1. sports
2. bobsled

ThammasatInt. J. Sc.Tech.,Vol.5, No.2, May-August2000
How to Estimateand Designthe Filter Run
Duration of a Horizontal-FlowRoughing
Filter
Dome Sittivate
Dept. of Rural Technology, Thammasat University,Pathumthani,Thailand'
Abstract
methodsusedbeforeslow sandfilters(SSF),horizontal-flowroughing
Among pre-treatment
for applicationin developing
filters(HRF) have beenfound to be the most effectiveand appropriate
parts,
have low capitalcost,can be
mechanical
require
no
countries.This is becausethey are simple,
with a wide varietyof raw
capacity,
and
retention
solids
their
high
to
long
time
due
for
a
operated
on turbidity (solids)removal
Most researchon HRFs has concentrated
surfacewater characteristics.
whereasthe way to estimateand designthe run durationtime beforefilter (HRF) cleaningnecessary
a largegravel(size l0-20 mm) filter mediawere placed
hasreceivedlittle attention.In this research,
filtersused
as a singlepackage,uncoveredand outletat the top. It was foundthat the laboratory-scale
over time,
of
filter
run
to
the
estimation
with
respect
physical
model
in this study provideda useful
This
modelis
practical
field
experience.
with
agreement
in
reasonable
be
found
to
were
andthe results
basisfor the designof full scaleHRFsand is easyto apply.
a reasonable
Keywords: Algae removal, motile algae,HRF (horizontal-flow roughing filter), filter run
duration, turbidity
l.Introduction
U n d e rt r o p i c acl o n d i t i o n so,n ep o s s i b i l i t y
which has been used as pre-treatmentbefore
slow sand filtration for reducing the algae
contentof raw wateris horizontal-flowroughing
f i l t r a t i o n( H R F ) t l l . H R F i s n o t o n l y u s e df o r
improvingthe physicalwaterquality in orderto
but also
meetthe slow sandfilter requirements
ranging
viruses
and
bacteria
for rernovingsome
pm
and 0.4
20
l0
to
in sizefrom approximately
rnost
cornrnon
The
to 0.02 pm, respectively[2).
type of HRF, which has beentrsedwidely, was
d e v e l o p ebdy W e g e l i n[ 3 ] i s s h o w ni n F i g u r el .
FurlhermoreSittivate [4] found that the large
gravelmedia(size l0-20 rnm) which were ttsed
in the filterswith the outletat the top of an HRF
( s i m i l a rt o W e g e l i nt 2 l , t 3 l ) p r o d L r c et dh e b e s t
resultsfor algae and turbidity retroval (tnore
than 95Yo and 90o/ootl average.respectively)
and sedimentationwas consideredto be the
for the rernoval
principalmechanismresponsible
of algaeand turbidity in raw water at the low
filtrationrate(0.3 m/h)
The objectiveof this researchwas to
a
devise simple method by which horizontalflow roughingfilterscould be designedfor field
applicationunder tropical conditionsand their
long-term performance predicted' This
essentiallyrequires an understandingof the
processvariables,knowledgeof the perlinent
of removal of particulatematter,
rnechanisms
includingalgae,andthe availabilityof equations
to describe the time-space variation and
accumulationof materials inside the filter
rnedia.
Many theoreticalequationshave been
developedto describethe filtrationprocessand,
particularly,to try to estimatethe time of run
duration for HRF design However, some
parameters
and constantsin theseequationsare
ruotuniversaland will, no doubt,vary with each
2000
Int.J. Sc.Tech.,Vol.5,No.2,May-August
Thammasat
combination of influent characteristicsand
fi ltrationconditions[5].
Moreover,when designershaveapplied
such equationsthey have had to make many
assumptions of those parameters for the
estimationof results.
It is known that the diversityof removal
mechanismscauses the flow of suspended
particlesthrough porous media to be a very
In the presentstudy,the
complexphenomenon.
run durationtime of the HRF model was rather
short. Therefore,it is logical to considerthe
main removal mechanismis sedimentationof
the incomingsolidsand algal massin the large
gravelmediafilter [4].
In spite of the difficulty of accounting
for all variables,it was decidedto usethe results
obtained in the experimentsof this study to
developanotherapproachto designthe HRF anc
estimatethe run duration before cleaning is
necessary.
i,li___f<-ffii__+_ffi_l
A H
+
I-+a
a
I"
Drainvalves
List of symbols
designguidelines
dg
(mm)
gravelsize
vr:
H
(m)
filter depth
Vd:
Lr,z.t
W
A
a
Qd
VF
Vd
(m)
(m)
(rn')
(m3/h)
(m3/h)
(m/h)
(m/h)
filter length
filter width
area
filter cross-section
flow rate
drainagerate
filtration rate
drainagerate
Q
: a
:0.3-1.5m/h
A
-60-90m/h
Q
(Lr+L2+L3).W
H.W
AH-30cm
H -0.8-1.20m
Figure l: Layout and Designof a Horizontal-flow Roughing Filter
(Wegelin,1996)
2. Experimental Procedure
2.1 Pilot Plant
The pilot plant system [4] slrown in
of the following:(l) A plastic
Figure2 consisted
filter box composed of two symmetrical
of 1.6m x
channelswith dirnensiorrs
rectangular
0.39 m x 0.195 m each:the lateralwalls of the
filter box were fitted with sampling ports. (2)
Two plastic cylindrical tanks (total volume of
one cubic metre) with an internaldiameterof
I .60 m and heightof 0.50m; thesewerethe feed
tankswhich containedalgaeandclay to simulate
ThammasatInt. J. Sc.Tech.,Vol.5, No.2, May-August2000
tropical surface water. (3) Two stirrers for
mixing the suspensionin the large tank. (4) A
light sourceconstructedfrorn lightweightsteel
bars to carry fluorescent tubes; this was
suspendedover the two large tanks. (5) Two
peristaltic pumps were used to feed the
from the algaeinoculationtanksinto
suspension
the filters.
The suspensionof turbidity and algae
from the largetankswas pumpedinto the inlet
chambersof each filter, passedthrough the
gravel media and dischargedinto the outlet
chambersof each filter. During the filter runs,
effluentwater sampleswere analysedfollowing
the method of Standard Methods for
Examinationof Water and Wastewater[6],
mainly to determine the DO concentration,
turbidity and pH. The chlorophyll a
concentrationwas also determinedas a measure
ofthe algaeconcentration.
The pilot studyplant
was installed in a specific room where the
temperaturewas controlled between 19 and,23o
so as to maintaingood environmental
condition
for algal growtlr in the largetanks [4].
Light
Figure 2: A schematicdiagram of filtration equipment(not to scale)
2.2 Design of Synthetic Raw Water
with a high algae
To preparea medir"rm
contentin tlre preparationof an appropriateraw
water, one factor which was consideredwas
turbidity. High turbidity in water inhibits
photosynthesis
and thus reducesthe production
of oxygenby algaeas well as algal growth [7].
Two types of motile green algae (Euglena
gracilis and Chlamydontonasreinhardtii) were
chosenas representative
algaefor this research
Wegelin
suggested
that the possibilityof
[4].
[3]
annualturbiditymaximumin raw watersources
in tropicalareasis normallynot higherthan 100
NTU. As a result, it was decidedto set the
tLrrbidityof the syntlreticraw waterat 100NTU
for every run in this research.
Algae inocLrlation
in the largetank usedthe rnethodssuggested
by
The
Freshwater Ecology
Institute of
( A m b l e s i d eU
, K ) . T h e r e q u i r e dt u r b i d i t yw a s
achievedby the additionof a known weight of
clay, which would produceturbidity 100 NTU
in the water in the largetank and maintainedin
suspension
by continuousstirring [4]. The type
of clay usedwas kaolin clay (Speswhitechina
clay) supplied by ECC International Ltd,
C o r n w a l lE
, ngland.
2.3 ExperimentalPlanning
Experiments as shown in Table I
and in the Filter diagram (Figure 3) were
performed on the 1.60 m long channel. The
filtration rate was 0.3 m/h throushout the
study.
Thammasatlnt. J. Sc.Tech.,Vol 5, No.2, May-August2000
Table 1: Planningof exPeriments
Filter
Run no.
lnfluent turbiditY
Algae
(NTU)
type
B
J
4
125 to 1 9 5
E
E
E
2 3 0 to 2 3 5
C
E&C
tz5
to 1 9 5
lnfluent algaecontent
Gravel
Outlet
during the run (pg/l)
size(mm)
position
L
T
t'
6.49to 3'7.66
uell
6.49ro 37.66p.p.ll
M
L
145to 198
l 9 . 4 8 t o1 5 1 . 9 5
ue/l
8 6 1 0t o 3 6 3 1 2 $ e / l
I 50to 260
23883to 360ue/l
L
L
T
T
reinhardtii'
Notes:A: Filter A, B = Filter B. E= Euglenagracilis,C= Chlamydomonas
- l0 mm'
5
s
i
z
e
d
g
r
a
v
e
l
,
M
e
d
i
u
m
M
:
r
n
m
.
L : L a r g e g r a v e ls, i z e d1 0 2 0
T : Outletat the top, 0.I 6 m from the bottomof the filter bed'
solids(SS)andgravelmedia
by suspended
Cloggedfilter volume= Filtervolumeaccumulated
:
by
algae
[4]'
Inleilurbidity(NTU) clay turbidity+ tLrrbiditvproduced
fFtffil
-+
I h tlter tt
I
->
Outlets
Run nos.l to 4.
passthrough filters
Figure 3: Plan view of filter showingflow of syntheticraw water
3. Results
Throughoutthe study'the filter volumes
i n w h i c h s u s p e n d esdo l i d sa n d m i c r o o r g a n i s m s
accumulatedin the pore spacesof the filter
media over time were measured Linear
of the dimensionsof the clogged
measurements
areaprofile along the side of both filters were
made at various times in every rlln and
converted into volumes of the filter b;'
multiplicationof the filter width (0 195 m)
Thesedataare given in Tables2 to 6 Note that
this total volume is composedof the volume of
solids
and
media
Particles
gravel
filter
the
occupying
clay)
and
fmicroorganisms
mediaporespaces.
media (size-l0-20 mm) filter
Table 2: Accumulationof solidsmassin the large gravel
:
volume in Run No'1: Algal culture Euglena gracilis
Accumulated
time
(h)
JJ
)o
96
126
242
365
484
55'7
Cloggedfilter
volume
(crn')
Time
difference
(h)
Clogged filter
volume
difference(cmi)
4,09).UU
7,020.00
9 . 74 0 2 5
1 3 . 3 0785
t4,820.00
22.985.63
28.255.50
35 , 0 1 2 . 3 5
23
40
30
Ir6
lL)
I l9
I J
2.925.00
2,720.25
3,568.50
1 . 5I 1 . 2 5
8 ,r 6 5 . 6 3
5,269.87
5
6.'756.'7
average :
Clogged filter
volume
difference /
differencetime
(cm3/h)
Influent
turbidity
(Nru)
1 2 ' 7t 7
68.01
r1 8 . 9 5
13 . 0 3
40
t70
44.28
92.56
t20
t45
125
15.17
152.14
66.19
ThammasatInt. J. Sc.Tech.,Vol.5, No.2, May-August2000
Table 3: Accumulationof solidsmassin the medium gravel media (size5-10 mm) filter
volume in Run No.l: Algal culture = Euglena gracilis
Accumulated
time
(h)
33
56
96
126
242
365
484
557
605
Clogged filter
volume
(.,n')
Time
difference
(h)
Clogged lilter
volume
difference
(cm')
Clogged filter
volume
difference /
difference time
(cm3/h)
3.968.25
6 1t1.t'l
z)
6.669.00
40
30
l16
I 1,407.50
14,430.00
17.788.88
2t ,498.7
5
23.75t.00
26,822.2s
I z)
ll9
48
2.344.88
355 . 8 7
4 , 7 38 . 5 0
3.022.50
3 , 35 8 . 8 8
3.709.87
2.252.25
3 , 0 7.12 5
average =
r01.95
8.90
t57.95
26.06
27.31
31 t 8
30.85
Influent
turbidity
(NTU)
95
70
40
70
zv
63.98
45
25
20
56.02
148.13
Table 4: Accumulationof solidsmassin the large gravel media (size10-20mm) filter
volume in Run No.2: Algal culture = Euglena gracilis
Accumulated
time
(h)
28
52
82
9'7
t45
Clogged filter
volume
(cm')
25.072.13
32,048.2s
44,850.00
49.530.00
6t,t61.75
Time
difference
(h)
Clogged filter
volume
difference
(cmr)
Clogged filter
volume/
difference time
(cm3/h)
Influent
turbidity
(Nru)
235
24
30
l5
48
69' 76.12
12,801.75
4.680.00
II,631.75
average:
290.67
426.73
3t2.00
z+z.) )
230
225
230
3r7.93
230
Table 5: Accumulationof solidsmassin the large gravel media (size10-20mm) filter
volume in Run No.3 : Algal culture : Chlamydomnasreinhardtii
Accumulated
time
(h)
53
79
t27
223
3t9
415
487
Clogged filter
volume
(.'n')
Time
difference
(h)
Clogged filter
volume
difference (cmr)
r6.32r .50
2t,67
4.25
29.250.00
33.783.75
38 . 2 8 8 . 2 5
55,282.s0
6 0 , 0 9 4l 3.
26
Clogged filter
volume
difference /
difference time
(cm3/h)
Influent
turbidity
(Nru)
I 57 . 8 3
98
45
50
4.533
. 75
lt.zJ
45
4.504.50
46.92
16.994.25
t77.02
66.83
45
45
50
116.95
54
5
'157
75
7 5 7 57 5
48
96
96
96
72
4 , 8 11 . 6 3
average =
205.88
2000
ihammasatInt.J. Sc.Tech.,Vol.5,No.2,May-August
Table 6: Accumulation of solids massin the large gravel media (size 10-20mm) filter
volume in Run No.5: Algal culture = Euglens & Chlamydomonos
Accumulated
time
(h)
78
05
30
50
9l
Time
difference
(h)
Clogged filter
volume
difference (cm3)
5.533.r3
42.237.00
47,s80.00
27
25
20
6l.l6l.7s
4l
Clogged filter
volume
(.'nt)
Clogged filter
volume
difference /
difference time
(cm3/h)
(Nru)
260
23,868.00
29.40r.13
Influent
turbidity
l2.835.88
5.343.00
l 3 . 5 8.17 5
average =
204.93
513.44
267.15
33t.26
2r0
329.20
201
200
185
150
b) Simplifiedword statement:
4. Discussion
4.1 Progressionof Filter Volume Occupied by
SuspendedSolids and Filter Media
In HRFs, the filtration rate is assumedto
remain constantthroughouta filter run and the
type of flow in filter media is plug flow
[3].Tchobanogloust8l suggested that, in
of the
general,the mathematicalcharacterisation
within
matter
particulate
of
removal
time-space
the filter is based on a considerationof the
equation of continuity, together with an
auxiliary rate equation. The equation of
continuity for the filtration operation may be
developedby consideringa suspendedsolids
mass balance for a section of filter of crosssectionalarea,and ofthe thickness,measuredin
the directionof flow. The massbalancewould
be as follows:
= [inflow - outflow] -
Accumulation
(2)
degradation
However, it was found that the degradationin
HRF occurred very little [a]. Hence, the last
term in equation(2) is negligible.
of equation(1) [8]:
c) Symbolicrepresentation
d V
ac
d
:-a
0c dx
t
(3)
A
where,
differential volume of reactor or
dV
filter (L3)
dcldt = rate of change of suspendedsolids
within the containeror
concentration
filter (ML-3Tr)
a) Generalword statement:
a
Rateof accumulation
of solidswithin the :
volumeelement
Rateof flow of
solidsinto the
volumeelement
Rateof flow of solids
out of the volume
element
(1)
:
dCl&
:
dx
C
:
:
volumetric rate of flow -into and out
of the reactoror filter (L' T-')
changein concentrationofsuspended
solids in fluid stream with distance
(MLr.L)
differentialdistanceof filter (L)
concentrationof suspendedsolids in
reactoror filter (ML")
Thammasat
Int.J. Sc.Tech.,Vol.5,No.2,May-August
2000
If A dx is substituted for dV. the resulting
This can be rewritten;
expressionis
vz) (ya),f" - (vrXy,'),f"
A d x 0 C : - a d C dx
dt
T
(4)
[(cssrz- cssorz)tz- (cssrr- cssorr)tr]
where; I = cross-sectional
areaof filter (L2)
Therefore,equation(4) may be rewritten:
A&dc :- Qac
dt
(s)
Vr
V2
Ttz
where;
"f"
ytr
At
:
At
:
weight of suspendedsolids retained in
the filter mediaat any time (M)
differenttime (T)
Therefore, equation(6) can be rewritten in a
form suitablefor numericalanalysis:
AP,
tz-tr
= QAC.
or
- ( C s s , rC
P-rz-:&': Q [(Css,z- sso,r)]
Csso,z)
tz-tr
(7)
where;
= weight of suspendedsolids retained
Pt
in the filter mediaat any time (g)
= flow rate through the cross-sectional
a
areaof filter (m'/h)
Css,r : concentration
of suspende-d
solidsin
the influentat time tr (g/m')
Css,z = concentration
of suspended
solids in
the influent at time tz(g/m')
Csso,r = concentration
of suspended
solids in
the effluentat time t, (g/mt)
Csso,z= concentration
of suspende^d
solids in
the effluentat time h (g/m')
: cross-sectional
A
areaof filter (m')
= filtration rate (m/h) , t: time (h)
Vp
(g)
where;
It can be said that the first term ofequation(5) is
the differentweight of solidsaccumulatedin the
pore spaceof the filter media over the different
time. Hence,equation(5) may be rewrittenas an
ordinarydifferentialequation:
dP:- Qdc
(6)
Pr
A vpx
:
filter volume occupied by suspended
solidsandfilter mediaat time t1(cmr)
: filter volume occupied by suspended
solidsand filter mediaat time t2 (cm3)
= total unit weight of solids mass
accumulatedin the pore spaceof filter
mediaat time ty(g/ cmr)
= total unit weight of solids mass
accumulatedin the pore spaceof filter
mediaat time t2(g/ cmr)
= porosity of the bed when clean (Vo /
v)
The time difference between t1 and t2 is not
much.Hence,it can be said that Ttz-Ttt.=ft
Therefore,equation(8) may be rewritten;
.f" (yJ [Vz - Vr] = A vr [(Cssrz- Cssorz)rztr-tr
(Cssrr- Cssor,)tr]
(e)
where;
It
total unit weight of
solids
massaccumulated
in the pore space
of filter media
The filter volume(Vz , V,) and porosity
(/" ) are easily measuredbut it is very difficult
to measurethe total unit weight of solids mass
accumulatedin the pore spaceof filter media
(yJ. This is becausethe suspendedsolidscontent
will be differentat any depth or at any length in
the pore spaceof the filter media of the HRF,
and at any time as influencedby the drag force
of fluid, velocity of flow throughthe pore space
of the filter media,gravity and the characteristic
diameterof the solidsparticles(Fair et al., 1968)
during the run duration. So the first term of
equation (9) cannot be used to estimate the
solids mass retainedin the filter. However, all
Int. J. Sc.Tech.,Vol.5,No.2, May-August2000
Thammasat
parametersin the last term of equation(9) can
be measuredand used to estimatethe solids
massretainedin the filter media
Hence, it follows that (V2 - V1) is
occupiedby gravelmediaand solidsmassA vp
tu- (Css,r- Csso,r)t1].Therefore,
[(Css,,- Csso,z)
the rate of filter volume occupiedby gravel
mediaand solid mass,per time, is equivalentto
the differencein weight of suspendedsolids
mass accumulatedin the pore space of filter
mediaover the time change,dividedby porosity
of gravel media and total unit weight of solids
mass accumulatedin the pore spaceof filter
the
media.Hence,equation(9) afterrearranging
terms.vields:
Vz - Vr
tz- tr
=
- Csso,z)
tz- (Css'rA vr [(Css'z
(y,)
,f"
Hence,with the outlet at the top and a
filtration rate of 0.3 m/h (at average influent
turbidify = [230 + 154 + 201] I 3: 195 NTU),
the averagechangeof filter volume over time
accumulatedby gravel media and suspended
solids mass(clay and algal mass)in the large
gravel media (size l0-20 mm) filter is 250
cm'/h. Hence, the average change of filter
volume accumulatedby gravel media and
solidsmass(clay and algal mass)in
suspended
the large gravel media filter pore space over
t i m e ( S e e e q u a t i o n( 1 0 ) ) i s 1 1 0 c m ' / h ( : l J Q
crn'/h * porosityof gravelmedia= 250 cmrlhv
0.4372)
This rate could be the basisof both the
design of HRFs, and estimatesof the run
durationtimesfor scaledup HRFs.
4.2 Confirmationof Results
(10)
c s s o , rt ), l
Sittvate [4] indicated that under the
same operatingconditionsas adoptedfor the
large gravel media filter, the medium gravel
(size 5-10 mm) filter with lesspore spacethan
the largegravelmediafilter, was blockedfaster
than the largegravelmediafilter. This resulted
in the duration of a filter run for a rnedir.rm
gravel media filter becomingshorterthan that
for a large gravel media filter. Furthermore,
becausetaking many samplesalong the bed of
the large gravel filter over time drew out clay
from the pore spaces,as observedin the first run
it was decided,therefore,not to use the result
from Table 2 and 3 in this analysis.For the
largegravelmediafilter model,the filter volume
solidsand filter mediaat
occupiedby suspended
differenttimes,from tables4 to 6, were 317.93,
I I 6.95 and 329.20 cm'/h, respectively.Irr
addition,Sittivate[4] found that the patternsof
the removal characteristicsfor Euglenagracilis
and Chlamydomonas reinhardtii in HRF were
similar. Hence, the average filter volume
solidsand filter rnediaat
occupiedby suspended
differenttimes
3t7.93+116.95+329.20
3
254.69 - 250cm3lh
C h e c k i n gt h i s v a l u e( l l 0 c m r / h ) w a s
achievedby estimatingthe run durationtime of
the large gravel filter model which normally
failed by solidsblockingalong the filter length
i n l 0 d a y s[ 4 ] ;
The effective volume pore space of filter
modeloccupiedby largegravelmedia
:
1 9 . 5x 2 0 x 1 5 3x 0 . 4 3 7 2
:
26,087.72cm3
..
R u nd u r a t i o nt i m e = 2 6 . 0 8 7 . 1 2 1 1 1x0 2 4
:
4,3
9 . 8 8d a y s - l 0 d a y s
Scale-upConsiderations
As shown in Section 4.2, for the
conditionsof filtration velocity (vr) : 0.3 m/h
and with the filter outlet at the top, it was
that:
calculated
R (day) = (Volurneof filter mode-loccupied
116 lsrn'/h) x 24
by gravelmedia(cmr)(/") )
(11)
where:
Run duration (days) before filter
cleaningis necessary.
Thammasat
Int. J. Sc.Tech.,Vol.5,No.2, May-August2000
where;
Let,
A,n =
L,n
w,n :
D r :
Qo :
The cross-sectional
areaof filter model
( : 1 9 . 5x 2 0 c m 2 )
The lengthof filter model(internal)
The width of filter model(internal)
The depthof filter model(internal)
porosityof the bedwhenclean(V"i V1)
flow-rate through the cross-sectional
areaof filter prototype(m3/h)
Q,n = flow-rate through the cross-sectional
areaof filter model(mr/h)
Equation( l4) / Equation( 15)gives,
0;
( 1 1 )w i l l b e ;
H e n c ee, q u a t i o n
Ao.vr
:
Qo
=
n*u'
Ao
A-
therefore;
Qo = (AolA,,).Q,
Ap/A,n = rnr
1 1 0( c m 3 / hx) 2 4
or
:
(W,n.D,n.)x L,,,(/o)
2,640(cm3lh)
a
therefore, Qo
and,
Q,n
=
(17)
(12)
where;
This filter model will also fit the
prototype (the large-scale design for field
application), on the basis of an irnplied
confidencein the principlesof geometricaland
dynamicalsimilarity. Henderson[9] suggested
that the detailed interpretationof model
measurementsrequires that scale ratios be
available for translating model values for
variousquantities,e.g.,velocity,discharge,
etc.,
prototypevalues.Scales
into the corresponding
can be deducedfor all physical quantitiesif
scalesare known for mass,length,tirne,and the
physical propertiesof prototype and rnodel
fluids.
The prototype will have conditions
similar to the filter model in this study, for
instance the low filtration rate (0.3 m/h),
position of the outlet, gravel media size, and
irrfluerrtturbidity.The scaleratio of this rnodel
and the prototype can be obtained frorn the
cross-sectional
areaof the prototypedividedby
the cross-sectional
area of the model. This is
becausethe filtration rate (ve) in the designof
the prototype is equal to that irr the model and
the depth of the filter is not significantin the
f i l t e r b e h a v i o r s[ 0 ] . H e n c e .t h i s s c a l er n o d e l
ratiocan be calculatedas follows:
From
(l6)
:
rn1
scaleratio of protofypeand model
Hence, for the large scale of HRF (prototype)
with tlre sarneconditionsas in the model, the
run durationcan be estimatedby;
:
R
Ao.Lp (,f")p
(t8)
2,640 (m1)
wnere:
: the lengthof filter prototype(m)
Lp
(,f")o = porosity of the bed of filter prototype
whenclean(V"/ Vr)
It was found that R ccli vr. and R cc l/
influentturbidity(NTU) [2], therefore,equation
(18) could be modified by these proportional
relationships
for use in the designof a HRF, as
f o l l o w s[ 9 ] :
R :
Ao.Lo (.f")o
(NTU)p/(NTU),"
2,640 rn1(v1)o/(ve),"
when;
Av
(13)
trz
Ap.vr
( 14)
rn: :
A,n.Vp
( 15)
= (ve)o/(vp),,,
(NTU)p/(NTU,.
(le)
ThammasatInt. J. Sc.Tech.,Vol.5,No.2, May-August2000
the filter media are assumedto not change.
This equationshould be used with the low
filtration rate of 0.3 m/h for the scaled up
designas in this study,becauseit was found
that, with thesefilters used in this study, both
algae and turbidity removal were more than
90oh at this filtration rate. Moreover, this
filtration rate was found to be suitable for
operationand has producedturbidify removal
efficiencies more than 80% in field
i n s t a l l a t i o n( [s3 ] a n d[ 1 l ] ) .
Equation( l9) becomes,
R
:
Ao.Lo.(/")o
2-640. mr "t" m,
(20)
4.4 Model validation
From equation(20), it can be saidthat
the filtration rate and turbidity are the main
parameterswhich affect the run duration and
turbidity removal efficiency as Wegelin [2]
found. If the gravel media is chosenas in
Wegelin [3]'s designguidelines,the porosity
of the gravel media size 10-20 mm will not
affectthis equation.As a result,this model is
valid for use in HRF design,for the filtration
rateadoptedand up to the influentturbidityof
raw water usedin the studY.
4.5 Confirmation of the Physical Modelling
Although equation (20) has been
developedto be a simple tool for the design
and estimationof the run durationtime of a
HRf', it is still not proven on the larger scale.
However,the horizontal-flowroughingfilters
usedin the Blue Nile HealthProjectin Sudan
[2], were similarto this modeland the results
reportedby Brown can be usedto confirm the
applicabilityof equation(20). A plan view of
the HRF seriesusedin the Sudanis shown in
F i g u r e4 .
Brown (1988) rePortedon two runs
wherethe run durationof the filter (Figure4)
in the
when solidsaccumulated
was terminated
filter media pore spaces to the point of
breakthrough.In the first run, with a filtration
rate of 0.4 m/h and averageinfluent turbidity
300 NTU, the run durationtime was 20 days.
In the secondrun (after cleaning),with average
filtration rate 0.45 m/h and averageinfluent
turbidity 312 NTU, the run durationtime was
I 4 days.
Lebcir[10] alsofoundthat:
- Particles are removed mainlY bY
and the removal is slowed by
sedimentation
increasingthe filtration rate or Reynolds
Number.
-Laminar flow conditions prevailed
when the filtration ratewas lessthan I m/lr and
2 mlh for large gravel media filter (LGF) and
small gravel filter (SGF), respectively(both
outlet at the top).
- The patternof the turbiditydistribution
insidethe bed is in conformitywith that of the
sedimentationtanks at flow velocitiesbetween
0.5 - I m/h. This is becausethese flow
velocities are low enough to not cause
turbulencein the gravelfilter media.
- For the depthofbed channel,studiesto
date merely state the problems of structr-rral
constraints that can be faced with deep
channelsandthey do not give any indicationof
whether the depth will influence filter
behaviour.
Hence,equation(20) could be valid if
the filtration rate doesnot exceedI m/h for the
largeand medium gravelfilter mediaand the
other conditionsare as in tlris study.This is
within this rangethe conditionsinside
because
From equation(20); for the first run,
Ao.Lo
(,f")o
2,640 m1.Flz.lrlr
np
1Tl1
l0
1 x 0 . 9 5x 1 0 , 0 0 0c m 2
0 . 9 5 x1 . 0 0 x1 0 , 0 0 0 : 2 4 . 3 6
19.5x 20
Ill2
( V F ) p / ( V F ) *= 0 . 4 / 0 . 3 :
lTlr
( N T U ) p / ( N T U ) .: 3 0 0 / 1 9 s :
1.33
1.54
ThammasatInt. J. Sc.Tech.,Vol.5,No.2, May-August2000
EE
BROKENNILE BRICKS
f.:0.575
{-VF
-rsffirsffi,,ffi,ul*_
Figure 4: Plan view of the Horizontal Flow Roughing Filter used in Blue Nile
Health Project in Sudan [12] (not to scale)
Note: The depth of this filter was 1 m.
Substituting
thesevaluesin equation(20),
R :
W
providesthe estimatedrun durationtirne for the
first run;
R :
I x 0 . 9 5x 1 , 0 0 0 , 0 0 0 ( 2 . 9 x 0 . 5 715x+
0.40+1x0.37)
I x 0 . 9 5x 1 , 0 0 0 , 0 0 0 x ( 2 . 9 x 0 . 5 7I 5x +
2 , 6 4 0x 2 4 . 3 6x 1 . 3 3x 1 . 5 4
:
15 days (cf. 14 daysin Brown's results)
0.40+1x0.37)
:
17.58days
(cf. 20 days in Brown's
results)
For the secondrun,
rnz :
( V e ) p (i V r ) . : 0 . 4 5 l 0 . 3 :
1.5
rn: = (NTU)'/(NTU),': 3l2l 195= 1.6
Substituting
thesevaluesin equation(20),
providestlre estimatedrun durationtime for the
secondrun;
It would appearthat equation(20) gives a
reasonable agreement with practical field
experienceand the estimation of HRF run
duration obtained from the equation is a
reasonable
basisfor design.As a resultof this.
Figure 5 was constructedin order to provide
someguidelinesfor designand estimationof the
run durationfor the large gravel media (10-20
mm) HRF operatingunder the conditionsas
shown in Figure5. For eachgiven turbidityand
filter length,the run durationin dayscan be read
off. For example,with inputturbidity(including
algaeand suspended
solid particles)150 NTU
and a filter lengthof 7 m, the corresponding
run
durationtime is sixty days(Figure5).
95
ThammasatInt. J. Sc.Tech.,Vol.5, No.2, May-August2000
Run duration (daY)
360
340
320
300
____
(T__________1_
ll
t n t . lt l
_____7
HRF ll
llouttet
280
260
240
220
200
180
160
140
120
100
80
60
t ----
40
20
0
9
1
0
1
1
1
2
1
3
1
4
15
16
17
18
19
Filter length (m)
5: Designguidelinesfor large gravel media filter (sizel0-20 mm) with outlet at the Top'
Figure
porosity (/") = 0.43,filtration velocity:0.3 m/h, depth of water level: 1.0 m'
grave|mediafi|terdepth-l.05m,andwidthofthefi|ter=1.00m.
5. Conclusion
Due to the fact that there was not
sufficient time over the period of the
programsto covera full statistical
experimental
evaluation of all design variables such as
areaof the filter,the
filtrationrate.cross-section
the
size of the filter
media,
filter
the
of
length
of raw water
rnedia,the irrfluentcharacteristics
( t u r b i d i t ye) t c . ,a c o m p r e h e n s i vd e s i g nm o d e li s
n o t p o s s i b l e .H o w e v e r .b a s e do n e x p e r i m e n t a l
findings and theoretical considerations,the
physicalmodelswhich have beendevelopedin
this study can provide significantimprovement
on previousdesign and could be used in the
designand to estimatethe run durationtime of
HRFs.
6. References
W e g e l i n ,M . , R o u g h i n gG r a v e l F i l t e r sf o r
SuspendedSolids Removal,In Slow Sand
Filtration, Recent Advances in Water
TreatmentTechnolog,',Edited by N' J' D'
Graham,Ellis Hardwood,UK, pp. 103-122,
I988.
M' Horizontal Flow Roughing
Wegelin,
t2l
Filtration, a Design, Construction and
Operation Manual, IRCWD, Switzerland,
1986.
t3] Wegelin, M., SurfaceWater Treatmentby
RoughingFilters:A Design.Construction
and Operation Manual, SANDEC Report,
2196, Swiss Centre for DeveloPment
Cupertinoin Technologyand Management
(sKAT), 1996.
t4l Sittivate, D., Algae Removalfrom Surface
Roughing
Il'ater by Horizontal-flow
ill
Thammasat
Int. J. Sc.Tech.,Vol.5,No.2, May-August2000
Filtration,Ph. D. Thesis,Civil Engineering
Department,University of Newcastleupon
T y n e ,U K . , 1 9 9 9 .
l-51 Mohammed,S.T., Roughing Filtration of
Surface Water for Village Supplies in
Developing Countries, Ph. D. Thesis,
Universityof Newcastleupon Tyne, UK.,
I987.
t6l APFIA, Standard Methodsfor Examination
of Water and LVastewater,
(19't' edition),
AmericanPublic Health Association,New
York, 1995.
17l Carncross, S and Feachem. R.G.,
Environmental Hectlth Engineering in the
Tropics,John Wiley and Sons,New York,
1993.
t8l Tchobanoglous,G. and Burton, F.L..
Biofogical Unit Process in Wastewater
Engineering Treatment, Disposal and
Reuse,McGraw Hill, New York, pp. 88-90,
1991.
f
-
I
t_l
[9] Henderson, F.M., Open Channel Flow,
MacmillanPublishingCo., Inc.,New york,
1966.
[10] Lebcir, R., Factors Controlling the
Performance
of HorizontalFlow Roughing
Filters,Ph. D. Thesis,The University of
NewcastleuponTyne,UK., 1992.
l
]
L
l o y d , 8 . , P a r d o n ,M . a n d W h e e l e r .D . .
I
Final Report on the Development,
Evaluationand Field Trials of a Small
Scale Multi-stage, Modular Filtration
Systemfor the Treatmentof Rural Water
S u p p l i e sO. D A , L o n d o n U
, K., 1986.
2]
Brown,
D.,
Horizontal
Flow Roughing
|
Filtration as an Appropriate Pretreatment
before Small Slow Sand Filters in
Developing Countries,M.Sc. Dissertation,
Civil EngineeringDepartment,University
of NewcastleuponTyne,UK., 1988.