United States
Environmental Protection
Agency
Water Engineering
Research Laboratory
Cincinnati OH 45268
Research and Development
EPA/600/S2-86/107 Mar. 1987
Project Summary
Nitrate Removal from Drinking
Water in Glendale, Arizona
Dennis Clifford, Chieh-Chien Lin, Liou-Liang Horng, and Joan Boegel
A 15-month pilot-scale study of ni-
trate removal from drinking water by
ion exchange (IX), reverse osmosis
(RO), and electrodialysis (ED) was car-
ried out in Glendale, Arizona, where 10
of 31 drinking water wells had been
shut down because of excess nitrates.
The raw water contained 18 to 25 mg/L
NO3-N, 43 mg/L surf ate, and 530 mg/L
total dissolved solids (IDS). The experi-
ments were carried out using the Uni-
versity of Houston/U.S. Environmental
Protection Agency (UH/EPA) Mobile
Drinking Water Treatment Research Fa-
cility (Mobile Inorganics Pilot Plant).
All three processes could readily re-
duce the nitrate level far below the
maximum contaminant level (MCL) of
10 mg/L NO3-N. However, anion ex-
change with chloride-form, strong-base
resins was studied in the greatest detail
because of the simplicity and low cost
of this method. Based on these studies,
the rough capital-plus-operating costs
estimated for the production of 1000
gal (3.8 m3) of 7.0 mg/L NO3-N product
water in a 1.0-mgd (160-m3/hr) plant
with bypass blending are $0.30 for IX,
$0.85 for ED, and $1.00 for RO. IX, how-
ever, produces a high-chloride product
water with about 500 mg/L TDS,
whereas the product waters from RO or
ED would contain only about 180 mg/L
TDS. Also, the disposal of the IX waste-
water containing excess NaCI, NaNO3,
and Na2SO4 is potentially a bigger prob-
lem than disposal of RO or ED brine.
None of the costs presented include the
cost of disposal of the resulting waste-
waters.
For the desalting processes, the
polyamide RO membrane performed
better than cellulose triacetate on the
basis of nitrate rejection—94% com-
pared with 76%, respectively. However,
ED performed better than either RO
membrane by producing 96% nitrate re-
jection at 76% recovery.
Various flow rates, commercial
resins, and sulfate concentrations were
tried during the IX exhaustion tests. Of
these, only the sulfate concentration
was important: As it increased, the
time to nitrate breakthrough was
sharply reduced, and the size of the ni-
trate elution peak sharply increased.
Nitrate always broke through before
sulfate; but fortunately, nitrate break-
through was usually signaled by a sig-
nificant pH rise in the effluent as a re-
sult of simultaneous carbonate elution.
When a completely regenerated resin
was used, a run was complete when
the effluent pH rose to become equal to
the influent pH.
Regeneration of the nitrate-laden
resin was studied extensively using
complete regeneration, partial regener-
ation, and regenerant reuse. For com-
plete regeneration, (i.e., the removal of
more than 95% of the sorbed nitrate),
the more dilute the regenerant, the
more efficient it was. For example,
0.25 N NaCI required 3.0 equiv. Cl~/
equiv. resin, whereas 1.0 N NaCI re-
quired 180% of this value. Partial regen-
eration (i.e., the removal of 50% to 60%
of the adsorbed nitrate followed by
thorough mixing of the resin bed and
high-nitrate-leakage exhaustion) con-
sumed 37% less NaCI than the most ef-
ficient complete regeneration. Regener-
ant reuse and counterflow regeneration
were not effective in these studies, but
more research is warranted. IX regener-
ant brine disposal remains an unsolved
problem that needs further study.
This Project Summary was devel-
oped by EPA's Water Engineering Re-
search Laboratory, Cincinnati, OH, to
-------
announce key findings of the research
project that is fully documented in a
separate report of the same title (see
Project Report ordering information at
back).
Introduction
In March 1981, the University of Hous-
ton, U.S. Environmental Protection
Agency (UH/EPA) Mobile Drinking
Water Treatment Research Facility was
moved to Glendale, Arizona, a suburb of
Phoenix. A 15-month project was ini-
tiated to study nitrate removal from well
water containing 530 mg/L total dis-
solved solids (TDS). The specific objec-
tive of the Glendale, Arizona, work was
to compare the following on the basis of
technical feasibility and process eco-
nomics: reverse osmosis (RO), electro-
dialysis (ED), and strong-base, chloride-
form anion exchange for the removal of
nitrate from Glendale's Well No. 1. For
ion exchange (IX), the objectives were
(a) to evaluate the effects of resin type,
sulfate concentration, and flow rate on
the exhaustion process, (b) to evaluate
the effects of NaCI concentration, re-
generation level, flow direction, and
spent regenerant reuse on the regener-
ation process, (c) to optimize the design
of a single-bed, chloride-IX process
using complete regeneration, and (d) to
optimize the design of a single-bed,
chloride-IX process using partial regen-
eration.
Several previous studies in the United
States and Europe have demonstrated
the technical feasibility and low-cost of
IX with chloride-form resins. Thus em-
phasis has been placed on the IX proc-
ess, particularly in regard to detecting
nitrate breakthrough and improving the
efficiency of regeneration. Not included
in this study, but nonetheless impor-
tant, is establishing an environmentally
sound means of disposal for the spent
regenerant brine containing NaCI and
NaNO3.
Experimental Procedures
Water Analysis
Table 1 presents a complete analysis
of the raw water from Glendale Well
No. 1. The data represent a single sam-
ple taken during the first week of the
study from a 120,000-gal (454 m3) ele-
vated storage tank isolated from the dis-
tribution system. The N03-N concentra-
tion was later found to vary from 18.3 to
25.5 mg during the 15-month study.
Standard Methods were used in most
cases, but nitrate was determined using
Table 1.
Raw Water Analysis for Glen-
dale, Arizona, Well No. 1
Analysis
Concentration
(mg/L)
pH 8.0
Conductivity (microSiemens) 820.
Silt density index (SDI) 5.6
Total dissolved solids 532.
Silica (SiO2) 23.7
Anions:
Total alkalinity (as CaCO3) 102.
Nitrate-N 19.2
Fluoride 0.52
Chloride 122.
Sulfate 42.5
Bicarbonate 124.
Cations:
Total hardness (as CaCO3> 198.
Calcium 43.0
Magnesium 28.0
Sodium (by difference) 76.0
Iron 0.26
a low-range Hach* field-test kit based
on the cadmium reduction method. The
accuracy of the Hach method was as-
sured by frequent standardizations and
was verified in EPA quality assurance
surveys. The method of standard addi-
tions was occasionally used to ensure
the accuracy of the analyses.
Desalting Tests
The RO system was made up of two
single-pass, hollow-fiber modules—
one with cellulose triacetate mem-
branes (Dowex, RO-4K) and the other
with polyamide membranes (DuPont
B9). Only one single-pass module was
operated during each test, which lasted
at least 100 hr at 50% to 75% conversion
to determine the rejection of the various
ions in the feed. Before actual system
operation, Dow or DuPont used their re-
spective computer programs to deter-
mine the optimum pH, pretreatment re-
quirements, and estimated percentage
of rejections. These programs made use
of the complete chemical analysis of the
raw water including its temperature.
Sulfuric acid (to reduce pH to 6.7) and
sodium hexametaphosphate (SHMP, 10
mg/L) were added to the feed water as
called for by the computer projections
to prevent CaCO3 scaling. The pre-
treated feedwater was filtered through a
deep-bed filter containing AG media
(granular aluminosilicate) and then
•Mention of trade names or commercial products
does not constitute endorsement or recommenda-
tion for use.
through a 10-u, cartridge filter befort
passage through the RO module. Be
cause the water was stored in an out
side tank and passed through outside
lines in the Arizona sun, summer feed
water temperatures reached 45°C
Hence the first RO test with the DuPom
module was performed by passing the
feedwater through an ice bath to reduce
the feedwater temperature to 31 ± 4°C
The Dow RO test was deferred unti
March, when the weather was cooler.
The ED tests were performed using an
Ionics Aquamite I reversible current EC
unit with internal brine recycling tc
achieve up to 80% recovery. To comply
with the manufacturers' recommenda-
tion, the raw water (pH 8.1) was not pH-
adjusted, nor was an antiscalant added.
Pressurized raw water was passed
through the standard granular activated
carbon (GAC) filter and 10-micron car-
tridge filter before entering the ED stack,
which contained 200 pairs of cation/
anion membranes. The dechlorinating
activated carbon filter was used, but il
was unnecessary since the feedwater
contained no chlorine.
IX Tests
The IX tests were performed using
nonpretreated raw water in either 10-
in.-diameter columns containing 47 L of
resin or 8-in.-diameter columns contain-
ing 24.3 L resin. The resin bed depth
was approximately 3 ft (0.91 m). Back-
washing of the exhausted columns was
accomplished using raw water and line
pressure. Regenerations were per-
formed using technical grade NaCI typi-
cally pumped downflow through the
resin after lowering the water level in
the columns to 1-in. above the resin. A
few upflow (countercurrent) regenera-
tions were done, but they were without
much success because of the poor flow
distribution. Waste brines from the IX,
RO, and ED systems were discharged to
the city sewer system.
Complete, partial, reuse, cocurrent,
and countercurrent regenerations were
studied in detail by monitoring 30 pilot-
scale exhaustion/regeneration cycles.
The efficiency of each regeneration was
calculated in terms of the gram equiva-
lents of chloride (equiv. CD required to
remove 1 gram equiv. of nitrate (equiv.
NOs) from the raw water or spent resin.
Various salt (NaCI) concentrations be-
tween 0.25 and 3.0 N were used for the
regenerations.
Complete regenerations were
achieved by pumping a considerably
-------
excess of salt down (cocurrent) through
the exhausted resin. To calculate the re-
generation efficiency, the chloride re-
quired to elute approximately 95% of
the nitrate during a complete regenera-
tion was determined by integration of
the area under the nitrate elution curve.
Each partial downflow regeneration
was achieved by pumping a less-than-
sufficient amount of salt down through
the spent resin bed and following that
with a displacement rinse using feed-
water. Upflow partial regenerations
were also studied.
Results
Desalting Results
The ED and RO test results are pre-
sented in Table 2, which shows that ni-
trate rejection by the ED membranes is
highest at 96%. This fact and the mini-
mum ED pretreatment requirements
make ED the process of choice for
achieving a combination of nitrate re-
moval and desalting in the test water.
Several additional factors should be
kept in mind, however. The RO tests
were made with single modules, but in
actual practice, reject staging would be
used to achieve greater water recovery
with similar nitrate rejection. Also, it is
doubtful that both acid and SHMP addi-
tion would be required for RO pretreat-
ment. In fact, during the DuPont RO test,
the acid addition pump failed and the
feedwater pH was unadjusted for 4 days
without apparent membrane fouling or
change in nitrate rejection. A final plus
for RO is that rapid advances are being
made in RO membrane technology. A
small community's choice of IX, RO, or
ED will also be influenced by the sim-
plicity of the equipment and its ability to
withstand frequent start/stop cycles.
Such design variables could not be eval-
uated in these 3- to 4-day continuous
tests.
Overview of IX Exhaustion
Results
Nineteen of the thirty IX exhaustion
runs were monitored for nitrate break-
through (see Table 3). Typical break-
through curves for nitrate and other
anions are presented in Figure 1, which
shows that up to 400 bed volumes (BV)
could be treated before nitrate broke
through and reached the 10-mg/L MCL
Following breakthrough, a nitrate peak
resulted in which the effluent concen-
tration exceeded that of the influent by
as much as 50%, depending on the sul-
Table 2. Desalting Results
Description
Nitrate rejection, %
Sulfate rejection, %
Chloride rejection, %
Bicarbonate rejection, %
TDS rejection, %
Feedwater pH
Product pH
SHMP antiscalant, mg/L
Product water flow, L/min
Operating pressure, psig
Water recovery, %
Maximum allowable temp., °C
Operating temperature, °C
ED
96.
97.
88.
93.
92.
8.1
5.9
0.0
1.2
45.
76.
45.
28-32
Hollow Fiber
Polyamide
RO
94.
98.
95.
95.
95.
6.7-8.1
5.6
10.0
4.6
350.
50.
35.
27-35
Hollow Fiber
Cellulose Triacetate
RO
76.
98.
—
95.
83.
6.7
6.0
10.0
11.1
280.
74.
30.
18-26
Table 3.
Run No.
1
2
3
4
5
6
7-16
17
18
19
20
21
22
23
24
25
26
27
28
30
Summary of IX Exhaustion Runs
BV to NO3-N
MCL Comments
422
410
410
345
245
177
Exhaustion only
346
323
320
338
322
30
0
420
410
330
340
370
115
Type 1 (ASB-1) resin, C0* = 18.6 mg/L
Type 2 (ASB-2) resin, C0 = 21.5 mg/L
ASB-2, low flow rate, C0 = 22 mg/L
Isoporous Type 1 (Dowex 1 1), C0 = 25.5 mg/L
ASB-2, Feed SOj = 140 mg/L
ASB-2 Feed SO; = 310 mg/L
for complete regeneration studies
Partial (1/1) regeneration*, leakage1 =11 mg/L
Partial (1/1) regeneration, leakage = 15 mg/L
Partial (0.64/1) regeneration, leakage > 25 mg/L
Partial (1/1) regeneration, leakage = 15 mg/L
Partial (1/1) regeneration, leakage = 16 mg/L
Upflow partial regeneration failed
Upflow partial regeneration failed
Complete regeneration, 0.25 N
Complete regeneration, 0.25 N
Partial (1/1) regeneration, leakage =17 mg/L
Partial (1/1) regeneration, leakage = 16 mg/L
Partial (2/1) regeneration, leakage = 12 mg/L
Partial (1/1) regeneration, feed SO; = 404 mg/L
*C0 = concentration of NOyN in feedwater; mg/L N = mg/L NO3-N.
f = Partial (1/1) regeneration = partial regeneration with 1 equiv. Cl~/equiv. resin.
* = Leakage values are NOj-N concentrations at approximately 10 bed volumes after start of
rt m
run.
-------
fate concentration. Also evident in Fig-
ure 1 is the fact that when the effluent
pH rose to become equal to the influent
pH, nitrate was beginning to break-
through. Subsequent analysis showed
that this pH increase was due to the
simultaneous elution of carbonate with
nitrate.
During the course of the 28 exhaus-
tion/regeneration tests, the nitrate ca-
pacity of the type 2 resin used did not
decrease measurably. This long-term
stability may be expected for any typical
styrene-diviny(benzene, strong-base
anion resin used on the Glendale water
or any other non-fouling groundwater.
Effect of Resin-Type
Based on prior laboratory studies
with simulated nitrate-contaminated
groundwaters, little variation in nitrate
removal performance was expected as
a result of the type of strong-based resin
used. This prediction was verified by
comparing the performances of three
different styrene divinyl-benzene
resins: lonac ASB-1 (a type 1 gel), lonac
ASB-2 (a type 2 gel), and Dowex 11 (a
type 1 isoporous resin). Resins with
higher capacity produced longer runs to
nitrate breakthrough, but after capacity
was taken into account, no significant
differences existed in exhaustion per-
formances as a result of the kind of resin
used. The type 2 gel resin was used for
the subsequent 25 exhaustion/regener-
ation runs because of its high capacity
and slightly greater preference for ni-
trate compared with chloride.
Effect of Sulfate Concentration
Increasing the sulfate concentration
of the Glendale groundwater from 42.5
to 310 mg/L by spiking shortened the
run length to nitrate breakthrough from
400 down to 180 BV. The earlier nitrate
breakthrough was due to two factors—
the increasing concentration of ions in
the spiked feedwater and the resin's
tendency to prefer sulfate to nitrate in
dilute solution. However, in one run the
sulfate/nitrate preference was inverted
to favor nitrate when the ionic strength
was increased to the point (0.030 N)
where sulfate was less preferred than
nitrate and broke through earlier. This
run demonstrated the well-known con-
cept of selectivity reversal, in which di-
valent ions become less preferred than
monovalent ions as ionic strength in-
creases.
IX Regeneration Results
Sulfate Elution
Regardless of the regeneration
method used, sulfate was much easier
to strip from the resin than nitrate. This
effect was evident in all the regenera-
tion elution curves, which are typified
by Figure 2. The most preferred ion dur-
ing exhaustion (divalent sulfate) is the
first to be stripped from the resin during
regeneration because of the selectivity
reversal that occurs in high ionic
strength salt solution.
Complete Regenerations
Although the elution of sulfate is con-
siderably easier than that of nitrate, the
elution of both anions becomes much
less efficient as the regenerant NaCI
concentration is increased (Figure 3).
Two reasons are proposed for the ineffi-
200
780
760
^ 14°
§ 720
O
700
80
60
40
20
360
320
280
- ^ 240
o
o
760
720
80
40
36
32
28
o. 24
-
-------
BedVolumes
345
1200 -
- 12000
40
60 80 100
Time, min
120 140 160
Figure 2.
Typical elation of nitrate and sulfate from ASB-2 resin during a complete
regeneration with 0.5 N NaCI in a cocurrent fashion.
ciency of concentrated regenerant solu-
tions: (1) inefficient mass transfer at
high concentration, and (2) the ten-
dency for the resin to release divalent
ions like sulfate at increasing ionic
strength. Although dilute regenerants
more efficiently use chloride to elute ni-
trate, they yield greater wastewater vol-
ume and increase the required regener-
ation time.
Reuse of Regenerant
Attempts were made to conserve salt
ky reusing the spent regenerant during
a roughing regeneration before a sec-
ond polishing regeneration with fresh
NaCI. Dilute (0.5 N) spent regenerants
reused in this manner did elute 20% to
50% of the nitrate on the resin, but the
subsequent polishing regeneration with
fresh NaCI was very inefficient. In fact, it
took as long for the polishing regener-
ant to elute the remaining nitrate from
the column as a single-step regenera-
tion would have taken. Thus no net salt
savings were realized when reusing
spent regenerant to achieve complete
regeneration. Partial regeneration was
not attempted using spent brine, but it
probably would have been more suc-
cessful and it deserves further study.
Partial Regeneration
Prior laboratory studies indicated that
the amount of salt required for regener-
ation is substantially reduced when the
resin is only partially (e.g., 50%) regen-
erated. This regeneration procedure is
efficient because it avoids the inefficient
tail of the nitrate elution curve (see Fig-
ure 2). The disadvantage of partial re-
generation is that it yields significant
and potentially excessive nitrate leak-
age on subsequent exhaustion of the
nitrate-contaminated resin. The partial
regeneration experiments in Glendale
produced four significant conclusions:
(1) Complete mixing of the resin bed is
required to avoid excessive nitrate leak-
ing following cocurrent regeneration,
(2) counterflow partial regeneration is
not nearly as efficient as cocurrent,
(3) the amount of salt required is signif-
icantly greater than that calculated
using another published methodology,
and (4) unlike complete regeneration,
there is no significant difference in the
efficiency of partial regeneration with
salt concentrations in the range of 0.25
to 1.0 N (1.5% to 6%) NaCI.
Partial Regeneration Level
Partial regeneration levels from 0.64
to 2.0 equiv. Cl~/equiv. resin were
tested in Glendale. Figure 4 compares
the Glendale nitrate breakthrough
curves following complete and partial
regenerations with 1.0 N salt. Clearly, a
regeneration level of 0.64 equiv. of salt/
equiv. resin is insufficient to produce a
reasonable run length, but a regenera-
tion level of 1.0 or higher may be ac-
ceptable if the excessive nitrate leakage
is disregarded during the initial 75 BV of
effluent.
Extensive backwashing of the par-
tially regenerated bed was used in an
attempt to eliminate the initial high-
nitrate leakage. This step was not suc-
cessful, however (Figure 4). The suc-
cessful elimination of excessive leakage
is apparent in Figure 5, where the per-
formance of a partially regenerated but
unmixed bed is compared with that of a
well-mixed homogeneous bed. The ex-
periments for eliminations of leakage
(Figure 6) were performed in small labo-
ratory columns using simulated Glen-
dale water. Following exhaustion and
partial regeneration with 1.0 equiv. Cl /
equiv. resin, the unmixed bed was sim-
ply exhausted again. By contrast, the
-------
I
0.00
O.SO 1.60
Normality of NaCI, N
2.40
3.20
Figure 3.
Inefficiency of nitrate and sulfate elution from exhausted ASB-2 resin as a function
of regenerant concentration.
resin in the other column was removed,
manually mixed, and placed back int«
the column before the exhaustion run
shown. Thorough mixing clearly elimi-
nates the excess nitrate leakage at the
beginning of an exhaustion following a
cocurrent (downflow) partial regenera-
tion. No amount of conventional back-
washing will achieve this mixing be-
cause the resin bed remains classified
with the largest resin beads on top and
the smallest on the bottom.
Comparison of Complete and
Partial Regeneration
After it was confirmed that the early
excess nitrate leakage could be elimi-
nated by efficiently mixing the resin fol-
lowing a partial regeneration, the
method became quite attractive. The
experimental data for 1.0 equiv. Cl~/
equiv. resin showed the partial regener-
ation to be more efficient than complete
regeneration, even with the dilute 0.25
N regenerant. For this reason, a detailed
comparison was made of the methods.
Although 1.0 equiv. Cl~/equiv. resin was
the partial regeneration level used in
most of the pilot runs, a careful exami-
nation of the nitrate effluent histories
led to the conclusion that slightly more
chloride would be needed to keep ni-
trate under the maximum desirable
NO3-N Cone, in Haw Water
Breakthroughs
for
Partial
Regenerations
Breakthroughs
for
Complete
Regeneration
700
750
200 250
Bed Volumes
300
350
400
450
Figure 4. Nitrate breakthrough curves for completely and partially regenerated beds of ASB-2 resin.
6
-------
Time hr
25
20
X
o
I
' 2 in
o
5
n
5 10 15 20 25
/- NOl-N Cone, in Raw Water
\ eq. cr
A 7 n pnn fjn ->c
\» e1- R (Heterogeneous Bed)
\\ ^ MCL
__. — Sj^*A Att-1 •• - •• - f~ »• • - - • C
O *"O— O O O — O-^-v ^pr^r^^T^r^^^
/ ^ ^^*O~~O~^~O — U — O v »
r /\^— o— /^
/ e<7 Cr
eQ- R (Homogeneous Bed)
• iiiii
30 35
1
n
// [ j
A/ D— « 10.0
o
5.0
0
1 1 1 I 1 1 1
0 Cr acfoa/ Cr predicted
•& HCOl actual HCOl predicted
-5-o-O-cK
\o o
\ ^ O
^^^° 0 o
-------
Table 4. Approximate Costs for Nitrate Removal by IX, ED, and RO'
Process
5/7000 gal
$/m3
Product Water
TDS (mg/L)
Single-bed chlorine-
form IX
ED
RO
(aramid membrane)
0.3
0.85
1.00
0.08
0.22
0.26
500
185
185
"Assumptions: Feedwater = 500 mg/L TDS, 21 mg/L NOy-N, 50 mg/L SOf, Bypass flow = 30%
of blended product flow. Desalting process = 90% overall TDS rejection. NO3-
N = 7 mg/L in blended product.
through than was the flow rate or kind
of commercial resin used. When sulfate
concentration was increased from 43 to
310 mg/L, the BV to nitrate break-
through decreased from 410 to 180.
Cocurrent complete regeneration
(i.e., the removal of more than 95% of
the sorbed nitrate) required 3 to 9 times
the stoichiometric amount of chloride,
with dilute regenerants being far more
efficient than concentrated ones.
Cocurrent partial regeneration (i.e.,
the removal of 50% to 60% of the sorbed
nitrate) required 1.0 to 1.2 times the
stoichiometric amount of chloride and
was not influenced by the regenerant
concentration. Complete mixing of the
resin bed is mandatory following partial
regeneration to reduce the NCVN leak-
age to about 7 mg/L.
A comparison of the optimum com-
plete and partial regeneration indicates
that partial regeneration consumes 37%
less salt and is therefore preferred if sig-
nificant nitrate leakage can be tolerated.
Recommendations
Laboratory- and pilot-scale studies
should be carried out using the recently
developed, nitrate-selective, styrene-
DVB strong-base anion resins. The lat-
ter are based on greater charge separa-
tion distance to reduce the sulfate
preference and greater hydrophobicity
to increase the nitrate affinity.
Further research and development
should be carried out on regeneration
._«_ x rn'.u ^ .•••••.-,-
efficiency>sperft rtfgehe'faht disposal?
and nitrate breakthrough detection—'
the key issues in nitrate removal by
chloride IX, an effective and relatively
inexpensive process.
Partial regeneration should be fur-
ther studied on a laboratory scale using
background waters containing various
levels of sulfate, bicarbonate, and chlo-
ride. The effects of suJfate concentration
on NaCI efficiency and pH detection of
nitrate breakthrough should be exam-
ined in these tests using conventional
and nitrate-selective resins.
The full report was submitted in fulfill-
ment of Cooperative Agreement No.
CR-807939 by the University of
Houston-University Park under the
sponsorship of the U.S. Environmental
Protection Agency.
Dennis Clifford, Chieh-Chien Lin, Liou-Liang Horng, and Joan Boegel are with
the University of Houston-University Park, Houston, TX 77004.
Thomas Sorg is the EPA Project Officer (see below).
The complete report, entitled "Nitrate Removal from Drinking Water in Glendale,
Arizona," {Order No. PB 87-129 284/AS; Cost: $18.95, subject to change)
will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Water Engineering Research Laboratory
U.S. Environmental Protection Agency
Cincinnati, OH 45268
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
BULK RATE
POSTAGE & FEES PAI
EPA
PERMIT No G-35
Official Business
Penalty for Private Use $300
EPA/600/S2-86/107
0000329
60604
------- |