=
2
z
D
z
u z v
f t

.

\

= 1 ) (
Exceeding Concentration
Reference Manual 35
u
f
Bottom friction velocity
D Water depth
z Vertical position in the water column
The settling velocity of the particles is adjusted by:
(5.3)
The formulation is based on the publication: Simulation of turbulent diffu
sion of suspended sediment by a random walk model, published by Rolf
Deigaard and Erik Asp Hansen, DTU, Dec. 1994.
5.7 Exceeding Concentration
If you want to include the exceeding concentration option, you enter the
Exceeding Concentration dialog and specify the concentration limit. This
threshold value should be given in units
z kg/m
3
(concentration) for a PA simulation
z mm (oil slick thickness) for a SA simulation
The program calculates the exceedance frequency between 0 and 100%,
and the results can be saved in the output area file.
5.8 Flood/Dry
If you run the MIKE 21 or MIKE 3 PA/SA module using a hydrodynamic
file where Flood/Dry has been included, the Flood/Dry facility will auto
matically be setup in the MIKE 21 or MIKE 3 PA/SA simulation. If a
particle reaches shallow waters, with a water depth below a certain thresh
old value specified in the hydrodynamic file, the particle position will be
unaffected until the water depth (tide) at this location reaches above this
threshold value.
For a PA simulation you should be aware of the fact that the concentration
field in dry points will be based on the threshold water depths.

.

\

=
D
z
u
dz
dv
f
t
2
1
Basic Parameters
36 M3&M21 PA/SA
5.9 Hydrodynamic Data
5.9.1 General Description
The PA/SA simulation needs information about the hydrodynamic condi
tions. These can be specified in two ways:
z as a full hydrodynamic field from a MIKE 21 or MIKE 3 HD simula
tion (varying in time and space 2D/3D) or in the very simple case as a
timeseries of current speed and direction (constant in space). In the lat
ter case also a bathymetry must be specified.
z as a limited number of flow patterns, a socalled database. The database
is given as a MIKE 21 or MIKE 3 HD file, but only with a few
timesteps. Each timestep contains a characteristic flow situation, and
each time hydrodynamic information is required, a flow situation is
pointed out by the timeseries given in the reference file. Lower and
upper limit defines the minimum and maximum values that are allowed
in the reference file.
5.9.2 Example of a Database
Establishment of a database is not an easy task as a onetoone correspon
deance between the surface current speed in a point and the flow pattern in
the entire domain should be defined.
From a measurement gauge surface current speed for a position in the
computational domain is available. The surface current speed typically
belongs to the interval 5m/s to 5m/s. If the characteristic situations are
northerly current, southerly current and almost no current 3 flow
fields are sufficient, corresponding to 5m/s, 0m/s and +5m/s. The fields in
the database must be sorted in ascending order. When the simulation runs
each value in the reference file is matched by a flow field which is interpo
lated linearly between the two surrounding flow fields.
Hints: Other timeseries than one of current speed can be used as reference
file. To an example a flow field that repeats itself can be constructed by
using a database file containing a flow field for every 6th hour (00:00,
06:00, 12:00, 18:00) and a reference file that points out the time of the day.
5.10 Line Discharge Calculations
The line transport facility is used to calculate the transport of the sub
stance across a userspecified line in the model area. At each time step
both the instantaneous and the accumulated transport from time zero is
calculated.
Logarithmic Velocity Profile
Reference Manual 37
The cross sections is specified in grid coordinates (floating values) and the
discharge will be taken as positive clockwise across the section, see
Figure 5.1.
Figure 5.1 Orientation of discharge line
You can make up to 3 discharge calculations, and it is allowed that the
cross section may cross land areas.
The output is saved in a type 0 file.
5.11 Logarithmic Velocity Profile
If you have included a logarithmic velocity profile in your simulation, you
should specify a bottom roughness. The bottom roughness can be constant
in all model points or vary in space (given as a type 2 file). The roughness
is specified in metres.
Basic Parameters
38 M3&M21 PA/SA
5.12 Maximum Particle Age
A maximum particle age can be given (in seconds). Particles reaching the
specified age (counted from the time of release, not from the start of the
simulation) are simply taken out of the simulation. This option can be use
ful e.g. under long sedimentation calculations, where only the rate of sedi
mentation is to be considered, not the amount of sedimented material.
5.13 Nested Grid
The following rules should be observed when running a nested PA/SA
simulation
z the advective current should be specified for each area
z bed resistance should be specified for each area (constant or 2D map)
z the type of Eddy viscosity formulation will be common for all areas,
whereas the actual value can be different in the respective areas
z spatial varying wind is only specified for the main area. The wind is
internally interpolated to the subareas
z sources can be specified in all model areas, but should be specified in
the model area of the finest resolution which covers the actual geo
graphical position of the source
z in the output specifications, each output area must be related to an
associated model area
See also Bathymetry.
5.14 Number of Particles
In the Sources dialog it is possible to specify the number of particles
released per time step.
If you specify a short simulation period or a large time step, it is some
times a good idea to release a number of particles at each time step,
thereby making the concentration fields more "smooth".
If you specify that 5 particles are released at each time step and constant
source flux of 5 m
3
/s, then each released particle will have the mass 1
m
3
/s/timestep. You can specify either a constant or a timevarying source
mass. Note that the source flux in PA has units kg/s and in SA m
3
/s.
Output Areas
Reference Manual 39
5.15 Output Areas
5.15.1 General Description
Computers are not yet so powerful that a simulation can be run each time a
plot of, for example, the concentration field is needed. Therefore it is nec
essary to store the basic results from the simulations. On the other hand,
the amount of output produced by a single simulation is often so large that
it is necessary to limit the amount of output saved. You therefore have the
option of saving up to 8 selected parts of the output.
5.15.2 Specifying the Output Area
In the Output Areas dialog you specify how many output data files you
wish to have produced from the simulation. You can specify the contents of
each data file under "Output Items".
Specify the name of the output file (including the path).
Give a short description of the content under "title"
Specify the area to be included in the data file. By default the whole area
is chosen, but if you are only interested in a part of the model area and you
have finished the calibration, you could specify the area of interest only.
You can specify a Nested Grid Output area by selecting a negative fre
quency in both horizontal directions and in that way get a better resolution
of your particles.
If, for example, you specify the grid frequency to 4, the grid spacing in
your output file will be a quarter of what is specified in the hydrodynamic
file.
Finally, you specify the range of time steps to be saved and whether every
time step should be included or only every second, third, etc. The aver
aged output results, which you can select in the Output Areas dialog
(under Output Items), are based on calculations from a period equal to the
time frequency for the output data files.
5.15.3 Remarks and Hints
If you run the model in foreground mode, the progress of your simulation
is shown on the screen. If the model is run in background mode (windows
icon), you can follow the progress of your simulation by following the
number of time steps written in your .log file.
Basic Parameters
40 M3&M21 PA/SA
5.16 Sources
5.16.1 General Description
The effects of rivers and outlets from power stations, etc. can be included
in a simulation. These sources have to be included in both the hydrody
namic and the PA/SA modules. The source positions and discharges are
specified in the hydrodynamic input part, whereas the source fluxes
(amount of particles/amount of oil) are specified in the particle input part.
5.16.2 Horizontal position
The horizontal position must be specified for each source. The position
should be given in grid points (floating point values) and it can either be
given as
z constant values
z a type 0 file containing the x and y coordinates of the source. The x and
y coordinate must be given as two separate items.
Furthermore the reference area for the source should be specified.
5.16.3 Vertical position
For a PA simulation also the source depth must be specified, either as
z a constant value
z a type 0 file
The source depth is given in metres and is counted positive from the sur
face.
In an SA simulation the source is automatically placed at the surface.
5.16.4 Source flux
The mass flux from each source can be given as
z a constant value
z a type 0 file
In SA the mass flux should be given in and in PA in .
If the source flux is given as a type 0 file, the timestep in this timeseries
should be equal to or greater than the simulation timestep. In the latter
case MIKE21/3 SA/PA automatically interpolates linearly. A timestep in
the source flux file that is less than the simulation timestep is not allowed.
m
3
/s   kg/s  
Stratification
Reference Manual 41
An instantaneous release can be achieved by defining a source flux file
with timestep equal to the simulation timestep and zero value at all places
except for the start time of the simulation.
5.17 Stratification
See Water Properties.
5.18 Time Exposition
By including Time Exposition, the model will create a map with informa
tion about how fast the particle cloud is moving. This information can be
interpreted as how long time a particle uses for being advected and dis
persed from the source point to a certain location. The results will be pre
sented as isolines with the unit seconds.
5.19 Time Step
Selecting the Time Step
As a rule of thumb a particle should not move more than one grid cell in
each timestep, otherwise the numerical integration of the stochastic differ
ential equation that controls the particle motion becomes unstable. If a
particle displacement in fact exceeds one grid cell MIKE 21&3 PA/SA
divides the displacement in halves until a displacement below one grid
cell is obtained.
The time step for your simulation is selected as follows:
1 First you determine the grid spacing, , as described under Bathyme
try, which is equal to the maximum allowable particle displacement
2 Secondly you determine the maximum current speed U
max
in the com
putational domain. In most cases U
max
is assessed after inspecting the
flow field.
3 From the definition of the Peclet number the maximum timestep can
now be calculated: . In nested simulations must
be calculated for all areas and the smallest one chosen for the PA/SA
simulation.
x
t
max
x/U
max
t
max
Basic Parameters
42 M3&M21 PA/SA
5.20 Water Properties
The density of the water is calculated from salinity and temperature .
In an oil spill simulation (SA), the water temperature itself is a parameter
that influences the properties of the oil. In SA the temperature and salinity
can be specified as constants or timeseries.
In a sedimentation calculation (PA) the particles are affected by the den
sity of the water. If 3D fields of salinity and/or temperature are available
these can be used in sedimentation simulations, otherwise constant values
for salinity and temperature must be given.
5.21 Wind
5.21.1 General Description
When wind is acting on the surface of water, energy is transferred to the
water through the shear stress. The energy is used for stirring/mixing the
water and accelerating the water in the direction of the wind.
The wind shear stress results in acceleration of the upper thin surface
layer, and gradually the zone, where the shear stress is active, progresses
in depth.
The depth of influence of the wind is defined as the depth where 90% of
the wind induced energy is used for the above mentioned processes. The
depth depends on the wind speed and direction history since the profile is
a result of the wind action over time, and furthermore, it depends on the
bathymetry and stratification of the water column.
The surface current velocity due to the wind is calculated as
V
s
= C
w
W (5.4)
where C
w
is the wind friction coefficient and W is the wind velocity in m/s
10 m above the sea surface. The wind friction coefficient is different from
what is specified in MIKE 21 & MIKE 3 HD.
Notice that the direction of the wind is given in degrees blowing from rel
ative to true North, see Figure 5.2.
Wind
Reference Manual 43
Figure 5.2 Definition of wind direction
5.21.2 Specifying the Wind Conditions
The wind conditions can be specified in three ways:
1 As a wind which is blowing from the same direction and with the same
magnitude over the whole model area for the entire simulation period
2 As a wind where the magnitude and direction varies during the simula
tion period, but is the same over the entire model area
you have to prepare a data file (type 0) containing the wind speed
and direction before you set up the PA/SA simulation.the wind
speed and wind direction must be given as two separate items in the
data file. The time step of the wind input data file does not, how
ever, have to be the same as the time step of the PA/SA simulation.
A cubic interpolation will be applied if the time steps differ. The
only requirement is that the wind data arespecified for the complete
simulation period
In both cases it is not necessary to specify a start up period during
which the wind speed is increased.
3 As a wind where the magnitude and direction varies during the simula
tion period and over the model area
Basic Parameters
44 M3&M21 PA/SA
you have to prepare a data file (type 2) containing the wind speed
components and air pressure before you set up the oil spill simula
tion. This can be done by either using one of the two MIKE 21 wind
generating programs: cyclone generated wind and pressure, or wind
generated on the bases of digitised pressure fields. (These programs
are found in the MIKE 21 Toolbox). Or you can enter the data
directly into a type 2 file using the grid editor.
the wind speed and pressure must be given as three separate items
in the data file. The first item should be the pressure in HPa, the two
next ones should be the wind speed in the xdirection and ydirec
tion, respectively. However, the time step of the wind input data file
does not have to be the same as the time step of PA/SA simulation
a linear interpolation will be applied if the time steps differ. The
only requirements are that the wind map matches the bathymetry
map and that the wind data covers the complete simulation period
5.21.3 Specifying the Wind Friction
The wind friction coefficient you specify in MIKE 21 & MIKE 3 PA/SA is
not comparable to what you can specify in MIKE 21 & MIKE 3 HD,
while in the HD the coefficient corresponds to the momentum of the fluid
and in the PA/SA the coefficient corresponds to the surface speed of the
fluid.
Normally a wind friction coefficient of 0.02 will give good results for
moderate and strong winds in the open sea. For weak winds, however,
smaller coefficients can be used.
If you specify a varying wind speed, you might also need to specify a var
ying friction coefficient. Consequently, the possibility of varying the fric
tion coefficient linearly as a function of the wind speed is included.
Remarks and Hints
You can use the wind friction factor as a parameter in you model calibra
tion.
5.21.4 Specifying the Wind Drift Angle
Due to the influence from the Coriolis force, the direction of the wind drift
vector is turned relative to the wind direction. The angle of deviation is
termed the wind drift angle. It turns to the right on the Northern Hemi
sphere and to the left on the Southern Hemisphere. The magnitude of the
wind drift angle varies in time if the wind friction is time varying, other
wise the wind drift angle is constant in time and is given in degrees.
Wind
Reference Manual 45
5.21.5 Shore Current Zone and Depth of Wind
The Shore Current Zone is the depth at which the calculation of the
wind influence on the current changes from the onshore formulation to
the offshore formulation. Setting Shore Current Zone to zero gives off
shore formulation all over. Onshore formulation at all depths can be
obtained by setting Shore Current Zone to a value that exceeds the larg
est depth.
The onshore expression calculates a correction over the entire depth, the
offshore expression corrects only the current down to Depth of Wind.
Thus Depth of Wind should always be greater than or equal to Shore
Current Zone.
Basic Parameters
46 M3&M21 PA/SA
Dissolved matter
Reference Manual 47
6 SEDIMENT PARAMETERS
6.1 Dissolved matter
If dissolved matter is chosen in the Type of Particles dialog the particles
are neutrally buoyant, which means that they follow the flow without
being settled. This option can be used for calculation of spreading of e.g.
bacteria, where it can be combined with a Linear Decay.
6.2 Light Attenuation
6.2.1 General Description
In clear water the light intensity is reduced exponentially from the water
surface down through the water column. Sediments in suspension result in
an additional dampening of the light caused by shading from the sediment
particles.
The background dampening from clear water depends on the water depth
and a dampening coefficient. The light attenuation caused by sediments in
suspension depends on the water depth, the concentration of suspended
sediments and a light attenuation coefficient describing the shading per
mass.
The light attenuation can be calculated both as an instantaneous value and
as an averaged value, where the averaging one is referring to the time fre
quency of saving output.
6.2.2 Recommended Values
The default light attenuation coefficient is based on analyses of measure
ments performed in the Great Belt, Denmark, 1991. Shading calculations
using this value have been verified against measurements of eelgrass
growth in the Great Belt.
The default light attenuation coefficient includes both the background
dampening of clear water and the light attenuation caused by sediments in
suspension.
It is, however, made possible that you can specify a separate background
dampening coefficient.
Sediment Parameters
48 M3&M21 PA/SA
6.3 Linear Decay
Decay as a result of chemical or biological processes is described in the
following way:
Every particle has attached a certain mass of the total released substance.
This mass may change during the calculation if a decay process is defined.
A first order decay is considered so that
(6.1)
M being the particle mass and k [s
1
] the decay factor. When the mass of a
certain particle becomes less than a user specified minimum mass, the par
ticle is taken out of the simulation.
The decay law yields after integration
(6.2)
The decay period T
90
is defined as
(6.3)
which brings
(6.4)
The decay period must be specified by the user.
6.4 Resuspension
6.4.1 General Description
Deposited material resuspends when the shear stress at the sea bed
exceeds a critical limit. The amount of material resuspended depends on
the grain size of the material, the degree of consolidation and the physico
chemical interactions between the particles (cohesive/noncohesive).
M k
dt
dM
=
( ) ( ) t k M t M
o
= exp
1 . 0 =
o
M
M
( )
90
10 ln
T
k =
Sedimentation
Reference Manual 49
6.4.2 Resuspension Within This Model
Resuspension is presently not available for the 1
st
order depth averaged
sediment description. Both the cohesive/noncohesive sediment calcula
tions include resuspension.
The resuspension of sedimented material is determined by the relationship
between the critical shear stress and the actual shear stress at the seabed.
The critical shear stress is a function of the critical shields' parameter and
depends on the material in question.
6.5 Sedimentation
6.5.1 1
st
Order Sediment Process
The sediment is regarded as fully mixed over the depth, and the settling of
sediment takes place with a constant or randomly generated settling veloc
ity, thereby reducing the remaining mass of each particle in suspension.
When the mass falls below a specified minimum value, the remaining
mass of the particle is deposited. The particles are advected with the local
velocity profile, which can be constant or logarithmic in the vertical. The
velocity profile can be superimposed with a windinduced current profile
if wind is applied to the model.
6.5.2 Mud and Sand
The sediment is released at a specified depth, and the particles settle with a
constant or randomly generated settling velocity. The particles are deposited
when they reach the bottom. This simulation type gives you the possibility
of calculating the concentration of suspended material at different depths.
The particles can be advected by a threedimensional velocity field or by
logarithmic velocity profiles established from the depthintegrated veloci
ties specified in the Hydrodynamic Data dialog. The velocity profiles can
be superimposed with a wind induced current profile if wind is applied to
the model.
Typical values for the critical shields' parameter for motion is 0.045 for
sand and 0.2 for clay.
NOTE: The sediment output will be zero at the first time step for saving
output.
Sediment Parameters
50 M3&M21 PA/SA
6.6 Settling Velocity
6.6.1 General Description
There are two ways of including settling velocities in a particle simulation.
The easiest way is to specify a constant settling velocity for each source
specified. A more advanced method is to include a particle size distribu
tion or a particle settling velocity distribution based on laboratory meas
urements of the material released.
6.6.2 Frequency function for settling velocity
The model expects the following input:
z A type 0 file with two items, where each timestep is a set of settling
velocity [m/s] and observed frequency.
A settling velocity chosen at random from this distribution is assigned to
each particle when it is released.
6.6.3 Frequency function for particle size
The model expects the following input:
z A type 0 file with two items, where each timestep is a set of particle
size [mm] and observed frequency
z relative density of sediment and a kinematic viscosity of water
The model calculates the frequency function for settling velocity by using
Stokes law. A settling velocity chosen at random from this distribution is
assigned to each particle when it is released.
Air Properties
Reference Manual 51
7 OILSPILL PARAMETERS
7.1 Air Properties
The air properties consist of the air temperature and the cloudiness, which
also can be specified as constant in time or as time varying (type 0). These
environmental data influence the weathering of oil, and are important in
the calculation of the properties in the waterinoil emulsion. The cloudi
ness is applied to correct the heat input from the sun on a clear day to the
actual conditions. The time series has to cover the specified simulation
period and the model will make the necessary interpolation to get specific
values at each time step during the simulation.
7.2 Emulsification
Emulsification of crude oils and/or refined products involves the disper
sion of water droplets into the oil medium. As the percentage of water
increases (up to a maximum of 7585%), the resulting emulsion undergoes
a dramatic increase in viscosity, its effective volume increasing by a factor
of up to 5, its density approaching that of the surrounding water, and col
our changes occurring ranging through browns, oranges and reds. These
properties of the resulting emulsions make cleanup operations more diffi
cult. Furthermore, evaporation of components from an emulsion will also
be diminished.
The mechanism by which water incorporates droplets into a slick is poorly
understood. Therefore, the process of emulsification can only be discussed
in a qualitative way. However, it can be stated that the mixing energy
through wave action is an important requirement and that emulsions are
less likely to be formed in calm conditions.
To estimate the significance of emulsification for a particular oil type
under given environmental conditions, it is necessary to know the follow
ing:
z first, whether an oil has the potential to form an emulsion
z secondly, how fast the emulsion is formed and how much water is
taken up, and
z finally, how stable the emulsion is
Furthermore, information about the emulsified oil is required.
OilSpill Parameters
52 M3&M21 PA/SA
The rate at which water is taken up in the slick is greatly influenced by the
type of oil (Lynch) and the environmental conditions. However, properties
and conditions that favour vertical dispersion are expected also to be
favourable for emulsification.
For cleanup purposes, information on the viscosity of an emulsion is nec
essary.
7.3 Evaporation
The evaporation processes are included in the MIKE 21 & MIKE 3 SA
models. The rate of evaporation of the lighter components is influenced by
the percentage of lighter components in the crude oil itself, the oil temper
ature, the oil thickness and area, and the physical forces of wind (and
wave) energy. The effect of evaporation is to increase the density of the
remaining oil and to increase its viscosity. Pour point may also be affected,
determining whether or not the oil remains a fluid or becomes a waxy
solid or semisolid.
In practice, as crude oil spreads out on the sea surface a large surface area
is exposed and the initial rapid loss of the volatile components occurs. The
main loss occurs during the first 2 or 3 hours and the process then slows
down with the increasing proportion of the higher boiling components
remaining in the residue. In theory, the higher boiling components also
have a vapour pressure and should eventually evaporate, but increasing
viscosity and pour point of the remaining oil lead to a complex process of
emulsification and solidification, which reduces the surface area of the
slick and hence the rate of evaporation.
During the first 24 hours after spillage, most crude oils will have lost
approximately 2530% of their lighter components and although the loss
continues, the rapid evaporation process can be assumed to have ended.
Subsequent "weathering", e.g. by wind action and thermal radiation from
the sun, then leads to a slow but progressive loss of the middle boiling
range components. The associated increase in the density and viscosity
takes up to about one week to simulate with the SA model.
The effect of evaporation on the physical properties of the resulting resid
ual has a major influence on the type of response required for cleanup of
the oil. It has already been shown how the pour point influences the
fluid/solid properties of the oil and the method of recovery.
Heat Transport
Reference Manual 53
7.4 Heat Transport
Five input parameters are used in the heat balance calculation for the oil.
The Albedo Value describes the fraction of the solar radiation, which is
reflected. This parameter is primarily depending on the colour of the oil.
The emissivity of the oil, water and air are constant values in time and space
and are applied for the calculation of the heat balance for the oil slick. The
evaporation constant is proportional to the amount of evaporated oil, and
is a calibration parameter.
7.5 Radiation
It is possible to take the solar radiation into account when calculating the
oil slick temperature. The emissivity constants related to the water, oil and
air have to be specified. The model calculates the variations in the solar
radiation with respect to the geographical position, the time of the day and
which season of the year.
7.6 Sedimentation
General Description
Only relatively few crude oils weather to produce residues that are denser
than seawater and sink. The majority of crude oils yield weathered oils
lighter than seawater, and even if they form stable waterinoil emulsions,
the resultant material is positively buoyant. Nevertheless, floating oil can
come into contact with heavy mineral particles, e.g. sand, silt, etc., espe
cially in coastal areas. In these instances solid particles may be incorpo
rated into the bulk of the oil or may become coated with a layer of oil. In
both circumstances, the resultant mixture becomes negatively buoyant and
sinks. The oil is therefore carried to the seabed and incorporated in bottom
sediments where the degradation processes are very slow.
Remarks and Hints
The above mentioned processes are not described in the Spill Analysis
module of MIKE 21 and MIKE 3, which not at all take the buoyant effects
into account.
7.7 Vertical Dispersion
Transport into the water column can be accomplished by a number of
mechanisms: dissolution, dispersion, accommodation, and sedimentation.
OilSpill Parameters
54 M3&M21 PA/SA
Dispersion and accommodation are by far the most important mechanism
during the first week of weathering.
Dispersion is a mechanical process. Water turbulence tears off globules of
oil and entrains them, forming an oilinwater emulsion. These emulsions
can be stabilised by surfaceactive agents coating the oilwater interface
and thus preventing the oil from coalescing. This formation of colloids is
called accommodation. In harsh weather, the dominant dispersion mecha
nism is probably wave breaking, while in calm weather, the most signifi
cant mechanism is probably the stretching compression of the slick
leading to droplet separation.
Vertical dispersion of crude and/or refined products after spillage at sea is
the process of forming small droplets of oil, which will be incorporated
into the water column. Besides evaporation, the rate of vertical dispersion
largely determines the lifetime of an oil spill.
Vertical dispersion reduces the volume of the slick at the surface and
reduces evaporative loss, but it does not lead to changes in the physico
chemical properties of the spilled material in the way that evaporation
does, for example.
In general, oilinwater emulsions are not stable, and the oil droplets tend
to coalesce and return to the surface, again forming a slick. Vertical emul
sifiers and/or suspended solids can stabilise them by reducing interfacial
tension, which leads to a reduction in droplet size.
Vertical dispersion is still a very poorly understood process, and the math
ematical description of it is still in an early phase of development
(Mackay). The importance of dispersion for a particular spill of crude oil
under certain weather conditions, and differences in dispersion behaviour
of different crude oils, can only be evaluated in a qualitative way. There
are fairly reasonable "best estimates" derived from observations of histori
cal spills (Blaikley et al.), which determine the rate of dispersion at sea to
be 05% oil lost per day depending on the sea state. More detailed infor
mation on vertical dispersion can be obtained from the comprehensive
review compiled by (Raj et al.).
The fraction of sea surface dispersed in the water column per unit time is
calculated as a lost fraction of sea surface, under nonwave breaking con
ditions. The input parameter is named oilinwater interfacial tension. The
default value is 20 [dyne/cm]
Viscosity
Reference Manual 55
7.8 Viscosity
If the viscosity of the original crude oil, determined at only one tempera
ture, is all that is available, it is possible to estimate the viscosity of the
weathering oil at various temperatures as a function of the time after spill
age. Although only a rough estimate of the viscosity of the remaining oil
is obtained it may be useful in deciding on the cleanup technique to be
applied, e.g. whether or not dispersants should be used. The oil viscosity
in the MIKE 21 & MIKE 3 SA is specified in Cs at a certain reference
temperature measured in degrees Celsius.
OilSpill Parameters
56 M3&M21 PA/SA
57
S C I E N T I F I C B A C K G R O U N D
58 M3&M21 PA/SA
Scientific Background 59
8 INTRODUCTION
The present Scientific Documentation aims at giving an indepth descrip
tion of the equations and numerical formulation used in the Particle Anal
ysis module (PA) and the Spill Analysis module (SA) of MIKE 21 &
MIKE 3.
Both modules are based on the random walk technique where an ensemble
of particles is followed instead of solving the Eulerian advectiondiffusion
equation.
The particles move due to the advective current and due to turbulent fluc
tuations. The advective velocities are (usually) obtained from a hydrody
namic simulation (MIKE 21 HD or MIKE 3 HD), whereas the turbulent
contributions are controlled by the dispersion coefficients.
The properties of the released particles are described by
z PA: Settling velocity, either constant or specified as a distribution of
velocities or grain sizes. Also nonbuoyant particles can be simulated.
z SA: The ratio between eight oil fractions that are charachterized by
density, vapour pressure, pour point etc.
The mass (PA) or the volume (SA) of the particle cloud can change due to
z PA: Sedimentation/resuspension. Futhermore a linear decay can be
employed.
z SA: Weathering of the oil: emulsification (water uptake), evaporation,
entrainment into the water column and dissolution.
Both models calculate concentration fields of the pollutant by tallying the
number of particles in each grid cell.
Introduction
60 M3&M21 PA/SA
Lagrangian discrete parcel method
Scientific Background 61
9 ADVECTION AND DISPERSION
The advectiondiffusion model is designed to simulate the surface and
subsurface transport of pollutants. The pollutant is divided into discrete
parcels and sets of spatial coordinates are assigned to each parcel. It is
assumed that these parcels advect with the surrounding water body and
diffuse as a result of random processes.
9.1 Lagrangian discrete parcel method
Typically in models using Lagrangian Discrete Parcels Method the pollut
ant is represented by a large ensemble of small parcels. The movement of
each parcel is affected by the physicochemical processes. Once the par
cels are released in the water body, their discrete path and mass are fol
lowed and recorded as function of time relative to the reference grid
system fixed in space. Then the density distribution of the ensemble can
be interpreted as the concentration of the pollutant.
The Lagrangian discrete parcel scheme calculates the displacement of
each particle as the sum of an advective deterministic component and an
independent, random Markovian component, which statistically approxi
mates the random and/or chaotic nature of timeaveraged tidal mixing.
9.1.1 AdvectionDiffusion equation
In a vertically mixed environment, twodimensional depthintegrated
mass transport models are commonly employed. An analysis of the trans
port of a conservative substance with no sources can be based on the fol
lowing time and spaceaveraged equation:
(9.1)
where c(x,y,t) is the depthaveraged concentration, u(x,y,t) and v(x,y,t) are
the x and y components of the depthintegrated velocities, h(x,y,t) is the
depth (mean sea level), and D
xx
(x,y,t), D
yy
(x,y,t), D
xy
(x,y,t) and D
yx
(x,y,t)
are dispersion coefficients which represent all of the mechanisms that
]
y
c
D
h +
x
c
D
[h
y
+ ]
y
c
D
h +
x
c
D
[h
x
=
y
h) c (v
+
x
h) c (u
+
t
h) (c
yy yx xy xx
v x t ( ) t , ( )
u x t ( ) t , ( )

\ .
 
atan =
) ( ) ( )
D

D
( =
D
) (
D
+ ) (
D
=
D
) (
D
+ ) (
D
=
D
T L xy
2
T
2
L yy
2
T
2
L xx
cos sin
cos sin
sin cos
t d
dx
A x t , ( ) B x t , ( ) t ( ) + =
t ( )
Lagrangian discrete parcel method
Scientific Background 63
Equation (2.4) becomes equivalent to the stochastic differential equation
(9.5)
where dW(t) is the random Wiener process. According to Gardiner /9/ the
mean value of dW(t) is zero and the mean square is proportional to dt.
The simplest discretization of equation (9.5) is the explicit Euler scheme
for stochastic differential equations. This scheme is of order
(Kloeden and Platen, /13/).
(9.6)
where Z
n
is a vector of one, two, or three independent random numbers
depending on the dimensionality of the problem from a distribution with
zero mean and unit variance.
9.1.3 FokkerPlanck equation
In the limit when the number of particles is infinite and the time step is
infinitely small, equation (9.6) is equivalent to the FokkerPlanck equa
tion
(9.7)
where f(x,tx
0
,t
0
) is the conditional probability density function for x(t) .
dx A x t ( ) t , ( )dt B x t ( ) t , ( )dW t ( ) + =
t
x
n
x
n
x
n 1
A x
n 1
t
n 1
, ( )t B x
n 1
t
n 1
, ( ) tZ
n
+ = =
t
f
x
i
A
i
f ( ) +
x
i
x
j
2
1
2
B
ik
B
jk
f
\ .
 
=
Advection and Dispersion
64 M3&M21 PA/SA
The partial differential equation (9.1) and the FokkerPlanck equation
(9.7) are equivalent if
(9.8)
(9.9)
and
(9.10)
In the special case where D
xy
= D
yx
= 0, i.e., the coordinate system is
aligned the principal axes of dispersion, the random walk analogue to
equation (9.1) becomes
(9.11)
(9.12)
Looking at equations (9.11) and (9.12) we see that the deterministic or
'advective' velocity of a parcel in a given direction consists of three com
ponents: (1) the true advective velocity, u, (2) the gradient of the disper
sion coefficient and (3) a term due to the gradient of the water depth. In
case of a constant dispersion coefficient the contribution from (2) is zero.
The choice of the random component is discussed in sec. 9.3.
,
x
D
+
x
h
h
D
y
D
+
y
h
h
D
u +
y
D
+
y
h
h
D
+
x
D
+
x
h
h
D
= A
yx yx yy yy
xy xy xx xx








.

\

+
v
(
=
yy yx
xy xx
T
D D
D D
B B
2
1
h c = f
Z
t
D
2 + t u) +
x
D
+
x
h
h
D
( = x
n xx
xx xx
Z
t
D
2 + t v) +
y
D
+
y
h
h
D
( = y
n yy
yy yy
Advection
Scientific Background 65
The time step must be compatible with the grid size to make sure that the
simulation is accurate. This means that the timestep must be chosen such
that a particle at most moves at most one grid cell during one timestep.
9.2 Advection
The combined effects of current and wind drag cause the advection of the
particles.
9.2.1 Current
A significant factor in the transport of the pollutant is the current field,
which can be obtained from a two or threedimensional hydrodynamic
model for tidal and winddriven currents in the region of interest, but a
time series of measurements can also be employed.
MIKE 3 HD
If the the current field is obtained from a MIKE 3 HD simulation it reads
(9.13)
where u, v, and w are the velocities in the x, y and z directions respectively,
x, y, and z are the Cartesian coordinates, and t is time. In MIKE 3 HD the
wind shear layer is resolved, thus the wind contribution should not be
added from the PA/SA simulation.
MIKE 21 HD
A current field from MIKE 21 HD is depth integrated, which means that
the vertical velocities are inherently assumed to be zero and the values of
the horizontal velocities are averaged over the depth
(9.14)
Even if the MIKE 21 HD simulation includes wind, the wind shear layer is
not resolved in the depth integrated field. MIKE 21 PA/SA includes the
possibility of parameterizing the wind shear, which is explained is section
9.2.3.
From the depth integrated velocities a logarithmic velocity profile can be
established, but the simulation can also be run with the homogeneous pro
file (rarely used as there is no reason for not using the extra information
that the logarithmic profile gives).
U u x y z t , , , ( ) v x y z t , , , ( ) w x y z t , , , ( ) , ,   =
U
a
u
a
x y t , , ( ) v
a
x y t , , ( ) ,   =
Advection and Dispersion
66 M3&M21 PA/SA
Logarithmic profile
The logarithmic profile is defined as:
(9.15)
where
z = vertical coordinate measured from sea surface [m]
V(z)= the logarithmic current profile
= von Krmn's constant (0.42)
h = water depth [m]
= Nikuradse roughness [m]
= friction velocity [m/s]
The friction velocity is given by
(9.16)
where
= mean current velocity [m/s] =
and the other variables are defined as above.
The expression in eq. (9.15) equals zero when
(9.17)
Below this depth the advective velocities are asumed to be zero.
V z ( )
U
f

h z
k
n
/30

\ .
 
ln =
k
n
U
f
U
f
V
mean
h
k
n
/30

\ .
 
ln 1

\ .



 
=
V
mean
u
a
2
v
a
2
+
z h
k
n
30
 =
Advection
Scientific Background 67
Figure 9.1 Logarithmic current profile and wind induced current profile
9.2.2 Weighting of current and wind
The total drift velocity can be determined by a weighting formula:
(9.18)
in which
= wind velocity vector usually at 10 m above the water surface
= depth integrated current velocity vector
c
w
= wind drift factor
c
a
= advective water current factor.
9.2.3 Wind
Wind drift vector
The magnitude of the wind drift vector is commonly assumed to be pro
portional to the magnitude of the wind speed 10 m above the sea surface,
U
w
. This factor c
w
*
is termed with the wind drift factor and the common
value varies from 3 to 4 percent by AlRabeh /1/.
The wind contribution is divided into an offshore contribution and an on
shore contribution. The onshore formulation is employed in shallow
waters with a water depth less than a specified depth, h
sep
, which is a posi
tive value measured from the free water surface (with unit [m]). The on
U
tot
c
w
z ( )U
w
c
a
z ( )U
a
+ =
U
w
U
a
Advection and Dispersion
68 M3&M21 PA/SA
shore expression acts over the entire depth. The offshore formulation
takes over when the water depth exceeds h
sep
. The offshore expression
acts only down to the (user) specified depth of wind influence, h
w
.
Figure 9.2 shows the principle.
Figure 9.2 The wind influence on the advective current. The onshore expres
sion acts on shallow depths, on deeper water the offshore expres
sion takes over.
If h
sep
= 0 the offshore expression applies all over. If h
sep
is greater than
the maximum depth the onshore expression applies at all depths.
The depth of wind influence (h
w
) should always be greater than
(or equal to) the depth that separates the onshore and off
shore formulations (h
sep
), otherwise a particle can move from a
wind influenced position to a position with no wind influence
by moving horizontally.
Offshore formulation
The vertical distribution of the offshore wind drift vector is given by
(9.19)
where
= 3/h
w
[m
1
]
c
w
z ( ) c
w
k
0
z ( ) exp = 0 z h
w
k
0
Advection
Scientific Background 69
= depth of wind influence [m] (input)
z = vertical coordinate measured from sea surface
= wind drift factor [] (input)
In order not to create an additional mass flux caused by wind shear stress,
the current profile is adjusted such as to have zero net flow. The contribu
tion can be expressed as:
(9.20)
Onshore formulation
The onshore distribution is based on a parabolic vertical profile and is able
to produce back flow at depth, where the offshore logarithmic profile does
not.
The vertical distribution of the parabolic onshore profile is given by
(9.21)
where
h = local water depth in meter
z = vertical particle coordinate, measured from sea surface
= wind drift factor (input)
The parabolic profile causes the windgenerated flow in the upper third of
the water column to be in the same direction as the current and the flow in
the lower part to be in the opposite direction of the wind. There is no net
depth averaged mass transport due to the wind.
Wind drift angle
Due to the influence from the Coriolis force, the direction of the wind drift
vector is turned relatively to the wind direction. The angle
w
of deviation
is termed the wind drift angle. It turns to the right on the northern hemi
h
w
c
w
( ) ( ) ( ) z c dz z c
h
z c
w
h
w
w
w
w
+ =
0
1
c
w
z ( ) c
w
1 3
z
h

\ .
 
1
z
h

\ .
 
= 0 z h
c
w
w
= kinematic viscosity of sea water [kg m
1
s
1
]
g = acceleration due to gravity [m s
2
]
The magnitude of the wind drift angle varies with the geographical loca
tion and wind speed and it is often estimated as 1215 degrees in the North
Sea decreasing to 0 at the equator.
9.3 Turbulent dispersion
9.3.1 Horizontal dispersion
Longitudinal and transversal dispersion
The pollutant elements diffuse in the medium. Diffusion occurs in the hor
izontal and vertical directions. An average element travels in time t
p
:
(9.23)
where S
rms,L
and S
rms,T
is the rootmeansquare distance for both the longi
tudinal and transversal direction respectively and D
L
and D
T
are the hori
zontal diffusion coefficients.
For any individual element, the diffusion step size S is generated ran
domly by
(9.24)


.

\

w
3
w
w
g

U

= exp
S
rms L ,
2 D
L
t
p
= S
rms T ,
2 D
T
t
p
=
  r S r R S
r
r
=
Numerics
Scientific Background 71
where is the random number in the interval r to r. The value of r is
chosen so that S
rms
must be equal to the mean square of all values of S.
Since the random number varies from r to r and
(9.25)
then the distance that any element travels by diffusion is
(9.26)
Neutral dispersion
The neutral dispersion is given by
(9.27)
where D
o
is the neutral dispersion coefficient.
9.3.2 Vertical dispersion
The vertical dispersion plays an important role in sediment calculations,
where it is a dominating factor for the vertical transport of the sediment.
This topic is dealt with in section 10.1.3.
Vertical dispersion is also an issue in oil spill calculations, but in MIKE
21&3 SA it is treated as en entrainment process, see section 11.3.
9.4 Numerics
9.4.1 Bilinear interpolation of current field
The current motion is in most cases obtained from a MIKE 21 or a MIKE
3 HD output. In these current files the velocities are stored a the grid
points with a given time frequency. As the particles in a Lagrangian dis
crete parcels model most of the time will be situated at offgrid points a
good interpolation procedure is needed.
R  
r
r
3
1
1
1
2
=
(
dS S
S
L
R  
1
1
6 D
L
t
p
= S
T
R  
1
1
6 D
T
t
p
=
 
p o
oy
ox
t D R
S
S
=
)
`
6
1
1
Advection and Dispersion
72 M3&M21 PA/SA
In order to interpolate the velocities in space the bilinear interpolation is
chosen. The twodimensional version of the bilinear interpolation scheme
is shown in equation (9.28)
(9.28)
where F
1
, F
2
, F
3
and F
4
are known grid point values and x ,y are dimen
sionless distances, see Figure 9.3.
A similar procedure is applied in 3D. In time a linear interpolation is
adopted.
Figure 9.3 Bilinear interpolation (2D case)
9.4.2 Reflection
Reflection at surface and bottom
This issue is only relevant in PA calculations, where particles are moving
in the vertical (as opposed to SA calculation, where the particles remain
on the surface)
F F
1
F
2
F
1
( ) y F
4
F
1
( ) x
F (
1
F
2
F
3
F
4
) x y +
+ +
+
=
Numerics
Scientific Background 73
If a particle during a timestep moves across upper boundary (z
i
<0) it is
reflected by the following rule:
(9.29)
If a particle moves below the lower boundary (z
i
>h) the reflection
depends on which type of particle the simulation is run with (dissolved,
1.order sedimentation, sand or mud).
Particles in a dissolved or 1.order sedimentation simulation are per
fectly reflected
(9.30)
Particles in a sand simulation are sedimented. When they are eroded
they are given a vertical position corresponding to the depth where the
logarithmic profile equals zero.
(9.31)
Particles in a mud simulation are sedimented too. When they reenter
the calculation they are placed at the bottom
(9.32)
This makes it easy for a sand particle to escape from the bottom, as it
more easily is caught by the advection, whereas a mud particle tends to
stay sedimented.
Reflection at land points
If a particle encounters land it is reflected by the perfect reflection algo
rithm. The main property of this algorithm is that the angle of incidence
equals the angle of reflection and that no (kinetic) energy is lost at the
impact. The principle is sketched in Figure 9.4.
z
i
z
i
=
z
i
2h z
i
=
z
i
h k
n
30 ( ) =
z
i
h =
Advection and Dispersion
74 M3&M21 PA/SA
Figure 9.4 Principle of perfect reflection. The sketch shows a coastline which
is parallel to the xgridlines. In the model a land boundary will
always be parallel to either xgridlines or ygridlines.
At the end of a timestep the particle is found at the position (x
new
,y
new
),
which turns out to be on land. The particle is then reflected to the position
(x
new,ref
,y
new,ref
)
(9.33a)
(9.33b)
9.4.3 Calculation of concentration from distribution of particles
PA calculations
The concentration of suspended sediment in a PA calculation is calculated
at the end of a timestep. The horizontal position of a particle defines which
grid cell it belongs to, thus the number of particles in each grid cell can
easily be found. The concentration is obtained by dividing the total mass
of the particles in a grid cell by the volume of the water in the grid cell.
This gives a depth averaged concentration. It is possible to consider only a
layer (as opposed to the entire water column)  in this case only particles
in the specified depth are counted and the averaging is over the layer
depth.
SA calculations
The SA module only tracks particles at the surface, and the concentra
tion of oil , the oil slick thickness, is measured in [mm]. The thickness of
the slick is calculated at the end of a timestep by tallying particles in the
(x
new
,y
new
)
(x
old
,y
old
)
(x
new,ref
,y
new,ref
)
y
bound
0
x
new,ref
x
new
=
y
new,ref
2y
bound
y
new
=
Numerics
Scientific Background 75
grid cells and then for each grid cell dividing the total volume of the parti
cles present in the cell by the area of the cell.
Advection and Dispersion
76 M3&M21 PA/SA
Settling
Scientific Background 77
10 SEDIMENT PROCESSES
The vertical transport of a particle is a result of a sedimentation compo
nent and dispersion components.
10.1 Settling
10.1.1 Dissolved matter
In the simplest case the particles act as dissolved matter, i.e. the parti
cles are nonbuoyant and follow the flow without settling. If the particles
reach the bottom they are reflected instead of being settled.
10.1.2 1.order sedimentation process
The 1.order sedimentation process is designed for a PA simulation which
is run on a depthintegrated velocity field (e.g. from MIKE 21 HD), where
the logarithmic velocity profile is not activated. As the velocity is uniform
all over the depth the sedimentation criterion described in the next section
cannot be used.
The sedimentation is calculated on the basis of a settling velocity, assum
ing that the sediment is fully mixed over the vertical. The amount of sedi
mentation at a given time and place is calculated from a first order
process, using the settling velocity divided by the depth as decay parame
ter. The suspended amount of material at time step i is given as:
(10.1)
where
S
i
0 =
suspended amount of sediments at time, i [kg]
t
p
= time step [s]
k = process constant,
where
V
set
= settling velocity for the individual particle [m/s]
h = water depth [m]
( )
p
o
i i
t k S S = exp
k
V
set
h
 =
Sediment Processes
78 M3&M21 PA/SA
10.1.3 Sand and mud sedimentation process
In the vertical direction the substance changes position as a result of the
advective velocity, the settling velocity of the particle and vertical disper
sion. The vertical displacement can be expressed as:
(10.2)
where
z
i
= vertical coordinate at time, i [m]
V
set
= settling velocity of the individual particle [m/s]
w = vertical current velocity produced by MIKE 3 HD [m/s]
t
p
= time step [s]
If a sand or mud simulation is run on a 2D current field (from a MIKE 21
HD simulation) the vertical current velocity is zero, i.e. w=0.
The dispersion D
z
consists of 2 contributions:
(10.3)
where
D
v
= vertical dispersion caused by turbulence, see equation (10.4).
D
w
= dispersion caused by wind acting on the surface.
D
w
is only active over a part of the depth due to the specified 'depth of
wind influence' from the wind.
Turbulent contribution
The vertical transport of the element, S
v
, is
(10.4)
new position old position settling dispersion + + =
z
i
z
i 1
w V
set
+ ( )t
p
D
z
+ + =
w v z
D D D + =
 
p v v
t D R S =
6
1
1
Settling
Scientific Background 79
where D
v
is the vertical dispersion coefficient. The form in equation (10.4)
covers the transport of the element upward and downward where the ran
dom number varies from 1 to 1.
Settling contribution
If only an average settling velocity is known this can be assigned to all
particles (constant settling velocity). If more detailed information is
available a distribution of the settling velocity can be specified as pairs of
velocity and frequency (varying settling distribution). Then for each
particle the program generates a random settling velocity based on the
specified distribution.
If the distribution of particle size is given the program calculates the corre
sponding settling velocities using Stokes' law /7/
(10.5)
where
S = specific gravity of sediment grain
g = gravity [m/s
2
]
d
p
= particle size [m]
w
= kinematic viscosity of water [m
2
/s]
Eddy profile
In sedimentation calculations the diffusion model is violated by introduc
ing the setttling velocity V
set
. The following approach is described by Dei
gaard and Hansen, /5/.
A parabolic eddyviscosity distribution is applied in accordance with the
assumed logarithmic velocity profile, Figure 10.1. This replaces the con
stant dispersion coefficient D
v
applied in nonsediment calculations. In the
following the turbulent dispersion coefficient D
v
is taken to equal the eddy
viscosity , i.e the vertical dispersion coefficient varies over the depth.
(10.6)
where
( )
w
p
set
d g S
V
18
1
2
t
( ) ( ) h z z U z
f t
/ 1 =
Sediment Processes
80 M3&M21 PA/SA
= von Karman's constant = 0.42
U
f
= friction velocity
h = water depth
z = vertical coordinate, (0 at the surface)
Figure 10.1 Theoretical eddy viscosity profile
This variation of the vertical dispersion coeffeicient gives rise to an artifi
cial contribution to the settling velocity, V
set,a
(10.7)
This correction ensures zero net transport of suspended sediment.
Wind contribution
The velocity distribution due to wind shear stresses on the free surface is
considered to be given by
(10.8)
where
V
set a ,
z
t
=
( ) ( ) z K V z v
s
= exp
Settling
Scientific Background 81
z = vertical coordinate, positive downwards (0 at the surface) [m]
K = velocity profile coefficient [m
1
]
V
s
= surface current velocity due to the wind [m/s]
Figure 10.2 Velocity profile due to a wind field. Vertical dispersion
The wind shear stress distribution is approximated in the following way
(10.9)
hence, considering the velocity distribution presented above, it can be
shown that
(10.10)
where
h
w
= depth of wind influence [m], user specified
= water density [kg/m
3
]
v = eddy viscosity [m
2
/s]
( ) ( )
w w
h z dz dV v z / 1 / = =
( ) ( ) ( ) ( ) =
s w w
V K z K h z z v / exp / 1
Sediment Processes
82 M3&M21 PA/SA
w
= shear stresses at the free surface [N/m
2
]
The current velocity at the surface due to the wind is by default considered
to be C
w
= 2% of the wind speed U
w
measured 10 m above datum.
The velocity distribution is then fixed by choosing
at the surface,
at z = h
w
,
which implies that K = 3/h
w
Hence
(10.11)
and noting that
(10.12)
we get
(10.13)
where
C
w
= wind friction coefficient.
Finally, the depthintegration can be performed, yielding
(10.14)
Finally, D
w
can be expressed:
(10.15)
v 0 ( ) V
s
C
w
W = =
v h
w
( ) 0,05V
s
=
z ( ) C
w
U
w
3
z
h
w

\ .
 
exp =
w
C
w
U
w
2
air
water
C
w
U
w
2
800 = =
z ( ) U
w
C
w
1
z
h
w

\ .
 
3
z
h
w

\ .
 
1
2400
 exp =
D
w
1
h
w
 t ( ) t d
0
h
w
7,447 10
4
h
w
U
w
= =
  m t D D
p w w
= 6
Settling
Scientific Background 83
Sedimentation and resuspension
The 3D module for sediment calculations includes descriptions of resus
pension for both noncohesive and cohesive materials. The following sec
tions give a brief overview of the 3D sediment specific description.
Sedimentation
Particles falling below the level where the local current velocity falls to
zero (z
0
=k
n
/30 above the bed) are assumed to be in a bed shear layer.
For noncohesive sediment, sedimented particles will remain on the bed,
provided the bottom shear stress is less than the critical shear stress for the
bed material in question, i.e.:
(10.16)
where
U
f
= friction velocity [m/s]
c
= critical Shield's parameter
S = relative sediment density [kg/kg]
g = acceleration due to gravity [m/s
2
]
d = particle fall diameter [m]
For a bottom shear stress slightly above the critical shear stress at the ero
sion threshold, particles are seen to jump along the bed. Correspondingly,
high instantaneous concentrations can be found in the bottom layer.
PLEASE NOTE:
The vertical dispersion is affecting the vertical transport of particles and
thereby the amount of particles in the bed shear layer for consolidation of
particles.
Resuspension
At higher shear stresses the possible turbulent suspension becomes larger
than the downward settling of the sediment and sediment may thus be
brought entirely into suspension. The rate of resuspension is proportional
( ) gd S U
c f
1 <
Sediment Processes
84 M3&M21 PA/SA
to the square root of the friction velocity. The limit at which resuspension
can occur is given by:
(10.17)
where
bed
=eddy viscosity at the apparent bed [m
2
/s]
V
set
= particle settling velocity [m/s]
t
p
= model time step [s]
10.2 Light Attenuation
The light extinction at the seabed caused by sediments in suspension is
described by LambertBeers law.
(10.18)
where
I
b
= light intensity at the bottom
I
s
= light intensity at the surface
K
b
= background dampening for clear water [m
1
]
h = water depth
a = light extinction coefficient [m
2
/kg]
c = concentration of suspended sediment
p set p bed
t V t > 6
( ) ( ) h c a h K I I
b s b
= exp exp
Decay
Scientific Background 85
10.3 Decay
To every particle a mass of the substance in question is attached. The mass
may change during the calculation if a decay is defined. First order decay
is included:
(10.19)
where
M = mass [kg]
k = decay factor [s
1
]
After integration the decay law yields:
(10.20)
The decay time T
90
is defined as the time where 90% of the mass is lost
(10.21)
from which the decay constant is calculated
(10.22)
t d
dM
kM =
( ) ( ) t k M t M
o
= exp
M T
90
( )
M
0
 0,1 =
k
10 ( ) ln
T
90
 =
Sediment Processes
86 M3&M21 PA/SA
Spreading
Scientific Background 87
11 OIL SPILL PROCESSES
11.1 Spreading
Oil discharged on a water surface will immediately start to increase its
surface area. Two different mechanisms cause this. The first is spreading
induced primarily by the oil properties (differences in the density of the oil
and seawater, surface tension). The sum of retardation and spreading
forces will determine the spreading rate. Fay /8/ has developed the well
known and commonly used three phase spreading theory. The three phases
are (1) the primary phase, in which only gravity (spreading) and inertia
(retardation) are important, (2) an intermediate phase, in which gravity
and viscous (retardation) forces dominate, and (3) a final phase, in which
surface tension (spreading) is in equilibrium with viscous forces.
The second mechanism is dispersion. Random motions induced by waves,
wind and tidal currents translocate the elements of the slick relative to
each other and to the centre of mass of the slick. When the slick is decom
posed to slicklets with stable surface areas, only dispersion will move the
oil elements away from the centre of mass.
Mackay et al. /14/ developed a modified gravityviscous formulation of
Fays theory for area growth based on the following assumptions:
z The oil can be regarded as a homogeneous mass.
z The slick is assumed to spread out as a thin, continuous layer in a cir
cular pattern. However, real oil slicks often have irregular shapes and
often break up into streaks.
z No loss of mass from the slick is assumed. This assumption is, of
course, not applicable to real oil slicks, where both evaporation and
dispersion are important.
With these assumptions the change of slick area, A
oil
, with time can be
expressed by
(11.1)
where:
K
a
= constant [s
1
]


.

\


.

\

A
V
A K
=
dt
dA
oil
oil
4/3
1/3
oil a
oil
Oil Spill Processes
88 M3&M21 PA/SA
t = time [s]
A
oil
= R
oil
2
[m
2
]
The oil slick volume is:
(11.2)
The initial oil slick thickness is estimated to be
h
s
=10 cm at t=0 (11.3)
It is well known that after certain time oil spilled on the sea ceases to
spread until the Pour Point of the individual oil particles exceeds the local
sea water temperature.
11.2 Evaporation
Evaporation from an oil spill is influenced by the composition of the oil,
air and sea temperatures, spill area, wind speed, solar radiation, and slick
thickness. Various attempts have been made to quantify the evaporation
rates.
The following assumptions are made:
z No diffusion limitation exists within the oil film. This is in general a
reasonable assumption of temperature above 0C and oil slick thick
ness below 510 cm.
z The oil forms an ideal mixture.
z The partial pressure of the components in the air, compared to the
vapour pressure, is negligible.
With these assumptions the rate of evaporation can be expressed as:
(11.4)
where
N
e
= the evaporation rate
s oil oil
h R V =
2
] s
m
/
m
[
X
M
RT /
P k
=
N
2 3
i
i
i SAT
i ei
e
i
Vertical dispersion
Scientific Background 89
k
e
= the mass transport coefficient
P
SAT
= the vapour pressure
R = the gas constant
T = the temperature
M = the molecular weight
X = the mole fraction
=the density of oil fraction
i = refers to the properties of component i
The estimate of k
ei
is based on Mackay et al. /14/ and is defined as
(11.5)
where
k = a constant (to be estimated)
A
oil
= the slick area [m
2
]
Sc
i
= vapour Schmidts number for component i
U
w
= wind speed [m/s]
11.3 Vertical dispersion
Transport into the water column can be accomplished by a number of
mechanisms: dissolution, dispersion, accommodation, and sedimentation.
Dispersion and accommodation are by far the most important mechanism
during the first week of weathering.
Dispersion is a mechanical process. Water turbulence tears off globules of
oil and entrains them, forming an oilinwater emulsion. These emulsions
can be stabilised by surface active agents coating the oilwater interface
and thus preventing the oil from coalescing. This formation of colloids is
called accommodation. In hard weather the dominant dispersion mecha
nism is probably wave breaking, while in calm weather the most signifi
[m/s]
U Sc A
k =
k
0.78
w
2/3
i
0.045
oil ei
Oil Spill Processes
90 M3&M21 PA/SA
cant mechanism is probably the stretching compression of the slick,
leading to droplet separation.
The fraction of sea surface dispersed in the water column per unit time is
calculated as a lost fraction of sea surface, under nonwave breaking con
ditions, and is calculated by
(11.6)
where D
a
is the fraction of the sea surface dispersed per second and D
b
is
the fraction of the dispersed oil not returning to the oil slick, expressed by
(11.7)
where
U
w
= wind speed
and
(11.8)
where
oil
= oil viscosity [cP]
h
s
= oil slick thickness [cm]
ow
= oilwater interfacial tension [dyne cm
1
]
The rate of upwelling of dispersed oil droplets is calculated by
(11.9)
D D
a
D
b
=
3600
)
U
+ (1 0.11
=
D
w
2
a
ow
s
oil
b
h
50 + 1
1
=
D
)
D
 (1
D
=
dt
V
d
b a
oil
Dissolution
Scientific Background 91
11.4 Dissolution
Assuming that the actual concentration of hydrocarbons is negligible com
pared to the solubility, the rate of dissotution is expressed by
(11.10)
where
= solubility of fraction i [mg/kg water]
= molar fraction of fraction i
M = molar weight of fraction i [kg/mol]
= density of fraction i [kg/m
3
]
= oil slick area [m
2
]
The mass transfer coefficient for dissolution is calculated by
(11.11)
where
(11.12)
11.5 Emulsification
One of the most important processes leading to the persistence of the oil at
the sea surface is the formation of a waterinoil emulsion (dispersed water
droplets in the oil), turning the oil into a very viscous mixture. Such an
emulsion can contain up to 80% water. The ease of formation and the sta
bility are very dependent on the oil type and environmental conditions.
For instance, the stability of the emulsion is closely related to the amount
of surfactants present in the oil and the rate of water uptake is closely
A
M
X C Ks
=
dt
dV
oil
i
i
mol
sat
i i
ds
i
i
sat
i
C
mol
X
oil
A
e 10
2.36 = K
i
6
S
i
e
i
1,4 for alkanes
2,2 for aromatics
1,8 for oilfines
\

=
Oil Spill Processes
92 M3&M21 PA/SA
related to sea state. A mathematical description of this process including
all relevant parameters is not yet available. Existing emulsification models
are rather empirical by nature.
A reasonable model describing emulsification can be obtained by regard
ing the reaction as an equilibrium reaction.
(11.13)
The change in water content with time can then be expressed by:
(11.14)
R
1
is thus the rate of wateruptake. It increases with increasing tempera
ture and wind speed. This is taken into consideration by expressing R
1
as
(11.15)
where
U
w
= the wind speed
oil
= the oil viscosity
y
w
max
= the maximum water content (input)
y
w
= the actual water content
K
1
= a coefficient to be estimated (input)
R
2
is the rate of water release. It decreases with increasing content of
asphaltenes, wax and surfactants in the oil and with increasing oil viscos
ity. This is expressed by:
(11.16)
Oil water +
Waterinoil emulsion
Oil water +
Waterinoil emulsion
t d
dy
w
R

R
=
dt
dy
2 1
w
( ) y  y
)
U
+ (1
K
=
R
w w
oil
2
w
1 1
max
y
Wax
A
1
K
=
R
w
oil
S
2 2
Heat transport
Scientific Background 93
where
As = content of asphaltenes in the oil (wt %)
Wax= content of wax in the oil (wt %)
K
2
= a coefficient to be estimated
The emulsification constants K
1
and K
2
were estimated to:
[kg/m
3
] (11.17)
[kg(wt%)/s] (11.18)
The coefficients K
1
and K
2
will be estimated using experimental data from
a controlled oil spill in Haltenbanken /12/.
11.6 Heat transport
Vapour pressure and viscosity are strongly temperature dependent. It is
furthermore observed that an oil slick can become considerably warmer
than the surrounding air and sea. It is therefore relevant to establish a
model for calculation of the slick temperature. Figure 11.1 shows a heat
budget for an oil slick.
K
1
5 10
7
=
K
2
1,2 10
5
=
Oil Spill Processes
94 M3&M21 PA/SA
Figure 11.1 Heat budget for an oil slick: 1 = heat transfer between air and oil, 2 =
emitted and received radiation, 3 = solar radiation,4 = evaporative
heat loss, 5 = heat transfer between oil and water, 6 = emitted and
received radiation.
11.6.1 Heat transfer between oil and air
The heat transfer between the oil slick and the atmosphere is expressed by
(11.19)
where
(11.20)
and
Sc = Schmidt's number
T
oil
= oil temperature [ Kelvin]
T
air
= air temperature [Kelvin]
a
= density of air [kg/m
3
]
C
pa
= heat capacity of air [j/kg/C]
)
T

T
(
k A
=
H oil air
air  oil
H oil
air  oil
T

.

\

P
S
C k
=
k
r
c
0.67
air
pa
a
m
air  oil
H
Heat transport
Scientific Background 95
The Prandtl's number for air is calculated by
(11.21)
where the heat capacity of air is given by equation (11.42). If there is no
evaporation, k
H
oilair
is simply calculated by Duffie and Beckmann /6/.
(11.22)
11.6.2 Heat transfer between oil and water
The heat transfer between the oil slick and the water is expressed by:
(11.23)
where k
H
oilwater
is the heat transfer coefficient calculated by Bird et al. /2/:
(11.24)
and the heat capacity of the water is given by equation (11.44).
The Prandtl's number of water is based on the following expression e.g.
Duffie and Beckmann /6/.
(11.25)
The Re is a characteristic Reynolds number for calculating the heat trans
fer coefficient of oilwater and is defined by
(11.26)
where
rel
is the kinematic viscosity of the oil slick.
)
T
0.003 + 8055 0.0241(0.1
C
=
P
air
a
a p
r
k
H
oil air
5,7 3,8U
w
+ =
( )
T

T k A
=
H oil water
water  oil
H oil
water  oil
T
Pr Re C
+ 0.332 =
k
2/3
w
0.5
w p
w
water  oil
H


.

\

273.15) 
T
( 0.000848 + 0.330
1
C
=
Pr
w
w
w w p w
w
oil
rel
A
4
= Re
Oil Spill Processes
96 M3&M21 PA/SA
11.6.3 Solar Radiation
The solar radiation received by the oil slick is dependent on a number of
parameters, the most important being the location of the spill, the date and
time of day, cloudiness, and the content of water, dust, and ozone in the
air.
The variation of solar radiation over a day is assumed to be sinusoidal:
(11.27)
The day starts at t
sunrise
and ends at t
sunset
t
sunrise
= time of sunrise (seconds after midnight)
t
sunset
= time of sunset (seconds after midnight)
t
sunset
can be calculated by adding the day length, T
d
, to t
sunrise
.
(11.28)
The day length is calculated by the following expression
(11.29)
where
= latitude (north positive)
= declination (the angular position of the sun at solar noon with
respect to the plane of the equator)
(11.30)

.

\

sin
max
otherwise 0,
t
< t <
t
,
t

t
t
 t
H K
= H(t)
sunset sunrise
sunrise sunset
sunrise
o t
[s]
T
+
t
=
t d
sunrise sunset
T
d
tan tan ( ) acos =
23,45 360
284 n +
365

\ .
 
sin
Heat transport
Scientific Background 97
H
0
max
is the extraterrestrial radiation at noon, calculated by Duffie and
Beckmann /6/
(11.31)
where
I
sc
= solar constant = 1.353 [W/m]
n = day number of the year
s
= sunrise hour angle, solar noon being zero, and each hour equalling
15 of longitude (mornings being positive)
and
If the sky is cloudfree, K
t
0.75; otherwise K
t
will decrease with increas
ing cloudiness.
A high fraction, a (albedo), of this solar radiation reaching the ground is
reflected. The net heat input from solar radiation will thus be:
[W/m
2
] (11.32)
11.6.4 Emitted and received radiation
The oil slick will lose and receive heat due to long wavelength emitted
radiation. The amount of heat received and lost due to radiation is easily
quantified using StefanBoltzman's law. The net amount of heat received
by the oil slick is calculated from
(11.33)
where
[W/(m
2
K)]
l
air
, l
water
, l
oil
= emissivities of air, water and oil, respectively
( ) ( ) ( ) ( ) ( ) ( ) sin sin sin cos cos
365
360
cos 033 . 0 1
12
max
0
+


.

\


.

\

+
=
s s
sc
sunrise sunset
t
n
I
t t
K
H
K
t
H
H
0

solar radiation reaching the surface
extra terrestrial radiation
 = =
1 a ( ) H t ( )
( ) ]
m
[W/
T l
2 
T l
+
T l
=
H
2 4
oil oil
4
water water
4
air air
rad
total
Boltzman constant 5,72 10
8
= =
Oil Spill Processes
98 M3&M21 PA/SA
T
air
, T
water
, T
oil
= temperatures of air, water and oil, respectively
11.6.5 Evaporative heat loss
Evaporative cooling will cause a heat loss of:
[W/m
2
] (11.34)
where
H
vi
= heat of vaporisation of component i [J/mol]
A dynamic heat balance for the slick thus gives the following:
(11.35)
where
= rate of water uptake [m
3
/s]
= amount of upwelling dispersed oil drops [m
3
/s]
C
po
= heat capacity of oil [J/kg C]
C
pw
= heat capacity of water [J/kg C]
11.7 Chemical and physical properties of the oil
11.7.1 Oil fractions
Oil is a mixture of hundreds of hydrocarbons, whose individual chemical
properties vary widely. The properties of the oil as a whole depend on the
H
vapor
N
i
H
vi
i
number of components
=
( ) ( )  
( ) ( )
( )
oil oil water poil oil
oil
pw w
water
vi i oil air oa oil water ow
oil oil water water air air
p
oil
A T T C
dt
dV
C
dt
dV
H N T T h T T h
T l T l T l H a
h C
t d
T d

.

\

+ +
+ +
+ +
=
4 4 4
2 1
1
t d
dW
water
t d
dV
oil
Chemical and physical properties of the oil
Scientific Background 99
properties of the individual constituents. Because these constituents
weather at different rates, the slicks properties will change with time.
The properties of the oil are described by dividing the oil into eight frac
tions, defined according to their distillation properties and their chemical
structur (alkane or aromatic). Table 11.1 summarises these fractions.
The most important physical and chemical properties (average) of each
fraction are listed in Table 11.2.
Table 11.1 Oil fractions
Fraction Description Boiling range
1 C
6
C
12
(Paraffin) 69230C
2 C
13
C
25
(Paraffin) 230405C
3 C
6
C
12
(Cycloparaffin) 70230C
4 C
13
C
23
(Cycloparaffin) 230405C
5 C
6
C
11
(Aromatic) 80240C
6 C
12
C
18
(Aromatic) 240400C
7 C
9
C
25
(Naphtenoaromatic) 180400C
8 Residual (incl. heterocycles) >400C
Table 11.2 Physical and chemical properties of oil fractions. The values for
vapor pressure are taken from Yong and Wang /15/, the value t is
the temperature in C.
Fractio
n
Mole
weight
[g/mole]
Density
[kg/m
3
]
Viscosit
y at
100F
[cs]
Vapor pressure
[mm/Hg]
Surface
tension
[103N/m]
1 128.0 715 0.536 10
(6.94
1417.61(t+202.17))
29.9
2 268.0 775 4.066 10
(7.01
1825.05(t+149.76))
35.2
Oil Spill Processes
100 M3&M21 PA/SA
11.7.2 Viscosity
Viscosity will increase during weathering, primarily due to evaporation
and emulsification. Furthermore, viscosity is strongly dependent on the
slick's temperature. The viscosity is calculated in three steps.
First, calculate "waterfree" oil slick viscosity at T
ref
= 100F, using the
KendallMonroe equation:
(11.36)
where
X
i
=the model fraction of fraction i.
3 124.0 825 2.000 10
(6.91
1441.79(t+204.7))
29.9
4 237.0 950 4.000 10
(6.99
1893.78(t+151.82))
35.2
5 110.5 990 0.704 10
(6.91
1407.34(t+208.48))
32.4
6 181.0 1150 6.108 10
(6.97
1801.00(t+162.77))
29.9
7 208.0 1085 3.000 10
(6.97
1789.85(t+164.56))
29.9
8 600.0 1050 458.0 0 47.2
Table 11.2 Physical and chemical properties of oil fractions. The values for
vapor pressure are taken from Yong and Wang /15/, the value t is
the temperature in C.
Fractio
n
Mole
weight
[g/mole]
Density
[kg/m
3
]
Viscosit
y at
100F
[cs]
Vapor pressure
[mm/Hg]
Surface
tension
[103N/m]


.

\

1/3
i i
8
=1 i
3
oil
T
X
=
REF
Chemical and physical properties of the oil
Scientific Background 101
Second, calculate the oil slick viscosity at the actual temperature e.g.
CONCAWE (1983).
(11.37)
where
T = temperature (K)
= is kinematic viscosity at the temperature T (Cs)
B = 3.98
Third, calculate the slick viscosity at the actual temperature and water
content, using an equation suggested by Hossain and Mackay /10/:
(11.38)
Evaporation also causes an increase in the viscosity which can be calcu
lated by
(11.39)
where
C
4
= dimensionless content in the oil [wt %]
F
e
= fraction of oil evaporated
The combined effect of emulsification and evaporation can be calculated
by the following equation, which is the sum of the two effects in differen
tial form
(11.40)
( )   ( )  
T
T
B  0.7 + = 0.7 +
ref
oil
T
oil
T
ref
log log log log log
y 0.654  1
y 2.5
=
w
w
oil
exp
( )
F C
=
e 4
oil
0
oil
exp
dt
dy
) y y  (1
2.5
+
dt
dV
V
1
C
=
dt
d
w
w w
2
oil
e
0
oil
oil
4
oil
max
,
T
10
119.3

T
0.133 + 998.73 =
C
2
air
5
air a p
10
273.15) 
T
( 3.3912
+ 1684.74 =
C
3 
oil
oil
o p
10
)
T
0.00061  (4.3684 =
C
3
w w p
F K
+
P
=
P e 1 p 0 p oil p,
y
K

P
 +
P
=
P 2 p oil p oil p water  oil p
103
A P P E N D I X A
104 M3&M21 PA/SA
AppenDIX A 105
A.1 REFERENCES
/1/ AlRabeh, A.H., 1994. Estimating surface oil spill transport due to
wind in the Arabian Gulf. Ocean Engng, Vol. 21 No. 5, pp 461465.
/2/ Bird, R.B., W.E. Steward and E.N. Lightfoot, 1960. Transport Phe
nomena. Wiley and Sons, New York.
/3/ CMFMWOS, 1985. Computer Model Forecasting Movements and
Weathering of Oil Spills. Final Report for the European Economic
Community, WQI and DHI, October 1985.
/4/ CONCAWE, 1983. Characteristics of Petroleum and its Behaviour
at Sea. Report no 8/83.
/5/ Deigaard, R. and E.A. Hansen, 1994. Simulation of Turbulent Diffu
sion of Suspended Sediment by a Random Walk Model. Prog. Rep
75, Institute of Hydrodynamics and hydraulic Engineering, technial
University of Denmark.
/6/ Duffie, J.A. and W.A. Beckmann, 1974. Solar Energy Thermal
Processes. Wiley Interscience, New York.
/7/ Engelund, F.A., Fl. Bo Pedersen, 1982. Hydraulik. Den Private
Ingenirfond, Danmarks Tekniske Hjskole, pp 6869.
/8/ Fay, J.A., 1969. The spread of Oil Slick on a Calm Sea. In Oil on
the Sea.
/9/ Gardiner, C.W., 1985. Handbook of stochastic methods for Physics
Chemistry and Natural Science. SpringerVerlag, Berlin.
/10/ Hossain, K and D. Mackay, 1980. Demoussifier  a new chemical
for oil spill countermeasures. Spill Technology Newsletter, v5, n6,
pp 154156.
/11/ Hunter, J.R., 1987. The application of Lagrangian particle tracking
technique to modelling dispersant in the sea. In Numerical Model
ling: Application to Marine Systems, ed. J. Noye, NorthHolland.
/12/ IKU  Institut for kontinentalsokkelundersgelser: The experimen
tal Oil Spill in Haltenbanken 1982, IKU Publication No. 112,
1984.
/13/ Kloeden, P and E. Platen, 1995. Numerical Solution of Stochastic
References
106 M3&M21 PA/SA
Differential Equations. Springer
/14/ Mackay, D, I. Bruis, R. Cascarenhus and S. Peterson, 1980. Oil
Spill Processes and Models. EPS, R&D Division, Canada.
/15/ Yong, W.E. and H. Wang, 1977. Modelling of Oil Evaporation in
aqueous environment. Water Research, Vol. 11 pp 879887.
107
I N D E X
Index
108 M3&M21 PA/SA
A
advection . . . . . . . . . . . . . . . . 65
C
current motion . . . . . . . . . . . . . 71
D
Dispersion . . . . . . . . . . . . . . . . 89
dispersion . . . . . . . . . . . . . . . . 70
dissolution . . . . . . . . . . . . . . . . 91
dissolved matter . . . . . . . . . . . . 77
E
Evaporation . . . . . . . . . . . . . . . 88
F
FokkerPlanck equation . . . . . . . . 63
L
Lagrangian Discrete Parcels Method 61
Langevin equation . . . . . . . . . . . 62
light extinction . . . . . . . . . . . . . 84
O
Oil fractions . . . . . . . . . . . . . . . 98
R
Reflection . . . . . . . . . . . . . . . . 72
Resuspension . . . . . . . . . . . . . 83
S
Settling . . . . . . . . . . . . . . . . . . 77
slick temperature . . . . . . . . . . . . 93
T
temperature . . . . . . . . . . . . . . . 18
V
vertical displacement . . . . . . . . . 78
Viscosity . . . . . . . . . . . . . . . . 100
W
waterinoil emulsion . . . . . . . . . . 91
wind . . . . . . . . . . . . . . . . . . . 67
wind friction . . . . . . . . . . . . . . . 17