Toxic release/Dispersion

Models
Instructor: Dr. Tamaghna Chakraborti
Ph. no. : +91 9892770980
Outline
• Introduction
• Parameters affecting dispersion
• Neutrally buoyant dispersions
• Pasquill-Gifford model
• Dense gas dispersions
Introduction to dispersion modeling
• Accident  process equipment can release toxic material
• Explosive rupture of a process vessel due to runaway reaction
• Rupture of pipeline containing toxic materials at high pressure
• Rupture of tank containing toxic material stored above its boiling point
• Rupture of train/truck transportation tank following accident
• Importance of toxic release model : Access the effects of the release
incident on the plant and neighboring community environments
• Excellent safety program : Identify problems before they occur, hence
it requires a toxic release model
• Important part of the consequence modeling procedure
Consequence modeling
• Third step of the consequence
modeling procedure
1. Identify release incident
2. Develop a source model
3. Estimate downwind concentration using
dispersion model
• Predictions of dispersion models used
to:
• Develop emergency response plan with
the surrounding community
• Develop engineering modifications to
eliminate the source of release
• Enclosing the source of potential release
• Adding area monitors to detect leaks
Dispersion models
• Describes the air-borne transport of toxic materials away from the
accident site
• Two mechanisms of dispersion into atmosphere :
• Plume (continuous discharge of materials)
• Puff (sudden/intermittent discharge of a large amount of material)
• Maximum concentration  at the point of discharge/release
• Downwind concentrations < maximum concentration
• Turbulent mixing and dispersion of the toxic substance with air
Parameters affecting dispersion
• Atmospheric dispersion of toxic
materials depend on a wide variety of
parameters
• Wind speed
• Atmospheric stability
• Ground conditions (buildings, water,
trees)
• Height of release above ground level
• Momentum and buoyancy of the initial
material released
 Parameters affecting dispersion (cont’d)
• ↑ wind speed, plume becomes longer and narrower
• Substance carried downward faster, but also diluted faster by larger quantity
of air
• Atmospheric stability  relates to vertical mixing of air
• During day, air temperature ↓ as height ↑ encouraging vertical motion
• At night, less decrease of temperature with height, so less vertical motion
• Classified according to three stability classes :
• Unstable  Air of lower density below air of higher density
• Neutral  air temperature difference does not affect mechanical turbulence
• Stable  Temperature near ground lower than temperature at higher
elevations
• Ground conditions  affect mechanical mixing at surface + wind
profile
Parameters affecting dispersion (cont’d)
• Release height  affects ground
level concentrations
• Buoyancy and momentum of
material released  changes the
effective height of the release
• Momentum of high velocity jet will
carry gas higher than point of release
• Buoyancy takes over after that
• Positively buoyant, lift upwards
• Negatively buoyant, slump to ground
• After certain distance, gas becomes
neutrally buoyant
• No buoyant forces
Neutrally buoyant dispersion models
• Used to model the concentration downwind of a release, in case gas is
mixed with fresh air thus becoming neutrally buoyant
• Typically apply to gases at low concentrations, in the ppm range
• Two types of vapor cloud dispersion models:
• Plume
• Puff
• Typical example of plume : continuous release of gases from a
smokestack
• Typical example of puff : sudden release of a fixed amount of material
because of rupture of storage vessel
Developing the model equations
•  Consider instantaneous release of a fixed mass of material,
• Co-ordinate system fixed at the source
• Assumption: no reaction or molecular diffusion
• From material balance, the concentration of material resulting from
the release given by:

Velocity of air

Subscript ‘j’ represents summation over all possible coordinates x, y

and z
Model equations (cont’d)
• Turbulence
  causes wind velocity to fluctuate from average values
• No model exists to adequately describe turbulence
• So the following approximate procedure is adopted :
• Velocity represented as average value + stochastic quantity
Random variable

• As a result, concentration can also be expressed as average value +

stochastic quantity
Random variable

• are random variables

What are random variables?
• A random/stochastic variable is described as a variable whose value
depend on the outcome of a random phenomena
• Tossing of a coin: ‘1’ if heads, ‘0’ if tails
• Rolling of a dice
• Classically, random variables used to model phenomena where the
relationship between the effect and the cause cannot be established,
e.g. : turbulence
Dispersion modeling
•  Stochastic fluctuations occur around mean values so:

• Gives rise to the following form of the species balance equation:

• Additional equation needed to represent the turbulent flux

• Done by defining an eddy diffusivity
Dispersion modeling (cont’d)
•  The model equation for dispersion thus stands as:

• Atmosphere is assumed to be incompressible (though it is a

compressible fluid – Why??)

• Thus, the dispersion equation becomes:

Coordinate system used in dispersion
modeling
• Origin of the co-ordinate system at the
ground level below the point of
discharge
• X-axis : Centerline directly downwind
from release point
• Rotated for different wind directions
• Y-axis : Distance off the center-line in the
plane of the ground
• Z-axis : Elevation above the origin and
the plane of the ground
 Different case-studies
•  Different case studies performed
• Each case study ratchets up the complexity level than the previous one
• and , in x-direction and , otherwise  in accordance with our defined
coordinate system, for all the case-studies
• Case-study 1 : Steady state continuous point release with no wind
• Release at origin
• Free expanse of gas (no ground to reflect gas particles)
• Constant mass release rate ( constant)
• No wind ()
• Steady state()
• Constant eddy diffusivity ( for ,
otherwise, in all directions)
 Case-studies (cont’d)
• Dispersion
  equation reduces to:

• The boundary conditions are:

Continuous release rate at the origin

• The first three boundary conditions impose the condition of spherical
symmetry on the problem
• Dispersion equation becomes tractable by defining the radius:
Case studies (cont’d)
•  Transforming the dispersion equation in terms of r and imposing the
condition of spherical symmetry we get :

• The boundary conditions for solving this differential equation are :

• The solution for the differential equation is:

Case-studies (cont’d)
• 
• Case-study 2 : Puff with no wind
• Applicable conditions
• Puff release  instantaneous release of fixed mass of material
• No wind ()
• Constant eddy diffusivity (similar to case-study 1)
• Dispersion equation reduces to:
Case-studies (cont’d)
• Solution
  of the above equation requires initial and boundary conditions
• Boundary condition :

Concentration is finite at origin

• Initial condition :

• Convert the differential equation to spherical coordinates like the

previous case-study
Solution to case-study 2
•  Solution to the above differential equation in spherical coordinates:

• Solution in terms of Cartesian coordinates:

• Case-study 3 : Non-steady state continuous point release with no

wind
• Applicable conditions:
Case-studies (cont’d)
•   • Constant mass release rate ()
• No wind (average value of velocity = 0)
• Constant eddy diffusivity
• Solution derived by integrating the instantaneous solution of the
previous case-study w. r. t. time
• Result in spherical coordinates:

• And in rectangular coordinates:

Case-studies (cont’d)
• 
• As , solution tends to case-study 1
• Case-study 4 : Steady-state continuous point release with wind
• Applicable conditions:
• Continuous release ( = constant)
• Wind blowing in x-direction only ()
• Constant eddy diffusivity in all directions (similar to case-study 1, 2 and 3)
• The dispersion equation reduces to:
Case-studies (cont’d)
• 
• The boundary conditions are the same as case-study 1
• However the solution not spherically symmetric due to the presence
of convection term
• Solution:

• For a slender plume,

Case-studies (cont’d)
•  So,

• Along the centerline of the plume,

• Case-study 5: Puff with no wind and eddy diffusivity is a function of

direction
• Relevant conditions
• Puff release ( same as case-study 2)
• No wind
Case-studies (cont’d)
•   • Each coordinate direction has a different but constant eddy diffusivity ( and )
• Dispersion equation can be written as:

• Solution:

• Case-study 6: Steady-state continuous point source release with

wind and eddy diffusivity is a function of direction
Case-studies (cont’d)
•  Applicable conditions:
• Continuous release as in case-study 4
• Steady state
• Wind blowing in x-diection only
• Each co-ordinate has a different but constant eddy diffusivity
• Slender plume approximation
• Form of the dispersion equation:

• Solution for the above equation for a slender plume:

Case-studies (cont’d)
• 
• Along the centerline,

• Case-study 7: Puff with wind

• Applicable conditions:
• Puff release (same as case-studies 2 & 5)
• Wind blowing in x-direction only
• Each coordinate has a different but constant eddy diffusivity
Case-studies (cont’d)
•  Solution to the equation found by a simple transformation of co-
ordinates
• Replace existing coordinate by a new coordinate,
• New coordinate system moves with wind velocity
• Solution to the dispersion equation:

• Case-study 8: Puff with no wind and source on the ground

• Same as case 5 but with the source on the ground
Case-studies (cont’d)
•  Ground represents an impervious boundary which elastically reflects
the toxicant molecules/particles and does not absorb/adsorb
molecules
• Solution is twice that of case study 5

• Case-study 9: Steady-state plume with source on ground

• Same as case-study 6 but with the release source on the ground
• Ground  impervious boundary which elastically reflects the toxicant
molecules/particles and does not absorb/adsorb them
Case-studies (cont’d)
•  Solution to case-study 9  2× that of case-study 6

• Case-study 10: Continuous steady-state source with source at height

above the ground
• Ground  an impervious, elastically reflecting boundary
• Solution obtained by considering an image source below the ground
along with the real source
• Solution:
Pasquill-Gifford model
•  Eddy diffusivity approach  very useful theoretically but not
convenient experimentally
•  not a constant but a function of:
• Position
• Time
• Wind velocity
• Prevailing weather conditions
• Solution: Gaussian dispersion model/Pasquill-Gifford model
• Assumption: Gaussian distribution of toxicants around the mean value
of position
• Three dispersion coefficients defined and
Pasquill-Gifford (cont’d)
• Dispersion coefficients  a function of:
• Atmospheric conditions
• Distance downwind from the release
• Atmospheric conditions classified according to six stability classes:

A  extremely unstable
B  moderately unstable
C  slightly unstable
D  neutrally stable
E  slightly stable
F  moderately stable
Dispersion coefficients
• Gaussian
  distribution:

• and tabulated as functions of downwind distance for both puff and plume
models
• Tabulated functions different for different stability classes of the atmosphere
• Case-studies 1 – 10 re-derived by Pasquill using the dispersion coefficients
• Re-derived equations along with correlations for dispersion coefficients
collectively called Pasquill-Gifford model.
 Different case-studies for Pasquill-Gifford
model
•  Case-study 11: Puff with instantaneous point source at ground level,
coordinates fixed at release point, wind only in x-direction with constant
velocity
• Case-study identical to case-study 7 but with twice the concentration on
account of impenetrable ground
• Concentration term:
• Compare with Gaussian distribution
• and similarly for dispersion along other directions
• So, according to Pasquill-Gifford model,
Pasquill-Gifford (cont’d)
•  Also gives a measure of the total integrated dose received by an
individual standing at fixed coordinates (x,y,z)

• Total integrated dose at ground level:

• Measure of cloud boundary defined by a fixed concentration is

sometimes required
• Done by solving the equation:
Pasquill-Gifford (cont’d)
• 
• Procedure for calculating isopleth (Lines of equal concentration)
a) Specify the concentration, wind velocity and time
b) Determine the concentration along the x-axis. Define the boundary of the
cloud along x-axis
c) Calculate the isopleth with the help of the above equation
• Case-study 12: Plume with continuous steady-state source at ground
level and constant wind velocity along x-direction
• Situation identical to eddy diffusivity case study 9
Pasquill-Gifford (cont’d)
• 

• So, according to the Pasquill-Gifford/Gaussian distribution model:

Pasquill-Gifford (cont’d)
•  Case-study 13: Plume with continuous steady state source at height
above ground level and wind moving along x-axis with constant
velocity
• Case identical to case-study 10
• Pasquill-Gifford distribution again found by comparing the eddy
diffusivity distribution function with the Gaussian distribution
function

 
2𝐾𝑦𝑥 2𝐾 𝑧 𝑥
𝜎 𝑦↔
√𝑢
,𝜎 𝑧 ↔
𝑢√
Pasquill-Gifford (cont’d)
• 
• Ground level concentration obtained by setting

• Ground level centerline concentration obtained by setting

Pasquill-Gifford (cont’d)
•  Maximum ground level concentrations determined using the
formulae:

• Use the first equation to compute the downwind distance and hence,
calculate the maximum concentration
Limitations of Pasquill-Gifford model
• Applies only to neutrally buoyant dispersions of gases
• Turbulent mixing dominant feature
• Typically valid 0.1-10 km from the release point
• Predicted concentrations are time averages  possible for
instantaneous local values to greatly exceed average values
• Important if emergency response is to be decided
• Models presented here assume 10 min time average
Dense gas dispersions – Britter & McQuaid
model
• Dense gas  Any gas whose density is greater than the density of
ambient air
• Reason: Contaminant may be a gas with molecular weight greater than that
of air
• Contaminant may also be a gas at low temperature resulting from auto-
refrigeration during release (Joule-Thomson effect)
• A puff release  formation of a cloud having similar dimensions 
slumps to the ground under gravity, diameter ↑, height ↓ 
considerable initial dilution because of gravity driven intrusion into
ambient air  subsequent increment in height due to entrainment of
air  neutrally buoyant after sufficient dilution
Britter and McQuaid model
• Empirical development by performing a dimensional analysis and correlating
existing data on dense cloud dispersions
• Best suited for instantaneous or continuous ground level releases of dense
gas
• Release assumed to occur at ambient temperature (no heat transfer effects)
and without aerosol/liquid droplet formation
• Little impact of atmospheric stability so this is not a part of the model
• Results obtained mostly on flat terrain  Model not applicable to areas
where terrain effects are significant
• Input to model:
• Initial cloud volume/plume volume flux
• Duration of release
• Initial gas density
Britter & McQuaid (cont’d)
•  Step 1: Determine whether the dense gas model is applicable
• Initial cloud buoyancy defined as:
Density of ambient
air (mass/volume)
  Initial buoyancy factor ()
Initial density of
 Acceleration due to released material
gravity () (mass/volume)
 •
Define a characteristic source dimension
• For continuous release,
  = Initial plume volume flux (volume/time)
• For instantaneous release,
= Wind speed at 10 m elevation (length/time)
= Initial volume of released material
 Britter & McQuaid (cont’d)
•  Criteria for a sufficiently dense cloud to require dense cloud
representation:
• For continuous release,
• For instantaneous release,
• If the above criteria are satisfied, go to Step 2
• Step 2: Estimate the downwind concentrations using the figures given
in the next slide
• Different criteria for continuous and instantaneous releases
• Criteria for determining whether release is continuous/instantaneous
Duration of release (time)
𝑢  𝑅 𝑑 ≥2.5 , Release considered continuous

Downwind distance (length) 𝑥 { ≤0.6 , Release considered instantaneous

In − between Maximum of instantaneous & continuous models
Estimation of downwind concentrations for
dense gas dispersions

For dense gas plumes For dense gas puffs

Limitations of the Britter & McQuaid model
• Dimensional analysis technique  based on correlation developed
from experimental data
• Correlated data obtained from flat rural terrain, so unable to predict
concentrations where terrain effects are significant
• Unable to a/c for effects of parameters such as:
• Release height
• Ground roughness
• Wind speed profiles
Problem
On an overcast day a stack with an effective height of 60 m is releasing
sulfur dioxide at a rate of 80 g/s. The wind speed is 6 m/s. The stack is
located in a rural area. Determine
a) The mean concentration of sulfur dioxide on the ground 500 m
downwind.
b) The mean concentration on the ground 500 m downwind and 50 m
crosswind.
c) The location and value of the maximum mean concentration of the
ground directly downwind.
Solution
• Continuous
  release
• Ground concentration directly downwind given by:

• How to know the dispersion coefficients?

• Function of:
• Atmospheric stability
• Downwind distance
• Overcast conditions during daytime:
• Stability class: D
• For D stability class in rural conditions:
Solution (cont’d)
• 

• All units are in m

• Calculate

• Compute the concentration

• For the second part of the problem:
Solution (cont’d)
• Compute
  the concentration

• Maximum concentration is found at:

• Compute the downwind distance from the above formula using trial
and error
• Compute using the value of the downwind distance
• Use formula to get