You are on page 1of 11

Magazine of Concrete Research Magazine of Concrete Research, 2014, 66(4), 175–185

Volume 66 Issue 4 http://dx.doi.org/10.1680/macr.13.00184


Paper 1300184
Evaluation of SCC yield stress from L-box Received 24/06/2013; revised 31/10/2013; accepted 31/10/2013
test using the dam break model Published online ahead of print 20/12/2013

Chamani, Hosseinpour, Mostofinejad and ICE Publishing: All rights reserved


Esmaeilkhanian

Evaluation of SCC yield stress


from L-box test using the dam
break model
Mohammad R. Chamani D. Mostofinejad
Associate Professor, Department of Civil Engineering, Isfahan University of Professor, Department of Civil Engineering, Isfahan University of
Technology, Isfahan, Iran Technology, Isfahan, Iran
M. Hosseinpour B. Esmaeilkhanian
Graduate Student, University of Sherbrooke, Sherbrooke, Canada Graduate Student, University of Sherbrooke, Sherbrooke, Canada

In this study, an analytical method for the dam break flow of non-Newtonian fluids is extended to model self-
compacting concrete flow in the L-box test. The rheology of fresh self-compacting concrete is expressed by a
Bingham model. To simplify the mathematical formulation, inertia forces and the effect of rebar are neglected and
the fluid is assumed to be homogeneous. Several self-compacting concrete mixtures with different yield stresses are
also tested to measure the L-box blocking ratio. The proposed model arrested profiles satisfactorily coincide with the
experimental results. A relation between yield stress and the L-box blocking ratio is also derived. A comparison
between the predicted values of yield stress and the experimental results shows that the proposed model
successfully evaluates the yield stress of the tested samples. Moreover, it is observed that the proposed model is
valid in the case of an L-box with rebar. Finally, a plot of the relative yield stress against L-box blocking ratio is
presented that can be practically used for workability design of self-compacting concrete.

Notation î (x  xy )/(xf  xy ) non-dimensional length along


B Bingham number x-axis
Dmax maximum recommended size of aggregate r density
G torque ô0 yield stress of fluid
g gravitational acceleration
H characteristic fluid depth
h(x, t ) depth of the flowing layer Introduction
h1 (î) final profile of SCC in L-box test Properties of self-compacting concrete (SCC) are distinguished
h1 height of SCC arrested profile at L-box back from those of normal concrete when it flows in a fresh state. Two
wall major properties of SCC are filling ability and passing ability
h2 height of SCC arrested profile at end wall (Goodier, 2003). These properties are two main parameters for
h2 /h1 L-box blocking ratio workability design of concrete to attain high quality and
~
h(î, t), ~xf (t) perturbation variables efficiency of the concrete during construction. An inexpensive
k, n coefficients and efficient workability design is to predict and further optimise
L characteristic length the workability of concrete by way of flow simulation for the
L0 L-box channel length selected construction processes; for example, transportation, cast-
l0 L-box channel width ing, compaction, finishing, and so on. Modelling the rheological
R2 correlation coefficient behaviours of fresh concrete, and developing a proper flow
Sf slump flow value analysis, are the most basic conditions of workability design.
t time SCC provides a good opportunity to develop and use workability
V volume of test sample design technology because its self-compacting characteristic can
x, z coordinates simplify the flow simulation. Laboratory tests such as slump flow,
xa active length from the lock-gate V-funnel and L-box are used to quantitatively assess the filling
xf instantaneous position of front edge from lock- ability and flowability properties. The outputs of the tests depend
gate on the geometry of the test apparatus and intrinsic characteristics
xy instantaneous position of yield point from lock- of the tested SCC. Using the principles of fluid dynamics, it is
gate possible to develop a relation between the rheological parameters
â coefficient of SCC, the geometry and the output results of rheological tests.

175
Magazine of Concrete Research Evaluation of SCC yield stress from L-box
Volume 66 Issue 4 test using the dam break model
Chamani, Hosseinpour, Mostofinejad and
Esmaeilkhanian

The relation between the yield stress of fresh concrete and the Obviously, the flow of SCC as a viscoplastic fluid inside the
results of rheological tests has been studied by many researchers. L-box apparatus can be treated as a laboratory dam break test. In
More reliable outputs are those recorded when the fresh concrete fact, the SCC mixture is poured into the vertical part of the
stops; that is, the slump flow diameter and the L-box blocking device while the gate is closed (see Figure 1). Then, the gate is
ratio. Considering the governing equations of fluid motion and removed, letting the sample inside the reservoir flow into the
existing models for the shear stress of non-Newtonian fluids, it is horizontal part. The governing equations of viscoplastic dam
concluded that the flow of fresh concrete stops when the shear break are then used to simulate the flow of SCC in the L-box test.
stress in the tested sample becomes equal to or smaller than the The L-box apparatus studied in this work corresponds to one
yield stress. The final shape of the tested sample is directly presented earlier (Precast/Prestressed Concrete Institute, 2003).
related to the fresh concrete yield stress. Earlier works (Murata, The dimensions of the apparatus are shown in Figure 1.
1984; Schowalter and Christensen, 1998) examined the cone test
of fresh concrete. It was shown that the slump test is a practical There are two major challenges for modelling SCC flow in the
measure to determine the yield stress of a material (Roussel and L-box test. In the normal L-box test, three rebars are placed in
Coussot, 2005; Roussel, 2006). It is concluded that in a general front of the gate to evaluate the potential of the SCC sample for
case, a numerical approach can estimate the yield stress based on segregation. The presence of rebars may affect the flow of the
the elasto-visco-plastic model of a non-Newtonian fluid. Recently, sample owing to the blocking of aggregates. The presented 2D
a limited number of research studies have been conducted models neglect the effects of the sidewalls. Numerical simulations
investigating the correlation between yield stress and L-box of the flow in the L-box with rebar and without rebar using a 2D
outputs (Dufour and Pijaudier-Cabot, 2005; Lashkarbolouk et al., and three-dimensional (3D) Galerkin finite element showed that
2013; Roussel et al., 2007). the effect of L-box sidewalls on flow profile is negligible (Thrane
et al., 2004). Previous studies on slump flow tests also showed
In the present work, a two-dimensional (2D) analytical model is that 2D approximation yields acceptable errors (Dufour and
developed to predict SCC flow in the L-box test. As the physical Pijaudier-Cabot, 2005; Roussel et al., 2007). The comparison
conditions of the flow in the L-box test resemble those in dam between the predicted values of the proposed model with experi-
break phenomena, the previous method for the analysis of dam mental results is further evidence of the validity of the proposed
break for a non-Newtonian fluid (Matson and Hogg, 2007) is 2D assumption.
modified. The proposed model predictions are compared with
previous experimental data. Finally, some practical applications Background
of the proposed model are provided and the predictions of the The inertia forces are often neglected in the Navier–Stokes
model are compared with the model from a previous work equations used in solving dam break problems (Hogg and
(Nguyen et al., 2006). In the present model, the SCC flow is Pritchard, 2004; Whitham, 1955). For large values of yield stress,
studied in the final intervals of fluid motion where the inertia this approximation is valid due to the domination of viscous
effects are already diminished due to reduction of the flow forces over inertia ones. In addition, the induced back-wave that
velocity. occurs when the fluid strikes the wall of the channel end

)
Characterisation of SCC flow in L-box test 0 mm 4 in (10
0 mm)
(20
7 in
L-box test and dam break phenomenon 78
The phenomenon during which the volume of fluid bounded in
the reservoir of a dam is instantaneously released due to the
collapse of the dam is called dam break. This results in the flow
23 85 in (600 mm)

3
of fluid through the downstream area, driven by gravitational Three no. 4 rebar distanced at 8 in (35 mm)
forces. The release of a viscous or viscoplastic fluid in a channel
as a result of dam break is a classical problem in shallow-water
fluid mechanics (Balmforth and Craster, 1999; Balmforth et al.,
5 in (150 mm)

2007; Hogg and Pritchard, 2004). The gravitationally driven


h1

spreading of a layer of viscoplastic fluid has a wide range of


h2

applicability in industrial processes such as lavas, mud and liquid


foods. Such non-Newtonian fluids show a non-linear relation
7
8

3121 in (8
between shear stress and strain rate. The major concern when 00 mm
)
analysing such processes is how to incorporate the yield stress in
calculations. It is assumed that the velocity is predominantly
horizontal and the pressure approximately hydrostatic. Laboratory Figure 1. L-box test apparatus (adapted from Precast/Prestressed
experiments have been conducted to verify the proposed models Concrete Institute, 2003)
(Chanson et al., 2006; Nsom, 2002; Zhou et al., 2004).

176
Magazine of Concrete Research Evaluation of SCC yield stress from L-box
Volume 66 Issue 4 test using the dam break model
Chamani, Hosseinpour, Mostofinejad and
Esmaeilkhanian

disappears for large values of yield stress. This indicates that taneous position of the front edge of the yield point. Note that
inertia forces can be neglected with good accuracy. xy (t ) is a point behind which the fluid does not take part in
motion located between 0 , x , 1. At x ¼ xf , h ¼ 0, boundary
The 2D slow flow of a sheet of incompressible viscoplastic fluid conditions are h(î ¼ 1, t ) ¼ 0 for all t; and for 0 < x < xy ,
down an inclined plane using a Herschel–Bulkley model and a h(x) ¼ 1 and the boundary condition is h(î ¼ 0, t ) ¼ 1. For
lubrication model of motion is analytically studied (Matson and notational convenience, the active length is introduced as
Hogg, 2007). Neglecting the inertia effect, the arrested state xa (t ) ¼ xf (t )  xy (t ).
occurs when the gravitational forces are in equilibrium with the
yield stress. The evolution of flows is derived by numerically The governing equations of the motion of the flowing layer with
integrating the equations of motion. Using an analytical solution, Reynolds number of order unity at most were then formulated.
the perturbations to the final state are calculated resulting in the Combining the mass and momentum equations, and using an
determination of flow profile from initiation to a stationary state. explicit finite-difference scheme, the governing equation for the
It is assumed that the fluid layer is shallow and slowly moving so fluid depth was derived as (Matson and Hogg, 2007)
that lubrication style asymptotic reduction of the governing
equations can be applied (Matson and Hogg, 2007). The pressure @h 1 @h n @
 (î_xa þ x_ y ) ¼ xa
was assumed to be hydrostatic. Although Matson and Hogg @t xa @î 1 þ 2n @î
derived a single equation for the fluid depth in a differential form "   1þ1=n  #
2
in the case of an inclined plane, only the case of a horizontal @h h @h h @h n
plane was comprehensively solved and discussed. The coordinate
3  B  B
1: @î xa @î xa @î 1 þ n
system was chosen such that the x-axis was aligned along the
channel bed and the z-axis was perpendicular to the channel bed
(see Figure 2). The depth of the flowing layer was denoted by where the dot denotes differentiation with respect to time, n is a
h(x, t ). The motion was studied in the ‘lubrication regime’, for coefficient depending on the rheology of the fluid, and B is the
which the characteristic length scale along the bed, L, was much Bingham number, which demonstrates how the viscous forces
greater than the characteristic fluid depth, H. and gravitational forces are interacting, given by

Fluid motion after instantaneous release from behind a lock-gate ô0 L



and flowing over a horizontal bed is analysed next. The fluid is 2: rgH 2
enclosed between an impenetrable back wall located at x ¼ 0 and
the lock-gate at x ¼ 1. Before moving the lock-gate, the depth of
the fluid behind the lock-gate is assumed to be unity. After lifting where ô0 is the yield stress of the fluid, r is the fluid density and
the lock-gate, the instantaneous position of the front edge is g is the gravitational acceleration.
shown by x ¼ xf (t ) and its depth by h(xf , t ) ¼ 0 for all times.
Over the mathematical procedure, a change of variables is made Equation 1 is only applicable in the region 0 , î , 1. As for SCC
and is defined as î ¼ (x  xy )=(xf  xy ) where xy (t ) is the instan- the range of B is less than 1/3 (owing to its very low ratio of yield
stress to specific gravity), only the case of B , 1/3 is explained
here. When B , 1/3, the yield point reaches the back wall before
z
the motion is arrested. For a large value of time, it is assumed that
Profile of the
xy ffi 0 and therefore xa ¼ xf (Matson and Hogg, 2007). Consider-
fluid ing the height profile as a function of î, it is obvious that h(î, t ) is
always greater than the final profile, h1 (î), and that the front
position xf (t ) must always be less than the front position of flow
when it stops, xf 1 : The final profile, h1 (x), and xf 1 are derived as
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u " 1=3 #
u 9
t
h1 (x) ¼ 2B x
3: 8B

θ
x  1=3
9
xf1 ¼
4: 8B
Figure 2. Schematic view of 2D flow of a sheet of incompressible
viscoplastic fluid down an inclined plane
To obtain the instantaneous profile of the flow, the positive

177
Magazine of Concrete Research Evaluation of SCC yield stress from L-box
Volume 66 Issue 4 test using the dam break model
Chamani, Hosseinpour, Mostofinejad and
Esmaeilkhanian

~(î, t) and ~xf (t), for these arrested states


perturbation variables, h since ô0 of SCC is relatively low, this assumption is no longer valid
are obtained as (Matson and Hogg, 2007) as the flow front reaches the end wall of the channel before the flow
stoppage. Hence, in the proposed model, Equation 9 is used without
5: h(î, t) ¼ h1 (î) þ ~
h(î, t) any approximation. Substituting Equation 7 and Equation 9 into
Equation 5 yields
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 1=3
xf (t) ¼ xf 1  ~xf (t) 8B
6: h(î, t) ¼ (3B) 1=3
1  xf î
9
 n  (1þn)=3
(1 þ n)k 3 E(î)
where it is assumed that ~h=h1  1, ~xf =xf 1  1: In the solution þ
10: 2 B2 tn
presented by Matson and Hogg, the function ~h(î, t) is found by
the separation of variables method. It is assumed that
~ 1=3
h(î, t) ¼ A(î)F(t) ¼ (B2 =3) E(î)F(t) and ~xf (t) ¼ F(t): The
function E(î) represents the variation of the profile that results Substitution of Equation 4 and Equation 8 into Equation 6 gives
only from the time constraint. Upon substitution of Equations 5
 1=3  n  (2þn)=3
and 6 into Equation 1, using the continuity equation and the 9 (1 þ n)k 3 1
perturbation method, and after a rigorous mathematical proce- xf (t) ¼ 
11: 8B 2 B2 tn
dure, the following equations were derived.

 n  (1þn)=3
~h(î, t) ¼ (1 þ n)k 3 E(î) The values of E(î) for n ¼ 1 are displayed in Figure 3. Table 1
7: 2 B2 tn shows the values of k for various values of n, where for n ¼ 1,
k ¼ 0.08663. Using these values in Equations 10 and 11, and
eliminating t n gives
 n  (1þn)=3
(1 þ n)k 3 1 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
~xf (t) ¼  1=3
8: 2 B2 tn 8B
h(î) ¼ (3B)1=3 1  xf î
9
"   1=3 #
where k is a constant depending on the value of n and E(î) is a 1=3
3B B2
perturbation function produced during the mathematical procedure. þ E(î)  xf
12: 8 3
Proposed model
In the proposed model, an attempt is made to derive a relation- 1·6
ship between the yield stress of the concrete and the L-box 1·4
blocking ratio using Equations 5–8. The value of n is assumed as 1·2
unity, which means the rheology model of the fluid is expressed
1·0
by a Bingham fluid. For a large value of time, it is possible to
E(ξ)

0·8
assume that xy ffi 0. With this assumption, Equation 3 is trans-
formed into a function of î as 0·6
0·4
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
xf 0·2
h1 (î) ¼ (3B)1=3 1  î
9: xf 1 0
0 0·2 0·4 0·6 0·8 1·0
ξ
pffiffiffiffiffiffiffiffiffiffiffi
Equation 9 is approximated as h1 (î) ¼ (3B)1=3 1  î since Figure 3. Perturbation function E(î) plotted against re-scaled
Matson and Hogg assumed that ~xf =xf 1  1, which means distance, î, when B , 1/3 for n ¼ 1
xf ffi xf 1 (Matson and Hogg, 2007). However, in the L-box test,

n 1/5 1/3 1/2 1 2 3 5

k 0.37580 0.29360 0.21590 0.08663 0.01419 0.00235 0.00007

Table 1. Values of constant k for different power law indices n

178
Magazine of Concrete Research Evaluation of SCC yield stress from L-box
Volume 66 Issue 4 test using the dam break model
Chamani, Hosseinpour, Mostofinejad and
Esmaeilkhanian

For a fixed value of B, Equation 12 gives the height profile for any located in the range of 0.85 < î < 0.95. Some samples of the
xf : This equation will be verified later by comparing with case studies are shown in Figure 4.
experimental data. To simplify Equation 12 for the arrested profile
of SCC in the L-box, one must determine values for xf and î. It was stated that the height profile of fluid in the L-box apparatus
after it is arrested can be considered almost linear (Nguyen et al.,
In the case of SCC, the arrested length of sample in the L-box, 2006). It is concluded that, after being arrested, the linear part still
xf 1 , would always be larger than the L-box channel length and remains linear and the non-linear part flows back such that the
the channel end wall disturbs the normal arrested profile. The height profile at the end point of the linear part does not vary.
height profile given by Equation 12 cannot predict the effect of Figure 5 shows an image taken from the final shape of a SCC
the L-box channel end wall on the arrested profile shape. As sample in an L-box test that clearly illustrates the fact that the
introducing the effect of the end wall into the governing equa- arrested profile of SCC inside the L-box apparatus can be assumed
tions of the profile is difficult, one must link the instantaneous linear. Therefore, h2 can be assigned to this point, which is a point
profile of flow when its front position reaches the end wall, to the inside the ‘transition zone’ and was selected at î ¼ 0.9 as an
real arrested profile shape of sample in which h1 and h2 are the appropriate average for all case studies. As a practical validation
heights of the profile at the back wall and at the end wall, of this assumption, Figure 6 presents a comparison between the
respectively. By considering L ¼ 100 mm, xf becomes equal to 8. height profiles at xf ¼ 8 and arrested time, respectively. It can be
observed that when xf ¼ 8, h(î ¼ 0.9) is almost equal to h2
Since h1 is recorded at the L-box back wall, h(î ¼ 0) is used as measured from the final shape of the sample in the L-box test.
an approximation for h1 : To verify this assumption, the values of
h(î ¼ 0) for two boundary limits, namely at xf ¼ 8 and xf ¼ xf 1 Letting î ¼ 0 (E(î) ¼ 1.0) and î ¼ 0.9 (E(î) ¼ 1.38) in Equation
are presented in Table 2. Note that for xf ¼ 8, the values are 12, and for xf ¼ 8, h1 and h2 are determined as
calculated from Equation 12, but for xf ¼ xf 1 , the experimental
values are used. The recorded h1 from the L-box test falls "   1=3 #
1=3
between the two boundary limits. As the values of |˜h(î ¼ 0)| are 3B B2
h1 ¼ h(î ¼ 0) ¼ (3B)1=3 þ 8
small, it can then be concluded that approximating h1 with 13: 8 3
h(î ¼ 0) is a rational assumption. Besides, in the case of SCC,
the value of |˜h(î ¼ 0)| is less than the nominal maximum 0
recommended size of aggregate; that is, Dmax ¼ 19 mm (BIBM et ⫺0·05
al., 2005) and it is the maximum possible error for recording h1 : ⫺0·10
⫺0·15
In the final step, h2 must be determined. The profile used to ⫺0·20
h⬘(ξ)

derive h1 was formulated such that h(î ¼ 1) ¼ 0. It is not possible ⫺0·25


⫺0·30 Yield stress ⫽ 35 Pa
to assign h2 to h(î ¼ 1), and a point behind h(î ¼ 1) should be
⫺0·35 Yield stress ⫽ 41 Pa
used as an approximation for h2 : By examining the differentiation Yield stress ⫽ 14 Pa
⫺0·40
of height profile with respect to î, h9(î), it is observed that h9(î) Yield stress ⫽ 61 Pa
⫺0·45
can be divided into two main parts. In the first part, h9(î) is ⫺0·50
almost constant, meaning that the height profile, h(î), is nearly 0·50 0·55 0·60 0·65 0·70 0·75 0·80 0·85 0·90 0·95 1·00
linear. In the second part, there is significant variation of h9(î) in ξ
a short distance, which means the height profile is completely Figure 4. Samples of case studies for determining î
non-linear. By doing a simple case study, it is observed that the correspondent to h2 (xf ¼ 8 and specific gravity ¼ 2.5)
‘transition zone’ between the linear and non-linear parts is

Experimental yield Theoretical h(î ¼ 0) Experimental data |˜h(î ¼ 0)| ¼ |hexp – |˜h(î ¼ 0)|/hTheo : %
stress, ô0 : Pa at xf ¼ 8: mm hTheo: |(î ¼ 0): mm
xf1 h(î ¼ 0) at xf ¼ xf1
(mm)

14 47 11.32 43 4 8.5
35 56 9.64 52 4 7.1
42 58 8.00 61 3 5.2

Table 2. Two boundary limits for h(î ¼ 0), comparison between


theoretical values and experimental data (Nguyen et al., 2006) at
the arrested state

179
Magazine of Concrete Research Evaluation of SCC yield stress from L-box
Volume 66 Issue 4 test using the dam break model
Chamani, Hosseinpour, Mostofinejad and
Esmaeilkhanian

h(ξ ⫽ 0·9) at xf ⫽ 8

h1
h2
(a)
Figure 5. Linear final shape of SCC sample inside L-box test
apparatus (h2 /h1 ¼ 0.303)

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 1=3
8B
h2 ¼ h(î ¼ 0:9) ¼ (3B)1=3 1  7:2
9
"   1=3 #
1=3
3B B2
þ 1:38 8
14: 8 3

Knowing h2 and h1 , it is possible to calculate the L-box blocking h2


ratio, h2 /h1 , from yield stress as

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
h2 5:31(B)1=3  1  6:92(B)1=3  0:69 (b)
¼
15: h1 3:85(B)1=3  1:5 Figure 6. Comparison between height profiles: (a) profile at
xf ¼ 8, h(î ¼ 0.9) ¼ 38 mm; (b) arrested profile, h2 ¼ 40.2 mm

Solving Equation 15 for B gives

h pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi i3
3:81 3 106 0:82 þ 21:70 (1:65  h2 =h1 )(0:36 þ h2 =h1 ) þ h2 =h1 (45:85 þ 24:70h2 =h1 )

16: (1:38  h2 =h1 )6

Upon substituting the equivalent value of B ¼ ô0 L=rgH 2 into


Equation 16, the value of yield stress, ô0 , is derived as

h pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi i3
rgH 3 :81
2 3 10 6 :
0 82 þ 21 : 70 (1 :65  h 2 =h 1 )(0 : 36 þ h 2 =h 1 )  h 2 =h 1 (45 : 85  24 :70h 2 =h 1 )
ô0 ¼
17: L (1:38  h2 =h1 )6

180
Magazine of Concrete Research Evaluation of SCC yield stress from L-box
Volume 66 Issue 4 test using the dam break model
Chamani, Hosseinpour, Mostofinejad and
Esmaeilkhanian

discharged into the mixer. Then the cementitious materials with


Experimental study and analysis
another one-third of the total mixing water accompanied by SP
Experimental set-up are added to the mixture. In the final step, the rest of the mixing
Physical properties of all classes of aggregates used in the water containing VMA content is introduced into the mixture.
experimental work of the present study are shown in Table 3 The whole process lasts 3 min. Slump flow and L-box tests are
and sieve analyses are shown in Table 4. Portland cement type conducted immediately after mixing. For each mix design, the
I with specific gravity of 3.15 and a mineral additive (lime- L-box test is executed both with and without rebar. The yield
stone powder) with mean size of 0–150 m, specific gravity of stress of each mix design is evaluated using the equation (Murata
2.5 and water absorption of 11.1% are used for making the and Shitayama, 1976)
concrete samples. Super-plasticiser (SP) and viscosity-modifying
admixture (VMA) with polycarboxylic basis are also used. Both
SP and VMA had 30% solid content and a specific gravity of 2rgV
ô0 ¼
1.05. 18: ðS 2f

Seven SCC mixtures are designed to obtain different L-box


blocking ratios. Water/cementitious materials ratios and VMA where V is the volume of test sample, and Sf is slump flow value.
content are 0.4 and 2 kg/m3 , respectively, for all mix designs Results of rheological measurements are shown in Table 6. The
(Table 5). The mixing procedure for all mix designs is the same. test results will be used in the next section.
First, all aggregates with one-third of the total mixing water are
Prediction of arrested profile in L-box apparatus
So far, the relationship between the yield stress and L-box
blocking ratio by using the height profile of fluid inside the
L-box apparatus has been derived. The accuracy of the governing
Aggregate class Silica sand River Coarse Equation 12 for the prediction of arrested fluid profile is
(75–150 m) sand aggregate evaluated next. In Figure 7, the theoretical height profiles
predicted by Equation 12 for fluids with different yield stresses
Specific gravity 2.6 2.8 2.6
are compared with experimental data (Nguyen et al., 2006). The
Water absorption: % 9 2.5 1
tested fluids were suspensions of limestone powder and water
Table 3. Physical properties of aggregates with specific gravities from 1.8 to 2.3 depending on the solid
volume fractions.

Sieve size: mm 0.15 0.3 0.6 1.18 2.36 4.75 9.5

River sand 3.72 21.2 38.5 50.8 74.8 96.8 100


Coarse aggregate — 0.6 0.8 1.3 5.8 55 98.1

Table 4. Grading of aggregates (passing percent)

Mix number Gravel: River sand: Silica sand: Limestone Cement: Silica fume: SP: kg/m3
kg/m3 kg/m3 kg/m3 powder: kg/m3 kg/m3 kg/m3

1 736.3 788.3 242.6 60 330.4 36.7 6.0


2 742.8 795.7 244.7 60 314.7 35.0 7.5
3 749.3 802.2 246.8 60 298.9 33.2 7.5
4 755.9 809.2 249.0 45.5 298.9 33.2 7.5
5 762.4 816.2 251.1 31 298.9 33.2 7.5
6 768.9 823.2 253.3 16.5 298.9 33.2 7.5
7 775.4 830.1 255.4 2 298.9 33.2 7.5

Table 5. Mixture proportions

181
Magazine of Concrete Research Evaluation of SCC yield stress from L-box
Volume 66 Issue 4 test using the dam break model
Chamani, Hosseinpour, Mostofinejad and
Esmaeilkhanian

Mix L-box L-box Slump Yield evaluate the yield stress of SCC is presented next. It must
number blocking blocking ratio flow: stress: Pa be noted that there are various types of rheometers that
ratio (without mm measure the yield stress of fresh concrete. Previous works
(with rebar) rebar) showed that these devices give different values for the yield
stress for a specific fresh concrete (Ferraris et al., 2004).
1 0.37 0.43 640 210 The predicted values of the yield stress, some available
2 0.68 0.70 800 134 experimental data and analytical models (Emborg, 1999;
3 0.35 0.43 790 138 Hwang et al., 2012; Nguyen et al., 2006; Petersson and
4 0.55 0.62 765 147 Billberg, 1999; Sonebi, 2003), and experimental results of the
5 0.52 0.56 730 161 current study as described in the earlier section on ‘Experi-
6 0.40 0.44 735 159 mental set-up’, are shown in Figure 8. Note that the yield
7 0.21 0.28 700 175 stress of each mix design in the present experiments was
evaluated using Equation 18. The following equation is
Table 6. Rheological test results for various mix designs proposed between h1 and h2 in the case of L-box with rebar
(Nguyen et al., 2006)
 
ô0 L0 l0 L0 2 ô0
h1  h2 ¼ þ þâ
90 Proposed model 19: rg V l0 rg
80 Present experiments, yield stress ⫽ 14 Pa
Present experiments, yield stress ⫽ 35 Pa
70 Present experiments, yield stress ⫽ 42 Pa
Present experiments, yield stress ⫽ 61 Pa where L0 is the L-box channel length, l0 is the L-box channel
Thickness: mm

60
width and â is a dimensionless coefficient linked to geometry of
50 obstacles – that is, the diameter and the spacing of the rebar in
40 the L-box test, obtained as 4.4 from experimental data (Nguyen
30 et al., 2006). Having assumed a linear arrested profile for the
20 flow, the mean height is introduced as havg ¼ (h1 þ h2 )=2 ¼
10 V =(l0 L0 ): For the L-box apparatus studied here, the values are
0 L0 ¼ 800 mm, l0 ¼ 200 mm, and V ¼ 12 l. Upon substituting
0 200 400 600 800 1000 1200 these values and using the concept of havg , the modified form of
x: mm Equation 19 is derived as (Nguyen et al., 2006)
Figure 7. Comparison between predicted arrested profiles  
rg 1  (h2 =h1 )with rebar
(Equation 12) and Nguyen et al. (2006) data ô0 ¼  
20: 154 1 þ (h2 =h1 )with rebar

Figure 7 shows that the arrested profile obtained from Equation Note that all the experimental data were obtained from L-box test
12 perfectly coincides with experimental profiles for different with rebar. The values of yield stress (ô0 ) in other works are also
tested samples. The values for correlation factor, R2 , for all expressed in the form of torque, G (Emborg, 1999; Sonebi,
data series are between 0.986 and 0.994. The only minor errors 2003). The United States National Institute of Standards and
occurred where the fluid hit the back wall (x ¼ 0 mm). This Technology (NIST) estimated the yield stress as (Ferraris et al.,
discrepancy basically appears as the effects of the back wall on 2004)
the motion of fluid in the vertical direction are not considered
in the proposed model. As a result, the experimental values are 21: ô0 ¼ 122G
larger than theoretical ones owing to the effects of frictional
forces which are exerted on the fluid, resulting in a lower
velocity. However, h1 is approximated somewhat larger than its Although the proposed model does not include the presence of
real value in the proposed model (see Table 2). This latter rebar, Figure 8 shows that the predicted values of yield stresses
effect could compensate the former assumption (neglecting the agree fairly well with experimental results recorded from tests
vertical motion at the back wall), since they have opposing with rebar (R2 ¼ 0.75, RMSE ¼ 37.3 Pa). Moreover, the predicted
effects. values give better assessment of yield stress in comparison with
the previous model (Nguyen et al., 2006) (R2 ¼ 0.71,
Applicability of proposed model to the case of SCC RMSE ¼ 72.0 Pa).
Although there is still no device that gives an absolute yield
stress for concrete, the capability of the proposed model to As a final output, a plot of the ratio of yield stress to specific

182
Magazine of Concrete Research Evaluation of SCC yield stress from L-box
Volume 66 Issue 4 test using the dam break model
Chamani, Hosseinpour, Mostofinejad and
Esmaeilkhanian

Proposed model, Equation 17


Nguyen et al. (2006) model, Equation 20
Present experiments
Petersson and Billberg (1999)
Emborg (1999)
Sonebi (2003)
Hwang et al. (2012)
250

200

150
τ0: Pa

100

50

0
0 0·1 0·2 0·3 0·4 0·5 0·6 0·7 0·8 0·9 1·0
h2 /h1

Figure 8. Comparison between proposed model (solid line) and


Nguyen et al. (2006) model (dashed line) and experimental data
of the yield stress of SCC

gravity for various L-box ratios is given in Figure 9. The ô0


(Pa) ¼
graph is considered as a practical feature of this study which SG
8  3  2  
can be used for quick evaluation of yield stress in on-site > h2 h2 h2
>
> : : :
works. As another alternative, Figure 9 is also approximated >
> 91 3  175 3  39 6 þ 104:8
>
> h 1 h 1 h1
by the following equation, which is easier to use than Equa- >
>
>
>
tion 17. >
> h2
>
> 0 < < 0:5
< h1
>
>  3  2  
>
> : h2 : h2 : h2
>
> 340 5  609 8 þ 214 1 þ 55:2
>
> h1 h1 h1
110 >
>
>
>
>
> h2
100 >
: 0:5 , < 1:0
22: h1
90
Yield stress/specific gravity: Pa

80

70
where SG is the specific gravity of the tested sample. Note
60 that for both ranges of h2 /h1 , the correlation coefficient,
50 R2 ¼ 0.99.

40
Conclusion
30 In this paper, a method is presented to evaluate the filling and
20 flowing abilities of SCC by developing a relationship between
the yield stress and the L-box blocking ratio. The fluid flow
10
inside the L-box apparatus is modelled using a dam break theory
0 for a viscoplastic non-Newtonian fluid whose shear behaviour
0 0·1 0·2 0·3 0·4 0·5 0·6 0·7 0·8 0·9 1·0
can be expressed by a Bingham model. The previous dam break
h2 /h1
model is modified and a relation between the yield stress and the
Figure 9. General relationship between L-box blocking ratio and L-box blocking ratio is derived. Experimental work is also
ratio of yield stress/specific gravity conducted using seven SCC mixtures with different L-box
blocking ratios.

183
Magazine of Concrete Research Evaluation of SCC yield stress from L-box
Volume 66 Issue 4 test using the dam break model
Chamani, Hosseinpour, Mostofinejad and
Esmaeilkhanian

Good correlation between the predicted and experimental arrested Hwang SD, Khatib R, Lee HK, Lee SH and Khayat KH (2012)
profiles shows that applying the concept of the dam break Optimization of steam-curing regime for high-strength,
phenomenon to the case of fluid flow analysis in the L-box test is self-consolidating concrete for precast, prestressed concrete
an appropriate procedure. Comparison between the proposed applications. PCI Journal 57(3): 96–107.
model predictions and the experimental SCC data proved that the Lashkarbolouk H, Chamani MR, Halabian AM and Pishehvar AR
proposed model can also be used in the case of fresh SCC. By (2013) Viscosity evaluation of SCC based on flow simulation
using this model and evaluating the value of yield stress, it is in L-box test. Magazine of Concrete Research 65(6): 365–
possible to predict the workability behaviour of SCC during the 376.
cast. Figure 9 or Equation 22 can be used as a simple way to Matson GP and Hogg AJ (2007) Two-dimensional dam break
assess the yield stress of SCC for on-site works. The only data flows of Herschel–Bulkley fluids: The approach to the
required for using Figure 9 or Equation 22 to evaluate the yield arrested state. Journal of Non-Newtonian Fluid Mechanics
stress are the L-box test results, which means this method is 142(1–3): 79–94.
cheap and easy to apply, in comparison with using an expensive Murata J (1984) Flow and deformation of fresh concrete.
and hard-to-work rheometer. Materials and Structures 17(2): 117–129.
Murata N and Shitayama Y (1976) The deformation of fresh
Acknowledgements concrete under static load. Annual Report of Cement
The authors would like to express their gratitude to the staff of Technology 30: 270–273 (in Japanese).
the Concrete Technology Laboratory of Isfahan University of Nguyen TLH, Roussel N and Coussot P (2006) Correlation
Technology (IUT), Ebrahimi and Rahnama, for their dedicated between L-box test and rheological parameters of a
help and support during the experimental work. homogeneous yield stress fluid. Cement and Concrete
Research 36(10): 1789–1796.
REFERENCES Nsom B (2002) Horizontal viscous dam-break flow: experiments
Balmforth NJ and Craster RV (1999) A consistent thin-layer and theory. Journal of Hydraulic Engineering 128(5): 543–
theory for Bingham plastics. Journal of Non-Newtonian Fluid 546.
Mechanics 84(1): 65–81. Petersson O and Billberg P (1999) Investigation on blocking of
Balmforth NJ, Craster RV, Perona P, Rust AC and Sassi R (2007) self compacting concrete with different maximum aggregate
Viscoplastic dam breaks and the Bostwick consistometer. size and use of viscosity agent instead of filler. Proceedings
Journal of Non-Newtonian Fluid Mechanics 142(1–3): of the 1st International RILEM Symposium on
63–78. Self-Compacting Concrete, pp. 333–344.
BIBM, CEMBUREAU, EFCA, EFNARC and ERMCO (2005) The Precast/Prestressed Concrete Institute (2003) Interim Guidelines
European Guidelines for Self-compacting Concrete: for the Use of Self-Consolidating Concrete in Precast/
Specification, Production and Use. Self-compacting Concrete Prestressed Concrete Institute Member Plants. Precast/
European Project Group, See www.efnarc.org/pdf/ Prestressed Concrete Institute, Chicago, IL, USA, Report
sccguidelinesmay2005.pdf (accessed 13/12/2013). TR-6-03, p. 122.
Chanson H, Jarny S and Coussot P (2006) Dam break wave of Roussel N (2006) Correlation between yield stress and slump:
thixotropic fluid. Journal of Hydraulic Engineering 132(3): Comparison between numerical simulations and concrete
280–293. rheometers results. Materials and Structures 39(4):
Dufour F and Pijaudier-Cabot G (2005) Numerical modelling of 501–509.
concrete flow: Homogeneous approach. International Journal Roussel N and Coussot P (2005) ‘Fifty-cent rheometer’ for yield
for Numerical Analytical Methods in Geomechanics 29(4): stress measurements: From slump to spreading flow. Journal
395–416. of Rheology 49(3): 705–718.
Emborg M (1999) Rheology tests for self-compacting-concrete: Roussel N, Geiker MR, Dufour F, Thrane LN and Szabo P
how useful are they for the design of concrete mix for (2007) Computational modeling of concrete flow: General
full scale production. Proceedings of the 1st International overview. Cement and Concrete Research 37(9): 1298–
RILEM Symposium on Self-Compacting Concrete, 1307.
pp. 95–105. Schowalter WR and Christensen G (1998) Toward a
Ferraris CF, Brower LE, Beaupre D et al. (2004) Comparison of rationalization of the slump test for fresh concrete:
Concrete Rheometers: International Tests at MB. National Comparisons of calculations and experiments. Journal of
Institute of Standards and Technology, Gaithersburg, MD, Rheology 42(4): 865–870.
USA, Internal Report, NISTIR 7154. Sonebi M (2003) Medium strength self-compacting concrete
Goodier CI (2003) Development of self-compacting concrete. containing fly ash: Modelling using factorial
Magazine of Concrete Research 156(4): 405–414. experimental plans. Cement and Concrete Research
Hogg AJ and Pritchard D (2004) The effects of hydraulic 34(7): 1199–1208.
resistance on dam-break and other shallow inertial flows. Thrane LN, Szabo P, Geiker M, Glavind M and Stang H (2004)
Journal of Fluid Mechanics 501(3): 179–212. Simulation of the test method ‘L-Box’ for self-compacting

184
Magazine of Concrete Research Evaluation of SCC yield stress from L-box
Volume 66 Issue 4 test using the dam break model
Chamani, Hosseinpour, Mostofinejad and
Esmaeilkhanian

concrete. Annual Transactions of the Nordic Rheology Society Zhou JG, Causon DM, Mingham CG and Ingram DM
12: 47–54. (2004) Numerical prediction of dam-break flows
Whitham GB (1955) The effects of hydraulic resistance in the in general geometries with complex bed topography.
dambreak problem. Proceedings of the Royal Society of Journal of Hydraulic Engineering 130(4):
London 227(Series A): 399–407. 332–340.

WHAT DO YOU THINK?


To discuss this paper, please submit up to 500 words to
the editor at www.editorialmanager.com/macr by 1
August 2014. Your contribution will be forwarded to the
author(s) for a reply and, if considered appropriate by
the editorial panel, will be published as a discussion in a
future issue of the journal.

185

You might also like