F I LT R AT I O N

Integrating
high-rate DAF technology
into plant design
Pilot-plant studies of DAF technology indicate
flocculation times can be shortened and hydraulic loading rates
significantly increased without loss of performance.

S
James K. Edzwald,
John E. Tobiason,
Tony Amato, and
Lawrence J. Maggi ince promulgation of the Sur-
face Water Treatment Rule, which effectively required
filtration of surface water supplies, more US water
providers are looking into dissolved-air flotation
(DAF) as a pretreatment
clarification step prior to
Compared with sedimentation, dissolved-air flotation (DAF) is a filtration. Because DAF is
more efficient clarification process for separating floc particles, efficient at removing par-
which are often low in density. This article investigates the use of ticles and turbidity, water
short flocculation times with high DAF and filter hydraulic utilities can integrate
loading rates and examines the feasibility of integrating high-rate more-economical filter
DAF technology into water facility design. Research was designs into new plants.
conducted at pilot scale using two water sources of varying Edzwald and associates1–5
quality. Numerous runs were carried out under extremely showed though model-
conservative cold water conditions of 3–5oC. The most important ing, laboratory research,
research finding is that integration of high-rate DAF treatment and pilot-plant research
technology is feasible. DAF treatment facilities may be designed that flocculation tanks in
and operated with flocculation times of 5 min, DAF hydraulic DAF plants can be de-
loading rates of 30–40 m/h (12–16 gpm/sq ft) depending on water signed with much shorter
temperature, and high filtration rates of 20 m/h (8 gpm/sq ft). At detention times than had
higher DAF loading rates, excess air bubbles can be eliminated by
internal or external air removal methods. For executive summary,
see page 179.

DECEMBER 1999 J.K. EDZWALD ET AL 41

© 1999 American Water Works Association, Journal AWWA December 1999
 the Interim Enhanced Surface
FIGURE 1 Schematic of the dissolved-air flotation pilot plant
Water Treatment Rule
(IESWTR) 6 and the Disin-
Static mixers Varied fectants/Disinfection By-prod-
length
ucts (D/DBP) Rule. 7 The
Dissolved-air IESWTR requires filtered water
flotation tank turbidities of < 0.3 ntu and 2-
Coagulants log removal of Cryptosporidium
by filtration. In addition, the
Two-stage Movable
flocculation Recycle water industry has set a vol-
inlet section
injection untary goal of 0.1 ntu for fil-
External air
tered water turbidity. The
removal IESWTR is an interim rule,
however, and the final rule is
expected to require even
Diffused stricter standards for turbidity,
air Giardia, and Cryptosporidium.
Diffuser Four filters DAF has been shown to
External air removal column was bypassed for experiments
with no air removal and internal air removal be particularly effective in re-
moving Cryptosporidium
oocysts.8–11 DAF and filtra-
TABLE 1 Raw water quality for Harwood’s Mill Reservoir, Newport News, Va.
tion provide two effective
(February–June 1997)
barriers for Cryptosporidium
Parameter Range Mean control. The D/DBP Rule
requires enhanced coagula-
Temperature—oC 7–22 tion as a best treatment tech-
pH 7.1–7.9 7.4
Turbidity—ntu 1.3–3.7 2.4 nology for control of DBP for-
Particle counts (2–199 µm)—number/mL 2,910–8,050 5,320 mation by removal of DBP
Apparent color—cu 20–50 40
Total organic carbon—mg/L 3.8–7.3 5.9
precursors. DAF is more effi-
Dissolved organic carbon—mg/L 3.6–7.0 5.7 cient than sedimentation in
Ultraviolet—cm–1 0.16–0.22 0.19 removing low-density floc
Specific ultraviolet absorbance—m–1mg–1 L 2.8–4.9 3.6
produced from coagulation
of total organic carbon (TOC).
TABLE 2 Raw water quality for Hemlocks Reservoir, Fairfield, Conn. The benefits provided by
DAF have sparked interest
Phase 1 Phase 2 within the US water treatment
10/6/97–11/6/97 12/16/97–1/20/98
industry, and a movement to
Parameter Range Mean Range Mean implement the technology is
under way. Currently, more
Temperature—oC 12–17 3–5
pH 7.0–7.3 7.2 7.5 7.5 than 20 US plants and a few
Turbidity—ntu 0.61–1.25 0.91 0.66–0.74 0.69 Canadian facilities are using
Particle counts (2–199 µm)—
number/mL NM* 5,800–6,080 5,970
DAF in their operations. 12
Apparent color—cu 20–25 23 20 20 Additional plants are under
Total organic carbon—mg/L 2.9–3.2 3.0 2.9–3.3 3.1 construction (e.g., Greenville,
Dissolved organic carbon—mg/L 2.7–3.1 2.8 2.7–3.0 2.9
Ultraviolet—cm–1 0.068–0.071 0.069 0.081–0.089 0.085 S.C., and Cambridge, Mass.)
Specific ultraviolet or are in the design stage (e.g.,
absorbance—m–1mg–1L 2.3–2.6 2.5 2.8–3.2 3.0 New York, N.Y. [Croton sup-
*NM—not measured ply], and Winnipeg, Man.).
The goal of this research
was to investigate the inte-
gration of short flocculation
been used (5–10 min versus 20–30 min). In addition times, high DAF hydraulic loading rates, and high
to treating high-quality water for previously unfil- filtration rates into an overall extension and devel-
tered water supplies, DAF is especially effective in opment of DAF technology. The approach was to
treating supplies with low-to-moderate turbidities, push DAF loading rates to very high levels but use a
supplies with natural color, and supplies with algae high filtration rate that was accepted in practice. The
problems. Furthermore, DAF is designed at higher objectives were to (1) confirm the use of short floc-
hydraulic loading rates than most settling processes culation times of 6–7 min, (2) evaluate the effects of
and is more efficient in removing floc particles. DAF hydraulic loading rates of 17–44 m/h (7–18
In December 1998, the US Environmental Protec- gpm/sq ft) on DAF and filtration performance, (3)
tion Agency issued two new sets of water regulations: investigate the subsequent possibility of air carryover

42 VOLUME 91, ISSUE 12 © 1999 American Water Works Association, Journal AWWA December 1999 JOURNAL AWWA
to filtration and the effectiveness of two FIGURE 2 Dissolved-air flotation (DAF) performance for short
air removal methods, and (4) evaluate flocculation times at Newport News and Fairfield
various filter designs including filter rates
of 10 and 20 m/h (4 and 8 gpm/sq ft). On-line turbidity Degassed turbidity
2.0
Research methods A
Scope and design. The research
examined the quality of DAF-clarified 1.5
water and filtered water as well as filtered
water production. The DAF tank was

Turbidity—ntu
intentionally tested without the use of air
removal methods at hydraulic loading rates 1.0

high enough to cause air bubbles to leave

the tank and enter the filter columns.
Two methods of mitigating air-bubble 0.5
carryover were investigated at high DAF
hydraulic loading rates. In some experi-
ments, an internal air removal (IAR) 0.0
device was placed inside the DAF tank. NAR IAR
Tubes (such as those used in tube settlers)
and plates (plate-settling) were used. They
On-line turbidity Degassed turbidity
served to collect excess air bubbles that 2.0
coalesced on the surfaces and rose to the B
top of the DAF tank. Dahlquist13 describes
an earlier IAR system that was modified
1.5
for the research reported here.
In other experiments, the DAF tank
Turbidity—ntu

effluent was directed though an external

air removal (EAR) process. This consisted 1.0
of downward water flow though a 6-m-
(20-ft-) high air-bubble stripping column
(0.2-m [8-in.] diameter) in which air was 0.5
injected at the base of the column (coun-
tercurrent operation) though a diffuser at
a rate of 4.7 standard L/min, such as prac- 0.0
ticed in countercurrent ozone contactors. NAR IAR NAR IAR
After exiting the EAR column, the water Phase 1 (12–15 C)
o
Phase 2 (5 C)
o

was applied to the filter columns. A—Newport News, 6-min flocculation time, DAF hydraulic loading rate of
An important aspect of the research o
17 m/h (7 gpm/sq ft), 16–19 C; B—Fairfield, 7-min flocculaton time, DAF
was the examination of the integration of hydraulic loading rate of 29 m/h (12 gpm/sq ft); NAR—no air removal;
IAR—internal air removal
high-rate treatment from flocculation
though DAF clarification and filtration.
For DAF clarification, project perfor-
mance criteria were turbidities of 0.5
(goal) to 1 ntu and total residual aluminum (Al) study period. The supply is characterized as one of
(coagulant) < 15 percent of the dose. For filtration, low-to-moderate turbidity and TOC. (Over the course
performance criteria were turbidities of < 0.1 ntu, of the pilot experiments, however, the water tem-
total residual Al of < 0.2 mg/L, and unit filter run vol- perature and raw water quality changed, especially
ume (UFRV)—a measure of filtered water produc- the concentration of organic matter.) The bulk of the
tion—of at least 200 m3/m2 (5,000 gal/sq ft) with a research examining DAF loading rates was conducted
goal of 500 m3/m2 (12,500 gal/sq ft). DAF turbidities between March and May 1997. During this period,
were measured via grab sampling and on-line sam- raw water TOC increased from 4 to 7.3 mg/L, and
pling, using a ratio turbidimeter and low-range tur- water temperature increased from 12 to 22oC. Specific
bidimeter,* respectively. Filtered water turbidities ultraviolet absorbance (SUVA) (UV absorbance per
were measured via grab sampling with the ratio tur- unit of dissolved organic carbon [DOC]) values of
bidimeter. Total particle counts (2–199 µm) were 2.8–4.9 indicate that DOC was a mixture of aquatic
measured on grab samples using a particle counter.† humic and nonhumic dissolved organic matter.14,15
Study sites. Newport News, Va. Pilot experiments Thus, DOC did influence coagulation.
were carried out at Newport News from February to
June 1997 using the Harwood’s Mill Reservoir supply. *Ratio/XR and 1720C, Hach Co., Loveland, Colo.
Table 1 summarizes raw water quality data for the † PC 2400PS, Chemtrac Systems Industries, Norcross, Ga.

DECEMBER 1999 J.K. EDZWALD ET AL 43

© 1999 American Water Works Association, Journal AWWA December 1999
 the two sites, although some fine-tuning
FIGURE 3 Filter turbidity and unit filter run volume (UFRV) at
was done. At Newport News, alum alone
Fairfield for a 7-min flocculation time and a dissolved-air
was used at a dosage of 45–50 mg/L and
flotation hydraulic loading rate of 29 m/h (12 gpm/sq ft)
pH of 6.4–6.7. At Fairfield, alum was used
at dosages of 10–17 mg/L with 1 mg/L of
A NAR IAR
a cationic polymer.* The pH was con-
0.15 trolled at 6.3–7 in the flocculation tank,
depending on the water temperature. The
Newport News plant was operated at a
flow rate of 22.7 m3/h (100 gpm) and a
0.10 total flocculation time of 6 min; the Fair-
Turbidity—ntu

field pilot plant was operated at a rate of

Filter 1

20–21 m3/h (90 gpm) and a 7-min floc-

Filter 1

Filter 2
Filter 1
Filter 2

Filter 2
culation time. The flocculation tank was
Filter 1

Filter 2

0.05
baffled into two stages, and mixing was
provided in each stage with vertical tur-
bine mixers at an intensity of 70 s–1.
The DAF tank had a movable inlet sec-
0.00 tion that allowed variation of the tank’s
horizontal area and hydraulic loading while
B
1,000 25,000 maintaining constant flow and floccula-
tion conditions. Air bubbles were produced
Filter 2

by recycling clarified water at 8 percent of

800 20,000
Filter 2

plant flow through an unpacked saturator

UFRV—gal/sq ft
Filter 2
Filter 2

operated at 520–550 kPa (75–80 psig).

2
UFRV—m /m

600 15,000
At Fairfield, the hydraulic connection
Filter 1
3

Filter 1
Filter 1

between the DAF tank and the filters was

Filter 1

400 10,000 direct gravity flow, similar to that found

in full-scale practice. At Newport News,
200 5,000 the filters were on a floor 8 m (26 ft)
below the DAF tank effluent, which
0 0 necessitated use of a valve between the
Phase 1 (12-15 C) Phase 2 (5 C)
o
DAF tank and filters to reduce the head.
o

Filter 1—low-rate, conventional dual media; filter 2—high-rate, coarse Filter media size, depth, and hydraulic
monomedia; NAR—no air removal, IAR—internal air removal loading were also investigated. DAF-clar-
ified water was applied to four parallel
150-mm- (6-in.-) diameter granular
media filters. Table 3 summarizes filter
media configurations.
Fairfield, Conn. Pilot experiments were also per- Filters 1–3 were identical for both field sites. Filter
formed from July 1997 to March 1998 using a high- 1 was operated at a low rate of 10 m/h (4 gpm/sq ft)
quality New England supply from Hemlocks Reservoir and had 0.60 m (24 in.) of 1-mm anthracite over 0.30
in Fairfield. Raw water data are summarized in Table m (12 in.) of 0.5-mm sand. Filters 2 and 3 were oper-
2. At this site, work was conducted in two phases. ated at a high rate of 20 m/h (8 gpm/sq ft). Filter 2 was
Phase 1 research was carried out in fall 1997 when the a monomedia filter containing 2 m (79 in.) of 1.3-
water temperature was 12–17oC. Raw water quality mm anthracite. Filter 3 was a dual-media design con-
was fairly constant with turbidities of 0.61–1.25 ntu taining 1.70 m (67 in.) of 1.3-mm anthracite over
and TOC of 2.9–3.2 mg/L. Phase 2 work was con- 0.30 m (12 in.) of 0.6-mm sand. Filter 4 was also
ducted in the winter at water temperatures of 3–5oC. operated at the high rate. At Fairfield, filter 4 was a
Turbidities and TOC were also low and constant in monomedia design tested using 2 m (79 in.) of 1.1-mm
phase 2, with turbidities of 0.66–0.74 ntu and TOC of anthracite. At Newport News, filter 4 was a dual-
2.9–3.3 mg/L. SUVA data (2.3–3.2) indicate that the media design tested using 2 m (79 in.) of 1.5-mm
supply contained some aquatic humic dissolved anthracite over 0.30 m (12 in.) of 0.8-mm sand.
organic matter; because DOC was low, however, coag-
ulant dosages were not high. Results and discussion
Pilot-plant setup. Figure 1 shows a schematic of Flocculation time. Prior modeling of the DAF
the pilot-plant setup. The pilot plant included chem- contact zone,1,16 laboratory experiments,2,3 and pilot-
ical addition, rapid mixing via static mixers, floccu- plant studies4,5 have shown that flocculation times as
lation, DAF clarification, and filtration. Coagulation low as 5 min are feasible in DAF plants. For this re-
conditions (type, dosages, and pH) were established
primarily by the practice of the full-scale plants at *Cytec 572C, Cytec Industries, West Paterson, N.J.

44 VOLUME 91, ISSUE 12 © 1999 American Water Works Association, Journal AWWA December 1999 JOURNAL AWWA
search, the objective was not to vary floc-
FIGURE 4 Bottom of whitewater blanket position as a function
culation time but to confirm that floccu- of dissolved-air flotation (DAF) hydraulic loading rate
lation times of about 5 min could be used
in an integrated high-rate DAF treatment
DAF Loading Rate—gpm/sq ft
strategy with high DAF and filtration 4 6 8 10 12 14 16 18
hydraulic loading rates. The total mean 0.0 0
flocculation time was fixed at 6 min at 0.2
Water level at top of tank
8
Newport News and 7 min at Fairfield.

Depth From Top of Tank—in.

Depth FromTop of Tank—m
0.4 16
During the course of the experimen-
tal research, water quality (especially TOC 0.6 24
and water temperature) changed at New- 0.8 32
port News, making it difficult to main- 1.0 40
tain constant coagulation conditions of 1.2 48
dosage and pH. Thus, coagulation was 1.4
Bottom of whitewater blanket
56
characterized as good, but not optimal.
1.6 Position of subnatant 64
In contrast, constant and optimal coagu- water exit
lation conditions existed for each phase of 1.8 72
the Fairfield research. 2.0 80
Figure 2 shows typical DAF turbidity 20 25 30 10
35 15 40 45
DAF Loading Rate—m/h
performance data for Newport News (part o
Data observed for 15–18 C
A) and Fairfield (part B), which confirm
that short flocculation times can be used
in high-rate DAF plants. Newport News
site results (Figure 2, part A) demon- FIGURE 5 Effects of dissolved-air flotation (DAF) loading rate
strated excellent DAF performance, with and air removal on turbidity and particle counts
turbidities < 1 ntu for a hydraulic loading at Newport News at 15–19oC
of 17 m/h (7 gpm/sq ft). DAF with IAR to
reduce air-bubble carryover performed
NAR IAR EAR
slightly better than no air removal (NAR). A
4.5
With or without the IAR device, how-
ever, performance was excellent, with 4.0
only 6 min of flocculation time prior to 3.5
DAF. Part A of Figure 2 shows on-line 3.0
DAF turbidity data as well as grab sam-
Turbidity—ntu

ples. The grab samples were degassed by 2.5

holding samples for 30 min to remove 2.0

any bubbles that might cause an artifi- 1.5
cial turbidity reading. These data show
1.0
little difference in the two readings, indi-
cating air bubbles were not escaping the 0.5
DAF tank at a hydraulic loading of 17 0.0
m/h at water temperatures of 16–19oC. 17 (7) 22 (9) 44 (18)
Part B of Figure 2 shows DAF perfor- B
DAF Loading Rate—m/h (gpm/sq ft)
mance data at Fairfield for phase 1 (fall 4,000
Particle Counts (2–199 µm)—number/mL

water temperatures) and phase 2 (winter 3,500

temperatures). At 29 m/h (12 gpm/sq ft),
the DAF loading rate was higher than at 3,000
Newport News, but the flocculation time 2,500
was approximately the same. For phase 1
(temperatures of 12–15oC), both on-line 2,000
turbidities (particulates and bubbles) and 1,500
degassed turbidities (particulates only)
were very good (<1 ntu). With bubbles 1,000
taken into account, the turbidities attrib- 500
utable to particulates were < 0.5 ntu, indi-
0
cating excellent DAF performance for floc 17 (7) 22 (9) 44 (18)
separation at short flocculation times and DAF Loading Rate—m/h (gpm/sq ft)
high DAF hydraulic loading rates. For Average data shown for more than one experiment; error bars
phase 2 (temperatures of ~5oC), on-line indicate the standard deviation; NAR—no air removal, IAR—internal
turbidities for the NAR case were > 1 ntu. air removal, EAR—external air removal
Degassed turbidities were < 1 ntu, indi-

DECEMBER 1999 J.K. EDZWALD ET AL 45

© 1999 American Water Works Association, Journal AWWA December 1999
 TABLE 3 Summary of filter media characteristics and loading rates

Depth Effective Size Uniform Ratios

Filter Media m (in.) mm Coefficient L/d

Filter 1—low-rate, conventional dual-media; Anthracite 0.60 (24) 0.9–1.0 1.4–1.5 630
10 m/h (4 gpm/sq ft)
Sand 0.30 (12) 0.50–0.55 1.3–1.4 570
Total L/d 1,200
Filter 2—high-rate, coarse monomedia; Anthracite 2.0 (79) 1.25–1.35 < 1.5 1,540
20 m/h (8 gpm/sq ft)
Filter 3—high-rate, dual-media; Anthracite 1.70 (67) 1.25–1.35 < 1.5 1,310
20 m/h (8 gpm/sq ft)
Sand 0.30 (12) 0.55–0.65 1.3–1.4 500
Total L/d 1,810
Filter 4 (Fairfield)—high-rate, fine monomedia; Anthracite 2.0 (79) 1.05–1.15 < 1.5 1,820
20 m/h (8 gpm/sq ft)
Filter 4 (Newport News)—high-rate, coarse dual-media; Anthracite 2.0 (79) 1.44–1.55 < 1.5 1,330
20 m/h (8 gpm/sq ft)
Sand 0.30 (12) 0.80–0.85 1.3–1.4 360
Total L/d 1,690

cating this was largely attributable to air-bubble car- at DAF loading rates of 40 m/h (16 gpm/sq ft) or
ryover. Because the IAR device removed some of the greater. The blanket was retained in the tank at load-
excess air bubbles, on-line turbidities were much less ing rates of about 30 m/h (12 gpm/sq ft) or less.
than with NAR. Degassed turbidities were ~ 0.7 ntu, Although the whitewater blanket was largely retained
indicating excellent performance given the short floc- in the tank at this DAF loading rate, some air bubbles
culation time, high DAF loading rate, and cold water. exited the tank. At rates of 22 m/h (9 gpm/sq ft) or
Data from Fairfield (Figure 3) are used to illustrate less, the blanket of bubbles penetrated less of the
that excellent filtered water quality and filtered water tank depth so air-bubble carryover was minor.
production were achieved with a short flocculation time. During phase 2 at Fairfield (when water temper-
These data are for a high DAF loading rate of 29 m/h (12 atures were 3–5oC), observations indicated the white-
gpm/sq ft) and for warm and cold water. Filtered tur- water blanket penetrated the full depth of the tank at
bidities of ~ 0.05 ntu were achieved for the low-rate the lowest DAF loading rate tested (29 m/h [12
(10 m/h [4 gpm/sq ft]) conventional filter (filter 1) and gpm/sq ft]). This indicates that for colder water tem-
the high-rate (20 m/h [8 gpm/sq ft]) monomedia filter peratures, whitewater blankets sink deeper in the
(filter 2). The other high-rate filters performed similarly. DAF tank. These observations are significant with
Differences in filtered water production are discussed respect to the highest DAF loading rate that can be
later; the authors note here that UFRVs exceeded the used without air-bubble mitigation measures and the
minimum criteria of 200 m3/m2 (5,000 gal/sq ft) for need for and effectiveness of air-bubble removal.
both filter loading rates, whereas the high-rate filter In summary, for warmer water conditions of 15oC
yielded UFRVs > 500 m3/m2 (12,000 gal/sq ft). or greater, the whitewater blanket was retained in
In summary, extensive data collected at both field the tank for loading rates up to 30 m/h (12.3 gpm/sq
sites showed that DAF performed well with floccu- ft). Under cold water conditions of 3–5oC, the blan-
lation times of 6–7 min, producing turbidities of 0.5–1 ket penetrated deeper for the same hydraulic loading
ntu for warm and cold water at DAF loading rates and exited the tank at 29 m/h (12 gpm/sq ft), the
up to 30 m/h (12 gpm/sq ft). Only at higher DAF lowest loading rate tested in phase 2 at Fairfield.
hydraulic loading rates under cold water conditions DAF hydraulic loading rate and air removal.
(3–5oC) did DAF turbidities exceed 1 ntu, but this Newport News. Figure 5 shows effects of hydraulic
was attributable to air-bubble carryover for the base loading rate and air removal method on DAF effluent
(NAR) and IAR cases. DAF turbidities exiting the air- turbidity and particle counts (2–199 µm) for the New-
bubble stripping column (EAR) were about 1 ntu for port News site. Data are for loading rates of 17, 22, and
these extreme conditions. Filter performance was not 44 m/h (7, 9, and 18 gpm/sq ft) at water temperatures
adversely affected by low flocculation times, main- of 15–19oC. Turbidity data are on-line readings that
taining turbidities < 0.1 ntu. reflect both particles and any air bubbles that may
DAF whitewater blanket position. The position be in the DAF effluent.
of the whitewater blanket in the DAF tank was mea- Results indicated excellent DAF turbidities (<1
sured as a function of hydraulic loading rate. These ntu) at 17 and 22 m/h (7 and 9 gpm/sq ft) and poor
measurements were made during experiments with- performance at 44 m/h (18 gpm/sq ft). Both IAR and
out air removal (NAR case). Figure 4 shows results at EAR methods reduced turbidities caused by air bub-
both field sites for water temperatures of 15–18oC. bles, although the effect was much more significant
The whitewater blanket penetrated the full depth at 44 m/h (18 gpm/sq ft). Particle count data in part
of the tank, and air-bubble carryover was extensive B of Figure 5 support the turbidity results, indicating

46 VOLUME 91, ISSUE 12 © 1999 American Water Works Association, Journal AWWA December 1999 JOURNAL AWWA
good performance at loading rates of 17
FIGURE 6 Dissolved-air flotation (DAF) turbidity at Fairfield
and 22 m/h (7 and 9 gpm/sq ft) and poor
as a function of loading rate and air removal for phase 1
performance at 44 m/h (18 gpm/sq ft). As at 12–17oC and phase 2 at 3–5oC
described previously, the whitewater
blanket filled the entire tank depth at a
DAF loading rate of 44 m/h (18 gpm/sq
ft) but not at rates of 17 and 22 m/h (7 NAR—on line IAR—on line
EAR—on line Degassed
and 9 gpm/sq ft). 2.00
Fairfield. The effects of DAF loading Phase 1

NAR
rate and air removal method on DAF efflu- 1.75
ent turbidity were also investigated at the
1.50
Fairfield site. Based on the similarity in
DAF performance for loading rates of 17 1.25

Turbidity—ntu
and 22 m/h (7 and 9 gpm/sq ft) at New-
port News and the apparent failure at 44 1.00

NAR
m/h (18 gpm/sq ft), loading rates of 21, 29,

IAR
and 40 m/h (8.6, 12, and 16 gpm/sq ft) 0.75

IAR
were studied in phase 1 work when water

NAR

EAR

EAR
0.50
temperatures were 12–17oC. Results are

EAR
IAR
shown for both on-line turbidity (partic- 0.25
ulates plus bubbles) and turbidity of
degassed grab samples (particulates). 0.00
Phase 1 results are summarized in 21 (8.6) 29 (12) 40 (16)
Figure 6. The figure shows DAF turbidi- DAF Loading Rate—m/h (gpm/sq ft)
ties < 1 ntu for all DAF hydraulic loading NAR—on line IAR—on line
rates with either IAR or EAR. Even for EAR—on line Degassed
8
NAR, turbidities were < 1 ntu for DAF Phase 2
loading rates up to 29 m/h (12 gpm/sq 7
ft). For NAR, some air-bubble carryover

NAR
occurred at 29 m/h (12 gpm/sq ft), 6
whereas at 40 m/h (16 gpm/sq ft), the
whitewater blanket penetrated the full 5
Turbidity—ntu

NAR
DAF tank depth (Figure 4), causing sig-

IAR
4
nificant air-bubble carryover. A com-
parison of on-line and degassed sample 3
IAR

turbidities (Figure 6) indicates that both

NAR

EAR

air-bubble removal methods significantly

EAR
2
IAR

reduced air bubble carryover at DAF

EAR

loading rates of 29 and 40 m/h (12 and 1

16 gpm/sq ft). As the loading rate
increased, the effect on the true turbid- 0
29 (12) 34 (14) 39 (16)
ity associated with floc particle carryover DAF Loading Rate—m/h (gpm/sq ft)
from DAF was significantly lower than
that for turbidity from air-bubble carry- NAR—no air removal, IAR—internal air removal, EAR—external air removal
over. Particle-related turbidity increased
somewhat with increasing DAF loading
rate yet remained at an excellent level of
approximately 0.3 ntu at the highest rate
of 40 m/h (16 gpm/sq ft).
Phase 2 pilot-plant work at Fairfield
was carried out at DAF loading rates of
29, 34, and 39 m/h (12, 14, and 16 gpm/sq ft) with Turbidities increased with DAF loading rates, although
water temperatures of 3–5oC, which represented a degassed samples indicated that most of the turbidity
conservative test of hydraulic loading rate effects. At was attributable to air bubbles, not particles. The true
these cold water temperatures, the whitewater blan- particle turbidities (degassed readings) were about 1
ket penetrated the entire depth of the DAF tank for ntu or less. For all DAF loading rates, IAR and EAR
all hydraulic loading rates. Phase 2 results (Figure 6) were effective in reducing the turbidity caused by
showed the same trends as in phase 1, but DAF tur- bubble carryover.
bidities were higher for the cold water. DAF effluent Residual Al. Another measure of DAF perfor-
turbidities were about 1 ntu or less at a loading rate mance is residual Al following DAF. Alum added
of 29 m/h (12 gpm/sq ft) with either IAR or EAR. under good coagulation chemistry should form par-

DECEMBER 1999 J.K. EDZWALD ET AL 47

© 1999 American Water Works Association, Journal AWWA December 1999
 At Fairfield, alum dosages were low
FIGURE 7 Filtered water turbidity at Newport News as a function of
(1.4 mg/L as Al), so only absolute total
dissolved-air flotation (DAF) loading rate and air removal
residual DAF Al concentrations were
used to evaluate performance. Soluble
Filter 1 Filter 2
Filter 3 Filter 4 Al was about 0.08 mg/L for all experi-
0.20 ments, indicating good coagulation chem-
NAR
istry. For a water temperature of 17oC,
0.15 total Al following DAF was 0.4 mg/L or
Turbidity—ntu

less for DAF loading rates of 21 and 40

m/h (8.6 and 16 gpm/sq ft) for NAR and
0.10
for both air-bubble carryover removal
methods. For colder water of about 12oC
0.05 and a DAF loading rate of 29 m/h (12
gpm/sq ft), total Al was 0.8 mg/L fol-
0.00 lowing DAF. Overall, data for Newport
17 (7) 22 (9) 44 (18) News and Fairfield indicate that DAF was
DAF Loading Rate—m/h (gpm/sq ft) able to effectively remove alum floc at
0.20 high DAF loading rates.
IAR Filtration performance. Filtration
0.15
performance was assessed by monitor-
ing filtered water turbidity and head loss
Turbidity—ntu

and by determining UFRVs. Each pilot

0.10
experiment included four parallel fil-
ters—one at 10 m/h (4 gpm/sq ft) and
0.05 three at 20 m/h (8 gpm/sq ft)—each with
a different media configuration (Table 3).
0.00 This section is organized as follows:
17 (7) 22 (9) 44 (18) • First, the key findings from New-
DAF Loading Rate—m/h (gpm/sq ft) port News are presented. (This is done
0.20 because the data from Newport News
EAR were used to refine study conditions of
0.15
DAF loading rate and filter design for the
Fairfield site.)
Turbidity—ntu

• Second, a more extensive summary

0.10
of findings is presented for Fairfield. This
sequence is used because of the reasons
0.05 given above and because the flow
hydraulics between the DAF tank and fil-
0.00 ters were better controlled at Fairfield (see
17 (7) 22 (9) 44 (18) the section “Pilot plant setup” on page 44).
DAF Loading Rate—m/h (gpm/sq ft) Coagulation chemistry was good for
Average data shown for more than one experiment; error bars indicate Newport News and very good for Fair-
the standard deviation; filter 1—low-rate, conventional dual media, filter
2—high-rate coarse monomedia, filter 3—high-rate dual media, filter
field; all filters showed typical behavior of
4—high-rate coarse dual media; NAR—no air removal, IAR—internal air ripening (occurred within 1 h), steady-
removal, EAR—external air removal state removal of turbidity and particles,
and breakthrough. Turbidity was mea-
sured on an hourly basis during the
steady-state portion of the filter run. Dur-
ticulate Al that is effectively removed by DAF. The ing this period, turbidity readings did not vary signif-
authors measured both soluble Al and total Al (sum icantly; therefore summary bar charts are presented.
of particulate and soluble). Total Al was used to Turbidity data over a filter cycle are presented later.
assess performance because soluble Al concentra- UFRVs were calculated based on turbidity breakthrough
tions were low. of 0.2 ntu or terminal head loss of 2.4 m (8 ft).
At Newport News, alum dosages were 4.3±0.2 Newport News. Figures 7 and 8 summarize per-
mg/L as Al. At these high dosages, DAF performance formance for the four filters for each DAF loading
was evaluated on a percent removal basis and by the rate and air removal method. If more than one exper-
absolute DAF residual concentration. Total Al iment was carried out at a specific loading and air
removals of 85–90 percent were achieved for all DAF removal method, then data were averaged, and error
loading rates, with or without air-bubble carryover bars (standard deviation) and the average are shown.
removal. DAF effluent total Al concentrations were Figure 7 shows that filtered water turbidities of
about 0.5–0.6 mg/L. 0.04–0.05 ntu were achieved consistently for DAF

48 VOLUME 91, ISSUE 12 © 1999 American Water Works Association, Journal AWWA December 1999 JOURNAL AWWA
loading rates of 17 and 22 m/h (7 and 9
FIGURE 8 Unit filter run volume (UFRV) at Newport News as a
gpm/sq ft) for all filter designs. This cor-
function of dissolved-air flotation (DAF) loading rate and
responds to good DAF-treated water air removal
being applied to the filters (as shown in
Figure 5). Good filtration performance
(<0.10 ntu) was found for IAR at a DAF
loading rate of 44 m/h (18 gpm/sq ft). Filter 1 Filter 2
Filter 3 Filter 4
For NAR and EAR at this rate, turbidi- 20,000
800
ties were higher for all filters except the NAR
low-rate filter. Filtration performance for
600

UFRV—gal/sq ft
15,000
filter 2 (high-rate, coarse monomedia)

2
UFRV—m /m
and filter 4 (high-rate, coarse dual-media)

3
400 10,000
was fair (NAR) or poor (EAR). For filter
3 (high-rate dual-media), performance
was good for NAR and poor for EAR. 200 5,000
UFRV data for the four filters are
shown in Figure 8. Runs with the high- 0 0
17 (7) 22 (9) 44 (18)
rate filters were generally terminated DAF Loading Rate—m/h (gpm/sq ft)
based on turbidity breakthrough (at 0.2 800 20,000
ntu), whereas the low-rate filter was IAR
terminated on head loss (2.4 m [8 ft]) 600

UFRV—gal/sq ft
15,000
2

for DAF loading rates of 17 and 22 m/h

UFRV—m /m
3

(7 and 9 gpm/sq ft). Data in Figure 8 400 10,000

show that DAF loading rate and air
removal affected UFRV. High UFRVs
200 5,000
(>400 m3/m2 [10,000 gal/sq ft]) were
found generally for the NAR and IAR
0 0
cases at DAF loading rates of 17 and 22 17 (7) 22 (9) 44 (18)
m/h (7 and 9 gpm/sq ft). Low UFRVs DAF Loading Rate—m/h (gpm/sq ft)
(ⱕ200 m3/m2 [ⱕ5,000 gal/sq ft]) were 800 20,000
EAR
found for a rate of 44 m/h (18 gpm/sq
ft) and for most experiments with EAR. 600 15,000

UFRV—gal/sq ft
2
UFRV—m /m

Filter media type had less effect on

3

UFRV, but some trends were observed. 400 10,000

Filter 4 (high-rate, coarse dual-media)
typically had lower UFRVs than the 200 5,000
other high-rate filters. This was because
turbidity breakthrough occurred earlier 0 0
for this filter, despite a lower rate of 17 (7) 22 (9) 44 (18)
head loss development. DAF Loading Rate—m/h (gpm/sq ft)
Fairfield. Based on the Newport News Average data shown for more than one experiment; error bars indicate
findings, the authors decided to investi- the standard deviation; filter 1—low-rate conventional dual media, filter
2—high-rate coarse monomedia, filter 3—high-rate dual media, filter
gate DAF loading rates between approx- 4—high-rate coarse dual media; NAR—no air removal, IAR—internal air
imately 20 and 40 m/h (8 and 12 gpm/sq removal, EAR—external air removal
ft) at Fairfield. Because of the poor per-
formance of filter 4 (high-rate, coarse
dual-media), it was replaced with a high-
rate, fine monomedia filter (see Table 3).
As previously noted, research at Fair-
field was divided into two phases. Phase
1 was conducted in the fall when water tempera- Figure 9 shows the effect of DAF loading rate on
tures measured 12–17oC. The experiments exam- filtered water turbidity for each air removal case.
ined the effect of DAF loading rate on DAF perfor- With NAR and IAR, DAF loading rates did not affect
mance and were run for 8 h. Phase 2 was conducted filtered water turbidity. Turbidity ranged between
at water temperatures of 3–5oC. Filter performance approximately 0.04 and 0.06 ntu, with the excep-
was evaluated over a 24-h period unless terminated tion of the high-rate dual-media filter (filter 3), which
because of turbidity breakthrough (0.2 ntu) or head produced slightly lower turbidity. The data for the
loss (2.4 m [8 ft]). Phase 2 filter data are summa- EAR case indicate poor turbidity performance and
rized in the following paragraphs because of the con- the onset of breakthrough for the high-rate mono-
servative water temperatures and the use of 24-h media filters (filters 2 and 4) at the two higher DAF
filter runs. loading rates.

DECEMBER 1999 © 1999 American Water Works Association, Journal AWWA December 1999 J.K. EDZWALD ET AL 49
 FIGURE 9 Filtered water turbidity at Fairfield as a function
iments for these two air removal cases
of dissolved-air flotation (DAF) loading rate and air resulted in UFRVs > 250 m3/m2 (6,140
removal for phase 2 at 3–5oC gal/sq ft) for all filters, indicating good
filtered water production, given the high
Filter 1 Filter 2 DAF hydraulic loading rates and cold
Filter 3 Filter 4 water conditions. UFRVs for the high-
0.20
NAR rate filters (filters 2–4) were also consis-
0.16 tently much greater than the UFRV for
the low-rate filter (filter 1). UFRV val-
Turbidity—ntu

0.12 ues were low for EAR and again showed

the effect of earlier turbidity break-
0.08 through with the EAR system.
0.04
Figure 11 shows the effects of air
removal and filter design on filtered water
0.00 turbidity for a DAF loading rate of 39
29 (12) 34 (14) 39 (16) m/h (16 gpm/sq ft). In all filter design
DAF Loading Rate—m/h (gpm/sq ft) cases, EAR had the earliest turbidity
0.20 breakthrough. It occurred earliest for the
IAR high-rate monomedia filters (filters 2 and
0.16 4), then for the high-rate dual-media fil-
Turbidity—ntu

ter (filter 3), and last for the low-rate

0.12
dual-media filter (filter 1). No turbidity
0.08 breakthrough was found for the low-rate
dual-media filter for the NAR and IAR
0.04 cases. Furthermore, for NAR and IAR,
the high-rate dual-media filter delayed
0.00
turbidity breakthrough the longest, fol-
29 (12) 34 (14) 39 (16)
DAF Loading Rate—m/h (gpm/sq ft)
lowed by the high-rate, fine monomedia
filter (filter 4).
0.20
EAR In summary, air removal methods
0.16 affected filtration. For NAR and IAR, sim-
ilar filtration performance was observed.
Turbidity—ntu

0.12 Air-bubble carryover was reduced by

IAR. Any air bubbles carried to the filter
0.08 columns from the DAF tank with IAR
and all the air bubbles leaving DAF for
0.04
the NAR case rose to the surface in the
0.00 water column above the filter media and
29 (12) 34 (14) 39 (16) did not penetrate the media. Thus, excel-
DAF Loading Rate—m/h (gpm/sq ft) lent filtration performance was observed
Filter 1—low-rate, conventional dual media, filter 2—high-rate coarse for all DAF and filter loading rates with
monomedia, filter 3—high-rate dual media, filter 4—high-rate fine regard to project criteria for turbidity and
monomedia; NAR—no air removal, IAR—internal air removal,
EAR—external air removal
UFRV. For the EAR case, air bubbles car-
ried over from DAF were removed in the
stripping column, but new air was dis-
solved at the base of the stripping col-
Figure 10 shows the effect of DAF loading rate on umn and was later precipitated as bubbles within the
UFRV for each air removal case. UFRV results are not filter bed as head loss increased during filtration. This
included for the 29-m/h (12-gpm/sq ft) DAF load- caused precipitated air accumulation in the filter and
ing rate with EAR. This particular experiment was earlier turbidity breakthrough.
ended 11.5 h from the start because of an equip- Residual Al. For both sites, filtered water total
ment problem with the DAF recycle system. At the residual Al was 0.05–0.10 mg/L, meeting project per-
time the experiment ended, the high-rate, deep- formance criteria and indicating good coagulation
bed monomedia filters (filters 2 and 4) were begin- chemistry and treatment.
ning to show signs of turbidity breakthrough. A
UFRV is not shown for filter 4 at a DAF loading rate Application of results to practice
of 34 m/h (14 gpm/sq ft) with IAR because this fil- This research demonstrates the concept of inte-
ter had to be taken off line early in the run because grating individual water treatment processes into an
of equipment problems. overall strategy that uses short detention times and
With NAR and IAR (Figure 10), UFRV tended to high hydraulic loading rates. The concept of inte-
decrease with increasing DAF loading rate. All exper- grated water treatment plant design is based on the

50 VOLUME 91, ISSUE 12 © 1999 American Water Works Association, Journal AWWA December 1999 JOURNAL AWWA
premise that DAF is an efficient and key
solid–liquid separation process and that FIGURE 10 Unit filter run volume (UFRV) at Fairfield as a function
of dissolved-air flotation loading rate and air removal
treatment prior to DAF (coagulation–floc-
for phase 2 at 3–5oC
culation) significantly affects its perfor-
mance. Likewise, DAF performance
affects downstream filtration. Filter 1 Filter 2
Pretreatment. Pretreatment consists Filter 3 Filter 4
of coagulant addition and flocculation. 1,000
NAR
24,000
The DAF tank has two functions. First, 800 19,000

UFRV—gal/sq ft
floc particles are contacted with air bub-

PHL

PHL
UFRV—m /m
bles; second, the floc particle–bubble ag- 14,000

3
600

HL

TB
HL

HL
TB
gregates rise to the surface. Pretreatment

TB
PHL

TB

PHL
9,000

PHL
400
is directly linked to the first step. 16
Coagulation conditions of dosage and 200 5,000
pH are critical in preparing floc parti- 0
0
cles so that air bubbles stick to floc par- 29 (12) 34 (14) 39 (16)
ticles in the contact zone of the flotation DAF Loading Rate—m/h (gpm/sq ft)
tank. Optimal coagulation conditions 1,000 24,000
IAR
are those that produce hydrophobic floc
800 19,000

UFRV—gal/sq ft
particles with little or no surface charge.
2
UFRV—m /m

PHL

PHL
Optimal coagulation conditions for flota-
3
600 14,000

HL

TB
HL

TB
tion are the same as for settling

HL
TB
PHL

PHL
9,000

PHL
processes, but the flocculation require- 400

ment is different. 200 5,000

Sedimentation requires large particles
0 0
(usually hundreds of microns in size) for
29 (12) 34 (14) 39 (16)
efficient settling and removal. Because DAF Loading Rate—m/h (gpm/sq ft)
DAF enables efficient contacts between air 1,000 24,000
bubbles and floc particles to be achieved EAR
with floc particles only tens of microns 800 19,000

UFRV—gal/sq ft
2
UFRV—m /m

in size,16 short flocculation times can be 14,000

3

600
used. The authors’ research verified that
flocculation times of nominally 5 min are 9,000

TB
400
TB

TB
Equipment
TB
TB
TB

feasible in DAF plants using high DAF

TB

TB

problem 5,000
200
and filtration loading rates. These find-
ings were demonstrated at two water 0 0
supplies and for cold water temperatures. 29 (12) 34 (14) 39 (16)
DAF Loading Rate—m/h (gpm/sq ft)
DAF. Historically DAF tanks have been
designed with hydraulic loading rates of UFRV basis: TB—turbidity breakthrough 0.2 ntu, HL—head loss of 2.4 m
(7.9 ft); PHL—projected head loss of 2.4 m (7.9 ft); filter 1—low-rate,
5–10 m/h (2–4 gpm/sq ft).17 These load- conventional dual media, filter 2—high-rate coarse monomedia, filter
ing rates stem from DAF experiences in 3—high-rate dual media, filter 4—high-rate fine monomedia; NAR no air
Scandinavia and the United Kingdom. In removal, IAR—internal air removal, EAR—external air removal
the United States, a large DAF facility has
been built in Fairfield, Conn., with a load-
ing rate of 15 m/h (6 gpm/sq ft), and
there is interest in designing DAF plants
in Boston and New York (Croton supply) with higher conditions and reaches the bottom when the water is
loading rates. In the United States and other countries, cold. In warm water, DAF loading rates as high as
differences in water quality and pretreatment may 30 m/h (12 gpm/sq ft) can be used and excellent DAF
make higher hydraulic loading rates feasible. performance can be achieved with or without mea-
This research set out to examine DAF performance sures to reduce air-bubble carryover. The loading rate
as a function of hydraulic loading rate. The authors can be increased to 40 m/h (16 gpm/sq ft), and DAF
also identified hydraulic loading rates when the white- turbidities of < 1 ntu can be achieved when either
water blanket exits the DAF tank and investigated IAR or EAR methods are used. For cold water con-
mitigation methods to reduce air-bubble carryover ditions, DAF turbidities of ~ 1 ntu are achieved with
to the filters. According to their findings, small quan- IAR or EAR.
tities of air bubbles exit the DAF tank, and DAF per- Filtration performance. In the integrated design
formance is excellent for hydraulic loading rates as concept, the effect of DAF performance on filtration
high as 20 m/h (8 gpm/sq ft). The whitewater blan- is an important consideration; consequently, the
ket approaches the bottom of the DAF tank at a load- authors examined the effect of DAF hydraulic load-
ing rate of 30 m/h (12 gpm/sq ft) under warm water ing rates on filtration. Filtered water quality and water

DECEMBER 1999 © 1999 American Water Works Association, Journal AWWA December 1999 J.K. EDZWALD ET AL 51
 ters as air bubbles, causing turbidity
FIGURE 11 Filter turbidity at Fairfield as a function of run time
breakthrough and additional head loss.
for the four filter designs and air removal
at a dissolved-air flotation loading rate of 39 m/h
Design measures must be taken to avoid
(16 gpm/sq ft) for phase 2 at 3–5oC this. In some cases, the dissolved air will
be released as water is conveyed to full-
NAR IAR EAR scale filters. However, the filters must
0.30
Filter 1 be designed to avoid negative head
0.25
within the filter media. This potential
Turbidity—ntu

0.20
problem of air precipitation within the
0.15
filters has important practical signifi-
0.10 cance for all water plants using inter-
0.05 mediate ozone prior to filtration,
0.00 whether they are settling plants or DAF
0.25
Filter 2 facilities.
Turbidity—ntu

Research results for the different fil-

0.20
ter designs suggest that higher water
0.15 production can be achieved with the
0.10 higher-rate filters with coarser media.
0.05 Whereas allowable head loss typically
0.00
governed filter run length for the low-
Filter 3 rate filters, turbidity breakthrough typ-
0.25
ically governed filter run length for the
Turbidity—ntu

0.20
high-rate filters, despite increasing the
0.15 ratio of total bed depth to media diam-
0.10 eter from 1,200 L/d for the 10-m/h (4-
0.05 gpm/sq ft) conventional dual-media fil-
ter to as much as 1,800 L/d for the
0.00
Filter 4 20-m/h (8-gpm/sq ft) filters. Head loss
0.25 data showed that particles penetrated
Turbidity—ntu

0.20 much deeper into the coarser media at

0.15 the higher filter loading rate. Applica-
0.10 tion of a filter aid polymer may have
0.05
been able to delay turbidity break-
through for the coarse media filters. The
0.00 deep-bed dual-media filter (filter 3) is a
0 5 10 15 20 25
Time—h good filter design for an integrated high-
rate DAF treatment plant. It delayed tur-
Filter 1—low-rate conventional dual media, filter 2—high-rate coarse
monomedia, filter 3—high-rate dual media, filter 4—high-rate fine bidity breakthrough and provided good
monomedia; NAR—no air removal, IAR—internal air removal, filter performance.
EAR—external air removal
Conclusions
The most important finding of this
research is that integration of high-rate
production were measured, as well as any effects of DAF treatment technology is feasible. DAF treatment
air-bubble carryover on performance. facilities may be designed and operated with floccu-
Filtration performance was examined under con- lation times of 5 min, DAF hydraulic loading rates
servative cold water conditions and met or exceeded of 30–40 m/h (12–16 gpm/sq ft) depending on water
criteria at DAF loading rates of 29, 34, and 39 m/h (12, temperature, and high filtration rates of 20 m/h (8
14, and 16 gpm/sq ft) for NAR and IAR. Unacceptable gpm/sq ft). This research also showed that the deep-
filtration performance was found for the EAR sys- bed dual-media filter—1.7 m (67 in.) of anthracite
tem at DAF loading rates of 34 and 39 m/h (14 and (1.3-mm effective size) over 0.30 m (12 in.) of sand
16 gpm/sq ft). The EAR stripping column removed air (0.6-mm effective size)—is a good filter design for
bubbles that carried over from DAF but dissolved an integrated high-rate DAF treatment plant.
additional air that precipitated within the filters, caus- At the higher DAF loading rates, removal of excess
ing earlier turbidity breakthrough. air bubbles can be accomplished by IAR or EAR meth-
This finding means that DAF plants that use ods. The EAR system did have a negative effect on fil-
intermediate ozone can be designed at high DAF tration, dissolving air because of water pressure at
loading rates at which any air bubbles leaving DAF the base of the stripping column. This potential prob-
will most likely be stripped out in the ozone con- lem is not unique, however, and may exist for water
tactors. However, the intermediate ozone contactors plants with intermediate ozone contactors, regard-
may dissolve air that is later precipitated in the fil- less of the upstream clarification process. The problem

52 VOLUME 91, ISSUE 12 © 1999 American Water Works Association, Journal AWWA December 1999 JOURNAL AWWA
can be solved in practice by maintaining a positive 8. HALL, T. ET AL. Cryptosporidium Removal During
head within the filter. Water Treatment Using Dissolved-air Flotation.
For major cities with large water treatment facil- Water Sci. & Technol., 31:3/4:125 (1995).
ities, high-rate treatment technology offers advan- 9. PLUMMER, J.D. ET AL. Removal of Cryptosporidium
tages. Integration of high-rate DAF treatment can parvum From Drinking Water by Dissolved-air
lead to lower capital costs and smaller plant foot- Flotation. Jour. AWWA, 87:9:85 (Sept. 1995).
prints. This research has led to the development of 10. K ELLEY, M.B. Removal of Cryptosporidium by
DAF RapideTM,* an integrated high-rate DAF treat- Selected Drinking Water Treatment Processes.
ment plant strategy. Doctoral dissertation, Rensselaer Polytechnic
Institute, Troy, N.Y. (1996).
Acknowledgment 11. EDZWALD, J.K. & KELLEY, M.B. Control of Cryp-
Funds for this research were provided by Purac tosporidium: From Reservoirs to Clarifiers to Fil-
Ltd., Anglian Water Innovation, and the University ters. Water Sci. & Technol., 37:2:1 (1998).
of Massachusetts at Amherst. The authors thank 12. NICKOLS, D. & CROSSLEY, I.A. State of the Art of
Peter Harvey of Purac Ltd. and Pat Green of Anglian Dissolved-air Flotation in the United States of
Water Engineering for initiating and facilitating the America. Dissolved-air Flotation. Chartered Insti-
research project; Newport News, Va., water treat- tution of Water and Envir. Mgmt., London
ment personnel (especially Brian Ramaley, Michael (1997).
Hotaling, and Randall Hawkins) for their coopera- 13. DAHLQUIST, J. The State of DAF Development and
tion and assistance during the pilot research exper- Applications to Water Treatment in Scandinavia.
iments at the Harwood’s Mill Water Treatment Plant; Dissolved-air Flotation. Chartered Institution of
Howard Dunn, Gary Kaminski, and John Schultz Water and Envir. Mgmt., London (1997).
at BHC Company for their assistance with the pilot 14. EDZWALD, J.K. ET AL. Surrogate Parameters for
studies at the William S. Warner Treatment Plant Monitoring Organic Matter and Trihalomethane
in Fairfield, Conn.; and Jan Dahlquist of Purac AB Precursors in Water Treatment. Jour. AWWA,
(Lund, Sweden) and Colleen Walsh for their con- 77:4:122 (Apr. 1985).
tributions to the work. 15. EDZWALD, J.K. & VANBENSCHOTEN, J.E. Aluminum
Coagulation of Natural Organic Matter. Chemical
References Water and Wastewater Treatment (H.H. Hahn and R.
1. MALLEY, J.P. JR. & Edzwald, J.K. Concepts for Klute, editors). Springer-Verlag, New York
Dissolved Air Flotation Treatment of Drinking (1990).
Waters. Jour. Water Supply Res. & Technol. (Aqua), 16. EDZWALD, J.K. Principles and Applications of Dis-
40:1:7 (1991). solved-air Flotation. Water Sci. & Technol., 31:3/4:1
2. EDZWALD, J.K. ET AL. Flocculation and Air Require- (1995).
ments for Dissolved-air Flotation. Jour. AWWA, 17. GREGORY, R. ET AL. Sedimentation and Flotation.
84:3:92 (Mar. 1992). Water Quality and Treatment. McGraw Hill, New
3. EDZWALD, J.K. ET AL. Dissolved-air Flotation: Pre- York (5th ed., 1999).
treatment and Comparisons to Sedimentation.
Chemical Water and Wastewater Treatment III (H.H. About the authors: James K.
Hahn and R. Klute, editors). Springer Verlag, Edzwald* is a professor in the Depart-
New York (1994). ment of Civil and Environmental
4. DAHLQUIST, J. ET AL. High-rate Flocculation, Fil- Engineering at the University of
tration in Potable Water Treatment. Chemical Massachusetts, Amherst, MA 01003-
Water and Wastewater Treatment IV (H.H. Hahn 5205. He holds a PhD from the Uni-
et al, editors). Springer-Verlag, New York versity of North Carolina at Chapel
(1996). Hill and MS and BS degrees from the
5. VALADE, M.T. ET AL. Pretreatment Effects on Par- University of Maryland at College Park. Edzwald has 35
ticle Removal by Flotation and Filtration. Jour. years’ experience in the water field and is well known for his
AWWA, 88:12:35 (Dec. 1996). research on coagulation and dissolved-air flotation and for
6. USEPA. National Primary Drinking Water Reg- development of the specific ultraviolet absorbance concept
ulations: Interim Enhanced Surface Water Treat- to characterize natural organics in water supplies. John E.
ment; Final Rule. Part V. EPA, 40 CFR Parts 9, Tobiason is an associate professor of civil and environmen-
141, and 142. Fed. Reg., 63:24:69478 (Dec. 16, tal engineering at the University of Massachusetts at
1998). Amherst. Tony Amato is technology manager for Purac Ltd.,
7. USEPA. National Primary Drinking Water Reg- Birmingham Rd., Kidderminster, Worcester, DY10 2SH
ulations, Disinfectants and Disinfection Byprod- England. Lawrence J. Maggi is a project engineer with
ucts; Final Rule. Part IV. EPA, 40 CFR Parts 9, CH2M Hill Inc., 13921 Park Center Rd., Herndon, VA
141, and 142. Fed. Reg., 63:241:69390 (Dec. 16, 20171; at the time of this research, Maggi was a graduate
1998). student at the University of Massachusetts, Amherst.

*Purac Ltd., Kidderminster, England *To whom correspondence should be addressed

DECEMBER 1999 © 1999 American Water Works Association, Journal AWWA December 1999 J.K. EDZWALD ET AL 53