Note: This message is displayed if (1) your browser is not standards-compliant or (2) you have you disabled CSS. Read our Policies for more information.
This is the twelfth article in a series about air quality in general and how it applies to Indiana. I am planning to publish an article every two weeks to cover air pollution topics. This article will discuss air quality dispersion modeling.
When a new major source is seeking to build or an existing major source is seeking to make a major modification in emissions, the source is obligated to demonstrate that the new emissions will not cause violations of the national ambient air quality standards. This is done by applying an air quality dispersion model. An air quality dispersion model (model) is a series of computer programs that relate emissions, topography, meteorology and plant characteristics to predict the impact on air quality.
Some of the key parameters from a facility that are needed to do modeling are: 1) emission rate (pounds per hour), 2) stack height, 3) stack diameter, 4) exit gas temperature, 5) exit velocity, 6) location of the stack in relation to property boundaries and 7) heights, lengths and widths of buildings near the stack. Predictions of air quality levels are only made for locations that are considered ambient air. If the source, has fenced off areas that are not accessible to the public, predicted air quality is only made outside of this area.
The model will also need meteorological data. This normally includes wind speed, wind direction, temperature and some other parameters. IDEM provides meteorological data for Indiana. Additional information on modeling is also found on this site. The current air quality model which is used is AERMOD. U.S. EPA has determined that this model is the acceptable model to use for most purposes. Links to this model and U.S. EPA modeling guidance can be found on the U.S. EPA site.
One key concept in dispersion is plume rise. The emissions released from a stack disperse more if they get higher into the air and can mix with clean air before they reach the ground and are breathed by people. There are three ways to increase plume rise. The first is to build a tall stack. U.S. EPA has rules that limit how tall a stack can be. The second is to have warm gas exiting the stack. On calm, cold winter days you can see the steam plume from a power plant rise very high into the atmosphere. The third is to have a high velocity of gas exiting the stack. Both of these last two methods use energy, so they cannot be increased without consideration of energy costs.
A model is run with five years of meteorological data. Modeling is conducted for each hour of the year. The idea behind running five years of data is that any worst case conditions of wind speed and direction will be covered in a five year period. The meteorological data does not need to be for the most recent five years, but should be for five recent consecutive years. Data is typically available for National Weather Service stations at major airports.
After model runs have completed, the output files are reviewed. They provide averages for the appropriate times to compare to the national ambient air quality standards. Some additional steps which may be required are to model all other sources in the vicinity of the new plant and to add background air quality which represents the level of air quality before the new source is built. If the total value, new source plus existing sources plus background, is less than the national ambient air quality standard, then the project can proceed. If not, the source will need to make modifications.
One of the concerns is how accurate is the air quality model. U.S. EPA believes that the model should be within a factor of two of the correct value. Some work by IDEM indicates that under the best conditions, AERMOD may achieve this level of accuracy. However, we believe that the model should do better and are working with U.S. EPA to do further work in comparing model estimates with real world measurements. Hopefully this will result in better model estimates.