EVALUATION OF THE POTENTIAL WATER SUPPLY IMPACTS
PROPOSED BY A GASOLINE PIPELINE
Richard C. Johnson, P.G., P.E.
Associate, Camp Dresser & McKee Inc.
Stergios A. Dendrou, Ph.D., P.E.
Senior Engineer, Camp Dresser & McKee Inc.
In January 1974, the Florida Gas Transmission Company proposed to convert
an existing 24-inch natural gas pipeline to a pipeline which transports
light petroleum products. This pipeline travels through Florida from the
Panhandle areas to Port Everglades in Fort Lauderdale. Camp Dresser &
McKee Inc. (CDM) was engaged to evaluate impacts on the ground water system
which would occur if the pipeline leaked. CDM conducted an extensive
hydrologic analysis to determine processes governing gasoline entry into
the ground water system and the expected area of contamination after entry.
Results of this analysis were incorporated into ground water flow and
pollutant transport models developed to simulate the movement of light
petroleum products in the Biscayne Aquifer. Four types of representative
pipeline spills or leaks were considered. In addition, public water supply
well fields threatened by a pipeline spill or leak were identified and
potential contamination was assessed in terms of risk probabilities and
pollutant travel time zones.
The existing Transgulf, Inc. pipeline runs from Louisiana to southeast
Florida (Figure 1). In southeast Florida, the pipeline lies within the
physical space of the Biscayne Aquifer, a sole-source water supply located
a few feet below the ground surface. This aquifer extends from the surface
to a depth of more than 300 feet along the coast, thins westward to a depth
of about 70 feet in central Broward County, and pinches out near the west
county line. The Biscayne consists of about 80 feet of medium permeability
sand which overlies at least 120 feet of highly permeable limestone
interbedded with sandstone and sand. It is underlain by a thick sequence
of relatively impermeable clayey materials which in turn overlie the
permeable limestone formations of the Floridan aquifer. The focus of this
study is the evaluation of the potential ground water system impacts that
may be caused by accidental spills or leaks from the pipeline.
Petroleum products (hydrocarbons) in public water supplies cause taste,
odor, and appearance problems before health or treatment problems occur.
On this basis, the maximum allowable hydrocarbon concentration is therefore
the minimum detectable concentration. The World Health Organization (1971)
International Criteria for hydrocarbons is 0.2 micrograms per liter (ug/l).
Any accidental spill or leak that results in detectable hydrocarbon
concentrations in Biscayne Aquifer waters will contaminate the aquifer and
TRANSGULF PIPELINE J p) ORT
Figure 1: Location of Transgulf Pipeline from Baton Rouge
to Port Everglades in Broward County, Florida
public water supplies. The greatest impacts from a potential pipeline leak
would occur in Broward County, which contains the largest area of the
aquifer traversed by the pipeline and the largest number of public water
supply well fields potentially affected.
A statistical evaluation of the existing pipeline in terms of potential
leakages, spill scenarios, volumes of gasoline spilled, rates of leakage,
etc. was performed by NDE Technology, Inc. (1982) and provided input to
this study. The NDE study determined that there will be an average of two
to three spill events in Broward County over a 20-year period. To evaluate
the potential impacts from these accidental spills or leaks, ground water
flow and pollutant transport models are used to simulate the movement of
light petroleum products in the Biscayne Aquifer. Head/velocity fields are
first calculated using finite difference computer modeling techniques that
incorporate complex variations in aquifer properties and the effects of an
extensive canal system designed for drainage and recharge control.
Pollutant movement is simulated using a random-walk, mass transport
computer model combined with the previously generated head/velocity fields.
The products that are to be conveyed through the pipeline are primarily
gasoline, and occasionally jet fuel and fuel oil. These materials have
high seepage rates, high volatilization rates, and high dissolution rates
into water. The gasoline-type products therefore reach the water table
very fast irrespective of the amount of leakage, and provide a long-term
source of dissolved hydrocarbon materials. Gasoline from a spill or leak
in the proximity of a well may contaminate the well and the surrounding
aquifer area directly, but the dissolved material can travel within the
ground water system and potentially contaminate distant wells and a much
larger area of the aquifer. The problem analysis must therefore evaluate
the potential impacts of both the gasoline mass contamination and the
dissolved materials contamination.
The underground movement of gasoline from a pipeline spill or leak takes
place in three stages: 1) downward seepage, 2) lateral spreading and
dissolution, and 3) convection/dispersion. Downward seepage takes place
through the pore spaces of the unsaturated zone and leaves a trail of
residual hydrocarbons to coat the surface of the soil particles. Lateral
spreading and dissolution into the water occur by gravity and capillary
forces when the gasoline reaches the water table, floats on the ground
water surface, and spreads influenced primarily by the presence of water
table gradients. Migration of the dissolved hydrocarbon constituents of
gasoline then occurs primarily by the processes of convection and
dispersion. These stages are described schematically on Figure 2. Each
stage takes place over a different time scale and involves different
volumes of the gasoline. An order of magnitude analysis presented
hereafter determines the relative importance of each stage in the potential
contamination of the aquifer and water supply well fields.
The velocity of downward gasoline movement through the unsaturated zone may
be described by Darcy's Law incorporating the gradient and the fluid
conductivity. The fluid conductivity is dependent on the pore size and
distribution, and on the fluid properties; including the intrinsic
permeability (a property of the soil matrix along), the specific weight of
the fluid, and the viscosity of the fluid. For gasoline, the specific
weight is 0.75 and the viscosity is 0.65 centipoises at normal ground water
temperature. A typical intrinsic permeability in the Biscayne Aquifer is
0.85 centimeters per second (cm/sec) or 0.018 mgd/ft which is indicative
of granular, sand and travel material (CDM, 1980). Fluid conductivity for
gasoline is then calculated as 860 gallons per hour per square foot
(gal/hr/ft ) and is 15 percent higher than the fluid conductivity for
At saturation, the rate of downward seepage per foot of head across a
cylinder approximating the gasoline core of Figure 2, is equal to 200,000
gal/hr/ft. Fluid conductivity in the unsaturated phase is smaller, but can
be as high as the saturated case during the propagation of the wetting
front. Less than one hour is required for the gasoline product to reach
the water table and about two hours is required to reach steady-state
Lateral Spreading and Dissolution
Upon reaching the water table, the gasoline product will start dissolving
into the water. The dissolution rate is presently believed to be a
function only of the total surface of contact between the product and
*6 .'i rums vur flULAWe
S GASOLINE CORE
.* VAPOR ZONE (EVAPORATION ENVELOPE)
S- DISiOLVED. '*.** '.
= "' ..N HYDROCARBONS '.* ,
SATURATED GROUND WATER FLOW ** /
Underground Movement of Gasoline from a leak or Spill
(adapted from Pfannkuch, 1982)
water, characterized by the mass transfer exchange coefficient, E, in units
of mass per unit area per unit time. For gasoline, the mass transfer
exchange coefficient is estimated (Convery, 1979) to be 0.1 milligrams per
square meter per second (mg/m /sec).
The surface of contact involves the full depth of the unsaturated zone when
the gasoline and water are in contact. Assuming that the surface of
contact is initially a circle of 20 feet in diameter, the dissolution rate
is approximately 10,500 mg/hr, or approximately one and one-half pounds per
day. This rate increases as the gasoline mass spreads laterally, drifts
downgradient, and the surface of contact increases. Lateral spreading and
drift primarily affect the amount of dissolved material. These processes
add relatively little to the gasoline migration, and they are confined to
the immediate vicinity of the spill. If a water supply well is located
within a range of about 200-1,000 feet from a spill, however, the gasoline
mass can reach and contaminate the well.
The dissolution and the ensuing dispersion are processes of a longer time
scale than the initial downward seepage. On the other hand, the dispersed
material can travel long distances. Therefore, the hydrocarbons that
dissolve into the ground water constitute the greatest threat to public
water supplies. The dissolved hydrocarbons move through the aquifer system
by several mechanisms. Convection is the best understood of all the
processes contributing to mass transport. This is the movement caused by
the mechanical action of the fluid flow on the pollutant. In the absence
of other processes, convection is the same as the average movement of the
fluid particles, and the corresponding flow time is the so-called hydraulic
travel time. Convection accounts for a large part of the pollutant
transport near well fields, where velocities are high as ground water flow
is accelerated toward wells.
In the absence of ground water flow (zero convection), a slug of pollutant
will expand around the initial point of release by the processes of
molecular diffusion and dispersion. Dispersion is the tendency for the
solute to mix and spread from the path of convection flow. This tendency
is caused by molecular diffusion and microscopic variations among
velocities within individual pores (microscopic diffusion), and by
large-scale heterogeneities macroscopicc diffusion) within the aquifer
Movement of the dissolved material by convection flow is described by the
Darcian velocity, which is the product of the hydraulic conductivity and
the hydraulic gradient divided by the effective porosity. Assuming a
gradient of eight feet per 200 feet or 0.004 ft/ft, the Darcian velocity is
calculated as 48 ft/day. Approximately 40 days are therefore required for
the dissolved material to travel 2,000 feet and reach the well at
steady-state, Dispersive effects will accelerate the movement so that the
dissolved material will start reaching the well much sooner. At
steady-state, all dissolved material will be fully mixed in the pumped
water. The 0.004 gradients are typically encountered in 2-mgd production
wells, equal to 315,000 liters per hour (1/hr). The full mixed
concentration would then be 10,500 mg/hr divided by 315,000 1/hr, or 0.033
mg/1. Thus, the 0.2 ug/1 (World Health Organization, 1971) threshold will
be exceeded much earlier than 40 days. On the other hand, major well
fields are comprised of a system of many wells, with the supply water being
completely mixed from all the wells. This can delay the time of detection,
thereby resulting in a larger permanent damage to the well field.
ANALYSIS OF SCENARIOS
The variety of physical processes previously discussed relates to
hydrocarbon migration in ground water flow systems and in particular to the
Biscayne Aquifer. Several pipeline spill scenarios are analyzed to fully
quantify the complete range of possible effects on the aquifer and public
water supplies from pipeline gasoline leakage. Several pipeline failure
scenarios were advanced in a statistical evaluation of the pipeline by NDE
technology, Inc. (1982). The spill volumes and durations for each
anticipated failure scenario are presented in Table 1.
The first scenario involves a major rupture of the pipeline and subsequent
release of large volumes of gasoline under high pressure. The resulting
geyser would initially spread about 1.5 million gallons of gasoline over
the land surface where it would flow into depressions, nearby canals and
lakes, and along the pipeline trench, and eventually enter the ground water
The second spill scenario represents a minor break and is classified in
Table 1 as an average spill. A leakage of this magnitude would presumably
be detected within about one month and cause an immediate shut down of the
pipeline. Rapid release of gasoline in the preliminary stages of the spill
ESTIMATED SPILLAGE FROM TRANSGULF PIPELINE IN BROWARD COUNTYa
Spillage Condition (Total Volume) Spill Rate Duration Spill Size
Gallons Gallons/Hour Hours Gallons
1. Pipeline Rupture
6-Inch Dia. Hole) 435,354 1,443,600 0 to 0.066 96,240
34,480 0.066 to 1.2 41,376
34,480 1.2 to 9.8 297,738
2. Average Spillage
2-Inch Dia. Hole) 164,415 160,380 0 to 0.066 10,692
4,875 0.066 to 1.2 4,062
4,062 1.2 to 37.84 148,848
3. Undetected Leakage
Dia.Hole) 244,480 24,486 10
Dia.Hole) 122,243 12,243 10
Dia.Hole) 400,000 500 1 Year
Dia.Hole) 140,160 16 1 Year
Dia.Hole) 200,000 9 Continuous
a From NDE Technology, Inc. (1982)
would force some of the gasoline to the surface or along the buried
pipeline, posing a threat to the surface water near the break. However,
most of the gasoline would enter the soil and seep rapidly towards the
ground water table.
The final scenario in Table 1 represents leakages that would not be
detected by the safety measures incorporated in the design and operation of
the pipeline (NDE, 1982). The volume of gasoline involved depends on the
size of the pipeline hole and the associated duration (time before
detection). These leakages would be detected by other means: obvious
ground water contamination of private wells, gasoline seepages into surface
depressions or surface water bodies, contamination of public water
supplies, etc. Undetected leakages product an equally serious threat to
the Biscayne aquifer since large areas of the aquifer can become
contaminated before the leak is detected.
Based on the proceeding discussion, three types of pipeline leakages can
occur (Table 1): pipeline rupture, average spill, or undetected leak. The
impacts from each of these leak types vary based on the type of leak and
the proximity to wells. For detectable leaks, well field proximity
increases the probability that major water supply contamination will occur.
For undetectable leaks, large areas of the aquifer can be contaminated
before detection and remedial action. To evaluate the different variations
under typical spill/location conditions, the following scenarios are
analyzed in detail in the next section:
o undetected gasoline spill near a well field,
o detected gasoline spill near a well field, and
o gasoline spill away from a well field.
The gasoline mass infiltrates rapidly (within one hour) and reaches the
water table which acts as a barrier. Under the influence of gravity and
capillary forces, the gasoline mass spreads laterally, while continuing to
dissolve into the water at a slow rate (0.1 mg/a /sec). Thus, the core of
the gasoline mass stays in the immediate vicinity of the spill site (within
about 200 to 1,000 feet), while the dissolved hydrocarbons slowly disperse
to greater distances. In most cases, the dissolved hydrocarbons reach and
contaminate water supply wells. To further quantify the potential extent
of well field contamination, a pollutant mass transport/dispersion model is
used. The dispersion model and the approach followed in the scenario
analysis are presented in this section.
Convection and dispersion are the fundamental processes that account for
the movement of dissolved hydrocarbons in ground water, and the notion of
travel time is also very useful in quantifying the effects of gasoline
migration. Existing land use ordinances in Dade County and Broward County
regulate sources of pollution within a 200-day pollutant travel time zone
to protect the public water supply from contamination. The size of the
200-day travel time zone varies between 200 to 1,000 feet, which is
approximately the size of the area directly affected by the spilled
gasoline core. Thus, gasoline spillage within the 200-day travel time zone
can be considered as catastrophic. It is denoted as scenario zero.
Larger travel time zones are considered to be zones of lesser risk. The
presence of regional inhomogeneities and anisotrophy (stratification),
canals and other influencing boundary conditions, and multiple well fields
in the study area require the use of a numerical model to quantify the size
and extent of these zones. The Prickett-Naymik-Lonnquist model (1981) was
used to simulate transient and steady-state flow conditions with variable
pumpage rates, rainfall recharge, canal leakage, evapotranspiration,
induced infiltration, transmissivity, and storage coefficients.
Potentiometric heads and Darcy velocities are computed by solving the
second order partial differential equation for ground water flow over a
discretized, spatial domain (two dimensional in the horizontal plane) at
variable time increments.
The hydraulic portion of the model provides the velocity field for a
compatible solute transport, random-walk code. This model portion computes
the convective movement of particles representing a constituent mass, and
adds a random longitudinal and transverse step to account for longitudinal
and transverse dispersion. The transport process is repeated over a large
number of particles and a large number of steps to replicate the dispersion
"spreading" process accurately and economically. The random-walk model was
used because of its accuracy, economy, and versatility in computing travel
This is the scenario of a leak in the immediate vicinity of a well field,
within the 200-day travel time zone. In this scenario, gasoline will
likely directly contaminate the well field, within a few hours following
the accidental leak. This case should be qualified as catastrophic for all
practical purposes. No additional analysis is required.
Scenario one is the rupture scenario. The total spilled volume is 435,000
gallons. Such an accident will be detected immediately and mitigative
actions undertaken within 10 hours. The gasoline is under high pressure
and the resulting geyser is likely to spread over an area about 300 feet in
diameter. Such an area will be producing dissolved hydrocarbons in the
aquifer for at least 10 days, assuming that the site of the spill can be
completely cleaned up and contained in 10 days. All well fields in the
area are likely to be immediately shut down. Since water table recovery of
the cone of depression can be achieved within a few hours, dispersion of
the dissolved material will take place under the effect of regional
This case was simulated using the mass transport model, and the resulting
movement of the dispersion plume is shown on Figure 3-A. The initial
circle centered around the "point of release," represents the extent of the
initial contamination. The remaining elliptic-shaped curves represent the
position of the plume after 100 days, 200 days, 500 days, 1,000 days, and
1,500 days, respectively. Assuming no significant loss or biodegradation,
the average concentration of the plume after 1,500 days is still higher
Go GROUND WATER --
Soo00 CONTOUR (FT)
"0O # DAYS I
)F 200 DAYS 1000 DAYS-
I 10so DAYS I
AREA 500,000 SO.FT.
rupture would result in a large dissolved hydrocarbon dispersion
if the gasoline is recovered immediately.
An average pipeline leak results
the gasoline is recovered within
in a relatively
0 100 200 300 400
SCALE IN FEET
AREA 430,000 SQ.FT.
small dispersion plume if
AREA 170,000 O.FT.
An undetected pipeline leak can contaminate as large an area of the aquifer
as a pipeline rupture.
than the 0.2 (ug/1) threshold specified by the international standards. As
shown on Figure 3-A, a large area of the aquifer, on the order of 500,000
square feet, is contaminated.
Scenario two is the scenario of an average spill. The total spilled volume
is 164,000 gallons. Such an accident could be detected within about 40
hours and the polluted site could be cleaned up or contained within about
one month. The average spill is likely to spread over an area 30 feet in
diameter producing dissolved hydrocarbons in the aquifer for one month.
This leak can occur away from well fields or close to well fields, but is
likely to be detected fairly early, so that the plume will disperse under
the influence of regional gradients only. This case was simulated as
illustrated on Figure 3-B. The total dissolved material is less than in
the analysis of scenario one, and the areal extent of the contamination is
reduced, affecting approximately 170,000 square feet of the aquifer.
Scenario three represents the case of an undetected leak with a total spill
of 400,000 gallons. Such a leak can go undetected for as long as one year.
The circle of contamination around the point of release increases
continuously over the year from a few feet to as, much as 300 feet. This
leak is likely to remain undetected for a year only if it is away from well
fields. Therefore, the plume will disperse under the effect of regional
gradients. This scenario was simulated as illustrated on Figure 3-C. The
total affected area is on the order of 430,000 square feet. This result
shows that despite the low rate of release, the undetected leak can prove
as disastrous as the rupture case (scenario one).
SUMMARY AND CONCLUSIONS
From the analysis of the dispersion plume in the three scenarios of Figures
3-A, 3-B, and 3-C, areas where the aquifer may become contaminated extend
at least 1,500 feet downgradient (i.e. east of the pipeline), assuming that
the prevailing regional. gradients are the same along the length of the
pipeline. Such conditions prevail along the pipeline in Broward County.
In addition, some contamination about 300 feet upgradient is indicated by
the figures due to the initial spread from the point of release. A zone of
potential hazard is thus delineated, approximately 1,800 feet wide, along
the length of the pipeline. Six major public water supply well fields are
located in the immediate vicinity of the potential hazard zone.
To further quantify the extent of the damage that may result from a
gasoline leak, three zones are delineated around these well fields, a zero
to 200-day travel time zone (zone of immediate danger), a 200 to 500-day
travel time zone (zone of high risk), and a 500 to 1,500-day travel time
zone (zone of moderate risk). The 1,500-day limit is used in the present
conservative calculation because beyond this period of time, biodegradation
may become important. These zones were developed using the dispersion
model by inverting the simulated head/velocity field and releasing
particles at the wells. The particles were allowed to move through the
velocity field for the desired travel time increased by 25 percent to
f account for dispersion. The locus of particles around the well field
defines the desired travel time zone.
The ratio of the length of the pipeline traversing each zone to the total
length of the pipeline in Broward County is a measure of the conditional
probability that if a leak occurs in Broward County, this leak will affect
a major public water supply. The total length of the Transgulf Pipeline in
Broward County is 25.6 miles. The length of the pipeline within each risk
zone, the corresponding probability of contamination, the estimated well
field replacement cost, and the population served by each well field are
presented in Table 2. The total risk of catastrophic contamination
(inrediate danger zone) of a well field in Broward County in the
eventuality of a pipeline leak is 9.4 percent. The total risk of
contamination in the high risk zone is 21.3 percent, and the total risk in
the moderate risk zone is 31.7 percent. That is, if there is a pipeline
leak in Broward County, there is at least a 1/3 chance that a major public
water supply well field will be affected.
Anderson, M.P., 1979, "Using Models to Simulate the Movement of
Contaminants through Groundwater Flow Systems," in Critical Reviews in
Environmental Control, v. 9, issue 2.
Camp Dresser & McKee Inc., 1980, "Prospect Well Field Impact Analysis,"
Convery, M. P., December 1979, "The Behavior and Movement of Petroleum
Products in Unconsolidated Surficial Deposits," Ph.D. Thesis, University of
NDE Technology, Inc., 1982, "LPP Pipeline Study," unpublished report.
Pfannkuch, H.O., 1982, "Problems of Monitoring Network Design to Detect
Unanticipated Contamination," GWMR.
Prickett, T.A., Naymik, T.G., and Lonnquist, C.G., 1981, "A Random-Walk
Solute Transport Model for Selected Groundwater Quality Evaluations,"
Illinois State Water Survey Bulletin 65.
World Health Organization, 1971, "International Standards."
Richard C. Johnson, P.E., P.G.
Camp Dresser & McKee Inc.
1945 The Exchange, N.W.
Atlanta, Georgia 30339
" Stergios Dendrou, Ph.D., P.E.
Camp Dresser & McKee Inc. -11-
7630 Little River Turnpike 313/19
Annandale, Virginia 22003