You are on page 1of 6

FUNDAMENTALS OF COHESIVE POWDER

CONSOLIDATION AND FLOW

J. Tomas
Mechanical Process Engineering, The Otto-von-Guericke-University
Universitätsplatz 2, D – 39 106 Magdeburg, Germany

Abstract - The fundamentals of cohesive powder consolidation


and flow behaviour are explained to combine reasonably particle
and continuum mechanics. The influence of elastic-plastic re-
pulsion and, consequently, stressing pre-history dependent adhe-
sion is demonstrated by the model “stiff particles with soft con-
tacts”. The flow properties of a cohesive limestone powder (d50
= 1.2 µm) are shown. These models are used to evaluate shear
cell test results as constitutive functions for computer aided ap-
paratus design for reliable flow.

1. INTRODUCTION

The well-known flow problems of cohesive particulate solids in storage and transportation con-
tainers, conveyors or process apparatuses - mainly mentioned by Jenike [1] - leads to bridging,
channelling, oscillating mass flow rates and particle characteristics associated with feeding and dos-
ing problems. Taking into account this list of technical problems, it is essential to deal with the fun-
damentals of cohesive powder consolidation and flow behaviour, i.e. to develop a reasonable com-
bination of particle and continuum mechanics.

2. PARTICLE CONTACT CONSTITUTIVE MODEL

The well-known failure hypotheses of Tresca, Coulomb-Mohr, the yield locus concept of Jenike
[1] and Schwedes [2] were supplemented from Molerus [3] by the cohesive steady-state flow crite-
rion. The consolidation and non-rapid, frictional flow of cohesive powders was explained by the
adhesion forces at particle contacts [3]. His advanced theory is the suitable basis of the extended
and generalised model approach [4, 5] which is shown now.
In principle, there are four essential deformation effects in particle-surface contacts and their
force-response behavior can be explained as follows, see review [4]:
• elastic contact deformation which is reversible, independent of deformation rate and consolida-
tion time effects and valid for all particulate solids,
• plastic contact deformation with adhesion which is irreversible, deformation rate and consolida-
tion time invariant, e.g. mineral powders,
• viscoelastic contact deformation which is reversible and dependent on deformation rate and con-
solidation time, e.g. bio-particles,
• viscoplastic contact deformation which is irreversible and dependent on deformation rate and
consolidation time, e.g. nanoparticles fusion.
This paper is intended to focus on the model of isotropic, stiff, linear elastic, mono-disperse spheri-
cal particles that are approaching to soft contacts by attractive adhesion forces of particle surfaces,
Fig. 1. Thus, this soft or compliant contact displacement is assumed to be small hK/d << 1 compared
to the size (diameter) of the stiff particle. During this surface stressing the stiff particle is not so
much deformed that it has a remarkable change of the particle shape. But by way of contrast, soft
particle matter such as biological cells or structured macromolecular organics do not behave so.
These particle mechanics should be explained by a typical force-displacement diagram [4, 6].
The origin of this diagram hK = 0 is equivalent to the characteristic adhesion separation of a direct
contact aF=0. After approaching FN ∝ a −2 from an infinite distance -∞ to a minimum separation a =
aF=0 the sphere-sphere contact without any contact deformation is formed by the long-range attrac-
tive adhesion force FH0 (the so-called jump in). Then the contact may be loaded FH0 – Y and, as the
response, is elastically deformed with an approximated circular contact area, Fig. 1b). With increas-
ing external normal load this soft contact starts at a pressure pf with plastic yielding at the point Y
Fig. 1c). The micro-yield surface is reached and this maximum pressure can not be exceeded and
results in a combined elastic-plastic yield boundary of the partial nanoplate-plate contact. This
force-displacement behaviour may be described by a very comfortable linear model with the parti-
cle centre approach (height of flattening or overlap) of both particles hK:
FN + FH 0 = π ⋅ r1, 2 ⋅ p f ⋅ (κ A − κ p ) ⋅ h K (1)

c) elastic-plastic deformation to
Fig. 1: Force-displacement
a nanoplate - plate contact
diagram of particle approach
a), loading path with elastic
contact deformation b), elas-
b) elastic contact
deformation hK tic-plastic deformation to a
nanoplate-plate contact c),
unloading path by particle
FN detachment with “frozen”
a) particle approach hK FN plastic contact flattening d),
which was recalculated from
loading
shear strength of the cohesive
U limestone powder (particles
- hK elastic-plastic as spheres with median di-
50 yield boundary ameter d50 = 1.2 µm).
unloading
normal force
FN in nN

reloading

HERTZ centre
approach
0
FH0 Y 0.1 0.2 0.3
hK in nm
elastic
recovery
d) particle detachment A
pull-off force FN,Z

adhesion boundary
-50
- hK

FN,Z
This line is shown in Fig. 1
as elastic-plastic yield boundary with the averaged particle radius at contact of two spheres 1 and 2
r1,2 = (1/r1 + 1/r2)-1. Thus predominant plastic yielding behaviour provided, the contact stiffness
decreases with smaller particle size d = 4.r1,2 (or micro-roughness radius of non-deformed contact)
especially of cohesive nanopowders:
= π ⋅ r1, 2 ⋅ p f ⋅ (κ A − κ p )
dFN
k N ,pl = (2)
dh K
A confined plastic field is formed inside of the contact circle. The elastic-plastic contact area coef-
ficient κA represents the dimensionless ratio of plastic particle contact deformation area Apl to total
contact deformation area A K = A pl + A el . This includes a certain elastic displacement mainly at the
perimeter of contact circle:
2 A pl
κA = + (3)
3 AK
The plastic repulsion coefficient κp describes a dimensionless ratio of attractive van der Waals pres-
sure pVdW to repulsive particle micro-hardness pf that is related to a plate-plate model:
p VdW C H ,sls
κp = = (4)
pf 6 ⋅ π ⋅ a 3F=0 ⋅ p f
The characteristic adhesion distance lies in a molecular scale aF=0 ≈ 0.3 - 0.4 nm and depends
mainly on the properties of liquid-equivalent packed adsorbed layers. This separation can be esti-
mated for a molecular interaction potential minimum − dU / da = F = 0 = Fat + Frep or force equilib-
rium. Provided that these molecular contacts are stiff enough compared with the soft particle con-
tact behaviour influenced by these mobile adsorption layers due to molecular rearrangement, this
separation aF=0 is assumed to be constant during loading and unloading in the interesting macro-
scopic stress range σ < 100 kPa. The Hamaker constant (index solid-liquid-solid) CH,sls due to Lif-
schitz theory is related to continuous media and depends on their dielectric constants and refractive
indices.
Constant mechanical properties provided, the finer the particles the smaller is the yield point Y
in Fig. 1 [6]. Thus, an initial elastic contact deformation has no relevance for cohesive powder con-
solidation and may be excluded. But after unloading beginning at point U along curve U – A, the
contact recovers elastically in the compression mode and remains with a plastic displacement hK,A:

⋅ E * ⋅ r1, 2 ⋅ (h K − h K ,A ) − FH ,A
2 3
FN ,unload = (5)
3
But along the symmetric curve A - U the contact may be reloaded:

FN ,reload = − ⋅ E * ⋅ r1, 2 ⋅ (h K ,U − h K ) + FN ,U
2 3
(6)
3
If one applies a certain pull-off force FN,Z = - FH,A, here negative,
FH ,A = FH 0 + π ⋅ r1, 2 ⋅ p VdW ⋅ h K ,A (7)

the adhesion (failure) boundary at point A is reached and the contact plates are failing and detach-
ing with the increasing distance a = aF=0 + hK,A - hK. This actual particle separation can be consid-
ered for the calculation by means of a long-range hyperbolic adhesion force curve FN,Z ∝ pVdW(a) ∝
a-3 from the plate-plate model as given from Eq.(4).
The slopes of elastic-plastic yield and adhesion boundaries in Fig. 1 are typically for irreversible
particle contact stiffness. Consequently, if one eliminates the displacement hK a non-linear adhe-
sion-normal force function FH = f(FN) is obtained [6] which can be linearised as shown in Fig. 2:
κA κp
FH = ⋅ FH 0 + ⋅ FN = (1 + κ ) ⋅ FH 0 + κ ⋅ FN (8)
κA − κp κA − κp

The dimensionless elastic-plastic contact consolidation coefficient κ = κp/(κA - κp) is given by the
slope of adhesion force FH influenced by predominant plastic contact failure. This displacement
coefficient κ characterises the irreversible particle contact stiffness or softness as well. A shallow
slope implies low adhesion level FH ≈ FH0 because of stiff particle contacts, but a large slope means
soft contacts, or i.e., a cohesive powder flow behaviour. This model considers, additionally, the
flattening of soft particle contacts caused by the adhesion force κ⋅FH0. Thus, the total adhesion force
FH consists of a stiff contribution FH0 and a displacement influenced component κ.(FH0 + FN), Fig. 2.

Fig. 2: Recalculated particle


125 contact forces of limestone
FH(FN) t = 0 h
FH,tot(FN) t = 24 h FH,tot = (1+κvis) .FH0 + κvis. FN according Eq.(8), median
diameter d50 = 1.2 µm,
adhesion force FH in nN

100 surface moisture XW = 0.5


FN %. The points characterise
the normal stress levels of
75 YL 1 to YL 4 according to
d

FH Fig. 4. The total adhesion


force FH,tot after a time con-
50 κΑ κ .F solidation is also shown in
FH =κ - κ . FH0 +
d

p
A p κA - κp N Fig. 2. This time variable

FN slope is given by the elastic-


25 plastic and viscoplastic con-
tact consolidation coeffi-
cient κvis which is obtained
0.0 by the sum of rheological
- 25 0 25 50 75 100 125
normal force F in nN models [6].
N

This Eq.(8) can be inter-


preted as a general linear particle contact constitutive model, i.e. linear in forces, but non-linear
concerning material characteristics.

3. COHESIVE POWDER FLOW CRITERIA

Using the linearised elastic-plastic particle contact constitutive model Eq.(8) the failure condi-
tions of particle contacts are formulated [5]. It should be noted that the stressing pre-history of a
cohesive powder flow is stationary (steady-state) and delivers significantly a cohesive stationary
yield locus in the τ-σ-diagram of Fig. 3:
τ = tan ϕ st ⋅(σ + σ 0 ) (9)
This is characterised by a dynamic equilibrium of simultaneous contact shearing, unloading and
failing, creating new contacts, loading, reloading, unloading and shearing again. The isostatic ten-
sile strength σ 0 = (1 − ε 0 ) / ε 0 ⋅ FH 0 / d 2 of an unconsolidated powder without any particle contact de-
formation is obtained from the adhesion force FH0 with the initial porosity ε0 of very loose packing.
For the combination of angle of internal friction ϕi for incipient contact failure (slope of yield lo-
cus) with the stationary angle of internal friction ϕst the relation was derived as follows [3, 5]:
tan ϕ st = (1 + κ ) ⋅ tan ϕ i (10)
The softer the particle contacts, the larger are the differences between these friction angles and con-
sequently, the more cohesive is the powder response.
The linearised yield locus, Fig. 3, that is obtained from resolution of a general square function
[5], is simply to use (σM,st, σR,st centre and radius of Mohr circle for steady-state flow):
 σ 
τ = tan ϕi ⋅ (σ + σ Z ) = tan ϕi ⋅  σ + R ,st − σ M,st  (11)
 sin ϕi 
Fig. 3: Yield characteristics of a
Yield Locus Stationary Yield Locus cohesive powder in the τ-σ-
diagram. In general, the steady-
state flow of a cohesive powder is
shear stress τ

end point
Consolidation Locus cohesive. Hence, the total normal
σR,st=(σ1-σ2)/2 stress consists of an external
ϕi contribution σ, e.g. by weight of
τc σR powder layers, plus (by absolute
ϕst σM ϕi value) an internal contribution by
0 σ2 σc pre-consolidation dependent ad-
σΖ σ0
σM,st=(σ1+σ2)/2 σ1 σiso
normal stress σ
hesion (tensile stress σZ).

The consolidation locus and instantaneous yield locus describe the boundaries of incipient plastic
deformation during consolidation and flow. The stationary yield locus is the envelope of all Mohr-
circles for steady-state flow (critical state line) with a certain negative intersection of the abscissa
σ0. With the derivation of the linear yield locus, Eq.(11) and Fig. 3, the uniaxial compressive
strength or unconfined yield stress σc is also found as a linear function of the major principal stress
σ1, Fig. 4:
2 ⋅ (sin ϕ st − sin ϕ i ) 2 ⋅ sin ϕ st ⋅ (1 + sin ϕ i )
σc = ⋅ σ1 + ⋅σ (12)
(1 + sin ϕ st ) ⋅ (1 − sin ϕ i ) (1 + sin ϕ st ) ⋅ (1 − sin ϕ i ) 0
Fig. 4: Constitutive con-
solidation function of lime-
25
uniaxial compressive strength σc in kPa

ffc < 1 stone powder, straight line


σc(σ1) t = 0 hardened 1 < ffc < 2
σct(σ1) t = 24 h non flowing very cohesive regression fit = 0.96, d50 =
20 1.2 µm, surface moisture
σ1 σc XW = 0.5 % accurately ana-
lysed by Karl Fischer titra-
15 tion. Additional flow prop-
2 < ffc < 4 erties according to this ba-
cohesive
sic model Eq.(12) are the
10 averaged angle of internal
friction ϕi = 37°, the sta-
10 < ffc < 4
easy flowing tionary angle of internal
5 friction ϕst = 43° and the
free flowing isostatic tensile strength of
ffc = σ1/σc ≥ 10 the unconsolidated powder
0
-5 0 5 10 15 20 25 30 35 40 σ0 = 0.62 kPa.
consolidation stress σ1 in kPa A considerable time con-
solidation under the major principal stress σ1 after one day storage at rest is also shown in Fig. 4.
Equivalent linear functions are also used to describe these time effects [5].

4. CONCLUSIONS

A complete set of physically based equations for steady-state flow, incipient powder consolida-
tion and yielding has been shown. Using this, the yield surfaces due to theory of plasticity may be
described with very simple linear expressions in centre and radius stresses σM, σR:
 σ R − sin ϕi ⋅ (σ M − σ M ,st ) − σ R ,st yield locus (YL)

Φ YL, SYL , CL = 0 =  σ R ,st − sin ϕst ⋅ (σ M ,st + σ 0 ) stationary yield locus (SYL) (13)
σ − sin ϕ ⋅ (− σ + σ ) − σ consolidation locus (CL)
 R i M M ,st R ,st

The consolidation and yield loci and the stationary yield locus are completely described only with
three material parameters, i.e. angle of internal friction ϕi, stationary angle of internal friction ϕst,
isostatic tensile strength of an unconsolidated powder σ0 plus the characteristic pre-consolidation
(average pressure at steady-state flow) influence σM,st.
This physical approach can be used for modern data evaluation of various of powder flow prop-
erties concerning particle size distribution (nanoparticles to granules), moisture content (dry, moist
and wet), material properties (minerals, chemicals, pigments, waste, plastics, food etc.), which have
been tested and evaluated for more than the last 20 years. Thus, these models are directly applied to
evaluate the test data of a new oscillating shear cell [7] and a press-shear-cell in the high-level pres-
sure range from 50 to 2000 kPa of liquid saturated, compressible filter cakes [8].
Finally, these models are used as constitutive functions for computer aided data evaluation of
shear cell test results and process apparatus design for reliable flow. For example, the limestone
powder, Fig. 4, delivers a practically reasonable, minimum outlet width for a conical hopper bmin,st =
0.77 m for steady-state flow, but bmin = 0.82 m for incipient yield and incredible bmin,t = 4.2 m after
time consolidation at rest t = 24 h for gravitational flow without any flow promotion.

5. REFERENCES

1 A.W. Jenike, Storage and flow of solids, Engng. Exp. Stat. Bull. No. 123, Univ. Utah, 1964
2 J. Schwedes and H. Wilms, Fließeigenschaften von Schüttgütern, pp. 39-58, in: Martens P, Silo
– Handbuch, Ernst & Sohn Berlin, 1988
3 O. Molerus, Effect of interparticle cohesive forces on the flow behaviour of powders, Powder
Technology 20, pp. 161-175, 1978
4 J. Tomas, Assessment of mechanical properties of cohesive particulate solids – part 1: particle
contact constitutive model, Particulate Sci. & Technology 19, pp. 95-110, 2001
5 J. Tomas, Assessment of mechanical properties of cohesive particulate solids – part 2: powder
flow criteria, Particulate Science & Technology 19, pp. 111-129, 2001
6 J. Tomas, Zur Mechanik trockener kohäsiver Schüttgüter, Schüttgut 8, pp. 522-537, 2002
7 Th. Kollmann and J. Tomas, Vibrational Flow of Cohesive Powders, in: A. Levy and H. Kal-
man, Handbook of Conveying and Handling of Particulate Solids, pp. 45-56, Elsevier, Amster-
dam 2001
8 B. Reichmann and J. Tomas, Expression behaviour of fine particle suspensions and the con-
solidated cake strength, Powder Technology 121, pp. 182-189, 2001

You might also like