vis fluid velocity [ft/min (m/min)], L is inlet solids concentration [Ib/ft* (kg/m*)], and
is time (min). An expression for the cake—fabric filter resistance coefficient using the
Kozeny-Carman procedure has been derived for determining flow through granular
media (2)
ky =(2% Te o
g e
where kis the Kozeny—Carman coefficient, which equals approx 5 for a wide variety of
fibrous and granular materials up to a porosity equal to about 0.8, € is the porosity or
fraction void v onless), His fluid viscosity [1b,,/(S [0] Pp
is the true density of solid material (Ih, /), and the S is the specific surface area/unit
volume of solids in the cake layer (£): This equation shows that as the particles being
filtered become smaller in diameter, the porosity of the cake decreases and consequently,
K, incteases. The net result of the larger cake—fabric filter resistance coefficient (K,) is
that the pressure drop increases as porosity decreases.
where G; is the dust loading and, along with V, is assumed co!
ing the filtration cycle; ¢ is the filtration time; and Ko, the dust
GReRte, is eotirnted from:
: 06
peavson (_te_) (2st) (_
Ky= ———, |-=— ] (—— —— 6-88}
2 de maal (25) ( vm» ) (0.0159 (6-88)
where d, is the geometric mass median diameter (m), 1, is the gas viscosity
ikg/m/sec|, op is the particle density |kg/m*), and V is the superficial velocity
m/sec}.
tant dur-4.1.1.2 Pressure Drop
During the mid-1800s, Darey formulated the followi
through a porous bed.”
1g law for flow of fluid
AP=LyV/K
whe
AP = pressure difference across the bed
L_ = bed thicknes
uf = fluid viscosity
Vs = superfi
K = bed permeability
‘This equation assumes that the fluid is essentially incompressible and steady,
the fluid ‘ewtonian, and the velocity is low enough that only
viscous effects occur, Over the past 100years, investigators have been trying
to find ways to predict K and to refine the Darey equation,
‘The basic Darey equation can be used to predict the pressure drop for an
operating fabric filter with dust cake accumulating on the fabric.
AP2SV+K.CV't
where:
AP = pressure drop, in. HO
Sp = effective residual drag, in, HO/fpm
v {pm
Ky = specific cake coefficient
G ntrations, gr/cubie foot
ration time, minutes
Energy loss through a fabric filter is composed of two parts. The first, SV,
represents drag or energy expended in pumping system gas through the
leaned equilibrium fabric of the fabric filter. The second part of the equation,
K.CV', represents energy required to pump gas through the filter eake that
builds up on the surface of the fabric, Gas velocity appears in both terms, but
hecause it is squared for the cake portion of the equation, it is especially
important for describing the energy consumed in pumping gas through the
filter cake. Another important part of the equation is K:, the specific cake
ccoelficient. This term is characteristic of the dust, varies for different dusts, and
is a measure of how rapidly pressure drop will build up in a system.
‘A fabrie filter in stable, cyclic operation will normally reach a point of con-
stant drag characteristics. That is, the resistance to gas flow of the freshly
leaned fabric is the same at the beginning of successive filtration cycles. In
practice, the value may change as the fabric ages. Residual drag is a measured
value, There is no useful predictive equation for resid4.1.1.3, Experimental Measurements of K,—Specific Cake Coefficient
Many researchers have conducted laboratory and pilot-seale fabric filter tests
to measure K; or specific cake coefficient, Billings and Wilder” reported an
extensive field survey of K; as a function of the air-cloth ratio (filtration veloc:
ity) and particle size. In this early work, K; was determined from the reported
values of operating air-cloth ratio (V), dust loading (C), filtration time (i), and
residual and maximum pressure drops (AP,,AP,.). While this earlier work was
quantitative, the wide range of dusts, quality of reported data, and configura
tion of the individual systems (single compartment, multiple compartment,
type of cleaning, etc.) led to considerable seatter. The relationship showed
order-of-magnitude variations in K; at a given particle size.
In more recent tesis, obtained under controlled conditions, the relationships
among K», particle size, and velocity have been shown more clearly.
Data from Dennis et al. and Davis and Kurzynske are shown in Figures
41.2 and 4-13, respectively." The solid lines represent each researcher's best
to the data, where available. The data reported by Dennis et al. were sum:
ed from eight different sources for ly ash, mica, and tale at 2-6fpm; the
1d Kurzynske were on tale dusts at a velocity of 4pm. Both
trong dependence of K; on the particle size,
It is evident from these data that velocity also has an effect on K. While
this observed effect may be partially attributed to the effect of velocity on dust
cake packing and/or Reynold’s number, most researchers have reported that
Ky isa function of velocity such that:
K.
ave
Dennis et al. reported that x had a value of 0.5 for fy ash and varied from 0.5
to 1.0, Davis and Frazier in a series of tests on fly ash using 1 different fier
materials, reported an average value of 0.7 for fly ash. The data in Figures 4-1.2
and 4-13, data by Davis and Frazier", and data by Prazier and Davis!” were
100
= om
= 50 ge Po
é 8
E of
= °
2
2° o |
= °
= 3
10 No
23 46 i020
MMO— jm
Figure 4-1.2, Data from Dennis et al, and Davis and Kuraynske lines of best fit.8
E
3
2 3 wm o
om
Figore 4-1.3, Ks serus MMD and face velocity (copyright ASTM),
normalized to a velocity of 3fpm and replotted in Figure 4-1.4, assuming an
average value for x of 0.6. The normalized data show that there is a well-
defined relationship between K> and particle s
A best-fit equation was determined for the data
184MMD""
where K, is measured in the English system and MMD is in microns, The best-
fit equation predicts the K; value within a factor of two. The agreement
4-14. Ky normalized to Spm vs. MMD (copyright ASTM)
between various sets of data is excellent considering that measurements
obtained under carefully controlled laboratory conditions for a constant par-
ticle size distribution have shown a facior-of-lwo variation within a single
laboratory.'