Anda di halaman 1dari 6

Journal of Mechanics Engineering and Automation 3 (2013) 29-34 D DAVID PUBLISHING

Flow Loss in Screens: A Fresh Look at Old Correlation

Ramakumar Venkata Naga Bommisetty, Dhanvantri Shankarananda Joshi and Vighneswara Rao Kollati
Engineering Aerospace, MCOE, Honeywell Technology Solutions Pvt. Ltd., Bangalore 560037, India

Received: December 7, 2012 / Accepted: January 5, 2013 / Published: January 25, 2013.

Abstract: Pressure losses in flow components are generally characterized either by pressure loss coefficients or by discharge
coefficients. The pressure drop for incompressible flow across a screen of fractional free area is often calculated from widely used
correlation provided in Perrys Handbook. This correlation was developed based on experimental work which have covered a wide
range of fractional free area ( = 0.14 to 0.79). The present work aims at validation for a flow in plain square mesh screen with a
particular fractional free area (porosity, ) of 0.25 using CFD (Computational Fluid Dynamics) approach. The simulations are carried
out for wide range of screen Reynolds number (Re = 0.1 to 105) covering both laminar and turbulent flow regimes. Initial simulations
are carried out for incompressible fluid (water) and further extended to compressible fluid (air). Discharge coefficients obtained from
the simulations are compared with experimental values. Effect of compressibility on discharge coefficients is described.

Key words: Pressure loss coefficient, discharge coefficient, screen, fractional free area, CFD (computational fluid dynamics),
compressibility.

1. Introduction flow through screens. They developed a pressure loss


correlation which predicts flow through screens for the
Fluid flow through screens takes place in a number
wire Reynolds number range of 104 to 104 using the
of technical areas including filtering, mining and
conventional orthogonal porosity and a function of
mineral processing, porous beds and a variety of flow
wire Reynolds number. The correlation was extended
straightening and turbulence reduction applications.
by the conventional cosine law to include flow which is
Information on flow characteristics of these screens is
not perpendicular to the screen.
very important factor while selecting the screens for
Sodr and Parise [4] designed an experimental
different applications. Several experimental studies
procedure to investigate the friction factor of
were carried out earlier to generate these flow
plain-square woven metal screen adopted in Stirling
characteristics for general purpose screens. Kays and
engine regenerator. They developed an equation to
London [1] did investigation on friction factor for four
evaluate the pressure drop in annular bed of screens.
woven metal screens. Armour and cannon [2]
Wu et al. [5] conducted experiments to measure
investigated hydraulic resistance of five types of
pressure drop of the flow through woven metal screens.
woven metal screens through experiments made in a
Four woven metals screens with different porosities of
circular channel with single layer of metal screen. They
the plain-square type were tested and based on the
provided an equation for the calculation of pressure
tested study, an empirical equation was developed for
drop based on the flow velocity, the porosity and the
friction factor characteristic of plain square type woven
geometry of the screen. Brundrett [3] did investigation
metal screens. They also developed five empirical
on the prediction of pressure drop for incompressible
equations respectively for five types of metal screens
(Plain square, fourdrinier, full twill, plain dutch and
Corresponding author: Ramakumar Venkata Naga
Bommisetty, Ph.D., technology specialist, research fields: twilled dutch types). Along with these resources,
turbine cooling, heat transfer and CFD. E-mail:
pressure drop across the screens is widely obtained
ramakumar.bommisetty@honeywell.com.
30 Flow Loss in Screens: A Fresh Look at Old Correlation

from Perrys Chemical Engineers Handbook [6] in incompressible and compressible flows. Effect of
which pressure drop (p) across the screens for compressibility on discharge coefficient is also
incompressible fluids (for constant densities) is explored.
expressed as
V 2 2. Numerical Set-Up
p = K (1)
2
In order to simplify the simulations, a screen with
where = fluid density, V = superficial velocity based
an array of 5 5 holes is considered, shown in Fig. 1.
upon the gross area of the screen, K = pressure loss
As shown, holes are square in section with side of
coefficient.
The relation between pressure loss coefficient and 0.015 mm and the gap between each hole is 0.015 mm.
discharge coefficients for such screens is obtained by This screen leads to fractional free area (porosity) of
1 1
2 0.25. Screen length of 0.015 mm is considered for
K = 2 2 (2) present study. This represents a typical screen used for
C
where C = discharge coefficient and = the fractional flow straightening application. In order to have some
free area or porosity which is defined as surface area of duct length for flow to develop, an entry and exit duct
opening to the total area. of lengths 1mm each are considered. The full domain
The discharge coefficient (C) for the screen with considered for the present study is shown in Fig. 2.
aperture Ds is given as a function of screen Reynolds ANSYS ICEMCFD (version 12.1) is used for model
number, Re = Ds(V/)/ in a graphical form. This generation and mesh generation. Tetrahedral elements
information is useful for plain square-mesh screens with prism layers are considered for grid
with wide range of porosity, = 0.14 to 0.79. This generation. For the grid refinement study the grid is
graphical curve fits most of the data within 20 percent. refined uniformly (by changing the maximum element
But this information is limited to incompressible flows. size) while the number of nodes have changed by 2-3
An attempt has been made in this paper to model a times.
screen and generate flow characteristics for a typical Regarding the boundary conditions for the
screen to cover a wide range of screen Reynolds simulations, the top, bottom and side faces
number (Re = 0.1 to 105) using CFD. Commercial (surrounding faces) are considered as walls. Velocity
software ANSYS ICEMCFD (version 12.1) [7] is used inlet condition is used for domain inlet. Pressure
for model and mesh generation. ANSYS CFX (version outlet with gage pressure of zero is applied for domain
12.1) [8] is used for solving and post processing. The outlet. Simulations are also performed with symmetry
reason for this tools selection is its flexibility and boundary condition for surrounding faces and no
robustness for creating unstructured mesh using considerable change is found in pressure drop with
tetrahedral elements with embedded prism layers at the change of boundary condition. High resolution
walls [7] and the solving capability of the tool with scheme is used to solve the continuity, momentum,
great accuracy [8]. All the simulations are performed energy and turbulence equations. The simulations are
under steady state conditions with quality and trust considered well converged when the monitored
regarding grid refinement and iterative errors. Initial properties got stabilized and the RMS residuals has
simulations are carried out with incompressible fluid dropped to 1e-4. For the simulations with screen
(with water) and then extended to compressible fluids Reynolds number less than 10, flow is considered as
(with air). Discharge coefficients are calculated from laminar [6]. At higher Reynolds number, flow is
pressure drop across the screen and are compared with considered as turbulent and standard k- model is used
the correlation available [6] for both the for turbulence closure.
Flow Loss in Screens: A Fresh Look at Old Correlation 31

3. Results
This part is divided into 3 sections dealing with grid
refinement, screen characteristics for incompressible
fluids and screen characteristics for compressible
fluids.

3.1 Grid Refinement Study

This grid refinement study is performed with


Reynolds number of 100 for three meshes. The
parameter of interest is static pressure drop across the
screen. Simulations are performed with nodes of
0.09M, 0.17M and 0.5M. The difference of static
(a) Front view of the screen with an array of 5 5
pressure drop across the screen is about 1.2% and the
value obtained with 0.17M is closer to the experimental
value compared to the other meshes. This accuracy is
by all means good enough and since the computational
time is reasonable with mesh of 0.17M nodes, this
mesh is used for further analysis.

3.2 Screen Characteristics for Incompressible Fluids

As mentioned earlier, simulations are performed


with incompressible fluid of water. Seven number of
simulations with screen Reynolds number of 0.1, 1, 10,
100, 1000, 10000 and 100000 are performed. Flow
with screen Reynolds number up to 10 is considered as
(b) Isometric view of the screen Laminar and the simulations with other Reynolds
Fig. 1 Geometric details of the screen. number are treated as turbulent. K- turbulence model
is used for turbulence closure. For the selected
(a) (b)
Reynolds numbers, inlet velocity is calculated from
screen opening dimension and fractional free area. This
value is supplied as the boundary condition at inlet.
Two rating stations, one at upstream of screen and
the second one at downstream of screen, are considered
for pressure drop calculation. Upstream station is
(c) (d) considered at a distance of two hydraulic diameter of
screen opening. Downstream rating station is
considered at a distance of five hydraulic diameters of
channel. Static pressures at these two rating stations are
obtained from the converged solutions and discharge
Fig. 2 (a) Mesh for the full domain; (b) Mesh distribution
near the screen; (c) Mesh in the fluid domain; (d) Mesh and coefficients are calculated using Eqs. (1)-(2).
prism layer distribution in the screen. Discharge coefficients for various screen Reynolds
32 Flow Loss in Screens: A Fresh Look at Old Correlation

number are compared with the available experimental


values [6], shown in Fig. 3. As is shown in Fig. 3,
predicted discharge coefficient matches very well with
experimental data up to Reynolds number of 10 (i.e., in
the laminar region).
For turbulent region, simulations under predicted the
discharge coefficients compared to the corresponding
experimental data. Discharge coefficient has increased
with increase in screen Reynolds number up to the
screen Reynolds number of 1000 and there is no
change in discharge coefficient with further increase in
Reynolds number. Simulations are also able to predict
the similar trend but the discharge coefficient got
(a) Velocity contours at selected planes
stabilized at 0.92 against the experimental value of 1.4.
along the streamwise direction
Discharge coefficients have crossed the value of unity
in the experiments which is not observed with the
simulations.
Velocity and pressure contours at selected planes
along the stream wise direction are shown in Figs.
4a-4b), for screen Reynolds number of 1000. Location
of selected planes and stream lines are shown in inset
of Fig. 4. Complete mixing and uniform flow is
observed on the upstream locations of the screen. As
expected, higher velocities are found near the screen
regions. Presence of screen is highlighted on the
downstream locations. A fully developed profile in
velocity is observed at outlet. From the contours of the
pressure, it can be noticed that huge pressure drop is
(b) Pressure contours at selected planes along
required to cross the screen. Pressure drop across the the stream wise streamwise direction
channel as a function of screen Reynolds number is Fig. 4 Contours of CFD results.
shown in Fig. 5.

Fig. 3 Effect of screen Reynolds number on discharge Fig. 5 Pressure drop across the channel for various screen
coefficients. Reynolds numbers.
Flow Loss in Screens: A Fresh Look at Old Correlation 33

The pressure drop across the screen is found to be


proportional to the Reynolds number (velocity) for
Reynolds number less than 10, and then increased as
square of the Reynolds number for Reynolds number
greater than 100. These observations are in line with
the assumption of laminar flow lower Reynolds
number (< 100) and turbulent for higher Reynolds
number.

3.3 Screen Characteristics for Compressible Fluids Fig. 6 Effect of compressibility on discharge coefficient.

Simulations with compressible fluid, air, have been in these simulations. Huge pressure drop is observed
carried out to cover the similar range of Reynolds across the screen because of high mach exist in the
number. Simulations are run with the geometry shown screen. Variation of discharge coefficient with screen
in Fig. 1 up to screen Reynolds number of ~ 100. It is Reynolds number is shown in Fig. 6. As there is a
found difficult to get converged solutions for further considerable variation of fluid density on upstream and
increase in Reynolds number because of high mach downstream locations, Screen Reynolds number is
numbers obtained in the screens. Dimensions of the calculated based on upstream location. Drastic
screen are scaled and simulations are performed for reduction in discharge coefficients is observed in these
further increase in Reynolds number. Using scaling, it simulations with increased screen Reynolds number.
has been possible to increase the Reynolds number Compressible effects have shown large impact on
even with lower mach numbers in the screen. Fig. 6 discharge coefficients. For the same Reynolds number
shows the comparison of predicted discharge for 1000, discharge coefficient has reduced from 0.8 to
coefficients with experimental and incompressible 0.41. It is noted that density variation across the screen
simulations. The maximum Mach number obtained at plays an important role in discharge coefficients. So,
the entrance of screen in these calculations is 0.4. The the available correlation is found suitable for density
maximum density variation across the screen is about variation below 20%.
20% (for Reynolds number of 100). With this low
4. Conclusions
Mach number and lower density variation, no
considerable differences are observed in the A numerical study is carried out to find the flow loss
dependence of discharge coefficient on Reynolds in screens. Commercially available CFD code ANSYS
number. But slightly lower discharge coefficients are CFX 12.1 is used for numerical study. Simulations are
found with compressible fluid compared to performed for both incompressible and compressible
incompressible fluids in turbulent region (Reynolds fluids. Laminar model is used for screen Re 10 and
number > 100). Discharge coefficient for compressible Standard k- turbulence model for higher screen
fluid became constant at 0.85 against the value of 0.92 Reynolds Numbers. Based on the present study, the
for incompressible fluid. following conclusions can be derived:
In addition to the aforementioned simulations, few y Predicted discharge coefficient values of
more simulations are also performed in the screen incompressible fluid matched with the available
Reynolds number range of 100-1000 with the initial experimental results up to screen Reynolds number of
screen dimensions to study the compressibility effects. 10. But for the turbulent region, predicted discharge
Mach number in the screen is observed to be transonic coefficients are lesser compared to the experimental
34 Flow Loss in Screens: A Fresh Look at Old Correlation

values; References
y Discharge coefficient has increased with increase
[1] W.M. Kays, A.L. London, Compact Heat Exchangers,
in Reynolds number up to screen Reynolds number of McGraw-Hill, 1964.
1000 and then constant discharge coefficient is obtained [2] J.C. Armour, J.N. Cannon, Fluid flow through woven
with further increase in Reynolds number. The constant screens, AIChE Journal 14 (3) (1968) 415-420.
[3] E. Brundrett, Prediction of pressure drop for
discharge coefficient obtained from the simulation is
incompressible flow through screens, Journal of Fluids
0.91 against the value of 1.4 in experiments;
Eng. 115 (2) (1993) 239-241.
y For compressible fluids, predicted discharge [4] J.R. Sodr, J.A.R. Parise, Friction factor determination for
coefficients are in line with incompressible fluids till flow through finite wire-mesh woven-screen matrices,
the density variation across the screen is about 20%; Journal of Fluids Eng. 119 (1997) 847-851.
[5] W.T. Wu, J.F. Liu, W.J. Li, W.H. Hsieh, Measurement and
y Considerable reduction (maximum of 50%
correlation of hydraulic resistance of flow through woven
reduction) is noticed in discharge coefficient with
metal screens, International Journal of Heat and Mass
higher density change for compressible fluids; Transfer 48 (2005) 3008-3017.
y Density change of fluid in the screen is found to [6] D.W. Green, R.H. Perry, Perrys Chemical Engineers
be stronger influencing parameter for flow loss in Handbook, 8th ed., McGraw-Hill, 2008.
[7] ANSYS, User Manual ANSYS ICEM CFD/AI*
screens compared to the screen Reynolds number.
Environment Release 12.1, ANSYS, Inc., Southpointe 275
Acknowledgments Technology Drive, Canonsburg, PA 15317, November
2009.
The authors are thankful to the management of [8] ANSYS, User Manual ANSYS CFX Release 12.1,
Honeywell and HTS for permitting them to share the ANSYS, Inc., Southpointe 275 Technology Drive,
findings. Canonsburg, PA 15317, November 2009.

Anda mungkin juga menyukai