1. VO Biophysik | G. Schauberger | Institut für Med. Physik & Biostatistik
Reconstruction of airborne emissions by inverse dispersion modelling and generation of synthetic emission data by a Monte Carlo model
G. Schauberger1, M. Piringer2, E. Petz2, W. Knauder2 and K. Baumann-Stanzer2
1 Medical Physics and Biostatistics Department of Biomedical Sciences Veterinärmedizinischen Universität Wien
2 Central Institute for Meteorology and Geodynamics Department of Environmental Meteorology

2. VO Biophysik | G. Schauberger | Institut für Med. Physik & Biostatistik
Content
Two-step procedure to re-construct the emissions of a thermal treatment plant for waste:
Step 1: Re-calculate emissions from measured ambient concentrations leeward of the pollution source (“inverse dispersion modelling”)
Step 2: Application of a Monte-Carlo model to supplement the emission flow rates by step 1 by synthetic ones to generate emission time series over the entire year
Model performance
Summary of results and application of the method

3. Site 265 235 Source Ambient concentration
VO Biophysik | G. Schauberger | Institut für Med. Physik & Biostatistik
Site
Ambient concentration
265
235
Source

4. Step 1: Inverse dispersion modelling
VO Biophysik | G. Schauberger | Institut für Med. Physik & Biostatistik
Step 1: Inverse dispersion modelling
Wind
Stability
Camb
Emission flow rate Q Q

5. Analysis of ambient concentrations (I)
VO Biophysik | G. Schauberger | Institut für Med. Physik & Biostatistik
Analysis of ambient concentrations (I)
265°
235°
235
Mean concentration (in µg/m³) as a function of wind direction for a wind velocity v >0.8m/s

6. Analysis of ambient concentrations (II)
VO Biophysik | G. Schauberger | Institut für Med. Physik & Biostatistik
Analysis of ambient concentrations (II)
Ambient concentration time series
Intermittence factor IF
Fraction of values above a certain threshold (=0 or detection limit)
Working days and weekends are taken into account
Emission flow rate
Time

7. Results for the intermittence factor
VO Biophysik | G. Schauberger | Institut für Med. Physik & Biostatistik
Results for the intermittence factor
265°
235°

8. Step 2: Monte Carlo model
VO Biophysik | G. Schauberger | Institut für Med. Physik & Biostatistik
Step 2: Monte Carlo model
To set up the model, a data analysis (data set of step 1) is carried out to determine predictors which describe the data set appropriately: First group: Influence of meteorology (Wind velocity, air temperature) Second group: Process flow of the plant (day of the week, time of the day)
Regression analysis of the two meteorological parameters shows a high regression coefficient for the logarithmically transformed values of the emissions of the seven species
The appropriate theoretical distribution function to describe the emissions is the log-normal distribution

9. Influence of meteorological parameters
VO Biophysik | G. Schauberger | Institut für Med. Physik & Biostatistik
Influence of meteorological parameters
Air temperature
Wind velocity

10. Influence of the process flow
VO Biophysik | G. Schauberger | Institut für Med. Physik & Biostatistik
Influence of the process flow
Day of the week
Daytime
80%
20%
Example: Ethyl acetate

11. Statistics of the emissions
VO Biophysik | G. Schauberger | Institut für Med. Physik & Biostatistik
Statistics of the emissions
CDF: log-normal distribution
Butyl acetate
Benzene

12. VO Biophysik | G. Schauberger | Institut für Med. Physik & Biostatistik
Monte-Carlo Model
Input parameters
Air temperature
Wind velocity
Day of the week
Time of the day
Model parameters
Intermittence factor IF (diurnal variation for working day and weekend)
3 x 3 matrix for air temperature and wind velocity with mean value and variance
Relative diurnal variation of mean value and variance for all days
Monte-Carlo Model
Evenly distributed random number RN1
log-normally distributed random number RN2
Output parameter
Emission mass flow E

13. Comparison synthetic – empirical emission data
VO Biophysik | G. Schauberger | Institut für Med. Physik & Biostatistik
Comparison synthetic – empirical emission data
Comparison of the synthetic
emission data Emod
(Monte-Carlo model) with
empirical emission data
Emeas (inverse dispersion
model) for ethyl acetate,
in form of a qq-plot

14. Comparison synthetic – empirical emission data
VO Biophysik | G. Schauberger | Institut für Med. Physik & Biostatistik
Comparison synthetic – empirical emission data
Comparison Emod (light grey) and Emeas (dark grey) for butyl acetate (mean values and standard deviations)

15. VO Biophysik | G. Schauberger | Institut für Med. Physik & Biostatistik
Summary
Two-step procedure to re-construct emission data of the thermal treatment plant for waste: Step 1: Inverse dispersion modelling Step 2: Monte Carlo model
Predictors for the Monte Carlo model: Environmental predictors: Air temperature and wind velocity Process flow predictors: day of the week and time of the day
Synthetic emission data were generated
A good agreement between measured and modelled emission data can be achieved. Both the range of emissions as well as their time course are well reproduced.

16. Applications Completion of fragmentary emission data
VO Biophysik | G. Schauberger | Institut für Med. Physik & Biostatistik
Applications
Completion of fragmentary emission data
Assessment of emission data derived from short-time measurements
Extension of measurements to long time series for the future (e.g. climatic change impacts) 16