MODELLING OF DISPLACEMENT WASHING OF PULP FIBERS USING THE HERMITE COLLOCATION METHOD

Abstract

A non-linear mathematical model for displacement washing of pulp fibers with suitable boundary conditions is presented. Model equations are divided into two phases: particle phase and external fluid phase. The Hermite collocation method (HCM) is employed to solve the non-linear model equations. The validity and efficiency of the proposed model is demonstrated by comparing the numerical results with the experimental ones. Stability of the proposed method is checked by evaluating L2 and L? norms at different time intervals. Industrial parameters such as displacement ratio and wash yield have been calculated to check the validity of the model. Numerical values obtained from the model equations are presented in terms of 2D graphs.

Keywords
Cubic Hermite polynomials; Peclet number; Biot number; Wash yield; Displacement ratio; Bed porosity

INTRODUCTION

Applications of mathematical techniques for the solution of mass transfer problems in the field of chemical and process industries has always been the centre of study for mathematicians, as well as for chemical engineers. The diffusion dispersion problems arising from packed bed reactors have always motivated the scientists for the development of new models (Pellet, 1966; Grähs, 1974; Neretnieks, 1974,1976; Perron & Lebeau, 1977; Al-Jabari et al., 1994; Kukreja et al., 1995; Eriksson et al., 1996; Potůček, 1997; Potůček & Pulcer, 2004; Arora et al., 2006,2008). It has given rise to the use of a variety of analytical and numerical techniques for the solution of these models such as Laplace transforms (Brenner, 1962; Pellett, 1966; Rasmuson & Neretnieks, 1980; Liao & Shiau, 2000; Aminikhah, 2012), orthogonal collocation method (Villadsen & Stewart, 1967; Michelsen & Villadsen, 1971; Raghvan & Ruthven, 1983; Adomaitis & Lin, 2000; Solsvik & Jakobsen, 2012), orthogonal collocation on finite elements (Carey & Finlayson, 1975; Ma & Guiochon, 1991; Arora et al.,2005,2006,2008), Galerkin method (Liu & Bhatia, 2001; Onah, 2002; Bhrawy & El-Soubhy, 2010; Nadukandi et al., 2010; Shen et al., 2011; Zhu et al., 2011; Solsvik & Jakobsen, 2012), Tau method (El-Daou & Al-Matar, 2010; Vanani & Aminataei, 2011; Solsvik & Jakobsen, 2012), Least Square method (Sporleder et al.,2010; Toledo et al.,2011; Wu, 2012; Solsvik & Jakobsen, 2012), spline collocation (Kadalbajoo et al.,2008; Khuri & Sayfy, 2009; Pedas & Tamme, 2011; Rashidinia & Ghasemi, 2011; Dhawan et al., 2012). In diffusion rate processes, usually the diffusion within particles pores and diffusion in the fluid phase between particles is considered. Diffusion dispersion problems in spherical catalyst particles have been studied widely as compared to particles having cylindrical geometry, e.g., fibers. Due to the highly porous nature and complex particle geometry of pulp fibers and rigidity in the computational approach, the majority of investigators have preferred the study of glass beads and granular particles instead of pulp fibers. The rate of removal of the soluble impurities from the pores of the particles is slower than that from the interstices of the particles, due to the relatively small pore size. Therefore, diffusion through the pores of particles is expected to control the overall washing rate (Pellett, 1966; Han, 1967).

The objective of the present study was to develop a mathematical model for the analysis of washing of the porous structure of particles, especially fibers. The prime motive of washing is to remove the soluble impurities adsorbed on the particle surface using external aid. For porous particles the solute must diffuse out before it is dispersed by the flowing liquid. The solute from the irregular void channels of the bed is displaced and coupled with diffusion-like dispersion of washing liquid in the direction of flow. Due to the adsorption of solute on the fiber surface, mass transfer occurs through the particle pores to particle surface and from particle surface to external fluid.

Dilution/extraction and displacement washing are basically two types of pulp-washing operations. In the former case the pulp slurry is diluted and thoroughly mixed with wash liquor and then filtered for thickening. In the latter case the liquor in the pulp bed is displaced with wash liquor. However, if the same amount of wash liquor is used, displacement washing is better than dilution/extraction (Potůček, 1997).

MODEL DEVELOPMENT

Fluid concentration in packed beds of porous particles is mainly based on two types of mechanisms. One is related to the transfer rate between fluid and particles and is given explicitly by the differential

This type of mechanism has been discussed widely for both linear and non-linear isotherms (Brenner, 1962; Sherman, 1964; Grähs, 1974; Potůček, 1997; Potůček & Pulcer, 2004; Arora & Potůček, 2012). In the second type of rate mechanism, the function f(n,q) is replaced by an integral equation signifying the solid phase diffusion into the interior of the particles. In this paper, second type of rate mechanism has been followed for a Langmuir adsorption isotherm to study the displacement washing of fluid flow through the packed bed of particles.

The packed bed, in the present study, is divided into two zones, namely a zone of flowing liquor and the zone of porous material. The packed bed is composed of unbleached Kraft pulp fibers having cylindrical geometry. The schematic representation of pulp fibers is given in Figure 1.

Figure 1
Schematic representation of fibers.

Flow of fluid through the bed is represented by the external fluid concentration c(x,t). intraparticle and interparticle solute concentrations are represented by n(r,x,t) and q(r,x,t), respectively. Particle and bed porosities are described by β and ε, respectively. Movement of solute within particle pores is defined by Fick's law. The representation of different zones of the packed bed is given in Figure 2.

Figure 2
Schematic representation of different zones.

Properties of the Mathematical Model

The proposed model is based on certain features which are discussed below:

a) The system is isothermal and the bed is macroscopically uniform.

b) The particles (fibers) are porous and of uniform cylindrical size.

c) Pore radius and particle length are very small as compared to the axial distance.

d) The dispersion in the fluid is described by an axial dispersion coefficient and diffusion in particles is described by an intraparticle diffusion coefficient and both are independent of axial distance, pore radius and solute concentration.

e) The adsorption equilibrium between interparticle and intraparticle solute concentrations is assumed to be Langmuir.

f) The average solute concentration is defined over the bed cross section.

Model for the Particle Phase

The diffusion equation describing the movement of solute within particle pores as defined by Pellett (1966), is given by:

where w is the local intraparticle solute concentration and does not distinguish between the solute adsorbed on the particle surface and the solute within the particle pores. To distinguish between these two, the surface diffusion effects are assumed to be negligible. The transport of solute within the particles is effectively described by a diffusion equation involving both the concentration of solute adsorbed on the particle surface and the intrapore solute concentration. The driving force for diffusion is assumed to be the intrapore concentration gradient and therefore the intraparticle diffusion equation takes the following form:

The local solute concentration applies to particles at any radial distance 'r', which gives w =β q + (1 -β ) n , and local equilibrium is assumed to prevail in the individual intraparticle pores.

Adsorption Isotherms

Immense literature is available for a linear adsorption isotherm (Sherman, 1964; Pellett, 1966; Neretnieks, 1974; Perron & Lebeau, 1977; Rasmuson & Neretnieks, 1980; Kukreja et al., 1995; Potůček, 1997; Potůček & Pulcer, 2004). This is due to the fact that a linear adsorption isotherm linearises the differential equation describing the behavior of fluid flow and condenses the mathematical complexities. Crotogino et al. (1987), Trinh et al. (1989), Al-Jabari et al. (1994) are among the prominent names to follow the study of the Langmuir adsorption isotherm. However, these authors discussed the bulk fluid phase only. In the present study, the intrapore solute concentration and the concentration of solute adsorbed on the particle surface are related by a Langmuir adsorption isotherm in the particle phase. The mass balance equation for the adsorption isotherm is:

Model for the Bulk Fluid

The transport phenomenon in the porous media with a bed porosity ε is described by a one-dimensional axial dispersion model involving an axial dispersion coefficient (Rasmuson & Neretnieks, 1980). The mass balance equation for the bulk fluid is:

where q is defined as the volume average concentration in the particles. To correlate the intrapore solute concentration and the bulk fluid concentration, it is necessary to define a relation between these two in terms of axial distance.

Rasmuson & Neretnieks (1980) and Raghvan & Ruthven (1983) have related the particle phase and bulk fluid phase by defining the condition at the particle boundary. Therefore, Eq. (5) reduces to the following form:

Using Eq. (6) in Eq. (4), the latter takes the form:

Initial and Boundary Conditions

The concentration gradient is assumed to be zero at the centre of the particle, i.e.,

It is assumed that external mass transfer resistance exists at r = R and mass transfer to the particle surface is controlled by the film resistance mass transfer coefficient.

It signifies the relation between Eq. (6) and Eq. (9).

At the inlet and creek of the bed, Danckwart's boundary conditions were applied.

Rearranging and introducing dimensionless variables given in the nomenclature, the mathematical Equations (2), (3) and Equations (7) to (12) can be converted into the following form:

HERMITE COLLOCATION METHOD

Orthogonal collocation is one of the weighted residual methods used to solve the system of boundary value problems. In this technique the residual is set equal to zero at collocation points. For a stiff system of equations, the method of orthogonal collocation does not give good results for small values of the parameters. To overcome this problem, Carey & Finlayson (1975) proposed the conjunction of finite elements with the orthogonal collocation method, termed as orthogonal collocation on finite elements (OCFE). In this technique, the trial function is approximated by using Lagrangian interpolating polynomials as base functions. With Lagrangian interpolating polynomials as base functions, an additional condition of continuity has to be imposed to make the approximating function continuous at the node points. To overcome this problem, the Hermite collocation method (HCM) was proposed, which is the combination of Hermite interpolating polynomials and the orthogonal collocation technique. In the Hermite collocation method cubic Hermite interpolating polynomials are used as base functions to approximate the trial function (Dyksen & Lynch, 2000; Lang & Sloan, 2002; Jung, 2003; Liu et al., 2005; Ma et al., 2006; Finden, 2007; Mazroui et al., 2007; Han, 2009; Ricciardi & Brill, 2009; Gülsu et al.,2011; Yalçinbaş et al., 2011; Orsini et al., 2011). To apply the collocation technique, the residual is set equal to zero at the collocation points. In Hermite interpolating polynomials, the additional condition of continuity at node points is not required due to the structure of Hermite interpolating polynomials, as the Hermite interpolating polynomials possess C¹ continuity. Therefore, the Hermite collocation method has the property to transform the mixed collocation method into an interior collocation method, which reduces the number of collocation equations.

The proposed model was solved using HCM by applying Hermite collocation on the radial and axial domain. In the proposed model, cubic Hermite polynomials were employed to discretize the approximating function. The cubic Hermite basis is defined (Dyksen & Lynch, 2000) as:

The piecewise cubic Hermite basis functions are designed such that:

P_j(ξ_i) = δ_ji, P_j'(ξ_i)=0, P_j(ξ_i) = 0, P_j'(ξ_i) = δ_ji

Orthogonal collocation is applied within each sub-domain in the radial and axial direction by intro ducing new variables η* and ζ, respectively, in such a way that h_y = ξ_y+1 - ξ_y such that ζ= 0 when ξ = ξ_γand ζ=1 when ξ = ξ_γ+1, and h_l=η_l+1-η_l such that η* = 0 when η*=η_l and η = 1 when η = η_l+1., where where

After rearranging the terms, P_j (ξ) and P_j(ξ) take the following form (Finlayson, 1980):

Therefore, the approximating function for C^γ(ζ, τ) can be defined as c_i^γ's are continuous functions of τ. Similarly, the approximating functions for Q ^{(l, γ)}(η,ζ,τ) and N ^{(l, γ)}(η,ζ,τ) can be defined in both the radial and axial directions. The structure of the (γ-1)^th , γ^th and (γ+1)^th elements in the axial domain for Hermite collocation is defined in Figure 3. where the

Figure 3
Structure of the (γ-1)^th , γ-1^th and (γ-1+1)^th elements in Hermite collocation.

COLLOCATION POINTS

Choice of the collocation points is the basis of the collocation technique. These are chosen in such a way to keep the error minimum. Therefore, the interior collocation points in both the axial and radial direction are taken to be the roots of the shifted Legendre polynomial. The collocation points are taken to be with 0 and 1 as boundary points by transforming the domain [-1,1] of a shifted Legendre polynomial of order 2 onto [0,1]. and

As given in the previous section, substituting

and radial domain, respectively. It converts the system of partial differential equations given in Eq. (13) to Eq. (20) into the system of ordinary differential equations. At the j^thcollocation point, the system of equations can be expressed as:

where B_ji and A_ji are discretization matrices for the second- and first-order derivatives, respectively, in the radial domain and B_ji and A_ji are discretization matrices for the second- and first-order derivatives, respectively, in the axial domain. The resulting system of differential equations can be put into the matrix form as H C'= M C, where H is the matrix of Hermite functions, C is the matrix of collocation functions and C'is the matrix of the derivative of these collocation functions with respect to time and M is the coefficient matrix.

After application of HCM, for 5 elements in the radial domain and varying elements in the axial domain, '46s₂' number of differential equations appear, where s₂ is the number of elements in the axial domain. This system of differential equations is solved using MATLAB with the 'ode15s' system solver, which uses the backward differentiation formula to solve the system of differential equations.

STABILITY ANALYSIS

The test of every numerical technique lies in how stable it is or how fast it converges. In present study, the stability analysis was made on the basis of the L₂-norm and L_∞-norm. In Table 1 to Table 4, the L₂-norm and L_∞-norm are presented for different values of Peclet number Pe and Biot number Bi at different time intervals. The average bed porosity ε and dimensionless parameter ψ are taken to be 0.6314 and 0.1, respectively. The values in Table 1 to Table 4 are found to be stable after 10 elements in the axial domain. However, the L_∞-norm is observed to be higher than the L₂-norm, but the stability in the results can be better judged by the L₂- norm as compared to the L_∞- norm.

Table 1
L₂-norm and L_∞-norm for Pe=5, Bi=4.5.

τ (-)	2 elements		5 elements		10 elements		12 elements
	||C||2	||C||∞	||C||2	||C||∞	||C||2	||C||∞	||C||2	||C||∞
0	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000
0.50	0.935	1.035	0.920	1.363	0.903	1.502	0.900	1.523
1.00	0.772	0.927	0.791	1.045	0.787	1.100	0.787	1.116
1.25	0.696	0.877	0.727	0.964	0.729	0.994	0.728	1.002
1.75	0.554	0.735	0.600	0.795	0.606	0.835	0.607	0.838

Table 2
L₂-norm and L_∞-norm for Pe=7, Bi=5.5.

τ (-)	2 elements		5 elements		10 elements		12 elements
	||C||2	||C||∞	||C||2	||C||∞	||C||2	||C||∞	||C||2	||C||∞
0	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000
0.50	0.935	1.083	0.945	1.496	0.940	1.568	0.938	1.604
1.00	0.732	0.999	0.782	1.120	0.786	1.146	0.786	1.161
1.25	0.636	0.892	0.702	0.987	0.710	1.028	0.711	1.038
1.75	0.458	0.655	0.539	0.828	0.558	0.854	0.556	0.857

Table 3
L₂-norm and L_∞-norm for Pe=10, Bi=6.7.

τ (-)	2 elements		5 elements		10 elements		12 elements
	||C||2	||C||∞	||C||2	||C||∞	||C||2	||C||∞	||C||2	||C||∞
0	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000
0.50	0.923	1.232	0.956	1.541	0.960	1.663	0.960	1.675
1.00	0.666	0.980	0.758	1.127	0.769	1.195	0.770	1.200
1.25	0.536	0.783	0.656	1.050	0.673	1.073	0.674	1.078
1.75	0.320	0.440	0.450	0.790	0.473	0.846	0.476	0.852

Table 4
L₂-norm and L_∞-norm for Pe=15, Bi=7.5.

τ (-)	2 elements		5 elements		10 elements		12 elements
	||C||2	||C||∞	||C||2	||C||∞	||C||2	||C||∞	||C||2	||C||∞
0	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000
0.50	0.938	1.391	0.981	1.665	0.992	1.836	0.993	1.850
1.00	0.609	0.907	0.761	1.286	0.778	1.302	0.780	1.311
1.25 1.75	0.438 0.200	0.616 0.359	0.637 0.385	1.095 0.823	0.666 0.425	1.174 0.890	0.669 0.429	1.182 0.893

EXPERIMENTAL

Displacement washing experiments simulated under the laboratory conditions were performed in a cylindrical glass cell with inside diameter of 35 mm under constant bed height of 30 mm. The fiber pulp bed occupied the volume between the permeable septum and a piston, covered with 45 mesh screens to prevent fiber losses from the bed.

Pulp beds were formed from a dilute suspension of unbeaten unbleached Kraft pulp in black liquor. Pulp was obtained from a blend of softwoods containing approximately ¾ spruce and ¼ pine. After compressing to the desired thickness of 30 mm, the consistency, i.e., mass concentration of moisture-free pulp fibers in the bed was about 12%. In order to characterise the pulp fibers used in the experiments, physical properties of Kraft pulp were determined. The Schopper-Riegler freeness had a value of 13.5°SR. The degree of delignification of the pulp was expressed by the kappa number of about 32. Using the Kajaani FS-100 instrument, the mean length of the fibers was also measured. The weighted average length was 3.35 mm, while the numerical average length was 1.46 mm. The specific surface of fibers having a coarseness of 0.15 mg m^-1 was 733 m² kg^-1.

Distilled water was used as wash liquid and was distributed uniformly through the piston to the top of bed at the start of the washing experiment. At the same time, the displaced liquor was collected at atmospheric pressure from the bottom of the bed v ia the septum. The washing effluent was sampled at different time intervals until the effluent was colourless. Samples of the washing effluent leaving the pulp bed were analysed for alkali lignin using an ultraviolet spectrophotometer operating at a wavelength of 295 nm. In Table 5, the input values for three runs of experiments are given. Displacement washing experiments with composition of pulp fibers, including washing equipment, were described in detail in Arora & Potůček (2012).

Table 5
Experimental data of pulp fibers.

Parameter	Value	Unit
Pe	7.921-10.651	-
ε	0.6075-0.6314	-
u	(0.1300-0.2071)×10-3	m/s
Consistency	11.737-12.303	%
DL	(0.611-1.145)×10-6	m2/s
β	0.7246	-
L	30×10-3	m
R	(10-25) ×10-9	m

RESULTS AND DISCUSSION

It can be observed from Table 1 to Table 4 that stability occurs after 10 elements in the axial domain. Therefore, the theoretical effect of different parameters on the exit solute concentration will be discussed for 5 elements in the radial domain and 10 elements in the axial domain.

Effect of Peclet Number (Pe)

In Figure 4, the theoretical effect of Peclet number is shown on the exit solute concentration for Bi= 5 and ε = 0.6314. It can be observed from this figure that, as the value of Peclet number increases, i.e., for Pe≥ 20, solution profiles converge rapidly to the steady-state condition and follow a Gaussian shape curve. However, for small values of Peclet number, i.e., Pe = 5, solution profiles take a long time to converge to a steady-state condition. This is due to the fact that, for small values of Pe, the axial dispersion coefficient (DL) is large. It causes more back mixing and, as a result, fewer impurities are washed out and a long time in the washing operation is required. Therefore, high Pe is preferred for better washing operations at small intervals of time. For Pe = 20 and Pe = 30, no considerable effect is observed on solution profiles. It authenticates the fact that better washing operations can be achieved for Pe ≤ 30 (Al-Jabari et al., 1994; Arora et al., 2006).

Figure 4
Behaviour of the solution profile for different values of Pe with Bi = 5 and ε = 0.6314.

Effect of Biot Number (Bi)

Biot number (Bi) signifies the mass transfer resistances inside and on the surface of the particle. Theoretical effects of Biot number on the solution profiles are shown in Figure 5for Pe = 10 and ε = 0.6314. For small values of Biot number, i.e., Bi = 5, solution profiles take a long time to converge to the steady-state condition as compared to large values of Biot number, i.e., Bi = 15, where the solution profile converges for τ nearly equal to 3. This is due to the fact that, for high Bi, the mass transfer rate increases, resulting in more removal of impurities adsorbed on the particle surface and hence better washing operations can be achieved.

Figure 5
Behaviour of the solution profile for different values of Bi with Pe = 10 and ε = 0.6314.

Effect of the Dimensionless Parameter ψ

The effect of pore radius of the particle, film resistance mass transfer coefficient and interstitial velocity can be observed effectively by studying the effect of the dimensionless parameter ψ on the solution profiles. In Figure 6 the theoretical effect of ψ is shown for Pe = 10, Bi = 6.7 and ε = 0.6314.

Figure 6
Behaviour of the solution profile for different values of the dimensionless parameter ψ for Pe = 10, Bi = 6.7 and ε = 0.6314.

It is observed from this figure that, for ψ = 0.05, more time is elapsed to converge to the steady-state condition, as compared to the value of ψ = 0.5. This is due to the fact that, with the increase in the value of ψ, the pore radius of particles swells, which in-creases the diffusion in the particle pores, resulting in better removal of impurities adsorbed on the particle surface and hence better washing operations can be attained. For ψ > 0.5 no significant effect can be analysed on the solution profiles.

Effect of Bed Porosity (ε)

The packed bed porosity, ε is the most significant and perceptive factor which affects the solution profiles extensively. As the value of bed porosity increases, permeability increases causing the amount of solute removed from interparticle voids by the convective mechanism to increase, and the washing process runs very rapidly. The effect of bed porosity ε for Pe = 10 and Bi = 6.7, is shown in Figure 7. It can be easily observed from the figure that the solution profiles converge to the steady-state condition more rapidly for ε = 0.8120 as compared to ε = 0.5561.

Figure 7
Behaviour of the solution profile for different values of ε with Pe = 10 and Bi = 6.7.

Wash Yield (WY_τ=1)

The most significant solute removal parameter used in the industry is wash yield. It is the proportion of dissolved solids removed relative to the dissolved solids entering with the unwashed pulp. If the solute is split in a similar portion as the fluid, it is easy to analyse how much solute is left in the thickened pulp. Wash yield defined for τ = 1 can be calculated using the following formula (Potůček, 1997):

Wash yield for Pe = 9, Bi = 6 and ε = 0.6075 was calculated at different time intervals for 5 elements in the radial domain and varying elements in the axial domain. Values of WYare given in Table 6. It is observed from this table that wash yield at any time lies within the range varying from 0.7 to 0.8 which is less than nearly seventy times the initial solute concentration. As time increases from τ = 5, no considerable effect is observed and the values of WYτ=1 remain stable. For pulp fibers, the influence of Peclet number on wash yield can be linked by Equation (28) given by Potůček (2005), which agrees well with the values given in Table 6.

Table 6
WY _τ=1 for different elements of the axial domain and at varying time periods.

τ (-)	2 elements	5 elements	10 elements	12 elements
2 3 4 5	0.7834 0.7566 0.7532 0.7528	0.7680 0.7394 0.7355 0.7349	0.7604 0.7306 0.7266 0.7260	0.7592 0.7294 0.7252 0.7247

COMPARISON OF EXPERIMENTAL AND NUMERICAL VALUES

Theoretically, one can vary a single parameter keeping the other parameters constant. However, practically, it is not possible to vary only a single parameter. As bed porosity changes, the pore radius of the particle, interstitial velocity and permeability vary, which varies the axial dispersion and intraparticle diffusion coefficient and, therefore, the Peclet number and Biot number also vary. In Figure 8 the behaviour of the solution profiles is shown for different values of Pe, Bi and ε. Although the range of bed porosity does not vary greatly, the solution profiles converge rapidly for Pe = 10.651 and Bi = 6.7 as compared to Pe = 7.921 and Bi = 5.5. It reflects the fact that, with the increase in Biot number, the pore radius of particles increases, which gives better washing results.

In Table 7, a comparison between experimental and numerical values is shown for different values of Pe, Bi and ε in terms of relative error:

where C_exp is the experimental value and C_num is the numerical value. It is observed from this table that the minimum error is of the order 10^-3, which is less than 1% in all cases. It signifies the validity of the model for the washing cell.

Figure 8
Behaviour of the solution profiles for different values of Pe, Bi and ε.

Table 7
Comparison of Experimental and Numerical values.

Pe=7.921, Bi=5.5,ε=0.6314			Pe=9.230, Bi=6,ε=0.6075			Pe=10.651, Bi=6.7,ε=0.6195
Experimental	Numerical	Relative	Experimental	Numerical	Relative	Experimental	Numerical	Relative
		Error			Error			Error
1.0000	1.00000	0.0000	1.00000	1.0000	0.0000	1.0000	1.00000	0.0000
1.0000	1.00048	4.8000×10-4	9.6737×10-1	9.6929×10-1	1.9766×10-3	1.0000	1.00157	1.5700×10-3
9.7173×10-1	9.7029×10-1	1.4838×10-3	9.3008×10-1	9.3093×10-1	9.1202×10-4	1.0000	1.00087	8.7000×10-4
8.8734×10-1	8.8681×10-1	6.0321×10-4	9.0763×10-1	9.0996×10-1	2.5747×10-3	9.6433×10-1	9.6509×10-1	7.8795×10-4
7.1077×10-1	7.1074×10-1	4.3998×10-5	6.4407×10-1	6.4714×10-1	4.7762×10-3	7.9303×10-1	7.9107×10-1	2.4727×10-3
6.1286×10-1	6.1219×10-1	1.1040×10-3	5.5085×10-1	5.5074×10-1	1.9326×10-4	7.7146×10-1	7.7267×10-1	1.5696×10-3
5.6985×10-1	5.6967×10-1	3.2013×10-4	4.3051×10-1	4.3232×10-1	4.2148×10-3	6.5989×10-1	6.6091×10-1	1.5424×10-3
9.2175×10-2	9.2648×10-2	5.1268×10-3	1.2712×10-2	1.2817×10-2	8.2313×10-3	6.5948×10-2	6.5857×10-2	1.3774×10-3
3.2773×10-2	3.2812×10-2	1.1700×10-3	8.0508×10-3	8.1164×10-3	8.1460×10-3	1.9909×10-2	1.9781×10-2	6.4219×10-3
2.6628×10-3	2.6742×10-3	4.2687×10-3	1.4407×10-3	1.4476×10-3	4.7769×10-3	2.3642×10-3	2.3717×10-3	3.2002×10-3
1.1471×10-3	1.1368×10-3	8.9802×10-3	8.4746×10-4	8.4127×10-4	7.2967×10-3	1.1199×10-3	1.1133×10-3	5.8375×10-3
6.9644×10-4	6.9108×10-4	7.6962×10-3	5.9322×10-4	5.8904×10-4	7.0502×10-3	6.6363×10-4	6.6479×10-4	1.7630×10-3
5.3257×10-4	5.3132×10-4	2.3370×10-3	4.6610×10-4	4.7017×10-4	8.7198×10-3	4.9772×10-4	4.9464×10-4	6.1858×10-3

Displacement Ratio (DR)

Displacement ratio is the most perceptive factor in the washing theory. It is calculated from the average solute concentration of the solute using the following formula:

In Figure 9, a comparison has been made for the displacement ratio calculated for different values of Pe, Bi and ε. It is observed from this figure that, with the increase in the value of Pe, Bi and ε, the solution profiles for DR converge more rapidly as compared to small values of Pe, Bi and ε. For different values of Pe, Bi and ε, the solution profiles converge at almost the similar times. This is because the bed porosity lies in same range in the three washing runs (Arora & Potůček, 2012). However, the solution profile converges slightly faster for Pe = 10.651 as compared to other solution profiles.

Figure 9
Behaviour of DR for different values of Pe, Biand ε.

CONCLUSION

A mathematical model for the displacement washing process was presented and solved using HCM. Numerical results were compared with the experimental ones of a washing cell for different values of displacement washing parameters. It can be concluded from all the experimental and theoretical findings that the proposed mathematical model agrees fairly well with the experimental data, in view of significant aspects of dispersion, diffusion and different porosities. Comparison of numerical and experimental values presented in the tables and graphs validate this information.

NOMENCLATURE

Greek Symbols

ACKNOWLEDGEMENT

Dr. Shelly Arora is thankful to UGC for providing financial assistance in the form of the major research project F. No. 41-786/2012(SR). Ms. Inderpreet Kaur is thankful to DST for providing a INSPIRE fellowship. Authors are thankful to reviewers for their valuable comments.

REFERENCES

• Adomaitis, R. A. and Lin, Y., A collocation/quadrature-based Sturm-Liouville problem solver. Applied Mathematics and Computation, 110, pp. 205-223 (2000).
• Al-Jabari, M., van Heiningen, A. R. P. and van De Ven, T. G. M., Modelling the flow and the deposition of fillers in packed beds of pulp fibers. Journal of Pulp and Paper Science, 20(9), pp. J249-J253 (1994).
• Aminikhah, H., The combined Laplace transform and the new homotopy perturbation methods for stiff systems of ODEs. Applied Mathematical Modelling, 36, pp. 3638-3644 (2012).
• Arora, S., Dhaliwal, S. S. and Kukreja, V. K., Solution of two point boundary value problems using orthogonal collocation on finite elements. Applied Mathematics and Computation, 171, pp. 358-370 (2005).
• Arora, S., Dhaliwal, S. S. and Kukreja, V. K., Simulation of washing of packed bed of porous particles by orthogonal collocation on finite elements. Computers and Chemical Engineering, 30, pp. 1054-1060 (2006).
• Arora, S., Dhaliwal, S. S. and Kukreja, V. K., Mathematical modelling of the washing zone of an industrial rotary vacuum washer. Indian Journal of Chemical Technology, 15, pp. 332-340 (2008).
• Arora, S. and Potůček, F., Verification of mathematical model for displacement washing of kraft pulp fibres. Indian Journal of Chemical Technology, 19, pp. 140-148 (2012).
• Bhrawy, A. H. and El-Soubhy, S. I., Jacobi spectral Galerkin method for the integrated forms of second-order differential equations. Applied Mathematics and Computation, 217, pp. 2684-2697 (2010).
• Brenner, H., The diffusion model of longitudinal mixing in beds of finite length: Numerical values. Chemical Engineering Science, 17, pp. 229-243 (1962).
• Carey, G. F. and Finlayson, B. A., Orthogonal collocation on finite elements. Chemical Engineering Science, 30, pp. 587-596 (1975).
• Crotogino, R. H., Poirier, N. A. and Trinh, D. T., The principles of pulp washing. Tappi Journal, 70(6), pp. 95-103 (1987).
• Dhawan, S., Kapoor, S. and Kumar, S., Numerical method for advection diffusion equation using FEM and B-splines. Journal of Computational Science, 3, pp. 429-437 (2012).
• Dyksen, W. R. and Lynch, R. E., A new decoupling technique for the Hermite cubic collocation equations arising from boundary value problems. Mathematics and Computers in Simulation, 54, pp. 359-372 (2000).
• El-Daou, M. K. and Al-Matar, N. R., An improved Tau method for a class of Sturm-Liouville problems. Applied Mathematics and Computation, 216, pp. 1923-1937 (2010).
• Eriksson, G., Rasmuson, A. and Theliander, H., Displacement washing of lime mud: Tailing effects. Separations Technology, 6, pp. 201-210 (1996).
• Finden, W. F., Higher order approximations using interpolation applied to collocation solutions of two-point boundary value problems. Journal of Computational and Applied Mathematics, 206, pp. 99-115 (2007).
• Finlayson, B. A., Nonlinear Analysis in Chemical Engineering. McGraw-Hill, New York (1980).
• Grähs, L. E., Washing of cellulose fibres: Analysis of displacement washing operation. Ph.D. Thesis, Chalmers University of Technology, Goteborg, Sweden (1974).
• Gülsu, M., Yalman, H., öztürk, Y. and Sezer, M., A new Hermite collocation method for solving differential difference equations. Applications and Applied Mathematics, 6(11), pp. 1856-1869 (2011).
• Han, X., A degree by degree recursive construction of Hermite spline interpolants. Journal of Computational and Applied Mathematics, 225, pp. 113-123 (2009).
• Han, C. D., Washing theory of the porous structure of aggregated materials. Chemical Engineering Science, 22, pp. 837-846 (1967).
• Jung, H. S., L_∞ Convergence of interpolation and associated product integration for exponential weights. Journal of Approximation Theory, 120, pp. 217-241 (2003).
• Kadalbajoo, M. K., Yadav, A. S. and Kumar, D., Comparative study of singularly perturbed two-point BVP's via: Fitted mesh finite difference method, B-spline collocation method and finite element method. Applied Mathematics and Computation, 204, pp. 713-725 (2008).
• Khuri, S. A. and Sayfy, A., Spline collocation approach for the numerical solution of a generalized system of second-order boundary-value problems. Applied Mathematical Sciences, 3(45), pp. 2227-2239 (2009).
• Kukreja, V. K., Ray, A. K., Singh, V. K. and Rao, N. J., A Mathematical model for pulp washing in different zones of a rotary vacuum filter. Indian Chemical Engineer Section A, 37(3), pp. 113-124 (1995).
• Lang, A. W. and Sloan, D. M., Hermite collocation solution of near-singular problems using numerical coordinate transformations based on adaptivity. Journal of Computational and Applied Mathematics, 140, pp. 499-520 (2002).
• Liao, H. T. and Shiau, C. Y., Analytic solution to an axial dispersion model for the fixed bed adsorber. American Institute of Chemical Engineers, 46(6), pp. 1168-1176 (2000).
• Liu, F. and Bhatia, S. K., Application of Petrov-Galerkin methods to transient boundary value problems in chemical engineering: Adsorption with steep gradients in bidisperse solids. Chemical Engineering Science, 56, pp. 3727-3735 (2001).
• Liu, X., Liu, G. R., Tai, K. and Lam, K. Y., Radial point interpolation collocation method (RPICM) for partial differential equations. Computers and Mathematics with Applications, 50, pp. 1425-1442 (2005).
• Ma, Z. and Guiochon, G., Application of orthogonal collocation on finite elements in the simulation of non-linear chromatography. Computers and Chemical Engineering, 15(6), pp. 415-426 (1991).
• Ma, H., Sun, W. and Tang, T., Hermite spectral methods with a time-dependent scaling for parabolic equations in unbounded domains. Society for Industrial and Applied Mathematics, 43(1), pp. 5875 (2006).
• Mazroui, A., Sbibih, D. and Tijini, A., Recursive computation of bivariate Hermite splines interpolants. Applied Numerical Mathematics, 57, pp. 962-973 (2007).
• Michelsen, M. L. and Villadsen, J., A convenient computational procedure for collocation constants. The Chemical Engineering Journal, 4, pp. 64-68 (1971).
• Nadukandi, P., Onate, E. and Garcia, J., A high-resolution Petrov-Galerkin method for the 1D convection-diffusion-reaction problem. Computer Methods in Applied Mechanics and Engineering, 199, pp. 525-546 (2010).
• Neretnieks, I., A mathematical model for continuous countercurrent adsorption. Svensk Papperstidning-Nordisk Cellulosa, 11, pp. 407-411 (1974).
• Neretnieks, I., Adsorption in finite bath and counter-current flow with systems having a non linear isotherm. Chemical Engineering Science, 31, pp. 107-114 (1976).
• Onah, S. E., Asymptotic behaviour of the Galerkin and the finite element collocation methods for a parabolic equation. Applied Mathematics and Computation, 127, pp. 207-213 (2002).
• Orsini, P., Power, H. and Lees, M., The Hermite radial basis function control volume method for multi-zones problems: A non overlapping domain decomposition algorithm. Computer Methods in Applied Mechanics and Engineering, 200, pp. 477-493 (2011).
• Pedas, A. and Tamme, E., Spline collocation methods for linear multi-term fractional differential equations. Journal of Computational and Applied Mathematics, 236, pp. 167-176 (2011).
• Pellett, G. L., Longitudinal dispersion, intraparticle diffusion and liquid-phase mass transfer during flow through multiparticle systems. Tappi Journal, 49(2), pp. 75-82 (1966).
• Perron, M. and Lebeau, B., A mathematical model of pulp washing on rotary drums. Pulp & Paper Canada, 78(3), pp. TR1-TR5 (1977).
• Potůček, F., Washing of pulp fibre bed. Collection of Czechoslovak Chemical Communications, 62, pp. 626-644 (1997).
• Potůček, F. and Pulcer, M., Displacement of black liquor from pulp bed. Chemical Papers, 58(6), pp. 377-381 (2004).
• Potůček, F., Chemical engineering view of the pulp washing. Papir a Celulóza, 60(4), pp. 114-117 (2005).
• Raghvan, N. S. and Ruthven, D. M., Numerical simulation of a fixed bed adsorption column by the method of orthogonal collocation. American Institute of Chemical Engineers, 29(6), pp. 922-925 (1983).
• Rashidinia, J. and Ghasemi, M., B-Spline collocation for solution of two-point boundary value problems. Journal of Computational and Applied Mathematics, 235, pp. 2325-2342 (2011).
• Rasmuson, A. and Neretnieks, I., Exact solution of a model for diffusion in particles and longitudinal dispersion in packed beds. American Institute of Chemical Engineers, 26(4), pp. 686-690 (1980).
• Ricciardi, K. L. and Brill, S. H., Optimal Hermite collocation applied to a one-dimensional convection-diffusion equation using an adaptive hybrid optimization algorithm. International Journal of Numerical Methods for Heat and Fluid Flow, 19(7), pp. 874-893 (2009).
• Shen, T. T., Xing, K. Z. and Ma, H. P., A Legendre Petrov-Galerkin method for fourth-order differential equations. Computers and Mathematics with Applications, 61, pp. 8-16 (2011).
• Sherman, W. R., The movement of a soluble material during the washing of a bed of packed solids. American Institute of Chemical Engineers, 10(6), pp. 855-860 (1964).
• Solsvik, J. and Jakobsen, H. A., Effect of Jacobi polynomials on the numerical solution of the pellet equation using the orthogonal collocation, Galerkin, tau and least squares methods. Computers and Chemical Engineering, 39, pp. 1-21 (2012).
• Sporleder, F., Dorao, C. A. and Jakobsen, H. A., Simulation of chemical reactors using the least-squares spectral element method. Chemical Engineering Science, 65, pp. 5146-5159 (2010).
• Toledo, Leal, R. C. and Ruas, V., Numerical analysis of a least squares finite element method for the time dependent advection-diffusion equation. Journal of Computational and Applied Mathematics, 235, pp. 3615-3631 (2011).
• Trinh, D. T., Poirier, N. A., Crotogino, R. H. and Douglas, W. J. M., Displacement washing of wood pulp-an experimental study. Journal of Pulp and Paper Science, 15(1), pp. J28-J35 (1989).
• Vanani, S. K. and Aminataei, A., Tau approximate solution of fractional partial differential equations. Computers and Mathematics with Applications, 62, pp. 1075-1083 (2011).
• Villadsen, J. and Stewart, W. E., Solution of boundary value problems by orthogonal collocation. Chemical Engineering Science, 22, pp. 1483-1501 (1967).
• Wu, J., Least squares methods for solving partial differential equations by using Bezier control points. Applied Mathematics and Computation, 219, pp. 3655-3663 (2012).
• Yalçinbaş, S., Aynigül, M. and Sezer, M., A collocation method using Hermite polynomials for approximate solution of pantograph equations. Journal of The Franklin Institute, 348, pp. 1128-1139 (2011).
• Zhu, P., Xie, Z. and Zhou, S., A uniformly convergent continuous-discontinuous Galerkin method for singularly perturbed problems of convection-diffusion type. Applied Mathematics and Computation, 217, pp. 4781-4790 (2011).

Publication Dates

• Publication in this collection
  Apr-Jun 2015

History

• Received
  09 Dec 2013
• Reviewed
  22 July 2014
• Accepted
  24 Aug 2014