- Email: [email protected]
Changes in concrete properties during pumping and formation of lubricating material under pressure
Cement and Concrete Research 108 (2018) 129–139
Contents lists available at ScienceDirect
Cement and Concrete Research journal homepage: www.elsevier.com/locate/cemconres
Changes in concrete properties during pumping and formation of lubricating material under pressure
T
⁎
Egor Secrierua, , Dario Cotardob, Viktor Mechtcherinea, Ludger Lohausb, Christof Schröfla, Christoph Begemannb a b
Technische Universität Dresden, Institute of Construction Materials, Dresden, Germany Leibniz Universität Hannover, Institute of Building Materials Science, Hanover, Germany
A R T I C LE I N FO
A B S T R A C T
Keywords: Fresh concrete Rheology Pumping Lubricating layer Filtrate Specific surface Temperature
This study quantifies the changes in the rheological properties of fresh concrete while focusing on lubricating layer (LL) formation during pumping. Full-scale pumping experiments were carried out on ready-mix concrete accompanied by the state-of-the-art rheological tests. Pumping markedly increased the yield stress. It also led to an increase in the air content, which contributed to a decrease in viscosity of fresh concrete. The dynamic loading from pumping generates a pressure gradient in concrete over the pipe cross-section. The pressure gradient is assumed to facilitate the movement of lubricating material to the concrete-wall interface, completing the formation of LL. This postulate is based on experimental evidence obtained by a portable high-pressure filter press and the extracted filtrate. The amount of filtrate depends on the specific surface of the fines, on concrete bulk viscosity, and on chemical admixtures. Finally, an increase in concrete temperature was observed depending on the concrete's composition and the properties of the LL.
1. Introduction Pumpability, in the context of concrete technology, is a mixture characteristic which describes its ability to be pumped through a pipeline. Pumpability is not an intrinsic concrete feature [1], but it is rather the result of a holistic approach involving the enhancement of concrete composition [2–4], the adaptation of pipeline geometry and pumping gear [5] under permanent monitoring and quality control on site [6]. Optimising the pumpability of concrete is a challenging task; it is always a search for a compromise between enhancing concrete pumpability and preserving the stability of the fresh mixture [7]. A complete blockage of the pumping line could be the ultimate consequence of unbalanced concrete rheological properties changing in time, even if the maximum capacity of the pump has not yet been reached [8]. An efficient pumping can be achieved if sufficient lubricating layer (LL) is generated at the pipe wall-concrete interface [9]. At the same time, the force transfer from pump to concrete shall occur through hydrodynamic interaction and not due to the frictional interaction among aggregate particles [10]. The hydrodynamic interaction can be achieved when in any cross-section along the pipeline the aggregates are fully covered with lubricating material (paste). Accordingly, the fresh concrete microstructure must be dense enough to prevent an
⁎
excessive water escape. As a result, a hydrodynamic stress transfer can be assured which is far less dissipative than the stress transfer though friction [11]. The hydraulic pressure gradient within concrete during pumping triggers the formation of the lubricating layer mainly due to flow-induced particle migration [12,13], but also due to the separation of water from the fines paste [14]. The water separation is facilitated by the fines of concrete matrix which act as a filter. A series of rheological tools to predict concrete pumpability, e.g. rheometers, have been applied and represent the background of the present research [1,15–17]. However, these tools can only partially reproduce the formation of lubricating material under pressure during pumping. In pursuing this challenge, a so-called portable high-pressure filter press (PHPFP) is applied to quantify the amount of filtrate that can be pressed out of concrete. It is considered that the total amount of filtrate after a given time interval and kinetics of water separation can help to estimate both the proneness of concrete to forming LL and its stability. Thus, the results of high-pressure filtration obtained by means of a hydraulic ram press are used in the article at hand to explain the findings gained from the pumping sequences. The availability of filtrate material in the mixtures under investigation and the influence on the pressure-flow rate relationship (P–Q curve) including curve slope and yintercept are analysed before and after pumping. Furthermore, the
Corresponding author. E-mail address: [email protected] (E. Secrieru).
https://doi.org/10.1016/j.cemconres.2018.03.018 Received 11 October 2017; Received in revised form 15 January 2018; Accepted 26 March 2018 0008-8846/ © 2018 Published by Elsevier Ltd.
Cement and Concrete Research 108 (2018) 129–139
E. Secrieru et al.
Table 1 Compositions of concrete mixtures under investigation. Material
Density [kg/m3]
Dosage [kg/m3] CVC mixtures
CEM III/A 42.5 N Fly ash Sand 0/2 Sand/gravel 2/8 Gravel 8/16 Water HRWRAa AEAb w/bc [−] Vol. aggregatesd [−] Vol. pasted [l/m3] a b c d
3075 2200 2650 2650 2650 1000 1040 1050 – – –
HPC mixtures
Grout
Gravel
M2A
M2B
M2C
M5A
M5B
M10A
M10B
SCC
360 – 781 508 526 180 2.88 – 0.50 0.69 300
310 50 781 508 526 180 2.88 – 0.50 0.69 306
310 50 781 508 526 180 2.50 – 0.50 0.69 306
310 50 781 508 526 180 3.60 – 0.50 0.69 307
360 – 735 479 496 180 2.88 1.80 0.50 0.65 302
360 – 735 479 496 180 2.52 1.80 0.50 0.65 301
360 120 724 471 488 180 5.04 – 0.38 0.64 356
360 120 724 471 488 180 4.08 – 0.38 0.64 356
360 220 667 434 450 180 6.06 – 0.31 0.59 402
400 280 1060 – – 340 – – 0.50 0.40 597
High-range water-reducing admixture as aqueous solution with 19 % content of active agent. Air-entraining agent. Water-to-binder ratio. Related to concrete unit volume.
capacity was only 1.5 m3. Each batch was immediately unloaded into a ready-mix truck for homogenisation and transport to the site of the pumping experiments. Rheological testing was performed inside a fully equipped warehouse for each concrete truckload as well as on the pumped concrete released from the duct. The recorded climatic conditions on site (including the warehouse) during experiments were: temperature of (20.5 ± 3.3) °C and a relative humidity of (69.2 ± 3.7) %. Each test was completed within a timespan corresponding to a maximum concrete age of 120 min. During testing, the mixtures remained visually stable, thus fulfilling the basic requirement of practical relevance.
changes in fresh concrete properties during pumping are experimentally captured and discussed in detail. As conclusion, two relatively simple tests for the estimating concrete pumpability and stability on site are proposed. 2. Experimental investigation 2.1. Materials and mixtures In total, nine concrete mixtures are investigated in the present paper. The concrete mixtures were purposefully developed for full-scale pumping experiments together with the involved industry partners. The mixtures were chosen according to their consistency and rheological properties, yield stress and plastic viscosity, so that various flow behaviours could be obtained: predominantly plug flow for conventional vibrated concretes (CVC) [18] and shear flow for self-compacting concretes (SCC) [19]. The compositions of the mixtures under investigation are given in Table 1. The mixtures Gravel (reference mixture) and M2A-M5B represent conventional vibrated concretes; M10 and M11 are flowable high-performance concretes (HPC) with increased paste content in comparison to ordinary concrete. Finally, SCC is a self-compacting mixture with high contents of paste and high-range water-reducing admixture (HRWRA). The physical and chemical characteristics of the cement and fly ash are given in Table 2. The parametric study included a variation of water-to binder ratio (w/b) and paste content. Mixtures M2A, M2B, M2C, M10A and M10B were compared in analysing the influence of the amount of HRWRA on concrete stability. Mixtures M5A and M5B were selected to investigate the effect of AEA; the designed air content was 5 %. The total amount of water as well as the sieving curve were kept constant for all the mixtures for the sake of easier comparison. Each concrete mixture was prepared at a ready-mix station in three replicate batches of 1.5 m3. This was necessary because a total volume of 4.5 m3 was required for the pumping experiments, but the mixer
2.2. Pumping circuit The horizontal pumping circuit had a length of 154 m, see Fig. 1. A truck-mounted concrete twin-cylinder hydraulic pump was employed to feed the pipe. The pipeline was made of high-pressure steel pipes of two diameters: DN125 and DN100, i.e., of inner diameter of 125 and 100 mm, joined through a DN125/100 reduction. The circuit ended with a distributor, from which concrete was discharged back into either the pump feed hopper or the disposal container. The data acquisition system was housed in a tent in the immediate vicinity of the pipeline. The pipeline was instrumented with eight pressure transducers of stainless steel for abrasive media (model BROSA type 0310, BROSA AG, Germany) with a measurement range up to 400 bar and a limit pressure of 300 %, Fig. 1c. The flow rate was monitored with an electromagnetic flow meter for pulsating flow (model Transmag 2 Sensor 911/E, Siemens AG, Germany) installed at the end of the pipeline, see Fig. 1d. All sensors were connected to the data acquisition system, thus recording and processing the data in real time by a portable computer. Before introducing the concrete into the pipeline, 0.75 m3 of a preparatory cement grout containing fine aggregates was pumped through to facilitate LL formation and the initial movement of the concrete; Table 1 gives the composition of the grout. Just after pumping
Table 2 Physical and chemical characteristics of the cement and fly ash. Type
CEM III/A 42.5 N Hard coal fly ash
Density [kg/m3]
3075 2200
Blaine fineness [cm2/g]
4780 3000
Chemical composition [%] SiO2
Al2O3
Fe2O3
CaO
MgO
SO3
K2O
Na2O
Cl
22.50 47.5
6.66 26
2.40 11.5
53.20 1.0
4.79 –
3.08 –
0.82 –
0.24 –
0.08 –
130
Cement and Concrete Research 108 (2018) 129–139
E. Secrieru et al.
Flow direction P1
P4 Reducer
DN125
P2
DN125
P3
DN100
P6
DN100
P7
P5 FM
P8
2.50 m
3.00 m
3.00 m
10X3.00 m
a) Concrete plant
Pump Pressure sensors Rubber hoses
Reducer
DN125
Flow meter
DN100
Wooden beams
b)
c)
d)
Fig. 1. a) Schematic representation of the pumping circuit and the location of eight pressure sensors (P1-P8) and flow meter (FM), b) overview of the pumping test setup; c) pressure sensor, d) electromagnetic flow meter.
respectively. In the above functions, τ0i is the yield stress [Pa] and μi is the viscosity parameter [Pa·s/m] of the LL, which can be measured by tribometer; τ0 is the yield stress [Pa] and μ is the plastic viscosity [Pa·s] of the concrete bulk; the parameters τ0 and μ can be measured by viscometer; k is the pipe filling coefficient (k = 1); R is the radius [m] and L is the length [m] of the pipe. The effective pressure predicted by the Sliper device and transposed onto the real pipeline geometry takes the form of Eq. (3):
this volume, the feeding of the pipeline was switched to the actual concrete mixture. The process corresponds to common practice on any construction site in pumping concrete [1,20,21]. Upon release at the end of the pipeline, the primary cement grout together with a small portion of “contaminated” concrete was dumped via the distributor into the container located next to the setup. The rest of the concrete, i.e. the relevant portion for later characterisation, was then pumped in circuit. Four different flow rates were imposed stepwise upwards (increasing) and downwards (decreasing), each step lasting 3 min. The total pumping time amounted to 24 min. The outcomes of each step were average values for flow rate and resulting pressure.
P = f (a, b, Q, L, R)
where L is the length [m] and R is the pipeline radius [m] [22,24]. The yield stress related parameter a [Pa] and viscosity related parameter b [Pa·s/m] are the outcomes of Sliper measurements. Further discussion on data processing with regard to employed rheological devices is available in [22].
2.3. Testing procedure 2.3.1. Rheological measurements Upon delivery of the fresh concrete, a sample was taken and its properties were tested by means of both empirical tests, i.e. slump flow and flow table tests, as well as by means of more the sophisticated rheological testing equipment: ConTec 5 viscometer (ConTec, Iceland), a so-called “tribometer” built on the basis of the viscometer [22] and the Sliding Pipe Rheometer, Sliper (Schleibinger GmbH, Germany); see Fig. 2. The pumpability of fresh concrete depends on the rheological properties of the concrete bulk and, above all, on those of the LL. Based on Kaplan's approach [18], the concrete flow can be of either plug or shear type. The pumping pressure P is described in [18,22,23] for the case of concrete plug flow type as
P = f (τ0i , μi , Q, k , L, R)
2.3.2. Filtrate amount and lubricating layer (LL) thickness The filtrate pressed out of the concrete was investigated using a portable high-pressure filter press (PHPFP) developed at the Institute of Building Materials Science, Leibniz Universität Hannover [14]; the test set-up is shown in Fig. 3a. The filtration cell has an inner diameter D = 150 mm and a filling height H = 300 mm and is used in combination with filter paper having a pore size of 4 to 12 μm. In this way, the rate of filtrate formation can be reproduced in a small-scale test. It is presumed to mimic the radial, friction-induced separation of the liquid phase from bulk concrete while being pumped through a long pipe. Here the flowing particles themselves act as the self-growing filter from which the LL is cast out. When the container is filled with freshly mixed concrete, under applied pressure the solution containing cement particles (filter material) tends to escape; see Fig. 3b. The applied force is increased linearly up to 100 kN (≈ 57 bar) at a rate of 3.33 kN/s.
(1)
and for the shear flow type as
P = f (τ0i , μi , τ0 , μ, Q, k , L, R)
(3)
(2) 131
Cement and Concrete Research 108 (2018) 129–139
E. Secrieru et al.
a)
b)
c)
Fig. 2. Rheological devices: a) viscometer, b) tribometer and c) Sliper with weights (courtesy of Stephan Falk).
Fig. 3c indicates the rate of filtrate formation in time for the investigated mixtures. The curves of the mixtures were compared based on the total filtrate amount. Depending on concrete composition, two main phases of the filter material formation are identified. In the first phase, the rate of filtrate formation is assumed to be nearly constant. In the second phase, the rate gradually falls to zero. For the mixtures under investigation, the maximum filtrate amount is attained within 5 min corresponding to the duration of the first phase. The transition time cut between those two phases was chosen for a filtrate rate formation below 0.3 g/s for 5.3 l of concrete sample subjectively as a compromise between the desire to keep measuring time short (with respect to applicability of the method in the praxis of construction) and the well measurable filtrate amount for various concrete mixtures. This choice resulted in the duration of the first phase of 300 s, or 5 min. The second phase, another 5 min, covers the remaining time to 600 s. The higher the filtrate amount obtained, the better the assumed pumpability. The main problem is the rate of filtrate material formation since intense “bleeding” is a clear indication of an unstable mixture [1]. The authors have therefore used the rate of filtrate formation as a measure of pumpability based on the amount of water with a maximum particle size ≤ 0.012 mm extruded out of the concrete samples. At this stage it should be remarked that if such a highly densified state occurred in the pipeline, the nature of the interactions among the particles would change from hydrodynamic to directly frictional and finally cause a blockage of the pipeline [10,25] even if a highly efficient HRWRA is used as a lubricating agent. Since the procedure is intended for use on site, the testing time was adjusted according to the total pumping time of concrete. On site, depending on pipeline length, the concrete leaves the pipeline within 2 to 15 min [26]. It is therefore meaningful to analyse the formation of filter material within the time range experimentally. Fig. 4 shows a schematic representation of the filtrate extraction process. According to this model, under high pressures corresponding to the pumping pressure (maximum 100 bar), the physically bound water,
i.e., rheologically effective water, is expected to be pressed out of the concrete. On the other side, the chemically bound water and interstitial water are expected to remain in the system, unless very high temperature is applied [27]. Theoretically, the entire filtrate amount is interpreted as being available for LL formation and enhancement. Potentially, less liquid volume could contribute to LL formation, since the maximum applied pressure in the filtration setup is higher than the pumping pressure. The relative amount of paste Vpaste [l] required to build a LL of a specific thickness e can be calculated using Eq. (4):
Vpaste = π [R2‐(R‐e )2] L
(4)
where R [m] and L [m] are pipeline radius and length, respectively. The necessary amount of lubricating material corresponding to a minimum LL thickness e for the full-scale pipeline is shown as an example in Table 3. 3. Results and discussion 3.1. Influence of pumping on fresh concrete properties The results of the rheological tests are summarised in Table 4. These are yield stress and viscosity related parameters as determined using the rheological devices; see Section 2.3.1. The plastic viscosity μ and the yield stress τ0 are those of the concrete bulk. The viscosity parameters μi and b and the yield stress parameters τ0i and a describe the properties of the lubricating layer between the concrete and the pipe wall. In all mixtures the viscosity parameters μ and μi decreased during pumping. The only exceptions were the CVC mixtures containing AEA, i.e., M5A, M5B, M2C, the last with a higher HRWRA content in comparison to M2A and M2B. For these mixtures the viscosity parameters μi and b slightly increased. The formation of filtrate can be expected to reduce the magnitude of the rheological parameters of the lubricating layer as measured with a tribometer or Sliper. The reason for that is additional water which decreases both yield stress and viscosity 132
Cement and Concrete Research 108 (2018) 129–139
E. Secrieru et al.
Press Barometer Pump
Filtration cell (container)
Scale
a)
b)
Filtrate formation rate [g/s]
3.5 Gravel M2A M2B M2C M5A M5B M10A M10B SCC
3.0 2.5 Phase 1 2.0 1.5 1.0
Phase 2 0.5 0.0 0
150
300
450
600
Time t [s]
c) Fig. 3. a) High-pressure filter press for stability tests; b) exemplary extracted aqueous filtrate sample for Gravel mixture; c) filtrate formation in time pressed out of the concrete mixtures.
Fig. 4. Schematic representation of filtrate extraction from concrete sample. 133
Cement and Concrete Research 108 (2018) 129–139
E. Secrieru et al.
average about 46 % at the beginning and 48 % at the end of pumping campaign independent of the mixture type; see Fig. 5. The filling degree represents the ratio between the real volume of aspirated concrete and the volume for a filled piston. Table 6 shows an example of the determination of the filling degree for the reference mixture Gravel based on the measured data and piston geometry. The measured values are significantly lower in comparison to 70 % mentioned in [1] or even 90– 100 % which were assumed in [8,29,33]. This means that in the present study the pistons, by default, mechanically induced air into the mixtures under investigation. The additional air is lost during pumping, especially in steep pipelines where concrete “free fall” on site occurs [8]. However, such “free fall” was not a part of the pipeline setup in the research at hand. That the effect of air content in concrete had much less effect on the viscosity parameters of LL can be probably traced back to high shear gradients in the layer which “expel” air bubbles towards the less sheared concrete bulk.
Table 3 Necessary amount of lubricating material depending on LL thickness. Pipeline section
DN125 DN100
Pipe diameter D [mm]
125 100
Radius R [mm]
62.5 50.0
Pipe volume [m3]
Total length L [m]
79 75
0.969 0.589
Paste volume Vpaste [l] for LL of assumed thickness e e = 1 mm
e = 2 mm
54.10a
107.24a
a
Total amount of paste for two pipeline sections under the assumption that e remains constant independent of pipe diameter.
parameters of the lubricating layer [28]. The increase in the viscosity parameter corresponds to a very pronounced decrease in filtrate amount for the pumped concrete in comparison to the same mixture before pumping; see Section 3.3. The yield stress parameters τ0, τ0i and a increased significantly. In other words, all concretes became stiffer upon being pumped. The sharp increase in yield stress, manifested in the form of a drop in slump flow or spread flow values, has previously been reported [26,29]. One explanation for the behaviour may be a temperature increase in concrete occurring during pumping due to permanent friction at the pipe wallconcrete interface; see Table 5. Three degrees Kelvin increase for CVC mixtures and up to eight degrees for HPC during the 24 min of dwell time of the material in the circuit certainly accelerate the cement hydration, which results in a significant rise of the yield stress as reported in [30,31] compared to isothermal conditions upon the mere time interval. Opposite the increase in yield stress-related parameters upon pumping, the plastic viscosity μ of bulk concrete decreased for all mixtures. The corresponding viscosity parameters μi and b of the lubricating layer decreased for the CVC mixtures Gravel, M2A, M2B and HPC mixtures M10A, M10B and SCC whereas they increased for CVC mixtures M2C, M5A and M5B; see Table 4. From the experience it is known that increase in the air content leads to decrease in the viscosity parameters. Such a relationship was observed for the mixtures Gravel, M2A, M2B, M10A, M10B and SCC for all viscosity related values but, interestingly, are not valid for M2C, M5A and M5B concerning the viscosity of LL; see Table 5. Furthermore, it is worth noting that in the bulk concrete the viscosity parameters changed more pronouncedly than those of LL. Previously, it was suggested that the changes in fluidity cannot be regarded as due only to the changes in the lubricating layer properties [32]. In the case of the viscosity parameter, the main reason for the decrease was likely the incomplete filling of the pump pistons, on
3.2. Influence of composition and rheological properties of concrete on filtrate kinetics Fig. 6a and b show the effect of the specific surface of fine particles (< 0.125 mm) on the filtrate amount pressed out of concrete before and after pumping. The specific surface of fines was determined by multiplying the mass of binder contained in the concrete sample with the corresponding Blaine fineness value; see Table 2. In general, the filtrate amount decreased in inverse proportion to the specific surface area of the fines. The mixtures in the ellipse are CVC; they differed only slightly in the total surface area due to cement replacement by fly ash. By comparison such replacement increased the filtrate amount for the constant water content in the initial system, cf. the CVC mixtures Gravel and M2A. This is not surprising considering different water demand of cement and fly ash particles due to the differences in the particle shape [34], the effect of which cannot be appropriately captured by Blaine measurements. The actual effect of HRWRA on the dispersion degree of fine particles can only be partially considered unless the particle surfaces are fully saturated. In such case the fine particles would decant as sediment [35], and the mixtures would become very unstable, which does not occur frequently. Therefore, the available and actually accessible total surface area of the fines is unknown. Nevertheless, the mixtures belonging to the group CVC have a noticeably lower total surface area of fines (due to the lower fines content) than HPC mixtures. Thus, a general trend can be well seen. For example, in comparison to CVC mixtures significantly lower filtrate amounts were released by the mixtures M10A, M10B and SCC upon applying the pressure in the PHPFP. Here, the filtrate amounts pressed out of CVC mixtures M2A,
Table 4 Rheological parameters of fresh concrete before and after pumping (left and right columns, respectively). Mix.
Yield stress τ0 [Pa] Gravel M2A M2B M2C M5A M5B M10A M10B SCC a b c d
LLa: tribometer
Concrete bulk: viscometer
112c 99 148 133 64 155 114 262 18
Viscosity μ [Pa∙s] 324d 252 356 319 206 329 404 460 332
30c 22 17 20 11 19 36 31 36
Concrete bulkb and LL: Sliper
Yield stress τ0i [Pa] 14d 15 13 12 8 13 15 15 18
48c 20 56 35 26 54 94 65 11
Viscosity μi [Pa∙s/m] 152d 91 153 125 106 124 126 140 108
LL = lubricating layer. Concrete bulk can be partially sheared. Before pumping. After pumping. 134
944c 737 591 530 531 927 1283 1438 2249
Yield stress a [Pa] 710d 522 531 536 578 982 1136 1164 1557
141c 107 200 131 156 213 97 193 21
Viscosity b [Pa∙s/m] 381d 174 299 229 190 242 309 270 265
587c 482 380 216 235 458 689 942 935
305d 367 303 306 272 471 471 630 747
Cement and Concrete Research 108 (2018) 129–139
E. Secrieru et al.
Table 5 Results of slump flow, flow table and other conventional tests on fresh concrete (left and right columns, respectively). Flow table, slump flow⁎ [mm]
Mix. Gravel M2A M2B M2C M5A M5B M10A M10B SCC⁎ ⁎ a b
530 615 565 605 640 490 655 485 673
a
430 495 470 480 530 510 435 430 318
Density [kg/dm3]
Air content [%] b
1.4 1.3 1.2 0.8 3.7 4.9 1.5 2.1 2.0
a
4.2 3.7 2.2 2.6 5.1 5.1 2.7 2.8 2.9
2.37 2.39 2.40 2.39 2.30 2.30 2.38 2.35 2.28
a
Temperature [°C] b
2.27 2.30 2.31 2.31 2.25 2.26 2.31 2.30 2.28
22.5a 20.5 25.5 27.5 23.8 20.0 21.5 23.0 23.0
26.2b 23.3 28.5 32.5 25.7 24.9 27.0 31.0 28.1
Slump flow. Before pumping. After pumping.
100
Filling degree [%]
Gravel mixture with slightly higher HRWRA content. Obviously, the cement replacement by fly ash at a dosage of 10 % could “overcompensate” the decrease in filtrate amount due to lower HRWRA content, due to the lower specific surface of the fly ash and thus higher amount of “free” water; see also the comments above. Regarding the CVC mixtures M5A and M5B, one can remark an increasing effect of adding air-entraining agent on the filtrate amount in comparison to the reference mixture. While showing the highest filtrate amount, the mixture M5A was prone to bleeding but remained pumpable. It seems that, in addition to HRWRA, AEA as well exhibited a fluidising effect that caused some bleeding. Since the mixture barely reached an air content of 3.7 %, it was decided to modify M5A by decreasing the amount of HRWRA and denominate it as M5B. The result was a more stable mixture with an initial air content of 5 %. Analysing the case of the pumped mixtures, for the CVC the filtrate amount decreased in comparison to the same mixtures tested before pumping, while the opposite was the case for the concretes with a high content of fines. When pumped the mixtures M10A, M10B and SCC changed their consistency from “sticky” to rather “stiff”, which “allowed” the filtrate to move within the system. Consequently, the filtrate retention in these concretes decreased in the course of pumping. The filtrate amount depended on concrete plastic viscosity μ and exerted a substantial impact on the resulting viscosity parameter μi; see Fig. 7. Indeed, it is not only the rheological properties but also the amount of LL forming at the concrete/pipe wall interface that determines the magnitude of the viscosity parameter μi. The rheological tests showed that the CVC mixtures Gravel and M2A-M5B with lower viscosity featured a higher filtrate release. Thus, the lubricating layer forming for the mixtures Gravel and M2A-M5B is postulated to be thicker than in the case of the HPC mixtures M10A, M10B and SCC due to the higher filtrate retention of the latter. However, also the properties of LL depend on the extent of the filtrate release.
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
80 60 40 20 0 5
15 25 35 Flow rate Q [m3/h]
45
Fig. 5. Variation of filling degree depending on the concrete type and flow. Table 6 Example of calculation the piston filling degree for the reference mixture Gravel. Number of strokes [−]
13 22 28 39 29 22 9 24
b
Piston geometry Diameter [mm]
Length [mm]
Flow rate Q [m3/ h]
230 230 230 230 230 230 230 230
2000 2000 2000 2000 2000 2000 2000 2000
10.52 16.47 18.04 29.67 21.95 14.12 7.93 15.01
Concrete volume [m3] per stroke
Piston volume [m3]
Filling degree k [%]
0.57 0.86 0.94 1.48 1.06 0.74 0.40 0.75
1.08 1.83 2.33 3.24 2.41 1.83 0.75 1.99
52.8 47.2 40.2 45.8 44.0 40.4 53.0 37.6
3.3. Influence of filtrate amount on pumpability The pumpability of fresh concrete depends on the rheological properties of concrete bulk and forming LL [17,18,22]. In practice it is best described as the relationship between pumping pressure and flow rate in a P–Q curve with A as y-intercept and B as curve slope; see Fig. 8 and Table 4. The parameter A is attributed to the initial pressure push necessary to initiate concrete flow in the given pipeline, similar to the yield stress parameters τ0 and τ0i. In the previous studies, it was shown that the linear correlation between the yield stress parameter of the lubricating layer τ0i and the parameter A is not always pronounced or straightforward [22,37], in the present investigation the coefficient of determination R2 reaches a value 0.50. The parameter B determines the concrete resistance against movement in the pipeline as a function of moving speed and depends on the
M2B and M2C were dominated by the HRWRA dosage. Among them, M2C containing fly ash, with the highest HRWRA dosage, “generated” the highest filtrate amount. On the other end M2B yielded the lowest filtrate amount, corresponding to the lowest HRWRA dosage. Obviously, by increasing the dosage of HRWRA at a constant water amount in the mixture, the percentage of the “rheologically effective” water also increases. At the same time, the highly dispersive effect of the HRWRA also increases the “active” surface area of fine particles in the mixture and, subsequently, the amount of chemically bound water [36]. However, the amount of water released from agglomerates is greater than the amount of chemically bound water on the surface of the particles. Even so, the amount of filtrate in M2B is higher than in the 135
Cement and Concrete Research 108 (2018) 129–139
E. Secrieru et al.
80
60 50 40 30 20
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
70 Filtrate amount [l/m3]
70 Filtrate amount [l/m3]
80
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
R² = 0.64
60 50 40 30
R² = 0.80
20 10
10
0
0
0
0 1 2 3 Sp. surface of fines Afines·109 [cm2/m3]
a)
1 2 3 Sp. surface of fines Afines·109 [cm2/m3]
b)
80
80
70
70
60
60
Filtrate amount [l/m3]
Filtrate amount [l/m3]
Fig. 6. Effect of the specific surface of fine particles (< 0.125 mm) in concrete on the filtrate amount a) before and b) after pumping.
50 Gravel M2A M2B M2C M5A M5B M10A M10B SCC
40 30 20 10 0 0
R² = 0.91
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
40 30 20 10
R² = 0.63
0
10 20 Plastic viscosity
30 [Pa·s]
40
0
a)
10 20 Plastic viscosity
30 [Pa·s]
40
b) 80
80
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
60 50 40 30 R² = 0.71
20
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
70 Filtrate amount [l/m3]
70 Filtrate amount [l/m3]
50
10
60 50 40 30 R² = 0.75
20 10
0
0 0
1000 2000 3000 Viscosity parameter i [Pa·s/m]
0
c)
1000 2000 3000 Viscosity parameter i [Pa·s/m]
d)
Fig. 7. Correlation between the filtrate amount and plastic viscosity μ for concretes a) before and b) after pumping as well as between the filtrate amount and viscosity parameter μi for concretes c) before and d) after pumping. 136
Cement and Concrete Research 108 (2018) 129–139
E. Secrieru et al.
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
40 30
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
50
Eff. pressure P [bar]
Eff. pressure P [bar]
50
20 10
40 30 20 10
0
0 0
10 20 30 Flow rate Q [m3/h]
40
50
a)
0
10
20 30 40 3 Flow rate Q [m /h]
50
b)
Fig. 8. P–Q curves obtained from the full-scale experiments on concrete mixtures a) at the beginning and b) at the end of pumping. Table 7 Parameters A (y-intercept) and B (slope) describing P–Q curve from full-scale pumping experiments and filtrate amount measured before and after pumping (left and right columns, respectively). Mix.
ΔA
y-Intercept A [kPa]
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
70b 68 234 258 236 313 247 324 −253a
868c 539 699 630 454 712 1233 1171 1411
ΔB
Slope B 3
[%]
[kPa∙h/m ]
+1137 +687 +199 +145 +93 +128 +399 +261 –
85b 74 64 56 44 76 146 161 163
60c 58 52 46 38 63 111 128 105
Filtrate amount
Change
3
[%]
[l/m ]
−29 −23 −19 −18 −13 −17 −24 −20 −35
45.3d 57.4 51.4 66.1 74.5 60.4 39.1 40.8 29.5
[%] 49.7e 56.1 51.4 56.0 63.3 55.2 39.6 41.1 33.1
+9.6 −2.3 −0.1 −15.3 −15.0 −8.6 +1.2 +0.6 +12.2
a The negative value is the result of a linear course of P–Q assumed by default; actually, SCC rather exhibits nonlinear P–Q behaviour due to shear thickening at elevated flow rates [38]. b Pumping at increasing flow rate. c Pumping at decreasing flow rate. d Before pumping. e After pumping.
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
300
200
100
300
200
100 Phase 1
Phase 1
Phase 2
0
Phase 2
0 0
a)
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
400
Filtrate amount [g]
Filtrate amount [g]
400
150 300 450 600 750 900 Time t [s]
0
150 300 450 600 750 900 Time t [s]
b) Fig. 9. Filtrate formation in time measured on concretes a) before and b) after pumping. 137
Cement and Concrete Research 108 (2018) 129–139
E. Secrieru et al.
4. Conclusions
Table 8 Lubricating layer thickness e [mm] corresponding to the available filtrate amount before and after pumping. e [mm]
ebefore eafter
The paper at hand has dealt with the influence of pumping on rheological properties of fresh concrete behaviour. For this purpose, full-scale pumping experiments were conducted on ready-mix concrete accompanied modern rheological tests before and after pumping. The focus was set on the formation of lubrication layer and on the parameters, which determine its properties. In this context a portative highpressure filter press was used to estimate the stability of the mixtures under conditions of pumping. The main outcomes of the study are:
Mixture Gravel
M2A
M2B
M2C
M5A
M5B
M10A
M10B
SCC
1.31 1.43
1.65 1.61
1.48 1.48
1.91 1.61
2.15 1.82
1.74 1.59
1.13 1.14
1.18 1.18
0.85 0.95
viscosity parameters μ and μi. An obvious correlation was observed between the viscosity parameters, especially μi and B, with R2 = 0.80; see Tables 4 and 7. Depending on concrete composition, the parameter B tendentially exhibits elevated values for the mixtures with increased binder (fines) content, i.e., M10A, M10B and SCC, and well correlates with the viscosity parameter μi, the latter mostly determining concrete pumpability [22]; see Table 7. As a result of pumping the magnitude of the A parameter is altered noticeably more than that of the parameter B. The measured change is a clear indication that during pumping concrete loses its consistency in terms of yield stress and slump flow/flow table spread, which necessitates the higher initial pumping pressure needed to start moving concrete in the pipeline. With regard to the filtrate amounts, mixtures M5A, M5B and M2C with their “highest” or “easiest” pumpability also generated the largest amount of expressible suspension of fines, i.e., lubricating material. At the other extreme were the HPC mixtures with considerably elevated fines content, i.e., M10A, M10B and SCC, which required more liquid to cover their particles' surfaces. The difference in filtrate amount measured for M2A and M2B had only a negligible effect on the slope of the P–Q curve. The same remark is also valid for the mixtures M10A and M10B: In both cases, the HRWRA amount was varied. It should be noted that based on the consideration of filtrate amounts it is only possible to qualitatively predict the pumpability of concretes with a similar composition. In the next step, the rate of filtrate formation was analysed; see Fig. 9. Depending on the concrete composition, the extruded water facilitates the formation and enhancement of the LL with respect to the available lubricating material. Since the filtrate material has an aqueous character, it must exert a dilutive effect on the composition of the LL and, thus, lead to decrease in its rheological parameters, i.e., yield stress and plastic viscosity. However, an excessively high filtrate formation rate indicates a reduction in stability [14]. M5A showed the highest filtrate formation rate and was prone to bleeding already before pumping. The mixture M5A could successfully be pumped without causing any trouble, the worst of which would be blockage upon an excessive loss of water due to bleeding. The other extremes with respect to both filtrate formation rate and the total amount of filtrate, i.e., HPC mixtures M10A, M10B and SCC, required the greatest effort in terms of effective pumping pressure. The HPC mixtures with their high fines contents, however, generated higher filtrate amounts after the pumping, while the filtrate formation rates remained low in Phase 1. A reason for the changed behaviour in Phase 2 might be the decrease in mixture consistency and increase in air content that facilitated the movement of the filtrate water out of the system. Since the volume of LL material cannot exceed the total amount of the available paste, the maximum physically possible LL thickness was calculated using Eq. (4). Table 8 shows the resulting LL thicknesses according to the available lubricating material. Accordingly, the LL thickness can be quantitatively improved through the filtrate material that is “pressed” towards the wall during pumping with up to 2 mm.
• Pumping increases yield stress and reduces the viscosity of fresh concrete. • An increase in paste volume by increasing the amount of binder • • • • • • • •
(fines), at constant water amount, generally leads to a marked increase in the required pressure for a given flow rate. The total amount of filtrate pressed out of concrete in a given time interval and the filtration rate influence the formation of lubricating layer and subsequently the pumpability of concrete to a great extent. The filtrate material can increase the thickness of the lubricating layer by up to 2 mm. The filtrate amount depends on the total surface area of the fines and on the related concrete viscosity parameters, especially on the plastic viscosity of the concrete bulk. As the surface area of the fines increases, the rheologically effective water and, thus, the amount of filtrate decrease. An increase in the dosage of HRWRA at a constant water content in concrete also increases the rheologically effective amount of water and the amount of filtrate. The degree of filling of pump pistons reaches an average value of 50 %. Such relatively low filling degree is considered responsible for the increase in air content in fresh concrete during pumping. The temperature increase due to pumping is more pronounced for HPC in comparison to CVC. Sliper and the portative high-pressure filter press have proven their applicability for quick assessment of concrete pumpability and stability on site.
Acknowledgements The authors are grateful to the Federal Ministry for Economic Affairs and Energy of the Federal Republic of Germany and to the German Federation of Industrial Research Associations, Collective Research Project grant number 18361 BR/1, for supporting this research through grant number 18361 BR/1 “Zielsichere betontechnische Gestaltung des Pumpens von Frischbeton”. The authors appreciate the important contribution of the German Ready-mixed Concrete Research Association (Forschungsgemeinschaft Transportbeton e.V.) in helping to organise and monitor the full-scale pumping experiments. The authors extend their gratitude to the German Association for Concrete and Construction Technology (Deutscher Beton- und Bautechnik-Verein e.V.) for the financial support in developing the portable high-pressure filter press (PHPFP) [39]. References [1] D. Kaplan, F. De Larrard, T. Sedran, Avoidance of blockages in concrete pumping process, ACI Mater. J. 102 (2005) 183–191. [2] M.S. Choi, Y.J. Kim, K.P. Jang, S.H. Kwon, Effect of the coarse aggregate size on pipe flow of pumped concrete, Constr. Build. Mater. 66 (2014) 723–730. [3] F. Chapdelaine, Fundamental and Practical Study on the Pumping of Concrete, PhD Thesis, Laval University, 2007. [4] T. Neumann, Einflüsse auf die Pumpbarkeit von Beton, Beton, Vol. 5 (2012), pp. 166–171. [5] Placing Concrete by Pumping Methods, ACI 304.2R-96 (Reapproved 2008), (1996). [6] O. Río, Á. Rodríguez, S. Nabulsi, M. Álvarez, Pumping quality control method based on online concrete pumpability assessment, ACI Mater. J. 108 (2011) 423–431. [7] D. Feys, G. De Schutter, K.H. Khayat, R. Verhoeven, Changes in rheology of self-
138
Cement and Concrete Research 108 (2018) 129–139
E. Secrieru et al.
[23] E. Secrieru, J. Khodor, V. Mechtcherine, C. Schröfl, Lubricating Layer and Its Relevance to Concrete Pumpability, (2017) (Submitt. to CBM). [24] V. Mechtcherine, V.N. Nerella, K. Kasten, Testing pumpability of concrete using Sliding Pipe Rheometer, Constr. Build. Mater. 53 (2014) 312–323. [25] D. Feys, Understanding the pumping of conventional vibrated and self-compacting concrete, in: N. Roussel (Ed.), Underst. Rheol. Concr, Woodhead Publishing Limited, Cambridge, 2011, pp. 331–353. [26] J. Aldred, Burj Khalifa – a new high for high-performance concrete, Proc. ICE Civ. Eng. 163 (2010) 66–73. [27] H.F.W. Taylor, Cement Chemistry, 2nd ed., Thomas Telford Publishing, 1997. [28] Ó.H. Wallevik, J.E. Wallevik, Rheology as a tool in concrete science: the use of rheographs and workability boxes, Cem. Concr. Res. 41 (2011) 1279–1288. [29] R. Khatib, Analysis and Prediction of Pumping Characteristics of High-Strength Selfconsolidating Concrete, PhD Thesis, Université de Sherbrooke, 2013. [30] W. Schmidt, H.J.H. Brouwers, H.-C. Kühne, B. Meng, Influences of superplasticizer modification and mixture composition on the performance of self-compacting concrete at varied ambient temperatures, Cem. Concr. Compos. 49 (2014) 111–126. [31] E. Secrieru, V. Mechtcherine, C. Schröfl, D. Borin, Rheological characterisation and prediction of pumpability of Strain-Hardening Cement-Based-Composites (SHCC) at various temperatures, Constr. Build. Mater. 112 (2016) 581–594. [32] J.H. Kim, S.H. Kwon, S. Kawashima, H.J. Yim, Rheology of cement paste under high pressure, Cem. Concr. Compos. 77 (2017) 60–67. [33] K.A. Riding, J. Vosahlik, D. Feys, T. Malone, W. Lindquist, Best Practices for Concrete Pumping, (2016). [34] P.-C. Aïtcin, S. Mindess, Modern binders, Sustain. Concr, CRC Press, 2011, pp. 87–90. [35] R.J. Flatt, I. Schober, Superplasticizers and rheology of concrete, in: N. Roussel (Ed.), Underst. Rheol. Concr, 1st ed., Woodhead Publishing Limited, 2012, pp. 144–208. [36] E. Sakai, T. Kasuga, T. Sugiyama, K. Asaga, M. Daimon, Influence of superplasticizers on the hydration of cement and the pore structure of hardened cement, Cem. Concr. Res. 36 (2006) 2049–2053. [37] E. Secrieru, M. Butler, V. Mechtcherine, Prüfen der Pumpbarkeit von Beton – Vom Labor in die Praxis, Bautechnik (11) (2014) 797–811. [38] D. Feys, R. Verhoeven, G. De Schutter, Why is fresh self-compacting concrete shear thickening? Cem. Concr. Res. 39 (2009) 510–523. [39] L. Lohaus, D. Cotardo, DBV 308 – Pumpstabilität von Beton. Prüfverfahren zur Mischungsstabilität bei hohem Druck, Schlussbericht, (2017).
consolidating concrete induced by pumping, Mater. Struct. 49 (2016) 4657–4677. [8] D. Feys, Interactions Between Rheological Properties and Pumping of Self-compacting Concrete, PhD Thesis, Gent University, 2009. [9] M. Choi, N. Roussel, Y. Kim, J. Kim, Lubrication layer properties during concrete pumping, Cem. Concr. Res. 45 (2013) 69–78. [10] R. Browne, P. Bamforth, Tests to establish concrete pumpability, ACI J. 74 (1977) 193–203. [11] J. Yammine, M. Chaouche, M. Guerinet, M. Moranville, N. Roussel, From ordinary rheology concrete to self compacting concrete: a transition between frictional and hydrodynamic interactions, Cem. Concr. Res. 38 (2008) 890–896. [12] J. Spangenberg, N. Roussel, J.H. Hattel, H. Stang, J. Skocek, M.R. Geiker, Flow induced particle migration in fresh concrete: theoretical frame, numerical simulations and experimental results on model fluids, Cem. Concr. Res. 42 (2012) 633–641. [13] M. Choi, Y.J. Kim, S.H. Kwon, Prediction on pipe flow of pumped concrete based on shear-induced particle migration, Cem. Concr. Res. 52 (2013) 216–224. [14] Y.A. Abebe, Flowable and Stable Concrete. Design, Characterisation and Performance Evaluation, PhD Thesis, Leibniz Universität Hannover, 2017. [15] E. Secrieru, Pumping behaviour of modern concretes - Characterisation and prediction, PhD Thesis Technische Unversität Dresden, 2018. [16] Y.A. Abebe, L. Lohaus, Rheological characterization of the structural breakdown process to analyze the stability of flowable mortars under vibration, Constr. Build. Mater. 131 (2017) 515–525. [17] D. Feys, K.H. Khayat, A. Perezschell, R. Khatib, Development of a tribometer to characterize lubrication layer properties of self-consolidating concrete, Cem. Concr. Compos. 54 (2014) 40–52. [18] D. Kaplan, F. de Larrard, T. Sedran, Design of concrete pumping circuit, ACI Mater. J. 102 (2005) 110–117. [19] D. Feys, G. De Schutter, R. Verhoeven, Parameters influencing pressure during pumping of self-compacting concrete, Mater. Struct. 46 (2013) 533–555. [20] C. Hazaree, V. Mahadevan, Single stage concrete pumping through 2.432 km (1.51 miles): weather and execution challenges, Constr. Mater. 3 (2015) 56–69 (Case Study). [21] H. Imaizumi, H. Jitousono, T. Kemi, K. Fujii, Study on the blockage in concrete pumping with preceding mortars, Cement Sci. Concr. Technol. 53 (1999) 732–737. [22] E. Secrieru, S. Fataei, C. Schröfl, V. Mechtcherine, Study on concrete pumpability combining different laboratory tools and linkage to rheology, Constr. Build. Mater. 144 (2017) 451–461.
139
Contents lists available at ScienceDirect
Cement and Concrete Research journal homepage: www.elsevier.com/locate/cemconres
Changes in concrete properties during pumping and formation of lubricating material under pressure
T
⁎
Egor Secrierua, , Dario Cotardob, Viktor Mechtcherinea, Ludger Lohausb, Christof Schröfla, Christoph Begemannb a b
Technische Universität Dresden, Institute of Construction Materials, Dresden, Germany Leibniz Universität Hannover, Institute of Building Materials Science, Hanover, Germany
A R T I C LE I N FO
A B S T R A C T
Keywords: Fresh concrete Rheology Pumping Lubricating layer Filtrate Specific surface Temperature
This study quantifies the changes in the rheological properties of fresh concrete while focusing on lubricating layer (LL) formation during pumping. Full-scale pumping experiments were carried out on ready-mix concrete accompanied by the state-of-the-art rheological tests. Pumping markedly increased the yield stress. It also led to an increase in the air content, which contributed to a decrease in viscosity of fresh concrete. The dynamic loading from pumping generates a pressure gradient in concrete over the pipe cross-section. The pressure gradient is assumed to facilitate the movement of lubricating material to the concrete-wall interface, completing the formation of LL. This postulate is based on experimental evidence obtained by a portable high-pressure filter press and the extracted filtrate. The amount of filtrate depends on the specific surface of the fines, on concrete bulk viscosity, and on chemical admixtures. Finally, an increase in concrete temperature was observed depending on the concrete's composition and the properties of the LL.
1. Introduction Pumpability, in the context of concrete technology, is a mixture characteristic which describes its ability to be pumped through a pipeline. Pumpability is not an intrinsic concrete feature [1], but it is rather the result of a holistic approach involving the enhancement of concrete composition [2–4], the adaptation of pipeline geometry and pumping gear [5] under permanent monitoring and quality control on site [6]. Optimising the pumpability of concrete is a challenging task; it is always a search for a compromise between enhancing concrete pumpability and preserving the stability of the fresh mixture [7]. A complete blockage of the pumping line could be the ultimate consequence of unbalanced concrete rheological properties changing in time, even if the maximum capacity of the pump has not yet been reached [8]. An efficient pumping can be achieved if sufficient lubricating layer (LL) is generated at the pipe wall-concrete interface [9]. At the same time, the force transfer from pump to concrete shall occur through hydrodynamic interaction and not due to the frictional interaction among aggregate particles [10]. The hydrodynamic interaction can be achieved when in any cross-section along the pipeline the aggregates are fully covered with lubricating material (paste). Accordingly, the fresh concrete microstructure must be dense enough to prevent an
⁎
excessive water escape. As a result, a hydrodynamic stress transfer can be assured which is far less dissipative than the stress transfer though friction [11]. The hydraulic pressure gradient within concrete during pumping triggers the formation of the lubricating layer mainly due to flow-induced particle migration [12,13], but also due to the separation of water from the fines paste [14]. The water separation is facilitated by the fines of concrete matrix which act as a filter. A series of rheological tools to predict concrete pumpability, e.g. rheometers, have been applied and represent the background of the present research [1,15–17]. However, these tools can only partially reproduce the formation of lubricating material under pressure during pumping. In pursuing this challenge, a so-called portable high-pressure filter press (PHPFP) is applied to quantify the amount of filtrate that can be pressed out of concrete. It is considered that the total amount of filtrate after a given time interval and kinetics of water separation can help to estimate both the proneness of concrete to forming LL and its stability. Thus, the results of high-pressure filtration obtained by means of a hydraulic ram press are used in the article at hand to explain the findings gained from the pumping sequences. The availability of filtrate material in the mixtures under investigation and the influence on the pressure-flow rate relationship (P–Q curve) including curve slope and yintercept are analysed before and after pumping. Furthermore, the
Corresponding author. E-mail address: [email protected] (E. Secrieru).
https://doi.org/10.1016/j.cemconres.2018.03.018 Received 11 October 2017; Received in revised form 15 January 2018; Accepted 26 March 2018 0008-8846/ © 2018 Published by Elsevier Ltd.
Cement and Concrete Research 108 (2018) 129–139
E. Secrieru et al.
Table 1 Compositions of concrete mixtures under investigation. Material
Density [kg/m3]
Dosage [kg/m3] CVC mixtures
CEM III/A 42.5 N Fly ash Sand 0/2 Sand/gravel 2/8 Gravel 8/16 Water HRWRAa AEAb w/bc [−] Vol. aggregatesd [−] Vol. pasted [l/m3] a b c d
3075 2200 2650 2650 2650 1000 1040 1050 – – –
HPC mixtures
Grout
Gravel
M2A
M2B
M2C
M5A
M5B
M10A
M10B
SCC
360 – 781 508 526 180 2.88 – 0.50 0.69 300
310 50 781 508 526 180 2.88 – 0.50 0.69 306
310 50 781 508 526 180 2.50 – 0.50 0.69 306
310 50 781 508 526 180 3.60 – 0.50 0.69 307
360 – 735 479 496 180 2.88 1.80 0.50 0.65 302
360 – 735 479 496 180 2.52 1.80 0.50 0.65 301
360 120 724 471 488 180 5.04 – 0.38 0.64 356
360 120 724 471 488 180 4.08 – 0.38 0.64 356
360 220 667 434 450 180 6.06 – 0.31 0.59 402
400 280 1060 – – 340 – – 0.50 0.40 597
High-range water-reducing admixture as aqueous solution with 19 % content of active agent. Air-entraining agent. Water-to-binder ratio. Related to concrete unit volume.
capacity was only 1.5 m3. Each batch was immediately unloaded into a ready-mix truck for homogenisation and transport to the site of the pumping experiments. Rheological testing was performed inside a fully equipped warehouse for each concrete truckload as well as on the pumped concrete released from the duct. The recorded climatic conditions on site (including the warehouse) during experiments were: temperature of (20.5 ± 3.3) °C and a relative humidity of (69.2 ± 3.7) %. Each test was completed within a timespan corresponding to a maximum concrete age of 120 min. During testing, the mixtures remained visually stable, thus fulfilling the basic requirement of practical relevance.
changes in fresh concrete properties during pumping are experimentally captured and discussed in detail. As conclusion, two relatively simple tests for the estimating concrete pumpability and stability on site are proposed. 2. Experimental investigation 2.1. Materials and mixtures In total, nine concrete mixtures are investigated in the present paper. The concrete mixtures were purposefully developed for full-scale pumping experiments together with the involved industry partners. The mixtures were chosen according to their consistency and rheological properties, yield stress and plastic viscosity, so that various flow behaviours could be obtained: predominantly plug flow for conventional vibrated concretes (CVC) [18] and shear flow for self-compacting concretes (SCC) [19]. The compositions of the mixtures under investigation are given in Table 1. The mixtures Gravel (reference mixture) and M2A-M5B represent conventional vibrated concretes; M10 and M11 are flowable high-performance concretes (HPC) with increased paste content in comparison to ordinary concrete. Finally, SCC is a self-compacting mixture with high contents of paste and high-range water-reducing admixture (HRWRA). The physical and chemical characteristics of the cement and fly ash are given in Table 2. The parametric study included a variation of water-to binder ratio (w/b) and paste content. Mixtures M2A, M2B, M2C, M10A and M10B were compared in analysing the influence of the amount of HRWRA on concrete stability. Mixtures M5A and M5B were selected to investigate the effect of AEA; the designed air content was 5 %. The total amount of water as well as the sieving curve were kept constant for all the mixtures for the sake of easier comparison. Each concrete mixture was prepared at a ready-mix station in three replicate batches of 1.5 m3. This was necessary because a total volume of 4.5 m3 was required for the pumping experiments, but the mixer
2.2. Pumping circuit The horizontal pumping circuit had a length of 154 m, see Fig. 1. A truck-mounted concrete twin-cylinder hydraulic pump was employed to feed the pipe. The pipeline was made of high-pressure steel pipes of two diameters: DN125 and DN100, i.e., of inner diameter of 125 and 100 mm, joined through a DN125/100 reduction. The circuit ended with a distributor, from which concrete was discharged back into either the pump feed hopper or the disposal container. The data acquisition system was housed in a tent in the immediate vicinity of the pipeline. The pipeline was instrumented with eight pressure transducers of stainless steel for abrasive media (model BROSA type 0310, BROSA AG, Germany) with a measurement range up to 400 bar and a limit pressure of 300 %, Fig. 1c. The flow rate was monitored with an electromagnetic flow meter for pulsating flow (model Transmag 2 Sensor 911/E, Siemens AG, Germany) installed at the end of the pipeline, see Fig. 1d. All sensors were connected to the data acquisition system, thus recording and processing the data in real time by a portable computer. Before introducing the concrete into the pipeline, 0.75 m3 of a preparatory cement grout containing fine aggregates was pumped through to facilitate LL formation and the initial movement of the concrete; Table 1 gives the composition of the grout. Just after pumping
Table 2 Physical and chemical characteristics of the cement and fly ash. Type
CEM III/A 42.5 N Hard coal fly ash
Density [kg/m3]
3075 2200
Blaine fineness [cm2/g]
4780 3000
Chemical composition [%] SiO2
Al2O3
Fe2O3
CaO
MgO
SO3
K2O
Na2O
Cl
22.50 47.5
6.66 26
2.40 11.5
53.20 1.0
4.79 –
3.08 –
0.82 –
0.24 –
0.08 –
130
Cement and Concrete Research 108 (2018) 129–139
E. Secrieru et al.
Flow direction P1
P4 Reducer
DN125
P2
DN125
P3
DN100
P6
DN100
P7
P5 FM
P8
2.50 m
3.00 m
3.00 m
10X3.00 m
a) Concrete plant
Pump Pressure sensors Rubber hoses
Reducer
DN125
Flow meter
DN100
Wooden beams
b)
c)
d)
Fig. 1. a) Schematic representation of the pumping circuit and the location of eight pressure sensors (P1-P8) and flow meter (FM), b) overview of the pumping test setup; c) pressure sensor, d) electromagnetic flow meter.
respectively. In the above functions, τ0i is the yield stress [Pa] and μi is the viscosity parameter [Pa·s/m] of the LL, which can be measured by tribometer; τ0 is the yield stress [Pa] and μ is the plastic viscosity [Pa·s] of the concrete bulk; the parameters τ0 and μ can be measured by viscometer; k is the pipe filling coefficient (k = 1); R is the radius [m] and L is the length [m] of the pipe. The effective pressure predicted by the Sliper device and transposed onto the real pipeline geometry takes the form of Eq. (3):
this volume, the feeding of the pipeline was switched to the actual concrete mixture. The process corresponds to common practice on any construction site in pumping concrete [1,20,21]. Upon release at the end of the pipeline, the primary cement grout together with a small portion of “contaminated” concrete was dumped via the distributor into the container located next to the setup. The rest of the concrete, i.e. the relevant portion for later characterisation, was then pumped in circuit. Four different flow rates were imposed stepwise upwards (increasing) and downwards (decreasing), each step lasting 3 min. The total pumping time amounted to 24 min. The outcomes of each step were average values for flow rate and resulting pressure.
P = f (a, b, Q, L, R)
where L is the length [m] and R is the pipeline radius [m] [22,24]. The yield stress related parameter a [Pa] and viscosity related parameter b [Pa·s/m] are the outcomes of Sliper measurements. Further discussion on data processing with regard to employed rheological devices is available in [22].
2.3. Testing procedure 2.3.1. Rheological measurements Upon delivery of the fresh concrete, a sample was taken and its properties were tested by means of both empirical tests, i.e. slump flow and flow table tests, as well as by means of more the sophisticated rheological testing equipment: ConTec 5 viscometer (ConTec, Iceland), a so-called “tribometer” built on the basis of the viscometer [22] and the Sliding Pipe Rheometer, Sliper (Schleibinger GmbH, Germany); see Fig. 2. The pumpability of fresh concrete depends on the rheological properties of the concrete bulk and, above all, on those of the LL. Based on Kaplan's approach [18], the concrete flow can be of either plug or shear type. The pumping pressure P is described in [18,22,23] for the case of concrete plug flow type as
P = f (τ0i , μi , Q, k , L, R)
2.3.2. Filtrate amount and lubricating layer (LL) thickness The filtrate pressed out of the concrete was investigated using a portable high-pressure filter press (PHPFP) developed at the Institute of Building Materials Science, Leibniz Universität Hannover [14]; the test set-up is shown in Fig. 3a. The filtration cell has an inner diameter D = 150 mm and a filling height H = 300 mm and is used in combination with filter paper having a pore size of 4 to 12 μm. In this way, the rate of filtrate formation can be reproduced in a small-scale test. It is presumed to mimic the radial, friction-induced separation of the liquid phase from bulk concrete while being pumped through a long pipe. Here the flowing particles themselves act as the self-growing filter from which the LL is cast out. When the container is filled with freshly mixed concrete, under applied pressure the solution containing cement particles (filter material) tends to escape; see Fig. 3b. The applied force is increased linearly up to 100 kN (≈ 57 bar) at a rate of 3.33 kN/s.
(1)
and for the shear flow type as
P = f (τ0i , μi , τ0 , μ, Q, k , L, R)
(3)
(2) 131
Cement and Concrete Research 108 (2018) 129–139
E. Secrieru et al.
a)
b)
c)
Fig. 2. Rheological devices: a) viscometer, b) tribometer and c) Sliper with weights (courtesy of Stephan Falk).
Fig. 3c indicates the rate of filtrate formation in time for the investigated mixtures. The curves of the mixtures were compared based on the total filtrate amount. Depending on concrete composition, two main phases of the filter material formation are identified. In the first phase, the rate of filtrate formation is assumed to be nearly constant. In the second phase, the rate gradually falls to zero. For the mixtures under investigation, the maximum filtrate amount is attained within 5 min corresponding to the duration of the first phase. The transition time cut between those two phases was chosen for a filtrate rate formation below 0.3 g/s for 5.3 l of concrete sample subjectively as a compromise between the desire to keep measuring time short (with respect to applicability of the method in the praxis of construction) and the well measurable filtrate amount for various concrete mixtures. This choice resulted in the duration of the first phase of 300 s, or 5 min. The second phase, another 5 min, covers the remaining time to 600 s. The higher the filtrate amount obtained, the better the assumed pumpability. The main problem is the rate of filtrate material formation since intense “bleeding” is a clear indication of an unstable mixture [1]. The authors have therefore used the rate of filtrate formation as a measure of pumpability based on the amount of water with a maximum particle size ≤ 0.012 mm extruded out of the concrete samples. At this stage it should be remarked that if such a highly densified state occurred in the pipeline, the nature of the interactions among the particles would change from hydrodynamic to directly frictional and finally cause a blockage of the pipeline [10,25] even if a highly efficient HRWRA is used as a lubricating agent. Since the procedure is intended for use on site, the testing time was adjusted according to the total pumping time of concrete. On site, depending on pipeline length, the concrete leaves the pipeline within 2 to 15 min [26]. It is therefore meaningful to analyse the formation of filter material within the time range experimentally. Fig. 4 shows a schematic representation of the filtrate extraction process. According to this model, under high pressures corresponding to the pumping pressure (maximum 100 bar), the physically bound water,
i.e., rheologically effective water, is expected to be pressed out of the concrete. On the other side, the chemically bound water and interstitial water are expected to remain in the system, unless very high temperature is applied [27]. Theoretically, the entire filtrate amount is interpreted as being available for LL formation and enhancement. Potentially, less liquid volume could contribute to LL formation, since the maximum applied pressure in the filtration setup is higher than the pumping pressure. The relative amount of paste Vpaste [l] required to build a LL of a specific thickness e can be calculated using Eq. (4):
Vpaste = π [R2‐(R‐e )2] L
(4)
where R [m] and L [m] are pipeline radius and length, respectively. The necessary amount of lubricating material corresponding to a minimum LL thickness e for the full-scale pipeline is shown as an example in Table 3. 3. Results and discussion 3.1. Influence of pumping on fresh concrete properties The results of the rheological tests are summarised in Table 4. These are yield stress and viscosity related parameters as determined using the rheological devices; see Section 2.3.1. The plastic viscosity μ and the yield stress τ0 are those of the concrete bulk. The viscosity parameters μi and b and the yield stress parameters τ0i and a describe the properties of the lubricating layer between the concrete and the pipe wall. In all mixtures the viscosity parameters μ and μi decreased during pumping. The only exceptions were the CVC mixtures containing AEA, i.e., M5A, M5B, M2C, the last with a higher HRWRA content in comparison to M2A and M2B. For these mixtures the viscosity parameters μi and b slightly increased. The formation of filtrate can be expected to reduce the magnitude of the rheological parameters of the lubricating layer as measured with a tribometer or Sliper. The reason for that is additional water which decreases both yield stress and viscosity 132
Cement and Concrete Research 108 (2018) 129–139
E. Secrieru et al.
Press Barometer Pump
Filtration cell (container)
Scale
a)
b)
Filtrate formation rate [g/s]
3.5 Gravel M2A M2B M2C M5A M5B M10A M10B SCC
3.0 2.5 Phase 1 2.0 1.5 1.0
Phase 2 0.5 0.0 0
150
300
450
600
Time t [s]
c) Fig. 3. a) High-pressure filter press for stability tests; b) exemplary extracted aqueous filtrate sample for Gravel mixture; c) filtrate formation in time pressed out of the concrete mixtures.
Fig. 4. Schematic representation of filtrate extraction from concrete sample. 133
Cement and Concrete Research 108 (2018) 129–139
E. Secrieru et al.
average about 46 % at the beginning and 48 % at the end of pumping campaign independent of the mixture type; see Fig. 5. The filling degree represents the ratio between the real volume of aspirated concrete and the volume for a filled piston. Table 6 shows an example of the determination of the filling degree for the reference mixture Gravel based on the measured data and piston geometry. The measured values are significantly lower in comparison to 70 % mentioned in [1] or even 90– 100 % which were assumed in [8,29,33]. This means that in the present study the pistons, by default, mechanically induced air into the mixtures under investigation. The additional air is lost during pumping, especially in steep pipelines where concrete “free fall” on site occurs [8]. However, such “free fall” was not a part of the pipeline setup in the research at hand. That the effect of air content in concrete had much less effect on the viscosity parameters of LL can be probably traced back to high shear gradients in the layer which “expel” air bubbles towards the less sheared concrete bulk.
Table 3 Necessary amount of lubricating material depending on LL thickness. Pipeline section
DN125 DN100
Pipe diameter D [mm]
125 100
Radius R [mm]
62.5 50.0
Pipe volume [m3]
Total length L [m]
79 75
0.969 0.589
Paste volume Vpaste [l] for LL of assumed thickness e e = 1 mm
e = 2 mm
54.10a
107.24a
a
Total amount of paste for two pipeline sections under the assumption that e remains constant independent of pipe diameter.
parameters of the lubricating layer [28]. The increase in the viscosity parameter corresponds to a very pronounced decrease in filtrate amount for the pumped concrete in comparison to the same mixture before pumping; see Section 3.3. The yield stress parameters τ0, τ0i and a increased significantly. In other words, all concretes became stiffer upon being pumped. The sharp increase in yield stress, manifested in the form of a drop in slump flow or spread flow values, has previously been reported [26,29]. One explanation for the behaviour may be a temperature increase in concrete occurring during pumping due to permanent friction at the pipe wallconcrete interface; see Table 5. Three degrees Kelvin increase for CVC mixtures and up to eight degrees for HPC during the 24 min of dwell time of the material in the circuit certainly accelerate the cement hydration, which results in a significant rise of the yield stress as reported in [30,31] compared to isothermal conditions upon the mere time interval. Opposite the increase in yield stress-related parameters upon pumping, the plastic viscosity μ of bulk concrete decreased for all mixtures. The corresponding viscosity parameters μi and b of the lubricating layer decreased for the CVC mixtures Gravel, M2A, M2B and HPC mixtures M10A, M10B and SCC whereas they increased for CVC mixtures M2C, M5A and M5B; see Table 4. From the experience it is known that increase in the air content leads to decrease in the viscosity parameters. Such a relationship was observed for the mixtures Gravel, M2A, M2B, M10A, M10B and SCC for all viscosity related values but, interestingly, are not valid for M2C, M5A and M5B concerning the viscosity of LL; see Table 5. Furthermore, it is worth noting that in the bulk concrete the viscosity parameters changed more pronouncedly than those of LL. Previously, it was suggested that the changes in fluidity cannot be regarded as due only to the changes in the lubricating layer properties [32]. In the case of the viscosity parameter, the main reason for the decrease was likely the incomplete filling of the pump pistons, on
3.2. Influence of composition and rheological properties of concrete on filtrate kinetics Fig. 6a and b show the effect of the specific surface of fine particles (< 0.125 mm) on the filtrate amount pressed out of concrete before and after pumping. The specific surface of fines was determined by multiplying the mass of binder contained in the concrete sample with the corresponding Blaine fineness value; see Table 2. In general, the filtrate amount decreased in inverse proportion to the specific surface area of the fines. The mixtures in the ellipse are CVC; they differed only slightly in the total surface area due to cement replacement by fly ash. By comparison such replacement increased the filtrate amount for the constant water content in the initial system, cf. the CVC mixtures Gravel and M2A. This is not surprising considering different water demand of cement and fly ash particles due to the differences in the particle shape [34], the effect of which cannot be appropriately captured by Blaine measurements. The actual effect of HRWRA on the dispersion degree of fine particles can only be partially considered unless the particle surfaces are fully saturated. In such case the fine particles would decant as sediment [35], and the mixtures would become very unstable, which does not occur frequently. Therefore, the available and actually accessible total surface area of the fines is unknown. Nevertheless, the mixtures belonging to the group CVC have a noticeably lower total surface area of fines (due to the lower fines content) than HPC mixtures. Thus, a general trend can be well seen. For example, in comparison to CVC mixtures significantly lower filtrate amounts were released by the mixtures M10A, M10B and SCC upon applying the pressure in the PHPFP. Here, the filtrate amounts pressed out of CVC mixtures M2A,
Table 4 Rheological parameters of fresh concrete before and after pumping (left and right columns, respectively). Mix.
Yield stress τ0 [Pa] Gravel M2A M2B M2C M5A M5B M10A M10B SCC a b c d
LLa: tribometer
Concrete bulk: viscometer
112c 99 148 133 64 155 114 262 18
Viscosity μ [Pa∙s] 324d 252 356 319 206 329 404 460 332
30c 22 17 20 11 19 36 31 36
Concrete bulkb and LL: Sliper
Yield stress τ0i [Pa] 14d 15 13 12 8 13 15 15 18
48c 20 56 35 26 54 94 65 11
Viscosity μi [Pa∙s/m] 152d 91 153 125 106 124 126 140 108
LL = lubricating layer. Concrete bulk can be partially sheared. Before pumping. After pumping. 134
944c 737 591 530 531 927 1283 1438 2249
Yield stress a [Pa] 710d 522 531 536 578 982 1136 1164 1557
141c 107 200 131 156 213 97 193 21
Viscosity b [Pa∙s/m] 381d 174 299 229 190 242 309 270 265
587c 482 380 216 235 458 689 942 935
305d 367 303 306 272 471 471 630 747
Cement and Concrete Research 108 (2018) 129–139
E. Secrieru et al.
Table 5 Results of slump flow, flow table and other conventional tests on fresh concrete (left and right columns, respectively). Flow table, slump flow⁎ [mm]
Mix. Gravel M2A M2B M2C M5A M5B M10A M10B SCC⁎ ⁎ a b
530 615 565 605 640 490 655 485 673
a
430 495 470 480 530 510 435 430 318
Density [kg/dm3]
Air content [%] b
1.4 1.3 1.2 0.8 3.7 4.9 1.5 2.1 2.0
a
4.2 3.7 2.2 2.6 5.1 5.1 2.7 2.8 2.9
2.37 2.39 2.40 2.39 2.30 2.30 2.38 2.35 2.28
a
Temperature [°C] b
2.27 2.30 2.31 2.31 2.25 2.26 2.31 2.30 2.28
22.5a 20.5 25.5 27.5 23.8 20.0 21.5 23.0 23.0
26.2b 23.3 28.5 32.5 25.7 24.9 27.0 31.0 28.1
Slump flow. Before pumping. After pumping.
100
Filling degree [%]
Gravel mixture with slightly higher HRWRA content. Obviously, the cement replacement by fly ash at a dosage of 10 % could “overcompensate” the decrease in filtrate amount due to lower HRWRA content, due to the lower specific surface of the fly ash and thus higher amount of “free” water; see also the comments above. Regarding the CVC mixtures M5A and M5B, one can remark an increasing effect of adding air-entraining agent on the filtrate amount in comparison to the reference mixture. While showing the highest filtrate amount, the mixture M5A was prone to bleeding but remained pumpable. It seems that, in addition to HRWRA, AEA as well exhibited a fluidising effect that caused some bleeding. Since the mixture barely reached an air content of 3.7 %, it was decided to modify M5A by decreasing the amount of HRWRA and denominate it as M5B. The result was a more stable mixture with an initial air content of 5 %. Analysing the case of the pumped mixtures, for the CVC the filtrate amount decreased in comparison to the same mixtures tested before pumping, while the opposite was the case for the concretes with a high content of fines. When pumped the mixtures M10A, M10B and SCC changed their consistency from “sticky” to rather “stiff”, which “allowed” the filtrate to move within the system. Consequently, the filtrate retention in these concretes decreased in the course of pumping. The filtrate amount depended on concrete plastic viscosity μ and exerted a substantial impact on the resulting viscosity parameter μi; see Fig. 7. Indeed, it is not only the rheological properties but also the amount of LL forming at the concrete/pipe wall interface that determines the magnitude of the viscosity parameter μi. The rheological tests showed that the CVC mixtures Gravel and M2A-M5B with lower viscosity featured a higher filtrate release. Thus, the lubricating layer forming for the mixtures Gravel and M2A-M5B is postulated to be thicker than in the case of the HPC mixtures M10A, M10B and SCC due to the higher filtrate retention of the latter. However, also the properties of LL depend on the extent of the filtrate release.
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
80 60 40 20 0 5
15 25 35 Flow rate Q [m3/h]
45
Fig. 5. Variation of filling degree depending on the concrete type and flow. Table 6 Example of calculation the piston filling degree for the reference mixture Gravel. Number of strokes [−]
13 22 28 39 29 22 9 24
b
Piston geometry Diameter [mm]
Length [mm]
Flow rate Q [m3/ h]
230 230 230 230 230 230 230 230
2000 2000 2000 2000 2000 2000 2000 2000
10.52 16.47 18.04 29.67 21.95 14.12 7.93 15.01
Concrete volume [m3] per stroke
Piston volume [m3]
Filling degree k [%]
0.57 0.86 0.94 1.48 1.06 0.74 0.40 0.75
1.08 1.83 2.33 3.24 2.41 1.83 0.75 1.99
52.8 47.2 40.2 45.8 44.0 40.4 53.0 37.6
3.3. Influence of filtrate amount on pumpability The pumpability of fresh concrete depends on the rheological properties of concrete bulk and forming LL [17,18,22]. In practice it is best described as the relationship between pumping pressure and flow rate in a P–Q curve with A as y-intercept and B as curve slope; see Fig. 8 and Table 4. The parameter A is attributed to the initial pressure push necessary to initiate concrete flow in the given pipeline, similar to the yield stress parameters τ0 and τ0i. In the previous studies, it was shown that the linear correlation between the yield stress parameter of the lubricating layer τ0i and the parameter A is not always pronounced or straightforward [22,37], in the present investigation the coefficient of determination R2 reaches a value 0.50. The parameter B determines the concrete resistance against movement in the pipeline as a function of moving speed and depends on the
M2B and M2C were dominated by the HRWRA dosage. Among them, M2C containing fly ash, with the highest HRWRA dosage, “generated” the highest filtrate amount. On the other end M2B yielded the lowest filtrate amount, corresponding to the lowest HRWRA dosage. Obviously, by increasing the dosage of HRWRA at a constant water amount in the mixture, the percentage of the “rheologically effective” water also increases. At the same time, the highly dispersive effect of the HRWRA also increases the “active” surface area of fine particles in the mixture and, subsequently, the amount of chemically bound water [36]. However, the amount of water released from agglomerates is greater than the amount of chemically bound water on the surface of the particles. Even so, the amount of filtrate in M2B is higher than in the 135
Cement and Concrete Research 108 (2018) 129–139
E. Secrieru et al.
80
60 50 40 30 20
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
70 Filtrate amount [l/m3]
70 Filtrate amount [l/m3]
80
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
R² = 0.64
60 50 40 30
R² = 0.80
20 10
10
0
0
0
0 1 2 3 Sp. surface of fines Afines·109 [cm2/m3]
a)
1 2 3 Sp. surface of fines Afines·109 [cm2/m3]
b)
80
80
70
70
60
60
Filtrate amount [l/m3]
Filtrate amount [l/m3]
Fig. 6. Effect of the specific surface of fine particles (< 0.125 mm) in concrete on the filtrate amount a) before and b) after pumping.
50 Gravel M2A M2B M2C M5A M5B M10A M10B SCC
40 30 20 10 0 0
R² = 0.91
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
40 30 20 10
R² = 0.63
0
10 20 Plastic viscosity
30 [Pa·s]
40
0
a)
10 20 Plastic viscosity
30 [Pa·s]
40
b) 80
80
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
60 50 40 30 R² = 0.71
20
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
70 Filtrate amount [l/m3]
70 Filtrate amount [l/m3]
50
10
60 50 40 30 R² = 0.75
20 10
0
0 0
1000 2000 3000 Viscosity parameter i [Pa·s/m]
0
c)
1000 2000 3000 Viscosity parameter i [Pa·s/m]
d)
Fig. 7. Correlation between the filtrate amount and plastic viscosity μ for concretes a) before and b) after pumping as well as between the filtrate amount and viscosity parameter μi for concretes c) before and d) after pumping. 136
Cement and Concrete Research 108 (2018) 129–139
E. Secrieru et al.
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
40 30
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
50
Eff. pressure P [bar]
Eff. pressure P [bar]
50
20 10
40 30 20 10
0
0 0
10 20 30 Flow rate Q [m3/h]
40
50
a)
0
10
20 30 40 3 Flow rate Q [m /h]
50
b)
Fig. 8. P–Q curves obtained from the full-scale experiments on concrete mixtures a) at the beginning and b) at the end of pumping. Table 7 Parameters A (y-intercept) and B (slope) describing P–Q curve from full-scale pumping experiments and filtrate amount measured before and after pumping (left and right columns, respectively). Mix.
ΔA
y-Intercept A [kPa]
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
70b 68 234 258 236 313 247 324 −253a
868c 539 699 630 454 712 1233 1171 1411
ΔB
Slope B 3
[%]
[kPa∙h/m ]
+1137 +687 +199 +145 +93 +128 +399 +261 –
85b 74 64 56 44 76 146 161 163
60c 58 52 46 38 63 111 128 105
Filtrate amount
Change
3
[%]
[l/m ]
−29 −23 −19 −18 −13 −17 −24 −20 −35
45.3d 57.4 51.4 66.1 74.5 60.4 39.1 40.8 29.5
[%] 49.7e 56.1 51.4 56.0 63.3 55.2 39.6 41.1 33.1
+9.6 −2.3 −0.1 −15.3 −15.0 −8.6 +1.2 +0.6 +12.2
a The negative value is the result of a linear course of P–Q assumed by default; actually, SCC rather exhibits nonlinear P–Q behaviour due to shear thickening at elevated flow rates [38]. b Pumping at increasing flow rate. c Pumping at decreasing flow rate. d Before pumping. e After pumping.
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
300
200
100
300
200
100 Phase 1
Phase 1
Phase 2
0
Phase 2
0 0
a)
Gravel M2A M2B M2C M5A M5B M10A M10B SCC
400
Filtrate amount [g]
Filtrate amount [g]
400
150 300 450 600 750 900 Time t [s]
0
150 300 450 600 750 900 Time t [s]
b) Fig. 9. Filtrate formation in time measured on concretes a) before and b) after pumping. 137
Cement and Concrete Research 108 (2018) 129–139
E. Secrieru et al.
4. Conclusions
Table 8 Lubricating layer thickness e [mm] corresponding to the available filtrate amount before and after pumping. e [mm]
ebefore eafter
The paper at hand has dealt with the influence of pumping on rheological properties of fresh concrete behaviour. For this purpose, full-scale pumping experiments were conducted on ready-mix concrete accompanied modern rheological tests before and after pumping. The focus was set on the formation of lubrication layer and on the parameters, which determine its properties. In this context a portative highpressure filter press was used to estimate the stability of the mixtures under conditions of pumping. The main outcomes of the study are:
Mixture Gravel
M2A
M2B
M2C
M5A
M5B
M10A
M10B
SCC
1.31 1.43
1.65 1.61
1.48 1.48
1.91 1.61
2.15 1.82
1.74 1.59
1.13 1.14
1.18 1.18
0.85 0.95
viscosity parameters μ and μi. An obvious correlation was observed between the viscosity parameters, especially μi and B, with R2 = 0.80; see Tables 4 and 7. Depending on concrete composition, the parameter B tendentially exhibits elevated values for the mixtures with increased binder (fines) content, i.e., M10A, M10B and SCC, and well correlates with the viscosity parameter μi, the latter mostly determining concrete pumpability [22]; see Table 7. As a result of pumping the magnitude of the A parameter is altered noticeably more than that of the parameter B. The measured change is a clear indication that during pumping concrete loses its consistency in terms of yield stress and slump flow/flow table spread, which necessitates the higher initial pumping pressure needed to start moving concrete in the pipeline. With regard to the filtrate amounts, mixtures M5A, M5B and M2C with their “highest” or “easiest” pumpability also generated the largest amount of expressible suspension of fines, i.e., lubricating material. At the other extreme were the HPC mixtures with considerably elevated fines content, i.e., M10A, M10B and SCC, which required more liquid to cover their particles' surfaces. The difference in filtrate amount measured for M2A and M2B had only a negligible effect on the slope of the P–Q curve. The same remark is also valid for the mixtures M10A and M10B: In both cases, the HRWRA amount was varied. It should be noted that based on the consideration of filtrate amounts it is only possible to qualitatively predict the pumpability of concretes with a similar composition. In the next step, the rate of filtrate formation was analysed; see Fig. 9. Depending on the concrete composition, the extruded water facilitates the formation and enhancement of the LL with respect to the available lubricating material. Since the filtrate material has an aqueous character, it must exert a dilutive effect on the composition of the LL and, thus, lead to decrease in its rheological parameters, i.e., yield stress and plastic viscosity. However, an excessively high filtrate formation rate indicates a reduction in stability [14]. M5A showed the highest filtrate formation rate and was prone to bleeding already before pumping. The mixture M5A could successfully be pumped without causing any trouble, the worst of which would be blockage upon an excessive loss of water due to bleeding. The other extremes with respect to both filtrate formation rate and the total amount of filtrate, i.e., HPC mixtures M10A, M10B and SCC, required the greatest effort in terms of effective pumping pressure. The HPC mixtures with their high fines contents, however, generated higher filtrate amounts after the pumping, while the filtrate formation rates remained low in Phase 1. A reason for the changed behaviour in Phase 2 might be the decrease in mixture consistency and increase in air content that facilitated the movement of the filtrate water out of the system. Since the volume of LL material cannot exceed the total amount of the available paste, the maximum physically possible LL thickness was calculated using Eq. (4). Table 8 shows the resulting LL thicknesses according to the available lubricating material. Accordingly, the LL thickness can be quantitatively improved through the filtrate material that is “pressed” towards the wall during pumping with up to 2 mm.
• Pumping increases yield stress and reduces the viscosity of fresh concrete. • An increase in paste volume by increasing the amount of binder • • • • • • • •
(fines), at constant water amount, generally leads to a marked increase in the required pressure for a given flow rate. The total amount of filtrate pressed out of concrete in a given time interval and the filtration rate influence the formation of lubricating layer and subsequently the pumpability of concrete to a great extent. The filtrate material can increase the thickness of the lubricating layer by up to 2 mm. The filtrate amount depends on the total surface area of the fines and on the related concrete viscosity parameters, especially on the plastic viscosity of the concrete bulk. As the surface area of the fines increases, the rheologically effective water and, thus, the amount of filtrate decrease. An increase in the dosage of HRWRA at a constant water content in concrete also increases the rheologically effective amount of water and the amount of filtrate. The degree of filling of pump pistons reaches an average value of 50 %. Such relatively low filling degree is considered responsible for the increase in air content in fresh concrete during pumping. The temperature increase due to pumping is more pronounced for HPC in comparison to CVC. Sliper and the portative high-pressure filter press have proven their applicability for quick assessment of concrete pumpability and stability on site.
Acknowledgements The authors are grateful to the Federal Ministry for Economic Affairs and Energy of the Federal Republic of Germany and to the German Federation of Industrial Research Associations, Collective Research Project grant number 18361 BR/1, for supporting this research through grant number 18361 BR/1 “Zielsichere betontechnische Gestaltung des Pumpens von Frischbeton”. The authors appreciate the important contribution of the German Ready-mixed Concrete Research Association (Forschungsgemeinschaft Transportbeton e.V.) in helping to organise and monitor the full-scale pumping experiments. The authors extend their gratitude to the German Association for Concrete and Construction Technology (Deutscher Beton- und Bautechnik-Verein e.V.) for the financial support in developing the portable high-pressure filter press (PHPFP) [39]. References [1] D. Kaplan, F. De Larrard, T. Sedran, Avoidance of blockages in concrete pumping process, ACI Mater. J. 102 (2005) 183–191. [2] M.S. Choi, Y.J. Kim, K.P. Jang, S.H. Kwon, Effect of the coarse aggregate size on pipe flow of pumped concrete, Constr. Build. Mater. 66 (2014) 723–730. [3] F. Chapdelaine, Fundamental and Practical Study on the Pumping of Concrete, PhD Thesis, Laval University, 2007. [4] T. Neumann, Einflüsse auf die Pumpbarkeit von Beton, Beton, Vol. 5 (2012), pp. 166–171. [5] Placing Concrete by Pumping Methods, ACI 304.2R-96 (Reapproved 2008), (1996). [6] O. Río, Á. Rodríguez, S. Nabulsi, M. Álvarez, Pumping quality control method based on online concrete pumpability assessment, ACI Mater. J. 108 (2011) 423–431. [7] D. Feys, G. De Schutter, K.H. Khayat, R. Verhoeven, Changes in rheology of self-
138
Cement and Concrete Research 108 (2018) 129–139
E. Secrieru et al.
[23] E. Secrieru, J. Khodor, V. Mechtcherine, C. Schröfl, Lubricating Layer and Its Relevance to Concrete Pumpability, (2017) (Submitt. to CBM). [24] V. Mechtcherine, V.N. Nerella, K. Kasten, Testing pumpability of concrete using Sliding Pipe Rheometer, Constr. Build. Mater. 53 (2014) 312–323. [25] D. Feys, Understanding the pumping of conventional vibrated and self-compacting concrete, in: N. Roussel (Ed.), Underst. Rheol. Concr, Woodhead Publishing Limited, Cambridge, 2011, pp. 331–353. [26] J. Aldred, Burj Khalifa – a new high for high-performance concrete, Proc. ICE Civ. Eng. 163 (2010) 66–73. [27] H.F.W. Taylor, Cement Chemistry, 2nd ed., Thomas Telford Publishing, 1997. [28] Ó.H. Wallevik, J.E. Wallevik, Rheology as a tool in concrete science: the use of rheographs and workability boxes, Cem. Concr. Res. 41 (2011) 1279–1288. [29] R. Khatib, Analysis and Prediction of Pumping Characteristics of High-Strength Selfconsolidating Concrete, PhD Thesis, Université de Sherbrooke, 2013. [30] W. Schmidt, H.J.H. Brouwers, H.-C. Kühne, B. Meng, Influences of superplasticizer modification and mixture composition on the performance of self-compacting concrete at varied ambient temperatures, Cem. Concr. Compos. 49 (2014) 111–126. [31] E. Secrieru, V. Mechtcherine, C. Schröfl, D. Borin, Rheological characterisation and prediction of pumpability of Strain-Hardening Cement-Based-Composites (SHCC) at various temperatures, Constr. Build. Mater. 112 (2016) 581–594. [32] J.H. Kim, S.H. Kwon, S. Kawashima, H.J. Yim, Rheology of cement paste under high pressure, Cem. Concr. Compos. 77 (2017) 60–67. [33] K.A. Riding, J. Vosahlik, D. Feys, T. Malone, W. Lindquist, Best Practices for Concrete Pumping, (2016). [34] P.-C. Aïtcin, S. Mindess, Modern binders, Sustain. Concr, CRC Press, 2011, pp. 87–90. [35] R.J. Flatt, I. Schober, Superplasticizers and rheology of concrete, in: N. Roussel (Ed.), Underst. Rheol. Concr, 1st ed., Woodhead Publishing Limited, 2012, pp. 144–208. [36] E. Sakai, T. Kasuga, T. Sugiyama, K. Asaga, M. Daimon, Influence of superplasticizers on the hydration of cement and the pore structure of hardened cement, Cem. Concr. Res. 36 (2006) 2049–2053. [37] E. Secrieru, M. Butler, V. Mechtcherine, Prüfen der Pumpbarkeit von Beton – Vom Labor in die Praxis, Bautechnik (11) (2014) 797–811. [38] D. Feys, R. Verhoeven, G. De Schutter, Why is fresh self-compacting concrete shear thickening? Cem. Concr. Res. 39 (2009) 510–523. [39] L. Lohaus, D. Cotardo, DBV 308 – Pumpstabilität von Beton. Prüfverfahren zur Mischungsstabilität bei hohem Druck, Schlussbericht, (2017).
consolidating concrete induced by pumping, Mater. Struct. 49 (2016) 4657–4677. [8] D. Feys, Interactions Between Rheological Properties and Pumping of Self-compacting Concrete, PhD Thesis, Gent University, 2009. [9] M. Choi, N. Roussel, Y. Kim, J. Kim, Lubrication layer properties during concrete pumping, Cem. Concr. Res. 45 (2013) 69–78. [10] R. Browne, P. Bamforth, Tests to establish concrete pumpability, ACI J. 74 (1977) 193–203. [11] J. Yammine, M. Chaouche, M. Guerinet, M. Moranville, N. Roussel, From ordinary rheology concrete to self compacting concrete: a transition between frictional and hydrodynamic interactions, Cem. Concr. Res. 38 (2008) 890–896. [12] J. Spangenberg, N. Roussel, J.H. Hattel, H. Stang, J. Skocek, M.R. Geiker, Flow induced particle migration in fresh concrete: theoretical frame, numerical simulations and experimental results on model fluids, Cem. Concr. Res. 42 (2012) 633–641. [13] M. Choi, Y.J. Kim, S.H. Kwon, Prediction on pipe flow of pumped concrete based on shear-induced particle migration, Cem. Concr. Res. 52 (2013) 216–224. [14] Y.A. Abebe, Flowable and Stable Concrete. Design, Characterisation and Performance Evaluation, PhD Thesis, Leibniz Universität Hannover, 2017. [15] E. Secrieru, Pumping behaviour of modern concretes - Characterisation and prediction, PhD Thesis Technische Unversität Dresden, 2018. [16] Y.A. Abebe, L. Lohaus, Rheological characterization of the structural breakdown process to analyze the stability of flowable mortars under vibration, Constr. Build. Mater. 131 (2017) 515–525. [17] D. Feys, K.H. Khayat, A. Perezschell, R. Khatib, Development of a tribometer to characterize lubrication layer properties of self-consolidating concrete, Cem. Concr. Compos. 54 (2014) 40–52. [18] D. Kaplan, F. de Larrard, T. Sedran, Design of concrete pumping circuit, ACI Mater. J. 102 (2005) 110–117. [19] D. Feys, G. De Schutter, R. Verhoeven, Parameters influencing pressure during pumping of self-compacting concrete, Mater. Struct. 46 (2013) 533–555. [20] C. Hazaree, V. Mahadevan, Single stage concrete pumping through 2.432 km (1.51 miles): weather and execution challenges, Constr. Mater. 3 (2015) 56–69 (Case Study). [21] H. Imaizumi, H. Jitousono, T. Kemi, K. Fujii, Study on the blockage in concrete pumping with preceding mortars, Cement Sci. Concr. Technol. 53 (1999) 732–737. [22] E. Secrieru, S. Fataei, C. Schröfl, V. Mechtcherine, Study on concrete pumpability combining different laboratory tools and linkage to rheology, Constr. Build. Mater. 144 (2017) 451–461.
139