Water Res. Vok 16. pp. 1319 to 1325. 1982 0043-135~ 82 081319-07503.

00 0
Printed in Great Britain. All rights reserved Copyright © 1982 Pergamon Press Ltd

EVAPORATION OF CRUDE OIL AT SEA

R. REIJNHART and R. ROSE
Koninklijke/Shell-Laboratorium. Amsterdam. The Netherlands

(Receired February 1981)

Abstract--A simple model for describing the simultaneous evaporation and spreading of oil spills at sea
has been developed and tested. A previously reported theory was used for calculating the aerodynamic
part of the evaporation process. Here, the thermodynamic part, i.e. the oil characterization and the
weathering processes, is described in detail. By comparison with the results of laboratory evaporation
experiments and sea trials the model is shown to work quite adequately. Also, the influence of the
mixing of oil into the water column, e.g. due to wave action, is shown to be small enough to be
neglected. The spreading model was purposely kept as simple as possible, in view of the uncertain
influences of wave action and emulsion formation. The velocity of the edge of the oil slick is prescribed
and a critical thickness beyond which the slick stops spreading is introduced. The latter turns out to
have a significant influence on the rate of evaporation. Examples of crude evaporation are given and the
influence of temperature and wind velocity is investigated.

NOMENCLATURE formula. The procedure for taking into account

Co = equilibrium concentration of vapour, kg m- a changes in composition and the method of character-
d+r~, = critical slick thickness, m izing the oil will be discussed in Section II.
g = acceleration of gravity, m s- 2 In order to test the model, batch experiments were
Gz = mass of evaporating liquid, kg carried out in which various oils were allowed to
M~ = average molar weight of liquid, kg kmol- '
M~ = average molar weight of gas, kg kmol- L evaporate under well-defined conditions of tempera-
M, = average molar weight of source, kg kmol- ture, stirring, etc. In particular, we investigated
Pl = vapour pressure of component i, N m -2 (Section IIl) how the evaporation of oil changes when
p,o, = total vapour pressure, N m-2 it is thoroughly mixed with water, as happens in
Q,, -- rate of evaporation, kg s-
stormy weather at sea. In Section Ill, the results of
Q, = source strength, kg s-t
S= surface area of evaporating liquid, m-2 some theoretical crude oil evaporation calculations
~:o = average wind velocity at 10 m height, m s- L are compared with experimental data and the rela-
V,d~c = velocity of edge of slick, m s- 1 tively simple thermodynamic mixture model is shown
xi = mol fraction of component i in liquid to work satisfactorily. Spreading of the oil is taken
xi.+ = mol fraction of component i in source
y+ = mol fraction of component i in gas into account in a rather simplified model. However,
~t = constant defined in equation (12), m s- as will be argued in Section V, the model is expected
Z= constant of proportionality in equation (11). to describe the various processes rather well, in spite
of the fact that reality is far more complicated.
I. INTRODUCTION
It. CHARACTERIZATION OF CRUDE OIL
In previous papers (Reijnhart & Rose, 1980; Reijnhart
et al., 1980, 1982a, b; Reijnhart & Toneman, 1982), Crude oil is a mixture of hundreds of hydrocarbon
physical models have been presented for calculating compounds which vary in molar weight from very
the evaporation of liquid mixtures and the dispersion light (C~) to very heavy (C~o). C~ is an artificial hy-
of vapours in the atmosphere. Special attention was drocarbon compound which represents all com-
given to the calculation of changes in liquid compo- ponents more volatile than n-C4, and, similarly, C~o
sition and temperature with time and procedures were represents all components heavier than n-C2o. In
introduced for calculating heat fluxes from different order to describe the evaporation of such a multicom-
kinds of ground towards spreading liquid pools. ponent mixture, we characterize it by 27 single com-
In this paper, evaporation of oil from the sea sur- ponents, including C~ and C~o (Table 1). The tool
face will be considered. The thermodynamics of fractions of these components are determined from a
changes in composition (Reijnhart et al., 1982b) will so-called true boiling point curve. In Fig. 1 an
be taken into account in a straightforward manner, example of an Ekofisk oil is given. The weight frac-
contrary to what has been published before. For tion of oil which boils off between, for example, 125.7
example, Blokker (1964) did not allow for changes in and 139.1°C (i.e. 3% w/w) is now represented by the
composition; Goodwin et al. (1976) and Makay & pure component 10 (i.e. m-xylene), which boils at
Matsugu (1973) took a transfer coefficient approach, 139.1°C. The next heavier fraction (2% w/w) is rep-
and Yang & Wang (19=/7) used a first-order decay resented by n-Cg, etc. The tool fraction of C.~ is deter-
1319
1320 R. REIJ:',.HART and R. ROSE

Table 1. Weathering of 166,800kg Ekofisk crude oil

Mot fractions after (days~

Component 0 I 2 4 7

CI 0.0362
2-Methylbutane 0.024
2-Methylpentane 0.048
Benzene 0.062
2-Methylhexane 0.036
Toluene 0.054
m-Xylene 0.046
1-Methyl-2-ethylbenzene 0.042
1,2.4-Trimethytbenzene 0.011 0 0 0 0
i- Methyl-4-isopropylbenzene 0.008
Butane 0.038
Pentane 0,020
Hexane 0.026
Heptane 0.033
Octane 0.050
Nonane 0.035
Decane 0.013
Undecane 0.043 0.018 0.002
Dodecane 0.038 0.068 0.045 0.0t7 0.010
Tridecane 0.035 0.085 0.079 0,063 0.055
Tetradecane 0.032 0.086 0.088 0.087 0.085
Pentadecane 0,027 0.074 0.077 0.080 0.081
Hexadecane 0.021 0.056 0.060 0.063 0.064
Heptadecane 0.022 0.060 0.063 0.067 0.068
Octadecane 0.020 0.054 0.057 0,061 0.062
Nonadecane 0,018 0.050 0.053 0,056 0.057
C_:0 0,163 0.448 0.476 0.506 0.517

T = 2WC, i~Lo = 10 m s- t, de,, = 0.003 m, Z = 1.0 (see also Fig. 5).

mined by a summation of the mol fractions of all is found from

components lighter than C,~ (which are determined x-,+o = S_ x, (3)
from a G L C analysis): i>~20

Y', x i ' Mi
x:, = y . x,. (I) M.;o - i'.2o (4)
i<a,
x,
The molecular weight of Cg is found from the equation i~20

x," M, IIL EVAPORATION AND SPREADING

M~ - i,:4 (2) Evaporation of crude oil may be described by the
Y~ xi following mass and mole balance equations:
i<4
dG~
Similarly, the mole fraction and molar weight of C~-o -- = Q°,(s) + O,, (5)
dt
%w
I L- CO 15[, 2~ 4- TRIMETHYL(W.NZENE
80
2 n-C s t4n-Cto .}U~30
3 L- CG 15I - METHYL-ISOPROPY LSENZIENE 29~"-
2e
4 n-C a t6n-Cll 27~,I"
5 Ce'He Z6 /
60 6 L-C 7 2 5 /
7 n- C7 Z4.,~"
8 TOLUENE 2 5 /
9 n - C• 2 I~:~ "
I0 m- XYLENE zO.jf
I I n-C 9 19
I I~,~
12 I - METHYL-2-ETHYLBENZ, 1 7 Y
,.,, ,5 ~ ,'~'~ 25 n-Czo ~
,ocj~ 26 o-C. i
20 IO~j~l 4 27 n-el. { ,
e 9~ IZ
'= 28
2e .n-C
-c. ~lO
~'C=o
6~f z9 .-c,. ]
3,.Y 30 o-C;oj
0 1 I 1
0 tOO 200 300 400 500
°C
Fig. I. True boiling curve of Ekofisk crude oil (4 m Vigreux) characterized by 30 pure components.
 Evaporation of crude oil at sea 1321

d {x i ~G~}=Y, Q.(S)
~+xi.,~-],
Q, i = I. . . . . N (6)
More sophisticated treatments of spreading for "'ideal
oils" on calm seas have been presented before (Fay,
1969, 1971; Hoult, 1972; Fannelop & Waldman.
in which N is the total number of components. Notice 1972), but we have purposely selected this simple
that the rate of evaporation Q,v is a function of the approach, because important effects such as wave
surface area S. A source term Q, has been included for action and emulsion formation are not taken into
completeness. However, in the numerical examples account even in the more advanced treatments.
presented later (Section VI) Q, is set equal to zero, Furthermore, spreading is in practice never homo-
and the initial configuration consists of a certain pre- geneous. Generally, two areas of oil slick can be dis-
scribed quantity of oil on a given surface area. We tinguished: a thick viscous, mostly emulsified area of
assume in this work that the oil remains isothermal at about 3 mm thickness (d¢,~,)surrounded by a very thin
the temperature of the sea water. In earlier work sheen of oil. The surface area of the latter is larger
(Reijnhart et al., 1982b) a third equation was added in than that of the thick region, but its volume is much
order to calculate changes in temperature as well. The smaller. In Section VI some case studies will be
average molar weights Mj and M, of liquid (oil) and presented in which the sensitivity of the model predic-
gas are derived from the simple linear relationships tions to Z is demonstrated.
N In the following sections the mass of oil GL is calcu-
M~ = Z xi' Mi (7) lated as a function of time by a numerical solution of
i=!
the equations (5) and (6) and by employing the oil
N characterization scheme of Section II.
M, = ~ yi'M~. (8)
l=l

IV. BATCH EVAPORATION EXPERIMENTS

In order to find the molar concentration Yi of
component i in the gas phase we apply Raoult's law The theoretical model which describes the aerody-
Pl namic part of the evaporation has previously been
Yi = x ~ - - , (9) verified and proven satisfactory for pure liquids
Pto|
(Reijnhart & Rose. 1980). In order to verify the
in which p, is the equilibrium vapour pressure of com- thermodynamic model, presented in Section III, we
ponent i and Ptot the total vapour pressure have now carried out batch evaporation experiments,
N which will be described below.
P,o, = Z p,. 00) A schematic sketch of the equipment is shown in
i=l
Fig. 2. The liquid in vessel A is evaporated by blowing
In order to keep the analysis as simple and straight- air through the four symmetrical pipes Pl-P4. Heat
forward as possible, activity coefficients are not taken losses due to evaporative cooling are compensated for
into account, primarily because they are not well by the thermostat B. Very intense mixing (turbulence)
known, but also because of uncertainties in the oil was accomplished by continually reversing the direc-
claaracterization procedures. tion of rotation of the stirrer. The rate of evaporation
The rate of evaporation Qc, depends both on aero- of the liquid is measured by weighing the whole vessel
dynamic parameters such as wind velocity and atmos- A periodically on an electronic balance.
pheric stability, and on liquid paramters such as the In order to simulate evaporation of oil at sea, water
total vapour pressure Pro, and the surface area S of the may be added to A. However, when measuring the
oil slick. For details of the calculation of Q~ we refer
to Reijnhart et al. (1980), where a discussion is given
of the physical model employed.
A very simple model is used to describe the time
dependence of the surface area S. The oil slick is
i ' II
taken to be circular and of uniform thickness d, while
the velocity of the edge of the slick is assumed to be
determined by gravity effects and to be given by the
expression i u, -
v,d,. = Z x~ 2 yd, ( 11 ) t
in which O is the acceleration of gravity. For "ideal"
oils, the factor X is of the order of one, but in reality it
IA ~-i- -'~-<" I
should depend on the oil viscosity, surface tension
and emulsion formation. If the inertial spreading of
< J --j
Fay (1969, 1976) is chosen, Z should be taken equal to
0.5. We assume here that the slick stops spreading
(e.g. due to emulsion formation and a much increased
viscosity) when a critical thickness d . , is reached. Fig. 2. Experimental set-up.
t322 R REIJNHaRTand R. ROSE

Vow EVAPO~,TED %N EVAPORATED

[3 ~'-

20 21-

3c-i ~ 31
40:- 4

60i 1 I i [ I 6 I I I I
0 2 4 6 O I0 0 20 40 60 80 I00
TIME, mtn TIMF., mLn

Fig. 3. Measured batch evaporation as a function of time for pentane (O), iso-pentane (O), toluene (F-I)
and n-decane (A). The calculated values are represented by the solid curve.

rate of oil evaporation in this situation, one has to sidered in Fig. 3 is not negligible. Although in Section
take precautions in order to avoid water evaporation III we considered only isothermal processes, which is
or condensation. This was done by using air which adequate for oil evaporation, we did take into
had been carefully saturated at the desired tempera- account evaporative cooling and heat transfer from
ture, and using a second control vessel which con- the thermostat in the theoretical results of Fig. 3.
tained water only. The humidity of the air was then In order to test the simple mixture model employed
controlled in such a way that no evaporation took in this work, we first measured the evaporation rate of
place from the second vessel. a simple mixture of pentane, toluene and decane. A
We would like to emphasize at this point that the comparison with the theoretical predictions is shown
experiments reported here serve only for checking the in Fig. 4 and the agreement is quite satisfactory.
thermodynamic mixture model and the oil character- Subsequently, by applying the oil characterization
ization scheme. Evaporation in an atmospheric wind scheme of Section II we calculated the evaporation of
field is obviously different from the batch situation different crude oils and made a comparison with
considered here. However, one may still legitimately batch experiments. Again, the proportionality con-
assume that the rate of evaporation in both cases is stant of 0.0009 m s- 1 was used, and the stirring speed
proportional to the thermodynamic equilibrium con- was also kept the same. In Fig. 5, data on Brent,
centration Co of the vapour at the liquid surface: Ekofisk, Kuwait and Burgan crude evaporation at dif-
ferent temperatures are shown. Evidently, the theor-
Q~, = :(Co, (12) etical model describes the experimental data of the
in which ~[m s - t ] depends on the wind velocity, the various oils quite well, in spite of the crudeness of the
geometric configuration, stirring, etc. Therefore, the mixing model and the oil characterization scheme.
thermodynamic theory can be tested by comparing In Fig. 6 evaporation results are shown of 0,050
the evaporation of pure liquids and mixtures. and 0.2 kg oil on water, with initial oil thicknesses of
3 and 12mm, respectively. The same conditions of
V. C O M P A R I S O N OF THEORY AND
stirring, temperature and air flow rate as in the case of
EXPERIMENT
pure oil were applied. Extreme care was taken to
avoid water condensation or evaporation from vessel
In Fig. 3 we show some evaporation results for A. The curves definitely show some retarding
different pure liquids without water. All theoretical influence of mixing of the oil into the water on the
curves were calculated by using the same value of ~( evaporation. However, considering the limited accu-
[equation (12)], viz. 0.0009 ms -t. Evidently, a very */.w EVAPORATED
satisfactory description of the experimental data is 0
found by this procedure. The experiments were all
done with the same stirring mode and the same air
20 -----. 0
flow rate (0.44 x 10-3m3s-t). When the stirring
speed was reduced, different data were found, which 3O

indicates that the evaporation process is influenced, 40 I r

0 40 80 120 160
among other' things, by the state of the liquid surface. TI ME I ml.n
Also, temperature gradients in the liquid surface layer,
Fig. 4. Measured batch evaporation as a function of time
which arise due to evaporative cooling, are influenced for a mixture of pentane (15~ w/w), toluene (35~ w/w) and
by the stirring process. We should mention here that n-decane (50?/0 w/w). Measured: (O); calculated: solid+
evaporative cooling of the very volatile liquids con- CUrVCS.
 Evaporation of crude oil at sea t323

% . EVAPORATED
0

(5)
13
|
(4)

20 ^. (3)

2~ 12)
o
(~)
3C

3~ I I I [ l I
0 i 2 3 4 5 6
TIME, rain
Fig. 5. Batch evaporation of crude oils. Measured: Brent (1), Ekofisk (2) and Kuwait (3), T = 20"C.
Burgan, T -- 30~C (4) and Burgan, T = 20°C (5). Calculated: solid curves.

% w EVAPORATED

,)o
I~
o

~0 0

2-=
o o
3C

35

40 f I I I I 1 I I I I I d
0 2 4 6 B I0 12 14 16 18 20 27. 24
TIME, h
Fig. 6. Batch evaporation of crude oil on water. 0.2 kg Brent on 1.5 kg water (@); 0.051 kg Brent on
1.5 kg water (A). Calculated: solid curves; T = 20.0°C.

racy of these experiments, estimated at + 15%, this calculating Q=v, a s m o o t h b o u n d a r y layer flow over
influence is quite small. For practical purposes, we the oil surface is assumed. This will give a conserva-
therefore believe that evaporation of oil on rough seas tive value as far as the evaporation is concerned, since
can be calculated without taking into account mixing in stormy weather the liquid surface will be in violent
of the oil into the water. However, the oil itself should */.w EVAPORATED
then be considered perfectly mixed. F o r calm seas, 10
this is almost certainly not the case, a n d a retardation
of the evaporation, due to concentration and perhaps 15
also to temperature gradients in the oil layer, must be
ZO
taken into account. The above findings are in agree-
ment with Cormack's et al. (1978) results.
3O
Vl. EVAPORATION OF SPREADING OIL (IA)
35
AT SEA
4O
The rate of spreading of an oil slick [V=d. in equa-
tion (11)] and the critical slick thickness d . . b o t h 4... I I I I I l
influence the rate of evaporation. In Fig. 7 we show 4 8 12 16 20 24
TIME, h
both the influence of Z at a fixed value of
Fig. 7. Calculated evaporation of 166,800 kg Brent crude
d:,i~ = 0.003 m a n d the influence of d:,, at a fixed
oil at sea. (1A) d=,, = 0.03 m, Z -- 1.0; (IB) doa, = 0.0010 m,
value of X = 1. The wind velocity in this example is Z = 1.0; (IC) dcr~ -- 0.0003 m, Z = .1.0; (O) d°,~, = 0.003 m,
chosen equal to 10 m s - t at 10 m height a n d the stab- ;~ -- 0.5; (O) d , , -- 0.003 m, 7. = 2.0; T -- 20.0°C; ~1o --
ility of the atmosphere is assumed to be neutral. In 1 0 m s -t.
I324 R. REIJX;HART and R, ROSE

{I) BURG,.%NT.,R~JDEL:~O°C (4) BRENTCRUDE ~ °C

C2) BuRGAN CRUDE4 0 % (5) EKOFISK CRUDE20%, T%0:~O~/s
C3) KL,W ~ " CRUOE20~C (6) EKOFISK CRUDELR3~'C,~0:20m/s

=I.,, EVAPOR-~TEC
Oi

rO
(~)
2C {2)

3O {3}

4C

60 r I I I I I I
0 I 2 3 4 fi 6 7
TI ME, DAYS

Fig. 8. Calculated evaporation of 166,800 kg crude oil at sea. Z = 1.0; d . . = 0.003 m.

motion and droplet formation may occur, which the uncertainties involved in both the experimental
speeds up the evaporation. Clearly, the influence of data and the theoretical model.
dcrit is very important, whereas Z has but little
influence. A small der~, leads to a large surface area of
VII. DISCUSSION
the slick and, therefore, to a relatively rapid evapor-
ation. However, oil at sea does not spread out as When we consider the oil evaporation curves of
smoothly as suggested by the model described in Fig. 8, two features are quite evident: (I) the initial
Section III. Often emulsion formation occurs and the evaporation of lighter oil components is quite rapid
spreading slows down and ultimately stops due to and takes place within 1 day; and (2) the "residue"
increased viscosity. According to various literature evaporation is very slow and continues for many
da,a (Cormack et al., 1978; Audunson et al., 1979) a days. The shape of the evaporation curves depends on
value of dot, = 0.003 m is a reasonable assumption. many variables, of which the most important are oil
In Fig. 8 two other effects are examined for differ- composition sea-water temperature, wind velocity and
ent crudes: the influence of oil temperature and wind spreading characteristics. The thermodynamics of the
velocity. Evidently, an increase in either temperature evaporation is believed to be described quite well by
or wind velocity accelerates the evaporation process. the model presented here, as was proven in the batch
However, it will also be clear from Fig. 8 that the experiments. The aerodynamics is also described ad-
more significant differences in evaporation are due to equately by the model presented earlier, but the rate
differences in composition. of evaporation during a storm might be underesti-
In Fig. 9, finally, a comparison is made with sea mated, due to increased surface roughness. The
trials. Evidently, Cormack's et al. (1978) results are spreading velocity is shown to have a rather small
better described by our model than Audunson's et al. influence on the slope of the evaporation curves.
(1979), although both fits are satisfactory, considering However, the critical thickness of the slick, beyond

*/*w EVAPORATED
O:

IC

2C

30
1975
40
eye
50
0
I
4 8
I ]
12 0 4
i"
8 12
TIME, DAYS
o b
Fig. 9. Evaporation of Ekofisk crude oil at sea. (a) comparison with Cormack's experiments [] July 1975
~=o=6.2ms -1, T = 11.6°C, 0.5t. O June 1976 ~1o= 1.3ms-L T = ll.0°C, 10.Ot. A Jan. 1977
Fro = 7.7 m s-1, T = 7.0°C, 10.0 t., calculated results with X = 1.0, d , , = 0.003 m. (b) Comparison with
Audunson's experiments • Observation from Bravo blow out. T = 6°C, ~to = 8 m s - t . Solid curve:
calculated evaporation of 10,000 kg oil. Z = 1.0, d , , = 0.003 m. O Observation from Tromsoflaket 1978,
T = 10°C, /~to = 8 m s -1. Dotted curve: calculated evaporation of t0,000kg oil, with Z = 1.0,
~lo =8ms -t ,
 Evaporation of crude oil at sea 1325

which spreading stops, turns out to be very impor- Fannelop T. K. & Waldman G. D. (1972) Dynamics of oil
tant. If the critical thickness could be reduced, for slicks. AIAA Jl 10, 506.
Goodwin S. R.. Makay D. & Shiu W. Y. (t9761 Character-
example, by adding a de-emulsifier to the oil, much ization of the evaporation rates of complex hydrocarbon
more rapid evaporation would occur. Wave action mixtures under environmental conditions. Can. J. chem.
and mixing of the oil into the water have a relatively Engng 54, 290--294.
small influence on the rate of evaporation. Batch ex- Hoult D. P. (1972) Oil spreading on the sea. A. Rer. Fluid
periments indicate a slightly reduced evaporation. Mech. 4, 341.
Makay D. & Matsugu R. S. (1973) Evaporation rates of
However, this reduction is certainly compensated for liquid hydrocarbon spills on land and water. Can. J.
by the greater roughness of the surface in the case of chem. Engng 51,434-439.
stronger wave action. Dissolution of hydrocarbons Reijnhart R. & Rose R. (1980) Vapour cloud dispersion
into the water and subsequent evaporation plays a and the evaporation of volatile liquids in atmospheric
role, as yet unknown, in this process. wind fields. II. Wind tunnel experiments. Atmosph. Ent, ir.
14, 759-762.
Reijnhart R. & Toneman L. H. (1982) Vapour cloud dis-
REFERENCES persion and the evaporation of volatile liquids in atmos-
Audunson T., Steinbakke P. & Krogh F. (1979) Fate of oil pheric wind fields. III. Atmospheric stability and lateral
spills on the Norwegian continental shelf. Oil Spill Con- dispersion. To be published.
ference, Los Angeles. Reijnhart R.. Hermans G. A. M. & Toneman L. H. (1982a)
Blokker P. C. (1964) Spreading and evaporation of pet- Vapour cloud dispersion and the evaporation of volatile
roleum products on water. Proceedings of the 4th Inter- liquids in atmospheric wind fields. IV. Density and grav-
national Harbour Conference, Antwerp. pp. 911-919. ity effects. To be published.
Cormack D., Nichols J. A. & Lynch B. (1978) Investigation Reijnhart R.. Piepers J. & Toneman L. H. (1980) Vapour
of factors affecting the fate of North Sea oils discharged cloud dispersion and the evaporation of volatile liquids
at sea. Part 1: Ekofisk crude oil, July 1975--Feb. 1978. in atmospheric wind fields. I. Theoretical model.
Warren Spring Laboratory Report LR 273 (OP). Atmosph. Era'Jr. 14, 751-758.
Fay J. A. (1969) The spread ofoil slicks on a calm sea. Oil Reijnhart R., Rose R. & Toneman L. N. (1982b) Vapour
on the Sea (Edited by Hoult D. P.). Plenum Press, New cloud dispersion and the evaporation of volatile liquids
York. in atmospheric wind fields. V. Time-dependent proper-
Fay J. A. (1971) Physical processes in the spread of oil on ties of pool and cloud.
water surface. Prevention and control of oil spills. Yang W. C. & Wang H. (1977) Modelling of oil evapor-
American Petroleum Institute pp. 463-467. ation in aqueous environment. Water Res. ! I, 879-887.

'~,'.~. 1 6 8 - - ~