Numerical simulation of formwork pressure while pumping self-compacting concrete bottom-up

Engineering Structures 70 (2014) 218–233

Contents lists available at ScienceDirect

Engineering Structures journal homepage: www.elsevier.com/locate/engstruct

Review article

Numerical simulation of formwork pressure while pumping self-compacting concrete bottom-up Serge Tichko a,⇑, Jens Van De Maele b, Niels Vanmassenhove b, Geert De Schutter a, Jan Vierendeels c, Ronny Verhoeven d, Peter Troch e a

Magnel Laboratory for Concrete Research, Ghent University, Belgium Ghent University, Sint-Pietersnieuwstraat 25, B-9000 Gent, Belgium Department of Flow, Heat & Combustion Mechanics, Ghent University, Belgium d Laboratory for Hydraulics, Ghent University, Belgium e Department of Civil Engineering, Ghent University, Belgium b c

a r t i c l e

i n f o

Article history: Received 15 April 2013 Revised 3 April 2014 Accepted 4 April 2014 Available online 8 May 2014 Keywords: CFD Numerical modelling Formwork pressure Self-compacting concrete Base filling

a b s t r a c t Self-compacting concrete (SCC) enables new casting techniques, filling formworks by pumping bottomup. However, fundamental questions remain concerning the formwork pressure when following this advanced filling procedure. In order to determine the maximum formwork pressures, a series of formwork filling tests, with SCC being pumped from the base of the formwork, have been performed at the Magnel Laboratory for Concrete Research of the Ghent University. Numerical simulations of these formwork filling tests have also been performed for comparison with the experiments. During the filling process, the formwork pressures were measured close to the base of the formworks, where the maximum pressures were expected to occur. The measured formwork pressures were finally compared with the computed formwork pressures. Both the experiments and the simulations in this study revealed that the formwork pressures during the filling tests were slightly higher than hydrostatic for SCC pumped from the base of the formworks. This was due to the additional occurring hydraulic losses. Ó 2014 Elsevier Ltd. All rights reserved.

Contents Research significance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1. Rheological behaviour of self-compacting concrete (SCC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2. Formwork pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2.1. Top filling process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2.2. Disadvantages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2.3. Base filling process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3. Numerical sensitivity analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Full-scale formwork filling tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Numerical simulation of the formwork filling tests. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. General overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Model equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Computational domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Boundary conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5. Solution of the model equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6. Formwork filling simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Comparison of the results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Comparison of the results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.

⇑ Corresponding author. Tel.: +32 (0)476 29 26 08; fax: +32 (0)15 71 82 79. E-mail address: [email protected] (S. Tichko). http://dx.doi.org/10.1016/j.engstruct.2014.04.008 0141-0296/Ó 2014 Elsevier Ltd. All rights reserved.

219 219 219 220 220 220 220 222 222 224 224 224 225 225 226 226 227 227

S. Tichko et al. / Engineering Structures 70 (2014) 218–233

5.

219

4.2. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230 Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232

Research significance Most of the available codes and guidelines for determining the formwork pressure have been developed for casting processes with normal vibrated concrete (NVC) [9–13]. Although the DIN 18218 standard [10] gives some design guidelines for use with SCC, these codes and guidelines are generally not suited for casting with SCC when pumped from the base at high casting rates (>7 m/h) [8,15,16]. NVC is traditionally cast from the top of the formwork in several layers, which are individually vibrated in order to remove the entrapped air as much as possible and to ensure good compaction around the steel rebars. As such, the casting rates are rather low. The base filling technique with SCC, which is presented in this article, allows for much faster casting rates with still good compaction. Although the formwork pressures are higher with base filling compared to top filling, the filling times can be noticeably reduced. For the precast industry, this could mean a more cost effective manufacturing process at a higher production rate. 1. Introduction 1.1. Rheological behaviour of self-compacting concrete (SCC) SCC has been developed in Japan during the 1980s. At that time, the Japanese construction industry encountered many problems due to a lack of skilled and qualified workmen, which slowed down the construction pace and impaired the durability of new concrete structures. During the 1990s, SCC gradually made its entrance into Europe through the Netherlands and the Scandinavian countries, and since then, the amount of SCC being applied in construction is continuously increasing, together with the number of countries where it is being used [1,2]. According to De Schutter [1], SCC can be defined as a concrete which needs to possess sufficient fluidity in order to be able to fill a formwork completely (filling ability) without the aid of other forces than gravity, even when having to flow through narrow gaps (passing ability), but also showing a sufficient resistance to segregation, during flow and in stationary conditions (stability). In order to achieve sufficient fluidity in SCC, without increasing the water content, super-plasticizers must be applied. Only adding superplasticizers to traditional concrete is not sufficient to create SCC, due to the large amount of coarse aggregates, which can form particle bridges when flowing through a narrow gap, causing blocking.

Therefore in order to fulfil the passing ability condition, the amount of coarse aggregates is reduced. On the other hand, extra amounts of fine materials, like limestone filler, fly ash or silica fume are added in order to increase the stability of SCC [1,2]. Several material models are available for describing the rheological behaviour of fresh concrete, such as the Bingham model [3], the modified Bingham model [2] or the Herschel-Bulkley model [4]. In all these material models a (dynamic) yield stress is defined, which is the minimum value of the applied shear stress needed to maintain flow. For the present study the HerschelBulkley model has been selected, because this model is able to capture the fresh behaviour of a wide variety of SCC mixes. The Herschel-Bulkley model is formulated mathematically in

sHB ¼ s0;HB þ K HB ðc_ ÞnHB

ð1Þ

where: the index HB stands for Herschel-Bulkley, sHB the shear stress in the material [Pa], c_ the shear rate in the material [1/s], s0,HB the yield stress [Pa], KHB the consistency factor[Pa sn] and nHB the consistency index [–]. The Bingham model can be considered as a special case of the Herschel-Bulkley model, for which the consistency index nHB equals one. Using the Herschel-Bulkley model to describe the steady state behaviour of SCC is not so straightforward though. The consistency factor KHB has no physical meaning. The dimension of KHB is Pa sn, meaning that the consistency factor is also dependent on the consistency index nHB. Only when nHB equals one (Bingham), the consistency factor KHB can be regarded as the plastic viscosity lp of the concrete. Furthermore, the apparent viscosity, defined as the ratio between the instantaneous shear stress and shear rate, is becoming infinite when the shear rate approaches zero. This singularity will have to be handled properly in numerical simulations, as will be explained in Section 3.2. Depending on the mix design, SCC in the fresh state can show thixotropic behaviour and shear thickening to various degrees. Thixotropy can be defined as a reversible build-up and breakdown of internal structure, due to flocculation or coagulation of cement particles for which the influence of inter-particle forces is still significant. Shear thickening is an increase in apparent viscosity with increasing shear rate, when no yield stress is present. When the fluid has a yield stress, the apparent viscosity will first decrease when the shear rate increases, and from a certain shear rate value on, the apparent viscosity will increase again when the shear rate further increases (see Fig. 1). The effect of shear thickening

Fig. 1. Typical shear stress vs. shear rate curve (left) or apparent viscosity vs. shear rate curve (right) for cementitious materials [5].

220

S. Tichko et al. / Engineering Structures 70 (2014) 218–233

is captured by the power law term of the Herschel-Bulkley model for nHB > 1, which becomes important at high shear rates. Roussel [6] proposed the following model for describing thixotropy in SCC. He introduced a structure parameter k, the flocculation state of the material, which influences the apparent rheological properties of the material. The model is defined by two equations, one constitutive equation based on the HerschelBulkley model (Eq. (2)) and one equation describing the rate of change of the internal structure (Eq. (3)). In these two equations three parameters, T, m and a, have to be identified through rheological measurements.

s ¼ ð1 þ kÞs0;HB þ K HB ðc_ ÞnHB

ð2Þ

@k 1 ¼  akc @t Tkm

ð3Þ

where: k is the flocculation state of the material [–], T the flocculation characteristic time [s], a the de-flocculation parameter [–] and m the parameter expressing the influence of the flocculation state on the rate of flocculation [–]. The structure parameter k can reach values ranging from 0 (no structure) to infinity (full structure). Typically, the characteristic time of flocculation T (several minutes) is much longer than the characteristic time of de-flocculation (several tens of seconds in the 1–10 s1 shear rate range). The de-flocculation parameter a is for SCC of the order of 0.01 [7]. Furthermore, many experiments conducted by researchers in other research programs have revealed so far that when fresh SCC is left at rest, the yield stress increases linearly with time, leading therefore to the conclusion that m has to be zero [6–8]. 1.2. Formwork pressure As already mentioned in the first section on the research significance, most if not all of the available codes and guidelines for determining the formwork pressure have been developed for casting processes with normal vibrated concrete (NVC) [9–13]. Although the DIN 18218 standard [10] gives some design guidelines for use with SCC, these codes and guidelines are generally not suited for casting with SCC when pumped from the base at high casting rates (>7 m/h) [8,15,16]. NVC is traditionally cast from the top of the formwork in several layers, which are individually vibrated in order to remove the entrapped air as much as possible and to ensure good compaction around the steel rebars. After some time during the top filling process, the first cast layers at the bottom of the formwork are left at rest. Due to the much higher stiffness of NVC in its fresh state compared to SCC, these lower layers are able to bear the weight of the upper concrete layers without increasing the lateral pressure. As already mentioned in the introduction, SCC is much more fluid than NVC, leading to higher formwork pressures. These formwork pressures are certainly much higher when casting at high casting rates, or when the formwork filling technique being used, does not allow thixotropy to perform its beneficial wall pressure reducing effect. Some of the previously mentioned codes and guidelines recommend that unless a method based on appropriate experimental data is available, a formwork should be designed to withstand the full hydrostatic pressure exerted by the ‘‘fluid’’ concrete. This guidance generally limits contractors to short walls or extremely strong and thus expensive formworks. The lack of adequate formwork design rules for use with SCC has been the primary motivation for starting many research programs, which aim at analysing the wall pressure evolution inside formworks during casting with SCC and deriving practical and relevant guidelines from the research results [8,14–21].

Two types of formwork filling processes are encountered when casting with SCC; filling from the top of a formwork and filling by pumping from the base of a formwork. Both filling processes have their advantages and disadvantages [35]. They are briefly summarized here: 1.2.1. Top filling process Advantages:  There is no inlet duct intersecting with the formwork wall. The surface finish quality is not impaired by the presence of an inlet duct.  This filling technique, if properly executed, can profit from the thixotropic property of the selected SCC. The maximum formwork pressure is then lower than the full hydrostatic pressure. 1.2.2. Disadvantages  In some cases the height of the cast elements can be limited due to necessity of long supply ducts. This is most unfortunate because one of the advantages of using SCC is that it is particularly suited for filling high formworks.  Filling from the top of a formwork increases the probability of having more air getting entrapped during the filling process, which negatively influences the concrete strength development and the durability.  Discontinuous filling, due to intermittent deliveries of concrete, leads to weak interfaces between the cast concrete layers. These interfaces, or ‘‘pour lines’’ are only noticed afterwards when the formwork is stripped.  Due to the fact that the concrete is falling down inside the formwork, the risk of segregation increases. Segregation in concrete occurs when the coarse granulates gather at the base of the formwork due to gravity, leading to a non-homogenous concrete mix. 1.2.3. Base filling process Advantages:  The risk of air getting entrapped during the filling process is avoided or limited, thus leading to an improved quality of the cast.  The occurrence of segregation is counteracted due to the continuous upward movement of the concrete. This also improves the final strength of the concrete element.  There is no need for long supply ducts.  Due to the continuous pumping of SCC from the base of the formwork, there are no weak interfaces or ‘‘pour lines’’ in the final cast.  High concrete elements can be cast at high casting rates without any interruption of the filling process, thereby drastically reducing the time needed for casting. Disadvantages:  Base filling requires the use of a shut-off valve, which is mounted at the base of the formwork. This valve prevents the fresh concrete from escaping the formwork, when the supply duct is disconnected after the formwork has been completely filled. It increases the complexity of the formwork.  Obtaining a good surface finish of the cast element near the inlet valve is a challenge. The design of the shut-off valve is crucial, because the moving part of the valve must perfectly seal the inlet opening and be completely coplanar with the formwork walls, hopefully leaving no surface marks when the formwork is stripped.

221

S. Tichko et al. / Engineering Structures 70 (2014) 218–233

 During the filling of the formwork, the SCC remains fluid all the time, thus leading to high formwork wall pressures, especially when casting at high casting rates. These high wall pressures require very strong and stiff formworks, making them more expensive. This fact will in some cases limit the casting height. Many researchers have already studied the top filling process with SCC and the evolution of the formwork wall pressure during this filling process. When the formwork is filled from the top, the freshly poured concrete layers in the upper part of the formwork do not disturb the SCC remaining in the lower part of the formwork during the filling process. Due to thixotropy, the SCC in the lower part of the formwork starts to build up a structure able to withstand pressure from the concrete above, without increasing the horizontal pressure against the formwork walls [6]. For this type of formwork filling, the horizontal formwork pressure for vertical concrete elements can be calculated using Eq. (4) [6,17].

 phor ¼ 1 



s0 H q gH T qSCC gRe SCC

ð4Þ

where: e is the cast wall thickness [m], H the height of the cast element [m], qSCC the density of the SCC [kg/m3], g the gravity acceleration [m/s2], R the casting rate [m/s], T the flocculation characteristic time [s] and s0 the yield stress [Pa]. Unlike casting with NVC, the SCC inside the formwork may certainly not be vibrated after each layer is cast. Any vibration will destroy the flocculation state and will lead to a drastic (and possibly catastrophic) increase of the formwork pressure. Very little research has been conducted on the base filling process with SCC so far. The objective of the present study is to determine the relevant casting parameters influencing the evolution of the formwork wall pressure during the base filling process with SCC. In a first stage of the research program, the base filling process of columns and walls without rebars is examined. The aim is to first gain a thorough understanding of very simple (and therefore maybe somewhat unrealistic) base filling processes with SCC, before studying the more complex ones with steel reinforcement being present. As already mentioned by Omran [20], so many factors do influence the development of the formwork wall pressure during casting. Only the filling rate, the placement method and the dimensions of a cast will be investigated in this study. When the SCC is pumped from the base of the formwork, the positive influence of thixotropy on the formwork pressure will be very limited, certainly in the vicinity of the inlet(s). In this filling

configuration, the formwork is simply an extension of the supply duct, through which SCC is flowing. Therefore a large part of the SCC inside the formwork is constantly in motion, prohibiting the concrete to develop some structure able to bear its own weight without increasing the lateral pressure exerted on the formwork walls. Thus, with base filling, almost no reduction of the formwork pressure due to thixotropy can be expected, certainly not when filling formworks at high casting rates. This will be further demonstrated in Section 4.1. During the base filling process, the formwork wall pressure will not only depend on the hydrostatic pressure, but also on the hydraulic losses DpL. In turn, these hydraulic losses depend on the filling velocity (v2), the Herschel-Bulkley model parameters (the yield stress, the consistency factor and the consistency index) [2,21] and also on the existence of local disturbances (flow bifurcations due to bends, sudden expansion and/or contraction of the formwork cross section, the presence of rebars). To clarify this, consider a first point just above the inlet of a column, where the SCC enters the formwork at the base, and a second point at the SCC–air interface. This configuration is shown in Fig. 2. The Bernoulli equation then states that:

p1 þ qSCC

v 21 2

þ qSCC gh1 ¼ p2 þ qSCC

v 22 2

þ qSCC gh2 þ DpL

ð5Þ

As the cross section of the column does not change over the entire height, so does the average filling velocity also remain constant, and therefore v1 equals v2. The resulting pressure at the SCC– air interface is equal to the atmospheric pressure. Referencing the static pressures p1 and p2 in Eq. (5) to the atmospheric pressure, and therefore only considering relative static pressures, p2 equals zero. Eq. (5) then becomes:

phor ¼ p1 ¼ qSCC gðh2  h1 Þ þ DpL

ð6Þ

This clearly shows the additional contribution of the hydraulic losses to the horizontal wall pressure, when the SCC is pumped in the formwork from the base. During the base filling process, when the concrete surface inside the formwork has already passed the inlet and is moving towards the top, in the lower part of the formwork just under the inlet, the SCC is almost at rest. Therefore, the maximum wall pressure at the base of the formwork can be estimated with Eq. (7). The results obtained from the numerical simulations performed in this study prove the validity of this equation. This will be illustrated further in Section 4.1.

pmax;base ¼ p1 þ qSCC gh1 ¼ qSCC gh2 þ DpL

Cross section 2, p2, v2, h2

Cross section 1, p1, v1, h1 Inlet of the formwork

Fig. 2. Flow configuration inside a formwork – definitions.

ð7Þ

222

S. Tichko et al. / Engineering Structures 70 (2014) 218–233

Wall pressure [105 Pa]

0.6 τ0 = 10.3 Pa τ0 = 30 Pa τ0 = 50 Pa Hydrostatic

0.5 0.4 0.3 0.2 0.1 0.0 0

1

2

3

4

5

6

7

8

9

10 11 12 13 14 15 16

Time during the filling process [s] Fig. 3. Influence of the yield stress value on the maximum formwork pressure (qSCC = 2314.4 kg/m3, KHB = 10 Pa sn, nHB = 1.35). Fig. 6. Layout of the formwork filling tests.

0.6

Wall pressure [105 Pa]

n

0.5

KHB = 10 Pa·s KHB = 17.7 Pa·sn

0.4

KHB = 25 Pa·sn Hydrostatic

performed with four different consistency indices for SCC, ranging from 1 to 1.5. For the value ranges of the Herschel-Bulkley model parameters used in this sensitivity analysis, the influence of the yield stress value is rather limited, whereas the influence of the consistency factor and the consistency index is more pronounced. It is therefore important to accurately assess the rheology of the SCC through measurement [2,8,24–26,37].

0.3 0.2 0.1 0.0

2. Full-scale formwork filling tests 0

1

2

3

4

5

6

7

8

9

10 11 12 13 14 15 16

Time during the filling process [s] Fig. 4. Influence of the consistency factor on the maximum formwork pressure (qSCC = 2314.4 kg/m3, s0 = 10.3 Pa, nHB = 1.35).

Wall pressure [105 Pa]

0.6

n=1 n = 1.2 n = 1.35 n = 1.4 Hydrostatic

0.5 0.4 0.3 0.2 0.1 0

0

1

2

3

4

5

6

7

8

9

10 11 12 13 14 15 16

Time during the filling process [s] Fig. 5. Influence of the consistency index on the maximum formwork pressure (qSCC = 2314.4 kg/m3, s0 = 10.3 Pa, KHB = 10 Pa sn).

1.3. Numerical sensitivity analysis In the present study, numerical simulations have been performed in order to quantify the influence of the Herschel-Bulkley material model parameters on the occurring hydraulic losses [21]. The details of these numerical simulations will be explained later in this article, but the results of the sensitivity analysis are already presented here, in order to clearly demonstrate the contribution to the hydraulic losses and to the formwork wall pressure. Fig. 3 shows the results of the simulations performed with three different SCC yield stress values, ranging from 10 Pa to 50 Pa. Fig. 4 presents the results of the simulations performed with three different consistency factors for SCC, ranging from 10 Pa sn to 25 Pa sn, while Fig. 5 shows the results of the simulations

This part of the article describes the four full-scale formwork filling tests that have been performed at the Magnel Laboratory for Concrete Research of the Ghent University in Belgium [21]. During these tests, two columns and two walls without steel reinforcements were cast. The full layout of the formwork filling tests is shown in Fig. 6. The dimensions of the two columns are identical. The columns have a height of 2 m, a depth of 21 cm and a width of 17.4 cm. The SCC is pumped from the base of the formwork. The two walls have also identical dimensions (a height of 2 m, a length of 4 m and a thickness of 21 cm) and the SCC is also pumped from the base of the formwork, but the positions of the SCC inlets are different for the two walls. A Schwing concrete piston pump of type P 2023 is used in all the filling tests for pumping the SCC inside the formworks. For these base filling processes, a mechanical shut-off valve has been designed at the Laboratory for Hydraulics and the Magnel Laboratory of the Ghent University. The formwork configurations are shown in Fig. 8. Due to handling requirements (connection and disconnection of the inlet duct, closing of the shut-off valve), the inlet of these formworks needed to be positioned at a certain limited height above the base. The position of the inlet above floor level (centre position) is 0.27 m for column type A, 0.285 m for column type B, 0.275 m for wall type A and 0.175 m for wall type B. While the formworks were filled, the SCC discharge rate was measured at the concrete pump. The column formworks were filled at a rate of 5 l/s (casting rate = 490 m/h), while the formworks of the walls were filled at a rate of 6 l/s (casting rate = 25 m/h). The time of each filling process was recorded and the progression was monitored with several cameras. The formwork pressure was also measured at several positions at the base of the formwork, where the maximum pressure on the formwork walls was expected. Two types of pressure measurements were used; manometers and electronic pressure transducers. The manometers were fixed on an intermediate water chamber, sealed with a rubber membrane, and flush mounted on the formwork. The

223

S. Tichko et al. / Engineering Structures 70 (2014) 218–233

Fig. 7. Pressure measurement system.

(Columns A,B)

(Wall B) (Wall A) Pos. 3

Pos. 4

Pos. 3

Pos. 1

Pos. 1

Pos. 2 Pos. 2

Fig. 8. Connections of the SCC inlet for the walls and for the columns.

Table 1 Mounting height of the pressure measurements systems. Formwork

Type

Location

Height (m)

Column A

Manometer + water chamber Manometer + water chamber Electronic pressure sensor Manometer + water chamber Manometer + water chamber Electronic pressure sensor Manometer + water chamber Manometer + water chamber Manometer + water chamber

Above the inlet

0.385

Above the inlet

0.385

Above the inlet (position 1) Position 2

0.4

Position 3

0.41

Position 4

0.08

Above the inlet (position 1) Position 2

0.35

Position 3

0.08

Column B Wall A

Wall B

0.41

0.12

electronic pressure transducers were flush mounted on the formworks, without the intermediate water chamber configuration. Fig. 7 shows the very simple design of the pressure measurement system with the intermediate water chamber, as well as the hydraulic testing equipment, which was used for the calibration of these pressure measurement units.

The locations of the pressure measurements (cross hared squares) for the walls and for the columns are shown on Fig. 8. The heights above the base at which these measurements were mounted on the formworks are presented in Table 1. At three different moments during the filling processes, a sample of concrete has been taken in order to investigate the fresh properties, by means of a Tattersall Mk-II rheometer and standard tests on SCC like slump flow, V-funnel, L-box, sieve stability and air content. A ready-mix company supplied the SCC mix used in the filling processes. Table 2 lists the type and amount of ingredients that were used in the concrete mix. Although the SCC mix design required a grading curve with a maximum aggregate size of 14 mm, it is still possible to have some aggregates with a larger size in the delivered SCC. After inspection, a limited number of aggregates with a maximum size of 20 mm have been observed (see also Section 4.2). The water-to-cement ratio W/C is 0.53 and the water-to-powder ratio W/P is 0.32. The standard tests and the rheometer measurements were performed on the chosen SCC mix at three different moments during the filling tests:  When the concrete mixing truck arrived at the Magnel Laboratory, before the start of the first filling test.  After the filling of the first wall formwork (wall type A).  After the filling of the last column formwork (column type B).

224

S. Tichko et al. / Engineering Structures 70 (2014) 218–233

Table 2 SCC mix design per m3. Ingredient

Comments

Quantity

Units

Sand 0/4

Grain size between 0 mm and 4 mm Aggregate size between 2 mm and 7 mm Aggregate size between 7 mm and 14 mm

805

kg/m3

450

kg/m3

265

kg/m3

Portland cement

100

kg/m3

Blast furnace cement - Low Alkaline Limestone filler Viscosity modifying Agent (VMA) Polycarboxylether (PCE) type superplasticiser -

265

kg/m3

235 0.5

kg/m3 kg/m3

7.2

l/m3

193

l/m3

Limestone aggregates 2/7 Limestone aggregates 7/ 14 Cement type C I 52.5 N Cement type C III/ A 42.5 LA Calcitec Rheomatrix Glenium 27 Water

 Modelling of the flow of SCC through steel bars during an L-box test, using the FVM in combination with a porous medium analogy [28].  Modelling of the flow of semi-dilute and dense suspensions (like SCC) through simple rebar configurations, using the Dissipative Particle Dynamics Method (DPDM) [29].  Modelling of casting processes with SCC when filled from the top of the formwork, using the FVM in combination with a mathematical model describing thixotropy [17,18].  Modelling of the particle migration during casting processes, using the FVM in combination with multiphase modelling techniques [30]. The following sections describe in detail the numerical models, which have been developed for simulating the base filling tests performed in the present study. 3.2. Model equations

The following results were obtained from the standard concrete tests: see Table 3. The results from the standard tests indicate that the delivered SCC mix was very fluid. The rheological measurements, performed with a Tattersall Mk-II rheometer, also confirm this observation. The SCC from sample no. 2 has been poured from the top of wall type A. This second sample consisted mainly of cement paste with a small amount of aggregates, indicating the occurrence of segregation during the filling process. The measurement results of the first and the last sample revealed that the rheology of the delivered SCC remained constant during the filling tests. The Herschel-Bulkley model parameters, measured with the Tattersall Mk-II rheometer, will be presented in Section 3.6.

The most general description of a fluid flow is obtained from the conservation laws, more commonly known as the Navier–Stokes equations. They represent mass conservation, conservation of momentum and conservation of energy [23,32]. The flow conditions observed during our filling tests are laminar. Because the density and the viscosity of the SCC are assumed to be independent of the typically very small temperature changes, the conservation of energy is not considered in the numerical simulations. Therefore for unsteady, isothermal, laminar flows these equations take on the following form: Conservation of mass: ! @q þ rðq v Þ ¼ 0 @t

ð8Þ

Conservation of momentum: 3. Numerical simulation of the formwork filling tests 3.1. General overview In order to understand more clearly the flow behaviour of SCC in its fresh state during the casting process, many research programs have been initiated during the past years. While reading through the substantial amount of technical reports and articles produced during these research programs, two things can be observed: the great variety of numerical modelling techniques being used, and the need for experimental validation. A small overview of the numerical discretization methods encountered is presented in the following list. This list is certainly not exhaustive:  Modelling of the flow of fresh concrete in a viscometer, using the Finite Difference Method (FDM) [24].  Modelling of the flow of SCC in rheometers, using the Finite Element Method (FEM) and the Finite Volume Method (FVM) [25,26].  Modelling of the flow of SCC during a slump flow test, relying on the Distinct (or Discrete) Element Method (DEM) [27,36].

! @ðq~ vÞ þ rðq~ v ~ v Þ ¼ rp þ rs þ q g @t

ð9Þ

where: q is the density or specific mass of the fluid (kg/m3), ~ v the  is stress tensor (Pa), ~ velocity vector (m/s), s g the gravity acceleration vector (m/s2) and  the tensor product of two vectors. Eqs. (8) and (9) are solved for the pressures and the velocities over the entire computational domain for each time step with the aid of the Ansys Fluent v.6.3.26 and v.12.0 software. The stress tensor s for the SCC is modelled using the Herschel-Bulkley material model (see Eq. (1)), as already mentioned in Section 1.1. The Herschel-Bulkley material model is discontinuous for shear rates close to zero. For zero shear rate, the apparent viscosity becomes infinite. This cannot be handled properly in a numerical code. In order to avoid this mathematical singularity, in Ansys Fluent, a modified Herschel-Bulkley model has been implemented [32]. The Herschel-Bulkley model has been transformed into a two-zone model: a first zone for shear rates lower than a critical shear rate (=s0/l0) where the ‘‘rigid’’ material behaves like a very viscous fluid with viscosity l0, and a second zone for shear rates higher than the critical shear rate where the material follows the original

Table 3 Results from the standard tests. SCC sample no. 1 Density (kg/m3) Sieve stability (%) L-box ratio (–) Slump flow (mm) V-funnel time (s) Air content (%)

SCC sample no. 2 2315 12.8 0.97 820 4.9 3.5

Density (kg/m3) Sieve stability (%) L-box ratio (–) Slump flow (mm) V-funnel time (s) Air content (%)

SCC sample no. 3 2329 18.3 1.00 935 3.3 4.2

Density (kg/m3) Sieve stability (%) L-box ratio (–) Slump flow (mm) V-funnel time (s) Air content (%)

2298 17.6 0.94 830 4.6 3.5

225

S. Tichko et al. / Engineering Structures 70 (2014) 218–233

Herschel-Bulkley model. The equations for this two-zone HerschelBulkley model are:

 nHB 1

For c_ > c_ c :

s c_ g ¼ 0;HB þ K HB c_ c_ c

For c_ 6 c_ c :

g ¼ s0;HB

ð10Þ

  ð2  c_ =c_ c Þ c_ þ K HB ð2  nHB Þ þ ðnHB  1Þ c_ c c_ c ð11Þ

where g is the apparent viscosity of the material [Pa s] and c_ c the critical shear rate in the material [1/s]. The critical shear rate, as defined in Ansys Fluent for the modified Herschel-Bulkley model, is not to be confused with the definition mentioned in other literature [2,5]. When a critical shear rate is mentioned in the remainder of this article, we refer to the Ansys Fluent definition stated above. In order to capture the evolution of the interface between the SCC and the air during the formwork filling process, the Volume of Fluid (VOF) method, developed by Hirt and Nichols [33], is used. The VOF method can model two or more immiscible fluids by solving a single set of momentum equations and tracking the volume fraction of each of the fluids throughout the domain.

Fig. 10. Computational domain for the wall type A filling process.

3.3. Computational domain The first step in a numerical simulation is the creation of the geometry of the computational domain, and while doing this, defining the boundaries of the system under study. For the numerical simulation of the SCC flow through the formworks, shown in Fig. 6, a 3D geometry is constructed for each cast element and meshed with the Ansys Fluent Gambit v.2.3.16 pre-processor [34]. Figs. 9–11 show the resulting high quality meshes for the columns and the walls. After performing a grid convergence study, these highly regular meshes provided the most accurate simulation results at an acceptable computation time. 3.4. Boundary conditions Boundary conditions need to be applied to the computational domain, in order to obtain solutions for the model equations mentioned in Section 3.2. At the flow inlet, a uniform velocity profile of the incoming SCC and the volume fraction of the SCC phase are

Fig. 11. Computational domain for the wall type B filling process.

imposed, while at the top of the formwork, the pressure is atmospheric, resulting in a zero gauge pressure condition at the outlet. At the pipe walls and the formwork walls, a no-slip condition is assumed. Other researchers [2,22,31] have observed the presence of a very fine layer of cement paste close to solid walls, in which large velocity gradients are present and for which the validity of this no-slip condition has been confirmed. However, further comments regarding the wall boundary condition are made in Section 4,2. The imposed inlet conditions, shown in Table 4, are derived from the flow measurements performed during the filling tests and the geometry of the inlet duct. The length of the inlet duct used in the present numerical models has been determined from small preliminary simulations, which aimed at obtaining for laminar flow conditions a fully developed parabolic velocity profile at the end of the duct, when the SCC is entering the formwork.

Table 4 Inlet conditions.

Fig. 9. Computational domain for the column filling process.

Structural element

Flow rate (l/s)

Inlet pipe diameter (mm)

Uniform inlet velocity (m/s)

Casting rate (m/h)

Columns Walls

5 6

106 106

0.57 0.68

490 25

226

S. Tichko et al. / Engineering Structures 70 (2014) 218–233

The solver settings used in the performed numerical simulations of the formwork filling tests are presented in Table 5.

Table 5 Solver controls.

a b

Controls

Columns

Wall type A and B

Multiphase VOF model/ scheme Flow regime VOF scheme Transient controls Unsteady formulation Body force formulation Momentum discretization Volume fraction discretization Pressure discretization Pressure–velocity coupling

2 Phases

2 Phases

Laminar Explicit NITAa 1st order implicit Implicit body force 2nd order upwind Geo-reconstruct

Laminar Explicit NITAa 1st order implicit Implicit body force 2nd order upwind CICSAMb

Body force weighted Fractional step method

Body force weighted Fractional step method

Non-iterative time advancement. Compressive interface capturing scheme for arbitrary meshes.

Table 6 Fresh properties of the SCC used in the tests. Property

Value

Units

Density qSCC Yield stress s0,HB Consistency index nHB Consistency factor KHB Critical shear rate

2314 10.3 1.35 17.7 0.36

(kg/m3) (Pa) (–) (Pa sn) (1/s)

3.5. Solution of the model equations The Navier–Stokes equations, formulated in Eqs. (8) and (9), are discretized using the Finite Volume Method and they are solved with a 3D, double precision, implicit pressure-based solver [32].

3.6. Formwork filling simulations Some material properties of the fresh SCC are required for performing the present numerical simulations. These properties were obtained from the measurements reported in Section 2. The value for the density is an average of the measurements of the three samples. The Herschel-Bulkley parameters were taken from the rheological measurements performed on the first sample. As already mentioned in Section 2, the rheology of the delivered SCC did not change during the filling tests. The values of the required properties are summarized in the following Table 6. The VOF method proves to be very effective in capturing the time evolution of the SCC–air interface during the formwork filling processes. Using this method, it is possible through simulation to optimise the formwork filling process during the design phase. The number of inlets and their positioning on the formwork can be determined in order to avoid the creation of unwanted air pockets during casting process. The self-levelling property of the chosen SCC mix during the filling process can also be verified. Figs. 12–14 show the simulated formwork filling processes of the two columns and the two walls respectively. Figs. 13 and 14 clearly show the evolution of the shape of the concrete surface during the filling processes. A small slope within the concrete surface can be observed during the initial stages of the wall casting processes. This slope gradually vanishes towards the end of the filling processes due to the self-levelling properties of the SCC. The simulated shape of the concrete surface was in good agreement with the observed shape during the formwork filling. This is qualitatively demonstrated in Fig. 15 for the columns and in Fig. 16 for the walls.

Fig. 12. Simulated filling process of the columns.

S. Tichko et al. / Engineering Structures 70 (2014) 218–233

227

Fig. 13. Simulated filling process of wall type A.

4. Comparison of the results and discussion 4.1. Comparison of the results Because in our experiments the SCC was pumped from the base of the formworks up to the top, it was constantly in motion during the filling process. The time needed to fill each of the formworks was also very short, around 16 s for both columns and around 280 s for both walls. According to Roussel [6] and Billberg [8], the characteristic time to build up a thixotropic structure in most SCC mixes is about 280 s and higher. This motivates our belief that a possible reduction of the formwork wall pressure due to thixotropy (see also Section 1.2) could be disregarded during the filling tests performed in this study. Without the presence of a thixotropic structural build-up, the formwork pressures are expected to be slightly higher than hydrostatic for low filling velocities (see Eq. (6)), because at these low filling velocities the resulting hydraulic losses will be very small compared to the hydrostatic pressure component. In our formwork filling experiments, formwork wall pressures that are slightly

higher than hydrostatic can therefore be expected during the filling of the two walls. The wall pressure during the filling of the two walls has been measured at several positions close to the base of the formworks, as shown in Fig. 8 of Section 2. The following graphs show the measured wall pressures and the simulated wall pressures at these locations. The type of the pressure measurement devices used during the filling of two walls is listed in Table 2 of Section 2. During these formwork filling tests, some of the pressure measurement devices were damaged by the course aggregates of the SCC right from the start of the experiment. This was the case for the electronic pressure sensor mounted at position 4 of wall type A, and the manometer with the intermediate water chamber mounted at position 3 of wall type B. For these two positions no wall pressures could be recorded. Due to practical difficulties, it was not possible to stop the filling process of the two walls exactly at a height of 2 m. Therefore the resulting height was 1.955 m for wall type A and 2.061 m for wall type B. As can be seen from Fig. 17, the wall pressure measured by the electronic pressure sensor at position 1 of wall type A remains

228

S. Tichko et al. / Engineering Structures 70 (2014) 218–233

Fig. 14. Simulated filling process of wall type B.

Top of the formwork

Fig. 15. Comparison of the simulated SCC–air interface with the on-site observation at t = 14 s for column type A.

229

S. Tichko et al. / Engineering Structures 70 (2014) 218–233

Inlet

Lateral wall of the formwork Fig. 16. Comparison of the simulated SCC–air interface with the on-site observation at t = 18 s for wall type B.

45 45

Wall pressure pos. 3 - meas.

40

Wall pressure pos. 3 - simul.

40

Hydrostatic component - theor.

35

Hydrostatic component - simul.

30 25 20 15 Wall pressure pos. 1 - meas. Wall pressure pos. 1 - simul. Hydrostatic component - theor. Hydrostatic component - simul.

10 5

Pressure [kPa]

Pressure [kPa]

35 30 25 20 15 10 5 0

0

50

100

150

200

250

300

0

Time [s]

0

50

100

150

200

250

300

Time [s] Fig. 17. Wall pressure at the electronic pressure transducer above the inlet of wall type A – position 1.

Fig. 19. Wall pressure at the manometer of wall type A – position 3.

45

45

40

40

35

35

30

30

Wall pressure pos. 1 - meas. Wall pressure pos. 1 - simul.

Pressure [kPa]

Pressure [kPa]

Hydrostatic component - theor.

25 20 15 Wall pressure pos. 2 - meas.

10

25 20 15 10

Wall pressure pos. 2 - simul. Hydrostatic component - theor.

5

Hydrostatic component - simul.

5

Hydrostatic component - simul.

0

0

50

100

150

200

250

300

Time [s]

0

0

50

100

150

200

250

300

Time [s]

Fig. 18. Wall pressure at the manometer of wall type A – position 2.

Fig. 20. Wall pressure at the manometer above the inlet of wall type B – position 1.

hydrostatic all the way during the filling process, and is therefore in agreement with the expected wall pressure derived from Eq. (6) when the hydraulic losses are small due to a low filling velocity (6 l/s or 0.006 m3/s divided by 0.84 m2 equals 0.00714 m/s). The

simulated wall pressure at the same location is also hydrostatic. On the other hand, the wall pressures at the measurement positions 2 and 3 of wall type A are lower than the expected and simulated hydrostatic pressure. For wall type B, Figs. 20 and 21 also

S. Tichko et al. / Engineering Structures 70 (2014) 218–233

show a wall pressure at the measurement positions which is lower than the hydrostatic pressure, although the difference between the expected hydrostatic wall pressure and the measured wall pressure is smaller than the wall pressure difference observed at position 2 and 3 of wall type A. In reality, are the occurring wall pressures at these locations really lower than the hydrostatic pressure or is this a measuring artefact? An answer to this question will be formulated in the discussion in the next section. Due to a much higher filling velocity (5 l/s or 0.005 m3/s divided by 0.03654 m2 equals 0.137 m/s) for the columns, we expect the wall pressures to be higher than the hydrostatic pressure, because the hydraulic losses in these filling cases should be much higher than during the filling of the two walls (see also Eq. (6)). Figs. 22 and 23 show the measured wall pressure evolution just above the inlet of the two columns. The wall pressures above the inlet were also measured with a manometer – intermediate water chamber device for both columns. As for the walls, it was also not possible to fill the two formworks of the columns exactly up to a height of 2 m. Therefore the resulting height was 2.012 m for column type A and 2.225 m for column type B. When looking at the measured wall pressure evolution of column type A, we notice that the measured wall pressure above

45

Wall pressure pos. 2 - meas. Wall pressure pos. 2 - simul.

40

Hydrostatic component - theor.

Pressure [kPa]

35

Hydrostatic component - simul.

30 25 20 15 10 5 0

0

50

100

150

200

250

300

Time [s] Fig. 21. Wall pressure at the manometer of wall type B – position 2.

45

Wall pressure manometer - meas. Wall pressure manometer - simul.

40

Hydrostatic component - simul.

Pressure [kPa]

35

Hydrostatic component - theor.

30 25

50 Wall pressure manometer - meas.

45

Wall pressure manometer - simul.

40

Hydrostatic component - simul. Hydrostatic component - theor.

35

Pressure [kPa]

230

30 25 20 15 10 5 0

0

2

4

6

8

10

12

14

Time [s] Fig. 23. Wall pressure above the inlet at the manometer of column type B.

the inlet is indeed higher than the hydrostatic pressure, and that the measurements are consistent with the simulated wall pressures at the same measuring location. On the other hand, the wall pressure recorded just above the inlet of column type B is lower than the hydrostatic pressure at the same location. So which measurement is telling us the truth? More reflections about these observations also in the next section. Last but not least some additional information regarding the flow of SCC inside the formworks during the performed filling tests. Fig. 24 shows the simulated wall pressure just above the inlet of column type A. Due to the SCC flow pattern in the 90° bend at the inlet, there is difference between the wall pressure at the front of the column and wall pressure at back of the column, where the entering SCC is hitting the wall. This wall pressure difference just above the inlet will be more pronounced at high filling velocities, and so, is only relevant in our column filling tests. The difference is about 130 Pa. Compared to the absolute value of the wall pressure above the inlet, this small pressure difference is rather insignificant. As already mentioned in Section 1.2, the concrete velocities in the formwork below the inlet are very small. This is illustrated in Fig. 25, where the velocity vector plot in the centre plane of column type A is shown. For the two walls the velocities in the lower zone are even much lower. Therefore, the contribution of the hydraulic losses to the wall pressure in the region below the inlet will be very small. Thus we may conclude that the difference between the maximum wall pressure at the bottom of the formwork and the wall pressure at the measurement location just above the inlet is the hydrostatic pressure due to the height of SCC below the manometer position. This is illustrated by the simulation results presented in the following Table 7, proving the validity of Eq. (7), derived in Section 1.2.

20

4.2. Discussion

15

When studying the results of the measurements performed with the electronic pressure sensors and the manometer – intermediate water chamber units carefully, we can summarize the following observations:

10 5 0

0

2

4

6

8

10

12

14

Time [s] Fig. 22. Wall pressure above the inlet at the manometer of column type A.

 The electronic pressure sensor presents a linearly varying wall pressure with time, whereas the measurements done with the manometer – intermediate water chamber units lead to a non-linear evolution of the wall pressure.

S. Tichko et al. / Engineering Structures 70 (2014) 218–233

231

Fig. 24. Simulated wall pressure just above the inlet of the columns (0 mm = pressure measurement – 210 mm = formwork wall facing the inlet).

Fig. 25. Vector plot of the simulated velocities below the inlet of column type A at t = 14.1 s.

 The wall pressure measured by the electronic pressure sensor above the inlet of wall type A and the manometer – intermediate water chamber unit above the inlet of column type A are both in agreement with the theory derived in Section 1.2 and with the simulation results.  The other wall pressure measurements, performed with manometer – intermediate water chamber units, are lower than the simulated wall pressures and even lower than the hydrostatic pressure which is related to concrete height being cast in the formwork. These measurements are not in agreement with the theory derived in Section 1.2.

 The measured wall pressures of the columns are higher than the measured wall pressures of the walls, although the hydrostatic pressure component remains equal for all casts. Although the thixotropic parameters of Eqs. (2) and (3) have not been determined through separate rheometer tests, as proposed by Roussel in [6], we believe that, motivated by the statements made earlier in Section 4.1, thixotropy cannot be important in our base filling tests, and as such is not an explanation for the discrepancy between theory/simulation and some of the measured wall pressures. Thixotropy could not have influenced the formwork

Table 7 Comparison of the simulated wall pressures of column type A (simulation time = 14.1 s) at two locations. Simulated wall pressure at the manometer location above the inlet (h = 0.385 m) Simulated wall pressure at the base of the column Difference between the simulated wall pressure of the two locations Hydrostatic pressure component of the SCC between the manometer and base

36,287 Pa 44,964 Pa 8752 Pa 9.81 m/s2  2314.4 kg/m3  0.385 m = 8741 Pa

232

S. Tichko et al. / Engineering Structures 70 (2014) 218–233

pressure, because of the relatively short filling times and the SCC being constantly sheared during the whole filling process (certainly for the columns, maybe less pronounced for the walls). However, the non-linear evolution of the measured formwork pressures of the walls, shown in Fig. 17–23, can be related to the experienced shortcomings of the manometer – intermediate water chamber units used in the filling tests. The calibration of these pressure measurement units turned out to be very difficult, due to the non-linear deformation of the plastic membranes being submitted to the concrete pressure, the variable surface tension of the used membranes and the unavoidable inclusion of air bubbles into the water chambers. After the SCC had hardened and the formwork walls were removed, we observed that the granulates of the SCC were completely blocking the rubber membrane of the pressure measurement units of the two walls. The diameter of the circular measuring surface with the rubber membrane was 20 mm. Although the nominal maximum size of the granulates was 14 mm, some of them did have a diameter of about the same size as the pressure measurement opening (see also Section 2). This blocking would surely result into lower wall pressures being measured during the formwork filling tests. We therefore carefully attribute the mere existence of the discrepancy between theory/ numerical simulations and some of the measured wall pressures to a measuring artefact. Further optimization of the design of the manometer – intermediate water chamber units is necessary in order to avoid the previously mentioned shortcomings. Although the present design of the manometer – intermediate water chamber units was simple and the costs for manufacturing these units were very low compared to the price of the electronic pressure sensors, it turned out that measuring the wall pressure with an electronic sensor is more accurate and more practical. It is also much more convenient to capture the measurement data electronically, instead of having to use camera’s and stop watches to monitor the evolution of the manometers. The simulation results and the pressure measurements clearly show the contribution of the hydraulic losses to the formwork wall pressures. For formworks being filled from the base at a high casting rate, the hydraulic losses become important, certainly for highly viscous SCC. The theory derived in Section 1.2 as well as the simulations reveal that the hydraulic losses add up to the hydrostatic pressure, thus increasing the formwork pressure even more. This is clearly seen in all the simulation results and in some of the pressure measurements. Although the hydrostatic pressure is the same for all the casted structural elements because of equal height, the formwork pressure at the base of the columns is higher than for the walls. This is because the cross section of the walls is much larger than the cross section of the columns, and therefore the filling velocity and the resulting hydraulic losses in the columns are much higher than the filling velocity and the related hydraulic losses in the walls. The hydraulic losses, when pumping SCC from the base of the formworks up to the top, will even increase more if the flow cross section of the formwork is reduced due to the presence of steel reinforcements or if the viscosity of the chosen SCC is higher. On the other hand, Feys [2] and Le [31] have clearly demonstrated that when pumping SCC through ducts, the hydraulic losses are much lower than expected, due to the presence of a thin cement paste layer near the walls with a much lower viscosity than the concrete bulk viscosity. The effect of the thin paste layer near walls on the occurring hydraulic losses can be expected to be much more important when the ratio between the sum of the wall surfaces (formwork and rebars) and the total formwork volume reaches a certain (yet undetermined) value. So for accurately determining the hydraulic losses in ducts while pumping SCC to the formworks [2], the correct implementation of the no-slip condition applied to the thin cement paste layer near the walls becomes essential. In

our filling tests, this influence is much less pronounced because of the relatively large cross sections (certainly for the walls) and of the absence of rebars in our filling tests. Further research is needed in order to determine the threshold value of this previously mentioned ratio of wall surface to formwork volume, for which a correct numerical implementation of the interaction between the thin cement paste layer and the walls, but also the interaction of the cement paste layer and the remaining concrete becomes necessary. These effects will be further analysed and quantified in future full-scale formwork filling tests with different SCC mixes and with steel rebars being included. In these future formwork filling tests, electronic pressure sensors will be used in order to obtain more accurate wall pressure measurements and hopefully to rule out any measuring artefact as experienced in the present study. When more details on the formwork filling process are required and accurate measurements cannot be implemented or are very difficult to perform, numerical simulations can be carried out using the implemented Herschel-Bulkley material model, describing the relation between the applied shear rate and the resulting shear stress in the SCC, in combination with the VOF method. These CFD simulations can predict the formwork pressures, the flow and time evolution of the casting process accurately, provided the correct input data is supplied. 5. Conclusions Large-scale formwork filling tests by pumping self-compacting concrete (SCC) bottom-up have been performed and combined with computational fluid dynamics (CFD) techniques. The following conclusions have been obtained:  When pumping SCC from the bottom, in principle hydrostatic formwork pressures can be expected increased with the hydraulic pressure losses. The hydraulic losses become more important for high casting rates, small formwork cross-sections (e.g. columns) or when using an SCC with a high viscosity.  When pumping SCC from the bottom it is possible to noticeably reduce the casting times by applying high casting rates. However, stronger and stiffer formwork wall systems will be required to sustain the higher wall pressures.  The use of a manometer, mounted on an intermediate water chamber and sealed with a rubber membrane, to measure wall pressures in a formwork leads to inaccurate measurements. The use of electronic pressure sensors to measure formwork wall pressures is therefore to be preferred. It is much more convenient to directly capture the wall pressure data in a digital format for further analysis.  In case of relatively high filling rate, the effect of thixotropy on the formwork pressure can be disregarded.  The correct determination of the rheological material parameters of the SCC mix used in the filling process is important, as these parameters have an influence on the resulting formwork wall pressures, especially the consistency factor and the consistency index [2,8,24–26,37]. The formwork filling process and the resulting formwork pressures can be accurately simulated by means of CFD techniques in combination with a Volume of Fluid (VOF) approach for the simulation of the free concrete surface. These simulations allow optimising the formwork filling process during the design phase. References [1] De Schutter G, Bartos P, Domone P, Gibbs J. Self-compacting concrete. Caithness: Whittles Publishing; 2008. [2] Feys D. Interactions between rheological properties and pumping of selfcompacting concrete. Ph-D dissertation, Ghent University; 2009.

S. Tichko et al. / Engineering Structures 70 (2014) 218–233 [3] Bingham EC. Fluidity and plasticity. New York: McGraw-Hill; 1922. [4] Herschel WH, Bulkley R. Measurement of consistency as applied to rubberbenzene solutions. Proc ASTM 1926;26:621–33. [5] Roussel N. Steady state flow of cement suspensions: a micromechanical state of the art. Cem Concr Res 2010;40:77–84. [6] Roussel N. A thixotropy model for fresh fluid concretes. Cem Concr Res 2006;36:1797–806. [7] Roussel N, Geiker MR, Dufour F, Thrane LN, Szabo P. Computational modelling of concrete flow: general overview. Cem Concr Res 2007;37:1298–307. [8] Billberg P. Form pressure generated by SCC – influence of thixotropy and structural behaviour at rest. Ph-D dissertation, Stockholm: Royal Institute of Technology; 2006. [9] CIRIA. CIRIA Report 108 – concrete pressure on formwork. London: Construction Industry Research and Information Association; 1985. [10] Deutsches Institut für Normung. DIN 18218 – Frischbetondruck auf lotrechte Schalungen. Berlin: Beuth Verlag; 2008. [11] ACI Committee 347–04. Guide to formwork for concrete. ACI Standard; 2004. [12] Adam M, Bennasr M, Santos Delgado H. Poussée du béton frais sur les coffrages. Annales de l’ITBTP; 1965. [13] Gardner NJ. Formwork pressure and cement replacement by fly ash. Farmington Hills: American Concrete International; 1984. [14] Puente L, Santilli A, Lopez A. Lateral pressure over formwork on large dimension concrete blocks. Eng Struct 2010;32:195–206. [15] Cauberg N, Desmyter J, Piérard J, Parmentier B. Bekistingsdruk bij zelfverdichtend beton – WTCB dossiers – Katern 7, no. 3, WTCB; 2006 (in Dutch). [16] Khayat KH, Bonen D, Shah S, Taylor P. SCC formwork pressure – task 1: capturing existing knowledge on formwork pressure exerted by SCC. National Ready-Mix Concrete Research Foundation – American Concrete Institute; 2007. [17] Ovarlez G, Roussel N. A physical model for the prediction of lateral stress exerted by self-compacting concrete on formwork. RILEM Mater Struct 2006. [18] Roussel N. Rheology of fresh concrete: from measurements to predictions of casting processes. Mater Struct 2007;40:1001–12. [19] Roussel N, Cussigh F. Distinct-layer casting of SCC: the mechanical consequences of thixotropy. Cem Concr Res 2008;38:624–32. [20] Omran AF, Khayat KH, Elaguab YM. Effect of SCC mixture composition on thixotropy and formwork pressure. J Mater Civ EngASCE 2012:876–88. [21] Van De Maele J, Vanmassenhove N. Vullen van een bekisting met zelfverdichtend beton door het pompen van onderaan. Master thesis, Ghent University; 2010 (in Dutch). [22] Tichko S, Feys D, De Schutter G, Vierendeels J, Verhoeven R, Troch P. Numerical modelling of the flow of self-compacting concrete in conveying pipes. In: 6th

[23]

[24] [25]

[26]

[27]

[28] [29] [30]

[31]

[32] [33] [34] [35]

[36]

[37]

233

International RILEM symposium on self-compacting, SCC 2010 Conference Proceedings, vol. 2, Montreal; 2010. p. 719–28. Hirsch Ch. Numerical computations of internal and external flows. Fundamentals of numerical discretization, vol. 1. New York: John Wiley & Sons; 1989. Wallevik JE. Rheology of particle suspensions. PhD dissertation, The Norwegian University of Science and Technology; 2003. Hu C, de Larrard F, Sedran F, Boulay C, Bosc F, Deflorenne F. Validation of BTRHEOM, the new rheometer for soft-to-fluid concrete. Mater Struct 1996;29:620–31. Le HD, Vierendeels J, Tichko S, Troch P, Kadri EH, Aggoun S, et al. Computational fluid dynamics calibration of Tattersall MK-II type rheometer for concrete. Appl Rheol 2013;23. Shyshko S, Mechtcherine V. Simulating the workability of fresh concrete. In: Proceeding of the international RILEM symposium of concrete modelling – CONMOD’08; 2008. p. 173–81. Vasilic K, Meng B, Kühne HC, Roussel N. Flow of fresh concrete through steel bars: a porous medium analogy. Cem Concr Res 2011;41:496–503. Martys NS. Study of a dissipative particle dynamics based approach for modeling suspensions. J Rheol 2005;49:401–24. Spangenberg J, Roussel N, Hattel JH, Stang H, Skocek J, Geiker MR. Flow induced particle migration in fresh concrete: theoretical frame, numerical simulations and experimental results on model fluids. Cem Concr Res 2012;42:633–41. Le HD, De Schutter G, Kadri EH, Aggoun S, Vierendeels J, Troch P. Velocity profile of self compacting concrete and traditional concrete flowing in a half open pipe. In: 3rd International conference on concrete repair, rehabilitation and retrofitting (ICCRRR-2012), CRC Press; 2012. p. 1382–7. Fluent 12.0 User’s Guide. Fluent Inc.; 2009. Hirt CW, Nichols BD. Volume of fluid method for the dynamics of free boundaries. J Comput Phys 1981;39:201–25. Gambit 2.3 Modelling Guide. Fluent Inc.; 2006. De Schutter G, Feys D, Verhoeven R. Ecological profit for a concrete pipe factory due to self-compacting concrete technology. In: Zachar J, et al, editors. Proceedings of the 2nd international conference on sustainable construction materials and technologies, vol. 2, Ancona; 2010. p. 1281–7. Mechtcherine V, Gram A, Krenzer K, Schwabe J-H, Shyshko S, Roussel N. Simulation of fresh concrete flow using discrete element method (DEM): theory and applications. Mater Struct 2014;47(4):615–30. Mechtcherine V, Nerella VN, Kasten K. Testing pumpability of concrete using Sliding Pipe Rheometer. Construct Build Mater 2014;53:312–23.