You are on page 1of 23

Materials and Structures (2013) 46:533555 DOI 10.

1617/s11527-012-9912-4

ORIGINAL ARTICLE

Parameters inuencing pressure during pumping of selfcompacting concrete


Dimitri Feys Geert De Schutter Ronny Verhoeven

Received: 7 November 2011 / Accepted: 12 July 2012 / Published online: 27 July 2012 RILEM 2012

Abstract The main difference between conventional vibrated concrete (CVC) and self-compacting concrete (SCC) is observed in the fresh state, as SCC has a signicantly lower yield stress. On the other hand, the placement of SCC by means of pumping is done with the same equipment and following the same practical guidelines developed for CVC. It can be questioned whether the ow behaviour in pipes of SCC is different and whether the developed practical guidelines can still be applied. This paper describes the results of full-scale pumping tests carried out on several SCC mixtures. It shows primarily that the slump or yield stress of the concrete is no longer a dominating factor for SCC, as it is for CVC. Instead, the pressure losses are well related to the viscosity and the V-funnel ow time of SCC. Secondly, bends cause

an additional pressure loss for SCC, which is in contrast to the observations of Kaplan and Chapdelaine and the estimation of the practical guidelines is not always on the safe side. Finally, due to the specic mix design of SCC, blocking is less likely to occur during pumping operations, but the same rules as for CVC must be applied during start-up. Keywords Self-compacting concrete Rheology Viscosity Pressure loss Pumping

1 Introduction 1.1 Research signicance Since the development of self-compacting concrete (SCC) in the late 1980s [1], the research on this concrete type has focused on several aspects: from raw materials, properties in fresh state up to mechanical and structural properties and durability. When focusing on the properties in fresh state, there is still a research gap between the characterisation of the fresh concrete properties when it leaves the mixer or the concrete truck and the ow of concrete in the formwork. In fact, SCC is mostly cast in the same way as conventional vibrated concrete (CVC): by means of large concrete buckets moved with a tower crane or a rolling bridge (in case of a precast plant), or by means of a concrete pump. As a result, although there has been very little research on the pumping of

D. Feys (&) de Concrete Division, Faculty of Engineering, Universite , Sherbrooke, Sherbrooke, 2500, Boulevard de lUniversite QC J1K 2R1, Canada e-mail: Dimitri.Feys@USherbrooke.ca G. De Schutter Magnel Laboratory for Concrete Research, Department of Structural Engineering, Faculty of Engineering, Ghent University, Technologiepark 904, 9052 Zwijnaarde, Belgium R. Verhoeven Hydraulics Laboratory, Department of Civil Engineering, Faculty of Engineering, Ghent University, SintPietersnieuwstraat 41, 9000 Ghent, Belgium

534

Materials and Structures (2013) 46:533555

SCC, it is done in practice, with the same equipment and following the same rules which have been developed for CVC. The main difference between CVC and SCC is observed in the fresh state, as SCC has a signicantly lower yield stress [2]. Therefore, it can be questioned how the ow of this concrete type in pipes is inuenced by its fresh properties and are the rules developed for CVC [3, 4, 5] still (partially) valid? This paper discusses full-scale experiments investigating the pumping of SCC. The main subject is the ow behaviour in straight pipes, while trying to give an estimation for the velocity prole. A comparison is made with the rules of thumb for CVC and the fullscale experiments on CVC, performed by Kaplan [6] and Chapdelaine [7]. In the nal stage, the behaviour in bends is briey discussed and compared to the existing literature. 1.2 Rheological properties of fresh concrete It is generally accepted that fresh concrete is a Bingham material, showing a yield stress and a plastic viscosity [8, 9]. The yield stress is the resistance to the initiation of ow, while the plastic viscosity is a measure for the resistance to a further increment in ow rate (Eq. 1). Due to thixotropy, structural breakdown and loss of workability caused by chemical reactions, the yield stress and plastic viscosity are not constant in time [2, 8, 10, 11]. Furthermore, the rheological properties depend on the shear-history the material has undergone. _ s s 0 lp c _n s s0 K c _ cc _ s s0 l c
2

but the modied Bingham model [14, 16] has also the capacity to describe shear-thickening behaviour and is preferred by the authors (Eq. 3). The modied Bingham model is applied on the results in this paper which show shear-thickening behaviour. Note that it is not in the scope of this paper to describe the physical causes of shear-thickening, as these are explained in [14, 17]. On the other hand, shear-thickening has a large consequence on the pressure during pumping, as shown in this paper.

2 Flow behaviour of conventional vibrated concrete in pipes In this section, an overview is given of the existing knowledge on the ow behaviour of CVC in pipes and it is indicated whether the described phenomena are applicable on SCC. 2.1 Flow or friction As concrete is a concentrated suspension of solid particles, the type of stress transfer during the movement of concrete in pipes can be different. If the stress transfer is dominated by direct contact between the (large) solid particles, namely the coarse aggregates, the friction between these aggregates will be the determining factor for the force necessary to move the concrete in a pipe. This can occur if insufcient cement paste or mortar is present to lubricate the coarse aggregates, as can be seen in Fig. 1 (right) [18, 19, 20]. This situation is dened by Browne and Bamforth [18] as unsaturated concrete, and the pressure during pumping of unsaturated concrete evolves exponentially with the length of the pipe, as shown in the right part of Fig. 2 [18]. In the opposite case, when sufcient cement paste or mortar is present (Fig. 1, left), direct contact between the coarse particles is avoided and the shearing takes place in the cement paste [18, 20]. As a result, the concrete can be regarded as a suspension and rheology can be applied to study the ow behaviour. The stress transfer is of the hydrodynamic type and Browne and Bamforth [18] dened such concrete as saturated. In this case, the pressure decreases (as it is the case for Newtonian liquids) linearly with the length of the pipe (Fig. 2, left) and the pressure loss is constant in a horizontal straight section

1 2 3

where s is the shear stress (Pa); s0 is yield stress (Pa); _ is shear rate lp is plastic viscosity (Bingham) (Pa s); c (s-1); K is consistency factor (H.-B.) (Pa sn); n is consistency index (H.-B.) (-); l is linear term (mod. Bingham) (Pa s); c is second order parameter (mod. Bingham) (Pa s2). In literature, it is stated that the rheological behaviour of fresh SCC can deviate from the linear, Bingham behaviour in some specic situations. In these cases, shear-thickening has mostly been observed, necessitating the application of a different rheological model [12, 13, 14, 15]. Most authors apply the Herschel-Bulkley equation on the results (Eq. 2),

Materials and Structures (2013) 46:533555

535

Fig. 1 Distinction between hydrodynamic interactions (paste suspends aggregates) (left) and friction (paste lls partly the voids between the aggregates) (right) during the ow of concrete through pipes

Pressure

Pressure

Length

Length

Fig. 2 The pressure decreases linearly over the length of a straight, horizontal pipe in case of hydrodynamic interactions (left), while friction causes an exponential decrease of the pressure with the length of the pipe (right). Figure after Browne and Bamforth [18]

with a constant diameter. Browne and Bamforth [18] have proven mathematically that the case of friction requires signicantly higher pressures to pump concrete. Friction needs thus to be avoided at all times. The recommendation of some practical guidelines to have a certain minimum paste or mortar content, or a certain minimum slump, is based on the avoidance of friction [35, 21, 22, 2325]. For SCC, friction is less likely to occur (in regular conditions), as by denition, contacts between coarse

particles must be avoided to full the lling and passing ability criteria [26]. It can also be observed that the amount of coarse aggregates is reduced in SCC mix design, compared to regular CVC [26]. Kaplan discussed in his thesis different causes of blocking [6, 27]. He states that blocking during startup is most common. It is caused by the loss of cement paste at the pipe walls and by inertia, relative to the viscous drag of the inserted cement paste, forcing the coarse aggregates to move ahead of the bulk concrete

536

Materials and Structures (2013) 46:533555

with each stroke of the pump [27]. As a result, the concentration of coarse aggregates increases and can transform a saturated concrete into an unsaturated concrete. The stress transfer switches from hydrodynamic to friction and if the pump cannot deliver the pressure needed to move the concrete any further, the pipe gets blocked. Blocking during start-up has been observed in this research project when inserting SCC into a 100 m long horizontal circuit. Figure 3 shows the evolution of the pressure in two measurement sections along the pipe. The sudden shocks in the pressure (down to zero and back up) are due to the working principle of the pump (which is explained further). With time, the pressure slightly increases as more concrete is inserted in the pipes. Suddenly, the pressure rises up to 55 bar, as the concrete blocks in the pipe. The pumping is consequently stopped and a front of aggregates, similar to the one depicted in Fig. 1 (right), must be removed before continuing. Although a mixture of water and cement was inserted in front of the concrete, blocking did occur in case of SCC. As a result, the same practical rules recommended for CVC must be applied during the insertion of SCC. 2.2 Lubrication layer For CVC, it is known that during pumping, the concrete moves as a large plug in the pipe, surrounded by a lubrication layer [6, 7]. This layer consists of
60 50
Section 1 Section 3

Pressure (bar)

40 30 20 10 0 1140 -10

1160

1180

1200

1220

1240

Time (s)

Fig. 3 The pressure at two different measurement sections in the long pumping circuit increases slightly as concrete is being inserted gradually in the pipes. Around 1,205 s, a transition takes place from hydrodynamic interactions to friction, as insufcient cement paste is available between the aggregates and the concrete blocks at 1,210 s, corresponding to the pressure peak

cement paste or mortar, as the coarse aggregates move away from the zone with the largest shear rate, which is at the wall. The entire velocity difference between pipe wall and concrete is concentrated in this layer. In literature, it is reported that the thickness of the lubrication layer varies between a few mm up to 1 cm [6]. As it is currently impossible to directly measure the thickness, although efforts have been made [28, 29], it is unsure what the exact thickness is. The lubrication layer facilitates the pumping of concrete through pipes. If no lubrication layer could be formed, the pumping pressure would be signicantly higher to pump the concrete at the same discharge rate. Some authors take the effect of the lubrication layer into account by introducing a slipping or sliding velocity [18, 19, 24, 30, 31]. The principle of slippage and lubrication layer is visualized in Fig. 4 [32]. The total velocity at a certain distance r from the centre of the pipe is then composed of the slipping velocity (which is constant) and a shearing velocity (which can vary). The shearing velocity depends on the applied shear stress (which is related to the pressure loss per unit of length and the pipe radius) and the yield stress of the concrete. For CVC, the shearing velocity can be zero across the pipe, reecting the plug ow. Kaplan [6] and Chapdelaine [7] related pumping pressures to the properties of the lubrication layer. To measure these properties, they both modied an existing concrete rheometer in a so-called tribometer. In concrete rheometry, the slip between the rheometer walls and the concrete is avoided by installing ribs. These ribs are removed in a tribometer and the ow properties of the concrete near a smooth surface are measured. Similar as in rheology, the yield stress (Pa) and the viscous constant (Pa s/m) of the lubrication layer are determined by changing the rotational velocity and measuring the resulting torque, which is transformed into a shear stress (Eq. 4) [6, 7, 33]. Note that the viscous constant has a different dimension than the plastic viscosity. Namely, the calculation of the shear rate in the lubrication layer is impossible, as its thickness is unknown. Therefore, the thickness of the lubrication layer is incorporated in the viscous constant, for an assumed linear velocity distribution in the lubrication layer. s s0;i gi v 4

where s is the shear stress (Pa); s0,i is yield stress of the lubrication layer (Pa); gi is viscous constant of the

Materials and Structures (2013) 46:533555

537

Fig. 4 Distinction between no-slip (left), slip (middle) and lubrication (right). In case of a lubrication layer, the velocity at the wall is zero (no slip), but the velocity gradient near the wall is larger, in this case over a distance h1 from the wall. This larger velocity gradient is caused by the lower viscosity of the lubrication layer (lm) compared to the viscosity of the concrete

(ls). When concrete is pumped, it could be that the yield stress of the concrete outside the lubrication layer is higher than the applied shear stress. In that case, the velocity in the concrete would be constant, while the velocity gradient is maintained in the lubrication layer. Figure from Thrane [32]

lubrication layer (Pa s/m); v is velocity difference over the lubrication layer (m/s).

3 Experimental work 3.1 Test-setup The experimental part of this research was carried out on full-scale pumping circuits. The total length of the rst circuit was 25 m, constructed with steel pipes with an inner diameter of 106 mm. After the exit of the pump, a 12 m straight, horizontal section was installed, followed by a 180 bend (composed of two 90 bends with a 1 m pipe in between). The second part of the circuit was inclined, in order to feed the concrete back to the pump (Fig. 5). In this way, the circuit is a loop circuit as the pumped concrete was reutilised several times. This circuit was used to determine the relationship between the rheological properties and the pumping pressure of the concrete. The main results discussed in this paper were obtained in this small circuit. A second and third circuit, with lengths of 100 m (Fig. 6) and 80 m respectively, consisted of extending the small 25 m circuit with four straight sections, connected with 180 bends in between. The last bend, before starting the inclined part, was composed of two 90 bends with a 1 m pipe in between. Several tests have been performed on these longer circuits, but only the results on blocking, discussed previously and the

results for the pressure losses in bends are included in this paper. The pump was a truck-mounted piston pump, capable of delivering a pressure of 95 bar or a discharge rate of 150 m3/h (or 40 l/s). The discharge rate can be controlled in 10 steps, varying between 4 to 5 l/s up to approximately 40 l/s. For safety reasons, only the ve lowest discharge rates were applied, with a maximum of 20 l/s. The pump itself has two pistons, which alternately push the concrete inside the pipes and pull concrete from the pumping reservoir. When the pushing piston is empty (and consequently the pulling piston is full), a powerful valve inside the reservoir changes the connection between the pistons and the circuit. This provokes a sudden decrease and increase in pressure during approximately one second. As shown in Fig. 7, it can be clearly seen in the measured pressure evolution. It can also be clearly heard on site. 3.2 Measurement systems 3.2.1 Pressure loss In the horizontal straight section of the 25 m circuit, two pressure sensors were installed at a distance of 10 m from each other. The pressure sensors are equipped with a metallic seal, resistant to abrasion, to avoid the insertion of cement or concrete in the pressure chamber. The pressure chamber is lled with oil and transfers the pressure applied on the seal to the sensor. The sensors have a maximum capacity of

538

Materials and Structures (2013) 46:533555

Fig. 5 25 m pumping circuit. The straight, horizontal section on the left contains the pressure sensors

Fig. 6 Extension of the pumping circuit to 100 m

35 bar, with a safety factor of 2 for accidental overload. With these sensors, the pressure difference between two points can be measured and the pressure

loss per unit of length can be calculated. The pressure sensors were connected to a data-acquisition system, registering the local pressures at a rate of 10 Hz.

Materials and Structures (2013) 46:533555 Fig. 7 Evolution of the upstream pressure (closest to the pump) with time, clearly showing the change of the valve of the pump (see inset). The discharge rate (Q) is decreased stepwise, maintaining each step for ve full strokes of the pump

539

Upstream pressure (bar)


8 7

Upstream pressure (bar)


20 18 16 14 12 10 8 6 4 2 0 0 50

6 5 4

strain gauges pressure sensor

High Q

3 2 1 0 72 72.5 73 73.5 74

74.5

75

Time (s)

Low Q

100

150

200

250

Time (s)

In the vicinity of each pressure sensor, a set of three strain gauges was attached to the outer wall of the pipe. The strain gauges allowed to measure the expansion and contraction of the pipe (which had a thickness of 3 mm), which is related to the locally applied pressure [6]. The strain gauges served as a back-up system in case something happened with one of the pressure sensors. The output of the strain gauges was calibrated with the measured pressures when the pressure sensor was working correctly, while during failure of the pressure sensor, the strain gauges served as full measuring units. For the longer circuits, the pressure sensors were installed in the last horizontal straight section. Several other pipes were also equipped with strain gauges in order to follow the pressure evolution through the circuit. In the 100 m circuit, such an equipped section was installed 0.5 m before the last 180 bend, while in the 80 m circuit, an equipped section was installed in between the two 90 bends before the inclined part. In this way, the pressure loss over a bend can be accounted for. The location of the measurement sections for each circuit is shown in Fig. 8. As can be seen in Fig. 7, the pressure evolution with time shows large peaks due to the change of the valve of the pump. Only the values of the pressure in equilibrium (during pushing of a piston) were taken into account for the analysis.

3.2.2 Discharge rate The determination of the discharge rate was not straightforward as no electromagnetic discharge meter was at disposal. Instead, the time needed for a certain number of strokes of the pump (the emptying of one piston) was recorded, both by hand with a stopwatch, and with the les delivering the pressure evolution with time (similar as Fig. 7). As the volume of one pumping piston is known (83.1 l), the discharge rate can be calculated. But the pistons are normally not completely full, inducing an error (over-estimation of volume) in this procedure. The lling coefcient of the pistons must be known to properly determine the discharge rate [6, 7]. Therefore, a special calibration procedure has been employed. It consisted of pumping concrete in a closed reservoir, which was connected to a load cell. Knowing the density of the concrete, the discharge rate can be calculated based on the load variation with time. As the load cell was connected to the same data acquisition system, measuring at 10 Hz, the discharge rate could be determined, for one stroke, during the period that the pressure was in equilibrium. As the total time measured by the stopwatch also includes the time of the change of the valve (the socalled dead time), a second error is induced in the manual measurements (over-estimation of time). By coincidence, both errors compensate each other and

540

Materials and Structures (2013) 46:533555

Fig. 8 Schematic presentation of the pumping circuits, showing the locations of the pressure sensors and strain gauges

the discharge rate measured by the stopwatch method is the same discharge rate applied when the pressure is in equilibrium, as demonstrated in Fig. 9. Note that in Fig. 9, the maximum discharge rate applied was 25 l/s. 3.3 Concrete In total, 19 concretes were tested in the described pumping circuits, of which 18 were SCC, and one was a pumpable CVC mixture. All concretes were prepared in a ready-mix concrete plant and transported to the laboratory. Usually, the production and transport of the concrete took 45 min1 h. Except for the mixtures developed by the concrete plant, all mixtures contained ordinary portland cement (CEM I) and limestone ller as powder materials. The maximum aggregate size of the SCC was 14 mm. The

superplasticizer employed was a poly-carboxylate with guaranteed workability retention of 100 min. In Table 1, the mix designs for all concretes for which results are used in this paper are shown. 3.4 Testing procedure Shortly after the delivery of the concrete, a sample was taken and the fresh properties were tested by means of the standard tests on SCC (slump ow, V-funnel, etc.) and by means of the Tattersall Mk-II rheometer (Fig. 10) [8, 14]. During the initial characterization of the SCC, the concrete was inserted in the pipes. The rst 250 l of pumped material, which was a mixture of the preparatory cement paste, aggregates and concrete, was removed from the site. In contrast to the 100 m circuit, no blocking was observed during the insertion of any of the concretes in the 25 m circuit. The

Materials and Structures (2013) 46:533555 Fig. 9 Calibration of the discharge rate shows that both methods with the stopwatch, as with the output le (similar as Fig. 7) represent the real discharge rate (measured with the load cell)
30

541

Stopwatch 25

Discharge rate from load cell (l/s)

Output File 20

15

10

0 0 5 10 15 20 25 30

Discharge rate from stopwatch or output file (l/s)

reservoir of the pump was lled and 250 l of concrete was left aside before the concrete truck left the lab. The total amount of concrete inside the pipes and the reservoir of the pump was approximately 1 m3. At a concrete age of 60 min (if possible), a rst pumping test was executed. This test consisted of pumping the concrete through the pipes at the ve lowest discharge rates available on the pump. These discharge rates corresponded to the steps of the pump, step 1 being around 5 l/s, step 2: approx. 8 l/s, step 3: 1213 l/s, step 4: 1516 l/s and for step 5, a discharge rate of 1820 l/s was obtained. For security reasons, the discharge rate was not increased above 20 l/s, except for some special cases. During the test, all steps were maintained for ve full strokes each, which means that for each discharge rate, the contents of ve pistons was pushed through the pipes. The discharge rate was decreased stepwise and the test had a total duration of around 4 min (Fig. 7). This testing procedure was repeated each 30 min, until it was decided to discard the concrete and clean the circuit. In most cases, three to four tests were executed for each concrete. In between these tests, the concrete was at rest or other types of tests were executed, such as the discharge calibration tests. Even after a rest period of approximately 25 min, the re-start, which could be compromised by thixotropic build-up, did not deliver any problems.

Before the start of a selected number of tests, concrete was pumped in the closed reservoir to take samples for the tests on fresh concrete. Initially, when taking the 12 l sample in the rheometer bucket directly at the outlet of the circuit, it was sometimes not fully representative for the concrete inside the pipes. Taking a sample when the valve of the pump changes delivers more aggregates, as they move forward due to inertia, while the paste in the concrete has stopped. Taking a sample as the pumping cylinder just starts pushing delivers more paste, as the paste moves almost instantly, while the aggregates, which have slowed down, need to be accelerated. As a result, the sample for the rheometer sometimes contained very few aggregates (as it was almost mortar), or too many aggregates to be considered as SCC. As a result, some rheological tests were conducted on a sample that was not representative for the concrete inside the pipes, and these results were not used in the analysis. The decision was made based on visual observations. In a later stage during the research, this problem was omitted by taking the sample for the rheometer from the 100 l sample taken from the pump for the tests on fresh SCC. The latter concrete was not visibly affected by the changes of the valve, as the concrete sample was sufciently large to be considered as homogeneous. As a result, it is advised to take a sufciently large volume of concrete on the jobsite (wheelbarrow instead of a bucket) when analysing or sampling concrete after pumping.

542 Table 1 Mix designs for pumped concretes (units in kg/m3) SCC 1 Gravel 8/16 Gravel 3/8 Sand 0/5 CEM I 52.5 N Limestone ller Water SP (l/m3) Powder content (kg/m3) W/C-ratio () W/P-ratio () Slump ow at plant (mm) Remarks SCC 7 Gravel 8/16 Gravel 3/8 Sand 0/5 CEM I 52.5 N Limestone ller Water SP (l/m3) Powder content (kg/m3) W/C-ratio () W/P-ratio () Slump ow at plant (mm) Remarks 434 263 853 360 239 165 12.69 599 0.458 0.275 650 SCC 8 434 263 853 360 239 165 14.44 599 0.458 0.275 680 328 0.538 0.521 Plant-Mix Contains FA SCC 13 Gravel 8/16 Gravel 3/8 Sand 0/5 CEM I 52.5 N Limestone ller Water SP (l/m3) Powder content (kg/m3) W/C-ratio () W/P-ratio () Slump ow at plant (mm) Remarks 434 263 853 360 239 165 ? 599 0.458 0.275 700 Target SF SCC 14 434 263 853 360 239 160 21.9 599 0.444 0.267 640 581 0.452 0.324 650 SCC 15 CVC 1 434 263 853 360 239 165 11 599 0.458 0.275 SCC 2 434 263 853 360 239 165 11 599 0.458 0.275 SCC 3 434 263 853 360 239 165 15.22 599 0.458 0.275 690

Materials and Structures (2013) 46:533555

SCC 4 459 278 901 300 200 165 12.16 500 0.550 0.330 710

SCC 5 434 263 853 360 239 165 20.95 599 0.458 0.275 710

SCC 6 434 263 853 360 239 165 13.33 599 0.458 0.275 720

SCC 9 410 248 805 400 300 165 18.15 700 0.413 0.236 700

SCC 10 434 263 853 360 239 165 11 599 0.458 0.275 650

SCC 12 434 263 853 360 239 165 ? 599 0.458 0.275 675 Target SF

SCC 16 434 263 853 360 239 165 ? 599 0.458 0.275 700 Target SF

SCC 17

581 0.452 0.324 700 Plant-Mix Target SF

Plant-Mix

Materials and Structures (2013) 46:533555

543

Fig. 10 Tattersall Mk-II rheometer used to measure the rheological properties of the pumped concretes

The rheological properties were measured with the Tattersall Mk-II rheometer. As the rheological properties need to be expressed in fundamental units when applying them on the pumping data, the torquerotational velocity data were transformed into shear stress and shear rate according to the procedure described in the PhD work of Feys [34]. Although the transformation procedure is not perfect, it is shown in literature that the Tattersall Mk-II rheometer delivers similar results as the ConTec rheometer [35]. Note furthermore that the same study concluded that the Tattersall Mk-II rheometer is not capable of correctly measuring the rheological properties of very uid concretes. As the concrete is pumped in a loop circuit, it is reused several times and it underwent changes in the rheological properties. As the rheological properties of the concrete are measured each time when executing a pumping test, the further derived relationship between the rheological parameters and the pumping pressure is independent of the changes occurring in the concrete. Each combination rheologypumping pressure is used as an independent data point.

4 Results for straight pipes The main results reported in this paper are valid for straight, horizontal sections and are based on the results obtained with the 25 m circuit. Due to a change in rheological properties of the concrete during pumping, the pressure loss for the test at 120 and 150 min of age (tests 3 and 4) was lower than the pressure loss for test 1 at 60 min and even test 2 at 90 min of age (Fig. 11) [34]. As a result, there is a discrepancy between the test results of the rst pumping test (at 60 min of age) and the measured rheological properties of the corresponding concrete. As the concrete sample was taken before the test, it did not undergo the same shear history as the concrete in the pipes. Furthermore, during the rst test (60 min) and potentially during the second test (90 min), the concrete was not in equilibrium conditions. This implies that these results cannot be employed in the analysis of a potential rheologypumping relation. The discussion on the changes in properties due to pumping is beyond the scope of this paper.

544

Materials and Structures (2013) 46:533555


19-20 l/s 15-16 l/s 12-13 l/s 8-9 l/s 5-6 l/s

50 45 40

Pressure loss (kP a/m)

35 30 25 20 15 10 5 0 60 80 100 120 140 160 180 200

Concrete age (min)

Fig. 11 The evolution of the pressure loss at each discharge rate decreases during the rst three tests, remains constant and increases afterwards. Test results from SCC 8

4.1 Velocity prole and existence of a lubrication layer for SCC 4.1.1 Existence of lubrication layer The existence of a lubrication layer during pumping of SCC has not been measured directly, but can be indirectly proven with the following mathematical procedure: The equilibrium of forces in a straight, horizontal pipe expresses the relationship between the pressure loss per unit of length (in Pa/m) and the shear stress at the inner wall of the pipe (in Pa), by means of Eq. 5. sw Dptot R L 2 5

where sw is the shear stress at the pipe wall (Pa); Dptot is total pressure loss over the length L (Pa), L is length of the considered section (m); R is radius of the pipe (m). It is further known from rheology that in a circular pipe, the shear stress decreases linearly from its

maximum value at the wall to zero in the centre, regardless of the rheological properties of the material [36] (Fig. 12). The shear stress distribution is thus only inuenced by the pressure loss per unit of length and the pipe radius. Knowing the rheological properties of the concrete (which are measured with the Tattersall Mk-II rheometer), the shear rate distribution across the pipe can be calculated. Integrating the shear rate over the pipe radius delivers the velocity prole. In this case, it is assumed that the velocity at the wall is zero. By integration of the velocity prole over the cross section of the pipe, the discharge rate corresponding to the pressure loss and rheological properties is obtained. In this procedure, two assumptions have been made: the velocity at the wall is zero (there is no slippage) and the material is homogeneous (the rheological properties are constant in the entire cross section of the pipethere is no lubrication layer). Following this procedure, the Poiseuille equation is obtained for Newtonian liquids [37, 38] and the BuckinghamReiner equation (Eq. 6) is concluded for Bingham liquids [8, 36, 39]:

  3 4 3 p 3 R4 Dptot 4 16 s4 0 L 8 s0 L R Dptot 24 Dptot 3 L lp

Materials and Structures (2013) 46:533555

545

CVC: Shear stress < Yield stress dp/dx fixed shear stress CC shear stress fixed SCC Yield stress

Plug flow + lubrication layer lubrication layer

CVC

velocity

SCC

velocity

shear rate Yield stress SCC: Shear stress > Yield stress
Fig. 12 Theoretical velocity proles for CVC and SCC. As a main difference, the plug in SCC is much smaller and a part of the concrete itself is sheared in the pipes. This is caused by the yield stress of the concrete. For CVC, the shear stress is in most

lubrication layer Plug flow + lubrication layer + shearing flow


cases lower than the yield stress. The ow is only made possible by the lubrication layer. For SCC, the yield stress is sufciently low to cause shearing in the concrete, but a lubrication layer is proven to be present

For modied Bingham materials, a more extended equation has been derived in [40] (Eq. 7): h p D3 3 Q l7 W l6 140 l c3 s3 0 sw : 4 3 6720 c sw
2 3 2 W l4 c sw 6 s0 14 l5 c s0 70 s2 0c l

8 W c3 sw s0 3 sw 4 s0 2 2 W l2 c2 3 s2 w 24 s0 8 sw s0 i 3 3 120 W c3 s3 w 64 W c s0 7 p with W l2 4 c sw 4 c s0 : If no lubrication layer were formed, these theoretical equations should match the experimentally obtained data. In the last three columns of Table 2, the discharge rate, experimental and predicted pressure loss respectively, are shown for different tests. The predicted pressure loss is based on the BuckinghamReiner equation (Eq. 6) if the concrete shows Bingham behaviour (c/l = 0). In case of c/l [ 0, Eq. 7 is used. From Table 2, it can be seen that the theoretical equations provide a signicantly higher estimation of the pressure loss at a certain discharge rate compared to the experiments [34], which is similar as

concluded for CVC by Kaplan [6]. This difference can sometimes attain one order of magnitude. Furthermore, the predictions based on Eq. 7 provide a higher overestimation than the predictions based on Eq. 6. This can be attributed to the extrapolation of the rheological data outside the shear rate range used in the rheometer, which is explained in Sect. 4.2.1. As a result, the most probable explanation is that a lubrication layer must be formed, also in case of SCC, to facilitate the pumping. These results are in line with the observations of Jacobsen et al. [41]. 4.1.2 Velocity prole for SCC On the one hand, the lubrication layer is formed in the vicinity of the pipe wall, but in the centre of the pipe, the concrete (especially SCC) is assumed to have the same rheological properties as measured in the rheometer. As the shear stress distribution in the pipe is known, the plug radius, which denes the boundary between sheared and unsheared concrete, can be calculated as rplug R s0 =sw (if s0 B sw). For CVC, the plug radius is in most cases almost equal to the radius of the pipe, as the concrete yield stress is higher

546 Table 2 Fresh properties and rheological properties of the concrete during different pumping tests, reported in Figs. 13, 14 and 15. The last three columns indicate the discharge rate, experimental and theoretical pressure loss. The theoretical

Materials and Structures (2013) 46:533555 pressure loss is based on the BuckinghamReiner equation if c/ l = 0, or it is based on the extended version for the modied Bingham model if shear-thickening is observed Q (l/s) Dp (kPa/m) experimental 53.9 41.0 28.5 15.7 9.2 63.5 50.2 35.9 21.7 11.8 62.4 48.1 33.9 22.7 13.3 86.6 65.7 48.4 31.5 18.9 91.5 73.2 55.3 36.2 22.4 53.9 41.7 27.7 18.3 10.1 53.7 39.5 29.9 19.0 10.6 52.3 38.3 26.3 17.2 10.1 Dp (kPa/m) theoretical 373 270 185 97 53 251 210 161 108 63 216 178 137 95 58 287 231 182 126 77 312 262 210 141 88 221 180 142 99 59

SCC 1 Age (min) 130

Slump ow (mm) V-Funnel (s)

710 5.3

Yield stress (Pa) l (Pa s) c/l (s) App visc at 10 s-1 (Pa s) Yield stress (Pa) l (Pa s) c/l (s) App visc at 10 s-1 (Pa s) Yield stress (Pa) l (Pa s) c/l (s) App visc at 10 s-1 (Pa s) Yield stress (Pa) l (Pa s) c/l (s) App visc at 10 s-1 (Pa s) Yield stress (Pa) l (Pa s) c/l (s) App visc at 10 s-1 (Pa s) Yield stress (Pa) l (Pa s) c/l (s) App visc at 10 s-1 (Pa s)

49.0 24.7 0.012 35.5 114.4 39.5 0 50.9 162.5 34.1 0 50.3 270.1 45.8 0 72.8 410.1 48.6 0 89.6 83.0 35.2 0 43.5

SCC 2 Age (min) 125

Slump ow (mm) V-Funnel (s)

660 4.5

SCC 2 Age (min) 180

Slump ow (mm) V-Funnel (s)

523 4.9

SCC 3 Age (min) 170

Slump ow (mm) V-Funnel (s)

470 5.3

SCC 3 Age (min) 195

SCC 4 Age (min) 90

SCC 4 Age (min) 115

Slump ow (mm) V-Funnel (s)

700 3.2

SCC 4 Age (min) 145

Slump ow (mm) V-Funnel (s)

645 4.1

Yield stress (Pa) l (Pa s) c/l (s) App visc at 10 s-1 (Pa s)

140.6 31.6 0 45.7

18.7 15.2 11.8 7.4 4.6 19.2 16.0 12.2 8.0 4.5 18.9 15.4 11.7 7.9 4.5 18.5 14.7 11.4 7.6 4.3 18.6 15.4 12.1 7.7 4.3 19.1 15.5 12.1 8.3 4.8 18.5 15.0 12.4 8.4 4.7 19.8 16.0 11.9 8.1 4.8

209 170 128 90 56

Materials and Structures (2013) 46:533555 Table 2 continued Q (l/s) SCC 5 Age (min) 105 Slump ow (mm) V-Funnel (s) 660 4.3 Yield stress (Pa) l (Pa s) c/l (s) App visc at 10 s-1 (Pa s) 73.4 28.2 0 35.5 18.7 14.5 12.2 8.3 4.7 18.9 15.4 12.3 7.8 5.4 20.2 16.9 12.4 8.0 5.5 20.2 15.8 12.3 8.5 5.4 19.4 16.5 12.8 8.3 5.9 20.1 16.3 12.5 9.2 5.5 20.6 16.5 12.6 8.9 5.2 19.3 15.6 12.0 8.8 8.6 19.6 15.8 12.6 7.9 6.1 Dp (kPa/m) experimental 53.9 35.5 26.5 16.1 9.6 49.7 38.9 29.2 18.6 12.2 40.8 30.5 20.6 11.7 7.5 40.0 29.7 21.1 13.8 8.7 31.8 24.8 17.1 10.6 7.6 33.9 24.4 16.5 11.6 7.6 37.1 27.4 18.3 11.1 6.7 25.6 18.7 14.6 9.5 7.7 20.9 14.6 7.7 2.3 2.7

547

Dp (kPa/m) theoretical 174 136 115 79 47

SCC 6 Age (min) 135

Slump ow (mm) V-Funnel (s)

688 3.4

SCC 7 Age (min) 120

Slump ow (mm) V-Funnel (s)

695 3.0

Yield stress (Pa) l (Pa s) c/l (s) App visc at 10 s-1 (Pa s) Yield stress (Pa) l (Pa s) c/l (s) App visc at 10 s-1 (Pa s) Yield stress (Pa) l (Pa s) c/l (s) App visc at 10 s-1 (Pa s)

31.2 16.4 0.023 27.1 29.4 14.7 0.021 23.8 21.8 3.5 0.134 15.1 21.4 9.4 0.033 17.7 50.0 16.2 0 21.2 11.9 0.6 0.484 7.6

SCC 7 Age (min) 150

Slump ow (mm) V-Funnel (s)

710 3.5

SCC 8 Age (min) 150

SCC 8 Age (min) 180

Slump ow (mm) V-Funnel (s)

693 3.1

Yield stress (Pa) l (Pa s) c/l (s) App visc at 10 s-1 (Pa s) Yield stress (Pa) l (Pa s) c/l (s) App visc at 10 s-1 (Pa s) Yield stress (Pa) l (Pa s) c/l (s) App visc at 10 s-1 (Pa s)

SCC 8 Age (min) 210

Slump ow (mm) V-Funnel (s)

570 3.5

SCC 9 Age (min) 105

388 320 191 96 55 365 241 160 90 47 397 290 178 79 42 328 226 142 85 33 110 89 68 49 30 233 153 91 50 48

SCC 10 Age (min) 240

Slump ow (mm) V-Funnel (s)

685 2.5

548 Table 2 continued

Materials and Structures (2013) 46:533555

Q (l/s) SCC 13 Age (min) 120 SCC 12 Age (min) 150 Slump ow (mm) V-Funnel (s) 750 2.0 Yield stress (Pa) l (Pa s) c/l (s) App visc at 10 s-1 (Pa s) Yield stress (Pa) l (Pa s) c/l (s) App visc at 10 s-1 (Pa s) Yield stress (Pa) l (Pa s) c/l (s) App visc at 10 s-1 (Pa s) Yield stress (Pa) l (Pa s) c/l (s) App visc at 10 s-1 (Pa s) Yield stress (Pa) l (Pa s) c/l (s) App visc at 10 s-1 (Pa s) 6.7 2.3 0.091 7.2 11.6 6.4 0 7.6 40.6 9.8 0 13.9 88.8 7.0 0 15.9 34.4 9.8 0 13.2 19.5 15.7 12.9 10.1 19.1 15.3 11.6 8.0 6.9 13.2 10.0 6.2 4.0 17.0 12.8 8.9 5.1 3.6 13.9 10.4 6.6 3.9 18.6 14.7 11.0 7.7 4.2 20.2 15.6 11.8 7.6 4.7

Dp (kPa/m) experimental 13.3 9.3 6.5 3.9 25.7 18.8 13.3 8.0 6.8 22.9 16.4 9.3 6.3 30.7 23.4 16.6 10.7 8.6 27.1 19.2 12.4 8.4 23.5 16.8 11.7 7.9 4.1 28.4 21.8 15.6 9.2 4.7

Dp (kPa/m) theoretical 183 121 84 53 40 32 25 17 15 44 34 22 15 43 33 25 16 13 46 35 23 14

Slump ow (mm) V-Funnel (s)

645 2.1

SCC 15 Age (min) 120

Slump ow (mm) V-Funnel (s)

570 3.4

SCC 15 Age (min) 210

Slump ow (mm) V-Funnel (s)

445 3.7

SCC 16 Age (min) 210

Slump ow (mm) V-Funnel (s)

535 3.9

SCC 17 Age (min) 120

Slump ow (mm) V-Funnel (s)

750 2.2

CVC 1 Age (min) 210

Slump (mm)

240

Yield stress (Pa) l (Pa s) c/l (s) App visc at 10 s-1 (Pa s)

122.7 15.2 0 27.5

105 83 64 44 29

than the shear stress at the pipe wall. As a consequence, CVC ows at uniform velocity, surrounded by the lubrication layer [] (Fig. 12). The calculations for SCC have shown that in all cases, even at the lowest discharge rates, a part of the SCC is sheared (Fig. 12), as the largest plug radius calculated for the entire set of experiments is 3.7 cm (SCC 3195 min). This is attributed to the low yield stress of this type of concrete, compared to the shear stress at the pipe wall.

As a result, the velocity prole of SCC is assumed to be composed of a small plug in the centre of the pipe, a lubrication layer near the wall and sheared concrete in between (Fig. 12). Note that this type of behaviour was predicted by Kaplan in [6]. When considering the concrete as a homogeneous material in the pipe, the shear rate can be calculated based on the pressure loss per unit of length (experimentally obtained) and the rheological properties of

Materials and Structures (2013) 46:533555

549

the concrete, according to the procedure explained in Sect. 4.1.1. The test results at the highest discharge rate indicate a maximum shear rate between 30 and 60 s-1 for homogeneous concrete. Increasing pipe diameter would lower these values, but increasing discharge rate increases these shear rates. In case a lubrication layer is considered, these shear rates would be signicantly higher. Assume that the lubrication layer has rheological properties that are 10 times lower than these of concrete, the shear rates would be approximately 10 times higher. (This is just to give an example as there is no proof for this statement.) No tribological measurements to characterize the lubrication layer properties were performed in this research project. In any case, performing tribological measurements on SCC would not be straightforward, as the basic assumption for concrete tribology is that the concrete itself is not allowed to be sheared [6, 7, 33]. Due to the low yield stress of SCC, this assumption is unlikely to be fullled, complicating signicantly the testing and data transformation procedure. 4.2 Inuence of rheological behaviour 4.2.1 Inuence of viscosity In Fig. 13, based on the results in Table 2 (except the CVC), the pressure loss per unit of length is plotted as
Fig. 13 For each discharge rate, the pressure loss per unit of length is correlated to the apparent viscosity of SCC, taken at a shear rate of 10 s-1
Pressure loss (kPa/m)
120

a function of the apparent viscosity at a shear rate of 10 s-1. This apparent viscosity represents the inclination of a straight line connecting the origin and the rheological curve at a shear rate of 10 s-1 [36]. A shear rate of 10 s-1 to calculate the apparent viscosity is chosen, as it represents approximately 2/33/4 of the maximum shear rate applied in the Tattersall Mk-II rheometer (which is between 12 and 14 s-1). Calculating the apparent viscosity at or beyond the maximum shear rate in the rheometer would make the results very sensitive to small errors due to the uctuations of the torque during the measurement. Therefore, it appeared more appropriate to calculate the apparent viscosity at 10 s-1. As stated above, the shear rate in the sheared part of the concrete during pumping can reach up to 60 s-1 (or even higher if a higher discharge rate is applied), resulting in a discrepancy in the range of shear rate between the rheometer and the ow in the pipes. On the other hand, the maximum shear rate obtained in the Tattersall MkII rheometer is already a very high value for a concrete rheometer. Increasing it further would signicantly increase the risk of (dynamic) segregation during testing. As a result, it was decided not to increase the maximum shear rate in the rheometer to maintain sufcient quality of the rheometer results and to keep the discrepancy between the rheometer and the pipe ow.

100

Q = 18 - 20 l/s Q = 15 - 16 l/s Q = 12 - 13 l/s Q = 8 l/s y = 0.85x + 19.60 R = 0.96

80

Q = 5 l/s

60

y = 0.67x + 14.01 R = 0.96

40

y = 0.50x + 9.24 R = 0.95 y = 0.33x + 5.40 R = 0.95

20 y = 0.18x + 4.48 R = 0.88 0 0 10 20 30 40 50 60 70 80 90 100

Apparent Viscosity at 10 s-1 (Pa s)

550

Materials and Structures (2013) 46:533555

Figure 13 shows that for each range of discharge rates, a good correlation can be found between the pressure loss per unit of length and the concrete apparent viscosity. This good agreement is the consequence of the shearing of the concrete. On the other hand, the relationships are empirical, as they are only valid for the range of discharge rates and for the pipe diameter used. It is not in the authors intention to provide prediction tools for the pressure, but only to show that in case of SCC, the concrete viscosity is a dominating factor. The practical guidelines for pumping CVC predict the total pressure based on the discharge rate, diameter of the pipe, the equivalent length of the pipeline and the spread value of the concrete [4, 25]. The latter value is a kind of measure, similar to the slump test, related to the capability of the concrete to form and maintain the lubrication layer. For SCC however, the spread value would be very high and if the guidelines are followed, very low pressures would be needed. The practical experience however indicates that in many cases, the pump has to work more in case of SCC. This means that the operators observe in general larger pressures needed to pump SCC. As a result, the practical guidelines to predict the pumping pressure cannot

be applied on SCC. Therefore, it would be better to modify the practical guidelines for pumping of SCC, in which the pressure loss is related to the viscosity of SCC. In the experimental work of Jodeh and Nassar [42], two different SCC were pumped in a 250 m circuit. Although both SCC had the same initial slump ow of 750 mm, a signicant difference in total pressure was monitored: 250 bar of SCC 1 compared to 92 bar for SCC 2. The V-funnel ow times for SCC 1 and SCC 2 were 20 and 10 s respectively, showing the importance of viscosity on the total pressure. Figure 14 shows a good agreement between the pressure loss and the V-funnel ow time of the tested concretes in this experimental work. In the work of Kaplan [6] and Chapdelaine [7] on CVC, the pumping pressure is well related to the viscous constant of the lubrication layer. They already showed the importance of a viscosity term in this casting process. The authors are convinced that the characteristics of the lubrication layer (mainly the viscous constant), together with the concrete viscosity, should be able to give a good prediction of the pressure needed to pump SCC. Only the difculties in performing tribological measurements, as stated in the previous section, must be solved.

100 90 Q = 18 - 20 l/s 80 70 Q = 15 - 16 l/s Q = 12 - 13 l/s Q = 8 l/s Q = 5 l/s 60 50 40 30 20 10 0 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 y = 11.82x - 10.55 R = 0.73 y = 8.71x - 8.70 R = 0.69 y = 5.60x - 6.04 R = 0.65 y = 2.92x - 1.65 R = 0.61 y = 15.38x - 11.21 R = 0.77

Pressure loss (kPa/m)

V-Funnel flow time (s)

Fig. 14 The pressure loss per unit of length can be related to the V-Funnel ow time of SCC

Materials and Structures (2013) 46:533555

551

4.2.2 Shear-thickening The SCC tested in this research project showed in many cases shear-thickening behaviour. This can be seen in Table 2 where the rheological properties of the concrete, measured at each test used in the analysis, are shown. The parameter c/l expresses the intensity of shear-thickening when applying the modied Bingham model. The larger c/l, the more severe the shear-thickening, while c/l equal to zero reects Bingham behaviour. The shear-thickening behaviour of the concrete is reected in the pressure lossdischarge rate curve. For each pumping test, the pressure loss at each discharge rate is shown in Table 2 and an example is shown in Fig. 15. This gure compares the pressure lossdischarge rate curves of SCC 7 and the only CVC which has been pumped. The SCC showed shear-thickening, while the CVC was a Bingham material, which is also observed in the pressure loss discharge rate curve. As a result, shear-thickening is a disadvantageous phenomenon, increasing pumping pressures, and should certainly be accounted for. Figure 15 also conrms the conclusion that the viscosity is a dominating factor for the pumping pressure, as the pressure loss is higher for the SCC compared to the CVC. Note that SCC 7 and CVC have a similar apparent viscosity at a shear rate of 10 s-1. Extrapolating the rheological curve to the shear rate range during pumping would deliver a larger apparent

viscosity for SCC 7, due to the shear-thickening behaviour.

5 Results for bends The practical guidelines take the inuence of bends and reducers into account by identifying an equivalent length. For example, the Schwing-guide [25] and Guptil et al. [4] state that a bend of 90 causes a pressure loss which is equivalent to 3 m of straight pipes. As a result, for each 90 bend, 3 m must be added to the total circuit length to calculate the pressure needed. In the research of Kaplan [6] and Chapdelaine [7], it is concluded, somewhat surprisingly, that the bends and reducers investigated in the respective researches do not cause an additional pressure loss. The bends and reducers can thus be considered as a straight section. In this project, the inuence of a 90 and a 180 bend have been investigated in the 80 and 100 m circuits. A pressure measurement section was installed just before and just behind the bend. The pressure loss in a section containing a bend was compared with the pressure loss in a straight section. By determining the pressure loss per unit of length (of straight pipes) from the latter section, the inuence of the bend was isolated in the former section. The bends have a centre line radius (CLR) of 17 cm, which implicates a rather short bend. The equivalent length for the bends is calculated as the pressure loss over the bend divided by the pressure loss per meter in a straight pipe. The equivalent length of a bend is thus the length of a straight pipe causing an equal pressure loss. Figure 16 shows an example of the pressure evolution with time, when applying a stepwise decrease in discharge rate. Figure 16a shows the pressure measured upstream (section A) and downstream (section B) of a straight section. The grey line represents the pressure measured after a 90 bend, downstream of the straight section (section C). Similarly, Fig. 16b shows the pressure measured in the same straight section (black linessection AB), and the pressure before a 180 bend, upstream of the straight section (section C). The pressure difference between the grey line and the closest black line includes the corresponding bend and one meter of straight pipes (as the pressure was measured in the middle of a 1 m straight section).

45 40

Pressure loss (kPa/m)

35 30
SCC

25 20 15 10 5 0 0 5 10 15 20 25
CVC

Discharge rate (l/s)

Fig. 15 The pressure lossdischarge rate curve reects the rheological behaviour of the concrete. While the CVC showed perfectly Bingham behaviour, the SCC (SCC 7) displayed shearthickening in the rheometer

552 Fig. 16 Example of the pressure evolution with time when pumping concrete at a stepwise decreasing discharge rate. The black lines represent the pressure in the two measurement locations in the straight section (sections A and B). In Fig. 16a, the grey line represents the pressure measured downstream of a 90 bend, downstream of the straight section (section C). In Fig. 16b, the grey line is the pressure is measured upstream of a 180 bend, upstream of the straight section (section C). Note that the pressure difference between the grey line and the corresponding upstream or downstream black line also includes 1 m of straight pipes

Materials and Structures (2013) 46:533555

(a)
Section A Pressure (bar)
9 8 7

Section C

Section B
6 5 4 3 2 1 0
0 50 100 150 200 250 300 350

Section A-B: Section B-C:

10.16 m straight 1.01 m straight + 90 bend

Section A Section B Section C


400

-1

Time (s)

(b)
Section A Pressure (bar)
14

12

Section C
10

Section B Section C-A: 1.01 m straight + 180 bend 16.16 m straight

Section A-B:
6

Section C Section A Section B

0 0 -2 50 100 150 200 250 300 350 400 450

Time (s)

Figure 17 shows the raw data for SCC 14, 15, 16 and 17. Each point corresponds to one full stroke of the pump at a certain discharge rate. Figure 17a shows the inuence of a 90 bend and is based on the results of SCC 16 and 17, while Fig. 17b shows the inuence of a 180 bend, based on SCC 14 and 15. From Fig. 17a

and b, it can be seen that for SCC, the equivalent length shows a very large scatter, but it is in most cases signicantly higher than the length of the bend, which is indicated by the grey dashed line in Fig. 17. The statement in the practical guidelines that a 90 bend is equivalent to 3 m of straight section (full black line) is

Materials and Structures (2013) 46:533555 Fig. 17 Pressure loss over a bend, expressed as the equivalent length. The black full line represents the statements in the practical guidelines, while the dashed grey line is the real length of the bend and corresponds to the results of Kaplan and Chapdelaine. For SCC, bends cause an additional pressure loss compared to a straight section. The additional pressure loss appears to decrease with increasing discharge rate and increasing concrete viscosity. In some cases, the rule of thumb of the practical guidelines is not sufcient to quantify the pressure loss over a bend

553

(a)

90 Bend

6 SCC 16 SCC 17

Equivalent length (m)

0 2.00

4.00

6.00

8.00

10.00

12.00

14.00

16.00

18.00

20.00

Discharge rate (l/s)


180 Bend

(b) 14
12

SCC 14 SCC 15

Equivalent length (m)

10

0 2.00

4.00

6.00

8.00

10.00

12.00

14.00

16.00

18.00

Discharge rate (l/s)

slightly above the average measured. On the other hand, it is not a safe statement as some measurement points indicate signicantly larger pressure losses. As a preliminary conclusion, it can be stated that the equivalent length decreases with increasing discharge rate and that for more viscous SCC, lower equivalent lengths are obtained, as SCC 14 is more viscous than SCC 15 and SCC 16 is more viscous than SCC 17. Generally, it can be concluded that bends cause an additional pressure loss in case of SCC, but more research is needed to determine the exact magnitude and the most important parameters.

6 Conclusions Based on full-scale pumping tests, the similarities and differences between pumping, CVC and SCC have been investigated. Furthermore, the applicability of the practical guidelines developed for CVC has been veried for SCC. In the practical guidelines, minimum values for the amount of nes, slump, etc. are dened to avoid the occurrence of friction in the concrete during pumping, as friction would lead to excessive pumping pressures and potentially blocking. Due to the adapted mix

554

Materials and Structures (2013) 46:533555 2. Wallevik JE (2003) Rheology of particle suspensions,.Ph-D dissertation, The Norwegian University of Science and Technology, Trondheim 3. Crepas RA (1997) Pumping concrete, techniques and applications, 3rd edn. Crepas and Associates, Inc., Elmhurst 4. Guptill NR et al (1998) (ACI-Comm 304), Placing concrete by pumping methods. American Concrete Institute, Farmington Hills M, Mailvaganam N, Malhotra VM, Jo5. Spiratos N, Page licoeur C (2003) Superplasticizers for Concrete: fundamentals, technology, and practice, Ottawa 6. Kaplan D (2001) Pumping of concretes. PhD dissertation, es, Paris Laboratoire Central des Ponts et Chausse 7. Chapdelaine F (2007) Fundamental and practical study on Lathe pumping of concrete. PhD dissertation , Universite bec val, Que 8. Tattersall GH, Banll PFG (1983) The rheology of fresh concrete. Pitman, London 9. Wallevik OH (2003) Rheology: a scientic approach to develop self-compacting concrete. Proc 3rd Int Symp on Self-Compacting Concrete, Reykjavik, pp 2331 10. Roussel N (2006) A thixotropy model for fresh uid concretes: theory, validation and applications. Cem. Conc. Res. 36:17971806 11. Wallevik JE (2009) Rheological properties of cement paste: thixotropic behavior and structural breakdown. Cem Concr Res 39:1429 12. Cyr M, Legrand C, Mouret M (2000) Study of the shear thickening effect of superplasticizers on the rheological behaviour of cement pastes containing or not mineral additives. Cem. Conc. Res. 30:14771483 13. De Larrard F, Ferraris CF, Sedran T (1998) Fresh concrete: a Herschel-Bulkley material. Mat Struct 31:494498 14. Feys D, Verhoeven R, De Schutter G (2008) Fresh self compacting concrete: a shear thickening material. Cem Concr Res 38:920929 15. Heirman G, Vandewalle L, Van Gemert D, Wallevik OH (2008) Integration approach of the Couette inverse problem of powder type self-compacting concrete in a wide-gap concentric cylinder rheometer. J Non-Newtonian Fluid Mech 150:93103 16. Yahia A, Khayat KH (2001) Analytical models for estimating yield stress of high-performance pseudoplastic grout. Cem Concr Res 31:731738 17. Feys D, Verhoeven R, De Schutter G (2009) Why is fresh self-compacting concrete shear thickening? Cem Concr Res 39:510523 18. Browne RD, Bamforth PB (1977) Tests to establish concrete pumpability. ACI-J 74:193203 19. Ede AN (1957) The resistance of concrete pumped through pipelines. Mag Concr Res 9:129140 20. Yammine J, Chaouche M, Guerinet M, Moranville M, Roussel N (2008) From ordinary rheology concrete to self compacting concrete: a transition between frictional and hydrodynamic interactions. Cem Concr Res 38:890896 21. ASTM C 33 (2003) Standard specication for concrete aggregates. American Society for Testing and Materials, Philadelphia 22. Jolin M, Burns D, Bissonnette B, Gagnon F, Bolduc L-S (2009) Understanding the pumpability of concrete. Proc 11th Conf Shotcrete Underground Support, Davos

design of SCC to full the criteria on lling and passing ability, blocking is less probable to occur during pumping. On the other hand, several blockings were observed during the insertion of SCC in the long pumping circuits due to a lack of cement paste at the concrete front. The preparation of a water-cement mixture to be inserted before the concrete remains necessary when SCC is employed. Due to the signicantly lower yield stress of SCC, the velocity prole in a pipe is different. The velocity prole of SCC is assumed to consist of a small plug in the centre of the pipe, a lubrication layer at the wall and a part of sheared concrete in between. In many cases during pumping CVC is not sheared. The practical guidelines for pumping concrete relate the pressure loss to the spread or slump of the concrete. This would imply that SCC should show very low pressure losses during pumping. On the other hand, as SCC itself is sheared (in addition to the shearing of the lubrication layer), the viscosity becomes a determining factor inuencing pumping pressures. Good correlations have been established between the pressure loss and the apparent viscosity of the concrete and between the pressure loss and the V-funnel ow time. Kaplan and Chapdelaine have found that a bend does not increase the pressure loss during pumping of CVC, while the practical guidelines state that a 90 bend is equivalent to 3 m of straight pipes. The preliminary results shown in this paper indicate that during pumping of SCC, an additional pressure loss occurs in the bends and that the pressure loss can even be larger than the rules of thumb. It also appears that the equivalent length of a bend is reduced with increasing discharge rate and increasing viscosity, but further research is needed to conrm these statements.
Acknowledgments The authors would like to acknowledge the Research Foundation in Flanders, Belgium (FWO) for the nancial support of the project and the technical staff of both the Magnel and Hydraulics laboratory for the preparation and execution of the full-scale pumping tests.

References
1. Ozawa K, Maekawa K, Kunishima M, Okamura H (1989) High-performance concrete based on the durability design of concrete structures. Proc 2nd East Asia: Pacic Conference on Structural Engineering and Construction, Chiang Mae, Vol. 1, pp 445456

Materials and Structures (2013) 46:533555 23. Kasami H, Ikeda T, Yamane S (1979) On workability and pumpability of superplasticized concrete. Proc 1st CANMET/ACI Conf Superplasticizers in Concrete, Ottawa, pp 6786 24. Sakuta M, Kasanu I, Yamane S, Sakamoto A (1989) Pumpability of fresh concrete. Takenaka Technical Research Laboratory, Tokyo 25. Schwing (1983) In: Eckardstein KEV (ed) Pumping, concrete and concrete pumps: a concrete placing manual, p 133 26. De Schutter G, Bartos P, Domone P, Gibbs J (2008) Selfcompacting concrete. Whittles Publishing, Caithness 27. Kaplan D, de Larrard F, Sedran T (2005) Avoidance of blockages in concrete pumping process. ACI Mat J 1023:183191 28. Jacobsen S, Haugan L, Hammer TA, Kalogiannidis E (2009) Flow conditions of fresh mortar and concrete in different pipes. Cem Concr Res 39:9971006 ssig M (1974) Pumping of fresh concrete, in particular 29. Ro lightweight concrete, through pipes. PhD dissertation, RWTH Aachen, Westdeutscher, Opladen, p 224 30. Chalimo T, Touloupov N, Markovskiy M (1989), Pecularities of concrete pumping (in Russian). Minsk 31. Morinaga M (1973) Pumpability of concrete and pumping pressure in pipeline. Proc RILEM Seminar on Fresh Concrete: Important Properties and their Measurement, Vol 7, Leeds, pp 139 32. Thrane LN (2007) Form lling with SCC. PhD dissertation, Technical University of Denmark, Denmark 33. Ngo TT (2009) Inuence of concrete composition on pumping parameters and validation of a prediction model

555 for the viscous constant. Ph-D dissertation, University Cergy-Pontoise Feys D (2009) Interactions between rheological properties and pumping of self-compacting concrete. PhD dissertation, Ghent University, Ghent Feys D, Heirman G, De Schutter G, Verhoeven R, Vandewalle L, Van Gemert D (2007) Comparison of two concrete rheometers for shear thickening behaviour of SCC. Proc. 5th Int RILEM Symp on SCC, Ghent, p 365370 Macosko CW (1994) Rheology principle, measurements and applications. Wiley-VCH, New-York rimentales sur le Poiseuille JLM (1840) Recherches expe ` s-petits dimouvement des liquides dans les tubes de tre ` tres. CR Acad Sci Paris 11:961967, 10411049 ame rimentales sur le Poiseuille JLM (1841) Recherches expe ` s-petits dimouvement des liquides dans les tubes de tre ` ters. CR Acad Sci Paris 12:112115 ame Buckingham E (1921) On plastic ow through capillary tubes. Proc. Am. Soc. Testing Mat. 21:11541161 Feys D, Verhoeven R, De Schutter G (2008) Extension of the Poiseuille formula for shear thickening materials and application to self compacting concrete. Appl Rheol 18(6): 62705 Jacobsen S, Vikan H, Haugan L (2010) Flow of SCC along surfaces. Proc 6th Int RILEM Symp Design, Prod Place al, Vol I, pp 63174 ment SCC, Montre Jodeh SA, Nassar GE (2009) Pumpability assessment of C90 SCC. Proc 2nd Int Conf Adv Concr Technol MiddleEast: Self-Consolid Concr, Abu Dhabi, pp 155176

34.

35.

36. 37.

38.

39. 40.

41.

42.

You might also like