RELATIONSHIPS BETWEEN RHEOLOGICAL PROPERTIES AND PUMPING OF FRESH SELF-COMPACTING CONCRETE

Transcription

1 RELATIONSHIPS BETWEEN RHEOLOGICAL PROPERTIES AND PUMPING OF FRESH SELF-COMPACTING CONCRETE Dimitri Feys (1),(2), Ronny Verhoeven (2) and Geert De Schutter (1) (1) Magnel Laboratory for Concrete Research, Department of Structural Engineering, Faculty of Engineering, Ghent University, Belgium (2) Hydraulics Laboratory, Department of Civil Engineering, Faculty of Engineering, Ghent University, Belgium Abstract In literature, the research topics on self-compacting concrete are numerous, almost covering the whole range from the raw materials to fully developed structures. In this research range, one particular item is still missing, namely the casting of SCC by means of pumping. Currently, not much is known on the relationship between the fresh concrete properties and the pumping process. This paper describes full scale pumping tests on self-compacting concrete. It contains two parts: the first part is dealing with the influence of the rheological properties of the concrete on the pumping process, showing that viscosity and shear thickening have a major importance. The second part contains the influence of pumping on the rheological properties of the concrete, where it is clearly shown that viscosity will decrease due to pumping. Structural breakdown and air content change the rheological properties of the SCC. If structural breakdown dominates the effects of the air content, the SCC will show a larger tendency to segregation. If the effects of the air content dominate, the yield stress of the SCC will increase, possibly leading to improper filling of the formwork. 1. INTRODUCTION The research performed on self-compacting concrete worldwide is very broad, dealing with several topics from the characteristics of the constituent materials, over the fresh properties to the hardening and hardened phase [1]. In practice, the advantages of self-compacting concrete are numerous, but one can be especially highlighted: SCC, and only SCC, provides the ability of filling formworks from the bottom by means of pumping. As the knowledge dealing with pumping of SCC is still quite restricted, a research project dealing with the interactions between the rheological properties and pumping of SCC has been executed at Ghent University [2]. This paper will describe some of the main findings dealing with pumping of SCC. It will cover two main subjects, dealing with the influence of the rheological properties on the pumping pressures and dealing with the influence of the 64

2 pumping action on the rheological properties of SCC. Finally, some recommendations will be given, but further investigation is definitely needed. 2. PUMPING OF CONCRETE In literature, pumping of concrete has been discussed both as a purely scientific phenomenon [3-6], and also from a practical point of view [7-8]. From a scientific point of view, a distinction has been made between saturated and un-saturated concrete [3-5]. In case of unsaturated concrete, at some locations in the pipes, the water has migrated to other regions in the concrete and the stress transfer in the concrete is occurring by friction between the aggregates. As a result, the pumping pressure needed increases significantly, and when the pump cannot deliver this pressure, blocking occurs. For the saturated concretes, the opposite is true and the concrete is supposed to flow (or move) in the pipes, without large problems. As a result, water migration should be avoided as much as possible in order to avoid friction and blocking. The movement of traditional concrete in pipes is supposed to occur as a large plug advancing at a uniform velocity in the centre of the pipe and a kind of slippage or lubrication layer near the wall, where the velocity gradient is concentrated [4-6]. This slippage or lubrication layer has a thickness between 1 and 8mm, according to different authors, although direct measurements are not straightforward. In case of self-compacting concrete, as the amount of aggregates is reduced and segregation is more or less prevented [1], the concrete is supposed to remain saturated during the flow in pipes, except during start-up. The assumed velocity field for traditional concrete can be questioned, as the radius of the plug is only determined by the pressure loss in the pipes, the geometry of the pipe (length and diameter) and the yield stress of the material [5]. The lower the yield stress, the smaller the plug radius, and it is possible that the a large part of the concrete in the pipes will undergo shearing. In this case, pumping of SCC must be studied in a different way compared to traditional concrete. 3. TEST SETUP 3.1 Pump and pipes The pump used is a truck-mounted piston pump (fig. 1), available in industry. This type of pump contains two cylinders which alternately push concrete inside the pipes and pull concrete from the concrete reservoir. A powerful valve changes the connection between the pipes and the cylinders. Theoretically, each cylinder contains 83.1 l of concrete in our case. 65

3 Figure 1: Left: concrete piston pump. Right: Short circuit (25m). Behind the pump, circuits have been built by means of steel pipes with an inner diameter of 106mm and a thickness of 3mm (fig. 1). Short (25m) and long circuits (81 105m) have been built, creating a loop circuit. At the end of the loop, a reservoir has been installed, capable of taking samples of the pumped concrete. During the tests, the valve at the bottom of this reservoir has been kept open and the concrete falls back inside the reservoir of the pump. In this way, the same concrete has been used continuously. One can argue that this procedure is not similar to any practical situation, but it is the only way to restrict the amount of concrete needed and to investigate the influence of pumping on the rheological properties. 3.2 Measurement systems The pressure loss over a certain length of pipes has been measured by means of two flush-mounted pressure sensors, which have been located at a fixed distance from each other in a horizontal, straight part of the circuits. Additionally, strain gauges have been attached to the outer pipe wall in order to provide a back-up for the pressure sensors (fig. 2) [6]. After the evaluation of this measuring principle, strain gauges have been installed in the long circuit as full measuring units. Discharge measurements have not been executed directly. On the other hand, as during the change of the valve of the pump, the pressure drops and rises in one second (fig. 3), the time needed to pump 83.1 l of concrete can be measured quite accurately. In this way, a method has been developed in order to measure the discharge, which has been verified by means of a calibration. 3.3 Concrete During the research project, 19 different concretes have been pumped, of which 18 were designed to be self-compacting. The standard SCC-mixes have been produced with CEM I 52.5 N, limestone filler as additional fine material, rounded aggregates with a maximal size of 16 mm and polycarboxylates with a long workability retention as superplasticizer. As the amount of concrete needed was large (1.5 to 3.5m³), it has been produced in a concrete manufacturing plant and transported to the lab. 66

4 Figure 2: Pressure sensor and strain gauges 4. INFLUENCE OF RHEOLOGY ON PUMPING 4.1 Testing procedure After the insertion of the concrete in the short circuit, several pumping cycles have been executed with the concrete. A pumping cycle consists of decreasing the discharge (Q) of the pump stepwise, from 20 to 5 l/s, keeping each discharge for five full strokes (fig. 3). In this way, an accurate determination of pressure loss and discharge can be achieved. Before each cycle, a sample of the concrete is taken and subjected to a rheometer test, with the Tattersall Mk-II rheometer [2,5,9] and to the standard tests on fresh SCC [1]. Pressure (bar) Pressure (bar) High Q Time (s) Low Q Time (s) 67

5 Figure 3: Plot of pressure as a function of time, clearly showing the pressure drop and rise during the change of the valve of the pump. 4.2 Parameters influencing pumping pressures Figure 4 shows the results of a pumping cycle, where the pressure loss per unit of length has been expressed as a function of the discharge. The black full line represents TC, while the grey dashed curve is a typical result for SCC. As can be seen, the SCC-curve is higher than the line for traditional concrete, indicating higher pressure losses for this SCC. On the other hand, the SCC was more fluid than the TC, which has been confirmed by the rheometer results. Secondly, the TC-line is a straight line, while the SCC-curve is no longer linear. This can be due to shear thickening, which has been observed in case of SCC composed with Belgian materials [2][9-10]. Pressure loss (kpa/m) SCC 7 TC Discharge (l/s) Figure 4: Pressure loss as function of discharge for TC (full black) and SCC (dashed grey). After elimination of all possible sources of errors, due to sampling, rheometer inaccuracy and non-equilibrium of the concrete in the pipes (see next section), for each discharge, a linear relationship between the pressure loss and the viscosity, determined at a shear rate of 10/s, has been found. Especially at the higher discharges, this relationship showed a remarkably good correlation (R² = 0.93), indicating that the pressure losses are very well related to the viscosity [2]. As a result, the viscosity and possible shear thickening mainly influence the pressure losses when pumping SCC (or fluid concrete), and not the yield stress, especially at high discharges. In case a concrete with a rather high yield stress is pumped, or in case of very low discharges, the influence of the yield stress on the pressure losses will increase. 68

6 4.3 Velocity profile Theoretical predictions of the pressure loss discharge curves, based on the principle of the capillary rheometer [11], have revealed that the concrete cannot be regarded as a homogeneous material flowing through the pipes [2]. Before studying slippage or lubrication, the geometrical wall effect should be incorporated in the theoretical calculations, because this effect certainly occurs. On the other hand, this calculation requires the exact knowledge of the evolution of the rheological properties of the material in the vicinity of the wall. It also requires very accurate (and correct) rheometer data, which can be doubted in any case. Furthermore, there is a discrepancy between the shear rate ranges of the pumping tests and the rheometer. Cement paste, or mortar, should be extracted near the wall, and their rheological properties should be determined very accurately [6]. As a result, it is currently impossible to model the influence of the geometrical wall effect, and any theory dealing with slippage and other lubrication layers (the geometrical wall effects causes in fact a kind of lubrication layer) cannot be proven. As a result, the correct velocity profile is unknown, but it is assumed to contain a (small) plug near the centre, a large velocity gradient near the wall and a smaller velocity gradient in between. A direct measurement of the velocity profile during the pumping of concrete is currently impossible at a reasonable price, but this technique would reveal the real physics behind the process. 5. INFLUENCE OF PUMPING ON RHEOLOGY As stated in the previous section, the concrete in the pump and in the pipes is not always in an equilibrium condition. Theoretically, it will never be, due to the permanent action of thixotropy and loss of workability [5][12-14], but from a practical point of view, the concrete appears to be in equilibrium between the third and fifth cycle. The pressure losses at a certain discharge are higher during the first two cycles. In case the concrete has been kept a long time, pressure losses increased again, after the fifth cycle. As this decrease in pressure loss between cycle 1 and 3 is impossible to explain by the evolution of the concrete properties alone, pumping must influence the rheological properties. 5.1 Testing procedure As a first idea, it was suspected that thixotropy caused this decrease in pressure losses with increasing times of pumping. As a result, instead of the regular procedure, in which the discharge has been decreased, an opposite scheme has been worked out (fig. 5). After insertion of the concrete (at the lowest discharge), the concrete was pumped at discharge step 1 (the lowest possible) until equilibrium in the pressure loss has been achieved. After sampling and a discharge calibration, the discharge has been increased by one step, to discharge 2 (say in general: n), which has been maintained until equilibrium has been achieved. Again, a new sample of concrete has been taken and discharge has been calibrated, which has been followed by a rather quick decrease from discharge n to discharge 1, maintaining each discharge for 5 full strokes (as in the regular cycles). After this down-curve, discharge is increased to step n+1, and the same procedure has been repeated. During the tests in the lab, for safety reasons, the maximal discharge did not exceed step 4 (15-16 l/s) or step 5 (19-20 l/s), depending on the pressure losses. By means of this procedure, the concrete evolves each time to a new state. In this new state, a rheometer test has been performed on a part of the taken sample, together with a slump flow, 69

7 V-funnel, sieve stability, density and air content measurement. Even by means of the pressure losses, an evolution can be seen. 5.2 Results The result of the pressure losses as a function of discharge are shown in figure 6, in which it can be seen that the pressure loss at a certain discharge is lower when a higher discharge has been applied before. Maintain discharge for equilibrium pressure loss Discharge step Quick stepwise decrease in discharge 5 Sampling + discharge calibration Time ± 1.5 minute ± 1 hour ±4 minutes Figure 5: Testing procedure in order to determine the influence of pumping on the rheological properties. 70

8 Pressure loss (kpa/m) Equilibrium Down from discharge 2 Down from discharge 3 Down from discharge Discharge (l/s) Figure 6: Pressure loss as a function of discharge, indicating a decrease in pressure loss at a certain discharge if a higher discharge has been applied before. The rheometer test results and the tests on fresh concrete partly confirm this trend: the viscosity and the V-funnel flow time have been found to decrease with increasing discharge. On the other hand, slump flow and yield stress do not follow a fixed pattern. Depending on the SCC, the slump flow remains constant or it decreases. 5.3 Physical causes It is suspected that at least two different phenomena are influencing the rheological properties. The first phenomenon is structural breakdown [5], in its most broad sense. This structural breakdown combines the thixotropic breakdown [12-13] due to an increase in shear rate and the breakdown due to the disruption of the chemical connections between cement particles, formed by hydration [5][14]. Due to structural breakdown, the concrete becomes more fluid, which is translated in a decrease in yield stress and viscosity. The second phenomenon which has been observed is an increase in air content with increasing number of times the concrete has been pumped (and possibly also with increasing discharge). A more detailed analysis and explanation of the influence of air on the rheological properties of SCC can be found in [15]. As a conclusion of this paper, it is stated that increasing the number of (small) air bubbles increases the yield stress and decreases the viscosity. Both structural breakdown and the increase in air content cause a decrease in viscosity, but they act in opposite ways on the yield stress. In case the effect of the air bubbles dominates the structural breakdown, yield stress increases. In the other case, yield stress decreases, or at least, it remains almost constant. 5.4 Consequences 71

9 Based on a very restricted amount of results, the following trend has been observed. In case the delivered SCC was rather fluid (high slump flow), the yield stress did not increase significantly. On the other hand, as viscosity decreases, segregation can be provoked by high speed pumping. For the less fluid SCC, the effect of the increasing air content dominates, increasing the yield stress due to pumping. This can result in a too stiff SCC, which can lose its ability to fill the formwork completely. Further research is needed in order to confirm this theory. 6. RECOMMENDATIONS As we had the impression that the effects of structural breakdown and increase in air content are becoming more significant with increasing discharge (shear rate), it is advised to reduce pumping velocities, which can be achieved by decreasing discharge or increasing the pipe diameter. This will reduce the risks for segregation or improper filling of the formwork and it reduces the pressure losses. Furthermore, two tests are recommended to be carried out on site. A slump flow and a sieve stability test (or another test capable of measuring segregation resistance immediately and accurately) of the non-pumped and pumped concrete should indicate in which way the properties of the concrete are evolving, especially in situations with high velocities and/or long conveying lines. If the sieve-(un)stability value increases after pumping and the slump flow remains constant, a less fluid SCC must be delivered to the pump. In case the slump flow decreases, the SCC should be made (a little) more fluid. 7. CONCLUSIONS By means of full scale pumping tests, the interactions between the rheological properties and pumping of SCC have been investigated. It has been observed that mainly the viscosity and shear thickening, if present, influence the magnitude of the pressure losses and not the yield stress in case of SCC. As a result, as SCC has in general a higher viscosity than traditional concrete, higher pumping pressure are needed in order to pump SCC at the same discharge. In an attempt to determine the real velocity profile, it has been proven that the concrete in the pipes cannot be regarded as a homogeneous material. There should be a layer of more fluid material in the vicinity of the wall, but whether this effect is purely caused by the geometrical wall effect or whether other effects like slippage or an additional lubrication layer are present is currently unknown. Measuring the velocity-profile would reveal the hidden phenomena, but this is currently impossible at a reasonable price. Due to pumping, two effects influence the rheological behaviour of SCC: structural breakdown and an increase in air content. Both effects cause a decrease in viscosity, but act in an opposite way on the yield stress. If structural breakdown dominates the effects of the increase in air content, the yield stress will not increase and segregation can be provoked. In the other case, the yield stress does increase and improper filling of the formwork can occur. From the obtained results, it appears that more fluid SCC tends to segregate and less fluid SCC tends to stiffen more. From a practical point of view, it is advised to reduce pumping velocities in order to reduce the pumping pressures and to minimize the risks of segregation and improper filling of the 72

10 formwork. By performing a slump flow and a sieve stability tests before and after pumping, the SCC can be optimized. ACKNOWLEDGEMENTS The authors would like to acknowledge Research Foundation in Flanders (FWO) for the financial support of the project and the technical staff of both the Magnel and the Hydraulics laboratory for the preparation and execution of the full scale pumping tests. REFERENCES [1] De Schutter G., Bartos P., Domone P., Gibbs J., Self-Compacting Concrete, Whittles Publishing, Caithness, 296pp, [2] Feys D., Interactions between rheological properties and pumping of self-compacting concrete, Ph-D-thesis, Magnel Laboratory for Concrete Research, Hydraulics Laboratory, Ghent University, [3] Ede A.N., The resistance of concrete pumped through pipelines, Magazine of Concrete Research, Vol. 9, No. 27, pp , [4] Browne R.D. and Bamforth P.B., Tests to establish concrete pumpability, ACI Journal, Vol. 74, , [5] Tattersall G.H. and Banfill P.F.G., The rheology of fresh concrete, Pitman, London, 356pp, [6] Kaplan D., Pumping of concretes, Ph-D-thesis (in French), Paris: Laboratoire Central des Ponts et Chaussées, [7] Crepas R.A., Pumping concrete, techniques and applications, 3rd edition, Crepas and Associates, Inc., Elmhurst (Ill.), [8] Guptill N.R. (Editor), Placing concrete by pumping methods, ACI-304, American Concrete Institute, Farmington Hills, [9] Feys D., Verhoeven R., De Schutter G., Fresh self-compacting concrete: a shear thickening material, Cement and Concrete Research, Vol. 38, , [10] Heirman G., Vandewalle L., Van Gemert D., An analytical solution of the Couette inverse problem for shear thickening SCC in a wide-gap concentric cylinder rheometer, Journal of non-newtonian Fluid Mechanics, Vol. 150, , [11] Feys D., Verhoeven R., De Schutter G., Extension of the Poiseuille formula for shear-thickening materials and application to self-compacting concrete, Applied Rheology Vol. 18, 62705, [12] Wallevik J.E., Rheology of particle suspensions, Fresh concrete, mortar and cement paste with various types of lignosulphonates, Ph-D-thesis, Trondheim: The Norwegian University of Science and Technology, [13] Roussel N., A thixotropy model for fresh fluid concretes: Theory, validation and applications, Cement and Concrete Reseach, Vol. 36, , [14] Wallevik J.E., Rheological properties of cement paste: Thixotropic behavior and structural breakdown, Cement and Concrete Research, Vol. 39, 14-29, [15] Feys D., Roussel N., Verhoeven R., De Schutter G., Influence of air content on the steady state rheological properties of fresh self-compacting concrete, without air entraining agents, Proceedings of the 2 nd International Symposium on Design, Performance and Use of Self-Consolidating Concrete, Beijing, China, June