Professional Documents
Culture Documents
Apac060 PDF
Apac060 PDF
Apac060 PDF
This paper presents new formulations for estimating the effective floc density (Δρf) and the relative
viscosity (μr) of mixture of fine suspended sediments. Fitting parameters of the new models were
calibrated using data available in the literature. Good comparisons were obtained when the settling
velocities estimated based on the proposed models were compared against the settling velocity
estimated from existing models found in the literature. The average value of the primary particle
size, Dp for the data used in the analysis, inferred from the new model for Δρf was found to vary
from 0.05 μm to 100 μm with a mean value of 2.5 μm. The new model for μr is applicable to
mixtures of both non-cohesive and cohesive sediments.
Introduction
where ρs and Dp are density and size of primary particles, respectively, ρw is density of
water, and F is a constant, fractal dimension. Khelifa and Hill [2] argued that F is not a
constant but varies with Df, given by:
β
⎛ Df ⎞
F = 3⎜ ⎟ (2)
⎜ Dp ⎟
⎝ ⎠
where β = log(Fc/3) / log(Dfc/Dp), Fc and Dfc are characteristic values of F and Df,
respectively.
Research on the estimation of relative viscosity of fluid-sediment mixtures (μr)
includes theoretical [4, 5], empirical [6, 7] and semi-empirical [8-13] approximations. For
non-cohesive sediments, for a dilute suspension where the volumetric concentration (φ) is
less than 0.1, the formula for μr proposed by Einstein [4]:
μ r = 1 + kφ (3)
where k = 2.5 is the intrinsic viscosity, is probably the most well-known; for higher
sediment concentration, μr can be better estimated by Dougherty- Krieger [11]:
− kφmax
⎛ φ ⎞
μ r = ⎜⎜1 − ⎟⎟ (4)
⎝ φ max ⎠
where φmax is the maximum volumetric concentration of particles. For cohesive
sediments, Winterwerp [14] adopted Eq. (3) with φ being volumetric concentration of
flocs. Camenen [7] modified Eq. (4) assuming k = 2.5 and proposed an empirical
formula:
−φmax / 2
⎛ φ ⎞
μ r = ⎜⎜1 − ⎟⎟ (1 − φ p )−2 (5)
⎝ φ max ⎠
where φp and φ are the volumetric concentrations of primary particles and flocs,
respectively, and φmax ≈ 0.8.
Although using the mentioned models for Δρf and μr in estimation of settling
velocity of particles gives promising results, however, there are still some limitations to
these models. Firstly, the assumption of self-similar fractal flocs is not realistic [2, 3].
Secondly, there is no general model for μr, which can be equally applicable to both non-
cohesive and cohesive sediments. Therefore, there is a need to improve existing models
for Δρf and μr.
Given that Δρf = ƒ(Df, Dp, ρs, ρw), applying Buckingham-Pi theorem [15], the
dimensionless groups are π1 = Δρf /(ρs - ρw), π2 = Df /Dp, and π3 = ρs /ρw. Introducing
gravitation acceleration (g), kinematic viscosity of water (ν) and floc size (Df) and
combining with π3 gives:
2
1/ 3
⎡⎛ ρ ⎞ ⎤
D # = ⎢⎜⎜ s − 1⎟⎟ g / ν 2 ⎥ Df (6)
⎣⎢⎝ ρ w ⎠ ⎦⎥
where the inclusion of g and ν is to obtain the new dimensionless group only. This
parameter accounts for the joint contribution of π3 and Df on π1. By trial, it can be shown
that a formula for Δρf /(ρs - ρw) = ƒ(Df/Dp, D#) can be expressed as:
−1.0
⎧ ⎡ βf ⎫
⎪ ⎛ Df ⎞⎤ ⎪
Δρ f = (ρ s − ρ w )⎨1 + ⎢(D# )α f ⎜ − 1⎟⎥ ⎬ (7)
⎜ Dp ⎟⎥
⎪ ⎢⎣ ⎝ ⎠⎦ ⎪
⎩ ⎭
where αf and βf are fitting parameters (subscript .f refers to floc) which will be
determined based on curve-fitting with existing data.
Figure 1. Equation (7) with αf = 0.7, βf = 0.8 and (i) Dp = 0.05 μm (ii) Dp = 2.5 μm (iii) Dp = 100 μm
(Note: original graph is from Khelifa and Hill [2]).
Khelifa and Hill [2] compiled 26 published data sets of settling velocity as a function
of floc size. The compiled data covers both field and laboratory measurements from 1975
to 2003. Following Khelifa and Hill [2], the following values of ρs, ρw and ν were
assigned: ρs = 2,300 kg/m3, ρw = 1,000 kg/m3 and ν = 1×10-6 m2/s. Setting a range for Dp
from 0.05 μm [2] to 100 μm [12], the values of αf and βf that gave the best fit to the data
were found to be αf = 0.7 and βf = 0.8, as shown in Fig. 1. The figure shows that the
3
mean value of Dp is 2.5 μm, essentially equal to the Dp of kaolinite, which is 3 μm
according to van Rijn [12].
The sensitivity of the model to αf and βf was tested and the results (not shown here)
indicated that the model is more sensitive to βf rather than αf for the range of Δρf
commonly encountered in practice. Moreover, the settling velocity data published in
Winterwerp [14] was used to compare with calculated settling velocity of flocs having
the effective density determined by the new model (Appendix A); the calculated settling
velocities agree with the measured data well.
μr, i
⎧
1 + ⎨α
(φ / n ) ⎫ ⎧
1 + ⎨α
⎧
1 + ⎨α
(φ / n ) ⎫
⎬ ⎬ ⎬
⎩ max − (φ / n ) ⎭
φ ⎩ max − 2(φ / n ) ⎭
φ ⎩ max − n(φ / n ) ⎭
φ
The expression for the relative viscosity is then obtained as a product of the relative
viscosity for each part. Thus:
β β
n ⎡ ⎛
⎢ ⎜
μ r = Π 1 + ⎜α
φ/n ⎞ ⎤
⎟ ⎥
n ⎡
⎢
⎛
⎜
= Π 1 + ⎜α
[φ / φ max ] / n ⎞ ⎤
⎟ ⎥ (9)
i =1⎢
⎝ φ max − iφ / n ⎟⎠ ⎥ i =1⎢
⎝ 1 − i[φ / φ max ] / n ⎟⎠ ⎥
⎣ ⎦ n →∞ ⎣ ⎦ n→∞
For non-cohesive sediments, various experimental data from many researchers were
used to compare with the calculated results of μr using Eq. (9), shown in Fig. 2. The
values of φmax adopted are φmax = 0.60 for the data and references used by Camenen [7]
and φmax = 0.55 for Robinson [8] data. The comparison shows that Eq. (9) can reproduce
the measured experimental data well, with β = 1.0 and α in the range from 1.375 to 3.5,
with α = 2.0 providing a good fit to a majority of the data points.
4
Figure 2. Equation (9) for various combination of α and β; n = 10; (i) β = 1.0, α = 1.375, (ii) β = 1.0, α = 1.5,
(iii) β = 1.0, α = 2.5, (iv) β = 1.0, α = 3.5, (v) β = 1.0, α = 2.0, and (vi) β = 0.6, α = 0.127. [Figure adapted from
Camenen [7]. Note that the data from Vand [6] (Δ), Nir and Acrivos [16] (+), Rutgers [17] (◊) and Robinson [8]
(•) have been included in the figure]
α = Av (D* )α v (10)
⎡ (s − 1)g ⎤
1/ 3
D* = ⎢ ⎥ Df (11)
⎣ ν2 ⎦
where s = ρf/ρw, and Av and αv are empirical coefficients (subscript .v refers to
viscosity) depending on the properties of the primary particles and surrounding fluid. In
order to ensure that the value of α will never be larger than the limit of α = 0.127. By
trial-and-error, it was found that Av ≈ 0.01 and αv relates to Df (μm) by:
5
Equation (12) was derived by using Eq. (7) for Δρf with αf = 0.7, βf = 0.8, ρs = 2,300
kg/m3, ρw = 1,000 kg/m3, ν = 1×10-6 m2/s, Dp,max = 100 μm, and Av varied from 0.001 to
0.1, and by comparing calculated settling velocity of flocs using the new model for μr
against measured data.
The settling velocity data of mud by Thorn [18], published in Camenen [7], was used
to compare with calculated hindered settling velocity using the new model for μr
(Appendix B). The calculated settling velocities agree with the measured data well. The
value of α was found to be 0.0471 and the relative viscosity of the mixture is smaller than
that by Einstein’s [4] model; this implies that, for a wide range of volumetric
concentration of flocs, the intrinsic viscosity of mud does not always equal to 2.5 as
suggested by Einstein [4].
Conclusions
New models for the effective density of flocs and the relative viscosity of sediment-fluid
mixtures have been developed. Fitting parameters of the new models were calibrated
using data available in the literature. Good comparisons were obtained when the settling
velocities estimated based on the proposed models were compared against the settling
velocity estimated from existing models found in the literature. Compared to existing
models for Δρf, Eq. (7) has several advantages: (i) it does not consider the concept of
self-similarity of flocs but makes use of dimensional analysis, and (ii) it has two fitting
parameters making the model more flexible. Compared to existing models for μr, Eq. (9)
is more flexible because (i) it contains φmax and two empirical coefficients, and (ii) it is
applicable to mixtures of both non-cohesive and cohesive sediments.
References
1. S.N. Liss, T.G. Milligan, I.G. Droppo and G.G. Leppard, “Methods for Analyzing
Floc Properties.” Flocculation in Natural and Engineered Environmental Systems,
I.G. Droppo, G.G. Leppard, S.N. Liss and T.G. Milligan, eds., CRC Press, 1-21
(2005).
2. A. Khelifa and P.S. Hill, “Models for effective density and settling velocity of
flocs.” J. Hydraulic Research, 44 (3), 390-401 (2006).
3. C. Kranenburg, “The fractal structure of cohesive sediment aggregates.” Estuarine,
Coastal and Shelf Science, (39), 451-460 (1994).
4. A. Einstein, “Eine neue Bestimmung der Molekul dimensionen.” Annalen der
Physiek, Leipzig (4), 19 (1906).
5. V. Vand, “Viscosity of solutions and suspensions. I – Theory.” J Physical and
Colloid Chemistry, (52), 277-299 (1948a).
6
6. V. Vand, “Viscosity of solutions and suspensions. II – Experimental determination
of the viscosity – concentration function of spherical suspensions.” J. Physical and
Colloid Chemistry, (52), 300-314 (1948b).
7. B. Camenen, “Settling velocity of sediments at high concentrations.” Sediment and
Ecohydraulics. INTERCOH 2005, T. Kusuada, H. Yamanishi, J. Spearman and J.Z.
Gailani, eds., Elsevier, (9), 211-226 (2008).
8. J.V. Robinson, “The viscosity of suspensions of spheres.” J. Physical and Colloid
Chemistry, (53), 1042-1056 (1949).
9. M. Mooney, “The viscosity of a concentrated suspension of spherical particles.” J.
Physical and Colloid Chemistry, (6), 162-170 (1951).
10. T.J. Dougherty, “Some problems in the theory of colloids.” PhD Thesis, Case
Institute of Technology (1959); I.M. Krieger, Surfaces and Coatings Related to
Paper and Wood, R. Marchessault and C. Skaar, eds., Syracuse University Press
(1967).
11. I.M. Krieger, “Rheology of monodisperse lattices.” Advances in Colloid and
Interface Science, Elsevier, (3), 111-136 (1972).
12. L.C. van Rijn, Principles of sediment transport in rivers, estuaries and coastal seas,
Aqua Publications, The Netherlands (1993).
13. N.-S. Cheng, “Effect of concentration on settling velocity of sediment particles.” J.
Hydraulic Engineering, ASCE, 123 (8), 728-731 (1997).
14. J.C. Winterwerp, “On the dynamics of high-concentrated mud suspensions.” PhD
Thesis, Delft University of Technology (1999).
15. S.A. Hughes, Physical models and laboratory techniques in coastal engineering,
World Scientific (1993).
16. A. Nir and A. Acrivos, “Experiments on the effective viscosity of concentrated
suspensions of solid spheres.” International Journal of Multiphase Flow, Pergamon
Press, (1), 373-381 (1974).
17. I.R. Rutgers, “Relative viscosity of suspensions of rigid spheres in Newtonian
liquids.” Rheologica Acta, (2), 202-210 (1962).
18. M. Thorn, “Physical processes of siltation in tidal channels.” Hydraulic modelling in
maritime engineering: Proceedings of the conference organized by the Institution of
Civil Engineers, held in London on 13-14 October 1981, Thomas Telford Ltd.,
London, 65-73 (1982).
19. N. Chien and Z. Wan, Mechanics of sediment transport, ASCE Press (1999).
7
Appendix A
where nf is the (invariable) fractal dimension, and A and B are shape parameters.
Winterwerp [14] assumed the average value of the primary particle to be 4 μm and
adjusted nf to range from 1.7 to 2.3, with an average value of 2, in order to fit the
calculated settling velocity from Eq. (A1) to the measured data (Fig. A1).
Figure A1. Settling velocity of flocs with αf = 0.7, βf = 0.8, ρs = 2,300 kg/m3, ρw = 1,000 kg/m3, ν = 1×10-6 m2/s
and (i) Dp = 1 μm (ii) Dp = 2.5 μm (iii) Dp = 7 μm (iv) Dp = 50 μm (Note: original graph is from Winterwerp
[14]).
For the present study, using the proposed model for Δρf [Eq. (7)] and the modified
Stokes’ law:
ws =
( )
4 ρ f − ρ w gD f
(A2)
3ρ w C D
8
for estimating the settling velocity. Following Winterwerp [14], the drag coefficient (CD)
is determined by Eq. (A3), where the particle Reynolds number (Rep) is defined by Eq.
(A4).
CD =
24
Re p
(
1 + 0.15 Re 0p.687 ) (A3)
ws D f
Re p = (A4)
ν
The calculated settling velocity of flocs for selected values of Dp ranging from 1 μm to
50 μm is shown in Fig. A1. The figure shows that the calculated settling velocities using
the proposed model agree well with the measured data and Winterwerp’s [14] results.
The average value of Dp obtained is about 7 μm; this value is close to the average value
of Dp (Dp = 4 μm) assumed by Winterwerp [14]. In addition, for Chesapeak Bay data
(denoted by the symbol “◊”), it was found that the value of Dp varies from 4.5 to 8 μm
with an average value of Dp of about 6 μm.
9
Appendix B
Camenen [7] used Thorn’s [18] settling velocity data for high concentration mud
obtained from the Severn estuary. The author found that Df ≈ 295 μm, ρf ≈ 1,053 kg/m3
provided a good fit to Thorn’s [18] data. The original graph in Fig. A2 is from Camenen
[7] showing results using Camenen’s [7] models, denoted as ‘R & Z modified’ and
‘mixture theory’, together with other models considered by Camenen [7].
Figure A2. The hindered settling velocity with Thorn’s [18] data and (i) Av = 0.001, α = 0.0191, (ii) Av = 0.01,
α = 0.0471, and (iii) Av = 0.1, α = 0.116 (Note: Camenen [7] assumed cgel ≈ 70 kg/m3; the present study
assumed cgel = 80 kg/m3; original graph is from Camenen [7]).
For the present study, following Chien and Wan [19], the hindered settling velocity
of flocs (Ws) was estimated by:
W s = (1 − φ )ws ,m (A5)
where φ is volumetric concentration of flocs; ws,m is the relative velocity between the
falling flocs and surrounding fluid (subscript .m refers to mixture), given by:
( )
4 ρ f − ρ m gD f
ws ,m = (A6)
3ρ m C D , m
10
The drag coefficient CD,m was determined by Eq. (A7), where Rep,m was defined by Eq.
(A8).
C D ,m =
24
Re p,m
(
1 + 0.15 Re 0p.,687
m ) (A7)
ws ,m D f
Re p ,m = (A8)
νm
The proposed model for μr will give unrealistic values when φ/φgel approaches 1.0
(μr → ∞, νm → ∞ and Ws → 0); therefore, the value of the νm at φ/φgel > 0.95 was
approximated by that at φ/φgel = 0.95. In addition, the calculation was carried out up to
mass concentration c/cgel = 0.99 only because the proposed model does not take into
account consolidation process when c ≥ cgel.
Figure A3. Variation of μr with φ using Thorn [18] data with (i) Einstein’s model and (ii) proposed model.
The calculated settling velocity using the new model is shown in Fig. A2. It can be
seen from the figure that the use of the proposed model reproduces the measured settling
velocity very well except at the range of mass concentration below about 3 kg/m3 where
the measured values may be affected by flocculation. In addition, viscosities of the
mixture calculated by Einstein’s model [Eq. (3)] and by the new model are presented in
Fig. A3. The figure shows that the viscosity of the mixture predicted by the new model is
smaller than that by Einstein’s model. This implies that, for a wide range of volumetric
11
t
concentration of flocs, the intrinsic viscosity of mud does not always equal to 2.5 as
suggested by Einstein [4].
12