131 views

Original Title: An Explicit Equation for the Terminal Velocity of Solid Spheres Falling in Pseudo Plastic Liquids-11

Uploaded by bill_c86110

- dieloil.pdf
- Industrial Flow Measurement_Basics and Practice
- Dimensionless Parameters
- Industrial Flow Measurement Practice
- 113851176 Fluid Mechanics and Machinery Notes 2
- ANALYSIS OF BLOOD FLOW THROUGH STENOSED VESSEL UNDER EFFECT OF MAGNETIC FIELD
- Rising Velocity for Single Bubbles in Pure Liquids
- CONTROL VALVE SIZING.pdf
- Ventilation Network Analysis
- Progresivo Pag. 1 - 45 Ingles
- Advance Fluid Mechanics Lectures 7-8
- r05220302 Mechanics of Fluids
- Fluid Assignment
- Viso Us Heating Solver
- Hmt Part 7 - Conv Governing Eqns and Solutions
- (Elasticity)
- A Model of the Development of a Periphyton Community
- Transition and Turbulence
- Chapter 2 (Lecture 2)
- Viscosity of Periodic Suspensions - P. Gondret - 1996

You are on page 1of 11

www.elsevier.com/locate/ces

falling in pseudoplastic liquids

Vassilios C. Kelessidis∗

Mineral Resources Engineering Department, Technical University of Crete, Polytechnic City, 73100 Chania, Greece

Received 18 February 2004; received in revised form 14 June 2004; accepted 7 July 2004

Abstract

An explicit equation is proposed which predicts directly the terminal velocity of solid spheres falling through stagnant pseudoplastic

liquids from the knowledge of the physical properties of the spheres and of the surrounding liquid. The equation is a generalization

of the equation proposed for Newtonian liquids. By properly deﬁning the dimensionless diameter, d∗ , a function of the Archimedes

number, Ar, and the dimensionless velocity, U∗ , a function of the generalized Reynolds number, Re, to account for the non-Newtonian

characteristics of the liquid, the ﬁnal equation relating these two variables has similar form to the Newtonian equation. The predictions

are very good when they are compared to 55 pairs of Re—CD for non-Newtonian data and 37 pairs for Newtonian data published

previously. The root mean square error on the dimensionless velocity is 0.081 and much better than the only other equation previously

proposed.

䉷 2004 Elsevier Ltd. All rights reserved.

Keywords: Non-Newtonian ﬂuids; Particle; Sedimentation; Suspension; Slurries; Explicit terminal velocity

1. Introduction so that

4 gd 0.5

Knowledge of the terminal settling velocity of solids in V = . (2)

liquids is required in many industrial applications. Typi- 3 CD

cal examples include drilling for oil and gas, geothermal Numerous attempts have been made to establish theoreti-

drilling, hydraulic transport systems, thickeners, mineral cal relationships of the terminal settling velocity of solid

processing, solid–liquid mixing, ﬂuidization equipment. spheres but the theoretical expressions are normally valid for

The type of movement of single solid sphere in Re, based on the particle diameter, less than 200. For higher

Newtonian and non-Newtonian liquids is well-known; after Re, resort must be made to experimental and empirical

a short acceleration time, it will fall at its terminal settling relationships.

velocity V . For an unbounded liquid, V can be calculated There are over 50 correlations published relating the

from the knowledge of the liquid and solid physical prop- Reynolds number to the drag coefﬁcient for the case of

erties and from the drag coefﬁcient, deﬁned by Newtonian ﬂuids. In these correlations the terminal ve-

4 gd locity is implicitly derived, hence, resort must be made to

CD = (1) trial and error procedure for deriving the terminal velocity.

3 V 2

There are not as many explicit relationships to predict V ,

with few equations for Newtonian liquids and even fewer

for non-Newtonian liquids.

∗ Tel.: +30-2821037621; fax: +30-2821037874. Turton and Clark (1987) proposed an explicit rela-

E-mail address: kelesidi@mred.tuc.gr (V.C. Kelessidis). tionship, where they deﬁned a dimensionless particle

0009-2509/$ - see front matter 䉷 2004 Elsevier Ltd. All rights reserved.

doi:10.1016/j.ces.2004.07.008

4438 V.C. Kelessidis / Chemical Engineering Science 59 (2004) 4437 – 4447

1/3

3 g 1/3 18 0.824

d∗ = CD Re 2

=d (3) Re=Ar 1/3

4 2 1

Ar 2/3

and a dimensionless terminal velocity, U∗ , 1.214

3 0.428 0.412

1/3 + . (12)

4 Re 1/3 2 4 Ar 1/3

U∗ = =V . (4)

3 CD g (s − )

This analysis shows that knowing the physical properties

They then combined the low Reynolds number range (Stokes and particle diameter one can estimate Ar (Eq. (9)) and then

law), where get Re from Eq. (12) from which V is readily evaluated.

24 Explicit relationships for the terminal velocity for Newto-

CD = (5) nian ﬂuids have also been reported by Hartman et al. (1989)

Re

and Nguyen et al. (1997).

with the high Reynolds number range, where

Hartman et al. (1989) modiﬁed the implicit equation, pro-

CD = K1 (6) posed by Zigrang and Sylvester (1981), and using their own

data they proposed,

in the form

K2 0.5∗K2 K12 log Re = P (A) + log R(A), (13)

18 3 K1

U∗ = 1 + (7) P (A)=[(0.0017795A − 0.0573)A + 1.0315]A

d∗2 4 d∗

−1.26222, (13a)

so that the velocity could be directly estimated. (It should be

noted that this equation is listed with a mistake in the paper R(A) = 0.99947 + 0.01853 sin(1.848A − 3.14), (13b)

of Turton and Clark (1987) as well as in the paper of Heider

and Levenspiel (1989), with the square root shown on the A = log Ar (13c)

parameter d∗ in the second term, rather than the correct form

as an accurate solution—approximation to the equation of

shown above).

Zigrang and Sylvester (1981) in the sense of least-squares

Using the 408 data points (the ones used by Turton and

ﬁt. They compared this equation together with the equa-

Levenspiel, 1986) and minimizing the RMS error, which is

tion proposed by Turton and Clark (1987), Eq. (12), with

deﬁned as in the following equation:

their own experimental data, which covered a range of

Q 5 < Re < 78 and 141 < Ar < 5720. They found relative de-

RMS_U∗ = , viations in velocity of (+1.7%)–(−1.3%) for their proposed

N

equation, while for the equation of Turton and Clark (1987)

Q= (log10 U∗ exp − log10 U∗pred )2 (8) they found relative deviations of (+3.14%)–(−5.2%).

Nguyen et al. (1997) proposed a three parameter expres-

they found K1 = 0.428 and K2 = 0.824 with the value of K1 sion relating the ratio of the Stokes velocity to the particle

comparable to the value of the drag coefﬁcient at high val- velocity with the Archimedes number, as in,

ues of Reynolds number (CD = 0.44). With these parameter

values, they found that Q = 0.244 and RMS_U∗ = 0.024. Vs Ar

=1+ [1 + 0.079Ar 0.749 ]−0.755 , (14)

It should be noted that Eq. (3) can be recast as V 96

3

d g g 2 d 3 where Vs is the terminal velocity for the Stokes regime

d∗ =

3

= = ∗ Ga = Ar (9)

2 2 (Re < 0.1) and given by

with the dimensionless number Ga normally referred to as d 2 g

the Galileo number and Ar is the Archimedes number. Vs = . (15)

18

Similarly, Eq. (4) can be rearranged as

2 This equation can be recast in the form Re–Ar as

V 3 V 3 2 d 3

U∗3 =V 3 = =

g g 2 Ar Ar 1

Re = Ar

. (16)

Re3 18 1 + [1 + 0.079Ar 0.749 ]−0.755

= (10) 96

Ar

The authors showed graphically that their data fall on the

from which ﬁnally,

predicted V -d curve, for the range of 1.68 < Re < 65, using

Re ballotini particles (diameters between 0.1 and 0.8 mm and

U∗ = . (11)

Ar 1/3 density of 2.5 g/cm3 ) falling in water.

V.C. Kelessidis / Chemical Engineering Science 59 (2004) 4437 – 4447 4439

Richardson, 1999; Kelessidis, 2004) have established that with

expressions relating the drag coefﬁcient to Reynolds number

proposed for Newtonian liquids could be extended to non- 0.51

a = 0.1 × exp − 0.73n , (21)

Newtonian liquids, provided that the Reynolds number is n

estimated based on the apparent viscosity as

0.954

b= − 0.16. (22)

V 2−n d n n

Re = Regen = . (17)

K Their equation covered the range 10 < Ar < 106 and

Others (Acharya et al., 1976; Kawase and Ulbrecht, 1981a,b; 1 < Re < 104 . (It should be noted that in the original publi-

Ceylan et al., 1999) have provided theoretical solutions valid cation of Chhabra and Peri (1991), the expression for a is

for creeping ﬂow of pseudoplastic ﬂuids around a falling erroneously printed without the multiplication factor of 0.1,

sphere, resulting in corrections to the Newtonian case of the and using the equation with the wrong coefﬁcients gives

form, CD = Re 24

Xn with Xn = f (n), while also providing always a considerably underestimated velocity. In a more

equations of the form CD = f (Re, n) for higher Reynolds recent publication (Chhabra and Richardson, 1999) the

numbers, derived from regression analysis of their own as value of a is correctly reported, as in Eq. (21)). For given

well as other investigators’ data, a procedure followed also values of d, , , K and n, Ar is calculated from Eq. (19)

by Matijašić and Glasnović (2001) using their own data. and Re is then predicted using Eqs. (20)–(22). Once Re is

However, all the proposed equations by the aforementioned calculated, the terminal settling velocity can be predicted

authors are implicit in velocity. Reynolds and Jones (1989) from Eq. (17).

proposed to plot the logarithm of the ratio of the drag force Briens (1991), using dimensionless diameter and velocity,

to the square of the sphere diameter against the logarithm of as in Newtonian ﬂuids, proposes a graphical procedure to

the ratio of the settling velocity to the particle diameter in determine directly the velocity from the ﬂuid and solid prop-

order to determine the form of the function f (n) referenced erties. However, the graphical procedure makes it impracti-

above and validated the procedure with a series of their own cal for computational purposes and the results are given in

data for Re < 0.1. terms of the ﬂow behavior index n.

Koziol and Glowacki (1988) extended the approach of Graham and Jones (1994) solved numerically the problem

Schiller and Nauman to non-Newtonian power law ﬂuids by of spheres moving through a quiescent power law ﬂuid and

forming the parameter provided relationships of the form CD = f (Re, n) valid for

0.2(2)n Re 100(2)n with the factor (2)n arising because

2−n

4 d 2+n g 2−n 2 they had deﬁned the Reynolds number based on the radius

2−n

A = CD Re2 = (18) of the sphere. When the proposed equation was compared

3 K2

to experimental data, different curves were produced for

which is independent of velocity. They presented a general different values of the power law coefﬁcient. They could not

plot of Re vs A based on their own data as well as data from ﬁnd closed form solutions and the authors proposed to follow

other investigators. The data and plot covered the range of the graphical method used by Koziol and Glowacki (1988)

0.001 < Re < 10. From the results it is evident that such a described above, in order to determine the settling velocity

presentation can be of merit, however different curves are given the particle size and density and the ﬂuid properties.

obtained for different n values, indicating that a unique equa- Darby (1996) proposed a methodology which determines

tion may not be possible to be obtained using the proposed not only the CD –Re relationship for spheres falling in power

parameters. law ﬂuids but also the velocity directly, although there is still

Chhabra and Peri (1991) extended the approach of Koziol a need for iteration in the ﬁnal proposed equation. He ex-

and Glowacki (1988) to higher Reynolds numbers, seeking tended the Dallavall equation proposed for Newtonian ﬂuids

a relationship of the form using correction factors, derived from various curves pro-

posed by Chhabra (1995), to account for the shear thinning

Re = f (Ar, n) behavior which were ﬁnally given as functions of the power

law exponent n.

with Re given by Eq. (17). They deﬁned the non-Newtonian

Ar as

4gd (2+n)/(2−n) 3. Comparison of explicit correlations for Newtonian

Ar = CD Re 2/(2−n)

= . (19) liquids

3K 2/(2−n) 2/(n−2)

They gathered 400 experimental data points pertaining to We attempt here to compare the predictions of the ter-

Re 1 and derived an equation, by minimizing the RMS minal velocity from the proposed explicit correlations for

4440 V.C. Kelessidis / Chemical Engineering Science 59 (2004) 4437 – 4447

Newtonian ﬂuids using Eqs. (12), (13), (16) and (20) and The results are shown in Fig. 1. It is evident that Eqs. (12),

quantify any differences. We include Eq. (20), proposed for (13) and (16) give very similar results, for the range of

non-Newtonian ﬂuids, since it is claimed to be valid for ﬂuid Ar mentioned, while Eq. (20) progressively diverges from

behavior index values of 0.38 n 1 (Chhabra and Peri, the results of the others for Ar > 1.5 × 104 . We compute

1991). The range of validity of the proposed equations is the RMS error on Reynolds and the mean relative absolute

listed in Table 1. We see that the Archimedes number covers error in velocity, as stated above, where as reference we get

a very extended range. the Reynolds number and velocity predicted from Eq. (12).

We will be comparing the predictions from the above These error values are shown in Table 2, for the range of

mentioned equations for 10 Ar 0.35 × 105 which covers 10 Ar 0.35 × 105 .

situations with particle diameters between 1–6 mm, with a The values of the error indicators show that all three Eqs.

density of 2500 kg/m3 , ﬂuid density of 1100 kg/m3 and (12), (13) and (16), give very similar results, evident also in

apparent liquid viscosities in the range of 10–30 cp. Fig. 1, while the errors from Eq. (20) are much larger, prob-

For a combination of values of , s , d, or a we com- ably because the parameter values were derived primarily

pute Ar and then Re from one of the equations listed above. from non-Newtonian data.

We compare the relative errors using two indicators, the Since Ar can take values outside the above range, we

Root Mean Square Error on Reynolds, deﬁned in a similar have made similar computations for an extended range of

manner as in Eq. (8), 2 Ar 1010 . For this extended range, we get the error val-

ues shown in Table 3. We see that for this extended range,

QR

RMS_Re = ,

N Table 2

QR = (log10 ReT − log10 Rei )2 (23) Errors between predictions from different correlations and from correlation

of Turton and Clark (1987) (10 Ar 0.35 × 105 )

and the relative absolute error in velocity, deﬁned as Hartman et al. Nguyen et al. Chhabra and Peri

N (1989), Eq. (13) (1997), Eq. (16) (1991), Eq. (20)

1 |VT − Vi | RMS_Re 0.20 0.18 0.36

EV = ∗ 100. (24) EV (%)

N VT 3.8 3.3 12.7

1

Table 1

Range of parameters for validity of proposed correlations Table 3

Errors between different correlations, for an extended range of Ar, com-

Source Ar Re n puted as per Table 2 (10 Ar 1010 )

Turton and Clark (1987) 1–108 0.1–2 × 104 1.0 Hartman et al. Nguyen et al. Chhabra and Peri

Hartman et al. (1989) 1–107 0.1–0.55 × 104 1.0 (1989), Eq. (13) (1997), Eq. (16) (1991), Eq. (20)

Nguyen et al. (1997) 1–8 × 104 0.1–450 1.0 RMS_Re 0.03 0.10 0.56

Chhabra and Peri (1991) 10–106 1–104 0.38–1.0 EV (%) 5.3 16.4 355

350

300

250

Reynolds number

200

150

100

50

0

0.0E+00 5.0E+03 1.0E+04 1.5E+04 2.0E+04 2.5E+04 3.0E+04 3.5E+04

Archimedes number

Turton & Clark, eqn.12 Hartman et al., eqn. 13 Nguyen et al., eqn. 16 Chhabra & Peri, eqn. 20

Fig. 1. Comparison of predictions of Reynolds number from Archimedes number from various correlations proposed for Newtonian ﬂuids.

V.C. Kelessidis / Chemical Engineering Science 59 (2004) 4437 – 4447 4441

Table 4 Table 5

Non-Newtonian data from other investigators Newtonian data from other investigators

Re CD Re CD Re CD Re CD

Kelessidis (2004), 15 pairs Kelessidis and Mpandelis Kelessidis (2004), 10 pairs Kelessidis and Mpandelis (2004),

(2004), 14 pairs 9 pairs

15.622 3.574 1166.04 0.46

0.112 174.572 9.04 3.79 55.243 1.849 657.44 0.57

0.568 35.534 4.09 7.01 67.611 1.463 215.09 0.67

0.723 26.059 0.78 20.35 120.236 1.174 2792.80 0.32

1.629 14.463 39.80 1.66 154.441 1.027 1.21 24.42

2.274 11.029 6.12 5.86 190.059 0.694 0.48 62.84

2.877 12.769 2.63 12.04 566.714 0.605 0.08 272.70

7.310 6.936 0.48 41.42 671.864 0.511 4.42 8.39

9.623 4.558 23.44 2.42 1049.610 0.531 4.65 7.97

11.969 3.481 23.34 2.40 1269.610 0.523

22.115 2.849 4.09 9.26

22.654 2.516 1.72 20.11

27.261 2.215 0.30 77.96 Di Felice (1999), 12 pairs Hartman et al. (1989), 6 pairs

29.595 2.130 15.71 3.33 0.016 1541.572 5.189 7.001

50.130 1.677 15.46 3.47 0.128 190.312 6.088 6.290

64.223 1.471 0.486 57.659 19.234 2.763

1.829 17.756 30.538 2.095

2.218 15.863 50.295 1.572

Miura et al. (2001), Pinelli and Magelli (2001), 7.523 6.010 78.249 1.246

18 pairs 8 pairs 29.574 2.558

0.445 59.698 1.246 16.222 53.740 1.800

2.613 12.639 1.889 9.885 98.780 1.161

6.561 6.253 4.779 5.447 237.122 0.675

7.408 5.405 5.640 4.197 365.800 1.004

8.614 4.007 9.902 3.366 928.550 0.500

24.024 2.527 14.075 2.305

32.464 1.542 19.264 1.922

43.644 1.402 28.012 1.287

53.654 1.497 Table 6

68.644 0.987 Range of Re and power law parameters for utilized data in this work

94.419 0.828

153.220 0.598 dynes∗ s n

Source Re K 2 n

174.157 0.633 cm

257.458 0.511 non-Newtonian

269.090 0.501 Kelessidis and Mpandelis 0.3–40 0.85–1.15 0.75–0.91

406.838 0.511 (2004)

407.534 0.500 Kelessidis (2004) 0.1–64 0.16– 2.7 0.75–0.92

770.472 0.378 Miura et al. (2001) 0.4–770 0.17–5.90 0.56–0.75

Pinelli and Magelli (2001) 1.0–30 0.47–0.52 0.73

Newtonian

Kelessidis and Mpandelis 0.1–2800 0.01–1.35 1.0

(2004)

the predictions from Eqs. (12) and (13) are still very similar, Kelessidis (2004) 16.0–1270 0.01–0.06 1.0

the errors from Eq. (16) are larger, while the errors from Di Felice (1999) 0.01–929 0.23–1.74 1.0

Eq. (20) are very large, primarily because the data used is Hartman et al. (1989) 5.0–78 13.2 1.0

outside the validity range of Eq. (20).

4. Scope of work (2004) and Kelessidis and Mpandelis (2004), which are

listed in Table 5. The range of parameters covered is listed

The purpose of this work was to evaluate the different in Table 6.

approaches for the direct determination of velocity of spher- We have shown above that Eq. (12) proposed for Newto-

ical particles falling in non-Newtonian liquids and to de- nian data provides similar predictions to the other equations

termine which equation better describes the experimental (Eqs. (13) and (16)). Hence, we will be extending (generaliz-

data, and if possible, propose a new equation. The data uti- ing) this equation to cover non-Newtonian ﬂuids, we would

lized include non-Newtonian data of Miura et al. (2001), then be comparing the predictions with measured data and

Pinelli and Magelli (2001), Kelessidis (2004), Kelessidis ﬁnally we will be comparing the predictions from this equa-

and Mpandelis (2004), listed in Table 4, and Newtonian tion to predictions derived with the explicit equation pro-

data of Di Felice (1999), Hartman et al. (1989), Kelessidis posed for non-Newtonian data (Eq. (20)).

4442 V.C. Kelessidis / Chemical Engineering Science 59 (2004) 4437 – 4447

5. Theory Chhabra and Peri (1991) nor from the approach of Koziol

and Glowacki (1988) nor from Briens (1991), pointing to

In accordance to the work of Turton and Clark (1987), we the fact that Eq. (25) is the proper generalization of Eq. (3)

deﬁne, using the generalized Reynolds number (Eq. (17)), a used for Newtonian liquids.

dimensionless diameter, d∗ = (Ar)(2−n)/(2+n) , as We seek therefore an expression similar to the Newtonian

(2−n)/(2+n) liquids, Eq. (7), but care should be exercised on the way the

(2−n)/(2+n) 3 generalization is made for the pseudoplastic ﬂuids. Accord-

d∗ =(Ar) = CD Re 2/(2−n)

4 ingly, we perform the generalization as per the following

g (2−n)/(2+n)
2/(2+n) steps.

=d . (25) We combine the low Reynolds range, Eq. (5),

K

24

For n = 1 this equation gives the value of d∗ , given above CD =

(Eq. (3)) and deﬁned by Turton and Clark (1987). Re

Similarly we deﬁne a dimensionless velocity, U∗ , as with the high Reynolds number range, Eq. (6),

1/(2+n) 1/(2+n)

Re n+1 CD = K1

U∗ = =V (26)

((3/4)CD )n g n K n since, as stated above, the Newtonian relationships could be

which for n = 1 gives the value of U∗ as in Newtonian ﬂuids extended to non-Newtonian ﬂuids provided that the gener-

(Eq. (4)). alized Reynolds number is used.

Eq. (26) can be recast to give From Eq. (26) and Eq. (5), we can get,

2+n Re1+n

V n 2+n

U∗1 = ,

U∗2+n = . (27) 18n

K g

(2+n)/(1+n) Re

Since U∗1 = (31)

18n/(1+n)

g 2/(2−n) d (2+n)/(2−n)

Ar = , and from Eq. (25),

K 2/(2−n)

g K 2/(2−n) (2+n)/(2−n)

d∗ =

3

CD Re 2/(2−n)

,

= (Ar) 2/(2−n) (2+n)/(2−n) . (28)

d 4

(2+n)/n

Combination of (27) and (28) gives, d∗ = 18(2−n)/n Re. (32)

Re1/(2−n) Combining (31) and (32) and eliminating Re, we get,

U∗ = . (29)

Ar n/(2+n) (2+n)/n

(2+n)/(1+n) d∗

It should be noted that Eq. (25) should be the proper way U∗1 = ,

18(2+n)/n(1+n)

of deﬁning the dimensionless diameter. If we deﬁne d∗ 1/n

(1+n)

as other investigators did, as for e.g. for the derivation of d∗

Eq. (20) (Chhabra and Peri, 1991) and denote it as dc∗ , it U∗1 = . (33)

18

can be shown that

3 gd (2+n)/(2−n) / Now, for the second branch and from Eqs. (6) and (25), we

dc∗ = CD Re2/(2−n) = . (30) get

4 K 2/(2−n) 2/(n−2)

(2−n)/(2+n)

If the published data mentioned above is plotted as U∗ − dc∗ , 3

d∗ = K1 Re2/(2−n) ,

as in Fig. 2, it is clear that there is not a unique relationship 4

between the two variables with dc∗ so deﬁned. There is vari- (2−n)/2

(2+n)/2 3

ation among the data from different investigators, indicating d∗ = K1 (Re). (34)

that the relationship is not independent of the value of n, a 4

fact also evident from Chhabra and Peri (1991), where the For the dimensionless velocity, from Eqs. (6) and (26) we

values of the correlation parameters a and b are functions get,

of n (Eqs. (21) and (22)). However, when we deﬁne the di- 1/(2+n)

mensionless diameter as in Eq. (25) above and plot the same Re

U∗2 = ,

data, we get Fig. 3. Here we see that all data, Newtonian [(3/4)K1 ]n

and non-Newtonian, fall along the same curve thus getting a

(2+n) Re

unique relationship between the two dimensionless param- U∗2 = . (35)

eters, something not possible neither from the approach of [(3/4)K1 ]n

V.C. Kelessidis / Chemical Engineering Science 59 (2004) 4437 – 4447 4443

Fig. 2. Relationship between experimental values of U∗ and dimensionless diameter, dc∗ , deﬁned by Chhabbra and Peri (1991). Open symbols=Newtonian

data, closed symbols = non-Newtonian data.

Fig. 3. Relationship between U∗ and dimensionless diameter, d∗ . Open symbols = Newtonian data, closed symbols = non-Newtonian data.

Hence, combining (34) and (35) and eliminating Re, we get, This can be rearranged in order to ﬁnally get,

4 d∗ 1/2 K2 /n K2 /2 1/K2

U∗2 = . (36) 18 3 K1

3 K1 U∗ = 1/ + . (38)

d∗1+n 4 d∗

Hence, in a similar fashion as in Turton and Clark (1987),

we combine the two dimensionless velocities, as in Eq. (38) is the generalization of Eq. (7) to cover the case

of pseudoplastic ﬂuids since we claim that the general-

1 1 1

= + . (37) ized equation should hold both for Newtonian and non-

U∗K2 K2

U∗1 K2

U∗2 Newtonian ﬂuids.

4444 V.C. Kelessidis / Chemical Engineering Science 59 (2004) 4437 – 4447

Fig. 4. Comparison of predicted with measured dimensionless velocities for non-Newtonian data.

The parameters K1 , K2 derived for Newtonian ﬂuids by at U∗,meas ∼ 23 underpredicted severely with an error of

Turton and Clark (1987) should be the same as the parame- (U∗,meas − U∗,pred )/U∗,meas of around −22%. This point

ters for the generalized equation. We therefore compare the corresponds to Re = 2800, a terminal velocity of 100 cm/s.

predictions from Eq. (38) using the Newtonian K1 , K2 val- For the next lower value of U∗ ∼ 15, we have a Reynolds

ues determined by Turton and Clark (1987) with previously number of 1200 and a velocity of 37 cm/s. Hence, if we

gathered data. We determine its validity by plotting mea- restrict the predictions to the range of U∗ between 0.02 and

sured and predicted values of the dimensionless velocity and 15, the comparison is very good and the root mean square

by calculating the RMS error in U∗ , as in Turton and Clark error for these data points is 0.041.

(1987) (Eq. (8)). Combining all data, except for the one of U∗ = 23, we

Using the Newtonian values of K1 =0.428 and K2 =0.824, see that for both Newtonian and non-Newtonian data the

Eq. (38) ﬁnally becomes predictions are satisfactory, particularly in view of the ranges

1.214 of the parameters K, n, which are listed in Table 6, with an

18 0.824/n 0.321 0.412 RMS error on U∗ of 0.081, indicating the good accuracy of

U∗ = 1 + . the predictions.

d∗1+n d∗

In order to compare the predictions from (39) with the

(39)

predictions from (20), we put them in comparable form. We

The predicted values of U∗ are compared with the values recast the new proposed equation, Eq. (39), in a similar form

of U∗ from experimental data, as computed using the data to Eq. (20),

from Table 4 for non-Newtonian data and from Table 5 for Re = (U∗ )2−n (Ar)n(2−n)/(2+n) . (40)

Newtonian data. This comparison is depicted in Fig. 4 for the

non-Newtonian data and in Fig. 5 for the Newtonian data. We compare then the Reynolds numbers predicted from Eq.

From Fig. 4 it is evident that the data is predicted well (40) with the Reynolds number predicted from Eq. (20) for

with the new proposed explicit equation for the range of pa- the non-Newtonian data of Table 4 and for the Newtonian

rameters which give dimensionless velocity values from 0.1 data of Table 5. The comparison is shown in Figs. 6 and 7,

to 16 corresponding to terminal velocities between 1.2 and respectively.

60.0 cm/s. The RMS error on U∗ (Eq. (8)) is 0.10, indicat- From Fig. 6 we see that for the non-Newtonian data, the

ing the good accuracy of the predictions. This value should predictions of Eq. (40) compare very well with the mea-

be compared to the value of 0.024 derived for the 408 New- sured Reynolds numbers. However, Eq. (20) does a poor job

tonian data points from Turton and Clark (1987). and over predicts the generalized Reynolds number for a

From Fig. 5 we see that we have good predictions for given Archimedes number for the whole range of Reynolds

the range of U∗ between 0.02 and 15, with the one point numbers studied (0.1–1000). On the other hand, for the

V.C. Kelessidis / Chemical Engineering Science 59 (2004) 4437 – 4447 4445

Fig. 5. Comparison of predicted with measured dimensionless velocities for Newtonian data.

Newtonian data, shown in Fig. 7, both equations predict very Eq. (20), a considerable improvement with respect to the

similar values to the measured values of Reynolds number, non-Newtonian data.

except for the higher Reynolds number range, where Eq. It has therefore been shown that Eq. (39), which is the

(20) tends to over predict the dimensionless velocity, while generalization of the Newtonian equation (12), describes ex-

Eq. (40) does as good a job for the whole range of Reynolds tremely well reported Newtonian data, as it should. It has

numbers tested. The RMS error on Reynolds number (com- also been shown that it describes very well reported non-

puted from Eq. (23)) for the non-Newtonian data is 0.12 for Newtonian data which cover a rather extensive range of

Eq. (40) and 0.54 for Eq. (20). The corresponding errors power law parameters and of generalized Reynolds numbers.

for the Newtonian data are 0.04 for Eq. (40) and 0.18 for Furthermore, this equation outperforms the only other ex-

4446 V.C. Kelessidis / Chemical Engineering Science 59 (2004) 4437 – 4447

plicit equation proposed for non-Newtonian data (Eq. (20)) while results from both equations are very good for Newto-

as the graphical comparison (Fig. 6) and the comparison of nian data.

the RMS errors of Reynolds number have shown.

Notation

6. Conclusions

a parameter, given in Eq. (21), dimensionless

The various explicit equations for the terminal velocity of b parameter, given in Eq. (22),

dimensionless

g 2/(2−n) d (2+n)/(2−n)

spheres falling through Newtonian liquids have been com- Ar Archimedes number, K 2/(2−n)

,

pared and it has been shown that all predict similar values of dimensionless

Reynolds numbers for typical ranges of Archimedes num- CD drag coefﬁcient, dimensionless

bers. d particle diameter, L

The Newtonian variables, dimensionless velocity and d∗ dimensionless diameter, Eq. (25), dimensionless

dimensionless diameter, proposed by Turton and Clark dc∗ dimensionless diameter from Chhabra and Peri

(1987), have been generalized to cover non-Newtonian (Eq. (30)), dimensionless

pseudoplastic ﬂuids. It has been shown that if reported ex- EV percent relative absolute error in velocity,

perimental data are recast in the proposed form, with the Eq. (24), %

dimensionless variables given by Eqs. (25) and (26), all g acceleration of gravity, L/T2

experimental data fall along the same curve, thus eliminat- 2/(2−n) (2+n)/(2−n)

Ga Galileo number, K 2/(2−n)

d

, dimensionless

ing the need for having multiple curves, functions of the n

power law index, as was the case from previously suggested K consistency index, M/LT

correlations. n power law index, dimensionless

The predictions from the new generalized equation (Eq. N number of samples used in Eq. (23) and Eq. (24),

(39)) are compared with previously reported experimental dimensionless

data from other investigators which cover non-Newtonian Re Reynolds number based on particle diameter, V d ,

and Newtonian liquids. The comparison is very satisfactory dimensionless

2−n n

and the root mean square error in the dimensionless velocity Regen generalized Reynolds number, V K d , dimen-

for all data points is 0.081, a similar order of accuracy as sionless

for the Newtonian data. ReT Reynolds number predicted from correlation of

Comparison of the predictions derived from the new equa- Turton and Clark (1987), dimensionless

tion with predictions derived from another correlation (Eq. Rei Reynolds number predicted from other correla-

(20)) shows that the new generalized equation outperforms tions, dimensionless

the predictions from Eq. (20) for the non-Newtonian data, V solid terminal velocity, L/T

V.C. Kelessidis / Chemical Engineering Science 59 (2004) 4437 – 4447 4447

Vs solid terminal velocity in the Stokes region Chhabra, R.P., Peri, S.S., 1991. Simple method for the estimation of

(Eq. (15)), L/T free-fall velocity of spherical particles in power law liquids. Powder

VT solid terminal velocity predicted from correlation, Technology 67, 287–290.

Chhabra, R.P., Richardson, J.F., 1999. Non-Newtonian Flow in the Process

Turton and Clark (1987), L/T

Industries. Butterworth-Heinemann, Oxford.

Vi solid terminal velocity predicted from other corre- Darby, R., 1996. Determining settling rates of particles. Chemical

lations, L/T Engineering 103, 109–112.

U∗ dimensionless velocity, Eq. (26), dimensionless Di Felice, R., 1999. The sedimentation velocity of dilute suspensions of

U∗,exp dimensionless velocity from experimental data, nearly monosized spheres. International Journal of Multiphase Flow

25, 559–574.

dimensionless

Graham, D.I., Jones, T.E.R., 1994. Settling and transport of spherical

U∗,pred dimensionless velocity predicted from Eq. (7) or particles in power-law ﬂuids at ﬁnite Reynolds number. Journal of

Eq. (39), dimensionless Non-Newtonian Fluid Mechanics 54, 465–488.

U∗1 dimensionless velocity for low Reynolds number, Hartman, M., Havlin, V., Trnka, O., Carsky, M., 1989. Predicting the

dimensionless free fall velocities of spheres. Chemical Engineering Science 44 (8),

1743–1745.

U∗2 dimensionless velocity for high Reynolds number,

Heider, A., Levenspiel, O., 1989. Drag coefﬁcient and terminal velocity

dimensionless of spherical and nonspherical particles. Powder Technology 58, 63–70.

Kawase, Y., Ulbrecht, J.J., 1981a. Newtonian ﬂuid sphere with rigid or

Greek letters

mobile interface in a shear thinning liquid: drag and mass transfer.

density difference, (s − )/, dimensionless Chemical Engineering Communications 8, 213–231.

Kawase, Y., Ulbrecht, J.J., 1981b. Motion of and mass transfer from an

s solid density, M/L3

assemblage of solid spheres moving in a non-Newtonian ﬂuids at high

liquid density, M/L3 Reynolds numbers. Chemical Engineering Communications 13, 55–64.

liquid viscosity, M/LT Kelessidis, V.C., 2004. Terminal velocity of solid spheres falling in

a apparent liquid viscosity, M/LT Newtonian and non-Newtonian liquids. Techn. Chron. Sci. J., Techn.

Chamber Greece, V, No 1&2.

Kelessidis, V.C., Mpandelis, G., 2004. Measurements and prediction of

Acknowledgements terminal velocity of solid spheres falling through stagnant pseudoplastic

liquids, Powder Technology (accepted).

The author would like to thank the two anonymous ref- Koziol, K., Glowacki, P., 1988. Determination of the free settling

erees for their valued suggestions. He would also like to parameters of spherical particles in power law ﬂuids. Chemical

acknowledge the continuous support from the Department Engineering Processing 24, 183–188.

Matijašić, G., Glasnović, A., 2001. Measurement and evaluation of drag

of Mineral Resources Engineering of Technical University

coefﬁcient for settling of spherical particles in pseudoplastic ﬂuids.

of Crete. Chemical and Biochemical Engineering Quarterly 15, 21–24.

Miura, H., Takahashi, T., Ichikawa, J., Kawase, Y., 2001. Bed expansion

in liquid–solid two-phase ﬂuidized beds with Newtonian and non-

References Newtonian ﬂuids over the wide range of Reynolds numbers. Powder

Technology 117, 239–246.

Acharya, A., Mashelkar, R.A., Ulbrecht, J., 1976. Flow of inelastic and Nguyen, A.V., Stechemesser, H., Zobel, G., Schulze, H.J., 1997. An

viscoelastic ﬂuids past a sphere I. Drag coefﬁcient in creeping and improved formula for terminal velocity of rigid spheres. International

boundary-layer ﬂows. Rheologica Acta 15, 454–470. Journal of Mineral Processing 50, 53–61.

Briens, C.L., 1991. Correlation for the direct calculation of the terminal Pinelli, D., Magelli, F., 2001. Solids falling velocity and distribution in

velocity of spherical particles in Newtonian and pseudoplastic (power- slurry reactors with dilute pseudoplastic suspensions. Industrial and

law) ﬂuids. Powder Technology 67, 87–91. Engineering Chemistry Research 40, 4456–4462.

Ceylan, K., Herdem, S., Abbasov, T., 1999. A theoretical model for Reynolds, R.A., Jones, T.E.R., 1989. An experimental study of the settling

estimation of drag force in the ﬂow of non-Newtonian ﬂuids around velocities of single particles in non-Newtonian ﬂuids. International

spherical solid particles. Powder Technology 103, 286–291. Journal of Mineral Processing 25, 47–77.

Chhabra, R.P., 1990. Motion of spheres in power law (viscoinelastic) Turton, R., Clark, N.N., 1987. An explicit relationship to predict spherical

ﬂuids at intermediate Reynolds numbers: a uniﬁed approach. Chemical particle terminal velocity. Powder Technology 53, 127–129.

Engineering and Processing 28, 89–94. Turton, R., Levenspiel, O., 1986. A short note on the drag correlation for

Chhabra, R.P., 1993. Bubbles, Drops and Particles in non-Newtonian spheres. Powder Technology 47, 83–86.

Fluids. CRC Press, Boca Raton, Fl. Zigrang, D.J., Sylvester, N.D., 1981. An explicit equation for particle

Chhabra, R.P., 1995. Calculating settling velocities of particles. Chemical settling velocities in solid–liquid systems. A.I.Ch.E. Journal 27,

Engineering 102 (9), 133–135. 1043–1044.

- dieloil.pdfUploaded byFelipe José de Oliveira
- Industrial Flow Measurement_Basics and PracticeUploaded byRicardo Alexis Fuentes Lufitt
- Dimensionless ParametersUploaded bythunderheadII
- Industrial Flow Measurement PracticeUploaded bypneuma110
- 113851176 Fluid Mechanics and Machinery Notes 2Uploaded byVeera Reddy
- ANALYSIS OF BLOOD FLOW THROUGH STENOSED VESSEL UNDER EFFECT OF MAGNETIC FIELDUploaded byIJBSS,ISSN:2319-2968
- Rising Velocity for Single Bubbles in Pure LiquidsUploaded byandrea
- CONTROL VALVE SIZING.pdfUploaded byOutofbox
- Ventilation Network AnalysisUploaded byrodrigo5239
- Progresivo Pag. 1 - 45 InglesUploaded byRodrigo Leal Guzman
- Advance Fluid Mechanics Lectures 7-8Uploaded byKhalid Yousaf
- r05220302 Mechanics of FluidsUploaded byandhracolleges
- Fluid AssignmentUploaded byYush Kumar
- Viso Us Heating SolverUploaded byVignaesh Subramaniam
- Hmt Part 7 - Conv Governing Eqns and SolutionsUploaded bySarabjit Singh
- (Elasticity)Uploaded byMd Rihan Maaz
- A Model of the Development of a Periphyton CommunityUploaded bykiwy10
- Transition and TurbulenceUploaded byTeodora Pinheiro
- Chapter 2 (Lecture 2)Uploaded byHood
- Viscosity of Periodic Suspensions - P. Gondret - 1996Uploaded byeduardo
- 48701Uploaded bykrishima
- Oscillating Flow in a Constricted TubeUploaded bysaek
- Exam 040113Uploaded byalialavi2
- Convection IIIUploaded byChurreya Chai Lom
- 46647 Final 00311Uploaded bykurniawan waskito
- 3Uploaded byKeshava Murthy A
- Me305-Hw01-2002FallUploaded byAhmet Burak Özcan
- 1Uploaded bynraiin
- tan.pptxUploaded byTanvir Bhagat
- pdf . dcf analisisUploaded byEduard Avila Meriño

- AIAA-2008-0315 GAW(1) HysteresisUploaded bytdfsks
- Lkoiyt Fred SweUploaded byTom Mark
- 51218699-modeling-low-reynolds-number-incompressible-flows-using-SPH.pdfUploaded byAbhiyan Paudel
- Chemistry in Chemical EngineeringUploaded bytan97
- Measurement of Flow in PipelinesUploaded bysatishchemeng
- Lab Manual BT3120Uploaded byAmit Kumar
- Chapter 8Uploaded byMohamed Tarek Kamar
- Heat Transfer in Evaporators & CondensersUploaded byjebman1
- Introduction to Numerical Methods in Chemical EngineeringUploaded byAshwini Prince
- Experiment-3-Reynold-and-osborne final version.docxUploaded bySionglin Hon
- Shuttlecock DynamicsUploaded byhokl88
- Boundary conditions at a naturally permeable wallUploaded byOscar A. Luévano
- Discussion OxbourneUploaded byafnan_lion94
- DNSUploaded byDamiano Baccarella
- 111111 Civil Related MCQsUploaded byShams Khattak
- Effect of the Moonpool on the Total Resistance of a DrillshipUploaded byNitin Shinde
- Analysis Comparing Performance of a Conventional Shell and Tube Heat Exchanger Using Kern, Bell and Bell Delaware MethodUploaded byInternational Journal of Research in Engineering and Technology
- Vladimir V. Riabov- Applications of Underexpanded Jets in Hypersonic Aerothermodynamic ResearchUploaded byFraosm
- shukla2010.pdfUploaded byDishant Beniwal
- Acoustics of Fluid-Structure InteractionsUploaded byPrateek Bahl
- Fluid Mechanics Lab ManualUploaded byPrasad Sai
- Stokes Analytical SolutionUploaded bylgrosche
- Proceedings of the Institution of Mechanical Engineers, Part G_ Journal of Aerospace Engineering-1999-Luke-337-46Uploaded bymail_krk
- SwingUploaded byThamaraikannan Seshachalam
- mainUploaded byRavi Kumar
- Thermo - Hydrulic Turbine - Dimensional AnlysisUploaded bybachayadav
- CAEA Windage Loss Reduction WhitepaperUploaded byrajnarayang
- Merlo.pdfUploaded bymuco
- Turbulent FlowUploaded byromeshmorey
- Marsh FunnelUploaded byAndré Mattos