The freezing and thawing (F–T) is one of the main threats for concrete in cold regions. The migration and phase transformation of moisture inside concrete induced by the F–T cycles is the origin of surface scaling and internal damage, which can induce the deformation of the matrix. The deterioration process of concrete can be further accelerated with the deicing salt. Previous studies usually followed the standardized methods. The deformation of concrete during the F–T cycles is free, which is different from the concrete in a service state. The concrete specimen is only equivalent to one unit in the concrete member. All concrete units undergo a deformation subjected to the F–T cycles, limiting the free deformation of adjacent concrete units. The limitation between adjacent concrete units is a form of restraint. Some restraint devices are developed to limit the deformation of concrete during F–T cycles. Conventional concrete frost resistance tests underestimate surface damage but overestimate internal damage due to the lack of consideration for restraint. However, the impact of restraint on the deterioration of concrete pore structure and the initiation/propagation of microcracks due to F–T cycles is unclear. Characterizing the internal damage of concrete is an important way to investigate the mechanism of concrete frost damage. The common method is to use digital image processing (DIP) to analyze the deterioration of concrete microstructure induced by F–T cycles, which has the advantages of convenient sample preparation and unrestricted sample size. In recent years, DIP methods have been also combined with morphological image processing to form a method of segmenting, extracting, and quantitatively analyzing for pores and microcracks that rely on "shape factors" to identify pores and microcracks. This paper was to focus on the restrained and unrestrained concrete as research objects. The pore structure and microcrack parameters were statistically analyzed after the single-side salt F–T test. The residual strain of concrete was calculated. The comparative analysis of the differences between two types of concrete in terms of pore structure deterioration and microcrack initiation was carried out to elucidate the influence of restraint on the deterioration of concrete pore structure and microcracks. The results could be expected to guide the mix design of frost resistant concrete in cold regions, ensuring the long service life of concrete materials.

P. I 42.5 cement, natural river sand with a fineness modulus of 2.68, and graded stone by uniformly mixing the crushed stone with 5–10 mm and 10–20 mm in a mass ratio of 4∶6 were used to prepare fresh concrete at a water/cement ratio of 0.60. The 28 d cube compressive strength of the concrete specimen was 28.3 MPa. Two groups of moulds were used to cast concrete specimens. For unrestrained specimens, cylindrical plastic moulds with a diameter of 100 mm and a height of 70 mm were used. For restraint specimen, a restraint device was used. The curing and pre saturation of concrete specimens were carried out according to the salt freezing method in the CIF test and the GB/T 50082—2009. The CDF/CIF TESTER produced by Schleibinger Geräte was used to provide single-side salt F–T cycles. After each 4 F–T cycles, the DIP sample was prepared. The processing of digital images included shooting, binarization, segmentation and extraction of region of interest (ROI), and skeleton. The obtained pore parameters included pore diameter distribution, most probable pore diameter (d_M), mean pore diameter (d), standard deviation of pore diameter (d_σ), and percentage of damaged pores (P_dam). The obtained microcrack parameters included the mean length (L_mean) and mean width (W_mean) of microcracks.

The d_M, d, d_σ and P_dam of both restrained and unrestrained concretes increase with the F–T cycles. After the 32 F–T cycles, the d_M, d, d_σ, and P_dam of the restrained concrete are 11%, 10%, 5%, and 7% smaller than those of the unrestrained concrete, respectively. The results indicate that the restraint limits the expansion of the pores due to the F–T cycles. During the cooling process, the restraint weakens the water migration inside concrete via limiting the shrinkage deformation of concrete, thereby reducing the pore diameter shrinkage induced by water migration, the saturation of larger pores and thereby reducing the crystallization pressure. The restraint reduces the pore expansion and microcrack initiation of restrained concrete. The of restrained and unrestrained concretes increases with the F–T cycles. After the 32 F–T cycles, the L_mean of restrained concrete is 7% less than that of unrestrained concrete. The W_mean of unrestrained concrete decreases, but the W_mean of restrained concrete increases with the increase of F–T cycles. The restraint changes the evolution of microcracks in concrete due to the F–T cycles, resulting in macroscopic differences in the frost damage of the two types of concretes. Furthermore, after the 32 F–T cycles, the S_n of restrained concrete is 30% less than that of unrestrained concrete. The grey correlation degree shows that the pore expansion and microcrack length induced by F–T cycles are the main factors leading to the increase of residual strain. A correlation between the expansion of pore diameter inside restrained concrete and the increase in residual strain is greater, which is related to the weakening of water migration under the restraint.

The restraint had a significant inhibitory effect on the pore diameter expansion and microcrack initiation of concrete subjected to single-side salt freezing and thawing. The restrained concrete was to release stress through the expansion of microcrack width, being a key mechanism for reducing internal damage induced by the single-side salt freezing and thawing cycles. The main reason for the increase in residual strain of concrete induced by the F–T cycles could be the expansion of pore diameter and the growth of microcrack length.

The rapid adoption of 3D printed cement-based materials (3DPC) in construction industry, particularly in sustainable buildings, highlights a critical importance of understanding its shrinkage characteristics. Different from conventional concrete, 3DPC is characterized due to the absence of coarse aggregates, the lack of templates support during printing, and immediately exposure to environmental conditions, which make it more susceptible to shrinkage-induced cracking. Shrinkage in cement-based materials leads to internal stresses and potential durability issues, affecting a long-term structural integrity. It is thus to investigate the shrinkage behavior of 3DPC. In particular, the effect of material compositions (i.e., water-binder ratio (W/B), sand-binder ratio (S/B), and supplementary cementitious materials such as fly ash (FA) and ground granulated blast furnace slag (GGBS)) on the shrinkage (total shrinkage, autogenous shrinkage, and drying shrinkage) is not fully clarified. This study was to investigate the shrinkage characteristics of 3DPC and compare it with other cement-based materials (i.e., UHPC and cast mortar).

To investigate the impact of material composition on 3DPC shrinkage, a series of laboratory experiments were conducted by a standardized contact shrinkage testing method. Four key factors evaluated were W/B, S/B, FA replacement ratio, and GGBS replacement ratio. These material variables were selected based on their known influence on the microstructure and shrinkage behavior of traditional cement-based materials.

Total shrinkage (TS), autogenous shrinkage (AS), and drying shrinkage (DS) were measured at specific intervals to capture early-age shrinkage behavior in 14 d. TS was calculated as a sum of AS and DS to represent the overall dimensional changes of the material for over time. The degree of influence of each material factor on the shrinkage was determined by grey relational analysis (GRA). GRA is a statistical method that enables the quantification of the relationship between multiple variables in complex systems, making it particularly suitable for determining the relative significance of material compositions in the context of 3DPC shrinkage.

For comparative purposes, the shrinkage characteristics of conventional cast mortar and UHPC were also analyzed under identical conditions, enabling a comprehensive evaluation of the shrinkage performance of 3DPC.

The results reveal the distinct shrinkage patterns based on the material composition of 3DPC. Increasing the W/B can significantly reduce AS. However, the influence of W/B on DS is less than that on AS. Also, the S/B has a measurable effect on drying shrinkage. After an initial curing period of 3 d, a higher S/B ratio leads to a marked reduction in DS likely due to the presence of fine sand particles, which restricts the movement of water within the matrix and limit capillary stresses.

Besides, the replacement of cement with FA has a beneficial effect on shrinkage reduction. The 14 d AS and DS both are significantly lower in 30% FA-modified 3DPC, compared to the control sample. Also, the incorporation of FA reduces the 14 d autogenous shrinkage and drying shrinkage of 3DPC in the same proportion. In contrast, the use of GGBS has a more complex influence on shrinkage. While GGBS replacement increases AS after 2 d, it reduced drying shrinkage before 5 d. The TS of 3DPC firstly decreases and then increases with the increase of GGBS dosage, and the TS of 3DPC is the minimum value when the dosage of GGBS is 10%.

The grey relational analysis shows the factors influencing 3DPC shrinkage in an order of significance, i.e., W/B, S/B, GGBS and FA. This highlights a predominant role of water content in governing both autogenous and drying shrinkage in 3DPC systems, which is consistent with those in other cement-based materials. However, the relative importance of sand and supplementary materials indicates that a further optimization of the mix design can lead to an improved shrinkage performance. Comparing the shrinkage behavior of 3DPC with that of UHPC and cast mortar, the shrinkage magnitudes of 3DPC are generally higher. These results indicate a necessity of tailored shrinkage control strategies in 3DPC, especially in large-scale applications where cracking risk is a major concern.

This study provided a comprehensive evaluation of the shrinkage behavior of 3D printed cementitious materials, especially the influence of material compositions such as W/B, S/B, FA, and GGBS. Increasing the W/B of 3DPC significantly reduced AS more than DS. The DS decreased with increasing S/B after 3 d. The replacement of FA proportionately decreased the 14 d AS and DS of 3DPC, whereas the replacement of GGBS replacement ratios increased AS after 2 d and DS before 5 d. The influence of material composition on 3DPC shrinkage was ranked in a decreasing order of W/B, S/B, GGBS, and FA. Compared to other cement-based materials like UHPC and cast mortar, 3DPC exhibited a higher shrinkage, indicating the need for more focused research and innovation in shrinkage reduction techniques for 3DPC. These findings contributed to a deeper understanding of the material behavior and provided a foundation for a future work on optimizing 3DPC for large-scale and crack-resistant applications.

Compared with the conventional moulded concrete process, the layer-by-layer stacking process produces a large number of interlayer interfaces in 3D printed concrete (i.e., 3DPC), leading to a low flexural and tensile strength of 3DPC and a poor structural integrity of 3D printed concrete structure, which restricts the development and large-scale application of concrete 3D printing technology. Reinforcement is considered as one of the most effective measures to improve the flexural and tensile properties of 3D printed structures and enhance the structural integrity. Various reinforcement techniques for 3DPC, such as fiber reinforcement and transverse reinforcement, are developed, but all of them fail to effectively reinforce the interlayers. A longitudinal reinforcing approach involving short-cut straight and n-shaped bars was firstly proposed in 2018, and some work showed that longitudinal reinforcing method of n-shaped bar could enhance the mechanical properties of 3DPC.

In this paper, the influences of the penetration degree of the n-shaped bars and the overlapping conditions on the flexural properties of 3DPC were investigated, and the damage process of 3DPC was analyzed via crack displacement and section morphology. The different reinforcing methods were compared.

Portland cement P·I 42.5 according to Chinese standard and natural river sand with a fineness modulus of 3.4 as a fine aggregate were used. Hydroxypropyl methylcellulose (HPMC) with an apparent viscosity of 198 Pa·s was selected as a viscosity enhancing agent and a bauxite-based powdered accelerator with a fineness (80 μm square-hole sieve residue) of 9% was used to adjust the thixotropy and buildability of 3DPC. A polycarboxylic acid-based high-performance superplasticizer with a solid content of 39% was used to adjust the flowability of the printed materials. In this work, a ratio of the layer number penetrated by n-shaped bar (L_P) to the total layer number (L_P) was defined as a penetration degree (P) including overlapped penetration degree P_ovr and nonoverlapped penetration degree P_non. Eight types of n-shaped bar layouts that meet the constraint requirements of the formula were investigated.

To ensure the accuracy of the n-shaped bar overlapping and positioning, a manual reinforcing method was used to accurately locate and reinforce the n-shaped bars. This process was repeated for ten layers of concrete, resulting in the completion of 3DPC specimens reinforced by n-shaped bars. For the unreinforced specimens, the preparation was completed via directly printing ten layers in a row.

After printing, the specimens with the sizes of 40 mm×200 mm×180 mm were placed in an indoor environment for 24 h and then transferred to a standard curing room for 28 d. In the curing period, the specimens underwent a three-point bending test using a CMT-300 universal pressure testing machine. The span between the specimen supports was 160 mm, and the loading was controlled using mid-span displacement mode at a loading speed of 0.01 mm/min. The flexural strength of each group with three specimens was obtained via calculating the average value.

When P_non=0.1, the addition of n-shaped bars fails to strengthen the 3DPC, and reduces the flexural strength of 3DPC. When P_non=1.0, the maximum flexural strength reaches 5.88 MPa, which is only slightly greater than that of the unreinforced 3DPC. Under the condition of nonoverlapped, the n-shaped bar is not effectively connected at the interface, and the n-bar has little enhanced effect.

P_ovr=0, Compared to the unreinforced 3DPC as P_ovr=0, the flexural strength of reinforced 3DPC increases when P_ovr=0.2, with an increase of 100%. As the penetration degree increases, the flexural strength increases by 128% when P_ovr=0.3 and by 152% when P_ovr=0.9. The flexural strength is highly improved by an overlapped n-shaped bar. The damage mode of overlapped n-shaped bar is mainly a ductile failure with obvious strain hardening characteristics.

The influence weights of f_t0, P and ρ of flexural strength of reinforced 3DPC are 0.896 2, 0.836 0 and 0.766 3, respectively. Based on the analysis of these influence weights, the advantages and disadvantages of the two reinforcement methods, i.e., the rivet method and chopped bar method, are evaluated. To ensure a fair comparison, each parameter is normalized to the same order of magnitude and unit, and transformed into equivalent flexural strength. The chopped bar and n-shaped bar (overlapped) methods both have the similar reinforcing effects, with equivalent flexural strengths of 27.74 MPa and 26.25 MPa, respectively, based on the proposed equivalent flexural strength formula.

The interface defects at the shoulder and bottom of the n-shaped bar were increased when the n-shaped bar was nonoverlapped, having little effect on the flexural strength. All the reinforced 3DPC exhibited a brittle damage. When the n-bar was overlapped, the flexural strength of reinforced 3DPC increased with the penetration degree, but there was a saturation value for the reinforcement of flexural strength by penetration degree, and there was a most economical penetration degree in the actual project. The damage mode of 3DPC reinforced by n-shaped bar under overlapped condition was a ductile failure with strain hardening characteristics. Based on the gray entropy system theory, the flexural strength of reinforced 3DPC (f_t) of the influence weight was obtained in a decreasing order, i.e., matrix flexural strength (f_t0)＞penetration degree (P)＞reinforcing ratio (ρ). To improve the flexural strength of current longitudinal reinforcement methods, the chopped bar and overlapped n-shaped bar were proven to be more effective rather than the rivet and nonoverlapped n-shaped bar techniques. It was essential to carefully choose an appropriate reinforcing method corresponding to the specific requirements of the printing project.