Rheol.

Acta 17, 446-453 (1978)

© 1978 Dr. Dietrich Steinkopff Verlag, Darmstadt
ISSN 0035-4511/ASTM-Coden: RHEAAK

Materials Handling Division, Warren Spring Laboratory, Stevenage, Herts (England)

Some observations on the rheological behaviour of dense suspensions

D. C.-H. Cheng and R. A. R i c h m o n d

With 5 figures
(Received December 20, 1977)

1. Introduction rheological property which we will describe

by the term "granulo-viscous". We will show
This paper deals with the rheology of "dense that both out results and those already noted
suspensions", by which we mean solid-liquid in the literature are examples of granulo-
mixtures in which the particle phase is present viscous behaviour.
in very high concentrations, well in excess of We first carried out numerous experiments
what is normally termed "concentrated sus- on a wide fange of materials. At first the results
pensions". The volume concentration could seemed quite chaotic, but on persisting distinc-
be in the region of 60% or more with the pores tive patterns began to emerge, not just on rePeat
completely filled with the liquid phase (saturated tests on the same sample, but also with different
systems), it could be such that the solids are in materials. From the standpoint of viscous be-
close-packed configuration or near to it, or it haviour, even non-Newtonian behaviour, such
could even be unsaturated so that the pores patterns would appear to be anomalous. Howev-
are only partially filled and air is present. er, it eventually became clear to us that these
Some isolated papers have been published on anomalies could be explained i n terms of
saturated systems. We shall refer to them below granular solids behaviour of the particles, which
in appropriate paragraphs. is well known in soil and powder mechanics.
In the past few years, Warren Spring Labora- Once this explanation is accepted the anomalies
tory has, mainly in cooperation with ifi'dustrial are seen to be predictable. In this paper therefore
sponsors, carried out extensive studies into the we would use this approach to introduce
properties of both saturated and unsaturated granulo-viscous phenomena. We will show how
dense suspensions. Examples are sand, mine one might have predicted them from a com-
tailings, coal, colliery waste, mineral concen- bination or inter-action of prior knowledge
trates, cement grouts, graphite-in-oil pastes. of viscous behaviour and that of granular
They also include fermentation broths, fruit behaviour, and we will cite experimental results
puree, barley meal-water animal feed and sewage from our work and published papers as il-
sludges. The results of these studies are ulti- lustration.
mately applied to technological problems such
as storage hopper discharge, pipeline design,
reactor scale-up, stability of tailings dumps, 2. Experiments
stockpiles and ships' cargoes, and manufacture The instruments used in the WSL work
of lead batteries, foundry moulds and welding included several rotational viscometers. In the
electrodes. Some of these technological in- Contraves Rheomat 15 FC Viscometer, samples
vestigations are still in progress, but the ob- were subjected to various shear cycles, using
servations on dense suspension behaviour are smooth-walled and roughened bobs and cups.
sufficiently advanced for a generalised descrip- In some experiments, the moisture concentra-
tion to be given. In this paper we will advance a tion was measured as a function of height i n
qualitative description of this behaviour, and the sample. The Ferranti-Shirley Cone-Plate
show that it constitutes a distinct category of Viscometer was used in the usual way and also
454
 Cheng and Richmond, Some observations on the rheological behaviour of dense suspensions 447

with the cone displaced from the plate, as an density, there is a set of critical normal and
approximation to parallel plate viscometer. shear stresses, which defines its dilation/con-
In the Weissenberg Rheogoniometer, serrated solidation behaviour. For stress within bounds,
parallel plates were used on unsaturated samples the solids are stable and do not undergo ir-
and in oscillatory testing, subjecting samples to reversible deformation. But on exceeding the
cycles of increasing and decreasing amplitudes bounds, the solids will deform irreversibly.
at different frequencies. A Davenport Extrusion On consolidation, the solids tend to densify en
Rheometer was used with tubular dies in masse, but dilation tends to occur locally,
conventional testing. All these instruments have resulting in the development of "slip planes", of
been described previously (1). Other tests several particle-diameters in thickness, within the
were carried out using a Ring Shear Cell (2), bulk. Over quite wide limits the shear stress
not usually used in investigations into fluid in the slip plane is independent of the shear
rheology. This last instrument is derived from rate or velocity difference across the thickness,
the soil mechanics shear cell and the powder :but increases as the normal stress in the slip
mechanics Jenike Flow Factor Tester. It meas- plane increases. Under steady shearing con-
ures the shear stress developed in granular ditions, the local bulk density in (the thickness
and particulate solids, when subjected to dif- of) the slip plane, varies with the normal stress
ferent normal stresses. The Atterberg Liquid also.
Limit Test (3) was also used. For another test, On e can easily see that ifa suspension contains
in order to have bettet control of the conditions a conCentration of particles high enough for
of test, carefully shaped samples of known bulk granular behaviour to be significant, this is
density were subjected to controlled vibration going to modify the viscous behaviour which
on a Ling Vibrator. This test was similar to the is obtained by virtue of the presence of the
standard flow or slump tests used for testing liquid phase. Hence the term "granulo-viscous
cement and concrete etc. (4) and that described behaviour" in describing dense suspensions.
in the IMCO Code of Safe Practice (5). Some examples will illustrate how granular
behaviour might affect viscous behaviour and
what might be expected of granulo-viscous
3. Viscous and granular bases of granulo-viscous behaviour.
behaviour
The discussion of granulo-viscous behaviour 4. Features of granulo-viscous behaviour
is best begun by recapitulating on viscous
behaviour. Under steady shearing, viscous be- 4.1. Stick-slip phenomenon
haviour is described by the shear stress-shear
The first example of granulo-viscous be-
rate relationship, or flow curve. This is charac-
haviour could be simply described as stick-
terised by the feature that the shear stress
slip, which is illustrated by fluctuating torque
increases as the shear rate increases, and the
in a rotational viscometer at steady rotation.
flow curve has a positive slope. Unless ex-
The relatively slow rise followed by rapid fall,
tremely high pressures are involved, the flow
and repeated at regular periodicity, can be
curve is independent of pressure or normal
readily explained in terms of jamming and
stress. Viscous fluids are usually taken to be
release of the particles. We have found this
incompressible.
behaviour to a greater or lesser extent in the
In contrast the behaviour of dry or nearly
materials we have studied. Kao et al. (7) observed
dry particulate solids (granular behaviour) is
fluctuating rotational speeds with glass beads
characterised, firstly, by the shear stress being
in a Stormer viscometer (their fig. 12).
dependent on normal stress and bulk density
and not on shear rate, and, secondly, by com-
4.2. Changes in flow curve and packing density
pressibility, i.e. the propensity of the lattice
variation
of particles either to dilate or to compact/
consolidate depending on the stresses current That granulo-viscous material shows viscous
at the time. These conclusions are well known behaviour (torque increasing as speed increases)
results from soff and powder mechanics (6~,~ is shown in figure I for a red lead oxide slurry
They also mean that, associated with any bulk in which are plotted the peak torques obtained
448 Rheolo9ica Acta, Vol. 17, No. 4 (1978)

100 I

R H E O M A T 15
SAMPLE D
SYSTEM Cs/C2s
RUN 22 UP O, DOWN •
RUN 23 UP In DOWN •

80

0
"N 6 0 -

D
0 40 -
n-
O
1-

~ /
1-2

20 -
.d"
# ,,,13//

0 I I
0 5 10 15
Fig. 1. Viscometric results for a red lead
SPEED NUMBER oxide slurry

during the stick-slip phenomenon. There are to give a different packing density (or even a
two distinct flow curves, one annotated by different lattice configuration). As long as the
letters of the alphabet, the other by numerals. density remains unchanged, the suspension
The figure also shows yield behaviour charac- shows viscous behaviour giving a flow curve of
teristic ofgranular solids, in the sudden transition positive slope. But depending on the hydro-
from one flow curve to the other. This can be dynamic forces at play, it can jump to a different
explained in terms of the response of the particles flow curve corresponding to a change in packing
to stresses acting on them. These stresses are density. Figure 1 illustrates a jump to a higher
derived from the hydrodynamic interaction density on decreasing speed. We have observed
between the solid and liquid phases and depend a j u m p to lower density in a variety of materials,
on the shear rate as well as the current packing as well as the opposite density changes with
density of the particles. It may be that at some speed changes. Hoffman (8) and Wagstaff and
set of conditions the stresses remain within the Chaffey (9) also observed sudden increase in
bounds for stability, then as the shear rate or viscosity in monodispersed spheres that can
viscometer speed is changed, the packing den- be interpreted in the same way.
sity remains unaltered. The viscometer response Gradual changes in dilatant behaviour are
would be expected to give a flow curve with observed in our work as in published data
positive slope. But if the speed is changed (8, 1J) equally as sudden changes. The main
outside the bounds, the particles would yield factor involved is solids concentration. The
 Chen9 and Richmond, Some observations on the rheological behaviour of dense suspensions 449

I I
RHEOMAT 15
SYSTEM DgbD29

80 MASTER MC 29'9°/o B
EST. MIDDLE MC 2 3 " 0 % BB
UP DOWN
ist CYCLE --o --
2nd CYCLE -D---mm-

6O
m ~ R "~~ «D'~D
.£
ul
I~ B "D--o..
>

«, 4 0
m-

oe ¢
£ [3

20

0 I I I I
I 5 10 15 Fig. 2. Viscometric results for a mine
SPEED NUMBER rfS-~ tailing

higher it is, the more one would encounter at the initial stages of test, but a series of equilib-
catastrophic behaviour. The exact response of rium densities may be reached which is repro-
the suspension depends on the way the visco- duced if the speed cycling is repeated.
meter is used, for example the rate at which the It is, of course, appreciated that there is no
speed is applied, cf. Strivens (12), and the exact easy way to measure directly the density
state of solids packing at the start of test. changes. In situ methods are required, if the
Reproducibte results, as illustrated in figure 1, suspension is not to be disturbed by sampling,
can obviously be obtained, but in other examples, and estimates suggest that discrimination of
repeat tests on different samples of the same much better than 1 part in 1000 is needed be-
suspension do not give the same results. But cause very small changes in packing density
whatever the details of the individual behaviour, have very pronounced effects. It is thought that
they can all be explained in terms of granulo- such changes in density will perhaps never be
viscosity. directly measured, but can only be inferred
Particularly with coarse materials, such as from observations on changes in stresses.
a sand of median size of 180 ~tm diameter
(fig. 2) the adjustment of packing density takes
4.3. Wall effects
place at every step of speed change, to such an
extent that a curve showing decreasing torque Other evidence for granulo-viscous behaviour
with speed increase is obtained. It is easily comes from investigations into wall effects,
appreciated that the packing density decreases when tests were carried out using both smooth
with speed to such an extent that its effect bobs as well as roughened bobs. Figure 3 refers
outweighs the viscous response of torque in- to results obtained on a fresh cement grout;
creasing with speed at constant density. Figure 2 containing an inert filler, which shows no
also shows that the density change is irreversible appreciable chemical thickening over the test
450 Rheologica Acta, Vol. 17, No. 4 (1978)

100
I I I Il I
-- 40

o 80

._
"o

-- 30

UPPER BOUND

RI

w 60
\
O
I-
Z~
I-
Z w
tu o~
ùJ
RHEOMAT 15 20 I---
> i/I
1 A CUP-1L BEAKER Iz
UP DOWN
8 • A MIX BOB C U R V E CURVE ul
-Il
40
i,o
O 31 B2s O •
A 36 B2s A •
O 35 B29 [] •

THE BASIS FOR EQUIVALENT

SPEED AND TOP,QUE IS B2S BOB. 1 0

20
LOWER BOUND

I I I I
1 5 10 15
SP£•D NUMBEA OR EOUIVALENT SPEED NUMBER

Fig. 3. Rheomat Viscometer results for cement grout (smooth and rough bobs)

period. On the whole, a smooth bob gave lower surprising phenomenon, that the rough-bob
torques than a rough bob, as one would expect torque was lower than the smooth-bob torque.
from the conventional concept of wall slip, This can be explained by noting that granular
see Chen9 and parker (13). But figure 3 shows solids flow by developing a slip surface. One
that on occasions, the smooth-bob torque was would expect this to develop certainly at the
as high as the rough-bob torque. These~ob- smooth-bob surface, where the shear stress is
servations can be explained by postulating that highest. But if the solids are positively captured
the torque is governed by the local solids on the rough bob, the slip plane could well
density at or near to the bob surface. With a form in the bulk of the sample. If then we
smooth bob, density tends to be low, whereas postulate that the torque is derived from
particles are caught by the rough surface frictional-type interaction in the slip surface,
whereupon the local density would be high. it could well be that the friction between barley
But at times, the density round a smooth bob meal and steel is higher than that between meal
could also be high because of its earlier history, and meal, which is reflected in the torque
then it Would give torques that equal the rough- observed.
bob torques. Additional support for frictional effects is
Figure 4 taken from Richmond (14) referring derived from pipe flow data obtained from the
to a 2:1 barley meal-water mixture, shows a same barley meal suspension (fig. 5). The pipe
 Chen 9 and Richmond, Some observations on the rheolooical behaviour o f dense suspensions " 451

1 I i I ] I I
100

,'1 I ~~g__---::
80 ù II B~,,__.

INITIAL / 1/ / /
PEAKS / / # //
/ , // , /
O

"Fn 6 0

8
ul

D 40
0

/ / /
20

~'~STEADY TORQUES

O L

0 2 4 6 8 10 12 14
SPEED NUMBER Fig. 4. Viscometric results for a meal-water mixture

103 i i i i i i Ii I i i i i i i i I i i i i i i i

• C2s[Bg I B2s/A 9 3" PlPELINE

• C 2 s p/Bg 0 B2g/Aq • ,4 h SOAK
O B2s[21 PEAKS • 2 h SOAK
O B2sp/21 X
A B2g/21 +
RH EOMAT 15

102
U
13.

i/)
• 0 •I ~
hl
Œ
t- o +Il<: •
0 ++)~ x (~ *~' o~
r~ x •, v , [:~ll
w °o~ .ql All
T
o A A ^'•~l~lö
II~n
10 O A []u
A cP
AAô [][]

1 I i i I I i il I I I I I IIII I I I I I I I

IO 10 2 10 3
SHEAR RATE, s-i
!8 8631

Fig. 5. Viscometric and pipeline results for the meal-water mixture of fig. 4
452 Rheologica Acta, Vol. 17, No. 4 (1978)

is made of ABS plastic. The intermediate shear stress should form an integral part of future
stress obtained hefe would indicate that the study on dense suspensions. At WSL, a variable
meal-plastic friction is intermediate between speed Ring Shear Cell has been developed and
that with steel and meal with itself. modifications of the rotational viscometers are
planned t o control the applied normal load.
4.4. Highly flocculated materials The former instrument has started to be used
on materials such as putty and findings will be
In the same class as barley meal, suspensions reported on another occasion.
such as fermentation broths, sewage sludges
and others already named, show dense suspen-
sion and granulo-viscous behaviour at solids 5. Theoretical investigations
contents as low as a few per cent by weight. Theoretical investigation into the viscosity
This is because the particles, being fibrous or of concentrated suspensions has received much
glutinous, form open flocs. These flocs with attention in the literature (16, 17) and equations
the entrapped liquid behave as single entities derived cover concentration levels up to the
that make frictional contact with each other, point when viscosity becomes infinitely large.
so imparting granulo-viscous characteristics on Although such concentrations should show
the suspension as a whole. We have observed granulo-viscous behaviour, none of the theories
in these suspensions the same features of predict this because no account was taken of
behaviour as found in suspensions of hard particle-particle frictional interaction and gran-
particles. ular behaviour. Metzner and Whitlock (10) cän
be said to have opened the discussion on dense
4.5. Liquefaction suspensions, as opposed to just concentrated
If one studies the behaviour of dense suspen- ones but they did not consider granular be-
sions as a function of composition, the pheno- haviour as our discussion above has done.
menon of liquefaction is encountered. A Satu -~ Other publications can be mentioned as rele-
rated or unsaturated material may be of a vant to the study of dense suspensions (18-21).
moisture content that will normally possess They used either a continuum approach or
significant strength to resist flow. But under mechanistic modelling. Of the latter, Hoffman
certain conditions, it will lose this strength (21) can be noted as predicting the critical shear
catastrophically. Practical examples of this are rate for sudden increase i n viscosity. But his
found in soil mechanics (sandy foundations and theory involves linearisation and does not other-
quick clays) and in ships' cargo stability, Green wise describe non-Newtonian, far less gran-
and Hughes (5). We have carried out funda- ulo-viscous behavmur. It would seem that the
mental studies into this phenomenon, Cheng whole area of dense suspension rheology is
(15), but limits of space prevent further discus- ripe for fundamental research.
sion here. Summary
A review is given of current knowledge of the rheology
4.6. Influence of normal stress of solidqiquid mixtures of very high solids concentra-
An important feature of granular behaviour tions, including unsaturated systems. Their behaviour
is traced to viscous behaviour of fluids and granular
is the dependence of shear stress and density behaviour of particulate solids, and termed "granulo-
on normal stress. One would therefore expect viscous". Various features of granulo-viscous behaviour
that this would have a counterpart in granulo- are described: stick-slip, ehanges in flow curve, packing
viscous behaviour. However all the published density variation, wall effects, behaviour of highly
work on dilatant viscosity and most of our flocculated materials, liquefaction, and influence of
normal stress. Theoretical work that exists in the
viscometric work were carried out under at- literature is referred to, but it would seem that granulo-
mospheric pressure. What work we have done viscous behaviour is open for fundamental research.
in the Ring Shear Cell, where the normal load
was varied, has mostly been at the same rota- Zusammenfassung
tional speed. Thus, it is clear that the investiga- Es wird ein Überblick über den gegenwärtigen
tion of the dependence of the flow curve or Kenntnisstand in der Rheologie von Feststoff-Flüssig-
shear stress-shear rate relationships on normal keits-Mischungen mit sehr hohen Feststoff-Konzen-
 Cheng and Richmond, Some observations on the rheological behaviour of dense suspensions 453

trationen einschließlich ungesättigter Systeme gegeben. 9) Wagstaj]i I., C. E. ChafJey, J. Colloid Interf.
Ihr Verhalten wird sowohl zum viskosen Verhalten Sci. 59, 53 (1977).
von Flüssigkeiten als auch zum Granulat-Verhalten 10) Metzner, A. B., M. Whitlock, Trans. Soc.
körniger Feststoffe in Beziehung gesetzt und als Rheol. 2, 239 (1958).
ùgranulo-viskos" bezeichnet. Verschiedene Merkmale 11) Umeya, K.~ in: S. Onog i (Ed.), Proc. 5 Intern.
des granulo-viskosen Verhaltens werden beschrieben: Congr. Rheol., Vol. 2. Univ. Tokyo Press, Tokyo,
Haft-Gleit-Phänomene, Veränderungen der Fließkurve, p. 295 (1970).
Variation der Pa'ckungsdichte, Wandeffekte, Verhalten 12) Strivens, T. A., J. Colloid Interf. Sci. 57, 476
von hochgradig ausgeflockten Stoffen, Verflüssigung (1976).
und Einfluß der Normalspannungen. In der Literatur 13) Cheng, D. C.-H., B. R. Parker, In: C. Klason,
vorgefundene theoretische Untersuchungen werden J. Kubat (Eds.), Proc. 7 Intern. Congr. Rheol. Gothen-
ebenfalls erwähnt, doch es scheint, daß das granulo- burg, p. 518 (1976).
viskose Verhalten noch auf eine grundlegende Er- 14) Richmond, R. A., WSL Res. Rept. No. LR
forschung wartet. 248 (MH), 1977.
15) Cheng, D. C.-H., Contribution to discussion
ReJèrence~ at Soc. Underwater Tech. and Instn. Mining Metallurgy
Meeting, WSL, 3 March 1977 (Ref. 5).
1) Cheng, D. C.-H., in: Proc. Symp. Phys. Props 16) Rutgers, R., Rheol. Acta 2, 202, 305 (1962);
Liquids and Gases for Plant and Process Design, 3, 118 (1963).
East Kilbride, 1968. Edinburgh : HMSO, 1970, p. C 48. 17) Goddard, J. D., J. Non Newtonian Fluid Mech. 2,
2) Sutton, H. M., WSL Res. Rept No. LR 148 (MH), 169 (1977).
1971. 18) Yerushalmi, J., S. Katz, R. Shinnar, Chem.
3) Taylor, D. W., Fundamentals of soil mechanics. Engng. Sci, 25, 1891 (1970).
Wiley & Sons, New York 1948; BS 1377: 1967; 19) Lindsey, G. H., S. A. Murch, Trans. Soc; Rheol.
ASTM D 423-66. 17, 487 (1973).
4) BS 890: 1972. Speeification for building limes; 20) Di Giovanni, P. R., S. L. Lee, J. appl. Mech. 41 E,
BS 1881: Part 2: 1970. Methods of testing fresh 35 (1974).
concrete; BS 4551 : 1970. Methods of testing Mortars. 21) Hoffmann, R. L., J. Colloid Interf. Sci 46, 491
5) Green, P. V., T. H. Hughes, Trans. Instn. Min. (1974).
Metall. (Sect. A), 86, A 150 (1977).
6) Ashton, M. D., D. C.-H. Cheng, R. Farley,
F. H. H. Valentin, Rheol. Acta 4, 206, 1965 ; D. C.-H. Authors' address:
Cheng, WSL Res. Repts. Nos. LR 48 (CE), LR 49 (CE) Dr. D. C.-H. Cheng, Mr. R. A. Richmond
1967. Materials Handling Division
7) Kao, S. V., L. E. Nielson, C. T. Hill, J. Colloid Warren Spring Laboratory
Interf. Sci. 53, 358 (1975). P. O. Box 20, Gunnels Wood Road
8) Hoffman, R. L., Trans. Soc. Rheol. 16, 155 (1972). Stevenage, Herts. SG 1 2 BX (England)