This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
Water soluble polymers such as cellulosic ethers or starch ethers are often included in the mix-design of Self Compacting Concretes (SCCs) in order to improve their stability and robustness. The stability, including resistance to liquid-solid separation and sedimentation, may be attributed to the increase of the viscosity of the liquid phase due to the thickening effect of the polymer. The later is then referred to as a Viscosity-Modifying Admixture (VMA). In the present study, we consider the influence of VMAs on the rheological properties of the material at cement scale level. In particular, the change in the thixotropic properties of the cement paste due to the inclusion of VMA is investigated. It is found that addition of VMA significantly enhances rebuild-up kinetics at rest following shearing at high shear-rate. The influence of VMA on the steady state rheological properties is also considered. As already reported in the literature, the yield stress is found to monotonically increase with VMA content, while the consistency presents a minimum indicating the existence of an optimum value of the VMA for which the workability of the cement paste is maximum.
Keywords: cement paste; self-compacting concretes; thixotropy; viscosity-modifying admixture.
(*)Author for correspondence: [email protected]
Viscosity-modifying admixtures (VMAs) are often used in mix-design of highly fluid cementitious materials, including self-compacting concretes (SCCs), pumpable concretes, etc., to avoid solid-liquid separation and to improve the robustness of the formulation. That is the primary objective when using these admixtures. However, one can expect some influence on the rheological behaviour since addition of VMA will change the rheology of the aqueous phase and the interactions between the solid particles. This problem has been investigated in the literature, in particular by considering the combination of VMAs and Superplastizers (SPs) [1-7]. Phan et al.  showed that the influence of VMA on the rheological properties of cement pastes is actually small compared to that of SPs. This has been explained by the fact that VMAs act mainly on the liquid phase, while SPs modify the granular configuration (deflocculation), which actually dominates the rheological behaviour in such a highly concentrated suspension.
The reported results concerning the influence of VMAs on the rheological properties of cementitious materials (the studies concerned in general cement pastes) converge to the conclusion that addition of this admixture decreases the fluidity (including yield stress and viscosity) of the material (see for instance  and references therein). The results presented here show that the influence of VMA on the rheological parameters is actually more complex, due the fact that this admixture may play an ambiguous role: on one hand it may increase the liquid phase viscosity, which would lead to an increase of the paste's viscosity, and on the other hand, it may lubricate the contacts between the solid particles, which would lead to a decrease of the viscosity.
A number of authors pointed out that self-compacting concretes (SCCs) show highly thixotropic properties [8-10], without however reporting clear experimental results. Thixotropy is often a crucial property for building materials. Thixotropy may reveal advantageous, and sometime necessary, for the application process of the material. This is the case for instance for paints or rendering mortars. On the other hand, thixotropy may be undesirable for instance when dealing with the process of successive concreting. The influence of mix-design of such material on their thixotropic behaviour is then a crucial issue. In the present study this rheological property is quantitatively investigated. We consider both microstructure breakage at a high shear rate and rebuild up at rest.
2. EXPERMENTAL PROCEDURE
The cement pastes are composed of tap water, Portland cement (CEM I 52,5 PM ES CP2 from Teil in France), fillers (Ground limestone with a similar granular size distribution than that of the cement), a Polycarboxylate-type Superplastizer (Glenium 27 from Degussa) and a polysaccharide-type viscosity-modifying admixture (Foxcrete from AVEBE). The role of the SP is to disperse the cement particles that are prone to aggregation due to colloidal interactions. In the case of the SP used here the dispersion is due to both steric and electrostatic effects of the adsorbed polymer. It is shown here that the cement pastes (even without VMA) are shear thinning, indicating that we do not have complete dispersion of the aggregation for the SP dosage rate used here.
The composition of the cement paste is reported in Table 1. The composition reported in this table corresponds to the reference paste, which is used in practice to proportion self-levelling concretes. Here we consider the change in the rheological properties when varying the concentration by weight of the VMA. Then, 4 other pastes are prepared by increasing or decreasing the VMA dosage rate. The 5 cement pastes considered are then: REF (reference paste), REF-50 (obtained by dividing the VMA dosage rate by 2), REF-100 (without VMA), REF+50 (obtained by increasing the reference dosage rate by 50%) and REF+100 (obtained by increasing the reference dosage rate by 100%).
Table 1 : Composition of the cement paste [6-7]
SuperPlastizer (SP) (g)
The mixing procedure consists of the same set of steps for all the pastes considered in order to improve the reproducibility of the tests. A laboratory paddle-pan mixer is used. The total duration of mixing is quite high (11.5 min) to insure homogeneous suspensions. The mixing procedure is described in Table 2.
Table 2 : Mixing procedure
Mixing at low speed (95 RPM)
Mixing at high speed (165 RPM)
Duration of mixing (min)
2.2 Rheological measurements
The rheological measurements were performed using a stress-controlled shear rheometer (AR2000 from TA Instruments) equipped with the vane geometry (Fig. 1). In such geometry, the tested material is not subjected to a uniform shear rate. This condition is usually required in rheological experiments in order to measure actual material properties, and to get a simple relationship between the measured torque/rotational velocity and the shear-stress/shear-rate. Vane geometry is nevertheless recognized to be appropriate for granular suspensions such as mortars [11-12] since slippage can be avoided and the material is sheared in volume.
Figure 1 : Rheological measurement system : a) vane, b) inner cylinder.
The gap (distance between the periphery of the Vane and the outer cylinder) is 5.5 mm, which is more than an order of magnitude larger that the maximum size of the cement or filler particles (about 0.1 mm).
The shear-rate and the shear-stress are inferred from the torque and the rotational velocity of the vane by calibrating with a Newtonian fluid.
The temperature was regulated at 25°C (to within 0.1°C) thanks to a circulating water system. In order to prevent evaporation of the paste's water the measurement system was sealed.
2.2.2 Measurement procedure
All the rheological measurements were undertaken during the induction period during which the hydration rate of the cement is very low and may have non-significant influence on rheology. In order to check that this was actually the case, two successive and same rheological measurements with the same sample were performed, indicating that there was no irreversible transformation (hydration) of the material up to 2 hours. The fact that the induction period is so long is due to the presence of the organic phases, including the SP and the VMA.
Before starting measurements, the samples are pre-sheared at 200 s-1 and then hold at rest during 1 min.
The steady state rheograms (shear-stress versus shear-rate) are determined by subjecting the material to cycles of increase-decrease of shear-rate. An approximate steady state is obtained within 4-5 loops. The flow curves reported here correspond to the descendant branch of the last loop. For each paste, the rheological parameters reported here correspond to average values over at least 5 tests undertaken using a freshly prepared sample.
The transient behaviour is considered separately in details here. Figure 2 represents the flow path used to investigate the break down at high shear rates and rebuild up at rest. In order to mimic rebuilding at rest (to fit the precise definition of thixotropy), the material is subjected to a very low shear-rate (0.01 s-1) during this period.
200 s-1 during 4min
0.01 s-1 during 45 min
Figure 2: Flow path used to consider transient properties (thixotropy), including breakdown at a high shear rate and rebuild up at rest (very low shear rate).
3. RESULTS AND DISCUSSION
3.1 Thixotropic behaviour
Figure 3 represents the temporal evolution of the stress when the paste is subjected to a relatively high shear-rate (200 s-1), corresponding the breakdown of the microstructure. As it can be seen in Figure 3, the breakdown kinetics can be fairly well fitted by the sum of two exponentials in the case of pastes containing VMA. The correlation coefficient of the fit is in all cases very close to 1 (0.999…). For pastes without VMA a simple exponential decay turned out to be enough to account for the breakdown kinetics.
Figure 3 : Evolution of the stress versus time for different VMA dosage rates at a relatively high shear rate (200 s-1), representing the microstructure break down. The continuous lines correspond to the best fit with the sum of two exponentials. ( ) REF-100, ( ) REF-50, ( ) REF, ( ) REF+50, (+) REF+100.
The breakdown kinetics of the pastes devoid of VMA can be understood in terms of a competition between shear-induced breakage of cement and filler aggregates and Brownian induced aggregation of the colloidal part of the particles. In the presence of VMA, the polymer is subject, on one hand, to alignment and disentanglement under flow and, on the other hand, to relaxation towards maximum entropy due Brownian motion. Due to the very high molecular weight of the VMA polymer (order of millions), its average relaxation time would be quite high. One can then assume that the breakdown dynamics is governed by two main characteristic times. Then it is straightforward to show, using a basic thixotropy model, that the temporal evolution of the stress can be modelled by the sum of two exponentials in agreement with the experimental results reported in Fig. 3.
A large number of more or less sophisticated models for thixotropy have been reported in the literature [10-16]. To deal with thixotropy the simplest model has to contain at least a parameter l that characterises the degree of interconnection of the material s microstructure at a given time and shear rate. The precise physical meaning of l depends upon the actual microstructure evolving under flow and that has significant effects on the measured rheological property (yield stress, apparent viscosity, etc.). In general it is assumed that l = 0 for a fully broken down microstructure and l=1 when the microstructure is fully built up. The evolution of the parameter l is governed by the competition between the microstructure rebuild up, that takes place with a characteristic time t, and the microstructure breakdown whose kinetics can be assumed to be proportional to the shear-rate.
Our experimental results (Fig. 3), as discussed above, strongly suggest that the rebuild up kinetics would be governed by at least two different characteristic times tg and tp, corresponding respectively to the granular and the polymer (VMA) average relaxation times. The fact that we have only one characteristic time to account for breakdown kinetics of the paste without VMA supports this vision. We can then write down the following kinetic equation for the structural parameter l :
This corresponds to the simplest model for thixotropy as proposed many years ago by Moore . In particular, this simple model does not take into account eventual shear-induced aggregation. More sophisticated model can be used [18-20], however Equation (1) is enough for a qualitative interpretation of our experimental results.
For a given shear-rate, the stress s can be assumed to scale to a first approximation with the parameter l. That is: s=al=a(lg+lp), where a is a constant and the indexes g and p refer respectively to the granular and polymer contributions to the degree of microstructure interconnection. Solving Equation 1 leads then to the transient behaviour of the stress:
s=sg+sp = (2)
where the superscripts 0 and ∞ refer respectively to the initial and steady state values of the stress.
Table 3 : Characteristic times for breakdown and that of rebuild up for different dosage rates of VMA.
The best fit of the experimental stress relaxation leads to the values of the relaxation times for the different pastes considered. They are reported in Table 3. One can notice that the polymer relaxation times are an order of magnitude higher than the granular ones. This can be expected since the molecular weight of the polymer is very high. The evolution of the relaxation times for microstructure breakage is not monotonous. One can observe a minimum value for the reference paste. The reason of such evolution of microstructure breakdown kinetics as a function of VMA content is not clear. This point deserves more investigation.
Figure 5 represents a typical temporal evolution of the stress (representing the microstructure rebuild up) when the paste is hold at rest (very low shear rate) following shearing at a high shear-rate (200 s-1). The example in Figure 4 corresponds to the reference paste. The results for the other pastes are similar.
Figure 4 : Typical temporal evolution of the stress at a very small shear rate (0.01 s-1) representing the microstructure rebuild up after break down at a high shear rate (200 s-1). The continuous line corresponds to the best fit with a stretched exponential.
In contrast with breakage the rebuild up kinetics cannot be described by the sum of two exponentials. In this case a stretched exponential is more suitable to fit the stress growth curves. A stretched exponential is akin to processes involving a whole distribution of relaxation times. One can understand physically such a behaviour by the fact that since the paste is subjected to a very low shear rate a full distribution of relaxation times may be mobilized, including those corresponding to the grains and the polymer chains. During the break down process only relaxation times on the order (or smaller) of the characteristic time of flow (inverse of applied shear-rate) may be mobilized. Stretched exponential response of thixtropic fluids has already been reported in the literature .
Once again a simple model of thixotropy, including an infinite set of independent relaxation times, can be used to account for a stretched exponential behaviour of the structure rebuild up.
The best fit of the experimental curves with stretched exponentials leads the characteristic times (tr) of the rebuild up process, that is:
where and are respectively the equilibrium and the initial stresses.
The evolution of rebuild up characteristic time for different dosage rates of VMA is reported in Table 3. Our experimental results clearly show that the VMA speeds up the rebuild up properties of the paste. The physical interpretation of this result is not straightforward and needs further investigation. In practise VMAs are sometimes referred to as thixotropic agents, our experimental results confirm that in a more quantitative way.
3.2 Steady state behaviour
Figure 5 represents the steady state flow curves of the different pastes considered. The influence of VMA dosage rate is quite small. This has been attributed (see Ref. ) to the fact that the rheological behaviour of such a highly concentrated granular suspension is dominated by the granular phase, whereas VMA has influence mainly on the liquid phase. The effect of VMA dosage rate depends qualitatively upon the shear rate interval considered, in agreement with the results reported by other authors . This will be discussed in more details below when considering the rheological parameters.
Figure 6 represents the evolution of the plastic viscosity, which is defined as the derivative of the shear-stress with respect to the shear-rate, in function of shear-rate for different dosage rates. It is to be noted that we use here plastic viscosity instead of apparent viscosity (shear-stress divided by shear-rate) because it is the slope of the flow curve that determines the sensitivity of the stress to the variation of shear-rate. Since we deal with yield stress fluids the two types of viscosity are different.
Figure 5 : Flow curves for different VMA dosage rates. ( ) REF-100, ( ) REF-50, ( ) REF
( ) REF+50, ( ) REF+100
Figure 6 shows that the cement pastes containing VMA are shear-thickening (plastic viscosity increases with shear-rate) throughout the whole shear-rate interval considered. On the other hand the inserted zoom-in graph in Figure 6 shows that the rheological behaviour of the cement paste without VMA is not monotonous. This paste is shear-thinning at low shear-rates and shear-thickening at high shear-rates.
Figure 6 : Evolution of the plastic viscosity for different VMA contents : ( ) REF-100, ( ) REF-50, ( ) REF, ( ) REF+50, ( ) REF+100. The inserted graph zooms in the behaviour of REF-100 at low shear rates.
Shear-thinning behaviour is akin to flocculated suspensions and entangled polymer solutions or melts. Shear-thinning is generally attributed to shear-induced deflocculation in the former case and to polymer-chains disentanglement and alignment in the later. In our case, the shear thinning may be attributed to both phenomena.
Shear-thickening is generally attributed to repulsive interactions between both colloidal and non-colloidal particles in the case of suspensions [21-22] and shear-induced structures in polymer solutions [23-24]. Both two phenomena may contribute to shear thickening in our case since aqueous solutions of polysaccharide are known to exhibit shear thickening at sufficiently high shear-rates .
The steady state rheological parameters can be determined by the best fit with a Herschel-Bulkley (H-B) model :
where σ is the shear-stress, the shear-rate, σy the dynamic yield stress, k the consistency and n the fluidity index.
In the case where the behaviour is not monotonous (paste without VMA) the shear thinning and shear thickening zones of the flow curve are fitted separately.
The evolution of the dynamic yield stress as a function VMA dosage rate is reported in Table 4. The cement paste yield stress monotonically increases with VMA content in agreement with previous studies reported in the literature [2, 4, 5]. Such behaviour is expected and may be attributed in particular to entanglement and intertwining of the VMA polymer chains at low shear rates.
In contrast to the yield stress, the evolution of the consistency (in the shear-thickening
zone) when increasing VMA dosage rate is non-monotonous (see Table 4). The consistency first decreases when adding VMA and then increases. Such behaviour has already been reported in the literature in the case another type of a cementitious material  and attributed in particular to air-entraining effects of the polymer admixture. One can also invoke the fact that the polymer may actually play an ambiguous role. On one hand, it may increase the liquid phase viscosity leading to an increase of the paste viscosity, and on the other hand, it may lubricate the contacts between the solid particles leading to a decrease of the paste viscosity. The competition between the two effects would lead to the behaviour observed here. A similar discussion has already been reported in the case of silica suspensions and cement pastes whose composition is close to that of ours .
Table 4 : steady state rheological parameters of pastes.
Similarly to the consistency, the behaviour of the fluidity index (in the shear thickening zone) is non monotonous, presenting a maximum value around the reference paste
(see Table 4). This may also be attributed a double role that may be played by the VMA: on one hand polysaccharide aqueous solutions are known to exhibit shear-thickening at high shear-rates  and, on the other hand, its lubricating effect would decrease the contribution of the granular contacts to shear-thickening.
The rheological behaviour of cement pastes involved in mix-design of highly fluid concretes was investigated experimentally. In particular the influence of VMAs on the rheological properties was considered. Both transient (thixotropy) and steady state behaviours were investigated. Without VMA the pastes exhibited complex steady-state flow-curves, including a shear-thinning branch at low shear-rate and a shear-thickening one at high shear-rates. It was found that the shear-thinning branch was absent for pastes with VMA. The evolution of the steady state rheological parameters was found to found to be complex : the dynamic yield stress monotonically increases with VMA content, while the consistency and the fluidity index present an extremum value around the reference dosage of VMA (the one used to mix-design self-levelling concretes).
The influence of the VMA on thixotropic behaviour was investigated by considering breakdown kinetics under high rates and rebuild up at rest or very low shear-rate. It was found that the breakdown kinetics was governed by two main characteristic times (the relaxation curves could be fitted with the sum of two exponentials) which differ by an order of magnitude. This was attributed to the two different constituents of paste, namely the VMA polymer and the granular phase, whose relaxation dynamics would take place at quite different timescales.
On the other hand, the rebuild up kinetics was found to follow a stretched exponential-like process. This was attributed to the fact that, at rest or very low shear-rate, a large set of relaxation times, including those corresponding to the polymer and the grains, may be mobilized.