You are on page 1of 18

FUNDAMENTALS AND THEORY OF AIR SCOUR

By A p p i a h Amirtharajah, 1 M . ASCE

ABSTRACT: An original theoretical analysis of the dynamics of air scour from a


fundamental and microscopic viewpoint is presented. Concepts in soil me-
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

chanics and porous media hydraulics are combined to explain the complex flow
patterns which emerge when air and water flow concurrently through porous
media. An equation which predicts the formation' and collapse of air pockets
within the bed is developed by equating the air pressure within a bubble to
the soil stresses in an active Rankine state plus the pore-water pressures. This
condition of collapse-pulsing which occurs at particular combinations of si-
multaneous air and subfluidization water flows has been associated with the
probable optimum condition for air scour. The theory compares well with four
sets of experimental data collected on two different graded sands. The theo-
retical expression has been condensed into a simple design equation for prac-
tical application.

INTRODUCTION

A companion paper (7) presents an experimental characterization of


air-water dynamics through filter media during backwashing of filters
with air scour. In that paper the results are analyzed conceptually using
a soil mechanics approach. The effectiveness of backwashing filters with
air scour is related to the abrasion between sand grains. The intensity
of abrasion is coupled to: (1) The effective stresses between the grains;
and (2) the magnitude of their relative movement. These variables are
considered to be the major parameters affecting the detachment of fil-
tered particles from the grains during backwashing. The largest com-
bination of relative motion and effective stress between sand grains
seemed to occur with simultaneous air and subfluidization water flow
at a condition termed collapse-pulsing, where air pockets formed and col-
lapsed within the bed. An empirical equation to predict collapse-pulsing
was developed and related to possible optimum backwash with air and
subfluidization water used together. However, the underlying reasons
for the different types of motion which occur or the reasons for the ex-
istence of the empirical equation are not considered in the previous paper.
This paper presents an original theoretical analysis of the dynamics of
air scour from a fundamental and microscopic viewpoint. Many of the
qualitative observations made in the previous paper are rationalized in
terms of basic concepts in soil mechanics and fluid flow through porous
media. A theoretical equation is developed to predict the condition of
collapse-pulsing. This equation is favorably compared with experimental
data as well as the empirical expression developed previously (7).
'Prof., Dept. of Civ. Engrg. and Engrg. Mechanics, Montana State Univ., Boze-
man, Mont. 59717.
Note.—Discussion open until November 1, 1984. Separate discussions should
be submitted for the individual papers in this symposium. To extend the closing
date one month, a written request must be filed with the ASCE Manager of Tech-
nical and Professional Publications. The manuscript for this paper was submitted
for review and possible publication on October 17, 1983. This paper is part of
the Journal of Environmental Engineering, Vol. 110, No. 3, June, 1984. ©ASCE,
ISSN 0733-9372/84/0003-0573/$01.00. Paper No. 18899.

573

J. Environ. Eng. 1984.110:573-590.


APPLICABLE CONCEPTS FROM SOIL MECHANICS

Consider a small cubical element of dimension V at a depth, Z, within


a sand layer, as shown in Fig. 1(a). Any force applied to the matrix of
sand is transmitted by contact forces between adjacent grains, which can
be resolved into normal and tangential forces N0 and P0 as shown (8).
It is assumed that the pressure within the pores is zero (i.e., atmospheric
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

pressure). These contact forces give rise to a normal stress, a, and a


shear stress, T, on the faces of the elements in which
2N0 SP0
and T= (1)
a -a a- a
Under conditions where the surface of the sand is horizontal and the
properties of the sand do not vary in the horizontal directions, there are
no shear stresses on vertical and horizontal planes within the sand bed
and the stresses are called geostatic. They are also principal stresses. The
vertical geostatic stress, cr„, is calculated by considering the weight of
soil above that depth, i.e.
ov = 7SZ (2)
in which ys = unit weight of sand; and Z = depth below sand surface.
Strain at the contact point or by the deformation of a mass of granular
soil is fundamentally due to: (1) Elastic or plastic strain of individual
sand particles at their points of contact; and (2) the relative sliding be-
tween particles. Experiments have shown that interparticle sliding with
a rearrangement of particles is the major contributor to soil strain and
failure (8). When a cohesionless granular soil fails, its strength is rep-
resented by the Mohr-Coulomb failure law
jf = <Tf tan 4> (3)
in which suffix / = failure condition; and c> | = friction angle. When the
horizontal stress on an element of soil is decreased to a specified mag-
nitude, the full shear strength of the soil will be mobilized. No further
reduction in horizontal stress is possible/ and the soil is in an active Ran-
kine state of stress. The ratio between the horizontal and vertical stresses
can then be obtained from a Mohr circle analysis as shown in Fig. 1(b),

Mohr-Coulomb failure condition:


•tr » (T t a n <t>
sand surface

lanklne active \
j 17 T 7 T j B t a i e of s t r e s s ^

I stress, <Xa

TTTTT
(a) (b)

FIG. 1.—(a) Contact Forces and Stresses in Soil; (b) Mohr Circle Analysis for Ac-
tive Ranklne State
574

J. Environ. Eng. 1984.110:573-590.


where the stress circle touches the Mohr-Coulomb failure condition. From
AOAB
(o> - <rH)
t AB 2
+= OB~ (ay + aH)
(4)
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

2
(CT V -0H)
or (5)
((Ty + CTH)

in which oy = vertical stress; and crH = horizontal stress. Adding and


subtracting the numerator and the denominator and rearranging
oH _ (1 - sin <>
|)
(6)
ay (1 + sin <)>)
With standard trigonometric identities, Eq. 6 can also be expressed as
°H l 2
— = tan 4 5 - * (7)
ov
Eqs. 2 and 7 enable the horizontal and vertical stresses to be deter-
mined when the soil is in an active Rankine failure state. It should be
emphasized that Eqs. 6 and 7 are applicable only under conditions of
zero pore-water pressure (8).
Consider the same element 'a' under a saturated soil at level a-b as
shown in Fig. 2(A). The total stresses, oy and a H , acting on the element
will now also include the pore-water pressure within the pores, u. The
force carried by the soil skeleton per unit area is reduced by the pore-
water pressure and is defined as the effective stresses (intergranular stresses)
5y and aH given by
ov - oy - u (8)

IS.
L. V
1 X

I

i
. ..
N
S X
+•* m b * *
a— .««. -2-b
1 4+4*&« 44

'

(a)
element axa (b)

FIG. 2.—Effective Stresses and Pore Water Pressure In: (a) Saturated Soil; and
(b) Soil with Vertical Fluid Flow
575

J. Environ. Eng. 1984.110:573-590.


*H = <TH - « • • (9)
The effective stress is the only stress which contributes to frictional re-
sistance. It is also independent of the height of water above the sand
bed. Under conditions where a pore-water pressure exists in a soil stra-
tum, Eq. 7 is only applicable to the effective stresses and not to the total
stresses (8). Thus, the relationship for the coefficient of lateral stress for
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

the active Rankine state is

K = -^ = tan2 ( 45 - ^ ) (10)

Under saturated conditions with no water flow, applying Eqs. 2 and 8


to the system in Fig. 2(a), it can be shown (7) that
*v = Z(lt -!«,) = Zyb (11)
in which Z = depth below sand surface; yt = unit weight of sand plus
saturated water; yw = unit weight of water; and yt, = submerged or buoy-
ant unit weight. The term Zyb includes the total stress minus the static
pore-water pressure.
The effective stress is significantly influenced by water flow through
the media. Under flowing water conditions, in addition to the static pore-
water pressure, the frictional drag force on the individual sand grains
also reduces the effective stress between the grains. Consider the effect
of flowing water shown in Fig. 2(b). As shown in the figure, the total
head loss across the sand bed of depth H is h. Therefore the head loss
across the partial depth of sand Z is given by h(Z/H) in length units
and yw h(Z/H) in stress units. If the hydraulic gradient is defined as i,
then i = (h/H) or (dh/dZ). The dynamic pore-water pressure ud, is re-
lated to the head loss through the sand layer and is expressed as

ud = Z yj = Z 7 J — I (12)

Under conditions of upward flow of water, the effective vertical stress


is then represented by
(dh\
ov = Zyb-Z "iw\ — \ (13)

Eq. 13 is used to analyze conditions in the solid-liquid matrix during


simultaneous air-water backwash of filters. In addition to the soil me-
chanics concepts summarized previously, it is also necessary to sketch
briefly the fluid mechanics of bubble formation in liquids in order to
analyze the changes in air motion that occur during backwash. This
analysis then leads to the development of a new theory for air scour.

MECHANICS OF BUBBLE FORMATION IN LIQUIDS AND FLUIDIZED BEDS

Davidson and Schuler (3,4) studied the formation and motion of bub-
bles in viscous and inviscid liquids, both theoretically and empirically.
Amirtharajah and coworkers (1) have applied these concepts and ex-
tended the studies to the formation and motion of air bubbles in liquids
576

J. Environ. Eng. 1984.110:573-590.


and fluidized beds from circular orifices and underdrain orifices. The
motion of these large bubbles was principally controlled by the inertia
of the liquid carried by the bubble, and the effect of surface tension was
negligible.
In contrast to the preceding, let us consider the forces controlling the
motion of small bubbles (i.e., at small air flow rates) in water. Davidson
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

and Schuler (4) distinguished two cases: (1) Bubble formation at constant
air flow rate; and (2) bubble formation at constant pressure, which is of
greater practical importance. The second case applied to most real world
systems where the gas is from a constant pressure supply and the rate
of gas flow during the time of bubble formation, At is variable, even
though the rate of gas supply over long periods of time, t, may be con-
stant. For the constant pressure case, they showed that the rate of flow
of gas is proportional to the square root of the pressure difference across
the orifice (4). Thus
/ 2T\ V 2
Q„ = *»(?! -pgh-—J (14)

in which Q„ = air flow rate; Px = gage pressure on lower side of orifice;


pgh = static water pressure; T = surface tension; r = radius of bubble;
and k" = empirical coefficient. Eq. 14 can be rearranged in the form
IT
P1 = Pgh + — + kQ2a (15)
r
in which k = coefficient.
Eq. 15 indicates that under constant pressure conditions, the total pres-
sure required for the formation of bubbles should be equal to three frac-
tions: (1) The static water pressure, pgh; (2) the surface tension require-
ment (2T/r); and (3) a term related to the kinetic energy of the air flow
into the bubble. It is shown in later sections of this paper how the pre-
ceding equation could be applied in an original fashion to analyze the
formation and motion of air bubbles in porous media during air scour.

ANALYTICAL RATIONALE FOR DIFFERENT PHASES OF AIR MOTION

The different phases of motion of concurrent air and water through a


filter bed is complex, but has characteristics which are reproducible for
different sizes of sand when the water flow rate is expressed in terms
of the percentage of the minimum fluidization velocity, % Vmf. Visual
and cine film studies indicated the general patterns of motion which are
shown macroscopically in Fig. 3. They have been described in detail in
a companion paper (7).
The sequence of the different phases of air motion is best analyzed
conceptually by considering a fixed rate of air flow at a constant pressure
with a variable water flow. Consider the concurrent water flow rate in-
creasing from velocities of 0 to Vmf. Microscopic views of the different
phases of motion shown in Fig. 3, are shown magnified in Fig. 4. They
are analyzed as follows:
Air Scour Alone.—Under conditions of zero water flow, the air can
move through the pores of the media as small bubbles surrounded com-
577

J. Environ. Eng. 1984.110:573-590.


a 0
o 0
, ' : • - • • < >
a
•:•
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

(a) <°>

note: single bubble formation


In theBe two c a s e s

a Q
O
0

' VV
W*' I:./

(c) (d) (e)

FIG. 3.—Air Dynamics Through Filter Bed; Approximate Values of V in Terms of


(%Vml) are: (a) 0; (b) 5%; (c) 10-20%; (d) 25-50%; (e) 75% and Above

pletely by water as shown in the lower sections of Fig. 4(a). Let us as-
sume that the pressure within the bubbles at formation is given by Eq.
15. The smaller the bubble, the higher the contribution of surface tension
to its internal pressure would be. As the bubbles rise, the volume would
increase due to two effects: (1) The reduction in water pressure, pgh; and
(2) the release of surface tension pressure due to increasing radius. As
soon as the bubbles grow larger than the diameter of the pore openings,
the bubbles transform into tubular channels of flow, as shown in the
top layers of Fig. 4(a).
As the air tubes reach the surface of the sand, their diameters would
tend to grow larger due to the effective stress between the sand grains
tending toward zero and, thus, the air tubes pushing the grains apart
(Eqs. 8-11). In the limit at the surface of the sand, the intergranular
stresses are zero, and the sand functions like a liquid. As soon as the
air leaves the sand surface, the constrictions due to the pores which
cause tubular formation no longer apply, and the air exits in the form
of spherical bubbles to conform to the minimum surface energy shape.
At increasing air flow rates, these phases of motion will be reproduced
with the bubbles and channels being larger [Fig. 3(a)].
It should be evident that since several small bubbles are formed at the
supply orifice, it is possible for the bubbles to move randomly through
the pores as they transform into tubes. Thus, it is expected that air scour
alone would form exiting bubbles distributed randomly at several loca-
tions on the surface. The only abrasion between the grains occurs at the surface
where the tubes form and collapse, but there is no general failure of the soil or
relative motion between a large number of grains.
578

J. Environ. Eng. 1984.110:573-590.


w a t e r How f o r m a t i o n of air
V = 0 - 1 0 % Vmf channels
above
r e g i o n of cavity
V = 1 0 - 2 0 % Vmf stagnant water
sand grains
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

" f o r m a t i o n of
rearrangement air c a v i t y
and c o m p r e s s i o n
of sand g r a i n s much larger M n „ through
than grains m a d | a

(schematic)
V = 1 0 - 2 0 % Vmf
(a) (6) (c)

region where soil at end of c o l l a p s e and


IB being c o m p r e s s e d s t a r t of p u l s i n g , Burface t e n s i o n f o r c e s
due l o men
/ f o r m a t i o n of new air c a v i t y s i z e s i m i l a r l " c f In p o r e s
b u b b l e above t o sand g r a i n s
/ Vmf /

soil collapsing
under a c t i v e
Ranklne s t a t e pore water preesur<
air c a v i t y collapsing

(d)

FIG. 4.—Microscopic View of Air Dynamics: (a) Air Scour Alone; (b) Formation of
Air Cavity; (c) Flow Lines Around Large Air Cavity; (d) Collapse-Pulsing; (e) Forces
on Air Cavity

Air Scour Plus Low Subfluidization Water Flows.—As water begins


to flow upward, the pore-water pressure increases due to the head losses
within the bed and the effective stresses decrease (Eqs. 9 and 13). It
should be emphasized that even though the effective stresses are re-
duced as the water flow rate increases, they do not reach zero until spe-
cific head loss conditions are reached in a unisize media filter, at which
time the effective stresses are simultaneously zero at all depths of the
filter. This is the condition of minimum fluidization. Due to the lower
effective stresses at all depths of the filter, the tubular channels will form
at lower and lower layers of the filter [see Fig. 3(b)].
In a graded sand filter, it is probable that the top layers composed of
smaller sized particles may be fluidized while the lower layers are in a
static bed condition and, thus, bubble formation would occur at the top
layers at these low overall subfluidization water flows. This, in fact, oc-
curs in a graded sand filter as shown in Fig. 3(b).
Abrasion and cleaning effectiveness is probably even less than under
air scour alone, since abrasion between grains at the top most layers
where all the filtered particles lie is less than for air scour alone.
Air Scour Plus Subfluidization Water Flows for Collapse-Pulsing.—
As the water flow rate is increased further to approximately 10-20% of
Vmf, air pockets form at the supply orifice and then collapse. The for-
579

J. Environ. Eng. 1984.110:573-590.


mation and collapse of these air cavities also seem to indicate a failure
of the soil around it [Fig. 3(c)]. These experimental observations are ra-
tionalized as follows:
Assume that the effective stresses have been reduced to such an extent
that the air pressure at the orifice is greater than the total strength of
the soil-water structure, plus the requirements for surface tension and
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

kinetic energy of airflow. Under these conditions a cavity forms at the


supply orifice as shown in Fig. 4(b). As the cavity grows to dimensions
greater than the media grains, it causes a rearrangement and possible
increase of effective stress at the grain surfaces due to local and limited
compression of the grain structure. As the horizontal soil stresses are
less than the vertical stresses, the cavity would tend to grow horizontally
as shown in Fig. 4(c). As the cavity grows horizontally, the simultaneous
water flow around it will reduce the velocities of the water at the upper
surfaces of the cavity near the stagnation point as shown in Fig. 4(c).
This reduction in water flow will reduce the pore-water pressure at the
top regions of the cavity which would then revert to conditions similar
to the upper regions of the bed shown in Fig. 4(a) [or Fig. 3(a)]. Thus,
the air flow tends to form a single tubular channel very close to the
forward stagnation point [Fig. 4(c)], As the tubular air channel forms,
the pressure within the cavity declines rapidly, the horizontal stresses
in the soil reduce, and the soil around the cavity must fail under con-
ditions similar to an active Rankine state. These events are shown in
Fig. 4(d). The air cavity may then collapse to dimensions of the grain
sizes and then reform again, cyclically. The formation and collapse of
these air cavities have been termed collapse-pulsing by Hewitt and
Amirtharajah (7).
At the initial stages of collapse-pulsing (i.e., at velocities in the range
of 10-20% Vmf) a reduction in air pressure and an increase in volume
occurs as the cavities form one above the other. This may result in an
elongated horizontal cavity, causing the water to flow around the cavity
and reduce it to such an extent that conditions near the forward stag-
nation point are similar to the lower reaches of Fig. 3(a). This is shown
in Fig. 3(c) with the formation of a static horizontal lens or cavity at the
middle sections of the bed. The air leaks from the lens at the same rate
at which it is supplied in the form of small bubbles within the pores
which may transform into tubes at the upper layers of the bed [Figs.
3(c) and 4(a)].
Further increases of water flow rate cause collapse-pulsing to occur
throughout the entire depth of the bed at subfluidization water flows of
25-50% of Vmf [Fig. 3(d)]. Two important characteristics define the exact
flow rate at which collapse-pulsing occurs throughout the bed. At this
condition: (1) The air flow exits from the bed at a single location almost
exactly above the air supply orifice; and (2) there is a collapse of the soil
under active Rankine conditions around all of the cavities up to the sur-
face [Figs. 3(c) and 3(d)]. These conceptual considerations suggest that
the optimum air-water combination for cleaning is also the point at which
collapse-pulsing occurs throughout the bed. Some literature evidence to
confirm this has been reported (7).
Air Scour Plus Almost Fluidization Water Flows.—As the water flow
rate is increased to values greater than 75% of Vmf, the effective stresses
580

J. Environ. Eng. 1984.110:573-590.


are reduced to such an extent that the soil-water matrix functions like a
liquid and the higher inertia of the liquid tends to transfer momentum
to the air cavities or bubbles that are formed. The result is that the bub-
bles move upward with the water and the dynamics of motion is similar
to that of bubbles in liquids or fluidized beds [Fig. 3(e)]. The mechanics
of air bubble motion under these conditions is related to buoyancy forces
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

and the inertia of the liquid carried with the bubbles (1).
It is clear from the preceding that the detailed experimental observa-
tions (7), shown in condensed form in Figs. 3(a-e), is well rationalized
from a microscopic viewpoint by combining concepts in soil mechanics,
porous media flow, and fluidization. The conceptual framework used
for the qualitative analysis done in this section is now developed into a
theory to explain quantitatively the experimental data for collapse-pulsing.

THEORY FOR COLLAPSE-PULSING DURING AIR SCOUR

Consider the formation and collapse of air cavities as shown in Fig.


A(b-d). As the air cavity is initially growing, the soil around it may be
compressed, and the actual pressure within the bubble and soil are dif-
ficult to define. However, toward the end of collapse, the stresses in the
soil are calculable since the soil is in an active Rankine failure state, as
shown in Fig. 4(d). This is the condition, shown for purposes of analysis,
in Fig. 4(e). It is also assumed that the excess surface energy required
for surface tension is related to the menisci in the pores of the grains
and not to the overall size of the air cavity which has no definable ra-
dius. The formation and collapse of the cavity is assumed to be con-
trolled by horizontal stresses.
Consider Eq. 15. pgh represents the total static water pressure against
which bubbles are formed in liquids. Within a soil, assume that the cor-
responding pressure against which an air cavity forms and collapses is
the total horizontal stressCTH(which also includes the pgh term). Rear-
ranging Eq. 15 for a soil-water matrix with the preceding assumption

<TH= -Pi - — - *Q? (16)


r
From Eqs. 10 and 13 for a soil in an active Rankine state

, dh
aH = tan2 ( 45 - - I ov = tan2 ( 45 - j - (17)
z7b-z7w(-
Substituting for aH in Eq. 9
,'dhX
OH = 6H + u = tan 45 Zyb - Z,a ( - 1 + u (18)

For water flowing upward through a bed of depth Z and static water
level, Hi, above it, as shown in Fig. 2(b)
dh_
« = yw(Z + H^ + ywZ (19)
Jz
Substituting Eq. 19 in Eq. 18
581

J. Environ. Eng. 1984.110:573-590.


aH = tan z I 45 - - Z7b-Z7w(£ 7.(Z + H1) + 7 l ( , Z l £ j (20)
Under conditions of quasi-equilibrium when the air cavity collapses and
is in a transition state before reforming [Fig. 4(e)], o-H given by Eqs. 16
and 20 are equal. Therefore
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

. dh
P i ~ - kQ2a = tan2 ^45 - £ Zyb-Zya[ —

(dh\
+ 7 „(Z + H1) + ywZ^—J (21)

The head loss gradient, dh/dZ, through porous media can be related to
the superficial water velocity, V, using the Carman-Kozeny equation as
shown in Fair, et al. (5):
dh k'ii(l-e0ff6^v
dZ „ ,.,, (22)
8 Pi £o \W/
in which \i = coefficient of viscosity; p( = density of water; e0 = porosity
of static bed; \\i = sphericity; k' = coefficient; and d = diameter of par-
ticles. Consider a fluidized bed at minimum fluidization, the Fair and
Hatch relation equating head loss to the buoyant weight of particles in
differential form is given (2,5) by
dh
= (S,-l)(l-6) (23)
dZ
in which Sg = specific gravity of grains; and e = expanded porosity. Both
Eqs. 22-23 apply at minimum fluidization since it is the boundary con-
dition between the static and fluidized bed. Since e = €mf = e„, equating
the right-hand sides of the equations at the minimum fluidization ve-
locity, Vmf

v (s
§Pl »
e y -'- ' 1)(1 - O (24)

k'ix(l-z0)2/6 (St - 1)(1 - O


Therefore :— (25)
8 Pi iK V,mf
From Eqs. 22 and 25 the head loss gradient at V < Vmf is
V
= (S,-l)(l-e0) (26)
dZ V,mf
From Eqs. 21 and 26
V
Pr~ - fcQ2 = tan2 (45 - ^ Zlb - Zyw(Sg - 1)(1 - e0) —-
V,mf.

ya(Z + HO + y„Z(Sg - 1)(1 - e0) — (27)


V,mf

582

J. Environ. Eng. 1984.110:573-590.


Eq. 27 is the final equation of the theory. Under constant pressure con-
ditions and a fixed depth of a given sand media, all of the parameters,
Pi, T, r, k, ()>, Z, yb, Ss, yw, e0 and H i , are measurable or have standard
values. The expression Eq. 27 is derived from fundamental considera-
tions. Grouping the similar terms together and expressing, V/Vmf, as a
percentage rather than a fraction, Eq. 27 can be transformed to
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

[%v)+aQl=h • ^
in which a and b = constants for a particular system.
The theory developed in the preceding assumes that the filter grains
are not stratified and that the hydraulic gradient, dh/dZ, is uniform at
all depths, Z. While these assumptions are rather unrealistic at the initial
stages of air scour of a dirty filter, they are certainly valid towards the
later stages of backwashing when a large fraction of the deposited par-
ticles have been carried away by the water flow rate, V, and when sig-
nificant mixing of the grains has occurred.
It is worthwhile emphasizing a comparison of the preceding theoret-
ical form of the equation at collapse-pulsing with the empirical equation
obtained by Hewitt and Amirtharajah (7):

(%•¥-) + 3.64Q„ = 49.0 (29)


\ "mf/
In the following sections, Eq. 27 with the independently measured val-
ues of coefficients is compared with experimental data.

EXPERIMENTAL MEASUREMENTS FOR DATA AND COEFFICIENTS

The experimental methodology and the data collected have been de-
scribed in the companion paper and elsewhere (6,7). The combinations
of air and water flow rates to produce collapse-pulsing were determined
for three different graded sands at two different depths for each of the
two sands and for a single depth for the third sand. The data were used
to obtain an empirical equation for collapse-pulsing. In all of the exper-
iments, the total depth of media and static water above it, Z + Hlf was
5.66 ft (1.73 m). In the following analyses the third sand was not used
since it was tested only at a single depth, and measurements for Pi and
k were not made for this sand.
In addition to the preceding data, it was also necessary that indepen-
dent measurements for Pi, k and § be made from additional experi-
ments, to validate the theory. The pressure, P, just below the orifice was
measured with a water manometer at various air-water flow rates for
the sand with D60% size = 0.62 mm at two different depths. The variation
of inlet air pressure versus water flow rate is shown in Fig. 5 for the
fixed air flow rate of 2.10 scfm/sq ft (38.41 m/hr). The water flow rate
for collapse-pulsing equal to 2.58 gpm/sq ft (6.30 m/hr), and its corre-
sponding air pressure equal to 3.56 psi (24.56 kN/m 2 ), is also shown in
Fig. 5. These data of air pressure at collapse-pulsing were plotted aganst
the square of the air flow rate, Q 2 , as shown in Fig. 6. For the 30 in.
(760 mm) depth of sand, the theoretically expected linear relation be-

583

J. Environ. Eng. 1984.110:573-590.


w a t e r f l o w r a t e at
collapse-pulsing

/ i
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

/ i 2
air f l o w r a t e = 2 . 1 0 s c f m / f t
sand d e p t h — 1 8 i n .
D = 0 . 6 2 mm
60

1 1 1 1 1 1 _
1.0 2.0 3.0 4.0 5.0

WATER FLOWRATE, V (gpm/ft )

FIG. 5.—Typical Variation of Inlet Air Pressure with Water Flow Rate (1 gpm/sq
ft = 2.44 m/hr, 1 psi = 6.9 kN/m2)

sand size , D = 0.62 mm


60
A = d e p t h of 3 0 i n .
0 = d e p t h of 1 8 I n ,

-3 2
P = 3 . 8 8 - 2.36 X10 Q.

-o*-
-O --„-._"__
a : "•
- <- j
<
3 T -3
P = 3 . 6 5 - 2.36 X10
2
O.

2 2 2 2
(AIRFLOW) , Qa ( s c f m / f t )

FIG. 6.—Air Pressure versus Air Flow Function for Collapse-Pulsing [1 psi = 6.9
kN/m2,1 (scfm/sq ft)2 = 334.52 (m/hr)2]

tween P and Q„ is obtained with a correlation coefficient of 0.77. For the


18 in. (460 mm) depth of sand, the scatter in the data is considerable.
Despite the scatter, a linear relation with the same slope seems reason-
able. From the lines in Fig. 6, the values for k and Pi in Eq. 27 are ob-
tained from the slopes and intercepts, respectively. These data were col-
lected (6) prior to the development of the theory in this paper. Un-
fortunately, similar data for the other two sands were not collected. For
the sand with D60 = 0.86 mm it was assumed that a similar slope would
be obtained (i.e., -2.36 X 10~3) as in Fig. 6, and that a similar difference
in the intercepts would occur for the 18 in. (460 mm) and 30 in. (760
mm) depths. The values of Px and k used for subsequent analyses are
summarized in Table 1. Also shown in the table are the minimum flui-
dization velocities, Vmf.
The angle of internal friction, <t>, was measured experimentally (6) by
using a rectangular box 6 in. x 6 in. X 12 in. (150 mm x 150 mm X 150
584

J. Environ. Eng. 1984.110:573-590.


TABLE 1 .—Summary of Values for <$>, Px and k for Use in Eq. 27
^ Minimum
k
fluidiza-
tion ve- In pounds
locity, In pounds per square
Vmf, in per square foot per
F
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

gallons inch per standard


per min- Angle of In pounds In pounds standard cu- cubic feet
Sand size, ute per internal per per bic feet per per minute
Deo. in Sand depth, Z, square friction, square square minute per per square
millimeters in inches foot <t>° inch foot square foot foot
(1) (2) (3) (4) (5) (6) (7) (8)
0.62 30 6.0 33.2 3.88 558.7 2.36 x 10" 3 0.34
0.62 18 6.0 33.2 3.65 525.6 2.36 x 10~3 0.34
0.86 30 11.0 33.9 3.62 521.3 2.36 x 10" 3 0.34
0.86 18 11.0 33.9 3.39 488.2 2.36 x 10" 3 0.34
Note: 1 in. = 25.4 mm; 1 gpm per sq ft = 2.44 m/hr; 1 psi = 6.9 kN/m 2 ; 1 psf = 47.9
N/m 2 ; 1 (scfm/sq ft)2 = 334.54 (m/hr) 2 .

mm) with a sliding gate as one of the sides. The box was filled with
dried and cooled (to room temperature) sand, and the gate was opened.
The sand collapsed with a very smooth failure plane, whose inclination
to the horizontal was calculated from a measurement of its tangent. The
results could be reproduced to an accuracy of ±0.5°. The results for the
two sands are tabulated in Table 1. The values of the parameters in Table
1 are used in Eq. 27 to calculate the theoretical equations which are then
compared with experimental data. All of the data were collected at 20° C
(±1°) and the calculations were made for a temperature of 20° C.

COMPARISON OF THEORY AND EXPERIMENT

In order to simplify Eq. 27 to a form similar to Eq. 28, it is necessary


that a rational means of calculating the radii controlling surface tension,
r, be developed. As shown in Fig. 4(e), the surface tension force differ-
ential arises across the air-water interface only. This occurs only within
the pores. Typical diameters of pores within spherical media are 7-10%
of the grain dimensions. Since the theory is developed for collapse-puls-
ing throughout the bed depth, and the D60% size of media has been used
for describing the features of graded sands in backwashing studies, this
size was chosen to represent the pore dimensions. When collapse-puls-
ing occurs throughout the bed there is significant destratification and
mixing and thus it is rationalized that the D60% size is the appropriate
choice for the model. Choosing a mean pore size of 8.2%, the value of
r used for the theoretical calculations is 0.082 x (D60/2). This particular
value of pore size was chosen so that a theoretical line matched one set
of experimental data reasonably. Thus the pore size could be construed
as an adjustable parameter; however, the identical value of pore diam-
eter is used for all the sands and the value is very consistent with typical
pore dimensions. Therefore, it is not a parameter that may be arbitrarily
adjusted. Typical numerical values are assigned for all of the other pa-
rameters as follows.

585

J. Environ. Eng. 1984.110:573-590.


An Illustrative Theoretical Calculation.—The following calculations
in English units are for the graded sand, D60 = 0.62 mm and depth Z
= 2.5 ft (760 mm). The values used for the constants are: Surface Tension
T at 20° C = 72.8 dynes/cm = 5.0 x 10"3 lbf/ft; r = [0.082 x (0.62/2)]
mm = 8.34 X 10' 5 ft (2.54 X 10"5 m); yw = 62.4 pcf (9.81 kN/m 3 ), 7, =
120.0 pcf (18.84 kN/m 3 ); yb = 57.6 pcf (9.03 kN/m 3 ); and e0 = 0.40 Sg =
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

2.52.
Using the preceding constants and other values from Table 1 in Eq.
27
2T , „/ d> V
Pi - — ~ kQl = tan21 45 - J Zyb - Zyw(Sg - 1){1 -60) V —
mf.
V
+ yw(Z + Hi) + ywZ(Sg - 1)(1 - €„) (30a)
V mf
2 X 5.0 X 10~3 , ,/ 33.2A
which gives
6 558.7 =- Q 2 = tan2 45
5 - 0.34 w ,
8.34 x 10~ \ 2
V
2.5 X 57.6 - 2.5 x 62.4 (2.52 - 1)(1 - 0.40) + 62.4(5.66) + 62.4
Vmf.
V
X 2.5(2.52 - 1)(1 - 0.40) (30b)
V,mf
V
Thus 0.34 Q2 + 100.75 — = 43.57 (30c)
Vmf
Converting (V/Vmf) into %(V/Vmf)
V I =43.2
0.34Q2+I% — • (31)
V VmfJ
In a similar fashion, the theoretical equations calculated for the other
sand depth and sand sizes using the parameter values in Table 1, are
summarized as follows:

D60 = 0.62 mm, sand depth = 18 in. (460 mm):


V
0.56 Q 2 + %— = 45.0 (32)
Vmf,
D60 = 0.86 mm, sand depth = 30 in. (760 mm):
V
0.34 Qi + % V, = 40.2 (33)
mf,
D60 = 0.86 mm, sand depth = 18 in. (460 mm):
V
0.56 Qj + %— = 39.3 (34)
V,mf,
The theoretical equations, Eqs. 31-34, are plotted against the experi-
mental data in Figs. 7-8. It is seen that the correspondence between
586

J. Environ. Eng. 1984.110:573-590.


aond depth = 3 0 In. D 0 0 = 0.62 mm


A
—*»»_ A 2 V
^^-****^ from theory: 0 . 3 4 Q . + (%n—) =43.2

£ 30
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

A
mo ^^"***,****^>*-fc.

experimental d a l a ^ ^

i i i i
D =
a and depth = 18 In. eo 0*82 mm

""
«»«w^ from theory: 0.S6 Q a 2 4- l%~~) = 4 5 , 0

®^^^/
®^@

\ \ experimental
^ ^d a ^ i *
ta^**^^^

i i i i

FIG. 7.—Comparison of Theory with Experiment for Collapse-Pulsing: Graded Sand


Media, D60 = 0.62 mm [1 (scfm/sq ft)2 = 334.52 (m/hr)2]

PERCENT MINIMUM FLUIDIZATION, ( * T71—)

O
1 1 71 •
- r _1 -T
r

i •
I
5
(AIR 1

o
II / . I
II
r o
o X
s -/\i S
erlm

theory

3 3 ? ?
9 / s

o
o
B /' is / o
W Q. 01 ? H/ w
s
o »o> tj
/f *
\ / o 0
m/ft

M
o \ / + D
2. // a
5 S 1
W
M

1
I \ s
I at
II

P •? // II

p
03 a
II a \\ // «
40.2
39.3

I I
'1
o

•y
FIG. 8.—Comparison of Theory with Experiment for Collapse-Pulsing: Graded Sand
Media, D60 = 0.86 mm [1 (scfm/sq ft)2 = 334.52 (m/hr)2]
587

J. Environ. Eng. 1984.110:573-590.


theory and experiment is quite good. It needs to be reemphasized that
possibly the only coefficient that may be termed adjustable in the theory
was the pore size; thus, the experimental data seem consistent with the
assumptions made and the theory developed.

PRACTICAL DESIGN EQUATIONS


Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

Hewitt and Amirtharajah (7) reported that all of the data are reason-
ably represented by a single empirical correlation Eq. 29 which could be
used as a practical design equation. A mean linear relationship between
Ql and %(V/Vmf) can also be estimated from the theoretical Eqs. 31-34
by averaging the coefficients. The resultant equation is

0.45 Ql + % i l l =41. (35)


v,mf,
This equation applies for %(V/Vmf) values of 20 to 41.9. This practical
design equation has been plotted in Fig. 9 with all of the experimental
data. Also included in Fig. 9 are the data for the third sand of size D60
= 1.54 mm.
As a comparison with the preceding theory, a linear regression of all
of the data in Fig. 9 gave the following equation with a correlation coef-
ficient of 0.88:
V_ (36)
0.45 Qi+[% = 43.0
V,mf,
This equation is also shown in Fig. 9 as the dashed line.
Any of the three equations, Eqs. 29, 35 or 36, may be used for pur-
poses of design of air scour systems for collapse-pulsing. They would
all give similar design values. However, Eq. 35 seems preferable since
it is derived from Eq. 27. As evident from Figs. 7-8, application of Eq.

D
@ 6o~ ° - 6 2 mm 18 In. sand depth
60
— A D
60= ° ' 6 2 mm
3 0 In. sand depth
A @ D 6 _ = : 0 . 8 6 mm 1 8 In. Band d e p t h
B D 6 Q = 0 . 8 6 mm 3 0 In. sand depth
40
^n 18 In. sand d e p t h
^ D 1 . 5 4 mm
"^•^•^ A
6 0

> ©
30 - -8
from theory ^

©
20
/ ^ " ^
from linear regression

10 I I i _ '

(AIR FLOW) , Q Ucfm/ft )

FIG. 9.—Comparison of Data with General Design Equations [1 (scfm/sq ft)2


334.52 (m/hr)2]
588

J. Environ. Eng. 1984.110:573-590.


27 is probably the best approach for different types of media and depth,
since it has a fundamental and general basis.
SUMMARY

The dynamics of air scour from a fundamental and microscopic view-


point has been analyzed theoretically. The complex flow patterns which
Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

emerge when air and water flow concurrently through porous media are
analyzed in terms of the air pressure within the bubble and the soil stresses
and pore water pressures within the solid-liquid matrix. This conceptual
analysis leads to an original theoretical equation (Eq. 27) which predicts
the condition of flow in 3-phase systems wherein air pockets form and
collapse within the bed (collapse-pulsing). Rather remarkable agreement
is obtained when the theory is compared with experimental data. The
possibly cumbersome theoretical equation has been condensed into a
simple design equation of the form

0 . 4 5 Q 2 + ( % - ] - ) =41.9 (Eq.35)
V Vnif/
in which Q„ = air flow rate in standard cubic feet per minute per square
foot and %(V/Vmf) = percent minimum fluidization velocity.

ACKNOWLEDGMENTS

This study was partially supported by the Engineering Experiment


Station at Montana State University and the National Science Founda-
tion under Grant No. ISP-8011449. All of the data used in the paper were
collected by Steven R. Hewitt, former Graduate Research Assistant at
Montana State University, whose contribution is gratefully acknowl-
edged. Helpful discussions with Alfred Scheer and Bruce Suprenant on
soil mechanics are also acknowledged.

APPENDIX I.—REFERENCES

1. Amirtharajah, A., Morrison, R. J., and Holnbeck, S. R., "The Mechanics of


Air Scour During Filter Backwash," 1981 Annual Conference Proceedings Part I,
American Water Works Association, May, 1982, pp. 387-404.
2. Amirtharajah, A., "Optimum Backwashing of Sand Filters," Journal of the En-
vironmental Engineering Division, ASCE, Vol. 104, No. EE5, Oct., 1978, pp. 917-
932.
3. Davidson, J. F., and Schuler, B. O. G., "Bubble Formation at an Orifice in an
Inviscid Liquid," Transactions of the Institution of Chemical Engineers (London),
Vol. 38, 1960, pp. 335-342.
4. Davidson, J. F., and Schuler, B. O. G., "Bubble Formation at an Orifice in a
Viscous Liquid," Transactions of the Institution of Chemical Engineers (London),
Vol. 38, 1960, pp. 144-154.
5. Fair, G. M., Geyer, J. C., and Okun, D. A., Water and Wastewater Engineering,
Volume 2, John Wiley & Sons, Inc., New York, N.Y., 1968.
6. Hewitt, S. R., "Air Motion Through Filter Media During Air Scour," unpub-
lished CE 575 Report, Department of Civil Engineering and Engineering Me-
chanics, Montana State University, Bozeman, Mont., 1982.
7. Hewitt, S. R., and Amirtharajah, A., "Air Dynamics Through Filter Media
During Air Scour," Journal of Environmental Engineering, ASCE, Vol. 110, No.
3, June, 1984, pp. 591-606.
589

J. Environ. Eng. 1984.110:573-590.


8. Lambe, T. W., and Whitman, R. V., Soil Mechanics, John Wiley & Sons, Inc.,
New York, N.Y., 1969.
APPENDIX II.—NOTATION

The following symbols are used in this paper:


Downloaded from ascelibrary.org by New York University on 05/10/15. Copyright ASCE. For personal use only; all rights reserved.

a = constant;
b = constant;
D60 — 60% passing sand size = effective size x uniformity
coefficient;
d = diameter of particles;
dh/dZ = head loss gradient or hydraulic gradient;
8 = acceleration due to gravity;
K = depth of water above media;
h = head loss;
i = hydraulic gradient;
K = coefficient of lateral stress;
k = coefficient;
k',k" = coefficients;
N0 = normal forces;
Po = tangential forces;
Pi = air supply pressure;
Qa = air flow rate;
r = radius of bubble;
ss = specific gravity;
T = surface tension;
u = pore-water pressure;
Ui = dynamic pore water pressure;
V = subfluidization water velocity;
vmf = minimum fluidization water velocity;
(%V/Vmf) = percentage of minimum fluidization velocity;
z == depth
lb
of sand;
submerged or buoyant unit weight;
Is = unit weight of sand;
1w = unit weight of water;
It = (it + yw) ~ unit weight of sand plus saturated water;
e = expanded bed porosity;
e
m/ = porosity at minimum fluidization;
e0 = static bed porosity;
V- = absolute viscosity;
Pi = mass density of liquid;
cr = total stress;
°y = normal stress at failure;
O-H = total horizontal stress;
* H = effective horizontal stress;
(JV = total vertical stress;
vv = effective vertical stress;
T = shear stress;
T
/ = shear stress at failure;
4» = angle of internal friction; and
* = sphericity.
590

J. Environ. Eng. 1984.110:573-590.

You might also like