A&WMA International Specialty Conference - DRI Desert
Transcript of A&WMA International Specialty Conference - DRI Desert
Applications of Accidental Release Modeling for Environmental Risk
Assessment and Emergency Response in China
Hui Guo, PhD
Chengzhi Wu, Weiping Dai, Lixian Dong
May 12, 2010Xi’an, China
trinityconsultants.com
A&WMA International Specialty Conference
Background
� 14,742 environmental pollution
accidents occurred from 1998 to
2006 in China; air pollution
accidents accounted for 34.4%.
� Emergency response plan and
environmental risk assessment are
now required in China.
� Both regulatory and technical
guidance on the subject is not as
comprehensive as needed.
� The role of air dispersion modeling
for accident release.
Objective of this Study
� Compare Chinese technical guidance of Risk
Management Plan (RMP) with that in the U.S.
� Discuss the applicability of four refined models to
Chinese RMP guidance.
� Introduce different levels of accidental release
models from relatively easy-to-use model to
sophisticated models.
Comparison of Regulatory and Technical Guidance in China and in the U.S.
� Accidental Release Prevention Program, also referred
to as the Risk Management Program (RMP) rule
released by U.S. EPA
� Hazard assessment
� Prevention program
� Emergency response
� Technical Guidelines for Environmental Risk
Assessment on Projects
� Current version– release in 2004
� New proposed version – proposed in 2009
Table 1. Detailed Comparison between Chinese and U.S. RMP Guidance (1)
Parameter
U.S. EPA RMP GuidanceNewly Proposed Chinese RMP
Guidance
Worst Case Scenario Alternative ScenarioMaximum Credible Accident
Scenario
Meteorological
Condition
Mete data source
The past three years’ mete data
at the site or from the nearby
mete station
The past three years’ mete data at the
site or from the nearby mete station
One year’s mete data in the past
three years at the site or from
station
Wind Speed/Stability 1.5 m/s and F stability
Most frequent stability class; the
most frequent non-calm wind speed
in that stability classConduct air dispersion modeling
by using whole year mete data to
get the worst hourly
meteorological condition.Ambient Temperature
/Humidity
Highest daily maximum
temperature and average
humidity
Average temperature and humidity
Wind Direction All directions (circle) All directions (circle) Wind direction at that hour
Topography Surface roughness
For obstructed terrain, use
urban – 1meter; For flat terrain,
use rural – 0.3 meter
For obstructed terrain, use urban –
1meter; For flat terrain, use rural –
0.3 meter
Not mentioned
Source
Release Quantity Largest quantity Reasonable quantity Reasonable quantity
Release Duration
10 minutes for gas release;
instantaneous spill for toxic
liquids
Be estimated based on the length of
time it would take to stop the release
5-30 minutes for release
15-30 min for pool evaporation
Height of release Ground level release Any appropriate release height Any appropriate release height
Temperature of released
substance
Highest daily temperature or
process temperature, whichever
is higher; For gases liquefied by
refrigeration, use chemical’s
boiling point.
Process or ambient temperature that
is appropriate for the scenario
Process or ambient temperature
that is appropriate for the scenario
Mitigation Mitigation system Passive mitigation system Passive and active mitigation system Not mentioned
Table 1. Detailed Comparison between Chinese and U.S. RMP Guidance (2)
Parameter
U.S. EPA RMP GuidanceNewly Proposed Chinese RMP
Guidance
Worst Case Scenario Alternative ScenarioMaximum Credible Accident
Scenario
Endpoints
Endpoints for toxic substances Be specified in the Regulation Be specified in the Regulation LC50 and IDLH
Endpoints for flammable
substances
Overpressure of 1 psi for vapor
cloud explosions
� Overpressure of 1 psi for vapor
cloud explosions;
� Radiant heat of 5kW/m2 for 40
seconds for fireballs or pool fires
� Lower flammability limit (LFL) for
vapor cloud fires.
Not required
Models
Dispersion model
Any appropriate model, including
plume models, puff models and
dense gas dispersion model
Any appropriate model, including plume
models, puff models and dense gas
dispersion model
Gaussian puff model, adjusted
puff model for dense gas
modeling
Fire model N/AVapor cloud fire, pool fire, jet fire and
BLEVENot required
Explosion model
Vapor cloud explosion (TNT-
equivalent model or other models
taking into confinement of the
vapor cloud)
Vapor cloud explosion (TNT-equivalent
model or other models taking into
confinement of the vapor cloud)
Not required
Modeling
Results
Maximum distances to the
endpoints
The geographical areas (the
circles) that could be affected
according to the endpoints
The geographical areas (the circles) that
could be affected according to the
endpoints
The geographical areas (the
circles) that could be affected
according to the endpoints
Maximum concentrations at
receptorsNot required Not required
Maximum concentration at each
receptor, time, and concentration
contour
Centerline concentrationConcentration distributions on
centerlineConcentration distributions on centerline
Maximum concentration on
centerline and the corresponding
location
Others
Incompletely combusted toxics Not required Not requiredIncomplete combustion fraction
based on empirical methods
Toxics generated from
combustionNot required Not required
SO2, CO and other toxic
substances from polymer
combustion
Application of Appropriate Models for Different Accidental Release Scenarios
Possible Release Scenarios
� One phase or two phases
� Choked or unchoked
� Cloud buoyancy
Source Term Analysis� Source Term Analysis:
� Basic equations in the technical guidelines
� Commercial software
� Accidental release of toxic aqueous solutions– heat and
mass transfer mechanisms need to be considered when
calculating emission rates
� Specific Source Term Model: SOURE5 model for
LNG Modeling
� In the U.S., computer simulations instead of the basic
equations is required to model LNG accidental releases.
� Source5 model – developed by Gas Technology Institute
(GTI) to predict the vaporization and spreading rate of
instantaneous and continuous LNG spills over land or
water.
Accidental Release Dispersion Models
Table 2. General and Specific Characteristics of Four Refined Accidental Release Dispersion Models in the U.S. (1)
INPUFF AFTOX SLAB DEGDIS
Sponsor U.S. EPA
U.S. Air Force,
supported by U.S.
EPA
U.S. Department of Energy
(DOE), supported by U.S.
EPA
U.S. Coast Guard,
U.S.EPA and Gas
Research Institute (GRI)
Algorithm Gaussian puff Gaussian puff
Conservation equations of
mass, momentum, energy
and species. The
conservation equation are
spatially averaged to treat
the cloud as a steady state
plume, a transient puff or a
combination of the two
depends on the release
duration
Jet plume model (Ooms
model) first, then dense
gas dispersion model is
used to model the
subsequent dispersion
when jet plume reaches
ground level
BuoyancyNeutrally/positively
buoyant gas
Neutrally/positively
buoyant gasDense gas Dense gas (or aerosol)
Source TypeSingle or multiple
point source
Evaporating pool,
or point source
Evaporating pool, elevated
vertical jet release, elevated
horizontal jet release and
instantaneous volume source
Ground-level area
source dense gas (or
aerosol) clouds, vertical
jet release plume
Release type
Finite duration, or
continuous plume
from a stack
Instantaneous, finite
duration and
continuous release
Instantaneous, finite duration
and continuous release
Transient (not for jet
release), finite duration
and continuous release
Table 2. General and Specific Characteristics of Four Refined Accidental Release Dispersion Models in the U.S. (2)
INPUFF AFTOX SLAB DEGDIS
Emission RateNot calculate, support
variable emission rates
Calculate the
evaporation rate of
liquid pool internally
Not calculate Not calculate
Mete Data
Constant mete condition
or variable mete
conditions for each mete
period with the same
length
Constant mete
condition for one run
Constant mete condition
for one run
Constant mete condition
for one run
Deposition/Decay Yes No No No
Output
Concentration vs. time
for a given receptor; and
puff trajectories after
each simulation period
Maximum distance to
specified endpoints at
each time step;
Concentration at
specified location and
time;
Concentration at specific
location at each time step;
time averaged maximum
concentration at different
downwind distance
Concentration field at
specified times;
concentration time
histories at specified
positions.
Other Strengths
Handle time-dependent
release rates, spatially
and temporally variable
wind field and moving
point source
Calculate 90%
confidence interval
(CI) for toxic corridor;
Also can simulate
neutrally buoyant release
including lofting of the
cloud if it becomes
lighter-than-air
Specified by U.S.GTI to
conduct LNG accidental
modeling combined with
SOURCE5 model;
account for heat and
water transfer
Limitations No consideration of chemical reaction, no complex terrain and no building downwash effects
Accidental Release Dispersion Models
� INPUFF & AFTOX – Gaussian Puff Model,
Neutrally/Positively Buoyant Gas
� Most applicable to Chinese newly proposed RMP
guidance.
� INPUFF – more suitable for modeling toxic gas or soot
generated from fire or explosion
� AFTOX – more suitable for modeling liquid evaporation
pool
� DEGADIS & SLAB – Dense Gas Model
� DEGADIS – vertical jet release
� SLAB – horizontal & vertical jet release
� Important – Concentration Averaging Time
INPUFF
AFTOX
SLAB
DEGDIS
Application of Accidental Release Models
� Easy-to-use models
� Aloha
� RMP*Comp
With a chemical database, simple source term
calculation module and simple dispersion, fire and
explosion models
� Refined models
� INPUFF, AFTOX, SLAB, DEGADIS etc.
� Sophisticated models
� CHARM (Complex Hazardous Air Release Model)
CHARM� A sophisticated 3D Eulerian grid model developed by
Mark W.Eltgroth.
� CHARM can treat: complex terrain (e.g. valleys);
complex meteorological condition (e.g., mountain-valley
flows, sea breezes); aerodynamic effects of nearby
buildings (building downwash) and chemical reactions.
� CHARM can calculate dispersion of dense gas, neutrally
buoyant and positively buoyant gas from all kinds of
release, and also predicts thermal radiation and
overpressures caused by fire and explosion.
� Grid wind field with multiple met sites, Convective heat
transfer, gas deposition and liquid flow over terrain.
CHARM
Summary
� When dispersion modeling analyses are
conducted for RMP, accurate source parameters
and appropriate models are important.
� This study provides discussions on consideration
and selection of appropriate models for accidental
release scenarios.
� Regulatory RMP requirements in China and the
U.S. are compared and discussed.
Contact Information
� Ms. Hui Guo, Trinity Consultants, China
86-0571-28828368
� www.trinityconsultants.com