ESTIMATION OF BINGHAM RHEOLOGICAL PARAMETERS OF SCC FROM SLUMP FLOW MEASUREMENT L. N. Thrane, C. Pade and T. Svensson Danish Technological Institute, Concrete Centre, Taastrup, Denmark Abstract Different rheometers yield results of relative poor correlation and are often impractical for use at production plants and job sites. The slump flow test is the most widely used test method for quality control of the flow properties of SCC. However, often only the slump flow is measured because operator dependency and measuring accuracy make it impossible to obtain accurate information on the transient flow of SCC. Performing the traditional slump flow test in an automated manner ensuring identical execution each time combined with simple image analysis has been developed to accurately measure the flow curve i.e. spread versus time. This paper discusses the estimation of the yield stress and plastic viscosity based on an automated slump flow test. The effect of particle size and volume fraction is discussed and the results are compared with measurements in a co-axial rheometer. 1. INTRODUCTION The development of SCC has resulted in new possibilities in relation to casting concrete. However, success is not a matter of course but depends strongly on the proper choice of rheological parameters and casting technique, which determine the form filling ability and flow patterns. The form filling ability expresses the ability of the concrete to homogenously flow out into every corner of the formwork, and the flow patterns express the flow characteristics e.g. the direction and rate of flow at any point during placing. Comparing experimental observations with simulations have shown that the flow patterns have a significant effect on e.g. the dynamic segregation resistance and the surface quality [1]. One of the challenges to SCC is to bring rheology from research and development into the field of practical applications e.g. as a tool in quality assurance procedures [2]. Today, practical experiences from castings are related to the results from standard test methods such as the slump flow test, which is a quick and easy-to-use method compared to commercial rheometers. However, the challenges are twofold: First of all, standard test methods are operator dependent and provide test specific results, and secondly comparative studies have shown that yield stress (τ 0 ) and plastic viscosity (η pl ) depend on the viscometer applied [3]. It has been concluded that more research is needed to obtain a good correlation between any two rheometers and especially the plastic viscosity cannot be easily and uniquely measured [4]. Much work have been carried out on estimating the yield stress from the slump and slump flow based on the homogeneity approach, see e.g. the extensive summary in [5], and recently a relation for the yield stress in the L-box was proposed [6]. For estimation of the plastic viscosity from the slump flow test, numerical simulations of the transient flow are needed taking into account the cone lifting velocity, cone orientation, and the base plate conditions. 353
This paper presents an automated slump flow test which has been developed to measure both the yield stress and plastic viscosity based on numerical simulations applying a homogeneity approach. The advantage of this method is that it is based on a well-known principle of testing, quick and easy to perform, operator independent, and that the estimated rheological parameters corresponds to the qualitative behaviour observed in the test. 2. EXPERIMENTAL 2.1 4C Auto Slump Flow The 4C Auto Slump Flow system is shown in Figure 1. The system is a PC automated slump flow test where the spread versus time is determined using digital video image analysis. The base plate is a dry sand blasted glass plate and the lifting of the Abrams cone is performed at a constant speed of 7.0 cm/s. The video recording is done at a frame rate of 15 s -1. For each frame the position of the concrete front is determined using a standard find edge image analysis algorithm. 800 700 Spread (mm) 600 500 400 300 200 100 Plastic viscosity = 158 Pa s Plastic viscosity = 42 Pa s 0 0 5 10 15 20 25 Time (s) Figure 1: The 4C Auto Slump Flow system and examples of flow curves. Note that flow curves with very different plastic viscosities may obtain identical values of t 500. The flow curve is subsequently compared to a database of simulated flow curves obtained for yield stresses in the range 10-250 Pa and plastic viscosities in the range 5-250 Pa s using boundary conditions as in the experimental setup, i.e. a no-slip surface and a cone lifting velocity of 7.0 cm/s. The yield stress is estimated directly from a relation between the slump flow, yield stress, and density. In case the flow is asymmetrical around the cone, the system can apply the manually measured slump flow. Estimating the plastic viscosity requires an accurate and detailed monitoring of the flow curve. Especially the initial part of the flow curve is important and the plastic viscosity is estimated within slump flows from 200 to 450 mm. Above 300 mm, the yield stress starts to become of importance, which is taking into account. It is the authors opinion that the t 500 value is a very questionable estimate of the deformation properties of SCC. The significant operator dependency makes it a very inaccurate measure and both the yield stress and the plastic viscosity will influence on the t 500 value e.g. see the two flow curves in Figure 1, 354
which obtain the same t 500 value but very different plastic viscosities. These are both examples from real applications, and behaved very different during casting. Also, empirical investigations correlating the t 500 value with the plastic viscosity from rheological measurements have not shown a clear relationship, see e.g. [7]. 2.2 Viscometer A co-axial cylinder viscometer, the ConTec viscometer 3, has been applied to measure the yield stress and plastic viscosity. The measuring procedure includes 8 different rotation velocities with a maximum and minimum value of 0.57 s 1 and 0.05 s 1, respectively [8]. To obtain steady state flow and limit segregation, testing have been undertaken for 15 s at each rotational velocity. The raw data files are analysed individually in order to leave out points without steady state and avoid e.g. negative yield stresses. 2.3 L-Box and Yield Stress Recently, analytical expressions have been derived relating the yield stress and the socalled H 2 /H 1 ratio in the L-box test [6]. In the L-box, the ratio of material thickness to particle size is higher than in the slump flow test at large slump flows. Relations have been derived with and without reinforcement, and it is assumed inertia is negligable so the gate must be lifted slowly especially at low plastic viscosities [9]. H 2 ρ g (1 ) Without reinforcement: H1 τ 0 = (1) H 2 84 ( + 1) H1 Where ρ is density and g is gravity. 2.4 Mix Design Investigated Parameters The mix design parameters of interest of each experimental series are provided in Table 1. Table 1. The five different experimental series investigated. Series Mix design parameters Yield stress 1 2 3 4 5 Three mortars and six concretes: Mortar: Fine aggregate volume fraction (0-4 mm) 0.60. Concrete: Aggregate volume fraction (4-16 mm): 0.32 0.42 Three concretes: Identical mix design in kg/m 3 with three different fine particle size distributions (same fine aggregate source). Three concretes: Identical mix design in kg/m3 with three different SP-dosages Three concretes: Identical aggregate proportions and paste content, but different w/c and SP-dosages. Target slump flow at 550 mm. Four concretes: Aggregate volume fraction (4-16 mm): 0.40 0.42 SCC mixed and rheology measured Very fine sand added and rheology measured Fly ash + SP added and rheology measured L-Box L-Box Plastic viscosity 355
Table 1 has been divided into a into a number of series specifically targeted at investigating a particular relationship, e.g. comparing yield stress estimated by and to yield stress estimated by L-Box (series 1 and 5), or at investigation of the influence of changing a particular material parameter on the rheological parameters (series 2, 3, and 4). 3. RESULTS AND DISCUSSION Figure 2 shows the yield stresses measured by the system and the plotted against the yield stresses measured in the L-Box. The yield stress of ten concretes and three mortars were measured using both the analytical expression for the L-Box experiment and the system. In addition, the yield stress of four of the concretes was also measured using the viscometer. Yield Stress, L-Box (Pa) 8 16 32 64 128 256 Serie5 mortar Serie1 concrete Serie3 concrete 8 16 32 64 128 256 Yield Stress, or (Pa) Figure 2: Comparison of the yield stress measured by and to the yield stress measured from the L-Box. There appears to be a fair correlation between yield stresses measured by the system and the L-Box. However, there seems to a trend towards the system measuring higher yield stresses than the L-Box. This trend is apparently more pronounced at low yield stresses perhaps due to inertia effects in the L-Box experiments leading to an underestimation of the yield stress by this method. For the, the yield stresses are lower than the estimation from the and L-Box. Regarding the homogeneity assumption, there is not a significant trend between the results of the concretes and mortars tested. Consequently, based on the performed experiments, it seems that the slump flow may be used as an estimate of the yield stress at least above approximately 30 Pa for SCC mixes with aggregate size up 16 mm. A yield stress of 30 Pa corresponds to a slump flow of approximately 650 mm. It is difficult to argue that in this yield 356
stress range one method of measuring yield stress is better than the others. However, further testing will be carried out, and include more testing in the low yield stress range and aggregate sizes larger than 16 mm. Figure 3 shows yield stress versus plastic viscosity for parameter studies measured both with the system and the viscometer. The scales are double up (8, 16, 32 etc.) as the simulations show that it has e.g. the same relative effect on the flow curves going from a plastic viscosity of 10 to 20 Pa s as from 60 to 120 Pa s, respectively. : decreasing w/c at constant yield stress : decreasing w/c at constant yield stress : increasing fineness of sand : increasing fineness of sand : increasing SPdosage : increasing SPdosage : series 5 : series 5 Yield Stress (Pa) 8 16 32 64 128 256 8 16 32 64 128 256 8 16 32 64 128 Plastic viscosity (Pa s) Figure 3: Comparison of the rheological parameters of SCC measured using the system and the -viscometer when different mix design parameters are changed. The results show that though the values of the rheological parameters obtained are not identical, the qualitative information obtained from the system and the viscometer are the same when evaluating the influence of different mix design parameters. Both methods suggest that increasing the dosage of superplasticizer brings down the yield stress while having a relatively minor decreasing effect on the plastic viscosity. Likewise both methods yield comparable responses to increased fineness of the sand fraction, decreased w/c at constant slump flow, and to series 5 including subsequent additions of first very fine sand and secondly fly ash and SP. In general, however, it appears that the system measures higher plastic viscosities and higher yield stresses than the viscometer. A possible explanation could be that a slight degree of particle migration occurs during testing in the viscometer due to shear rate gradients inherent to a co-axial viscometer resulting in an underestimation of the rheological parameters. 256 357
4. CONCLUSION The rheological parameters measured in a controlled automated slump flow test using the system have been compared to those obtained using the -viscometer and an analytical expression for the yield stress in the L-Box. For yield stresses above approximately 30 Pa, it seems difficult based on the performed experiments to argue that one method of measuring yield stress is better than the others at least for SCC mix compositions with maximum size aggregates of 16 mm and coarse aggregate volume fractions of up to 0.40. For the effect of mix design parameters, the results show that the system and viscometer provide the same qualitative information. Consequently, the rheological parameters of SCC determined by appear to be as applicable and trust-worthy as the rheological parameters provided by other methods available to the concrete technologist. Improved test methods for an easy and quick determination of the rheological parameters should improve the quality control, ensuring that decisions are based on science rather than qualitative interpretations on the job site. It will also bring further understanding into the effect of the yield stress and plastic viscosity on the form filling characteristics which will be a major step towards bringing rheology from research and development into the field of practical applications. For instance, was used for acceptance testing at a full-scale SCC bridge casting and proved applicable in relation to controlling the dynamic segregation resistance, the moving front of the concrete, and finishing with sloping surfaces. REFERENCES [1] Thrane, L. N., Simulation and Verification of Form Filling with Self-Compacting Concrete, in Proc. of the Nordic Concrete Research Meeting, pp. 89-91, 2005, Sandefjord, Norway. [2] Skarendahl, A., The Present - the Future, in Proceedings of the 3 rd International RILEM Symposium on Self-Compacting Concrete, Reykjavik, Iceland, Ed. O. Wallevik and I. Nielsson, RILEM Publications S.A.R.L, pp. 6 14, 2003. [3] Beaupre, D., Chapdelaine, F., Domone, P., Koehler, E., Shen, L., Sonebi, M., Struble, L., Tepke, D., Wallevik, O., Wallevik, J. E., Comparison of concrete rheometers: International tests at MB (Cleveland Oh, USA) in May, 2003, Ed. C. F. Ferraris and L. E. Brower, NISTIR 7154. [4] Ferraris, C. F., Martys, N., Relating Fresh Concrete Viscosity Measurement from Different Rheometers, Journal of Research of the National Institute of Standards and Technology, vol. 108, no. 3, pp. 229 234, 2003. [5] Roussel, N., Coussot, P. Fifty-Cent Rheometer for Yield Stress Measurements: From Slump to Spreading Flow, Journal of Rheology, vol. 49, no. 3, pp. 705 718, 2005. [6] Nguyen, T.L.H., Roussel N., Coussot P., Correlation between L-box Test and Rheological Parameters of a Homogeneous Yield Stress Fluid, Cement and Concrete Research, vol. 36, pp. 1789-1796, 2006. [7] Nielsson, I., Wallevik, O.H, Rheological Evaluation of some Empirical Test Methods- Preliminary Results, in Proc. of the 3 rd International Symposium on Self-Compacting Concrete, Reykjavik, Ed. O. Wallevik and I. Nielsson, RILEM Publications S.A.R.L, pp. 59 69, 2003. [8] Geiker, M., Brandl, M., Thrane, L. N., Bager, D. H., Wallevik, O., The Effect of Measuring Procedure on the Apparent Rheological Properties of Self-Compacting Concrete, Cement and Concrete Research, vol. 22, no. 11, pp. 1791 1795, 2002. [9] Thrane, L. N., Szabo, P., Geiker, M., Stang, H., Pade, C., Simulation and Verification of Flow in Test Methods, Proc. of the 2 nd North American Conf. on the Design and Use of SCC and the 4 th Int. RILEM Symp. on SCC, 551-556 (1), 2005, Chicago, USA. 358