STABILIZING SELF-LEVELLING CONCRETE WITH POLYSACCHARIDE ADDITIVES Jacob Terpstra AVEBE Industrial Specialties The Netherlands Abstract One of the high-performance concretes made possible by superplasticizers is self-levelling concrete (SLC). A distinct property of SLC is its very low (plastic) viscosity. This allows for a very fast and large flow so that the concrete becomes self-levelling. The low viscosity introduces the problem of instability of the concrete leading to segregation and bleeding. Stabilization of the concrete is necessary but should not impede the flow properties by increasing the viscosity. A new starch-based stabilizing agent has been developed which provides such a rheology to the concrete that bleeding and segregation is effectively prevented without affecting significantly the ability to flow. The mechanism of action of the stabilizer allows for its use in combination with any type of superplasticizer with a preference for the melamine and polycarboxylate-ether types. The stabilizer allows for a more robust concrete as it prevents failing due to small deviations in the mix design. Therefore, both the workability and the ease of production of SLC are increased by the stabilizing agent. 1. INTRODUCTION Over the past few years there has been much development of self-compacting concrete (SCC) and self-levelling concrete (SLC). Advantages of SCC and SLC compared to normal concrete relate especially to the on-site workability, meaning less heavy labour and ease of application. Despite these significant advantages production of SCC and SLC is only limited. One of the reasons is that very high demands are set on the exact composition of SCC and SLC in view of stability against bleeding and segregation. The traditional SCC has a high powder content (500-600 kg/m 3 of cement and filler such as fly ash or limestone powder).[i] The resultant SCC has a high coherence contributing to its stability. However, the exact mix design is critical. Aggregate grading, choice of filler and the addition of chemicals (including water) are important. A 1% deviation in actual water content can cause a mixture to fail. Determination of the concrete rheology can shed more light on the flow and stability properties. To this end use is often made of rheological parameters as defined in the Bingham Page 1
model: plastic viscosity and yield value.[ii] For obtaining a stable concrete either the yield value or the plastic viscosity (or both) need to be sufficiently high.[ii] For powder-stabilized SCC the plastic viscosity of the SCC is high. Also stabilizing admixtures generally improve the stability by increasing the plastic viscosity of the concrete. For maintaining the required fast flow properties of SLC the plastic viscosity must be sufficiently low. Therefore, a high powder content or a viscosity enhancing stabilizing agent cannot be used. In this article the focus is on the stabilization of SLC using a new starchbased stabilizer developed particularly for this type of concrete in order to maintain its specific flow properties. 2. INTRODUCTION TO STARCH TERMINOLOGY Starch is a natural polysaccharide obtained from plants like potato, maize, wheat, tapioca, rice, etcetera. In these plants starch is deposited in water-insoluble grains. In general, starch exists of two types of molecules: amylopectin that is strongly branched and the linear amylose. Potato starch contains 80% amylopectin and 20% amylose. Figure 1 shows an overview and a detail of the strongly branched amylopectin structure. A B Figure 1: Model of amylopectin structure (A) and detail showing branching point (B). The developed stabilizing agent is based on potato starch. This starch is very pure compared to e.g. cereal starches that contain large fractions of fats and proteins. In order to obtain the functionality of stabilization the starch is both chemically and physically modified. These modifications also ensure that the stabilizer is safe to apply in concrete. Page 2
3. EXPERIMENTAL 3.1 Materials Cement: Portland cement type CEM I 52.5 (Vicat) Filler: (a) Class F Fly ash, (b) Powdered limestone with a specific gravity of 2700 kg/m 3 and a Blaine specific surface area of 385 m 2 /kg. Aggregate: River sand (0-5 mm) and gravel (a) 4-16 mm and (b) 5-12 mm. Admixtures - Superplasticizers: (a) Melamine sulphonate superplasticizer (Rheobuild 2500, MBT), (b) Polycarboxylate ether superplasticizer (Glenium 27, MBT) - Stabilizer: starch-based stabilizer (Foxcrete, AVEBE), (a) Foxcrete S20, 20% solids content, (b) Foxcrete S100F (powder). Concrete mixtures with the compositions mentioned in table 1 were investigated varying the type of filler, the sand to gravel ratio and the type of stabilizer. For the mixtures nr. 5 and 6 a high dosage rate of the superplasticizer was set in order to make the concrete critical concerning the stability. Table 1: Compositions of SLC mixtures. The type of component is coded as listed above. Amounts (Am.) of components are given in kg/m 3. Mix Cement Filler Aggregate Total Admixture nr. Sand Gravel water Superplasticizer Stabilizer Amount Type Am. Type Am. Type Am. Type Am. 1 280 a 120 900 a 900 170 a 4.8 a 1.3 2 280 a 120 1100 a 700 180 a 4.5 a 0.9 3 280 b 120 900 a 900 170 a 4.5 a 1.3 4 280 b 120 1100 a 700 195 a 4.5 a 0.9 5 280 a 120 1000 b 800 210 b 6 a 2.0 6 280 a 120 1000 b 800 210 b 6 b 0.13 3.2 Methods - The flow of the SLC was measured with a truncated cone of 120 mm height and having an upper and lower diameter of 170 and 225 mm, respectively, standing on a flat surface. After filling the cone is slowly lifted. The resulting diameter of the concrete mass is measured in two directions and averaged. - The funnel time was determined using the O-funnel test as a measure of the speed of the flow. The time is measured which is required for 10 litre of concrete to flow without vibration through a funnel with an aperture diameter of 50 mm. - The amount of bleeding was measured on five litres of concrete in a basin being 245 mm long, 245 mm wide and 100 mm high. After 90 minutes the bleeding water at the surface was removed and weighed. The bleeding rate is defined as the ratio of bleeding water and the total water in the concrete. Page 3
- The resistance to segregation was determined by filling a column with the concrete to be tested. It was left to stand until it started to set after which the surrounding column was removed and cross-sections were sampled from the upper, middle and lower part of the concrete column. Each sample was washed and sieved through a five millimetre mesh screen and weighed.[iii] The segregation S is calculated as: Wlower Wupper S (%) = 100 W average where W lower and W upper are the amounts in kg/m 3 of gravel larger than 5 mm in the lower and upper cross-section from the column, respectively, and W average is the average amount of gravel of all three sections. Segregation is defined as a segregation S larger than 5%. - The compressive strength was measured on three cylinders with a diameter of 110 mm and a length of 220 mm. The rate of loading was 5 kn/s. - The unrestrained drying shrinkage was measured according to the French standard NFP15-433 on prismatic samples having dimensions of 70 70 280 mm. The specimens were demoulded after 24 hours and kept at 20 ± 2 C and 50 ± 10% RH. - The water permeability of the concrete was determined after 90 days on discs with a diameter of 110 mm and a height of 50 mm using Darcy s relationship. 4. RESULTS AND DISCUSSION 4.1 SLC properties before setting The SLC mixtures as mentioned in table 1 contain a relatively low amount of cement and fly ash in order to prevent a too high plastic viscosity. The results on flow and stability with these SLC mixtures are shown in table 2. For the mixtures 1-4 in which the type of filler and the sand to gravel ratio were varied, the quantity of mixing water was adapted in order to obtain approximately the same flow. Table 2: Flow, funnel time, specific gravity, segregation and bleeding rate of the concrete mixtures including the stabilizer unless mentioned otherwise. nd denotes not determined. The rate of bleeding for the mixes 5 and 6 was determined after 3 hours compared to 90 minutes for the mixes 1-4. Mix nr. Flow (mm) O-funnel time (s) Specific gravity Segregation (%) Rate of bleeding (kg/m 3 ) without with (%) stabilizer stabilizer 1 630 nd 2360 9 2 0 2 600 nd 2340 8 2 0 3 620 nd 2340 8 4 0 4 570 nd 2290 5 3 0 5 610 7 nd nd 3 0.6 6 620 7.6 nd nd 3 0.5 Page 4
A first note on the flow results is that the cone used has smaller dimensions than the usual Abrams cone. Because of this the reported values of the flow are smaller than usually encountered. Figure 2 shows that the concrete has spread totally and that the smaller value of the flow is only caused by the smaller amount of concrete used in the test. Assuming the same average thickness of the concrete after spreading on the flow table the flow can be calculated to be 750-800 mm using an Abrams cone. The funnel used has a smaller opening than a V- funnel resulting in longer funnel times and a more critical assessment of the flowing properties. Figure 2: Flow result of SLC mix nr. 6. The tested concrete mixtures had a flow around 610 mm and a flow time between 7 and 8 seconds (for mixes nr. 5 and 6). Concrete mixtures with these flow characteristics have a strong tendency to segregate as is clear from the results for mixtures 1-4 (see also figure 3). 10 8 Segregation S (%) 6 4 2 Without stabilizer With stabilizer 0 1 2 3 4 Mix nr. Figure 3: Segregation without and with stabilizer for the mixtures 1-4. Page 5
For these four concrete mixtures the use of the stabilizer decreases significantly the segregation resulting effectively in stable concrete (segregation S smaller than 5%). Next to the effect on segregation the stabilizer also leads to a negligible bleeding for all the concrete mixtures. The combination of limestone powder as a filler and the high amount of sand (mix nr. 4) leads to a relatively stable concrete also without the stabilizer. Adsorption of water by the limestone powder and by the fines in the sand contributes to the stability of the concrete. However, this also leads to a decrease of the flow despite the higher w/c. The stability of the concrete can also be observed from figure 2. The gravel is spread homogeneously over the whole diameter and no bleeding is observed at the edges. The edges further give an indication on the mechanism of action of the stabilizing agent. The picture clearly shows that the edges have a definite height which is a sign of a certain yield value. This suggests that the stabilizing agent stabilizes the concrete by introducing a yield value which is large enough to keep the concrete homogeneous. At the same time the yield value is small enough to enable a large flow of the concrete under its own weight. The fast flow indicates that the stabilizer does not have a significant effect on the plastic viscosity of the concrete. Results of rheological investigations confirm this view on the mechanism of action.[iv] Comparison of the SLC mixtures nr. 5 and 6 shows that both types of stabilizing agents have a good stabilizing functionality leading to similar fresh properties of the concrete. Calculating the dosage rate of the stabilizers based on solid content reveals that the stabilizer used in mixture 6 is more effective. The dosage of 0.13 kg/m 3, meaning 0.046% on (the low amount of) cement, makes it as effective as viscosity enhancing agents such as welan gum.[v] 4.2 Characterization of hardened concrete The development of the compressive strength is good as shown in table 3 for the mixtures 1-4 at different times of hydration. The early strength of the mixtures 3 and 4 is higher than that of mixtures 1 and 2 because of the contribution of the powdered limestone. In contrast, the long-term strength of mixtures 1 and 2 is higher which can be attributed to the puzzolanic property of the fly ash. The lower compressive strength for mixture 4 can be attributed to the higher water/cement ratio. Table 3: Compressive strength in MPa for the concrete mixtures 1-4 containing the stabilizing agent. Concrete nr. 18 hrs 2 days 7 days 28 days 90 days 1 8.7 21.9 40.6 49.9 62 2 8.1 21.3 40.4 47.6 60 3 12.6 23.1 41.3 45.7 51 4 9.8 20.9 34.4 40.4 42 The drying shrinkage and water permeability of the hardened concrete mixtures were investigated for the SLC mixtures 1-4. The shrinkage is 10-25% larger than is usual for normal concrete (table 4). Possibly, this increase can be at least partly ascribed to an effect of the superplasticizer.[vi] The water permeability is in the order of 10-12 to 10-13 m/s which points at a good durability. The permeability is lower for the concrete mixtures 1 and 2 which can be due to the presence of the puzzolanic fly ash. Page 6
Table 4: Unrestrained drying shrinkage and water permeability both at 90 days for the mixtures 1-4. Concrete nr. l/l (µm/m) Water permeability (10-12 m/s) 1 550 0.46 2 640 0.95 3 630 2.0 4 660 1.9 5. CONCLUSIONS The starch-based stabilizing agents tested in this investigation are very effective in stabilizing SLC while maintaining the flow properties (both the size and the speed of the flow). Prevention of segregation and bleeding also has a positive effect on the hardened properties of the concrete. Therefore, the stabilizing agents allow for the delicate tuning of the rheological properties of the concrete towards optimum stability, workability and hardened concrete properties. Finally, as confirmed by practical application of the stabilizing agent, the stabilizing action will also increase the robustness of the concrete stability against variations in the exact mix design. This contributes to the ease of production, the pumpability of the concrete and to the workability in general. 6. REFERENCES [i] Takada, K., Influence of Admixtures and Mixing Efficiency on the Properties of Self-Compacting Concrete, PhD-thesis, Delft University of Technology, (Delft University Press, ISBN: 9040725012, 2004). [ii] Níelsson, I. and Wallevik, O.H., Rheological evaluation of some empirical test methods preliminary results, in Self-Compacting Concrete, Proceedings of the 3rd International RILEM Symposium, Reykjavik, August 2003 (RILEM Publications, 2003), 59-68. [iii] Rols, S., Ambroise, J. and Péra, J., 1999. Effects of different viscosity agents on the properties of self-levelling concrete. Cement and Concrete Research 29: 261-266. [iv] Terpstra, J., to be published. [v] Khayat, K., Ghezal, A., Effect of viscosity-modifying admixture-superplasticizer combination on flow properties of SCC equivalent mortar, in Self-Compacting Concrete, Proceedings of the 3rd International RILEM Symposium, Reykjavik, August 2003 (RILEM Publications, 2003), 369-385. [vi] Neville, A.M., Properties of concrete, (Addison Wesley Longman Limited, 1998). Page 7