Effect of Nitrate Addition on Biorestoration
of Fuel-Contaminated Aquifer:
Field Demonstration
by S. R. Hutchins’, W. C. Downs’, J. T. Wilson’, G. B. Smith’, D. A. Kovacs’,
D. D. Fine’, R. H. Douglass’, and D. J. Hendrix®
Abstract
A spill of JP-4 jet fuel at the U.S. Coast Guard Air Station in Traverse City, Michigan, contaminated a water-table
aquifer. An infiltration gallery (30 ft X 30) wasinstalled above a section of the aquifer containing 700 gal JP-4. Purge wells
recirculated three million gallons of ground water per week through the infiltration gallery at a rate designed to raise the
‘water table above the contaminated interval. Ground water containing ambient concentrations of oxygen and nitrate was
first recirculated for 40 days. Concentrations of benzene in monitoring wells beneath the infiltration gallery were reduced
from 760 to< 1 ug/l. Concentrations of toluene, ethylbenzene, m,p-xylene, and o-xylene were reduced from 4500 to 17, 840
to 44, 2600 to 490, and 1400 to 260 ug/l, respectively. Average core concentrations of benzene, toluene, ethylbenzene,
‘msp-xylene, and o-xylene were reduced from 0.84 00.032, 3310013, 1810 0.36, S810 7.4, and 26 to 3.2 mg/kg, respectively.
Ground water amended with nitrate (10 mg/l nitrate-nitrogen) and nutrients was then recirculated for 76 days. Final core
concentrations of benzene, toluene, ethylbenzene, m,p-xylene, and o-xylene were 0.017, 0.036, 0.019, 0.059, and 0.27 mg/kg,
respectively. Final aqueous concentrations were < 1 ug/lfor benzene and toluene, 6 ug/ for ethylbenzene, and 20 to 40 g/1
for the xylene isomers, in good agreement with predicted values based on residual fuel content and partitioning theory.
Although alkylbenzene concentrations have been substantially reduced, the test plot istill contaminated with the weathered
fuel. Based on stoichiometry, approximately 10 times more nitrate was consumed than could be accounted for by BTX
icating that other compounds were also degraded under denitrifying conditions.
Introduction
Leaking underground storage tanks area major source
of ground-water contamination by petroleum hydrocar-
bons. There are approximataely two million underground
tanks storing gasoline in the U.S., and there have been
90,000 confirmed releases reported in the last two years
(OUST, 1990). Gasoline and other fuels contain benzene,
toluene, ethylbenzene, and xylenes (BTX) which are
hazardous compounds regulated by the U.S, Environmen-
tal Protection Agency (EPA, 1977). Although these aro-
matic hydrocarbons are relatively water-soluble, they are
contained in the immiscible bulk fuel phase which serves as a
slow-release mechanism for sustained ground-water con-
tamination. Pump-and-treat technology alone is economi-
cally impractical for renovating aquifers contaminated with
bulk fuel, because the dynamics of immiscible fluid flow
result in prohibitively long time periods for removal of the
*U.S. EPA, Rober S. Kert Environmental Research Lab,
‘Ada, Oklahoma 74820.
NSH Technology Services, Robert S. Keer Environmental
Research Lab, Ada, Oklahoma 74820.
"The Traverse Group, Ine, 2525 Acro Park Drive, Traverse
City, Michigan 49684.
“Solar Universal Technologies, Inc, Ground Water Remedi
ation Division, Traverse City, Michigan 49684
Received May 1990, revised September, November 1990, and
February 191, accepted February 1991
Discussion open until January 1, 1992
Vol. 29, No, 4GROUND WATER—July-August 1991
organic phase (Wilson and Conrad, 1984; Bouchard etal,
1989), Biorestoration has been recommended as treatment
alternative and involves enhancing the activity of the native
subsurface bacteria to degrade fuel hydrocarbons through
addition of nutrients and other compounds. Aerobie bio-
restoration has been shown to be effective for many fuel
spills (Thomas et al,, 1987; Lee et al., 1988). However,
success is often limited by the inability to provide sufficient
oxygen to the contaminated zones due to the low water
solubility of oxygen (Wilson etal, 1986).
Nitrate can also serve as an electron acceptor and
results in anaerobic biodegradation of organic compounds
via the processes of nitrate reduction and denitrification
(Tiedje, 1988). Because nitrate is less expensive and more
soluble than oxygen, it may be more economical to restore
fuel-contaminated aquifers using nitrate rather than oxy-
gen, Several investigators have observed biodegradation of
aromatic fuel hydrocarbons under denitrfying conditions
(Kun eta, 1988; Major et al. 1988; Miheleic and Luthy,
1988; Hutchins etal, 1991). However, the process is not well
understood at field scale where several other processes,
including aerobic biodegradation, can proceed concomi
tantly. There have been only a few field studies on nitrate-
mediated biorestoration of fuel-contaminated aquifers.
Results include complete removal of benzene and toluene
with the xylenes being more recalcitrant (Battermann,
1986), a 95 to 98% reduction in purgeable alkylbenzenes
(Sheehan et al, 1988), and complete removal of toluene
with benzene, ethylbenzene, and the xylenes being unaf-
371fected (Lemon et al, 1989). Although most of these field
studies have demonstrated partial success, the complexity of
the field sites and the limited monitoring data preclude a
thorough evaluation of the proces.
The use of nitrate to promote biological removal of fuel
aromatic hydrocarbons was investigated for a JP- jet fuel
spillat Traverse City, Michigan, through afield demonstra-
tion project in cooperation with the U.S. Coast Guard
Laboratory ests had indicated that denitrification would be
a suitable alternative for biorestoration of the aquifer
(Hutchins etal, 1991). The Coast Guard plans to evaluate
nitrate-based biorestoration, aerobic biorestoration using
hydrogen peroxide, and soil vapor extraction to determine
which technology is most cost-effective for bringing the site
tothe closure limit ofS g/l benzene in the ground water as
specified by the Michigan Department of Natural Resources
The objective ofthis study was to evaluate whether nitrate-
based biorestoration could successfully attain this closure
criterion and to quantitate the extent of remediation.
Site Description
In February 1985, four large underground storage
tanks at a US. Coast Guard facility in Traverse City,
Michigan were found to be leaking JP-4 jet fuel through
faulty piping to ashallow underlying aquifer. By the time the
leaks were discovered and the tanks and contaminated soil
were excavated, several thousand gallons of JP-4 et fuel had
been lost to the subsurface. A series of interdiction wells
were installed to contain the contaminated ground water
and equipped with product recovery pumps to recover free
product overlying the water table (Figure 1). This sucoess-
fully retained the plume within the facilty’s boundaries,
however, large area predominantly underlying the con-
crete apron and runway had been contaminated. Moreover,
the seasonal rise and decline of the water table had effec-
tively smeared the contamination over a 5 to 8 ft depth
interval, resulting in residual saturation of JP-4 (up to
32,000 mg/kg) in the unsaturated zone as well as free
product at the water table
The aquifer is composed of thick glacial deposits with
the upper portions being lacustrine in origin (Twentereta.,
1985). These lacustrine glacial deposits consist of an upper
sand and gravel unit and an underlying clay unit. The
thickness of the sand ranges from 50 to 60 ft in the
‘immediate veinity ofthe spill and consist of Fine to medium
grained sand in the upper 15 to 20 ft and coarse to very
coarse sand with gravel in the lower section. The water table
varies seasonally and ranges from 12to 8 f below the land
surface. The average linear ground-water velocity is esti-
mated to be 5 ft per day.
Field Demonstration Project
The focus of the field demonstration project is a 30 ft
30 ft infiltration study area located within the larger area
contaminated by the JP-4 plume (Figure 1), An infiltration
gallery was constructed to perfuse the study area with
‘ground water supplemented with nitrate and nutrients.
Computer modeling of the hydraulics was conducted 10
design the infiltration system (Downs etal, 1989). Results of,
the model indicated an infiltration rate of 200 GPM over the
sm
study area was required to create a water-table mound
encompassing the contaminated zone. Thisenabled the con-
taminated unsaturated zone within the study area to be
completely saturated and thus allowed even distribution of
nitrate and nutrients. The preexisting line of interdiction
wells PP-5 through PP-8 were pumped at 15 GPM each
With the effluents routed to a carbon treatment system.
‘These wells are screened throughout the aquifer and include
the contaminated zone. In addition, five pumping wells,
PP-SA through PP-9A, were installed to recirculate the
recharge water back to the infiltration gallery. These wells
are screened below the contaminated zone to avoid influx of
free product. The pumping wells recirculate the water at 200
GPM, and the interdiction wells provide a net discharge of
60 GPM to retain nitrate and contaminants within the
facility boundaries.
‘The infiltration gallery consists of a primary 4-in, PVC
pipe equipped with twelve 2-in. PVC pipe arms, each arm
being 150 in in length and containing 3/16-in. holes spaced
‘one in. apart to ensure even hydraulic distribution. The
gallery is contained within a 3-f¢ gravel pack buried $ ft
below land surface to prevent freezing during winter
months. The bottom ofthe gravel pack is sloped downwards
towards the center to concentrate the mound at the center of|
the study area (Figure 2). The study area contains five
cluster wells to monitor ground-water quality and 1
piezometers to monitor water levels and head loss across the
contatninated zone, Each cluster wel is constructed of Yin.
stainless steel tubing and contains six diserete sampling
points with 40 X 40 mesh stainless stel screens. The sample
points for each cluster well are spaced 2 ft apart with
interspersed bentonite seals to prevent channeling of ground-
‘water recharge. The sample points for each cluster well are
located both within and beneath the contaminated zone to
allow monitoring of biodegradation and contaminant
transport within the different regions. The lines for the
cluster wells are routed below land surface to & sample
building. The sample building also contains the equipment
required for recirculating the ground water obtained by the
pumping wells and the chemical addition facilites.
a
oom
Ls
Fig. 1. Plan view of field demonstration project. Ground.
water Is recirculated to infiltration gallery at 200 GPM.
Interdiction wells dispose of contaminated ground water at
60 GPM. Locations of original uel tanks are shown.Fig. 2. Cross section of infltration gallery showing location
‘of monitoring wells and plezomelers.
Infiltration, using recirculated ground water without
‘amendments, commenced April 20, 1989 and was continued
until the design recireulation rate was attained. This was
maintained for two weeks to achieve hydraulic equilibrium.
A tracer study indicated that vertical flow was uniform
beneath the infiltration gallery with a hydraulic residence
time of 8 hrs within the contaminated zone. The total
hydraulic residence time was 7 to 12 days, depending upon
the distance to each well and its respective pumping rate
Nitrate and nutrients were then applied at design concentra-
tions of 62 mg/| sodium nitrate, 10 mg/ monobasic pot
sium phosphate, 10 mg/l disodium phosphate, and 20 mg;
‘ammonium chloride, and full operation of the system com-
‘menced May 31, 1989 and continued for two months. Per
formance of the pilot project was assessed by monitoring
both aqueous BTX and inorganic chemical concentrations
throughout the system as well as by periodically coring
through the contaminated zone in the study area. The final
sample cores were taken August 9, 1989, shortly after nitrate
addition ceased.
Water Quality Analysis
Water quality data were collected weekly during the
project. These analyses included dissolved oxygen, nitrate,
nitrite, and BTX from selected cluster monitoring well
points, As shown in Figure 2, there are six well points for
each cluster monitoring well, with the shallowest well point
being designated Level | and the deepest being Level 6. The
following discussion will focus on water quality at Level 2
(within the contaminated zone, above the original water
table), Level 4 (jst beneath the contaminated zone), and
Level 6 (further beneath the contaminated zone); the other
well points were not sampled on a routine basis. Data were
averaged for each of the five cluster monitoring well points
across a given level. This was done to allow a concise
description of changes in water quality during infiltration
across the contaminated zone. A more complete dataset has
been published (Hutchins et al., 1990)
‘The dissolved oxygen profiles for the injection water
and the water at different levels are shown in Figure 3
Infiltration began on Day 3 and nitrate and nutrient addi
tion began on Day 44, Samples were collected from the well
points at Day 0 to provide background information.
Because the aquifer naturally contains low organic carbon
and is highly transmissive, oxygen levels were high except
‘where contamination occurred, and in fact were even sub-
stantial at the bottom ofthe contaminated zone with values
approximating 3 mg/l, However, oxygen concentrations
dropped rapidly as water was recirculated through the con-
taminated zone (Figure 3). This is probably due to aerobic
biodegradation occurring within the contaminated zone
which had originally been oxygen-limited. Oxygen concen-
trations stabilized at 0.5 to 1.0 mg within the monitoring,
wells by Day 7. By Day 20, the infiltration water consisted
‘mostly of water which had passed through the contaminated
zone, causing influent oxygen levels to drop and leading to
less oxygen consumption as recharge water moved across
the contaminated zone. It appears that limited oxygen con-
sumption occurred from Day 20 to Day 64 with litte oxygen
removal afterwards, but this may not be significant given the
resolution of the data,
Based on the measured infiltration rates, the mass of
oxygen added can be calculated using the following equation:
M=/CFat
‘where M is the total mass in mg, C is the concentration in
‘mg/l, F is the infiltration rate in L/d, and tis the time in
days. The integral is evaluated numerically using a trapezoi-
dal algorithm. Mass estimates can be generated for each
level throughout the vertical profile, and the differences
between oxygen mass supplied in the infiltration water and
that found at Level 6 corresponds to oxygen uptake within
the contaminated zone, assuming that oxygen removal is
due primarily to aerobie biodegradation within the contam-
inated zone. This also assumes vertical flow through the
contaminated zone. In fat, there isthe potential for lateral
flow from the edges of the study area; however, because
there are no monitoring well points along the lateral flow
path, the effect of lateral flow cannot be quantified. This
equation was also used to calculate mass fluxes for the other
parameters (Table 1).
=== ection Water
Level?
Une
sgt” uso]
0 2 40 60 ab 00 BD
Tome 2098)
Fig. 3. Dissolved oxygen profile beneath infiltration gallery.
LSD is the least significant difference between means al the
90% confidence level.
1Table 1. Calculated Mass of Nutrients and Contaminants in infiltration Water Before and Ater Nitrate Addition,
and Changes in Mass After Passage Through Contaminated Zone
‘Before nitrate addition
After nitrate addition
(day = 310 44) (ay = 4410 112)
‘Total mass supplied Mass change from feed Total mass supplied Mass change from feed
Parameter in recharge 10 Level In recharge 10 Level 6
Dissolved oxygen 2 35 ra 4
Nitrate-iteogen 3 25 0 499
Nitvte-nitrogen <2 = 66 25
‘Ammonis-nitrogen 2 301 “18
Phosphate-phosphorus <2 7 9 2
Benzene 34 29 0 0
Toluene 16 “5.1 59 58
Ethylbenzene 88 456 43 ay
mp-Xylene a +34 B 39
o-Xylene 20 +64 " +12
(Data are in ke)
Priorto nitrate addition, 92 kg oxygen were supplied in
the recharge water and 55 ke oxygen were removed by the
time the infiltrating water reached Level 6 (Table 1). After
nitrate addition, oxygen removal was minimal. Oxygen
levels were low during this time, and most of the oxygen
removal occurred prior to the onset of rapid denitrification,
‘Thus, there was significant oxygen available for aerobic
biorestoration atleast 20 days prior to nitrate addition, but
not 20 days afterwards.
‘When nitrate and nutrients were added to the infiltra-
tion water, there was a lag period of about 20 days before
significant nitrate removal occurred (Figure 4). This is con-
sistent with the observation that denitrifying bacteria gener-
ally require a finite time for enzyme induction. However,
nitrate removal was rapid and consistent after this time, with
‘removal rate of about 1-mg nitrate-nitrogen/Ifhr. This
served to keep nitrate-nitrogen concentrations well below 10,
‘mg/l once the water reached the interdiction field, in com-
pliance with the permit granted by the Michigan Depart-
‘ment of Natural Resources. Low concentrations of nitrate in
the Level 6 wells indicate that the microorganisms in the
contaminated zone were not nitrate-limited, and that the
system could be managed to minimize nitrate contamina-
tion of ground water. There was essentially no nitrate
removal between Levels 4 and 6, that i, after the infiltrating
water had already passed through the contaminated zone.
This implies that most of the nitrate was consumed during
oxidation of organic compounds associated with the con-
taminated interval
‘There were significant nitrate concentrations in the
feed prior to nitrate addition; a net total of 25 kg nitrate-
nitrogen were consumed as compared to 5S ke oxygen
(Table 1). Thus, there may have been some denitrtying
activity prior to exogenous nitrate addition, but it is not
possible to distinguish this from nitrate uptake for cellular
synthesis. After nitrate addition, however, approximately
500 kg nitrate-nitrogen were consumed. Given the limited
oxygen availability at this time, itis probable that most of
this uptake was in conjunction with denitrification, Sup-
porting evidence is provided by analysis of selected core
samples before and after nitrate addition; although total
m4
‘microbial numbers did not increase appreciably, there was a
icant increase in the denitrfier population, especially
in those intervals which had been contaminated with fuel
(Figure 5). Nitrite production was transitory and concentra-
tions stabilized to0.1 to 0.5 mg; Initrite-nitrogen by Day 70
(Figure 6). In addition, approximately 20 kg of ammonia-
nitrogen and phosphate-phosphorus were also consumed
throughout the study, presumably for cellular synthesis
(Table 1).
TIT Ineeton Water
O— Uavele
ti
w=]
=
EP] Boainnivate
00
mo 20
co 80
Tine (ays)
Fig. 4. Nitrate-nitrogen profile beneath infiltration gallery.
LLSDis the least significant difference between means al the
‘90% confidence level.
Background
Ty Dewi
By Tosa Couns
ley (Pe MSL
aaoeae
ing Cal Nob
Fig. 8. Distribution of total bacterial populations and denitri-
fier populations at selected depths.TET nection water
= Une?
os
y as
rato (gf
Begin Nate
ineeson
“ime (ey)
Fig. 6. Nitrite-nitrogen profile beneath infiltration gallery.
LSD is the least significant difference between meansat the
‘90% confidence level.
Time cay,
Fig. 7. Benzene profile beneath infiltration gallery. LSD is
the least significant difference between means at the 90%
confidence level.
In the initial part of the test, then, both nitrate and
‘oxygen are available for respiration whereas aerobic metabo-
lism is probably minimal after influent oxygen levels
dropped and nitrate was added. The stoichiometry of BTX
biodegradation can be described as follows:
CoHor + 77.75 0: ~ 61 COs + 33.5 H20
CaHor + 62.2 NOy + 62.24 =
61.00; + 31.1 N+ 646 H20
Based on these equations, about 3.1 kg oxygen are required
for each kg of BTX mineralized under aerobie conditions,
and 1.1 kg nitrate-nitrogen are required for each kg BTX.
mineralized under denitrifying conditions. These estimates
can then be compared with mass removal of selected alkyl
benzenes to assess the potential for biodegradation,
Benzene removal was rapid and complete prior to
nitrate addition (Figure 7). By Day 14, benzene levels were
typically less than 1 yg throughout the contaminated zone.
Significant benzene concentrations were present in the
recirculated infiltration water up to Day 35; these concen-
trations are the result of continued contact of the ground
‘water with the fuel downgradient of the study site. Removal
of benzene occurred before the water reached the Level 2
well points. There was clearly enough oxygen removal in the
system to account for aerobic biodegradation of benzene.
tga tiale ET tone
‘concentration (mai)
Fig. 8. Toluene profile beneath infiltration gallery. LSD is,
the least significant difference betwaen means at the 90%
confidence lev
Toluene removal was more complex (Figure 8). By Day
21, the system had equilibrated hydraulically and toluene
concentrations were similar throughout. Between Day 21
and Day 50, some toluene removal became apparent as the
infiltration water reached the Level 2 well points, After this
time, toluene concentrations dropped rapidly and were
below I gi! at all levels even though concentrations per-
sisted in the recirculated infiltration water. Prior to nitrate
addition, approximately 5 kg toluene were “added” to the
recirculated water asa result of leaching from the soil in the
contaminated interval (Table 1). There was leaching of each
Of the other target compounds at this time as well. The
leaching effect was probably not observed for benzene
because it attained equilibrium more rapidly and/or was
‘more quickly degraded. The simultaneous occurrence of
biodegradation and leaching complicates the interpretation
ofthe relative effects of nitrate and oxygen in the initial part
of the test, For toluene, however, there was a net loss of kg
after nitrate addition, compared to 4 kg oxygen consumed
(Table 1). Given the other sinks available for oxygen, it
seems likely that denitrification accounted for most of the
toluene removal at this time.
Leaching and biodegradation were also apparent for
xylene (Figure 9). Unlike toluene, however, there was
Begia Nate =
oor Iopton ate
E
a
z
i
Time (days)
Fig. 9. m,p-Xylene profile beneath infiltration gallery. LSD is
the least significant difference between means at the 90%
confidence level.
375Begin Nirate -
Tepson
ene
2
concen
af é wo]
Tine (ay) :
Fig. 10. 0-Xylene profile beneath infiltration gallery. LSD is,
‘the least significant difference between means at the 90%
confidence level.
xylene prior to nitrate addi-
tion. After nitrate addition, m,p-xylene concentrations
rapidly dropped by the time the infiltrating water reached
the Level 2 well points, but then rose thereafter as it was
leached from the contaminated zone. A similar pattern was
seen with ethylbenzene (data not shown). There was a net
release of each of these compounds prior to nitrate addition
and a net removal afterwards (Table 1). It is surprising that
concentrations in the Level 6 well points were generally
‘greater than those in the Level 4 well points, considering that
both of these levels are below the original contaminated
zone. This may be due to mobilization of residual fuel
leading to chromatographic movement of xylene through
the contaminated zone, Alternately, it may be due to leach-
ing of xylenes in the lower, more contaminated interval with
concomitant biodegradation occurring in the upper section.
The profiles for o-xylene were unique, indicating no
loss of the compound during infiltration (Figure 10), There
was a gradual decline in total mass due to dilution effects
during recirculation ofthe infiltration feed, but no biologi-
cal activity was observed during this time period, Labora-
tory work has also shown o-xylene to be more recalcitrant
‘than the other xylene isomers under denitrifying conditions
(Hutchins et al, 1991). Mass analysis indicates that there
‘was a small release of o-xylene after nitrate addition as well
as before (Table 1). Although this is within experimental
error, these data indicate that the system might still have
potential for leaching o-xylene.
Core Analyses
Coring is required o correct for inadequacies in moni-
toring well data because the slow release of BTX from
residual saturation inthe contaminated material may cause
aan underestimate, based on water quality analyses alone, of
aqueous BTX concentrations at equilibrium. Core samples
were obtained using aseptic sampling techniques under
nonoxidizing conditions to preserve the microbial commu-
nity structure and chemical integrity of the cores (Leach et
al, 1989). Selected samples were analyzed for total micro-
bial direct counts (Ghiorse and Balkwill, 1983), denitrifier
population counts (Tiedje, 1982), and fuel carbon and JP-4
content (Vandegrift and Kampbell, 988). Core samples
‘were obtained from the study area prior to hydraulic toad-
56
ing (April 5, 1989), prior to nitrate and nutrient addition
(May 23, 1989), and after two months of operation (August
9, 1989). This corresponds to Days -13, 36, and 112, respec-
tively, on the previous infiltration profiles. Two different
locations within the infiltration gallery area were cored from
597.7 to 607.5 ft mean surface elevation for each sampling
event (Figure 2). Fach core was 18 to 36 in. in length and
provided several 3- to 4-in. subcores. These subcores were
collected anaerobically into sterile half-pint Mason jars.
Sterile 10-ml tuberculin syringes, with the tips removed,
were then used to obtain duplicate minicores from each jar;
this provided replicate composites for chemical analyses.
The minicores were transferred to sterile, nitrogen-flled
OA bottles inside the glovebox and sealed with Teflon-
lined silica septa. These samples were analyzed for fuel
carbon and JP-4 content, In addition, these extracts were
analyzed by GC/MS to verify concentrations obtained by
GC analysis and to provide semiquantitative information on
the major classes of hydrocarbons present in the cores.
The distribution of JP-4 within the contaminated zone
at various times during the project is shown in Figure Il. The
contaminated zone covers a depth of about 6 feet and
concentrations vary considerably within narrow depth
intervals The data for each replicate eore profile were aver-
aged to provide mean concentrations of JP-4 throughout
the contaminated interval, and the total mass of JP-4 was
Fig, 11. JP-4 distribution in field demonstration project at
selected time intervals. Elevation is in ft mean sea level and
ind surface is at 615 fi MSL. Original waler table is shown
at 600 ft MSL.calculated for the contaminated zone outlined by the
recharge boundaries, assuming that the separate core loca-
tions were representative of the study area.
Alter one month of floodingto achieve hydraulic equi-
librium, the mass of JP-4 was reduced from a background of|
2250 kg to 1970 kg. Based on statistical analysis of the
replicate minicores in relation to the separate core profiles,
this difference is significant at the 90% confidence level.
Considerable JP-4 remains trapped within the study area
‘The pore volume ofthe affected study area containing the
original contamination was calculated to be 1.1 10°, and
an estimated 600 pore Volumes of recirculated water were
passed through the study area prior to nitrate addition.
After an additional two months of flooding using nitrate
and nutrients, approximately 700 pore volumes were recir-
culated through the study area and the concentrations of
JP-4 were further reduced to 1420 kg. Although this reduc-
tion is significant, it indicates that the aquifer still contains
residual fuel hydrocarbon,
Selected cores were further analyzed by GC/MS to
ascertain changes in composition of the residual fuel hydro-
carbon during the course of the study (Table 2). Samples
were selected from three depth intervals: (1) unsaturated,
contaminated zone; (2) highly contaminated zone at the
water table; and (3) saturated, contaminated zone. The
reconstructed ion chromatograms were evaluated using
selective ion searches for alkanes, cycloalkancs, alkylben-
zenes, and indans,
Priorto hydraulic loading, the distribution was similar
to that of JP-4 in the unsaturated zone with the proportion
of alkylbenzenes increasing inthe saturated zone (Table 2)
This would be expected if the more water-soluble alkylben-
zenes were leached from other regions ofthe site and were in
equilibrium with the residual fuel hydrocarbon, After flood-
ing began, and before nitrate addition, much of the alkyl-
benzenes were removed and the alkane fraction increased
proportionally. Estimates of total fuel by GC analysis for
JP-4 and GC/MS analysis for total class agreed reasonably
‘well, and indicated thatthe bulk mass decreased in the upper
regions ofthe contaminated zone and increased inthe lower
region, probably in response to mobilization of the fuel
After two months of nitrate addition, there was generally a
further decline in concentrations of each class although the
proportions did not change appreciably. Review of the
reconstructed ion chromatograms reveals that most of the
Jower molecular weight compounds ofeach class have been
removed by this time.
All of the core samples obtained from the originally
contaminated area were analyzed for BTX using GC/MS to
provide quantitation of total mass available in the con-
tinuous core for the individual compounds (Table 3). These
data provide definitive evidence of the extent of BTX
removal during the demonstration project. Although
measurable concentrations of each compound remains, the
mass of BTX has been reduced by 9.7% compared 10 36.9%
for JP-4, The concentrations of BTX and JP-4 in each
subcore were used to calculate the expected equilibrium
concentrations of BTX in the pore water within each sub-
core according to the following equation:
Ce = Mi (kin Vet Vn)
‘where Cw is the concentration of the analyte, M is the mass
of analyte per kg core, kw is the partition coefficient
between the fuel and water phases, and Ve and Vy are the
‘Table 2, Changes in Concentrations and Proportions of the Major Classes of Hydrocarbons in Selected Cores
Treatment
Core elevation (ft MSL) Prior to Afier flooding. ‘Afier two months
‘and description Class ‘hydraulic loading ‘before nitrate of nitrate
‘Aligylbenzenes 809 Dt 175( 4) ra)
Indans 133 (2) 63( 2) 12 )
603-604, Alkanes 5460, (74) 3,210 (80) 1,500 (83)
Unsaturated, contaminated Cycloalkanes 97813) ‘586 (14) 266 (15)
Total of above® 7,400 4,090 1,800
Pa 6.720 1,610 1310
Alkgylbenzenes 2010 (13) 124 1) 258 ( 3)
Indans 236 ( 2) 34( 0) 1o1c 1)
600-601, ‘Alkanes 9,170 67) 8,550 (85) 7,080 (80)
‘Water table Cycloalkanes 2,130 (16) 1,380 (14) 1,450 (16)
Total of above 13,500 10,100 8,890
Pa 22,100 6,060 8,300
Aliylbenzenes 65 (28) 230( 4) 704)
Indans 16¢.7) 89(.2) 144)
599.600, ‘Alkanes 11.8 (50) 4,110 (79) 281 (73)
Saturated, contaminated Cycloalkanes 3.6(13) ‘808 (15) B09)
Total of above 25 5,230 385
sP4 46.1 8,204 386
(Data are in ma/kg dry weight, mean of wo replicate subsamples of single cores)
*Total is semiquantitative estimate based on GC/MS analysis of major classes only; JP-4 concentrations based on GC analysis
‘Proportions are expressed in percent
srTable 3. Changes in Average Concentration of BTX Across the Contaminated Zone (598 to 605 ft MSL)
Treatment
Prior io “fier flooding, After wo months
Compound Parameter hydraulic loading before nitrate of nitrate
Mean (mg) ke) 0.838 00315 ‘0.0165
Benzene Range (mg) kg) (0.000-5 49) (0.001-0.183) (0:000-9.173)
Mass (kg) 0357 00135 0.0071
Mean (mg/kg) 226 0.134 0.0360
“Toluene Range (mg/ ke) (0.003.216) (0.005-1.09) (0.010-9.106)
Mass (ke) n9 00872 0.0154
Mean (mg/kg) 183 0.356 0.0190
Ethylbenzene Range (mai ka) (0.000-106) (0.002-3.60) (0.002-0.057)
Mass (kg) 181 0.152 0.0081
Mean (mg/kg) 58.0 137 0.0587
‘mp-Xylene ange (mg/kg) (0.003-325) (0.079.282) (0.003-0,188)
Mass (kg) 248, 3S 0.0251
Mean (mg/kg) 255 32 0.272
o-Xylene ‘Range (mg/kg) (0.02:136) (0077-156) (0.028-4.77)
Mass (kg) 09 1.38 0.116
‘Mean (mg/kg) 13s 2 0.401
Total BTX. Range (mg/kg) (0.008-780) (0.116575) (0.092-5.29)
Mass (kg) 316 478 0.71
‘Mean (mg/kg) 5260 4620, 3310
sp4 Range (mg) ke) (1,3-29,400) (629-18,300) (4.2-25,800)
Mass (kg) 2250 1970 1420
(Concentration data are in mg/kg dry weight, mean of entire core segment from two continuous cores. Data in parentheses are range of
Values. Mass values are calculated from mean concentrations and represent total mass in affected study area.)
volumes of fuel and water, respectively, per kg core. The
concentrations of each analyte were weighted for the
sampled interval and averaged to provide an equilibrium
concentration of the analyte for the entire contaminated
interval. This provides an indication of the maximum BTX
concentrations in solution once flow stops and BTX parti-
tions into the pore water from residual fuel.
‘These predicted values were compared with the actual
concentrations in the monitoring wells (Table 4). The data
agree reasonably well, although actual concentrations are
much higher than predicted for the samples obtained prior
to hydraulic loading, This is because the Level 2 well points
were in the unsaturated zone at this time, and this necessi-
tated the use of the Level 3 well points located at the water
table within the most contaminated interval. This causes the
actual mean BX concentrations to be higher than those
averaged fo the other monitoring wells.
These data indicate thatthe previously contaminated
Table 4. Comparison of Actual BTX Aqueous Concentrations with Predicted Values at Equilibrium, Based on Partition
Theory Using Residual BTX and Fuel Mass
Treatment
Panition (concentrations in ug)
coeffcient* Prior to Afr flooding, ‘fier p90 months
Compound (og kw) ydraulic loading before nitrate of nitrate
Benzene 339 ‘Actualh 760,