Experimental Analysis of Creep and Shrinkage of Self-Compacting Concrete with Recycled Concrete Aggregates

Featured Application

The loading of specimens during the creep deformation test was performed at an age of 2 days in order to simulate the process of introducing the prestressing force into precast, prestressed concrete elements. The results presented can serve as input for the calibration and verification of existing models for predicting the creep behaviour of concrete with recycled concrete aggregates.

Abstract

The recycling of old concrete from the demolition of concrete structures is necessary for the rational use of natural aggregate resources. Recycled concrete aggregates (RCAs) are the highest quality recycled aggregates as they are the closest to natural aggregates. However, the use of RCAs is always associated with greater fluctuations and usually with a deterioration in workability, mechanical properties and long-term properties. The use of RCA in self-compacting concrete (SCC), where the proportion of aggregate is lower than in conventional concrete, is one way of mitigating the effects of RCAs. In this paper, the effects of coarse and fine RCA are investigated, focusing on dimensional changes due to shrinkage and creep. SCC mixes were developed in which the dolomite aggregates were partially or completely replaced by RCAs and additionally mixes in which 50% of the cement was replaced by fly ash. The average shrinkage strain measured after 180 days increased from 0.34 mm/m for a mix with natural aggregates to 1.04 mm/m for a mix made entirely with RCAs, showing an almost proportional increase in strain with RCA content. At the same age, the creep compliance ranged from 0.07 GPa⁻¹ for the mix with natural aggregates to 0.34 GPa⁻¹ for the mix made entirely with RCAs, and is most strongly correlated with hardened concrete density.

1. Introduction

Nowadays, the construction sector is confronted with two specific circumstances that influence its sustainability and will become increasingly important in the future. On the one hand, intensive construction activity requires larger quantities of concrete, which leads to an increased consumption of natural resources and energy. It is estimated that 14 billion m³ of concrete was produced worldwide in 2020 [1]. In terms of natural resources, there is a particular focus on aggregates for concrete, the increasing use of which is leading to environmental degradation. Sand, gravel and crushed stone for construction alone account for almost 24% of the global extraction of non-metallic materials, and volumes are expected to double by 2060, reaching around 60 Gt [2]. In 2019, a total of around 3 Gt of aggregates was produced in EU countries, including UK and EFTA countries, of which 9.3% were recycled aggregates [3]. The production of concrete consumed around 5.5% of total primary energy demand in 2015 and was responsible for 8.5% of global greenhouse gas emissions [2]. On the other hand, the high intensity of construction means that more and more construction waste is being generated. For this reason, the recycling and reuse of construction waste as a new building product is becoming increasingly important in many industrialised countries around the world. Data show that around 30 to 50% of construction waste consists of concrete [4,5]. It is estimated that 50 million tonnes of concrete waste are generated in China every year [6]. In the EU, concrete accounted for the largest share of construction and demolition waste (CDW) in 2020, estimated at around 56% or 74 million tonnes [7].

One of the sustainable ways to reuse concrete waste is to produce recycled concrete aggregate (RCA) and use it as a substitute for part of the aggregate in new concrete. In this way, several benefits are achieved, such as reducing the consumption of natural aggregates, reducing the amount of construction waste, and reducing the cost of transportation, energy consumption and overall CO₂ emissions [8,9,10]. Today, however, the majority of concrete waste is downcycled, by being used mainly for road construction and backfilling [7], or disposed of in landfills [6]. Whether this practice will change depends on national regulations.

Numerous studies on RCA show that its quality can vary greatly, but also that its properties always fall short of those of natural aggregates. The density of RCA is up to 10% lower than that of natural aggregates, the water absorption of recycled aggregates is several times higher, the crushing coefficient according to the Los Angeles method is lower etc. [11,12,13]. In addition, recycled aggregates often contain microcracks due to the crushing process and are considerably more porous [14]. The results of experimental studies on concrete with RCA have shown that the incorporation of RCA into concrete reduces workability, reduces strength and stiffness, and increases porosity. The main reason for this is the poorer properties of the recycled aggregate due to the presence of old cement paste, which leads to higher absorption and the formation of several different interfaces (contact zones) in the concrete [15,16,17].

Since the matrix of self-compacting concrete (SCC) is adjusted to have better workability properties than normal concrete, the idea was brought up that the SCC paste matrix would serve as a better medium for RCA [18]. Compared to normal concrete, SCC contains more fine particles originating from cement, finest aggregate fraction and fillers, which affect the inter-particle forces in the fresh state and result in a refined microstructure and pore network in hardened concrete. Much research, particularly in the last twenty years, has focused on determining the effects of partial or complete replacement of natural coarse aggregate with RCA. In general, the addition of RCA decreases the flowability, reduces the mechanical properties and increases the permeability properties [18,19,20,21,22]. It has been shown that the flow properties of SCC can be maintained even if all natural aggregates are replaced by RCA [23,24,25,26]. The research conducted has shown that RCA–SCC can be used for structural reinforced concrete and that the incorporation of RCA into SCC is ecologically and economically feasible [9,27,28,29].

When designing structural concrete elements, the long-term structural deformation due to shrinkage and creep must be considered. The incorporation of RCA in concrete increases both shrinkage and creep, which is due to the porosity and reduced stiffness of the mortar adhering to the aggregates [30,31,32]. To date, there are only a minimal number of studies addressing the shrinkage [23,33,34] and creep [35,36] of SCC with RCA, and it should be emphasised that no research has yet fully investigated the deformation properties of this material.

This paper presents the results of testing the mechanical (compressive strength, static Young’s modulus) and deformation (shrinkage, creep) properties of SCC with RCA obtained by crushing precast concrete elements. The tests were carried out with nine concrete mixtures. The influence of the water/cement ratio (w/c), the type of cement matrix, the amount of recycled aggregate and the cement replacement by fly ash were investigated. The knowledge gained can serve as input for the calibration and verification of existing models for predicting creep behaviour and contribute to an increased application of RCA in the construction industry.

2. Materials and Methods

2.1. Materials

The cement used for the production of concrete meets the conformity criteria of the European standard EN 197-1:2011 for rapid hardening Portland cement without mineral additives and is designated as CEM I 42,5 R [37]. In two concrete mixtures, the cement was partially replaced by siliceous fly ash from pulverised coal combustion. The density of the cement was 3.17 g/cm³ and that of the fly ash was 2.11 g/cm³.

The natural aggregate was crushed dolomite, and the recycled aggregate was produced from prefabricated concrete elements. Three fractions—0/4, 4/8 and 8/16 mm—of both aggregates were used for the tests (Figure 1). The density, absorption and particle size distribution of the aggregates are shown in

Table 1. Density and absorption of natural and RCA.

Property	Natural Aggregate Fraction			Recycled Aggregate Fraction
	0/4 mm	4/8 mm	8/16 mm	0/4 mm	4/8 mm	8/16 mm
Density (kg/dm3)	2.62	2.77	2.78	2.25	2.30	2.34
Absorption (% of mass)	0.5	0.4	0.3	7.5	6.1	4.7

and Figure 2. Recycled aggregates produced from concrete elements exhibit less variation in physical and mechanical properties than recycled aggregates containing brick or masonry mortar particles in addition to concrete.

In SCC mixtures, a dolomite filler was used that was produced in the same quarry as the natural aggregate. The dry density of the filler is 2.85 g/cm³ and its particle size distribution is given in

Table 2. Results of sieve analysis of dolomite filler particles.

Sieve (mm)	0.063	0.09	0.125	0.25	0.5	1.0
Passing percentage (%)	76	84	93	97	100	100

. Polycarboxylate superplasticizer was used to achieve adequate flow properties of SCC. According to the production specification, the density of the superplasticizer is in the range of 1.06–1.1 g/cm³.

2.2. Concrete Mix Design

Nine concrete mixtures were designed (

Table 3. Concrete mix composition quantities per 1 m³.

Component	Concrete Mixture
	NC 1	NC 2	SCC0	SCC10	SCC20	SCC30	SCC40	SCC50	SCC60
Cement (kg)	400	400	400	400	400	200	400	400	200
Natural aggregate 0/4 mm (kg)	930	930	970	970	970	970	970	970	-
Natural aggregate 4/8 mm (kg)	267	-	310	310	-	-	310	-	-
Natural aggregate 8/16 mm (kg)	-	-	320	-	-	-	-	-	-
Recycled aggregate 0/4 mm (kg)	-	-	-	-	-	-	-	-	970
Recycled aggregate 4/8 mm (kg)	-	267	-	-	310	310	-	310	310
Recycled aggregate 8/16 mm (kg)	644	644	-	320	320	320	320	320	320
Water (l)	200	200	200	200	200	200	160	160	160
Superplasticizer (kg)	2.4	2.0	16.1	7.6	10.5	8.6	12.6	16.9	20.6
Filler (kg)	-	-	200	200	200	200	200	200	200
Fly ash (kg)	-	-	-	-	-	200	-	-	200
Water/cement ratio	0.5	0.5	0.5	0.5	0.5	0.5	0.4	0.4	0.4
Aggregate/paste ratio (by volume)	2.21	2.28	1.75	1.86	1.91	1.75	2.08	2.13	2.09

). The empirical mix design method for SCC was used, in which cohesion is achieved by the addition of fillers [34]. All mixtures had the same binder quantity of 400 kg. The mixes NC1 and NC2 were normal concrete mixes with a w/c ratio of 0.5, in which the coarse aggregates were partially or completely replaced by RCA. The SCC0, SCC10, SCC20 and SCC30 mixes were SCC mixes with a w/c ratio of 0.5. The SCC0 mix was produced with natural aggregate, while in the SCC10, SCC20 and SCC30 mixes, the coarse aggregates were partially or completely replaced by RCA. In the SCC30 mix, 50% of the cement is replaced by fly ash. The SCC40, SCC50 and SCC60 mixes were SCC mixes with a w/c ratio of 0.4. In the SCC40 and SCC50 mixes, the coarse aggregates were partially or completely replaced by RCA. The SCC60 mix was an eco-mix, as it was produced entirely with RCA and 50% of the cement was replaced by fly ash. When calculating the aggregate/paste volume in Table 3, the volume of the aggregate is calculated as the volume of three aggregate fractions (0–4 mm, 4–8 mm and 8–16 mm) and the volume of the cement paste as the sum of the volumes of cement, fly ash, water and superplasticizer.

2.3. Specimen Mixing, Casting and Curing

Due to its high absorption capacity, the recycled aggregate was partially saturated with water. This reduced the amount of additional water that had to be added during mixing to compensate for the saturated surface dry condition of the moisture. All concrete constituents were conditioned at 20 ± 2 °C. Mixing was carried out in a compulsory mixer with a capacity of 75 litres. Since the amount of concrete for each mix was 108 litres, two batches were made for each mix, first the batch of 48 litres and then the second batch of 60 litres. Care was taken to reduce the time between mixing the two batches to around 45 to 60 min.

In the normal concrete mixes, the aggregates were first mixed with about half of the mixing water for two minutes. At the end of the two-minute standing time, the cement was added and mixed with the aggregate, followed by the addition of the remaining water mixed with superplasticizer. For SCC mixtures, aggregate and filler were mixed with about one-third of the water. After 30 s, cement or cement and fly ash were added, followed by the addition of the second third of the water. After two minutes of mixing, the remaining water, premixed with superplasticizer, was added and mixed for 1.5 min. Mixing was then interrupted for two minutes and continued for a further three minutes. For SCC mixtures, a slump flow test was carried out to check whether sufficient flowability of the mixture was achieved. An additional amount of superplasticizer was added to the first batch, if necessary, to achieve the desired slump or slump flow, and it was mixed for a further two minutes. In all cases, the second batch of mix was prepared with the same total amount of superplasticizer as the first batch, which was added premixed with water. Therefore, there is a difference in the type of superplasticizer dosage between the first and second batches.

The concrete specimens made from the NC1 and NC2 mixes were compacted on a vibrating table, while no vibration was used for the specimens made from SCC. After casting, the specimens were stored at a temperature of 20 ± 5 °C and covered with a plastic sheet. After 24 h, the specimens were demoulded. The specimens for the compressive strength test and Young’s modulus test were stored in the curing chamber at a relative humidity of >95% and a temperature of 20 ± 2 °C. The specimens for the shrinkage and creep tests were prepared for testing and transported to the test room where the relative humidity was kept at 60 ± 5%.

2.4. Testing Methods

After mixing, the following properties of the fresh concrete were evaluated: density (EN 12350-6), temperature and air content (EN 12350-7) [38,39]. For the NC1 and NC2 mixes, the consistency was evaluated using a slump test (EN 12350-2) [40]. For the SCC mixtures, the workability was evaluated by testing the flowability with the slump flow test (EN 12350-8), the viscosity with the V-funnel test (EN 12350-9), the passing ability with the L-box test (EN 12350-10) and the resistance to segregation with the sieving test (EN 12350-11) [41,42,43,44]. Figure 3 shows the L-box test and the measurement of the flowability of SCC.

Concrete cylinders with a diameter of 150 mm and a length of 300 mm were used to determine the compressive strength of the concrete, the static Young’s modulus in compression and to measure shrinkage and creep strain. For each test, one specimen was taken from the first batch and two specimens from the second batch of concrete. The compressive strength was tested at a concrete age of 2 and 28 days in accordance with standard EN 12390-3 [45]. The Young’s modulus was determined at the age of 28 days as the secant modulus after three loading–unloading pre-cycles in accordance with EN 12390-13 [46]. The compressive strength at the specimen age of 2 days was determined with the aim of assessing the development of the early compressive strength and to obtain the characteristic value of the strength required for the compressive loading of the specimens during the creep deformation test.

Drying shrinkage length change was determined in accordance with ISO 1920-8 [47]. After demoulding, measuring pins were glued to opposite sides of each cylinder at a distance of 200 mm and the initial value was recorded. The change in length was measured with a length comparator and the displacement was measured with a digital indicator. Parallel to the samples for shrinkage measurement, pins were also attached to the samples for creep test. The samples were loaded at an age of 2 days to a value of 30% of the achieved compressive strength. The load after 2 days was chosen to simulate the process of introducing the prestressing force into prefabricated, prestressed structural elements. The creep test was carried out in accordance with the ISO 1920-9 standard [48]. The load frame for the creep test consisted of a spring system to maintain the load and a permanently installed load cell to read and adjust the load force. Figure 4 shows some details of the measurement procedures. During the creep and shrinkage measurements, the ambient temperature was kept at 20 ± 2 °C and the relative humidity at 60 ± 5%.

The measurement of shrinkage and creep was carried out up to a concrete age of 180 days. An exception is the SCC50 mixture, for which the measurements were carried out up to a concrete age of 90 days. Previous investigations of the deformation properties of SCC with recycled aggregate have shown that the shrinkage and creep deformations become stable after 180 days and approach the asymptotic final value [35].

The creep coefficient of concrete under compressive load is calculated as follows:

$$\Phi \left(t,t_0\right)={\displaystyle \frac{{\epsilon }_p\left(t,t_0\right)}{{\epsilon }_{el}\left(t_0\right)}}$$

(1)

$${\epsilon }_p\left(t,t_0\right)={\epsilon }_{tot}\left(t,t_0\right)-{\epsilon }_{el}\left(t_0\right)-{\epsilon }_s\left(t\right)+{\epsilon }_s\left(t_0\right)$$

(2)

where Φ(t,t₀), ε_p(t,t₀) and ε_tot(t,t₀) are the creep coefficient, the creep strain and the total strain of the specimens under load, respectively, at concrete age t, which were subjected to a constant load at age t₀. Ε_el(t₀) is the elastic strain at loading, while ε_s(t) and ε_s(t₀) are the shrinkage strains at age t and t₀ respectively.

3. Results

3.1. Fresh Concrete Properties

The results of the fresh concrete properties are shown in

Table 4. Fresh concrete properties.

Fresh Concrete Properties	Concrete Mixture
	NC1	NC2	SCC0	SCC10	SCC20	SCC30	SCC40	SCC50	SCC60
Density (kg/m3)	2330 (±10)	2295 (±9)	2460 (±1)	2365 (±31)	2299 (±40)	2259 (±20)	2392 (±40)	2378 (±7)	2163 (±22)
Temperature (°C)	19.8 (±1.6)	20.1 (±2.1)	23.7 (±0.3)	23.1 (±1.4)	23.1 (±0.6)	22.2 (±0.5)	23.2 (±0.6)	23.1 (±1.5)	21.7 (±0.2)
Air content (%)	3.5 (±0.8)	2.8 (±0.1)	1.6 (±0.3)	2.2 (±0.7)	3.0 (±0.9)	1.5 (±0.2)	1.8 (±0.2)	2.3 (±0.6)	3.1 (±0.5)
Slump (mm)	90	95	-	-	-	-	-	-	-
Slump flow (mm)	-	-	745 (SF2)	735 (SF2)	690 (SF2)	715 (SF3)	800 (SF3)	755 (SF3)	725 (SF2)
Slump flow time, t500 (s)	-	-	1.7 (VS1)	0.5 (VS1)	1.4 (VS1)	0.6 (VS1)	1.0 (VS1)	1.9 (VS1)	2.4 (VS2)
L-box (3 bars) (-)	-	-	0.81 (PL2)	0.82 (PL2)	0.75 (−)	0.72 (−)	0.71 (−)	0.78 (−)	0.80 (PL2)
V-funnel (s)	-	-	5.6 (VF1)	2.5 (VF1)	2.7 (VF1)	3.1 (VF1)	7.4 (VF1)	11.8 (VF2)	9.4 (VF2)
GTM sieve (%)	-	-	5.2 (SR2)	11.6 (SR2)	6.0 (SR2)	8.8 (SR2)	11.0 (SR2)	3.5 (SR2)	0.0 (SR2)

For the density, the temperature and air content numbers in parentheses represent the half-span of two measurements. Classification of SCC mixes according to standard EN 206 is given in parentheses. For the L-box test, (−) means that the criteria are not met.

. The density of the fresh concrete decreases with the increase in the proportion of recycled aggregate, but also with the addition of fly ash, which is due to the lower density of these constituents compared to natural aggregate and cement. A reduction in the w/c ratio increased the density, as the reduction in the w/c ratio was achieved by reducing the water content. The air content varied between 1.5% and 3.5%. In SCC mixtures, the air content increased with the increase in RCA content, except in the SCC30 mixture, which contained fly ash.

The results listed in Table 4 for the slump of the NC mixtures and the workability properties of SCC refer to the second batch, in which the entire amount of superplasticizer was added premixed with water. This allowed for better dispersion of the superplasticizer and resulted in better flowability than the first batch. Therefore, the authors believe that these results are more meaningful for evaluating the effects of RCA on workability.

The slump values of the NC1 and NC2 mixes show that the concrete has a plastic consistency. Taking into account the amount of superplasticizer and the slump, RCA reduced the water requirement in the NC2 mix. The slump flow of the SCC mixes was classified according to the European standard for concrete HRN EN 206:2021 [49], which shows that the criteria for SCC were met (Table 4). From the dosage of superplasticizer and the results of the slump flow and V-funnel test, it can be seen that replacing the 8/16 mm fraction with recycled aggregate increases flowability and decreases viscosity. This is particularly evident in the SCC40 mix, where the amount of water was reduced by 40 litres compared to the SCC0 mix. The replacement of the 4/8 mm fraction increased the water requirement and viscosity in the SCC20 and SCC50 mixtures. It can be seen from Table 4 that the density of the SCC50 mix was almost the same as that of the SCC40 mix, although it contained a greater proportion of lower density aggregates. The results of the compressive strength tests (

Table 5. Hardened concrete properties at an age of 2 and 28 days.

Hardened Concrete Properties	Concrete Mixture
	Unit	NC 1	NC 2	SCC0	SCC 10	SCC 20	SCC 30	SCC 40	SCC 50	SCC 60
Density	kg/m3	2343 (±18)	2290 (±6)	2478 (±10)	2374 (±24)	2323 (±17)	2297 (±23)	2407 (±24)	2389 (±12)	2150 (±34)
Compressive strength, 2 days	MPa	27.3 (±0.7)	24.3 (±0.3)	46.0 (±0.4)	29.4 (±0.6)	24.6 (±0.2)	9.4 (±0.2)	37.6 (±0.9)	40.4 (±3.5)	9.3 (±0.1)
Compressive strength, 28 days	MPa	45.1 (±1.3)	45.9 (±0.7)	57.6 (±0.7)	51.1 (±0.6)	52.6 (±0.4)	25.3 (±0.7)	56.3 (±0.4)	65.4 (±1.0)	30.0 (±2.2)
* Young’s modulus, 2 days	GPa	15.4 (±1.9)	17.8 (±3.1)	36.8 (±1.4)	23.9 (±1.9)	19.5 (±1.9)	8.7 (±1.3)	24.3 (±0.8)	32.8 (±2.0)	9.3 (±1.0)
Young’s modulus, 28 days	GPa	31.0 (±0.6)	27.5 (±0.3)	40.6 (±2.9)	30.5 (±0.1)	30.6 (±0.2)	25.8 (±2.0)	35.3 (±0.8)	34.9 (±0.6)	23.5 (±0.9)

* Calculated from load and deformation measured at initial loading of samples for creep test. Numbers in parentheses represent standard deviation.

) support the possibility that mixtures in which both coarse aggregates were replaced with recycled aggregates have a denser packing of particles. It is known that denser packing of coarse aggregates can increase flow resistance [50]. Most of the results reported show that recycled concrete aggregates reduce the flowability of SCC [19,51,52,53,54]. The reduction in flowability is often attributed to the increased internal friction between the aggregates due to the increased angularity. The comparison of natural and recycled aggregates shown in Figure 1 indicates that this was not the case with the RCA used in this work.

The partial replacement of cement with fly ash improved fluidity, mainly due to the reduced friction caused by the spherical shape and smooth surface of the fly ash particles [55]. Apart from the SCC0 mixture, only mixtures with fly ash were able to fulfil the criteria for passing ability. It has been shown that the addition of pulverised coal fly ash increases the flowability and passing ability of SCC RCA mixes [23]. In the SCC60 mix, the highest superplasticizer dosage was required to achieve adequate flow properties. The particle size distribution of natural and recycled aggregate fractions differs, with the greatest difference being in the fine aggregates (Figure 2). The quantities of the individual fractions in the concrete mix were optimised for natural aggregate, which was then replaced by the same quantity of recycled aggregate. This leads to deviations from the optimised particle grading and can result in changes in the packing density and, consequently, changes in the water requirement.

3.2. Hardened Concrete Properties

Table 5 shows the results of the compressive strength and Young’s modulus tests. Each test was performed on three samples and the average value is given. The density was determined from the measured dimensions and mass of the samples for the compressive strength test. Figure 5 shows averaged values of the shrinkage strain and the total strain under compressive load. Both normal concrete and SCC appear to exhibit similar trends in strain development, regardless of the type of aggregate. It was found that the hyperbolic Equation (3), which is often used to represent the increase in the compressive strength of concrete [56], could be fitted fairly accurately to the shrinkage and total strain data.

$$y\left(t\right)=A{\displaystyle \frac{B\left(t-C\right)}{1+B\left(t-C\right)}}$$

(3)

where y(t) is the strain at age t (shrinkage strain or total strain), A is the limiting strain, and B and C are coefficients that determine the shape of the curve. Figure 5 also shows the best fit equations. The coefficients of determination were above 0.95 in all cases. The primary purpose of fitting the equations to the measured strains here is to smooth out the fluctuations in the average values that normally occur in this type of measurement. It is not intended to use these fitted curves to extrapolate data, particularly to estimate long-term strains, as these values would likely underestimate the actual strains that would be measured in a long-term experiment of 1 year or longer [57].

The creep coefficients calculated according to Equation (1) are shown in Figure 6a–c. The creep coefficient was calculated using measured data for ε_tot(t,t₀) and ε_s(t), with linear interpolation between the data points to match the age of the concrete. The measurement uncertainty intervals of the creep coefficient are shown in Figure 6 and are determined according to the methodology described in JCGM 100:2008 [58]. These intervals represent the combined standard uncertainty of the creep coefficient u_c(Φ) given by Equation (4).

$$u_c^2\left(\Phi \right)={\displaystyle \sum _{i=1}^N{\left(\frac{\partial \Phi }{\partial {\epsilon }_i}\right)}^2{u}^2\left({\epsilon }_i\right)}$$

(4)

where ε_i are the parameters of the measurement model and u(ε_i) are the associated standard uncertainties. The measurement model used here is assumed to be equivalent to Equation (2), where the parameters used to calculate the creep coefficient are ε_tot(t,t₀), ε_el(t₀), ε_s(t) and ε_s(t₀) and their standard uncertainties are determined from the standard deviations of the experimental data. It should be emphasised that the measurement uncertainty intervals shown in Figure 6 are mainly derived from the differences between the samples and are used here to assess the confidence in the averaged creep coefficients. Analysing the measurement uncertainty of the creep or shrinkage tests would require a more comprehensive discussion, which is beyond the scope of this paper.

4. Discussion

4.1. Impact of Variations in Concrete Mix Composition on Mechanical Properties

The mechanical properties at the age of 28 days show that replacing natural aggregate with recycled aggregate had different effects on the Young’s modulus and compressive strength (Figure 7). The SCC0 mix had the highest Young’s modulus. Replacing the coarse aggregate resulted in a reduction in Young’s modulus by 25% for mixes SCC10 and SCC20, regardless of whether only one or both fractions of the coarse aggregate were replaced. This reduction is primarily due to the lower stiffness of the recycled aggregate compared to the natural aggregate. The compressive strength in the SCC10 mix was reduced by 11% and in the SCC20 mix by 9%. Replacing 50% of the cement with pulverised coal fly ash in mix SCC30 resulted in a further 36% reduction in stiffness and 56% reduction in compressive strength compared to mix SCC0. This could be due to the slower pozzolanic reactions that dominate at an age of more than 28 days [59], but also to the lower stiffness of fly ash particles compared to cement particles [60,61]. The reduction of the cement paste volume by lowering the water content by 40 litres and the reduction of the capillary porosity due to the lower w/c ratio both increased the stiffness of the cement paste matrix in the SCC40 and SCC50 mixes. However, the introduction of recycled aggregate resulted in a reduction in Young’s modulus by 13% and 14% for the SCC40 and SCC50 mixes, respectively, compared to SCC0 mix. At the same time, the compressive strength was reduced by 2% in the SCC40 mix and increased by 14% in the SCC50 mix. It is noteworthy that the replacement of 4/8 mm aggregate in the SCC20 and SCC40 mixes had almost no effect on the Young’s modulus compared to the replacement of only 8/16 mm aggregate, but at the same time it increased the compressive strength (Figure 7a). For the normal concrete mix NC2, the replacement of both coarse aggregates also resulted in an increase in compressive strength compared to the NC1 mix. This could be related to the lower stiffness of the recycled aggregate, which reduces the stress concentrations, i.e., reduces the difference in the stiffness of the cement matrix and the aggregate, resulting in a more favourable stress distribution under loading [62]. Xiao et al. numerically determined that the magnitude of tensile and shear stresses in the interfacial transition zone increases rapidly with the increase in the modulus of elasticity of the aggregate [63]. This also means that the introduction of fly ash into the cement matrix in the SCC30 mix increased the difference between the stiffness of the aggregate and the matrix, resulting in higher stress concentrations compared to the SCC20 mix and consequently contributing to lower compressive strength. In the SCC60 mix, all the above effects are present, resulting in a 42% reduction in Young’s modulus and 48% reduction in compressive strength compared to the SCC0 mix. Carro-Lopez et al. compared the compressive strength of SCC with natural fine aggregate and with 100% replacement of the fine aggregate with recycled aggregate and found a 47% reduction in strength [52].

At an age of 2 days, the SCC10 and NC1 mixtures exhibited a higher compressive strength than the SCC20 and NC2 mixtures. The “crossover” in compressive strength for these mixtures therefore took place in the period between 2 and 28 days. This could also be related to the gradual increase in stiffness of the cement matrix caused by cement hydration and the associated redistribution of stress between the matrix and aggregates.

The increase in compressive strength in the mixtures SCC20, SCC50 and NC2 compared to the mixtures SCC10, SCC40 and NC1 shows that the replacement of all coarse aggregates by RCA is more favourable than the replacement of only one coarse fraction. The results allow the conclusion that incorporation of RCA reduces the compressive strength. However, this reduction can be partially compensated for if the natural aggregates are replaced by recycled aggregates in a sufficiently large particle size range. On the other hand, the workability and density results indicate that, in addition to differences in stiffness, the effect of denser particle packing on compressive strength should not be ruled out.

To achieve the transformation from normal to self-compacting concrete, part of the coarse aggregate contained in normal concrete mixes is replaced by dolomite filler. The volume of total coarse aggregate in SCC mixes is approximately equal to the volume of 8/16 mm aggregate in normal concrete. Therefore, the recycled aggregate content (by volume) in the NC1 mix is approximately equal to the volume of recycled aggregate in the SCC20, SCC30 and SCC50 mixes (Figure 7). Comparing the mechanical properties of the SCC20 and NC1 mixes, which both have the same w/c ratio of 0.5, it can be seen that replacing the normal cement matrix with the cement matrix of SCC leads to an increase in compressive strength (Figure 7a), while the Young’s modulus is approximately the same (Figure 7b). This could be related to the increased stiffness of the SCC cement matrix due to filling effects, i.e., increased cohesion and improved particle packing due to the addition of fine mineral admixture [62,64].

4.2. Shrinkage Deformation

Dimensional changes measured on unloaded concrete cylinders are not only caused by the evaporation of water from the surface, but also by self-desiccation, which is present throughout the specimens. Self-desiccation phenomena are associated with cement hydration and lead to so-called autogenous shrinkage [65]. Autogenous shrinkage develops faster than drying shrinkage and occurs mainly at an early age. Gabrijel et al. measured the autogenous shrinkage of SCC produced with the cement type CEM I 42,5 R with a w/c ratio in the range 0.4–0.47 and found that the autogenous deformation was most intense within the first 20 h and only a moderate increase was measured in the following 4 days [66]. It was also found that the change in length of concrete specimens subjected to drying 1 day after casting was mainly due to drying shrinkage for a mix with a w/c ratio of 0.4 [67]. The same study also found that the contribution of autogenous deformation to overall shrinkage increased with decreasing w/c ratio [67]. Therefore, the change in length of the unloaded specimens in this work can be largely attributed to drying shrinkage.

Regardless of the shrinkage mechanism, the internal restraints that counteract deformation are the same and are usually related to the quantity of aggregates and their drying shrinkage [65,68,69]. For concrete produced with natural aggregates, shrinkage depends only on the shrinkage of the cement paste and the volume of the aggregates, as most aggregates are non-shrinking [65]. RCAs contain hardened cement paste, which is the reason for the higher water absorption (Table 1). Since the recycled aggregate was partially saturated before mixing and further saturated during mixing, it is expected that this aggregate will show a reduction in volume due to water loss. Fathifazl et al. used pre-saturated RCA and concluded that the higher early shrinkage of the RCA mixtures could be due to the relatively easier moisture loss from the RCA particles [70]. This explains why the concrete mix SCC0 exhibited the lowest shrinkage strain. In addition, the SCC0 concrete mix exhibited the highest Young’s modulus (Figure 7) and thus the greatest internal restraint to shrinkage. The lowest Young’s modulus and the highest recycled aggregate content can also explain the highest shrinkage of mix SCC60. NC mixes contain a larger volume of recycled aggregate compared to SCC mixes with the same w/c ratio, which could explain their slightly larger shrinkage strain.

The regression line fitted to the experimental data shows that an increasing RCA content leads to an increase in shrinkage strain. As a first approximation, it can be assumed that this increase is proportional to the volume of RCA in the concrete mix, which can be seen from the slope coefficient of the regression line (Figure 8). The slope of the regression curve is almost the same when the 180-day shrinkage values are plotted. Domingo-Cabo et al. found that the shrinkage strain of mixes containing 100% recycled coarse concrete aggregate was 70% higher than that of the control mixture at 180 days [71].

The comparison of the shrinkage curves of the mixes SCC10, SCC20, SCC40 and SCC50 shows a higher shrinkage rate for mixes with a w/c ratio of 0.4 during the first weeks (Figure 5). This could be related to the greater contribution of the self-desiccation to the shrinkage strain in mixes with a lower w/c ratio. Mixes with a w/c ratio of 0.5 have a lower internal restraint and contain a greater amount of evaporable water, and at a later age, their shrinkage rate becomes greater than that of mixes with a w/c ratio of 0.4.

The replacement of cement with fly ash in the SCC30 mix increased the initial shrinkage strain, possibly due to a slow stiffness development. However, the long-term shrinkage strain is not significantly different from the shrinkage strain of the SCC20 mix. Qin et al. measured the shrinkage strain of concrete containing 50% and 100% recycled coarse concrete aggregate and in which 50% of the cement was replaced with fly ash [35]. They found that the presence of fly ash reduced the shrinkage strains by approximately 15% after 420 days compared to mixes containing recycled aggregate without fly ash.

4.3. Load-Induced Strain

The interpretation of the creep deformation of concrete based only on the creep coefficient Φ(t,t₀) can be misleading, especially when these coefficients are used for the creep analysis of concrete structures [57]. It should be noted that the creep coefficient is a normalised quantity. It depends not only on the actual creep strain ε_p(t,t₀), but also on the elastic deformation ε_el(t₀), which in turn depends on the magnitude of the applied stress and the Young’s modulus of the concrete at the age of loading E(t₀). The creep coefficients in Figure 6 are calculated assuming that the initial strain measured immediately after loading the test cylinders to the prescribed stress level σ(t₀) corresponds to the elastic deformation ε_el(t₀). The Young’s moduli at the age of 2 days (Table 5) are calculated as E(t₀) = σ(t₀)/ε_el(t₀). For a complete description of concrete material under load, at least one more parameter must be known in addition to the creep coefficient.

Another way of expressing the response of a viscoelastic material to a long-term load is the compliance function J(t,t₀). Using the compliance function, the time-dependent strain of samples under constant compressive load is expressed as ε(t) = J(t,t₀) σ(t₀), where σ(t₀) is the compressive load applied at age t₀. ε(t) consists of the instantaneous (or elastic) strain at the moment of loading and the time-dependent strain, i.e., ε(t) = ε_el(t₀) + ε_p(t,t₀). It follows that J(t,t₀) = [1 + Φ(t,t₀) ]/E(t₀). The compliance function, or simply compliance, is the sum of the elastic compliance, which is constant and equal to 1/E(t₀), and the creep compliance, which is time-dependent and is expressed as Φ(t,t₀)/E(t₀). In contrast to Φ(t,t₀), compliance function is a material property that provides a complete description of the concrete under the applied load, at least for linear creep analysis [57]. Figure 9 shows the compliance function for the concrete mixes analysed in this work. Error bars in Figure 9 represent the combined measurement uncertainty of the compliance function calculated from the standard measurement uncertainty of the creep coefficient Equation (4) and the standard deviation of the 2-day Young’s modulus (Table 5).

The physical interpretation of the compliance function is that it represents the strain per unit stress and is expressed in units 1/unit stress. Therefore, at the same stress level, greater compliance corresponds to greater strain. The initial value of compliance measured during the experiment appears at about 15 min (0.01 days), which corresponds to the time required to apply a load and measure the initial change in length. Since the measured compliance spans several orders of magnitude of time, these values are plotted on a logarithmic scale (Figure 9). For SCC mixes with a w/c ratio of 0.5, the lowest compliance is measured for mixture SCC0 (Figure 9b). The initial compliance depends on the concrete stiffness, so it is expected that the mix with the highest Young’s modulus will have the lowest initial compliance. Recycled aggregate increases compliance, especially when both coarse aggregate fractions are replaced. The effect of replacing both coarse fractions with recycled aggregate became evident after the 10th day under load, when the slope of the compliance curve started to deviate from the compliance measured for the SCC0 and SCC10 mixes (Figure 9b). This shows that the rate of creep strain after an initial period of 10 days is higher in the mix SCC20 than in the mix SCC10. The replacement of cement with fly ash in the mix SCC30 resulted primarily in an increase in initial compliance while the shape of the compliance curve is very similar to that of the mix SCC20. For normal concrete mixes, the effect of the amount of recycled aggregate is also visible after the initial 10-day period under load (Figure 9a). Even if only the creep compliance is analysed, the replacement of cement with fly ash is found to increase the creep strain. Kou and Poon tested concrete with 25% and 35% fly ash replacement and found that for samples loaded at 28 days, the creep strain measured after 120 days decreased with increasing fly ash content [72]. This phenomenon was attributed to the greater increase in strength due to pozzolanic reactions after loading. The contribution of fly ash to strength development depends on both the w/c ratio and the fly ash replacement ratio [73]. With a high proportion of fly ash (>45%) and a high w/c ratio, the compressive strength of concrete with fly ash cannot exceed the compressive strength of control concrete [73].

The comparison of the creep coefficients shown in Figure 6b shows that the creep coefficients are lowest for the mix SCC30, which could lead to the false conclusion that strain under load is lowest for this mixture. But as already mentioned, this is because the creep coefficient does not contain complete information about the behaviour of the concrete under load.

Since creep originates from hydrated cement paste [62], mixes with a lower cement paste content and less water available for hydration exhibit a lower creep strain. This led to a lower load-induced strain of the SCC40 and SCC50 mixes compared to the SCC10 and SCC20 mixes. However, this cannot explain the lower strain of mix SCC50 compared to mix SCC40, as the test results of the other mixes clearly show that the incorporation of a higher amount of recycled aggregate increases the creep strain. A look at the results of the compressive strength test after 2 and 28 days (Table 5) shows that the mix SCC50 exhibited a greater increase in strength of 25 MPa compared to 18.7 MPa measured on samples from mix SCC40 in that period. In addition, the mix SCC50 has a higher aggregate/cement ratio compared to the mix SCC40 (Table 3). Both effects, the lower cement paste content and the greater increase in strength, counteracted the effect of the greater volume of recycled aggregate, so the measured load-induced strain of the mix SCC50 was lower overall than that of the mix SCC40.

For the mix SCC60, the initial compliance is comparable to the stress-induced strain measured for the mix SCC30, which is due to the lower initial stiffness at 2 days of age due to the cement replacement by fly ash. At longer load duration mix SCC60 shows the highest rate of creep strain which can be attributed to the incorporation of fine recycled aggregate. Although very little data are available on the influence of fine RCA on creep, studies have shown that it increases creep strain [31].

Manzi et al. found that the lowest creep in SCC with RCA occurred in the mix with the highest compressive strength and the smallest pore size distribution [36]. Figure 10 shows the creep compliance after 28, 90 and 180 days under load together with the 28-day compressive strength. The initial compliance can be derived from Table 5 as the reciprocal value of the 2-day Young’s modulus. The data in Figure 10 confirm that creep was lowest in the mix with the highest compressive strength. However, the mixes NC2 and SCC20 exhibited higher creep values than the NC1 and SCC10 mixes despite their higher compressive strength. Both mixes, NC2 and SCC20, had a lower cement paste content and consequently a higher aggregate/cement paste volume (Table 3), which should also contribute to the lower creep strain. This indicates that the increased creep strain in these mixes can only be attributed to the presence of recycled aggregate. Replacing cement with 50% fly ash increased the creep strain in the mix SCC30 but not as much as the decrease in compressive strength would suggest. On the other hand, replacing the fine aggregate with RCA doubled the creep strain compared to the mix SCC30 despite the lower cement paste content. When considering whether the influence of the fine RCA is greater than the influence of coarse RCA, it should be borne in mind that the fine RCA in the SCC mix accounted for 60% of the total aggregate. Therefore, it is not obvious that fine RCA makes a greater contribution to creep than coarse RCA.

Figure 10 also shows the average density of the concrete samples for the compressive strength test. With the exception of the mix SCC50, the decrease in concrete density corresponds well with the increase in creep strain. The linear correlation between density and 28-day, 90-day and 180-day creep compliance has coefficients of determination of 0.81, 0.87 and 0.90, respectively. In comparison, the linear correlation between compressive strength and creep compliance has coefficients of determination of 0.64, 0.62 and 0.56 for 28, 90 and 180 days under load, respectively.

Most models for predicting the creep of concrete use the 28-day compressive strength as input [74]. ACI 209R-92 requires the concrete density in addition to the compressive strength to evaluate Young’s modulus [74]. It is therefore known that both compressive strength and density are strongly correlated with creep and both are governed by porosity. What concretes with recycled aggregates have in common is their increased porosity, which is caused by the porosity of the recycled aggregate. For the concrete mixtures analysed in this work, the volume of the pores contained in the aggregate can be calculated from density and absorption properties (Table 1) and the concrete mix composition (Table 3). This calculation shows that the volume of pores contained in the aggregate alone, excluding air voids and cement paste porosity, ranges from 0.69% for the SCC0 mix to 9.45% for the SCC60 mix. The creep strain measured in this work is the total creep, which is a sum of basic creep and drying creep which is strongly influenced by moisture movement [57]. This implies that the incorporation of RCA in concrete affects the moisture movement, which also affects the creep deformation, and that this process could be the origin of differences in the measured creep deformation between concrete with natural aggregate and concrete made with RCA.

5. Conclusions

Recycled concrete aggregates are considered “high quality” recycled aggregates for concrete production in terms of their mechanical and physical properties and their variability. Of particular interest in this work was the influence of RCA on long-term dimensional stability, i.e., shrinkage and creep deformation. The addition of RCA instead of natural crushed aggregate leads to a change in the fresh and hardened concrete properties:

• When only the 8–16 mm fraction was replaced, the flowability increased, most likely due to the fact that the RCA used contains a certain amount of river gravel particles that were detached from the cement matrix during the crushing process, reducing the overall angularity of the coarse aggregate. A further increase in the RCA content led to a decrease in the flowability of SCC mixes. The greatest increase in the water demand was observed when the fine aggregate was replaced with RCA.
• RCA reduced the Young’s modulus and generally also reduced the compressive strength. The increase in compressive strength in the mixes SCC20, SCC50 and NC2 compared to the mixes SCC10, SCC40 and NC1 indicates that replacing all coarse aggregate with recycled aggregate concrete is more favourable than replacing only one coarse fraction. The main reasons for this are a more favourable stress distribution under load due to a smaller difference in the stiffness of the cement matrix and aggregate, or a higher packing density.
• An increasing RCA content leads to an approximately proportional increase in shrinkage strain. This is due to the greater absorbed water content in the RCA and the simultaneous reduction in stiffness. The presence of fly ash did not cause a significant deviation in shrinkage strain.
• An increase in the coarse and fine fraction of RCA increases creep deformation. Although the highest creep strain was measured on test specimens made from a mix that contained fine RCA, it is not obvious that fine RCA makes a greater contribution to creep than coarse RCA.
• The replacement of aggregate and cement with RCA and fly ash led to a relatively large change in concrete density. Density had a stronger linear correlation with creep strain than compressive strength.

The European standard for concrete production EN 206 allows the use of up to 50% replacement of coarse aggregates with RCA. Nevertheless, the data show that RCA is rarely used as an aggregate in concrete. This is understandable, as there are a number of parameters that need to be considered when deciding whether the use of RCA is appropriate. One technological limitation is that relatively small quantities of RCA come from a demolition site. Or if concrete with natural aggregates meets a certain compressive strength class, it is not sufficient to simply replace the natural aggregates with recycled aggregates, as RCA concrete is likely to have a lower strength. To compensate for the lower strength, some modification of the concrete composition is required, probably by increasing the cement content.

However, it is possible that the use of recycled aggregates in concrete will be enforced by regulations. Concrete already accounts for more than 50% of demolition materials, and the need to recycle these into concrete aggregates will direct demolition activities towards prior characterisation and more careful separation of each type of material.

The experimental investigations carried out in this work have confirmed that SCC–RCA is feasible. The mechanical properties of SCC–RCA, even when 100% of the aggregate is RCA, can be utilised for structural applications. The w/c ratio range of 0.4–0.5 covers most concrete mixes used in construction, so the results presented in this paper can be widely applied. The design rules for concrete structures allow the use of RCA and the results presented in this paper can be used to re-evaluate the resistance reduction factors, especially for structural elements where deformations are of concern.