Parallel-plate rotational rheometry of cement paste: Influence of the squeeze velocity during gap positioning

Cement and Concrete Research 75 (2015) 66–74

Contents lists available at ScienceDirect

Cement and Concrete Research journal homepage: http://ees.elsevier.com/CEMCON/default.asp

Parallel-plate rotational rheometry of cement paste: Influence of the squeeze velocity during gap positioning Fábio A. Cardoso a, Alessandra L. Fujii b, Rafael G. Pileggi a,⁎, Mohend Chaouche b a b

Department of Construction Engineering, Escola Politécnica, University of São Paulo, 05508 900 São Paulo, Brazil CNRS, École Normale Supérieure de Cachan, Laboratoire de Mécanique et Technologie, 61 Avenue du Président Wilson, 94235 Cachan, France

a r t i c l e

i n f o

Article history: Received 28 November 2014 Accepted 23 April 2015 Available online xxxx Keywords: Rheology (A) Parallel-plate Cement paste (D) Phase separation Squeeze-flow

a b s t r a c t This paper reports an experimental investigation regarding the influence of the squeeze velocity, during positioning of the parallel-plate gap, on the rotational shear flow behaviour of a cement paste. The descent of the upper plate was performed using diverse speeds while the normal force generated due to the compression of the paste was recorded. The slower the plate speed, the higher the resulting normal force. This behaviour was caused by liquid– solid separation, which is more likely to occur at slow squeeze velocities. Phase separation was confirmed by assessing, via microwave drying, the water contents of the trimmed portion of the paste sample and of the portion actually subjected to rotational shear cycles. Owing to the variation of water/cement ratio induced by liquid radial migration, paste's Bingham yield stress and plastic viscosity were significantly affected by squeeze speed, and both rheological parameters presented an inversely proportional relationship with this experimental variable. © 2015 Elsevier Ltd. All rights reserved.

1. Introduction Concentric cylinders, cone-plate, and parallel-plate geometries are normally used either in rotational flow or in oscillatory modes for the rheological evaluation of concentrated suspensions. The first is used for relatively low viscous suspensions as cement slurries, whereas the parallel plate geometries are more appropriate for assessing the rheological behaviour of pastes [1–3]. Cone-plate setup has the advantage of shearing the material at a constant rate along the sample radius, while in the plate–plate geometry the shear rate varies (being zero at the centre and maximum at the edge) [1–4]. However, the former geometry also presents a main drawback for the evaluation of granular suspensions, which is the jamming of particles under the apex of the cone [1–6]. The truncated cone minimizes this effect allowing for its utilisation for suspensions [1–6], though only for a limited maximum gap, which also restricts the maximum particle size present in the suspension (due to the minimum required gap/particle size ratio of 10). On the other hand, the parallelplate geometry does not have the jamming problem; the gap can be adjusted from tens of microns to a few millimetres according to the material [1–3] or to the goal of the experiment [4,5,7,8], and it provides a large shear rate range that is adjustable by changing gap and plate diameter [1–4]. Because of its suitable features, the parallel-plate geometry has been extensively used for the rheometry of cement pastes in flow [4,6–18] ⁎ Corresponding author. Tel.: +55 1130915442. E-mail address: [email protected] (R.G. Pileggi).

http://dx.doi.org/10.1016/j.cemconres.2015.04.010 0008-8846/© 2015 Elsevier Ltd. All rights reserved.

and small amplitude oscillatory [4,16–22] modes. Most of the studies aimed to evaluate the effects of composition and admixtures [9,13–18, 20–22]; while others focused on the influence of processing parameters like mixing method [4] and temperature [17]; understanding the setting kinetics and microstructural development [11,16,17,19–22]; or the relationship between rheology of cement paste and that of concrete [7,15,18]. The technique was also utilised for the development of a reference material intended for calibration of cement paste rheometers [12]. Roughness [8] and gap distance [4,7,8] were the test parameters systematically investigated to date. Rough plates tend to increase the measured Bingham yield stress compared to the values obtained with smooth plates; on the other hand, plastic viscosity decreases with plate roughness [8]. The use of smaller gaps resulted in higher flow resistances [4,7,8], especially when the gap approached the size of the particles/agglomerates of the paste, which was considered as a limiting gap related to a sharp increase of flow resistance [4,7]. Despite the extensive employment of the parallel-plate technique for cement paste rheometry, there is no mention whatsoever in the consulted references on cement pastes about the procedure used to adjust the gap prior to the rotational or oscillatory measurements. As general rules for most materials, the documents from rheometer manufacturers suggest the use of excess material and, then, trimming the squeezed-out portion at a gap 5% higher than the measuring position to ensure a correct filling and total contact between sample and plates [1,23]. In addition, it is recommended to attempt disturbing the sample as little as possible when loading, in order to maintain its original structure [1,23–25]. This mainly consists of descending the upper plate slowly to keep low shear rate and to avoid excessive

F.A. Cardoso et al. / Cement and Concrete Research 75 (2015) 66–74

increase of the normal force and possible modification/destruction of the sample structure during squeezing [1,25]. This suggestion is worthy particularly for structured fluids, gels, and highly viscoelastic materials. Nevertheless, this may not be the best choice for concentrated suspensions of particles in the micron range like cement pastes. Actually, the squeeze-flow is a widely used rheometry technique, which has been applied for food, pharmaceuticals, polymer composites, ceramic pastes, and other concentrated suspensions [26–30]. Its use is becoming more frequent as an alternative/complementary method for cement-based pastes [16,31,32] and mortars [33–37], especially to simulate flow situations associated to geometric restrictions (extrusion, spreading, brick laying, flow through a nozzle during pumping or spraying). The geometry change inherent to the method can induce liquid–solid phase separation, because the liquid may flow radially outwards through the porous structure of packed particles (filtration or drainage), thus having a substantial influence on the squeeze-flow behaviour of concentrated suspensions [27–30,32–36]. The occurrence and intensity of phase separation depend on material characteristics (liquid viscosity and permeability of packed particles) and test parameters, mainly speed and gap [27–30,32–36]. As the oscillatory and flow parallel-plate rheometry of pastes must be preceded by a squeeze-flow, it is important to understand how this pre-test stage may affect these measurements, thus providing useful information to the development of experimental procedures more suitable for concentrated suspensions. For this reason, the objective of this work is to determine the influence of the squeeze velocity, during positioning of the parallel-plate gap, on the rotational shear flow behaviour of a cement paste. 2. Experimental

67

The Peltier plate maintained the temperature of the lower plate at 25 °C during the tests. 2.4. Sample loading The samples were moulded immediately before the tests. A metallic ring with 40 mm diameter and 2.8 mm height was used for moulding the cement paste on the lower plate (Fig. 1a). The paste was placed with a spoon and the excess material removed during levelling the sample with a spatula; the ring was then removed. The upper plate was lowered until it gently touched the flat surface of the sample at a gap distance of ≈ 2800 μm (Fig. 1b). This preparation procedure ensured that initial geometry and volume of the samples were constant for all tests. Gap positioning was performed in two steps: ❖ 2800–1050 μm: four constant squeezing speeds (50, 100, 250 and 750 μm/s) and the automatic exponential decay mode (average speed of 24 μm/s) were employed (Fig. 2). The automatic exponential mode was adopted because it is the default mode of the equipment and was used in previous publications [14,16,17,20,21]. The normal force was measured at the lower plate. At 1050 μm (Fig. 1c), the portion of the sample squeezed out from the plates was removed and placed in a small silicone container for the phase separation test. ❖ 1050–1000 μm: only the automatic exponential decay mode (average speed of 1.4 μm/s) was used, as the displacement was small and the gap had to be reached very precisely. At this stage (Fig. 1d) no further manipulation was done in the sample and, after coupling the solvent trap device on the lower plate, the system was ready to start the rotational test.

2.1. Cement paste Cement pastes with water/cement ratio of 0.40 were prepared with Portland cement type CEM I 52.5 N (Lafarge Ciments — Usine Du Havre, France) specified according to EN 197-1. Specific gravity (by Helium picnometry), BET specific surface area (by Nitrogen adsorption) and particle size distribution (by laser diffraction in deionized water) of the cement are shown in Table 1. 2.2. Mixing procedure The paste batches were prepared with 50 g of cement and 20 g of water both at room temperature of 23 °C. The materials were manually mixed with a small metallic laboratory spoon in a 150 mL plastic container for 180 s. The cement and 10 g of water were mixed for 30 s and, then, the whole paste for additional 150 s. Immediately after mixing, apparent density was measured using a 20 mL cylindrical container. The average values of fresh apparent density and air content were 1.89 g/cm3 and 2.4%, respectively. Solid concentration of the paste was 44.6 vol.% without considering the entrained air. 2.3. Rheometer The rheological measurements were performed using a shear rheometer AR2000 Ex (TA Instruments) with stainless steel parallelplate geometry with 40 mm in diameter and crosshatched surfaces.

Table 1 Physical characteristics of the Portland cement: ρ = specific gravity; SSA = BET specific surface area; D10, D50, and D90 = particle diameters associated to values of 10%, 50%, and 90% of the cumulative size distribution curve. Cement

ρ (g/cm3)

SSA (m2/g)

D10 (μm)

D50 (μm)

D90 (μm)

CEM I 52.5 N

3.10

1.05

4.1

21.2

55.0

2.5. Shear cycles The rheological evaluation consisted of two consecutive shear cycles with no rotational pre-shear step. The shear rate varied in ramp mode from 0 to 300 s−1 in 60 s and then back to 0 s−1 [5]. The total testing time was 240 s. Tests were performed at 5 and 35 min after the beginning of the mixing and the duration of the routine — from squeeze to the end of second cycle — was approximately 8 min. For each squeeze speed used for the positioning of the gap, all tests were performed three times. A different mixing batch was used for each repetition. 2.6. Phase separation evaluation In order to assess the occurrence of liquid−solid phase separation induced by squeezing the sample during loading, the water contents of the trimmed portion of the paste sample and of the portion actually subjected to the rotational shear cycles were determined. For this purpose, microwave drying was employed due to its successful utilisation on mortar samples in previous studies [35,36]. A reference portion was also used in every test for control; a similar amount of material (≈4 g) was taken from the bulk paste in the flask and placed directly in the silicone container for drying. The pastes were spread all over the container to increase surface area and avoid spalling when drying, which could invalidate the measurements due to loss of material. This problem was worse with the tested samples, as some of them presented drier and more rigid lumps than the other portions. All three portions — trimmed, tested sample, and reference — were weighed before and after microwave drying for 7 min at maximum power of the oven (Micro-ondes 23 L 800 W, MS23F301EFS — Samsung France). The period of 7 min was more than enough to reach constant mass of the samples. Additional time just heated the oven excessively without drying the samples any further.

68

F.A. Cardoso et al. / Cement and Concrete Research 75 (2015) 66–74

a) Moulding setup

b) Initial gap = 2800 μm

c)

d)

Trim gap = 1050 μm Test gap = 1000 μm

Fig. 1. Images of the moulding setup and sample loading procedure.

3. Results and discussion 3.1. Squeeze-flow Fig. 3 shows the curves of the cement pastes squeezed using the exponential automatic mode and at a constant rate of 750 μm/s, at 5 and 35 min after mixing. The force-displacement curves of the automatic tests (graphics a and c) present two distinct regions of squeeze behaviour: (i) viscous flow or plastic deformation at low loads, for which the normal force displays a slightly increase followed by a quasi-plateau; (ii) then a transition to strain hardening occurs with a very sharp raise of the normal force for small displacements of the upper plate. The first stage is dominated by the viscous/plastic response of the material, while the latter is associated to the increase of friction between the particles. Consequently, two of the samples almost reached the 50 N force limit of the rheometer before the plate got to the trimming gap of 1050 μm. In these cases, the upper plate was slightly rotated; just enough to relieve the normal force and, then, the tests proceeded as programmed. On the other hand, the curves of the samples squeezed at 750 μm/s (Fig. 3 graphics b and d) only show, besides the initial placing phase, the viscous/plastic deformation stage, in which a quasi-linear small increase of load occurred. The maximum normal forces of 2–3 N reached for the 750 μm/s samples were substantially lower (at least one order of magnitude) than the ones measured for the automatic mode. After a first look at these results, such substantial difference between the curves at low and high squeezing rates might be interpreted as a very intense shear-thinning behaviour of the cement paste. However, as mentioned in the Section 1, phase separation must be considered when analysing squeeze-flow results of concentrated suspensions. The test can generate liquid phase radial migration, which raises the solid concentration in the central region of the sample as well as the required force to keep squeezing the material [27–30, 3000

2500

Gap (μm)

Exponential mode Automatic Gap (μm) v (μm/s) 2800-2500 730 2500-2000 212 2000-1750 82 1750-1500 57 1500-1250 31 1250-1150 14 1150-1100 7 1100-1070 3 1070-1050 < 1 2800-1050 avg ≈24

(i) Automatic (ii) 50 μm/s (iii) 100 μm/s (iv) 250 μm/s (v) 750 μm/s

2000

1500

ii i v iv iii

1000 0

10

20

30

40

50

60

70

Time (s) Fig. 2. Curves of the upper plate position vs. time from 2800 to 1050 μm. The variation of the average speed of the automatic mode at diverse gap ranges is detailed.

32–36]. The strain-hardening behaviour is related to phase separation when it takes place at gaps considerably larger than the maximum particle size of the suspension being tested [35], which is the case in the present study. At slow squeeze rates, the liquid has a longer time to flow outwards than it has when fast rates are used. Therefore, the likelihood of occurrence of phase separation is expected to increase when the displacement rate of the upper plate is decreased [27–30,32–36]. The probability of liquid–solid separation also increases with gap reduction and is inevitable at some stages when the gap approaches the dimensions of the larger particles (or agglomerates) of the material. At 750 μm/s, the speed was fast enough so that the liquid did not have time to flow separately from the particles. Therefore, the samples flowed homogeneously and the result is associated solely to a viscous/ plastic response of the material [27–30,32–36]. For the paste characteristics and experiment setup used in the present work, the speed of 750 μm/s seems to be greater than the critical velocity that limits the occurrence of liquid–solid segregation [27–30,32]. It is also worth noting that the repeatability of the curves at 750 μm/s looks better than the repeatability of the curves obtained in the automatic mode. Apparently, the phase segregation not only increased the normal force considerably, but also caused a more unpredictable response, since the formation of regions with higher concentration may have occurred differently in each sample. The squeeze-flow behaviour considering all the velocities tested can be observed in Fig. 4. Results show that the results of the automatic mode and 50 μm/s are very similar, with intense strain hardening and, consequently, high normal forces reached. The curves at 100 μm/s present the strain hardening profile, but with a lower maximum force (20 N) owing to the lower liquid–solid separation tendency at a higher displacement rate when compared to the results of automatic mode and 50 μm/s. The curves of 250 and 750 μm/s are very similar, with the absence of strain hardening and maximum force below 4 N. The similarity of the results between 250 and 750 μm/s is probably caused by the rather homogeneous flow response of the paste at these speeds. The critical velocity for the occurrence of phase separation is likely located within the 250–750 μm/s range. Hence, as 750 μm/s is above this transition value, the force should become an increasing function of velocity above this value, as it is the actual case for a homogeneous fluid whatever its rheological behaviour is. This transitional behaviour had already been demonstrated for concentrated suspensions [27–30], including cement pastes [32]. 3.2. Phase separation evaluation The data of phase segregation tests in terms of water/cement ratio (w/c) of the trimmed portion of the pastes and of the samples actually subjected to the shear cycles are displayed in Fig. 5. The results prove that liquid migration occurred during the positioning of the gap and that the phenomenon was more intense at slow speeds, as reported in previous works [28,30,33,36]. For the automatic mode (average speed of 24 μm/s), the difference of w/c between trimmed and tested sample

F.A. Cardoso et al. / Cement and Concrete Research 75 (2015) 66–74

50

5

a) Automatic_5min S3

30 20 10

S1 S2

0

S1

3

S3

S1

2

S2 1 0 5

c) Automatic_35min

S3

d) 750 μm/s_35min

4

Normal force (N)

40

Normal force (N)

b) 750 μm/s_5min

4

Normal force (N)

Normal force (N)

40

69

30 20 10

S2

0 1000

1500

2000

2500

3

S3 S2

2

S1

1 0 1000

1500

Gap (μm)

2000

2500

Gap (μm)

Fig. 3. Squeeze-flow curves of the pastes tested in the automatic mode (a, c) and at 750 μm/s (b, d), 5 min (a, b), and 35 min (c, d) after mixing. The repetitions for all three samples are shown. The odd shape of the curves of sample 1 at 750 μm/s (b, d) was caused by low data acquisition rate.

was about 0.06. A clear trend towards an equal value of w/c with the increasing squeeze velocity is observed for the samples tested at both 5 and 35 min. The w/c of the samples squeezed at 750 μm/s and 5 min matched in 0.363, while at 35 min the difference between the values of trimmed portion and tested sample was 0.01. The fact that all the measured values were below the nominal w/c of 0.40 is probably associated with experimental error of the method. The 50

a) 5min 5

Normal force (N)

40

50

4

30

3

20 100

2

100 250

Aut.

10

50

Aut. 750

1 0 1000

1500

2000

2500

amount of paste for each portion subjected to drying was about only 3 g. Thus, it is reasonable to consider that any amount of water lost in contact with the rough plates, the smooth surroundings of the lower plate (for the trimmed portion) and the tool used to manipulate the samples may have had a significant impact on the quantity of water measured by microwave drying. Moreover, w/c results of the reference samples — extracted from the bulk paste and directly subjected to microwave drying (hence much less manipulation) — were 0.390 ± 0.007 and 0.386 ± 0.005 for 5 min and 35 min, respectively. The difference from the nominal w/c value of 0.400 is probably due to a small amount of water that remained in the cement as hydrated phases. Nevertheless, despite the fact that the absolute values of w/c measured are lower than the effective ones, the absolute difference of w/c between trimmed and tested samples is valid. If we consider 0.363 (w/c at 750 μm/s) as the homogeneous w/c and assumed it equal to the nominal value of 0.40, thus, the curves can be elevated by 0.037. In this manner, an estimation for the samples tested with the automatic mode, results in effective w/c values of 0.42 and 0.36 for the trimmed and tested portions, respectively. This denotes that if the

0

0.40

50

b) 35min

5

50 Aut.

100

4

30

Water / cement (-)

Normal force (N)

40

50

3

Aut. 750

2

20

250

1

10

0 1000

100 0 1000

1500

2000

2500

0.38 0.36 0.34

Trimmed_5min Tested_5min Trimmed_35min Tested_35min

0.32 0.30 10

1500

2000

2500

100

1000

Squeeze velocity (μm/s)

Gap (μm) Fig. 4. Squeeze-flow curves of the pastes tested in the automatic mode and at 50, 100, 250, and 750 μm/s: (a) 15 min; (b) 35 min. Small graphics zoom in the 0–5 N range.

Fig. 5. Phase segregation results: water/cement ratios of the trimmed portions and of the tested samples. Average values and standard deviation bars are based on three measurements. For the automatic mode, its equivalent constant velocity of 24 μm/s is adopted.

70

F.A. Cardoso et al. / Cement and Concrete Research 75 (2015) 66–74

automatic mode or a very slow speed is used to load the sample, the paste is subjected to the rotational rheological measurements with a w/c of 0.36 instead of being tested with the intended value of 0.40. In addition that is only an average value; in some parts of the squeezed sample (in particular the central region), the local w/c can be even lower. This may have a significant impact on the measured (average) shear rheological properties as discussed hereafter. 3.3. Shear cycles Fig. 6 shows the repetitions of the shear cycles for the samples loaded by the automatic mode. In the graphics (a) and (c), for the first shear cycles, it can be seen a huge and unusual difference between the up and down curves. The hysteresis was substantially large because of the shear stress response in the up curve, which started at excessively high stresses and gradually diminished as the shear rate was increased. Then, during the down curve, the shear stress-shear rate curve assumed the typical positive average slope with an almost perfect Bingham linear fashion. The behaviour observed during shear increase was associated to the breakdown of more packed, heterogeneous structures formed as arches during the squeeze step, especially in the central areas of the samples, since the water content was reduced first in these areas, and the particles got closer to each other. Furthermore, as shearing progressed, the redistribution of water within the sample may have contributed for reducing the shear stress as well. The second shear cycles are depicted in the graphics (b) and (d) of Fig. 6. The shear stress-shear rate curves were at lower stresses compared to the values of the first cycles, and with a much smaller hysteresis. The obvious effect of time on the shear stress of both cycles can be seen comparing the upper and lower graphics; the scale of the graphics at 35 min had to be increased to show all the curves. Repetitions at 5 min were good, while the 35 min curves varied considerably. Fig. 7 shows the repetitions of the shear cycles for the samples squeezed at 750 μm/s. At 5 min, two of the samples did not present any hysteresis at all during the first shear cycle (graphic a); while the curve of sample 1 showed a moderate hysteresis, but with a normal shape and significantly smaller than the ones observed for the samples 300

loaded with the automatic mode. Shear stress levels for these samples were also lower than the ones on the automatic mode. After loading at 750 μm/s, the samples remained more homogeneous and with a higher local w/c ratio than the samples on the automatic mode. At 35 min and higher stresses, only two of the three samples presented moderate hysteresis, because of some structural breakdown/thixotropy effect. In the second shear cycle, Fig. 7 (graphic b), stress levels decreased compared to the results of the first cycle and hysteresis was small for all samples. All curves presented lower shear stress values than those obtained for the samples squeezed with the automatic mode. Repeatability was good for the samples of both cycles and testing times, except for the up curves of the first cycle that varied to some extent. The results of the shear cycles considering all the squeeze velocities used for plate loading are shown in Fig. 8. The results demonstrate for both shear cycles, as well as for both testing times, consistent agreement with the w/c results of Fig. 5, which pointed out that the w/c of the tested samples increased with the squeeze speed. The shear stress-shear rate curves are ranked in correlation with the w/c ratio: the higher the w/c, the lower the shear stress levels. It is worth mentioning that the shear history of the samples was not exactly the same (thixotropy effect), since the pastes squeezed at fast speeds were subjected to higher shear rates than the ones loaded with slow speeds. Nevertheless, even if this effect was significant, it was probably erased by the 120 s of shearing during the first cycle. In addition, the pastes are shear-thinning at relatively low shear rates, which is the case under the squeeze-flow conditions considered here. Shear history effects should be opposite to that of phase separation. Consequently, it does not interfere in the analysis, especially because the results concerning the second cycle corroborate the remarks in the same way as the ones from the first shear cycle. 3.4. Shear rheological parameters For the sake of simplicity, we ignore the shear-thinning part of the rheograms at low shear rates, and consider that the rheological parameters, yield stress, and plastic viscosity, are enough to characterise 300

a) Shear cycle1_Automatic_5min

150 100

Sample1_5min

Shear stress (Pa)

200

S2_5min

50

400 300 200

S1_35min S2_35min S3_35min

0 0

100

200

Shear rate (s-1)

150 100

S1_5min S2_5min

300

S3_5min

0 500

c) Shear cycle1_Automatic_35min

100

200

50

S3_5min

0 500

Shear stress (Pa)

b) Shear cycle2_Automatic_5min

250

Shear stress (Pa)

Shear stress (Pa)

250

d) Shear cycle2_Automatic_35min

400

S1_35min S2_35min

300

S3_35min 200 100 0 0

100

200

300

Shear rate (s-1)

Fig. 6. Shear cycles of the pastes squeezed in the automatic exponential mode: (a, b) 5 min; (c, d) 35 min. The repetitions for all three samples are shown.

F.A. Cardoso et al. / Cement and Concrete Research 75 (2015) 66–74

200

a) Shear cycle1_750μm/s

150

100

Sample 1_5min S2_5min S3_5min S1_35min S2_35min S3_35min

50

0 0

100

200

Shear stress (Pa)

Shear stress (Pa)

200

71

b) Shear cycle2_750μm/s

150

100

S1_5min S2_5min S3_5min S1_35min S2_35min S3_35min

50

0 0

300

100

Shear rate (s-1)

200

300

Shear rate (s-1)

Fig. 7. Shear cycles of the pastes squeezed at 750 μm/s: (a) shear cycle 1; (b) shear cycle 2. The repetitions for all three samples are shown.

the overall rheological behaviour of the pastes. These parameters were determined using the Bingham model by applying linear fitting on the downward flow curves [6]. The flow curves were used for fitting over the entire shear rate interval and presented a R2 value ranging from 0.971 to 0.989. The points at low shear rate (0–20s− 1) were slightly below the linear fitting in all curves. Fig. 9 displays the results of the rheological parameters as a function of the squeeze speed used during the positioning of the gap. Both yield stress (graphics a and c) and plastic viscosity (b and d) decreased with the squeeze velocity, and the curves of 5 and 35 min were practically parallel to each other. Graphics (e) and (f) display, respectively, the plots of relative yield stress (τ0_v/τ0_750) and relative plastic viscosity (μ_v/μ_750) using the values obtained at 750 μm/s (τ0_750 and μ_750) as the basis to compare with the values obtained with other velocities (τ0_v and μ_v). The results indicate that the yield stress values of the samples loaded at slow speed were 30 to 50% higher than the values of the pastes loaded at 750 μm/s. The respective values for the plastic viscosity were in the 40–65% range.

Automatic v50 v100 v250 v750

300

500

a) Shear cycle 1_5min

200

100

0

0

v100 v250 v750

d) Shear cycle 2_35min

250

Automatic v50 v100 v250 v750

150

Automatic v50

300

c) Shear cycle 2_5min

200

b) Shear cycle 1_35min

300

100

250

850Pa

400

200

300

Shear stress (Pa)

In the present work, only one paste mix-design was studied and yet the variation of the rheological behaviour measured was very significant. The effects of the squeeze speed may be even more critical when the influence of the composition and/or the use of admixtures are involved. Modifications in the paste composition that change the particle size distribution (like the use of supplementary cementitious materials or dispersing admixtures) can also alter the particle packing and, consequently, its permeability and separation tendency. Admixtures that act on the viscosity of the liquid (like cellulose ether for example) certainly influence the segregation behaviour as well. Therefore, if the squeeze velocity applied to set the gap is not taken into account in the

Shear stress (Pa)

Shear stress (Pa)

400

3.5. Practical consequences

Shear stress (Pa)

500

These results imply that, if a slow squeeze velocity is applied during paste loading; both rheological parameters may be significantly overestimated, because the measured values will actually be of a paste with a lower w/c ratio than the one of the original composition.

100 50

200 150

Automatic v50 v100 v250 v750

100 50

0

0 0

100

200

Shear rate (s-1)

300

0

100

200

Shear rate (s-1)

Fig. 8. Shear cycles of the paste samples squeezed at different speeds: (a, c) 5 min; (b, d) 35 min; (a, b) cycle 1; (c, d) cycle 2.

300

72

F.A. Cardoso et al. / Cement and Concrete Research 75 (2015) 66–74

a) Yield stress-Down curve 1

100

95

35min 5min

92

91

81

80

67

62

69

63

60

53

Plastic viscosity (Pa.s)

Yield stress (Pa)

120

b) Viscosity Down curve 1

0.6

0.57 0.49 0.5

40

0.3

c) Yield stress-Down curve 2

0.36

35min 5min

0.31

d) Viscosity Down curve 2 35min 5min

90

87

88 80

77 64

60

58

66

61 51

44

40

e) Yield stress comparison

Plastic viscosity (Pa.s)

Yield stress (Pa)

0.44

0.41

0.2

100

1.6

0.45 0.40

0.4

45 120

0.52

0.50

0.5

0.48 0.43

35min 5min 0.44

0.41

0.37

0.4

0.33 0.35

0.37

0.3

0.30

C1_5min C1_35min C2_5min C2_35min

1.6

1.5

1.5

1.4

1.4

1.3

0.27

0.2

1.3

C1_5min C1_35min C2_5min C2_35min

1.2 1.1 1.0 10

100

1.2 1.1 1.0 1000

f) Plastic viscosity comparison 10

100

Squeeze velocity (μm/s)

1000

Squeeze velocity (μm/s)

Fig. 9. Bingham rheological parameters as a function of the squeeze velocity: (a, b) yield stress; (c, d) plastic viscosity. Average values and standard deviation bars are based on three measurements. (e) Relative yield stress (τ0_v/τ0_750); (f) Relative plastic viscosity (μ_v/μ_750). Graphics (e) and (f) consider values at 750 μm/s as the basis for comparison.

500

(i) Ref_up curve (iii) HEMC_up (v) EVA_up

(ii) Ref_down - Bingham fit (iv) HEMC_down - Bingham fit (vi) EVA_down-Bingham fit

(vii) Ref* - Bingham recalculated i

400

Shear stress (Pa)

testing method, the occurrence of phase separation can lead to imprecise or even incorrect conclusions. Two previous studies [14,16] that investigated the effects of admixtures on the rheological behaviour of cement pastes are good examples to be re-analysed considering the findings of the present work and the experimental similarities among the researches. All three investigations employed parallel plate rheometry using the same model of equipment, geometry, and gap, as well as, similar shear cycle ramps and w/c ratio (0.38 for [14,16]). Most importantly, the same exponential automatic mode to descend the upper plate, which is the default of the equipment, was also used. Both works [14,16] share the same pure cement paste as the reference material. Its rheogram (Fig. 10, curves i and ii) is very similar to the results of the paste samples squeezed at low speeds (automatic and 50 μm/s, Figs. 6 and 8). They present large hysteresis and very high shear stress values during the beginning of the up curve (accelerating), indicating the occurrence of phase separation and the evaluation of a paste with lower water content than the intended value. The shear cycle of the paste modified with polyethylene vinyl acetate (EVA) copolymer [14] (Fig. 10, curves v and vi) did not present such behaviour, while the one from the paste with hydroxyethyl methylcellulose (HEMC) [16] (Fig. 10, curves iii and iv) still displayed considerable hysteresis (possibly owing to some agglomeration caused by the HEMC). The down curves (Bingham fit) show that the admixtures increased

iv

iii

y = 3.5x + 112.4

300

y = 3.6x + 124.5 200

vi y = 1.9x + 76.5

ii

v

vii

100

y = 1.2x + 51.7

0 0

20

40

60

80

100

Shear rate (s-1) Fig. 10. Shear cycles of the pure (Ref) and modified (HEMC and EVA) cement pastes tested 15 min after mixing, adopted from [14,16]. Both studies have employed the same pure paste. Up curves are plotted as scattered points, while the Bingham fits of the down curves are represented as solid lines. (vii) Ref* — Bingham fit of the reference down curve recalculated using ratio values of Fig. 9, τ0_24/τ0_750 and μ_24/μ_750 for cycle 1 at 5 min.

F.A. Cardoso et al. / Cement and Concrete Research 75 (2015) 66–74

both yield stress (≈55%) and viscosity (≈87%) of the pastes to a similar extent when compared to the pure paste, Table 2. Because the admixtures can change phase separation likelihood, it is reasonable to assume that the modified pastes were less prone to segregate than the pure pastes, since the HEMC [16] increases the viscosity of the water; while the EVA [14] dispersed the cement and added further fine particles (polymer) to the overall particle size distribution of the paste, which could lead to a better particle packing. As a result, the modified pastes were probably tested with higher w/c ratios (equal or at least more similar to the nominal) than the pure paste. In a hypothetical situation that considers the extreme case in which the modified pastes does not undergo phase separation at all, whereas the pure paste does, the effects of the admixtures on the rheological parameters would be even more intense. An estimate can be drawn by recalculating the Bingham curve for the pure cement paste (Ref), by applying the ratio values of Fig. 9 (τ0_24/τ0_750 and μ_24/μ_750 for cycle 1 at 5 min) to the parameters of curve ii. The recalculated Bingham curve (vii) represents the probable rheological behaviour of the reference paste with its nominal w/c value (or water content). As the reference curve is reduced (Ref*), the effects of the admixtures become more significant; indicating an increase of ≈ 130% in yield stress and ≈195% in plastic viscosity of the pastes, which are substantially higher than the original values (Table 2). The outcome of this analysis suggests that the results published in both papers [14,16] are not compromised qualitatively, but the differences between the rheological behaviour of the pure and of the modified pastes are probably bigger than the measured ones. 4. Conclusions A rheological investigation regarding the influence of the squeeze velocity, during gap positioning of the parallel-plate geometry, on the shear rheological behaviour of a cement paste has been undertaken. The normal forces required to squeeze the paste samples down to the trimming gap of 1050 μm were higher than the instrument limit 50 N at slow speeds, including when the exponential automatic default mode of the equipment was used. On the other hand, fast squeeze velocities in the range of 250–750 μm/s resulted in maximum forces of 3 N. This difference of behaviour was caused by the occurrence of liquid–solid phase segregation at slow speeds. The phase separation phenomenon has been evaluated by quantifying the amount of liquid that migrated radially and, thus, showed that the higher the speed, the lower the phase separation induced by squeeze stage. The effective water/cement ratio of the samples subjected to the rotational shear cycles ranged from 0.36, when squeezed with the automatic mode, to 0.40 (the nominal value), when loaded at 750 μm/s. Consequently, the rheological behaviour of the paste was significantly influenced by the squeeze speed during plate loading, due to the variation of w/c ratio. Yield stress and plastic viscosity of the samples loaded in the automatic mode were, respectively, 30–50% and 40–65% higher than those of the paste samples squeezed at 750 μm/s. These numbers denote the importance of considering the squeeze speed during gap positioning in parallel-plate rheometry (either in flow or oscillatory modes) of cement pastes and other concentrated suspensions. The scenario can be even more complex when comparing Table 2 Percentage increase of plastic viscosity (μ) and yield stress (τ0) due to the use of admixtures. μ/μ Ref and τ0/τ0Ref were calculated from the original values of [14,16], while μ/μ Ref ⁎ and τ0/τ0Ref⁎ were recalculated using the new values of the reference paste estimated by applying the parameters of Fig. 9e and f (cycle 1 at 5 min). Paste

0.25% HEMC [16] 5% EVA [14]

100




μ;τ0 μ Re f; Re f
; τ0 Re f; Re f


 −1 ð%Þ

μ/μ Ref

μ/μ Ref ⁎

τ0/τ0Ref

τ0/τ0Ref ⁎

84 90

192 200

47 63

117 141

73

pastes with modifications in the composition that not only influence their rheological behaviour, but also modify their phase separation tendency. Hence, if this experimental parameter is neglected and liquid–solid separation occurs, the subsequent shear rheological parameters measured may be overestimated and possibly induce incorrect conclusions. The guideline for the development of experimental protocols for the evaluation of concentrated suspensions by parallel-plate rheometry should consider the following: (a) the squeeze step should be performed always with constant volume of material and preferably with the same geometry as well, since these parameters can affect squeezeflow and separation behaviours; (b) preliminary squeeze tests should be conducted with different speeds while the normal force is monitored: if the normal force decreases as the squeeze speed is increased, phase separation is probably taking place and the fastest possible speed should be chosen, as long it does not compromise the accuracy of the positioning of the upper plate; (c) the descent of the upper plate should be performed in two steps: firstly, the suspension should be squeezed with the chosen speed down to a gap 5% higher than the measuring gap, and then trimmed; finally, the last 5% height can be squeeze using the exponential automatic mode of the equipment to ensure that the plate is perfectly positioned at the desired gap for the rotational measurements. Acknowledgements The authors would like to thank the support from the Brazilian research funding agencies FAPESP (Grants n. 2011/00948-9, 2012/ 18952-5, and 2013/27121-2) and CNPq; and from the company Parex Group (France). References [1] TA Instruments, Ten steps to a better rheological measurement — TA rheology training seminar, http://www.chem.mtu.edu/~fmorriso/cm4655/TAInstruments/2013TA_ 10stepstogoodrheolmeasurements.pdf2005. [2] Anton-paar, Cones, plates and cylinders: spoilt for choice — Anton-Paar learning material, http://anton-paar-rheology.com/wp-content/uploads/2014/04/Cones_ plates_and_cylinders_Spoilt_for_choice.pdf2014. [3] Malvern, Gemini & CVO rheometers — accessories guide, http://www.malvern.com/ en/2010. [4] M. Yang, H.M. Jennings, Influences of mixing methods on the microstructure and rheological behavior of cement paste, Adv. Cem. Based Mater. 2 (1995) 70–78. [5] D.A. Williams, A.W. Saak, H.M. Jennings, The influence of mixing on the rheology of fresh cement paste, Cem. Concr. Res. 29 (1999) 1491–1496. [6] P.F.G. Banfill, Rheology of fresh cement and concrete, Rheol. Rev. (2006) 61–130. [7] C.F. Ferraris, J.M. Gaidis, Connection between the rheology of concrete and rheology of cement paste, ACI Mater. J. 89 (1992) 388–393, http://dx.doi.org/10.14359/2575. [8] M. Nehdi, M.A. Rahman, Estimating rheological properties of cement pastes using various rheological models for different test geometry, gap and surface friction, Cem. Concr. Res. 34 (2004) 1993–2007, http://dx.doi.org/10.1016/j.cemconres. 2004.02.020. [9] C.F. Ferraris, K.H. Obla, R. Hill, The influence of mineral admixtures on the rheology of cement paste and concrete, Cem. Concr. Res. 31 (2001) 245–255, http://dx.doi. org/10.1016/S0008-8846(00)00454-3. [10] S. Amziane, C.F. Ferraris, Cementitious paste setting using rheological and pressure measurements, ACI Mater. J. 104 (2007) 137–145. [11] G. Sant, C.F. Ferraris, J. Weiss, Rheological properties of cement pastes: a discussion of structure formation and mechanical property development, Cem. Concr. Res. 38 (2008) 1286–1296, http://dx.doi.org/10.1016/j.cemconres.2008.06.008. [12] C.F. Ferraris, Z. Li, M.-H. Zhang, P. Stutzman, Development of a reference material for the calibration of cement paste rheometers, Adv. Civ. Eng. Mater. 2 (2013) 20120003, http://dx.doi.org/10.1520/ACEM20120003. [13] H. Vikan, H. Justnes, Rheology of cementitious paste with silica fume or limestone, Cem. Concr. Res. 37 (2007) 1512–1517, http://dx.doi.org/10.1016/j.cemconres. 2007.08.012. [14] A.M. Betioli, P.J.P. Gleize, V.M. John, R.G. Pileggi, Effect of EVA on the fresh properties of cement paste, Cem. Concr. Compos. 34 (2012) 255–260, http://dx.doi.org/10. 1016/j.cemconcomp.2011.10.004. [15] D. Feys, R. Verhoeven, G. De Schutter, Why is fresh self-compacting concrete shear thickening? Cem. Concr. Res. 39 (2009) 510–523, http://dx.doi.org/10.1016/j. cemconres.2009.03.004. [16] R.G. Pileggi, A.M. Betioli, F.A. Cardoso, V.M. John, Extended rheological characterization of cement pastes: squeeze flow plus rotational rheometry, Proc. 12th Int. Congr. Chem. Cem., Montreal — Canada, 2007.

74

F.A. Cardoso et al. / Cement and Concrete Research 75 (2015) 66–74

[17] R.C.O. Romano, R.G. Pileggi, Temperature's role in the rheological behavior of cementitious pastes prepared with air-entraining admixtures, Appl. Rheol. 22 (2012) 24333–24338, http://dx.doi.org/10.3933/ApplRheol-22-24333. [18] S.H. Jang, Identification of Concrete Incompatibilities Using Cement Paste Rheology(Doctoral Thesis,) Texas A&M University, 2009. (hdl.handle.net/1969.1/ ETD-TAMU-2009-05-349). [19] L. Nachbaur, J.C. Mutin, A. Nonat, L. Choplin, Dynamic mode rheology of cement and tricalcium silicate pastes from mixing to setting, Cem. Concr. Res. 31 (2001) 183–192. [20] A.M. Betioli, P.J.P. Gleize, D.A. Silva, V.M. John, R.G. Pileggi, Effect of HMEC on the consolidation of cement pastes: isothermal calorimetry versus oscillatory rheometry, Cem. Concr. Res. 39 (2009) 440–445, http://dx.doi.org/10.1016/j. cemconres.2009.02.002. [21] R.C.O. Romano, C. Liberato, M. Montini, J.B. Gallo, M.A. Cincotto, R.G. Pileggi, Evaluation of transition from fluid to elastic solid of cementitious pastes with bauxite residue using oscillation rheometry and isothermal calorimetry, Appl. Rheol. 23 (2013) 23830–23838, http://dx.doi.org/10.3933/ApplRheol-23-23830. [22] R. Haldenwang, I. Masalova, R. Elmakk, The effect of different Portland cements on initial hydration reaction of a self-compacting concrete cement paste, Annu. Trans. Nord. Rheol. Soc. 22 (2014) 139–143. [23] Anton-Paar, Sample preparation: shaken or stirred? — Anton-Paar learning material, http://www.world-of- rheology.com/fileadmin/public/rheology/Tips_Tricks_Joe_ Flow/tips-and-tricks-joe-flow-02-sample-preparation.pdf2014. [24] Malvern, Kinexus series user manual, user man, http://www.malvern.com/en/2014 (accessed September 5, 2014). [25] TA Instruments, Normal force measurements on the AR 1000-N Rheometer PART II, (2005), http://www.tainstruments.com/2005. [26] J. Engmann, C. Servais, A.S. Burbidge, Squeeze flow theory and applications to rheometry: a review, J. Non-Newtonian Fluid Mech. 132 (2005) 1–27, http://dx. doi.org/10.1016/j.jnnfm.2005.08.007.

[27] A. Poitou, G. Racineux, A squeezing experiment showing binder migration in concentrated suspensions, J. Rheol. 45 (2001) 609, http://dx.doi.org/10.1122/1. 1366717. [28] N. Delhaye, A. Poitou, M. Chaouche, Squeeze flow of highly concentrated suspensions of spheres, J. Non-Newtonian Fluid Mech. 94 (2000) 67–74. [29] J. Collomb, F. Chaari, M. Chaouche, Squeeze flow of concentrated suspensions of spheres in Newtonian and shear-thinning fluids, J. Rheol. 48 (2004) 405, http:// dx.doi.org/10.1122/1.1645514. [30] N. Roussel, C. Lanos, Particle fluid separation in shear flow of dense suspensions: experimental measurements on squeezed clay pastes, Appl. Rheol. 14 (2004) 256–265. [31] B.H. Min, L. Erwin, H.M. Jennings, Rheological behaviour of fresh cement paste as measured by squeeze flow, J. Mater. Sci. 29 (1994) 1374–1381. [32] T.H. Phan, M. Chaouche, Rheology and stability of self-compacting concrete cement pastes, Appl. Rheol. 15 (2005) 336–343, http://dx.doi.org/10.3933/ApplRheol-15-336. [33] Z. Toutou, N. Roussel, C. Lanos, The squeezing test: a tool to identify firm cement-based material's rheological behaviour and evaluate their extrusion ability, Cem. Concr. Res. 35 (2005) 1891–1899, http://dx.doi.org/10.1016/j.cemconres.2004.09.007. [34] F.A. Cardoso, V.M. John, R.G. Pileggi, Rheological behavior of mortars under different squeezing rates, Cem. Concr. Res. 39 (2009) 748–753, http://dx.doi.org/10.1016/j. cemconres.2009.05.014. [35] F.A. Cardoso, V.M. John, R.G. Pileggi, P.F.G. Banfill, Characterisation of rendering mortars by squeeze-flow and rotational rheometry, Cem. Concr. Res. 57 (2014) 79–87, http://dx.doi.org/10.1016/j.cemconres.2013.12.009. [36] F.A. Cardoso, F.C. Lofrano, V.M. John, R.G. Pileggi, Influence of experimental parameters of the squeeze-flow test on the rheological behaviour and phase segregation of cement mortars, Annu. Trans. Nord. Rheol. Soc. 22 (2014) 79–85. [37] R. Hendrickx, The Adequate Measurement of the Workability of Masonry Mortar(Doctoral Thesis) Katholieke Universiteit Leuven — Faculty of Engineering, 2009. (bwk.kuleuven.be/bwf/PhDs/Hendrickx.).