New Techniques for Modeling Air Quality Impacts of DoD Activities

1
New Techniques for Modeling Air Quality Impacts
of DoD Activities

• Talat Odman and Ted Russell
• Environmental Engineering Department
• Georgia Institute of Technology

2
Abstract

• This is a feasibility study to determine if two
  recently developed techniques can improve the
  accuracy of regional-scale air quality models to
  a level where they can be used to assess the
  impacts of DoD activities. Current large-scale
  models cannot differentiate a DoD facility from
  its surroundings on the other hand, small-scale
  models cannot track the long-range impacts.
• The first technique, adaptive grid, dynamically
  reduces the grid size to better resolve the
  evolution of the plumes from the source to the
  receptor.
• The second technique, direct sensitivity
  analysis, efficiently yields the sensitivity of
  pollutant levels to emissions from various
  sources it can accurately detect the effect of
  even very small sources.
• In this study, we will target the prescribed
  burning emissions from Fort Benning and their
  impact on the air quality of the Columbus
  metropolitan area. In particular, ozone air
  quality standards may be exceeded in Columbus if
  there is significant impact of NOx and VOC
  emissions from the controlled fires.

3
Adaptive Grid Modeling and Direct Sensitivity
Analysis for Predicting the Air Quality Impacts
of DoD Activities
4
Adaptive Grid Algorithm

• Inadequate grid resolution may be an important
  source of uncertainty in air quality models.
  Adaptive grids offer an effective and efficient
  solution.
• Our adaptive grid technique is a mesh refinement
  algorithm where the number of grid cells remains
  constant and the structure (topology) of the grid
  is preserved.
• A weight function controls the movement of the
  grid nodes according to user-defined criteria. It
  automatically clusters the nodes where resolution
  is most needed.
• Grid nodes move continuously as shown in the
  movie of the simulation. Grid cells are
  automatically refined/coarsened to reduce the
  solution error.

5
Snapshot of the Continuously Adapting Grid
6
Adaptive Grid Air Quality Model

• A simulation with the adaptive grid air quality
  model can be viewed as a sequence of two steps.
• Adaptation step The concentration field is
  frozen in time while the grid nodes are moved.
  The following tasks are performed
• Computation of the weight function
• Repositioning of the grid nodes
• Redistribution of the concentration field to new
  grid node locations
• Solution step The grid nodes are held fixed
  while the concentration field is
  advanced in time. The tasks are
• Processing of the meteorological and emissions
  data
• Coordinate transformation
• Numerical solution
• The adaptation step takes a fraction of the CPU
  time required by the solution step.

7
Ozone Simulation in Tennessee Valley

• We simulated ozone air quality in the Tennessee
  Valley for the July 7-17, 1995 period using
  the same model for physical and chemical
  processes, but with different grids
• Two static grids with uniform resolutions of 8 km
  and 4 km, and
• An adaptive grid that has the same number of
  cells as the 8-km resolution static grid.
• Meteorological inputs were prepared using the
  Regional Atmospheric Modeling System (RAMS) at
  4-km resolution.
• Emission inputs were prepared using the SAMI
  emissions inventory.
• There are over 9000 point sources in this domain
  including some of the largest power plants in the
  Unites States.
• Area sources are mapped onto an 8-km emissions
  grid.
• The adaptive grid is adapted to surface layer NO
  concentrations through the simulation.
• The cell size may drop to few hundred meters from
  the original 8 km.

8
Outstanding Resolution of NOx Plumes
Fixed Grid
Adaptive Grid
9
Superior O3 Predictions with Adaptive Grid
Sumner Co., TN
Graves Co., KY
10
Sensitivity Analysis with Direct Decoupled
Method (DDM)

• In DDM, sensitivity is defined as the first
  derivative (local slope) of the pollutant
  concentration with respect to an emission.
• The equations for sensitivities are similar to
  the equations governing pollutant concentrations.
  Therefore, they can be solved efficiently.
• Several sensitivities can be calculated
  simultaneously in a single model run, along with
  concentrations.
• Assuming linear response, the change in
  concentration resulting from a change in
  emissions can be approximated as
• For small changes in emissions, not only this
  assumption is valid but DDM is more accurate than
  the brute-force response. So, DDM is ideal for
  regional impact assessment of smaller sources.

11
PM2.5 Simulation in the Southeast

• Using an integrated air quality model, we
  simulated the future PM2.5 levels in the eastern
  United States focusing on Class I areas of the
  Southern Appalachian Mountains.
• Using DDM in these simulations, we calculated the
  sensitivity of PM2.5 levels to
• SO2, elevated NOx, ground-level NOx, and NH3
  emissions
• from 8 states in the Southeast and 5 surrounding
  regions.
• Some of these sensitivities are shown below.
• DDM sensitivities can be used in designing
  effective emission control strategies.
• More information about this study can be found at
  http//environmental.gatech.edu/SAMI

12
SO4 on July 15, 2010 and its Sensitivity to 10
Reductions of SO2 Emissions from 8 States
13
SO4 Sensitivity to SO2 Emissions at 10 Class I
Areas
14
Conclusion

• Currently, we are merging adaptive grid and
  sensitivity analysis techniques in a
  comprehensive air quality model and preparing
  emissions data for the simulations. Here, we
  presented results from recent applications of the
  two techniques.
• The adaptive grid was used to better simulate the
  fate of power-plant plumes in the Tennessee
  Valley during an ozone episode in July, 1995.
  When compared to observations, the adaptive grid
  produced significantly more accurate results than
  the classical static-grid models.
• The direct sensitivity analysis was used to
  determine the impact of emissions from
  neighboring states and regions to the air quality
  in the Southern Appalachian Mountains. For
  different receptor areas, the adverse
  contribution of each state or region was ranked.
  This information can be used in the design of
  emission control strategies.
• These results show that both techniques have
  upside potential for use in impact assessments of
  DoD facilities and operations.

15
Related Publications

• Srivastava, R. K., McRae, D. S. and Odman, M. T.,
  Simulation of dispersion of a power plant plume
  using an adaptive grid algorithm, Atmospheric
  Environment, vol. 35, no. 28, pp. 4801-4818,
  October 2001.
• Srivastava, R. K., McRae, D. S. and Odman, M. T.,
  Simulation of a reacting pollutant puff using an
  adaptive grid algorithm, Journal of Geophysical
  Research, vol. 106, no. D20, pp. 24245-24258,
  October 2001.
• Srivastava, R. K., McRae, D. S. and Odman, M. T.,
  An adaptive grid algorithm for air quality
  modeling, Journal of Computational Physics, vol.
  165, no. 2, pp. 437-472, December 2000.
• Yang, Y.-J., Wilkinson, J. G. and Russell, A. G.,
  Fast, direct sensitivity analysis of
  multidimensional photochemical models,
  Environmental Science and Technology, vol. 31,
  no. 10, pp.2859-2868, October 1997.