Effects of redispersible polymer powders on the structural build-up of3D printing cement paste with and without hydroxypropylmethylcellulose

Yi Zhang a, Zhengwu Jiang a,⇑, Yanmei Zhu a, Jie Zhang b, Qiang Ren a, Tao Huang c

aKey Laboratory of Advanced Civil Engineering Materials of Ministry of Education, School of Materials Science and Engineering, Tongji University, Shanghai 201804, ChinabWacker Chemicals (China) Co. Ltd., Shanghai 200233, Chinac Sichuan Research Institute of Building Materials, Chengdu 610081, China

h i g h l i g h t s

� Effects of typical RPPs on the structural build-up of cement pastes were studied.� RPPs show different effects on the structural build-up of cement paste under dynamic and static shear tests.� Effects of RPPs on the evolution of ultrasonic pulse velocity and shear modulus of cement pastes are similar.� In the presence of HPMC, VAE based RPP has different effects.

a r t i c l e i n f o

Article history:Received 21 May 2020Received in revised form 10 August 2020Accepted 12 August 2020Available online xxxx

Keywords:Structural build-up3D printingCement pastePolyvinyl acetate-ethylenePolyvinyl acetate-vinyl versatate-ethyleneHydroxypropyl methylcellulose

a b s t r a c t

This paper studies the effects of polyvinyl acetate-ethylene (VAE) based redispersible polymer powder(RPP) and polyvinyl acetate-vinyl versatate-ethylene (VVE) based RPP on the structural build-up of 3Dprinting cement paste within several hours of hydration. At the same time, two cases of the presenceor absence of hydroxypropyl methylcellulose (HPMC) were considered. The build-up performance ofcement paste was assessed by the hysteresis area, dynamic yield stress ratio, static yield stress, shearmodulus and ultrasonic plus velocity from the dynamic shear test, static shear test and ultrasonic wavetransmission test. In addition, the limit layer thickness and printing velocity were calculated by dynamicand static yield stresses to quantitatively characterize shape stability and printing efficiency of cementpaste. Results indicate that both VAE with higher ethylene content (VAE1) and VVE decrease the struc-tural build-up rate of cement paste, independent of shear test modes. However, VAE with lower ethylenecontent (VAE2) increases the structural build-up rate of cement paste under dynamic shear test whereasdecreases the growth rate of static yield stress. The dynamic yield stress ratio of cement paste containing2% VAE2 is 0.57, which is 83.9% higher than that of pure cement paste. In addition, HPMC alters the effectsof VAE2 on the structural build-up of cement paste. VAE1 and VVE decrease the shape stability of cementpastes with and without HPMC, while VAE2 is beneficial to the shape stability of cement paste. Limitlayer thicknesses of cement pastes with 4% VAE2 are 7.3 mm and 14 mm in the absence and presenceof HPMC, which are about 3.0 and 5.8 times that of pure cement paste. In the absence of HPMC, VVEdecreases the limit printing velocity of cement paste in the first tens of minutes. VAE1 and VAE2 canimprove the printing efficiency due to the increase of limit printing velocity in the first tens of minutes.In the presence of HPMC, only VAE2 significantly improves the limit printing velocity of cement paste.

� 2020 Elsevier Ltd. All rights reserved.

1. Introduction

There are many challenges in the construction industry, such aslow efficiency, scarcity of resources, high energy consumption and

CO2 emission [1–4]. Due to the potential in solving these chal-lenges, 3D printing concrete technology which builds objects viadeposition of cement-based materials layer upon layer hasreceived widespread attention in recent years [5–7]. Given thatthere is no mold supporting during the printing process, the yieldstress of 3D printing concrete at the moment of extruding from thenozzle (called initial yield stress, s0) should be sufficient to prevent

https://doi.org/10.1016/j.conbuildmat.2020.1205510950-0618/� 2020 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.E-mail address: jzhw@tongji.edu.cn (Z. Jiang).

Construction and Building Materials xxx (xxxx) xxx

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier .com/locate /conbui ldmat

Please cite this article as: Y. Zhang, Z. Jiang, Y. Zhu et al., Effects of redispersible polymer powders on the structural build-up of 3D printing cement pastewith and without hydroxypropyl methylcellulose, Construction and Building Materials, https://doi.org/10.1016/j.conbuildmat.2020.120551

the deformation by the flow of itself. The s0 of 3D printing concreteis dependent on the instantaneous response of structural build-up.Another key factor to achieve 3D printing of concrete is to ensurethe structural build-up rate fast enough to overcome the deforma-tion dominated by the gravity of subsequent layers [8–10]. In addi-tion, it is of great significance to match the deposition rate to thestructural build-up rate. For instance, if the structural build-up rateis too high, it will increase the difficulty of pumpability anddecrease adhesion strength between layers of 3D printing concrete[10,11]. Since aggregates are inert non-colloidal particles, thestructural build-up of 3D printing concrete is contributed bycement paste [12]. The structural build-up of pure Portland cementpaste is hard to meet printing requirements. Consequently, supple-mentary cementitious materials, setting accelerator and viscositymodifying admixtures (VMA) like cellulose ether and nanoclay,have been used to improve the structural build-up of 3D printingconcrete (Fig. 1) [13–15,54].

Redispersible polymer powder (RPP) is a type of powder madefrom polymer dispersion in water that can form a dispersion againwith basic properties comparable to the original dispersion [16–18]. RPP is generally added to cement-based material to improveits properties, such as fracture toughness, impermeability, worka-bility and adhesion strength on various substrates [19–23]. Giventhe above effects of RPP on cement-based materials, it may alsobe a potential admixture for 3D printed cement-based materials.Polyvinyl acetate-ethylene (VAE) and polyvinyl acetate-vinylversatate-ethylene (VVE) are the most widely used RPPs, occupy-ing over 80% shares in global market together with polyethylene-vinyl chloride-vinyl laurate [23]. Wang et al. [24] found that poly-vinyl acetate-vinyl versatate based RPP exhibits excellent water-reduction and water-retention effects in cement mortar. Silvaet al. [25] reported that acetate groups of VAE copolymer undergoalkaline hydrolysis and interact with Ca2+ ions of cement pastes toform an organic salt (calcium acetate). It was further demonstratedthat this interaction occurred in the first 15 min of hydration in[26]. Betioli et al. [27] found that the shear thinning behavior ofcement paste at 15 min after mixing altered to shear thickeningdue to the addition of VAE. They also found that VAE decreasesthe thixotropy of cement paste.

To date, the studies available in literature related to effects ofVAE on structural build-up are limited to non-printable cementpastes. The impact of VVE on structural build-up has not beenreported. Therefore, this study attempts to provide an insight ineffects of typical RPPs on structural build-up of 3D printing cementpastes (hereinafter referred to as cement paste) without and withHPMC at early age. The instantaneous response of structuralbuild-up in cement pastes is measured by dynamic shear test.Besides, the evolution of structural build-up in cement pastes atrest is analyzed and discussed based on static shear test and ultra-sonic wave transmission experiments.

2. Experimental program

2.1. Materials and mix proportions

The P�II 52.5 Portland cement produced by Nanjing Onoda inaccordance with the Chinese standard GB 175–2007 [28] was used.Chemical compositions of the cement are listed in Table 1. Thethickener was AD300-B hydroxypropyl methylcellulose (HPMC)

produced by Jiangsu Subote New Materials Co., Ltd., with a mois-ture content of 2.1%, ash content of 3.0%, PH value of 6.5 (1% solu-tion, 25 �C), and a viscosity of 100000 mPa�s (2% aqueous solution,20 �C). Three types of RPPs were supplied by Wacker Chemicals(China) Co., Ltd. Two of them are polyvinyl acetate-ethylene(VAE) synthesized with different ratios of ethylene and vinyl acet-ate monomers. VAE1 represents the VAE with a glass transitiontemperature (Tg) of �4 �C synthesized from a higher content ofethylene monomer (20% ± 2%). While VAE2 stands for the VAE witha Tg of 17.5 �C synthesized from a lower content of ethylene mono-mer (10% ± 2%). The third one is polyvinyl acetate-vinyl versatate-ethylene (vinyl acetate 77% ± 2%, vinyl versatate 15% ± 2%, ethylene8% ± 2%). Basic properties of RPPs are given in Table 2. Molecularstructures of RPPs and HPMC are illustrated in Fig. 2. Tap waterwas used as the mixing water.

The water-cement ratio was kept at 0.35 while the ratio of RPPto cement varies from 2% to 6%. Mix proportions are shown inTable 3.

2.2. Test methods

2.2.1. Rheological experimentRheological methods applying for structural build-up can be

divided into dynamic shear test method and static shear testmethod based on the applied shear rate (or applied shear stress)and the condition of measured cement pastes [29,30]. Becausethe applied shear rate or shear stress is so small that it causes littleor no damage to the structure of cement pastes. It can be consid-ered that there is only structural build-up in cement-based mate-rials during the static shear test. Since the cement-based materialis in a flowing state during the dynamic shear test, there is not onlystructural build-up but also structural breakdown. Obviously,dynamic shear test is suitable for characterizing the instantaneousresponse of structural build-up in cement pastes at flowing andfrom flowing to standing [12,31]. The static shear test is suitablefor evaluating the evolution of structural build-up in cement pastesover time during rest. Therefore, dynamic and static shear testswere adopted to characterize the instantaneous response of struc-tural build-up and the evolution of structural build-up in cementpastes respectively.

(1) Dynamic shear test

The cement paste samples were prepared by NJ-160 mixer. Thepaste was mixed at the rotational speed of 140 rpm for 60 s, andthen the mixer was stopped for 30 s to manually homogenize thepaste sample. This was followed by mixing at the rotational speedof 280 rpm for 60 s. After mixing, mixtures were immediatelyplaced in the sample cup. The rheological experiment of cementpaste was carried out by Brookfield RST-SST rheometer withMBT-40F-0018 sample cup and VT-50–25 vane (Fig. 3). The innerwall of the sample cup was covered with 400 mesh sandpaper toavoid slipping. The test range of shear stress was between 1.7and 1700 Pa under the selected measuring system (including vaneand sample cup). The rheological test was conducted at 20 �C usinga water-circulating device. Dynamic shear test started at 7 minafter the contact of cement and water. The shear history of cementpastes is shown in Fig. 4(a). Fig. 4(b) presents the typical shear

Table 1Chemical components of cement (wt.%).

Compound Na2O MgO Al2O3 SiO2 P2O5 SO3 Cl K2O CaO Fe2O3 TiO2 SrO LOI

Amount 0.08 0.65 4.56 20.9 0.12 2.65 0.05 0.87 65.00 3.23 0.22 0.03 1.64

Y. Zhang, Z. Jiang, Y. Zhu et al. Construction and Building Materials xxx (xxxx) xxx

2

stress-shear rate curve measured and hysteresis area underdynamic shear test.

(2) Static shear test

After the dynamic shear test, the static test started at 15 min ofcement hydration. This test was performed at a constant shear rateof 0.015 s�1 for 30 s. Static shear tests were repeated every 15 minuntil 120 min. However, if the shear stress reaches the rheometerstress limit earlier, the test stops. The typical shear stress versusshear strain curve is shown in Fig. 4. The cement paste appearsas elastic deformation before point B. Thereafter, it appears as plas-tic deformation. It is generally considered that point B indicates theend of the elastic deformation, while point A indicates the begin-ning of flow [32–34]. Point A is often called yield point. The shearmodulus (G) of the cement paste was obtained by linear fitting ofthe data from 20% to 60% of peak stress, as illustrated in Fig. 5.

2.2.2. Ultrasonic wave transmission experimentThe ultrasonic wave transmission test was conducted via a

commercially available system (IP8, UltraTest GmbH, Germany)at 22 �C ± 3 �C (Fig. 6). The mixing method of cement paste sampleswas the same as that in rheological experiment. Samples were cast

Table 2Basic properties of RPPs.

RPPs Code Appearance Protective collides Particle size/lm Moisture content/wt.% PH Tg/�C MFT/�C

Vinyl acetate-ethylene VAE1 Free-flowing,white powder

PVOH Median:1 <1 6–8 �4 0

VAE2 Free-flowing,white powder

PVOH Median:1 <1 6–8 17.5 3

Vinyl acetate-vinyl versatate-ethylene

VVE Free-flowing,white powder

PVOH Median:1 <1 6–8 13 0

Table 3Mix proportions of 3D printing cement pastes (g).

NO. P�II 52.5 VAE1 VAE2 VVE HPMC Water

Ref. 500 0 0 0 0 175VAE1-2 500 10 0 0 0 175VAE1-4 500 20 0 0 0 175VAE1-6 500 30 0 0 0 175VAE2-2 500 0 10 0 0 175VAE2-4 500 0 20 0 0 175VAE2-6 500 0 30 0 0 175VVE-2 500 0 0 10 0 175VVE-4 500 0 0 20 0 175VVE-6 500 0 0 30 0 175H-Ref. 500 0 0 0 1 175H-VAE1-4 500 20 0 0 1 175H-VAE2-4 500 0 20 0 1 175H-VVE-4 500 0 0 20 1 175

Fig. 1. 3D printing cement paste incorporating HPMC.

Fig. 2. Schematic illustration of molecular structure of polymers: (a) VAE (R = H), VVE (R = H or CH3), (b) HPMC (R = H or CH3 or CH2CH(OH)CH3).

Y. Zhang, Z. Jiang, Y. Zhu et al. Construction and Building Materials xxx (xxxx) xxx

3

into silicon molds with an inner diameter of 50 mm and a height of50 mm. A transmitter and a receiver were placed with a mutualdistance of 40 mm on each side. An ultrasonic pulse was transmit-ted every 1 min during the hydration until 200 min.

3. Results

3.1. Instantaneous response of structural build-up under dynamicshear test

3.1.1. Effects of VAE1 based RPPFig. 7 and Fig. 8 show shear stress-shear rate curves and appar-

ent viscosity-shear rate curves of cement pastes modified by VAE1.It can be found that cement pastes without HPMC behave as typicalyield-pseudoplastic fluids, indicated by positive intercepts of shearstress-shear rate curves and the decreasing apparent viscosity withshear rate. VAE1 slightly decreases the shear stress at lower shearrate, while it increases the shear stress at the shear rate higherthan 10 s�1, as presented in Fig. 7. Apparent viscosity of cementpastes with VAE1 is higher than that of pure cement paste at each

shear rate (Fig. 8). Additionally, the shear stress and apparent vis-cosity are increasing with VAE1. The apparent viscosity decreasescontinuously in the acceleration phase due to the higher structuralbreakdown rate than the structural build-up rate. In the decelera-tion phase, the apparent viscosity increases continuously owing tothe lower structural breakdown rate than the structural build-uprate. Because the structure cannot be restored as before in a shorttime, the shear stress of acceleration curve is higher than that ofdeceleration curve, resulting in the hysteresis (thixotropic) area.The hysteresis area is usually used for evaluation of structuralbuild-up in cement paste (thixotropy of cement paste). It shouldbe noted that the hysteresis area represents the difference betweenthe energy absorbed by structural breakdown and the energy con-sumed by structural rebuild-up, i.e. the stored energy which willcontribute to rebuild-up of structure in cement pastes. The hys-teresis area depends on the degree of structural breakdown andthe rate of structural rebuild-up in cement pastes. A larger hystere-sis area means a higher degree of structural breakdown, or a lowerstructural build-up rate, or both. Table 4 lists the dynamic yieldstress (sd) determined by linear extrapolation of the shear stress-shear rate curve to zero shear rate and hysteresis area of the shearstress-shear rate curve. It can be observed that the dynamic yieldstress in acceleration curve (sac) is higher than that in decelerationcurve (sdc) in Fig. 7. Additionally, when VAE1 is used, the hysteresisarea increases. Hysteresis areas of cement pastes incorporatedwith 2%, 4% and 6% of VAE1 are 4031.44 Pa�s�1, 3488.75 Pa�s�1

and 3897.61 Pa�s�1, exhibiting 68.6%, 45.9% and 63.0% higher com-pared with pure cement paste, respectively. The increase in hys-teresis areas indicates that VAE1 decreases the instantaneousstructural build-up rate of cement paste. Another manner todemonstrate this behavior is the decrease of the dynamic yieldstress ratio (sdc/sac) compared with pure cement paste.

The shear stress-shear rate curve of the cement paste withHPMC is different from that of pure cement paste, showing twostages in Fig. 7(a). The first stage is at the low shear rate(<60 s�1), the cement pastes with HPMC presents shear thickeningbehavior, though the apparent viscosity reduces first owing tobreakdown of the initial structure (Fig. 8(a)). The second stage isat a high shear rate (>60 s�1), cement pastes with HPMC presentshear thinning behavior. This phenomenon can be explained thatin the case of low shear rate, the HPMC adsorbed on cement parti-cles is not released, the probability of collision and adsorption

Thermostatic control system

Measuring system

Fig. 3. Brookfield RST-SST rheometer and water-circulating device.

Fig. 4. Dynamic shear test: (a) Shear history, (b) Typical curve and hysteresis area.

Y. Zhang, Z. Jiang, Y. Zhu et al. Construction and Building Materials xxx (xxxx) xxx

4

between HPMC and cement particles increase as well. Cementpastes containing 0.2% HPMC show higher shear stress and appar-ent viscosity compared with those without HPMC. It is noted thatin the presence of HPMC, the addition of VAE1 does not furtherincrease the shear stress and apparent viscosity of cement pastes.Actually, the addition of VAE1 in cement paste with HPMCdecreases the shear stress and apparent viscosity when the shearrate is below 80 s�1 in acceleration curve and below 40 s�1 indeceleration curve. Otherwise, the decrease of shear stress andapparent viscosity is not obvious over these above shear rates.HPMC increases the sdc/sac of cement pastes. Moreover, HPMCincreases the hysteresis area. The hysteresis area of cement pastewith HPMC is 9287.04 Pa�s�1, about four times that of pure cementpaste. The hysteresis area of the cement paste containing HPMCand VAE1 decreases to 7268.95 Pa�s�1. The sdc/sac decreases aswell. These results suggest that VAE1 reduces the average struc-tural rebuild-up rate of cement paste in the presence of HPMC.

3.1.2. Effects of VAE2 based RPPAs shown in Fig. 9 and Fig. 10, the shear stress and apparent vis-

cosity of cement pastes containing VAE2 increase compared withpure cement paste. Especially, the shear stress and apparent vis-cosity of cement pastes containing 4% and 6% VAE2 increase signif-icantly. VAE2 modified cement pastes also show a yield stressshear thinning behavior, yet the decrease rate of apparent viscosityof cement pastes containing 4% and 6% VAE2 doesn’t reduce obvi-ously till shear rate over 30 s�1 (Fig. 10(a)). The hysteresis area ofcement paste containing 2% VAE2 is 1143 Pa�s�1, reducing by 52.2%compared with that of pure cement paste. The anti-thixotropicbehavior of cement paste containing 4% and 6% VAE2 is observed,where the deceleration curve is above the acceleration curve(Fig. 9). This phenomenon can be explained by the fact that shear-ing promotes temporary aggregation rather than breakdown dueto the collision of these attractive particles in the cement paste[35,36]. These results suggest that the higher instantaneous struc-tural rebuild-up rate of cement paste with VAE2 compared withpure cement paste. Another evidence is the increased sdc/sac ofcement paste with VAE2. Even though both VAE1 and VAE2improve the shear stress and apparent viscosity of cement pastes,effects of VAE2 on hysteresis area and sdc/sac of cement pastes areopposite to those of VAE1. This phenomenon may be attributed tothe different acetate group contents between VAE1 and VAE2.Since the acetate groups undergo alkaline hydrolysis will form car-

Fig. 5. Typical shear stress-shear strain curve and the calculation range of shearmodulus.

Signal transceiver systemUltrasonic wave transmission system

(b)

(c)

(a)

(d)

Fig. 6. Ultrasonic wave transmission test of 3D printing cement pastes conductedby IP8 system: (a) Sample, (b) Transmitter, (c) Receiver, (d) Temperature sensor.

Fig. 7. Shear stress-shear rate curves of cement pastes modified by VAE1: (a) Acceleration curve, (b) Deceleration curve.

Y. Zhang, Z. Jiang, Y. Zhu et al. Construction and Building Materials xxx (xxxx) xxx

5

boxylate ions that can combine with the Ca2+ ions of cement pastesby ionic bond [25,37]. Therefore, the rate and degree of interactionbetween VAE2 (with higher acetate group content) and cementparticles will be higher than those of VAE1.

The addition of VAE2 does not change the profile of flow curveof cement paste in the presence of HPMC, as shown in Fig. 9. In thepresence of HPMC, the incorporation of VAE2 does not lead the

cement paste to exhibit anti-thixotropic behavior. On the contrary,the hysteresis area increases slightly. Besides, the sdc/sac of cementpaste with HPMC hardly changes after the addition of VAE2(Table 5). Unlike VAE1, the addition of VAE2 increases the shearstress and apparent viscosity of cement pastes in the presence ofHPMC. The shear thinning stage also occurs earlier in cement pasteincorporated with HPMC and VAE2.

Fig. 8. Apparent viscosity-shear rate curves of cement pastes modified by VAE1: (a) Acceleration curve, (b) Deceleration curve.

Table 4Rheological parameters of VAE1 modified cement pastes measured by dynamic shear test.

No. Dynamic yield stress/Pa Hysteresis area/Pa/s

sac sdc sdc/sac

Ref. 157.10 47.99 0.31 2391.56VAE1-2 160.60 32.92 0.20 4031.44VAE1-4 132.95 35.10 0.26 3488.75VAE1-6 124.09 33.15 0.27 3897.61H-Ref. 476.50 224.57 0.47 9287.04H-VAE1-4 277.10 114.49 0.42 7268.95

Fig. 9. Shear stress-shear rate curves of cement pastes modified by VAE2: (a) Acceleration curve, (b) Deceleration curve.

Y. Zhang, Z. Jiang, Y. Zhu et al. Construction and Building Materials xxx (xxxx) xxx

6

3.1.3. Effects of VVE based RPPExcept in the shear stress-shear rate acceleration curve of

cement paste containing 6% VVE, the shear stress and apparent vis-cosity of cement pastes with VVE is lower than those of purecement paste, as shown in Fig. 11 and Fig. 12. This phenomenonindicates the increase of fluidity of cement paste with VVE underthe effect of shear. The shear stress of cement paste containing

6% VVE is higher than that of pure cement paste below the shearrate of 40 s�1, probably due to higher dosage VVE leading to ahigher amount of absorbing water and lower particle space inthe cement paste. The hysteresis area of cement paste with VVEis higher than pure cement paste, especially when the content ofVVE is 6% (Table 6). Furthermore, the sdc/sac decreases significantlywhen the VVE is added into the cement paste. The sdc/sac of cement

Fig. 10. Apparent viscosity-shear rate curves of cement pastes modified by VAE2: (a) Acceleration curve, (b) Deceleration curve.

Table 5Rheological parameters of VAE2 modified cement pastes measured by dynamic shear test.

No. Dynamic yield stress/Pa Hysteresis area/Pa/s

sac sdc sdc/sac

Ref. 157.10 47.99 0.31 2391.56VAE2-2 203.05 115.80 0.57 1143.00VAE2-4 264.43 129.13 0.49 �9667.65VAE2-6 271.45 96.08 0.35 �11308.73H-Ref. 476.50 224.57 0.47 9287.04H-VAE2-4 481.52 227.45 0.47 9501.87

Fig. 11. Shear stress-shear rate curves of cement pastes modified by VVE: (a) Acceleration curve, (b) Deceleration curve.

Y. Zhang, Z. Jiang, Y. Zhu et al. Construction and Building Materials xxx (xxxx) xxx

7

paste containing 6% VVE is 0.07, only about a fifth of the purecement paste. These results suggest the negative effect of VVE onthe instantaneous structural build-up rate of cement paste.

VVE does not alter the flow curve profile of cement paste in thepresence of HPMC, as presented in Fig. 11. Nevertheless, the addi-tion of VVE results in the decrease of shear stress and apparent vis-cosity of cement paste incorporated with HPMC. Furthermore, inthe presence of HPMC, VVE reduces the average rate of structuralrebuild-up of cement paste which is proved by the decrease ofsdc/sac.

3.2. Evolution of structural build-up under static shear test

Static shear stress-time curves of cement pastes measured areshown in Fig. 13. It is observed that the static yield stress (ss)increases with resting time, indicating the growth of internal struc-ture in cement pastes. Additionally, the time to reach the peakvalue increases with resting time during the test, indicating theenhanced deformation resistance of cement pastes [29]. It can alsobe found that the shear stress of cement pastes with RPP is lowerthan that of pure cement paste at the same time. Compared withpure cement paste, the number of tests of cement paste withVAE1 or VVE increases within the test range of the rheometer.These results all demonstrate that the structural build-up rate ofcement paste decreases when RPPs are used.

As illustrated in Fig. 13, the time of cement paste with HPMC toreach the peak shear stress increases obviously compared to thepure cement paste. Moreover, HPMC affects the profile of the shearstress-time curve, resulting in a smoother peak. These changesmean that HPMC improves the deformation resistance of cementpastes.

It can be found that RPPs play different roles on the shear stress-time curve when they are added in cement paste incorporated withHPMC. The shear stress of cement paste with both HPMC and VAE1is lower than that only with HPMC, and the number of shear testsincreases to seven. A similar phenomenon is observed in thecement paste with HPMC and VVE. However, effects of VAE2 onthe shear stress development of cement paste with HPMC can bedivided into two stages. In the first stage, the addition of VAE2reduces the shear stress of the cement paste within the first hourof standing. In the second stage, the shear stress of the cementpaste with HPMC and VAE2 has a sharp rise after one hour ofstanding, even surpassing the cement paste only with HPMC.

3.2.1. Effects of VAE1 based RPPStatic yield stress (ss) and shear modulus (G) of VAE1-modified

cement pastes varied with time are presented in Fig. 14. Asexpected, the ss of cement pastes increases with resting time.The initial (15 min) static yield stress (ss0) of the cement paste sin-gly doped with VAE1 almost equals that of pure cement paste.However, the ss evolution of VAE1 modified cement paste is signif-icantly slower than that of pure cement paste, causing the differ-ence between the ss of the two cement pastes to increase withtime (Fig. 14(a)). The slower evolution of ss suggests that the struc-tural build-up rate of cement paste decreases due to the addition ofVAE1. It can also be found that the incorporation of HPMC resultsin both a higher ss0 and a lower structural build-up rate of cementpaste. The ss0 of cement paste singly doped with HPMC is overtwice that of pure cement paste. The increase of ss0 of cement pastecontaining HPMC can be attributed to interaction between theHPMC and cement particles, i.e. a relatively rigid network formeddue to the adsorption of polymer onto particles [38,39]. However,

Fig. 12. Apparent viscosity-shear rate curves of cement pastes modified by VVE: (a) Acceleration curve, (b) Deceleration curve.

Table 6Rheological parameters of VVE modified cement pastes measured by dynamic shear test.

No. Dynamic yield stress/Pa Hysteresis area/Pa/s

sac sdc sdc/sac

Ref. 157.10 47.99 0.31 2391.56VVE-2 121.74 22.44 0.185 3382.89VVE-4 137.14 14.97 0.11 2937.25VVE-6 211.44 14.0 0.07 7435.42H-Ref. 476.50 224.57 0.47 9287.04H-VVE-4 396.87 159.07 0.40 8666.39

Y. Zhang, Z. Jiang, Y. Zhu et al. Construction and Building Materials xxx (xxxx) xxx

8

the adsorption of cellulose ether on the surface of cement particlesreduces the contact between cement particles, thus slowing downthe formation of C-S-H network [40]. In addition, the ss0 of cementpaste incorporated with HPMC and VAE1 is lower than that ofcement paste singly doped with HPMC. This phenomenon may

relate to the competitive adsorption between HPMC and VAE1[41].

In Fig. 14(b), it can be seen that the G of cement paste contain-ing VAE1 and cement paste containing HPMC decreases comparedto that of pure cement paste. In addition, the G of cement paste

Fig. 13. Static shear stress-time curves of cement pastes measured at 0.015 s�1.

Fig. 14. Rheological parameters of VAE1-modified cement pastes varied with time: (a) Static yield stress, (b) Shear modulus.

Y. Zhang, Z. Jiang, Y. Zhu et al. Construction and Building Materials xxx (xxxx) xxx

9

with HPMC and VAE1 is the lowest among all the cement pastes.This phenomenon indicates that VAE1 decreases the rigidity ofcement paste, i.e. the shear resistance of cement paste, regardlessof the presence of HPMC. It can be explained that the absorptionof VAE1 on cement particles prevents the flocculation of cementparticles and delays the cement hydration. Though, the G ofcement pastes generally increases with time. It is interesting tonote that the G of pure cement paste and cement paste withVAE1 reaches the peak value at 30 min. However, the G of cementpaste containing HPMC increases with time continuously.

3.2.2. Effects of VAE2 based RPPIn Fig. 15, despite little effect on the ss0 of cement paste, VAE2

delays the development of ss0 over time. Besides, the decrease inthe G of the cement paste with VAE2 indicates that VAE2 reducesthe rigidity of the cement paste. It can be observed that there isa sharp rise in the modulus-time curves of cement pastes afterone hour except for that of pure cement paste (Fig. 15(b)).

Though the ss0 of cement paste containing HPMC is muchhigher than that of pure cement paste, the initial shear modulus(G0) of it is much lower than the G0 of pure cement paste. VAE2has little effect on the ss0 and G of cement paste in the presenceof HPMC in the first hour of hydration. However, VAE2 increasesthe development rate of ss0 and G of cement paste in the presenceof HPMC after one hour of hydration.

3.2.3. Effects of VVE based RPPAs can be seen from Fig. 16(a), VVE decreases the ss0 of cement

paste in the presence of HPMC. However, it has little effect on thess0 of cement paste without HPMC. It can also be seen that VVEslows the development of ss0 of cement pastes without and inthe presence of HPMC. As illustrated in Fig. 16(b), the G of cementpaste singly doped with VVE develops in fluctuation and shows arising trend in general. And the G of cement paste doped withHPMC and VVE increases with time. In addition, the G of cementpastes without and with HPMC decreases when the VVE is addedin.

3.3. Evolution of structural build-up under ultrasonic wavetransmission test

Ultrasonic wave propagation through materials relies on thevibration. The vibration deformation of a material depends on its

elastic mechanical properties, which are related to its microstruc-ture [42,43]. Ultrasonic pulse velocity is often utilized to character-ize the build-up of structure in cement pastes at rest [44,45]. Thewave acceleration therefore can reflect the structural build-up rateof cement paste.

The test results of cement pastes are demonstrated in Fig. 17. Asexpected, the ultrasonic pulse velocity (Vp) of cement pastesincreases with time during the test, indicating the developmentof structure in cement pastes (Fig. 17(a) and (b)). The Vp of cementpastes with polymers is lower than the velocity of sound in the air(340 m/s at 20 �C), which is attributed to the air entrapped incement pastes by polymers [44,46]. In accordance with the Vp ofcement paste from high to low, the order is pure cement paste,VAE1 modified cement paste, VAE2 modified cement paste, VVEmodified cement paste and HPMC modified cement paste. In addi-tion, all RPPs decrease the Vp of cement pastes in the absence andpresence of HPMC. This phenomenon indicates that RPPs decreasethe stiffness of cement pastes. The air-entraining effect of RPPs isdemonstrated as well.

It can be seen that the wave acceleration of pure cement pastedecreases with time. On the contrary, the wave acceleration ofcement pastes singly doped with one of the three RPPs developsin fluctuation first and then increases with time (Fig. 17(c)). Thisdevelopment trend can be explained that the absorption of poly-mers on cement particles inhibits the flocculation and reaction ofcement particles. Additionally, the wave acceleration of cementpaste containing HPMC is much lower than that of pure cementpaste. The wave acceleration of cement paste containing HPMCdecreases further as RPP is used (Fig. 17(d)).

Though the temperature of cement pastes varies between 27 �Cand 29 �C during the test, the development of temperature appearsto proceed in two stages, as illustrated in Fig. 17(e) and (f). Thetemperature drops in the first stage, followed by rising in the sec-ond stage. These two stages may correspond to the induction andacceleration period of hydration, respectively. It is found that theend of the induction period of cement pastes with RPPs is delayedcompared with the pure cement paste, which is consistent with[47]. This phenomenon demonstrates the retarding effect of RPPs.However, in the presence of HPMC, only the VAE2 delays the endof the induction period of cement paste. Though the induction per-iod of cement pastes doped with the other two RPPs has not beenprolonged and even shortened instead. The temperature of cementpastes doped with the other two RPPs during the induction period

Fig. 15. Rheological parameters of VAE2-modified cement pastes varied with time: (a) Static yield stress, (b) Shear modulus.

Y. Zhang, Z. Jiang, Y. Zhu et al. Construction and Building Materials xxx (xxxx) xxx

10

Fig. 16. Rheological parameters of VVE-modified cement pastes varied with time: (a) Static yield stress, (b) Shear modulus.

Fig. 17. Ultrasonic wave transmission test results: (a) and (b) Evolution of ultrasonic pulse velocity in the absence and presence of HPMC, (c) and (d) Evolution of waveacceleration in the absence and presence of HPMC, (e) and (f) Evolution of temperature in the absence and presence of HPMC.

Y. Zhang, Z. Jiang, Y. Zhu et al. Construction and Building Materials xxx (xxxx) xxx

11

is lower than that of cement paste singly doped with HPMC (Fig. 17(f)).

4. Discussions

4.1. Effects of RPPs on the instantaneous response of structural build-up

4.1.1. Instantaneous response of structural build-upThe three RPPs have different effects on the microstructure of

pure cement paste, leading to differences in flow behavior undershearing, i.e. different profiles of the acceleration curve (Fig. 7(a)–12(a)). Generally, RPP can improve the fluidity of cement-based material due to its ball bearing action [17]. Nevertheless,the acetate groups of RPP will be hydrolyzed in alkaline mediumto produce carboxylate ion (RCOO�). RCOO� combines with Ca2+

by ionic bond to form Ca2+ bridge cross-linking structure on thesurface of C-S-H gel or Ca(OH)2 crystal [25,37]. This Ca2+ bridgecross-linking structure will cause the increase of flow resistancein cement pastes. VAE2 with higher acetate group content has ahigher degree of hydrolysis. Therefore, it has a higher rate anddegree of interaction with cement particles, which results in ahigher structuring rate under dynamic shear test. Since the versa-tate group is difficult to hydrolyze in alkaline medium [48], thestructural build-up rate of cement paste with VVE is the lowest.It is noted that a similar profile of deceleration curves can be attrib-uted to the full destruction of the microstructure in cement pasteduring the accelerated shear stage. Although the profile of flowcurve does not alter with RPPs, the three RPPs play different roleson the rheological performance of cement pastes in the presence ofHPMC. HPMC can complex with Ca2+ to form cross-linking on thesurface of cement particles in the form of covalent bonds [41].Therefore, the different effects of RPPs may be attributed to the dif-ferent competitive adsorption between HPMC and RPPs. Effects ofRPPs on rheological properties and structural build-up of cementpastes are summarized in Table 7. It can be found that effects ofVVE on rheological properties of cement pastes are independentwith the presence of HPMC. However, in the presence or absenceof HPMC, effects of VAE1 and VAE2 on rheological properties ofcement pastes are not consistent.

There are various rheological models describing the flow behav-ior of cement pastes. Common models including Bingham model,modified Bingham model and Herschel-Bulkley (H-B) model wereused to fit the test data. The coefficient of determination (R2) isusually used to evaluate the fitting results. As the increase of inde-pendent variables in the model, the R2 increases. This may lead tomisjudge the model with more independent variables as the bestmodel. Therefore, R2 is only effective for comparative analysis ofmodels with the same number of independent variables. In viewof this phenomenon, the adjusted R2 was proposed [49,50], whichcan be used not only for the comparison of goodness of fit betweenmodels with the same independent variable but also for the com-parison between models with different independent variables. Theadjusted R2 values of different rheological models are listed inTable 8. It can be seen that the adjusted R2 of Bingham model islower than the other two, suggesting the worse goodness of fit

by the Bingham model. The adjusted R2 obtained from shearingdata fitting of cement paste incorporated with VAE1 is obviouslyhigher than that of pure cement paste. However, the adjusted R2

of cement pastes incorporated with other RPPs decreases. It isinteresting to note that the flow behavior index (n, a fitting param-eter of H-B model) of group VAE2-4 and VAE2-6 is higher than 1,which is usually used as evidence to prove that the cement pasteis a dilatant fluid. It may be able to partially explain the behaviorof the apparent viscosity drop rate reduction (Fig. 10). These resultsindicate that the incorporation of polymers does affect the internalmicrostructure of the cement paste and its response to shear.

4.1.2. Limit layer thicknessThe shape stability of cement pastes after extrusion relying on

the instantaneous structural build-up, which in return relates tothe dynamic yield stress and its recovery. The gravity of cementpastes is the load needed overcoming to keep the shape of cementpastes after extrusion [8,9]. The relation between dynamic yieldstress and gravity is shown as Eq. (1).

sd P mg=A ¼ qvh=A ¼ qgh ð1Þ

hl ¼ sd=qg ð2Þ

where A is the section area (m2), q is the apparent density ofcement paste (kg/m3), h is the layer thickness (m), m is the weightof cement paste (kg), v is the volume of extruded cement paste (m3),g is the gravity constant (here is 9.8 m�s�2), hl is the limit layerthickness (m).

The apparent density (q) of cement paste was tested by a grad-uated cylinder of 500 ml. It can be found that all RPPs decrease theapparent density of cement pastes (Fig. 18). This decrease effect isalso positively related to the content of RPPs. The apparent densityof cement pastes decreases significantly when RPPs and HPMC areadded at the same time. According to Eq. (2), the limit layer thick-ness (hl) calculated by sdc can be considered as the limit layerthickness after extrusion. It can be found that the hl of pure cementpaste is 2.4 mm, while the hl of cement paste with 0.02% HPMCincreases to 12.5 mm, about 5 times of pure cement. This is the rea-son that HPMC is usually used in 3D printing cement-based mate-rials. VAE1 decreases the hl of cement pastes, especially in thepresence of HPMC (Fig. 18(a)). In the presence of HPMC, the hl ofcement paste with 4% VAE1 is 7.2 mm, decreasing 42.4% comparedwith that of cement paste singly doped with HPMC. The decreasedhl of cement paste with VVE is also observed in Fig. 18(c). The hl ofcement paste with VVE and HPMC is only slightly lower than thatof cement paste singly doped with HPMC due to its significantlydecreased apparent density. Contrary to effects of VAE1 and VVE,VAE2 increases the hl of cement pastes whatever the HPMC existsor not (Fig. 18(b)). The hl of cement pastes with 2%, 4% and 6% VAE2increase 159.1%, 203.3% and 134.8% compared to that of purecement paste, respectively. The calculated results suggest thateffects of RPPs on the hl of cement pastes is consistent with itseffects on the dynamic yield stress, though it is influenced by theapparent density of cement pastes. Additionally, the increase ofhl of cement pastes with VAE2 in the absence and presence of

Table 7Effects of RPPs on rheological properties and structural build-up of cement pastes.

RPPs In the absence of HPMC In the presence of HPMC

sd Apparent viscosity Structural build-up rate sd Apparent viscosity Structural build-up rate

VAE1 decrease increase decrease decrease decrease decreaseVAE2 increase increase increase not obvious increase not obviousVVE decrease decrease decrease decrease decrease decrease

Y. Zhang, Z. Jiang, Y. Zhu et al. Construction and Building Materials xxx (xxxx) xxx

12

HPMC means that VAE2 is beneficial to the shape stability ofcement paste.

4.2. Effects of RPPs on the evolution of structural build-up

4.2.1. Evolution of structural build-upThere are numerous models describing the structural build-up

of cement pastes. Roussel et al. [12] introduced a linear modeland defined the increasing rate of static yield stress over time atrest as the structural build-up rate (Athix). Their model is com-monly used to describe the structural build-up of cement pastes.Despite this, it should be noted that one assumption of this linearmodel is that the flow of cement-based materials conforms to theBingham model. In addition, this model is only suitable for

describing structural build-up till to the characteristic time (tc).The irreversible evolutions in cement-based materials can beneglected prior to tc. Perrot et al. [9] reported that the structuralbuild-up is exponential after tc and proposed a model whichasymptotically tends to the Roussel’s model previous to tc. Bothof the two models were utilized for analyses of measured staticyield stresses.

Fitting results and parameters of the two models are shown inFig. 19 and Table 9. It can be found that fitting results of ss0obtained from Roussel’s model are slightly higher than that fromPerrot’s model before tc. The prediction of ss0 obtained from Rous-sel’s model is lower than the measured data after tc. However, pre-diction of ss0 obtained from Perrot’s model is similar to themeasured data. This phenomenon demonstrates that the ss0

Table 8Adjusted R2 of different rheological models.

No. Bingham model s ¼ sd þ l _c Modified Bingham model s ¼ sd þ l _cþ c _c2 H-B model s ¼ sd þ k _cn

Acceleration curve Deceleration curve Acceleration curve Deceleration curve Acceleration curve Deceleration curve

Ref. 0.925 0.904 0.979 0.957 0.988 0.997VAE1-2 0.970 0.930 0.990 0.958 0.995 0.993VAE1-4 0.908 0.920 0.988 0.946 0.998 0.990VAE1-6 0.956 0.942 0.993 0.961 0.998 0.990VAE2-2 0.834 0.871 0.826 0.973 0.839 0.994VAE2-4 0.947 0.569 0.963 0.949 0.961 (n > 1) 0.910VAE2-6 0.986 0.754 0.992 0.948 0.992 (n > 1) 0.975VVE-2 0.930 0.933 0.933 0.948 0.946 0.984VVE-4 0.521 0.940 0.679 0.965 0.669 (n > 1) 0.989VVE-6 0.208 0.953 0.306 0.964 0.179 0.987H-Ref. 0.903 0.939 0.941 0.945 0.922 0.972H-VAE1-4 0.940 0.929 0.990 0.961 0.983 0.990H-VAE2-4 0.881 0.887 0.910 0.926 0.888 0.976H-VVE-4 0.953 0.956 0.982 0.970 0.969 0.989

Note: l represents plastic viscosity. _c represents shear rate. c, k and n are constants.

Fig. 18. Limit layer thickness after extrusion and apparent density of cement pastes: (a) VAE1, (b) VAE2, (c) VVE.

Y. Zhang, Z. Jiang, Y. Zhu et al. Construction and Building Materials xxx (xxxx) xxx

13

increases exponentially after tc. From Table 9, though HPMCincreases the ss0 of cement pastes. Athix-l and Athix-e obtained fromRoussel’s model and Perrot’s model both decrease when HPMC isadded in cement pastes. These results indicate that HPMCenhances the initial structure but decreases the structural build-up rate. In addition, the influence of HPMC on the tc is not obvious.

As listed in Table 9, VAE1 and VAE2 have little effect on the ss0in the absence of HPMC. However, VVE significantly decreases thess0. In the presence of HPMC, all the RPPs decrease the ss0. Thisphenomenon can be attributed to the competitive adsorption

between RPPs and HPMC. VVE decreases the Athix-l and Athix-e ofcement pastes with and without HPMC. On the contrary, VAEbased RPPs have different effects on the Athix-l and Athix-e of cementpastes whether there is HPMC or not. The Athix-l of cement pastessingly doped with VAE based RPPs is lower than that of purecement paste. But the Athix-l increases when the VAE based RPPsand HPMC both added in cement pastes compared to that of thecement paste singly doped with HPMC. For the Athix-e, RPPs havenegative effects in the presence and absence of HPMC. Thedecrease of structural build-up rate is due to the adsorption of RPPs

Fig. 19. Experimental results fitted by Roussel’s model and Perrot’s model.

Table 9Parameters of different models describing structural build-up.

No. ss0/Pa Roussel’s model Perrot’s model

ssðtÞ ¼ ss0 þ Athix�l � t ssðtÞ ¼ ss0 þ Athix�e � tcðet=tc � 1ÞAthix - l/(Pa/min) Adjusted R2 tc/min Athix - e/(Pa/min) Adjusted R2

Ref. 153.3 17.45 0.982 110.5 13.00 0.996VAE1-4 146.7 14.82 0.967 83.0 9.94 0.990VAE2-4 157.3 7.60 0.973 47.6 5.26 0.992VVE-4 117.2 11.01 1.00 61.8 7.35 0.994H-Ref. 424.7 10.96 0.959 110.0 8.15 0.966H-VAE1-4 219.5 11.31 0.969 63.6 6.39 0.948H-VAE2-4 384.3 12.21 0.936 58.4 7.69 0.949H-VVE-4 283.8 8.43 0.924 79.7 5.79 0.975

Y. Zhang, Z. Jiang, Y. Zhu et al. Construction and Building Materials xxx (xxxx) xxx

14

on cement particles. Generally, stronger interaction between RPPsand cement particles results in a thicker adsorption layer, whichcauses the lower structural build-up rate [41,51]. Furthermore, itis worth to note that all the RPPs decrease the tc of cement pasteswith and without HPMC. In particular, VAE2 greatly shortens the tc.Therefore, the exponential model of Perrot is more suitable for thess0 evolution prediction of cement pastes with RPPs consideringthe underestimation of the ss0 by the Roussel’s model of after tc.

4.2.2. Limit printing velocityThe gravity-induced stress on the bottom layer is the largest

and gradually increases during printing. If the material with non-linear evolution of structural build-up, the static yield stress ofthe bottom layer should consist with the expression (3) to ensurethe stability of printing object [52,53,55]. Therefore, the limitprinting height of printing object can be calculated by Eq. (4) with-out considering the buckling failure of printing object, while thelimit average printing velocity can be calculated by Eq. (5).

ssðtÞ > qgHt=ffiffiffi

3p

ð3Þ

Hl ¼ffiffiffi

3p

ss=qg ð4Þ

hlv ¼ Hl=t ¼ffiffiffi

3p

ss=qgt ð5Þwhere Ht is the height of printing object at time t (m), Hl is the limitprinting height (m), hlv is the limit printing velocity (m/min).

The limit printing velocity (hlv) of cement pastes calculated byEq. (5) are shown in Fig. 20. When there is no measured staticyield stress, the prediction value by Perrot’s model is used to cal-culate the hlv. It can be found that both VAE1 and VAE2 increasethe hlv of cement pastes in the first tens of minutes, but the dis-crepancy of hlv between cement pastes with VAEs and purecement paste gradually decreases. The hlv of cement paste withVAE1 is higher than that of pure cement paste before 55 min.The hlv of cement paste singly doped with VAE2 is also higherthan that of pure cement paste previous to 25 min. The increaseof hlv is because VAEs decrease the apparent density and have lit-tle effect on the initial static yield stress, while the decrease ofdiscrepancy of hlv is due to the lower structural build-up rate ofcement pastes with VAEs. Though VVE decreases the apparentdensity of cement paste, it also significantly decreases the initialstatic yield stress, which results that the hlv of cement pastes sin-gly doped with VVE is not higher than that of pure cement pasteuntil 90 min. Until about 60 min to 75 min, the increase rate of

hlv of cement pastes with RPPs is higher than that of pure cementpaste. The hlv of cement paste with VAE1 is equivalent to that ofpure cement paste at approximately 85 min, while it is about120 min for the cement paste with VAE2. From the results, itappears that VAE1 and VAE2 are both beneficial to the efficiencyof printing after extrusion in the first tens of minutes. Addition-ally, VAE1 has less influence on the efficiency of later printingcompared with VAE2.

It can be noted that the limit printing velocity evolution can bedivided into two stages in the presence of HPMC. The hlv of cementpastes decreases with time in the first stage. This is followed by theincrease of hlv in the second stage (Fig. 20). In the first stage, the hlv

of cement pastes with HPMC is higher than that of cement pasteswithout HPMC, while the discrepancy of hlv between themdecreases with time. This phenomenon can be explained by thefact that HPMC significantly increases the ss0 but decreases thestructural build-up rate in the first stage. In the second stage,though hlv of cement pastes with HPMC increases with time, theirlimit printing velocities are lower than that of cement pastes with-out HPMC except for cement paste with HPMC and VAE2. In addi-tion, only VAE2 increases the hlv of cement paste in the presence ofHPMC, while the other two RPPs decrease the hlv of cement paste.These results suggest that the printing efficiency of cement pastescan be improved when VAE2 and HPMC are both added.

5. Conclusions

In this study, effects of typical RPPs on the structural build-up of3D printing cement pastes with and without HPMC in the first fewhours of hydration were investigated. Based on the above results,the following conclusions can be drawn:

(1) VAE1 reduces the dynamic yield stress of cement pasteswithout HPMC but improves its apparent viscosity. VAE2increases both dynamic yield stress and apparent viscosityof cement pastes in the absence of HPMC. In the presenceof HPMC, effects of VAE1 and VAE2 on the dynamic yieldstress and apparent viscosity of cement pastes are notexactly the same with cement pastes without HPMC. Never-theless, VVE decreases the dynamic yield stress and appar-ent viscosity of cement pastes whether the HPMC exists ornot.

(2) Under dynamic shear test, VAE1 and VVE decrease the struc-tural build-up rate in the absence and presence of HPMC,which results in their adverse effects on the shape stabilityof cement pastes. However, VAE2 increases the structuralbuild-up rate of cement paste without HPMC and shows lit-tle effect on the structural build-up rate of cement pasteswith HPMC. Furthermore, the cement pastes with a highdosage of VAE2 behave as anti-thixotropic fluids in theabsence of HPMC. VAE2 can improve the shape stability ofcement paste after extrusion. The limit layer thickness ofcement paste containing 4% VAE2 increases 203.3% com-pared to that of pure cement paste.

(3) Under static shear test, VAE2 and VAE1 have little effect onthe initial static yield stress. Conversely, VVE significantlydecreases the initial static yield stress of cement pastes. Inspite of different effects on the initial static yield stress, RPPsdecrease the structural build-up rate of cement paste. VAE1and VVE show similar effects on structural build up rate inthe presence of HPMC. However, VAE2 has little effect onthe structural build-up rate of cement paste containingHPMC in the first hour of hydration, then increases the struc-tural build up rate significantly. Furthermore, all the RPPsshorten the tc of cement paste with and without HPMC.

Fig. 20. Limit printing velocity of cement pastes.

Y. Zhang, Z. Jiang, Y. Zhu et al. Construction and Building Materials xxx (xxxx) xxx

15

VAE1 and VAE2 can improve the printing efficiency in thefirst tens of minutes in the absence of HPMC. In the presenceof HPMC, only VAE2 can improve the printing efficiency.

(4) Shear modulus of cement pastes decrease when RPP is addedin, which is attributed to the absorption of RPPs on cementparticles resulting in the delay of hydration. In the presenceof HPMC, VAE1 and VVE also reduce the shear modulus ofcement pastes. However, VAE2 almost does not affect theshear modulus of cement pastes in the presence of HPMC.

(5) Effects of RPPs on ultrasonic pulse velocity are agreed withtheir effects on the shear modulus measured by static sheartest. The evolution of ultrasonic pulse velocity shows thatRPPs decrease the structural build-up rate of cement pastewith and without HPMC at rest.

CRediT authorship contribution statement

Yi Zhang: Conceptualization, Methodology, Investigation, For-mal analysis, Writing - original draft. Zhengwu Jiang: Conceptual-ization, Writing - review & editing. Yanmei Zhu: Formal analysis,Writing - review & editing. Jie Zhang: Formal analysis, Writing -review & editing. Qiang Ren: Formal analysis, Writing - review &editing. Tao Huang: Formal analysis, Writing - review & editing.

Declaration of Competing Interest

We declare that we do not have any commercial or associativeinterest that represents a conflict of interest in connection withthe work submitted.

Acknowledgements

The authors gratefully acknowledge the financial supports pro-vided by National Key Research and Development Projects(2018YFC0705404), National Natural Science Foundation of China(51878480, 51678442, 51878481, 51878496), and the Fundamen-tal Research Funds for the Central Universities.

References

[1] T. Wangler, N. Roussel, F.P. Bos, T.A.M. Salet, R.J. Flatt, Digital concrete: areview, Cem. Concr. Res. 123 (2019) 105780, https://doi.org/10.1016/j.cemconres.2019.105780.

[2] W. Li, B. Dong, Z. Yang, J. Xu, Q. Chen, H. Li, F. Xing, Z. Jiang, Recent advances inintrinsic self-healing cementitious materials, Adv. Mater. 30 (17) (2018)1705679, https://doi.org/10.1002/adma.201705679.

[3] B. He, X. Zhu, Q. Ren, Z. Jiang, Effects of fibers on flexural strength of ultra-high-performance concrete subjected to cryogenic attack, Constr. Build. Mater. 265(2020), https://doi.org/10.1016/j.conbuildmat.2020.120323 120323.

[4] H. Zhao, K. Jiang, R. Yang, Y. Tang, J. Liu, Experimental and theoretical analysison coupled effect of hydration, temperature and humidity in early-agecement-based materials, Int. J. Heat Mass Transf. 146 (2020) 118784, https://doi.org/10.1016/j.ijheatmasstransfer.2019.118784.

[5] R.A. Buswell, W.R.L. de Silva, S.Z. Jones, J. Dirrenberger, 3D printing usingconcrete extrusion: 3D printing using concrete extrusion: a roadmap forresearch, Cem. Concr. Res. 112 (2018) 37–49, https://doi.org/10.1016/j.cemconres.2018.05.006.

[6] S. Lim, R.A. Buswell, T.T. Le, S.A. Austin, A.G.F. Gibb, T. Thorpe, Developments inconstruction-scale additive manufacturing processes, Autom. Constr. 21(2012) 262–268, https://doi.org/10.1016/j.autcon.2011.06.010.

[7] P. Wu, J. Wang, X. Wang, A critical review of the use of 3-D printing in theconstruction industry, Autom. Constr. 68 (2016) 21–31, https://doi.org/10.1016/j.autcon.2016.04.005.

[8] N. Roussel, Rheological requirements for printable concretes, Cem. Concr. Res.112 (2018) 76–85, https://doi.org/10.1016/j.cemconres.2018.04.005.

[9] A. Perrot, D. Rangeard, A. Pierre, Structural built-up of cement-based materialsused for 3D-printing extrusion techniques, Mater. Struct. 49 (4) (2016) 1213–1220, https://doi.org/10.1617/s11527-015-0571-0.

[10] V.N. Nerella, M.A.B. Beigh, S. Fataei, V. Mechtcherine, Strain-based approachfor measuring structural build-up of cement pastes in the context of digitalconstruction, Cem. Concr. Res. 115 (2019) 530–544, https://doi.org/10.1016/j.cemconres.2018.08.003.

[11] N. Roussel, F. Cussigh, Distinct-layer casting of SCC: The mechanicalconsequences of thixotropy, Cem. Concr. Res. 38 (5) (2008) 624–632, https://doi.org/10.1016/j.cemconres.2007.09.023.

[12] N. Roussel, A thixotropy model for fresh fluid concretes: theory, validation andapplications, Cem. Concr. Res. 36 (10) (2006) 1797–1806, https://doi.org/10.1016/j.cemconres.2006.05.025.

[13] A.V. Rahul, M. Santhanam, H. Meena, Z. Ghani, 3D printable concrete: mixturedesign and test methods, Cem. Concr. Compos. 97 (2019) 13–23, https://doi.org/10.1016/j.cemconcomp.2018.12.014.

[14] R. Arunothayan, B. Nematollahi, R. Ranade, S.H. Bong, J. Sanjayan,Development of a 3D-printable ultra-high performance fiber-reinforcedconcrete for digital construction, Constr. Build. Mater. 257 (2019), https://doi.org/10.1016/j.conbuildmat.2020.119546 119546.

[15] B. Panda, S. Chandra Paul, M. Jen Tan, Anisotropic mechanical performance of3D printed fiber reinforced sustainable construction material, Mater. Lett. 209(2017) 146–149, https://doi.org/10.1016/j.matlet.2017.07.123.

[16] M.U.K. Afridi, Y. Ohama, M.Z. Iqbal, K. Demura, Water retention and adhesionof powdered and aqueous polymer-modified mortars, Cem. Concr. Compos. 17(2) (1995) 113–118, https://doi.org/10.1016/0958-9465(95)00007-Y.

[17] Y. Ohama, Polymer-based admixtures, Cem. Concr. Compos. 20 (2-3) (1998)189–212, https://doi.org/10.1016/S0958-9465(97)00065-6.

[18] Y. Pei, X. Zhang, Y. Jiang, H. Huang, H. Chen, Redispersibility of acrylatepolymer powder and stability of its reconstituted latex, J. Dispersion Sci.Technol. 32 (9) (2011) 1279–1284, https://doi.org/10.1080/01932691.2010.505799.

[19] R.u. Wang, P.-M. Wang, L.-J. Yao, Effect of redispersible vinyl acetate andversatate copolymer powder on flexibility of cement mortar, Constr. Build.Mater. 27 (1) (2012) 259–262, https://doi.org/10.1016/j.conbuildmat.2011.07.050.

[20] D.G. Walters, VAE redispersible-powder hydraulic-cement admixtures, Concr.Int. 14 (4) (1992) 30–40.

[21] M. Ramli, A. Akhavan Tabassi, Effects of polymer modification on thepermeability of cement mortars under different curing conditions: acorrelational study that includes pore distributions, water absorption andcompressive strength, Constr. Build. Mater. 28 (1) (2012) 561–570, https://doi.org/10.1016/j.conbuildmat.2011.09.004.

[22] J.V. Brien, K.C. Mahboub, Influence of polymer type on adhesion performanceof a blended cement mortar, Int. J. Adhes. Adhes. 43 (2013) 7–13, https://doi.org/10.1016/j.ijadhadh.2013.01.007.

[23] N. Tarannum, K.m. Pooja, R. Khan, Preparation and applications of hydrophobicmulticomponent based redispersible polymer powder: a review, Constr. Build.Mater. 247 (2020) 118579, https://doi.org/10.1016/j.conbuildmat.2020.118579.

[24] R.u. Wang, P.-M. Wang, Action of redispersible vinyl acetate and versatatecopolymer powder in cement mortar, Constr. Build. Mater. 25 (11) (2011)4210–4214, https://doi.org/10.1016/j.conbuildmat.2011.04.060.

[25] D.A. Silva, H.R. Roman, P.J.P. Gleize, Evidences of chemical interaction betweenEVA and hydrating Portland cement, Cem. Concr. Res. 32 (9) (2002) 1383–1390, https://doi.org/10.1016/S0008-8846(02)00805-0.

[26] A.M. Betioli, J. Hoppe Filho, M.A. Cincotto, P.J.P. Gleize, R.G. Pileggi, Chemicalinteraction between EVA and Portland cement hydration at early-age, Constr.Build. Mater. 23 (11) (2009) 3332–3336, https://doi.org/10.1016/j.conbuildmat.2009.06.033.

[27] A.M. Betioli, P.J.P. Gleize, V.M. John, R.G. Pileggi, Effect of EVA on the freshproperties of cement paste, Cem. Concr. Compos. 34 (2) (2012) 255–260,https://doi.org/10.1016/j.cemconcomp.2011.10.004.

[28] GB 175, Common Portland Cements, Beijing, China, 2007.[29] Q. Yuan, D. Zhou, B. Li, H. Huang, C. Shi, Effect of mineral admixtures on the

structural build-up of cement paste, Constr. Build. Mater. 160 (2018) 117–126,https://doi.org/10.1016/j.conbuildmat.2017.11.050.

[30] N. Roussel, Understanding the Rheology of Concrete, Woodhead, Oxford, 2012.[31] P.F.G. Banfill, D.C. Saunders, On the viscometric examination of cement pastes,

Cem. Concr. Res. 11 (3) (1981) 363–370, https://doi.org/10.1016/0008-8846(81)90108-3.

[32] N. Roussel, G. Ovarlez, S. Garrault, C. Brumaud, The origins of thixotropy offresh cement pastes, Cem. Concr. Res. 42 (1) (2012) 148–157, https://doi.org/10.1016/j.cemconres.2011.09.004.

[33] J. Kruger, S. Zeranka, G. van Zijl, An ab initio approach for thixotropycharacterisation of (nanoparticle-infused) 3D printable concrete, Constr.Build. Mater. 224 (2019) 372–386, https://doi.org/10.1016/j.conbuildmat.2019.07.078.

[34] Y. Qian, S. Kawashima, Flow onset of fresh mortars in rheometers:Contribution of paste deflocculation and sand particle migration, Cem.Concr. Res. 90 (2016) 97–103, https://doi.org/10.1016/j.cemconres.2016.09.006.

[35] H.A. Barnes, Thixotropy—a review, J. Nonnewton. Fluid Mech. 70 (1-2) (1997)1–33, https://doi.org/10.1016/S0377-0257(97)00004-9.

[36] N. Roussel, A. Lemaître, R.J. Flatt, P. Coussot, Steady state flow of cementsuspensions: a micromechanical state of the art, Cem. Concr. Res. 40 (1) (2010)77–84, https://doi.org/10.1016/j.cemconres.2009.08.026.

[37] S. Chandra, P. Flodin, Interactions of polymers and organic admixtures onportland cement hydration, Cem. Concr. Res. 17 (6) (1987) 875–890, https://doi.org/10.1016/0008-8846(87)90076-7.

[38] K.H. Khayat, Viscosity-enhancing admixtures for cement-based materials — anoverview, Cem. Concr. Compos. 20 (2-3) (1998) 171–188, https://doi.org/10.1016/S0958-9465(98)80006-1.

Y. Zhang, Z. Jiang, Y. Zhu et al. Construction and Building Materials xxx (xxxx) xxx

16

[39] C.M. Fellows, W.O.S. Doherty, Insights into Bridging Flocculation, Macromol.Symp. 231 (1) (2005) 1–10, https://doi.org/10.1002/masy.200590012.

[40] J. Pourchez, A. Peschard, P. Grosseau, R. Guyonnet, B. Guilhot, F. Vallée, HPMCand HEMC influence on cement hydration, Cem. Concr. Res. 36 (2) (2006) 288–294, https://doi.org/10.1016/j.cemconres.2005.08.003.

[41] B. Ma, Y.i. Peng, H. Tan, S. Jian, Z. Zhi, Y. Guo, H. Qi, T. Zhang, X. He, Effect ofhydroxypropyl-methyl cellulose ether on rheology of cement paste plasticizedby polycarboxylate superplasticizer, Constr. Build. Mater. 160 (2018) 341–350,https://doi.org/10.1016/j.conbuildmat.2017.11.010.

[42] K.V. Subramaniam, X. Wang, An investigation of microstructure evolution incement paste through setting using ultrasonic and rheological measurements,Cem. Concr. Res. 40 (1) (2010) 33–44, https://doi.org/10.1016/j.cemconres.2009.09.018.

[43] Z. Sun, T. Voigt, S.P. Shah, Rheometric and ultrasonic investigations ofviscoelastic properties of fresh Portland cement pastes, Cem. Concr. Res. 36(2) (2006) 278–287, https://doi.org/10.1016/j.cemconres.2005.08.007.

[44] J. Keating, D.J. Hannant, A.P. Hibbert, Comparison of shear modulus and pulsevelocity techniques to measure the build-up of structure in fresh cementpastes used in oil well cementing, Cem. Concr. Res. 19 (4) (1989) 554–566,https://doi.org/10.1016/0008-8846(89)90007-0.

[45] D.J. Hannant, J. Keating, Equipment for assessing the development of structurein fresh cement pastes by the measurement of shear modulus, Cem. Concr. Res.15 (4) (1985) 605–612, https://doi.org/10.1016/0008-8846(85)90059-6.

[46] N. Robeyst, E. Gruyaert, C.U. Grosse, N. De Belie, Monitoring the setting ofconcrete containing blast-furnace slag by measuring the ultrasonic p-wavevelocity, Cem. Concr. Res. 38 (10) (2008) 1169–1176, https://doi.org/10.1016/j.cemconres.2008.04.006.

[47] Y.u. Jin, D. Stephan, Hydration kinetics of Portland cement in the presence ofvinyl acetate ethylene latex stabilized with polyvinyl alcohol, J. Mater. Sci. 53(10) (2018) 7417–7430, https://doi.org/10.1007/s10853-018-2074-1.

[48] A. Victor, V. David, Y. Zegui, N. Lan, K. Ore, High Quality Vinyl Ester BasedBinders for Elastomeric Roof Coatings. www.abrafati2017.com.br/2013/Dados/PDF/Paper_118.pdf. Accessed 2020

[49] J.P. Barrett, The Coefficient of Determination—Some Limitations, Am. Stat. 28(1) (1974) 19–20, https://doi.org/10.1080/00031305.1974.10479056.

[50] O. Harel, The estimation of R2 and adjusted R2 in incomplete data sets usingmultiple imputation, J. Appl. Stat. 36 (10) (2009) 1109–1118.

[51] X. Kong, S. Emmerling, J. Pakusch, M. Rueckel, J. Nieberle, Retardation effect ofstyrene-acrylate copolymer latexes on cement hydration, Cem. Concr. Res. 75(2015) 23–41, https://doi.org/10.1016/j.cemconres.2015.04.014.

[52] J. Engmann, C. Servais, A.S. Burbidge, Squeeze flow theory and applications torheometry: a review, J. Nonnewton. Fluid Mech. 132 (1-3) (2005) 1–27,https://doi.org/10.1016/j.jnnfm.2005.08.007.

[53] T. Wangler, E. Lloret, L. Reiter, N. Hack, F. Gramazio, M. Kohler, M. Bernhard, B.Dillenburger, J. Buchli, N. Roussel, R. Flatt, Digital concrete: opportunities andchallenges, RILEM Tech. Lett. 1 (2016) 67–75, https://doi.org/10.21809/rilemtechlett.2016.16.

[54] Q. Ren, Z. Jiang, H. Li, X. Zhu, Q. Chen, Fresh and hardened properties of self-compacting concrete using silicon carbide waste as a viscosity-modifyingagent, Constr. Build. Mater. 200 (2019), https://doi.org/10.1016/j.conbuildmat.2018.12.163 32432.

[55] J.A. Chamberlain, S. Clayton, K.A. Landman, J.E. Sader, Experimental validationof incipient failure of yield stress materials under gravitational, J. Rheol. 47 (6)(2003) 1317–1329, https://doi.org/10.1122/1.1619376.

Y. Zhang, Z. Jiang, Y. Zhu et al. Construction and Building Materials xxx (xxxx) xxx

17