1 Introduction
Three-dimensional (3D) printed concrete is becoming a more widely applied material in the concrete industry. In this technology, cementitious materials are used to construct elements layer by layer. The benefit of the production process is the increased form freedom leading to the possibility to optimize the construction. This could significantly reduce the concrete content and waste [
1–
3]. On the other hand, printable mixtures need a large quantity of binder material [
4] and are lacking a proper reinforcement technique [
5]. Therefore, some researchers have suggested the combination of 3D printed concrete and conventional concrete to harvest the benefits of both materials [
6–
9].
Combining two types of concrete in one construction is a difficult task. To obtain compatible behavior of these composite elements, different concrete properties should align with each other [
10]. A mismatch will result in stress concentrations or residual stresses in the final structure. One of these influencing factors is the long-time shrinkage behavior of concrete [
11].
Excessive shrinkage strains may induce tensile stresses exceeding the maximum tensile strength at an early age [
12]. This leads to small cracks which drastically decrease the durability of concrete structures [
13]. Knowledge and understanding of this time-dependent behavior are of high importance for obtaining more durable structures.
Where the concrete is cast at various ages or different concrete types are used, a difference will occur in shrinkage. The least shrinking material will at that point restrain the shrinkage of the other one. Restraining of the shrinkage leads to stresses, developing parallel to the interlayer between the two materials [
10]. In the restrained material, tensile stresses will occur while compressive stresses develop in the other. When the tensile stresses exceed the tensile strength, cracking or debonding is possible [
11]. Lowering the shrinkage difference reduces the restraining and thus the formation of micro cracks. Therefore, shrinkage compatibility between the different materials should be taken into account to successfully design a composite structure [
10].
Intensive research has already been performed on the compatibility of two cementitious materials in contact with each other. However, most of this research is related to repair mechanisms [
14–
16] involving a thin fresh mortar layer applied on an old concrete substrate. In most of these cases, the effect of restraining the shrinkage of the fresh mortar is investigated to improve cracking resistance. It is accepted that curing conditions and repair mortar properties both significantly influence the shrinkage development of the repair material [
10]. For example, Beushausen [
14] concluded that by improving the curing of the repair mortar, the tensile strength and elastic modulus increase and tensile relaxation decreases.
Existing 3D printed concrete contains mainly fine sand as an aggregate fraction. This means that, despite the name, these mixtures will have material properties similar to mortars rather than concrete [
15]. The total shrinkage of mortars is significantly larger than that of conventional concrete. The main factor is the low quantity of large aggregates, leading to a larger cement paste volume and increased shrinkage [
16]. The lack of formwork protection at an early age facilitates water evaporation and can also exacerbate shrinkage problems for 3D printed concrete [
17]. For example, Markin and Mechtcherine [
18] reported significant shrinkage-induced cracks propagating throughout the 3D printed concrete walls of a residential building (“Project Milestone, Eindhoven, Netherlands”). The formation of these cracks resulted in high repair costs and possible reduction of the service life of the overall structure.
Plastic shrinkage is of particular importance for 3D printed concrete during ages up to two hours. Moelich et al. [
17] concluded that a substantial amount of free shrinkage happens during the first 2 h after printing due to evaporation. However, several techniques to reduce moisture loss and consequent shrinkage have been investigated. Federowicz et al. [
19] tested several internal (shrinkage reducing admixtures) and external curing techniques (foil covering) to reduce the total shrinkage at early ages. Results showed that external curing was the most effective with a shrinkage reduction from 5178 to 1031 µm/m after 7 d. The usage of shrinkage reducing admixture caused a shrinkage reduction of only 7% to 23%, depending on the used quantity. van Der Putten et al. [
20] evaluated the effectiveness of adding superabsorbent polymers (SAPs) to reduce early age shrinkage involving internal curing. The introduction of SAPs showed a mitigation of 200% of total shrinkage compared to the reference mixture after 5 d.
It is not only the mixture composition that affects the shrinkage behavior of 3D printed concrete, but also the environment. Temperature and relative humidity (RH) control the evaporation rate of water from the printed concrete. Shahmirzadi et al. [
21] investigated the effect of different environmental conditions on the total shrinkage. The study showed that specimens cured at a RH of 85% compared to an RH of 35% had a 55% reduction in total shrinkage. Moreover, the contribution of drying shrinkage in the case of printed concrete cured at an RH of 85% became insignificant. Le et al. [
22] measured shrinkage on specimens submerged in three environments: in water, covered in foil and exposed to the air (RH 60% and temperature 20 °C) for 6 months. Printed concrete specimens cured in foil had drying shrinkage reduced by a factor of 3 to 4 compared to specimens exposed to the air. An explanation was found in the sensitivity of the cementitious system to changes in RH. Kovler and Zhutovsky [
23] explained that a disturbance in vapor pressure outside the element will change the equilibrium between absorbed water and vapor pressure within the system. As a result, any change in equilibrium will cause either net evaporation or net condensation depending on the RH, leading to shrinkage or swelling. In the case of an RH higher than 40%, studies attribute these volumetric changes to capillary tension, whereas RH lower than 40% are related to variation in surface energy and movement of interlayer water. However, there is no agreement on the active mechanism at particular humidity levels [
23].
The main difference between the application of 3D printed concrete as formwork and repair mortars is the sequence of application. While repair mortars are applied on an old concrete substrate, the conventional concrete is cast against the printed formwork. In other words, the printed concrete is the substrate and the cast concrete is the overlay. This makes the time gap between printing and casting an important factor. In the literature, little is mentioned about this time gap in the case of 3D printed formwork application. Wang et al. [
24] suggested that a long period between casting and printing would lead to the formwork constraining the cast concrete. This can result in microcracks near the interlayer zone affecting the bond strength. Zhu et al. [
9] printed cylindrical formwork and cast concrete inside after 24 h of curing of the formwork, for an axial loading tests, but no reports were made regarding shrinkage cracking. An earlier study by the authors [
25] tested the effect of casting on the printed formwork. Concrete was cast in cylindrical formwork with time delays of 7, 14, and 28 d. During the hydration period of the cast concrete, the deformation at the outside of the formwork was monitored. It was observed that during the first hours after casting, the printed formwork swelled on the outside. Once the hydration peak was passed, it started to contract again.
In this work, the mutual restraining effects of the printed and cast concrete and the effects on the outside of a concrete structure are investigated. In addition, the shrinkage profile throughout the printed layers is studied. Different environmental conditions and different time gaps between printing and casting are considered, as well as the influence of the commonly used ‘prewetting’, a pre-treatment technique to enhance the bond strength.
2 Materials and methods
2.1 Materials
The printed concrete was an in-house designed mixture by Mohan et al. [
26]. The binder was a blend of ground granulated blast-furnace slag (GGBS) provided by Ecocem Benelux B.V. and Portland Cement CEM I 52.5 N. The chemical composition can be found in a previous study by Mohan et al. [
26]. A washed sea sand with a maximum size of 1 mm was utilized as aggregate. To improve rheological properties, a methyl hydroxypropyl cellulose viscosity modifying agent (VMA) was added (Tylose® MOT 60000 YP4). The mixture has a sand-to-binder ratio of 1, which made it less eco-friendly. However, increasing the sand-to-binder ratio would have made the mixture more vulnerable to variations during printing and decreased the open-time of the printed mixture. This would have lead that not all samples could be printed at once.
The cast concrete, a self-compacting concrete (SCC) and conventional vibrated concrete (CVC), were based on the results of Desnerck et al. [
27]. Modifications were applied to the superplasticizer content of the SCC to obtain the desired flowability. The incorporated limestone filler was Calcitec 2001 M commercial, available from Carmeuse N.V. For all concrete mixtures, the same polycarboxylic ether-based superplasticizer was used (Masterglenium 51%–35% solid content). Both cast concretes had the same binder content and a water-cement ratio of 0.46. The particle size distribution can be found in Fig.1. The mix composition and properties of all mixtures are shown in Tab.1 and Tab.2, respectively.
2.2 Sample preparation
2.2.1 Cast concrete procedure
To properly disperse all the materials, the aggregates and binder materials were dry-mixed for 60 s. Water was added during the first 60 s after dry mixing. In the case of SCC, the superplasticizer was added after 60 s while continuously mixing for up to an extra 180 s. After this first mixing stage, any dry materials stuck to the bottom or side of the mixing pan were loosened. Afterwards, the concrete was mixed for another 120 s. Once the mixing was finished, the material was cast immediately in the molds.
2.2.2 Printing procedure
The printing procedure has been extensively explained in Ref. [
26]. Printing was performed with an Asea Brown Boveri (ABB) robot arm (Fig.2(a)) and Rudolf pump. During printing, the temperature was 19.3 °C and a RH of 45.2% was measured. The printed layers had a width of 40 mm and a height of 15 mm. A rectangular nozzle with side trowels was used to obtain a flat surface (Fig.2(b)). Printing was performed with a nozzle speed of 100 mm/s. Linear specimens were printed with a length of 1000 mm.
After printing, the specimens were cured in a climatizing room with a temperature (20 ± 1) °C and RH of at least 95%. After 24 h, the specimens were dry-cut into the desired dimensions. Afterwards, specimens were stored in different curing conditions: water cured; air cured with RH: 95%; air cured with RH: 60%.
2.2.3 Combined elements for differential shrinkage
Combined elements, with printed formwork and the cast concrete, were made to test the differential shrinkage. Printed specimens were prepared as described in the section ‘Printing procedure’. After designated curing periods (1, 7, or 28 d), three specimens were taken from each curing condition (water cured, air cured with RH 95%, and air cured with RH 60%). In the case of the water-cured specimens, excess water on the surface was first removed with a wet cloth. To simulate ‘pre-wetting’ pre-treatment, three printed samples were placed with their surface, that was to be in contact with the cast concrete, in water. The specimens were exposed to the water for 45 min. Afterwards, excess water was removed with a wet cloth. All specimens were placed in oiled steel molds (100 mm × 100 mm × 400 mm) as depicted in Fig.3. As the length of the molds was longer than the printed samples, partitions were placed. Once in place, concrete (prepared as in the section ‘cast concrete procedure’) was cast as shown in Fig.4. SCC was cast from a casting height of 1 m. CVC was placed in two different layers with compaction in between using a vibration table. After casting, the specimens were stored in a climatizing room with a temperature (20 ± 1) °C and RH of at least 95%. 24 h after casting, the combined elements were demolded. In Tab.3, an overview is given of all combined elements that were produced. For each combination, three specimens were prepared.
2.3 Methods
2.3.1 Free shrinkage
To observe the effect of shrinkage on the cast concrete, the free shrinkage of specimens covered in aluminum foil (sealed) and specimens exposed to the air (exposed) was tested. The free shrinkage of the SCC is measured according to NBN EN 12390-16 (2019). Six prisms of each concrete type (400 mm × 100 mm × 100 mm) were cast and cured for 24 h in a climatizing room with a temperature of (20 ± 1) °C and RH of at least 95%. After demolding, on all four sides of the specimens, two Demec-points were applied with a base length of 300 mm. The shrinkage was measured for three specimens exposed to the air (temperature (20 ± 2) °C and RH of (60 ± 10) °C), and for three specimens fully covered in aluminum foil. The change in length was measured at 1, 4, 7, 14, 28, and 56 d after demolding.
The free shrinkage of printed material had been investigated in a previous study by the authors [
25]. The previously reported results were further used in this research as the printed material was the same.
2.3.2 Differential shrinkage
The differential shrinkage was measured on the combined elements, as inspired by the tests performed by Beushausen [
14]. To measure the length change, Demec points were applied on the top and side of the printed specimen with a base length of 200 mm on the day of casting. One day after casting, the combined specimens were demolded and Demec-points were glued on the cast concrete. The exact location of the Demec-points can be seen in Fig.5 and Fig.6. Afterwards, the specimens were fully covered in aluminum foil with only one side exposed to the air (the side of the printed concrete) as shown in Fig.7. After preparation, the specimens were stored in a climatizing room (temperature (20 ± 2) °C and RH of (60 ± 10) °C) and the length change was measured for 56 d after casting. The starting point of the measurements (day 0) was the day of casting the concrete, so that only the printed material was measured at that point. The reference points of the cast concrete were measured on day 1. In all graphs, the average measured values are reported together with the standard error on the error bars.
Three additional printed specimens were prepared, similar to the combined specimens of 1-RH90-SCC, but without the cast concrete. In that case, one side was exposed and the other 5 sides were sealed with tin foil. In this way, no restraining effect due to the cast concrete was present.
3 Results
3.1 Shrinkage at the exposed side
In Fig.8, the results of the shrinkage of the printed [
6] and cast concrete are shown, up to 56 d after printing or casting. At first sight, a significant difference can be seen in the magnitude of shrinkage between the printed and cast concrete. In the case of total shrinkage, the printed concrete experiences a contraction which is at least 10 times larger than that of the cast materials in the first 7 d. The main reason for the difference between the printed and cast concretes is the lack of formwork protection during hydration and curing. Previous research has already shown that not protecting printed material after printing leads to a high level of evaporation of the water and induces high drying shrinkage rates of the printed material at the beginning of the curing period [
19]. Additionally, the high area-to-volume ratio of the printed elements compared to that of the cast concrete specimens leads to faster drying of the printed specimens. Over time, the ratio of (total shrinkage of printed concrete/total shrinkage of cast concrete) converges to a constant value of around 3. The high total shrinkage of the printed material compared to traditional mold cast concretes is not unusual, due to a high cement paste content and the lack of large aggregates of the printed concrete [
16,
17].
Not only the shrinkage behavior of the drying cast and printed specimens, but also the shrinkage behavior of the sealed cast and printed specimens is significantly different. The total shrinkage of the cast concretes is lower than that of the printed material. This is not unexpected as mixtures with higher w/c-ratios are less prone to autogenous shrinkage. Here, the cast concretes have a w/c-ratio of 0.45, while that of the printed material is only 0.35. At 56 d, the printed material under sealed conditions experiences a shrinkage which is a factor 2.85 times larger than that of the cast concretes.
Fig.9 shows the shrinkage effect of the printed material when measured at the outside of the formwork (base length indicated in Fig.5). Note that for the restrained specimens after 7 and 28 d, the free shrinkage during the curing period is not taken into account. The reference length is measured with respect to the first day of restriction.
When comparing the fully exposed and the one-sided exposed specimens, a reduction in shrinkage is visible for the one-sided exposed specimens. One-sided exposed specimens show a reduction of 22% ± 2% compared to the specimens fully exposed to the air, measured over all dates. Aluminum foil around the specimen prevents evaporation of water at most surfaces. This reduces the moisture flux within the specimens and creates a non-uniform distribution of the RH within the element. As a result, differential shrinkage starts to occur throughout the thickness of printed specimens [
28,
29]. It has to be noted that the one-sided exposed specimens could freely shrink. The differential contraction within the printed material could therefore result in the bending of the specimen [
14]. Curvature due to the bending is hard to determine and is therefore not reported in this study.
When comparing the shrinkage of the one-side exposed specimens with the ones which were cast (1-RH90-SCC), an additional reduction in shrinkage is observed. The reduction is 22% ± 3% smaller than in the case of the free specimens exposed at one side, on all measurements after 7 d. The reduction is believed to be a result of the restraining effect of the cast concrete [
11]. During the curing of the cast material, physical bonding will occur between the printed and cast concrete. In theory, this perfect bond results in an equal length change of the printed and cast concrete at the interface. As the cast concrete shrinks significantly more slowly than the printed concrete, a restraining of the free shrinkage will occur. According to Beushausen [
11], this restriction of free shrinkage would lead to tension within the printed material.
In between printing and casting of infill concrete, the formwork can freely shrinkage. The longer the formwork can freely shrink, the less chance the printed material will be restrained after casting. This can be seen in the results. Having a casting delay of 7 d gives a drastic reduction in strain, relative to that in the case of same-day casting, once restrained. When having a casting delay of 28 d, not shrinkage but expansion is measured at the outside of the element. Therefore, the shrinkage history of the printed formwork will influence the interaction between the two materials.
3.2 Shrinkage profile-effect of curing condition
Fig.10–Fig.14 show the shrinkage profiles of the printed and cast concrete over the thickness of composite specimens after curing for 56 d, in different curing conditions and with different cast materials. The values on the Y-axis show the distances toward the interlayer in centimeters. Negative values of distance represent the thickness of the cast concrete and positive values of the printed material. Note that the values measured on Day 1 for the cast concrete are zero, as this was the reference point.
Fig.10 shows the profile of 28-RH95-SCC. One day after casting, a small expansion at the interlayer occurs. The expansion seems not to manifest throughout the whole length of the printed material and is limited toward the interlayer. This is likely the result of water absorption by the printed concrete. As the printed material is exposed to a material with higher moisture content/RH (being in contact with the freshly cast concrete), the printed concrete will start to swell. During curing at 95% RH, only a small quantity of the water is evaporated, which reduces limits the absorption and swelling of the printed material.
Over time, both materials start to shrink. As only one side of the specimen is exposed to the air, a difference in the RH profile will occur within the specimen. Therefore, a difference in shrinkage rate at the exposed side and interlayer is seen for the printed concrete [
21]. The outside of the printed material shrinks much faster than the inner part. Next to this, there is already a significant difference in shrinkage rate between the printed and cast concrete. When observing the results as shown in Fig.7 and Fig.8, the SCC shrinks up to 10 times less than the printed material. As a result, the shrinkage of the printed material is restrained. This can be observed in Fig.10 and Fig.8 and also in Fig.9. However, it is worth mentioning that due to the restraining effect, the cast concrete is pulled along at the interface.
Fig.11 and Fig.12 depict the results of 28-RH60-SCC and 28-RH60-CVC. The results show that the evolution of the shrinkage profile and the form of the profile both follow a similar trend within the printed material, independent of the type of concrete. It is clear that after first contact with the cast material (day 1), the formwork starts to swell at the interlayer. Within the printed material, a linearly decreasing shrinkage profile with increasing distance of the interlayer is formed. As a result, an increase in strain is only slightly observable on the outside of the printed specimens. Wei et al. [
30] suggested that by changing the RH at one side, different warping deformation can occur, analogously to a temperature gradient. Xue et al. [
31] indicated that dry mortar absorbs water from freshly applied mortar. Knowing that the water uptake of dry concretes leads to swelling of the dried C-S-H structure [
32], it was expected that the printed concrete would start to swell after casting.
With an increase in time, the printed materials of 28-RH60-SCC and 28-RH60-CVC shrink. Where, before casting, the printed materials dried already for 28 d then most of the drying shrinkage has passed already (Fig.8). At this stage, most of the residual shrinkage is autogenous shrinkage which has a uniform distribution. As a result, the shrinkage profile in the printed material exhibits a rather constant horizontal shortening in time.
Comparing the two cast concretes, similar behavior was expected. Covering the exposed surface with thin foil was performed to limit the effect of drying shrinkage. Therefore, a uniform shortening profile would be expected. In Fig.11, this effect is not visible. The cast material develops a differential shrinkage profile rather than a uniform one. The printed material may be attracting the water of the SCC. Therefore, some drying shrinkage is introduced, creating an uneven deformation. Additionally, in Fig.8 it can be seen that after 28 d curing of the printed concrete, the shrinkage of the SCC is larger than that of the printed concrete. In that case, the printed concrete restrains the shrinkage of the SCC. Lowering the shrinkage of the cast concrete would improve the compatibility with the shrinkage of the printed concrete and a less uneven shrinkage profile would be expected. In Fig.12, a uniform shrinkage profile is forming over the depth of the cast material. Thus, optimal compatibility between the printed concrete and CVC seems to be obtained.
Fig.13 and Fig.14 show the shrinkage profiles of specimens 28-W-SCC and 28-W-CVC. The results show again a development of the shrinkage profile in the printed material which is independent of any influence of cast material. In both cases, no expansion is observed at the interface after casting. Moreover, the printed material immediately starts to shrink. Xue et al. [
31] observed that water migration from the fresh material toward the old substrate is limited when the substrate is fully saturated. Moreover, a small water flow from the substrate toward the hydrating repair mortar can be seen. This is a result of the changing capillary pressure of the newly formed structure. These printed specimens were cured underwater, where no evaporation could happen. This external curing method fills the pores with water from the start. As the pores are saturated, there is no water to flow from the freshly cast concrete toward the printed material during the casting procedure. Therefore, no swelling is observed. After casting, the printed specimens are partly exposed to the air and evaporation will eventually start. This creates a moisture gradient, resulting in an uneven shrinkage profile within the printed material [
21] as seen in both figures.
Concerning the cast concrete, both SCC and CVC experience differential shrinkage through the thickness of the element. This is a result of the moisture gradient in the printed material. It is believed that moisture can freely flow between the printed and cast materials. The development of the moisture gradient in the printed material, due to the evaporation surface, will extend into the cast material. This will lead to a moisture flow from the cast material toward the printed one. The water will flow until an equilibrium is reached between the internal and external moisture condition. However, this makes the water content through the depth of the cast concrete variable. As a result, the cast material can experience some kind of drying shrinkage which is higher at the interlayer than at the covered outside of the cast concrete.
3.3 Differential shrinkage-effect of casting delay
Fig.15 shows the shrinkage profiles measured after one day on specimens 1-RH90-SCC, 7-RH60-SCC, and 28-RH60-SCC. A significant difference can be observed between the profile created with a casting delay of 1 day and those with the casting delays of 7 and 28 d. Shrinkage of the printed material throughout the whole depth is observed when the casting delay is only 1 d. In the case of delays of 7 and 28 d, swelling of the printed material can be observed. The casting delay seems to influence the behavior of the printed specimens. The cause may be related to the moisture content of the printed material. When printed concrete is exposed to the air for a short duration after printing, only a little of the pore water will evaporate. As drying shrinkage of the printed material is related to the moisture content, only a limited part of the drying shrinkage will have occurred after one day. This can be seen in Fig.8. Therefore, when the casting delay is small, only a small quantity of water exchange and a high magnitude of shrinkage of the printed material is expected. In cases of longer period between casting and printing, drying shrinkage has mostly been finished. This means that pores inside the printed material are only partly filled with water. When the printed concrete becomes in contact with the freshly cast concrete, extraction of free water from the cast concrete by the printed one is expected [
33]. By filling the pores of the printed material, the printed material swells [
34]. Note that the swelling at the interface at 7 and 28 d is similar. It is presumed that the amount of swelling of the printed material is related to the drying shrinkage at the time of exposure to water. For this printed concrete, there is no significant difference in drying shrinkage between 7 and 28 d [
25]. Therefore, the expansion is expected to be similar.
Comparing Fig.16 with Fig.15, a different development of the shrinkage profile is obtained based on the casting delay. In case of a casting delay of 1 d, the printed concrete has a significantly higher shrinkage than the cast material. The difference in shrinkage results in differential strain along the depth of the composite element. It seems that the length reduction changes uniformly throughout the depth of the material. However, the gradient changes at the interface of the two materials.
The shrinkage profile of the specimens with a casting delay of 7 d shows a linear change. The linear shrinkage profile in the printed material does not change over time as it does for the specimens with a casting delay of 1 d. Similar behavior can be observed for the cast material. An explanation can be found in the shrinkage compatibility of both materials. In Fig.8, it can be observed that the total shrinkage profile of the printed material with a casting delay of 7 d and the total shrinkage of SCC have similar value after 28 d (285 µm/m for the printed material and 265 µm/m for the cast concrete). Therefore, no differential shrinkage is observed between the two materials. Any differential shrinkage that is observed within the printed material is related to the first day and to the swelling during casting.
In case the delay of casting is 28 d, a small change can be seen in the profile of the cast material. It is believed that the shrinkage of the printed material is less than that of the cast material. An opposite reaction to that is observed with a casting delay of 1 d. This means that the printed material is restraining the shrinkage of the cast concrete. Therefore, a differential shrinkage profile is observed within the cast material.
In Fig.17, the shrinkage profiles are shown after 56 d of curing. In the graphs, only a linear displacement can be observed. As observed in Fig.8, the increase in shrinkage of the exposed printed specimens after 14 d is similar to that of the sealed specimens. Since the sealed specimens are not subjected to any outside conditions, it is believed that the length reduction is uniform over the cross-section of the specimens. Therefore, it is reasonable to suppose that in the measurements after 28 and 56 d, only a linear displacement is visible.
3.4 Differential shrinkage–effect of prewetting
Xue et al. [
31] demonstrated for repair mortars that prewetting the substrate, reduces its water absorption thereafter. According to different sources [
35,
36], the consequent reduction in water flow between substrate and mortar can improve the bond strength between the two materials. As a larger quantity of water is present at the interlayer, better hydration of the new mortar is possible. Additionally, a reduction in water absorption from the fresh material could be beneficial in cases of large casting delays. In case of the 3D printed formwork application, lower water uptake would lead to a lower differential shrinkage between the cast and printed concrete and would also assure enough water remains available for the hydration products to form in the fresh cast concrete. In this study prewetting was performed on the side of the printed material against which concrete is cast (cured at RH: 95% and 60% for 28 d). The specimens were exposed for 45 min to water before casting.
In Tab.4, the water absorption and strain increase results for the side exposed to water for 45 min are shown for 28-RH90-SCC-PRE and 28-RH60-SCC-PRE. The printed concrete in all cases will absorb free water. The quantity of water absorbed is dependent on the environmental conditions during curing as expected. Higher RH during curing will lead to a lower absorption rate than occurs with low-humidity curing. However, the short duration of exposure of the specimens to water seems to have no significant influence on the expansion. By performing neutron radiography on printed concrete van Der Putten et al. [
37] observed that capillary water absorption was faster in the outer regions of printed concrete than in the bulk material. When these regions take up water significantly faster than the bulk material, any developing expansion will be limited to this small outer region. As the expanding part is rather small compared to the total element, the expansion due to the water uptake will be restrained and not visible. Therefore, it is likely that the initial uptake of water would not immediately lead to an increase in observable strain.
Comparing the results of prewetting (Fig.18) with no-prewetting (Fig.11), some similarities and differences are observed. First of all, the printed concrete swells in both cases after 1 d. The swelling is the largest at the interface. In both cases, the magnitude of swelling is similar (around 150 µm/m). However, the gradients in shrinkage within the materials seem to differ on day 1. For the printed material, the gradient is significantly lower in wetted samples (33 µm/m) than in non-wetted samples (55 µm/m) at an early age. Prewetting increases the quantity of water present in the printed material. As the prewetting time is short, most of the absorbed water is stored near the interface. Due to the higher water content at the cast-printed interface, a larger moisture gradient is present. Therefore, it is likely that more water can flow toward the outside of the printed concrete than occurs for the non-prewetted specimens. This results in a higher swelling much closer to the outside and a lower shrinkage gradient will be observed. On the inside of the printed concrete, less water is absorbed by printed material as indicated by Xue et al. [
31]. Therefore, a reduction in differential shrinkage within the printed concrete is expected compared to that in non-prewetted specimens.
Specimens printed and cured at a higher RH (RH = 95%) will obtain similar results as for the once cured at a low RH (RH = 60%). However, the higher quantity of water present in the pores after curing will reduce the capillary uptake of water. As a result, the influence of prewetting will be less pronounced than for a dryer specimen. This can be seen in Fig.19.
4 Conclusions
In this experimental research, the effect of curing conditions on the differential shrinkage of 3D printed and cast concrete was evaluated. Specimens were produced with two different types of cast concrete (self-compacting and vibrated concrete), combined with 3D printed concrete. Three different delay times between printing and casting (1, 7, and 28 d) and three different curing conditions (RH: 60%; RH: 95%, and submerged) were investigated. The main conclusions of this research are as follows.
1) The moisture content of a highly shrinking printed concrete is of the main importance for the development of the shrinkage of the composite specimen. A high moisture content during casting will lead to high shrinkage of the printed concrete after casting, while a low moisture content during casting results in minimal shrinkage of the printed concrete.
2) Printed concrete cured at lower RH for longer periods swells during casting. This swelling is most pronounced at the interface between cast and printed concrete. Swelling is a result of absorption by the printed concrete of moisture from the cast concrete.
3) Printed concrete cured at higher RH or submerged in water shows a linear shrinkage profile throughout the specimen depth.
4) The casting delay has a significant influence on the differential shrinkage between the two materials. Determining the optimal casting delay based on the shrinkage behavior of the printed material is recommended to reduce the differential shrinkage. Large casting delays tend to make the formwork a restraining factor, while for small casting delays, it is the cast material that tends to restrain.
5) Prewetting of samples cured at lower RH decreases the differential shrinkage at early ages within the printed material. Due to a lower water absorption, the material will swell less than when the specimen is dry.
Note that the results and conclusions are based on one tested printed concrete which is highly prone to shrinkage. In the case of printable concrete mixtures with limited shrinkage, different behavior could be expected. Suggestions for further research is the design of a model, taking into account the hydroscopic behavior of both printed and infill concrete. Results from such a model could improve the practical implementation of 3D printed concrete as a formwork in the construction industry.
PDF
(2771KB)
