You are on page 1of 10

2nd Mercosur Congress on Chemical Engineering 4th Mercosur Congress on Process Systems Engineering

NON-NEWTONIAN FLUID FLOW IN DUCTS: FRICTION FACTOR AND LOSS COEFICIENTS


Adelson Belizrio Leal, Luis Amrico Calada, Cludia Mirim Scheid*
Departamento de Engenharia Qumica - Universidade Federal Rural do Rio de Janeiro Abstract. The knowledge of the head loss in the flow of non-Newtonian fluids is very important for the execution of pipelines and pumping systems designs, common in plants of almost all kinds of industries. The determination of the total head loss involves the establishment of the friction factor corresponding to pressure drop in the straight section and the loss coefficients caused by each fittings and valves existent in piping systems. In this context, the head loss in globe valve, sudden contraction, 90 degree standard elbow and straight smooth circular pipe was studied. The fluids used in the tests were aqueous solutions of carboxymethylcellulose (CMC) and xanthan gum, in different concentrations, flowing in fully-developed turbulent flow. The rheology was determined by a capillary rheometer, in the same temperature condition of the head loss experiments. All solutions employed in this work showed non-Newtonian and pseudoplastic behavior. The rheological data were adjusted for the Ostwald-de Waele model and the correlation coefficients were above 0.99. The experimental data of friction factors were compared with predictive correlations proved in literature. On the other hand, the experimental data of loss coefficients were compared to that obtained to water flow and non-Newtonian flow those data available in literature.

Keywords: Fittings, pressure drop and valve.

1. Introduction
The non-Newtonian fluids, unlike the Newtonian, are defined as materials which do not conform to a direct proportionality between shear stress and shear rate. Among different models to represent this relationship, the Ostwald-of Waele model, though empirical, represents a great many non-Newtonian fluids given by,

= k n ,

(1)

where e are shear stress and shear rate respectively. While n is flow behavior index and k is fluid consistency index, both rheological parameters. Due to the necessity of the knowledge of the head loss in the execution designs of piping systems and pumping, the study of the flow of fluids non-Newtonian in ducts is very important for engineering. These fluids are present in many processes of the plants of almost all of the types of industries. The total head loss, ht, caused by some piping system consists of the sum of losses regarding to the straight pipe section, hd, and the losses regarding to the several fittings and valves present in this system, hs. In the case of a steady and isothermal flow of the incompressible fluid, without pump and no work involved, the

To whom all correspondence should be addressed. Address: LSP/DEQ/IT, UFRRJ, BR-465, Km 07, 23890-000 Seropdica Rio de JaneiroBrazil E-mail: scheid@ufrrj.br

2nd Mercosur Congress on Chemical Engineering 4th Mercosur Congress on Process Systems Engineering
total head loss can be given by the macroscopic mechanical energy balance applied to two points any of the piping (Bird et al, 1960), is given by,
2 p1 p 2 v1 v 2 2 ht = + + z1 z 2 , g 2g

(2)

where p1 and p2, v1 and v2, z1 and z2, are pressures, average velocities, the heights at points 1 and 2 respectively. is the fluid density and g is the gravitational acceleration. 1.1. Friction factor The turbulent flow in straight cylindrical duct of length L and diameter D, lying in a horizontal plane so that z1=z2 and v1=v2, the eq. (2) can be simplified to the equation given by

ht = hd =

p1 p 2 P = , g g

(3)

where P is the static pressure difference between 1 and 2. The dimensional analysis of the fully turbulent flow in straight cylindrical tube leads to the definition of the Fanning friction factor, f, given by,
P D . L 2v 2

f=

(4)

It is common to find in the literature the definition of Darcy friction factor; relationship by fD=4f. An important review about friction factor was made by Kemblowski and Kolodziejski (1973) and Coelho (1982), in which it is shown the three trends followed by the researchers in the attempt of correlating the factor friction in the turbulent flow of non-Newtonian fluids. The main correlations mentioned by these authors are presented in the table 1, with their respective definitions of Reynolds number and range of behavior index. All the equations, except the eq. (8), use the Fanning friction factor definition given by the eq. (4). The first trend, which maybe the oldest, uses the equations developed for the turbulent flow of Newtonian fluids to correlate the flow fluids of non-Newtonian, using several viscosity definitions in the expression for the Reynolds number. The second correlates the experimental data of friction factor and Reynolds number to the equations of the type Blasius, given by,
f = a Re b ,

(9)

where a and b are functions of the rheological properties fluids. For Newtonian fluids, a is 0.079 and b is -0.25. The correlation proposed by Shaver and Merrill (1959), eq. (5), is an exemple of this trend and was developed 2

2nd Mercosur Congress on Chemical Engineering 4th Mercosur Congress on Process Systems Engineering
with base in the experimental data obtained in the turbulent flow of aqueous solutions of CMC, carbopol and polyisobutylene in cyclohexeno. The last of those trends makes use of the "Prandtl law" to predict the friction factor. Equations of this type have the general form given by,

1 = A log Re f f

)+ C ,

(10)

where the parameters A, B and C are functions of the rheological properties of the fluid. The correlations given by the equations 6, 7, 8 are considered the most important for this case. Dodge and Metzner (1959), through a theoretical and experimental study using solutions of CMC, carbopol and clay, presented a correlation in which the hypothesis of the Newtonian turbulent flow was extended to nonNewtonian turbulent fluid flow. In the special case of fluids that follow the model of Ostwald-of Waele, this correlation is given by the eq. (6), where the Reynolds number used was defined by Metzner and Reed (1955) with base in the laminar flow. The deviation observed is this case were smaller than 2,5%, except for CMC solutions that present viscoelastic behavior. Clapp (1961), working with the flow of pseudoplastic fluids, presented, as well as Dodge and Metzner (1959), a correlation where the parameters A, B and C are functions of the rheological properties of the fluid. The correlation of Clapp is given by the eq. (7) with a maximum deviation of 4% in the range worked (Skelland, 1967). Tomita (1959) proposed a correlation for the turbulent friction factor based on the similarity criteria and Prandtls mixing length theories (Skelland, 1967). Unlike Dodge and Metzner (1959) and Clapp (1961), the correlation proposed by Tomita presents the parameters A, B and C independent of the rheology of the fluid and another definition for friction factor. This definition for friction factor consists of an extension of the Fanning friction factor. In the case of fluids that obeyed the Ostwald-of Waele model, Tomita correlation is given by the eq. (8) and Tomita friction factor corresponding is defined by

fT =

2DP 2n + 1 . 3Lv 2 3n + 1

(11)

It is valid to remind that the correlations of Dodge and Metzner, eq. (7), and Tonita, eq. (9) assume, when n equal to the unit (Newtonian fluids), the well-known expression of Karman-Nikuradse given by,

1 = 4 , 0 log Re f

f 0,4 .

(12)

2nd Mercosur Congress on Chemical Engineering 4th Mercosur Congress on Process Systems Engineering
Table 1. Correlations for turbulent flow of non-Newtonian fluids

Author
Shaver and Merril (1959) Dodge and Metzner (1959) Clapp (1961)

n
0.51.0

Reynolds number
ReSM = D n v 2 n 4n 8n 1 k 3n + 1
n

Correlation
f= 0,079 n (Re SM )10.5
5 2.63
n

(5)

0.41.0

Re MR

D n v 2 n 4n = n 1 8 k 3n + 1

1 4 .0 = 0 .75 log Re MR f f n

2n 2

0 .4 1 .2 n

(6)

0.70.8

Re Cl =

D n v 2n 8 n 1 k
1 n

2 n 0.45n 2.75 1 4.53 = log Re Cl f 2 + f n n

(7)

Tomita (1959)

0.20.9

3n + 1 6 D n v 2n n Re T = k 2n + 1 2n n

1 = 4 . 0 log Re T fT

f T 0 .4

(8)

2.2. Loss Coefficient The head loss in fittings and valves, hs, is resulted of the wall friction, of changes in the cross-section and/or direction of the flow and is expressed in function of the loss coefficients K, given by,

hs = K

v2 . 2g

(13)

The loss coefficient, K, is characteristic of each fitting and valve type while, hs, it can be obtained through the eq. (2). However, it should be taken into account the particularities of each fitting and valve during the simplifications of this equation, for example: the variation in the cross-section flow and the heights. It is valid to observe that in the case of the fittings, as a sudden contraction, in which there is variation in the average velocity before and after the fitting, the loss coefficient is related with the velocity regarding to the smallest diameter. The Crane Company (1976) provided a very extensive tabulation of loss coefficients for turbulent flow thorough various types and sizes of fitting and valves for Newtonian fluids. In the case of non-Newtonian fluids, Turian et al. (1998) studied the pressure drop in different types and sizes of fittings, valves and venturi meters for the flow of non-Newtonian slurries. These authors obtained loss coefficients similar to the flow of water for majority of the fittings and valves. Already Etmad (2004), accomplishing similar study with aqueous solutions of the CMC, obtained loss coefficients dependent of the rheology of the non-Newtonian fluid. In this context, this work seeks to test the correlations of friction factor and to compare the loss coefficients existent in the literature to that obtained experimentally, all in turbulent regime. For such an objective, polymeric solutions of xanthan gum and CMC were used in different concentrations, flowing through smooth cylindrical tube and two basic types of fittings (90 elbow and 1-3/4 sudden contraction ) and one globe valve.

2nd Mercosur Congress on Chemical Engineering 4th Mercosur Congress on Process Systems Engineering

2. Materials and Methods


The experimental apparatus used for the tests is showed in figure 1. It consists of a reservoir tank, with 500 liters of capacity, a centrifugal pump of 1.5 HP (model CAM W-6C made by Dancor, Brazil) and 6 m of pipe system in a horizontal plane working in loop. The friction factor was studied in a section of length 387cm with 2.65 cm of internal diameter. The friction loss coefficient was determined for globe valve and 90 degree standard, elbow with 1 diameter both and a sudden contraction 1 to . All the parts, except valve were made of PVC. The pressure drop in the fittings and the straight section was obtained by tube in U manometer in turbulent flow without end effects. The pressure drops in the fittings and the straight section were measured by means of U-tube manometers containing mercury and tetrachloromethane. The flow rate was calculated using a measuring cylinder and stop watch. For the tests, aqueous solutions of carboxymethylcellulose (CMC) and xanthan gum in different concentrations were used. The rheological behaviors of the solutions were obtained by the capillary rheometer in the average temperature and in the same range of shear rate of each experiment. All the fluids presented non-Newtonian and pseudoplastic behavior allowing the adjustment for the model of Ostwald-of Waele with coefficients of correlation over than 0.99 for all worked concentrations. The results of the rheology are presented in table 2.
Table 2. Concentration, temperature, rheological parameters and density of the polymeric solutions

Solutions

% (p/v) 0.010 0.012

T (C) 26.0 28.0 26.0 28.5 26.9 28.5 25.4 24.5 26.5 28.0 30.0 31.0 26.0 27.5 29.0 29.5 30.0 32.7 32.9

n 0.965 0.979 0.907 0.906 0.786 0.799 0.579 0.399 0.397 0.921 0.933 0.895 0.858 0.840 0.832 0.817 0.830 0.825 0.816

k
(dyna s /cm ) 0.0144 0.0121 0.0219 0.0215 0.0590 0.0520 0.3185 1.8490 1.8680 0.0227 0.0206 00314 0.0521 0.0645 0.0869 0.1109 0.1964 0.2404 0.3160
n 2

(s-1) 100 - 2044 110 - 2028 85 - 1739 88 - 2052 45 - 2012 52 - 1720 27 - 1063 2 - 496 2 - 485 81 - 1527 85 - 1570 64 - 1544 43 - 1198 36 - 1100 25 - 818 20 - 682 6 - 251 5 - 209 3 - 158

(g/cm3) 1.01 1.01 1.01 1.01 1.01 1.01 1.01 1.01 1.01 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00

XG

0.020 0040 0.100 0.100 0.150 0.200 0.250 0.400 0.500 0.900 1.000 1.350

CMC

n: Flow behavior index; k: Fluid consistency index; : Shear rate; T: Temperature and : Fluid density.

2nd Mercosur Congress on Chemical Engineering 4th Mercosur Congress on Process Systems Engineering

Sudden contraction 90 Elbow 50 80 50 50 50 50

Straight duct 387 50 V=500L Globe Valve Manometer Manovacuumeter 120

Distances are in cm.

Centrifugal Pump

Fig. 1. Schematic of the experimental apparatus.

3. Results and Discussions


In order to check the validity of the friction factor and loss coefficients data obtained in the unit of flow tests were done using water as standard fluid. The friction factors and the loss coefficients were evaluated throught the eq. (4) and eq. (13) and the data of P e v obtained experimentally. The deviations for the friction factor were always smaller than 5% in module for every range of Reynolds studied, indicating that the unit of test can be used in the determination of the friction factors for other fluids. In case of loss coefficients, the table 3 shows the values provided by the references and obtained in the experimental unit, with their respective standard deviations and correlation coefficients, for all the fittings and valve studied. The deviations differences observed can be explained by the different geometric configurations of the fittings employed in the determination of the loss coefficients while the low standard deviations and the correlation coefficients near the unit, confirm the capacity of the experimental unit to supply data for other fluids.
Table 3. Comparison of Newtonian experimental data with available results in the literature

Fittings and Valve K Globe valve 90 Elbow 1-3/4 Sudden contraction 0.41 10.40 1.33

Experimental 0.10 0.05 0.03 R2 0.99 0.99 0.99

Crane (1976) K 7.82 0.69 0.20

Turian (1998) K 10.000,75 1.110,163 -

K: Average loss coefficients, R2: Correlation coefficients and : Standard deviation.

2nd Mercosur Congress on Chemical Engineering 4th Mercosur Congress on Process Systems Engineering
The friction factors obtained experimentally through the eq. (4) were compared with the friction factors predicted by the eq. (5), eq. (6), eq. (7). and eq. (8), with the purpose of evaluating the performance of each correlation in turbulent flow, ReMR > 4000. The figure 2a shows that the deviations between the experimental and predicted friction factors by the correlation of Shaver and Merrill (1959), eq. (5), are less than 15% in module for the solutions of CMC. In the case of xanthan gum solutions, the deviations vary among -15% for solutions with n=0.965 to value larger than +90% for solutions with n=0.399. It is pertinent to remind that Shaver and Merrill (1959) observed deviations among +33% to -15% for solutions with 0.53n1.0 and it is not indicated the use of their correlation to solutions with n <0.40. The performance of correlation of Dodge and Metzner (1959), eq. (6), is presented in the figure 2b. It can be observed in this figure that the predicted friction factors are larger than obtained experimentally, for all of the solutions of CMC and xanthan gum studied. In the case of the xanthan gum, the deviations were between -35% and -15% for the solutions with n=0.399 and n=0.965 respectively. Taking into account the case of CMC flow with 0.81<n<0.92, the deviations between the experimental and predicted by the eq. (6) were smaller than -10%. The figure 2c shows the deviations between the experimental friction factors and predicted by correlation of Clapp (1961), given by the eq. (7). It is pertinent to point out that this correlation is valid for 0.698n0.813, but it was used to predict frictions factors of xanthan gum and CMC solutions with 0.399n0.965 showing deviations which are among +5% and-20%. The figure 2d shows the deviations between the experimental and predicted friction factors by the correlation of Tomita (1959), eq. (8). This equation overestimates the frictions factors for all the solutions. In the case of CMC solutions, the deviations are among -10% to -20% while for the xanthan gum solutions the deviations were between -20% and -70%. The table 4 shows the results of the statistical treatment applied to the experimental friction factors (totalizing 151 points) and calculated by the equations 5, 6, 7 and 8. It can be observed in this table that the correlation of Clapp (1961), eq. (7) is the best one while the correlation of Shaver and Merrill (1959), eq. (5), supplies the worst results for the friction factor. In this table, y is average, is standard deviation and DMA is average absolute deviation.
Table 4. Statistical treatment for several correlations

Statistical y

Shaver and Merrill 1.49

Dodge and Metzner 0.89

Clapp 0.93

Tomita 0.79

Equation

y=

1 n f exp . n i 1 f pred.

1.66

0.09

0.06

0.14

1 n f exp . y = n 1 i =1 f pred. i
DMA =

0.5

DMA (%)

15.98

13.66

8.66

33.26

100 n f exp . f pred. n i =1 f exp .


7

2nd Mercosur Congress on Chemical Engineering 4th Mercosur Congress on Process Systems Engineering

0.012
+90% +40% +20% +10% -10%

0.010
n=0.816 - CMC n=0.817 - CMC n=0.825 - CMC n=0.830 - CMC n=0.832 - CMC n=0.840 - CMC n=0.858 - CMC n=0.895 - CMC n=0.921 - CMC n=0.399 - XG n=0.579 - XG n=0.773 - XG n=0.907 - XG n=0.965 - XG

+10% -10%

0.010
0.008

-20% -30%

0.008

fExperimental

0.006

0.004

0.002

n=0.816 -CMC n=0.817 -CMC n=0.825 -CMC n=0.830 -CMC n=0.832 -CMC n=0.840 -CMC n=0.858 -CMC n=0.895 -CMC n=0.921 -CMC

n=0.399 -XG n=0.579 -XG n=0.773 -XG n=0.907 -XG n=0.965 -XG

fExperimental

0.006

0.004

0.002

0.000 0.000

0.002

0.004

0.006

0.008

0.010

0.012

0.002

0.004

0.006

0.008

0.010

fPredicted-Shaver and Merrill

fPredicted - Dodge and Metzner

(a)
0.012

(b)
0.012
+10%

0.010

0.008

f Experimental

fExperimental

n=0.816 - CMC n=0.817 - CMC n=0.825 - CMC n=0.830 - CMC n=0.832 - CMC n=0.840 - CMC n=0.858 - CMC n=0.895 - CMC n=0.921 - CMC

n=0.399 - XG n=0.579 - XG n=0.773 - XG n=0.907 - XG n=0.965 - XG

-10% -20%

0.010

0.008

n=0.816 - CMC n=0.817 - CMC n=0.825 - CMC n=0.830 - CMC n=0.832 - CMC n=0.840 - CMC n=0.858 - CMC n=0.895 - CMC n=0.921 - CMC

n=0.399 - XG n=0.579 - XG n=0.773 - XG n=0.907 - XG n=0.965 - XG

-10% -20% -30%

-50%

0.006

0.006

0.004

0.004

-70%

0.002

0.002
0.002 0.004 0.006 0.008 0.010 0.012

0.002

0.004

0.006

0.008

0.010

0.012

fPredicted-Clapp

fPredicted-Tomita

(c)

(d)

Fig. 2. Comparison of experimental friction factor and the factor predicted by eq. (5), (6), (7) e (8).

The curves of hs versus v2/2g for elbow 90, globe valve and sudden contraction are presented in the figures 3a, 3b and 3c respectively, for all the solutions. In these plotters the slope represents the loss coefficients of each fitting and it can be observed that the rheological behavior of the fluid is not important in the loss coefficient results. For the globe valve and the elbow 90, the dispersion of the points was smaller than in the sudden contraction presented deviations smaller than 5%. For sudden contraction this deviations were in maximum 15%. The table 5 presents a summary of the medium loss coefficients for non-Newtonian fluids and water with their respective deviations.

2nd Mercosur Congress on Chemical Engineering 4th Mercosur Congress on Process Systems Engineering

120 100 80
n=0.816 - CMC n=0.817 - CMC n=0.825 - CMC n=0.830 - CMC n=0.832 - CMC n=0.840 - CMC n=0.858 - CMC n=0.895 - CMC n=0.933 - CMC n=0.397 - XG n=0.579 - XG n=0.786 - XG n=0.906 - XG n=0.979 - XG

800 700 600 500

h s (c m )

n=0.817 - CM C n=0.825 - CM C n=0.830 - CM C n=0.832 - CM C n=0.840 - CM C n=0.858 - CM C n=0.895 - CM C n=0.933 - CM C

n=0.397 - X G n=0.579 - X G n=0.799 - X G n=0.906 - X G n=0.979 - X G

60 40 20

hs (cm)
30
2

400 300 200 100

0 0 10 20 40 50 60 70 80

0 0 10 20 30
2

40

50

60

70

v /2g (cm)

v /2g (cm )

(a)
120 100 80
n=0.816 - CMC n=0.817 - CMC n=0.825 - CMC n=0.830 - CMC n=0.832 - CMC n=0.840 - CMC n=0.858 - CMC n=0.895 - CMC n=0.933 - CMC n=0.397 - XG n=0.579 - XG n=0.786 - XG n=0.906 - XG n=0.979 - XG

(b)

h s(cm)

60 40 20 0 0

20 40 60 80 100 120 140 160 180 200 220

v /2g(cm)

(c)
Fig. 3. Loss coefficients for: a: 90 elbow, b: globe valve and c: 1-3/4 sudden contraction.

Table 5. Mean losses coefficients for turbulent flow of xanthan gum and CMC

Fittings and Valve Globe valve 90 Elbow 1-3/4 Sudden contraction K

XG 0.40 0.071 0.062 K

CMC 0.420 0.067 0.051

XG - CMC K 10.67 1.42 0.45 0.457 0.067 0.062

Water K 10.40 1.33 0.41 0.10 0.05 0.03

10.95 1.42 0.49

10.53 1.42 0.43

K: Average loss coefficients and : Standard deviation.

2nd Mercosur Congress on Chemical Engineering 4th Mercosur Congress on Process Systems Engineering

4. Conclusions
For the studied non-Newtonian solutions, it can be concluded that the correlations of Clapp (1961), Dodge and Metzner (1959) and Tomita (1959), in the average, overestimate the friction factor. While the opposite occurs with the correlation proposed for Shaver and Merrill (1959). It is important to point out that the correlation of Clapp (1961) was the one that presented better performance and Shaver and Merrill (1959) is the worst. In relation to the loss coefficients, it can be concluded that these do not depend on the rheological behavior of the fluid, what it indicates that the defined loss coefficients for Newtonian fluids can be used in the calculation of pressure drop for the non-Newtonian fluid flow.

References
Bird, R. B., Stewart, W. E., Lightfoot, E. N. (1960). Transport Phenomena, John Wiley & Sons, NY. Clapp, R. M. (1961). International Developments in Heat Transfer, Part III, 652-61; D-159: D-211-5. A.S.M.E., NY. In Skelland, A. h. P. (1967). Non-Newtonian Flow and Heat Transfer. John Wiley & Sons, NY. Coelho, G. L. V., (1982). Reologia e Escoamento Turbulento de Suspenses de Minrio de Ferro, Thesis de M. Sc., Campinas, Brasil. Crane Co. (1976). Flow of Fluids Through Valves, Fittings and Pipe. Tech. Paper No. 410, 16th printing. Crane Co., 300 Park Avenue, NY. Dodge, D. W., Metzner, A. B. (1959). Turbulent Flow of non-Newtonian Systems. A.I.Ch.E. Journal, 5, 191. Etmad, S. Gh. (2004). Turbulent Flow Friction Loss Coefficients of fittings for Purely Viscous non-Newtonian Fluids. Int. Comm. Heat Mass Transfer, 31, 763. Kemblowski, Z., Kolodziejski, J. (1973). Flow Resistances of non-Newtonian Fluids in Transitional and Turbulent Flow. Int. Chem. Eng., 13, 265. Metzner, A. B., Reed, J. C. (1955). Flow of non-Newtonian Fluids-Correlation of the Laminar, Transition and TurbulentFlow Regions. A.I.Ch.E. Journal, 1, 434. Shaver, R. G., Merrill, E. W. (1959). Turbulent Flow of Pseudoplastic Polymer Solutions in Straight Cylindrical Tubes. A.I.Ch.E. Journal, 5, 181. Skelland, A. h. P. (1967). Non-Newtonian Flow and Heat Transfer. John Wiley & Sons, NY. Tomita, Y. (1959). A study on non-Newtonian Flow in Pipe Lines. Bulletin of J.S.M.E, 2, 10. Turian, R. M., Ma, T. W., Hsu, F. L. G., Sung, M. D. J., Plackmann, G. W. (1998). Flow of Concentrated non-Newtonian Slurries: 2. Friction Losses in Bends, Fittings, Valves and Venturi Meters. Int. J. Multiphase Flow, 24, 243.

Acknowledgments
The authors acknowledge the financial support of the CAPES.

10

You might also like