-RESEARCH & TECHNOLOGY II

Using Powdered Activated Carbon: A Critical Review Issam N. Najm, Vernon L. Snoeyink, Benjamin W. Lykins Jr., and Jeffrey Q. Adams Because the performance of powdered activated carbon (PAC) for uses other than taste and odor control is poorly documented, the purpose of this article is to critically review uses that have heen reported and to analyze means of employing PAC more efficiently. The extent of adsorption of synthetic organic chemicals on PAC is strongly dependent on the type of compound being removed. The reported removals of trihalomethanes and trihalomethane precursors by PAC range from poor to very good. In selecting the point of addition of PAC, consideration must be given to the degree of mixing, the contact time between the PAC and the water, the PAC residence time, and the minimization of interference of adsorption by treatment chemicals. One of the main advantages of PAC is its low capital cost.

Synthetic and natural organic con- oxidation require that an adsorption taminants are often found in drinking process be used. Granular activated car- water sources. Categories of these com- bon (GAC) and powdered activated car- pounds include taste- and odor- (T&O) bon (PAC) can both be used to remove causing compounds, synthetic organic organic compounds from water. Pow- chemicals (SOCs), pesticides, herbicides, dered activated carbon is widely used in color, and trihalomethane (THM) pre- the United States for T&O removal,l~z cursors. The low removals of many and GAC filter-adsorbers arecommonly undesirable compounds by coagulation, used for the same purpose.3 Granular

of choice for removing SOCs, including volatile organic chemicals (VOCs). There are inherent advantages to applying activated carbon in fixed beds, including lower carbon usage rates for many applications and ease of spent carbon regeneration. However, PAC has the advantages of being a cheaper material and of requiring minimal capital ex- penditure for feeding and contacting equipment; also, it can be applied only when needed.

A full report of this project, published under the title Control of Organic Com- pounds With Powdered Activated Carbon (catalognumber90581). isavailablefrom the American Water W&ks Association Research Foundation, 66I66 W. Quincy

sedimentation, filtration, and chemical activated carbon has been the adsorbent Ave., Denver, CO 80235. ?_s,

TABLE 1 Manufacturersspecifications of some commercially available PACs

Parameter PAC l* PAC2t PAC3$ PAC45 PAC5** PAC6tt

Iodine number--mg/g 800 1,199 600 900 1,000 550 Molasses decolorizing index

z 14 18

Moisture as packed--percent 3 5 10 4 Apparent density-g/cmg 0.64 0.54 0.74 0.38 0?8 0.50 Ash content--percenf 6 3-5 3-5 Passing 100 mesh-@rent 99 95-100 95-100 99 Passing 200 mesh--percent :; 97 85-95 85-95 Passing 325 mesh-percent 90 98 90 65-85 65-85 i i

*Aqua-Nuchar, Westvaco, Covington. W.Va. tWPH, Calgon Corp., Pittsburgh, Pa. *Aqua, Westvaco, Covington. W.Va. Nuchar S-A, Westvaco, Covington, W.Va. **Nuchar SA-20, Westvaco, Covington, W.Va. ttHydrodarco B, American Norit, Jacksonville, Fla.

TABLE 2 Advantages and disadvantages of different points of addition of PAC

Point of Addition

Intake

Rapid mix

Filter inlet Slurry contactor preceding the

rapid mix

Advantages Disadvantages

Long contact time, good mixing Some substances may be adsorbed that would otherwise probably be removed by coagulation, thus increasing carbon- usage rate (this still needs to be demonstrated).

Good mixing during rapid mix and flocculation, Possible reduction in rate of adsorption because of interference reasonable contact time by coagulants, contact time may be too short for equilibrium

to be reached for some contaminants, some competition may occur from molecules that would otherwise be removed by coagulation.

Efficient use of PAC Possible loss of PAC to the clearwell and distribution system Excellent mixing for the design contact time, no A new basin and mixer may have to be installed; some

interference by coagulants, additional contact time competition may occur from molecules that may otherwise possible during flocculation and sedimentation be removed by coagulation.

JANUARY 1991 ISSAM N. NAIM ET AL 65

Copyright (C) 1991 American Water Works Association

TABLE 3 Isotherms for selected oraanic contaminants

Compound Water+

Alachlor

Aldicarb

Atrazine

Benzene

Bromodichloromethane Bromoform Carbofuran

Carbon tetrachloride Chlorobenzene

Chloroform Cyclohexane Dibromochloromethane Dibromochloropropane

o-Dichlorobenzene

cis-1.2.dichloroethylene

truns-1,2-dichloroethylene

2.4.Dichlorophenol

1,2-Dichloropropane

Dieldrin Endrin 2,4,5-T ester Ethylbenzene

Ethylene chloride Ethylenedibromide

Geosmin

Heptachlor Heptachlor epoxide Hexachlorobutadiene

DDW GW

DsDw SW

DDW

MGDw

MDW MDW DDW

SW MDW MDW DDW

SW MDW

MDW DDW

SW DDW

SW DDW

SW

iii?. SW GW SW

DDW

DD

L!i?W MDW

DDW MDW DDW

SW DDW DDW DDW DDW MDW

Kt 80.2

62 10.5 8.4

4.52 38.2 24.4

1.4E-9 3.57 0.10 0.65 16.2 11.9 0.12 0.11 9.0

0% 23.9 0.10 6.61 3.84 19.1

::2 0.15 0.12 0.62 0.63 23.3 13.4 0.3 0.4 3.0 5.0

ii3 16.2 4.5

0.0011 0.9

0.77 13.5 0.18 1.78

11.92 4.64

T t

l/n

0.26 0.33 0.38 0.4

0.41 0.29 0.36

0s 0:76

El 0:36 0.84 0.98 0.35 0.31 0.84 0.2

0.93 0.51 0.46 0.38 0.58 0.59

::ii 0.45 0.39 0.29 0.32 0.59 0.48

E 0.79 0.79 0.29 0.53 1.5

0.46 0.54 0.39 0.83 0.92 0.75 0.63

PAC Dose-mg/L

c, = 100 i&/L

10 rg/L

0.6 0.7 3.6 4.3 7.7 1.2 1.6

>lOO.OOO 9.6

156.4 20.5 2.2 3.3

108.4 85.7 4.5 4.4

433.6 2.4

105.7 4.2 8.1 2.0 2.8

115.7 137.4 176

51.5 58.2 2.0

7; 74.5

2::

6Y5 2.8

25j:7 34.7 33.7 2.7

74.0 6.1

1.34 4.5

- C,=

1 M/L 1.2 1.6 9.4

11.8 21.9 2.6 4.0

>lOO,OOO 27.7 990

152.3 6.1 8.3 825 900 11.0 10.0

3.300 4.14 990 15.0 25.8 5.2 11.8 495 660 825 160 157 4.2 7.4 330

247.5 33

19.8 28.5

430.4 6.11

902ko ii0

128.6 7.33 550 55.6 9.6

21.3

co = 10 cg/I

C,= 1 MI/L

0.1 0.14 0.86

1:: 0.24 0.37

>lOO,OOO 2.5 90

13.8 0.6

0.76

8?8

2 309 0.4 90

i:: 0.5 1.1

ti 75

14.5 14.3 0.4

0.67

22:: 3

1.8 2.5

39.1 0.56

8.:82 10

11.7 0.67

50

C% 1.90

T C,=

Timet

14 days 18 days 11 days 13 days 12 days

lh 15 min lh lh 7 days 7 days lh lh 11 days 10 days lh 15 min lh 8 days 8 days 11 days 10 days 3 days 31 days 15 days 10 days 10 days 7 days 30 min 3 days 10 days

lh 15 min 3 days lh 4 days 10 days 5 days 5 days

lh

-

Ranges

512-0.6 477-0.2 46-2.6

274-1.3 171-19

785-4.2 317-0.6

8,000-3.000 iO,OOO-3,OOt

57-1.3 90-2

111-2.9 365-2.2 120-2.5

17,500-1,4Of 732-15.4 401-14.1

75-5 io,OOO-3,001

40-1.5 69.2-0.2 85.8-0.2

1.056-14.4 293-59 615-5.5 230-19 214-9.7

415-13.5 406-59 300-l

600-50 196-4.1 126-4.6

!2,500-2,100 io,oOO-3,OuC

565-5.6 65-17

123-0.2 139-1.1

0.05-2.0 0008-0.05

40-0.16 30-0.04 117-1.3

In general, the performance of PAC in drinking water treatment for uses other than T&O control is poorly documented. The purpose of this article is to critically review those uses that have been re- ported and to analyze means of using PAC more efficiently. With this infor- mation, it should be possible to deter- mine more easily those situations in which PAC can be used more effectively than GAC. Characteristics of PAC

The primary characteristic of PAC that differentiates it from GAC is its particle size. Typically, 65-95 percent of commercially available PAC passes through a 325-mesh (44-pm) sieve. For comparison, Kruithof et a14 gave the particle size distribution of two powdered

66 RESEARCH AND TECHNOLOGY

activated carbons available in the Neth- erlands that show 23-40 percent by mass smaller than 10 pm diameter and lo-18 percent larger than 74 pm. The manufacturers specifications, including particle size distribution, of several PACs commercially available in the United States are summarized in Table 1. The particle size distribution is important because the smaller PAC particles adsorb organic compounds more rapidly than larger particles.5

Powdered activated carbon is made from a variety of materials including wood, lignite, and coal. Its apparent density, ranging from 0.36 to 0.74 g/cm3 (23 to 46 lb/cu ft), depends on the type of material and the manufacturing process. Iodine number and molasses number are often used to characterize PAC (Table 1).

For example, the AWWA standard for PAC specifies a minimum iodine number of 500.

Factors affecting performance of PAC Several design and operational pa-

rameters affect the performance of PAC for drinkingwater treatment. Important criteria for selecting the point with addi- tion of PAC include (1) the provision of good mixing or good contact between the PAC and all the water being treated, (2) sufficient timeof contact with adsorption of the contaminant, (3) minimal interfer- ence by treatment chemicals with ad- sorption on PAC, and (4) no degradation of finished water quality.

The PAC must be added in a way that ensures its contact with all of the flow. Addition at locations other than those

JOURNAL AWWA Copyright (C) 1991 American Water Works Association

TABLE 3, continued hems for selected organic contaminants 6

K-f

iot

T -

L

L CW

-

I

L

-

L

- -

c,= Time$ Ranges 1 Reference

6 days 170-2.1

7 days

5 days 5 days

149-0.6 460-0.016 l,OOO-100

500-40 0.04-5

0.02-0.04

28 days 28 days lh 7 days 7 days lh ,

12 days 15 min 4 days 15 days 30 min 28 days 28 days lh 7 days 7 days

5 ,

7 days 7 days 3 days 15 min 8 days 10 days lh 9 days 15 days

0.17-0.011 0.8-0.058

8,180-220 3.408-0.4

322-5.3 3,360-2,150

16.5-0.02 177-0.2

148-11.8 ~0,000-3,000

104-2.3 672-2.5 180-14

0.5-0.064 2.2-0.32 63-0.15 31-2.88 32-3.9

l,OOO-6 1 ,OOO-6 350-10 400-4

166-5.6 o,oOO-3,000 1,795-6.9 522-10.5

2,000-1,500 32.7-1.6 232-2.9

5

1

-tap water, HI 4-humic al cid

- PAC Dose-mg/

T L

10 fig/L

3.6 2.2

132.7 2.5 3.8

2.16 5.4

13.7 0.73 3.4 4.2 1.0 2.7

44.8 1.7 2.0

::i 7.1 6.1 3.7 2.9 4.7

43.1 20.6 30.9 11.3 11.3 4.2

2.13 12.3 2.4 4.3 3.1 2.2 3.1 4.1

P&?/L c,=

1 /ML 22

6.4 319.4

6.8 9.9

3.53 17.0

27.5 8.6

36.7 12.0 2.3 6.9 160 3.2 4.4 8.2

6.22 22

15.7 11.4 17.3

33.33 150

73.3 110.0 99.0 99.0 9.6 4.7 76

5.0 13.9

10

::9 13

c, = 10 fig/1

c,= 1 la/L

-

Compound Water

DDW DDW

:z DDW

TZ DDW DDW DDW DDW

HA MDW DDW

SW MDW DDW

GW DDW

DDW GW

DDw HA

MDW DDW

GW DDW

GF DDW DDW

DDW GW

MDW DDW

GW

l/n

0.74 0.43 0.34 0.39 0.36 0.17 0.46 0.5

1.65 0.26 1.03 0.99 0.42 0.34 0.36 0.51 0.23 0.31 0.48 0.28 0.45 0.37 0.45 0.74 0.81 0.50 0.51 0.51 0.9 0.9

0.32 0.3

0.75 0.27 0.47 0.47 0.16 0.42 0.46

Lindane

Methoxychlor

Methylisoborneol

Parathion PCB (Aroclor 1254)

Pentachlorophenol

Phenol Simazine

Styrene Toluene

Toxaphene

Trichloroethylene

2,4,6-Trichlorophenol

m-Xylene o-Xylene

p-Xylene

4.5 15.4 0.31 14.5 10.0

28.0** 5.83it

5.03 169.3

3.6 11.46

2.7 8.24 42.4 14.4 0.62 30.8 22.3 12.1 15.9 4.5

8% 5.725 2.97 0.66 1.35 0.9 1.0

:dp3 21.2 1.3

20.13

2 28.15 11.1 7.6

2: iii 28 74 74 27 27

2:

:7

i i

i i 38 28 72 28 28 73 28 28 47 12 12

23 23 12 12 28

;: 28 47 28 28

2.0 0.6 29 0.6 0.9

0.32 1.5

2.5 0.8

3.33 1.1 0.2 0.6

14.5 0.3 0.4 0.7

0.57 2

1.4 1.0

1.57 3.0

13.6 6.67 10.0 9.0 9.0

0.87 0.4 6.9

0.45

A:9 0.3 0.8 1.2

J

*MDW-mineralized distilled water, DDW-distilled deionized water, GW-groundwater. SW-surface water, : (5 mg/L as TOC) tFreundlich K values are listed for C in pg/L and Q in mg/g. SCIontact time in the isotherm bottle Range of remaining aqueous concentration in pg/L **co = 5,000 pg/L ttc,= l,~O /a/L

listed in Table 2 may not achieve this objective.

Sufficient time of contact is also necessary, and the time required is an important function of the characteristics and concentration of the molecule to be adsorbed.7 In the absence of competition and coagulation floe interference, 15 min is sufficient time for molecules such as dimethylphenol (molecular weight = 122, C0=90mg/L,carbondose=250mg/L)to equilibrate with a 325-mesh (44-pm-di- ameter) particle if adequate mixing is used.8 As the molecular size increases, the rate of diffusion into the pores of the PAC particle decreases. For example, rhodamine B dye (molecular weight = 422) requires approximately 5 h to come to equilibrium, a lO,OOO-molecular- weight fulvic acid requires 17 h, and a

50,000-molecular-weight humic acid re- quires about two days. The adsorption kinetics and equilibrium capacity depend on the type of carbon used, and, therefore, these values should only be taken as rough estimates. If insufficient time is allowed for equilibration, more PAC must be used to achieve the desired removal.

in the test with the composite sample was still decreasing after 120 min.

Means of applying PAC

The effect of PAC particle size on required contact time in the absence of floe interference was shown by Najm et al5for a continuous stirred tank reactor. In I5 min, 500 pg/L trichlorophenol was reduced to 25 pg/L by 16pm-diameter PAC but only to 275 pg/L by loo-pm- diameter PAC. The composite PAC sample, with an average diameter of 40 pm, reduced the concentration to 180 pg/L. In no case was equilibrium achieved in 15 min, and the concentration

Powdered activated carbon can be fed as a powder using dry feed machines or as a slurry using metering pumps.g In a conventional treatment plant, the com- mon points of PAC addition are the plant intake, rapid mix, and filter influent. Another point of addition that should be considered, although it is not commonly used, is a continuous-flow slurry con- tactor that precedes the rapid mix. The PAC can be mixed intensely with the water in the absence of floe to obtain rapid adsorption kinetics and then in- corporated into the floe in the rapid mix for subsequent removal by sedimentation andfiltration.5Table2summarizes some

JANUARY 1991 ISSAM N. NNM ET AL 67

Copyright (C) 1991 American Water Works Association

of the important advantages and dis- advantages of PAC addition at each of these points.

When PAC is added at the rapid mix, incorporation of PAC into the coagulant floe particles may reduce the rate of adsorption.@IzGauntlett and PackhamO conducted jar tests showing that the removal rate of chlorophenol by PAC in the absence of alum addition was most rapid and that the addition of PAC after alum addition gave a better rate of re- moval than when applied just before alum. Apparently, PAC added after alum floe had formed adhered to the outer surface of the floe rather than being incorporated into the floe. Gauntlett and Packham argued that the reason for the reduction in removal rate is that the adsorbate must diffuse through the part of the floe surrounding the PAC particle and then into the particle itself in order to be adsorbed. Najm et al,** however, found little reduction in the rate of adsorption of trichlorophenol on PAC in spite of the incorporation of the carbon particles into coagulant floes.

TABLE 4 PAC dose required to reduce the pesticide level in distilled and Little Miami river water

Addition of PAC at the intake has the advantage of providing extra contact time, but there is the possibility of the PAC adsorbing many compounds that would otherwise be removed by coagula- tion, flocculation, and sedimentation. On-site tests are recommended to de- termine whether one factor outweighs the other.

PAC Dose-mg/L

co= 10 pg/L c, = 1.0 )&g/L

Pesticide Method c, = 1.0 pg/L c, = 0.1 pg/L c, = 0.1 )&g/L c, = 0.05 pg/L

Parathion JT* 2.5 5 0.5 0.6 10 0.9 1.1

2,4,5-T ester JF+ i.5 17 1.5 3

JT 14 44 3

Endrin 1.8 14 1.3 i.5

JT 11 126 11 23

Lindane 2 12 1.1 2

JT 29 70 6 9

Dieldrin 3 isi 1.1 1.7 P 18 7 12

*PAC dosage in jar test in which pesticide is removed from distilled water by PAC alone, with a contact time of 1 h +PAC dosage in the plant in which pesticide is removed from river water by conventional treatment and activated carbon

TABLE 6 Percentage removals of THMprecursors with the addition of PAC to New Orleans water

cornDared with those of an ag2ated control* Addition of PAC just before the filter

is advantageous because the PAC can be retained in the filter and kept in contact with the water longer, thereby better using its capacity. The average PAC residence time is equal to one half of the time between two successive backwash- ings, assuming that PAC is continuously added to the influent water. However, the PAC must be added carefully to avoid its penetration into thedistribution system. The maximum dosage of PAC is limited by the ability of the filter to retain the PAC and by the rate of head loss buildup in the filter, which is ex- pected to increase as PAC dosage in- creases. Sontheimer et aI3 reported a lo-20 percent reduction in filter run time as a result of the addition of PAC. Dougharty and MorrisI reported that filter velocities >8-9 m/h resulted in a breakthrough of PAC when it was added at a dose of 60 mg/L. In a direct filtration pilot-plant study, Gifford et al*5 reported a 40 percent reduction in filter run time as a result of the addition of 30 mg PAUL. With a hydraulic loading of 4.9 m/h, no breakthrough of PAC was observed in the filter effluent.

Treatment Sequence

- 1 Ca

THM Precursor Removal-

5 mg PAUL

0 7 8 50 mg PAUL

lercent

500 mg PAUL

Mode 1 PAC addition, coagulation- flocculation-sedimentation, filtration

Mode 2 PAC addition, coagulation- flocculation-sedimentation, chlorination, filtration

Mode 3 Coagulation-flocculation- sedimentation, PAC addition, chlorination, filtration

31 84

41 77

*Same treatment but with no PA

0

ddition

27 90

TABLE 7 Effect of PAC dosage on the removal of chloroform precursors from Ohio River water*

PAC Dose-wag/L

0 10

Chloroform Formation Potential

rcg/L Reduction--percent 236 0 215 9

25

I

170 28 50 127

I 46

100 85 GA

Sontheimerll suggested adding a separate PAC reactor between the sedi- mentation basin and the filter to increase the time of contact and take advantage of coagulation and sedimentation for eliminating competing organics to the

Loo0 48 3OC, pH = 10.5-11.2, Clzresidual = 2-3 mg/L, stored three days

68 RESEARCH AND TECHNOLOGY JOURNAL AWWA Copyright (C) 1991 American Water Works Association

Effect 4

Treatment Type PAC Dose

mg/L

Settled water Settled water + Cl, Settled water + PAC +

Cz 2 4 8

20 50

100

of T

t

TABLE 5 PAC on trihalomethane formation potential (THMFP) concentration

CHCl,

NF* 27

22 16.4 20 16 11

9

*NF-none found

0.35 r

Trihalomethanes-us/L

15.1 8.0 16.4 10.2 15.8 9.4 16.9 12.2 13.0 10.0

9.5 8.8

TTHM rmol/L

NF 0.37

0.31 0.36 0.32 0.29 0.22 0.18 t

Removal of TTHM

percent

0

16 2

14 22 41 51

Figure 1. Preliminary cost estimates for PAC treatment as a function of dosage and system size

TABLE 8 Reduction of THMFP and TOC by coagulation and adso@tion on 50 mg/L of two types of PAC in water

from Daytona Beach Aquifer, Fla., and Ilwaco Reservoir, Wash. (after Amy and Chadiks7) After Adsorption

Coagulant Type

Al Al Al Fe Fe

Coagulant After Coagulation Carbon B Carbon C

Dose THMFP TOC THMFP:TOC THMFP TOC THMFP:TOC THMFP TOC THMFP:TOC mg/L b&L mg/L &mg P&?/L mg/L Irg/w l&L mg/L Wmg

Daytona Beach Aquifer, Fla.

0 667 9.2 72.5 433 7.96 54.4 369 7.31 50.5 ;o 509 6.88 74 297 6.13 48.4 309 5.77 53.6

418 6.54 ::8

220 4.98 4.2 242 4.78 50.6 30 383 4.5 218 4.22 51.7 265 4.18 63.4 15 548 7.65 71:6 345 6.21 55.6 321 5.97 53.8

30 401 6.76 59.3 257 5.41 47.5 328 5.09 64.4

Ilwaco Reservoir. Wash. ~ I

0 1,087 8.04 135.2 771 7.11 108.4 696 7 99.4 A; 2.5 1.25 579 371 5.34 3.33 111.4 108.4 569 240 2.86 5.64 100.9 83.9 442 258 5.14 86

2.91 88.7 Fe 2.5 894 6.9 129.6 788 6.83 115.4 755 6.78 111.4 Fe 5 350 3.0 116.7 347 4.18 83.0 321 3.73 86.1 Fe 7.5 245 2.27 107.9 127 2.38 53.4 124 2.06 60.2

JANUARY 1991 ISSAM N. NAJM ET AL 69 Copyright (C) 1991 American Water Works Association

maximum extent possible. A major disadvantage, however, is that an addi- tional coagulation step would be required to remove the PAG.

Improving adsorption efficiency. Various techniques for applying PAC to improve its adsorption of large, slowly diffusing compounds are available. For example, adding PAC to solids contact clarifiers has the potential for improved adsorption efficiency because the carbon can be kept in contact with the water for a longer time than when it is added to the rapid mix of a conventional plant. Kas- sam et ali6 reported mean carbon resi- dence times ranging from 9 h to 8.5 days when PAC was added to a solids contact slurry recirculating clarifier. Hoehn et all7 studied theaddition of PAC to a pilot- scale floe-blanket clarifier. They reported PAC retention times of one to two days. In a study of the addition of PAC to a bench-scale floe-blanket reactor, Najm et all2 measured steady-state PAC reten- tion times ranging from 9 h to two days. They determined that the maximum adsorptive capacity of PAC for 2,4,6- trichlorophenol (TCP) was achieved during this time. Richardi noted that the PAC dosage for detergent removal could be reduced by 25-40 percent if the carbon was added to the influent of a floe-blanket clarifier instead of to a conventional system. More research is needed to optimize this process and to determine coagulation procedures that can be used to increase adsorption kinetics, such as using a polyelectrolyte alone instead of alum or ferric salts.Jg

one contactor. To determine the mini- mum values of ml and m2 required to drop the adsorbate concentration from an initial concentration, C,, to a final concentration, Cz, the intermediate aqueous concentration, Ci, between the two reactors was calculated. Assuming that the PAC reached equilibrium with the effluent of each contactor and that the same Freundlich isotherm equation describes equilibrium in both contactors, Ci was calculated by

Cl 0 1/n C G- -flP=l-l Cl (1 ) n in which C, = influent concentration to contactor 1, Ci = effluent concentration of contactor 1 and influent concentration to contactor 2, Cz = effluent concentra- tion of contactor 2, and l/n = Freundlich equation constant.

pounds on PAC under both laboratory and field conditions. Various types of organics, activated carbons, and waters were included in these studies. The most common means of evaluating adsorption on activated carbon, especially PAC, has been through the use of adsorption iso- therms. The experimental bottle-point technique is widely used to achieve an isotherm relationship between theaque- ous adsorbate concentration and the carbon surface loading (see Randtke and Snoeyinkg for a description of this test). Because these isotherms are conducted under laboratory conditions, the ac- curacy of performance predictions based on them needs to be examined.

The PAC dose to each reactor can then be calculated using Eq 2, described later. The limitation to this approach is that equilibrium is not always reached when PAC is added to conventional contactors, and the same Freundlich equation may not apply in each contactor. When competing organics are present, the isotherm function depends on PAC dosage.

Table 3 lists some of the compounds for which isotherms have been deter- mined along with their Freundlich iso- therm constants. Also listed are the predicted carbon dosages required to reduce the concentration of each of these compounds from 100 pg/L to 10 pg/L and 1 pg/L, and from 10 pg/L to 1 pg/L. Each of these calculations used the iso- therm of that compound and assumed that equilibrium is achieved at the final concentration. The isotherm will change as the type of carbon and the type of water treated change. The following equation was used to calculate the carbon dosages required:

cc, - C,) cc, - c,, *=-z- Qe K Ccl

(2)

In the Roberts-Haberer process,zo-zz buoyant polystyrene spheres l-3 mm in diameter are coated with PAC. The spheres are held in the reactor by means of a screen, and the water to be treated is passed upflow through the medium. After saturation, the PAC is removed from the beads by backwashing the medium (downflow) with a high flow rate. New PAC is then applied to the beads. The PAC at the entrance to the process can be equilibrated with the influent concentration, thereby increas- ing its adsorption efficiency. Recovery of the PAC for regeneration may also be possible. In addition, the Roberts-Ha- berer filter may be used as a roughing filter to remove some suspended solids before filtration. However, careful studies are needed to assess the efficiency of adsorption for specific organics and the ability of the polystyrene beads to retain the PAC in order to determine the conditions under which the process will be more cost-effective than GAC.

Reactions with treatment chemicals and calciumcarbonate. Careful attention must be paid to the interaction of PAC with water treatment chemicals. Activated carbon will chemically reduce com- pounds such as free and combined chlo- rine, chlorine dioxide, chlorite, ozone, and permanganate; the demand for oxi- dants and disinfectants is thereby in- creased. The reaction of activated carbon with chlorine will reduce its adsorptive capacity for compounds such as phenol and substituted phenols.24.25 Lalezary- Craig et aI26 found a reduction in the ability of PAC to adsorb both geosmin and methylisoborneol in water con- taining free chlorine and monochlor- amine. Surprisingly, the effect of the monochloramine appeared to be greater than that of chlorine.

in which m is the carbon dosage (mg/L), C,, and C, are the initial and equilibrium concentrations (pg/L), respectively, and K and n are the Freundlich equation constants.

Sontheimer et alz3 suggested the addi- tion of PAC to a multistage contactor. They showed that addition of two doses of PAC, ml and m2, to contactors 1 and 2, respectively, in series theoretically re- sults in a lower usage rate than addition of the total carbon dose, mr= ml + m2, to

Addition of PAC to a water that is supersaturated with calcium carbonate or other precipitates, or treatment that causes an increase in pH to greater than the saturation pH just after PAC is added, such as in lime softening or aeration of water with a high carbon dioxide content, may lead to coating of the PAC particle with precipitates and to a corresponding decrease in adsorption efficiency. Also, adsorption at high pH is often poorer than at low pH because many organic contaminants are weak acids that ionize at high pH, rendering them more hydrophilic.

All the isotherm data listed in Table 3, except those of Lalezary et a127 and Najm et al, were conducted using GAC pulverized to a smaller size to enhance kinetics and reduce the time required for equilibration. Some of the isotherms of Miltner et a128 were conducted with 50 X 200-mesh pulverized GAC (74-297-pm diameter), whereas the rest of the iso- therms were conducted using pulverized GAC of ~200 mesh.

A major concern with some of the isotherm data listed in Table 3 relates to the contact time allowed between the carbon and the water before sampling. Although higher adsorption rates are achieved when GAC is pulverized, some large-molecular-weight organic com- pounds may not reach equilibrium in less than oneor twodays. The isotherms for large molecules determined with one hour or less of contact time can be used as rough estimates of adsorbability but probably indicate a capacity lower than achievable at equilibrium.

PAC performance Many studies have been conducted to

assess the adsorption of organic com-

Caution should be exercised when using the bottle-point isotherm of a trace compound conducted with natural water. The competitive effect of the background

70 RESEARCH AND TECHNOLOGY JOURNAL AWWA Copyright (C) 1991 American Water Works Association

natural organic matter on the adsorption isotherm of the trace compound on acti- vated carbon has to be carefully consid- ered when the isotherm is being used to predict the performance of an activated carbon process. When conducted in a multisolute system, the bottle-point iso- therm of a trace compound depends on the initial concentration used. Frickzg determined that the isotherm capacity of activated carbon for p-nitrophenol (PNP) decreased with decreasing initial PNP concentration in the presence of a constant background concentration of p- hydroxybenzoic acid (HBA) and phenol- 4sulfonic acid (PSA). With an initial concentration of 5 mg/L of each of HBA and PSA, a decrease in initial PNP con- centration from 10 to 2 mg/L resulted in a 50 percent reduction in the carbons capacity for PNP. Similarly, the authors research showed that the carbon capacity for TCP present in natural water and determined using the bottle-point iso- therm technique depended on the initial concentration of TCP. At a certain TCP equilibrium concentration, the carbon capacity decreased with decreasing ini- tial concentration as a result of the competitive effect of the background natural organics.

Therefore, the choice of the proper isotherm significantly affects the eval- uation of the performance of PAC in a continuous process. Although the effect of competitive adsorption behavior on the kinetic models is not clearly under- stood, the authors believe that the iso- therm to be used should be that con- ducted with an initial concentration close to the concentration of the influent to the continuous process.

Published reports of PAC performance are discussed subsequently and, when possible, are compared with performance predictions based on equilibrium iso- therms. The data on PAC performance are categorized according to type of compound as well as according to impact on drinking water quality.

T&O-causingcompounds. Powdered acti- vated carbon is widely used in the United States for T&O control. It has been estimated that about 90 percent of water utilities worldwide that use activated carbon use it in the form of PAC.ii.30In 1977 PAC was used in 25 percent of the 683 US water utilities surveyed, which included the 500 largest utilities.1 The number of utilities using PAC is con- stantly on the rise.2 Compounds causing T&O problems vary widely from the strongly adsorbed, low-molecular-weight compounds, such as 2,4-dichlorophenol (2,4-DCP), to the weakly adsorbed, high- molecular-weight compounds, such as some types of humic substances. This has resulted in a wide variation in PAC adsorption efficiency.

The concentration of organic com- pounds that cause a T&O problem is

JANUARY 1991

TABLE 9 Quantity ojPAC required as a junction of dosage and system size

System Size Quantity of PAC Required-lb/day

tnsd 5 mg PAWL 10 mg PAUL 2.5 mg PAUL 50 mg PAUL 75 mg PAUL

0.1 2.9 5.84 14.6 29.2 43.8 0.5 14.6 29.2 73.0 146.0 219.0 1.0 29.2 58.4 146.0 292.0 438.1 2.5 73.0 146.0 365.0 730.0 1,095.2 5.0 146.0 292.0 730.0 1,460.3 2.190.4

:: 292.0 438.1 584.1 876.1 2,190.4 1.460.3 2,920.5 4.380.8 4,380.8 6,571.2 25 730.1 1,460.3 3,650.6 7,301.3 10,951.g 50 L460.3 2.920.5 7.301.3 14.602.6 21.903.9 75 2.190.4 4,380.8 10,951.9 21,903.g 32855.9

100 2,920.5 5.841.0 14.602.6 29,205.2 43.807.8 150 4.380.8 8,761.5 21,903.g 43.8073 65.711.8

TABLE 10 Annual cost estimates jorpurchase ojPAC

System Size Cost of PAC-.$/year

rnsd 5 mg PAUL 10 mg PAUL 25 mg PAUL 50 mg PAUL 75 mg PAUL

i:: 2.132 426 4,263 853 10,658 2,132 4,263 6,395 21,316 31,974

1.0 4P63 8,526 21,316 42,632 63,963 2.5 10,658 21,316 53,290 106,580 159,899 5 21,316 42,632 106.580 213,204 319,798

10 42,632 86,279 213,204 426,393 639,597 15 63,963 127,911 319,798 639,597 959,395 25 106,595 213,204 532,988 1,065,989 1,598,977 50 213,204 426,393 1,065,989 2,131,979 3,179,969 75 319.798 639,597 1.598,977 3.197969 4,796,961

100 426,393 852,786 2,131,979 4,263,959 6,395,938 150 639,597 1,279,179 3,197,969 6,395,938 9,593,923

TABLE 11 Annual cost estimates of PAC feed equipment

System Size Capital and O&M Costs of Feed Rquipment+/year

msd 5 mg PAUL 10 mg PAUL 25 mg PAUL 50 PAUL mg 75 PAUL mg

0.1 2,125 2,135 2.165 2,365 2,595 0.5 2,420 2,620 3,155 4,065 4,665 :i 3,270 5,095 3,655 4,715 5,925 6,980

6,005 7,820 11,130 14.180 5 7,555 8,765 12.680 18,810 23,480

10 11.215 13,620 21,260 29,095 36,990 E 20,295 14,080 18,290 27,740 38,800 48.290

26,425 37,580 53,315 67,865 50 32,105 39,940 58,995 86,115 113,065 75 43,230 54,290 80,000 119,520 152,320

100 48,980 61,795 95,155 138,685 174,905 150 68,495 85,885 133,725 186,525 249,135

highly dependent on the type of com- pound. The threshold odor number

each compound. For example, geosmin is

(TON) is widely used as a measure of known to cause odor problems in the low

taste and odor in drinking water. The nanograms-per-litre range of concentra-

TON produced by a given concentration tionsz7 whereas 2 pg/L of PNP causes

of contaminant is usually different for objectionable odor, and 8 pg/L causes a taste problem.31 Sigworth32 presented a

ISSAM N. NAJM ET AL 71 Copyright (C) 1991 American Water Works Association

list of T&O-causing herbicides and pesticides alongwith theconcentrations that caused detectable odor. The con- centrations varied from 17 pg/L for aldrin to 4.7 mg/L for methoxychlor.

The concentration of PAC required to remove T&O varies widely, consistent with the diversity of T&O-causing com- pounds. Researchers33 found that 1 mg PAC/L reduced the TON by 1-24 units in the water supplies of four US cities, indicating a wide variability in adsorb- ability of T&O-causing compounds. Bench-scale studies of three types of carbon showed that 75 mg/L of the best carbon was required to achieve humic odor reduction from a TON of 15 to 3, and 30 mg/L of the best PAC was re- quired to reduce the concentration of 2,4- DCP from 1.75 too.24 mg/L, correspond- ing to a TON reduction from 22 to 3.34 The isotherms of Najm et all2 and Aly and Faust35 (see Table 3), however, pre- dict PAC doses of 13 and 20 mg/L, respectively, to achieve the same removal of 2,4-DCP.

A full-scale study was conducted on the performance of PAC for the removal of T&O at the East Drive Well in North Miami Beach, Fla.36 Doses of PAC up to 27 mg/L failed to remove a hydrogen- sulfide-like odor. In another full-scale study,s7 an average PAC dose of 23 mg/L reduced the TONof influent water to the Nitro, W.Va., water plant from 30-325 to 14.6, which was still much higher than the acceptable value of 3 or less. The specific compounds causing the T&O were not determined, so comparison with predicted dosages based on the isotherm is not possible.

Laboratory tests were conducted by Lalezary et alz7 to study the efficiency of two types of PAC (PAC l* and PAC 2t) for the removal of the following odor- causing compounds-geosmin, 2,3,6-tri- chloroanisole (TCA), 2-isopropyl-3-meth- oxypyrazine, 2-isobutyl-3-methoxypyra- zine, and 2-methylisoborneol (MIB)-at the nanograms-per-litre concentration level. The effects of chlorination, coagu- lation, and filtration on the adsorption of these compounds on the two types of PAC were investigated. In a pilot study that followed, Lalezary-Craig et alz6 investigated the effects of initial concen- tration, chlorine and chloramine resid- ual, presence of background organics, contact time, and filtration rate on the removal of geosmin and MIB by PAC 2. The authors reported that carbon doses as low as 10 mg/L reduced 66 rig/L each of geosmin and MIB to 2 and 7 rig/L, respectively, for surface coverages of approximately 6 pg/g. However, data are not available in this range to deter- mine whether equilibrium had been achieved. The presence of chlorine and chloramine, as well as background or- ganics, was found to decrease the re- moval of the two compounds by PAC.

72 RESEARCH AND TECHNOLOGY

The removals of geosmin and MIB achieved by PAC in the pilot plant com- pared well with those predicted by the bench-scale studies for similar initial concentration, chlorine dosage, and co- agulation conditions.26,27

Pesticides and herbicides. A growing problem facing many water treatment facilities is theincreasingpublicconcern, and the related regulations, regarding the concentrations of pesticides and herbicides (collectively termed biocides) in drinking water sources. Biocides are transported from agricultural fields into lakes and rivers by water runoff. Miltner et a13s reported the presence of several biocides in the Sandusky River in north- western Ohio during a survey from May to July 1984. These biocides included alachlor, metolachlor, atrazine, cyana- zine, metribuzin, carbofuran, and sima- zine at mean concentrations ranging from 0.44 pg/L for simazine to 8.17 pg/L for metolachlor. The maximum concen- trations, however, ranged from 1.15 to 22.3 pg/L.

Six biocides-endrin, lindane, meth- oxychlor, toxaphene, 2,4-D, and 2,4,5- TP-are included in the current drink- ing-water standards.3g New standards, proposed in May of 1989,40 do not include endrin, but include eleven additional biocides. Table 3 lists isotherm data for some of these compounds, showing them to be strongly adsorbed. Table 3 shows that a PAC dose of ~5 mg/L should be capable of reducing the concentration of any of the b&ides listed from 10 to ~1 pg/L and that the type of water affects the isotherm capacity, an effect probably caused by natural background organics.

A comparison of the carbon dosages required to reduce the pesticide concen- tration to a predetermined value in a jar test using distilled and natural water is shown in Table 4.41 The data show that the adsorptive capacity of PAC in raw river water is 50-93 percent lower than that in distilled water.

Miltner et a138 investigated the removal of several pesticides that were included in a national pesticide survey. Alachlor, atrazine, carbofuran, linuron, metola- chlor, metribuzin, and simazine were found in the influent water to three water treatment plants in Ohio. The performance of PAC for their removal was investigated in full-scale as well as bench-scale studies. The isotherms for these compounds on pulverized GAC are listed in Table 3. Although the adsorption capacities of PAC in jar tests and full- scale treatment were in good agreement, they were approximately two orders of magnitude lower than those predicted by the pulverized GAC isotherms listed in Table 3. One possible reason for this difference is that the initial concentra- tions in the isotherm studies were much higher than those used in the jar tests and full-scale studies. This would lead to

a greater reduction in biocide adsorption capacity in the jar tests and full-scale plants, as discussed earlier in this article. Other possible reasons for thedifference include (1) the short contact time allowed (1.5 h$ compared with several days for the pulverized GAC isotherms) and (2) some interference by coagulants.

The study of Miltner et a138 concluded that all pesticides detected in the San- dusky and Maumee rivers in Ohio will not be removed by conventional water treatment processes unless an activated carbon adsorption process is employed. Similar conclusions were reached by Richard et a142 for the removal of atrazine from the Des Moines, Iowa, water supply. Poor removals of the pesticide atrazine by PAC, however, were observed at New Orleans, La.43144

SO&. Synthetic organic compounds are the products of a wide range of manufacturing processes, and some are classified as hazardous to animal and human health. Occasionally, these com- pounds find their way into natural water sources by means of accidental spills, dumping, or leaching to groundwater aquifers. The US Environmental Pro- tection Agency (USEPA) has set maxi- mum contaminant levels (MCLs) for several VOCS,~~ and many other com- pounds are currently being added to the list (see Federal Register46 for the pro- posed listing). Many of these compounds are included in Table 3.

The adsorbability of SOCs is highly dependent on the type and properties of the compound being removed. A bench- scale study on PAC adsorption of four SOCs present in Mississippi River water was conducted at New Orleans, La.43 The removals of 1,2-dichloroethane, benzene, toluene, and atrazine by con- ventional treatment processes (coagula- tion, flocculation, sedimentation, filtra- tion, and chlorination) were investigated, along with their adsorption on 5,50, and 500 mg PAC/L. The raw water concen- trations of the compounds, however, never exceeded 1 pg/L and continuously varied throughout the period of the study. These facts made it difficult to draw solid conclusions concerning the removal of these compounds. The data, however, suggest that removal did not occur unless PAC was present.

The removal of TCP and trichloro- ethylene (TCE) from groundwater by PAC added to a bench-scale floe-blanket reactor was investigated by Najm et a1.i2 With the long carbon residence times supplied (one to two days), complete utilization of the adsorptive capacity of the carbon, as indicated by the isotherm, was achieved.

*Aqua Nuchar, Westvaco Corp.. Covington, W.Va. tWPH. Calgon Corp., Pittsburgh, Pa. $The totalcontact timeincludedaperiodofsedimentation

when mixing was not employed. The amount of adsorption that occurred during this time was not established.

JOURNAL AWWA Copyright (C) 1991 American Water Works Association

TABLE 12 Total cost estimates of PAC treatment

PAC Treatment Cost Estimates-$/l,OOOgal

System Size 5 mg PAUL 10 mg PAC/L 25 mg PAUL 50 mg PAUL 75 mg PAUL

mgd A* B C A B C A B C A B C A B C - 0.1 0.083 0.017 0.100 0.083 0.033 0.116 0.085 0.083 0.168 0.092 0.167 0.259 0.101 0.250 0.351 0.5 0.019 0.017 0.036 0.021 0.033 0.054 0.025 0.083 0.108 0.032 0.167 0.199 0.036 0.250 0.286

i.5 0.013 0.008 0.017 0.017 0.025 0.030 0.009 0.014 0.033 0.033 0.047 0.042 0.012 0.018 0.083 0.083 0.101 0.095 0.023 0.017 0.167 0.167 0.184 0.190 0.027 0.022 0.250 0.250 0.277 0.272 5 0.006 0.017 0.023 0.007 0.033 0.040 0.010 0.083 0.093 0.015 0.167 0.182 0.018 0.250 0.268

10 0.004 0.017 0.021 0.005 0.033 0.038 0.008 0.083 0.091 0.011 0.167 0.178 0.014 0.250 0.264 15 0.004 0.017 0.021 0.005 0.033 0.038 0.007 0.083 0.090 0.010 0.167 0.177 0.013 0.250 0.263 25 0.003 0.017 0.020 0.004 0.033 0.037 0.006 0.083 0.089 0.008 0.167 0.175 0.011 0.250 0.261 50 0.003 0.017 0.020 0.003 0.033 0.036 0.005 0.083 0.088 0.007 0.167 0.174 0.009 0.250 0.259 75 0.002 0.017 0.019 0.003 0.033 0.036 0.004 0.083 0.087 0.006 0.167 0.173 0.008 0.250 0.258

100 0.002 0.017 0.019 0.002 0.033 0.035 0.004 0.083 0.087 0.005 0.167 0.172 0.007 0.250 0.257 150 0.002 0.017 0.019 0.002 0.033 0.035 0.003 0.083 0.086 0.005 0.167 0.172 0.006 0.250 0.256

*A-capital cost plus O&M feed equipment cost, B-PAC material cost only, C-total cost

A major spill of carbon tetrachloride aliphatics, includingdichloroethenes and of THMFP than of TOC with the addition of PAC. (CCIJ in a tributary of the Ohio River

during early 1977 resulted in several studies of its removal by PAC. Bench- scale studies at the USEPA laboratory in Cincinnati utilized untreated river water containing approximately 70 pg CCL&. The water was dosed with varying amounts of PAC, mixed rapidly for 1 min, and then contacted for 1 h in a jar. The highest PAC dose of 30 mg/L re- sulted in CC& reduction of only 20 percent. The amount adsorbed on the carbon surface, 1.86 mg/g, was higher than the equilibrium capacity predicted by the l-h isotherm of Dobbs et aP7given in Table 3. During this period, 9.6 mg PAC/L was used at the Cincinnati, Ohio, water treatment plant. Using conven- tional treatment and this PAC dosage, the CC& concentration in the finished water was virtually equal to that in the influent water to the plant.48

chloroethanes (l,l-dichloroethane and l,l,l-trichloroethane), was virtually unaffected by the addition of up to 26 mg PAC/L. The chlorobenzene concentra- tion in the finished water was reduced from 0.5 to ~0.1 pg/L with the addition of 10 mg PAC/L. Nonvolatile synthetic organic chemicals were reduced from a combined finished water concentration of 45 to ~5 pg/L with the addition of 7.5 mg PAC/L. Comparison of the removal data, however, with those predicted by the isotherms listed in Table 3 was not possible because the concentrations of the compounds were far lower than those used in the isotherm tests.

Following the CC& incident, another spill of toxic chemicals occurred in the same river at a different location. The spill consisted of hexachlorocyclopen- tadiene and octachlorocyclopentadiene. Although the concentrations of the com- pounds were quite low (cl pg/L), up to 30 mg PAC/L was added as a precaution at the Evansville, Ind., plant.49 Both compounds were virtually absent (x0.1 pg/L) in the finished water.

After consumers complaints of a pesticidelike taste and odor in the drink- ing water of the Sunny Isles water treatment plant, North Miami Beach, Fla., in September 1977, laboratory analysis of the raw and finished water revealed the presence of 42 SOCs rang- ing in concentration from 0.01 to 73 ~g/L.36.50~5i A study of the removal of natural total organic carbon (TOC) and SOCs by 7.5,15, and 30 mg PAC/L was initiated. The PAC was added at the well intake to achieve a maximum of 2 h of contact time. A finished water concen- tration of 40 pg/L of volatile halogenated

JANUARY 1991

THMs and THMFP. The discovery that THMs are formed during drinking water treatment52*53 has caused major concern because of the potential health hazard associated with thesecompounds. Rook52 attributed the formation of these com- pounds to the reaction between natural organic matter and chlorine. Organic compounds that undergo this reaction are referred to as THM precursors. The amount of THMs that will be produced if the reaction goes to completion is referred to as the THM formation potential (THMFP). In 1979 the USEPA estab- lished an MCL for total THMs at 100 pg/L as an annual average.54 This MCL is currently being reviewed by USEPA.

Adsorption of THMFP. The reduc- tion of THMFP can be determined by measuring THMFP directly or by mea- suring TOC. Although there is usually a good correlation between the TOC and THMFP for a given type of water, there is nocorrelation that applies toall waters because of the differences in the charac- teristics of the organics. For example, Lange and Kawczynski55 observed sig nificant removal of TOC with virtually no reduction in THMFP. Hentz et a156 observed similar results for another water. Amy and Chadik,57 however, re- ported a substantially greater reduction

Duringconventional water treatment, part of the natural organic matter com- prising the THMFP is removed by chem- ical coagulation. Although these re- movals are usually low,55 some have been reported to be as high as 86 per- cent.58 Amy and Chadik57 reported a reduction in the THMFP of Ilwaco Reservoir (Wash.) water from 1,087 to 371 pg/L with an alum dose of 2.5 mg/L (a 66 percent reduction) and to 245 pg/L with an optimum ferric dose of 7.5 mg/L (a 77 percent reduction).

The removal of THM precursors by adsorption on activated carbon has varied from poor to very good. Many studies have reported the need for very high PAC dosages to achieve reductions in THMFP. In a pilot-scale study of conventional treatment processes, Love et a15g investigated the reduction of THMFP by PAC (Table 5). A PAC dose of 100 mg/L was required to reduce the THMFP from 53 to 25 pg/L. In a bench- scale study at New Orleans, La.,43 re- searchers determined that 90 percent reduction of THMFP could be achieved, compared with that obtained by conven- tional treatment, with a PAC dose as high as 500 mg/L (Table 6). Blank runs of conventional treatment with no PAC added resulted in THMFP reduction efficiencies varying from 27 percent in the spring to 54 percent in the fall. Hentz et aP6 determined that PAC doses ranging from 75 to 125 mg/L were required to achieve a slight reduction of 7.5-29 per- cent of a THMFP of approximately 425 pg/L. Their results agree with those of Wood and Demarco,60 Symons,Gi and Lange and Kawczynski,55 who also found PAC to be ineffective for the removal of THM precursors.

The use of PAC was evaluated for the removal of chloroform precursors in a study involving 15 of Kentuckys larger water utilities.62 After the Ohio River

ISSAM N. NAJM ET AL 73 Copyright (C) 1991 American Water Works Association

water had been chlorinated in a jar with a dose of 2-3 mg chlorine/L and then stored for three days, the terminal chloroform concentration was 236 erg/T. (Table 7). As shown in Table 7, a PAC dose of 50 mg/L applied prior to chlori- nation resulted in a 50 percent decrease

in the terminal chloroform concentra- tion. With a PAC dose of 1,000 mg/L, however, 20 percent of the chloroform formation potential (47.5 pg/L) was still not reduced. Addition of PAC to a pilot- scale floe-blanket reactor for the control of THMFP was investigated by Hoehn et al. The addition of 25 mg/L of PAC* to a water containing 362 pg THMFP/L improved the reduction efficiency by a mere 4 percent over the 56 percent reduction achieved by coagulation alone. However, the addition of 21 mg/L of another PAC.t a higher-grade carbon. resulted in an added 1% percent reduction in THMFP.

The PAC adsorption of THM precur- sors present in six US waters was investigated by Amy and Chadik.57 The raw water THMFP varied from 313 E.rg/L in the Mississippi River (La.) to 1,087 pg/L in the Ilwaco Reservoir (Wash.). With optimum alum doses of 15 and 2.5 mg/L as Al, THMFP in the Mississippi River and Ilwaco Reservoir waters was reduced to 130 and 370 &L, respectively. The THMFP was then reduced to 81 and 196pg/L, respectively, with the addition of 50 mg PAC/L. Table 8 lists the results of Amy and Chadik5 for reduction of THMFP in water from the Daytona Beach Aquifer (Fla.) and the Ilwaco Reservoir by using two types of PAC after coagulation. In the aquifer water, the value of THMFP:TOC for the raw water, 72.5 rg/mg, was virtually un- changed after coagulation. The ratio, however, dropped to an average of 50 pg/mg after treatment with 50 mg PAC/L. This suggests that the portion of the natural organic matter that produces THMs is selectively removed by PAC. As for the Ilwaco Reservoir water, Table 8 shows that the THMFP: TOC value dropped from 135 pg/mg for the raw water to 108 pg/mg with the addition of 7.5 mg ferric chloride/l, followed by a drop to 53 rg/mg after treatment with 50 mg PAC/L. This also suggests the selective removal of THM-forming natural organics by PAC.

A review of these findings clearly sug gests that the extent of THMFP reduc- tion by PAC largely depends on the types of PAC and water being used. In the absence of a reliable surrogate param- eter, the THMFP of a water sample needs to be measured directly. A standard method has been proposed for the de- termination of THMFP.63

Adsowtion of THMs. Prechlorina- tion is commonly practiced for the control of bacterial and algal growth in the treatment plant. In such a case, the

74 RESEARCHANDTECHNOLOGY

THM formation reaction starts prior to the addition of PAC. By the time PAC is added, the water will contain some re- maining THM precursors, THMs, and some residual chlorine. However, the oxidation-reduction reaction between chlorine and the surface of the activated carbon will result in the reduction of the adsorption capacity of activated carbon for phenolic compounds, as well as the destruction of the residual chlorine.@% The same decrease in capacity may occur for other compounds. Therefore, the effect of PAC addition on the fin- ished water THM concentration will not necessarily be due to the adsorption of the THM precursors or the THMs; the decrease in the residual chlorine concen- tration will result in lower THM forma- tion,56 which can be misinterpreted as higher THM removals by PAC. On the other hand, the destruction of the carbon surface will result in lower than expected removals of THMs per unit mass of PAC because of the reduction in carbon capacity.

Anderson et ala investigated the con- trol of THMs by PAC addition to the influent to the flocculation basin of a 40-mgd water treatment plant. In the first part of the study, chlorine was added to the raw water prior to the point of addition of PAC. In the second part, the chlorination point was moved to the flocculation basin effluent, allowing some contact between the water and the carbon in the absence of chlorine. With prechlorination, the THM concentration in the filter effluent was reduced from 40.2 to27.5 pg/L with theadditionof 7.3 mg PAC/L, and from 37.4 to 23.8 pg/L with the addition of 21.6 mg PAC/L. When the chlorine was added after the PAC, the THM concentration in the filter effluent dropped from 51.4 rg/L when no PAC was added to 28.2 ccg/L with 19.4 mgPAC/L added. The removals were higher than those predicted using the isotherms of Dobbs et al47 listed in Table 3. However, a good comparison cannot be made because chlorine was present during the adsorption of THMs.

In another full-scale study, Singley et al51 investigated the performanceof PAC added to a softening plant for the control of THMs. Prechlorination of the raw water was also practiced during part of this study. The finished water THM concentration of 40 I*g/L was virtually unchanged after the addition of 14.3 and 26.6 mg PACK.. The three-day THMFP, however, decreased from 135 to 40 rg/L with the addition of the two carbon dosages, and the average TOC of the water was reduced from 14 to 10.4 mg/L (26.6 percent reduction). The THM re- sults may have been affected by the removal of chlorine by the PAC. When the chlorination point was moved to the recarbonation units, the finished water THM concentration was still unchanged

with the addition of 7.1 mg PAC/L. However, the three-day THMFP was reduced from 135 to 110 pg/L, and the TOC was reduced from 12.6 to 4.4 mgiL (65.1 percent reduction). The much higher removals of TOC achieved when the chlorination point was moved (65.1 percent), compared with those achieved during prechlorination (26.6 percent), are consistent with a reduction in PAC capacity for organic compounds in the presence of free chlorine. Cost analysis of PAC use

The most important component of the cost of using PAC is the cost of the PAC itself. Over the last 30 years, the price of PAC has risen from about $O.O7/lb to as high as $.45/lb, an increase of >500 percent. During this time, the wholesale and producers price indexes increased by about 200 percent. The largest in- crease in PAC costs occurred in the mid- 1970s following improvements in PAC products. In addition to the cost of the carbon, treatment costs for PAC usage include capital and operation and main- tenance requirements for process feed equipment.67-70$ Preliminary cost esti- mates for PAC treatment have been developed for a range of system sizes and PAC dosage rates. Table 9 presents the quantity of PAC required (lb/day) as a function of system size and dosage, assuming 70 percent utilization of plant capacity and continuous application of PAC. Design requirements for feed equipment were based on the maximum flow rate and PAC dosage rate. Tables 10 and 11 present annual cost estimates ($/year) for the purchase of PAC and process feed equipment, respectively.$ Table 12 summarizes the unit cost esti- mates for PAC treatment ($/l,OOO gal).

Figure 1 presents unit cost estimates for PAC treatment ($/l,OOO gal) as a function of dosage rate for four system sizes. A significant decrease in cost occurs with increasing size for small systems up to about 1 mgd in capacity. Systems with capacities in the range of l-100 mgd show only small gradual decreases in unit costs with increasing size, given a fixed PAC dosage. For example, assuming a PAC dose of 5 mg/L, the unit treatment cost drops from about $0.10 to $0.03/1,000 gal with increasing system size from 0.1 to 1 mgd. By increasing system capacity from 1 to 100 mgd, the unit costs gradually de- crease from $0.03 to $0.02/1,000 gal. In this analysis, unit costs increase nearly linearly with respect to increasing dosage rate, given a fixed system size.

A typical maintenance dose of PAC for taste and odor control is approximately 5

*Aqua Nuchar. Westvaco Corp.. Covington. W.Va. tNuchar SA. Westvaco Corp.. Covington. W.Va. $All cost figures were updated to 1988. 5An interest rate of 10 percent and an equipment life span

of 20 years were assumed for this analysis.

JOURNALAWWA Copyright (C) 1991 American Water Works Association

mg/L. At this rate, quantities of PAC required for systems of l.O- and 100-mgd. size are about 1,000 and 1,070,OOO lb/ year, respectively. Many systems must increase the PAC dosage rate for a period of time to treat seasonal and occasional variations in water quality. A utility may have to increase the PAC dose from 5 to as much as 75 mg/L for several weeks to reduce the concentrations of synthetic organic contamination result- ing from an occasional chemical spill or seasonal agricultural runoff. For exam- ple, if a lo-mgd system increases its typical maintenance PAC dose rate from 5 to 50 mg/L for four weeks during the year, the additional PAC required is about 73,600 lb above the usual main- tenance dosage. This results in a total PAC requirement of about 180,200 lb/ year. This short-term but significant increase in the quantity of PAC required affects the treatment cost somewhat proportionally. The total cost estimate for the system, assuming only the 5- mg/L dose rate, is about $53,85O/year. By including the four-week increase in PAC dose to 50 mg/L., the total cost increases to about $84,66O/year, an in- crease of about 60 percent. The reduction in the cost of applying PAC can be calculated in a similar manner if there are periods of time when no PAC is required.

Summary and conclusions Powdered activated carbon can be

applied at different points in drinking water plants. In selecting the point of addition of PAC, consideration must be given to the degree of mixing and to minimizing short-circuiting, the ade- quacy of the time of contact between the PAC and the water, the PAC residence time, and the minimization of inter- ference of treatment chemicals with the adsorption process. For example, good mixing can be achieved in the rapid mix and flocculation basin, but PAC in the sedimentation basin may not beeffective because of poor mixing. When PAC is added to some treatment processes, such as floe-blanket reactors, significantly longer carbon residence times are achieved compared with conventional treatment. These longer residence times increase the removal of slowly adsorbing compounds and decrease the carbon usage rate.

The efficiencies of removal of taste- and odor-causing compounds by PAC vary, depending on the type and concen- tration of the compound causing the problem. High efficiencies of removal of compounds such as 2,4-dichlorophenol, geosmin, and 2-methylisoborneol were achieved with relatively low PAC doses (lo-25 mg/L). Doses of PAC ranging from 75 to 620 mg/L, however, were required to remove p-nitrophenol and humic odor.

JANUARY 1991

Powdered activated carbon has been effective for the removal of several biocides, depending on their type and concentration as well as the type of contactor. However, PAC doses as high as 475 mg/L were required to reduce the concentration of metolachlor from 50 to 1 fig/L.

The extent of adsorption of SOCs on PAC is strongly dependent on the typeof compound being removed. In general, PAC has a relatively low capacity for VOCs compared with nonvolatile or- ganics. It should be emphasized, how- ever, that the means of application of PAC, as well as the contact time avail- able, largely determine the percentage removal of these and other trace organics from drinking water.

The reported removals of THM pre- cursors and THMs from drinking water by PAC range from poor to very good. The fraction of natural organic matter that forms THMs varies in different waters. Some studies have shown that the ratio of THMFP to TOC of a natural water decreases after treatment with PAC, indicating the preferential removal of THM precursors by PAC. However, depending on the type of water being treated, few studies have reported re- duction in the TOC of the water with an insignificant drop in THM precursor concentration. Existing data indicate that no good correlation exists between the concentration of THM precursors and TOC in different types of water.

One of the main advantages of PAC is its low capital cost. Cost estimates of the addition of PAC at various dosages to different plant sizes are presented in this article. The annual cost of PAC decreases with increasing plant size. For a lo-mgd plant operating at full capacity and using a continuous PAC dose of 10 mg& the total cost is approximately $0.038/1,000 gal, which amounts to $380/day or $138,7OO/year.

Acknowledgment This work was supported by Lyonnaise

des Eaux, Paris, France; the AWWA Research Foundation; and the US Envi- ronmental Protection Agency. The opin- ions expressed in this article are thoseof the authors and not necessarily those of the supporting agencies.

References AWWA CommitteeRePort. Measurement and Control of Organic Contaminants by Utilities. JOUY. AWWA 69:5:267 (May 1977). AWWA. 1984 Utility Operating Data. Denver, Colo. (1986). GRAESE, S.L.; SNO~YINK, V.L.; & LEE, R.G. Granular Activated Carbon Filter- Adsorber Systems. Jour. A WKA, 79112~3 (Dec. 1987). KRUITHOF, J.C. ET AL. Selection of Brands of Activated Carbon for Adsorptive Prop erties. Activated Carbon in Drinking

5

6.

7.

8.

9.

10.

11

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

Water Technology. KIWA-AWWARF Rept. AWWARF, Denver, Cola. (1983). NAJM, LN. ET AL. Effect of Particle Size and Background Organics on the Ad- sorption Efficiency of PAC. JOUY. A WWA, 82:1:65 (Jan. 1990). AWWA. AWWA Standard for Powdered Activated Carbon. B600-78. Denver, Cola. (1978). MEIJERS, J.A.P. & VANDERLEER, R.C. The Use of Powdered Activated Carbon in Conventional and New Techniques. KIWA-AWWARF Rept. AWWARF, Den- ver, Colo. (1983). RANDTKE. SJ. & SNOEYINK, V.L. Evaluat- ing GAC Adsorption Capacity. Jour. A WWA. 75:8:406 (Aug. 1983). AWWA. Water Q,al& and Treatment. McGraw-Hill, New York (3rd cd., 1971). GAUNTLETT. R.B. & PACKHAM. R.F. The Use of Powdered Activated Carbon in Water Treatment. Proc. Conf. on Acti- vated Carbon in Water Treatment, Univ. of Reading, Water Res. Assn., Medmen- ham, England, Apr. 1973. SONTHEIMER, H. The Use of Powdered Activated Carbon. Translation of Reports on Special Problems of Water Technology, Volume 9-Adsofltion. EPA-600/g-76-030 (Dec. 1976). NAIM, I.N. ET AL. Powdered Activated Carbon in FloeBlanket Reactors. 1989 AWWA Ann. Conf., Los Angeles, Calif. SONTHEIMER, H.: K~LLE, H.; & SPINDLER, P. Rohiil und Trinkwasser, Vertifx de.. Bereichs fiir Wasserchemie. Univ. Karls- ruhe, FRG (1967). DOUGHARTY, I.D. & MORRIS, R.L. Studies on the Removal of Actinomycete Musty Tastes and Odors in Water Supplies. Jour. A WWA, 59:10:1320 (Oct. 1967). GIFFORD, J.S.; GEORGE, D.B.; & ADAMS, V.D. Svnergistic Effects of Potassium Permanganate and PAC in Direct Filtra- tion Systems for THM Precursor Re- moval. Water Res.. 23:lO (Oct. 1989). KASSAM, K. ET A;. Accumulation and Adsorption Capacity of Powdered Acti- vated Carbon on a Slurry Recirculation Clarifier. 1989 AWWA Ann. Conf., Los Angeles, Calif. HOEHN, R.C. ET AL. THM Precursor Control WithPowderedActivatedCarbon in a Pulsed-Bed, Solids Contact Clarifier. 1987 AWWA Ann. Conf., Kansas City, MO. RICHARD, Y. Personal communication (1986). LETTINGA. G.; BEVERLOO, W.A.; & VANLIER, WC. The Use of Flocculated Powdered Activated Carbon in Water Treatment. P~ogy. Water Technol., lo:537 (1978). HABERER, K. & NORMANN, S. Unter- suchungen zu einer Neuartigen Pulver- kohle-Filtrationstechnik fib die Wasser- aufbereitung. Vom Wassey, 49:331(1979). HABERER, K.&Z NORMANN, S. Entwicklung eines Kurztalt-Filtrations Verfahrens zum Einsatz vor Puiverkohle in der Wasseraufbereitung. Gas und Wasser- fach, Wasser/Abwa.&er, 118:393 (1980). HOEHN. R.C. ET AL. A Pilot-Scale Evalua- tion of the Roberts-Haberer Process for Removing Trihalomethane Precursors From Surface Water With Activated Carbon. 1984 AWWAAnn. Conf., Dallas, Texas. SONTHEIMER, H.; CRITTENDEN, J.C.; & SUMMERS, R.S. Activated Carbon for

ISSAM N. NAJM ET AL 75 Copyright (C) 1991 American Water Works Association

Water Treatment. DVGW-Forschungs- stelle, Engler-Bunte-Institut, Universitit Karlsruhe. FRG (2nd ed. lEnelish1 1988).

From Drinking Water. Res. Rept. 111. Univ. of Kentucky Water Resources Res. Inst., Lexington aune 1978). Standard Methods for the Examination of Water and Wastewater. APHA, AWWA, and WPCF. Washington, D.C. (17th ed., 1989). SUIDAN, M.T.; SNOEYINK, V.L.; & SCHMITZ, R.A. Performance Predictions for the Removal of Aqueous Free Chlorine by Packed Beds of Granular Activated Car- bon. Water-1976: I. Physical Chemical Wastewater Treatment. AICHE Sym. Ser., 73:18 (1976). SUIDAN, M.T.; SNOEYINK, V.L.; & SCHMITZ, R.A. Reduction of Aqueous HOC1 With Granular Activated Carbon. Jour. Envir. Engr. Div.-ASCE, 103:677 (1977). ANDERSON, M.C. ET AL. Control of Tri- halomethanes Usine Powdered Carbon. Natl. Conf. on Envi; Engrg. ASCE (July 1980). LEE, R. Personal commun. (Nov. 1988). GUMERMAN, R.C.; RUSSELL, L.; & HANSEN, S.P. Estimating Water Treatment Cost, Vol. II, Cost Curves Applicable to 1 to200 rngd Treatment Plants. EPA-600/2/79- l&b (1979). GUMERMAN, RX.; BURRIS, B.E.; & HANSEN, S.P. Estimation of Small Svstem Water Treatment Costs. USEPA 6bO/S2-84-184 (Mar. 1985). FISHER, J.L. Personal communication (Nov. 1988). DOBBS, R.A. & COHEN, J.M. Carbon Adsorption Isotherms for Toxic Organics. EPA-60018-80-023 (Apr. 1980). EL-DIB, M.A.; MOURSY, A.S.; & BADAWAY, M.I. Role of Adsorbents in the Removal of Soluble Aromatic Hydrocarbons From DrinkingWater. WaterRes., 12:1131(1978). COHEN, J.M. ETAL. Effects of Fish Poison on Water Supplies, Part 1, Removal of ToxicMaterials. Jour. AWWA,52:12:1551 (Dec. 1960). STEINER, J. IV; SINGLEY, J.E.; & EDWARD, J. Methoxvchlor Removal From Potable Water. journal AWWA, 71:5:284 (May 1979).

24. SNOEYINK,.V.L. & SUDAN, CT. &chl& rination by Activated Carbon and Other Reducing Agents. Disinfection: Waterand Wastewater U.D. Johnson, editor). Ann Arbor Sci. P&l., Arm Arbor, Mich.il975).

25. MCGUIRE. MJ. & SUFFET. I.H. Aoueous Chlorine/Act&ated Carbon Interactions. Jour. Envir. Engrg. Div.-AXE, 119: 3:629 (June 1984).

26. LALEZARY-CRAIG, S. ETAL. Optimizing the Removal of Geosmin and 2.Methyliso- borneol by Powdered Activated Carbon. Jour. A WWA, 80:3:73 (Mar. 1988). nn

I. LALEZARY. S.; PIRBAZARI, M.; & MCGUIRE, M.J. Evaluating Activated Carbons for Removing Low Concentrations of Taste- and Odor-Producing Organics. Jour. A WWA, 78:11:76 (Nov. 1986).

29.

30.

31.

32.

33.

28. MILTNER, RJ. ET AL. Final Internal Report on Carbons Use Rate Data. USEPA, Cincinnati. Ohio (Tune 1987). FRICK, B. Adsorptionsgle&hgewichte zwischen Aktivkohle und organischen Wasserinhaltsstoffen in Mehrstoffgemi- schen bekannter und unbekannter Zusammensetzung. Doctoral disserta- tion, Univ. of Karlsruhe, FRG (1980). HANSEN, R.E. The Costs of Meeting the New Water Quality Standards for Total Organics and Pesticides. 1975 AWWA Ann. Conf.. Minneanolis. Minn. BURTSCHEL~, R.H. ET AL. Chlorine De- rivatives of Phenol Causing Taste and Odor. Jour. A WWA, 51:2:205(Feb. 1959). SIGWORTH, E.A. Identification and Re- moval of Herbicides and Pesticides. Jour. A WWA, 57:8:1016 (Aug. 1965). Westvaco Corp. Taste and Odor Control in Water Purification. Westvaco Corp., Covington, W.Va. (1970). BROWN. G.N.: HYNDSHAW. D.B.: & VAN DONGE~. D.B: Powdered Activated Car- bon Removal of Tastes and Odors Plus. Proc. of AWWA Ann. Conf.. 97:2 (1977). ALY, O.M. & FAUST, S.D. Removal of 2,4- Dichlorophenoxy-acetic Acid Derivatives From Natural Waters. Jour. AWWA, 51:2:221 (Feb. 1965). SINGLEY, J.E. & ERVIN, A.L. East Drive Well Field Powdered Activated Carbon Study. Rept. to the City of North Miami Beach, Fla. (Sept. 1978). SYMONS, J.M. ET AL. Interim Treatment Guide for Controlling Organic Contam- inants in Drinking Water Using Granular Activated Carbon. Water S