Adsorption of acid dyes on to granular activated carbon in fixed beds

Adsorption of acid dyes on to granular activated carbon in fixed beds

~ Pergamon S0043-1354(97)0003%0 War. Res. Vol. 31, No. 8, pp. 2093~101, 1997 © 1997 ElsevierScienceLtd. All rights reserved Printed in Great Britai...

692KB Sizes 0 Downloads 63 Views

~

Pergamon

S0043-1354(97)0003%0

War. Res. Vol. 31, No. 8, pp. 2093~101, 1997 © 1997 ElsevierScienceLtd. All rights reserved Printed in Great Britain 0043-1354/97 $17.00 + 0.00

A D S O R P T I O N OF ACID DYES O N TO G R A N U L A R A C T I V A T E D C A R B O N IN FIXED BEDS G. M. W A L K E R * A N D L. R. W E A T H E R L E Y Department of Chemical Engineering, The Queen's University of Belfast, Belfast BT9 5AG, Northern Ireland, U.K.

(Received April 1996; accepted in revised form February 1997) Abstract--This work involved the treatment of industrial wastewater from a nylon-carpet printing plant in Northern Ireland which currently receives no treatment and is discharged straight to sea. As nylon is particularly difficult to dye, acid dyes are required for successful colouration, but they cause major problems with the plant's effluent disposal. Granular activated carbon Filtrasorb 400 was used to treat this effluent in a fixed-bed column system. Breakthrough curves from the fixed-bed column were shallow, even at low flow rates, which indicated a large mass transfer zone and inefficient use of adsorbent. Decrease in adsorbent particle size and decrease in linear flow rate produced a better bed performance. The bed depth service time (BDST) model proved effective for comparison of column variables, with calculated BDST constants providing a useful indication of bed performance. The BDST model also gave good approximation in predicting a bed performance using the relationships postulated by Hutchins (1973). © 1997 Elsevier Science Ltd

Key words--adsorption, dyestuffs, activated carbon, fixed beds, breakthrough curve, bed depth service time

NOMENCLATURE Co = Cb = C,- = C~ = F= K, = rnx = m~ = No = t= Z = Z0 =

initial dye concentration (kg m 3) breakthrough dye concentration (kg m -3) abscissal intercept in BDST plot (min) new abscissal intercept in BDST plot (min) volumetric flow rate (m 3s -~) BDST adsorption rate constant (m 3 kg-~s ~) slope of BDST plot (min m -~) new slope of BDST plot (rain m t) adsorption capacity in BDST model (kg m -3) time (s, rain) bed height (m) critical bed height (m) INTRODUCTION

The global consumption of textiles is currently around 30 million tonnes, with expected growth at 3 % per annum. The colouration of this total needs approximately 800,000 tonnes of dye, promoting an industry of £2500 million per annum (Glover and Pierce, 1992). This results in the textile industry having the potential to discharge vast amounts of synthetic colourant into the environment. Adsorption systems are rapidly gaining prominence as treatment processes which produce good quality effluents that are low in concentrations of dissolved organic compounds, such as dyes (McKay et al., 1983). Activated carbon is the most effective and widely used adsorbent for wastewater treatment (Davies *Author to whom all correspondence should be addressed [Fax: (01232)381753].

et al., 1978) and was used in fixed-bed columns for the treatment of textile effluent. Granular activated carbon ( G A C F400, Chemviron Carbon, Pittsburg, PA, USA) was used as the adsorbent in this work and is one of the most widely used grades of activated carbon in the watertreatment industry. G A C F400 is produced by the gas activation of bituminous coal in a two-step process: carbonisation and activation. This process produces an activated carbon with a substantial transport pore structure and good mechanical hardness. The effluent produced by the carpet printing process contains numerous contaminants, such as dyestuffs, surfactants, thickeners and finishing agents. However, the principal aim of this work was the reduction in effluent colour produced by acid dyestuffs. Three specific dyes, Tectilon Blue 4R-01 (TB4R), Tectilon Red 2B (TR2B) and Tectilon Orange 3G (TO3G) from Ciba Giegy (Bale, Switzerland), made up 75% of the total coloured effluent discharge of a carpet printing plant in Northern Ireland and were therefore chosen as the adsorbates for this work. These dyes are classed, respectively, as anthraquinone, azo and di-azo structures and contain acid functional groups which makes them difficult to adsorb using cheaper adsorbent materials. Comprehensive reviews of fixed-bed models have been presented by M c K a y (1984), with models being proposed having film-pore and combined film-poresolid diffusion theories. The bed depth service time (BDST) model proposed by Bohart and Adams

2093

2094

G . M . Walker and L. R. Weatherley

(1920) and later linearised by H u t c h i n s (1973) has been reported by M c K a y a n d Bino (1990) as offering the simplest a p p r o a c h a n d most rapid prediction of a d s o r b e r performance. The B D S T model has also been successfully used in describing a n d predicting dye c o l u m n a d s o r p t i o n using various adsorbents (Murray, 1995). It was intended t h a t the fixed-bed column design data obtained from B D S T modelling o f laboratory-scale test results will be used in a scale-up to a n industrial-scale pilot plant. The pilot-plant adsorbers will treat a fraction o f the effluent from the carpet printing p l a n t in order to estimate the feasibility o f reducing the colour c o n t e n t in the effluent by activated c a r b o n adsorption. EXPERIMENTAL PROCEDURE

The equipment used during these fixed-bed column studies is described by McKay and Bino (1990). It consisted of a 120-din3 PVC make-up tank in which aqueous dye solutions were formulated. The dye solution was formulated to concentrations ranging from 50 to 200mgdm -3 to simulate the varying dye concentration from the textile plant. The pH of the formulated dye solutions was held at neutral, again to simulate the textile plant effluent. The solution was then gravity fed to a feed tank, having a volume of 120 dm 3, and from which the effluent was pumped using centrifugal or peristaltic pumps through a valved rotameter to monitor the flow rate. The effluent solution was fed through a bed of GAC F400 in up-flow mode. As the total flow rate through these beds was a fraction of the total output of the pump, a recycle line from the pump was installed which delivered dye solution back into the feed tank. This had the dual effect of ensuring that the pump did not work against a dead head and also ensured that the dye solution was constantly mixed, thus providing a uniform initial dye concentration.

The carbon beds were contained in Perspex columns, the design of which is shown in Fig. l. Column diameters were selected at 25 and 50 mm (i.d.) to which an end-plate with an inlet nozzle was attached. The carbon bed was supported in the column by several layers of glass beads of various sizes located on a support plate, which ensured good liquid distribution. Each column had five 5 mm (i.d.) Perspex sample ports stoppered with an Suba-seal bung. This gave a standard bed height of 250 ram, which resulted in beds with carbon masses of approximately 75 g for the 25 mm i.d. column and 300 g for the 50 mm column (which varied with particle size). Samples were drawn at regular time intervals from these ports using a syringe with a hypodermic needle capable of taking a sample from the centre of the bed, Carbon particles with diameters less than 150/xm were removed by wet-sieving using distilled water. This ensured that fine particles would not block the needle of the syringe when samples were being taken. One of the most common causes of poor fixed-bed column performance is that the carbon is not de-aerated prior to the adsorption test. If de-aeration is not carried out, air pockets form in the column which can lead to channelling, increase in pressure drop and premature breakthrough due to the air pockets reducing the surface available for mass transfer between the carbon and the solution. To counteract this problem, carbon particles were wetted for 24 h in distilled water. This was then fed into the test column (which was also filled with water) as a slurry and then allowed to settle for an additional 24 h. The effect of process variables on bed performance was investigated; these included effluent flow rate, effluent concentration and carbon particle size. Dye concentrations were determined using spectrophotometric analysis. BED DEPTH SERVICE TIME MODEL

In the o p e r a t i o n of fixed-bed adsorbers, the objective is to reduce the c o n c e n t r a t i o n in the effluent so t h a t it does n o t exceed a pre-defined b r e a k t h r o u g h value (Cb). Initially, when the activated c a r b o n is

25 or 50 m m column outlet 2 m m Persnex mm

250 n~n

sample p granular activated c a r b o n

Scuba-seal stopper

a h u n i n i u m gauze

50 m m -~"

--

" g l a s s beads

cohillm inlet Fig. 1. Column design and sampling procedure.

nic . . . . . . . . . . ld syringe

Adsorption of acid dyes unsaturated the actual effluent concentration is lower than Cb, but as the effluent is pumped through the bed the carbon becomes saturated and the effluent concentration approaches Cb, i.e. the breakthrough point is reached. The original work on the BDST model was carried out by Bohart and Adams (1920), who proposed a relationship between bed depth (Z) and the time taken for breakthrough to occur. The service time (t) was related to the process conditions and operating parameters:

(Co )

In Cbb-- 1 = ln(e xouoz/e- 1) -- KaCot

2095

and the new intercept as

(7)

McKay and Bino (1990) stated that, if design data are required for a change in volumetric flow rate of solute to the adsorption system, the new gradient, with the intercept remaining unchanged, can be written as

(1) m.l~= mx ff~

(8)

Hutchins (1973) proposed a linear relationship between the bed depth and service time: EXPERIMENTAL RESULTS AND DISCUSSION

No Z

t = CoF

1 -- K ~

, /'C0 ) ln~ - 1

(2)

The critical bed depth (Z0) is the theoretical depth of adsorbent sufficient to prevent the adsorbate concentration from exceeding Cb at t = 0. By letting t = 0, Z0 is obtained from equation (1) by solving for Z. Since the exponential term is usually much larger than unity, the unity term within the brackets in the right-hand side of equation (1) is often neglected, leaving

The most important criterion in the design of fixed-bed adsorption systems is the prediction of column breakthrough or the shape of the adsorption wave front, which determines the operating life-span of the bed and regeneration times. There are many factors which affect column breakthrough, such as the nature of the adsorbate and adsorbent. As these factors were already fixed in this work, the effect of process variables on bed performance was evaluated by measurement of the breakthrough curves and their comparison with mathematical models which compare and predict the bed performance. E f f e c t o f b e d height

(3) Equation (2) enables the service time (t) of an adsorption bed to be determined by a specified bed depth (Z) of adsorbent, t and Z are correlated with the process parameters and initial pollutant concentration, solution flow rate and the adsorption capacity. Equation (2) has the form of a straight line: t = mxZ-

C~.

(4)

and the intercept of this equation represents C~ = f[ln(Co/Cb -- 1)]

(5)

The gradient of equation (4) may be used to predict the performance of the bed, if there is a change in the initial solute concentration Co to a new value Cg. Hutchins (1973) proposed that the new gradient (m)0 can be written as

mx - m.~

(6)

The adsorption columns used during this study contained sampling ports at 50-mm intervals along the height of the column, which provided access to determine the effect of bed height on adsorption. A typical single solute breakthrough curve is illustrated in Fig. 2 as a plot of dimensionless concentration versus volume of liquid treated. From this plot, a tracking of the adsorption wave front can be seen through the bed, with carbon coming into initial contact with the solute at the bottom of the bed, showing an increase in dimensionless concentration with volume treated. This increase is repeated over the entire duration of the experiment at all bed heights. However, results indicate that the profile of the breakthrough curve varies with bed height. Curves in the lower section of the bed had a concave profile, linear profiles in the centre of the bed and convex profiles at the column outlet. These trends indicate that the breakthrough does not follow the characteristic "S" shape profile produced in ideal adsorption systems, behaviour which is associated with adsorbates of smaller molecular diameter and more simple structure. Results reported by McKay and Bino (1990) show near ideal breakthrough curves for the adsorption of phenol onto activated carbon, with almost identical

2096

G. M. Walker and L. R. Weatherley

Ct/Co

1T

o.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

0

20

40

60

80

100

120

140

160

Volume Treated (dm 3) -.--

Scm - - o - l O e r n

~

15cm - - ~ - - 20cm - - x - - 25cm

]

Fig. 2. Effect of bed height on breakthrough curve. (TR2B, Co = 100 mg/dm 3, flow rate = 0.050 dm3/min, particle size = 355-500 #m). profiles being produced at different bed heights. However, most research carried out on the adsorption of dyes shows breakthrough curves similar in profile to the profiles reported here (McKay and Allen, 1983; McKay et al., 1984; Kulkarni et al., 1986; Walker and A1-Duri, 1994). The variation in concentration profile is due to the relatively large adsorption zone, i.e. carbon near the top of the column comes into contact with dyestuff before carbon near the bottom of the column is saturated. The rate at which the adsorption zone travels through the bed decreases with bed depth, suggesting that beds of an increased height may be required for dye adsorption. This phenomenon occurs in the adsorption of dyestuffs because, due to their large molecular structure, resistance to internal diffusion is much higher than smaller molecules such as phenol. This results in dye molecules not having enough contact time to diffuse from the surface of the particle to the adsorption sites and therefore being pumped further up the column to fresh activated carbon where the rate of dye uptake is higher. Results from experiments using 25 and 50 mm i.d. columns, operated under identical process conditions of linear flow rate, dye concentration and particle size, revealed no discernible difference in column performance in terms of breakthrough time and calculated BDST constants.

Effect of volumetric .flow rate In a packed column of uniform cross-sectional area, the volumetric flow rate is directly proportional to the overall linear flow through the bed; in this work columns of 25 and 50 mm i.d. were used in conjunction with centrifugal pumps, rotameters and

peristaltic pumps to vary the flow rate. Results indicate that an increase in flow rate decreases the volume treated until breakthrough and therefore the service time of the bed. This is due to decreased contact time between the dye and the carbon at higher flow rates; as the adsorption rate is internal diffusion controlled, premature breakthrough occurs.

Effect of dye concentration An increase in inlet sorbate concentration increased the slope of the breakthrough curve, reducing the volume treated before carbon regeneration. An increased inlet dye concentration at constant flow rate decreases the throughput until breakthrough. This may be caused by high adsorbate concentrations saturating the activated carbon more quickly, thereby decreasing the breakthrough time.

Effect of carbon particle size The effect of varying the carbon particle size used during these fixed-bed experiments had a marked increase on column performance. In the finer particle size ranges, adsorption breakthrough curves followed a much more efficient profile than larger particle size ranges, in that the breakthrough time increased and the curves tended towards the classic "S" shape profile. This correlated well with results from batch kinetic studies, in that the rate of adsorption appeared to increase with reduced particle size. These observations are also consistent with equilibrium isotherm data which stated that adsorption capacity appeared to increase with particle size (Walker, 1995). This increase in volume to breakthrough correlated well with results reported by McKay et al.

Adsorption of acid dyes (1983) of dye adsorption with activated carbon and Desai et al. (1992) of the adsorption of solvents onto alumina. This enhanced bed performance is probably due to a higher overall rate of diffusion because of the higher available interfacial area. However, the effect of pressure drop needs to be taken into account in the design of adsorbers packed with finer particles. B D S T with variation in flow rate

The effect of volumetric flow rate on bed performance was described using the BDST model, where the bed depth versus service time for Tectilon Orange at constant concentration and particle size is plotted on Fig. 3. The breakthrough point for BDST calculations in this work was fixed at 20% of the inlet feed concentration, i.e. at an inlet feed concentration of 200 mg dm -3, Cb in equation (1) is 40 mg dm -3. As the breakthrough time decreases with increase in flow rate, the slope of the BDST plot decreased accordingly and beds operated at low linear flow rates produced much higher service times. The predicted BDST equations calculated from the slope of the median for the higher and lower flow rates, using equation (8), show reasonable agreement. However, with an under-estimation of the lower flow rate performance and a over-estimation of the higher flow rate performance, it could be said that the BDST equation did not compensate enough for the variations. The BDST constants calculated for single dye component systems in 25mm i.d. columns are reported in Tables 1-3, with trends being established. For all three dyes, the critical bed depth (Z0) showed an increase with increase in flow rate, with the

2097

minimum bed depth required for adequate dye adsorption increasing by approximately 25, 50 and 75% for Tectilon Blue, Tectilon Orange and Tectilon Red, respectively, with a fourfold increase in flow rate. Calculated values of adsorption capacity (No) in general showed a decrease with increasing flow rate, with reductions of approximately 50% for Tectilon Red and Tectilon Orange with a fourfold increase in flow rate. These results correlate well with the observed performance in the breakthrough curves and go some way in explaining the reasons for poor adsorber performance at increased flow rates. However, the rate constant (K,) did not show consistent trends for variation in flow rate in these systems. The correlation coefficient r 2 values calculated for the linearisation of the experimental data were generally quite high, with values all above 0.97. B D S T with variation in dye concentration

The effect of inlet dye concentration on bed performance was compared with the BDST model in Fig. 4, with a plot of bed depth versus service time for Tectilon Red at constant flow rate and particle size. Figure 4 shows an increased slope at lower inlet concentrations, which is indicative of better bed performance with increased volume until exhaustion. The predicted BDST equations calculated from the slope and intercept of the median for the higher and lower inlet concentrations, using equation (7), show good agreement with experimental results. The slight variation may be due to error in the slope of the 200 mg dm -3 plot, as a slightly lower gradient of this line would produce predicted lines for the 100 and 300 mg dm -3 series almost identical to experimental

Service Time (rain) 700 ~600 + 500

.,'"

i

# •

?

q'#

400 300 + i

o..O~O"

100 +

.....

""

0 p, -100 ~

{

7

0.1

0.2

0.3

0.4

Bed Height (cm) •

Flow = 0.025 dmZ/min o Flow = 0,050 dmZ/min A Flow = 0.100 dm3/rnin ....... Predicted performance Linearisation

Fig. 3. Effect of flow rate on experimental and predicted BDST lines. (TO3G, Co = 200 rag/din 3, particle size = 1000-1400 Hm).

2098

G. M. W a l k e r a n d L. R. W e a t h e r l e y Table 1. Effect of flow rate on BDST parameters: Tectilon Blue (Co = 200 mg dm 3, particle size 355-500 #m)

Flow rate (dm 3 rain i) 0.050 0.100 0.200

Zo (m)

No × 10-4 (kg m -3)

BDST capacity (rag g t)

0.084 0.092 0.106

4.60 5.02 3,67

72.1 78.6 57.5

/~, × 10-2 (m 3 kg -t s ~)

r2

29.7 623 11.9

0.974 0.968 0.997

Table 2. Effect of flow rate on BDST parameters: Tectilon Red (Co = 200 mg dm -3, particle size 1000 1400 #m) Flow rate (dm3 min I) 0.025 0.050 0.100

Z0 (m)

No x 10-4 (kg m -3)

BDST capacity (rag g-~)

Ks x 10-2 (m3 kg 1s ~)

r2

0.076 0.078 0.133

3.90 2.40 1.67

61.1 37.6 26.2

1,96 6.19 10.4

0.974 0,982 0.987

Table 3. Effect of flow rate on BDST parameters: Tectilon Orange (Co = 200 mg dm -3, particle size 1000-1400/~m) Flow rate (dm 3 min -I) 0.025 0.050 0.100

Z0 x l0 -4 (m)

No x 10-4 (kg m 3)

BDST capacity (mg g i)

K~ x 10 2 (m3 kg -I s -~)

r:

0.077 0.089 0.115

3.03 2.13 1.50

47,5 33.4 23.5

2.46 6.08 1.34

0.994 0.966 0.996

data. Therefore, it could be concluded that the BDST equation can produce accurate predictions for variation in inlet concentration for acid-dye-activated carbon systems. Variations in the inlet concentration also affected the BDST constants, with some trends being established (Tables 4-6). In general, an increase in dye concentration increased the critical bed depth (Z0) of the adsorption column, with a 9% increase in depth with a threefold increase in concentration for Tectilon Blue and Tectilon Red. An increase in dye concentration also appeared to increase the adsorption capacity (BDST model), with this variation being

most evident in Tectilon Blue, with a 58% increase in No with a threefold increase in inlet concentration. As in the previous section, consistent trends in the BDST rate constant were hard to establish, although results tend towards an increase in K~ with increased dye concentration. The linearised plots of the BDST experimental data again produced a high degree of linearity with r 2 values around 0.98. B D S T with variation in carbon particle size

The effect of particle size on bed performance is shown in Fig. 5 and the data correlated using the BDST plot for the adsorption of Tectilon Orange

Service Time (min) 2500 y

t 2000+

o~ °o p o

15oo

o6,

1000~ °°,.t~

500 ;

o°"

L

oq

I

0~

0

.~

01

0 15

02

0 25

I

I

0.3

0.35

-500 ~ Bed Height (cm) •

100 mg/dm 3

r~

200 mg/dm 3

Predicted performance Fig.

4.

A

300 mg/dm 3

Linearisation

Effect o f dye c o n c e n t r a t i o n on e x p e r i m e n t a l a n d predicted B D S T rate = 0.025 dm3/min, particle size = 1000-1400/~m).

lines.

(TR2B,

flow

Adsorption of acid dyes

2099

Table 4. Effect of dye concentration on BDST parameters: Tectilon Blue (flow rate 0.100 dm3min -~, particle size 1000-1400,um) Concentration (rag I i) 100 200 300

Zo (m)

No × 10-4 (kg m -3)

BDST capacity (rag g-i)

Ka × 10-z (m3kg-~ s-i)

r2

0.078 0.080 0.083

1.77 2.25 2.79

27.7 35.3 43.7

1.67 5.15 9.92

0.984 0.987 0.994

Table 5. Effect of dye concentration on BDST parameters: Tectilon Red (flow rate 0.025 dm3min -~, particle size I000-I400/~m) Concentration (rag 1 i) 100 200 300

Z0 (m)

No × 10-4 (kg m -J)

BDST capacity (rag g-i)

K, x 10-2 (m3kg-~ s-t)

r2

0.079 0.076 0.086

3.28 3.90 3.00

51,4 61.1 47.0

2.22 1.96 3.31

0.994 0.994 0.991

TabIe 6. Effect of dye concentration on BDST parameters: Tectilon Orange (flow rate 0.025 dm3rain t, particle size 1000-1400/~m) Concentration (rag 1-~) 100 200 300

Zo (m)

No × 10-4 (kg m -3)

BDST capacity (rag g-~)

K, × 10-z (m3kg ~s-~)

r2

0.058 0,077 0.122

2.58 3.03 1.50

40.4 47.5 23.5

3.85 2.46 3.I 5

0.997 0.994 0.964

( T O 3 G ) . This plot s h o w s an increased slope o f the B D S T plot w i t h smaller particle size w h i c h indicates better b e d p e r f o r m a n c e a n d increased v o l u m e until e x h a u s t i o n . T h e p r e d i c t e d B D S T e q u a t i o n s calculated f r o m the slope o f the m e d i a n for the larger a n d smaller particle sizes, using a r e l a t i o n s h i p p r o p o s e d b y M c K a y et al. (1983) w h e r e the slope o f the p r e d i c t e d lines is a f u n c t i o n o f the s q u a r e o f the ratios o f t h e particle m e a n d i a m e t e r s , s h o w g o o d a g r e e m e n t with e x p e r i m e n t a l results. A s with the c o n c e n t r a t i o n predictions, the slight v a r i a t i o n m a y be due to e r r o r in the slope o f the m e d i a n plot, as a slightly lower g r a d i e n t o f this line w o u l d p r o d u c e p r e d i c t e d lines for

the 3 5 5 - 5 0 0 / ~ m a n d 7 1 0 - 1 0 0 0 # m series a l m o s t identical to e x p e r i m e n t a l data. T h e r e f o r e , it m a y be c o n c l u d e d that the B D S T e q u a t i o n c a n p r o d u c e a c c u r a t e p r e d i c t i o n s for v a r i a t i o n in particle size for a c i d - d y e - a c t i v a t e d c a r b o n systems using the relationship p r o p o s e d b y M c K a y et al. (1983). As with the previous p r o c e s s p a r a m e t e r s , variation in particle size also affected t h e B D S T c o n s t a n t s . T h e critical b e d d e p t h (Zo) a p p e a r e d to increase with increases in particle size, w h i c h c o r r e l a t e d well with b r e a k t h r o u g h curve o b s e r v a t i o n s . Bed particle size also a p p e a r e d to have a very significant effect o n the B D S T a d s o r p t i o n capacity, with a d s o r p t i o n

Service Time(min) 2000 o

°

o o

1500

0 p

1000 500 . ° o ° " .bp~

O, 0

~ ~

" 0.2

°°

t" 0.3

I 0.4

-500 2 Bed Height (cm) •

355-500 microns [] 500-710 microns ~ 710-1000microns ....... Predicted performance Linearisation

Fig. 5. Effect of carbon particle size on experimental and predicted BDST lines. (TO3G, Co = 100 mg/dm 3, flow rate = 0.050 dm3/min.)

2100

G. M. Walker and L. R. Weatherley Table 7. Effect of particle size on BDST parameters: Tectilon Blue (flow rate 0.100 dm 3 rain -~, Co = 200 mg dm 3)

Particle size (~tm) 500-710 710-1000 1000-1400

Z0 (m)

No x 10-4 (kg m -j)

BDST capacity (mg g-i)

Ka x 10-2 (m3 kg-~ s-~)

r2

0.106 0.122 0.140

2.53 0.50 0.053

39.6 7.8 0.8

86.1 37.8 31.0

0.947 0.964 0.995

Table 8. Effect of particle size on BDST parameters: Tectilon Red (flow rate 0.050 dm 3 min -~, Co = 100 mg dm -3) Particle size (gm) 355-500 500-710 710-1000 1000-1400

Zo

No × 10-4

(m)

(kg m -3)

BDST capacity (rag g ~)

K~ x 10 2 (m 3 kg ~s -~)

r2

0.076 0.090 0.110 0.128

7.4 2.1 1.67 1.08

116.0 32.9 26.2 16.9

2.06 6.13 6.30 8.32

0.974 0.979 0.999 0.998

Table 9. Effect of particle size on BDST parameters: Tectilon Orange (flow rate 0.050 dm ~rain -~, Co = 100 mg dm -3) Particle size (~tm) 355-500 500-710 710-1000 1000-1400

2"o (m)

No × 10-4 (kg m -3)

BDST capacity (mg g-~)

K, x 10-2 (m3 kg -~ s -~)

r2

0.064 0.096 0.095 0.133

6.17 3.80 2.20 0.75

96.7 59.6 34.4 11.7

2.95 3.17 5.50 11.60

0.996 0.988 0.968 0.964

capacity (No) decreasing rapidly with an increase in particle size with an average 86% reduction in capacity resulting from an increase in mean particle diameter from 427.5 to 1200 #m. Trends in the BDST rate constant were again hard to judge (Tables 7-9), with the data for Tectilon Blue showing a decrease with increase in particle size and Tectilon Red and Tectilon Orange showing an increase with increased particle size; the latter results correlate well with a batch kinetic HSDM model which showed an increase in diffusivity in large adsorbent particles (Walker, 1995). BDST with variation in dye t YTe

The BDST capacity converted to milligrams per gram for each dye type was calculated from fixed beds operated under identical process conditions of flow rate, initial dye concentration and bed particle size, with the results shown in Table 10. These indicate that the greatest dye adsorption occurs with TO3G, followed by TB4R and TR2B. The BDST capacity expressed as a fraction of the equilibrium capacity indicates slight differences between each of the dye types. CONCLUSIONS

The breakthrough curves calculated from the adsorption of acid dyes on to activated carbon

columns indicate that dye removal is strongly dependent on the system kinetics. A low linear flow rate increases the breakthrough time of the beds, as does a small particle size. The BDST capacity expressed in terms of mg of dye adsorbed per gram of carbon shows an increase at low linear flow rates and small particle sizes. The equilibrium capacities of TO3G, TR2B and TB4R are 852, 535 and 537 mg g-~, respectively (Walker, 1995). Therefore, operating the fixed-bed systems under advantageous conditions increases the BDST adsorption capacity to 10-15% of the equilibrium capacity. These observations lead to the conclusion that this process is internal diffusion controlled. The relatively large molecular size of the dye molecules makes diffusion within the carbon particles tortuous compared to smaller aromatic compounds such as phenols. The acid functional groups on the dyes also have an influence, in that acid dyes show a much poorer adsorption performance than basic dyes in the same system. The BDST model is a useful tool for comparing the performance of beds operating under different process variables. The relationships proposed to predict the effect of these parameters gave reasonable approximations to experimental results. The calculated BDST constants of Zo and No showed well-defined trends, although the BDST rate constant (/~) did not appear to follow any correlation. In

Table 10. Effect of dye on BDST capacity (flow rate 0.100 dm ~min ~, Co = 200 mg dm -3, particle size 1000-1400 ~m) Dye TO3G TR2B TB4R

Dye structure

BDST capacity (mg g-~)

Fraction of equilibrium (%)

Di-azo Azo Anthraquinone

97.9 44.5 66.3

11.5 8.3 12.3

Adsorption of acid dyes conclusion, this simple model can accurately describe and predict fixed-bed column performance in the adsorption of acid dyes. Acknowledgements--This work was funded by the Department of Education for Northern Ireland and Carpets International (UK) Ltd. REFERENCES

Bohart G. S. and Adams E. Q. (1920) Adsorption in columns. J. Chem. Soc. 42, Davies R. A., Kaempf H. J. and Clemens M. M. (1973) Removal of organic material by adsorption on activated carbon. Chem. Ind. 9, 827-831. Desai R., Hussain M. and Ruthven D. M. (1992) Adsorption on activated alumina. II. Kinetic behaviour. Can. J. Chem. Eng. 70, 707-715. Glover P. and Pierce J. H. (1992) Are natural products good for your health? Soc. Dyers and Color. Symp., "'Technology for Change", Oct. Hutchins R. A. (1973) New method simplifies design of activated carbon systems. Chem. Eng. 80 (19), 133-138.

WR 3 [/8--J

2101

Kulkarni P. B., Bal A. S., Bhole A. G. and Bhoi A. V. (1986) Colour removal by granular activated carbon. J. I W W A 18, 237 244. McKay G. (1984) Mass transfer processes during the adsorption of solutes in aqueous solutions in batch and fixed bed adsorbers. Chem. Eng. Res. Des. 62, 135-243. McKay G. and Allen S. J. (1983) A mathematical model for the fixed bed adsorption of Telon Blue dye on peat. J. Sep. Proc. Teehnol. 4, 8-15. McKay G. and Bino M. J. (1990) Fixed bed adsorption for the removal of pollutants from water. Wat. Environ. Pollut. 66, 33-53. McKay G., Blair H. S. and Gardner J. (1983) The adsorption of dyes in chitin, III. Intraparticle diffusion processes. J. Appl. Polym. Sci. 28, 1767-1778. Murray M. (1995) Adsorption and desorption of basic dyes in the aqueous phase. PhD thesis, The Queen's University of Belfast. Walker G. M. (1995) Industrial wastewater treatment using biological activated carbon. PhD thesis, The Queen's University of Belfast. Walker G. M. and A1-Duri B. (1994) Acid dye adsorption using fixed GAC columns. Proc. I. Chem. E. Res. Event, London, VoI. 2, pp. 109-110.