Indian Journal of Engineering & Materials SciencesVol. 4, December 1997, pp. 237-244

Defluor idation and empirical models in column studies usingfishbone charcoal

D S Bhargava" & D J Killedarb

a Department of Civil Engineering, University of Roorkee, Roorkee 247 667, India.

"Department of Civil Engineering, SGSITS, Indore, India

Received 24 September 1996; accepted 17 March 1997

Column bed adsorption studies have been carried out to study the fluoride removal on fishbonecharcoal, and to determine the effects of the various operating variables. The useful (or effective) treatedeffluent volume (corresponding to the desired breakthrough concentration of 1.0 mg/L of fluoride) is foundto be a function of the effluent flow rate, initial solute concentration and column bed depth. The usefultreated effluent volume decreased with an increase in the flow rate and initial fluoride concentration, but itincreased with the column bed depth. Empirical relationships have been developed to predict the stateduseful treated effluent volume for the known values of flow rate, column bed depth and initial fluorideconcentration for the observed test conditions. The relationships evolved manifest high correlationcoefficients. The studies are useful in small installations.

A comprehensive account of fluoride implicationsand permissible concentration in drinking wateralongwith a summary of fluoride removalmethods, materials, interferences, mechanisms,etc. is readily available elsewhere'<", The authorshave presented the numerous aspects of fluorideremoval through adsorption on fishbone charcoalin a batch system 16·22 as well as in a moving media

23-26 f dsornti . Thsystem 0 a sorption reactor operations. easpects of such an adsorption process using afixed-bed column operation involving a continuouseffluent flow which is more practical in real lifesituations, are not readily available in all itsvariations and analysis. This paper is intended tobridge such a gap.

In a continuous flow, fixed-bed columnoperation, the efficiency and system cost depend onthe removal capacity of the media, i.e., the amountof solute adsorbed per gram of the adsorbent. Thiscapacity is a function of several factors such as flowrate, column bed depth, initial solute concentration,pH temperature, regenerant used, and desiredquality of the treated water. In this study, the effectof some of these parameters, viz., the flow rate (Q),the initial solute concentration (Co), and the columnbed depth (D). on fluoride removal by fishbonecharcoal in a fixed bed with continuous flowoperation have been investigated. An attempt has

also been made to evolve possible inter relationshipsamong these parameters.

Experimental ProcedureA description for preparing the fishbone charcoal

h b d 'b d . I 16 20-26 d has een escn e previous y , an t echaracteristics of the prepared charcoal woulddepend on the kind of bone, its processing,carbonisation (temperature and time) and coolingscheme. Bhargava and Killedar24 presented acomprehensive economic analysis to show that thefishbone charcoal costs very little, and can beregarded as an economical (Rs. 11.00 per kg)

co.~C••uc::ou

~..::! 6.- G\

;; E 4:> •- ""'u

Co = 10.0 mg/L Ws = 8.0 Il/LTap water base (pH =8,0)

Distilled water ban (pH" 7.0)

e:>~..••cr:

2

O~~ __ ~ __ ~~~~~~ __ ~~~-L~o 120 140 360 480 ~OO

Time. min

Fig.I-Variation of the residual fluoride concentration withtime during the trial tests for interference by calcium and othercompeting ions, with Wsb=8.0 g/L, Co=1 0.0 mg/L, pH=8.0 (tapwater base) and 7.0 (distilled water) corresponding to batchrun tests

238 INDIAN J. ENG. MATER. SCI., DECEMBER 1997

adsorbent material in India. To simulate fieldconditions, the test fluoride solutions were preparedusing the tap water. The interference of thecompeting ions (such as sulphates and chlorides)present in tap water and also the presence of calciumin tap water was found to be negligible through atrial batch test run16.20.26(Fig. 1) . The test runs wereconducted at a temperature of 20±2°C. Thetest condition details for batch studies have beenstated elsewhere for studying the effects of variation. tu 16,21 d HI6,25In tempera re an p .

Fixed bed column studies were conducted with2.5 cm diameter glass column filled with fishbonecharcoal. Four sets of column studies wereperformed. In each set, for a given initial fluorideconcentration (Co= 2.5, 5.0, 10.0, or 20.0 mg/L) thedownward flow rates were varied and maintained at15 mLimin (3.06 mlzrnin/cm"), 30 mLimin (6.12ml.zrnin/cm"), 50 mLimin (J 0.2 mlzmin/cm"), and75 mLimin (15.3 ml.zrnin/cm") corresponding toeach of the various Co and column bed depth (D)values. Additional runs were also conducted withfishbone charcoal filled to a column bed depth (D)of 15 em with (i) the initial fluoride concentration ofthe test solution kept constant at 5 mg/L and theflow rate maintained at either 15, 25, 50 or 75mLimin; (ii) a constant flow of 15 mLiminmaintained and the initial fluoride concentration ofthe test solution kept at 2.5, 6.0, 10.0, and 20.0mglL; and (iii) the initial solute concentration andthe flow rate kept constant at 10 mg/L of F- and 15mLimin, respectively, while the depth of the mediabed varied from 15 em to 30 ern to 50 cm.

The fishbone charcoal was filled in glass columnsto the required depths in such a way that the bul kdensity of the filled absorbent material was 0.61 ~g/m' for example, a 15 ern column bed depth of ::Jfishbone charcoal weighed 45 g such that {45/[ (1t/4) ~

s Co= IS.om9 L..J Vi. = 45 9IJ'E. 4 0 = 15c:m

•••• pH = 8.0~ U (Tap wat. r)- • 3o c:~ .5!.•.•• 2•. ec: c~ ~ 1

;: c:; s 0 ~c....--A"-:=-:::::::"'-I....--:---=:===~---;~--'-~

o 2

Fig.2-Breakthrough curves (C, vs t) corresponding to thedifferent sorbate influent flow rates (Q)

x 2.S2xI5] = 0.61 g/cm }. The top of the columnconnected to a constant head maintaining tankwhich was charged by an overhead tank containingthe feed solution.

In each of these tests, the effluent concentrationwas monitored at different times and samples wereanalysed for fluoride concentration by the abovestated Ion Analyser. The various runs wereterminated when the effluent fluoride concentrationat the bottom of the column beds exceeded 1 mg/L(the permissible concentration, designated as thebreakthrough concentration) . The volume of theeffluent treated prior to the breakthroughconcentration, was designated as the 'useful (oreffective) treated effluent volume'.

Results and DiscussionVariation of breakthrough with influent flow rate

(Q)-Based on the observed data, the breakthroughcurves for the initial fluoride concentration (Co ) of5.00 mg/L corresponding to the different influentflow rate (Q = 15, 25, 50 and 75 mLimin) aredepicted in Fig. 2. The adsorbent material weight(Ws = 45 g) and bed depth (D = 15 em) in thecolumn were kept constant for all the runs. The bulkdensity of the adsorbent material is 0.61 g/cm", Theretention time or hydraulic residence timecorresponding to the flow rates (Q) of 15,25,50 and75 mLimin are 4.92, 2.95, 1.47 and 0.98 min,respectively.

The volume of treated effluent at the differentflow rates at the chosen effluent flourideconcentration at breakpoint, Ce of 1.0 mg/L wasevaluated from the breakthrough curves as 9.45,

2

O~~~ __~~ __~~~L--L~o 0·4 0.8 1.2 1.6Flow ratlZ per unit wei'3ht of adsorbent.

Qw.ml/min/9

Fig. 3-Variation of useful treated effluent volume (/I) withinfluent flow rate per unit weight of absorbent (Qw), based ondata of Fig. 2

10

8..CtiII..J

::J '->......till

iotill...~

4

6

oCo =S.Omg/LWs = 4S 9o = 1Scm

pH = 8.0 (Tap wo t e r )

BHARGA VA & KILLEDAR: FLUORIDE REMOVAL ON FISHBONE CHARCOAL 239

8.62, 3.60 and 3.15 L corresponding to the flowrates of 15,25,50 and 75 mLimin, respectively. Thevariation of treated effluent volume with respect tothe flow rate per unit weight of adsorbent (Qw ) isshown in Fig. 3. The volume of treated effluentdecreases with an increase in flow rate per gram ofthe adsorbent. This variation may be occurringbecause at higher flow rates, the available retentiontime (or hydraulic residence time) of solution in thecolumn height decreases and might be inadequate ascompared to the optimum time of retention timerequired for completing the sorption reactionbetween the solute and the media. This is causing ashorter breakthrough (at C, = 1.0 mg/L) for thehigher flow rates and thus less quantity of effluentvolume is treated.

Mass balance calculations were done todetermine the amount of fluoride removed atdifferent flow rates (Fig. 2) till the breakthroughconcentration point (C, = 1.0 mglL) . The effluentconcentration upto the breakthrough point variedfrom 0 to 1.0 mg/L as seen in Fig. 2. During thebreakthrough time, an average breakthrough effluentfluoride concentration (Ce-.v) of 0.5 mglL wasassumed for the time interval in which Ce variedfrom 0 to 1.0 mgIL instead of using the procedure ofintegration of breakthrough curve or consideringsmall time intervals. To illustrate, for a flow rate (Q)of 15 mUmin and breakthrough time (tb ) of 10.5 h(C, assumed 0 upto 7 h), the amount of fluorideremoved equals [(15/1000) x (7x60) x 5 + (15/1000)x (lO.~ ~ 7)?0 ,x (5 - 0.5)] = 31.5 + 14.175 = 45.7mg. Similarity, the amount of fluoride removed forflow rates of 25, 50 and 75 mLimin worked out tobe 38.6, 17.1 and 11.0 mg, respectively. The lesserfluoride removal at higher flow rates may be due toshorter available retention or contact time (orhydraulic residence time) .

The plot presented in Fig. 3 follows ar~lationship of the type shown in Eq. (1).V=a,+b,(I/Qw) (1)In Eq. (1), V represents the useful volume of thetreate~ effluent, in litre at a flow rate of Qw, inmLlmm per gram of the absorbent which yields aneffluent concentration of fluoride, C, of 1.0 mgIL,and a, and b, are coefficients.

The values of the coefficients al and b, weredete~in~d as 1.748 and 2.80, respectivelymamfestmg a coefficient of correlation of 0.931,through the regression of data shown in Fig. 3. Thus,Eq. (1) is rewritten as Eq. (2)

V==1.748+2.80 (l/Qw) '" (2)Eq. (2) can be used to determine the total weight ofabsorbent required for treating a known influentvolume at a given flow rate at an initial fluorideconcentration of 5.0 mg/L under the observed testconditions.

Variation of breakthrough with initial fluorideconcentration (C(J-Based on the observed databreakthrough curves at the different initial fluorideconcentrations (Co) are shown in Fig. 4 .. Thesecurves are obtained corresponding to a constant flowrate of 15 mLimin. At the chosen breakpointconcentration of Ce == 1.0 mg/L, the useful volumeof the treated effluent decreases with an increase inCo value. This is due to the fact that for a given flowrate and quantity of the adsorbent, the adsorption orexchange sites of the adsorbent are exhausted earlierwhen a higher initial fluoride concentration influentis encountered. Therefore, the time to thebreakthrough point rs less. Mass balancecalculations were carried out, as before to determinethe amount of fluoride removal at different initialfluoride concentrations (Fig. 4). The amount offluoride removed at Coof2.5, 6.0, 10.0, and 20.0 mg/Lwas 28.2, 44.6, 51.30and 63.18 mg, respectively.The breakthrough times are shorter for higher initialfluoride concentrations, but the amount of fluorideremoved increases with Co . This is probably due tohigher concentration gradients at higher Co values.

From Fig. 4, the values of the useful volume ofthe treated effluents corresponding to Ce = 1.00mgIL were calculated as 18.8, 8.1, 5.4 and 3.24 Lcorresponding to Co values of2.5, 6.0, 10.0 and 20.0

O.91--"-----~----:------.1<;: 0.8

_\0o :: 0.7

c: c:.s ~ 0·6(; ~~ -c c: 0·5.. ..•.. •..s ~ 0·4•.. •..

~ ; 0·3~ .-o ~~ a;: 2 0·2

Q: 15ml/min

W, = 45 9o = 15 em

pH = 8.0 (Top water)- ..; -~ .!:! 0.1f .'!:w So

21.6 26.6 35.0 42·2

r r eo t e d eflluent volume, 'I.L

Fig. 4-Breakthrough curves (C vs t) corresponding '0 thedifferent initial fluoride concentrations (Co)

240 INDIAN J. ENG. MATJ;:R. SCI., DECEMBER 1997

rng/L, respectively. The variation of volume of thetreated effluent versus the initial fluorideconcentration (Co) is shown in Fig. 5. The curve inFig. 5 follows the relationship of the type shown inEq. (3) .

Vt=av+b, (l/Co) (3)

In Eq. (3), V represent the useful volume of thetreated effluent in litre at an initial fluorideconcentration, Co corresponding to Ce = 1.00 mg/L,Q2 and b2 are coefficients. The values of thecoefficients a2 and b, were determined as 0.897 and44.58, respectively, manifesting a coefficient ofcorrelation of 0.999 through the regression of thedata shown in Fig. 5. Thus, Eq. (3) can be rewrittenas Eq. (4) .

V=0.897+44.58 (lICo)

Eq. (4) can predict the useful volume of the treatedeffluent (corresponding to C, = 1.0 mg/L) at anyinitial fluoride concentration (in the range of 2.5-20.0 mg/L) for the observed test conditions.

The coefficients al and b, (Eq. (1» and a2 and b,(Eq. (3) ) are appropriate for a pH of 8.0. Aspreviously presented, the fluoride removal increaseswith a decrease in the pH. Therefore, the usefultreated effluent volume will increase withdecreasing pH at a given flow rate or initial fluorideconcentration. Thus, the variation curvescorresponding to pH values lower than 8 would beplaced above the curve shown in Figs 3 and 5. It istherefore. easily inferred that the coefficients a I. b.,a-, and b2 would increase with decreasing pH.

..J 36>

Q =15 mllmin

-c::"::::J

Ws=4S 9

o = 115 em

pH= 8.0 (TO p wa t e r ).•...•.." 12"0

"..o"•..~ O~~ __~ __~ __~ __~~ __~ __-L~

o 8 16 32 40Initial flouridc concentration C mg/L• 0,

Fig. 5-Variation of useful treated effluent volume (V) withinitial fluoride concentration (C.,l based on data of Fig. 4

Variation of V with Q. D. and C-Based on theobserved data, the sample plots of the variation of V(L) (estimated by multiplying the flow rate by thetime of breakthrough and by the column bed area),with Qa (ml.zrnin/cm' of the column bed area) forthe various D and Co values are presented in Figs 6and 7, respectively. For any Co and D value, themagnitude of V decreases with an increase in theflow rate. This decreasing trend of V is relativelymore pronounced for the lower flow rates (Qa = 3.06

(4)

..J

>w 112E::::J

0 96>..r:: 80w2-.•.. 64..'""0

'" 48..0

'"•.... 32::J 16.•..'"".::J 0

0

o = 65 cm

pH = 8.0 (Tap water)

Co,mg/L2.5

~

5.0i 10.0 :.

20.0 -2 4 6 8 10 12 14 16

Influent flow rate,Qa,rftl/min/em2

Fig. 6-Sample variation plot of useful treated effluentvolume (V) with influent flow rate per unit column bed area(QJ corresponding to the different column bed depths (D). forCo=2.5 mg/L

4 6 8 10 12 14 16Influent flow, Qa mil min I cm2

Fig. 7-Samplc variation plot of useful treated effluentvolume (V) with influent flow rates per unit column bed area(Q.) corresponding to the different initial fluorideconcentrations (( ',,) for 1}"(,5 em

..J

~'"E 112::J"0> 96..e'"~ 80....••• 64"0

'"..0 48•••....

32::::J.•..••• 16'"::J

00 2

Co = 2.5 mg/L

pH= 8.0 (Tap water)

BHARGA VA & KlLLEDAR: FLUORIDE REMOVAL ON FISHBONE CHARCOAL

and 6.12 mlzrnin/cm") . The decrease in the usefultreated effluent volume with increasing flow ratesmay be occurring because at higher flow rates (Q. =10.2 and 15.3 mlzrnin/cm'), the shorter contact timein the column bed might be insufficient for asignificant fluoride removal and thus thebreakthrough occurs early. At lower flow rates thecontact time is increased, resulting in greaterfluoride removal. The overall effect is an increase inthe useful treated effluent volume V. For higher flowrates (Q. = 10.2 and 15.3 rnlzrnin/cm'), thisdecreasing trend more or less stabilizes and thevariation curves appear to become asymptotic to theflow rate axis. Such a stabilization in the V valuesoccurs beyond a high flow rate of 15.3 mg/min/cm'for most of the cases reported in this study. Athigher flow rates, V remains more or less sameprobably because the fresh adsorbent takes up thatmuch material instantly even in the shorter contacttime (thus early breakthrough) at the still higherflow rates. This indicates that operating the columnbed at flow rates greater than 15.3 mlzmin/cm' doesnot significantly change the useful treated effluentvolume, V

For a given Co and Q. values, the useful treatedeffluent volume increases with an increase in beddepth (Fig.6) . This is obviously expected becausethe increase of the column bed material depthprovides more grain surfaces, more sorption sites,and greater contact opportunities to enable anincreased useful volume of the treated effluent. Aplot between V and D is depicted in Fig.8 for thevarious Q. values. The various straight line plotspass through the origin as expected. The slopes of

..J 120 Q,m mln =>:I Co = Z.5 mg/L IS:J 100 pH = 8.0(Tap water)"0•..

SOI:":J 30;:•..

60 SO"" 7S"..Q 40"•....-; 20.."••::I 0

0 10 20 30 40 SO 60 70 80Column bed depth. 0, em

Fig. 8-Variation of useful treated effiuent volume (I') withcolumn bed depths (D) corresponding to the different sorbateinfluent flow rates (Q). for Co=2.5 mg/L.

241

such straight lines indicate the V value per unitdepth and is determined as 1.68, 1.05, 0.92 and 0.83Llcm depth for 15, 30, 50 and 75 mL/min flow rates,respectively. The close scatter of the observedpoints to the straight line plots indicateobservational reliability.These figures indicateslightly more V value per unit depth for low flowrates, as seen earlier.

Fig. 9 depicts a sample plot (based on observeddata) of the useful treated effluent volume versus theinitial fluoride concentration for one value of flow

2 .rate (Q. = 10.2 mLiminlcm ) and column bed depth(D= 42 ern) . The same trend is observed for othertest Q. and D values. The useful treated effluentvolume V, decreases very significantly for higherinitial fluoride concentration values. The useful (oreffective) volume of treated effluent has decreasedby about 60% when the initial fluoride concentrationhas increased from 2.5 to 5.0 mgIL while it hasdecreased by about 75% when initial fluorideconcentration was increased from 10.0 to 20.0mgIL. This indicates a marked role of the initialfluoride concentration on the amount of the usefultreated effluent volume. The decrease in the usefultreated effluent volume with an increase in initialinfluent fluoride concentration is due to the fact thatfor a given flow rate and bed depth, the adsorption

...J

>CI/ 42 Qa = 10.2ml/min/em2E:l 0 = 42 em-0 36> pH = 8.0 (Tap water)..cCI/ 30:l.•....CI/ 24"0

CI/18..

0CI/•.... 12:l...CI/ 6II'~

02 6 10 14 ZZ

Initial fluoride concentration ,c'o, mg/L

Fig. 9-Sample variation plot of the useful treated effiuentvolume (I') with initial fluoride concentrations (Co) at Q.=lO.2mLiminlcm2 and D=42 ern

242 INDIAN J. ENG. MATER. SCL, DECEMBER 1997

sites are exhausted earlier. This results in a shorterbreakthrough time and hence, a smaller usefulvolume of the effluent is yielded.

Predictive empirical models-The observed datawere analysed to determine possible correlationsbetween the useful treated effluent volume andvariables such as the flow rate, bed depth, and initialsolute concentration. Each plot (Vvs Qa) of the typeshown in Figs 6 and 7 follows a relationship of thetype shown in Eq. (5).log V=V-KQa ... (5)In Eq. (5), V is the useful treated effluent volume

in litre, Qa is the flow rate in mlzrnin/cm", and Vxand K are the regression coefficients whichrepresent the intercept and slope, respectively.Through a regression analysis of the observed data,the values of Vx and K were determined and aretabulated in Table I along with their coefficients ofcorrelation (r).

The values of Vx (Table I) increase withincreasing D, while the values of K decrease withincreasing D values for all Co values. Trial plots ofthe values of Vx and K versus D were prepared foreach set of Co and Qa values, for a possiblecorrelation. Two sample plots, one relating thevalues of Vx and D and the other relating the K andD values are shown in Fig.l0 for Co = 5.0 mg/L andQa = 10.2 mLiminlcm2

. The plots presented inFig.l0 follow the relationship of the type shown inEqs. (6) and (7), respectively.

Vx=e+f(D) (6)InK=-g-h(D) (7)In Eqs (6) and (7), e. f, g. and h are coefficients.

Their values were determined through the regression

of the data given in Table 1 and are presented inTable 2 along with their coefficients of correlation(r). The values of coefficients e, g, and h are varyingwith Co values. The value of coefficient f does notvary with Co, to any significance. Therefore, anaverage value of f is computed for further use(!average = 0.0112) .

Plots were prepared between the coefficients e, g.and h versus Co for a possible correlation. The plotsshowing the variation of coefficients e, and g and hversus Co are presented in Figs. 11 and 12,respectively, and are seen to follow the relationshipsrespectively for e. g, and h shown in Eqs. (8), (9)and (10).

1e=--- (8)i+jCo

g=m=n Cnh=p+qCo

(9)... (10)

-3 2Co = s.o mg/L33)(10Qo =10·2 ml/min/emZ

2027 pH = 8.0 (ToP water)

~ .16•• ·21 >0 b12III

~ 1S '""- :l0 "0 8> 9 >

43

o0·~-~~~20~-~~-40~~S~0--6~0~~70

Column bed depth,D. em

Fig. 100Sample variation plot of k and V. with the columnbed depth (D) for Co=5.0 mg/L and Q.=1O.2 mUmin cm2

Table 1 - Values of coefficients V. and k for Eq (5)

Initial Fluoride Column bed Coefficients for Eq. (5) Coefficient ofconcentration depth correlation,Co.mg/L (ern') V. k r

Co=2.5 D=20 1.530 0.0223 -0.931D=42 1.837 0.0212 -0.851D=65 2.033 0.0211 -0.866

Co=5.0 D=20 1.143 0.024 -0.973D=42 1.385 0.018 -0.977D=65 1.580 0.016 -0.965

Co=1O·0 D=20 1.018 0.0239 -0.996D=42 1.321 0.0235 -0.991D=65 1.484 0.01 \3 -0.992

Co=20.0 D=20 0.764 0.1094 -0.913D=42 1.187 0.0731 -0.998D=65 1.388 0.0348 -0.985

BHARGA VA & KILLEDAR: FLUORIDE REMOVAL ON FISH BONE CHARCOAL 243

Table 2 (Values of coefficients e.f, g, and h for Eqs (6) and (7) along with the respectivecoefficients of correlation (r)

Co Values of CoefficientsValue mglL e f r g h r

2.5 1.327 0.0111 0.990 1.644 5.30x 10" -0.9005.0 0.958 0.0097 0.997 1.555 3.90x 10.3 -0.96910.0 0.837 0.0103 0.983 1.424 7.28x 10.2 -0.88120.0 0.527 0.0138 0.977 0.716 11.07x 10.3 -0.987

In Eqs. (8) - (10), i. j, m, n, p, and q arecoefficients. Their values are 0.640, 0.062, 1.836,0.0535,4.013 x 10'4, and 5.647 x 10.4, respectively.The coefficients of correlation vary between 0.967to 0.989.

Substitution of Eqs. (6) - (10) in Eq. (5) yields arelationship presented in Eq. (11) .

log V={[( I ) +0.0112 n]0.640 + 0.062 Co

-[exp{(-1.836+0.0533Co) }-(4.013 x 10-4 + 5.647 X 10-4 Co )D} ]Qa

. . .( II)In Eq. (11), V is the useful volume of treated

effluent at breakthrough concentration of 1.0 mg/Lof fluoride (F) in liter, Co is the initial fluorideconcentration in mg/L, D is the column bed depth inem, and Qa is the flow rate in mUmin ern'.

Alternate model-As an alternative, thepolynomial based fitting concept was used todevelop another empirical relationship (Eq. (12))between the useful treated effluent volume and thevariables, flow rate, colunm bed depth and initialfluoride concentration.

V=[ r,+ Y2 log Qa + Y3D+ Y4 log Co]3 ... (12)Using the data presented in Table 1 and multilinear

regression analysis, the values of coefficients Yl, Y2'

1·0

If 0·8.•..oe 0·6:Je> 0·4 •

o~~ __~ __~ __~~ __~o 4 8 12 16 20 24

Initial fluor ide: conce: nt rationJ

Co,mg/L

Fig. II-Variation of the values of coefficient e with initialfluoride concentrations (Co)

Y3 and Y4 in Eq. (12) are 3.8227, -0.8869, 0.0263and -1.949, respectively. With these values ofcoefficients, the coefficient of correlation for Eq.(12) worked out as 0.970.

The relationships obtained in Eq. (11) and Eq.(12) can be exploited to estimate the amount ofuseful volume of treated effluent for the desired Co,Qa, and D values under the observed test conditions.The high coefficients of correlation ranging from0.950 to 0.995 for Eq. (11) and 0.958 to 0.998 forEq. (12) for various test conditions manifest thejustification of the relationships presented in Eqs(11) and (12). This is also reinforced in Fig. 13which compares sample observed data with thepredicted values [through the use of Eqs (11) and(12)] . Similar empirical relationships can beobtained for the different adsorbate-adsorbentsystems for any other chosen conditions, and alsofor effluent fluoride concentrations other than 1.0mg/L.Practical applications-Fishbones can beconsidered to be a comparatively cheaper and

-312xl0

10

.•..o 6tjI

":Jo 4>

2

o2 6 10 14 18 22

Ini tial fl uoride: c onc e:nt rat ion ,Co.mg/L

Fig. 12-Variation of the values of coefficients g and h withinitial fluoride concentrations (Co)

244 INDIAN J. ENG. MATER. SCI., DECEMBER 1997

effective material for the preparation of bonecharcoal. Defluoridation of drinking water withfishbone charcoal is feasible for isolatedcommunities of small sizes as well as at domesticlevel in the form of home filters.

The empirical relationships16,23,24obtained in thisstudy can be used to predict the fraction of fluorideremaining in solution at any pH and time. Theempirical relationships developed in this study canbe exploited to estimate the quantity of the usefultreated effluent volume corresponding to the statedbreakthrough concentration (1.0 mgIL of fluoride inthis study) for the known values of flow rate,column bed depth, and initial fluoride concentrationas per the stated test conditions. Such relationshipscan be of practical help in the design of smallinstallations.

The operating cost of a fixed bed column reactormainly depends upon the service time (the time atwhich the desired breakthrough concentrationoccurs) . Such studies can be useful in determiningthe economical service time to be provided for agiven set of variables to minimise the operating costof the reactor.

ConclusionsThe treated effluent volume corresponding to the

effluent concentration of 1.0 mgIL of fluoridedecreased with an increase in flow rate per gram ofadsorbent and also with an increase in initialfluoride concentration of solution. The volume oftreated effluent increases with an increasing columnbed depth. A relationship is developed to predict (i)the total weight of adsorbent required for treating aknown influent volume at a given rate at an initialfluoride concentration of 5.0 mgIL and (ii) thetreated effluent volume at any initial fluorideconcentration for a flow rate of 15 mUmin.

The useful treated effluent volume decreasedwith an increase in flow rate per unit area for a

12

C.' 20.018 11: 6S •

.•.o.•~ 27C>

"~ 21

•••.8 15 ,,--A •.....• -....-'-....1...- ....•15 21 27 33

predicted value~ of V,L

• Predicted values from eq.". x Predicted values from eq.12

••12 .'

••

Ie 24 30

Fig. 13--Comparison of the observed and predicted valuesthrough the use of Eqs (II) and (12), for (iD·d=2.5 mg/L andD~20 em, and (ii)-CooF20 mg/L and D=65 en.

given initial fluoride concentration and column beddepth. An empirical relationship was developed topredict the useful effluent (treated) volume for anygiven values of initial fluoride concentration (in therange of 2.5-20.0 mgIL), flow rate (in the range of3.06-15.3 ml./min/cm") and column bed depth (inthe range of 20-65 em) . The relationship manifestedhigh coefficients of correlation when projectedagainst the observed values,

ReferencesI SmithH V & Smith M C, Water Works Eng; 97 (1937)600.2 Mantell C L, Adsorption (McGraw Hill Book Co., New

York, NY, USA), 1945.3 Amphlett C B, Inorganic ion-exchange (Elseveir Publishing

Co., New York, NY, USA),1964.4 Harmon J A & Kalichman S'C, JAm Water Works Assoc, 57

(1965) 245.5 Mantell C L, Carbon and graphite handbook (John Wiley

and Sons, Inc, USA),1968, 137.6 Maier F J, Public Works, 6 {1971) 70.7 Weber W J Jr, Physicochemical Processes for Water Quality

Control (Wiley Interscience, New York, USA), '1972.8 Sorg T J, JAm Water Works Assoc, 70 (1978) 105.9 Choi W W & Chen K Y, J Am Water Works Assoc, 71

(1979) 562.10 Benefield L D, Judkins J F & Weand B L, Process

Chemistry for Water and Wastewater Treatment (Prentice-Hall, Inc, Englewood Cliffs, New Jersey, USA), 1982 .

II CPHEEO,Manuai on Water Supply and Treatment (CentralPublic Health and Environmental Engineering Organisation,Ministry of Works and Housing, New Delhi, India), 1984 .'

12 Hao 0 J & Huang C P, J Environ Eng Div, Am Soc Civ Eng,112 (1986) 1054.

13 Faust S T & Aly 0 M, Adsorption Processes for WaterTreatment (Butterworths, USA), 1987 .

14 Killedar D J & Bhargava D S, Part I, J Inst Public HealthEng, India, 1988 (1988) 6.

15 Killedar D J & Bhargava D S, Part II, J Inst Public HealthEng, India, 1988 (1988) 37.

16 Bhargava D S & Killedar D J, Res J Water PollutionControl Federation, USA, 63 (1991) 848.

17 Bhargava D S & Killedar D J, Indian J Environ Health, 33(1991)464.

18 Killedar D J & Bhargava D S, J Environ Eng Div, Inst Eng,India, 70 (1990) 47.

19 Bhargava D S & Killedar D J, J Environ Eng Div, InstEngrs, India, 72 (1991) 9.

20 Bhargava D S & Killedar D J" J Environ Eng Div, InstEngrs, India, 73 (1993) 59 .

21 Killedar D J & Bhargava D S, Indian J Environ Health, 35(1993) 81.

22 Bhargava D S & Killedar D J, Indian J Engg & MaterSciences, 2 (1995) 157 .

23 Bhargava D S & Killedar D J, Water Res, England, 25(1991) 953.

. 24 Bhargava D S & Killedar D J, Water Res, England, 26(1992) 781.

25 Bhargava D S & Killedar D J, J Environ Eng Div, Inst Eng,India, 73 (1992) II.

26 Bhargava D S & Killedar D J, J Instn Eng (India), EnvironEng Div, 75 (1994) 13.