r:
OF
ERIE,
Report
ENTAL TREATMENT
ERIE WATER
ENNSYLVANIA9
U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE
Federal Water Pollution Control Administration
-------
tiM vil
PROJECT REPORT
EXPERIMENTAL TREATMENT OF LAKE ERIE WATER
ERIE,PENNSYLVANIA, WATER PLANT
by
Kenneth A. Dostal and Gordon G. Robeck
Engineering Activities
Basic and Applied Sciences Program
Federal Water Pollution Control Administration
Cincinnati Water Research Laboratory
March 1966
U.S. Department of Health, Education, and Welfare
Robert A. Taft Sanitary Engineering Center
Cincinnati, Ohio
-------
-------
LIST OF TABLES
Table Page
1. Raw water characteristics 8
2. Characteristics of media used in experimental filters 17
3. Summary of filter runs 20
4. Incremental head losses for run 5 37
5. Summary of algae data 68
6. Influence of filtration rate on water production 69
7. Influence of filter media on water production 70
8. Influence of alum dose on water production 71
9. Unit water production for runs 5-8 of spring series 72
10. Expected length of run for various assumed production values 73
-------
LIST OF FIGURES
Figure Page
1. Frequency distribution of raw water turbidity values, May 5, 1964 3
to May 5, 1965.
2. Frequency distribution of coliform organisms in raw water, 4
May 5, 1964 to May 5, 1965.
3. Frequency distribution of plankton in raw water, Jan. 19, 1965 to 6
May 8, 1965.
4. Frequency distribution of plankton in raw water at Buffalo, N.Y., 7
October 1960 to Aug. 1965.
5. Frequency distribution of daily averages of raw water turbidity during 10
test periods.
6. Frequency distribution of plankton in raw water during test periods. 11
7. Details of dual-media filter head loss taps. 13
8. Size distribution of sand used in dual-media filter. 15
9. Size distribution of coal used for dual-media filter. 16
10. Head loss and turbidity values for run 1 of summer series, 19
Aug. 13, 1964.
11. Head loss and turbidity values for run 2 of summer series, 24
Aug. 17, 1964.
12. Head loss and turbidity values for run 3 of summer series, 25
Aug. 22, 1964.
13. Head loss and turbidity values for run 4 of summer series, 27
Aug. 24, 1964.
14. Head loss and turbidity values for run 5 of summer series, 28
Aug. 25, 1964.
15. Head loss and turbidity values for run 6 of summer series, 29
Aug. 26, 1964.
16. Head loss and turbidity values for run 7 of summer series, 30
Aug. 27, 1964.
17. Head loss and turbidity values for run 1 of fall series, 33
Nov. 18, 1964.
18. Head loss and turbidity values for run 2 of fall series, 35
Nov. 19, 1964.
19. Head loss and turbidity values for run 3 of fall series, 36
Nov. 20, 1964.
-------
LIST OF FIGURES (Contd.)
Figure Page
20. Head loss and turbidity values for run 4 of fall series, 38
Nov. 21, 1964.
21. Head loss and turbidity values for run 5 of fall series, 39
Nov. 22, 1964.
22. Head loss and turbidity values for run 6 of fall series, 41
Nov. 23, 1964.
23. Head loss and turbidity values for run 7 of fall series, 43
Nov. 24, 1964.
24. Head loss and turbidity values for run 8 of fall series, 44
Nov. 25, 1964.
25. Head loss and turbidity values for run 9 of fall series, 46
Nov. 26, 1964.
26. Head loss and turbidity values for run 10 of fall series, 47
Nov. 26, 1964.
27. Head loss and turbidity values for run 1 of winter series, 49
Feb. 11, 1965.
28. Head loss and turbidity values for run 2 of winter series, 51
Feb. 12, 1965.
29. Head loss and turbidity values for run 3 of winter series, 52
Feb. 13, 1965.
30. Head loss and turbidity values for run 4 of winter series, 54
Feb. 14, 1965.
31. Head loss and turbidity values for run 5 of winter series, 55
Feb. 15, 1965.
32. Head loss and turbidity values for run 6 of winter series, 56
Feb. 16, 1965.
33. Head loss and turbidity values for run 1 of spring series, 57
May 18, 1965.
34. Head loss and turbidity values for run 2 of spring series, 58
May 19, 1965.
-------
LIST OF FIGURES (Contd.)
Figure Page
35. Head loss and turbidity values for run 3 of spring series, 60
May 20, 1965.
36. Head loss and turbidity values for run 4 of spring series, 61
May 21, 1965.
37. Head loss and turbidity values for run 5 of spring series, 62
May 22, 1965.
38. Head loss and turbidity values for run 6 of spring series, 64
May 23, 1965.
39. Head loss and turbidity values for run 7 of spring series, 65
May 25, 1965.
40. Head loss and turbidity values for run 8 of spring series, 67
May 26, 1965.
-------
Project Report
EXPERIMENTAL TREATMENT OF LAKE ERIE WATER
by
Kenneth A. Dostal and Gordon G. Robeck
This study resulted from a discussion with representatives
of the Bureau of Water for Erie, Pennsylvania, concerning various
possible approaches for expansion of their municipal water plants.
Two small experimental filters were installed in one of the water
treatment plants by the Public Health Service as part of a continuing
study in which various aspects of water treatment are being reviewed.
The study was designed to check the influence of high-rate and
dual-media filtration coupled with the elimination of conventional
flocculators and settling tanks on the finished-water quality.
A series of 61 filter runs was made during the 1-year study
2
at filtration rates varying from 2 to 6 gpm/ft . The addition of
5 to 15 mg of alum per liter to the raw water just prior to filtration
through dual-media filters resulted in an effluent of very good quality,
Both turbidity and algae were consistently reduced to low levels.
Length of filter run depended upon the raw water turbidity, alum
dose, filtration rate, and size of coal used in the dual-media
filters.
Prior to this study, segments of the conventional water
treatment process (coagulation, flocculation, sedimentation, and
filtration) had been reviewed in three cooperative field studies.
The first field study , at a 3-mgd water treatment plant in Gaffney,
This work was performed while the authors were members of the Division
of Water Supply and Pollution Control, Public Health Service,
Department of Health, Education, and Welfare.
-------
- 2 -
South Carolina, was designed to determine the effect of shortened
flocculation and sedimentation times plus higher filtration rates
2
on the quality of the finished water. A similar study was
conducted in a 4-mgd water treatment plant located in Easley,
South Carolina. The Easley plant utilized sand as the filter
media, and the Gaffney plant utilized anthracite. In the second
study several runs were made with a small experimental filter
containing a coarse coal layer over a layer of fine sand. The
3
third study involved small experimental filters to check the
influence of both high-rate and dual-media filtration on the
finished-water quality and was conducted at a water plant that uses
a cold lime-soda softening process.
RAW WATER SOURCE AND QUALITY
Lake Erie serves as the raw water source for both of Erie's
municipal water plants, the West Plant and the Chestnut Street
Plant. Each plant has a separate intake, about a mile offshore
and 24 feet below the surface of the water. From May 5, 1964
to May 5, 1965, plant personnel made 259 turbidity measurements
on the raw water entering the treatment plant. A frequency distribution
of these determinations is shown in Figure 1. Fifty percent of
the daily turbidity measurements were 3.5 Jackson units or less,
and 90 percent were below 15 J.u. During the same 12-month period
197 measurements were made on the coliform content of the raw water;
this information is shown in Figure 2. Fifty percent of the determinations
yielded a value of 3.5 coliforms per 100 ml or less, and 90 percent
of the reported values were less than 50 per 100 ml. On January 19, 1965,
-------
- 3 -
50
00
OC
20
-; 10
T T
I I I I
NUMBER OF OBSERVATIONS = 259
I I
I I
I I I I
I I I I I I II I
10 20 30 40 50 60 70 80 90 95
PERCENT EQUAL TO OR LESS THAN
1
98 99 99.8
Figure 1. Frequency distribution of raw water turbidity values, May 5, 1964
to May 5, 1965.
-------
_ 4
en
CD
CJ
100
50
20
10
NUMBER OF OBSERVATIONS = 197
I
I I
I
I I
10 20 30 40 50 60 70 80 90 95 98 99
PERCENT EQUAL TO OR LESS THAN
Figure 2. Frequency distribution of coliform organisms in raw water, May 5, 1964
to May 5, 1965.
-------
- 5 -
the water treatment plant personnel started a routine program of
algae identification and enumeration. Figure 3 presents a frequency
distribution for the 58 measurements made prior to May 8, 1965. Fifty
percent were less than 350, and ninety percent of the values were
less than 800 per ml.
Inasmuch as the algae data shown in Figure 3 covered only a
4-month period, additional information was desirable. The only other
algae data found for Lake Erie were from the Public Health Service
Water Pollution Surveillance System, which has a sampling station
at the water treatment plant in Buffalo, New York. The Buffalo
intake is slightly over a mile from shore in about 12 feet of water.
A frequency distribution of 98 counts over nearly a 5-year period is
shown in Figure 4. Water characteristics at this plant are very
similar to those encountered at the water plants in Erie, Pennsylvania.
A comparison of Figures 3 and 4 shows that the algae counts are
fairly similar; therefore, the 4-month data (Figure 3) may be
reasonably representative of what would be expected throughout a
year or longer.
The raw water characteristics, listed in Table 1, for the
four periods in which the experimental work was completed indicate
the turbidity ranged from 2 to 58 J.u., algae counts ranged from
30 to 960/ml, and the temperature varied from 34 to 69°F.
-------
2,000
o
C3
I
ec.
I,POO
500
200
100
I I
I I I I | I I
I I
NUMBER OF OBSERVATIONS = 58
I I
I I I
10 20 30 40 50 60 70 80 90 95 98 99
PERCENT EQUAL TO OR LESS THAN
Figure 3. Frequency distribution of algae in raw water, Jan. 19, 1965 to
May 8, 1965.
-------
5,000
2,000
1,000
cu
«=c
5OO
200
100
T T
I I I I I I I
II III
NUMBER OF OBSERVATIONS = 98
I I
I I I 1 I
I I
10 20 30 40 50 60 70 80 90 95
PERCENT EQUAL TO OR LESS THAN
I I
98 99
99.8
Figure 4. Frequency distribution of algae in raw water at Buffalo, N.Y.,
Oct.,1960 to Aug.,1965.
-------
- 8 -
Table 1. RAW WATER CHARACTERISTICS
Hardness
Date Temp., (as CaCO ) Alkalinity Turbidity, Algae,
From To p_H ^F mg/liter mg/liter J.u. No./ml
8/13/64 8/28/64 7.8-8.0 68-69 128-131 93-98 1-16 430-840
11/18/64 11/27/64 7.4-7.8 47-52 124-135 91-94 5-58 220-960
2/11/65 2/19/65 7.7-8.2 34-41 123-137 86-95 3-18 30-220
5/18/65 5/27/65 7.3-7.8 52-58 130-139 91-95 2-24 80-190
-------
- 9 -
In order to see whether the algae numbers and turbidities
encountered during the test periods were representative of those
measured by the Erie water plant personnel, frequency distributions
were also plotted for the study periods. Figure 5 presents the frequency
distribution of the raw water turbidities (daily averages) found during
the various filter runs. Based on this information the filtration
experiments handled influents slightly higher in turbidity than
expected for the 12-month period (Figure 1). Algae counts
during the test periods were slightly lower than expected, as shown
by the frequency distribution on Figure 6. This was probably due to
the facts that the water plant algae data (Figure 3) covered only 4
consecutive months whereas the experiments were conducted during each
of the four seasons and that both sets of observations were somewhat
limited in number.
DESCRIPTION OF WATER PLANTS
In 1914 the Chestnut Street water treatment plant was completed
with a design rate of 24 mgd. Expansion in 1925 increased the capacity
to 32 mgd, which sufficied until 1932. At that time West Plant with
a design rate of 15 mgd was completed; it was expanded in 1951 to 28 mgd,
which gave a total capacity of 60 mgd.
Since only raw water was used at the West Plant, a brief description
of this plant will suffice. Raw water is treated with alum prior to
passing over weirs into the flocculation basins. Mechanical flocculation
of about 30 minutes is provided prior to sedimentation. Design
detention time in the sedimentation basins is about 3.5 hours. The
2
filters were designed to operate at a rate of 2 gpm/ft . Both pre-
and post-chlorination is practiced, and powdered activated carbon is
-------
100
i I i l I T
50
NUMBER OF DAILY AVERAGES * 34
20
± 10
(FROM FIGURE I)
I I I / I I I i I I I I I
2 5 10 20 30 40 50 60 70 80 90 95
PERCENT EQUAL TO OR LESS THAN
98 99 99.8
Figure 5. Frequency distribution of daily averages of raw water turbidity
during test periods.
-------
11.
I.OOO
500
200
100
50
20
10
I I I I
1 I I
I I I
I I
(FROM FIGURE 3)
I I
NUMBER OF OBSERVATIONS = 26
I i
12 5 10 20 30 40 50 60 70 80 90 95 98 99
PERCENT EQUAL TO OR LESS THAN
Figure 6. Frequency distribution of algae in raw water during test periods.
-------
- 12 ._
added when necessary for control of tastes and odors. When tannins
and lignins are encountered in the raw water, potassium permanganate
is added. Soda ash is added when necessary to raise pH.
TEST EQUIPMENT AND PROCEDURES
The equipment used in this study consisted of two small
experimental filters, pumps for feeding chemicals and controlling the
rate of flow, a head loss recorder for continuous measurement of the
head loss at four points in each of the filters, and two turbidimeters.
One turbidimeter continuously measured and recorded the raw water
turbidity, and the other was used to measure and record the effluent
turbidities from the two small filters and one of the full-scale filters.
Each of the experimental filter containers was constructed of 4-inch-I.D.,
clear plexiglass. The 24 inches of filter media was supported by a
perforated plate and screen. Figure 7 shows the locations of the
four points in the filter selected for head loss measurement. Incremental
head loss measurements are required with dual-media filters to determine
the location of floe deposition.
Throughout this study media sizes were varied in the dual-media
filter. About half of the filter runs were conducted with a media
combination referred to as "standard." This combination, used often in
past experimental work, consists of 18 inches of coal with an effective
size (e.s.) of 1.00 mm and a uniformity coefficient (u.c.) of 1.11
over 6 inches of sand (e.s. = 0.49 mm, u.c. = 1.14). Since both media
size ranges in the standard were very narrow by design and thus not
readily available for full-scale plants, other media sizes were
selected from commercially available materials. All filter media other
than the standard contained 18 inches of coal over 6 inches of sand
-------
. 13 .
4= TOTAL
Figure 7. Details of dual-media filter headless taps.
-------
- 14 -
with an e.s. = 0.43 mm and a u.c. = 1.62 (see Figure 8). At various
times the size of the coal was altered. The initial size distribution
of the coal is shown in Figure 9 (e.s. = 0.93 mm, u.c. = 1.71). For
the filter labeled "1.0", the coal was sieved and everything smaller
than 1.0 mm was discarded. In a full-scale filter this would be
accomplished by removing the top 13 percent of the coal layer following
backwashing. Other filters with the nomenclature of "1.2", "1.4", and
"1.7" contained commercially available coal with material smaller than
1.2, 1.4, and 1.7 mm, respectively, removed prior to adding the 18
inches of coal to the filter. With the size of coal purchased, the
amounts of coal discarded ranged from 13 percent for the "1.0" filter
media to 70 percent for the "1.7" media. Regardless of media size,
the same weights of sand and coal were added to every filter. Table 2
lists the effective size and uniformity coefficient for the various
coal and sand layers used in the filters.
Prior to starting each filtration run, the experimental
filters were backwashed thoroughly to assure cleanliness and the media
was then packed to a predetermined level. In addition to the continuous
measurement of incremental head losses and influent and effluent
turbidities, grab samples were taken for temperature, pH, alkalinity,
and total algae determinations. Each run was continued until one or
both of the filters had a turbidity breakthrough or the total
head loss exceeded 7 feet.
-------
I I
- 15 -
I I I
1.0
0-9
0.8
- 0.7
CO
ca
0.6
CJ
CO
0.5
0.4
0.3
EFFECTIVE SIZE = 0.43 mm
UNIFORMITY COEFFICIENT = 1.62
I I I I I I
I I I
I
I 2
10 20 30 40 50 60 70 80 90 95
PERCENT EQUAL TO OR SMALLER THAN
98 99
Fipre 8. Size distribution of sand used in dual-media filter.
-------
16
2.0
1.5
r-j
oo
CJ
1.0
0.5
I I
I 2
I I
I I I I
EFFECTIVE SIZE = 0.93 mm
UNIFORMITY COEFFICIENT = 1.71
I_ I I I I I I
10 20 30 40 50 60 70 80 90 95 98 99
PERCENT EQUAL TO OR SMALLER THAN
Figure 9. Initial size distribution of coal used for dual-media filter.
-------
17-
Table 2. CHARACTERISTICS OF MEDIAS USED IN
EXPERIMENTAL FILTERS
Nomenclature
Standard
1.0
1.2
1.4
1.7
Media
type
Coal
Sand
Coal
Sand
Coal
Sand
Coal
Sand
Coal
Sand
Depth,0
inches
18
6
18
6
18
6
18
6
18
6
Effect ive
size , mm
1.00
0.49
1 .14
0.43
1.29
0.43
1.46
0.43
1.72
0.43
Uniformity
coefficient
1.11
1.14
1.45
1 .62
1.33
1.62
1.23
1.62
1.15
1.62
a -
Similar weights added in each case: f&eai, 1,960 grams based on specific
gravity = 2.65 and 40 percent porosity; and feffiffty 2,880 grams based on
specific gravity = 1.55 and 50 percent porosity.
-------
- 18 -
RESULTS
Turbidity
Summer Series - Table 3 contains a summary of the results from
61 filter runs made during four seasons in 1964 and 1965. In all of
the runs, alum was added to the raw water entering the top of the
filter. The first series or "summer series" of seven runs, conducted
from August 13-27, 1964, was designed to check the influence of
filtration rate on both the length of run and the effluent quality
using the "standard" filter.
Filter influent and effluent turbidities along with the
incremental head losses for the seven runs are presented in Figures 10
through 16. In the first run, shown in Figure 10, both filters were
fed 20 mg of alum per liter and were operated at filtration rates of
2
4 and 6 gpm/ft . During this run, the influent turbidity averaged
16 J.u. and the effluent turbidities were consistently below 0.1 J.u.
Because of the high alum dose and poor head-loss distribution, the
2
run lengths were very short, 6 hours at the 4 gpm/ft filtration rate
and only 2 hours at the higher rate. As seen by the head-loss curves
for both filters, most of the removal took place in the top 6 inches
2
of the filter. Filter 2 at 6 gpm/ft did not result in any appreciable
2
improvement in floe distribution over filter 1 at 4 gpm/ft . As
shown later in this study by runs with similar influent turbidity
levels, 20 ing/liter was probably an excessive amount.
-------
20
10
CD
CC.
- 19
RAW
CD
CC.
1.0
0.5
0
8
CO
c-o
2
0
FILTER I EFFLUENT
_ FILTER I AT 4 gpm /sq ft
1.0
0.5
CO
cc
c/o
c/o
CD
a:
LLJ
FILTER 2 EFFLUENT
0
FILTER 2 AT 6 gpm/sq ft
ALUM = 20 mg/ I
2 3
TIME, hours
Fi£ure 10. Head loss and turbidity values for run 1 of summer series,
Aug. 13, 1964.
-------
- "20 -
TABLE 3. SUMMARY OF FILTER RUNS
Alum
Run Filter Filter Rate, Dos&
No. No. Media gpm/ft mg/liter
Filter
Raw Effluent Length
Turbidity, Turbidity, Final of run,
J.u. J.u. Head loss,ft hr.
Summer Series
1
2
3
4
5
6
7
1
2
3
4
5
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
1
2
1
2
1
2
1
2
std.a
Std.
Std.
Std.
Std.
Std.
Std.
Std.
Std.
Std.
Std.
Std.
Std.
Std.
i.ob
1.0
Std.
1.2C
Std.
1.2
Std.
1.2
Std.
4
6
4
6
2
4
6
2
6
2
4
2
4
4
4
4
4
4
4
6
6
6
6
20
20
10
10
10
10
10
10
10
10
10
10
10
5
10
10
10
10
10
5
5
10
10
16
15
3
3
3
3
4
4
3
4
3
3
2
2
Fall Series
25
8
7
28
24
> 35
> 35
> 35
> 35
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
1
1
1
1
1
1
1
1
1
1
1
1
I
1
1
L
1
1
1
8
2
1
1
7
7
7
7
7
7
7
5
7
7
7
7
7
4
7
7
6
7
8
7
7
7
7
.4
.0
.8
.8
.6
.8
.0
.4
.6
.5
.8
.6
.6
.0
.4
.6
.8
.4
.0
.2
.2
.2
.2
4.
2.
9.
9.
29.
9.
6.
17.
6.
19.
8.
23.
9.
21.
8.
13.
9.
9.
6.
8.
6.
5.
3.
9
0
0
0
5
0
0
0
0
0
5
0
0
5
5
0
0
0
4
0
0
0
0
-------
- 21 -
TABLE 3. SUMMARY OF FILTER RUNS (Contd.)
Filter
Run
No.
6
7
8
9
10
Filter
No.
1
2
1
2
1
2
1
2
1
2
Filter
Media
1.4d
Std.
1.4
Std.
1.76
Std.
1.7
Std.
1.7
Std.
Rate 2
gpm/f t
2
2
6
6
2
2
6
6
4 -
4
Alum
Dose
mg/liter
10
10
10
10
10
10 ;
10
10
10
10
Raw
Turbidity
J.u.
> 35
> 35
22
20
13
13
> 35
> 35
> 35
> 35
Effluent
Turbidity,
J.u.
0.1
0.2
0.1
0.1
0.3
0.1
0.1
0.1
0.1
0.1
Final
Head loss, ft
7.2
7.0
7.3
7.9
3.0
7.6
7.2
7.2
7.3
7.0
Length
of run,
hr.
22.0
13.0
6.4
5.0
20.0
20.0
6.5
5.0
13.5
8.0
Winter Series
1
2
3
4
5
6
1
2
3
4
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
Std.
1.0
Std.
1.0
Std.
1.0
Std.
1.4
Std.
1.4
Std.
1.4
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
2
2
6
6
4
4
2
2
4
4
6
6
3
3
5
5
3
3
5
5
10
10
10
10
10
10
10
10
10
10
10
10
Spring
10
5
10
5
7.5
12.5
7.5
12.5
4
4
4
4
9
9
8
8
7
7
8
8
Series
5
5
20
20
7
7
7
7
0.1
0.1
0.1
0.2
0.1
0.2
0.2
0.2
0.1
0.2
0.1
0.3
0.1
1.0
0.4
4.0
0.4
0.2
0.5
0.2
7.0
3.8
7.0
7.0
7.1
6.8
7.0
4.8
6.7
7.0
7.0
7.0
6.9
2.8
7.1
2.8
7.0
7.0
6.8
7.0
19.5
20.0
3.1
4.9
7.5
9.5
21.0
24.0
5.6
10.0
3.6
7.2
18.5
24.5
6.8
7.0
24.0
18.0
9.6
6.8
-------
. 22 _
TABLE 3. SUMMARY OF FILTER RUNS (Contd.)
Filter
Alum Raw Effluent Lengtli
Run
No.
5
6
7
8
Filter
No.
1
2
1
2
1
2
1
2
Filter
Media
1.0
1.4
1.0
1.4
1.0
1.4
1.0
1.4
Rate
gpm/ft
5
5
3
3
5
5
3
3
Dose
mg/liter
12.5
12.5
7.5
7.5
7.5
7.5
12.5
12.5
Turbidity
J .u.
5
5
3
3
2
2
3
3
Turbidity,
J .u .
0.1
0.2
0.3
0.3
0.2
0.2
0.1
0.1
Final
Head loss, ft
7.2
7.2
7.3
7.0
7.0
7.0
7.0
6.1
of run
hr.
6.0
9.6
25.6
38.0
9.3
14.1
16.0
20.0
a - Std. = 18 inches of coal (e.s. = 1.00 mm, u.c. = 1.11) over 6 inches of sand
(e.s. = 0.49 mm, u.c. = 1.14).
b - 1.0 = 18 inches of coal (originally: e.s. = 0.93 mm, u.c. = 1.71) nfter all
particles less than 1.0 mm removed, over 6 inches of sand (e.s. = 0.43 mm,
u.c. = 1.62).
c - 1.2 - same as b except particles less than 1.2 mm removed.
d - 1.4 - same as b except particles less than 1.4 mm removed.
e - 1.7 - same as b except particles less than 1.7 mm removed.
-------
- 23 -
In the second run, shown in Figure 11, the raw water turbidity
dropped to 3 J.u. and the alum dose was reduced to 10 rag/liter while
2
the filtration rates were held at 4 and 6 gpm/ft . Both filters again
produced an effluent with a turbidity less than 0.1 J.u., but there was
an improvement in the length of run. The importance of floe distribution
throughout the bed depth is shown by this run. For example, the top
6 inches of coal in filter 1 had a head loss of 5.7 feet upon termination
of the run, whereas the top 6 inches of filter 2 had a head loss of 4.0
feet. Both filters had the same total head loss after 9 hours, thus
2
filter 2 operating at a 6-gpm/ft rate produced 50 percent more water
than filter 1.
In run 3 (Figure 12) the filtration rates were changed to 2 and
2
4 gpm/ft and the alum dose was held at 10 mg/liter. Both effluent
turbidities were consistently below 0.1 J.u. with a run length for
2
filter 1 at 2 gpm/ft of 29.5 hours and for filter 2 of 9 hours.
One of the reasons for filter 1 lasting more than twice as long as
filter 2 was probably due to the lower raw water turbidity during the
12- to 24-hour period.
A very poor effluent was produced during the first 3 hours of run
4 by filter 1, but this was a result of a breakdown in the alum feed.
Filter 2 produced an effluent with a turbidity less than 0.1 J.u. and
reached a total head loss of 5.3 feet in 17 hours, with nearly all of
the removal occurring in the top 6 inches of the coal.
-------
10
_ 5
CD
- 24 -
RAW
1.0
0.5
8
OO
00
C3
=£
2
0
FILTER I EFFLUENT
i
MEDIA = STANDARD
CO
cc.
1.0
0.5
0
8
00
oo
CO
FILTER 2 EFFLUENT
FILTER 2 AT 6gpm/sq ft
ALUM = 10 mg/ I
MEDIA= STANDARD
4 6
TIME, hours
8
10
Figure 11. Head loss and turbidity values for run 2 of summer series,
Aug. 17, 1964.
-------
. 25 .
CD
0=
10
5
0
1.0
0.5
C/0
CO
RAW
FILTER I EFFLUENT
I
FILTER I AT 2 gpm/sq ft
ALUM = 10 mg/l
MEDIA = STANDARD
CD
CC.
1.0
0.5
0
8
FILTER 2 EFFLUENT
FILTER 2 AT 4 gpm/sq ft
ALUM = 10 mg/|
MEDIA = STANDARD
12 18
TIME, hours
24
30
Figure 12. Head loss and turbidity values for run 3 of summer series,
Aug. 22, 1964.
-------
- 26 -
Figure 14 shows run 5 in which filter 1 produced a very good
2
effluent for 6 hours at a filtration rate of 6 gpm/ft . Filter 2 at
2
2 gpm/ft reached the same head loss after 19 hours and produced a
similar effluent. Even though all the removal took place in the top
of the coal for filter 2, it lasted slightly more than three times as
long as filter 1. This was probably a result of a lower initial head
loss for filter 2.
In run 6 (Figure 15) both filters produced an effluent with a
2
turbidity less than 0.1 J.u. Filter 1 at 4 gpm/ft operated 8.5 hours,
2
and filter 2 at 2 gpm/ft was stopped at a similar head loss after
23 hours. The higher filtration rate resulted in a slightly deeper
penetration of the floe, but not enough to appreciably lengthen the
run since the length of run for filter 2 was more than twice that
of filter 1.
In the last run of the summer series (Run 7, Figure 16) both
2
filters were operated at the same rate, 4 gpm/ft with filter 1 receiving
10 mg/liter of alum and filter 2 receiving 5 mg/liter. Although there
was no noticeable difference in the effluent turbidity, the floe distribution
in the filter was markedly affected by the lower alum dose of filter 2.
After 9 hours of operation, filter 1 had a total head loss of only
2.1 feet. Throughout the run the distribution of floe removal in
filter 2 utilized much more of the bed depth than filter 1 or any
of the other filter runs in this series. If filter 2 had been continued
to a 7-foot head loss, there may have been a breakthrough during the
run inasmuch as a considerable amount of floe was being removed by
the sand layer. This is shown by the incremental head loss of 1.6 feet
-------
10
CO
cc.
CO
cc.
1.0
0.5
0
8
6
C/3
CO
o
4
RAW
FILTER I EFFLUENT
1 ,
0
FILTER I AT 6 gpm/sq ft
ALUM = 10 mg/l
MEDIA - STANDARD
CD
cc.
1.0
0.5
0
8
FILTER 2 EFFLUENT
FILTER 2 AT 2 gpm/sq ft
ALUM - I0mg/l
MEDIA = STANDARD
Figure 13 Head loss and turbidity values for run 4 of summer series,
Aug. 24, 1964.
-------
- 28 -
=! 10
± 5
o
CO
oc.
1.0
0.5
DQ
CC.
FILTER I EFFLUENT
00
CD
LU £
1 I
FILTER I AT 6 gpm/sq ft
ALUM -- 10 mg/l
MEDIA = STANDARD
= 1.0
^ 0.5
FILTER 2 EFFLUENT
FILTER 2 AT 2 gpm/sq ft
ALUM = JO mg/l
MEDIA = STANDARD
Figure 14. Head loss and turbidity values for run 5 of summer series,
Aug. 25, 1964.
-------
- 29 -
10
CO
cc.
CO
CC.
1.0
0.5
0
8
c/o
CO
C3
RAW
I I
FILTER I EFFLUENT
FILTER I AT 4gpm/sq ft
ALUM = 10 mg/l
MEDIA > STANDARD
FILTER 2 EFFLUENT
MEDIA - STANDARD
10 15
TIME, hours
Figure 15. Head loss and turbidity values for run 6 of summer series,
Aug. 26, 1964.
-------
- 39 -
CO
oc
10
5
0
i
= i.o
!= 0.5
CD
CC.
8
6
c/o
c/o
es
RAW
FILTER I EFFLUENT
FILTER I AT 4 gpm/sq ft
ALUM -- 10 mg/l
MEDIA = STANDARD
1.0
5 0.5
CO
CC
? 0
8
c/o
c/o
FILTER 2 EFFLUENT
2
0
FILTER 2 AT 4 gpm/sq ft
ALUM = 5 mg/l
MEDIA = STANDARD
10 15
TIME, hours
Figure 16. Head loss and turbidity values for run 7 of summer series
Aug. 27, 1964.
-------
- 31 -
between pressure taps 2 and 3. Judging from this run, it is quite
probable that a lower alum dose in many of the first six runs of
this series would have achieved the desired clarity and markedly lengthened
the filter run. Another point illustrated by this run is that alum
in excess of the amount needed for obtaining the desired clarity
strengthens the floe much as a filter aid does. The extra alum also
results in more suspended solids, which have to be removed by the filter.
Fall Series
From November 18 through 26, 1964, the second or fall series of
runs was conducted. This is the period of the year when more turbidity
is generally encountered in the raw water because high winds accompany
frequent storms.
One change was made in the experimental conditions used in the
first series of runs. Since both the coal and sand size ranges were
very narrow in the first series of runs, the media in one of the filters
was replaced with coal and sand of more practical size ranges. The media
in filter 2 was replaced with 6 inches of commercially available filter
sand (e.s. = 0.43 mm, u.c. = 1.62) and 18 inches of coal. Initially
the commercially available coal had an effective size of 0.93 mm and
a uniformity coefficient of 1.71; but before the coal layer was added
to the small filter, some of the fines were removed. In the first two
runs, filter 1 contained 18 inches of coal from which all of the
particles smaller than 1.0 mm had been removed by sieving and 6 inches
of sand. In later runs in this series coal sizes varied because coal
particles finer than 1.2, 1.4, and 1.7 mm were removed at different
times prior to adding the 18 inches of coal.
-------
- 32 -
Figure 17 shows the first run of the fall series. Filter 2 was
not operated because of difficulties with the effluent rate-of-flow
controller. Even though the raw water or filter influent turbidity was
higher than for any run in the summer series, 10 mg/liter of alum and a
2
4 gpm/ft filtration rate still resulted in an effluent turbidity below
0.1 J.u. Because of the wider particle size range of the coal (see Table 2),
a more effective use was made of the coal layer in this run than in
any of the previous runs. As shown by the head-loss curves, the top
6 inches of the coal had a head loss of 2.9 feet upon termination of the
run and the next 12 inches of the coal (between head-loss taps 1 and 2)
had an incremental head loss of 3.3 feet.
Both filters in run 2 (Figure 18) produced a very clear water,
with filter 1 lasting over 30 percent longer than filter 2 because of
the coal size. The combination of coal size, filtration rate, and alum
dose for filter 1 resulted in a much better distribution of head loss.
The first 6 inches of coal had a headless of 2.6 feet, or 0.43 foot of
head loss per inch of bed depth, and the next 12 inches of coal had an
incremental head loss of 4.1 feet (between taps 1 and 2), or 0.35 foot per
inch of bed depth.
Before the start of run 3, the media from filter 1 was replaced
with new sand plus coal from which particles less than 1.2 mm had been
removed. As with filter 1 in the first two runs, the larger coal particles
allowed a deeper penetration of the floe into the coal layer without any
noticeable deterioration in the effluent quality as measured by
turbidity (Figure 19).
-------
- 33 -
CD
40
30
20
10
0
RAW
i+- »35 J.u
± 0.5 |
CD
oe
1
_ FILTER
I
1
I
1 1 1
1 EFFLUENT _
1 1 1
r i r
00
e/i
CD
8
7
6
4
3
I I
FILTER I AT 4 gpm/sq ft
ALUM * I0mg/|
MEDIA = 1.0
4 6
TIME, hours
8
10
Figure 17. Head loss and turbidity values for run 1 of fall series
Nov. 18, 1964.
-------
- 34 -
The influence of raw water turbidity on length of run is
readily seen by comparing runs 2 and 3 (Figure 18 and 19). Both runs
2
at 4 gpm/ft and 10 rag/liter alum produced an effluent with a turbidity
of about 0.1 J.u. The average raw water turbidity in run 3 was about
three times that of run 2, and this resulted in a shortening of the
run for filter 2 from 9 to 6.4 hours. Filter 1 in run 3 contained a
larger size coal than it did in run 2, but the length of run was still
reduced from 13 to 9 hours.
For run 4, which is shown in Figure 20, the alum was reduced
2
to 5 mg/liter and the filtration rate was increased to 6 gpm/ft . The
reduction in alum coupled with an increase in both raw water turbidity
to over 35 J.u. and filtration rate resulted in filter effluent turbidities
higher than those for any of the previous runs. Filter 2 produced an
effluent with a turbidity of 0.5 J.u. or less, but filter 1 with the
larger media produced an effluent with a turbidity range of 0.3 to 1.9 J.u.
This passage of turbidity by filter 1 is also indicated by the incremental
head-loss curves, which show a penetration of floe into both the upper
(between taps 2 and 3) and lower half (between taps 3 and 4) of the sand
layer.
After the alum dose was increased 10 mg/liter, run 5 (Figure 21)
2
was started with the filtration rate held at 6 gpm/ft . Even though the
raw water turbidity was greater than 35 J.u., both filters produced
an effluent with a turbidity of 0.1 J.u.; but the length of run was
reduced markedly for both filters as compared to the previous run.
The floe penetrated deeper in filter 1 than in filter 2, and, thus, an
increase in length of run resulted for filter 1. Table 4 shows the
incremental head losses for both filters for run 5.
-------
- 35 -
20
10
DO
= 1.0
£ 0.5
OQ
oe n
CO
CO
~~ FILTER 1 EFFLUENT ~~
\
'
FILTER I AT 4gpm/ft2
ALUM = 10 mg / I
MEDIA = 1.0
CD
CC.
OO
FILTER 2 AT 4 gpm / sq ft
I I
FILTER 2 AT 4 gpm / sq f t
ALUM = I0mg/l
TIME, hours
Figure 18. Head loss and turbidity values for run 2 of fall series,
Nov. 19, 1964.
-------
40
20
- 36 -
-|>35 J.u.
1.0
0.5
= 0
i
8
^ f*
CD
00
c/o
2
0
FILTER I EFFLUENT
\
FILTER I AT 4 gpm / sq ft
ALUM = 10 mg/l
MEDIA = 1.2
= 1.0
£ 0.5
CD
CD _
0= 0
=>
8
1
FILTER 2 EFFLUENT
J
1
00 A
CO 4
£=1
-=£
I I
FILTER 2 AT 4 gpm / sq ft
ALUM = 10 mg/ I
4 6
TIME, hours
8
10
Figure 19. Head loss and turbidity values for run 3 of fall series,
Nov. 20, 1964.
-------
- 37 -
Table 4. INCREMENTAL HEAD LOSSES FOR RUN 5
Bed
depth ,
inches
0-6
6-18
18-21
Filter 1
^L
1.3
3.5
1.7
Ah
L/in.
0.22
0.29
0.57
Filter 2
^L
3.0
2.5
1.0
^T/'
L/in.
0.50
0.21
0.33
After 5 hours of operation, the top 6 inches of coal in filter 1 had
a head loss of 1.3 feet, or 0.22 foot per inch of bed depth. The same
portion of filter 2 had a head loss of 3.0 feet after 3 hours of operation,
or 0.50 foot per inch of bed depth. Between pressure taps 1 and 2, the
6- to 18-inch segment of the bed, the head loss was 3.5 feet for filter 1,
or 0.29 foot per inch of bed depth. The corresponding incremental head
loss for filter 2 was 0.21 foot per inch. Filter 1 gave the first filter
run in this 1-year study in which the average head loss per inch of
bed depth was greater for the 6- to 18-inch layer than it was for the
0- to 6-inch layer. The increased removal by the lower portion of the
coal layer was attributed to a more uniform size of coal in the upper
layer of coal and to a zone in which the coal and sand were mixed.
After the media in filter 1 was replaced again, this time with a
coal layer from which all particles smaller than 1.4 mm had been removed,
2
run 6 was started. Both filters were operated at 2 gpm/ft , the alum
dose was 10 mg/liter, and the raw water turbidity varied from 25 to over
50 J.u. As shown in Figure 22, the effluent turbidity was less than 0.5 J.u,
Filter 1 with the coarser coal lasted 22 hours as a result of a good
-------
^ 40
± 20
0
CO
oe
- 38 -
> 35 J.u.
RAW
= I
CD
CC
C/O
CO
CD
0
8
FILTER I EFFLUENT
FILTER I AT 6 gpm / sq ft
ALUM = 5 mg / I
MEDIA = 1.2
CD
ce
8
CO
0/3 A
<=> 4
FILTER 2 EFFLUENT
FILTER
ALUM
2 AT 6 gpm / sq
MEDIA = STANDARD
4 6
TIME, hours
8
10
Figure 20. Head loss and turbidity values for run 4 of fall series,
Nov. 21, 1964.
-------
- 39 -
40
20
CO
0=
>35 J.u.
RAW
40 J.u.
CD
03
1.0
0.5
0
8
CO
CO
2
0
FILTER I EFFLUENT
FILTER I AT 6 gpm / sq ft
ALUM = 10 mg/l
MEDIA = 1.2
= 1.0
0.5
CO
OC.
CO
CO
0
8
FILTER 2 EFFLUENT
FILTER 2 AT 6 gpm / sq
ALUM = 10 mq / I
ft
MEDIA = STANDARD
2 3
TIME, hours
Figure 21. Head loss and turbidity values for run 5 of fall series,
Nov. 22. 1964.
-------
- 40 -
distribution in floe removal, whereas filter 2 with the "standard"
media lasted only 13 hours. Both filters were stopped because of
excessive head loss, not turbidity breakthrough.
A lower raw water turbidity during run 7 resulted in a reduction
in the filter effluent turbidities, especially during the early part
of the run, as shown by Figure 23. In this run, filter 2 containing
the standard media had a higher utilization of the coal layer in
removal of the floe than previous runs in which similar media were used.
The top 6 inches of coal had a head loss of 2.3 feet when the total
head loss was 7 feet, whereas in run 6 the head loss in the top 6 inches
was 4.2 when the total was 7 feet. Filter 1 had a deep penetration of
the floe into the bed; much of the removal was in the zone of mixed coal
2
and sand. The filtration rate of 6 gpm/ft resulted in excessive penetration
of the floe into the lower 6 inches of the bed. Between pressure taps
2 and 3 the incremental headless was 1.7 feet, or 0.57 foot per inch
of bed depth as compared to 0.32 foot per inch of depth between taps
1 and 2. If the floe had been strengthened and retained higher in the
bed, the length of run could probably have been increased.
Prior to startup of run 8, the media in filter 1 were changed
again. This time only coal particles equal to or greater than 1.7 mm
were used. Figure 24 shows that this larger coal allowed a greater
passage of material by filter L than the standard media of filter 2,
although the effluent turbidity was still held at or below 0.5 J.u.
by filter 1. This increase in coal size had an additional effect on
-------
- 41 -
40
»- 20 I
ca
03
1
t t
50 52
. 1
V-"**^ .
f RAW ~~
45 J.u.
1
1
1.0
!= 0.5
CD
CO
OC
CO
CO
0
8
FILTER I EFFLUENT
FILTER I AT 2 gpm/sq ft
ALUM = 10 mg/ I
MEDIA = 1.4
1.0
± 0.5
0
8
CO
CO
1 1 1 1
FILTER 2 EFFLUENT _
1
1 1
FILTER
~AL(JM
T
2 AT 2 gpm
ft
MEDIA = STANDARD
10 15
TIME, hours
20
25
Figure 22. Head loss and turbidity values for run 6 of fall series
Nov. 23, 1964.
-------
- 42 -
the length of run over the coal size used in the previous run. After
20 hours of operation, filter 1 had a total head loss of 3.0 feet
compared to 7.6 feet for filter 2. The larger effective size and
smaller uniformity coefficient (see Table 2) of this coal layer allowed
the upper portion of the sand layer to penetrate higher into the coal
layer; this difference accounts for the much more efficient use of the
filter.
In run 9, shown in Figure 25, filter 1 with the coal greater
than 1.7 mm allowed the floe to penetrate to the sand layer when
2
operated at 6 gpm/ft . The head loss per inch of bed depth for the
bed layer between taps 2 and 3 (3 inches of bed depth) was much higher
than for any.other portion of the bed. This in turn reduced the length
of run advantage of filter 1 over filter 2, although both filters
produced an acceptable effluent from a turbidity standpoint.
2
A reduction in filtration rate to 4 gpm/ft yielded the data
shown in Figure 26 (run 10). During this run, the last of the fall
series, the highest influent turbidity of the study was recorded. The
value of 58 J.u. was measured after 6 hours of operation. Filter 2
produced an effluent with 0.1 J.u. of turbidity for the 8-hour run.
Filter 1 produced a similar effluent for 9 hours, and then a breakthrough
occurred. A small amount of filter aid would probably have eliminated
the breakthrough and lengthened the run by keeping more of the material
in the upper portion of the bed.
-------
=> 40
± 20
CD
QQ
cr: rt
- 43 -
RAW
3. 1.0
!= 0.5
DQ
QC
CO
00
<=>
0
8
FILTER I EFFLUENT
2
0
I
1
MEDIA = 1.4
CD
CC
OO
C/0
CD
1.0
0.5
0
8
FILTER 2 EFFLUENT
2
0
FILTER 2 AT 6 gpm / sq ft
ALUM = 10 mg / I
MEDIA = STANDARD
4 6
TIME, hours
8
10
Figure 23. Head loss and turbidity values for run 7 of fall series
Nov. 24, 1964.
-------
- 44 -
= 20
!= 10
CQ
CC
= i.o
± 0.5
CD
0=
00
00
O
1=1
=£
0
8
FILTER I EFFLUENT
FILTER I AT 2 gpm / sq ft
ALUM = 10 mg/1
MEDIA = 1.7
CO
oo
0
8
FILTER 2 EFFLUENT
1
FILTER 2 AT 2 gpm / sq ft
ALUM = 10 mg / I
MEDIA = STANDARD
10 15
TIME, hours
20
25
Figure 24. Head loss and turbidity values for run 8 of fall series
Nov. 25, 1964.
-------
- 45 -
Winter Series
The winter series of six runs was conducted from February 11
through 16, 1965. This period was indicative of the winter season in
that the raw water was cold (Table 1) and low in turbidity. Filter
media used in this series were similar to those used during the fall
series. Filter 1 contained the standard media, and filter 2 at
different times contained sand plus coal larger than 1.0 or 1.4 mm.
In the first run, shown in Figure 27, the alum dose was 10 mg/liter
2
and the filtration rate for both filters was 2 gpm/ft . Both filters
produced an effluent with a turbidity of about 0.1 J.u. After 19.5
hours of operation, filter 1 had a total head loss of 7.0 feet whereas
the head loss for filter 2 was only 3.8 feet. Although filter 1
appeared to have a distribution of removal throughout the coal layer,
filter 2 had a much lower head loss at the same point in time. This is
attributed to the zone of mixed coal and sand in filter 2 and the larger
coal size, both of which allowed floe removal to take place throughout
the upper 18 inches of the filter bed. The incremental head loss between
taps 1 and 2 for filter 1 probably was in the upper 3 to 6 inches of
that 12-inch segment.
A comparison of the incremental head-loss distribution in
filter 1 with the distributions obtained during the summer series in
filter 1, run 3 (Figure 12); filter 2, run 4 (Figure 13); filter 2,
run 4 (Figure 13); filter 2, run 5 (Figure 14); and filter 2, run 6
(Figure 15) is of interest. All of these filter runs used similar filter
media; and the filtration rate, alum dose, raw water turbidity, and filter
-------
- 46 -
3 40
H 20
CO
DC
J- >35 J.u. -4-
RAW
t
42 J.u.
1.0
0.5
CO
QC
CO
00
en
a:
I
FILTER I EFFLUENT
FILTER I AT6gpm/sq ft
AUJM = IOmg/l
1.0
- 0.5
CD
CC
8
6
CO
CO
FILTER 2 EFFLUENT
T
FILTER 2 AT 6gpm/sq ft
ALUM ' I0mg/l
MEDIA = STANDARD
8
TIME, hours
10
Figure 25. Head lossand turbidity values for run 9 of fall series,
Nov. 26, 1964.
-------
- 47 -
40
20
CD
±
42
1
t
37
1
_ ae i .. 1 1
f _
58 J.u. RAW
1 1
CD
OC.
00
8
FILTER I EFFLUENT
FILTER I AT 4 gpm/sq ft
ALUM = IOmg/l
MEDIA = 1.7
FILTER 2 EFTLUENT
FILTER 2 AT 4 gpm/sq ft
ALUM= 10 mg/l
MEDIA = STANDARD
2
Figure 26. Head loss and turbidity values for run 10 of fal
Nov. 26, 1964.
series,
-------
- 48 -
effluent turbidity were nearly identical. These runs also had a similar
influent as measured by pH, alkalinity, and hardness. Temperature was
the only measured characteristic that varied markedly; during the summer
runs the influent was 68°F, whereas run 1 of the winter series had an
influent temperature of 36°F. The observed difference between the four
summer runs and this run was the increased penetration of the floe into
the coal layer during the cold-water conditions.
The raw water turbidity remained at about 4 J.u. during run 2,
and the alum dose was held at 10 mg/liter. Increasing the filtration
2
rate to 6 gpm/ft gave the results shown in Figure 28. Both filters
produced an effluent with a turbidity below 1.0 J.u., but the effluent
from filter I was consistently better than that from filter 2. The
length of run for both filters was much less than one-third of the
lengths obtained in run 1, although the only known difference was the
higher filtration rate of run 2. Evidently the better distribution of
floe removal due to the higher rates was more than offset by the higher
2
initial head losses at the 6 gpm/ft rate. As with the first run of
the winter series, filter 1 appeared to have a deeper penetration of the
floe than filter 2, run 2 (Figure 11); filter 1, run 4 (Figure 13);
and filter 1, run 5 (Figure 14). These runs were similar in rate, media,
etc., with the principal difference between the two seasons being
temperature of the raw water.
2
In run 3 (Figure 29), the filtration rate was reduced to 4 gpm/ft
and the alum dose was maintained at 10 mg/liter. Even though the raw
water turbidity increased to about 10 J.u., both filters produced
an effluent generally below 0.5 J.u. in turbidity. Filter 1 effluent
was slightly better than filter 2 effluent, but filter 2 lasted 2.5 hours
-------
- 49 -
10
CO
-; i.o
| 05
CQ
CC
^ 0
8
CO
4
2
0
FILTER I EFFLUENT
FILTER I AT 2gpm/sq ft
MEDIA - STANDARD
4
3
2
± 0.5
ca
QC
CO
CO
0
8
FILTER 2 EFFLUENT
I I I
FILTER 2 AT 2 gpm/sq ft
ALUM = I0mg/l
MEDIA = 1.0
10 15
TIME, hours
20
25
Figure 27. Head lossand turbidity values for run 1 of winter series,
Feb. 11, 1965.
-------
- 50 -
longer than filter 1 to the same total head loss. There were no runs
during the summer series with a similar raw water turbidity for purposes
of comparison.
For the last three runs of the winter series, the media in
filter 2 was changed to 6 inches of sand plus 18 inches of coal greater
than 1.4 mm in size. Figure 30 shows run 4, for which the filtration
2
rate was 2 gpm/ft and the alum dose 10 mg/liter. The effluent
turbidities were similar, with filter 1 reaching a 7.0-foot head loss
in 21 hours. After the same length of operation, filter 2 had a head
loss of only 4.2 feet.
2
At 4 gpm/ft rate, filter 1 produced an effluent with a slightly
lower turbidity than that of filter 2 in run 5 (Figure 31). Filter 2
produced about 70 percent more water than filter 1 when they are
compared to the same final head loss.
The last run of this series is shown in Figure 32. Filter 1
2
at 6 gpm/ft produced a better effluent than filter 2 at the same
rate, but it reached a 7-foot head loss in one-half the time that it
took filter 2 to reach the same head loss.
Spring Series
The last set of eight runs was conducted from May 18 to 27, 1965.
In the first four runs, both filters contained the commercial sand and
coal (> 1.0 mm). The purpose of the first four runs was to check the
influence of alum dose on effluent clarity and length of run. Prior
to the fifth run, the media in filter 2 was replaced with coal larger
than 1.4 mm and commercial sand. Runs 5 through 8 compared filter media
as it influenced effluent quality and length of run. During this
2
series filtration rates of 3 and 5 gpm/ft were used instead of
-------
- 51 -
=. io
>-
± 5
ca
ce
i.o
0.5
DQ
ce
0
8
c/o
c/i
2
0
1.0
0.5
ca
or
CO
RAW
FILTER | EFFLUENT
FILTER I AT6gpm/sq ft
ALUM= IOmg/1
MEDIA* STANDARD "
I
FILTER 2 EFFLUENT
Figure 28. Head loss and turbidity values for run 2 of winter series,
Feb. 12, 1965.
-------
- 52 -
= 20
10
QQ
QC
RAW
I
3. 1-0
± 0.5
oa
cc.
t/)
C/0
1 I
L
V ,
T
i
1
FILTER 1 EFFLUENT _
- ,1
1
i . |
8
6
FILTER I AT 4gpm/«q ft
ALUM - 10 mg/l
MEDIA' STANDARD
=. i.o
± 0.5
FILTER 2 EFFLUENT
1
FILTER 2 AT 4gpm/sq ft
ALUM = 10 mg/l
MEDIA = 1.0
10
Figure 29. Head lossand turbidity values for run 3 of winter series
Feb. 13, 1965.
-------
- 53 -
2
2, 4, and 6 gpm/ft to allow more time for checking the influence of
alum dose and filter media.
The first run, shown in Figure 33, shows the influence of 5 and
10 mg of alum per liter on effluent turbidity and length of run at a
2
3 gpm/ft filtration rate. Ten mg of alum per liter resulted in an
effluent turbidity of 0.1 J.u., but 5 mg/liter gave an effluent turbidity
of about 1 J.u. Of interest is the influence these two doses had on
the rate of head-loss buildup. After 18 hours, filter 1 with 10 mg/liter
of alum had a total head loss of 6.8 feet whereas filter 2 had a
headless of only 1.8 feet. This points up again the importance of adding
just enough alum to achieve the desired effluent clarity. Generally,
additional alum reduces the length of run markedly because of the
added solids to be removed.
Figure 34 (Run 2) shows the effect of the same alum doses at a
2
flow rate of 5 gpm/ft , but with the raw water turbidity increased from
about 4 J.u. in run 1 to about 20 J.u. As a result neither filter
produced an effluent of the quality of run 1. Filter 1 effluent turbidity
averaged about 0.4 J.u., and the effluent from filter 2 averaged about
3.5 J.u. After 6 hours of operation, filter 1 had a head loss of
6.3 feet and filter 2 had 2.6 feet. The incremental head-loss curves
for both filters of run 2, and especially filter 1, reveal deep penetration
of the floe into the media. Some of this increased penetration is
attributable to the higher filtration rate. In this case the penetration
was detrimental as shown by the head-loss curves for filter 1 of run 2.
Penetration into the upper part of the sand layer (between taps 2 and 3)
resulted in a shortening of the filter run. Filter 2 of run 2 also
showed more removal between pressure taps 2 and 3 than was removed by
filter 2 of run 1.
-------
- 54 -
= 10
- 5
CD
CC
= 1.0
^ 0.5
CD
CC.
C/3
C/O
^ 1 - ! L
t ^
"~ RAW
1 1 1
I
.-
1
0
8
FILTER I EFFLUENT
FILTER I AT 2gpm/sqft
ALUM = IOmg/1
.MEDIA = STANDARD
1.0
0.5
CD
CO
cc
00
CX3
CD
8
6
FILTER 2 EFFLUENT
I I I
FILTER 2 AT 2 gpm / sq ft
ALUM = 10 mg/I
MEDIA - 1.4
10 15
TIME, hours
20
25
Figure 30. Head loss and turbidity values for run 4 of winter series,
Feb. 14, 1965
-------
- 55 -
10
± 5
DO
O=
1.0
0.5
CO
ce
8
6
C/3
CO
t
RAW
V
1 1 1 1
FILTER 1 EFFLUENT _
1 ^ 1 - 1 1
1 1 1 1
FILTER i AT4gpm/$q ft
ALUM = IOmg/1
MEDIA = 1.0
-.' 1.0
>-
I
5 0.5
CD
OS
CO
CO
0
8
FILTER 2 EFFLUENT
I I
FILTER 2 AT4gpm/sqft
4 6
TIME, hours
8
10
Figure 31. Head loss and turbidity values for run 5 of winter series,
Feb. 15, 1965.
-------
10
- 5
a
CO
ce
RAW
1.0
CD
ce
GO
CO
T FILTER
V
h-1 1
1 1
1 EFFLUENT
1 1
8
FILTER i AT 6 gpm/sq ft
ALUM = I0mg/l
MEDIA = STANDARD
1.0
0.5
CO
GO
I
I
FILTER 2 EFFLUENT
I .
4\
FILTER 2 AT 6 gpm/sq ft
ALUM = 10 mg/l
MEDIA = 1.4
Fi
10
gure 32. Head loss and turbidity values for run 6 of winter series,
Feb. 16, 1965.
-------
20
- 57 -
10
CD
CC.
RAW
1.0
± 0.5
OQ
cc.
0
8
FILTER I EFFLUENT
FILTER (AT 3flpm/sq ft
ALUM = 10 mg/l
MEDIA = 1.0
CO
CC.
FILTER 2 EFFLUENT
8
p 4
FILTER 2 AT 3gpm/sq ft
ALUM - 5mg/|
MEDIA = 1.0
10 15
TIME, hours
Figure 33. Head loss and turbidity values for run
May 18, 1965.
20
25
1 of spring seri es,
-------
- 58 -
=3
_; 40
5 20
CO
oe
^ o
C/O
c/o
8
6
FILTER I EFFLUENT
1
FILTER I AT 5 gpm/sq ft
ALUM = IOmg/1
MEDIA - 1.0
4
3
CO
ce
c/o
oo
o
8
8
I
FILTER 2 EFFLUENT
T
FILTER 2 AT 5gpm/sq ft
ALUM = 5 mg/l
MEDIA -- 1.0
Figure 34. Head loss and turbidity values for run 2 of spring series,
May 19, 1965.
-------
- 59 -
After changing the alum doses to 7.5 mg/liter in filter 1 and to
2
12.5 in filter 2, run 3 (Figure 35) vas started at a rate of 3 gpm/ft .
Filter 1 effluent averaged about 0.4 J.u. and filter 2 about 0.2 J.u.
of turbidity in the effluent. After 18 hours, filter 2 had a total head
loss of 7 feet, but filter 1 continued another 6 hours before it reached
a 7-foot head loss. As shown by the incremental head losses, the additional
alum tended to strengthen the floe and retain it higher in the filter bed.
Filter 2, upon termination, had a head loss of 5,4 feet across the coal
layer, whereas filter 1 had a head loss of 3.9 feet across the coal
layer, with more of the material penetrating into the sand layer.
2
Run 4, shown in Figure 36, was conducted at a 5-gpm/ft flow rate
with the same alum doses as those in run 3. Effluent turbidity from
filter 1 averaged about 0.5 J.u. and from filter 2 about 0.1 J.u. Filter 2
with the higher alum dose reached a 6.8-foot head loss in 6.5 hours, and
filter 1 lasted 3 hours longer to the same head loss. Comparing the
incremental head losses of run 4 with run 3 shows that the increase in
filtration rate had very little influence on floe penetration. Even though
the higher alum dose appeared to strengthen the floe and improve the
head-loss distribution, the length of run was shortened because of the
added solids to be filtered out.
After the media in filter 2 was changed to coal greater than 1.4 mm,
2
run 5 (Figure 37) was started at a filtration rate of 5 gpm/ft and an
alum dose of 12..5 mg/liter. Although filter 1 produced a slightly better
effluent, the run length was only 6 hours as compared to 9.5 hours for
filter 2 with the larger coal. Although both filters appear to have
relatively unfirom distribution of head loss, more head-loss taps between
the 6- and 18-inch taps (numbers 1 and 2)
-------
- 60 -
20
10
CO
0=
RAW
± I
CO
cc
c/o
C/O
0
8
6
4
FILTER I EFFLUENT
FILTER I AT 3 gpm/sq ft
ALUM - 7.5mg/l
MEDIA = 1.0
± I
CD
OC
c/o
c/o
CD
8
6
4
2
0
FILTER 2 EFFLUENT
1
I I
FILTER 2 AT 3gpm/sq ft
ALUM = I2.5mg/l
MEDIA = 1.0
10 15
TIME, hours
20
25
Figure 35. Head loss and turbidity values for run 3 of spring series,
May 20, 1965.
-------
=1 20
± 10
CD
OC
0
- 61 -
1 1 1
RAW
- - * '"'"~ *
1 1 1
- ,
CO
ce
00
ca
ef.
0
8
1
V FILTER
1
1 1 1
1 EFFLUENT
1 - ,
1 I -I
4
0
1 I
FILTER I AT 5 gpm / sq ft
ALUM = 7.5 mg / I
MEDIA = 1.0
FILTER 2 AT 5 gpm /sq ft
ALUM = I2.5mg / I
MEDIA = 1.0
Figure 36. Head loss and turbidity values for run 4 of spring series,
May 21, 1965.
-------
10
CO
OC
- 62 -
RAW
1.0
± 0.5
CQ
CC
00
00
0
8
FILTER I EFFLUENT
FILTER
'ALUM
MEDIA
I AT 5
= 12
. 1.0
^ 0.5
CQ
CC.
0
I I
FILTER 2 EFFLUENT
I I
FILTER 2 AT 5 gpm / sq ft
ALUM =12.5 mg / I
MEDIA = 1.4
Figure 37. Head loss and turbidity values for run 5 of spring series,
May 22, 1965.
-------
- 63 -
might have revealed a more uniform distribution in filter 2 than
filter 1. This could be explained by the larger zone for filter 2
in which the sand and coal are mixed because of the larger average
size of coal used in filter 2.
Following a reduction in both the alum dose and the filtration
rate, run 6 was conducted. As shown in Figure 38, the effluent turbidity
from both filters increased somewhat over run 5. Although the
filtration rate was reduced by two-fifths from run 6 to run 5, the
length of run for both of the filters was increased by a factor
of 4, with filter 2 again lasting much longer than filter,1. The
lower alum dose caused a deeper penetration of floe as shown by the
larger headloss increments between taps 2 and 3 for run 6 as compared
to run 5.
2
Run 7 (Figure 39) at 5 gpm/ft and 7.5 mg/liter of alum produced
2
effluents similar to run 5 at 5 gpm/ft and 12.5 mg/liter of alum.
Both filters operated longer in run 7 than in run 5, probably because
of the lower alum dose in run 7, even though the raw water turbidity
was slightly higher in run 5. This was true in spite of a better
head-loss distribution in run 5.
Increasing the alum to 12.5 mg/liter and reducing the filtration
2
rate to 3 gpm/ft improved on the effluent quality, especially in
the first part of the run as shown in Figure 40 for run 8. The
increased alum dose, although it improved both the effluent turbidity
somewhat and the head loss distribution, markedly reduced the length
of run. This is seen again by comparing run 6 and run 8. Five mg/liter
-------
^ 10
>-
!= 5
CO
cc.
- 64 -
RAW
. 2
1
V FILTER
r 1 .
- | .
1
1 1
EFFLUENT
1
* r
I- , ,
"* I-* 1
8
6
2
0
FILTER I AT 3 gpm / sq ft
ALUM = 7.5 mg/ I
MEDIA = 1.0
4
3
FILTER 2 EFFLUENT
1
16 24
TIME, hours
FILTER 2 AT 3 gpm / sq ft
ALUM = 7.5 mg/
MEDIA = 1.4
Figure 38. Head loss and turbidity values for run 6 of spring series,
May 23, 1965.
-------
_; io
0
- 65 -
1 f
RAW
. I
CD
CC.
FILTER I EFFLUENT
FILTER I AT 5 gpm / sq ft
E 0
8
6
c/o
oo
_\ FILTER 2 EFFLUENT
I
FILTER 2 AT 5 gpm / sq ft
ALUM = 7.5 mg / I
MEDIA = 1.4
6 9
TIME, hours
12
15
Figure 39. Head loss and turbidity values for run 1 of spring series,
May 25, 1965.
-------
- 66 -
of additional alum shortened the run by about 40 percent because of
the added solids that had to be removed.
Algae
During many of the runr in the summer and fall series, influent
and filter 1 effluent grab samples were taken for algae enumeration.
Influent and effluent samples from both filters-were taken once each
run in the winter and spring series. All of the above-mentioned samples
were taken at about the midpoint in the run and counted on site very
shortly thereafter. Algae were counted by the numerical or clump
count method, in which colonies and other associations of cells were
counted as units, the same as isolated cells. For each sample three
separate microscopic fields were examined. Beca\ise of the size of the
microscopic field the algae content per ml was obtained by multiplying
the average count by a factor of 31 or 27, depending on the microscope
used.
A summary of the data is shown in Table 5. The algae content
of the raw water varied from a low of 31 per ml found in run I of the
winter series to a high of 960 per ml found in run 6 of the fall series.
All of the effluent values reported for the winter and spring series
were less than 31 algae per ml. Five effluent samples during the
fall series contained some algae. The highest effluent count, resulting
in a value of 100 algae per ml was found during run 6 in the fall
series. This may have been a result of a marginal alum dose, 10 mg/liter,
for the high turbidities encountered in the raw water.
-------
= to
± 5
0
CO
cc
- 67 -
RAW
CO
CC.
C/5
(Si
e
FILTER I EFFLUENT
FILTER I AT 3 gpm/sq ft
ALUM .= 12.5 mg/l
_ MEDIA = 1.0
FILTER 2 EFFLUENT
FILTER 2 AT 3 gpm/sq ff
ALUM = 12.5 mg/l
8 12
TiME, hours
16
20
Figure 40. Headless and turbidity values for run 8 of spring senes,
May 26, 1965.
-------
- 68 -
Table 5. SUMMARY OF ALGAE DATA
Run
No.
5
6
7
8
2
3
4
5
6
7
8
9
Algae/ml
Raw
835
465
775
434
217
310
310
434
961
558
434
838
Filter-1
Effluent
Summer se
< 31
< 31
-
-
Filter-2
Effluent
ries
-
-
-
-
Fall series
< 31
< 31
55
31
102
< 31
31
53
-
*
«>
-
-
-
-
-
Run
No.
1
2
3
4
5
6
1
2
3
4
5
6
7
8
Algae/ml
Raw
31
186
186
217
217
217
135
135
108
135
189
135
81
81
Filter-1
Effluent
Winter G
< 31
< 31
< 31
< 31
< 31
< 31
Filter-2
Effluent
eries
< 31
< 31
< 31
< 31
< 31
< 31
Spring series
< 27
< 27
< 27
< 27
< 27
< 27
< 27
< 27
< 27
< 27
< 27
< 27
< 27
< 27
< 27
< 27
-------
- 69 -
The limited number of algae analyses revealed no differences in
removal due to filtration rate, alum dose, media size, or combinations
thereof. As with filtration rate, any influence of algae content on
the length of filter run was probably masked by variations in alum dose
and media size.
Production
To sort out the influence of some of the parameters on the length
of run, a production number was calculated for each of the runs. This
number was obtained by multiplying the filtration rate by the length of
run and then dividing this product by the total head loss when the run
was terminated. The values obtained are in gallons per square foot of
surface area per foot of final head loss. As the value decreases for
any set of conditions, the length of run is shorter and backwashing
becomes more frequent and costs associated with backwashing increase.
Table 6 summarizes this information for the runs in which the standard
filter was used along with 10 mg of alum per liter:
Table 6. INFLUENCE OF Fin RATION RATE ON WATER PRODUCTION
Rate2
gpm/ft
2
4
6
No. of
Runs
8
9
8
Average production
gal/ft2/ft
342
258
246
2 d
gpd/ft
2,810
5,400
7,820
a - Based on an average terminal head loss of 7 ft and 30 min.
down time during each backwash.
-------
- 70 -
As shown, the production per foot of final head loss decreases with
increasing rate. This is probably a result of the higher initial head
losses at the higher filtration rates coupled with the inability of the
flow at higher rates to pull the floe deeper into the filter with this
filter media. Although the production per foot of head loss decreased,
the production per day per square foot increased from 2,810 gallons at
the 2-gpm rate to 7,820 gallons at the 6-gt>m rate. The percent of
water used for backwash should be similar for all three rates, but
labor costs associated with backwashing would increase at the higher
rates inasmuch as backwashing is more frequent. All of the above runs
produced an effluent with an average turbidity below 0.5 J.u.
The next comparison was the production from parallel operation of
the standard filter and the other filter containing the commercial media
(Table 7).
Table 7. INFLUENCE OF FILTER MEDIA ON WATER PRODUCTION
Pairs of
Parallel
Runs
4
3
5
3
2
Average production - gal/ft /ft
Filter Media
Std. 1.0 1.2 1.4
264 408 (55)""
216 318 (47)
240 402 (68)
282
1.7
536 (85)
a - ( ) Percent increase over standard filter.
As expected, the production number increases with increasing size of coal.
Table 7 is based on the fall and winter series runs since only the standard
filter was used in the summer series, but the standard filter was not
used at all in the spring series.
-------
- 71 -
To show the influence of alum dose on the production at
2
filtration rates of 3 and 5 gpm/ft , the Table 8 was developed.
Table 8. INFLUENCE OF ALUM DOSE ON WATER PRODUCTION
Production, gal/ft
Rate,
gpm/ft
3
5
Alum Dose,
5 7.5
1570 618
750 426
mg/liter
10
480
288
2/ft
12.5
462
294
Each of the eight numbers represents one of the individual filter runs
during the first four runs of the spring series. Since they are not
averages of several runs, less weight should be attached to any specific
value; but they do show a marked trend. As with the standard filter,
higher filtration rates result in lower production. The main point to
be mentioned here is the importance of selecting the alum dose necessary
to give the desired effluent quality. Both filter runs with a 5-mg/liter
alum dose produced an effluent with an average turbidity in excess of
0.5 J.u., whereas the other three doses all resulted in an average
effluent turbidity below 0.5 J.u. Alum doses higher than required to
achieve the desired effluent quality markedly reduced the length of run.
The production numbers calculated for the last four runs of the
spring series are shown in Table 9.
-------
- 72 -
Table 9. UNIT WATER PRODUCTION FOR RUNS 5-8 OF SPRING SERIES
2
Production, gal/ft /ft
Rate,
gpm/ ft
3
3
5
5
Coal
size
1.0
1.4
1.0
1.4
Alum
7.5
630
978
396
606
dose, mg/liter
12.5
360
588
252
402
Again, each number is the result of only one filter run. The trends
mentioned previously are further confirmed with these data. Lower
filtration rate, lower alum dose, and larger coal size all tend to increase
the production.
Throughout the 1-year study the production values for runs in which
the average effluent turbidity was 0.5 J.u. or less ranged from 150 to
2
980. The value of 150 was from a standard filter operated at 6 gpm/ft,
and the upper value was obtained for filter 2, run 6 of the spring series.
For the filter runs in which the commercial media combinations were used,
nearly all the calculated production values fell in the 300 to 900 range.
A total of 17 runs was made using the filter with coal greater than 1.0 mm.
Fifteen of these runs produced an effluent with an average effluent
turbidity below 0.5 J.u., and the average production number for these
runs was 408. Just considering the runs at filtration rates from 2 through
2
4 gpm/ft (nine runs), the average was 474. Nine runs were conducted with
-------
- 73 -
the coal media larger than 1.4 mm, and the average production was 510
2
gal/ft /ft. At rates from 2 through 4 the average (5 runs) was 576.
Alum doses were higher than necessary in several of the runs with the
greater than 1.0-mm coal and 1.4-mm coal; thus, the expected production
with proper alum doaes would average higher than reported above. To
show what this means in terms of run length, the Table 10 was derived.
A final head loss of 8 feet was assumed.
Table 10. EXPECTED LENGTH OF RUN FOR VARIOUS ASSUMED PRODUCTION VALUES
Expected length of run, hours
Rate, 2
gpm/ft
2
4
6
Assumed production,
300 600 900
20 40 60
10 20 30
7 13 20
gal/ft2/ft
With an expected production value of 600, the average length of
2
run would be 40 hours at a 2-gpm/ft filtration rate and 20 hours at a
2
4-gpm/ft rate.
DISCUSSION
During this 1-year study 61 filter runs were made without the use
of flocculators and sedimentation basins. Raw water conditions encountered
during the four seasons were representative of those encountered by the
water plant during the 12-month period. Various combinations of filter
media, alum doses, and filtration rates were tested to check their
influence on the finished water quality and length of run. For this study,
finished water quality was determined by the continuous measurement of
turbidity and the periodic checking of grab samples for pH, alkalinity,
hardness, and algae populations. No attempt was made to measure coliform
organisms in the effluent because of their low numbers in the raw water
4 5
and previously published work, ' which showed a good correlation between
removal of turbidity and removal of coliform organisms, viruses, etc.
-------
- 74 -
Of the 61 filter runs, 56 produced an effluent with an average
turbidity of less than 0.5 J.u. Three of the five runs with effluent
turbidities greater than 0.5 J.u. were purposely dosed with a low or
marginal amount of alum. One of the runs had a breakthrough before a
7-foot head loss was reached, and the fifth run had no alum feed during
the first 2 hours of the run because of pump trouble.
Although no attempt was made to follow the full-scale filters
continuously, periodic comparisons of the effluents from the small
filters with the effluent turbidity from the full-scale filter disclosed
that the small filters were producing a water as good or! better than the
full-scale filter. This was expected since the experimental filters
were being fed an equal or higher alum dose than was being used by
the plant. No comparisons in length of run between the small and large
filters could be made since the large filters were backwashed routinely
on a time cycle since the head-loss gages did not work.
There is a noticeable difference in the effluent turbidity in the
initial part of the runs of the summer and fall series compared to the
winter and spring series. In the latter runs the alum feed pump and
the pump acting as the rate-of-flow controller were started nearly
simultaneously. This resulted in poor coagulation of some of the initial
raw water and higher effluent turbidities during the early part of
these runs. During the summer and fall series the alum feed was started
about a minute before filtration began, thus the Likelihood of passing
some poorly coagulated water through the filter was decreased. Generally
the filters were backwashed thoroughly so that the possibility of an
initial washout would be eliminated or the amount reduced.
-------
- 75 -
The length of run was influenced by the alum dose, the filtration
rate, and the media size. An alum dose higher than required to produce
a desired effluent quality created more solids to be removed and also
appeared to strengthen the floe, both of which caused shorter runs.
Most of the time, increasing the filtration rate accomplished very
little increase in the distribution of removal in the filter bed.
Usually any such benefit was more than offset by the increased head
loss upon startup. In other words, as the filtration rate was
increased, the quantity of water produced per unit of surface area per
foot of final head loss decreased. This does not mean, however, that
2
the 2-gpm/ft filtration rate is the best since the total annual costs
may be lower at higher filtration rates as a result of decreased
capital costs.
Although no runs were made using a small amount of filter aid,
4 5
based on previous work, ' some of the runs could have been lengthened
by the use of an aid. Upon termination of several runs, the upper
part of the sand layer (between taps 2 and 3) had over 2 feet of head loss.
A small amount of filter aid would have strengthened the floe and
allowed more removal in the upper 18 inches of the bed and, thus, produced
a longer run.
Alterations in the coal size used in the dual-media filter had
a marked influence upon the length of run. Although no comparisons
were made between a sand filter and any of the dual-media combinations,
2
previous work showed that the run length fibr the standard filter was
about three times that of a sand filter operated in parallel. This study
gave run lengths for the filter containing commercially available
media of 45 to 70 percent longer than the run lengths of the standard
-------
- 76 -
filter operated in parallel. The difference in run lengths between
the standard filter and the filter containing the commercial material
was a result of several factors. As the coal size at the top of the
filter was increased, the size of the pores also increased, which
allowed a deeper penetration of floe into the filter bed. Table 2
shows that the uniformity coefficient decreased as the coal size
increased. The lowering of the uniformity coefficient resulted in a
more uniform sized coal and a larger zone in which the sand and coal
were mixed, both of which allowed a more efficient use of the filter
bed for storage of floe.
When the coal size used in the dual-media filter was greater
than 1.7 mm, a considerable amount of the removal took place in the
lower half of the filter. If more runs had been made with this size
coal, breakthroughs would have been expected more frequently. In fact,
this is why additional runs with the greater than 1.7-mm coal were not
made during the winter and spring series.
Cold water encountered during the winter season appeared to
influence the distribution of floe removal. Similar runs in the
summer series, except for raw water temperature, removed the floe higher
in the filter media than the corresponding runs during the winter series,
No other comparisons could be made because of differences in raw water
turbidity, filtration rate, or media size.
Algae data obtained during this study, although limited in
quantity, did not reveal any significant problems associated with
their removal. Thirty to 960 algae per ml were relatively easy to
remove with the various dual-media combinations tried. Any influence
of algae content on length of run could not be demonstrated because
of the overriding influence of filtration rate and alum dose.
-------
- 77 -
Based on this 1-year study, treatment of Lake Erie water using
5 to 15 rag/liter of alum and dual-media filters will produce a high-
quality water. A filter with a coal size greater than either 1.2 mm or
1.4 mm appears to be optimum for this water. With 18 inches of coal
this size over 6 inches of sand and the proper alum dose, the expected
2
run length would be about 40 hours at a 2-gpm/ft filtration rate or
2
about half that value at a 4-gpm/ft rate. This run length could
probably be increased by the use of more than 18 inches of coal,
especially when the coal size is greater than 1.2 or 1.4 mm. This coal
size results in a fairly uniform coal layer and a good distribution of
the floe that is removed; thus, a deeper layer would give more storage
space.
-------
- 78 -
CONCLUSIONS
1. The addition of 5 to 15 mg/liter of alum to Lake Erie water
followed by filtration through a dual-media filter consisting of 18
inches of coal over 6 inches of sand resulted in high-quality effluent,
as measured by turbidity.
2. Effluent turbidity from the experimental filters was equal
to or lower than the effluent turbidity from a full-scale filter.
3. The effluent quality of the dual-media filters was maintained
2
at filtration rates ranging from 2 to 6 gpm/ft .
4. Size of coal used in the dual-media filter had a small influence
on effluent turbidity and a large influence on length of run.
5. The optimum coal size at the surface of the dual-media filter
appears to be in the range of 1.2 to 1.4 mm.
6. Alum doses in excess of that necessary for the desired
effluent clarity markedly reduced the length of run.
7. Algae encountered during the test periods were easily
removed and did not appear to have much influence on the lengtn of run.
-------
- 79 -
ACKNOWLEDGMENTS
Staff members Gene Sommerville and James F. Kreissl performed
the filter runs and recorded the data with meticulous care. Excellent
cooperation was obtained from many personnel of the Bureau of Water,
Erie, Pennsylvania.
-------
- 80 -
REFERENCES
1. Project report. High-rate filtration study at Gaffney, South
Carolina, water plant. USPHS, Robert A. Taft Sanitary Engineering
Center, May 1963.
2. Project report. High-rate filtration study at Easley, South
Carolina, water plant. USPHS, Robert A. Taft Sanitary Engineering
Center, August 1965.
3. Project report. High-rate and dual-media filtration study in a
northwestern Ohio water plant. USPHS, Robert A. Taft Sanitary
Engineering Center, January 1966.
4. Robeck, G.G., N.A. Clarke, and K.A. Dostal. Effectiveness of
water treatment processes in virus removal. JAWWA. 54:1275,
October 1962.
5. Robeck, G.G., K.A. Dostal, and R.L. Woodward. Studies of
modifications in water filtration. JAWWA, 56: 198, February 1964.
------- |