The Ohio River Oil Spill:
A Case Study
Robert M. Clark, Alan H. Vicory, and James A, Goodrich
A dime-sized flaw contributed to the rupture of a 45-year-old diesel oil storage tank
releasing nearly 800,000 gal of oil into the Monongahela River.
The spill of diesel oil fuel in January 1988 raised a number of
technical, legislative, and administrative issues—such as
assessing long- and short-term environmental damage,
evaluating regulations regarding oil tanks, and examining
spill response procedures. An important research need is the
development of computer models that can better predict the
travel time and the concentration of contaminants should a
spill occur.
Although industrial discharges
from point sources are regulated
by the National Pollutant Discharge
Elimination System, some toxic pollutants
are still detected in US surface waters. Pol-
lutants may come from nonregulated or
MARCH 1990
ROBERT M. CLARK ET AL 39
-------
toxic substances allowed in low concen-
tration in permitted discharges, acci-
dental or deliberate spills, nonpoint
sources, or stormwater runoff. Fre-
quently, these same surface waters are
major sources of drinking water. The -.
National Organics Monitoring Survey
(MOMS) conducted by the'US Environ-
mental Protection Agency fUSEPA)
examined 113 water systems and found
129 organic compounds—including car-
bon tetrachjoride, 'benzene, trichloro--^
ethylene, vinyl chloride, styrene, and
l,2»dichloroethane—in drinking-water.1
The Safe Drinking Water Act (SDWA)
and its amendments have increased the
number of maximum contaminant levels
(MCLs) that drinking water, utilities
must meet.2 In some cases, the target
levels for MCLs are being lowered. The
amendments and recent concern over
toxic discharges from wastewater treat-
ment plants have forced an increasing
awareness of the impact of upstream
discharges on drinking water quality.
This awareness has led to a realization
that drinking water utilities are highly
vulnerable to upstream point and non-
point sources of pollution. A water utility
that finds one or more MCLs violated at
its raw water intake may in the future
seek to identify the upstream discharger
and request that regulatory agencies
force the discharger to install controls,
rather than having the utility pay for
expensive water treatment processes.3
A massive spill of diesel oil on the
Monongahela River provided a striking
example of this vulnerability. A storage
tank containing more than 3.8 mil gal of
diesel oil collapsed Jan. 2, 1988, near
Pittsburgh, Pa. Nearly 800,000 gal of
diesel fuel breached an earthen barrier
surrounding the tank and entered the
Monongahela River through storm sew-
ers, 25 mi upstream from Pittsburgh.
Normal procedures used to control oil
spills were only partially successful, and
the diesel fuel soon began to mix with
the water. The spill also pushed through
'several locks and dams, causing the
diesel oil to mix vertically in the water
column. Approximately 30 percent of the
spilled fuel entering the river was recov-
ered with booms and vacuums. Figure 1
shows the Ohio River, including the
Monongahela and Allegheny rivers,
which meet to form the Ohio River at
Pittsburgh.
As the slick moved slowly past Pitts-
burgh, then into the Ohio River, water
plants prepared to close their water
intakes. The first utilities affected were
the West Penn Water Company, just
downstream of Pittsburgh, and West-
view Water Authority, just downstream
of West Penn. By Monday, January 4, the
slick was within 10 mi of the East
Liverpool, Ohio, water plant at mile
40 MANAGEMENT AND OPERATIONS
Penn«ylv«n!i
ALLEGHENY
RIVER
pmiburgti
1 L
WhMlIno "
MONONGAHELA
P.rtc.r.burg'' RIVER
Figure 1. ORSANCO area of responsibility in the Ohio River Basin
Figure 2. Daily flows at Wheeling, W. Va.
JOURNAL AWWA
-------
Monongahata
Rl.er
„ i i i i i i I i i i i—i—i—I—i—I—I—I—I—I—I—I—I—I—1—I—I—1—I—I—1—I—I—I—I—I—I
-30 -10 0 10 30 50 70 90 110 130 150
Figure 3. Oil concentration versus Ohio River mile on day 12
January 17
January II
January 19
January 20
January 21
(0 120 160 200 240 290
Ohio Rlvar Milt
Figure 4. Travel distance and change in concentration of Ohio River spill from
January 16 through January 21
TABLE 2
Comparison matrix for major water Quality models
TABLE 1
Municipal water intakes along
the Ohio River
Location of Intake
West View, Pa.
Robinson Township, Pa.
Coraopolis, Pa.
Sewickley, Pa.
Moon Township, Pa.
Edgeworth, Pa.
Ambridge, Pa.
Aliquipps., Pa.
Baden, Pa.
Conway, Pa.
Freedom, Pa.
Monaca, Pa.
Beaver, Fa.
Midland, Pa.
East Liverpool, Ohio
Chester, W.Va.
Wellsville, Ohio
Toronto, Ohio
Steubenville, Ohio
Mingo Junction, Ohio
Wheeling, W.Va.
Martins Ferry, Ohio
New Martinsville, W.Va.
Sisterville, W.Va.
Gallipolis, Ohio
Huntington, W.Va.
Ashland, Ky.
Ironton, Ohio
Greenup, Ky.
Portsmouth, Ohio
Maysville, Ky.
Cincinnati, Ohio
Covington, Ky.
Newport, Ky.
Oldam County, Ky.
Louisville, Ky.
Mile Point
4.5
8.6
10.2
11.4
11.7
12.8
17.4
19.3
20.1
21.5
23.8
25.3
26.0
36.3
40.2
42.1
47.2
59.0
65.2
71.0
86.3
88.6
128.1
137.1
268.6
304.2, 306.9
319.7
327.2
334.7,336.2
350.8
408.5
462.8
462.9
463.5
582.2
594.5, 600.6
Parameter
Applicable aquatic
systems
Number of dimensions
in aquatic system
Applicable toxic pollutants
Kinetic representation
Dynamic pollutant loading
Integrated sediment or
benthic nodules
Dynamic hydraulic
transport
Integrated hydraulics
Water Quality Model
TOXIWASP
Unlimited*
2t
Most
2nd order,
process
Yes
Sediment,
benthic
Yes
Not
WASP
Unlimited*
2t
Most
2nd order,
process
Limited
Benthic
Yes
Not
EXAMS
Streams and
rivers
1
Most
2nd order.
process
Limited
Sediment,
benthic
No
Yes
HSPF
Streams and
wells, mixed
1
Some
1st order.
gross
Yes
Sediment,
benthic
Yes
Yes
QUAL-II
Streams and
, lakes
1
Some
1st order,
gross
No
Benthic
No
Yes
DYNHYD-
DYNQUAL
Unlimited*
2t
Some
1st order,
gross
Yes
Benthic
Yes
Yes
*Applicable for most stratified lakes and reservoirs, large rivers, estuaries, and coastal waters
fDYNID-DYNQUAL can only represent the horizontal and longitudinal dimensions; TOXIWASP and WASP can represent the longitudinal
and the vertical or horizontal dimensions.
tDYNHYD has been used for the hydraulic part of WASP and TOXIWASP.
MARCH 1990
ROBERT M. CLARK ET AL 41
-------
point 30 (30 mi downstream from the
origination of the Ohio River at Pitts-
burgh). The intake valves were tempo-
rarily shut down while samples were
tested. On Tuesday, when the valves
were closed, the slick was approximately
28 mi long, and oil was found as deep as
16 ft. East Liverpool had a three-day
emergency water supply in reserve. By
Wednesday, the valves were reopened
and normal treatment was successfully
initiated. Although other treatment
plants along the Ohio dealt with the
problem similarly, treatment varied
because of the sudden changes in
weather, including the river freezing
and subsequent lack of movement of the
water past the intakes, necessitating
extended shutdown of some water
intakes. By January 27 the spill had
reached Louisville, Ky., 600 mi down-
stream, with diesel oil concentrations
(based on fluorometric measurements)
returning to background levels.
The spill area
The Ohio River begins at the con-
fluence of the Allegheny and Monon-
gahela rivers just below Pittsburgh, Pa.
(Figure 1). The Ohio River is nearly
1,000 mi long and flows through or
borders six states. It carries the waters
of a myriad of tributary streams that
stretch into 13 states, and its drainage
area covers more than 200,000 sq mi.
Approximately 10 percent of the popula-
tion of the United States lives in the
Ohio River Valley, with approximately
3.5 million people depending on the Ohio
River as a source of raw water supply.
Table 1 lists the water intakes that are
located along the river from Pittsburgh,
Pa., to Louisville, Ky.
Flows in the Ohio River normally
range from 35,000 to 220,000 cfs at
Cincinnati. At the time of the spill, the
flow in the Ohio River was 95,000 cfs.
Several days after the diesel oil spill
occurred, temperattires dropped into the
single digits, causing freezing in the
upper 100 mi of the Ohio River. River
velocities were reduced to a rate of less
than 0.5 mph, and river flow was reduced
to 25,000 cfs. On January 19, however,
the entire Ohio River Valley experienced
heavy rainfall and an extended warm
spell, increasing the river flows to more
than 200,000 cfs and the river velocities
to 3.1 mph. Figure 2 shows average daily
flows at Wheeling, W. Va., during the
spill period and illustrates one of the
major problems in predicting the move-
ment of the spill—the wide variation in
flow rate.
Spill conditions
As stated earlier, the cause of the spill
was the collapse of a 3.8-mil-gal diesel oil
tank. A number of possibilities were
January 27
Ohio River Mile
Figure 5. Travel distance and change in concentration of Ohio River spill January
27 at Louisville, Ky.
• Fluorescence
O QC concentration
• EMImated concentration
Figure 6. Estimate of reduction in peak concentration
investigated as to the cause of the tank
failure. Shifting of the clay and limestone
foundation was suspected as a possible
cause of the tank collapse that fouled
two rivers and triggered a drinking water
crisis in three states. Specifications
require a 3- to 4-ft-thick base of com-
pacted clay, topped by 1 ft of crushed
limestone. Before the tank was erected,
the foundation was rolled and soft spots
were filled and compacted. It was not
known whether the stone and clay pad
under the tank sank before the collapse
or was eroded by escaping oil. Inves-
tigators began taking soil samples from
the 120-ft-diameter foundation that
supported the steel tank bottom. There
was no visible settlement on the tank
pad, but when the tank collapsed, the
bottom ripped out and washed away a
great deal of stone. Portions of the tank's
foundation appeared to have sunk be-
tween 12 and 18 in.
Other possible causes of the failure
were fatigue of the 45-year-old steel used
for the tank, subfreezing temperatures
42 MANAGEMENT AND OPERATIONS
JOURNAL AWWA
-------
on the day of the accident, or problems
with vents in the tank.
A team of researchers ultimately dis-
covered that a dime-sized flaw on the
45-year-old tank contributed to the rup-
ture. A combination of factors aggra-
vated the flaw and caused the tank to
split. The tank was assembled in Cleve-
land in 1940 and rebuilt in 1986 at the
company's terminal south of Pittsburgh.
The original welds and those used during
the reassembly made the tank's steel
walls more brittle. Other contributing
factors were the relative weakness of the
World War II steel, pressure from the
diesel fuel, and cold weather.
The flaw, which consisted of a small,
rusting cavity about 0.125 in. deep, was
caused by a torch and was in the steel
before Ashland Oil personnel assembled
the tank.
Institutional and monitoring conditions
Once the spill occurred, procedures
for spill notification went into effect. An
Ashland Oil representative notified the
Pennsylvania Department of Environ-
mental Resources, which, in turn, noti-
fied USEPA's Region III office in Phila-
delphia. From that point the proper
procedures were followed to inform the
other states, USEPA regional offices,
other federal agencies, and water utilities
that would potentially be involved in
tracking and monitoring the spill. The
Ohio River Valley Water and Sanitation
Commission (ORSANCO) played a major
role in coordinating the dissemination of
information and emergency and remedial
action.
The commission maintains an organ-
ics detection system and a spill response
system. Both of these systems played a
major role in monitoringand tracking of
the spill.4
Organics detection system. The detec-
tion system consists of 13 laboratory
stations operated in cooperation with 11
water utilities and 2 industries. River
samples are collected daily and analyzed
for 16 halogenated compounds. The iden-
tification and control of toxic substances
is a high priority concern of ORSANCO.
Spill notification. An emergency re-
sponse directory that provides informa-
tion and telephone numbers concerning
the reporting of spills to the appropriate
agencies is issued twice a year. The
commission also maintains an emer-
gency response manual that details the
response procedures for spills affecting
the Ohio River and that lists the location
of water intakes, discharges, locks and
dams, and river terminals. The response
manual lists specific responsibilities for
each agency concerning notification,
tracking the spill, monitoring water
quality, treatment modification, inter-
agency communication, and notification
MARCH 1990
of the public. The manual is updated
annually.
The commission keeps an electronic
bulletin board for timely dissemination
of spill information. River flow forecasts
and information concerning spills are
posted daily at ORSANCO headquarters.
Water users and agency personnel can
have instant access to the system by
computer to learn the status of a spill.
The electronic bulletin board supple-
ments the 24-hour telephone service for
receiving notification reports.
Monitoring the spill
A major problem in tracking a spill
such as diesel fuel oil is measuring
concentrations of the contaminant and
its constituent parts. Diesel fuel is made
upof a number of volatile and nonvolatile
compounds, so that at any time the
individual constituents and associated
concentrations in the spill plume may
change. For example, soon after a spill,
the plume will have a much higher
proportion of volatile compounds. This
ratio changes the longer the plume is
exposed to the atmosphere. In this case,.
the plume passed through three locks
and dams within the first 50 mi on the
upper Ohio, and the fuel oil was thor-
oughly mixed in the water column.
Because diesel fuel fluoresces, a fluo-
rometer was obtained soon after the spill
for most of the tracking, based on indi-
vidual grab samples. The US Army Corps
of Engineers later provided a flow-
through fluorometer that proved to be
more convenient.
Variables such as hydrologic condi-
tions, ice, and the processes of dispersion,
emulsification, microbial degradation,
and volatilization made it difficult to
predict the movement and fate of the
spill. Figure 3 shows the fluorometer
reading for Jan. 13,1988 (day 12). As can
be seen, the peak of the spill had passed
Wheeling, W.Va. (mHe point 87). The tail
of the spill still showed measurable fluo-
rometric readings all the way into the
Monongahela.
Figure 4 shows a series of fluorometric
peaks from January 16 through January
21. Several effects can be seen. The
height of the peaks reduces fairly rapidly
between January 16 and January 21.
This is no doubt a result of several
factors, including dilution as a result of
increased stream flow, volatilization, and
the confluence of several tributaries.
Another interesting phenomenon is the
rapid increase in velocity of the wave
front resulting from increased stream
flows. Figure 5 shows the concentration
of the spill plume as it passed Louisville,
Ky., and shows that the fluorescent
measurements had almost returned to
background levels.
There was a great deal of interest in
predicting; the propagation of the spill
plume in i:he Ohio River. One approach
was based on the following equation for
peak concentration:3-5
Peak = 0.74 + 0.04236
x Init x e - K x TOT (#• = o.93)
(1)
in which Peak = peak concentration of
contaminant (ng/L), Init = initial con-
centration of the spill (fig/L),K = decay
rate of the contaminant (L/d), and
TOT = time of travel of the contami-
nant to the utility in days. Equation 1
was developed from the results of exten-
sive computer simulation runs based on
an Ohio River Basin case study. The
equation is intended to model concentra-
tion within a stream reach and is based
on 296 runs of the model. Figure 6 shows
the predicted peak concentrations of the
spill as measured by fluorescence and
gas chromatography.
The peak concentration may vary sub-
stantially if a large tributary enters at
the beginning of a reach and is not
accounted for. Equation 1 also assumes
complete mixing of the contaminant in
the river. To use Eq 1, spill and flow
information, most of which may or may
not be supplied by a regional authority,
must be available to the utility manager.
Although there is little empirical infor-
mation regarding the fate of priority
pollutants as they travel downstream, if
the pollutant is known, it should be
possible to know whether the pollutant
is conservative, highly volatile, or subject
to some other process affecting its disap-
pearance rate. With this information the
manager could perform several calcu-
lations incorporating a range of disap-
pearance rates. In this way the utility
manager would have ageneral idea as to
when the spill will reach the utility and
whether the level of concentration will
require extra water treatment or closure
of the intakes. Although the equation
was derived from QUAL-II simulations
for the Ohio River case-study area,
similar analyses could be applied to
other river basins.
Peak concentration is not the only
measurement of interest; spill duration
is important as well. A slow-moving
spill, although it may be of lower peak
height and lower average concentration
than a more intense spill that passes
quickly, may pose a major problem to a
utility with limited storage capability.
Damage from the spill. At this time the
ultimate damage of the spill is not known.
Many water utilities were forced to alter
their treatment and to close intakes as
the spill passed. Many communities used
bottled water as an emergency measure.
Some obvious damage, such as the
staining of concrete walls, occurred when
ROBERT M. CLARK ET AL 43
-------
the diesel oil passed through the locks
and dams on the upper Ohio. One of the
hardest hit water utilities was in Steu-
benville, Ohio. Icing conditions caused
the spill to virtually halt over Steuben-
ville's water intakes. The Ohio National
Guard and two breweries delivered water
to the residents. Commercial activity
came to a virtual standstill.
A number of downstream water utili-
ties have billed Ashland Oil for expenses
incurred in dealing with the spill. Ohio
and Pennsylvania officials estimated that
10,000 fish and 2,000 ducks were killed.
Dead fish were found floating in the
upper Ohio after the spill passed.
Ashland Oil Inc. has agreed to a long-
term cleanup program and will reim-
burse the federal government $680,000
in connection with the spill, based on the
terms of a proposed consent decree filed
in the US District Court in Pittsburgh.
The settlement requires Ashland Oil to
clean up water and soil contamination.
Ashland Oil will also be required to
reimburse the government for its costs
in carrying out emergency response work
related to the initial cleanup efforts and
for any future cleanup costs. Under the
terms of the proposed settlement, the
company is required to remove contam-
ination from the soil or to excavate and
dispose of all soil that is fouled. In
addition, Ashland Oil must build and
operate a system to pump and treat all
contaminated groundwater to bring it
up to federal drinking water standards.
Ashland received about 4,000 other
claims related to the spill. These claims,
if all were determined to be valid, would
total approximately $17 million.
Shortly after the spill, definition of the
overall extent of damage to the Ohio
River began according to the Natural
Resource Damage Assessment (NRDA)
procedure. This is a process made avail-
able by the Comprehensive Environ-
mental Response, Compensation and
Liability Act.6 In addition, five technical
committees, under the auspices of the
US Department of Interior, were estab-
lished- to study various effects of the
damage that occurred during the spill.
The five workgroups were aquatic life-
fish tissue residues, sediments, water
users, terrestrial life, and water quality.
The workgroups are no longer active
because Ashland Oil has been reim-
bursing various states and agencies for
their costs.
Institutional considerations
In general, the institutions designated
to monitor and respond to the conditions
of a spill worked relatively well. The
state and federal governments and
ORSANCO coordinated these activities
during the crisis. One issue that was
raised as a result of the spill is a need to
develop aboveground storage tank legis-
lation requiring inspection and certifica-
tion as to tank integrity.
Research needs
The spill clearly illustrated the need
to have better information regarding the
time of passage versus discharge levels
for various stages of Ohio River flow.
Computer models should be developed
that can better predict both travel time
and concentration of contaminants. Im-
proved ability to predict the arrival of a
spill and to estimate the fate of a contam-
inant as it proceeds downstream would
have immeasurable benefits for water
resource managers. Spills and other
emergencies require that public agencies
be able to estimate the consequences
quickly and use computer simulations of
the events as they are unfolding. The
prediction of the arrival and departure
times of a spill such as the one that
occurred on the Monongahela requires
some type of computerized model, even if
the spill is tracked en route. Modeling
also can be used to estimate the fate of
the spill as it travels downstream and
can be used to perform postspill assess-
ments on the regions where most of the
contaminant may have entered sedi-
ments or food chains.
These computer models should be
interactive. Unfortunately, many models
are available and it may be difficult to
select the appropriate ones. The USEPA
has done some research on computer
models both on the Ohio River and the
lower Mississippi. Table 2 lists some of
these models and their characteristics.7
Other needed research relates to such
items as a better classification of the
mixing and volatilization characteristics
of the river and techniques for estimating
sediment adsorption. There is a need for
better damage assessment procedures,
for information on treatment techniques,
and for removing contaminants from
source water. As illustrated by this spill,
research needs to be conducted to develop
emergency response plans for utilities
and other involved agencies.
Summary and conclusions
A number of technical, legislative, and
administrative issues were raised during
the Monongahela spill. Both long- and
short-term environmental damage as-
sessments need to be made. These in-
clude quantifying damage to fish and
wildlife and contamination of sediments
and possibly groundwater. Correlation
of fluorometric readings and gas chro-
matography-mass spectrometry data
should also be undertaken. Legislative
changes are being studied to assess the
strengths and weaknesses of regulations
regarding oil tanks. Spill response, coor-
dination, and communication procedures
are being reviewed by ORSANCO.
Another issue identified during the spill
is the need for an on-line water quality
and quantity model to estimate travel
times and concentrations not only for
spills, but also for planning purposes in
developing National Pollutant Discharge
Elimination System permits. A great
deal of information will have to be gen-
erated to calibrate a sophisticated model
for the Ohio River. The river is difficult
to model hydraulically because of nu-
merous dams and tributaries and barge
traffic. Data on volatilization, dispersion,
and emulsion are also critical parameters
that need clarification.
References
1. LINGG, R.D. ETAL. Quantitative Analysis
of Volatile Organic Compounds by GC-
US.Jour. A WWA, 69:11:605 (Nov. 1977).
2. GOODRICH, J.A.; CLARK, R.M.; GRAYMAN,
W.M. Toxic Screening Model for Drinking
Water Utility Management. Proc. AWWA
1987 Ann. Conf., Kansas City, Mo.
3. CLARK, R.M.; GRAYMAN, W.M.; & GOOD-
RICH, J.A. Toxic Screening Models for
Water Supply. Jour. Water Resources
Planning and Management Div.—ASCE,
112:2:149 (Apr. 1986).
4. ORSANCO 1986 Ann. Rept. (Mar. 1987).
5. GOODRICH, J.A. & CLARK, R. M. Predicting
Toxic Waste Concentrations in Commu-
nity Drinking Water Supplies: Analysis
of Vulnerability to Upstream Industrial
Discharges. USEPA Munic. Envir. Res.
Lab., EPA-600/52-84-112, Cincinnati,
Ohio (Sept. 1984).
6. US Dept. of the Interior, Natural Resource
Damage Assessments: Final Rule. Fed.
Reg.,Part III, 43 CFRPart II, 51:148:27674
(Aug. 1,1986).
7. GRAYMAN,' W.M. ET AL. Surface Water
Screening Model—A Case Study for
Water Utility Management. AWWA Res.
Fdn., Denver, Colo. (July 1986).
: About the authors:
Robert M. Clark is
director of the Drink-
ing Water Research
Division (DWRD), US
Environmental Pro-
tection Agency, 26 W.
I Martin Luther King
I Dr., Cincinnati, OH
45268. Clark has been a US Public Health
Service officer since 1961, detailed to the
USEPA in 1970. He has been director of
the DWRD since 1985 and is a member of
A WWA, ASCE, and AAEE. Clark's work
has been published by JOURNAL AWWA,
Journal Environmental Engineering
Division—ASCE, Journal Water Re-
sources Planning and Management Div-
ision—ASCE, and Environmental Sci-
ence and Technology. Alan H. Vicory is
executive director and chief engineer for
ORSANCO in Cincinnati, Ohio. James
A. Goodrich is a physical scientist with the
DWRD in Cincinnati.
44 MANAGEMENT AND OPERATIONS
JOURNAL AWWA
-------
-------
EPA/600/J-90/002
JOURNAL AMERICAN WATER WORKS ASSN.
VOL. 82 NO. 3, MARCH 1990
&U.S. GOVERNMENT PRINTING OFFICE:
1*90 - 7«I-I5*/2
------- |