# Effect of Fluid Bypassing on the Experimentally Obtained Darcy and Non-Darcy Permeability Parameters of Ceramic Foam Filters

- 1.3k Downloads
- 2 Citations

## Abstract

Ceramic foam filters (CFFs) are used to remove solid particles and inclusions from molten metal. In general, molten metal which is poured on the top of a CFF needs to reach a certain height to build the required pressure (metal head) to prime the filter. To estimate the required metal head, it is necessary to obtain permeability coefficients using permeametry experiments. It has been mentioned in the literature that to avoid fluid bypassing, during permeametry, samples need to be sealed. However, the effect of fluid bypassing on the experimentally obtained pressure gradients seems not to be explored. Therefore, in this research, the focus was on studying the effect of fluid bypassing on the experimentally obtained pressure gradients as well as the empirically obtained Darcy and non-Darcy permeability coefficients. Specifically, the aim of the research was to investigate the effect of fluid bypassing on the liquid permeability of 30, 50, and 80 pores per inch (PPI) commercial alumina CFFs. In addition, the experimental data were compared to the numerically modeled findings. Both studies showed that no sealing results in extremely poor estimates of the pressure gradients and Darcy and non-Darcy permeability coefficients for all studied filters. The average deviations between the pressure gradients of the sealed and unsealed 30, 50, and 80 PPI samples were calculated to be 57.2, 56.8, and 61.3 pct. The deviations between the Darcy coefficients of the sealed and unsealed 30, 50, and 80 PPI samples found to be 9, 20, and 31 pct. The deviations between the non-Darcy coefficients of the sealed and unsealed 30, 50, and 80 PPI samples were calculated to be 59, 58, and 63 pct.

## Keywords

Computational Fluid Dynamic Permeability Coefficient Superficial Velocity Permeability Parameter Filter Holder## Introduction

Ceramic foam filters (CFFs) are commonly used in various industries, including chemical, automotive, and metallurgy, for catalyst support, to control exhaust emissions, and for liquid metal filtration.[1, 2, 3, 4] In general, solid foams can be divided into open and closed cell materials, where each material has a different structure, property, and application.[2,5,6] Depending on the type and production method, ceramic foams may possess adequately high mechanical properties such as high thermal and chemical resistance, high structural uniformity, and strength.[6]

*via*a ceramic material,[2,7] as shown in Figure 1(a). Due to open pores, the foams have high permeability and, therefore, are able to capture solid particles. This makes CFFs suitable for filtration in metal production.[5,6] On the other hand, closed cell foams are polyhedron-like cells connected

*via*a solid interface while each cell is isolated from the others,[2,7] as illustrated in Figure 1(b). Such structure creates a suitable material for thermal insulation, fire protection, gas combustion burners,

*etc*.[5,6] In this research, the focus was on the open cell 30, 50, and 80 pores per inch (PPI) commercial alumina CFFs.

*i.e.*, filtration. In order to prime the filters, the melt has to reach a certain pressure on the top of the filter to initiate passage through the filter.[9, 10, 11, 12,14] To estimate, adjust, and maintain the required pressure and to regulate the melt velocity, it is essential to determine the Darcy and non-Darcy permeability coefficients of the filter.[8,9,15] The permeability coefficients can be obtained using permeametry experiments.[16, 17, 18] In general, permeametry is based on a gas or liquid flow through a porous media. Thus, the pressure drop

*∆P*(Pa) along the height of the porous media

*L*(m) as a function of the superficial velocity

*V*

_{s}(m/s) needs to be measured. Thereafter, the Darcy and non-Darcy permeability coefficients

*k*

_{1}(m

^{2}) and

*k*

_{2}(m) can be estimated using the Forchheimer equation for incompressible fluids (Eq. [1]),[2,8,9,15,17,19, 20, 21] if the fluid dynamic viscosity

*µ*(Pa s) and fluid density

*ρ*(kg/m

^{3}) values are known:

Different Methods That Have Been Used to Avoid Fluid Bypassing

No. | References | Year | Fluid | Filter Type | Sealing Method |
---|---|---|---|---|---|

1 | 1985 | any | any | a holder with thin rubber like sleeves | |

2 | 1999 | air | ceramic foams of 30 to 90 PPI | refractory cement | |

3 | 2000 | gas | aluminum based metal foams | thin, soft, cushionlike layer | |

4 | 2000 | air | Al | thin paper | |

5 | 2000 | air | aluminum metal foams | styrofoam | |

6 | 2002 | air | aluminum metal foams | styrofoam | |

7 | 2008 | air | Al | thin glass fabric | |

8 | 2008 | air-water | Ni, Ni-Cr, and Cu metal foams | careful machining and placement | |

9 | 2011 | air | periodic open cell foam | ceramic mat | |

10 | 2011 | air | SiC and Al | aluminum paper | |

11 | 2012 | air | Al | rubber band | |

12 | 2013 | water | Al | grease-impregnated cellulose fiber | |

13 | 2015 | argon | Al 5 PPI | alumina slurry coating | |

14 | 2015 | water | Al | blinding and grease-impregnated cellulose fiber |

## Method

### Experimental Procedure

Permeametry characteristics of commercial alumina CFFs of 30, 50, and 80 PPI were examined using water in the temperature range of 282 K to 284 K (9 °C to 11 °C). The experimental setup includes a submersible pump, a pressure transducer, a Plexiglas filter holder, 49.8-mm-diameter smooth pipes (outer diameter), a pressure transducer, a digital multimeter, a T-type thermometer, a National Instruments NI USB-TC01 data logger, an OHUAS T31P scale, a container, and CFFs of different PPI. Nine filter samples, three ~51-mm-diameter samples from 9-in. ~50-mm-thick filters for each PPI, were cut by using a computer numerical control water jet machine. Samples were divided in three groups. Thereafter, the samples were manually resized to about 49.5-mm diameter to fit into the filter holder, as explained elsewhere.[9] Group 3 of the samples was used to study the effect of fluid bypassing on permeametry parameters of the 30, 50, and 80 PPI CFFs. Therefore, the liquid permeametry experiments were performed first on the unsealed samples. Then the samples were dried at room temperature for 2 days before being sealed. The sealing procedure was done in three steps: (1) blocking of the side walls of the samples, (2) resizing, and (3) wrapping in grease-impregnated cellulose fibers.[9,15]

In total, two sets of permeametry experiments were performed using both the unsealed and well-sealed 30, 50, and 80 PPI samples. The pressure gradient profiles as a function of the mean fluid superficial velocity for both the well-sealed and unsealed samples were obtained, and the Darcy (*k* _{1}) and non-Darcy (*k* _{2}) permeability coefficients were empirically calculated, based on the Forchheimer’s equation (Eq. [1]). The Ergun’s approach, *i.e.*, dividing the Forchheimer equation (Eq. [1]) by superficial velocity and applying a linear regression, was used to estimate the permeability coefficients. The approach was found to be the appropriate method for obtaining the coefficients.[8,15,33]

### Mathematical Modeling

### Solution Technique, Transport Equations, and Assumptions

The Reynolds numbers for both the well-sealed and unsealed permeametry experiments were calculated to be in the range of 2500 to 26,000 and 5750 to 32,400, respectively.[15] Therefore, the turbulent flow computational fluid dynamics (CFD) simulations were performed.

*k–ɛ*,

*k–ω*,

*etc*., were not yet available in the porous media domain, as elucidated in COMSOL Multiphysics 5.1. Therefore, the following governing transport equations in the free flow and porous region for an incompressible fluid in a steady-state condition need to be solved.

- (1)
Free flow region—Reynolds-Averaged Navier–Stokes (RANS) equations for incompressible fluids, containing continuity (Eq. [2]) and conservation of momentum (Eq. [3]).[34, 35, 36, 37, 38]

- (2)
Porous region—continuity equation (Eq. [2]) together with the Reynolds-Averaged Brinkman–Forchheimer equation (Eq. [4]):[34,35,38, 39, 40, 41, 42]

*ρ*is density,

*U*

_{ i }is the time-averaged mean velocity in the

*x*

_{ i }direction,

*U*

_{ j }is the time-averaged mean velocity in the

*x*

_{ j }direction,

*P*is pressure,

*µ*is the dynamic viscosity,

*ɛ*is the filter porosity (here, open pore porosity was used), \( \left( { - \rho \overline{{u^{\prime}_{j} u^{\prime}}}_{i} } \right) \) is the Reynolds stress tensor

*τ*

_{ xy },

*k*is the Darcy drag term, and

*β*is the Brinkman–Forchheimer drag coefficient. Equation [3] consists of convection in the left-hand side and the following terms in the right-hand side: pressure gradient, viscous diffusion, and turbulent diffusion. In addition to the terms explained for Eq. [3], the two additional terms in the right-hand side of Eq. [4] are the Darcy and Forchheimer terms. The terms represent resistance to fluid flow in the porous media.

*τ*

_{ xy }) can be expressed as a function of eddy or turbulent viscosity,[36, 37, 38] as presented in Eq. [5]. According to the Prandtl’s mixing length hypothesis, the eddy viscosity is also expressed as a function of mixing length, as in Eq. [6]. In this hypothesis, it is assumed that a lump of fluid repositioning in the transverse direction retains its mean properties for the characteristic length of

*l*

_{mix}until it mixes with its surroundings.[15,36,37,43] In this theory, the mixing length (

*l*

_{mix}) is also correlated to the distance

*y*from the wall (Eq. [7]) with the proportionality constant

*κ*,

*i.e.*, Kármán constant, to be ≈0.41.[15,36,37,43]

*β*) in the Brinkman–Forchheimer equation (Eq. [4]) is defined by fluid density

*ρ*(kg/m

^{3}), geometric function

*F*(dimensionless), porosity

*ɛ*(dimensionless), and Darcy permeability coefficient

*k*(m

^{2}), as shown in Eq. [8]. Based on Ergun’s experimental findings on packed beds, the geometric function

*F*and Darcy permeability coefficient

*k*can be related to the porosity and particle diameter (Eqs. [9] and [10]).[44, 45, 46, 47] Here, the Darcy permeability coefficient (

*k*) is the same Darcy permeability coefficient (

*k*

_{1}) in Eq. [1].

*β*) cannot accurately model the experimentally obtained pressure drop values of the well-sealed 30, 50, and 80 PPI alumina CCFs. The average errors were reported to be in the range of 85.14 to 87.29 pct for the well-sealed 30, 50, and 80 PPI alumina CCFs. On the contrary, the empirically derived Darcy and non-Darcy terms, the same first- and second-order coefficients in the Forchheimer equation (Eq. [1]), could model the experimentally obtained pressure gradients with the average error of only 4.15, 1.56, and 4.41 pct for the well-sealed 30, 50, and 80 PPI alumina CCFs, respectively. Therefore, the same approach was employed to simulate fluid flow in the unsealed samples. As a result, the Forchheimer drag term (

*β*

*F*) defined by Eq. [11] was applied instead of the Brinkman–Forchheimer drag coefficient (

*β*) in Eq. [4].

In addition, the following assumptions were made in the statement of the mathematical model: (1) the fluid density, temperature, and dynamic viscosity were assumed to be constant; (2) the filters in the 2D axisymmetric models were assumed to be fully cylindrical; (3) the gravitational force was neglected (the filters were positioned horizontally in the experiment); and (4) the pipe surface was assumed to be smooth and frictionless.

### Boundary Conditions and Mesh Optimization

Boundary Conditions[15]

Inlet | Outlet | Wall |
---|---|---|

| | |

Minimum and Maximum Mesh Size Options[15]

Mesh No. | Element Size (mm) | Boundary Layer Size (mm) | Average Mesh Quality | ||
---|---|---|---|---|---|

Min. | Max. | Min. | Max. | ||

1 | 2.49 × 10 | 0.872 | 2.49 × 10 | 0.872 | 0.8437 |

2 | 9.96 × 10 | 0.697 | 9.96 × 10 | 0.697 | 0.8727 |

3 | 3.74 × 10 | 0.324 | 3.74 × 10 | 0.324 | 0.9332 |

4 | 4.98 × 10 | 0.167 | 4.98 × 10 | 0.167 | 0.9348 |

Effect of Mesh Size on CFD Predicted Pressure Gradient of Sealed 80 PPI Model

Fluid Velocity (m/s) | Exp. Pressure Gradient (Pa/m) | CFD Predicted Pressure Gradient (Pa/m) | Deviation to Experiment (Pct) | ||||
---|---|---|---|---|---|---|---|

Mesh 1 | Mesh 2 | Mesh 3 | Mesh 4 | Min. | Max. | ||

0.13 ± 0.008 | 184,457 ± 1310 | 191870 | 191946 | 192066 | 192090 | 4.02 | 4.14 |

0.21 ± 0.005 | 460,688 ± 3161 | 484312 | 483610 | 483961 | 484036 | 5.13 | 5.07 |

0.26 ± 0.005 | 690,917 ± 4088 | 718357 | 718649 | 719195 | 719321 | 3.97 | 4.11 |

## Results

### Permeametry Experiments

*et al.*[8] labeled as P (the light solid curves). Each presented number in the new experimental work represents an average of minimum 24, 26, and 31 readings with a confidence interval of 95 pct for the 30, 50, and 80 PPI filters, respectively. As presented in Figure 5 for an 80 PPI filter, the minimum and maximum margins of error for the pressure gradient data lie in the range of 0.59 to 0.85 pct and 1.29 to 6.16 pct for the fluid velocity.

The current experimental procedure is similar to Kennedy’s work. The only differences compared to the recent work were applying a slightly different sealing procedure, and taking experimental samples of alumina CFFs from the same manufacturer but from different batches, and different sized commercial filters.[15]

*k*

_{1}) and non-Darcy (

*k*

_{2}) permeability coefficients of the well-sealed and unsealed CFFs were estimated based on the experimentally obtained pressure gradients and the calculated mean fluid superficial velocity rates. The permeability coefficients were empirically estimated in accordance with the Forchheimer equation (Eq. [1]). As explained in Section II, the Darcy and non-Darcy coefficients for both the well-sealed and unsealed samples were acquired using the Ergun’s approach,

*i.e.*, dividing the Forchheimer equation (Eq. [1]) by velocity and applying a linear regression.[8,15,33] The calculated

*k*

_{1}and

*k*

_{2}values and the deviations between the well-sealed and unsealed coefficients are summarized in Table VI.

Calculated *k* _{1} and *k* _{2} Values of the Fully Sealed and Unsealed Single 50-mm-Diameter Samples

Filter Type | Calc. Values | Sealed Sample | Unsealed Sample | Deviation (Pct) |
---|---|---|---|---|

30 PPI | | 3.705 × 10 | 4.068 × 10 | 8.94 |

| 6.434 × 10 | 1.577 × 10 | 59.21 | |

50 PPI | | 1.961 × 10 | 2.446 × 10 | 19.83 |

| 1.136 × 10 | 2.696 × 10 | 57.85 | |

80 PPI | | 8.692 × 10 | 1.258 × 10 | 30.88 |

| 1.074 × 10 | 2.893 × 10 | 62.88 |

### Mathematical Modeling

- (1)
modeling the well-sealed permeametry experiments using the empirically derived

*k*_{1}and*k*_{2}values from the permeametry experiments of the well-sealed samples;[15] - (2)
modeling the unsealed permeametry experiments based on the empirically obtained

*k*_{1}and*k*_{2}values from the permeametry experiments on the unsealed samples; and - (3)
modeling a condition when the empirically derived

*k*_{1}and*k*_{2}values of the unsealed experimental trials are used in a model representing the sealed condition.

Scenarios 1 and 2 were used to simulate the actual experimental conditions. However, scenario 3 aimed at investigating a condition when no sealing procedure is applied and the empirically derived Darcy and non-Darcy coefficients are treated as the “adequate” permeability parameters of the filters. This approach, *i.e.*, using unsealed or inadequately sealed samples in permeametry experiments, was observed in several articles during the literature review.

## Discussion

The pressure gradient profiles as a function of superficial velocity for the well-sealed and unsealed 30, 50, and 80 PPI CFFs were experimentally obtained. A considerable deviation exists between the two pressure gradients in all filter types (Figure 4). The average deviations between the two gradients were calculated to be 57.2, 56.8, and 61.3 pct for the 30, 50, and 80 PPI CFFs, respectively. The lower pressure gradients observed in the unsealed CFFs are due to fluid bypassing from the gap between the filters and filter holder. The gap provides a path of least resistance to fluid flow, causing bypassing along the wall of the unsealed samples. Therefore, no sealing or inadequate sealing would result in underestimation of the pressure gradient at any given flow rate.[8,9,15] The figure also compares the recent results to the previous permeability studies performed on samples taken from alumina CFFs by Kennedy *et al.*,[8] as explained elsewhere;[15] the previous results could be reproduced. The empirically obtained Darcy (*k* _{1}) and non-Darcy (*k* _{2}) permeability coefficients of the unsealed 30, 50, and 80 PPI CFFs reveal deviations to the permeability coefficients of the well-sealed samples. The deviations are caused by underestimated pressure gradients due to fluid bypassing in the unsealed samples.

The CFD estimated pressure gradients of the well-sealed filters were found to be in good agreement with the experimentally obtained pressure gradients (Figure 8). The deviation was calculated to be in the range of only 0.3 to 5.5 pct for all three PPI types of filters.[15] The bias between the experimental and CFD estimated pressure gradients is believed to be due to the assumption made for CFD studies; *i.e.*, the modeled filters possess a perfectly cylindrical shape.[15]

The CFD calculated pressure gradients of the unsealed 30, 50, and 80 PPI samples also showed good agreement with the experimental results (Figure 9). The deviations to the experiment were calculated to be in the range of 3 to 8 pct for 30 and 50 PPI samples and 17 to 21 pct for 80 PPI samples. The main source of the deviations is believed to be due to the application of the empirically obtained Darcy (*k* _{1}) and non-Darcy (*k* _{2}) permeability coefficients in the CFD module. Here, the empirically derived *k* _{1} and *k* _{2} values represent both the filter and gap. However, in the CFD module, the values could be assigned only to the filter. In addition, the higher deviations in 80 PPI might be due to the dimensions of the 80 PPI filter sample as well as the nature of the filter. More specifically, the 80 PPI sample has slightly smaller diameter and larger thickness than the 30 and 50 PPI samples, as shown in Table II. Therefore, more fluid bypassing and lower pressure gradients could result, *i.e.*, the gap available for fluid bypassing would be to some extent larger and the pressure gradient would be calculated based on a larger thickness. Moreover, the 80 PPI filters contain much smaller openings, compared to 30 and 50 PPI filters.[8,15,48] In an 80 PPI filter, fluid requires higher pressure to pass through the openings. This could result in more fluid bypassing from the gap between the filter and filter holder if the filter is not fully sealed.

The mathematically obtained pressure gradients according to scenario 3, *i.e.*, using the permeametry experimental data of the unsealed sample in a model representing the sealed condition, were compared to the experimentally obtained pressure gradients of the well-sealed and unsealed 30, 50, and 80 PPI filters. The CFD estimated pressure gradients based on scenario 3 deviate from the experimentally obtained pressure gradients of the well-sealed samples, as one may expect. The deviation was calculated to be in the range of 50 to 60 pct. However, the CFD estimated pressure gradients of scenario 3 lie on or very close to the experimentally obtained pressure gradients of the unsealed samples (Figure 10). This also can be considered as confirmation that neglecting the sealing procedure results in underestimation of the pressure gradients.

## Conclusions

- 1.
The pressure gradients of the unsealed 30, 50, and 80 PPI CFFs revealed a considerable deviation of 57.2, 56.8, and 61.3 pct to the well-sealed CFFs.

- 2.
Sealing procedure was found to be necessary for accurate estimation of the permeability parameters of the filters. In order to avoid bypassing, it is essential to completely seal the specimen and to fill the gap between the sealed samples and filter holder.

- 3.
The empirically obtained Darcy coefficients (

*k*_{1}) of the unsealed 30, 50, and 80 samples showed a 9, 20, and 31 pct deviation to the well-sealed 30, 50, and 80 samples. - 4.
The empirically obtained non-Darcy coefficients (

*k*_{2}) of the unsealed 30, 50, and 80 samples showed a 59, 58, and 63 pct deviation to the well-sealed 30, 50, and 80 samples. - 5.
The CFD predicted pressure gradients based on the empirically obtained Darcy and non-Darcy permeability coefficients are in good agreement with the experimental data for both the well-sealed and unsealed CFF samples. Specifically, the deviations to the experimental data were in the range of 0.3 to 5.5 pct for all three PPI types of the well-sealed filters. The deviations to experimental data were found to be in the range of 3 to 8 pct for 30 and 50 PPI and 17 to 21 pct for 80 PPI unsealed filters.

## Notes

### Acknowledgments

The authors express their gratitude for the financial support from the Swedish Steel Producers Association “JERNKONTORET.” The laboratory and technical supports from the materials science departments at the Norwegian University of Science and Technology (NTNU) and Royal Institute of Technology (KTH) are also acknowledged.

## References

- 1.M.D.M. Innocentini, V.R. Salvini, V.C. Pandolfelli, and J.R. Coury:
*Am. Ceram. Soc. Bull*., 1999, p. 78.Google Scholar - 2.M.V. Twigg and J.T. Richardson:
*Ind. Eng. Chem. Res*., 2007, vol. 46, p. 4166.CrossRefGoogle Scholar - 3.F. Scheffler, P. Claus, S. Schimpf, M. Lucas, and M. Scheffler: in
*Cellular Ceramics: Structure, Manufacturing, Properties and Applications*, Wiley, Weinheim, 2005, pp. 454–83.Google Scholar - 4.R.A. Olson and L.C.B. Martins: in
*Cellular Ceramics: Structure, Manufacturing, Properties and Applications,*2005, pp. 403–15.Google Scholar - 5.L.J. Gibson and M.F. Ashby:
*Cellular Solids: Structure and Properties,*Cambridge University Press, Cambridge, United Kingdom, 1999.Google Scholar - 6.L. Montanaro, Y. Jorand, G. Fantozzi, and A. Negro:
*J. Eur. Ceram. Soc*., 1998, vol. 18, p. 1339.CrossRefGoogle Scholar - 7.M.V. Twigg and J.T. Richardson:
*Chem. Eng. Res. Des*., 2002, vol. 80, p. 183.CrossRefGoogle Scholar - 8.M.W. Kennedy, K. Zhang, R. Fritzsch, S. Akhtar, J.A. Bakken, and R.E. Aune:
*Metall. Mater. Trans. B*, 2013, vol. 44B, p. 671.CrossRefGoogle Scholar - 9.S. Akbarnejad, M.W. Kennedy, R. Fritzsch, and R.E. Aune: in
*Light Met*., 2015, p. 949.Google Scholar - 10.R. Fritzsch, M.W. Kennedy, J.A. Bakken, and R.E. Aune:
*Miner. Met. Mater. Soc.*, 2013, pp. 973–79.Google Scholar - 11.M.W. Kennedy, S. Akhtar, J.A. Bakken, and R.E. Aune:
*Metall. Mater. Trans. B*, 2013, vol. 44B, p. 691.CrossRefGoogle Scholar - 12.M.W. Kennedy, S. Akhtar, J.A. Bakken, and R.E. Aune: in
*Light Met*., 2011, p. 763.Google Scholar - 13.J.W. Brockmeyer and L.S. Aubrey:
*Ceram. Eng. Sci. Proc*., 1987, vol. 8, p. 63.Google Scholar - 14.M.W. Kennedy, R. Fritzsch, S. Akhtar, J.A. Bakken, and R.E. Aune: 2013, Metall. Mater. Trans. B
*44*(3), 671-690.CrossRefGoogle Scholar - 15.S. Akbarnejad, L. Jonsson, M.W. Kennedy, R.E. Aune, and P. Jönsson:
*Metall. Mater. Trans. B*, 2016.Google Scholar - 16.J. Banhart:
*Progr. Mater. Sci*., 2001, vol. 46, p. 559.CrossRefGoogle Scholar - 17.M. Daniel, D.M. Innocentini, P. Sepulveda, and S. Ortega: in
*Cell. Ceram. Struct. Manuf. Prop. Appl.*, 2005, pp. 313–41.Google Scholar - 18.F.A.L. Dullien,
*Porous Media: Fluid Transport and Pore Structure*, Academic Press, Inc., Cambridge, MA, 1992.Google Scholar - 19.P. Forchheimer:
*Z. Vereins Dtsch. Ing*., 1901, p. 1781.Google Scholar - 20.M.D.M. Innocentini, V.R. Salvini, and V.C. Pandolfelli:
*J. Am. Ceram. Soc*., 1999, vol. 82, p. 1945.CrossRefGoogle Scholar - 21.M.D.M. Innocentini, P. Sepulveda, V.R. Salvini, V.C. Pandolfelli, and J.R. Coury:
*J. Am. Ceram. Soc*., 1998, vol. 81, p. 3349.CrossRefGoogle Scholar - 22.J.P. Bonnet, F. Topin, and L. Tadrist:
*Transp. Porous Media*, 2008, vol. 73, pp. 233–54.CrossRefGoogle Scholar - 23.B. Dietrich:
*Chem. Eng. Sci*., 2012, vol. 74, p. 192.CrossRefGoogle Scholar - 24.A. Inayat, J. Schwerdtfeger, H. Freund, C. Körner, R.F. Singer, and W. Schwieger:
*Chem. Eng. Sci*., 2011, vol. 66, p. 2758.CrossRefGoogle Scholar - 25.A. Banerjee, R. Bala Chandran, and J.H. Davidson:
*Appl. Therm. Eng*., 2015, vol. 75, p. 889.CrossRefGoogle Scholar - 26.K.M. Gupta: US Patent No. 4,495,795, Jan. 29, 1985.Google Scholar
- 27.S. Akbar Shakiba, R. Ebrahimi, and M. Shams:
*J. Fluids Eng*., 2011, vol. 133, p. 111105.CrossRefGoogle Scholar - 28.G. Incera Garrido, F.C. Patcas, S. Lang, and B. Kraushaar-Czarnetzki:
*Chem. Eng. Sci*., 2008, vol. 63, p. 5202.CrossRefGoogle Scholar - 29.A. Bhattacharya, V.V. Calmidi, and R.L. Mahajan:
*Int. J. Heat Mass Transfer*, 2002, vol. 45, p. 1017.CrossRefGoogle Scholar - 30.V.V. Calmidi and R.L. Mahajan:
*J. Heat Transfer*, 2000, vol. 122, p. 557.CrossRefGoogle Scholar - 31.J. Paek, B. Kang, S. Kim, and J. Hyun:
*Int. J. Thermophys.*, 2000, 21(2), p. 453–64.CrossRefGoogle Scholar - 32.M.D.M. Innocentini, W.L. Antunes, J.B. Baumgartner, J.P.K. Seville, and J.R. Coury:
*Mater. Sci. Forum*, 1999, vol. 19, pp. 299–300.Google Scholar - 33.S. Ergun and A.A. Orning:
*Ind. Eng. Chem*., 1949, vol. 41, p. 1179.CrossRefGoogle Scholar - 34.B.V. Antohe and J.L. Lage:
*Int. J. Heat Mass Transfer*, 1997, vol. 40, p. 3013.CrossRefGoogle Scholar - 35.D. Getachew, W.J. Minkowycz, and J.L. Lage:
*Int. J. Heat Mass Transfer*, 2009, vol. 42, p. 2909.Google Scholar - 36.D.C. Wilcox:
*Turbulence Modelling for CFD*, DCW Industeries, La Cañada Flintridge, CA, 1993.Google Scholar - 37.F.M. White:
*Fluid Mechanics*, 7th ed., McGraw-Hill, New York, NY, 2010.Google Scholar - 38.P. Prinos, D. Sofialidis, and E. Keramaris:
*J. Hydraulic Eng*., 2003, vol. 129, p. 720.CrossRefGoogle Scholar - 39.K. Vafai and S. Kim:
*Int. J. Heat Fluid Flow*, 1990, vol. 11, p. 254.CrossRefGoogle Scholar - 40.K. Vafai and J. Kim:
*Int. J. Heat Fluid Flow*, 1995, vol. 16, p. 11.CrossRefGoogle Scholar - 41.D.A. Nield:
*Int. J. Heat Fluid Flow*, 1991, vol. 12, p. 269.CrossRefGoogle Scholar - 42.K. Vafai and C.L. Tien:
*Int. J. Heat Mass Transfer*, 1981, vol. 24, p. 195.CrossRefGoogle Scholar - 43.J.R. Holton:
*An Introduction to Dynamic Meteorology*, 4th ed., Elsevier, Amsterdam, 2004.Google Scholar - 44.A. Amiri and K. Vafai:
*Int. J. Heat Mass Transfer*, 1998, vol. 41, p. 4259.CrossRefGoogle Scholar - 45.A. Amiri and K. Vafai:
*Int. J. Heat Mass Transfer*, 1994, vol. 37, p. 939.CrossRefGoogle Scholar - 46.S. Ergun:
*Chem. Eng. Progr*., 1952, vol. 48, p. 89.Google Scholar - 47.K. Vafai:
*J. Fluid Mech*., 1984, vol. 147, p. 233.CrossRefGoogle Scholar - 48.N. Keegan, W. Schneider, and H. Krug:
*Light Metals*, TMS, Warrendale, PA, 1999, pp. 1031–41.Google Scholar - 49.J.T. Richardson, Y. Peng, and D. Remue:
*Appl. Catal. A Gen*., 2000, vol. 204, p. 19.CrossRefGoogle Scholar

## Copyright information

**Open Access**This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.