Pumping of Concrete and Mortar
Pumping of Concrete and Mortar
Pumping of Concrete and Mortar
no
SINTEF Building and Infrastructure Stefan Jacobsen, Jon Hvard Mork, Siaw Foon Lee , Lars Haugan
Keywords:
Materials technology, concrete, Pumpability, Rheology, Rheology, Flow
ISSN 18911978 (online)
ISBN 9788253610696 (pdf)
Address:
Tel:
Fax:
Forskningsveien 3 B
POBox 124 Blindern
N-0314 OSLO
+47 22 96 55 55
+47 22 69 94 38 and 22 96 55 08
www.sintef.no/byggforsk
www.coinweb.no
NTNU
Contact: Terje Kanstad
Email: [email protected]
Tel: +47 73594700
Spenncon AS
Contact: Ingrid Dahl Hovland
Email: [email protected]
Tel: +47 67573900
Rescon Mapei AS
Contact: Trond Hagerud
Email: [email protected]
Tel: +47 69972000
maxit Group AB
Contact: Geir Norden
Email: [email protected]
Tel: +47 22887700
Unicon AS
Contact: Stein Tosterud
Email: [email protected]
Tel: +47 22309035
Norcem AS
Contact: Terje Rnning
Email: [email protected]
Tel: +47 35572000
Skanska Norge AS
Contact: Sverre Smeplass
Email: [email protected]
Tel: +47 40013660
Summary
The pumpability of concrete, or its ability to move through pipes and hoses by the help of a pump while
maintaining its fresh and hardened properties, can be quantified as being better the lower the necessary
pressure to obtain a given flow in a specific configuration and set-up (pump type, -capacity, pipes/hoses,
diameter/length etc). The practical pumping process including the principles of piston- and screw pumps
show that for a wide range of concretes the flow in a given set-up is approximately proportional to pumpfrequency whereas resulting pressure depends on concrete technological parameters (concrete
composition, rheology). Based on review and some preliminary pumping trials in the NTNU concrete
laboratory we propose concrete pressure gradient over the pipe length as a measure of pumpability for a
given pump set-up.
We then present some results on flow measurements in experiments on commercial premix products
and open lab concretes and mortars. Measurements of rheological parameters in the BML-viscometer,
concrete pressure in pipe , concrete flow and energy in an instrumented full-scale pump-set up with a
screw pump indicate that plastic viscosity is the main rheological parameter affecting pumpability.
Analysis of the degree of plug flow indicates a wide variety of flow profiles. A simplified 2D FEM
analysis with Navier Stokes equation using Papanastasious solution for Bingham fluid gives convergent
flow with similar plug flow profiles as the analytical Buckingham Reiner equation. An analytical slip layer
model was developed applying the Bingham model. It can be fitted to plug flow but more experiments are
needed on what the boundary flow looks like.
Continued research should study plug flow, slip layers and concrete rheology effects (admixtures,
matrix volume and properties) on pumpability including effects of pump frequency, increasing vs
decreasing flow, hose characteristics (material, diameter, length, height etc). This could also be used in
form filling experiments. A reliable numerical model must describe both flow through the pump and
pipes/hoses as well as the form filling satisfactorily.
Oslo, 2008
Foreword
COIN - Concrete Innovation Centre - is one of presently 14 Centres for Research based
Innovation (CRI), which is an initiative by the Research Council of Norway. The main objective
for the CRIs is to enhance the capability of the business sector to innovate by focusing on longterm research based on forging close alliances between research-intensive enterprises and
prominent research groups.
The vision of COIN is creation of more attractive concrete buildings and constructions.
Attractiveness implies aesthetics, functionality, sustainability, energy efficiency, indoor climate,
industrialized construction, improved work environment, and cost efficiency during the whole
service life. The primary goal is to fulfill this vision by bringing the development a major leap
forward by more fundamental understanding of the mechanisms in order to develop advanced
materials, efficient construction techniques and new design concepts combined with more
environmentally friendly material production.
The corporate partners are leading multinational companies in the cement and building industry
and the aim of COIN is to increase their value creation and strengthen their research activities in
Norway. Our over-all ambition is to establish COIN as the display window for concrete
innovation in Europe.
About 25 researchers from SINTEF (host), the Norwegian University of Science and Technology
- NTNU (research partner) and industry partners, 15 - 20 PhD-students, 5 - 10 MSc-students every
year and a number of international guest researchers, work on presently 5 projects:
COIN has presently a budget of NOK 200 mill over 8 years (from 2007), and is financed by the
Research Council of Norway (approx. 40 %), industrial partners (approx 45 %) and by SINTEF
Building and Infrastructure and NTNU (in all approx 15 %). The present industrial partners are:
Aker Kvrner Engineering and Technology, Borregaard LignoTech, maxit Group, Norcem A.S,
Norwegian Public Roads Administration, Rescon Mapei AS, Spenncon AS, Unicon AS and
Veidekke ASA.
For more information, see www.coinweb.no
Table of contents
Concrete pumping.................................................................................................................. 6
Conclusions ......................................................................................................................... 31
Acknowledgement ................................................................................................................ 32
References
......................................................................................................................... 33
1 Introduction - pumpability
The pumpability of concrete has been defined as its ability to flow through a pipe by the help of a
pump (ACI 1998), and also the ability of confined concrete to flow under pressure while
maintaining its initial properties (Jolin et al 2006). Some examples of low pumpability due to lack
of maintenance of initial properties are forward segregation in the pipe leading to blocking,
negative effects on air void content and structure, as well as reduced compressive strength
compared to concrete that has not been pumped. Due to varying pump-and-pipe configurations
from one site to another (capacity, pipe length, -height, -diameter, -material etc), the term
pumpability yields for a site-specific pump set-up. Quantification of pumpability of a concrete
mix should therefore be obtained by resulting flow or pressure of fresh conrete, or pumping
energy required for the specific mix in the actual set-up. Pumpability may also be described more
qualitatively in terms of ease or problems with pumping operations from proportioning, grouting
of pipe lines, pump trials, blocking etc. Pumping of concrete and mortar is a cost effective way of
industrializing the construction industry as indicated by present market figures. In Norway
approximately 50 % of the concrete is pumped at some stage of the placing process (Fabeko 2007,
www.fabeko.no). In addition mortars, grouts etc are pumped, often in a combined premix
silo/continuous mixer/pump system. Obviously most concretes today are pumpable. There are,
however, certain types of concretes where pumpability is still not completely resolved. Examples
are slab-on-grade (K Tuutti pers comm. Oct 2007), certain light weight aggregate concretes,
(EuroLightcon 2000), possible uncontrolled increase of pressure during high speed pumping of
certain self compacting concretes (DeSchutter, personal communication December 2007), fibre
concrete with uncontrolled fiber orientation etc. To proceed in the development of pumpable
concrete we have reviewed pumping and investigated rheological properties and compositions
important for pumpability quantified as concrete flow, -pressure and pump energy consumption.
We also conducted some preliminary tests on the effect of concrete rheology on pumpability, the
effect of pumping on rheological properties of concrete as well as effects on hardened properties
such as strength, homogeneity and air void content. Finally we look at some simplified ways of
modelling the pumping process.
2 Concrete pumping
2.1 Piston pumps
Piston pumps dominate the concrete industry due to their high capacity. Figure 1 below shows an
example with two pistons working by filling one while emptying the other via a valve shifting
opening towards the feeder and shutting towards the pipe.
The pressure in the concrete shifts with the frequency of the strokes of the pump, see figure 2
above. The maximum concrete pressure can become quite high and even higher than the 7 MPa
shown in figure 2. The minimum pressure at retreat of piston before closure of valve and opening
for stroke of the other piston can even be negative. In figure 3 below the suction is seen to be very
strong so as to possibly reverse the flow in pace with the pump frequency. Clearly this will affect
the flow conditions both locally and globally in the pipe. The measured pressure gradient over the
length of the pipe is normally linear, as seen in figure 4.
Figure 3. Suction in fresh concrete in pipe observed at the turn of the stroke of a piston pump (Kaplan
2001)
Figure 4. Pressure gradient along steel pump pipe for different concrete mixes pumped with piston pump
(Kaplan 2001)
bottom of the hopper. Figure 6 shows capacity and energy use vs. concrete pressure. Both figures,
courtesy maxit Group AB R&D/m-Tec (Mork 2007a)
capacity/power = f(pressure)
8000
capacity (l/h)
7000
6000
5000
4000
3000
2000
1000
0
0
9
8
7
6
5
4
3
2
1
0
power (KW)
Figure 5. Screw pump with steel rotor and rubber stator, concrete entering at left and conveyed via moving
conveying chamber (maxit Group/m-Tec)
10 15 20 25 30 35 40
pressure (bar)
Figure 6. Concrete flow and energy consumption in screw pump as function of concrete pressure for two
different materials (maxit Group/m-Tec)
The capacity of screw pumps is generally lower than that of piston pumps, but the equipment is
lighter and easier to handle. The pump used for the preliminary experiments given later on in this
3
report has a maximum capacity of about 7 m /h at 50 Hz and 45 mm hose diameter with length
26,6 m. Depending on the material, the flow out of the pump seems to reach some kind of
threshold beyond which the capacity is reduced as the pressure increases, see figure 6. Even by
almost doubling the energy consumption, it is usually not possible to keep up the capacity due to
backflow as the pressure rises on the downstream side of the pump and the relatively soft rubber
in the stator cannot withstand the downstream pressure. The effect seems to be depending on the
material, as seen in figure 6 for two different cement based materials tested by maxit Group.
For both piston- and screw pumps we may thus assume that the flow out of the pump is constant
and depends primarily on the pump frequency (rotations or strokes per unit time). The pressure in
the fresh concrete in the hose will primarily depend on the rheological properties of the concrete
and the conditions in the pipe (length, diameter, geometrical configuration, friction). Due to the
steady screwing the pressure is constant without the alternation seen with a piston pump, see also
figure 5 in section 5 and earlier findings (Mork and Jacobsen 2008).
Figure 7. Effect of concrete slump on dp/dx as function of flow for different pipe diameters (Sakuta et al
1989). We postulate better pumpability the lower dp/dx for a given flow without negative effects on the
concrete
In the nomogram in figure 8 (ACI 1998) a similar relationship is obtained, but it contains more
necessary information embedded in a practical manner taking into account the effect of pipe
material (rubber giving 3 times the pressure of steel), number and different types of curves
(increasing the pressure), vertical sections etc.
10
Figure 8. Diagram for dimensioning concrete pumping set-up including effect of concrete slump on
pumpability (ACI 1998)
The rate of bleeding has been found to be a major factor for pumpability (Browne and Bamforth
1977, Tattersall and banfill 1983, Kaplan 2001, Kaplan and deLarrard 2005). High bleeding rate
may cause water to be easily squeezed out and cause blocking. Increased bleeding due to the
pressure in the pump and pipe has, as mentioned, been termed pressurized bleeding (Wallevik
2002). The testing and control of pressurized bleeding has been proposed as pumpability criterion
(Browne and Bamforth 1977, Kaplan 2001) together with slump since the latter obviously is an
insufficient measure of pumpability. Pressurized bleeding was measured in a special test cell as
volume water squeezed out from a sample of fresh concrete at 3.5 MPa pressure after 10 and after
140 seconds; V10 and V140. Figure 9 from (Browne and Bamforth 1977) shows acceptable
combinations of (V140 V10) and slump to ensure adequate pumpability.
Figure 9. Pumpability as function of slump and bleeding (Browne and Bamforth 1977)
11
For materials with too low workability and too high pressurized bleeding, the resulting blocking
of the coarse aggregate depends on the impulse of the piston pump, paste viscosity etc. This was
described and quantified (Kaplan 2001, Kaplan and deLarrard 2005). They developed a more
practical bleeding test using the standard pressurized air-void apparatus for routine testing of air
void content in fresh concrete. They recommended that bleeding should be measured over a
longer period than recommended in (Browne and bamforth); up to 60 minutes since it may take
some time from mixing until pumping. They expressed the rate of bleeding in cm3/hour, and based
on full-scale pumping experiments they recommended a bleeding rate lower than approximately
3
20 cm /hour measured under pressure in the air void apparatus to ensure pumpability, see figures
10 and 11 below.
Figure 10. Bleeding (cm ) vs time, 5 litres of concrete in standard air void apparatus (Kaplan 2001)
Figure 11. Limits for pumpability as function of bleed rate (cm /h) (Kaplan 2001)
For self compacting concrete (SCC) other practical tests than the original Abrams slump cone
must be used. At the construction site, Slump-flow and T500 seem to dominate. A reinforcement Jring can be used to indicate the ability of the SCC to pass reinforcement bars. In addition, test
methods like V-funnel test, L-box test and sieve segregation resistance exist (Barthos et al 2002,
European Guidelines for SCC 2005, Norsk betongforening 2007). It appears, however, that no
criteria for pumpability of SCC based on these tests exist (Watanabe et al 2001, Lu, Laijun 2005,
Feys, DeSchutter et al 2007). This is partly caused by the varying conditions in varying pump setups. The flow conditions in the pipe are for example unknown; in particular the flow profile. In
12
the research work on pumping of concrete (Rssig 1974), observations of flow profiles were made
by pumping coloured concrete into un-coloured concrete and letting it harden. Similar studies and
observations have been made in Russia and referred in (Kaplan 2001). From figure 12 below, it
seems that a flow profile with some kind of slip layer exists for a stiff-plastic concrete, having
around 40 cm flow spread measured with the German DIN-test.
Flow
Figure 12. Flow profiles observed after pumping concrete into coloured concrete, 40 cm flow-spread
concrete pumped at 1.77 m/s flow rate in 125 mm steel pipe (Rssig 1974)
The existence of a slip layer has also been discussed by several (Tattersall & Banfill 1983, Kaplan
2001, Haist & Muller 2005). The latter study extracted the pumped material near the wall,
measured its rheological properties and concluded that there is such a slip layer. There could be
several reasons for a paste- or matrix rich layer at the wall: one being the kind of dynamic
segregation (Thrane 2007 p.53). Apparently, coarse particles in a fluid matrix tend to migrate
towards the centre, that is; towards minimum rate of shear. For concrete pumped in a pipe, this
means towards the centre where the shear rate is zero or shifting. In an ongoing work
(Kalogiannidis 2008), a similar approach has been taken as in (Rssig 1974) on modern, self
compacting materials. Preliminary studies of SCC mortar or mini concretes with 8 mm maximum
aggregate size have shown that in plastic pipes ( = 45 mm) with expected low friction, clear
flow profiles can be seen even at low flow velocities < 0.5 m/s (Kalogiannidis 2008).
3.2 Rheological approach
More advanced rheological measurements on fresh concrete exist based on that concrete can be
treated as a Bingham fluid. Using a viscosimeter, the Bingham material parameters plastic
viscosity , and yield shear 0 (see appendix 1), which govern the flow in a pipe, can be measured.
The Bingham model is very useful even though it does not describe such phenomena as non
linearity during movement (shear thinning/-thickening and thixotropy) (Helland 1982, Mork
1994). The viscosimeters in the NTNU concrete laboratories have been further developed from
its predecessor (Tattersal 1976). The BML viscosimeter has been used extensively for more than a
decade to study and 0 (Wallevik 1990, Wallevik & Gjrv 1990, Barthos et al 2002), and is now
in commercial use in many concrete laboratories. Relations have been found between slump and
yield strength and between flow and plastic viscosity (Hu, deLararrd 1996, Sedran and deLarrard
1999, Wallevik C&CR 2006). The fluid flow conditions in the coaxial viscosimeter have also
been studied experimentally and numerically in great detail (Wallevik 2003).
Even though the flow during the coaxial movement of the viscosimeter is different from the flow
in a pipe, it is reasonable to assume that it can be used to evaluate the pumpability of a concrete
mix before full scale pumping is carried out. The first large scale study of the concrete Bingham
parameters as measured in a coaxial viscosimeter (and also slump) vs. pumpability, measured as
pressure in the hydraulic oil of a concrete piston pump, were made on the very flowable high
performance concrete (slump 190 - 240 mm) for the Gullfaks C Condeep platform (Hansen
3
1988). The investigated 2.7 m concrete batches had w/b in the range 0.42-0.45 with 420 kg of
13
cement, 2 % silica fume and around 1.2 % superplastcizer by weight of cement. The results
showed that both yield strength and plastic viscosity related to the hydraulic oil pressure.
Increasing yield strength gave the best relation to increasing pressure. Also increased slump
showed a relation to reduced pressure. Yield strength increased due to pumping and gave
improved stability as observed from viscosimeter measurements made before and after pumping.
It seems that the rheological data obtained before pumping correspond best to the measured pump
pressures in the long pipeline, rising more than 180 metres vertically in the shafts of the large
concrete structure. It was recommended that pumpable concrete should have as low yield strength
and plastic viscosity as possible without loosing stability. Furthermore, largest effect of rheology
on pumpability was expected for small pipe diameter, high pump flow and horizontal pipes with
many bends. Later studies on-site of the Bingham parameters vs. concrete pressure measured in
piston pumping were made in (Hu and DeLarrard 1996). The Bingham parameters were measured
with a BTHREOM viscosimeter and showed essentially the same as was found earlier in (Hansen
1988) on a similar number of mixes.
Since the pressure- and movement conditions in a pipe, particularly with piston pumps, alternate
as discussed above, it seems probable that the complete shear rate vs. shear cycle is of interest.
The hysteresis observed during BML-viscosimeter testing, see figure 13, will probably affect flow
and pressure in the pipe depending on pressure/suction and the local rate of shear in the flow
profile. For piston pumps, where the shear may be changing constantly due to the alternation of
the pump (see figures 2 and 3), the complete test cycle should therefore be investigated when
comparing pumpability with viscometer measurements.
Figure 13. Hysteresis between increasing and decreasing rotation velocity in the coaxial BML-viscosimeter
(Barthos et al 2002)
In a preliminary study of pump flow and pressure in screw pumps as related to rheology, we
compared a simplified Hagen Poiselle Bingham fluid without plug flow with some measurements
on pump flow, -pressure and rheology performed on the same materials (Jacobsen and Mork
1
2007). The expression in eq. (1) was used (Jacobsen and Mork 2007) by integrating eq. (A5) in
Appendix 1 ignoring the formation of a plug. This is a simplified integration termed v(r)no-plug in
appendix 3 and 4 compared to the Buckingham Reiner equation, see eq. (A9) in appendix 1 and
velocity distribution v(r)plug in appendix 3 and 4:
dp 8 Gx 8 0
a 4 dp 8 0
G
(
)
=
+
=
(1)
x
dx a 4
3a
8 dx 3a
0:
:
dp/dx:
a:
Gx:
Note; error in equation in (Jacobsen and Mork 2007) is corrected in eq.(1), see also appendix 1.
14
The simplification of omitting the plug compared to eq.(A9) may result in inward plugs as seen
in appendix 3. However, based on some excel-exercises with varying combinations of realistic
pipe radii, plastic viscosities and yield strengths, it was found that the contribution from 0 in
eq.(1) can be ignored for highly flowing concretes with low 0. Therefore, eq.(1) may be used as a
simplified indication of pumpability as function of concrete rheology, pipe diameter and pipe
length. Figures 14 and 15 below show the kind of relations eq.(1) give between diameter of pipe,
plastic viscosity, flow and pressure gradient.
10
dp/dx 1
[bar/m]
0,1
0,05
0,015
.
[Pa s]
[Pa s]
40
0,01
0,0012
0,01
20
20
11
0,1
Gx [m /s]
35
dp/dx
[bar/m]
10
0,0001
a [m]
Figures 14 and 15 illustrate the very strong effect of pipe diameter, with small pipes giving very
high pressure for the range of viscosities of some typical flowing concretes. The effect of plastic
viscosity on pressure is also quite strong and the use of logarithmic scales indicates what kind of
power law relationships between the main variables we expect to observe in pumping of concrete.
We should also keep in mind that unless the pressure gets too high we can assume that screw
pumps (and then also piston pumps since the backflow problem does not exist there) produce
quite constant flow. The resulting pressure or gradient (dp/dx) can be used as parameter for
pumpability.
The flow in fig.15 is of course very low for large pipes, but 1 litre/second is realistic for the
smaller diameters in figure 14 where there is a very strong effect on concrete pressure of changing
the diameter. From figure 15 we also see that the simplification of constant flow is reasonable for
our screw pump equipment. A range of concrete qualities tested at 25 Hz frequency had fairly
constant flow 0.23 0.28 m/s in a = 0.015 m and 0.41 0.92 m/s in a = 0.0225 m hoses. The
plastic viscosity for these materials varied from < 1 to > 25 Pa.s measured in the co-axial BMLviscosimeter. (Similar plots as those in figures 14 and 15 with yield shear 0 instead of from our
experiments showed very little effect on pressure).
The existence of a plug has been discussed and it has often been argued that, depending on a
number of factors, most of the concrete is moving in that fashion (Ede 1957, Johansson&Tuutti
1976, ACI 1998, Kaplan 2001, concrete society 2005, Haist and Muller 2005, Watanabe et al
2007). The plug flow concept is however not univocal since visualization with coloured concrete
hardened in pipes have shown flow profiles (Rssig 1974, Kaplan 2001, Kalogiannidis 2008). The
consequence of the Bingham model (and any model with a yield shear) is, however, that there will
be a gradual transition from plug to shear profile as increasing parts of the fluid in the cross
section is brought into shear at increasing flow rate. At the equivalent radius R0 of a plug in a
Bingham fluid, the shear acting on the tubular plug wall is equal to the yield shear 0, see figure
A2 in appendix 1:
15
R0 = (2L/p
)0
(2)
Appendix 1 is limited to a solution of the flow of concrete through a pipe based on Hagen Poiselle
flow combined with the Bingham material model resulting in the Buckingham-Reiner equation.
Figure 16 below shows an example of calculated plug flow (plug radius = 3.8 mm according to
eq.(2)) for a concrete pumped through a 45 mm diameter hose based on observed data on
rheology and pressure gradient from table 5 (Mork and Jacobsen 2008), see details in appendix 3.
0 = 57 Pa, = 8.6 Pa.s, = 45 mm, L= 26,6 m,
dp/dx = 30 kPa/m
25
20
15
10
r [mm]
5
0
-50,00
0,25
0,50
0,75
1,00
1,25
1,50
-10
-15
-20
-25
v(r)w plug [m/s]
Figure 16: Calculated plug flow profile with measured rheology and pump pressure (appendix 3)
The measured flow of the actual concrete in figure 16 was 0,60 m/s whereas Buckingham Reiner
predicts only 0,17 m/s in line with earlier research. The Bingham model was used to describe
concrete rheology in pipe flow in (Rssig 1974), but measurements of and 0 on pumped
concrete in a concentric viscosimeter were made for the first time a while later (Hansen 1988, Hu,
deLarrard 1996, Kaplan 2001). Pumped high performance concretes with in the range 70 500
Pa.s and 0 in the range 50 2000 Pa were studied in the latter two studies. Based on that 0 in the
actual range has little effect on the last terms within the brackets of the Buckingham Reiner
equation, eq.A9 in appendix 1, the plastic viscosity was expected to be the main parameter
affecting pumpability (Hu, deLarrard 1996). Pursuing the effect of viscosity on pumpability, it
.
was found that pumpability relates best to the plastic viscosity in the slip zone ( SL spec - Pa s/m)
(Kaplan 2001). The slip zone viscosity was calculated from pumping experiments making a
number of assumptions. It was also measured with a tribometer developed from the BTRHEOM
viscosimeter (deLarrard et al 1993, Kaplan 2001). The slipzone viscosity calculated from
pumping trials related well to tribometre experiments on the same concrete. The correlation
between the slip zone viscosity and the plastic viscosity of the bulk concrete, measured with the
viscosimetre, was however weak. Yield shear in general gave weaker correlation to pumpability
in the study (Hu, deLarrard 1996) as opposed to the observations (Hansen 1988). Based on our
own results, presented and discussed in section 5, it might at present seem that the plastic
viscosity may be a more important parameter than the yield shear stress parameter. It could thus
be that the main use of the yield shear stress is to evaluate the degree of plug flow using for
example equation (2). Similar tribometers as the one made in (Kaplan 2001) have later been
developed and successfully used to quantify power-laws for pumpability through screw pumps
(Hellman 2007, Koski 2007).
In Figure 17 the rheological properties of the slip layer around the 3.8 mm radius plug in figure 16
is adjusted into a Bingham material to give the measured flow vx = 0,60 m/s. This results in a
specific plastic viscosity of 131 Pa.s/m of the layer between the plug and the wall, which is in the
same range as found in (Kaplan 2001), see details in appendix 4.
16
r [mm]
10
5
0
-50,00
0,25
0,50
0,75
1,00
1,25
1,50
-10
-15
-20
-25
v(r)slip plug [m/s]
Figure 17. Calculated profile with Bingham properties of slip layer adjusted so that mean flow rate fits the
vx = 0,60 m/s flow measured in the pumping experiments, appendix 4.
Note that if there is a flow profile with or without a plug, it should have developed over some
length at some initial flow stage after the pump. This would then lead to some kind of radial inline mass transfer that could lead to some kind of in line mixing which is not explained or
included in the simplified uniaxial flow analysis in Appendices 1, 3 and 4. Kaplan (2001) claimed
that there is no radial transfer, whereas Thrane (2007, p.53) seems to be in favour of such radial
transfer, at least of coarse particles. Experimentally one should observe a radial pressure gradient
to have a radial flow. By looking closely at the longitudinal and transversal pressures in Figure 3
no such signs are seen, but admitting that this does not represent a very detailed study of the
existence of radial flow. From experimental observation it seems that coarse aggregate tends to
concentrate in the centre of the pipe (Haist and Muller 2005). From a fluid mechanics point of
view, it is expected that flow profiles such as in figure 17 would produce lower pressure towards
the centre due to the higher flow rate there. This follows from the reduced pressure in stream lines
going faster than neighbour lines, viz. Bernoullis equation and stream lines curving over a wing
compared to those passing straight under the wing.
Flowable concretes may show considerable deviation from the simple linear Bingham model
(Tattersall 1976, Mork 1994, Ferraris, de Larrard and Martys 2001). Examples are concretes
containing quite high volumes of fillers/powders and chemical admixtures such as 3rd generation
water reducing agents (WRA) and viscosity modifying agents (VMA). The Bingham model:
=0 +
dv
dr
0:
:
(3)
yield shear strength [Pa]
plastic viscosity [Pa.s]
dv dv
, :
dy dr
may then not be sufficient to describe the rheological behaviour. Rheology may then be better
described by other non-newtonian models such as Herschel-Bulkley and power law models (Mork
1994) to describe the dependence of shear stress upon rate of shear:
=0 + K
n 1
= K
or = K
17
(Herschel-Bulkley)
(4)
(power law)
(5)
( ) = + ( 0 ) 1 + ( ) 2
( n 1)
2
(Carreau model)
(6)
Where is viscosity, is a time constant and n is the power law index (n = 0.01 to fit the
Bingham model at high shear rate). These rheological models will of course result in other kinds
of velocity distributions compared to Buckingham Reiner, as discussed with the numerical
simulations in chapter 6. However, the Carreau model gives similar plastic viscosity as the
Bingham model at high rate of shear, presumably beyond the yield shear:
( ) =
based on = ( ) = 0 + (Bingham)
(7)
As seen from equation (7) there will be a convergence problem for a fluid with low rate of shear
or at rest applying the Bingham model on this form. Appropriate (convergent) material models for
numerical simulation of concrete flow are further discussed in section 6.2.
18
main parameter in the proportioning of concrete. Table 1 below is a simplification of the gradation
recommended for pumpable concrete.
Table 1 Recommended grading (ACI 1998)
Material diameter (mm)
0.125 0.25 0.5
1
2
4
8
16
Sand (%)
6
18
40
67
85
93 100 100
Total (%) (Dmax=16mm (min.))*
3
4
12
20
28
37
48 100
*: Natural material, for crushed coarse material the amount of 8-16 should be reduced
There are some recommendations concerning content of different types of fines but so far no
composite- or particle-matrix approach has been presented (Smeplass og Mrtsell 2003). The
recommendation below (Concrete society 2005) takes into account the effect of maximum
aggregate size by reducing amount of cement at increasing aggregate size. Density differences of
the materials in table 2 will not be reflected in the volumetric proportioning of concrete. This is a
serious drawback of table 2 that may cause large variations in volumetric composition. Clearly a
particle-matrix approach is needed. There is however a lack of empirical data on the effect of
some pumpability quantity of varying matrix composition and fractions. Furthermore, pozzolans,
filler and cement may have density differences in the order 400 kg/m3 resulting in significant
volumetric variations between the particle and matrix phases. It is therefore at present not possible
to make recommendations on pumpability of concrete mixes based on volumetric proportions and
matrix content, and more data are needed.
Table 2 Recommended amount of fines (< 0,3, 0,25 and 0,15 mm) (ACI 1998, Concr. Soc. 2005)
Material
Amount/limit/spec.
Source
Fines < 0.3mm
15 30 % of aggregate*
ACI 1998
Fines < 0.15mm
5 10 % of aggregate*
ACI 1998
* only cement content > 280 kg/m3 included in fines < 0,15mm
ACI 1998
3
cem+pozz+filler (Dmax = 10) < 0.25mm > 450 kg/m concrete
Con soc 2005
cem+pozz+filler (Dmax = 20) < 0.25mm > 420 kg/m3 concrete
Con soc 2005
3
cem+pozz+filler (Dmax = 40) < 0.25mm > 380 kg/m concrete
Con soc 2005
Admixtures will in general improve the workability and pumpability (ACI 1998). Increasing
dosage of water reducing admixtures may reduce the pressure in the pump line compared to
reference tests, as observed in full scale tests (Kasami et al 1979, Spiratos et 2003). This is in line
with the effect of reduced yield shear stress with addition of super plasticizer (Wallevik 1990,
Wallevik 2002). The effect will rest on that bleeding or segregation is not increased to become a
problem. For high performance concretes with low w/b, pumpability quantified as pressure, might
be worsened even with very high dosages of super plasticizers (Lepage et al 1998). A much
higher concrete pressure was observed during pumping of a w/b= 0.30 concrete than in a w/b =
0.38 concrete in the same pump configuration in spite of the former having much higher slump
than the latter. The loss of slump due to pumping was higher for the w/b = 0.30 concrete. Air
entraining agents work positively by reducing the tendency to aggregate segregation and bleeding.
The effect of pumping on the air content has been investigated and different effects have been
reported. Traditional concrete from the 80-ies with 320 kg/m3 was reported to have increased air
void content by pumping (Johansson & Tuutti 1976) whereas reduced air content was reported
after pumping of superplasticized high performance concrete (Pleau et al 1998). The latter
investigation found that the air void structure may not necessarily be destroyed by the pumping. A
large part of the lost air content during pumping consisted of dissolution and/or gashing of the
large voids. Therefore the air void spacing factor was not so seriously affected. The coalescence
or dissolution of the small voids with highest air pressure in addition to the pumping pressure was
found to have less effect on the air void structure. This is probably due to that the number of very
small (frost protective) voids was still very high after pumping. In (Kaplan 2001) reduced air void
19
content was observed after pumping. In (Lepage et al 1998) pumpability and effect of pumping on
air void content was investigated on w/b=0.38 and w/b=0.30 concretes with hydroxycarboxylic
acid water reducer and fatty acid based air entrainers, the latter concrete also with superplasticizer.
Air void structure was not affected by pumping the w/b = 0.38 concrete, whereas the w/b=0.30
concrete had serious damage to its air void system by the pumping. So far it seems that a general
conclusion is to use air entrained concrete that can be pumped at as low pressure as possible to
avoid damage to the air void system.
Mineral admixtures and/or fines content are mainly used to reduce bleeding, also increasing the
matrix volume. Small amounts of silica fume may for example replace a larger quantity of fines
and still improve pumpability (Maage 2003). To reduce the pumping pressure and avoid blockage
the matrix should, in addition to give as little bleeding and as good slip as possible, have high
enough plastic viscosity to slow down the acceleration of coarse aggregate particles from the
pump impulse, for example from a piston pump. At the same time the plastic viscosity should be
low enough to avoid high pressure gradients. In general more concretes are pumpable than the
limitations indicated by the recommended compositions.
4.2 Admixtures for pumped concrete pumping aids
American Concrete Institute report 212.3R-21 (ACI 2004) defines pumping aids as lubricants and
fine fillers used to overcome friction or segregation that cannot be controlled by changes in the
concrete mixture proportions. Many pumping aids are thickeners that increase the cohesiveness of
concrete.
In Australian Standard (Australian Standard) five categories of thickening admixtures for
concrete and mortar are identified:
1. Water-soluble synthetic and natural organic polymers that increase the viscosity of water
2. Organic flocculants
3. Emulsions of various organic materials
4. High-surface-area inorganic materials
5. Finely divided inorganic materials that supplement cement in cement paste-fly ash and
various raw or calcined pozzolanic materials, hydrated lime, natural or precipitated
calcium carbonates and various rock dusts.
In (Rixom and Mailvaganam 1999) the use of admixtures in pumped concrete was split into three
classes dependent on the cement content in the mix:
1. Low < 200 kg/m3
3
2. Medium 200-300 kg/m
3. High > 300 kg/m3
Admixtures connected to these three categories are:
1. Water-soluble synthetic and natural organic polymers, organic water-soluble flocculants,
emulsions and inorganic materials of high specific surface area.
2. Retarding water reducer and air-entraining agents
3. Dispersing agents as calcium lignosulfonates and sodium salts of hydroxycaboxylic acid.
According to the authors, mixes in both low- and high-cement-content classes are generally more
prone to problems concerning pumpability than the medium range. In low-cement-content mixes,
poor cohesion results in segregation, and in high-cement-content mixes, thixotropy causes
pipeline friction.
The use of admixtures in lightweight concrete is also briefly discussed, with recommendation of
using air-entraining agent, superplasticizer or thickener for improved pumpability.
20
21
tolerated will of course depend on the matrix volume and its rheology. The figures indicate that
the surplus of water needed can be relatively large so that there will sometimes be a need for
special efforts to reduce instability problems (segregation, bleeding) associated with the high
water content. In addition, precaution with respect to excess moisture loss, shrinkage etc. in
hardened concrete is needed.
Table 3. Compilation of pre- and post pumping properties of LWA concrete pumped in full scale
(EuroLightcon 2000 (EL), Johansen 1991 (J91), Mork and Jacobsen 2008 (MJ))
fc,pre
flowpre
fc,post pre
Flowpost
Data Successful Pumping
Ref
post
MPa
MPa kg/m3 kg/m3 cm
cm
P: piston, R; screw
H: horizontal; V: vertical
22
(1650) 75
P, 125mm, 10m H, 10 m V EL p.31
30.6
24.6 (1700) P
EL p.32
24.3
28.7 1540
1623
50
49
P, 25m H, 10m V
EL p.33
22,2
22
1445
1430
47
45
P, 25m H, 5m V
EL p.34
42.5
49.8 (2000) 53
P
EL p.35
51.1
55.6 1870
1930
22 (slump) 20 (slump) P, 250m H, 50m V
EL p.45
30
1835
16 (slump) 8 (slump)
P, 60 m ?, 4 pumps 75 m3/h
EL p.49
17.5 1800
28 (slump) 24 (slump) P, 30m ?
EL p.51
22 (slump) 21 (slump) P, 154m V, 40 - 60 m3/h
EL p.54
27.5
26.8 1470
1470
>20(slmp) >20(slmp) P, 32 m V
J91
27
31.1 1620
1700
80
71
R, 26,6 m H
52.7
50.6 1920
1930
73
69
R, 26,6 m H
27.7
24.7 1670
1660
70
73
R, 26,6 m H
M&J08
52.9
45.6 1890
1820
90
76
R, 26,6 m H
29.5
28.7 1700
1730
56
54
R, 26,6 m H
In a publication from the Norwegian concrete association it is recommended that LWA concrete
can only be pumped if there is less than 10 % loss of 28 day compressive strength after pumping
compared to concrete placed without a pump (Norsk Betongforening 1999). Two of the five
LWAs in (Mork and Jacobsen 2008) failed to pass this criterion by comparing strength of
specimens made before and after pumping. A closer look at minimum pump pressures and
necessary re-mixing required to succeed could be a way of proceeding with this. Perhaps the
pressure in the fresh concrete should be measured on routine basis during pumping for certain
concretes.
22
All preliminary tests referred in this report were run on 12 different mixes previously reported
(Mork 2007b, Mork and Jacobsen 2008). These mixes consisted of 4 ND and 4 LWA mortars
with total w/c in the range 0,43 0,69 and 4 commercial premix products (one LWA concrete,
one SCC, one and one screed). The compositions of the 4 ND and 4 LWA mortars are given in
(Mork and Jacobsen 2008) whereas product details of the premix products are given in (Mork
2007b, Mork and Jacobsen 2008).
Figure 18 and table 4 show measurements of pressure, energy consumption and flow on a
concrete with initially dry LWA. It was found that the screw pump produced quite constant flow,
pressure and energy consumption during the 4 consecutive 30 second pumping measurements. As
can be seen the age had an influence on the rheology, observed as increased pressure and pumping
energy at a quite constant pump flow. The presented measurements were made on a premix LWA
concrete with initially dry LWA with pumping and measurements repeated over a 49 minute
period.
3
90
70
50
trykk 1 /
bar
trykk 2 /
bar
y = 1,665x - 16,439
R2 = 0,999
30
Motorstr
m/A
10
vekt / Kg
-10
-10
10
30
50
70
time [sec]
y = 1,4476x - 3,6445
R2 = 0,9938
50
Linear
(vekt / Kg)
trykk 1 /
bar
trykk 2 /
bar
30
Motorstr
m/A
10
vekt / Kg
-10-10
10
30
time [sec]
50
70 Linear
(vekt / Kg)
-10
-10
90
10
30[sec]
time
50
trykk
1/
3
repeat 2, h = 3 m, LWA 1600 kg/m
bar
70
50
trykk 2 /
bar
y = 1,455x - 294,16
R2 = 0,9989
30
Motorstr
m / A
10
-10190
70
Linear
(vekt / Kg)
210
230
250
vekt / Kg
time [sec]
Linear
Figure18. Concrete flow, -pressure and electric current during pumping of LWA concrete (1600 kg/m )
over 49 minutes through 26.6 m 45mm rubber hose at 0 and 3 m height
The flow through the screw pump working at constant frequency of 25 Hz (50 % of max capacity)
reduced somewhat over the 49 minute period. Two possible causes are; reduced workability with
23
increasing time and reduced driving pressure difference at 3m height. The hydrostatic component
is p3m = 1600 kg/m3*9.81 m/s2 *3m 0,5 bar, p3m =(dp/dx) 3m = 50000 Pa/26,6m = 1880
1
Pa/m. ( p + f = 0 , see Appendix 2). This is only around 10 % of the lowest measured pressure
gradient in table 4: 4.4 bar/26.6 m = 16500 Pa/m. Since this hydrostatic component only applies in
two of the four measurements, ageing is probably the most important cause for increased pressure
and reduced flow. The ageing includes normal cement dissolution/coagulation/hydration effects in
addition to reduced paste volume and -flow since water is absorbed in LWA. Water absorbed in
LWA also increases concrete density, contributing to reduce the volumetric flow even more than
the flow based on constant density and gravimetric measurements. The increasing electric current
indicates that the required pump energy increases to be able to keep the flow almost constant as
concrete pumpability is reduced in line with figure 6 and the observations of increased plastic
viscosity etc in (Mork 2007, Mork and Jacobsen 2008).
In addition to studying the pressure, flow and energy consumption, a large number of
measurements of properties of fresh concrete were made. At varying stages of the
mixing/pumping process measurements were made of pumpability (flow, pressure, energy use)
and rheology (Bingham parameters (0 and ) with BML-viscometer (Jacobsen and Mork 2007,
Mork and Jacobsen 2008), slump, flow, T50, density and air void content). The measurements on
fresh concrete were made to a maximum age of approximately 45 minutes, depending on how
many steps of the pumping process that were investigated (mixing, pumping, pumping through 1
or 2 hoses), and whether batch- or continuous mixer was used.
12 different normal density and lightweight aggregate mortars and concretes were investigated.
This included 4 premix materials and 8 ordinary mixes. All materials were mixed for 5 minutes in
batches of 150 litres in a counter current horizontally rotating laboratory mixer. The four premix
materials were in addition tested in the continuous mixer with only about 20 seconds mixing.
After mixing all materials were poured into the pump feeder for pumpability studies combined
with rheology measurements.
In spite of some variations in age of fresh concrete at time of measurement, some comparisons
could be made to investigate the relation between the pumpability and the rheological properties.
This could best be made with measurements made at the same place, which is after pumping
through 26,6 m of hose.
The water content of continuously mixed pre-mix mortar was checked with a microwave oven
immediately after mixing and the w/c variation of the premix in the continuous mixer was found
to vary in the range +/- 0.03. Tests on batch mixes were then made with the same materials at a
constant mean w/c.
A large amount of results on rheology and pump flow and pressure were obtained in these initial
pump trials. We concentrated on making the rheological measurements as close as possible before
and after the measurements of pump flow and energy consumption, and pressure in the hose.
However, usually there was a time difference in the order of 10 minutes between rheology- and
pumpability measurements. The main effect of pumping on rheology was little change or slightly
improved workability measured as slump, T500, flow, 0 and after pumping compared to after
only 20 sec continuous mixing. Over longer time (up to approximately 45 minutes) the pumping
process in general seemed to act like some kind of remixing keeping the material workable though
stiffening.
For materials with widely varying rheological properties the pump gave approximately constant
flow, whereas pressure in the fresh concrete in the hose during pumping varied considerably
between the different materials (Mork 2007, Jacobsen and Mork 2007, Mork and Jacobsen 2008).
24
The largest comparable test series of this kind for the same pump set up was made on the 4 + 4
ND and LWA mortars and the premix LWA concrete (Mork and Jacobsen 2008). These 9
materials all had flowable or close to self compacting consistencies. In the pumping experiments
26.6 m of 45 mm rubber hose was used at 25 Hz pump frequency. The results showed hose flow
varying between 0.42 and 0.92 m/s, which is small compared to the variation in plastic viscosity
and yield shear 0. A similar test on 30 mm hose diameter with 6 mixes showed hose flow varying
between 0.23 and 0.28 m/s which is even better in terms of low flow variation for materials with
varying rheological properties. In table 5 below the main results and rheological data of the
concrete after pumping over 26.6 m are tabulated.
Table 5. Rheological data and pumpability measurements of 12 materials pumped at 0,16 0,20
lit/s (= 0,23 0,28 m/s) in 30mm and 0,66 1,47 lit/s (0,41 0,92 m/s) in 45mm simplified
analysis of plug flow and turbulence (Jacobsen and Mork 2007, Mork and Jacobsen 2008)
Material
BML
Measured
BML
(2L/p)0
Meas
Meas
vx
Gx
a (radius)
m
Pa.s
Pa
0,015
0,015
0,015
0,015
0,015
0,015
0,0225
0,0225
0,0225
0,0225
0,0225
0,0225
0,0225
0,0225
0,0225
16,5
14,2
25,5
23,1
21,4
7,4
3,7
5,78
2,05
8,57
4,6
2,7
1,33
0,98
2,66
30000
170000
2000000
2000000
100000
10000
735000
726000
476000
797000
332000
489000
283000
293000
529000
EXM m-Tec
EXM batch
JMS m-Tec
JMS batch
REP m-Tec
REP batch
Reference 4mm 30 MPa
Reference 4mm 60 MPa
Reference 8mm 30 MPa
Reference 8mm 60 MPa
LWA 4mm 30 MPa
LWA 4mm 60 MPa
LWA 8mm 30 MPa
LWA 8mm 60 MPa
maxit ton 915 m-tec
dp
Pa
Rplug
mm
m/s
lit/s
4,4
13,3
873
875
130
55
146,05
144,81
61,42
57,05
228,49
82,74
21,82
5,59
96,19
8
4
23
23
69
293
11
11
7
4
37
9
4
1
10
0,28
0,27
0,25
0,23
0,25
0,25
0,42
0,66
0,71
0,60
0,89
0,41
0,64
0,92
0,64
0,20
0,19
0,18
0,16
0,18
0,18
0,67
1,05
1,13
0,95
1,42
0,66
1,01
1,47
1,01
Re
1
1
1
1
1
2
10
10
31
6
14
11
34
68
17
In the tests presented in table 5 all kinds of consistencies of the material coming out of the hose
end were observed from apparent plug flow, to highly flowing self compacting materials.
Following the above discussions on the low variation of pump flow we have plotted pressure as
function of the two Bingham parameters in figures 19 and 20 below.
25
100000
dp/dx [Pa/m]
y = 8197,62x
2
R = 0,30
0,19
10000
0,75
y = 224,41x
2
R = 0,57
1000
measured dp/dx 30
measured dp/dx 45
100
1
10
100
0 [Pa]
Power (measured
dp/dx 30)
1000
Power (measured
dp/dx 45)
Figure 19. Relation dp/dx vs yield shear (0) at approximately constant flow 0,16 0,20 lit/s (= 0,23 0,28
m/s) in 30mm and 0,66 1,47 lit/s (0,41 0,92 m/s) in 45mm hoses
meas dp/dx ( 45, LWA,
ND, Dmax 8)
meas. dp/dx (30, SCC,
exp, thix)
calc dp/dx 45 (measured
v, mu and to)
calc dp/dx 30 (measured
v, mu and to)
Power (meas dp/dx ( 45,
LWA, ND, Dmax 8))
Power (meas. dp/dx (30,
SCC, exp, thix))
1000000
dp/dx [Pa/m]
100000
10000
y = 11513,20x
2
R = 0,59
0,45
1000
3,90
y = 0,10x
2
R = 0,67
100
0,1
10
100
[Pa.s]
Figure 20. Relation dp/dx vs plastic viscosity () at approximately constant flow 0,16 0,20 lit/s (= 0,23
0,28 m/s) in 30mm and 0,66 1,47 lit/s (0,41 0,92 m/s) in 45mm.
The plots in figures 19 and 20 indicate that pumpability quantified as pressure gradient relates
better to plastic viscosity than to yield stress. Figure 20 also shows that equation (1) predicts too
high pressure for the measured values of flow, plastic viscosity and yield stress. The deviation
seems to be larger the smaller the hose radius and the larger the plastic viscosity. The results are
in line with earlier findings that Buckingham Reiner predicts too low flow (Hu, de Larrard 1996,
Kaplan 2001). Furthermore, the flow was most sensitive to variations of within the observed
range of values of 0, and dp/dx. One probable reason for the weakness of Buckingham Reiner
is that the plug size and -shape and also probably the rheological properties of the slip layer are
unknown. Clearly more experiments and data on the properties of the flow profile are needed.
26
A simplified analysis of the degree of plug- or laminar/profile flow can be made based on
equation (2), pressure measurements in pumping experiments and measurements of 0. In fact, the
flow of concrete through a pipe can itself be used as a viscosimeter measuring plastic viscosity
and yield shear 0 provided enough information is at hand; pressure difference, flow and flow
profile (Hans 2003, Wiklund 2007). Our simplified analysis of the degree of plug flow according
to equation (2) and appendix 1 presented in table 5 shows that we have all kinds of variations
from plug flow to highly developed flow profiles. In most cases the plug radius is lower than the
pipe radius indicating that there is some degree of flow profile. Visual observations were made of
the surfaces of the stiff plug materials coming out of the hose. The surface of the dispatched
plug seemed wetter than the bulk. However, they appeared to dry up or reabsorb the surface
water quickly. Sawing of one hardened plug revealed even distribution of aggregate.
An evaluation of the flow profile can also be made with the Reynolds number (Re) to get an idea
whether the profile is laminar or perhaps turbulent if Re >> 1:
Re =
v x d
(8)
Here d is a characteristic size, in this case pipe diameter [m]. For Bingham materials we take the
plastic viscosity, , as a conservative estimate of the Newtonian viscosity (Thrane 2007). As seen
from table 5, Re is in the range 1 2 for the 30 mm diameter hose. This indicates laminar flow
and thus plug flow seems reasonable. However for the 45mm diameter hose, Re is in the range 6
68 indicating more viscous or liquid like behaviour and less probability at least for pure plug flow
for the concretes tested in this hose. We also note that the materials tested for pumpability in the
45mm hose were more flowable than the 30mm materials as judged from their lower plastic
viscosities. More direct quantitative use of the Reynolds number seems difficult since most
experience stems from fluids with viscosity orders of magnitude lower than our materials
(Helland 1982, Welty et al 2001).
Finally it should be mentioned that the ACI method in figure 8 to determine pressure for a given
pump capacity, set up, pipe length etc with concrete slump as the only rheology parameter, is too
rough. For our experiments many concretes with slump in the range 200 mm had much lower
pressure than the 17 bar predicted by the use of figure 8 for 8 inch slump and the specifics of our
pump-set up. There seems thus to be a need for more work on models and experiments for
concrete pressure as function of workability for example for SCC which has much more fluid
consistency than the workability range of figures 7 and 8 (Rodum, Hammer 2008). In addition the
diagrams cannot predict the effect on pumpability of changing pipe-cross section, which is a
matter of practical concern (Smeplass 2008).
27
explained in section 3.2 and appendix 2, the Buckingham Reiner type plug flow can be modified
with a surrounding Bingham slip layer as illustrated by fitting the plug flow in figures 16 and 17
to measured flow. This approach may be further developed assuming a no slip boundary were the
velocity of the plug is varied by adjusting the rheology and thickness of the slip- or lubricating
layer. We still require that the total integrated flow in eq.(A8) in Appendix 1 is equal to the
measured flow, and then adapt the plastic viscosity of the Bingham slip layer. In the figures above
we used the plug radius based on the shear acting on the concrete near the pipe wall, i.e the plug
radius calculated from eq (2) and Appendix 1. In figure 21 below we try a larger and perhaps
more realistic plug radius surrounded by a slip layer of only 2.1 mm thickness. Furthermore we
assume that both yield shear and plastic viscosity of the slip layer are low compared to the
Bingham parameters of the plug material, i.e. some kind of viscous paste or grout lubricating the
layer between the plug and the pipe. The detailed calculations are given in a worksheet in
appendix 3 and 4.
0SL = 2 Pa, 0SLspec = 952 Pa/m, SL = 1 Pa.s SL,spec = 476 Pa.s/m, 45
mm, L= 26,6 m, dp/dx = 30 kPa/m, layer = 2.1 mm
25
20
15
r [mm]
10
5
0
-50,00
0,25
0,50
0,75
1,00
1,25
1,50
-10
-15
-20
-25
v(r)slip plug [m/s]
Figure 21. Flow profile with vx = 0,60 m/s and very viscous 2.1 mm Bingham slip layer (app.4)
In fact we could let the plug radius and rheological properties of the slip layer vary within any
combination of layer thickness and layer plastic viscosity with total flow fitted to the observed
flow of the pumping experiment. This was done for the above example by integrating the flow
profile numerically in 1.5 mm layer steps. Figure 22 below shows a plot of specific plastic
viscosity of the (fictive) slip layer vs layer thickness for a constant yield shear (Appendix 4).
vx = 0,60 m/s, 0 = 2 Pa
Layer thickness
(mm)
12
10
8
6
4
2
0
0
100
200
300
400
500
600
Figure 22. Specific plastic viscosity of slip layer vs layer thickness for plug in figure 21 (app.4)
The relation slip layer viscosity - layer thickness in figure 22 is limited by the coarse integration
steps at the surface (1.5 mm). Presumably the specific viscosity of the slip layer will increase
towards infinity as the layer thickness is reduced towards zero. However, at present the most
28
important work lacking is on observing the actual slip layer; both its thickness and its rheological
properties. From figure 22 it seems dubious to apply the concept specific plastic viscosity of slip
layer when it in the present model actually varies depending on the layer thickness, and
presumably also on flow rate.
In order to describe the slip layer properties it could perhaps be possible to use the apparent linear
part of figure 22 resulting in a slip layer equivalent visco-thickness coefficient equal to the
absolute value of the ratio d(a-R0)/d(SLspec). Figure 22 above gives:
d(a-R0)/d(SLspec) = ABS((10-1.5)*10-3 m/(285-507) Pa.s/m) = 3.8 * 10-5 m2/(Pa.s) (9)
The proposed visco-thickness can be calculated from plots like figure 22 for all materials in this
study to characterize the slip layer. The advantage of such analyses is that the necessary
experimental data can be measured in the existing viscosimeter instead of having to use the kind
of specially designed tribometer developed in (Kaplan 2001). If tribometres are to be used it
would be better to develop an apparatus operating under pressure, perhaps where the pipe material
(steel, rubber, plastic etc) can be tested, in addition to measuring the plastic viscosity and yield
shear of the slip layer. However, the main objectives of testing and computations are to predict
whether the pumping will be successful, and to predict the flow behaviour during the casting
process. That is, the whole process from mixing to form filling including pumping should be
predictable. Therefore, variations in flow due to differences between pipe- and form surface
should not be size dependant so that the same surface should give predictable flow whether in a
small or big pipe or in a wide or slender mould with little or much reinforcement. Such an
experimental verification or check requires an extension of the flow simulations presented in the
next section of this report into realistic 3D models including the casting process and additional
experiments following the pumping studies. We therefore recommend that any model for flow of
concrete during pumping also should be verified in full scale moulding into varying forms.
6.2 Numerical simulation of flow of pumped concrete
We have chosen to investigate the applicability of an existing commercial software, COMSOL,
for simulation of flow of fresh concrete. Aware of other flow models such as the numerical model
of flow in the coaxial viscosimetre (Wallevik 2003) and other works on concrete flow such as
(Thrane 2007, Roussel 2007) we chose COMSOL for its availability and low user threshold. The
generalised Navier-Stokes equation for steady-state flow may be written:
v
(v + (v ) T ) + v v + p = F
t
(10)
(11)
where v is the velocity vector (m/s), the density (kg/m3), the dynamic viscosity (Pa.s), p the
pressure vector (Pa) and F is a volume force field (N/m3) such as gravity so that f = F/ (m/s2).
The equation is explained a bit more in Appendix 2.
Various rheological models, such as the Bingham model, Power-law model, Herschel-Bulkleyand Carreau model, can be related to the Generalized Navier-Stokes equation by establishing the
relationship between the plastic or dynamic viscosity (, ) and rheological parameters, see
section 3.2. Incorporating such models into Navier Stokes equation makes it practically
29
impossible to solve the resulting differential equations analytically for flow as function of time
and space for given boundary conditions, pressure and fluid material parameters.
The FEM-based Comsol Multiphysics Fluid dynamics module is a rather user friendly program
solving equation (10). We have here for simplicity made some simulations with Bingham-type
fluids in 2 dimensional openings with the same width and length as the diameter and length of the
30 and 45 mm hoses in the experiments. As boundary condition we used only no slip which can
be seen in all results as zero flow at the wall. Furthermore we used only a simple 2D
representation of the pipe in the form of two parallel border lines with width = 45 and 30 mm and
length = 26.6m.
Concrete in its fresh state may be considered to behave like a viscoplastic fluid with the rheology
of normal fresh concrete described by the Bingham model or Hershel-Bulkley model as
mentioned already in section 3.2, see also Appendix 1. However, both models, see eq.(3), (7) and
(4) respectively, have problems with convergence at low or zero strain rate when simulations
based on them are run in COMSOL Multiphysics. To overcome the convergence problem the
Bingham model was modified by introducing a term of (1 e m& ) with the constant m 100, to the
constitutive equation as below (Papanastasiou 1987):
(& ) =
0
(1 e m& ) +
&
(12)
In this way problems with low or zero rate of shear is avoided. Simulations based on the
Papanastasiou Model (user defined model) and the Carreau Model (default model of Comsol see
eq.(6)) were run in COMSOL Multiphysics by using the rheological parameters shown in Table 5.
The results are given in table 6, examples of flow profiles are shown in figures 23 and 24. In
Appendix 5 all flow profiles are shown.
Running simulations by the Carreau model are mainly based on the approach that at high strain
rates and o , the Carreau model becomes the Bingham model. This approach works well on the
shear region but not on the plug region in the flow. If the shear region in the flow involves high
strain rate, then the simulation results obtained in this region will be closer to the reality.
Therefore the velocity profiles obtained by the Carreau model are not so perfect when compared
to those by the Papanastasiou model. Although the Papanastasiou model, which is the modified
Bingham model, can give a better velocity profile, it takes much longer to converge to a solution
of equal accuracy as the Carreau model. However, from figures 23 and 24 shown below, the flow
velocity by the Papanastasiou model is lower than that by the Carreau model.
From the comparison of velocity profiles obtained from both models, clear plug flows can be seen
for most cases when using the Papanastasiou Model, see Appendix 5. The radii of the plugs can
be estimated from the velocity profiles and compared with the calculated values using the
Buckingham Reiner equation. In table 6 below we have collected measured flow together with
calculated flow using the analytical solution (Buckingham reiner) as well as the numerical
solution with Navier stokes equation and the Papanastasiou model.
30
0,42
0,66
0,71
0,60
0,89
0,41
0,64
0,92
0,64
0,18
0,12
0,33
0,60
0,20
0,20
0,38
0,67
0,21
0,8203
0,5176
1,1340
0,5127
0,0223
0,8049
1,1616
1,8640
0,8569
Plug radius
R0 estimate
FEM 2D
4 mm
3 mm
12 mm
12 mm
0
0
6 mm
7 mm
5 mm
3 mm
20 mm
5 mm
3 mm
1 mm
6 mm
From table 6 and Appendix 5 the main conclusions are that convergent numerical solutions for
Bingham materials are obtained and with similar plug shaped velocity profiles as in the analytical
solutions. Furthermore it seems that the numerical simulations are closer to the measured flow for
the larger pipe diameter. By comparing with the analytical plug-radii of table 5 we see that the
numerical model seems to predict somewhat smaller plug radii. However, since the geometry
differs between the two models we do not expect any match between simulations and analytical
solutions. (The simplified geometry in the FEM model should make its flow less influenced by the
slip conditions the larger the width since the ratio (Flow Area/Aspect) for a slit approaches 0.5 at
increasing width. For a circular pipe the same ratio is monotonically increasing with increasing
diameter as half the radius; r/2. Therefore we would also expect smaller plugs for the slit
geometry than for the tube geometry).
Further investigations should aim at making a rotational symmetric numerical model that can be
solved with reasonable computer power where different slip conditions are studied and compared
with experiments. The experiments should aim at studying the basic rheological parameters
expected to affect pumpability as well as some pipe and pump characteristics. Rheology,
pumpability and flow profiles of matrix and concrete with varying plastic viscosity and yield
strength should be investigated as function of pump frequency and pipe materials. Later the
pressure conditions in larger pipes and piston pumps should be investigated, and the flow during
form filling should also be checked. Also non-bingham behaviour such as shear thinning and
thixotropy should be investigated (Wallevik 2003, Roussel 2007) to find possible effects on
pumpability.
31
7 Conclusions
The pumpability of concrete, or its ability to move through pipes and hoses by the help of a pump,
while maintaining its fresh and hardened properties, can be quantified as being better the lower
the necessary pressure to obtain a given flow in a specific configuration (pump type, -capacity,
pipes/hoses, diameter, length etc). A range of properties can be specified to define pumpability
including the effect pumping has on hardened properties.
The practical pumping process including the principles of piston- and screw pumps show that for
a wide range of concretes the flow in a given set-up is approximately proportional to pumpfrequency whereas resulting pressure depends in addition on concrete technological parameters
(concrete composition, rheology). Based on review and some preliminary pumping trials in our
lab we propose concrete pressure gradient over the line as a measure of pumpability for a given
pump set-up.
The traditional pumpability criteria based on sufficient slump and low rate of bleeding still yield,
and a new simple method measuring pressurized bleeding in the conventional air void apparatus is
a simple and inexpensive test that can be used on site on 5 litre samples with air pressure. 20
cm3/hour bleeding is proposed as maximum rate of bleeding during the first 60 minutes in this
simple pumpability test.
Existing recommendations on composition of pumpable concretes are inadequate since they are
mainly based on mass or -percentage of various binder and filler fractions without consideration
of volumetric variations that may arise from density differences. Consequently the basis for
proportioning concrete is violated. Clearly a test programme with rigorous proportioning based on
particle matrix approach including adequate correction for measured air void content and
measured density combined with the criteria of ACI and Concrete Society and full scale
pumpability tests should be conducted to verify criteria for more flowable concretes than those in
present guidelines. So called pumping aids are usually divided into organic and inorganic
32
materials including thickeners and inorganic powders, where improvement of stability against
separation is one important effect of these.
We present some results on flow calculations and pumping experiments on mortar and concrete
mixes; both commercial premix products and open lab mixes. Measurements of rheological
parameters in the BML-viscometer and pressure, concrete flow and energy in an instrumented
full-scale pump-set up with a screw pump, indicate that plastic viscosity is the main rheological
parameter affecting pumpability. This includes LWA concrete, but there special considerations
must be taken to ensure sufficient matrix volume to secure the plastic viscosity in case pumping
pressure gives volume reduction and loss of pump flow due to too much water going into the
LWA. Also possible negative effects on strength must be controlled according to existing LWA
recommendations.
Finally we analyse the degree of plug flow and find that within our tests there are probably a wide
variety of flow profiles from plug to near-turbulence. Analytically calculated flow in pipe is lower
than measured flow, probably more so the higher the degree of plug flow. Numerical simulations
were made for a simplified 2D case solving Navier Stokes equation using a commercial FEM
software with a convergent solution of the Bingham model. The numerical simulations gave
similar plug flow as the analytical solutions.
Continued research on the pumping process should study flow characteristics like velocity profile,
plug, slip layer properties and concrete rheology effects (admixtures, matrix volume and
properties) on pumpability. Studies of pump frequency, increasing vs decreasing flow, hose
characteristics (material, diameter, length, height etc) on concrete pressure, flow profile and slip
layer should be carried out. In addition numerical models should be developed and various
practical cases should be verified such as form filling, effect of pipe material and dimensional
changes. Any discrepancies between reality and numerical predictions using Navier Stokes
equation should be further investigated to give generalized solutions for both pumping and
casting.
8 Acknowledgement
The work is part of an ongoing research project in collaboration between maxit Group and The
Norwegian University of Science and Technology (NTNU), Department of structural engineering
and supported by the Research Council of Norway as well as the concrete innovation centre
(COIN) where maxit Group and NTNU Dept of Structural Engineering are two of the partners.
The experiments described were carried out in the concrete laboratories at NTNU dept of
structural engineering but due to their labour intensive nature would not have been possible
without assistance from colleagues and partners at NTNU, maxit Group and SINTEF; Laboratory
engineer Ove Loraas, Andreas Gurk, Jon Elvar Wallevik, Gran Lors, Geir Norden, Mrten
Ekelyn, Bruno Jensen (maxit Group), in addition to help and support from Dr Hedda Vikan and
Dr Tor Arne Hammer, task responsible of COIN/TG 2 at Sintef and COIN centre leader,
respectively.
33
9 References
ACI Comm. 304 (1998) Placing concrete by pumping methods, ACI 304.2R-96, 2nd Print 25 p.
American Concrete Institute (ACI) (2004) rep. 212.3R-04, Chem. Admixtures for Concrete, 30 p.
Australian Standards MP 20.2, Thickening Admixtures for Use in Concrete and Mortar
Bartos P.J.M., Sonebi M., Tamimi, A.K.(2002) Workability and rheology of fresh concrete:
compendium of tests, Report of Rilem TC 145-WSM, Workability of special concrete
mixtures, ISBN 2-912143-32-2, 127 p.
Browne Roger.D, Bamforth Phillip B. (1977) Tests to establish concrete pumpability, ACI Journal
May 1977, 193-203
Concrete Society (2005) Pumping concrete, Good Concrete Guide 2, Surrey, 14 p.
De Larrard F., Szitkar J.C., Joly M., Claux F. and Sedran T. (1993) Design of a rheometer for
fluid concretes, RILEM Workshop, Ed.: P.Bartos, E&FN Spon 201-208
DeSchutter (2007) December 4, personal communication
Ede A.N.(1957) The resistance of concrete pumped through pipelines, Magazine of Concrete
Research, November, 129-140
EuroLightcon (1999) Methods for testing fresh LightWeight Aggregate Concrete, Document
BE96-3942/R11, 51 p.
EuroLightcon (2000) Pumping of Light weight Concrete based on expanded clay in Europe,
Document BE96-3942/R11, 65 p.
European guidelines for Self compacting concrete (2005) Specification, production and use.
ERMCO/CEMBUREAU/EFNARC, 63 p.
Ferraris C., de Larrard F, Martys N.(2001) Fresh Concrete Rheology : Recent developments, in
Materials Science of Concrete VI, Amer. Cer. Soc. Ed.S.Mindess, J.Skalny 215-241
Feys, D, Calie B., Verhoeven R., De Schutter G.(2007) Evaluation of gravitational flow tests with
SCC in pipes as a forecast for pumping pressures, 5th Int. symp. on SCC, Rilem PRO54
ISBN 978-2-35158-047-9, Vol.1, 515-520
Haist M., Mller H.S.(2005) Optimization of the pumpability of self-compacting Light Weight
Concrete, Proc SCC 2005 ACBM, NWU, Evanston, 6 p.
Hansen J.K.B.(1988) Characterisation of concrete pumpability using the two-point apparatus,
High Performance Concrete Materials Development report 3.2, report no Sintef STF 65
F89046, Norway, 45 p.
Hans A.(2003) Entwicklung eines In-line viskosimeters auf basis eines magnetisch-induktiven
Durchflussmessers, Dr.diss. Schriftenreie Institut fur Mess- und Regelungstechnik,
Universitt Karlsruhe TH (No.001),105 p.
Helland K.(1982) Fluidmekanikk 3.opplag, Tapir, Trondheim, 151 p.
Helland S, Maage M.(1993) Strength loss in un-remixed LWA Concrete, Utilization of High
Strength Concrete, Symp. Lillehammer Norway june 20-23 Proceedings vol 2 Ed. I.Holand,
E.Sellevold ISBN 82-91341-00-1, Norw Concrete Association Oslo, 744-751
Hellman F.(2007) Comparison of the power law pressure drop modell versus actual pressure drop
in the Vingker laboratory pump rig, Cementa Res. / Maxit Rep. 68105-08G 11 p.+ 2 app
Hu, C, De Larrard F., Sedran T.(1996) A new rheometer for high performance concrete, 4th Int
Symp on High Strength/High performance concrete, Presses de LCPC, 179-186
Jacobsen S., Mork J.H.(2007) Pumping of concrete and mortar, Norwegian Concrete day,
SINTEF, 18 p.
Johansen Kjetil (1991) Pumping of LWA concrete with density 1450 (In Norwegian), Norwegian
Concrete day, 15 p.
Johansson A, Tuutti, K.(1976) Pumpbetong och betongpumpning (Del 1), Betongmassas
strmning i rr (Del 2), CBI rapport 7:76, Stockholm, 10 p.
Jolin M., Chapdelaine F., Burns D., Gagnon F., Baupr D.(2006) Pumping concrete; a
fundamental and practical approach, presentation Shotcrete for underground support X,
Sept.12-16 (2006) 9 p.
34
Kalogiannidis E. (2008) Coloured mortar and flow profiles, NTNU/Dept of structural engineering,
Sintef, preliminary report, Norway, 27 p.
Kaplan D.(2001) Pompage des btons, Thse de doctorat ENPC/ LCPC, ISBN 2-7208-2010-5,
Paris, 228 p.
Kaplan D, deLarrard, F., Sedran T.(2005) Avoidance of Blockages in Concrete Pumping Process,
ACI Materials Journal, 102 (3) 183-191
Kasami H., Ikeda T. and Yamane S (1979) On Workability and pumpability of superplasticized
concrete, Proc. 1st CANMET/ACI Conf on Superplasticizers in Concrete, Ottawa,
V.M.Malhotra Ed., ACI SP62, 67-86
Koski Piiparinen, P.(2007) Pumping trials on premix concrete products and the use of the
viskomat to evaluate pumpability, Cementa Research / Maxit Report 68105-08J, 14 p.+2 app
Lepage S., Baalbaki M., Lessard M. and Aitcin P.C.(1998) Pumping air-entrained high
performance concreteProc Int symp. High-performance and reactive powder concretes
Vol.4, Univ.Sherbrooke, 393-404
Lu, Laijun.(2005) Construction Technology of Self-Compacting Concrete, Proc 1st Int Symp on
design, performance and use of SCC, SCC2005 China, Yu, Shi, Khayat and Xie (Eds)
RILEM PRO 042 ISBN: 2-912143-61-6, 617-625
Maage M.(2003) Concrete for various applications, Chap. 1 in Course on Concrete Technology,
NTNU, Dept Structural Engineering, 76 p.
Mork J.H.(1994) Effekt av sementens forhold mellom gips og hemihydrat p den ferske
betongens reologi dr.ing avhandling 1994:04, NTH, 286 p.
Mork J.H (2007a) Module on fresh concrete,Course Concrete Tecnology 3 NTNU Dept of Struct
Engineering 39 p.
Mork J.H.(2007b) More efficient use of mortar on the building site. Pumping and rheology, Maxit
Group/NTNU Draft report April 23, 9 p.
Mork J.H., Jacobsen S. (2008) More efficient use of mortar at the building site. Pumping and
rheology Pumping and rheological properties of LWA concrete, Maxit Group/NTNU Dept
of Structural Engineering/COIN TG2, 23 s.
Norsk Betongforening (1999) publikasjon 22, Lettbetong Spesifikasjoner og
produksjonsveiledning, NB Oslo 70 s.
Norsk Betongforening (2007) publikasjon 29, Spesifikasjon og produksjonsveiledning for
selvkomprimerende betong, NB, Oslo 48 s.
Papanastasiou, T.C.(1987) Flows of Materials with Yield, Journal of Rheology, 31(5)385-404
Pei, M., Wang, D., Hu X. and Xu D. (2000), Synthesis of sodium sulfanilate-phenolformaldehyde condensate and its application as a superplasticizer in concrete
Pleau R., Boulet D., Atcin, P.C.(1998) Influence of pumping on the air-void system of HPC, Proc
Consec 98, Gjrv, Sakai, Banthia (Eds.) E&FN Spon, 2047-2056
Rixom, R. and Mailvaganam, N. (1999), Chemical Admixtures for Concrete, 3rd edition, E & FN
Spoon, London, pp. 297-299
Rodum E., Hammer T.A., Jacobsen S. (2008), Veiledning for pumping av betong, SINTEF
Byggforsk Betong Rapp.utk 15.jan, 3D0005.02, 17 s.
Roussel N. (2007) Rheology of fresh concrete: from measurements to predictions of casting
processes, Materials and Structures (49) 1001-1012
Rssig M.(1974) Frdern von Frischbeton, insbesondere von Leichtbeton, durch Rohrleitungen,
Dr.diss, RWTH, Westdeutscher Verlag, ISBN 3-531-02456-6, 132 p. + 92 p. fig/tab
Sakuta M., Kasanu I., Yamane S., Sakamoto A. (1989), Pumpability of fresh concrete, Takenaka
Technical Research laboratory, Tokyo, pp. 125-133 (ref 50 in Kaplan 2001)
Sedran T., de Larrard F (1999) Optimization of self compacting concrete thanks to packing
model, Proceedings 1st SCC Symp, CBI Sweden, RILEM PRO7, 321-332
Smeplass S. (1998) Pumpability of LWA concrete with impregnated, pre-wetted and untreated
LWA, Report STF22 A98751, Sintef,(In Norwegian) 28 p.
Smeplass S.(2004) Betongproporsjonering, Utdelt Materiale (UM) 4, Skanska/NTNU fag TKT
4215 Betongteknologi 1 34 s.
Smeplass S. (2008) April 11, pers. Comm., Trondheim, Norway
35
Spiratos N., Pag M., Mailvaganam N., Malhotra V.M., Jolicoeur C.(2003) Superplasticizers for
Concrete : Fundamentals, Technology, and Practice, Ottawa, ISBN 0-9731507-1-8, 322 p.
Stephenson T.O., Concrete pumping Pumpable concrete, Australian civil Engineering 5 June
(1968) 24-27
Tattersall G.H., Banfill P.F.G.(1983) The Rheology of Fresh Concrete, Pitman Books Ltd.,
London, 356 p.
Tattershall G.H.(1976) The workability of concrete, Viewpoint Publications, Cement and
Concrete Association UK, 138 p.
Thrane L.H. (2005) Simulation and verification of form filling with self-compacting concrete,
Nordic Concrete Research Publ. No.33, ISBN 82-91341-91-5, 89-91
Thrane L.H.(2007) Form Filling with SCC, Concrete Centre DTI/DTU Denmark, 269 p.
Tuutti K.(2007) October 11 - 12, pers. comm. Stockholm, Sweden
Wallevik O.(1990) Rheology of fresh concrete and utilization on concrete with and without silica
fume, Dr.ing dissertation 1990:45 Norw.Univ.Sc.&Tech, 182 p. (In Norw.)
Wallevik O., Gjrv O.E.(1990) Development of a coaxial cylinders viscosimeter for fresh
concrete, Proc RILEM Int conference on properties of fresh concrete, 213-224
Wallevik O.(2002) Pumping of concrete, in Rheology course, IBRI Concrete Industry, Rheology
of cement suspensions, Practical applications, Iceland, 623-627
Wallevik J.E.(2003) Rheology of particle suspensions. Fresh concrete, mortar and cement paste
with various types of lignosulfonates, Dr.ing dissertation 2003:18 Norw.Univ.Sc.&Tech,
397 p
Wallevik J.E.(2006) Relationship between the Bingham parameters and slump, Cement and
Concrete Research (36) 1214-1221
Watanabe K., Kuroiwa S., Ono H., Tanigawa Y.(2001) Evaluation of pumpability of high-fluidity
concrete, Consec 01, Univ of BC, Canada, ISBN 0-88865-792-7, Vol.2, 1642-1649
Watanabe K., Ono H., Katou K. and Tanigawa Y.(2007) Analytical and experimental study on
flow of fresh concrete in conveying pipe, 5th Int. symp. on SCC, Rilem PRO54 ISBN 9782-35158-047-9, Vol.1, 393-398
Welty JR, Wicks CE, Wilson RE, Rorrer GL (2001) Fundamentals of momentum, heat and mass
transfer, 4th ed., John Wiley and Sons, 759 p.
Wiklund J.(2007) Ultrasound Doppler Based In-Line Rheometry, development, validation and
application, Doctoral thesis Lund University ISBN 978-91-628-7025-6, 81 p. + 5 papers
Internet:
www.comsol.com
www.fabeko.no
www.maxit-Group.com
www.putzmeister.com
www.schwing.com
36
APPENDIX 1
No-slip tube flow of Bingham liquid
The high-friction rubber hose wall makes it reasonable to start by assuming no-slip flow. This is
also justified by the recommendation (ACI 1998) 3 times higher pressure in rubber tubes than in
steel pipes of equal length and dimension.
For a newtonian fluid flowing between a fixed plate with area A and an equal parallel plate at
distance y moving at velocity v due to a force F, the shear, = F/A. is proportional to the shear
rate dv/dy:
= =
dv
dv
(or = (tube geometry ))
dy
dr
(A1)
shear [Pa]
dv dv
,
,
dy dr
:
v:
r:
y:
-1
rate of shear [s ]
viscosity [Pa.s]
maximum flow rate equal to speed of moving plate [m/s]
radius (m)
distance from surface (m)
For a Bingham fluid like concrete there is a yield shear that must be exceeded whereafter the
shear is proportional to the rate of shear:
=0 +
dv
dr
0:
:
(A2)
yield shear strength [Pa]
.
plastic viscosity [Pa s]
vx
v(r)x
p
dx
pressure [Pa]
length [m]
total concrete flow [m/s]
concrete flow as function of radius [m/s]
radius of pipe [m]
shear in flowing concrete [Pa], assumed > 0 at infinitesimal distance from pipe wall and
inwards
Mechanical equilibrium along the x-axis of the pipe element in figure A1 requires: Fx = 0:
r 2 dp 2rdx ( 0 +
dv
)=0
dr
37
(A3)
r dp
dv
0
=0
dr
2 dx
(A4)
The velocity distribution v(r) as the fluid flows through a pipe with radius a is found by separating
and integrating (A4) with boundary condition v(a) = 0 (no slip):
v ( r ) = dv =
1 dp
rdr 0
2 dx
dr
(A5)
dp
dx
Figure A2: Plug with radius R0 as shear load from pump pressure exceeds the yield 0.
Mechanical equilibrium in the x-direction gives the radius of the plug, R0 :
R0 2 dp 2R0 dx 0 = 0 R0 =
2 0
dp
(
)
dx
(A6)
r R0
r R0
(A7)
Eq.(A7) does not allow any transport if R0 a. If there is transport in such a case it is highly
probable that the concrete is flowing through the pipe more or less like a plug with a lubricating
slippage layer between the plug and the pipe wall. The rheological properties of the layer are
different from those of the bulk concrete. Visual inspection of the output from the hose-end in
such cases has sometimes revealed a moist surface of a plug-like material. In general there seems,
however, to be transitions between plug-like and flow-profile like output. Detailed studies
(Kaplan 2001) show that the plastic viscosity of the slippage layer describes the plug flow better
than yield shear strength of this fluid layer. Here we shall treat this in a simplified manner by
estimating the layer thickness and assuming that it has Bingham behaviour.
38
The total (integrated) flow, Gx [m3/s] is obtained from (A7) by multiplication of the plug
contribution (v(R0)) and integration of the rest of the cross section between the plug and the pipe
wall (v(R0) to v(a)):
a
Gx =
v ( R0 )R02
( a 2 r 2 ) dp 0
( a r ) 2rdr
dx
4
R0
(A8)
The resulting total flow, the so-called Buckingham-Reiner equation (Tattersall & Banfill 1983)
can, after integration and some algebra, be written (Kaplan 2001):
4
4 2 0 1 2 0 a 4 dp 4 R0 R0
a dp
Gx =
=
+
1
1
+
8 dx 3 a ( dp ) 3 a ( dp ) 8 dx
3a 3a
dx
dx
(A9)
a 2
Gx
(A10)
Eq.(A7) and (A9) normally underestimate concrete flow. However, by evaluating R0 from eq.(6)
and calculating the flow profiles and total flow with eq.(A7) and (A9) one can assess whether a
concrete is far from, or close to plug flow. One may then simply fit the total flow to an estimated
plug radius R0 est. This means estimating the layer thickness (a-R0 est) and adjusting the rheological
properties of the slip layer, 0,SL and SL to some properties more viscous than of the bulk fresh
concrete (0, ). One may also calculate the specific plastic viscosity and yield shear of the slip
layer: 0,SL spec = 0,SL/(a-R0) and SL spec = SL/(a-R0) but perhaps even better; to use a slip layer
equivalent visco-thickness coefficient: (a-R0)2/SL. Important further work will be to obtain
experimental data on the existence, possible thickness and rheological properties of the slip layer.
39
APPENDIX 2
The Navier Stokes equation
Hydrostatic equilibrium of a fluid at rest is described from the mass force in equilibrium with
hydrostatic pressure:
p + f = 0
(A11)
f:
:
p :
p y
p x
p
+ k z and i,j,k unit
+ j
y
x
z
+ ( v ) = 0
t
(A12)
v:
t:
v :
+ vy
+ vz
x
y
z
vy
v
v
+ vy
+ vz
+ x +
+ z
Differentiating the product ( v ) = v x
x
y
z
y
x
z
Introducing the friction within the liquid due to the viscous behaviour results in Navier Stokes
equation:
v
1
= p + f ( v )v + 2 v
t
(A13)
2vx
x
2v y
y
2vz
z
Note that using (A13) with cylindrical coordinates for the uniaxial simple case in figure A1 one
can arrive at eq. (A4). In numerical solutions of (A13) using commercial FEM-software, we may
introduce other rheological models, thus solving for non-newtonian fluids.
-10,5
-12
0,25
-1,5
0,30
0,27
-9
1,5
0,30
0,29
-7,5
0,30
0,28
4,5
0,31
0,29
0,30
-6
7,5
0,29
0,30
0,28
-3
10,5
0,27
-4,5
12
0,25
0,31
13,5
0,22
0,30
15
0,20
19,5
0,09
18
21
0,05
16,5
22,5
0,00
0,13
(mm)
(m/s)
0,16
29962,4
(Pa/m)
dp/dx
v(r)no-plug
Material:
(r = a ved rand)
22,5
(mm)
Diameter:
Pa s
8,57
(bulk concrete)
57
Pa
0
.
mm
45
3,8
(mm)
R0
0,25
0,27
0,28
0,29
0,30
0,31
0,31
0,31
0,31
0,31
0,31
0,31
0,30
0,29
0,28
0,27
0,25
0,22
0,20
0,16
0,13
0,09
0,05
0,00
(m/s)
v(r)w plug
NB: error at R0
>a
R0=20/(dp/dx)
-12
-10,5
-9
-7,5
-6
-4,5
-3
-1,5
1,5
4,5
7,5
10,5
12
13,5
15
16,5
18
19,5
21
22,5
(mm)
Gx (lit/s)
0,025187925
0,027201562
0,028246179
0,028154961
0,026761096
0,023897769
0,019398167
0,013095476
0,004822882
0,272486973
2,66
0,98
1,33
2,7
4,6
8,57
2,05
5,78
3,7
7,4
21,4
23,1
25,5
14,2
16,5
Pa.s
plast.visc.
BML
529000
293000
283000
489000
332000
797000
476000
726000
735000
10000
100000
2000000
2000000
170000
30000
Pa
press.
Measured
96,19
5,59
21,82
82,74
228,49
57,05
61,42
144,81
146,05
55
130
875
873
13,3
4,4
Pa
yield str
BML
10
37
11
11
293
69
23
23
mm
Rplug=(2L/p)0
Velocity distribution over hose with no-slip plug flow, plug-radius R0 (Hagen Poiselle Bingham flow, which integrated -> Buckingham Reiner equation (Tatt&Ban 1983, Kaplan 2001)):
0,64
0,92
0,64
0,41
0,89
0,60
0,71
0,66
0,42
0,25
0,25
0,23
0,25
0,27
0,28
m/s
1,01
1,47
1,01
0,66
1,42
0,95
1,13
1,05
0,67
0,18
0,18
0,16
0,18
0,19
0,20
lit/s
Gx
vx
meas. meas.
0,27
0,17
G (lit/s) vx (m/s)
40
19,5
0,94
-6
-7,5
-9
-10,5
-12
3,34
3,16
2,94
2,69
3,73
3,49
1,5
3,69
-4,5
3,60
3,60
4,5
3,49
-3
3,34
3,69
7,5
3,16
2,94
-1,5
10,5
2,69
3,75
12
2,41
3,73
15
13,5
2,09
18
21
0,49
16,5
22,5
0,00
1,74
(mm)
(m/s)
1,36
v(r)no-plug
29962,4
(Pa/m)
dp/dx
22,5
(mm)
(a-
20,4
(mm)
Pa s
R0 est =
-9
-7,5
-6
-4,5
-3
-1,5
0,67
-12
REP batch
Reference 4mm 30 MPa
REP m-Tec
JMS batch
JMS m-Tec
EXM batch
EXM m-Tec
2,66
0,98
1,33
2,7
4,6
8,57
2,05
5,78
3,7
7,4
21,4
23,1
25,5
14,2
16,5
529000
293000
283000
489000
332000
797000
476000
726000
735000
10000
100000
2000000
2000000
170000
30000
Pa
Pa.s
press.
Measured
plast.visc.
BML
0,06163
0,07111
0,08059
0,09007
0,09955
0,10903
0,11851
0,11035
0,04979
0,96126
0,6 m/s
vx (m/s)
SL, spec
0,60
0,0225
0,0225
0,0225
0,0225
0,0225
0,0225
0,0225
0,0225
0,015
0,015
0,015
0,015
0,015
1,5 Gx (lit/s)
4,5
7,5
10,5
12
13,5
15
16,5
18
19,5
21
22,5
(mm)
0,67 -10,5
0,67
0,67
0,67
0,67
0,67
0,67
0,67
0,67
0,67
0,67
0,67
0,67
0,67
0,67
0,67
0,67
0,67
0,67
0,67
0,67
0,49
0,00
v(r)plug
0,015
R0 est =
SL
mm
(Slip Layer)
Pa
0 SL
Diameter: 45
(r = a at wall)
10
37
11
11
293
69
23
23
mm
460
417
372
Pa s/m
285
SL, spec
0,64
0,92
0,64
0,41
0,89
0,60
0,71
0,66
0,42
0,25
0,25
0,23
0,25
0,27
0,28
m/s
10
4,2
0,95
G (lit/s)
41
0,60
vx (m/s)
1,01
1,47
1,01
0,66
1,42
0,95
1,13
1,05
0,67
0,18
0,18
0,16
0,18
0,19
0,20
lit/s
vx meas.Gx meas.
(R0-a) mm
Rplug=(2L/p)0
476,1905Pa s/m
96,19
5,59
21,82
82,74
228,49
57,05
61,42
144,81
146,05
55
130
875
873
13,3
4,4
Pa
yield str
BML
Velocity distribution in hose with plug-flow and slip-layer. Plug-radius R0 fitted to measured flow by estimateing layer thickness (a-R0), yield and plastic viscosity of slip-layer
Appendix 4 Excel sheet for plug flow with adjustable slip layer modified Buckingham Reiner figure 17, 21, 22
JMS m-Tec
JMS batch
REP m-Tec
REP batch
Ref 4mm30MPa
42
43
44
SINTEF Building and Infrastructure is the third largest building research institute in Europe. Our objective is to promote environmentally
friendly, cost-effective products and solutions within the built environment. SINTEF Building and Infrastructure is Norways leading
provider of research-based knowledge to the construction sector. Through our activity in research and development, we have established
a unique platform for disseminating knowledge throughout a large part of the construction industry.
COIN Concrete Innovation Center is a Center for Research based Innovation (CRI) initiated by the Research Council of Norway. The
vision of COIN is creation of more attractive concrete buildings and constructions. The primary goal is to fulfill this vision by bringing
the development a major leap forward by long-term research in close alliances with the industry regarding advanced materials, efficient construction techniques and new design concepts combined with more environmentally friendly material production.
www.sintef.no