Chemical Engineering

International Review of

Chemical Engineering Rapid Communications

(IRECHE) Contents Modeling of Clay Paste Extrusion Through a Rectangular Die by Fernando A. Andrade, Hazim A. Al-Qureshi, Dachamir Hotza

478

Analysis of Irradiator Tube Purging for Fiber Drawing by Showkat J. Chowdhury, Tyler Kirby

484

Removal of Copper (II) from Aqueous Solutions with Activated Carbon Obtained by Chemical Activation of Orange Peel by Liliana Giraldo, Juan Carlos Moreno-Piraján

494

Environment Protection – A COMSOL Simulation of Secondary Clarifiers in the Activated Sludge Process by Camelia Gavrila, Ion Gruia

503

Reaction Kinetics and Operation in a Packed Bed Reactive Distillation Column - Condensation Reaction of 3-Bromo Benzaldehyde by Kiran Prajapati, Raj Arora, Jigisha Parikh

508

Fed-Batch Biosurfactant Production in a Bioreactor by Frederico A. Kronemberger, Cristiano P. Borges, Denise M. G. Freire

513

Replacing Conventional Oxygenation Systems in Bioreactors by Membrane Contactors: Modeling and Performance Investigation by Frederico A. Kronemberger, Denise M. G. Freire, Cristiano P. Borges

519

Cascaded PSA Process for the Purification of Argon Using an Oxygen Selective Zeolite X Adsorbent by Sunil A. Peter, Pramathesh R. Mhaskar, Arun S. Moharir, Raksh V. Jasra

529

Energy for Sustainable Development by Mikhail Krasnyansky

539

Extraction and Characterization of Extracellular Polymeric Substances (EPS) from Waste Sludge of Pulp and Paper Mill by M. Pervaiz, M. Sain

550

Copyright © 2010 Praise Worthy Prize S.r.l. - All rights reserved

International Review of Chemical Engineering (I.RE.CH.E.), Vol. 2, N. 4 July 2010

Modeling of Clay Paste Extrusion Through a Rectangular Die Fernando A. Andrade, Hazim A. Al-Qureshi, Dachamir Hotza Abstract – An important aspect of clay paste extrusion is the pressure as a function of extruded velocity/distance traveled. Therefore, the present theory concentrates on deriving basic equations that control the clay paste flow behavior of rectangular extrusion process during the steady-state flow. The main parameters which control the extrusion pressure are the effective stress in compression of the clay paste, the geometry of extrusion tools and billet, operational conditions and the coefficients of friction between the barrel and the die land surfaces. The theoretical extrusion pressure results were analyzed and discussed thoroughly. Finally, it can be concluded that the present theoretical analysis serves to place the present approach in the context of work on other extrudable materials of different geometry. In addition, it can be used as an effective tool for evaluating and obtaining clay paste materials with optimized properties for extrusion process in the ceramic industry. Copyright © 2010 Praise Worthy Prize S.r.l. - All rights reserved.

Keywords: Clay, Extrusion, Materials Processing, Mathematical Modeling, Paste, Unit Operations

Nomenclature

I.

Abbreviations and symbols Initial area A0 Final area A1 Initial diameter D0 F Applied external force h Height Initial height of body test h0 Final height of body test hf Final height of extruder l1 Critical length Lc Average extrusion pressure Pave R Radius Initial radius R0 Final radius Rf w Width z Punch travel Greek Letters Semicone angle of the die α ε Effective strain Coefficient of friction between the punch and µ the clay paste Coefficient of friction between the wall and µw the clay paste Radial stress σr Axial stress σz Average extrusion pressure σave σ Effective stress Effective stress of compression σ comp

Introduction

Extrusion is a technique of conformation of powders, used for the processing of ceramic products since the 18th century. It is a technique of production associated with a low cost and high productivity, for products of crosssection mainly constant. Extrusion of clay has been employed successfully for mass production, for traditional construction materials, refractories, electronic substrates, composites and others [1]-[3]. Generally the extrusion process consists of several stages such as, feeding into the cylinder/barrel, extruding through the die, flow through the die-land and the ejection of the product. These phases can be carried out by employing different types of machines. An extrudate so formed can then be dried and sintered, removing the liquid phase and burning out any binders that may have been added, to form a finished ceramic product [4]. It is known that the extrusion mechanisms may occur in a laminar flow type which takes place in the container, and plug flow occurs in the die-land. These mechanisms depend on the geometry of the tooling and the flow properties of the clay paste, and consequently an expression has been put forward to predict the extrusion pressure. The success of the extrusion process of clay paste is controlled by tool geometry, moisture and lubricants, type and amount of binder, ingredients particle size and distribution, temperature and others. Needless to say, the plasticity of clay paste is an extremely important factor for the success of the extrusion process. Plasticity, in this case, and particularly in clay mineral systems, is defined as a property that shows shape changes without rupture when a clay body with added water is submitted to an external force. Furthermore, when the force is removed or

Manuscript received and revised June 2010, accepted July 2010


478

Fernando A. Andrade, Hazim A. Al-Qureshi, Dachamir Hotza

reduced below a value corresponding to the yield stress, the shape is maintained. However, the plasticity determination is not always an easy task since it cannot be immediately interpreted and applied. In fact, there are several direct and indirect methods for measurement and characterization of the plasticity of a clay body, although its experimental determination, in some cases, is operator dependent, which in turn may produce different results between different methods. Among these methods, the Atterberg’s plasticity index [5], the Pfefferkorn’s plasticity index [6], the stress/strain curves [5], the indentation [6]-[8], and rheological measurements [9] are the most used. Several works have been published concerning the different aspects of the extrusion, such as flow behavior, die and die-land geometries, velocity factor and shearing index among others [10]-[14]. In this context, a mathematical model for evaluation of the plasticity of clay bodies was developed from applied concepts of the plasticity theory [15],[16] by using the stress/strain diagram under compression [17],[18]. An analogy between plasticity of metals during forming and clays during extrusion has been already demonstrated in the literature [19]. Certain factors, such as the coefficients of friction and the effective flow stress of the clay could be then estimated from this model. In the derived expressions proposed in this work, empirical constants determined from the composition of the ingredients are not needed. The main objective here is to present semi-theoretical models to predict the average extrusion pressure based on the modified plasticity equations.

dσ z =

4µ wσ r dz D0

(2)

Since the diameter is constant, d ε r = d εθ , and from the Levy-Mises relationships [15], σ r = σ z : ln σ z =

4µw z +C D0

(3)

Fig. 1. Diagrammatic sketch of extrusion of clay paste through a rectangular die

Using the boundary condition where σz=σave (µw=0) at z=Lc and considering that the total pressure is given by (σ z )total = (σ z )µ ≠ 0 + (σ z )µ =0 , the average extrusion pressure with wall friction can be expressed as: Ptotal = (σ z )total = ln

II.

⎧⎪ ⎡ 4µ ⎤ ⎫⎪ = ⎨1 + exp ⎢ w ( z − Lc ) ⎥ ⎬1.5 ⎣ D0 ⎦ ⎭⎪ ⎩⎪

Theoretical Analysis

II.1.

Approximate Treatment

The extrusion process consists of forcing the clay paste through a die thereby reducing its cross-sectional area from A0 to A1 and increasing its length. In this analysis it is assumed that the steady-state has been reached, thus the average ideal extrusion pressure (µcomp=µ=0) for uniform deformation is given by: ln

σ ave

µ =0

= 1.5

A0 A1

∫

σ dε

A0 A1

∫

(4)

σ dε

0

It has been demonstrated experimentally, especially in metal processing operations [16], that at a certain distance Lc there is a rise in the pressure near the end of the extrusion process. At this point, funneling effect becomes apparent, the so-called “coring point”. It is fundamental to predict theoretically the average extrusion pressure from Eq. (4), then flow stress of the clay paste from compression test must be known. However, this has been treated and analyzed mathematically by the authors for clay paste and can be expressed as follows [17],[18]:

(1)

0

where σ is the flow stress, ε is the flow strain, µ is the coefficient of friction between the cylinder wall and the clay paste, and a shear factor (1.45 to 1.55) can be introduced into Eq. (1), which tends to improve the prediction of the extrusion pressure over a wide range. The above relation does not take into account the cylinder wall friction; therefore, to take this parameter into account, then consider the equilibrium in the extrusion direction (Eq. (2)), as shown in Fig. 1:

⎡ 2µ ⎤ rf − r ⎥ h ⎣ ⎦

σ = σ z = −σ comp exp ⎢

(

)

(5)

The axial compressive load can be evaluated from F=


rf

∫0

2π rσ z dr and using Eq. (5), thus:

International Review of Chemical Engineering, Vol. 2, N. 4

479


II.2.

⎡ h ⎛ h ⎞ ⎤ ⎢− ⎜ rf + ⎟+ ⎥ 2µ ⎝ 2µ ⎠ ⎥ ⎢ F = −2πσ ⎢ 2 ⎛ 2µ r f ⎞ ⎥ ⎢ + h exp ⎜ ⎟⎥ ⎢⎣ 4µ 2 ⎝ h ⎠ ⎥⎦

The extrusion process of clay paste and the free body in equilibrium are shown schematically in Fig. 1. The average pressure in the left end of the rectangular portion zone B is taken from Sachs’s theory [15] and is given by:

(6)

This expression will be used for the determination of the load vs. the variation in diameter for each compressively tested clay paste. In this way, a more accurate approach, in contrast to the other existing methods, such as the Atterberg’s and Pfefferkorn’s plasticity indices, is expected for obtaining ceramic bodies with optimized plasticity for a given application. In addition Eq. (5) can be inserted in Eq. (4) and integrated, and also by replacing h by z and D0 by R0 to maintain consistency, the expression for average extrusion pressure when α = 90º can be expressed as (7): Ptotal = 0.75

σ comp

( l f − R0 )

⋅

z

µcomp

Pave

⎡ 2σ (1 + B ) ⎢⎛ l0 = ⎜ 3 B ⎢⎜⎝ l f ⎣

B ⎤ ⎞ ⎟ − 1⎥ ⎟ ⎥ ⎠ ⎦

(8)

where B = µW cot α, α is the semicone angle of the die and σ is the flow stress of clay paste in compression test, Eq. (5). On the other hand, the equilibrium of forces in the tapered zone A can easily be established, and assuming a biaxial state of stress, then the equation can be integrated. Now, in the integrated expression putting the value z=0 , and the average ideal extrusion pressure given in Eq. (9) [20], the final theoretical expression for the average extrusion pressure of clay paste through a rectangular die can be determined and is expressed as follows:

⎡ 2µw ⎤⎫ ⎪⎧ ( z − Lc )⎥ ⎪⎬ ⋅ ⎨1 + exp ⎢ ⎣ R0 ⎦ ⎭⎪ ⎩⎪

⎡ 2µcomp ⎤ ⎫⎪ ⎪⎧ ⋅ ⎨1 − exp ⎢ l f − R0 ⎥ ⎬ ⎪⎩ ⎣ z ⎦ ⎪⎭

(

Analysis of Extrusion Through a Rectangular Die

)

σ ave =

Pave = 3

σ comp

z

( l f − R0 ) 2µcomp

σ comp

z

( l f − R0 ) µcomp

⎧⎧ ⎡ 2 (1 + B ) ⎢⎛ R0 ⎪⎪ ⋅ ⎨⎨1 + ⎜ 3 B ⎢⎜⎝ l f ⎪⎪⎩ ⎣ ⎩

⎡ 2µcomp ⎤ ⎪⎫ ⎪⎧ l f − R0 ⎥ ⎬ ⎨1 − exp ⎢ ⎣ z ⎦ ⎭⎪ ⎩⎪

(

)

(9)

⎧⎪ ⎡ µcomp ⎤ ⎫⎪ l f − R0 ⎥ ⎬ ⋅ ⎨1 − exp ⎢ ⎣ z ⎦ ⎪⎭ ⎩⎪

(

)

B ⎤⎫ ⎞ ⎛ lf ⎪ ⎪⎧ ⎡ L ⎟ − 1⎥ ⎬ exp ⎨2 B ⎢ − tan α + ⎜1 − ⎟ ⎥⎪ ⎝ R0 ⎪⎩ ⎢⎣ R0 ⎠ ⎦⎭

⎫ ⎞ ⎤ ⎪⎫ ⎪ ⎟ ⎥ ⎬ − 1⎬ ⎠ ⎥⎦ ⎪⎭ ⎪ ⎭

(10)

mechanical mixing and subsequently, left in a sealed plastic container for 24 h for moisture equilibrium. Those moisture contents were chosen between the Atterberg’s plastic and liquid limits. The clays used are all high plastic clays [21], as defined by Reeves et al. [22], since they present liquid limits between 50 and 70 wt.%. In a later stage, cylindrical specimens were manually prepared in PVC molds with 47.7 mm diameter and 49.8 mm height. In order to evaluate qualitatively the stress state along the longitudinal section, some of the specimens were sectioned and marked vertically with fine lines. Those samples were submitted to different uniaxial compressive loadings as shown in Fig. 2. It can be observed that barreling has taken place, and this is due to the friction between the ends of the specimen and the deforming tools/punch. Due to this, the coefficient of friction was taken into account as one of the mathematical parameters to analyze the forces that act on a cylindrical clay compact and was also assumed to be constant.

III. Materials and Methods III.1. Materials The prediction of the theoretical results will be determined by compressibility curves of the clay paste with different humidity levels. The materials used for the execution of the experiments were the AC12, AC39 and kaolin clays, whose chemical composition is given in Table I [21]. III.2. Methods The as-received clays were disaggregated in a ball mill and then sieved (20 mesh/840 µm), so that 1 kg of powder for each type of clay was obtained. The moisture content of the clays was determined from samples with about 50 g of the material introduced into porcelain crucibles, which were placed in the oven at 120 ±5°C for 24 h. Samples having a moisture content of 52 wt.%, 56 wt.% and 60 wt.% for the AC12, AC39 and kaolin clays, respectively, were prepared and homogenized by Copyright © 2010 Praise Worthy Prize S.r.l. - All rights reserved


480


TABLE I CHEMICAL COMPOSITION (WT.%) OF CLAYS AC12, AC39 AND KAOLIN, OBTAINED BY XRF Composition

AC12

AC39

Kaolin

SiO2

69.41

51.61

32.57

Al2O3

18.51

32.57

38.50

Na2O

0.08

0.89

0.10

K2O

2.91

1.57

0.25

MgO

0.82

1.59

Traces

CaO

0.05

1.48

Traces

TiO2

0.73

0.08

–

Fe2O3

2.20

1.04

0.40

P2O5

0.14

0.12

–

Loss on Fire (1000 ºC)

5.15

9.02

13.57

IV.

Results and Discussion

Before any comparison between the theoretical and the experimental results, it is fundamental to determine the flow stress (effective stress) of the clay paste subjected to uniaxial compression as explained previously. Typical load-radial expansion curves for different moisture levels for each type of the tested clay are shown in Fig. 3.

Fig. 3. Comparison between theoretical curves and experimental points for compression of kaolin paste having different levels of moisture

Initially a small elastic strain occurs followed by a relatively high plastic strain, which terminates with the material’s rupture. In addition, all the curves shown demonstrated similar behavior to that in literature [1],[2]. It is evident that the elastic limit of these materials is relatively small, and was not included in the theoretical model. Similar behavior was obtained for the other tested clay pastes. Therefore, by applying the iterative process [23], it was possible to determine the average coefficient of friction and the effective compressive flow stress for each type of clay with a specific moisture content, using Eqs. (3) and (4). Fig. 3 demonstrated that the theoretical curves (Eq. (4)) correlated well with the experimental data for µW =0 from which it could be interpreted that this simple model can be used for different types of clay with variable amounts of moisture content. Previously published experimental extrusion curves had been carried out with pastes and reproduced here for comparison purposes [4]. The experimental curve shown in Fig. 4 demonstrates the variation of the extrusion pressure as a function of punch travel. This curve consists of two principal and easily recognizable phases: (i) the coining phase, where the initial compression of the clay paste caused a rapid build-up of pressure; (ii) the steady-state phase, where the pressure stabilizes noticeably as the extrusion process proceeds steadily. The theoretical analysis of extrusion pressure for tapered die which is given in Eq. (7) involves several process parameters and projected a better understanding of the paste extrusion process.

Fig. 2. Sectioned clay paste sample showing strain lines representing the stress distribution

Four specimens of each type of clay having different moisture contents were tested. This procedure was repeated for the other types of clay as given in Table I. The average diameters were measured by means of digital photographs performed at each pause of the machine, using the software Corel Draw 11. Copyright © 2010 Praise Worthy Prize S.r.l. - All rights reserved


481


Fig. 6. Comparison between theoretical and experimental curves for the extrusion process of kaolin paste for defined coefficients of friction

Fig. 4. Comparison between theoretical and experimental curves for extrusion of kaolin paste at different coefficients of friction for the compression test

V.

Conclusion

The factors that influence the behavior of the clay during its processing were used to evaluate the average extrusion pressure. Among these factors, it was clear that the effective flow compressive stress of the clay subjected to uniaxial compression is a predominant factor. This being in addition to the coefficients of friction between the cylinder wall and the compression surface, which were estimated by trial and error technique. The involved parameters in the equation have remarkable influence on the predicted results of the extrusion pressure, such as the moisture content, type of clay, die dimension, and others. The agreement between the experimental and the theoretical results makes the present theory more reliable and a potentially useful tool for the evaluation of clay materials with optimized properties for a given application. The use of this theory will help reduce the number of experiments needed for qualifying preparation of the required moisture and chemical compositions of the clay paste. Also, it is worth mentioning, that care must be taken in estimating the effective flow stress and the coefficients of friction otherwise this may lead to unrealistic results of the pressure.

Fig. 5. Comparison between theoretical and experimental curves for the extrusion process of kaolin paste at different coefficients of friction of the cylinder wall

It appears that the predominant factors which govern the extrusion pressure are the coefficients of friction µW and µ, which were determined by the curve fitting technique. This becomes clear from the plot of the extrusion pressure as a function of the punch travel for various values of µW and µ, as shown in Figs. 4 and 5. On close examination of these figures, it becomes evident that certain values of the coefficient of friction can be selected so that the best fit occurs between the theoretical and the experimental curves. Needless to say, the chosen values were µW = 0.50 and µ= 0.10. These values were obtained from the curve fitting technique since they are not available so easily in published literature. It must be mentioned that, although the coefficient of friction can really be a dynamic variable as a function of the stress, processing parameters and porosity, which is an acceptable procedure, here it is assumed to be a constant, due to the difficulty of its formulation [23]. A theoretical curve using these values and α= 30º demonstrated remarkable agreement as shown in Fig. 6. The coefficient of friction for the cylinder wall is rather high, this compensates for the fact that the die land effect was not considered in the present analysis.

Acknowledgements The authors wish to thank UFSC’s Departments of Chemical and Mechanical Engineering for the use of the facilities and CNPq for partially financing the project and grants for the authors.

References [1] [2] [3] [4]


J. Benbow, J. Bridgwater, Paste Flow and Extrusion (Oxford University Press, New York, 1993). J.S. Reed, Principles of Ceramics Processing, 2nd ed. (Wiley, New York, 1995). F. Händle, (Ed.), Extrusion in Ceramics (Springer, New York, 2007). S. Amarasinghe, I. Wilson, Interpretation of paste extrusion data, Transactions of IChemE 76-A (1998) 3-8.


482


[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14] [15] [16] [17]

[18]

[19]

[20]

[21]

[22] [23]

M.J. Ribeiro, J.M. Ferreira, J.A. Labrincha, Plastic behaviour of different ceramic pastes processed by extrusion, Ceramics International 31 (2005) 515-519. C.O. Modesto, A.M. Bernardini, Determination of clay plasticity: Indentation method versus Pfefferkorn method, Applied Clay Science 40 (2008) 15-19. V. Doménech, E. Sánchez, V. Sanz, J. García, F. Ginés, Assessing the plasticity of ceramic masses by determining indentation force, 3rd World Congress on Ceramic Tile Quality, Castellón, Spain (1994). T.W. Feng, Using a small ring and a fall-cone to determine the plastic limit, Journal of Geotechnical and Geoenvironmental Engineering 130 (2004) 630-635. J.L. Amorós, G. Mallol, B. Campos, M.J. Orts, M.C. Bordes, Study of the rheological behaviour of different ceramic powder materials, 10th World Congress on Ceramic Tile Quality, Castellón, Spain (2008). D.B. Price, J.S. Reed, Boundary conditions in electrical porcelain extrusion, American Ceramic Society Bulletin 6 (1983) 13481350. J.J. Benbow, E.W. Oxley, J. Bridgwater, The extrusion mechanics of paste - The influence of paste formations on the extrusion parameters, Chemical Engineering Science 42 (1987) 2151-2162. J.J. Benbow, T.A. Lawson, E.W. Oxley, J. Bridgwater, Prediction of paste extrusion pressure, American Ceramic Society Bulletin 68 (1989) 1821-1824. J. Zheng, W.B. Carlson, J.S. Reed, Flow mechanics on extrusion through a square entry die, Journal of the American Ceramic Society 75 (1993) 3011-3016. A.S. Burbidge, J. Bridgwater, The single screw extrusion of pastes, Chemical Engineering Science 50 (1995) 2531-2543. O. Hoffman, G. Sachs, Introduction to the Theory of Plasticity for Engineers (McGraw-Hill, New York, 1953). B. Avitzur, Analysis of metal extrusion, Transactions of ASME B87 (1965) 57-70. O.J.U. Flores, A.P.N. Oliveira, M.C. Fredel, H.A. Al-Qureshi, D. Hotza, Experimental analysis of cylindrical clay bodies subjected to compressive axial loading, Brazilian MRS Meeting 2006, Florianópolis, SC, Brazil (2006). O.J.U. Flores, A.P.N. Oliveira, M.C. Fredel, H.A. Al-Qureshi, D. Hotza, Mathematical model applied for evaluation of plasticity in clays with different moisture contents, 50th Brazilian Ceramics Conference, Blumenau, SC, Brazil (2006). Baran, B., Erturk, T., Sarikaya, Y., Alemdaloglu, T., 2001. Workability test method for metals applied to examine a workability measure (plastic limit) for clays. Applied Clay Science 20 (1-2), 53–63. F.A. Andrade, H.A. Al-Qureshi, D. Hotza, Theoretical analysis of rectangular clay paste extrusion, 11th International Conference on Advanced Materials, Rio de Janeiro, RJ, Brazil (2009) S.L. Correia, K.A.S. Curto, D. Hotza, A.M. Segadães, Clay from Southern Brazil: Physical, chemical and mineralogical characterization, Materials Science Forum 498-499 (2005) 447452. G.M. Reeves, I. Sims, J.C. Cripps (Ed.). Clay Materials Used in Construction (Geological Society, London, 2006). H.A. Al-Qureshi, M.R.F. Soares, D. Hotza, M.C. Alves, A.N. Klein, Analyses of the fundamental parameters of cold die compaction of powder metallurgy. Journal of Materials Processing Technology 199 (2008) 417-424.

Authors’ information Fernando A. Andrade is Bachelor and M.Sc. in Materials Engineering from the Federal University of Santa Catarina (UFSC). He is currently a Ph.D. student at the Graduate Program in Materials Science and Engineering (PGMAT/UFSC), Florianópolis, SC, Brazil. Hazim A. Al-Qureshi is Mechanical Engineer with a Ph.D. in the area of Metal Forming from the University of Birmingham, United Kingdom. He is currently Visiting Senior Professor at the Mobility Engineering Centre of the Federal University of Santa Catarina (UFSC), Joinville, SC, Brazil. Dachamir Hotza is Chemical Engineer and earned a Dr.-Ing. degree from the Technical University of Hamburg-Harburg (TUHH), Germany. He is currently Associate Professor at the Department of Chemical Engineering at Federal University of Santa Catarina (UFSC), Florianópolis, SC, Brazil.



483


Analysis of Irradiator Tube Purging for Fiber Drawing Showkat J. Chowdhury, Tyler Kirby Abstract – This paper presents the numerical and experimental investigation of fluid flow and mass transfer in an irradiator tube. After applying the synthetic coating material for cushioning and protection, optical fibers or wires are drawn through a glass tube where UV light from the surrounding irradiators cause curing of the coating material. During this curing process, fumes are generated, which needs to be flushed out of the tube before it diffuses to the glass wall and causes tube smoking, a problem faced by the industries. Analytical calculations are first done to study the effect of laminar and turbulent flow, flow rate and tube diameter. Numerical simulation for the purging air flow and mass transfer of the generated chemical species is then done using a computational model and results presented. Irradiator tube flows with regular and enhanced lip at exit and without lip are all simulated. The mass concentration of generated species near the tube wall is found high for regular and extended lip (causes tube smoking), but low for no lip (suggests no smoking and hence desirable). Computations are also done for larger tube diameter, which gives lowest species concentration. Experiments are performed and they support the simulated results. Copyright © 2010 Praise Worthy Prize S.r.l. - All rights reserved.

Keywords: Fiber Coating, Irradiator Tube, Turbulence, Mass Diffusion, Tube Smoking, Numerical Analysis

σε

Nomenclature a CA C1, C2, Cµ DAB dij k L N A

P Q r R Re Rij S cT tc td ui v* V x ε εm ν νT ρ σk

Ratio of tube length over tube radius Molar concentration of species A Coefficients in turbulence model Binary mass diffusion coefficient Mean deformation rate tensor Turbulent kinetic energy Length of the irradiator tube Generation of species A due to chemical reaction Mean pressure Flow rate of purging air Radial coordinate Radius of the irradiator tube Reynolds number Reynolds stress tensor Turbulent Schmidt number Convective time Diffusive time Mean velocity Friction velocity Average velocity in the tube Axial coordinate Dissipation rate of turbulence kinetic energy Turbulent mass diffusivity Kinematic viscosity Turbulent or eddy viscosity Constant mass density Turbulent Prandtl number corresponding to k

ΤO

Turbulent Prandtl number corresponding to ε Shear stress at the wall

I.

Introduction

Fume generation during curing of the coating material of optical fibers and wires is a common problem faced by the industries. Optical fibers for telecommunication have to apply two layers of synthetic coating of polymers, inner layer for cushioning and outer layer for protection of the fibers. In this modern technological world, huge amount of data is being transferred through optical fibers. As optical fibers are basically made of glass of very small diameter through which telecommunication data passes at tremendous speed in the form of photons or light, so the quality of the protective coating materials is very important. In the draw tower, the fibers or wires are drawn continuously through coating dies for applying the coating material, and then they are passed through transparent glass tubes surrounded by UV irradiators. The UV light from the irradiators causes curing of the coating materials while generating fumes of some chemical species. Air or Nitrogen gas flows through these transparent irradiator glass tubes to purge or flush out the fumes. But some of these generated species travel through mass diffusion across the tube, from the optical fiber or wire at the center to the wall of the glass tube. At the wall, due to the heat generated by the UV radiation, these species are



484

Showkat J. Chowdhury, Tyler Kirby

pyrolyzed (burnt) and the glass tube wall loses its transparency. This is known as tube smoking. Due to this tube smoking, sufficient UV light from the irradiators can not pass through the glass tube, and the curing of the coating material of the optical fiber or wire becomes incomplete. Consequently, the continuous drawing process has to be stopped frequently for changing the irradiator tubes. This in turn reduces production and increases cost, which is undesirable. The size of the irradiator glass tube used, laminar or turbulent flow of gases through the tube for purging the generated species, inlet and exit condition of the irradiator tube, all have different effects on the mass diffusion of the species across the tube to cause tube smoking. Hence, an in-depth knowledge of the properties is required to eliminate the tube smoking problem. In industrial product development, designers are aided by experiments, but as a supplement to them, economical design and operation can be greatly facilitated by the availability of prior properties of the flowfield through some computational model. Due to proprietary or confidential nature of the problem, much literature is not available on this topic for reviewing. It has been mentioned in the literature that, the standard k – ε turbulence model has some limitations in predicting turbulent recirculating flows [1]-[3]. Because the streamline curvature in the recirculating zones produces large changes in the higher order quantities of the turbulent structure, and makes it anisotropic. Numerous modifications using additional empirical constants are proposed [4], but none of them is found to work for almost all geometries. As an alternate, higher order turbulence models, like the Algebraic Stress Model (ASM) and the Reynolds Stress Model (RSM) are considered. But the improvement in predicting the flow field of recirculating flows using ASM are not found to be significant [4],[5]. Though the Reynolds Stress Model has been able to predict the major features of the recirculating flow [6],[7], but greatly increases the computational complexity and time requirement. Lookwood and Shen [8] found no consistent advantage of the RSM predictions over the standard k – ε model results. Despite the benefits of the higher order models, the standard k – ε model is still widely used in predicting complicated turbulent flows [9]-[11]. The reason is, flows having heat transfer, mass transfer and chemical reaction in addition to fluid flow are so complex that the shortcomings of the standard k – ε model may be overlooked compared to its advantages like simplicity and economy. The objective of the present study is to obtain numerical predictions for the laminar and turbulent flow and mass diffusion in the irradiator tube for various flow rates and tube diameters, and to analyze the effect of the exit condition, to minimize tube smoking and improve performance, and to facilitate continuous drawing process. As mentioned above, though sophisticated turbulence models have been developed, yet the k – ε

model is still widely used in the industry due to its simplicity and as it has already been accommodated into many commercially available computer codes. Hence, in the present study the k – ε model [12] is used for turbulence calculations.

II.

Problem Statement

The flow of gas through a typical irradiator tube is shown in Fig. 1. The optical fiber or wire being drawn is shown at the centerline. In a vertical draw tower, a thin controlled layer of synthetic coating is applied to the fiber or wire while passing through a coating die, just before entering this tube. The transparent glass tube has a radius R and a length L. The UV light of specific intensity from the irradiators around the transparent tube is used to cure the coating material as it passes through the tube. During this curing process fumes of some chemical species are given out from the coating material. Purging air or nitrogen gas enters the tube through the annular space at inlet and tries to flush these species through the outlet. Honeycomb flow straightners are used at inlet to straighten the gas flow and reduce vibration of the fiber or wire, to maintain a uniform thin layer of coating on the fiber. For holding the vertical irradiator glass tubes in the draw tower, extended metallic annular strip or lips are used, which might affect the flow of purging gas. Some of these species generated during the curing process of the coating material, travel across the tube by mass diffusion to the glass wall. Due to the heat of UV light from the irradiators, the species at the glass tube wall are burnt and sticks to the wall, and the glass tube loses its transparency, which becomes worse with time. This is called Tube Smoking. After a few hours, the transparency of the irradiator glass tubes becomes so low that the UV light coming from the irradiators through the glass tubes are not sufficient to complete the curing process of the coating material, and the optical fiber or wire being drawn has to be discarded. As a result, the drawing process has to be stopped frequently for changing the irradiator glass tubes, which means loss of production capability and increase in manufacturing cost. To solve this problem, irradiator tubes are to be designed which does not cause tube smoking even after extended hours of use and allows continuous draw of optical fiber or wire in the draw tower, assuring product quality.

III. Simplified Modeling and Analysis For understanding the irradiator tube smoking, as preliminary or first round analysis of the problem, the convective time required to purge or flush the generated species out of the tube and the diffusive time required for mass diffusion of the species to the tube wall are calculated.



485


Here, νT is the turbulent or eddy viscosity and ScT is the turbulent Schmidt number. An irradiator glass tube of diameter 0.022 m and length 1.2 m is first considered. The flow of gas through the tube for Reynolds number Re ≥ 2300 is assumed turbulent. As the properties of the materials being used at the industries are proprietary or confidential and can not be published, so the binary mass diffusion coefficient for preliminary analysis is considered as DAB = 0.28×10-4 m2/s, corresponding to ammonia and air. The convective and diffusive time for various flow rates having laminar flow are calculated using Equs. (1) and (4) above, and their ratio tc/td plotted in Fig. 2 for different flow rates. For calculating the Reynolds number:

UV Radiation Glass Tube

Lip Calculation Domain

Air Inlet

R Outlet

r

Honeycomb

Fiber

L

x

Fig. 1. Geometry of an irradiator tube

Nitrogen or air at a flow rate Q enters the irradiator tube at inlet, to flush the species generated from the coating materials of the fiber or wire during the curing process, out of the tube before it diffuses to the tube wall. The convective time tc required for this, which is actually the average residence time is calculated as:

tc =

L aR = V V

Re =

1 8

τ o = λρV 2

where L is the length of the irradiator tube, R is the radius of the tube, a is the ratio of the tube length over tube radius, and V is the average velocity in the tube found as: Q A

(2)

a=

L R

(3)

R2 (laminar) DAB

(4)

td =

R2 R2 ≈ (turbulent) DAB + ε m ε m

(5)

λ=

νT SCT

(8)

0.3164 Re0.25

(9)

Hence, the friction velocity v* becomes: v* =

τo λ = V ρ 8

(10)

For estimating the eddy viscosity, from Schlichting [13] it is found:

νT v* R

= 0.08

(11)

From the data of Groenhof [14] and Notter and Sleicher [15] the turbulent Schmidt number is taken as: ScT = 0.8

where, R is the radius of the tube, DAB is the binary mass diffusion coefficient or mass diffusivity, and εm is the turbulent or eddy mass diffusivity. The turbulent mass diffusivity can be written as:

εm =

(7)

For frictional resistance of smooth pipes, Blasius gave [13]:

Here, A is the cross-sectional area of the tube. The diffusive time td required by the generated species to travel through mass diffusion from the center of the irradiator tube to the glass wall for laminar or turbulent flow is calculated as: td =

ν

the kinematic viscosity was assumed as ν = 15.86×10-6 m2/s. Shear stress at the wall for turbulent flow through pipes can be calculated as [13]:

(1)

V=

VD

(12)

The convective and diffusive time for various flow rates having turbulent flow are calculated using Eqs. (1), (2) and (5) – (12), and their ratio tc/td plotted in Fig. 2 for different flow rates. The above calculations are also repeated for larger tube diameters 0.033 m and 0.044 m, but having same tube length of 1.2 m, at different flow rates and the results plotted in Fig. 2.

(6)



486


For turbulent flow through a particular tube diameter, as the flow rate increases the velocity also increases, and from Equ. (1) convective time tc reduces. But with the increase of velocity, turbulent or eddy mass diffusivity εm also increases to some extent, which causes a decrease of diffusive time td, from Equ. (5). Consequently, for turbulent flow through a particular tube diameter, with the increase of flow rate the ratio tc/td does not decrease appreciably. This means by increasing the flow rate alone, might not remove the tube smoking problem. On the other hand, excessive increase of flow rate might cause fiber vibration, which again will cause nonuniformity in the coating thickness and fiber defect. For a particular turbulent flow rate, when the tube diameter increases, the velocity and eddy mass diffusivity reduces. From Equs. (1) and (5), convective and diffusive time tc and td will both increase, but td will increase more than tc. As a result, the convective / diffusive time ratio tc/td will decrease appreciably. For example, at a flow rate Q = 4 cfm = 1.888×10-3 m3/s, for tube diameters D = 22mm, 33mm and 44mm, the tc/td ratios are 0.719, 0.504 and 0.392, respectively. This means that increasing the tube diameter from 22mm to 44mm will help in eliminating the tube smoking problem.

D = 22 mm D = 33 mm D = 44 mm

Conv. Time / Diff. time., tc / td

1.6

1.2

0.8

0.4

0.0 0.0

2.0

4.0 3

6.0 3

Flow rate, Q (m /s X 10 ) Fig. 2. Comparison of convective / diffusive time profiles for laminar and turbulent flow

For laminar flow from Eqs. (1) and (4) it can be shown that: tc 2a DAB = td Re ν

(13)

IV.

Detailed Computational Simulation

In order to have a better understanding of the tube smoking problem, the fluid flow and mass diffusion in the irradiator tube is simulated using a CFD tool as described below.

This means for a particular tube radius having laminar flow the convective to diffusive time ratio varies inversely with the Reynolds number. In Fig. 2, the horizontal axis shows the variation of flow rate, and the vertical axis shows the variation of the convective to diffusive time ratio, tc/td for three different irradiator tube diameters. For laminar flow, the curves for all three diameters merge into a single one, and decreases rapidly with the increase of flow rate, until reaches transition to turbulent flow. When the flow becomes turbulent, the convective to diffusive time ratio, tc/td suddenly jumps. A low value of tc/td means the convection time required for the fume to travel along the tube length and flush out is smaller than the diffusion time required for traveling across the flow by mass diffusion to the tube wall. Hence, a low value of tc/td is desirable to eliminate the tube smoking problem. For turbulent flow, the eddy mass diffusivity εm is much larger compared to the binary mass diffusion coefficient DAB, and as can be seen from Eq. (4) the diffusive time required td becomes much smaller. Though the convective time required tc also decreases with increase of velocity, but not at the same rate as td, hence the ratio tc/td increases as the flow transition from laminar to turbulent. The flow rate of purging gas corresponding to laminar flow might not be sufficient to remove all the fumes generated, and so in most cases the flow has to be turbulent.

IV.1. Basic Equations and Numerical Details The equations governing the mean turbulent motion of an incompressible fluid using the k – ε model closure of Launder and Spalding [12] are given as: ∂ui =0 ∂xi uj

uj

∂ui ∂ 2ui 1 ∂P ∂ =− +ν − Rij ∂x j ∂x j ∂x j ∂x j ρ ∂xi

∂k ∂ = ∂x j ∂x j

uj

⎡⎛ νT ⎢⎜⎜ν + k σ ⎣⎢⎝

∂ε ∂ = ∂x j ∂x j

⎞ ∂k ⎤ ∂ui −ε ⎥ − Rij ⎟⎟ ∂ ∂ x xj ⎠ j ⎦⎥

⎡⎛ ν T ⎞ ∂ε ⎤ ⎢⎜⎜ν + ε ⎟⎟ ⎥+ σ ⎠ ∂x j ⎦⎥ ⎣⎢⎝

∂u ε2 − C1 Rij i − C2 ∂x j k k

ε


(14)

(15)

(16)

(17)


487


the outlet, an annular metallic strip or Lip for holding the tube with a radial thickness of 0.003 m is also considered to study its effect on the flow and mass diffusion. A staggered mesh finite volume method is used to discretize the differential equations written as Equs. (14) – (22). The computational domain is divided by a 110 X 38 non-uniform grid with finer spacing in the regions of large spatial gradients. The above differential equations are integrated over their appropriate staggered control volumes and discretized using a hybrid differencing scheme [16]. In order to avoid the need of extremely fine mesh and detailed calculations near the wall, logarithmic wall functions with no slip boundary conditions are applied at solid walls [12]. The turbulence kinetic energy k and the dissipation rate ε at inlet are calculated as:

Here, the flow is assumed to be steady. Equations (14) – (17) are the statements of conservation of mass, balance of linear momentum, conservation of turbulence fluctuation energy, and transport equation for turbulence dissipation rate. In these equations ρ is the mass density, ui is the mean velocity, ν is the kinematic viscosity, ν T is the turbulent or eddy viscosity, P is the mean pressure, ε is the dissipation rate, k is the fluctuation kinetic energy, σk and σε are the turbulent Prandtl numbers corresponding to k and ε. In all these equations regular tensor notation with Latin subscripts is employed. The Reynolds stress tensor is given as: Rij =

2 kδ ij − 2ν T dij 3

(18)

The mean deformation rate tensor is: dij =

1 ⎛ ∂ui ∂u j + ⎜ 2 ⎜⎝ ∂x j ∂xi

⎞ ⎟ ⎟ ⎠

kin = λ1 ⋅ uin2 ; ε in =

(19)

k2

ε

(20)

where Cµ is a constant. The turbulence model constants are assigned the following values [12]: Cµ = 0.09, C1 = 1.44, C2 = 1.92

σ k = 1.0, σ ε = 1.3

(21)

The conservation of chemical species or transport equation for species concentration, for the steady mean turbulent flow can be written as: uj

∂C A ∂ = ∂x j ∂x j

⎡ ∂C ⎤ ⎢( DAB + ε m ) A ⎥ + N A ∂x j ⎥⎦ ⎢⎣

(23)

where uin is the axial velocity of air at inlet, λ1 and λ2 are two constants having the values of 0.015 and 0.005, respectively. As the properties of the materials being used at the industries are proprietary or confidential and can not be published, so the binary mass diffusion coefficient is considered as DAB = 0.28×10-4 m2/s, corresponding to ammonia and air. The turbulent Schmidt number is taken as, ScT = 0.8. At the outlet boundary zero gradient conditions are applied. The discretized equations with boundary condition modifications are solved using the SIMPLE [17] and TDMA algorithm. Here, semi-implicit line-by-line relaxation method is employed to obtain converged solutions iteratively. Under relaxation factors of 0.5 is used for the linear momentum equations, while 0.7 is used for the k, ε, effective viscosity and species concentration equations to promote computational stability.

and the eddy viscosity is given as:

ν T = Cµ

kin3 / 2 λ2 R

IV.2. Results and Discussions

(22)

The turbulent flow of the purging air and the mass transfer of the species generated during curing of the coating materials in the irradiator tube, as shown in Fig. 1, are analyzed numerically using the above computational code and the results are presented below. In order to validate the capability of the model, the above computational model was also previously used to simulate the turbulent flow in a combustor for the experimental condition of by Brum and Samuelsen [18], and the results were published by Chowdhury and Kirby [19]. The simulated results were in good agreement with the experimental data, which shows that the present numerical model has the capability of predicting complex turbulent flows with reasonable accuracy.

Here, CA is the molar concentration of species A, DAB is the binary mass diffusion coefficient, εm is the turbulent or eddy mass diffusivity as given by Equ. (6), and N A is the generation of species A due to chemical reaction. The computational domain for the irradiator tube flow is shown in Fig. 1. A jet of air with a flow rate, Q = 4 cfm = 1.888×10-3 m3/s enters the tube through the annular space close to the wall at inlet. The optical fiber or wire with the coating material is drawn through the centerline. The function of the air is to purge or flush the species generated from the fiber coating material during curing before it diffuses to the glass wall. The irradiator glass tube has a length L = 1.2 m, and two radius of R = 0.022 m and 0.044 m, are considered for comparison. At



488


of the flow near the wall. With time the concentration of these burnt sticky substance near the downstream end of the tube increases and they extend upstream. After a few hours of operation, the entire glass tube becomes dark, and the transparency becomes so low that sufficient UV light from the irradiators can not enter for curing the coating material any more. As mentioned above, this problem is known as tube smoking. The fiber or wire drawing process has to be stopped for changing the tube, which means a loss of production and increase of manufacturing cost, and is undesirable.

x/L = 0.1 x/L = 0.5 x/L = 0.9

Mass Conc., CA (kmol/m3 X106)

1.6

1.2

0.8

x/L = 0.1 x/L = 0.5 x/L = 0.9

0.4

0.2

0.4

0.6

0.8


0.0 0.0

1.6

1.0

Radius, r/R Fig. 3. Comparison of mass concentration profiles for 22mm diameter tube and with Normal Lip condition

The computational model is first used to analyze the fluid flow and mass transfer of Fig. 1, having a metallic lip for holding the glass tube in vertical position in the draw tower, which is normally done in the industry. The tube diameter (D = 2R) is taken as 0.022 m, the length (L) as 1.2 m, and the lip is extended 0.003 m inside the tube. The annular jet of air at inlet is assumed to enter through the flow straightners with a uniform inlet axial velocity and a flow rate, Q = 4 cfm = 1.888×10-3 m3/s. The corresponding Reynolds number in the tube is Re = 6890, which means the flow is turbulent. The molar mass concentration of the species, CA generated during curing of the fiber coating material at the tube centerline is computed and presented in Fig. 3. In this figure, the radial distribution of the mass concentration at three different axial locations, x/L = 0.1, 0.5, and 0.9 are shown. Here, x is the axial distance from the inlet of the tube and R is the tube radius. It is found that, at x/L = 0.1, close to the inlet the mass concentration, CA near the tube wall is very low. At x/L = 0.5 and 0.9, the mass concentration near the tube center increases significantly compared to that at x/L = 0.1. The mass concentration from the tube center to the wall decreases sharply, but it is still quite large near the wall at larger x/L. As the species generated during curing of the coating materials are being flushed downstream, their concentration would increase along the centerline. But high values of concentration near the wall indicates that a large amount of species travel by mass diffusion across the flow to the tube wall before they are being flushed out. Due to the heat of the UV lamps, some of the species attached to the wall near x/L = 0.9, where the concentration is high enough will pyrolyze (burn) and produce some dark gummy substance which stick to the wall. These sticky substances change the characteristics

1.2

0.8

0.4

0.0 0.0

0.2

0.4

0.6

0.8

1.0

Radius, r/R Fig. 4. Comparison of mass concentration profiles for 22mm diameter tube and with Larger Lip condition

To study the effect of the metallic lip further, a larger metallic lip extending 0.005 m inside the tube is considered. The tube diameter (D) is taken as 0.022 m, length (L) as 1.2 m, and the flow rate of purging air, Q = 4 cfm = 1.888×10-3 m3/s, at a Reynolds number of Re = 6890, for the above flow condition. The computational model is used to simulate this flow and the predicted molar mass concentration of the species generated during curing of the coating material inside the irradiator tube is presented in Fig. 4, for three different axial locations. At x/L = 0.1, close to the inlet of the tube, it is seen that, the mass concentration CA near the tube wall is low, but more than that in Fig. 3. At further downstream, x/L = 0.5 and 0.9, the mass concentration near the tube center has increased drastically compared to that at x/L = 0.1, which means more accumulation of the species generated as it moves downstream. At x/L = 0.5 and 0.9, though the mass concentration near the wall is lower than that near the tube center, but they do not seem to decrease exponentially from the center to the wall, and also they are significantly higher compared to those in Fig. 3. This means that the species generated during curing of the coating materials are not flushed out adequately and a



489


large amount travel by mass diffusion to the tube wall before being flushed out. Due to the heat of the UV lamps, some of the species attached to the wall will pyrolyze or burn and the dark gummy substances produced will stick to the tube wall. As mentioned above, this will reduce the transparency of the glass tube and adversely affect the curing process, which is undesirable. As the concentration of the species near the wall are significantly higher compared to Fig. 3, so the chances of getting the problem of tube smoking at an even shorter time, is higher.


1.6

x/L = 0.1 x/L = 0.5 x/L = 0.9


1.6

0.8

0.4

0.0 0.0

0.2

0.4

0.6

0.8

1.0

Radius, r/R Fig. 6. Comparison of mass concentration profiles (22mm dia tube): solid line for No Lip, dashed line for Normal Lip condition, and dash dot line for Extended Lip condition

1.2

For comparing the above three cases, having normal lip, larger lip, and without lip, the radial distribution of mass concentration at different axial locations, x/L = 0.1, 0.5 and 0.9 are again plotted in Fig. 6. Here the solid lines correspond to no lip condition, while the dashed lines correspond to the usual condition of having lips (regular lip), and the dash-dot lines correspond to the larger lip condition. From the figure it is found that, the presence of the regular lip causes the mass concentration near the glass tube wall to rise considerably and start tube smoking. The presence of the lip causes an adverse pressure gradient in the irradiator tube and a recirculation close to the exit. The axial velocity of the purging air near the wall becomes small, which increases the convective time tc. Due to recirculation, the turbulence kinetic energy k and eddy mass diffusivity εm increases. This reduces the diffusive time td. As a result, the ratio tc/td becomes larger. From Fig. 2, it can be seen that increase of ratio tc/td means that the chances of the mass to diffuse to the wall before it is flushed out are higher. With the larger lip, the adverse pressure gradient in the irradiator tube and recirculation close to the exit is found to be even more. The axial velocity of the purging air near the wall becomes smaller, which increases the convective time tc. Also, due to larger recirculation, the turbulence kinetic energy k and eddy mass diffusivity εm increases further, and makes the diffusive time td smaller. As a result, the ratio tc/td becomes even larger, which means that the chances of the mass to diffuse to the wall before it is flushed out are worse, compared to the regular lip condition. Hence, to lower the mass concentration of generated species near the glass tube wall and eliminate the tube smoking problem, the lip should be avoided.

0.8

0.4

0.0 0.0

1.2

0.2

0.4

0.6

0.8

1.0

Radius, r/R Fig. 5. Comparison of mass concentration profiles for 22mm diameter tube and No Lip condition

For finding a possible solution to the tube smoking problem, the computational model is then used to analyze the flow of Fig. 1, but without the lip. As above, the tube diameter (D) is taken as 0.022 m and the length (L) as 1.2 m. The flow rate of purging air, Q = 4 cfm = 1.888×10-3 m3/s, at a Reynolds number of Re = 6890. The molar mass concentration of the species generated during curing of the coating material inside the irradiator tube is again calculated at different axial locations and are shown in Fig. 5. It is found that, at x/L = 0.1, close to the inlet the mass concentration, CA near the tube wall is almost zero. At x/L = 0.5 and 0.9, the mass concentration near the tube center increases significantly compared to that at x/L = 0.1, but near the wall it is still low. The mass concentration from the tube center to the wall decreases exponentially. This means that the species generated during curing of the coating materials are being flushed downstream before they can diffuse to the tube wall, which is desirable. Though the mass concentration near the wall at downstream (x/L = 0.9) is non-zero, but it is quite low compared to Fig. 3, and might not be enough to cause tube smoking.



490


1.6


x/L = 0.1 x/L = 0.5 x/L = 0.9


1.6

1.2

0.8

0.4

1.2

0.8

0.4

0.0 0.0

0.2

0.4

0.6

0.8

1.0

Radius, r/R 0.0 0.0

0.2

0.4

0.6

0.8

Fig. 8. Comparison of mass concentration profiles (No Lip condition): solid line for 22mm dia tube and dash dot line for 44mm dia tube

1.0

Radius, r/R Fig. 7. Comparison of mass concentration profiles for 44mm diameter tube and No Lip condition

V.

Experimental Verification

To verify the above numerical findings, some experiments are done. Optical fiber is drawn vertically in a draw tower through a coating die and a glass irradiator tube as shown in Fig. 1. The diameter of the tube (D) is taken as 0.022 m, length (L) is 1.2 m, flow rate Q = 4 cfm = 1.888×10-3 m3/s (Reynolds number Re = 6890), and the usual or regular lip for holding the tube has a thickness of 0.003 m. The UV light from the surrounding irradiator caused curing of the coating material and generated fumes. The experiment is carried out for eight hours. The intensity of light passing through the irradiator glass tube at the end of every two hours is measured and normalized with the intensity of light passing through a clean glass tube. The results are plotted in Fig. 9. It is found that tube smoking occurred, and got worse with time. Consequently, the intensity of light passing through the glass tube dropped, and at the end of eight hours the normalized intensity became only about 52%. Hence, the fiber drawing process has to be stopped after every few hours for changing the glass tube. The experimental result supports the computational simulation results presented in Fig. 3. As already mentioned, the presence of the lip causes an adverse pressure gradient in the tube and a recirculation near the exit. This in turn increases the species mass diffusion to the wall and causes tube smoking. The above experiment is then repeated with a larger lip of 0.005 m thickness, while having the diameter of the tube (D) as 0.022 m, length (L) as 1.2 m, flow rate Q = 4 cfm = 1.888×10-3 m3/s at Reynolds number Re = 6890. Optical fiber is drawn vertically in a draw tower through a coating die and the glass irradiator tube as shown in Fig. 1. The UV light from the irradiator caused curing of the coating material and generated fumes. The experiment is carried out for eight hours, and normalized

From the preliminary calculations in Fig. 2, it is found that increasing the tube diameter at the same flow rate would reduce the convective/diffusive time ratio tc/td, which means a reduction in chances that the mass will diffuse to the wall before it is purged out. So to further eliminate the tube smoking problem, using the above model computations are repeated for the flow of Fig. 1, but without the lip, and a larger irradiator tube diameter (D) of 0.044 m. The tube length (L) is taken as 1.2 m and the flow rate of purging air is maintained at Q = 4 cfm = 1.888×10-3 m3/s, at a Reynolds number of Re = 3445. The molar mass concentration of the coating fume in the irradiator tube is calculated and presented in Fig. 7. The plot shows the mass concentration profiles in the radial direction at three axial locations. Similar to Fig. 5, as the flow proceeds downstream, the mass concentration near the centerline increases as the generated species are washed out. But the mass concentration decreases exponentially from centerline to the tube wall, even faster than that in Fig. 5. This means that the probability of the species being washed out of the irradiator tube before they can diffuse to the wall is even higher. Fig. 8 shows the mass concentration profiles at axial locations x/L = 0.1, 0.5 and 0.9, for the above two no lip conditions with same air flow rate, but having tube diameters (D) of 0.022 m and 0.044 m. The solid lines correspond to tube diameter of 0.022 m, while the dashdot lines represent the larger tube diameter of 0.044 m. From the figure it is observed that, for D = 0.044 m the radial distribution of the mass concentration decreases faster than that for D = 0.022 m, similar to the findings of Fig. 2. Hence, to eliminate the tube smoking problem completely, 0.044 m diameter tubes without lip may be considered having the same flow rate of purging air.



491


reversal and smoke is not found to come out through the space near the fiber inlet. Increasing the purging air flow rate to 6 cfm = 2.832×10-3 m3/s increases fiber vibration, this causes fiber coating defect, and so is discarded.

intensity of light passing through glass tube is plotted in Fig. 9. It is found that tube smoking occurred in a short time compared to the previous case and got worse with time. The experimental result justifies the computational simulation results presented in Fig. 4. As mentioned earlier, due to the larger lip the adverse pressure gradient in the tube and the recirculation near the exit will be bigger. This in turn will further increase the species mass diffusion to the wall and cause faster tube smoking.

VI.

Based on the computational results and experimental verification, it may be concluded that, the present computational model is capable of predicting the fluid flow and mass transfer in the irradiator tube with reasonable accuracy. The convective time / diffusive time ratio, tc/td for laminar flow of purging air in an irradiator tube decreases exponentially with increase in flow rate. Low values of tc/td ratio means, the species generated during curing of the coating material are more likely to be washed out from the tube before they can diffuse to the glass wall and cause tube smoking. For turbulent flow, increasing air flow rate for the same tube diameter does not lower the tc/td ratio appreciably. Presence of lip at the tube exit increases the mass diffusion across the irradiator tube, which increases the concentration of the generated species close to the glass wall and initiates tube smoking. Removing the lip reduces mass diffusion across the tube, and lowers the concentration of the species close to the glass tube to such extent that prevents tube smoking, and allows continuous draw of fiber. Increasing tube diameter for turbulent flow of air at constant flow rate causes large reduction in the tc/td ratio, reduces the concentration of generated species close to the glass wall, and helps further in the elimination of tube smoking problem.

Normalized UV Intensity (%)

120

100

80

60

40 0.0

2.0

4.0

6.0

Conclusion

8.0

Draw Time (hrs) Fig. 9. Normalized Intensity of UV light entering for fiber curing: solid line for Regular Lip, dash dot line for Extended Lip, and dashed line for No Lip condition

The above experiment is again repeated, for same tube diameter of 0.022 m, length of 1.2 m, and purging air flow rate Q = 4 cfm = 1.888×10-3 m3/s, but without the lip. The experiment is continued for eight hours. It is observed that there is no tube smoking, and the normalized intensity of the UV light from the irradiator passing through the glass tube for curing the optical fiber coating material is about 100%, even after eight hours of drawing in the draw tower. The computational results obtained by the numerical model and presented in Fig. 5, explain this experimental observation. In the absence of the lip, there is no recirculation or adverse pressure gradient in the irradiator tube. Consequently, the species mass diffusion to the tube wall is less and is not enough to initiate tube smoking. To perform flow visualization and understand the problem further, smoke is mixed with the incoming purging air at a flow rate Q = 4 cfm = 1.888×10-3 m3/s, and the above conditions are studied. The same tube diameter of 0.022 m, and length of 1.2 m is taken and the flows with regular lip, larger lip and without lip are observed. It is found that, for larger lip of 0.005 m thickness, there is flow reversal and some smoke came out through the open space near the fiber inlet due to large adverse pressure gradient developed, which also supports the above computational results. But for regular lip of 0.003 m thickness and for no lip condition, the adverse pressure gradient is not sufficient to cause flow

References [1] [2] [3]

[4]

[5] [6] [7] [8]


M. Nallasamy, Turbulence Models and their Applications to the Prediction of Internal Flows: A Review, Computers Fluids 15 (2) (1987) 151–194. P. Koutmas, J.J. McGuirk, Isothermal Modeling of Gas Turbine Combustors: Computational Study, J. Propulsion Power, 7 (1991) 1064–1091. S.C. Favoloro, A.S. Nejad, S.A. Ahmed, Experimental and Computational Investigation of Isothermal Swirling Flow in an Axisymmetric Dump Combustor, J. Propulsion Power 7 (1991) 348–356. K.C. Chang, C.S. Chen, Development of a Hybrid k–ε Turbulence Model for Swirling Recirculating Flows Under Moderate to Strong Swirl Intensities, Internat. J. Numer. Methods Fluids, 16 (1993) 421–443. G.J. Sturgess, S.A. Syed, Calculation of Confined Swirling Flows, Internat. J. Turbo Jet Engines 7 (1990) 103–121. W.P. Jones, A. Pascau, Calculation of Confined Swirling Flows with a Second Moment Closure, J. Fluids Engrg., Trans. ASME 111 (1989) 248–255. M. Nikjooy, H.C. Mongia, A Second Order Modeling Study of Confined Swirling Flow, Internat. J. Heat Fluid Flow 12 (1991) 12–19. F.C. Lockwood, B. Shen, Performance Predictions of PulverisedCoal Flames of Power Station Furnaces and Cement Kiln Types, Twenty Fifth (Int.) Symp. On Combustion, The Combustion Inst., Pittsburgh, (1994) 503–509.


492


[9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19]

W. Yang, J. Zhang, Simulation of Swirling Turbulent Combustion in the TECFLAM Combustor, Computers and Chemical Engineering, 32(10) (2008) 2280-2289. P.J. Foster, J.M. Macinnes, F. Schubnell, Isothermal Modelling of a Combustion System with Swirl: A Computational Study, Combust. Sci. Tech., 155 (2000) 51–74. J.L. Xia, B.L. Smith, A.C. Benim, J. Schmidli, G. Yadigaroglu, Effect of Inlet and Outlet Boundary Conditions on Swirling Flows, Comput. Fluids 26 (8) (1997) 811–823. B.E. Launder, D.B. Spalding, The Numerical Computation of Turbulent Flows, Computer Methods in Applied Mechanics and Engineering 3 (1974) 269-289. H.T. Schlichting, K. Gersten, Boundary Layer Theory (SpringerVerlag Berlin, 8Rev Edition, 1999). H.C. Groenhof, Eddy Diffusion in the Central Region of Turbulent Flows in Pipes and Between Parallel Plates, Chemical Engineering Science 25 (1970) 1005-1014. R.H. Notter, C.A. Sleicher, The Eddy Diffusivity in the Turbulent Boundary Layer Near a Wall, Chemical Engineering Science 26 (1971) 161-171. S.V. Patankar, Numerical Heat and Fluid Flow (McGraw-Hill, New York, 1980). A.D. Gosman, W.M. Pun, Calculation of Recirculating Flows, Rept. No. HTS/74/12, Dept. of Mechanical Engineering, Imperial College, London, 1974. R.D. Brum, G.S. Samuelsen, Aerothermal Modeling Program Phase I Final Report, NASA CR-168243, NASA-Lewis Research Center, Cleveland, Ohio, 1983. S.J. Chowdhury, T. Kirby, Numerical Analysis of Swirling Turbulent Flow in a Model Combustor, Intern. Rev. Chemical Engineering, 2(3) (2010) (In press).

Authors’ information Dr. Showkat Jahan Chowdhury is a Professor in the Department of Mechanical Engineering at Alabama A&M University, Huntsville, Alabama, U.S.A. His Mailing Address is: 2511 Clifton Drive, Huntsville, AL 35803, U.S.A. Tel: (256)-372-8401, Fax: (256)-372-5888 E-mails: [email protected] [email protected] Mr. Tyler Kirby is a student in the Department of Mechanical Engineering at Alabama A&M University, Huntsville, Alabama, U.S.A.



493


Removal of Copper (II) from Aqueous Solutions with Activated Carbon Obtained by Chemical Activation of Orange Peel Liliana Giraldo1, Juan Carlos Moreno-Piraján2* Abstract – Activated carbons (ACs) were prepared by pyrolysis of orange peel in the presence of zinc chloride (ZnCl2) (chemical activation). Orange peel from the Colombian orange cultivar was impregnated with aqueous solutions of ZnCl2 following a variation of the incipient wetness method. Different concentrations were used to produce impregnation ratios of 40, 70, 110 and 160 wt%. Activation was carried out under an argon flow at a temperature of 823 K with a 4 h soaking time. The porous texture of the obtained ACs was characterized by physical adsorptions of N2 at 77 K and CO2 at 273 K. The impregnation ratio had a strong influence on the pore structure of these ACs, which could be easily controlled by simply varying the proportion of ZnCl2 used in the activation. Thus, a low impregnation ratio led to essentially microporous ACs. At intermediate impregnation ratios, ACs with a wider pore size distribution (from micropores to mesopores) were obtained. Finally, high impregnation ratios yielded essentially mesoporous carbons with a large surface area and pore volume. The four best-fit three-parameter isotherms Sips, Toth, Prausnitz-Radke and Vieth-Sladek suggest that the adsorption capacity of activated carbons from orange peel for copper ions was 63 mg/g. Copyright © 2010 Praise Worthy Prize S.r.l. - All rights reserved. Keywords: Orange Peel, Adsorption, Surface Area, Activated Carbon, Langmuir, PrausnitzRadlke

I.

Chemical activation is known as a single-step method of the preparation of activated carbon in the presence of chemical agents. Physical activation involves carbonization of carbonaceous materials followed by activation of the resulting char in the presence of activating agents such as CO2 or steam. The chemical activation method usually takes place at a temperature lower than that used in physical activation, therefore it can improve pore development in the carbon structure because of the effect of the chemicals. The carbon yields of chemical activation are higher than physical activation yields [8-37]. The potato is the most important foods in Colombia, an agricultural country, and is usually used for traditional foods and cakes, amongst other uses. Various agricultural products are widely used as the basis of Colombian food, such as cassava, maize, onion, rice, sugar, potato and especially the orange. Orange production in Colombia is very large and only small amounts are utilized by traditional food industries; the rest is used as a raw material for cassava starch industries. Orange starch making operations produce a large amount of solid waste (orange peel) and direct discharge of this solid waste will cause environmental problems. Here we report on the porous texture characteristics of ACs prepared by ZnCl2 activation of orange peel, i.e. the shells covering the orange. These constitute a by-product from orange processing following harvest with few practical applications and

Introduction

At present, adsorption is widely accepted in environmental treatment applications throughout the world. Liquid-solid adsorption systems are based on the ability of certain solids to preferentially concentrate specific substances from solutions onto their surfaces. This principle can be used for the removal of pollutants, such as metal ions and organics, from wastewaters [1-4]. Extensive research has been carried out during the last ten years to find low-cost, high-capacity adsorbents for the removal of metal ions. A wide range of adsorbents have been developed and tested, including several activated carbons [5-8]. A number of low-cost agricultural wastes: mud, tyre rubber and fly ash, have been used for the removal of a range of metal ions. Activated carbons are materials with complex porous structures and associated energetic and chemical inhomogeneities. Their structural heterogeneity is a result of the existence of micropores, mesopores and macropores of different sizes and shapes. Activated carbon is one of the most important adsorbents from an industrial point of view. The main application of this adsorbent is for the separation and purification of gaseous and liquid phase mixtures [1-7]. There are two processes for the preparation of activated carbon: chemical activation and physical activation.



494

Liliana Giraldo, Juan Carlos Moreno-Piraján

whose uncontrolled spill (e.g. into rivers) causes environmental concerns. In fact, the applications of orange peel are limited to use as a fuel or as feedstuff for animals. Solid and liquid wastes are generated in large amounts in the wastewater of several industries such as metal cleaning and plating baths, refineries, paper and pulp, fertilizers and wood preservatives, which are dumped or released into the water causing detrimental effects not only to humans but also upon the environment, therefore it has become imperative to develop methods for treating such wastes. Studies have proven that heavy metals such as lead, zinc, cadmium, chromium and copper are very toxic elements [8, 9]. The excessive intake of copper by humans leads to severe headaches, hair loss, hypoglycaemia, increased heart rate, nausea and damage to the kidney and liver. It may also cause psychological problems, such as brain dysfunction, depression and schizophrenia [9, 16-37]. There are various methods for removing heavy metals, including chemical precipitation, membrane filtration, ion exchange, liquid extraction and electrodialysis [8, 9]. However, these methods are not widely used due to their high cost and low feasibility for small-scale industries [9-13]. In contrast, the adsorption technique is one of the preferred methods for the removal of heavy metals because of its efficiency and low cost. Conventional adsorbents such as granular or powdered activated carbon are not always popular as they are not economically viable or technically efficient [8]. Nonconventional materials have been tested on a large scale for this purpose, such as fly ash [10], lignite [11] and tree fern [12], for example. In this paper, orange peel was used as an adsorbent to remove Cu2+ from an aqueous solution. Orange peel is often considered as a solid waste of agriculture and it is widely available [18, 20]. In fact, this material causes a significant disposal problem. Efforts have been made to use the cheapest and most unconventional adsorbents to adsorb heavy metals such as Cu(II) from aqueous solutions. In this study we synthesized activated carbons with large surface areas and pore volumes obtained from orange peel, and showed that development of the porous structure can be modulated by changing the relative proportions of feedstock and ZnCl2.

II.

inorganic impurities, and then oven dried for 24 h at 393 K to reduce the moisture content. Peel from Colombian oranges was impregnated with aqueous solutions of chloride zinc following a variation of the incipient wetness method; similar cases have been described in literature [38-42]. This consists of adding the amount of aqueous solution (2.0 ml g−1 orange peel), drop by drop (while stirring the solid to facilitate homogeneous absorption of liquid), necessary to produce swelling until incipient wetness is reached. Different concentrations of ZnCl2 in aqueous solution were used to vary the content of the impregnation agent, expressed as the impregnation ratio (Xp, wt%), defined as (g ZnCl2 per g orange peel)×100. Impregnation ratios of 40, 70, 110 and 160 wt% were used. After impregnation, the samples were dried for 8 h at 383 K in air. Pyrolysis treatments (chemical activation) were carried out in a vertical tubular reactor made of quartz in furnace (Carbolite™), (Fig. 1), using 25 g of impregnated and dried material in all cases.

Fig. 1. Vertical furnace for obtaining activated carbon from orange peel

All treatments were performed at a constant heating rate of 10 K min−1 and with an argon (99.999% pure) flow of 30 STP cm3 min−1, which was continued during heating and cooling. An activation temperature of 823 K and a soaking time of 4 h were used. After cooling the solid pyrolysis residue to room temperature, it was washed with milli-Q distilled water until conductivity of the washing liquids was lowered to