Environ. Sci. Technol.

1984, 18, 834-840

Evaporation Rate of Spills of Hydrocarbons and Petroleum Mixtures

Warren Stlver and Donald Mackay"
Department of Chemical Engineerlng and Applied Chemistry, University of Toronto, Toronto, Ontario, M5S 1A4 Canada

evaporated, as used by Mackay et al. (9). In the pseudo-

Procedures are described for quantifying the rate of component approach, a number of real or pseudocompo-
environmentalevaporation of liquid mixtures such as crude nents are selected and their relative quantities estimated
oils and petroleum products under a variety of environ- such that they reproduce the oil's evaporation or distilla-
mental conditions. Equations are derived from mass
transfer theory, which relate the fraction evaporated to an tion characteristics or gas chromatographic analysis. When
evaporative exposure and Henry's law constant. The di- the pseudocomponent approach is used to calculate
mensionless evaporative exposure characterizes the expo- evaporation rates, it is essential to compute and store the
sure conditions of atmospheric turbulence, spill volume, change in composition. A pseudocomponent or actual-
and area. The dimensionless Henry's law constant is a component approach may be required if information is
function of the liquid composition and temperature and required on the evaporation of specific components, for
of the fraction of liquid which has evaporated. Experi- example, for toxicity or forensic purposes. The analytical
mental procedures are described for measuring evaporation approach, on the other hand, permits easier calculations;
characteristics of a liquid mixture and for producing sam- however, this may result in less accuracy. This work in-
ples of partially evaporated material. vestigates the analytical approach to evaporation. It does
not treat any chemical conversions or the formation of
water-in-oil emulsions.
Introduction I t is first useful to review the experimental methods
When crude oils are spilled on land or water, evaporation which can be used to obtain information about volatility.
is often a significant process of mass loss from the spill.
Thus, a knowledge of the evaporation rate can be useful Experimental Configurations
in several respects. For example, the evaporation rate The evaporation characteristics for a mixture may be
controls the atmospheric vapor concentration and hence obtained under two limiting experimental conditions,
the threat of toxic exposure to those in the vicinity. isothermal, nonisobaric conditions and isobaric, noni-
Evaporation may modify the physical, chemical, and tox- sothermal conditions. Figure 1illustrates the configura-
icological properties of the liquid, notably density, vis- tions.
cosity, and the fraction of lower molecular weight sub- In isothermal systems, there are two limiting cases de-
stances. pending on how the oil is exposed to evaporation and how
There have been several reports describing the evapo- the evaporated vapors are removed. Surface or tray
ration behavior of spills. Reviews have been compiled by, evaporation in, for example, a constant temperature wind
for example, NAS (I),Malins (2), and Jordan and Payne tunnel is controlled by the mass transfer rate limitation
(3),and attempts have been made to develop expressions at the liquid surface, and the air leaving the system is far
quantifying evaporation rates, notably those of Yang and from saturated with oil vapor. A typical evaporation rate
Wang (4), Butler (5), Sivadier and Mikolaj (6), Mackay and curve is illustrated in Figure 1. An alternative is equilib-
Leinonen (7), Mackay and Matsugu (a), Mackay et al. (9, rium control in which the liquid is contacted with a slow
IO), Reijnhart and Rose (II), Payne et al. (I2),Drivas (13), flow of air such that the air leaving the system is fully
and Feigley (14). None of these approaches is completely saturated with oil vapor by, for example, bubbling air
satisfactory in enabling evaporation rates to be calculated through a column of the oil. The evaporation rate is then
from easily measured experimental properties and being controlled entirely by the flow rate of air through the
applicable to a wide variety of oils and environmental system, and mass transfer rates are unimportant. A curve
conditions. similar to Figure 1 is obtained.
This paper reviews the thermodynamic and kinetic pa- An example of an isobaric system is the familiar boiling
rameters which influence oil evaporation rates. Experi- point vs. fraction distilled curve shown in Figure 1. As the
mental procedures are presented which can be used to mixture evaporates, the temperature rises to the "bubble
determine evaporation properties, thus providing a basis point" temperature at which it exerts a constant vapor
for calculating evaporation rates under a variety of envi- pressure, normally atmospheric pressure. Kinetic con-
ronmental conditions. The work starts with the simple siderations are irrelevant in this system since the rate of
case of describing the evaporation characteristics of a pure composition change is determined by the rate of heat
liquid and then extending the description to include supply, and the temperatures achieved are dependent on
evaporation of multicomponent systems. In doing so, the composition and independent of time.
concept of a dimensionless "evaporative expcsure" is in- A primary objective of this work is to establish theo-
troduced as a criterion for determining extent of evapo- retical links between the data obtained from the distilla-
ration. tion, tray evaporation, and air bubbling configurations and
The difficulty with multicomponent systems is primarily show how data can be used to calculate environmental
that of expressing the liquids vapor pressure as a function evaporation rates.
of its changing composition. As a liquid evaporates, the
more volatile materials are lost preferentially and the Theory
mixture's total vapor pressure falls. There are two ap- Surface Evaporation. If a liquid, of vapor pressure
proaches for calculating the changing vapor pressure: the P (Pa), is spilled over an area of a (m2),the rate of evap-
pseudocomponent approach used by Yang and Wang ( 4 ) , oration is given by
Mackay and Leinonen (3, Reijnhart and Rose (II), and N = kaP/(RT) (1)
Payne et al. (12) or an analytical expression in which the
vapor pressure is expressed as a function of fraction where N is the molar flux (mol/s), k is the mass transfer
834 Envlron. Sci. Technol., Vol. 18, No. 11, 1984 0013-936X/84/0918-0834$01.50/0 0 1984 American Chemical Society
 it is possible that there is a contributing liquid-phase re-
/ I t -7 sistance, especially if the substance has a high air-liquid
partition coefficient (i.e., a high vapor pressure) or if the
liquid is viscous. We assume here that the air-phase re-
sistance dominates.
Stripping. The second approach is to use a gas strip-
ping technique with an exit gas rate G (m3/s). If the exit
gas is saturated, the evaporation rate will be GP/(RT)
(molls) and
dF,/dt = [GP/(RT)I(u/Vo) (6)
or
I 0Temperature Indicator I I dF, = HdO (7)
The evaporative exposure 0 is now defined as Gt/ Vo and
is the actual ratio of vapor volume to liquid volume.
The identical nature of eq 4 and 7 suggest that if surface
Heat
(tray) and stripping experimental data are plotted as F,
Figure 1. Schematic diagrams of evaporation configuratlons: (top) vs. 8, the points should lie on a common line, 0 being de-
by tray evaporation; (middle) by gas stripplng; (bottom) by batch dis- fined either as kat/ Vo or Gt/ V,. This is later verified
tillatlon. experimentally.
Distillation. In principle, if enthalpy of vaporization
coefficient under the prevailing wind conditions (m/s), R (A%)data are available, it is possible to calculate a pure
is the gas constant [8.314 Pa*m3/(mol.K)],and T is the component vapor pressure (at a lower temperature) from
environmental temperature (K). Equation 1 can be ar- a boiling point. For mixtures, composition rigorous cal-
ranged to give culation requires a knowledge of the changing values for
dF,/dt = kaPu/(V&T) (2) A7f for each component. A plot of boiling point, TB, as
a function of F, could then be converted into a plot of
where F, is the volume fraction evaporated, t is time (s), vapor pressure, P, as a function of F,. This would permit
u is the liquid’s molar volume (m3/mol), and Vo is the the calculation of Henry’s law constant, H, as a function
initial volume of spilled liquid (m3). Rearranging gives of F, and hence the definition of the evaporation curve,
dFv = [Pu/ (RT)1 (kadt / Vo) (3) F, vs. 0. Unfortunately few A% values are available, es-
pecially for mixtures as complex BS crude oils. In addition
or there is doubt about a changing molar volume u in the
dF, = H-dO (4) calculation of H. These difficulties make it impossible, at
present, to calculate a rigorous relationship between H and
The right-hand side of eq 3 has been separated into two TB for crude oils. However, it is impossible to use the
dimensionless groups. The group kat/Vo is termed the Clapeyron Clausius equation and Trouton’s rule to suggest
“evaporative exposure” and is denoted as 0. The evapo- a suitable empirical expression as follows:
rative exposure is a function of time, the spill area and
volume (or thickness), and the mass transfer coefficient In (P/PA) = B ( 1 - TB/T) (8)
(which is dependent on the wind speed). The evaporative where PA is atmospheric pressure (101 325 Pa) at which
exposure can be viewed as the ratio of exposed vapor the boiling point is measured, P is the lower vapor pressure
volume to the initial liquid volume. at temperature T ( K ) , and B is a dimensionless term
The group Pu/(RT) or H is a dimensionlessHenry’s law equivalent to A7f/(RTB)and, from Trouton’s rule, with
constant or ratio of the equilibrium concentration of the a fairly constant value of approximately 10.6.
substance in the vapor phase [P/(RT)]to that in the liquid It follows that H [which equals Pu/(RT)]can be ex-
( l / u ) . It is a function of temperature but not of other pressed as
environmental conditions. In H = In (PAu/(RT)]+ B - BTBT
The product OH is thus the ratio of the amount which (9)
has evaporated (oil concentration in vapor times vapor In H = A - BTB/T (10)
volume) to the amount originally present.
If the liquid is pure, H is independent of F,, and eq 4 where A and B are dimensionless constants which can be
can be integrated directly to give obtained by fitting experimental data. It is emphasized
that this is only a suggested empirical equation; it is not
F, = HO (5) rigorously derived for mixtures.
If k , a, and temperature are constant, the evaporation Now, the distillation data yield an equation relating F,
rate is constant and evaporation is complete (F, is unity) to TB which in its simplest form could be
when 0 achieves a value of 1/H. TB = To + TJ;, (11)
If the liquid is a mixture, H depends on F,, and eq 4 can
only be integrated if H is expressed as a function of F,; i.e., where Tois the initial boiling point at F, of zero and TG
the principal variable of vapor pressure is expressed as a is the gradient of the TB, F, line. Other more complex
function of composition. The evaporation rate slows as equations could be fitted, but only this simple equation
evaporation proceeds in such cases. is considered here. Combining eq 7, 10, and 11 gives
If the liquid is pure, the resistance to mass transfer must dF, = exp[A - B(To+ TGF,)/T]d0 (12)
lie entirely in the air phase because there is no necessity
for the substance to diffuse in the liquid phase to the which can be integrated to give
interface. The mass transfer coefficient k is then entirely F, = 1n [1 + B(TG/T)0 exp(A - BTo/T)I[T/(BTG)I
an air-phase resistance term. If the liquid is a mixture, (13)
Environ. Sci. Technot., Vol. 18, No. 11, 1984 835
This equation can be used to calculate the F, vs. 0 curve +,AIR.. . IN
.. . -_
from the distillation curve, if estimates of A and B are
available.
Equations 9-13 can also be written in an equivalent but
simpler form by redefining the two adjustable parameters SORBENT -._
A and B in terms of two new parameters, Hoand Q, where 1

In Ho = A - B T o / T (14) GAS TRAP

INTEGRATING FLOW METER

Q = BTG/T (15)
It follows that eq 10 and 11 yield GRADUATED CYLINDER

In H = In Ho- QF, (16) Flgure 2. Gas stripping apparatus.

Thus, the inherent assumption is that a plot of In H is Samples (0.5 pL) were analyzed by using a Hewlett-
linear in F,, extending from an initial (fresh oil) value of Packard Model 700 gas chromatograph. The column was
Ho to lower values as F, increases with a slope of -&. For 3 m long, 0.125 in. 0.d. stainless steel, coated with 10%
a pure substance Q is zero. SE-30 ultraphase on chromosorb P, A/W, DMCS, mesh
Equation 12 then becomes 60/80. The injection port temperature was 300 "C. The
#, = exp(Ho - QF,)d0 (17) oven started at 50 "C and was temperature programmed
at 7.5 "C/min to a maximum temperature of 280 "C. A
which when integrated gives the equivalent to eq 13, Shimadzu Model C-RIA integrating recorder was used.
namely Tray Evaporation. A volume of oil was weighed and
F, = In (1 + HoQ6)/Q (18) placed onto a tray (area = 0.082 m2) to obtain a slick
thickness of 1-5 mm. The tray was then placed into a wind
In summary, the key characterization of a liquid's tunnel and subjected to wind speeds between 4 and 12
evaporative behavior is a plot of F, vs. 0, different lines m/s. The mass transfer coefficient for the various wind
applying at different temperatures because of the depen- speeds was measured separately by toluene evaporation.
dence of P, and hence H, on temperature. These curves The mass of the toluene or oil was measured as a function
may be obtained from wind tunnel or gas stripping ex- of time by using an in situ top loading balance. The length
periments or (if A and B are known) from the boiling point of the experiments varied from 6 h to 2 weeks. During the
curve. If environmental estimates are available for the synthetic oil runs the samples taken were used for density
mass transfer coefficient k and the spill initial thickness measurement by pycnometer as well as for GC analysis.
(Vo/a),it is then possible to calculate 0 at any time and Thin-FilmTray Evaporation. A small volume (1 mL)
thus estimate F,. If the spill thickness or k change with of weathered oil was pipetted onto a small circular tray
time, it is necessary to calculate F, by analytical or nu- (area = 64 cm2). This oil was then spread over the tray
merical integration. For oil spill models in which other by using a spatula to obtain a uniform thin film. The
processes are occurring simultaneously, it is normally spatula was weighed before and after to account for the
preferable to use a finite difference form of eq 12, namely oil sticking to its surface. The tray was then clamped into
the wind tunnel and subjected to wind speeds of between
AF, = exp[A - B(To+ T$,)/rl(kaAt/V0) (19) 4 and 12 m/s. The mass of the tray was measured as a
AF, = HkaAt/Vo (20) function of time, but in this case a separate analytical
balance was used. The runs lasted approximately 7 days.
It may be desirable to obtain an equation to fit the Gas Stripping. The apparatus is shown in Figure 2.
experimentally determined F,-0 evaporation curve. The Either a 250-mL or a 1-L graduated cylinder was used.
form of this curve fit is suggested by the distillation ex- The gas trap was loosely packed with 3 M sorbent and
pression (eq 13): placed in an ice bath. A integrating wet test meter was
F, = (T/Kl) In (1 + KlO/T) exp(K2 - K d T ) (21) used to measure the cumulative air flow. For experiments
at other than room temperature a water bath ( i 2 "C) was
A value for K1 can be obtained from the slope of the F, used for both the graduated cylinder and the air feed.
vs. log 0 curve. For 0 greater than lo4,Kl is approximately The graduated cylinder was filled with the oil. The mass
2.3T divided by the slope. The expression exp(K2- K3/!19 of the cylinder before and after and the volume of the oil
can then be calculated, and K2 and K3 may be determined were recorded. Air was then bubbled through the oil
individually from evaporation curves at two different taking care to prevent liquid entrainment. Periodically,
temperatures. the air flow stopped, and measurements were made of the
mass and volume of oil as well as the volume of air passed
Experimental Procedures through the oil. Before the volume of oil was recorded,
The evaporation characteristics of five crude oils and sufficient time was taken to allow all the air bubbles to
one synthetic oil were studied. The crude oils used were escape from the oil. The evaporation curve was then ob-
Norman Wells, La Rosa, Murban, Lago Medio, and Pru- tained on both a mass and volume fraction basis.
dhoe Bay. The synthetic oil was a mixture of n-alkanes, Distillation. The distillation apparatus consisted of
the mole fraction composition being the following: pen- standard laboratory equipment. A known volume (usually
tane, 0.130; hexane, 0.087; heptane, 0.104; octane, 0.071; between 150 and 200 mL) was poured into the 500-mL
decane, 0.084; undecane, 0.045; dodecane, 0.086; tridecane, flat-bottom distilling flask. Small (1-2-mm diameter) glass
0.042; tetradecane, 0.108, hexadecane, 0.107; octadecane, beads were added to reduce bumping. The oil was stirred
0.048; eicosane, 0.050. with a Teflon-coated magnetic bar on a hot-plate stirrer.
The procedures for weathering the crude and synthetic The flask was partly insulated as shown to minimize heat
oils were the same; however, during the synthetic oil runs losses and give more consistent results. The oil was heated
frequent samples were taken for analysis by gas chroma- slowly, with the heat input increased slowly to give a
tography. uniform boiling rate in the flask. The condensate was

898 Environ. Sci. Technol., Vol. 18, No. 11, 1984

 250
e

NUMBER OF

STAGES

W
I ,”/ ’ o EXPERIMENTAL DATA
0 2003 5000 10003 lo5 lo“ 10’
EVAPORATIVE EXPOSURE e
Flgure 3. Plot of volume fraction evaporated vs. evaporative exposure
for the synthetic oil.

collected in 10-mL graduated cylinders (gradations 0.1

mL). The liquid temperature was measured by using a
thermocouple. The volume distilled was recorded as a
,ooy’’1 1
008 0.24 0.40 056 0 72
function of liquid boiling temperature. FRACTION DISTILLED Fy

Results and Discussion Figure 4. Distillation curves for the synthetic oil showing computed
curves assuming 1.0, 1.5, and 2.0 theoretical stages.
Wind Tunnel Mass Transfer Coefficient. The mass
transfer coefficient for the wind tunnel experiments was the same compositional changes. In addition, the exper-
measured by the evaporation of toluene. The experimental imental results are compared with those which can be
procedure was the same as that for the synthetic and crude computed from reported vapor pressures (16).
oils. The mass transfer coefficient k was calculated from The tray evaporation results are shown in Figure 3 as
eq 1 by using literature values for the vapor pressure of a plot of volume fraction evaporated (F,)vs. evaporative
toluene. The resulting mass transfer coefficient charac- exposure (e). Note the linear and logarithmic scales. The
terizes the gas-phase resistance only since a pure liquid evaporative exposure was calculated from the time of ex-
has been used. It is well established that k depends also posure based on a mass transfer coefficient of 1.4 X
on the Schmidt number of the evaporating component. No m/s and an initial slick thickness (Vola)of 2.5 mm.
correction was made here because its magnitude is not Measurements in the tray evaporation experiments are on
large and because an oil is made up of many components; a mass basis. From the initial and final densities (po and
thus, there is doubt about the appropriate Schmidt num- pF) the mass fraction F, can be converted to a volume
ber. A further complication is that the k value for mixtures fraction Fv by using
may be reduced by the presence of a liquid-phase resist-
ance corresponding to the diffusion of the evaporating 1- Fv = (1 - F , ) ( P ~ / P F ) (22)
component from the bulk of the liquid to the interface. The gas stripping results and the computed evaporation
Such diffusion implies a concentration gradient; thus, the curve based on vapor pressures from Zwolinski (16)are also
bulk and interfacial concentrations may differ. This effect shown in Figure 3. The coincidence of the curves verifies
is believed to be negligible in most cases with three ex- the evaporationrates as measured by the two experimental
ceptions. If the oil layer is deep, viscous, and poorly mixed, techniques when expressed in terms of 8. Error bar esti-
there may be appreciable concentration gradients. In some mates are given for the gas stripping and tray evaporation
cases a “skin” of evaporation residue (probably consisting curves. The gas stripping technique is believed to be more
of asphaltenes and waxes) may form at the surface and accurate.
block subsequent evaporation. Components of high vol- The gas flow rate (G)in the gas stripping experiments
atility (e.g., pentane) have high air-oil partition coeffi- was 1.3 X m3/s (800 mL/min), and the initial oil
cients; thus, the air resistance is very low, and the normally volume (Vo)was 240 mL, giving a rate of exposure (G/ Vo)
small liquid resistance may become significant. This will or de/dt of 3.2 per min. The rate of exposure for the tray
tend to occur early in the evaporation process when evaporation technique (ka/Vo)was 340 per min. The
evaporation is very rapid, and the effect becomes less slower rate of gas stripping exposure is one reason for its
significant a t later times which are of primary environ- greater accuracy. Interestingly, under typical environ-
mental interest. In general, the approach adopted here mental oil spill situations the rate of exposure is approx-
is valid only if the oil-phase “mixing time” is very much imately 600 per min or 1 million per day (k of 0.01 m/s,
shorter than the evaporation time. Several experimental Vola of 1 mm). Thus, to achieve a wide time range of
runs were performed, which give a relation for the mass evaporation data, a three regime approach is useful. For
transfer coefficient as a function of the air velocity as 0 from 0 to 50 OOO, gas stripping is ideal. Tray evaporation
measured by a pitot tube in the wind tunnel. Relations
are available in the literature for mass transfer coefficients
in environmental situations, e.g., Mackay and Matsugu (8)
-
is suitable for 6 from 50 000 to lo7. For exposures greater
than lo7 a thin-film (Vola 0.1 mm) technique is best.
The distillation curve for the synthetic oil is given in
and Mackay and Yuen (25). Figure 4. A correction was applied for the volume of
Synthetic Oil. The synthetic oil was used to test the uncondensed vapor in the system. The theoretical batch
equivalence of the three evaporation techniques. Equiv- distillation curve, assuming one theoretical stage separa-
alence implies that the experimental techniques give the tion, is also shown. A considerable temperature difference
same evaporation rates on a evaporative exposure basis and of some 20 “C exists between the theoretical and experi-
Environ. Scl. Technol., Vol. 18, No. 11, 1984 837
 o EXPERIMENTAL
o EX PER1MENTAL
corrected for loa volume

’i\ I

2000 5000 10000 io5 1o6 10’

EVAPORATIVE EXPOSURE e
Flgure 7. Evaporation curves for Norman Wells crude oil at 22 OC.

0.14 0.20 0.30 0.4 0

VOLUME FRACTION EVAPORATED F,
-\
o
a50
t LL’
0
Q30 t ’
,
I
I /
Flgure 5. Experlmental and computed fractlons of ndecane as a
function of F , for the distliiatlon technique. 2 azo
W

100 “ ,
L o ;GASSTRIPPING

t
/TRAY EVAPORATION
E 0.10 - /DISTILLATION PREDlCTlOh
” w I&

”I
\A
801 I I
J
Id

O’ 2000 5600 10000 lo5

W
a .
V
0
I
GAS STRIPPING
TRAY EVAPORATION
EVAPORATIVE EXWSURE 8
Flgure 8. Evaporation curves for Kuwalt crude oil at 22 OC.
W
n A DISTILLATION
40. n-decane remaining vs. the fraction distilled. The com-
+
z
W
COMPUTED\ A puted results based on 1.0, 1.5, and 2.0 theoretical stage
v
a
. separations are plotted along with the experimentalresults.
t The experimental compositions indicate a 1.8 equilibrium
20. stage separation.
Compositional comparison of the three experimental
techniques is shown in Figure 6. It is evident that the tray
0. evaporation and gas stripping techniques are giving the
same compositionalchanges but that the distillation gives
different compositions. The effect of the apparent 2-stage
separation and the higher temperature in the distillation
experiments will alter the relative volatilization rates of
the various classes of components present in a crude oil.
The distillation technique is believed to be less accurate
for predicting evaporation rates than stripping or tray
evaporation, but it can provide the data rapidly.
The gas stripping and tray evaporation techniques are
preferred for producing “typical” samples of weathered oil,
but distillation may be used if samples are required rap-
idly.
Crude Oils. The tray and gas stripping evaporation
curves for Norman Wells and Kuwait crude oils are shown
in Figures 7 and 8. The tray evaporation curve lags
slightly behind the gas stripping curve for the Kuwait
crude, as was the case for the other three oils studied. It
is possible that this is caused by a liquid-phase resistance.
For example, Yang and Wang ( 4 ) observed the formation
of a thin waxy film on the oil surface, significantly reducing
their measured evaporation rates. Other factors which may
be contributing to the lag are evaporative cooling or a
diffusivity (Schmidt Number) effect on the mass transfer
coefficient. In any event the lag is small, and it is con-
cluded that the tray and stripping techniques also give
 0
0

0
0
0 o GAS STRIPPING
c
a TRAY EVAPORATION
h
x THIN-FILM TRAY EVAPORATION

o NORMAN WELLS
x KUWAIT
I , . . . !
0 4003 10000 io5 1o6 10' It
EVAPORATIVE EXPOSURE e
0
100; Flgure ll. Extended evaporation of Murban crude oil at 22 "C.
0.10 a20 a30
VOLUME FRACTION DISTILLED 5
Flgure 9. Atmospheric pressure distillation curves for Kuwait and
Norman Wells crude oil.
LL'
-61 0.30.
x L A ROSA W / o.
o KUWAIT 5
A MURBAN 8 /"
0 NORMAN WE1.LS 4
LAG0 MEDIO
z azo. ,/
-10-
EXPERIMENTAL 22'C
x
I a O'C
0 EXPERIMENTAL
-c L L X -DISTILLATION
-12- PREDICTION O'C
I n H = 6.3 - 103TB/T A**

1 . I
0.04//
4 00 450 500 550
BOILING TEMPERATURE T, ( K )
600
Y 10000 20000 30 000
Figure 10. Plot of natural log of H a t 293 K vs. boiling temperature EVAPORATIVE EXPOSURE e
(K). Figure 12. Evaporation curves of Norman Wells crude oil at 0 and 22
"C.
consistent results for crude oils. Ignoring liquid-phase
resistance may result in errors for highly viscous oils. signed to elucidate quickly the extent of evaporation that
The distillation curves for the two crude oils are given may be reached. Figure 11 shows the evaporation curve
in Figure 9. To convert these distillation curves into on a mass basis for Murban crude oil by using the thin-film
evaporation curves, it is necessary to determine the con- technique. The exposure of 108 was attained in just 4 days.
stants A and B for eq 10. From the Fv4eurves, the mean The significant effect of temperature on the evaporation
slopes (H)were measured at selected F, points. The rates of crude oils is illustrated in Figure 12 which gives
corresponding boiling points were read off the distillation the evaporation curves of Norman Wells at 0 and 22 "C.
curves. Figure 10 was then compiled of In H vs. TB,a linear The extent of evaporation of the 0 "C Norman Wells oil
relationship being expected (eq 10). The correlation a t 0 of lo4 is only 70% of that for the evaporation at 22
coefficient was 0.96 for the linear regression, fitting A and "C.Equation 13 applies and predicts the evaporation rate
B to give at 0 "C accurately as shown in Figure 12.
In H = 6.3 - 10.3TB/T (23) Conclusions and Recommendations
These values of A and B were then used to predict the The theoretical background for quantifying evaporation
FV-0 curve for all the crude oils from their boiling point rates of crude oils has been developed and related to three
curves (i.e., To and TG)by using eq 13. The predicted experimental techniques, tray evaporation, gas stripping,
evaporation curves for Norman Wells and Kuwait are and distillation. The concept of a dimensionless evapo-
shown in Figures 7 and 8. In all cases the extent of rative exposure has been devised. The evaporative expo-
evaporation, at a 0 of 10OO0, was predicted correctly within sures provide characterization of environmental or labo-
4%. As shown in Figure 10 the correlation is based only ratory exposure conditions. It incorporates the oils' area
up to a boiling point of 550 K (280 "C). Caution must be and volume, the prevailing wind speed, and the time of
used when fractions evaporated are predicted that boil exposure. Experimental procedures to obtain fraction
above 280 "C. It is believed that the correlation tends to evaporated as a function of evaporative exposure have been
overpredict the rate of evaporation in this region. discussed. A synthetic oil was used to verify the evapo-
Some oil spills are exposed to very rapid evaporation rative behavior of these experiments. Crude oils were
(high exposure 0 caused by high winds and turbulence (i.e., studied, and a correlation was developed to predict the oil's
high 12) or by the oil being spread out very thinly (i.e., low evaporation curve from its distillation curve. The gas
Vo/a).In these situations it is useful to know the extent stripping and tray evaporation techniques are ideal for
of evaporation that may be attained. The thin-film tech- measuring the evaporation rates of crude oils and for
nique described under Experimental Procedures was de- producing weathered samples of crude oil. The distillation
Environ. Sci. Technol., Vol. 18, No. 11, 1984 839
 Environ. Sci. Technol. 1084, 18, 840-845

technique can be used for the prediction of the evaporation ( 7 ) Mackay, D.; Leinonen, P. J. “Mathematical Model of the
curve only; in general it should not be used for producing Behaviour of Oil Spillson Water with Natural and Chemical
samples. Dispersion”, prepared for Fisheries and Environmental
It is recommended that several Arctic oils be subjected Canada, 1977, Economic and Technical Review Report
to evaporative behavior determination as described in this EPS-3-EC-77-19.
report. I t may also be useful to extend this work to (8) Mackay, D.; Matsugu, R. S. Can. J. Chem. Eng. 1973,51,
434-440.
evaporation of chemical spills. (9) Mackay, D.; Paterson, S.; Nadeau, S. Natl. Conf. Control
Hazard. Mater. Spills 1980, 361-368.
Literature Cited (10) Mackay,D.; Stiver,W.; Tebeau, P. A. F’roc.-Oil Spill Conf.
National Academy of Sciences “Petroleum in the Marine 1983,331-337.
Environment”;National Academy of Sciences: Washington, (11) Reijnhart, R.; Rose, R. Water Res. 1982, 16, 1319-1325.
DC, 1975. (12) Payne, J. R.; Kirstein, B. E.; McNabb, G. D.; Lambach, J.
Malins, D. C. ”Effects of Petroleum on Arctic and Subarctic L.; de Oliveira,C.; Jordan, R. E.; Ham, W. Proc.-Oil Spill
Marine Environments and Organisms”, Academic Press: Conf. 1983, 423-434.
New York, 1977; Vol. I. (13) Drivas, P. J. Environ. Sci. Technol. 1982, 16, 726-728.
Jordan, R. E.; Payne, J. R. “Fate and Weathering of Pe- (14) Feigley, C. E. Enuiron. Sci. Technol. 1983, 17, 311-312.
troleum Spills in the Marine Environment”; Ann Arbor (15) Mackay, D.; Yuen, A. Environ. Sci. Technol. 1983, 17,
Science: Ann Arbor, MI, 1980. 221-217.
Yang, W. C.; Wang, H. Water Res. 1977,11,879-887. (16) Zwolinski, B. J.; Wilhoit, R. C. API Publ. 1971, No. 101
Butler, J. N. In “Transfer of Petroleum Residues from Sea (44-TRC).
to Air: Evaporative Weathering. Marine Pollutant
Transfer”; Windom, H. L.; Duce, R. A., Eds.; Lexington
Books: Lexington, MA, 1976; pp 201-212. Received for review September 12,1983. Accepted May 21,1984.
Sivadier, H. 0.;Mikolaj, P. G. Symp. Prev. Control Oil Financial support was provided by the Arctic Marine Oilspill
Spills 1973, 475. Program of Environment Canada.

Aspects of the Polycyclic Aromatic Hydrocarbon Geochemistry of Recent

Sediments in the Georges Bank Region
Paul D. Boehm’
Batteile New England Marine Research Laboratory, Duxbury, Massachusetts 02332

John W. Farrington
Coastal Research Center and Chemistry Department, Woods Hole Oceanographic Institution,
Woods Hole, Massachusetts 02543

bined PAH sources (e.g., stormwater runoff, municipal

rn Polycyclic aromatic hydrocarbon (PAH) concentrations sewage effluents, harbor oil spillages, and industrial inputs)
and compositions were determined seasonally over 1year in estuarine systems; (iii) direct introduction of waste
at 22 sampling stations on and adjacent to Georges Bank, materials via pipeline or barge disposal; (iv) resuspension
off the Northeastern U.S. coast. PAH concentrations are
directly related to the total organic carbon (TOC) content of materials reaching coastal marine sediments via (i) and
of the sediments and to the slit/clay content of these (ii) and deposition in settling areas.
sediments. Several groupings of two-ringed to five-ringed The Georges Bank region of the continental shelf off the
PAH compounds were quantified, and concentrations of Northeastern United States is an area of high productivity
these groups ranged from 1 to 100 ppb, dry weight. The (11, 12) and an area of dynamic sediment transport
overall compositions of the PAH compounds indicate that mechanisms (13,14). It is also an area of interest to the
pyrogenic (combustion-derived) PAH dominate the as- offshore petroleum industry. This study focuses on de-
semblage at higher levels while lower levels are often as- fining the levels and generic sources of PAH compounds
sociated more with a fossil fuel PAH origin. The rela- in the sediments of Georges Bank and the areas imme-
tionship of PAH to TOC, the ratio of fossil fuel to com- diately adjacent to the bank. Such information is im-
bustion PAH, and the absolute concentrations of PAH portant to the development of diagnostic parameters for
form a useful set of monitoring parameters in these and use in marine monitoring programs to assess any change
other continental shelf regions. in the organic geochemistryof the recent sediments of the
region. We also present results on relatively small scale
Introduction spatial and temporal variations in PAH levels and com-
The ubiquity of polycyclic aromatic hydrocarbon (PAH) positions in continental shelf sediments for the first time.
residues in sediments and soils has been well established Aspects of the saturated hydrocarbon geochemistry of the
on a global basis. Previous reports of PAH distributions region have been presented elsewhere (15).
(1-10) have addressed aspects of the distribution of PAH
in marine sediments. The distribution of PAHs, in marine
Experimental Section
sediments is governed by four transport processes: (i) Sampling Methods. Samples of surface sediments
aeolian transport of fossil fuel and wood combustion from the region were obtained over a period of five years
products followed by deposition on the sea surface and (1977-1982). On several cruises, Smith-MacIntyre (0.1 m2)
subsequent sedimentation; (ii) riverine transport of com- grab samplers were used for all collections, and a 0-3-cm
840 Environ. Sci. Technol., Vol. 18, No. 11, 1984 0013-936X/84/0918-0840$01.50/0 0 1984 American Chemical Society