different impact on the leaching test results. According to the results, coating D can effectively protect concrete from the acidity of wine and at the same time can effectively protect the wine from the leaching of metals in concrete.

4. Conclusions

The design of self-compacting concrete for the construction of winemaking tanks led to the selection of the optimum mixture that showed sufficient workability, no segregation tendency, and the desired mechanical properties. Due to the fact that this composition contained a percentage of natural pozzolan in replacement of cement to improve the environmental profile of the tank, a series of additional tests were carried out to determine whether the addition of pozzolan affects—and to what extent—the physical, mechanical and chemical properties of the concrete. For this purpose, the optimum mixture was compared with a corresponding reference mixture which had exclusively cement as a binder.

The study revealed the following:

• It is possible to manufacture SCC using 20% natural pozzolan as cement replacement and quartz as filler, in order to properly manufacture an oval-shaped tank with the required reinforcements.
• The use of natural pozzolan does not affect negatively the physical and mechanical characteristics of concrete, as well as its resistance to freeze-thaw cycles.
• The required strength class (C20/25) and the desired impermeability to water vapor were achieved, while the thermal characteristics of the concrete were considered satisfactory for the proposed use.
• The leachability test did not show the leaching of significant amounts of heavy metals from concrete into wine, after 30 days of partial immersion. Additionally, calcium and potassium leaching from plain concrete to wine were below the acceptable limits for concrete stability and wine preservation. After the present approach, it would be valuable to test the leachability of calcium and potassium over time, in a context of ongoing research. Furthermore, over time records will validate the present results and the suitability of the tank composition and the coatings effectiveness.
• The acidity of the wine was influenced by the treatment of the concrete surface and specifically, the composition of the coating.
• The use of pozzolan and silicate aggregates reduced the amount of calcium that can leach into the wine, but surface treatment or the use of an appropriate coating is recommended to further reduce this amount.