Toward Mix Design for Rheology of Self-Compacting Concrete

S.G. Oh, T. Noguchi and F. Tomosawa
University of Tokyo, Japan

Self-compacting concrete (SCC) can produce much higher fluidity with no occurrence
of segregation, thanks to its lower yield value and higher viscosity than conventional
concrete. However, the fresh property of SCC cannot be evaluated by conventional
consistency test such as slump test. It must be evaluated from the rheological point of
view, of which SCC is considered as a Bingham plastic fluid with two parameters, the
yield value and plastic viscosity. A new type of rheometer was developed to measure
these two parameters and experimental works were conducted on the imitative concrete
as well as the real one, in which the rheological properties of the matrix and the volume
fraction of aggregates were variously changed. The result of this research allow us to
develop an important parameter in establishing a new mix design method for the SCC.

1. Introduction
In recent years, a lot of study is done on how to improve the performance of concrete,
especially on topics regarding how to increase the strength, durability, and flowability of
concrete. At the same time, there are a number of reports published on how to evaluate
and predict those performances. High Strength concrete has become one of the hottest
topic since 1980s, and it is now possible to have structures that are built with concrete
over 100 MPa compressive strength, something we would not have even thought of
before. This kind of excitement has also triggered further development on the
construction techniques and materials used for such concrete. At the same time, there
were concern on the maintenance of concrete structure. How is it possible to minimize
the cost of maintenance, and to prolong the life of concrete. Thus, a lot of the research
shifts to the development on the durability of concrete.

OH, Toward Mix Design for Rheology, 1/12

1). which allows us to investigate on how rheology affects the behavior of the aggregates. As we can see. adding cement paste will change the interaction among aggregates. a new evaluation method must be developed. since normal slump method could not be used effectively. However. we have decided to just use the coarse aggregates and see how an increase in size and volume will change the viscosity and yield value of concrete. we have also adopted Kennedy’s theory of Excess Paste. Applying the theory of Multi-Phase flow. 2/12 . and fluid mechanics. we considered concrete as a two-phase flow. filled up with cement paste..1 Excess paste In 1918. it is necessary to have not only enough cement paste to cover the surface area of the aggregates. Abrams proposed a theory regarding reasonable mixing method for concrete in his paper “Design of Concrete Mixes”. There is still a lot of uncertainty concerning the appropriate mixing proportions and method. And about 20 or so years later. This is sort of like a dispersion effect. During the formulation. i. D. And there are a number of activities at the moment on how to combine both strength and durability into this third main property of concrete.The only thing that got left out until recently is research relating to the flowability of concrete.e.A. Moreover. the left side model shows that the aggregates closely contacted to each other. which gives a relationship on the interaction between the cement paste and the aggregates. We need to develop a method or a way of measure to quantify the flowability of this so called self-compacting concrete (SCC). it is necessary to quantify the most basic properties of the concrete with the use of the mechanism of the particles. His theory explained the fact that to attain workability. which is crucial to the understanding of the mechanism of the workability of fresh concrete. we chose the theory of Rheology as a tool to help attain such a general proportioning system. aggregates are pushed away from each other. It is about the water and cement ratio and the proportion of coarse aggregate. with void in between them. as the model on the right side show. Since doing an analysis with fine aggregates would be rather difficult to perform. so as to minimize the friction between them. 2. But if we mix this model with cement paste. OH. this closely packed aggregates are then separated by this thin film of cement paste around them. Also we could notice that the void is gone. high workability concrete is still something quite unknown to many researchers. For the purpose of having a right prediction on the flowability of concrete during fresh state. Toward Mix Design for Rheology. In the figure below (Fig. Kennedy proposed the “Excess Paste Theory” [1]. With all these uncertainties. Application of excess paste theory 2. but also more of it to give better flowability.

Pe = Vp – P c (1) P t p = ------es all (2) However. in order to get a more accurate value of tp(mm) we could include several other terms in the calculation.2 (b).Without a film of cement paste around them. as shown in fig.2 (a) shows a concrete sample with a good spacing between the aggregates that were covered by cement paste. the movement between aggregates would generate much friction and make workability impossible. as shown by Eq. as to minimize the friction between them. it is necessary to have a good spacing between the aggregates. Putting them together as the Eq. 2. as G. the total volume of paste. and squeeze out the excess cement paste surrounding them. What left is a top layer with just the paste itself. (2). We could then compact the aggregates. 1 Excess paste theory OH. And the cement paste that wraps around the aggregates is the so called ‘excess paste’ (Pe). To attain high workability on concrete. and denoted that as Va. with just enough cement paste to fill in the void space.(1). (3) shown would give a better result. and below it a compact state of aggregates. Fig. we need to consider the ratio of volume of aggregate to volume of concrete. which is calculated by subtracting Pe from Vp .2 The computation method for excess paste Using the assumption of the two-phase flow theory. 3/12 . First of all. This cement paste in between the voids is what I called the ‘compact paste’ (Pc). Toward Mix Design for Rheology. Aggregate Excess paste Compacted paste Add paste Void Thickness of excess paste (Compacted) (Dispersion) Fig. we considered that there is a continuous (good) grading of aggregates. as shown in Eq. We can also calculate the thickness of the excess paste (tp) by just simply dividing the volume of the excess paste (Pe) by the total surface area of the aggregates (sall). then we need to consider the percentage of solid volume.

5 . f = π ⁄ 6 . However. Heywood has proposed to use a direct method to measure aggregates in terms of length (L). crushed aggregates’ is f ⁄ k = 7. In case of sphere. Toward Mix Design for Rheology. he can formulating them into the equations below that will finally allow him to obtain the surface area S. By doing so.5 . the sieve size Do. and f ⁄ k = 6 .Aggregate Paste Concrete Pe Vp Pc Compacting Va (a) (b) (c) Fig. Notice that in his equations. we have considered a good grading on the aggregates used. there are several additional values. 2 Excess paste theory V 10 t p =  1 – -----a × 100 × --------------  S × Va G (3) where S is the specific surface area (mm2/mm3) For this research. width (B). and thickness (T). k = π .5. B m = --T (4) L n = --B (5) OH. but it is still quite hard to say their method is a good and proper one. the projected diameter Dp. Researchers like Powers [2] and others have proposed different ways to attain such a value. And assuming that they would turn out to have the same percentage of volume. And gravel’s is f ⁄ k = 6. it is quite difficult in reality to calculate or even to predict the total surface area of aggregates. The ratio of fine and coarse aggregates are in a ratio of 0. We have proposed such a method based on the theory done by research in particle engineering by Heywood [3]. 4/12 . the shape volume coefficient k and the shape area coefficient f.

and from this. we can sum up the product of each individual si and Ni and then divided by Va to get the specific surface area S. Toward Mix Design for Rheology.T L B T(厚さ):粒子をはさむ平行平板の間隔の最小値 Fig. 5/12 (10) . (10). For this reason. we will calculate the total surface area (sall) of aggregates by the following: s all = ∑s × N i i (9) i where si is the surface area.=  --------------2 Do  m + 1 2 1⁄2 (6) 3 v = kD p (7) 2 s = fD p (8) However. we have changed the equations around. This Ni value is a result from Eq. 3 Shape of aggregate B(幅) :厚み(T)を決定した平面に直角な方向で粒子をはさむ平行平板       の間隔の最小値 (長さ):厚み( )と幅( )を決定した平面に互いに直交する方向で粒子  2nm  Dp - -----. which will shown in the following equations: Va N i = ---------3ikD pi OH. what Heywood has proposed would only work for an individual grain. First of all. Ni is the numbers of a particle of size x. so that it would allow one to calculate the total surface area of aggregates. and would not work for a continuous grading of aggregates.

slump flow is first check. and with careful proportioning. The model paste were spread from a flow cone specified in JIS R 5201 (bottom diameter = 100. all the mixtures for self-compacting concrete were prepared with normal protland cement. But the obtained result will then verify later with using real concrete. and with volume ratio of aggregates ranging from 0. 3. which has a aggregate volume ratio of 0.V a 0. When polymers mix with water. we are able to simulate the viscous state of high workability concrete.1 Factors and levels Each of the different fractions of aggregate size and volume are shown in Table 1.0 +/.3 Experimental subject and method The subject concern here is to measure the slump flow and rheology constant of both the model paste and concrete.3. We are examining concrete models with aggregate size ranging from 5 mm uniformly to one with a good continuous grading.0. The flow tests were conducted on models and concrete.5 mm.5 mm. Aggregates are then replaced by artificial light weight aggregates. it is actually better to use a model concrete. Toward Mix Design for Rheology. The one highlighted with black color is the control. In the case of concrete.2 Materials used and mix proportions Since result obtained from using real concrete may vary due to the fact that concrete hydrates through time. Table 3 shows mix combinations of different aggregate size and volume ratio for each experiment trial. height = OH.15 kD p 20 n s all = ∑ i f Va -. as shown in Table 2. 3. we have selected polymers.× -------i k D pi (11) (12) n Va = ∑V ai (13) i s all S = -----Va (14) 3.0 +/. 6/12 . River sand was used as fine aggregate. For this model concrete.0. top diameter = 70. The experiment is done in the following order: after mixing the model paste and concrete.+ … +  fD 2p × -----------s all =  fD p 0. a viscous liquid will form.1 to 0. The experimental method 3.15 × -------------3  3  20   kD p0. whereas crushed stone was used as coarse aggregate.35.15  V a20   2  .

we are able to obtain 2 important rheology properties of these model pastes.3 %.F. and the shear stress and torque will also be measured.G.68 ) Coarse agg.M.= Viscosity agent Table 2: Factors and levels Factor Model (Flow of paste (cm)) Rheological properties of paste Concrete (Water cement ratio) Model Volume of aggregate (Fine : Coarse = 1 : 1) Concrete Level Symbol 20 25 30 35 P1 P2 P3 P4 0.0.26 0.5 mm) without any dropping motion.0 +/.47 ) Cement Portland cement (S.Table 1: Materials and properites Materials Model Concrete 1.Percentage of solid volume = 64.45 ) S.A.53 61C 59C 57C 53C 60. Artificial light-weight fine aggregate (S. to test the flow and rheology properties.45 ) Coarse agg.32. River sand (S.45 0.55 0. =2.= 2.F.Percentage of solid volume = 61. The speed of the rheometer will increase gradually from 1 rpm to 4 rpm.40 N26 N32 NV40 0. 2.G.F.P.60 0. =2.Percentage of solid volume = 58. Toward Mix Design for Rheology.G. Properties Paste Water + Polymer Fine agg.A. In the case of concrete.M.16) Fine agg. 3.= 2. Then if the mix shows no sign of segregation.F.P.= 3. Result and discussion From the experiments.61 0.32 0. which would allow our specially designed cylindrical rheometer machine.40 60 55 50 45 40 0. Crushed stone (S.0 %.59 0.= 1. = Specific gravity S. It is possible to draw a linear OH.3 Cellulose type S. = Superplasticizer V. =6. the model will be poured into a container.1 = 1.G. 7/12 . 4.57 0. and the diameters of the spread were measured as the flow values.66. the plastic viscosity and the yield value. slump tests stipulated in ASTM C 143-78 (Test method for slump of portland cement concrete) were conducted to measure the diameters of the horizontal spread.7.M. =2. Artificial light-weight coarse aggregate (S.66.50 0. 4 (Rheometer for concrete).2 Poly-carboxylic acid V. as described earlier in Fig.M.5 %.G.G.

Motor 1084.5 mm Gear Handle Revolving Cylinder    .




5. 8/12 . the lesser the extent of the flow. the slower the speed of the flow. OH. as shown in Fig. We can see that the rheology constants increase in proportion with increasing amount of polymer. and the higher the viscosity.   regression line for each of the models. Table 3 shows the mix proportion and the resulted flow of the models. It is obvious to see that the higher the yield value. Toward Mix Design for Rheology. which is in consistence with the Bingham’s plastic fluid.

The results obtained above are too spread out to generalize quantitatively. OH. From this. As noticed from the figures.70 450 G G 400 60 E . causes an increase in the value of the rheology constants. we could generalize the data into a result that would be consistent for all models by using relative rheology constants and relative thickness of excess paste [4]. And we can rearrange the equation into Eq. 9/12 . The method of obtaining such a relative value will be explained in the following. we could find a relationship between the resulted rheology constants and the thickness of excess paste. 350 50 A G E 30 C 20 A G E E P2 250 C P3 200 G P1 C P3 150 G E G E C E 50 C A C 0 0 Thickness of excess paste (μ m ) G E C A 0 50 100 150 200 250 300 350 400 450 A P4 A 100 A 0 E P2 E G A P4 C A 10 300 G P1 C 40 G C E A G E 50 100 150 200 250 300 350 400 450 Thickness of excess paste (μ m) Fig. However. which is when the ratio is about 0. 7. Relative thickness is defined as diving the thickness of excess paste t p x . This. the thickness of excess paste will at the same time decreases. (16) (Fig. we work on for using real concrete.40 to 0. 6 Thickness of excess paste and plastic viscosity in model concrete Fig.5. since each model shows a quite different rheology constant. 8). of which Px signifies a particular model with particle size x. the increase in rheology constants shoots up dramatically at around the 250 μm. as shown in both figures. (15). by the projected Diameter of aggregate D px . Toward Mix Design for Rheology. 7 Thickness of excess paste and yield value in model concrete x Aggregate Thickness of excess paste of particle x y z Fig.60. As we increase the ratio of aggregates. as shown by Eq. 8 Thickness of excess paste If we mix the model paste with aggregates with a volume ratio ranging from 0. 6 and Fig. as shown in Fig.

Toward Mix Design for Rheology. and we can see that all models fall into one generalized regression curve. P e x . as shown by Eqs. 9 and Fig. (17)). but the elaboration of the method is out of the scope of this paper. (22). which are calculated by dividing that of the concrete with that of the paste. can be calculated by multiplying t px with sx. p ex = t px × s x (17) P ex = n x × p ex (18) P ex = n x × Γ × D px × s x (19) This total volume of excess paste can be extended to a continuous array of aggregates. This number n can be estimated quite accurately by using the weight method. (18) and (19) shows the equations for total volume of excess paste around n numbers of aggregate x. just by simply summing up the P ex for each aggregate of size i. From these two curves. we established two generalized equations. n Pe = ∑P (20) ex i n Pe = Γ ∑n s D i i pi (21) i By rearranging the above Γ = -------xDpx (15) t px = Γ × D px (16) Then the volume of the excess paste around a particular aggregate x (Eq. (20) and (21). 10/12 . And Eqs. we get the equation for the relative thickness of excess paste. p ex . 10 show the result obtained from using this relative relationship. Pe Γ = -----------------------n n i s i D pi ∑ (22) i The only thing left now is the method for computing the relative rheological constants. which signifies the surface area of aggregate x. which are in the same shape with that of OH. as shown by Eq. Fig.

2 0. as shown by Fig.2 Relative thickness of excess paste Relative thickness of excess paste Fig.69+1 A 14 r 12 τyr = 0.04 0.22 +1 (24) τ yr = 0.92) 50 10 8 E A C G 6 G P1 E P2 C P3 A P4 A G 30 A 20 E C 4 E C G 40 E G C A 2 E G G P1 E P2 C P3 A P4 E C G 10 A C G E A C G G E E 0 0 0 0. 11/12 . 12 Relative thickness of excess paste and relative yield value in concrete Einstein’s equation of two-phase flow.97) E G C (R2 = 0.06 0.16 70 η = 0.05 0.12 Relative thickness of excess paste Fig. 11 and Fig.02 0. OH.0525Γ -2. 10 Relative thickness of excess paste and relative viscosity in model concrete and relative yield value in model concrete 125 35 30 ? H B 20 H 100 P 25 M 15 > 10 H N26-59C P N32-61C B N32-59C M N32-57C > N32-53C ? P ? 75 B M 50 > NV40-61C H N26-59C P N32-61C B N32-59C M N32-57C > N32-53C ? NV40-61C 25 5 0 0 0 0.22+1 A 60 (R2 = 0.1 0.05 0.1 0. Toward Mix Design for Rheology.02 0.04 0.1 0. η r = 0. 11 Relative thickness of excess paste and relative viscosity in concrete 0 0.08 0. that will allow us to estimate accurately any other models.0525Γ After applying this relative method to the real concrete model. 9 Relative thickness of excess paste Fig.15 0 0.12 Relative thickness of excess paste Fig.1 0. as shown by Eqs.69 +1 (23) – 2.0705Γ -1.06 0.0705Γ – 1. similar results obtained and we could generalized the result by the above 2 equations.08 0. (23) and (24).15 0. 12.

Co. and Tomosawa.69 +1 – 2. Noguchi. Powers.. 985-988 OH. A-1 Materials and Construction. which will be explained in another paper. T. 1940 2. which is shown as follow. 1. however. ‘Evaluation of rheological properties of concrete by thickness of excess mortar’.. John Wiley & Sons. 7. S. M.T. References 1. 373-400. 1968 3.G. Vp Γ = -----------------------n ni si Dpi ∑ i 2. 1998. 1980 4. ‘The Properties of Fresh Concrete’. For a continuous grading of aggregates. the relative thickness of excess paste can be calculated by the following equation. C. New York.5.0525Γ From the concepts above. T.K. Chemical Publ. η r = 0. Conclusion The results of this research are summarized as follow.. F.22 +1 τ yr = 0.. New York. Acknowledgement The authors thank Barry Cheong from UC Berkeley for his participation in this research and his support into writing this paper. Proceedings of the American Concrete Institute. 12/12 . Kanematsu.. A consistent equation for estimating accurately the relative yield value and relative plastic viscosity of SCC is established. ‘Particulate of fresh concrete’.0705Γ – 1. Summaries of Technical Papers of Annual Meeting of Architectural Institute of Japan. J. we can establish a new design method for the SCC. ‘The Design of Concrete Mixes’.C. 6.. Beddow..36. Kennedy. Oh. Toward Mix Design for Rheology. Vol.