Correlation between Paste and Concrete Flow Behavior

Transcription

1 ACI MATERIALS JOURNAL Title no. 105-M33 TECHNICAL PAPER Correlation between Paste and Concrete Flow Behavior by Julissa Hidalgo, Chun-Tao Chen, and Leslie J. Struble The correlation between cement paste rheology and concrete workability in mixtures containing a polycarboxylated acrylate ester was initially seen as poor, in that the admixture dosage required to produce full dispersion in cement paste, as determined using dynamic rheology, was approximately four times the dosage required to produce full dispersion in concrete, as determined using slump and slump loss. A preshear protocol using a highspeed blender to reduce differences in shear history between cement paste and concrete provided a much improved correlation between paste rheology and concrete workability. Adsorption isotherms verified that both pastes and concretes were at full admixture saturation when fully dispersed. Keywords: cement paste; high-range water-reducing admixture; rheology; shear; slump; yield. INTRODUCTION Concrete workability is defined as the amount of mechanical work, or energy, required to produce full compaction of the concrete without segregation 1 and is generally related to the concrete fluidity. Because cement paste is the liquid phase in concrete, its flow properties are reasonably expected to correlate with concrete workability. Rheological properties of cement paste, a viscoelastic material, include yield stress, viscosity as a function of strain rate, and dynamic shear modulus as a function of strain. For many years, the authors research group has been using dynamic rheology to study cement paste, in particular, the progressive stiffening due to cement hydration reactions 2 and the interactions between portland cement and dispersing admixtures. 3 Because paste plays a major role in concrete workability, it was assumed throughout this work that there is a direct relationship between paste and concrete flow behavior. The goal of this study was to explore that relationship in mixtures containing highrange water-reducing admixtures. The correlation between paste and concrete flow behavior is especially important because measuring rheological properties of concrete is very difficult. Slump, the standard field measurement of concrete workability, has been shown to correlate with yield stress 4 but provides no information about viscosity. Designing a rheometer to measure concrete viscosity is very complicated because the large gap required for concrete makes slip very likely, and the use of vanes to reduce slip lead to uncertainties in the rheological properties. A recent study showed a very poor correlation between several concrete rheometers currently available, probably reflecting these complications. 5 Rheological properties of cement paste, on the other hand, can be measured with considerable accuracy and precision. In particular, dynamic rheometry was found to provide a highly accurate and sensitive measure of flow behavior and allows one to measure yield stress without the slip often associated with large strains. 2 Using paste to study flow behavior offers many potential benefits over concrete, not the least of which is the ability to use dynamic rheometry. In a previous paper, Struble and Chen 4 reported only a weak correlation in flow behavior between concrete and paste, and the research reported herein was designed to explore the relationship more carefully in mixtures containing a high-range water-reducing admixture. Dynamic (oscillatory shear) measurements on paste were used to determine the viscoelastic properties and the admixture dosage response (dosage needed for full dispersion). Concrete slump and slump loss were used to determine the concrete workability and the admixture dosage response. The dosage response between paste and concrete was compared. In preliminary experiments, Struble and Chen 4 found that the cement paste and concrete did not correlate well the admixture dosages for full dispersion were much lower in concrete than in paste. It was then proposed that agreement in admixture dosage could only be obtained when the cement paste was given the same shear history as it experiences in concrete. Therefore, in the main experiments, the paste was presheared using a high stress protocol shown by Helmuth et al. 6 to reproduce the contribution of aggregates in shearing cement paste in concrete. Because slump is an uncertain measure of concrete yield stress, adsorption isotherms were also used to determine whether the admixture dosage that produced high slump and slump loss corresponded to the dosage that produced full dispersion. This work is presented in greater detail in two theses. 7,8 RESEARCH SIGNIFICANCE The development of new technologies for achieving highly workable concrete demands knowledge of paste and concrete rheology, so research in these areas is very important in concrete science and technology. Use of a high-shear preshear protocol for paste produced a direct correlation between the dosage of dispersing admixture in paste and concrete. This correlation suggests that paste rheology can be used to set the target dosage of water-reducing admixtures in concrete and to determine the nature of cement-admixture interactions, thereby using less material and taking advantage of the greater accuracy and precision of paste rheology. EXPERIMENTAL PROCEDURE Materials Several commercial portland cements were used during this study. In the preliminary experiments, Cements C1, C2, and C3 were used, which were obtained from a single manufacturer and used in previous studies. 2-4 In the main ACI Materials Journal, V. 105, No. 3, May-June MS No. M R1 received March 15, 2007, and reviewed under Institute publication policies. Copyright 2008, American Concrete Institute. All rights reserved, including the making of copies unless permission is obtained from the copyright proprietors. Pertinent discussion including authors closure, if any, will be published in the March- April 2009 ACI Materials Journal if the discussion is received by December 1, ACI Materials Journal/May-June

2 Julissa Hidalgo is an Assistant Materials Technologist at CTLGroup, Skokie, IL. She received her BS in civil and environmental engineering from the University of Puerto Rico-Mayaguez Campus, Mayaguez, Puerto Rico, and her MS in civil and environmental engineering from the University of Illinois at Urbana-Champaign, Urbana, IL. Her research interests include rheology of cement and concrete. ACI member Chun-Tao Chen is an Assistant Professor at the National Taiwan University of Science and Technology, Taipei, Taiwan. He received his BS in engineering from National Taiwan University; his MS in harbor and river engineering from National Taiwan Ocean University, Keelung, Taiwan; and his PhD in civil and environmental engineering from the University of Illinois at Urbana-Champaign. His research interests include rheology of cement and concrete and the interactions between cement and dispersing admixtures. Leslie J. Struble, FACI, is a Professor in civil and environmental engineering at the University of Illinois at Urbana-Champaign. She received her BS in chemistry from Pitzer College, Claremont, CA, and her MS and PhD in civil engineering from Purdue University, West Lafayette, IN. She is a member of ACI Committee 236, Material Science of Concrete. Her research interests involve the chemistry, microstructure, and engineering properties of cement and concrete, including rheology of cement paste and concrete. experiments, Cement B (Type I/II) was used. The compositions of all these cements are given in Table 1. The surface area of Cement B, determined using nitrogen gas adsorption, was m 2 /g (454.7 ft 2 /oz). The dispersing admixture was a commercial carboxylic acrylate ester (CAE), which met the requirements of ASTM C494, Standard Specification for Chemical Admixtures for Concrete, for Type F high-range water-reducing admixture. In the preliminary experiments, the same CAE was used that was used in a previous study. 3 In the main experiments, a different CAE was used: a liquid with a solids Table 1 Chemical composition of cements * Chemical Percent by weight constituent Cement C1 Cement C2 Cement C3 Cement B SiO Al 2 O Fe 2 O CaO MgO SO Calculated composition C 3 S C 3 A Total alkali (Na 2 O equivalent) * Provided by manufacturers. Table 2 Aggregate properties 282 Coarse aggregates Nominal maximum size 25.4 mm (1.0 in.) Unit weight (OD) 1630 kg/m 3 (102 lb/ft 3 ) Degree of absorption 1.46% Free water 0.4% Specific gravity (SSD) 2.67 Specific gravity (OD) 2.63 Fine aggregates Fineness modulus 2.63 Degree of absorption 1.52% Free water 0.54% Specific gravity (SSD) 2.62 content of 28.58% measured according to ASTM C1017, Standard Specification for Chemical Admixtures for Use in Producing Flowing Concrete. The aggregates used to make concrete were a crushed limestone and a natural siliceous sand. Their physical properties are summarized in Table 2. Paste rheology During this study, the oscillatory shear technique was used to measure the rheological properties of pastes containing variable amounts of CAE. An oscillatory shear stress was applied to the material and the resulting strain was measured, from which one can compute the storage modulus. By varying the amplitude of the stress and examining the storage modulus as a function of stress, one can determine the yield stress; on increasing the stress, the yield stress is the stress when the storage modulus decreases, typically by several orders of magnitude. The paste experiments were conducted at a fixed watercement ratio (w/c) (by mass) of Dosages of CAE varied from 0.01 to 0.70% (herein the admixture dosage is given as solid CAE relative to the mass of cement). In all experiments, zero time was taken when cement and water first came into contact. In the preliminary experiments, approximately 2 ml (0.07 fluid oz) of paste were mixed by hand for approximately 2 minutes. The paste was then transferred to the rheometer, an additional preshear applied, and rheological measurements were begun. In the main experiments, a larger volume of paste was mixed using a high-shear, water-cooled, speed-controlled blender. The blender had a minimum speed of 4,000 rpm and a maximum speed of 12,000 rpm. The protocol was based on that developed by Helmuth et al. 6 to produce paste with similar set behavior as concrete by applying a mixing shear rate of approximately 2200 s 1 for several minutes. The blender protocol was as follows: the cement, 500 grams (17.6 oz), was first sheared for 1 minute to break any agglomerates; the blender was then stopped and the appropriate amount of water and admixture solution were added within a period of 30 seconds; the paste was sheared for 30 seconds at low speed; the blender was stopped for 1.5 minutes, during which the walls were scraped with a spatula to provide uniform mixing; and the paste was sheared for another 1.5 minute at high speed. Approximately 2 ml (0.07 fluid oz) of the paste was then transferred to the rheometer, a preshear applied, and rheological measurements were begun. The rheometer used for this study was operated using stress-controlled oscillatory shear. The experiments used a Couette geometry with a 14.0 mm (0.551 in.) diameter smooth bob and a 15.4 mm (0.606 in.) diameter smooth cup, for a 0.7 mm (0.028 in.) gap between the bob and cup. The temperature was controlled at 25 C (77 F) by water that circulated around the cup. The rheometer and data acquisition were controlled by a computer. Regardless of the type of mixing, paste was presheared in the rheometer at 200 Pa (0.029 psi) for 45 seconds and then allowed to equilibrate for 300 seconds to provide a uniform starting condition, a procedure that is used routinely in these types of measurements. Oscillating stress was then applied using a frequency of 1 Hz (1 s 1 ) and an amplitude that was increased from 0 to 1000 Pa (0 to psi). ACI Materials Journal/May-June 2008

3 Concrete workability The concrete mixture was designed using the absolute volume method described in the ACI , Standard Practice for Selecting Proportions for Normal, Heavyweight, and Mass Concrete. The w/c was 0.35 the same as for the paste. The target slump was 25 mm (1.0 in.). The mixture design is summarized in Table 3. The concrete experiments used Cement B, the same as in the paste. For the preliminary experiments, concrete was prepared using a drum mixer and a slightly different procedure from the main experiments (as follows). Cement and fine aggregate were mixed for 1 minute, water and admixture solution were added, and the mortar was mixed for approximately 2 minutes, and then coarse aggregate was added and the concrete mixed for approximately 3 minutes. For the main experiments, concrete was prepared using a 0.1 m 3 (3.5 ft 3 ) pan mixer, with the mixing pan rotating at 10 rpm. The mixing procedure for concrete was based on ASTM C192-02, Standard Practice for Making and Curing Concrete Test Specimens in the Laboratory, (mixing for 3 minutes, resting for 3 minutes, and mixing for 2 minutes) except for minor variations in the mixing time and the admixture addition. Both aggregates, fine and coarse, were placed in the pan mixer and mixed for 1 minute to obtain good uniformity. To mimic the mixing procedure used for paste, the entire water and admixture solution was added at the beginning and mixed for 1 additional minute. The pan mixer was stopped, the cement was added, and the concrete was mixed for 3 minutes, allowed to rest for 3 minutes (during which time the pan was scraped), and then mixed again for 2 minutes. As soon as the concrete was mixed, concrete slump tests were conducted according to ASTM C143-03, Standard Test Method for Slump of Hydraulic-Cement Concrete. Four cones were filled. One slump was measured as soon as the cone was filled, and the others were measured at 10- to 15-minute intervals to determine slump loss. X-ray diffraction The mixing procedures for paste and concrete differed slightly, in that the paste was mixed for 4 minutes and the concrete for 8 minutes. The difference may be negligible because all tests were carried out at the beginning of the dormant period. Powder X-ray diffraction (XRD) was used to determine whether the difference in mixing time caused differences in the formation of hydration products between paste and concrete. If there is no difference between these hydration times, it is reasonable to conclude that this short difference in time between paste and concrete did not affect the rheology study. The cement pastes (made using Cement B) were mixed by hand for 1 minute at 25 C (77 F) using deionized water, w/c of 0.35, and 0.13% of CAE. The hydration was stopped at 4 minutes and 8 minutes to mimic the two mixing times explained previously, using the following procedure. A 20 ml (0.7 oz) glass bottle was filled approximately onethird with paste and filled to the top with methanol (shaking to mix and dilute the water). The cement was allowed to settle, the methanol-water liquid was carefully removed by pipette, and fresh methanol was added (shaking to mix). After 1 hour, the liquid layer was again removed by pipette and fresh methanol was added (shaking to mix). The sample was stored in this way until it was analyzed. Just before analysis, the methanol was removed by pipette, the specimen was ground in a mortar and pestle to a particle size of approximately 10 μm ( in.), methanol was allowed to evaporate, and the ground sample was packed into the sample holder from the back with modest hand pressure. The X-ray diffractometer collected data from 5 to 70 degrees (2 θ) with a step size of 0.02 degrees and a dwell time of 1.5 seconds. The working voltage was 40 kv and the current was 40 ma. Adsorption measurements As part of the main experiments, adsorption isotherms for CAE in paste and concrete were conducted to determine whether the dosage producing high slump and slump loss in concrete corresponds to the dosage required for full dispersion. These measurements were made by filtering the sample and then measuring the total organic carbon (TOC) in the liquid to determine the amount of adsorbed admixture. It was first necessary to perform a calibration to determine the relationship between TOC and CAE concentration. First, g (0.112 oz) of the thick CAE solution were transferred to a 100 ml (3.520 oz) volumetric flask and brought up to volume with deionized water, giving g of CAE per ml solution ( oz CAE per fluid oz solution). Dilutions of this solution were prepared volumetrically to provide eight solutions of various concentrations for calibration. A linear relationship between the concentrations of carbon and CAE was observed (Fig. 1). For cement paste, collecting the solution for TOC was begun within 5 minutes after mixing. Paste was centrifuged for 5 minutes at 6000 rpm and the resulting solution was transferred into a syringe and filtered through a 0.45 μm ( in.) paper filter. This solution was then centrifuged for a second time at 6400 rpm for 5 minutes to eliminate very small particles and finally filtered again with a syringe and a 0.45 μm ( in.) filter. For concrete, the coarse aggregate was first removed by sieving the concrete through a 4.75 mm Fig. 1 Calibration curve for determination of CAE concentration from results of total organic carbon (TOC) (x-axis ranges from 0 to oz C per lb solution and y-axis ranges from 0 to oz C per fluid oz solution). Table 3 Concrete mixture design (saturated surface dry basis) Cement, kg/m 3 (lb/ft 3 ) Water, kg/m 3 (lb/ft 3 ) Coarse aggregate, kg/m 3 (lb/ft 3 ) Fine aggregate, kg/m 3 (lb/ft 3 ) 510 (31.8) 179 (11.2) 1120 (69.9) 590 (36.8) ACI Materials Journal/May-June

4 (No. 4 mesh) sieve, and the resulting mortar was centrifuged and the solution filtered as described for paste. Because concrete was made using tap water, a sample of this water was also analyzed by TOC. Several aliquots of each sample were used to measure TOC and the adsorbed CAE was computed using the calibration curve. Fig. 2 Paste stress sweeps in preliminary experiments using varying amounts of CAE and: (a) Cement C1; (b) Cement C2; and (c) Cement C3 (from Reference 3, used with permission) (x-axes range from to psi and y-axes range from to psi). (Note: 1 psi = Pa.) RESULTS Paste rheology Paste rheology results are shown as stress sweeps (plots of dynamic storage modulus as a function of stress). In general, the pastes with 0% CAE had a high storage modulus at low stress and a high yield stress (stress at which the storage modulus decreased, indicating flow) characteristic of a flocculated microstructure. The pastes with high dosages of CAE had a very low storage modulus at all stress levels and no yield stress (at least not in the stress range that can be measured), characteristic of a fully dispersed microstructure. Pastes with intermediate dosages of CAE showed intermediate values of modulus and yield stress, characteristic of microstructures with intermediate flocculation. As CAE dosage was increased, the pastes became progressively more dispersed. As has been observed previously, 3,9 the storage modulus decreased by several orders of magnitude with increasing dosage of CAE. As noted previously, 3 the authors consider this type of behavior, a progressive increasing dispersion with increasing admixture dosage, to represent classical dispersion. The rheology results of the three cements in the preliminary experiments, which were reported previously, 3 are shown in Fig. 2(a), (b), and (c). With no CAE, stress sweeps were all characteristic of flocculated paste, and with high dosage of CAE, the stress sweeps became characteristic of dispersed paste. Two of the cements used in the preliminary experiments, Cement C2 and C3 (Fig. 2(b) and (c)), showed a slight gelation at intermediate dosages of CAE. For Cement C2, the gelation at 0.10 to 0.20% CAE was relatively strong. The gelation is indicated by: 1) increased storage modulus, similar to the modulus observed with no CAE; and 2) high yield stress, higher than the yield stress observed with no CAE and higher than the maximum value of 300 Pa (0.044 psi) used in these stress sweeps. Higher dosages of 0.25 to 0.40% CAE showed decreasing flocculation and increasing dispersion. For Cement C3, the gelation was relatively weak at 0.10% CAE, indicated by high storage modulus, similar to the modulus of 106 Pa (145 psi) observed with no CAE, and high yield stress, 100 Pa (0.015 psi), higher than the stress of 20 Pa (0.003 psi) observed with no CAE. Higher dosages of 0.15 to 0.40% showed decreasing flocculation and increasing dispersion. Cement B in the main experiment (Fig. 3) showed the same behavior described for Cement C1. With no CAE, the paste was flocculated, with a low-strain storage modulus of approximately 105 Pa (14.5 psi) and a yield stress of 10 Pa (0.001 psi). With an increasing dosage of CAE, the pastes became decreasingly flocculated and increasingly dispersed. There was no gelation at intermediate CAE dosages. Full dispersion was observed at 0.13% CAE. The CAE dosage required for full dispersion of Cement B was considerably lower than the dosage required for full dispersion of Cements C1, C2, and C3. This difference in dosage is attributed to the high-shear preshear protocol used with Cement B. It was previously reported 3 that this preshear protocol reduced both the storage modulus and the yield stress of pastes, especially pastes containing high-range waterreducing admixtures. Concrete workability Concrete slump results are shown in Fig. 4 for the preliminary experiments and Fig. 5 for the main experiments. Slump values increased dramatically with addition of CAE, from 284 ACI Materials Journal/May-June 2008

5 approximately 10 mm ( in.) for concretes with no CAE (somewhat higher for C2) to 200 to 250 mm (7.87 to 9.84 in.) for concretes with 0.1% CAE. In mixtures with dosages <0.1% CAE, the slump value was initially high but decreased substantially during the 1-hour test duration. High slump and slump retention were obtained only at the highest CAE dosages (approximately 0.13% for Cement C1, 0.11% for Cement C2, 0.11% for Cement C3, and 0.13% for Cement B). Above these dosages, segregation was observed, indicated by the separation of aggregate and paste and sometimes by bleeding, so higher dosages were not tested. Hydration Because paste and concrete were mixed for different periods of time (4 minutes for paste and 8 minutes for concrete), it was important to look for any difference in the degree of hydration. Especially important was the formation of ettringite (AFt), which may affect the rheological properties of cement paste and concrete, and the formation of calcium hydroxide (CH), which may accompany formation of poorly crystalline calcium silicate hydrate, which may also affect rheological properties of paste and concrete. To determine whether the difference in time affected the rheological properties, two samples of hydrated cements (one hydrated for 4 minutes and 8 minutes) were analyzed using XRD. As shown in Fig. 6, the XRD patterns appear to be the same and neither AFt nor CH was observed in either sample. Therefore, the difference in mixing time between paste and concrete can be ignored. Adsorption Figure 7 shows the amount of CAE adsorbed by Cement B in paste and concrete plotted as a function of CAE dosage (commonly called an isotherm). The dashed line indicates the admixture dosage required for full dispersion according to the stress sweeps, 0.13%. Kirby and Lewis 9 reported a similar isotherm for CAE in paste. They reported that the amount of CAE adsorbed relative to the cement surface area at the critical weight fraction was 1.35 mg/m 2 (59 oz/in. 2 ). The adsorption at the critical weight fraction in Fig. 7 was Fig. 4 Concrete slumps in preliminary experiments using varying amounts of CAE and: (a) Cement C1; (b) Cement C2; and (c) Cement C3 (y-axes range from 0 to 11-3/4 in. [0 to 300 mm]). Fig. 3 Paste stress sweeps in main experiment using varying amounts of CAE and Cement B (x-axes range from to psi and y-axes range from to psi). (Note: 1 psi = Pa.) Fig. 5 Concrete slumps in main experiment using varying amounts of CAE and Cement B (y-axis ranges from 0 to 11-3/4 in. [0 to 300 mm]). ACI Materials Journal/May-June

6 somewhat higher, 3.75 mg/m 2 (165 oz/in. 2 ), not surprising because different admixtures were used. It is expected that the saturation of paste with admixture would produce a dispersed microstructure and that additional Fig. 6 XRD patterns (Cu-Kα) for Cement B paste hydrated for 4 minute and 8 minutes showing absence of peaks attributed to AFt and CH (at positions indicated by dotted vertical lines) (8-minute pattern has been offset in intensity). Fig. 7 Adsorption isotherms of CAE admixture in paste and concrete, Cement B (y-axis ranges from 88 to 439 oz/in. 2 ). (Note: 1 oz/in. 2 = mg/m 2.) Fig. 8 Types I and II adsorption isotherms (often called BET classification), where W/W m is weight adsorbed (relative to weight at monolayer coverage) and P/P o is vapor pressure (relative to saturation pressure) (calculated using BET equations from Reference 10 with C = 100). admixture above saturation would remain in solution. Such behavior produces a Type I adsorption isotherm as described by Brunauer et al. (reported by Lowell and Shields 10 ) (Fig. 8(a)), characterized as chemisorption and encountered when adsorption is limited to few molecular layers, causing the adsorption to level off when P/P 0 is increased such that all adsorption sites are filled. Kirby and Lewis 9 reported that sulfonated naphthalene formaldehyde adsorption showed such Type I behavior. The CAE, on the other hand, appears to follow a Type II adsorption isotherm (Fig. 8(b)), encountered when multiple layers are adsorbed and the adsorption at saturation is infinite. Although it is not easy to identify precisely the dosage at full saturation in a Type II isotherm, it can be seen in Fig. 7 that the isotherm has an inflection point or knee of approximately 0.13% CAE for both paste and concrete, which can reasonably be taken as an estimate of the completion of the first adsorbed monolayer. DISCUSSION ON COMPARISON OF PASTE AND CONCRETE BEHAVIOR In the preliminary experiments, there was no agreement in the behavior of paste and concrete or in the admixture dosage for full dispersion in paste and concrete (Table 4). Although the CAE dosages used in concretes were within the recommended range, the dosage for full dispersion in concrete was approximately one-fourth that in paste. A similar difference in dosage between paste and concrete for sulfonated naphthalene formaldehyde admixture was observed. 8 Quite different behavior was observed in the main experiments. When pastes were first presheared using the high-shear protocol, results from paste rheology and concrete workability were very similar. Both paste and concrete were classically dispersed by CAE. Paste rheological properties, yield stress and elastic modulus, decreased as the dosage of CAE increased (Fig. 3) and concrete slump increased as the dosage of CAE increased (Fig. 5). The dosage for full dispersion in paste was 0.13% CAE, as indicated by zero yield stress, and the dosage for full dispersion in concrete was 0.13% CAE, as indicated by maximum slump and no slump loss. The objective of the adsorption measurements was to test our interpretation that the admixture dosage in concrete producing maximum slump and no slump loss indicates full dispersion, in part because the authors could not check slumps at higher CAE dosages due to concrete segregation. The segregation made slump measurements unreliable but was not expected to affect the adsorption behavior. The adsorption results showed that the interpretation was valid. The admixture dosage that provided maximum slump and slump retention (Fig. 5) also produced full saturation (Fig. 7). Lower dosages produced slump loss, much lower dosages produced lower maximum slumps, and higher dosages produced segregation. It should be noted that the coincidence of adsorption behavior between concrete and paste would probably not Table 4 Dosages for full dispersion in preliminary experiments (percent solid CAE by weight of cement) Cement Paste, % Concrete, % C C C ACI Materials Journal/May-June 2008

7 have been observed if the mixing procedures had not been so closely duplicated. Figure 9 summarizes the paste rheological parameters (storage modulus and yield stress) and concrete workability (initial slump) as a function of admixture dosage. From such plots it is possible to estimate the admixture dosage that produced full dispersion. Because yield stress shows a linear relationship with dosage, it allows the most precise estimation of admixture dosage for full dispersion. These data provide an opportunity to evaluate the relationship between concrete slump and the yield stress of paste. According to Tattersall and Banfill, 11 concrete slump shows a negative linear correlation with yield stress. In earlier work, Struble and Chen 4 reported an inverse but curvilinear (exponential) relationship between concrete slump and yield stress. Figure 10(a) shows a relationship between concrete slump (initial slump value) and paste yield stress similar to that reported by Struble and Chen, 4 and fitting with an exponential trendline gave a good correlation (R 2 = 0.98). Figure 10(b) shows a similar relationship between concrete slump and paste storage modulus (average of each measurement in the linear viscoelastic region), with a downward trend and a good linear correlation (R 2 = 0.97) when plotted in this way. Modulus values were plotted using a logarithmic axis because the authors previously 4 found such a relationship for several types of dispersing admixtures, though CAE mixtures had poor reproducibility. In the present study, the CAE mixtures showed a very strong linear relationship, perhaps because the authors were more careful to use the same mixing protocol for paste and concrete. The focus of this research was to explore the correlation between the rheological properties of paste and the slump of concrete and to consider whether the effects of dispersing admixtures are the same in paste and in concrete. Using a single concrete mixture, a clear relationship between paste Fig. 9 Paste and concrete values as function of admixture dosage, Cement B: (a) paste storage modulus; (b) paste yield stress; and (c) concrete initial slump (y-axes range from: (a) to psi; (b) 0 to psi; and (c) 0 to 11-3/4 in.). (Note: 1 psi = Pa; 1 in. = 25.4 mm.) Fig. 10 Initial concrete slump as function of paste (Cement B): (a) yield stress; and (b) storage modulus (x-axes range from: (a) 0 to psi; and (b) to psi and y-axes range from 0 to 11-3/4 in. [0 to mm]). (Note: 1 psi = Pa.) ACI Materials Journal/May-June

8 rheological properties (yield stress and dynamic storage modulus) and concrete slump was observed. There are numerous additional parameters that influence the slump of concrete (for example, concrete constituents, mixture proportions, and mixing procedure). These parameters need to be considered in any attempt to predict concrete slump based on measured rheological properties of paste. SUMMARY AND CONCLUSIONS The experimental findings and conclusions in this study are summarized as follows: 1. The yield stress and storage modulus of paste both decreased as the dosage of CAE was increased, although some cements showed a modest increase (gelation) at intermediate admixture dosages and a decrease only at higher dosages; 2. The slump of concrete increased and slump loss reduced as the dosage of CAE increased; 3. With hand mixing, full dispersion of paste was observed at approximately 0.4 to 0.7% CAE, based on the conversion from linear viscoelastic to viscous behavior; 4. With high-shear preshear, full dispersion in paste was observed at approximately 0.13% CAE; 5. Full dispersion in concrete was observed at dosages approximately 0.11 to 0.13% CAE, based on high slump and low slump loss; 6. The dosage giving full dispersion in concrete was the same as the dosage giving full dispersion in paste when paste was treated using a high-shear pre-shear protocol; 7. Admixture adsorption isotherms indicated that concretes were fully dispersed at dosages that provided high slump and low slump loss; 8. Concrete and paste gave similar adsorption responses. Monolayer adsorption in paste was observed at the same dosage at which the dynamic rheology behavior converted from linear viscoelastic to viscous. Monolayer adsorption in concrete was observed at the same dosage that produced maximum slump with no slump loss; 9. Concrete slump had an inverse and exponential relationship with the paste yield stress. Concrete slump had a linear downward trend on a semi-logarithmic scale with the storage modulus of paste; and 10. The correspondence between paste and concrete, the observation of full dispersion at the same admixture dosage and the correlation between paste rheological parameters and concrete slump, were only observed when the paste experiments used a high-shear preshear protocol. ACKNOWLEDGMENTS The authors are grateful to several organizations for funding this study: Consolis Technology Oy supported most of the work, the Center for Advanced Cement Based Materials supported CTC s preliminary studies, and SURGE and IMGIP at UIUC provided a fellowship for J. Hidalgo. Special thanks go to C. Zukoski for allowing use of facilities in the Colloid Interfacial Laboratory in the Department of Chemical & Biomolecular Engineering at UIUC, to the Waste Management Research Center at UIUC for conducting the total organic carbon tests, and to L. Shen in the Department of Civil and Environmental Engineering at UIUC for determining the cement surface area. Thanks also to R. Komarek and D. Corcoran for assistance in the concrete tests. A portion of Table 1 and all of Fig. 3 were reprinted, with permission, from the Journal of ASTM International (JAI), V. 3, No. 3, copyright ASTM International, 100 Barr Harbor Drive, West Conshohocken, PA The authors also gratefully acknowledge the following companies for their contributions: Essroc Inc. (Type I/II cement); W.R. Grace, Inc. (ADVAFlow, CAE used in preliminary experiments); and Degussa (Glenium 3200 HES, CAE used in main experiments). The following companies manufactured the equipment used in this study: Bohlin (CS-10 rheometer); Monarch Industries (Model RLX-3 Type D drum mixer); Lancaster (30-DP pan mixer) Rigaku (Geigerflex D-Max II X-ray diffractometer); and Shimadzu (Model VCPN total organic carbon analyzer). REFERENCES 1. Mindess, S.; Young, J. F.; and Darwin, D., Concrete, second edition, Prentice Hall, Upper Saddle River, NJ, 1996, p Struble, L. J.; Zhang, H.; Sun, G.-K.; and Lei, W.-G., Oscillatory Shear Behavior of Portland Cement Paste during Early Hydration, Concrete Science and Engineering, V. 2, No. 7, 2000, pp Chen, C.-T.; Struble, L. J.; and Zhang, H., Using Dynamic Rheology to Measure Cement-Admixture Interactions, Journal of ASTM International, V. 3, No. 3, 2006, 13 pp. 4. Struble, L. J., and Chen, C.-T., Effect of Continuous Agitation on Concrete Rheology, Journal of ASTM International, V. 2, No. 9, 2005, 19 pp. 5. Brower, L., and Ferraris, C. F., Comparison of Concrete Rheometers, Concrete International, V. 25, No. 8, Aug. 2003, pp Helmuth, R.; Hills, L.; Whiting, D.; and Bhattacharja, S., Abnormal Concrete Performance in the Presence of Admixtures, PCA R&D Serial No. 2006, RP333, Portland Cement Association, Skokie, IL, 1995, 88 pp. 7. Hidalgo, J., Effect of Shear History of Paste on Concrete Properties, MS thesis, University of Illinois at Urbana-Champaign, Urbana, IL, 2005, 61 pp. 8. Chen, C.-T., Interactions Between Portland Cements and Carboxylated and Naphthalene-Based Superplasticizers, PhD thesis, University of Illinois at Urbana-Champaign, Urbana, IL, 2007, 424 pp. 9. Kirby, G. H., and Lewis, J. A., Rheological Property Evolution in Concentrated Cement-Polyelectrolyte Suspensions, Journal of American Ceramic Society, V. 85, No. 12, 2002, pp Lowell, S., and Shields, J. E., Powder Surface Area and Porosity, third edition, Chapman and Hall, New York, NY, 1991, p Tattersall, G. H., and Banfill, P. F. G., The Rheology of Fresh Concrete, Pitman Advanced Publishing Program, Boston, MA, 1983, 356 pp. 288 ACI Materials Journal/May-June 2008