dmug 2016 - scott hamilton, ricardo energy & environment

45
CALPUFF for odour assessment A recent case study Scott Hamilton, PhD Technical Lead Urban Air Quality Modelling April 19 th 2016

Upload: ies-iaqm

Post on 15-Jan-2017

199 views

Category:

Environment


0 download

TRANSCRIPT

PowerPoint Presentation

CALPUFF for odour assessmentA recent case studyScott Hamilton, PhDTechnical LeadUrban Air Quality ModellingApril 19th 2016

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

This presentationCALPUFF for odour impacts in the literatureMethodologyWRFCALWRFCALMETCALPUFF modellingSome resultsComparison with Warren Springs modelPros and cons

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

CALPUFF in the literature for odour impacts

Despite the differences between CALPUFF and field inspection estimates, their general agreement suggests that both methods provide reasonable estimates of the real odor nuisance, so that their applied use is justified.

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

CALPUFF in the literature for odour impacts

...the odour impacts resulting from the application of the field inspection turned out to be quite comparable with those obtained by simulating the dispersion of emissions by means of a suitable dispersion model (CALPUFF), thus indicating that both approaches may be effective and complementary for odour impact assessment purposes.

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

CALPUFF can work well in the near field

On the basis of the combined results of the four-part validation (i.e., weight of evidence), the performance of CALPUFF was judged to be superior to that of AERMOD.

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

Modelled meteorology (WRF and CALMET)Nested WRF grid, using ARW coreBoundary meteorology from NCEP Climate Forecast System ReanalysisTerrain and coastal effects includedDetailed consideration of building effects (BPIP)One year modelled meteorologyCALPUFF model to explore effect of increasing stack height and velocity on odour dispersionAdvantages of CALPUFF include treatment of coastal effects, non-steady state meteorology, build up of pollutants over time, and handling of calm wind conditions (important for odour)

Main aspects of methodology

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

The site

Large manufacturing facility with waster water treatment and odour control unit. Short stack with very cool emissions.

History of odour complaints from nearby towns and villages.

Client was interested in knowing the emissions limits that would avoid future complaints.

We back calculated the emissions required to comply with 98%ile 1.5 ou/m3 limit at residential locations.

The computed values were then used to provide design values for a new OCU

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

Stack parameters modelledBase case1) Height- 14m from ground level 2) Efflux temperature- 15 to 25oC 3) Stack diameter- 1.3 m with an exit nozzle of 1 m 4) Efflux flow rate- 15000 m3/h 5) Efflux velocity- 5.3 m/s 6) Design velocity- >15 m/s

Scenarios (from client) 1) 14 m stack with stack velocity set to: a. 5 m/s b. 15 m/s c. 25 m/s 2) 25 m stack with stack velocity set to: a. 5 m/s b. 15 m/s c. 25 m/s 3) 30 m stack with stack velocity set to: a. 5 m/s b. 15 m/s c. 25 m/s

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

The Weather Research and Forecasting model (WRF)

The Weather Research and Forecasting (WRF) Model is a next-generation mesoscale numerical weather prediction system designed to serve both atmospheric research and operational forecasting needs.

We used the WRF-EMS tool (available from http://strc.comet.ucar.edu/software/newrems/), which at the time was based on WRF 3.4.1. WRF-EMS comes pre-set with useful defaults form NOAAs meteorologists.

WRF has a large worldwide community of registered users (over 25,000 in over 130 countries)

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

WRF model domain

The domain was designed so as to provide data in a format similar to that provided by commercial vendors of WRF data.The WRF model domain was prepared with nested grid configuration 25 x 25 cells in each direction with boundary and initialisation conditions provided by the Climate Forecasting System Reanalysis* datasets.

The model was set with 30 pressure levels which are more densely arranged near the earths surface.* http://cfs.ncep.noaa.gov/cfsr/downloads/

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

January d01- 300 x 300 km , winds and temperature

Visualisation prepared in UCARs IDV packageWRF modelling

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

January domain (d02)- 50 x 50 km- winds and temperature

Visualisation prepared in UCARs IDV packageWRF modelling

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

12

July domain 1- winds and temperatureWRF modelling

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

13

July d02- winds and temperatureWRF modelling

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

14

Processing the WRF dataThe WRF model provided a series of netCDF files each of which contained one day of meteorology for the Irvine bay area. The modelling yielded about 20GB of daily files which were then processed using the CALWRF code to derive suitable fields for use in the CALMET model. CALWRF produces CALMET.DAT files which form the input to the next stage of the model, the CALMET diagnostic meteorological model.

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

Comparison of modelled and measured winds

Observed wind direction

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

Comparison of modelled and measured wind speedObserved wind direction

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

Comparison of modelled and measured wind speed

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

Most complaints in early summer, this seems to coincide with a drop in dilution.

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

CALMET modelling- post CALWRF process

Prognostic met data nested grid 60kmCALMET domain16km @200m res.

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

23

TerrainPrognostic met data nested grid 60km

Gridded terrain elevation for the modelling domain were derived from the 3 arc-second digital elevation models (DEMs) produced by the United States Geological Survey (USGS). Elevations in the DEM files are in metres relative to sea level.

The spacing of the elevations along each profile is 3 arc-seconds which corresponds to approximately 90 metres.

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

24

Land use

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

25

CALMET modellingPrognostic met data nested grid 60kmCALMET domain16km

With terrain height and wind field from CALMET at 200m resolution

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

26

CALMET modellingPrognostic met data nested grid 60kmCALMET domain16kmClose up of wind field around site at 200m resolution

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

27

CALMET modellingPrognostic met data nested grid 60kmCALMET domain16kmGeophysical data example- surface roughness

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

28

CALMET modellingPrognostic met data nested grid 60km3D wind fields

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

29

CALPUFF modellingPrognostic met data nested grid 60kmCALMET domain16kmSite buildings included in the run

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

30

CALPUFF modellingPrognostic met data nested grid 60kmCALMET domain16kmNested receptor grids

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

31

Receptor set up

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

32

Plume tracking over 48 hours

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

33

ResultsOutputs from the dispersion model were processed using the CALRANK algorithm.

CALRANK extracts the relevant percentile values for the receptors from the hourly time series (CONC.DAT) CALPUFF output file for each case.

The assessment is based around the 98th percentile The 1000 oue/m3 baseline emission factor is set to test the dispersion of a unit emission in the domain around the GSK site.

To facilitate decision making we also provided results based on 5000, 10000 and 20000 oue/m3cases- based on the same flow rate (4.17m3/s) this corresponds to emissions of 4170, 20850, 83400 and 208500 ou/s from the OCU stack.

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

14m stack, 5ms velocity

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

35

30m stack, 5ms velocity

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

36

Base case- allowable odour units to eliminate complaints14m high, 5m/s velocity

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

37

Hamilton, Scott (HS) - The max emission density is equal to 15000m3 per hour/3600 = 4.17 per second.Hamilton, Scott (HS) - Divide by this value to get the max emission densityCase 220m high, 5m/s velocity

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

38

Case 325m high, 5m/s velocity

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

39

Case 330m high, 5m/s velocity

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

40

Agrees quite well with the CALPUFF results

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

Conclusions of the workThe on-site and boundary odour profile at the site would be improved significantly with a taller stack. This is evidenced by the dramatic reduction in concentrations around the site across runs 1 to 4, 5 to 8, and 9 to 12 (each set of runs corresponding to 14, 20, 25 and 30 m stacks with the groupings representing velocity).

The results suggest that odour concentrations are likely to be quite insensitive to release velocity enhancements at the distances where complaints are noted.

The impact of increasing the stack height is much less pronounced at more distant receptors. This is evidenced by the lack of sensitivity of the size of the 1.5 oue/m3 contour line for cases where the stack height is stepped up from 14m to 30m.

A design value of around 60,000 oue/m3 should be enough to prevent complaints at nearby properties.

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

42

Conclusions of the workA balanced approach of engineering solutions would appear to offer improvements to both of these situations (on site and off site), which as explained are quite distinct. Reducing emissions will improve the odour profile at residential receptors, whilst increasing the stack height will improve the odour profile on and very close to the site.

To inform the engineering decisions that will follow, we provided numerical results across 37 receptors around the site. These results essentially provide estimates of the carrying capacity of the local dispersion environment and the receptors within it to emissions from the OCU.

The results of the work were used to inform the specification for odour abatement at the facility, the engineering work is now underway.

Interestingly the Warren Springs calculation provided quite similar results!!

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

43

Pros and cons of the approachPROSThe modelling framework is set up now and can be turned to other applications at the same siteUsing prognostic data avoids issues of data quality at surface stations- data capture is 100%We learned a lot during the projectThe agreement between the modelled and measured meteorology was pretty good.Client was happy that unusual met had been considered- we covered all basesCONSExtremely data intensive- project folder is about 50GB of data. This presents problems for archiving if we do a lot of this sort of work.Probably takes about 5-10x longer from start to finish than using either AERMOD or ADMSQuite difficult operationally when things dont work as they should- not many people to turn to!

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence

Scott Hamilton, PhDRicardo Energy & [email protected]

# Ricardo-AEA Ltd Ricardo Energy & Environment in Confidence