 |
||
| Scientists' Contributions | ||
USE OF DISPERSION MODELS FOR ASSESMENT OF ACCIDENTAL GAS RELEASES
JOAO GOMES, Chem Eng, BSc, MSc, PhD
CENTRE FOR ENVIRONMENTAL TECHNOLOGIES
INSTITUTO DE SOLDADURA E QUALIDADE
OEIRAS, PORTUGAL
Email: jpgomes@isq.pt
-
Contents of the document
Abstract
Dispersion models describing the effect of sudden gas releases into the atmosphere are quite useful in the prediction of effects and analysis of the magnitude of industrial accidents, and therefore represent an important tool for risk analysis of industrial plants and installations.
This work describes some of the dispersion models currently used by ISQ in relation
with this subject as well as some studied cases and industrial applications. The models used in these
studies are numerical gaussian models approved by USEPA as part of the UNAMAP series.
1. Use of dispersion models
Since long the prediction of pollutants concentration has been estimated by
empirical formulae or by numerical models. Dispersion phenomena in the atmosphere were studied by
Pearson, Pasquill and Sutton (1) which formed the theoretical basis for a series of analytical
expressions describing dispersion of pollutants that could be sequentially organised resulting thus in
dispersion models. Some of these dispersion models are based on the transverse dispersion of the plume
corresponding to the repartition of pollutant concentrations accordingly to a gaussian field and
therefore are referred to as gaussian models. Following an axis defined by the source (the point where
the pollutants are released to the atmosphere) and the wind direction, a field very close to the source
where the concentration of pollutant is nil appears. From a very short distance, the concentration of
pollutants suddenly rises quickly, reaches a peak and decreases afterwards with the distance from the
source. Concerning the transversal dispersion a maximum of concentration is obtained within the axis
referred before and a decay according another axis, usually, normal to the first one, following a Gauss
curve.
The analytical expressions which form the solution of the dispersion equation allow
the determination of the pollutant concentration in a point defined by the coordinates (x,y), and are
related with the formula developed by Sutton (1) :
where Q is the volumetric flow rate of pollutants
being released by the source ; H the effective height of source, comprising plume elevation;
Un the mean velocity of release ; a parameter n for
accounting turbulence and Cy and
Cz as the coefficients for lateral and transversal dispersion,
respectively.
In fact, these models are the solution of the dispersion equation using analytical
or iterative techniques based on certain assumptions for the dispersion coefficients. Calculations are
usually made considering different meteorological conditions and for effective concentrations on
emission.
The use of these models is one of the less tedious and less expensive methods for
predicting the concentration of pollutants released to the atmosphere by continuous or discontinuous
processes. It is also a precise method, depending on model assumptions and considerations.
2. The segmented plume model
The Gaussian steady-state formula described before is valid during transport
conditions in fairly stationary and homogeneous situations. In order to treat time-varying transport
conditions and, especially, changes in wind direction, several authors (e.g. Hales at al.
(2); Bankley and Bass (3); Chan et al. (4))
have developed and used segmented Gaussian plume models. In the
segmented plume approach, the plume is broken up into independent elements (plume segments or sections)
whose initial features and time dynamics are a function of time-varying emission conditions and the
local time-varying meteorological conditions encountered by the plume elements along their motion.
Segments are, in fact, sections of the Gaussian plume and each segment generates a concentration field
that is still basically computed by the Gaussian equation presented before and that represents the
contribution of the entire virtual plume passing through that segment. Therefore, only one segment (the
closest) affects the concentration computation at each receptor, except that the occurrence of a
180o wind direction change can create a condition where the
contribution of two segments, that is two virtual plumes, should be superimposed at some receptors.
3. Puff models
Puff models, such as the ones developed by Lamb (5) and Roberts
(6), have, like segmented models, been developed to treat nonstationary emissions in
non homogeneous dispersion conditions. Puff methods, however, have the additional advantage of being
able, at least theoretically, to simulate calm or low wind conditions. The Gaussian puff model assumes
that each pollutant emission of duration which is often expanded to incorporate reflection and
deposition/decay terms. It should be noted that the analytical integration of this equation in
stationary, homogeneous transport conditions give the traditional plume formula. This equation requires
the proper evaluation of the horizontal ( Several studies have discussed the puff modelling approach in detail, improving its
application features. In particular, algorithms were proposed and evaluated for incorporated wind shear
effects (7), virtual distance (8) and virtual distance (9).
4. Emission modelling of accidental spills
Puff models described before are currently used nowadays to estimate the
concentrations resulting from accidental spills. The most important parameter in the simulation of
accidental spills of hazardous materials is the "source" term, i.e., the quantitative evaluation of the
dimension, rate and duration of the spill. Source emission models are divided then in five main groups
according to the type of source and occurrence of physical phases, as follows:
5. Computer systems for chemical emergency planning
In countries where a precise description of emergency plans are mandatory for
obtaining industrial working permits such as the United States, puff models are currently used in order
to evaluate these problems. The USEPA (1989) has identified the computer systems applicable to SARA
Title III of the Superfund Amendments and Reauthorization Act of 1986, i.e., those packages that are
suitable for local planning and for assistance in emergency response planning such as hazard
identification, vulnerability analysis through modelling of the releases, risk analysis and regulatory
requirements. The main models used for simulation of accidental releases are summarised in Table 2.
6. References
C(x,y) = 2Q / (CyCz2-nUn) exp((-1/x2-n) ((y2/Cy2) + (H2/Cz2)))
t injects into the atmosphere a
mass M = Qt, where Q is the
time-varying emission rate. The center of the puff containing the mass M is
advected according to the local time-varying wind vector. If, at time t, the center of a puff is
located at p(t)=(xp,yp,zp), then the concentration due to that puff at the receptor r=(xr,yr,zr) can be
computed using the basic Gaussian puff formula:
h) and vertical
(z) dynamics of each puff's growth. The total concentration in a
receptor at time t is computed by adding the contribution c from all
existing puffs generated by all sources. Note that this last equation differs from the traditional Gaussian equation
because an extra horizontal diffusion term has been substituted for the transport term, with the
consequent disappearance of the wind speed u. In other words, in a puff model, the wind speed affects
the concentration computation only by controlling the density of puffs in the region, that is, the lower
the wind speed, the closer a puff is to next one generated by the next source. Therefore, at least in
theory, a puff model can handle calm or low-wind conditions, and this approach represents the most
advanced and powerful application of the Gaussian formula.
Type of model Name of model Authors
Evaporation model - Ille and Springer
Evaporation model Army Whitacre
Evaporation model Shell Spills Fleischer
Evaporation model USAF ESL Clewell
Evaporation model AWS AWS
Evaporation model Illinois EPA Kelty
Evaporation model - Stiver and McKay
Evaporation model Monsanto Wu and Schroy
Evaporation model - Shaw and Briscoe
Jet model - Wilson
Jet/evap. model CHARM Eltgroth
Jet/evap. model Ontario MOE Moe
Jet/evap. model AIRTOX Paine
Jet/evap. model - Kunkel
Jet/evap. model DENZ Fryer and Kaiser
Jet/evap. model COBRA Alp
Acronym System name Author
ARCHIE Aut. Resource for Chemical Hazard Incident Evaluation US Dept. of Transportation
CAMEO II Comp. Aided Management of Emergency Operations US Dept. of Commerce
CARE Comp. Airborne Release Evaluation Env. Systems Corp.
CHARM Complex Hazardous Air Relesase Model Radian Corp.
MESOCHEM Chemical Atmospheric and Hazard Assessment System Impell Corp.
MIDAS Meteorological Inf. and Dispersion Assessment System Pickard, Lowe and Garrick Inc.
TECJET Advanced Jet Dispersion Model Technica International
TRACE II Toxic Release Analysis of Chemical Emissions Safer Emergency Systems Inc.
WHAZAN World Bank Hazard Analysis Technica International
 |
||
| Scientists' Contributions | ||