Examples and good practice

Industrial Air Quality Management

Presented by:
David Segersson
SMHI (Swedish Meteorological and Hydrological Institute)
[email protected]
Scope
Air Quality Management
Components

Incidental releases of harmful substances - risk assessment

Characteristics
Available methods
Recommendations

Meteorological measurements
Monitoring
AQM-systems
Recommendations
Air Quality Management
Objective
Limit the impact on environment and human
health caused by the emission of harmful
substances

Tools
Monitoring
Emission inventories
Dispersion models
 Detailed
Dispersion modelling mapping
Mapping of air quality makes it
possible to address:
- Population exposure
- Health impact
- Economical impact

Scenario calculations
- Evaluating measures
- ”What if?” scenarios

Air Quality Forecasts

Environmental Impact Assessments
Identifying the sources of air pollution

Impact assessment
Receptor modelling
Statistical analysis of modelling results
and monitoring data
Results in correlation maps showing
Position of emissions
Kohtla-
Emission intensity Järve
station
Could be used together with fence-line
monitoring, e.g. to estimate H2S
emissions and provide input for Liikuv-
scenario calculations Öhulabor
station

Emission localisation
Kohtla Järve, Oil shale industry, Estonia
 Incidental releases of harmful substances
- Characteristics
Sudden event, fast course
Short distances - mainly affects the people within the plant and right outside
Large emissions due to tank bursts, ruptures of pipes etc.
Substances are often more dense than air
Quantity of emissions are often very uncertain
Dispersion is difficult to predict due to turbulence
Wind speed [m/s]

Time
Meteorological measurements
Measured wind direction can be used directly as an indication on in which
direction emissions are dispersed
Detailed met data is required for dispersion models
Important to place meteorological mast in a undisturbed and representative
location
A meteorological mast should be at least
10 m.
Placed away from large buildings and trees
Example of wind flow pattern around
buildings

• CFD (Computational Fluid Dynamics) can be used to find

an optimal location

• Example of wind flow pattern around large buildings

 Meteorological measurements
Parameter Sensor
Horizontal wind A propeller anemometer of “Air Quality” approved type
(according to the US EPA)
Horizontal and vertical A three axis ultra sonic anemometer
wind at the top of the
mast
Temperature Platinum sensor PT-100 in a traditional radiation shield or
fan aspirated radiation shield.
Temperature difference A bridge configuration of thermo-couples for direct
measurement of temperature difference, mounted in fan
aspirated radiation shields.
Precipitation A traditional tipping bucket sensor
Global radiation A pyranemometer based on a photo diode with a spectral
response similar to the visual light.
Pre-processing and wind modelling
For dispersion modelling, a meteorological pre-processor is used to estimate:
- The boundary layer height ( zi )
- Wind speed profile U (z)
- Temperature profiles T(z)
- The lateral and vertical components of turbulence
- Stability parameter (Richardson number or by Obukhovs length)

A diagnostic wind model can be used to estimate a wind field from a number of
meteorological masts

Ideal configuration in areas with varying winds:

- One principal mast with complete instrumentation
- Complementary measurements of wind speed and wind direction at different
locations
Dispersion scenarios
Basis for decisions regarding evacuation etc.
Define different emission scenarios
Define meteorological “worst case”
- Low wind speeds
- Wind towards populated areas
- Stable atmospheric conditions - low mixing
- Unstable atmos. conditions – high mixing

+ Simple
- Innumerable possible scenarios
- rough - easy to overreact
Example – scenario calculations at Nynas
refinery in Sweden
2 worst case scenarios
- Large SO2 emissions from flare
- Pipe-burst

Objectives
- Updated radiuses of limit values on
site map
- Determine if the plume from the flare
can reach ground level during
incidents
- Determine time for plume to reach
nearby power plant
- Determine maximum concentrations
at nearest neighbour Nynas refinery, Nynäshamn, Sweden
- Increased understanding of risks
 Emission scenarios – pipe burst example
0.2
0.18
Release rate [kg/s]

0.16
Unlimited
0.14
0.12
0.1
0.08
0.06 Limited
0.04
0.02
0
0 4 9 13 17 22
Seconds
If no action is taken Hydrogen Sulphide gas will flow continuously

When the valve on the outlet pipe is closed, the release becomes limited
and the pressure in the system will decrease
 Threshold concentrations evaluated for
H2S at Nynas refinery
Conc. (ppm) Type of threshold Description
0.03 Olifactory (smell) threshold Very individual (0.008-0.2 ppm)
15 TGV Swedish threshold for short-term exposure (15 min)
30 ERPG-2 Max conc. during 1 hour not causing serious health
effects
100 ERPG-3 Max conc. during 1 hour not causing life-threatening
health effects.

600 Immediate risk of death (250 pulmonary oedema)

Summary of exposure threshold concentrations for H2S, at:

http://www.ivhhn.org/index.php?option=com_content&view=article&id=83
Dispersion modelling
Preliminary modelling for torch using a standard Gaussian plume model
Special consideration of plume-lift from flare due to high temperatures

State-of-the-art modelling performed for a few selected cases

- Large Eddy Simulation used to describe the wind flow field
- Density differences caused by temperature and gas concentrations included

Visualization of SO2 plume from torch, odour threshold

Air Quality Management systems?
Using professional tools for data-collection, validation, analysis etc. improves
quality and is an assurance for more correct conclusions

Makes quality control and analysis more efficient and let you focus on the
conclusions

Provides a platform that can be gradually improved or modified, according to

evolving requirements and ambitions

By combining tools and data from different sources, it makes more advanced
analysis possible:
Forecasts, dispersion modelling, people exposure calculations, real-time
dispersion modelling, inverse modelling, receptor modelling
Web publishing, ,
AQM-systems - incident scenarios
For use during incidents – everything
needs to be running!
Combines real-time meteorological
measurements with emissions to run
dispersion models
Continuous or predictive emission
monitoring can be used to feed the
dispersion model with detailed information
regarding emissions
Can raise alarms, and results showing
areas with concentrations above
thresholds

Fictive scenario for continuous

gas leak
PETROX SA, Talcahuano, Chile
AQM-system - early warnings
Exchange of data with Plant Information System
Data from Continuous Emission Monitoring Systems is connected to the
emission database to provide emissions updated in real time
Air Quality Monitoring
Desirable components of a fully functioning monitoring network:
Data available in real-time
Data validation (automatic and manual)
Data easily accessible for the public, for authorities and the private sector
Verified evaluation of Air Quality Standards, using only validated data
Automatic alarms for data collection problems, unrealistic values etc.
Stable technical solutions to minimize down-time and data gaps
Low operational costs
Minimal hardware requirements – less maintenance
Good practice guidance
- General
High quality meteorological monitoring

Real-time monitoring of main air pollutants in ambient air

Up-to-date emission inventory, if possible including continuous emission

monitoring or predictive emission monitoring

Impact assessment including population exposure to contaminants

Dispersion models used as a tool to analyse scenarios

Data-collection, validation and presentation of monitoring data easily available

for any user according to defined privileges

The emission inventory should be verified by comparison between dispersion

modelling results and monitoring data
Good practice guidance
- Incidents
High quality meteorological monitoring

Emission scenarios defined for sources causing incidents

Impact assessment for worst case scenarios

Gas detectors

Real-time dispersion modelling in combination with Continuous Emission

Monitoring or Predictive Emission Monitoring
- Alarms connected to emission monitoring
- Dispersion results shows expected concentrations and exposed areas

Fence-line monitoring
Good practice guidance
- Scenario calculations
Define worst case regarding emissions and meteorology

Define your objectives well and plan how you will use the scenarios

For most critical scenarios, use CFD (Computational Fluid Dynamics) with
turbulence described by Large Eddy Simulation to get a more realistic example
of the expected dispersion.

Use a simpler (faster) model to model different meteorological scenarios and

emissions. This will give an understanding of the variability, and provide a
knowledge bank for future situations.

If possible, using the local Air Quality Management system will make it easier to
reproduce and update the scenarios – not updated scenarios quickly become
irrelevant