Properties of Self-Consolidating Concrete Containing Class F Fly Ash

Transcription

1 Properties of Self-Consolidating Concrete Containing Class F Fly Ash Raissa P. Douglas ACBM, Northwestern University USA Van K. Bui Former Research Associate, ACBM, Northwestern University, USA Yilmaz Akkaya Research Associate, ACBM, Northwestern University, USA Surendra P. Shah ACBM, Northwestern University, USA Abstract An experimental program was performed to investigate self-consolidating concrete (SCC) mixes by altering the water/binder ratio and using different substitution rates of fly ash. In total, six mixes were evaluated. All mixtures were characterized using measures of segregation and flow. In addition, compressive strength and mold-finish were evaluated. The objective of the investigation is to develop a cost-effective, high performance SCC containing Class F fly ash, and to compare the experimental results with the predicted theoretical results. The results of this project should provide information that will help reduce the material cost of SCC, further sustainable development in the concrete industry, and contribute to the development and usage of SCC in the United States and the global community. Keywords: Self-consolidating concrete, rheology, fly ash 1. Introduction Since the introduction of SCC in Japan during the late 198 s, acceptance and usage of this concrete in the concrete construction industry has been steadily gaining momentum. Good SCC must possess the following key fresh properties: filling ability, passing ability, and resistance to segregation. Filling ability is ability of the concrete to flow into and fill completely all spaces within the formwork under its own weight. Passing ability is the ability of the concrete to flow through tight openings (ex. dense reinforcement) under its own weight without blocking. Resistance to segregation is the ability of the concrete to meet the filling ability and passing ability requirements while maintaining uniform composition (hence, no separation of aggregate from paste, or water from solids). In order to reduce segregation, SCC mixes typically are designed with high powder contents, and chemical admixtures such as superplasticizers and viscosity modifying admixtures (VMA), which tends to increase the material cost of SCC. One way to reduce the material cost is through adequate mix proportioning and the addition of supplementary cementitious materials such as fly ash. The incorporation of Class F fly ash in self-consolidating concrete as a means to replace portions of Douglas, Self-Consolidating Concrete, Page 1 of 11

2 cement can serve to decrease the cost of SCC, as well as further the sustainable development of concrete. An experimental program, which aims to investigate behavior of SCC containing Class F fly ash, has been carried out. This paper presents the test results of properties of the SCC in fresh and hardened states. The test results on rheology of the SCC s pastes are also discussed and compared with the previously developed model. 2. Experimental Program 2.1 Materials Ordinary (Type 1) Portland Cement (OPC) and Vermillion (Type F) Fly Ash were used. The physical and chemical properties are listed in Table 1. Fine river sand and 19mm diameter gravel were used as the aggregates. A polycarboxylate-based superplasticizer (Catexol- Superfluc 2 PC) was used. Table 1: Fly ash composition Property Class F Fly Ash Silicon Oxide (SiO2), % Aluminum Oxide (Al2O3), % Ferric Oxide (Fe2O2), % Calcium Oxide (CaO), % 5.4 Sulfur Trioxide (SO3), % 1.6 Magnesium Oxide (MgO), %.9 Loss on Ignition (LOI), % 2.17 Specific Gravity 2.27 Fineness (+325 Mesh), % 15.9 Moisture Content, %.5 Soundness, % -.2 S.A.I, 7 days, % Mix Proportions A total of six concrete mixes were tested. The proportions of the concrete mixes are summarized in Table 2. Concrete mixes were prepared by varying w/b while keeping the paste volume constant and gravel/total aggregate ratio (Nga) constant. The paste volume for all mixes is 364L/m 3 with an Nga =.52. Fly ash replacement was conducted on a mass basis. Table 2: Mix proportions Mix Gravel (kg/m 3 ) Cement (kg/m 3 ) Constituent Fly Ash (kg/m 3 ) Water (kg/m 3 ) SP (kg/m 3 ) S35-F S35-F S35-F S45-F S45-F W/B Douglas, Self-Consolidating Concrete, Page 2 of 11

3 S45-F Mixing and Test Methods Through the use of slump cone and plate, penetration apparatus, and L-box testing were undertaken to investigate the flowability, segregation resistance, and passing ability of SCC. The details of the tests can be found in references <1-3>. The compressive strength (using ASTM C-39) was determined after 1 day, 7 days, and 28 days. The viscosity and paste flow diameter of the cement pastes were also measured Concrete Mixes All mixes were prepared by mixing aggregates (fine and coarse), fly ash, and cement for one minute. Then, the mixing water was added, and the batch was mixed again for one minute. After waiting 3 seconds, the batch was mixed for two minutes, and then the superplasticizer was added and the batch was mixed for an additional two minutes. The tests of flowability, passing ability, and segregation resistance were then performed. For the flowability test, the time that it took the concrete to reach a diameter of 5 cm (T 5 ) and the final diameter of the concrete after it stopped flowing were recorded (see Fig. 1.a). The concrete s mold surface quality was observed after concrete filled into the vertical portion of the L-Box, which was made of flexi-glass. Then the passing ability test with an L-box was undertaken, and the H2/H1 ratio was calculated (higher H2/H1 ratio indicates greater passing ability of SCC). The concrete was poured into 1 x 2mm cylindrical plastic molds and covered with a plastic sheet in order to prevent water evaporation. The cylinders were demolded after one day and placed in a curing room at 1% relative humidity until the compressive strength test was performed. 1a 1b Fig. 1: a) Concrete slump flow; b) Paste flow test Fresh Paste: All mixes were prepared by placing the water into a standard Hobart planetary mixer. The cement and fly ash were weighed out together and added to the mixer over a one minute interval. The paste was mixed for 2.5 minutes at speed 1. Then, the mixer was stopped and the sides of the mixing bowl were scraped over a 3 second interval. The superplasticizer was added to the paste and the paste was mixed at speed 2 for an additional 2.5 minutes. Immediately, after mixing a portion of the paste was placed into a HAAKE-RS15 rheometer for viscosity testing and flow of paste tests (see Fig. 1.b) were conducted. See reference 4 for the paste flow tests measurement procedure. 2.4 Previously Developed Rheological Model Douglas, Self-Consolidating Concrete, Page 3 of 11

4 A rheological model for self-consolidating concrete was developed by Bui, Akkaya, and Shah <4>. This model, which takes into account aggregate particle interaction, is based on the assumption that there is a minimum flow, minimum apparent viscosity, and optimum flowviscosity ratio needed in order to produce a good quality SCC with satisfactory deformability and resistance to segregation. Therefore, there is a window of opportunity or satisfactory zone for obtaining a good performing SCC. It has been shown that segregation resistance is affected by aggregate diameter. Hence the average aggregate diameter (D av ) and average aggregate spacing (D ss ) were used as parameters to account for aggregate influence. See Reference 4 for an in depth discussion of these two parameters. It was proposed that this model would be valid for concrete mixes with different aggregate ratios, cement contents, fly ash types and content, and water/binder ratios. The model characterizes the paste matrix by its paste flow apparent viscosity at 1s -1. The apparent viscosity was determined using a measuring protocol developed by Saak, Jennings, and Shah <5> in which the shear rate is ramped from to 6s -1 over a 1s time interval, then the shear rate is held constant for 12s, and finally the shear rate is ramped back down from 6 to s -1 over a 3s time interval. The apparent viscosity, taken from the up curve of the hysteresis loop, at a shear rate of 1s -1 is used to characterize the pastes. (For simplicity, the apparent viscosity will be called viscosity for the remainder of this paper.) In addition, concrete properties were quantitatively characterized by passing ability and segregation resistance and qualitatively characterized by its mold surface. The model was developed with concrete mixes with water/binder (w/b) ratios ranging from.31 to.39 (only one mix was tested with a w/b =.31, most mixes had a w/b =.38) and paste volumes ranging from to L/m3. For high water to binder ratios (greater than.4), there is a high potential for sedimentation <6, 9>, slippage <7, 8>, and centrifugal separation <6>, when centric cylinder rheometers are used. Also, it is believed that the difference between the densities of paste and aggregates affects the requirement of rheology for SCC paste <4, 5>. However, in the recently developed rheological model <4>, this factor was neglected because the range of w/b used in test program was small. Therefore, in the study presented in this paper, the application of the rheological model will be examined for mixes with high water to binder ratio (w/b =.45), for which the difference between densities of paste and aggregates is significant and may affect the correlation of paste rheology with SCC s properties. 3. Results and Discussion 3.1 Concrete Mixes: Overall data for the fresh concrete properties are presented in Table 3 and Figure 2. Compressive strength results are shown in Figure Properties of Fresh SCC It is commonly accepted that a T 5 < 12s, concrete flow diameter (F d ) > 6mm, and H2/H1.8, will produce a good performing SCC <3, 4>. All mixes had similar slump flow times (average 3 seconds), and thus all mixes had good flowability. As indicated by the H2/H1 ratio, all mixtures exhibited good passing ability and no blockage was exhibited in any of the mixes. Previous research has shown that a penetration depth greater than 8 mm is an indication that an SCC would be prone to high segregation <4>. All mixes displayed good resistance to segregation, and good mold-surface finishability, except mix S45-F, which had a little honeycombing in the cast cylinders. Douglas, Self-Consolidating Concrete, Page 4 of 11

5 Table 3: Overall fresh properties Mix T5 (s) Flow Diameter (F d ) (mm) Penetration Depth (mm) H2/H1 S35-F S35-F S35-F S45-F S45-F S45-F As seen in Figure 2, for a given w/b, addition of fly ash generally results in a reduction of superplasticizer for a similar workability (flow diameter). Furthermore, as expected, for similar workability, increasing the water/binder ratio reduces the amount of superplasticizer required. Lastly, at higher w/b, superplasticizer requirement is less sensitive to the addition of addition of fly ash S35-F S35-F2 S35-F3 S45-F S45-F2 S45-F3 Mix SP (kg/m^3) W/C Figure 2: Superplasticizer Requirement and water/cement ratios for mixes Compressive Strength As expected, addition of fly ash resulted in a reduction of strength. However, all mixes still exhibited relatively high strengths, which ranged from 15 MPa to 35 MPa, and from 4 MPa to 7 MPa at ages of 1 day and 28 days, respectively. For the lower w/b mixes, addition of fly ash also resulted in a greater reduction of strength compared with that of the control mix (S35- F). The reason for this phenomenon is not clear. As shown in Figures 2 and 3, S35-F has the lowest water/cement ratio and thus it can be expected that its rate of strength development would proceed faster than the other mixes. In general, the overall trend of strength development was similar for mixes with similar water/binder ratios. Douglas, Self-Consolidating Concrete, Page 5 of 11

6 Compressive Strength (MPa) S35-F S35-F2 S35-F3 S45-F S45-F2 S45-F3 Mix 1 day 7 day 28 day Fig. 3: Compressive Strength 3.2. Comparison of Rheological Test Results with Developed Model Replication testing for viscosity measurements was performed on the each mix, resulting in an average standard deviation of.31 PaS and an average coefficient of variation of 13.%. Rheology test results are shown in Table 4. As expected, there is some degree of experimental error, but overall within a given mix the results are consistent. Table 4: Repeatability testing for viscosity. Mix Trial 1 Trial 2 Trial 3 µ (PaS) µ (PaS) µ (PaS) Average Viscosity (PaS) Standard Deviation (PaS) S35-F S35-F S35-F S45-F S45-F S Coefficient of Variation (%) The mixes in this study all have a D av and D ss of mm and.45 mm, respectively. The test results for flow diameter, viscosity, and flow-viscosity ratio are presented in Tables 5 and Fig. 4 through Fig. 7. The minimum lines depicted in Fig. 5 to Fig. 7, indicates the limit for a satisfactory SCC. The limit values were obtained from Reference <4>. Table 5: Paste flow diameter and viscosity Mix Flow Diameter (mm) Ave. Viscosity (PaS) Flow/Viscosity (mm/pas) S35-F S35-F S35-F S45-F S45-F S45-F Douglas, Self-Consolidating Concrete, Page 6 of 11

7 Paste Flow Diameters: Previous studies have shown that the stress vs. shear rate of paste for SCC follows the Power Law or Herschel-Bulkey model, and not the Bingham relationship <4, 1>. Hence, the paste flow, which appears partly related to the yield stress was used as an indicator of the concrete s yield stress.. Higher values for paste flow correspond to lower values of yield stress <11>. According to the rheological model <4>, for a D ss of.45 a minimum flow diameter of 357 mm is needed in order to have a SCC with good deformability. All of the mixes, except for S35-F3 (which is marginal), meet this criterion. In addition, although mix S45-F has a paste flow higher than the minimum flow, the respective concrete exhibits a little honeycombing. This confirms the findings of a previous study <4> that meeting the minimum paste flow does not guarantee good mold-surface finishing of concrete (see also further discussion in section Effect of aggregate dimension and difference between densities of aggregates and paste ). Apparent Viscosity: As expected, for a given w/b, the addition of fly ash decreases the viscosity. This supports previous findings <6, 11> that there are rheological benefits to using fly ash without increasing the superplasticizer dosage or increasing the water demand. According to the rheological model <4>, for a D ss of.45, a minimum apparent viscosity of.275 PaS is needed in order to have a SCC with good deformability and segregation resistance. All of the lower w/b mixes meet this criterion, but all of the higher w/b mixes have viscosities that are below the minimum. Hence the model predicts that these mixes will result in concrete that will segregate. The reason may be due to the fact that higher w/b mixes have a higher difference between densities of the paste and aggregates (see further discussion in section Effect of aggregate dimension and difference between densities of aggregates and paste ). Optimum Flow-Viscosity Ratio: According to the rheological model <4>, for a D ss of.45 the lower and upper bounds for a good quality SCC is 994 mm/pas and 1283 mm/pas, respectively. These values were used to determine the satisfactory zone depicted in Figure 4. Therefore, it is expected that the higher w/b mixes would produce a concrete that is prone to segregate, and the lower w/b mixes are likely to produce a good quality SCC. Hence, there is a discrepancy between the rheological model prediction for the concrete based on the cement pastes and the actual performance of the corresponding concrete determined from the testing methods described in Section According to the model, for the given paste flow diameters, the viscosity of the higher water/binder mixes should be in the range of about.35.5 PaS. Thus, the difference between the model predications and the actual experimental results appears to be related to the low viscosity values for the high w/b mixes. Although the viscosity values listed in Table 5 appear to be rather low, they are within the 5- fold range reported by Tattersall and Banfill <6>. Therefore, it is possible that the true tendency of viscosities of the mixes is being captured. However, it is well known that measurement procedure influences rheological properties <1, 11>, and as a result, it was initially assumed that the large decrease in viscosity may have been due to the rheological protocol. Therefore, the viscosity was also measured using different protocols (step up with and without pre-shearing, hysteresis loops with and without out an equilibrium period, and decreasing the maximum shear rate for the hysteresis loop testing), but these techniques yielded similar results and it can be concluded that the low viscosity values were not due to the rheological protocol. Several experimenters have reported the existence of slippage <7-8>, sedimentation when using Douglas, Self-Consolidating Concrete, Page 7 of 11

8 concentric cylinders rheometers for water/cement ratios over.4 <6, 9> and centrifugal separation <6>. Also, as mentioned section 2.4, for high water/binder ratios, the difference between densities of paste and aggregates is large and may influence the correlation of paste rheology with fresh SCC s properties. Therefore, although the previous model does an adequate job of predicting the performance of the lower water/binder mixes, in order to make it for general use, additional factors such as the difference between densities of paste and aggregates (i.e. the volumetric ratio between solid particle and liquid phase) are needed to be considered. Further discussion about this effect is in the following section. Paste flow diameter (mm) S45-F2 S45-F3 S45-F Segregation Zone Satisfactory zone S35-F2 S35-F3 S35-F Low Deformability zone Dav = Paste Viscosity (PaS) Figure 4: Satisfactory zone for SCC Effect of aggregate dimension and difference between densities of aggregates and paste For a single spherical particle, the segregation resistance under dynamic conditions is related to falling velocity (ν) of the particle in a matrix <5>. The falling velocity is expressed in the following equation: 8 g ρrp ν = (Eq. 1) 3 CDρm where ν: Falling velocity of a spherical particle in a matrix g: Gravity coefficient ρ: Difference between density of the particle and matrix r p : Particle radius ρ m : Matrix density C D : Drag coefficient, which is related to Reynolds (R e ) number and matrix viscosity (η) 1 η ( CD ) Re ρm From Eq. 1, it can be seen that a higher product between ρ and r p leads to a higher falling velocity. In order to avoid segregation, the falling velocity (ν) should be minimized. In the previous work <4>, the models for minimum paste flow, minimum viscosity, and optimum paste flow-viscosity ratio were developed. These criteria are related to average aggregate spacing (D ss ) and average aggregate diameter (D av ) (see Fig. 5, 6, and 7). As Douglas, Self-Consolidating Concrete, Page 8 of 11

9 mentioned, difference between densities of aggregates and paste ( ρ) affect the correlation of paste rheology with fresh SCC s properties; however, this effect was neglected in previous work since the density difference ( ρ) was small for the tested mixes. In this study, the values of ρ are large and needed to be considered. The average values of D av, ρ, and product between average aggregate radius (r av ) (r av = D av /2) and ρ of mixes in previous works and new tests are indicated in Table 6. The effects of the product between ρ and r av are illustrated in Fig. 5, 6, and 7. The product between ρ and r av hereafter is noted as P r. Table 6: Product (P r ) of ρ and r av Coarse-total aggregate ratio (Nga) D av (mm) ρ (kg/m3) P r = ρ*r av (g/m2) Previous experiment Case A Previous experiment Case B New tests-w/b= New tests-w/b= Paste flow (mm) Mix S45-F exhibits little honeycombing Aggregate spacing, Dss (mm) Minimum paste flow for case A (Dav=4.673mm, Density difference x Radius = 1.96 g/m2) Minimum paste flow for case B (Dav=5.675 mm, Density difference x Radius = kg/m2) New Tests - w/b=.35 (Dav=5.675mm, Density difference x Radius = 2.31 g/m2) New Tests - w/b=.45 (Dav=5.675mm, Density difference x Radius = g/m2) Fig. 5: Paste flow and different products of ρ and average radius (r av ) of aggregate Apparent viscosity (Pa.S) Aggregate spacing, Dss (mm) Minimum paste viscosity for case A: (Density difference x Radius = 1.96 g/m2) Minimum paste viscosity for case B (Density difference x Radius = kg/m2) New Tests - w/b=.35 (Dav=5.675mm, Density difference x Radius = 2.31 g/m2) New Tests - w/b=.45 (Dav=5.675mm, Density difference x Radius = g/m2) Fig. 6: Viscosity and different products of ρ and average radius (r av ) of aggregate Douglas, Self-Consolidating Concrete, Page 9 of 11

10 Paste flow-viscosity ratio (mm/pa.s) Mix S45-F exhibits little honeycombing Aggregate spacing, Dss (mm) Max. for good segregation resistance for case A (Dav=4.673mm, Density difference x Radius = 1.96 g/m2) Minimum for satisfactory flowability for case A (Dav=4.673mm, Density difference x Radius = 1.96 g/m2) Max. for good segregation resistance for case B (Dav=5.675mm, Density difference x Radius = g/m2) Minimum for satisfactory flowability for case B (Dav=5.675mm, Density difference x Radius = kg/m2) New Tests - w/b=.35 (Dav=5.675mm, Density difference x Radius = 2.31 g/m2) New Tests - w/b=.45 (Dav=5.675mm, Density difference x Radius = g/m2) Fig. 7: Paste flow-viscosity ratio and different products of ρ and average radius (r av ) of aggregate Fig. 5 and Fig. 6 indicate that, for the same aggregate spacing, higher values of P r tend to increase the minimum paste flow required in order achieve good flowability of SCC, and decrease the minimum viscosity needed avoid segregation of concrete. Fig. 7 shows the tendency that, for a given aggregate spacing, greater values of P r require a greater paste flowviscosity ratio in order to achieve satisfactory flowability and good segregation resistance of SCC. The test s results suggest that there is a close relation between optimum paste flowviscosity ratio and falling velocity (ν) as seen in Eq. 1 since higher values of P r also results in greater falling velocity in order to avoid segregation of a particle in a matrix. However, in order to confirm this suggestion, and incorporate the effect of P r into a universal model, further study is needed. 4. Conclusions In summary, the following conclusions can be drawn from this study: It is possible to produce a high performance SCC incorporating with Class F fly ash. For similar concrete slump flows, addition of fly ash generally resulted in a reduction of superplasticizer for a similar workability. Although compressive strengths of mixes containing fly ash were lower than mixes without fly ash, they were relatively high at both 1-day and 28-day ages. The application of the previously developed paste rheology model to mixes with w/b ratios of.35 and.45 was verified. The model does an adequate job of predicting the performance of the lower water/binder mixes. For the higher w/b, the effect of difference between densities of aggregates and paste cannot be Douglas, Self-Consolidating Concrete, Page 1 of 11

11 5. Acknowledgments neglected. Both aggregate spacing (D ss ) and the product (P r ) between average aggregate radius and difference between densities of aggregates and paste influence the correlation of paste rheology with fresh SCC s properties. A further study is needed to investigate the influence of difference between densities of paste and aggregates, and the behaviors of possible slip, sedimentation, or centrifugal separation when determining the viscosity using a concentric cylinder rheometer. Financial support for this project was obtained from the Portland Cement Association and Illinois Clean Coal Institute. The fly ash was obtained from Dynegy Energy. 6. References 1. Bui, V.K., and S.P. Shah, Rapid Methods for Testing Quality of Fresh Self- Consolidating Concrete, Proceedings from First North American Conference on the Design and Use of Self-Consolidating Concrete, Chicago, USA, Nov , Billberg, P., Mix Design Model for Self-Compacting Concrete, Proceedings from First North American Conference on the Design and Use of Self-Consolidating Concrete, Chicago, USA, Nov , Bui, V.K., Montgomery D., Hinczak, I., and K. Turner, Rapid Testing Method for Segregation Resistance of Self-Compacting Concrete, Cement and Concrete Research, V. 32, 22, pp Bui, V.K., Akkaya, Y., and S.P. Shah, Rheological Model for Self-Consolidating Concrete, ACI Materials Journal, V.99, No. 6, Nov.-Dec 22, pp Saak, W.A., Jennings, H.M., and S.P. Shah, New Methodology for Designing Self- Compacting Concrete, ACI Materials Journal, V. 98, No. 6, Nov.-Dec 21, pp Tattersall, G.H., and P.F.G. Banfill, The Rheology of Fresh Concrete, Pitman, (1983), pp Saak, W.A., Jennings, H.M., and S.P. Shah, The Influence of Wall Slip on Yield Stress and Viscoelastic Measurements of Cement Paste, Cement and Concrete Research, V. 31, 21, pp Struble, L.J., Puri, U., and X. Ji, Concrete Rheometer, Advances in Cement Research, 13, No. 2, April 21, pp Bhatty, J.I., and P.F.G. Banfill, Sedimentation Behaviour in Cement Pastes Subjected to Continouse Shear in Rotational Viscometers, Cement and Concrete Research, V. 12, 1982, pp Geiker, M.R. et. al., The Effect of Measuring Procedure on the Apparent Rheological Properties of Self-Compacting Concrete, Cement and Concrete Research, V. 32, 22, pp Ferraris, C.F., Obla, K.H., and Russell Hill, The Influence of Mineral Admixtures on the Rheology of Cement Paste and Concrete, Cement and Concrete Research, V. 31, 21, pp Douglas, Self-Consolidating Concrete, Page 11 of 11