Vous êtes sur la page 1sur 9

Materials and Structures (2006) 39:501–509

DOI 10.1617/s11527-005-9035-2

Correlation between yield stress and slump: Comparison


between numerical simulations and concrete rheometers
results
N. Roussel

Received: 29 June 2005 / Accepted: 21 July 2005



C RILEM 2006

Abstract Results of numerical flow simulations for 1. Introduction


two slump geometries, the ASTM Abrams cone and a
paste cone, are presented. These results are compared Fresh cementitious materials, as many materials in in-
to experimental results in the case of a cone filled with dustry or nature, behave as fluids with a yield stress,
cement pastes in order to validate the proposed nu- which is the minimum stress for irreversible deforma-
merical method and the chosen boundary conditions. tion and flow to occur. This yield stress is an unique
The correlation between slump and yield stress ob- material property and may, in the case of cement pastes
tained numerically for the ASTM Abrams cone is then (i.e. fine particles), be measured using conventional
compared to the experimental correlations obtained rheological tools. For example, Couette Viscometer [1]
by testing concrete with different rheometers dur- or parallel plates rheometer [2] are used in the labo-
ing comparative studies that were organized at LCPC ratory to measure the yield stress value. In the case
Nantes (France) in 2000 and MB Cleveland (USA) in of concretes containing coarse aggregates, large scales
2003. rheometers had to be developed (the BTRHEOM [3],
the BML [4] or the two-point test [5]). Even if, in situ,
Résumé Des résultats de simulations numériques simpler and cheaper tests such as the slump test [6]
d’écoulements sont présentés pour deux géométries are still often preferred, these apparatus represent a big
d’essais, le cône d’Abrams et le mini cône. La valida- step forward in the field of concrete science. However,
tion de la méthode numérique et des choix portant sur there still exists a discrepancy between the various con-
les conditions aux limites de l’écoulement est réalisée crete rheometers, [7,8]. These apparatus give the same
par comparaison avec des résultats expérimentaux rheological classification of materials but they do not
sur pâtes de ciment dans le cas du mini cône. La give the same absolute values of the rheological pa-
corrélation obtenue numériquement dans le cas du rameters. On the other hand, the slump test does not
cône d’Abrams est ensuite comparée aux corrélations give any value of a physical parameter at all. Its result
expérimentales obtenues pour différents rhéomètres à can not be expressed in physical rheological units but
bétons lors des campagnes de comparaisons orga- it has also proved through the years to be able to clas-
nisées au LCPC Nantes (France) en 2000 et à MB sify different materials in terms of their ability to be
Cleveland (USA) en 2003. cast.
The aim of this paper is to propose a theoretical cor-
relation between slump and yield stress and to compare
N. Roussel it to the experimental correlations obtained in the two
LCPC Paris, France rheometers comparison campaign [7,8]. In the first part
502 Materials and Structures (2006) 39:501–509

2. Literature study

2.1. Relations between slump and yield stress

Several attempts can be found in the literature in or-


der to relate slump to yield stress. Murata [11] and
Schowalter and Christensen [12] wrote a relation be-
tween slump and yield stress by assuming that the cone
could be divided into two parts. In the upper part, the
shear stress does not reach the yield stress and no flow
occurs. In the lower part of the cone, the shear stress
induced by the self-weight of the material is higher
Fig. 1 Initial cone shape and cylindrical coordinates. than the yield stress and flow occurs. The height of the
flowing lower part decreases until the shear stress in
this zone becomes equal to the yield stress, after which
Table 1 Cone geometries the flow stops. Schowalter and Christensen [12] wrote
Cone ASTM Abrams cone paste cone
a relation between the final total height of the cone
and the yield stress that did not depend on the mould
H0 (mm) 300 50 geometry. This relation or similar ones were success-
Rmin (mm) 200 35 fully validated by Clayton and co-workers [13] or Saak
Rmax (mm) 100 50
and co-workers [14] in the case of cylindrical moulds.
However, in the case of conical moulds, a discrepancy
between predicted and measured slumps was systemat-
ically obtained. In the above studies, the experimental
of the present work, results of numerical simulations results suggested that these relations and the fact that
are presented for the ASTM Abrams cone [6] and a they did not depend on the mould geometry are valid
paste cone test [9]. The geometries of these molds are for high slumps (i.e. low yield stress) (Clayton and co-
given in Figure 1 and Table 1. It should be noted that workers [12], Pashias and co-workers [15], Saak and
we are not using here what is called mini slump but co-workers [13]). In the case of the ASTM Abrams
the “ASTM C230 flow table conical mold” This cone cone, Hu and co-workers [16] gave a semi empirical
is used usually in conjunction with the flow table. It correlation between the yield stress τ0 (Pa) measured
is not the case here. The originality of the present ap- using the BTRHEOM, the density ρ and the slump
proach lies in using a rigorous three dimensional ex- (mm):
pression of the behaviour law and plasticity criterion
(read 2.2). (τ0 − 212)
s = 300 − 347 (1)
The numerical results are first compared to the ρ
paste cone test results and standard Vane test mea-
surements [10] on cement pastes. The good agree- 2.2. Yield criterion
ment between the obtained numerical results and ex-
perimental values over a wide range of yield stress It should be noted that all the above analytical ap-
confirms the validity of the numerical approach and proaches involve a unidimensional expression of the
allows us to use the numerical results obtained for yield criterion and behaviour law: flow occurs or stops
ASTM Abrams cone to predict slump in terms of yield when the shear stress becomes higher or lower than the
stress. yield stress. The other components of the stress ten-
In the last part of this paper, the numerical correla- sor are not taken into account when writing a scalar
tion between slump and yield stress is compared to the yield criterion. This greatly simplifies the analysis of
correlation between slump and yield stress obtained for the flow but is valid only if the flow is dominated by
three concrete rheometers: the BTRHEOM, the BML shear stresses (i.e. the diagonal terms of the devia-
and the two-point test. toric stress tensor can be neglected compared to the
Materials and Structures (2006) 39:501–509 503

shear stress). In fact, this assumption is true only in literature, a Bingham model is, in this case, a possi-
the ideal two-dimensional case studied by Coussot and ble rheological model. On the other hand, when the
co-workers [17] (very high slumps). This simplifica- above conditions are not fulfilled, it is then needed to
tion is similar to the use of the lubrication theory in take into account the presence of the particles [23, 24,
the squeezing flow literature [18]. The squeezing flow 25]. This type of approach is a lot more complex but
test is a simple compression test carried out on cylin- is the only suitable technique when confined flows or
drical samples with reduced slenderness. The appa- flows between steel bars are studied. In this paper, we
ratus consists in two coaxial circular parallel plates, will only use an homogeneous approach. As a con-
without any rotation. By using the lubrication theory, sequence, the validity of the presented results will be
only the shear stress is considered and the flow can be doubtful when the smallest characteristic dimension of
easily studied but this theory generates what is called the flow becomes lower than 5 times the biggest parti-
the “squeezing flow paradox”. On the plane of sym- cles or, in other words, when the slump is higher than
metry between the two plates, the shear stress equals 25 cm.
zero. The unidimensional yield criterion is of course Other authors have previously developed homoge-
not fulfilled and the material should flow as a solid neous numerical simulations of the slump flow. Tani-
body (i.e. “plug flow”). But Lipscomb and Denn [19] gawa and Mori [26] developed an innovative visco-
have demonstrated that plug regions can not exist in plastic finite element analysis introducing a frictional
squeezing flow of yield stress fluids. Clearly a solid interface law at the base of the slumping cone. They
body can not move radially outward with a velocity calculated the slump in terms of the yield stress but,
that increases with the radial coordinate as the con- as they did not have any experimental way to measure
servation equations demand. Wilson [20] has pointed the rheological parameters of concrete, they did not
out that this paradoxical “yielded/unyielded” region is compare their results to experimental measurements.
due to the neglect of the extensional stresses close to Later, Schowalter and Christensen [12] compared their
the centre plane where they are of a higher order than analytical prediction to Tanigawa and Mori numeri-
the shear stress. A proper three-dimensional criterion is cal results and found a good agreement. It should be
thereafter needed to avoid this paradox and Adams and noted that both predictions were based on a unidimen-
co-workers [21] can be quoted “a comprehensive yield sional plasticity criterion. Hu [27] assumed that the
criterion is one which is based upon a combination of shape of the deposit stayed conical and calculated the
all the acting components of the stress”. In this paper, state of stress using an elastoplastic finite element anal-
a 3D yield criterion will be used as it is the only way to ysis. Once again, a unidimensional yield criterion was
obtain correct quantitative results when the flow is not considered.
a purely shearing flow. Recently, Chamberlain and co-workers [28] calcu-
lated rigorously stresses in a purely plastic cylindrical
2.3. Numerical simulations sample using either von Mises or Tresca plasticity crite-
rion in order to determine the height of incipient failure,
Several authors have also developed numerical simu- which is the height of material required to just initiate
lations of this free surface stoppage flow. Two types flow for a given cylinder radius. This is equivalent to
of method are available when trying to simulate the calculating the critical yield stress for which flow does
flow of a rough suspension such as concrete. The first start or does not for a given cylindrical geometry. They
method consists in only considering an homogeneous found a discrepancy between the mono dimensional
fluid whereas the second one takes into account the approximation written by Schowalter and Chistensen
presence of the particles. The homogeneous approach [12] and their three dimensional approach. They also
is easier to implement but is valid only when the small- studied the dependency of the critical yield stress on
est characteristic dimension (thickness of the flow, size the cylinder radius.
of the mould, spacing between bars) of the flow is high
compared to the size of the biggest particle (5 times 2.4. Behaviour at the interface
larger for example in [22]) and when the material stays
homogeneous (which seems to be the case during a Most analysis in the literature are carried out assum-
slump test on a standard concrete). As accepted in ing sticky flow at the base of the deposit. If we go into
504 Materials and Structures (2006) 39:501–509

further details, the following distinction could be made. assumption, the influence of the lifting speed of the
In the case of fluid concretes (i.e. low yield stress, high mould (that depends on the operator) or of the plastic
slumps), this assumption is probably valid as these con- viscosity (if a Bingham model is chosen to describe
cretes behaves just like suspensions but, in the case of the material behaviour) on the measured slump have
high yield stresses, this assumption should be ques- not been studied: the velocity of the flow and its ki-
tioned. A concrete with a high yield stress may be ob- netic energy are not taken into account and the final
tained by two different trends in the mix proportioning. shape is calculated as a quasi static state assuming it is
On one hand, the amount of cement or fine particles reached slowly enough. In other words, what happens
may be high. The colloidal force network that can be before stoppage of the flow does not influence the shape
built between these fine particles increases the yield at stoppage. Tatersall and Banfill [29] experimentally
stress of the mixture. The concrete is similar to a dense concluded that the slump of fresh concrete is indeed
fine suspension and, in this case, the assumption of a highly correlated with yield stress but is not signifi-
sticking flow is also valid as experimentally obtained by cantly affected by the plastic viscosity. This conclusion
Pashias [15]. On the other hand, the amount of coarse was also reached by Murata [11]. Let us check here that
particles may be high. The behaviour of the obtained neglecting inertia effects is a correct assumption on a
concrete becomes closer to the behaviour of a cohesive theoretical point of view: let us roughly compare the
granular material. In this case, as for a granular ma- typical inertia stress (I = ρV 2 ) to the material yield
terial, the behaviour at the interface may be frictional. stress. For a small slump the flow duration is of the or-
Because of the uncertainty in the behaviour at the inter- der of magnitude of 1s for a slump of the order of 10 cm.
face, the results obtained with a sticky flow assumption We thus have I ≈ 20 Pa, a value much smaller than the
should be considered with care in the case of high yield material yield stress in that case (typically larger than
stress concretes as the validity of the assumption of a several hundreds of Pa). For a large slump the flow du-
sticking flow depends on the tested material aspect and ration is of the order of magnitude of 2–3 s for a slump
mix proportioning. of the order of 20 cm. We thus have again I ≈ 20 Pa, a
More recently, Chamberlain and co-workers [28] value once again much smaller than the material yield
studied the influence of the plate roughness on the crit- stress in that case (typically larger than several tenths
ical yield stress assuming a coulomb type friction law of Pascals). This confirms the experimental deductions
at the interface involving a friction coefficient μ equal of [29] and [11].
to zero for a perfect slip case. They showed that, above As the final shape only depends on the yield stress
a critical value μc depending on the cylinder radius, and as the final shape is the only point of interest for
there was no influence of the friction parameter on the the engineer, we will only study this aspect of the flow
height of incipient failure (or critical yield stress) and in his paper.
that the interface could then be considered as perfectly
rough. They also showed that the difference between
height of incipient failure predicted for the perfect slip 3. Numerical simulations
(μ = 0) and perfectly rough (μc = 0) cases increased
from zero for small radii to 18% for a radius equal to The cylindrical frame of reference (O, r, θ, z) is shown
2τ0 /ρg. In the case of typical concrete, this reference on Figure 1. p is the pressure,  is the stress tensor
radius becomes 0.16 m (yield stress of the order of mag- and  (d) is the deviatoric stress tensor. s is the slump.
nitude of 2000 Pa and density around 2500 kg/m3 ). The s is the dimensionless slump and τ0 the dimensionless
radius of the ASTM Abrams cone being equal to 0.1 yield stress as defined by Showalter and Christensen
m, the error made while neglecting the friction at the [12].
interface in the case of concrete should be lower than
18% for low slumps and high yield stress. s = s/H0

2.5. Effect of the inertia τ0 = t0 /ρg H0

In all the above analytical studies, inertia (or dynamic) It can already be noted here that, although this scal-
effects are neglected. Because of this simplifying ing was suitable for Schowalter results, it does not apply
Materials and Structures (2006) 39:501–509 505

so well to the present results as they appear to depend


on the cone geometry. We chose to use this scaling
however as it allows us to plot the predicted results for
the two studied geometries on the same figure.
As already stated, in this work, we wanted to im-
plement a proper three dimensional yield criterion (see
2.2). A 3D Bingham model was thus used to describe
the tested fluid behaviour. Moreover, in order to avoid
Fig. 2 Calculation grid and initial hydrostatic pressure in the
the undetermination of the strain state when the yield
cement paste before lifting of the paste cone mould. Note that
criterion is not fulfilled, the material was assumed to be- the horizontal to vertical scale ratio is not equal to 1.
have as an incompressible elastic solid up to the yield
stress, beyond which it behaves as a Bingham fluid. The initial pressure is hydrostatic and the speed at
Other methods exist to avoid this undetermination such which the mould is numerically lifted is infinite as
as the biviscosity model introduced by O’Donovan and the mould simply disappeared at t = 0 s. This “bru-
Tanner [30] or the exponential model proposed by Pa- tal” lifting could have generated non negligible iner-
panastasiou [31] but, as whether or not flow starts is tia effects taken into account by the code. The final
concerned, it was of particular interest to use a type of predicted shape could have been affected by the high
model where a situation of no flow at all (just deforma- mean kinetic energy of the flow that is not, as in real-
tion) can exist. This was the case here. The computa- ity, dissipated by the lifting of the mould. To solve this
R
tional fluid mechanics code Flow3D [32] was chosen problem, we chose to study numerical fluids with high
R
to solve the fluid mechanics equation. Flow3D is a plastic viscosity. This indeed generated slow flows, in
general purpose computer program with many capa- which inertia effects were negligible no matter the lift-
bilities. Using input data, the user can select different ing speed. The calculations were thus carried out with
physical options to represent a wide variety of fluid a plastic viscosity equal to 300 Pa.s for concrete in the
flow phenomena. The program can be operated in sev- ASTM Abrams cone and 10 Pa.s for cement paste in the
eral modes corresponding to different limiting cases paste cone (far above the traditional plastic viscosities
of the general fluid equations. In the opinion of the of these materials). It may be reminded here that, as
present author, FLOW3D is very user friendly when the final shape only depends on the yield stress when
dealing with otherwise complex free surface transient inertia effects are negligible which is the case of the
flow of non Newtonian fluids. real ASTM Abrams cone (read 2.5), the values chosen
The invariant generalization of a Bingham fluid used for the plastic viscosity do not matter. The obtained nu-
here is the one proposed by Oldroyd [33] based on the merical results are plotted on Figure 3. Examples of
three dimensional von Mises yield criterion: two dimensional predicted shapes are shown on Figure
4 for the ASTM Abrams cone. The presence of an un-
  1/2  
 1  yielded zone (usual in this type of simulation) can be
 (d) = τ0  d : d  + η d, 1  (d) :  (d) ≥ τ 2
2  2 0 noted.
The calculated values of the slump confirm the fact
where d is the strain rate tensor and η the plastic vis- that slump depends of course on yield stress and den-
cosity. When the flow is dominated by shear stress and sity but also on the tested volume and initial height.
shear rate, the previous relation simplifies to the famous Indeed, the predicted dimensionless slump is differ-
Bingham scalar model: ent for the two cone geometries. As already stated, the
scaling suggested by Schowalter and Christensen [12],
τ = τ0 + η · γ , τ ≥ τ 0 although suitable for his own experimental results, does
not apply here.

The generated grid in the case of the paste cone is 4. Comparison with experimental results
shown in Figure 2. The one used for the ASTM Abrams
cone is similar. The cell size is smaller in the zones Measurements were carried out using the paste cone
where the shearing is the highest. geometry given in Table 1 while the yield stress
506 Materials and Structures (2006) 39:501–509

Fig. 3 Dimensionless slump in terms of dimensionless yield


stress for the ASTM Abrams cone and paste cone. Both numerical
predictions and experimental results are plotted for the paste cone
test.

was measured using the Vane procedure described by


N’Guyen and Boger [10]. It can be noted that two dif-
ferent Vane test geometries were used on the HAAKE

R
ViscoTester VT550 to measure the yield stresses of
the tested mixtures. Indeed, as the yield stresses of the
studied mixtures varied from 0.6 Pa to 300 Pa, it was
necessary to change the geometry of the rotating tool
in order to measure an acceptable torque with a suffi-
cient precision. Several types of cements and mix pro-
portioning were tested but the mixing procedure was
Fig. 4 Examples of obtained shapes for the ASTM Abrams
always the same: the dry ingredients are first mixed for cone (a) yield stress = 2600 Pa (b) yield stress = 2000 Pa. Den-
2 min at the lowest mixer rpm setting (260 rpm), then sity = 2500 kg/m3 for both simulations.
the fluids are added and all are mixed for 2 min. The
mixer is then stopped to scrape its edges. A higher ro-
tation speed (700 rpm) is applied for 15 min followed time between the filling of the mould and its lifting to
by a 15 min final mixing phase at the lowest mixer rpm prevent any thixotropic effect from increasing the yield
setting. This chosen mixing procedure had two advan- stress. The mould was slowly lifted in order to elimi-
tages: it ensured good particle dispersions in the fluid nate any inertial effects that could take place. But, as
phase and gave any chemical binders enough time to different operators realized these tests, this slow lift-
act. ing speed was probably not constant. The spread and
The cone experimental results were obtained by var- height measurements were done after a 2 min waiting
ious users over a year period. The plate surface was the time. For each test, two perpendicular diameters and
same for all the tests. The maximum particle size of the maximum thickness of the collapsed sample were
these materials was 100 μm. It was largely smaller measured. The obtained results are plotted on Figure 3.
than the characteristic size of the mould. It was also The agreement between the numerical simulations and
smaller than the minimum height measured after col- the experimental results is very good. The experimental
lapsing of the sample (4 mm). There was no waiting critical dimensionless yield stress seems to be around
Materials and Structures (2006) 39:501–509 507

0.58. It is however difficult to precisely check the value be emphasized that we do not have at the moment any
of this critical yield stress as it is impossible not to de- way to measure the real value of the yield stress. It has
form the cone while lifting the mould. The good agree- been proven in [7] and [8] that the rheometers did not
ment between numerical and experimental results on give the same results. This does not mean that one is
the entire studied range of yield stress confirm the fact correct and the others are wrong, this could mean that
that no sliding occurs at the base of the deposit and that they are all wrong or at least that none of them is cor-
the use of a proper three-dimensional plastic criterion rect on the entire range of yield stress tested. Thus, the
allows a correct quantitative numerical prediction of yet unanswered question is “What is the real correla-
the final slump. tion between yield stress and slump?” In the frame of
this paper, this means that we do not have any way to
conclude on the validity on the simulations in the case
of concrete unless by reminding that they proved to be
5. Astm abrams cone and concrete rheometers
very efficient to simulate the flow of cement pastes and
that all the theoretical assumptions to carry valid simu-
The validation of the proposed numerical method ob-
lations in the case of concrete are fulfilled in the range
tained in the previous section allowed us to use the
of 5 to 25 cm slump.
proposed numerical method to predict the ASTM cone
It should be emphasized that, as the behaviour at the
slump in terms of the tested concrete yield stress. Apart
base of the deposit is unknown for high yield stresses
from the possible sliding at the interface that may occur
(possible sliding, read Section 2.4.), the validity of the
in the case of high yield stress concretes (see §2.4), the
obtained numerical results should be limited to slumps
flow in the case of the ASTM Abrams cone is identi-
higher than 5 cm. Moreover, for slumps higher than
cal. The numerical correlation between slump in mm
25 cm, the thickness of the flowing layer of concrete
and the ratio yield stress/density is plotted on Figure 5
becomes of the same order as the size of the biggest
along with the experimental results obtained in the two
aggregate and the homogeneous fluid mechanics ap-
rheometers comparison campaign. MBT-LCPC com-
proach proposed here is not valid any more (read Sec-
parison carried out at LCPC (Nantes, France) in 2000
tion 2.3.).
[7] and at MB (Cleveland, USA) in 2003 [8]. It has to
However, from the numerical predicted results, a
simple linear approximation may be written for slumps
between 5 cm and 25 cm.
τ0
s = 25.5 − 17.6 (2)
ρ

It is not the aim of the present work to conclude about


the efficiency of any concrete rheometer compared to
another. However, it seems that, in the 5–25 cm slump
range, the BTRheom correlation between measured
yield stress and measured slump is the closest to the
numerical correlation. For the lowest yield stresses and
the highest slumps, it seems that the BTRheom correla-
tion overestimates the yield stress. This may be linked
to the fact that the sliding assumption at the peripheral
interface needed in the analysis of the BTRheom data is
correct in the traditional concrete range (see §2.4) but,
when the yield stress becomes lower, the behaviour
at the interface becomes sticky and the shear stress
at the interface may not be neglected any more com-
Fig. 5 Yield stress-slump correlations. Experimental results for pared to the measured yield stress. The spontaneous
various rheometers and numerical correlations. formation of a limit layer (made up with water and fine
508 Materials and Structures (2006) 39:501–509

elements) limiting the friction may not occur for low presented. The obtained numerical results have been
yield stress concrete that behave just like traditional first compared to paste cone test results and standard
suspensions and the flow pattern may differ from the Vane test measurements on cement pastes. The good
theoretical one used to calculate the yield stress from agreement between the obtained numerical results and
the torque measurements. During the comparison cam- experimental values over a wide range of yield stress
paign at MB (Cleveland, USA) [8], this specificity of have confirmed the validity of the proposed numerical
the BTRHEOM was already spotted when comparing approach and its associated boundary conditions and
the various rheometers by measuring purely viscous allowed us to use the numerical results obtained for
oils. Concerning the two other apparatus, their main ASTM cone to predict slump in terms of yield stress. A
potential drawbacks may be reminded here: simple numerical correlation between slump and yield
stress has then been proposed.
- The ratio between the gap between the BML cylin- In the last part of this paper, this numerical correla-
ders and the maximum particle size is small. In the tion has been compared to the experimental correlation
geometry used in [8], the gap was only 45 mm, thus between slump and yield stress obtained for three con-
only allowing, on a theoretical point of view, the test- crete rheometers: the BTRHEOM, the BML and the
ing of concretes with particles smaller than 9 mm two-point test during comparison campaigns held in
[22]. When it is not the case, the consequence, if 2000 and 2003. It seems that, in the range of traditional
there is one, is very difficult to predict. However, in concretes (slump between 50 mm and 250 mm), there
the case of fluid concretes such as Self Compacting is a good agreement between the BTRHEOM measure-
Concretes (SCC) often prepared with smaller parti- ment and the theoretical prediction.
cles, the BML could prove the most suitable of the
three apparatus in the opinion of the present author.
It can be noted that the BML correlation between
yield stress and slumps gets closer to the numerical
predictions obtained here when the yield stress is of References
the order of several hundreds Pa.
- Because of the complexity of the flow in the two- 1. Shaughnessy R, Clark PE (1988) The rheological behaviour
of fresh cement pastes. Cem Concr Res, 18:327–341.
points test, the stress (or velocity) field can not be rig- 2. Nehdi M, Rahman M-A (2004) Estimating rheological prop-
orously calculated. The rheological parameters are erties of cement pastes using various rheological models for
simply extrapolated from the global measurements different test geometry, gap and surface friction. Cement
without taking into account the fact that some non Concrete Res., 34:1993–2007.
3. De Larrard F, Hu C (1996) The rheology of fresh high-
flowing zones, the sizes of which depend on the ro- performance concrete. Cem Conc Res, 26(2):283–294.
tation speed and of the tested material itself, might 4. Operating manual (2000) the BML viscometer, the viscome-
appear in the sample closed to the rotating impeller ter 4, Con Tec.
or to the external walls. However, in the case of 5. Tatersall GH, Bloomer SJ (1979) further development of the
two-point test for workability and extension of its range.
fluid concretes or SCC, this phenomenon may be Magazine of Concrete Research 31:202–210.
neglectible. As for the BML, the two points test cor- 6. ASTM Designation C-143-90 (1996) Standard test method
relation between yield stress and slumps gets closer for slump of hydraulic cement concrete. Annual Book of
to the numerical predictions obtained here when the ASTM Standards, 04.01, Am. Soc. Test. Mat., Easton, MD,
pp. 85–87.
yield stress decreases below the order of several hun- 7. Ferraris CF, Brower LE editors (2001) Comparison of
dreds Pa. concrete rheometers: International tests at LCPC (Nantes,
France) in October, 2000. National Institute of Standards
and Technology Interagency Report (NISTIR) 6819.
8. Ferraris CF, Brower LE editors (2004) Comparison of con-
6. Conclusion crete rheometers: International tests at MB (Cleveland OH,
USA) in May, 2003. National Institute of Standards and
In the first part of the present work, results of numer- Technology Interagency Report (NISTIR) 7154.
9. ASTM Designation C230/C230M-03, Standard Specifica-
ical simulations using a three dimensional expression tion for Flow Table for Use in Tests of Hydraulic Cement.
of the behaviour law and plasticity criterion for the Annual Book of ASTM Standards, 04.01, Am. Soc. Test.
ASTM Abrams cone and a paste cone test have been Mat., Easton, MD (2004).
Materials and Structures (2006) 39:501–509 509

10. Nguyen QD, Boger DV (1985) Direct yield stress measure- 22. Coussot P, Ancey C (1999) Rhéophysique des pâtes et des
ment with the vane method. J. Rheol., 29:335–347. suspensions, EDP Sciences, (in French).
11. Murata J (1984) Flow and deformation of fresh concrete. 23. Petersson O (2003) Simulation of Self-Compacting
Materials and Structures RILEM, 98:117–129. Concrete- Laboratory experiments and numerical modelling
12. Schowalter WR, Christensen G (1998) Toward a ratio- of testing method, Jring and L-Box test’, Proceedings of the
nalization of the slump test for fresh concrete: compar- 3rd international RILEM Symposium on Self-Compacting
isons of calculations and experiments. J. Rheol., 42(4):865– Concrete, RILEM PRO33 Reykjavik, Iceland, 202–207.
870. 24. Martys NS (2005) Study of a dissipative particle dynamics
13. Clayton S, Grice TG, Boger DV (2003) Analysis of the slump based approach for modeling suspensions. Journal of Rhe-
test for on-site yield stress measurement of mineral suspen- ology 49(2):401–424.
sions. Int. J. Miner. Process., 70:53–21. 25. Wallevik JE (2003) Rheology of particle suspensions; Fresh
14. Saak AW, Jennings HM, Shah SP (2004) A generalized ap- Concrete, Mortar and Cement Pastes with Various Types of
proach for the determination of yield stress by slump and Lignosulfonates. Ph.D. Thesis, Department of Structural En-
slump flow. Cem Concr Res 34:363–371. gineering, The Norwegian University of Science and Tech-
15. Pashias N, Boger DV, Summers J, Glenister DJ (1996) a nology.
fifty cent rheometer for yield stress measurements. J. Rheol. 26. Tanigawa Y, Mori H (1989) Analytical study on deforma-
40(6):1179–1189. tion of fresh concrete, Journal of Engineering Mechanics
16. Hu C, de Larrard F, Sedran T, Boulay C, Bosc F, Deflo- 115(3):493–508.
renne F (1996) Validation of BTRHEOM, the new rheometer 27. Hu C (1995) Rheologie des bétons fluids (rheology of fluid
for soft-to-fluid concrete. Materials and Structures, RILEM, concretes), thèse de doctorat de l’ENPC (PhD Thesis) France
29(194):620–631. (In French).
17. Coussot P, Proust S, Ancey C (1996) Rheological interpre- 28. Chamberlain JA, Clayton S, Landman KA, Sader JE (2003)
tation of deposits of yield stress fluids. Journal of Non- Experimental validation of incipient failure of yield stress
Newtonian Fluid Mechanics 66(1):55–70. materials under gravitational loading, Journal of Rheology,
18. Covey GH, Stanmore BR (1981). Use of the parallel 47(6):1317–1329.
plate plastometer for the characterisation of viscous fluids 29. Tatersall GH, Banfill PGF (1983) The Rheology of Fresh
with a yield stress, J. Non-Newtonian Fluid Mech. 8:249– Concrete, Pitman, London.
260. 30. O’Donovan EJ, Tanner RI (1984) Numerical study of the
19. Lipscomb GG, Denn MM (1984) Flow of Bingham fluids in Bingham squeeze film problem. J. Non-Newtonian Fluid
complex geometries. J. Non-Newtonian Fluid Mech. 14:337– Mech, 15:75–83.
346. 31. Papanastasiou TC (1987) Flows of Materials with yield. J.
20. Wilson SDR (1993) Squeezing flow of a Bingham material. Rheol., 31:385–404.
J. Non-Newtonian Fluid Mech. 47:211–219. 32. Flow3D version 8.1, User’s manual, volume 1, 2004.
21. Adams MJ, Aydin I, Briscoe BJ, Sinha SK (1997) A finite 33. Oldroyd JG (1947) A rational formulation of the equations of
element analysis of the squeeze flow of an elasto-viscoplastic plastic flow for a Bingham solid. Proc. Camb. Philos. Soc.,
paste material. J. Non-Newtonian Fluid Mech. 71:41–57. 43:100–105.