Comparison of Air Dispersion Models including ADMS, AERMOD and CALPUFF

Transcription

1 Comparison of Air Dispersion Models including ADMS, AERMOD and CALPUFF by Dr David Carruthers ADMS User Group Meeting Vilnius 19 January 21

2 Well Known Dispersion Models Short range dispersion model s (upto 5km) ADMS (ADMS4 Industrial, Roads, Urban, Airports) AERMOD, ISC, OML, AUSTAL Industrial releases CALINE Road sources OSPM Street canyons AirViro Urban air quality Medium range dispersion models CALPUFF - Regional haze

3 Comparison of ADMS, AERMOD and CALPUFF Model Features Modelling Feature ADMS AERMOD CALPUFF APPLICATIONS Applications Up to 5km from sources; local and urban scale. Up to 5km from sources. Local and Regional Pollution Impacts. SOURCE TYPES Source types Point, line (including road, rail), area, volume, grid, jet. Point, line, volume and area sources. Point, line, volume, area METEOROLOGY Meteorology DISPERSION Boundary layer structure ADMS Pre-processor AERMET Pre-processor CALMET Pre-processor h, L MO scaling h, L MO scaling h, L MO scaling Plume rise Advanced integral model Briggs empirical expressions Concentration distribution Advanced Gaussian plume and puff model Advanced Gaussian plume model Briggs empirical expressions Non-steady Gaussian puff model

4 Comparison of ADMS, AERMOD and CALPUFF Model Features Modelling Feature ADMS AERMOD CALPUFF COMPLEX EFFECTS Buildings Complex terrain Deposition (wet and dry) Chemistry Based on flow model with near and main building wakes. Based on calculation of flow field and turbulence filed by FLOWSTAR model. Uses PRIME buildings model. Interpolation between neutral flow approximate solution and stable flow impaction solution. Based on ISC building model. Effects of complex flow input via meteorological fields. YES YES YES GRS (Generic Reaction Scheme) 8 reaction scheme for NO x chemistry, parameterised sulphate chemistry. Ozone limiting model, assumes maximum conversion of NO to NO 2. NO x and SO 2 chemistry for particle generation.

5 Comparison of ADMS, ARMOD and CALPUFF Model Features Modelling Feature ADMS AERMOD CALPUFF OTHER OPTIONS Street canyon model YES NO NO Emissions system EMIT system NO NO Short term fluctuations for odours, explosions etc Visibility Model Radioactive decay model YES NO YES Condensed plume visibility NO Visibility Impairment (haze/smog) YES; includes γ-dose NO NO Puff Model YES NO Puff release default Coastline YES NO YES Input of vertical profiles of met data VALIDATION YES YES Uses meteorological fields. Extensive industrial point sources, area sources, road sources, urban areas, airports. Extensive industrial point sources, area sources. Validation of meteorological f ields, concentrations and visibility impacts.

6 Flat Terrain Validation I Major study 24 Field and Wind Tunnel Experiments Summary Scores for ISC3, ADMS and AERMOD (Different model input parameters) Table 1 ISC3 ADMS AERMOD Best Middle Worst 17 7 Table 2 ISC3 ADMS AERMOD Best Middle Worst Table 1 from Hanna et al, 6 th Workshop on Harmonisation, France Oct 1999 Table 2 from Hanna et al, AWMA Meeting, US, June 2

7 Flat terrain II Kincaid power plant Site flat farmland with some lakes (z = 1 cm) Met 171 hours, neutral to convective Release 187-m stack, SF 6 Results ns/m 3 (normalised by emission rate, quality 3 data) Data Mean σ Bias NMS E Corr Fac 2 Observations ADMS AERMOD

8 Flat terrain III Kincaid power plant Scatter plots (ns/m 3 ) 35 ADMS 4 35 AERMOD modelled 2 15 AERMOD observed Observed

9 Flat terrain IV Kincaid power plant Quantile-quantile plots (ns/m 3 ) 35 ADMS 4 35 AERMOD modelled 2 15 AERMOD observed Observed

10 Flat Terrain V - CALPUFF and ISC: Kincaid Q-Q plot for CALPUFF and ISCST3 (quality 3 data)

11 Flat Terrain VI - Prairie Grass Prairie Grass: scatter plot of concentrations ADMS 4.1 Prairie Grass: scatter plot of concentrations AERMOD 2222 Prairie Grass: scatter plot of concentrations ISCST modelled modelled modelled observed observed observed

12 Flat Terrain VII - Prairie Grass Prairie Grass: q-q plot of concentrations ADMS 4.1 Prairie Grass: q-q of concentrations AERMOD 2222 Prairie Grass: q-q of concentrations ISCST modelled observed modelled observed modelled observed

13 Flat Terrain VIII Power Plant Comparison: H = 2 m; Exit velocity = 22 m/s ADMS ADMS Met/AERMOD Dispersion Mean Conc. 1th percentile

14 Flat Terrain IX Comparing ADMS and ADMS/AERMOD (converter 1) Long term runs: Maximum normalised concentration (µg/m 3 /(g/s))

15 Building Effects I Two plume approach

16 Building Effects II: ADMS, AERMOD and ISC PRIME model used in AERMOD (and ISC) is similar in approach to the ADMS buildings model. Differences between ADMS buildings module and PRIME ADMS Box model for source in cavity Main wake velocity field: wake dimension, velocity and turbulence fields from wall-wake theory Main wake has 6 zone dispersion model Model applied at all downstream distances PRIME Modified Gaussian for source in cavity Main wake velocity field: wake dimension from experiment, velocity and turbulence fields from free-wake theory Main wake as 2 zone dispersion model Virtual source model applied far downstream

17 Building Effects III Robins & Castro Experiment K Maximum ground-level concentration as a function of source height θ= and Ws/Ue=3.1 Experimental ADMS 4. ADMS 4.1 ISC-Prime Zs/l

18 Building Effects IV Robins & Castro Statistics

19 Building Effects V Snyder Experiment Scatter plot of normalised concentrations ADMS 4.1 Scatter plot of normalised concentrations ISC-Prime ADMS y=x y=2x y=x/2 ISC-Prime y=x y=2x y=x/ modelled 15 modelled observed observed

20 Complex Terrain I ADMS Complex Flow Model based on FLOWSTAR Example Askervein: Change in speed over hill Fractional speedup ratio delta S Distance from HT (m) AERMOD and ISC use idealised approaches CALPUFF uses 3D time dependent flow field

21 Complex Terrain II: ADMS and AERMOD Comparison in Neutral flow US EPA Wind Tunnel Data Lawson, Snyder and Thompson (1989) Ratio of complex terrain to flat terrain maximum concentrations as function of stack height and location ADMS AERMOD

22 Complex Terain III ADMS and AERMOD Comparison Maximum Concentration (ug/m3) Long Term Average Concentration (ug/m3) ADMS (Max=178) ADMS (Max=4.) Stack and surrounding terrain, Ribblesdale Valley, North-West England Stack height = 1m Terrain = up to 3m AERMOD (Max=1162) AERMOD (Max=1.3)

23 Complex Terrain IV, CALPUFF: Wyoming study Meteorology 4 upper air stations 22 surface stations 44 precipitation stations MM5 fields Terrain 4 km resolution Receptors in Class 1 Wilderness area

24 Complex Terrain V: CALPUFF, Wyoming case

25 Road Traffic Emissions I US CALTRANS Experiment Layout of roads and receptors

26 Road Traffic Emissions II ADMS-Roads and CALINE-4 Comparison of trendlines calculated using ADMS Roads and CALINE4 concentrations Calculated SF 6 concentration (ppb) ADMS Roads CALINE4 y=x y=.5x y=2x Monitored SF 6 concentration (ppb) Figure 1 Comparison of trendlines calculated from ADMS-Roads and CALINE4 concentrations

27

28 Summary Dispersion models in use in Europe include ADMS, AERMOD, CALPUFF, OML and AUSTAL. Key features of the dispersion models ADMS, AERMOD and CALPUFF been have presented and contrasted. Where data are available the models are compared with each other and with field and wind tunnel data. CALPUFF was developed for assessing medium range impacts of major pollution sources. It requires meteorological fields as input.

29 ADMS-Roads Model Capabilities ADMS-Roads (Part of ADMS-EIA) is designed to model dispersion scenarios from single or multiple roads. Calculates emissions from traffic flows or accepts calculated emissions Allows many road sources Fully integrated street canyon model based on Danish OSPM model Includes impact of traffic induced turbulence on dispersion Integrated with Geographical Information Systems (GIS) and an Emissions Inventory Database

30 ADMS-Roads M4 calculated and monitored PM1 concentration ADMS Roads Monitored 12 Concentration (µg/m3) Jan Mar-97 3-Apr Jun-97 8-Aug-97

31 Validation Results ADMS-Urban 14 NOx Annual Average PM1 Annual Average 12 NOx Standard Deviation NO2 Annual Average NO2 Percentile 1 PM1 9.4 Percentile PM Percentile PM1 Standard Deviation NO2 Standard Deviation 1 O3 annual Average O3 Standard Deviation 8 Predicted Data (ppb) NOx Percentile Predicted Data (ug/m3) Monitored Data (ppb) Monitored Data (ug/m3)