Odour Assessment for Harvest Power (Richmond, BC): 2014 Modelling Update

Transcription

1 Harvest Power Inc. Submitted by: LEVELTON CONSULTANTS LTD. Levelton File # R

2 LEVELTON CONSULTANTS LTD Clarke Place Richmond, BC V6V 2H9 T: F: Levelton File # R Metro Vancouver Environmental Regulation and Enforcement Division 4330 Kingsway Burnaby, BC V5H 4G8 Attention: Johanna van den Broeke, Officer, Air Quality Permitting Specialist Subject: Dear Johanna: Levelton Consultants Ltd. (Levelton) is pleased to submit the attached odour assessment report as a requirement of permit GVA1054 for Harvest Power Canada Ltd. s facility in Richmond, BC. Yours truly, Levelton Consultants Ltd. Per: Chris Koscher, B.Sc.(Hons.), EP Senior Air Quality Specialist ckoscher@levelton.com

3 TABLE OF CONTENTS 1 Introduction Ambient Air Quality Objectives Background Ambient Air Quality Source Parameters Area Sources Volume Sources CALMET and CALPUFF Modelling Methodology Modelling Domain Dispersion Modelling Results Odours Discussion Model Evaluation Model Limitations Summary and Conclusions References Appendix A Modelling Methodology... A 1 A.1 Model Selection... A 2 WRF... A 2 CALMET... A 2 A CALMET Modelling Domain... A 2 A Terrain Elevation and Land Use Data... A 4 A Meteorological Data... A 6 A CALMET Model Options... A 7 CALPUFF... A 9 A CALPUFF Model Options... A 10 A Model Domain and Receptors... A 11 A.2 CALMET Quality Assurance and Control... A 12 Vancouver International Airport... A 14 A Temperature... A 14 A Wind Speed... A 15 A Wind Roses... A 17 A Radar Charts... A 20 Richmond South (T17)... A 22 Levelton File # R (Richmond, BC): 2014 Modelling Update Page ii

4 A Temperature... A 22 A Wind Speed... A 23 A Wind Roses... A 25 A Radar Charts... A 28 Harvest Power Site... A 30 A Temperature... A 30 A Wind Speed... A 31 A Wind Roses... A 33 A Radar Charts... A 35 Wind Fields... A 37 Stability Classes... A 49 Mixing Heights... A 49 WRF Vertical Profile... A 54 Levelton File # R (Richmond, BC): 2014 Modelling Update Page iii

5 List of Figures Figure 4 1 Area and Volume Source Locations... 5 Figure 5 1 CALMET and CALPUFF Modelling Domains... 9 Figure 6 1 Predicted Maximum 1 hour Odour Units (Based on July / August 2013 Test) Figure 6 2 Predicted Maximum 1 hour Odour Units (Larger View) (Based on July / August 2013 Test) Figure 6 3 Predicted Maximum 10 minute Odour Units (Based on July / August 2013 Test) Figure 6 4 Predicted Maximum 10 minute Odour Units (Larger View) (Based on July / August 2013 Test) Figure 6 5 Predicted Maximum 1 hour Odour Units (Based on December 2013 Test) Figure 6 6 Predicted Maximum 1 hour Odour Units (Larger View) (Based on December 2013 Test) Figure 6 7 Predicted Maximum 10 minute Odour Units (Based on December 2013 Test) Figure 6 8 Predicted Maximum 10 minute Odour Units (Larger View) (Based on December 2013 Test) Figure 6 9 Predicted Maximum 1 hour Odour Units (Based on April 2014 Test) Figure 6 10 Predicted Maximum 1 hour Odour Units (Larger View) (Based on April 2014 Test) Figure 6 11 Predicted Maximum 10 minute Odour Units (Based on April 2014 Test) Figure 6 12 Predicted Maximum 10 minute Odour Units (Larger View) (Based on April 2014 Test) Figure 6 13 Location of the Top 50 Predicted Odour Units, December 2013 Test (Blue Crosses) Figure 7 1 Locations Evaluated for Source Apportionment (December 2013 Testing) Figure 7 2 Relative Source Contribution at the Locations Shown in Figure 7 1, December 2013 Testing. Each Location is Based on the Predicted 1 hour Maximum Odour Unit Figure 7 3 Wind rose for the Top 50 1 hour Odour Units (Based on December 2013 Testing) Appendix Figures Figure A 1 Map Displaying the CALMET and CALPUFF Modelling Domain... A 3 Figure A 2 Terrain Data Used in CALMET... A 5 Figure A 3 Land Use Data Used in CALMET... A 6 Figure A 4 Receptor Grid for Harvest Power Modelling... A 12 Figure A 5 Locations used for the CALMET and WRF QA/QC... A 13 Figure A 6 Annual Average Temperature Variation at YVR... A 14 Levelton File # R (Richmond, BC): 2014 Modelling Update Page iv

6 Figure A 7 Diurnal Variation of Temperature at YVR... A 15 Figure A 8 Wind Speed Frequency at YVR... A 16 Figure A 9 Average Monthly Wind Speed at YVR... A 16 Figure A 10 Diurnal Variation of Wind Speed at YVR... A 17 Figure A 11 YVR Observed Wind Rose... A 18 Figure A 12 YVR CALMET Extracted Wind Rose... A 19 Figure A 13 YVR WRF NMM Extracted Wind Rose... A 20 Figure A 14 YVR Radar Chart Showing Average Wind Speed for Each Direction... A 21 Figure A 15 YVR Radar Chart Showing Frequency of Occurrence for Each Direction... A 21 Figure A 16 Annual Average Temperature Variation at Richmond South (T17)... A 22 Figure A 17 Diurnal Variation of Temperature at Richmond South (T17)... A 23 Figure A 18 Wind Speed Frequency at Richmond South (T17)... A 24 Figure A 19 Average Monthly Wind Speed at Richmond South (T17)... A 24 Figure A 20 Diurnal Variation of Wind Speed at Richmond South (T17)... A 25 Figure A 21 Richmond South (T17) Observed Wind Rose... A 26 Figure A 22 Richmond South (T17) CALMET Extracted Wind Rose... A 27 Figure A 23 Richmond South (T17) WRF NMM Extracted Wind Rose... A 28 Figure A 24 Figure A 25 Richmond South (T17) Radar Chart Showing Average Wind Speed for Each Direction... A 29 Richmond South (T17) Radar Chart Showing Frequency of Occurrence for Each Direction... A 29 Figure A 26 Annual Average Temperature Variation at the Harvest Power Site... A 30 Figure A 27 Diurnal Variation of Temperature at the Harvest Power Site... A 31 Figure A 28 Wind Speed Frequency at the Harvest Power Site... A 32 Figure A 29 Average Monthly Wind Speed at Richmond South (T17)... A 32 Figure A 30 Diurnal Variation of Wind Speed at Richmond South (T17)... A 33 Figure A 31 Harvest Power Site CALMET Extracted Wind Rose... A 34 Figure A 32 Harvest Power Site WRF NMM Extracted Wind Rose... A 35 Figure A 33 Harvest Power Site Radar Chart Showing Average Wind Speed for Each Direction... A 36 Figure A 34 Figure A 35 Figure A 36 Harvest Power Site Radar Chart Showing Frequency of Occurrence for Each Direction... A 36 Example Wind Field Generated by CALMET for the 10 m Level for the Hour Ending at 05:00 on February 3, A 37 Example Wind Field Generated by CALMET for the 170 m Level for the Hour Ending at 05:00 on February 3, A 38 Levelton File # R (Richmond, BC): 2014 Modelling Update Page v

7 Figure A 37 Figure A 38 Figure A 39 Figure A 40 Figure A 41 Figure A 42 Figure A 43 Figure A 44 Figure A 45 Figure A 46 Figure A 47 Example Wind Field Generated by CALMET for the 3000 m Level for the Hour Ending at 05:00 on February 3, A 39 Example Wind Field Generated by CALMET for the 10 m Level for the Hour Ending at 02:00 on August 11, A 40 Example Wind Field Generated by CALMET for the 170 m Level for the Hour Ending at 02:00 on August 11, A 41 Example Wind Field Generated by CALMET for the 3000 m Level for the Hour Ending at 02:00 on August 11, A 42 Example Wind Field Generated by WRF for the Surface Sigma Level for the Hour Ending at 05:00 on February 3, A 43 Example Wind Field Generated by WRF for the Mid Sigma Level for the Hour Ending at 05:00 on February 3, A 44 Example Wind Field Generated by WRF for the Upper Sigma Level for the Hour Ending at 05:00 on February 3, A 45 Example Wind Field Generated by WRF for the Surface Sigma Level for the Hour Ending at 02:00 on August 11, A 46 Example Wind Field Generated by WRF for the Mid Sigma Level for the Hour Ending at 02:00 on August 11, A 47 Example Wind Field Generated by WRF for the Upper Sigma Level for the Hour Ending at 02:00 on August 11, A 48 Frequency Distribution of Stability Classes Calculated for Vancouver Airport (YVR) and the CALMET Extracted Points... A 49 Figure A 48 CALMET Extracted Monthly Mixing Height Variation... A 50 Figure A 49 CALMET Extracted Mixing Height Diurnal Variation... A 51 Figure A 50 CALMET Extracted Mixing Height Frequency Distribution... A 52 Figure A 51 CALMET Extracted Mixing Heights for February 3, A 53 Figure A 52 CALMET Extracted Mixing Heights for August 11, A 53 Figure A 53 WRF Extracted Wind Speeds versus Vertical Height... A 54 Figure A 54 WRF Extracted Ambient Temperatures versus Vertical Height... A 55 Levelton File # R (Richmond, BC): 2014 Modelling Update Page vi

8 List of Tables Table 4 1 Area Source Parameters... 2 Table 4 2 Area Source Odour Emissions... 4 Table 4 3 Volume Source Parameters... 6 Table 4 4 Volume Source Odour Emissions... 7 Table 6 1 Maximum Predicted Odour Units for 10 minutes and 1 hour Table 6 2 Maximum Predicted Odour Units at August 8, 2013 through August 14, 2014 Complaint Locations, based on December 2013 Test Appendix Tables Table A 1 Heights of CALMET Model Layers... A 4 Table A 2 Surface Meteorological Stations Used for CALMET Input... A 7 Table A 3 Selected CALMET Model Options... A 7 Table A 4 Selected CALPUFF Model Options... A ALL RIGHTS RESERVED THIS DOCUMENT IS PROTECTED BY COPYRIGHT LAW AND MAY NOT BE REPRODUCED IN ANY MANNER, OR FOR ANY PURPOSE, EXCEPT BY WRITTEN PERMISSION OF LEVELTON CONSULTANTS LTD. Levelton File # R (Richmond, BC): 2014 Modelling Update Page vii

9 1 INTRODUCTION Harvest Power Inc. (Harvest Power), also known as Fraser Richmond Soil and Fibre (FRSF)), operates a composting and energy garden facility in Richmond, BC. The facility is located at 7028 York Road (latitude 49⁰ N, 123 ⁰ W; or me, mn in Universal Transverse Mercator (UTM), grid zone 10 coordinates, NAD83 reference), and operates under Metro Vancouver Air Quality Management Permit GVA1054 (May 11 th, 2013) (the Permit). Harvest Power holds contracts with a number of municipalities around the Lower Mainland to collect food scraps and other compostable materials such as yard trimmings. The Harvest Power site consists of multiple piles of compost at various stages of aging. As the piles reach maturity, they are then moved to a screening area where they are separated by size and either sold as soil or re introduced into the aging piles. In addition to the composting operations, there is a cogeneration power plant which converts biogas produced by the decay of food waste into electricity. Levelton Consultants Ltd. (Levelton) was engaged by Harvest Power to conduct updated air dispersion modelling assessment to predict potential odour impacts for Harvest Power s operation based on three recent odour sampling tests as a requirement of the Permit. Specifically, the Permit requires a written report updating the 2013 CALPUFF dispersion modelling results (Levelton, 2013b) using the most current version of the CALPUFF air dispersion modelling system and current meteorological data. Odour emissions are based on three tests from: the 2013 summer season (Levelton, 2013c); the 2013 fall season (Envirochem, 2013); and, the 2014 spring season (Envirochem, 2014). In addition, supplemental data for a number of sources was obtained from previous testing data (Opus DaytonKnight, 2013), the latest testing information available to characterize these sources. This report and appendix serve to satisfy the above noted permit requirement. 2 AMBIENT AIR QUALITY OBJECTIVES The federal and provincial governments, as well as Metro Vancouver, have developed ambient air quality objectives to promote long term protection of public health and the environment. Metro Vancouver s ambient air quality objectives are based on current knowledge regarding air quality and health, and take into consideration relevant objectives from around the world and the ability to achieve those objectives within the Metro Vancouver region. To our knowledge, there are currently no ambient air quality or standards explicitly for odours in place at the federal, provincial (BC), or regional (Metro Vancouver). 3 BACKGROUND AMBIENT AIR QUALITY Background ambient air quality data for odour was not considered or available for inclusion in the assessment. Levelton File # R (Richmond, BC): 2014 Modelling Update Page 1

10 4 SOURCE PARAMETERS 4.1 AREA SOURCES Area sources included in the assessment are listed in Table 4 1. Release heights of the various area sources included in the assessment were estimated and provided to Levelton by Harvest Power. The initial sigma z was calculated using guidance from the SCREEN3 Model User s Guide (US EPA, 1995), Table 1, for an elevated source. The initial sigma z is estimated by dividing by 2.15 as the piles are located close to other sources which act as buildings in terms of air dispersion characteristics. The summarized modelling parameters for each area source are presented in Table 4 1 and defined as orange polygons in Figure 4 1. Certain pile heights have decreased since the previous assessment (Levelton, 2013b). The decrease in pile heights is consistent with FRSF s efforts to reduce on site and off site odours. These efforts are consistent with compost Best Management Practices (BMPs). The receiving area is lower as a result of FRSF s shortened pre processing time. When material is tipped, it is processed and loaded onto the CASP as quickly as possible; resulting is smaller piles in the receiving area. The aging piles have been lowered in an effort to maintain higher levels of oxygen in the material. This ultimately results in better decomposition and reduced odour production. A sensitivity analysis of predicted odour to the initial sigma z assumption was conducted testing an initial sigma z ranging from 20%, 46.5% (US EPA guidance) and 100% of the effective source height. The maximum predicted odour from the sensitivity test ranged from approximately 12% less to 5% greater as compared to predicted odour when using the US EPA guidance to determine the initial sigma z. The contour plots of the predicted odour concentrations from the sensitivity tests were very similar with respect to the geographic extent. Based on the sensitivity test conducted, a change in initial sigma z would not result in a significant increase in predicted odour. It should be noted that while the overs and middlings piles have been modelled independently, the fines piles are modelled as part of the product piles. This is because the fines material is directly transferred into the product piles area of the facility and has therefore been captured in this source. The modelled areas listed in Table 4 1 were estimated from a geo referenced site plan provided to Levelton from Harvest Power and based on Harvest Power s input of the current site configuration. Table 4 1 Source Name Area Source Parameters Modelled Source Area (m 2 ) Height to Top of Source (Effective Height) (m) Initial Sigma z (m) Receiving and Mixing East CASP West CASP Pile Tear Down Aging Piles Screening Product Piles Northeast Biofilter Southwest Biofilter Levelton File # R (Richmond, BC): 2014 Modelling Update Page 2

11 Source Name Modelled Source Area (m 2 ) Height to Top of Source (Effective Height) (m) Initial Sigma z (m) Screening Biofilter Energy Garden Biofilter Portable Screening Overs Pile Portable Screener Middlings Pile Portable Screener Overs Pile Permanent Screener Middlings Pile Permanent Screener Odour emission rates used for air dispersion modelling were developed based on three tests from: the 2013 summer season (July / August, 2013) (Levelton, 2013c); the 2013 fall season (December, 2013) (Envirochem, 2013); and, the 2014 spring season (April, 2014) (Envirochem, 2014). In addition, supplemental data for a number of sources was obtained from previous testing data (Opus DaytonKnight, 2013), the latest testing information available to characterize these sources. Levelton has considered all of the sources to be operating (emitting) continuously throughout the entire modelled period. This approach is consistent with the language contained in Harvest Power s permit (GVA1054) and provides a conservative estimate of predicted odour impacts. With the exception of the biofilters, all other sources do not have a measureable flow. Therefore, to determine an emission rate, the testing sweep gas flow rate was used to estimate the OU/m 2 /s based on the following methodology. A 40 L Tedlar bag was filled with sample air in 5 minutes, or a sampling rate of 8 L/min. These samples were further diluted with clean air at a 35:1 ratio, thus the diluted sampling rate was L/min. The sampling area of the flux chamber was 960 cm 2, based off a diameter of 35 cm, therefore the estimated flow per unit area was m 3 /s/m 2. This estimated flow per unit area was used to transform the OU/m 3 provided into OU/m 2 /s for the purposes of dispersion modelling. The area source odour emissions are presented in Table 4 2 below. Levelton File # R (Richmond, BC): 2014 Modelling Update Page 3

12 Table 4 2 Source Name Area Source Odour Emissions Modelled Source Area (m 2 ) Odour Units (OU/m 3 ) Odour Emissions July / August 2013 December 2013 April 2014 Emission Rate (OU/m 2 /s) Odour Units (OU/m 3 ) Emission Rate (OU/m 2 /s) Odour Units (OU/m 3 ) Emission Rate (OU/m 2 /s) Receiving and Mixing * * * * East CASP ** 0.2** 2300** 0.2** 2300** 0.2** West CASP ** 0.2** 2300** 0.2** 2300** 0.2** Pile Tear Down ** 2** 23000** 2** 23000** 2** Aging Piles * * * * Screening ** ** 95** ** 95** ** Product Piles * * * * Northeast Biofilter Southwest Biofilter Screening Biofilter Energy Garden Biofilter Portable Screening ** ** 95** ** 95** ** Overs Pile Portable Screener * * * * Middlings Pile Portable Screener * * * * Overs Pile Permanent Screener * * * * Middlings Pile Permanent Screener * * * * * Based on December 2013 test results as only the biofilters were tested in July / August 2013 and April ** 2013/2014 test results were not available for these sources, therefore supplemental data from previous testing was used (Opus DaytonKnight, 2013) Levelton File # R Page 4

13 Notes: The fence line is shown by the red dashed line The area sources are shown by the orange triangles The volume source locations are noted by x Figure 4 1 Area and Volume Source Locations Levelton File # R (Richmond, BC): 2014 Modelling Update Page 5

14 4.2 VOLUME SOURCES In addition to the modelled area sources, conveyors are used to transfer and separate aged material into product piles. Seven volume sources have been identified in relation to the screening operation. The two screeners (permanent and portable) are associated with three pile sizes: overs, middlings and, fines for a total of six transfer points. The seventh point is the transfer from the front end loader into the portable screener s hopper. The permanent screener is also operated with loaders, but a partial enclosure has been constructed and this air is vented through the screening biofilter. Therefore, this material transfer point has not been considered. Volume source dimensions were based on information provided by Harvest Power which included estimated drop heights and conveyor dimensions. Table 4 3 lists the volume sources (material transfer points) considered in this assessment. The initial sigma z and initial sigma y values were determined using the guidance provided in the SCREEN3 Model User s Guide (US EPA, 1995), Table 1 for an elevated source. The initial sigma z values were determined using the drop height of the sources and dividing it by The initial sigma y values were determined using the length of the source (conveyor width and width of loader bucket) and dividing by 4.3. The loader is considered to be operating continuously, which is an overly conservative assumption as the loaders operation will be intermittent and limited to specific hours during the day, but provides an upper bound scenario. Table 4 3 Volume Source Parameters Height to Top of Source Source Name (Effective Height) (m) Drop Height (m) Initial Sigma z (m) Initial Sigma y (m) Overs Pile Portable Screener Middlings Pile Portable Screener Fines Pile Portable Screener Overs Pile Permanent Screener Middlings Pile Permanent Screener Fines Pile Permanent Screener Loader Transfer Portable Screener CALPUFF volume source emissions (material transfer points) were modelled based on an odour unit per unit time basis (OU/s) developed from the odour test results for the overs, middlings and fines piles from the December 2013 test (5508 OU/m 3 ) (Envirochem, 2013). The methodology for determining the volume source emission rates is as follows: 1. Determine the time required for material to fall from drop height: Where, 1 2 = material drop height (m) = initial fall velocity (m/s), assumed to be 0 m/s Δt = material fall time (s) = acceleration (gravity) (m/s 2 ) Levelton File # R (Richmond, BC): 2014 Modelling Update Page 6

15 2. Using the above time, calculate the fall velocity of the material: Where, = material fall velocity (m/s) = material drop height (m) = material fall time (s) 3. Calculate the flow of material: Where, = Flow rate of material (m 3 /s) = Area of column of material (m 2 ) = material fall velocity (m/s) 4. Use the material flow rate and the odour unit concentration to determine the emission rate for use in modelling: Where, = Odour emission rate (OU/s) = Flow of material (m 3 /s) = Odour Unit concentration (OU/m 3 ) The odour emission rates developed using this methodology were then used as the emission rates for the volume sources. Table 4 4 shows the volume source odour emission rates based on the above described methodology. The flow rate for the loader which is inputting material into the screener was based on the flow rate from the screener output (0.47 m 3 /s multiplied by 3) representing the flows to the overs, middlings, and fines piles. Table 4 4 Volume Source Odour Emissions Source Name Odour Units Estimated Flow Rate Odour Emission Rate (OU/m 3 ) (m 3 /s) (OU/s) Overs Pile Portable Screener Middlings Pile Portable Screener Fines Pile Portable Screener Overs Pile Permanent Screener Middlings Pile Permanent Screener Fines Pile Permanent Screener Loader Transfer Portable Screener Levelton File # R (Richmond, BC): 2014 Modelling Update Page 7

16 5 CALMET AND CALPUFF MODELLING METHODOLOGY The updated air dispersion modelling was conducted following the methods recommended in the Guidelines for Air Quality Dispersion Modelling in British Columbia (AQMG) (BC MOE, 2008) and the methodology followed for the previous Harvest Power modelling (Levelton, 2013b), which followed input from Metro Vancouver on the Detailed Model Plan for Harvest Power s Richmond Facility (Levelton, 2013a). This section presents a brief summary of the modelling methods. A detailed description is provided in Appendix A. The CALPUFF model suite was used for this analysis. CALPUFF is a suite of numerical models (CALMET, CALPUFF, and CALPOST) that are used in series to determine the impact of emissions in the vicinity of a source or group of sources. Detailed three dimensional meteorological fields were produced by the diagnostic computer model CALMET Version (121203), which is an updated version of the United States Environmental Protection Agency (US EPA) approved CALMET Version and contains numerous bug fixes. Input information used by CALMET includes: surface weather data, digital land use data, terrain data, and prognostic meteorological data. The three dimensional fields produced by CALMET were used by CALPUFF Version 6.42 (110325), an updated version of the US EPA approved CALPUFF Version CALPUFF is a three dimensional, multi species, non steady state Gaussian puff dispersion model that can simulate the effects of time and space varying meteorological conditions on pollutant transport. Finally, CALPOST Version 6.292, an updated version of the US EPA approved CALPOST Version 6.221, is a statistical processing program which was used to summarize and tabulate the pollutant concentrations calculated by CALPUFF. The three dimensional CALMET meteorological fields were generated using meteorological data from numerous surface stations and prognostic meteorological data from the Weather Research and Forecasting (WRF) Nonhydrostatic Mesoscale Model (NMM) model, and digital terrain and land use data. Building downwash was not considered and does not apply for this study as the sources are area and volume sources. 5.1 MODELLING DOMAIN The CALMET modelling domain is a 32 km by 32 km domain area centred approximately 2.1 km to the south south east of Harvest Power s facility, with a 250 m grid resolution and nine vertical layers. This domain was approved as part of the detailed modelling plan submitted to Metro Vancouver for the original modelling (Levelton, 2013a). Details of the CALMET modelling methodology are provided in Appendix A. The CALPUFF modelling domain was similar to the CALMET modelling domain and covers a 30 km by 30 km area extending to within 1 km of the CALMET boundaries to avoid edge effects. Within the domain, a nested sampling grid of receptors was created with the following special distributions: 20 m spacing along the facility boundary; 50 m spacing within 500 meters of the centre of the facility; 250 m spacing within 2 km of the centre the facility; 500 m spacing within 5 km of the centre of the facility; and 1000 m spacing to beyond 5 km of the centre of the facility. Levelton File # R (Richmond, BC): 2014 Modelling Update Page 8

17 In addition to the nested grids described above, Metro Vancouver requested a higher resolution receptor grid over portions of Metro Vancouver. These areas are considered to be areas with a high population density and will be used in the odour dispersion modelling to assess the potential for odour impacts. Receptors were not included within the facility boundary, where ambient air quality objectives are not applicable. A 1.5 m receptor height was used to simulate the average height of human air intake. Figure 5 1 shows the CALMET and CALPUFF domain including the receptors. Figure 5 1 CALMET and CALPUFF Modelling Domains Levelton File # R (Richmond, BC): 2014 Modelling Update Page 9

18 6 DISPERSION MODELLING RESULTS 6.1 ODOURS The maximum predicted odour units (OU/m 3 ) are shown in Table 6 1 for the three different tests described previously in Section 1 and Section 4. Each of the three test periods were modelled separately using the entire modelling period. The December 2013 test period had the highest maximum predicted odour concentrations, which is likely due to the high odour concentration from the Energy Garden Biofilter during this test period. The 10 minute average was calculated following the Ontario Ministry of Environment s Air Dispersion Modelling Guideline for Ontario formula in Section 4.4 (Ontario Ministry of Environment, 2009) To convert a 1 hour averaging period concentration to a 10 minute averaging period concentration, the Ontario Ministry of Environment recommends using a factor of 1.65, which means n = Table 6 1 Maximum Predicted Odour Units for 10 minutes and 1 hour Maximum Predicted Concentration Stack Testing Period 10 minute Average (OU/m 3 ) 1 hour Average (OU/m 3 ) July / August December April For the July / August 2013 testing, Figure 6 1 shows the maximum predicted 1 hour odour units. The 1 odour unit contour is highlighted in orange and denotes the region in which half of the population can detect an odour. A larger view of the same contour plot in Figure 6 2. The maximum distance to the 1 hour, 1 odour unit contour is approximately 5.6 kilometres from the facility. Figure 6 3 shows the calculated 10 minute odour units based on maximum predicted 1 hour odour units. These results were calculated by applying the Ontario guideline listed above. A larger view of the same contour plot is shown in Figure 6 4. The maximum distance to the 10 minute, 1 odour unit contour is approximately 6.4 kilometers from the facility. Similar contour plots for the December 2013 testing are shown in Figure 6 5 through Figure 6 8. The maximum distance to the 1 hour, 1 odour unit contour is approximately 5.9 kilometres from the facility. The maximum distance to the 10 minute, 1 odour unit contour is approximately 7.0 kilometers from the facility. Contours for the April 2014 testing are shown in Figure 6 9 through Figure The maximum distance to the 1 hour, 1 odour unit contour is approximately 4.3 kilometres from the facility. The maximum distance to the 10 minute, 1 odour unit contour is approximately 5.3 kilometers from the facility. Odour complaints received by Metro Vancouver for the period from August, 2013 through August, 2014 are shown as blue x s in Figure 6 1 through Figure Levelton File # R (Richmond, BC): 2014 Modelling Update Page 10

19 Notes: Contours are 0.5, 1, 5, 10, and 25 OUs Complaints received by Metro Vancouver from August, 2013 to August, 2014 indicated by blue x s Figure 6 1 Predicted Maximum 1 hour Odour Units (Based on July / August 2013 Test) Levelton File # R (Richmond, BC): 2014 Modelling Update Page 11

20 Notes: Contours are 1, 5, 10, and 20 OUs Complaints received by Metro Vancouver from August, 2013 to August, 2014 indicated by blue x s Figure 6 2 Predicted Maximum 1 hour Odour Units (Larger View) (Based on July / August 2013 Test) Levelton File # R (Richmond, BC): 2014 Modelling Update Page 12

21 Notes: Contours are 0.5, 1, 5, 10, and 25 OUs Complaints received by Metro Vancouver from August, 2013 to August, 2014 indicated by blue x s Figure 6 3 Predicted Maximum 10 minute Odour Units (Based on July / August 2013 Test) Levelton File # R (Richmond, BC): 2014 Modelling Update Page 13

22 Notes: Contours are 1, 5, 10, 20 and 40 OUs Complaints received by Metro Vancouver from August, 2013 to August, 2014 indicated by blue x s Figure 6 4 Predicted Maximum 10 minute Odour Units (Larger View) (Based on July / August 2013 Test) Levelton File # R (Richmond, BC): 2014 Modelling Update Page 14

23 Notes: Contours are 0.5, 1, 5, 10, and 25 OUs Complaints received by Metro Vancouver from August, 2013 to August, 2014 indicated by blue x s Figure 6 5 Predicted Maximum 1 hour Odour Units (Based on December 2013 Test) Levelton File # R (Richmond, BC): 2014 Modelling Update Page 15

24 Notes: Contours are 1, 5, 10, 20, 40, and 60 OUs Complaints received by Metro Vancouver from August, 2013 to August, 2014 indicated by blue x s Figure 6 6 Predicted Maximum 1 hour Odour Units (Larger View) (Based on December 2013 Test) Levelton File # R (Richmond, BC): 2014 Modelling Update Page 16

25 Notes: Contours are 0.5, 1, 5, 10, and 25 OUs Complaints received by Metro Vancouver from August, 2013 to August, 2014 indicated by blue x s Figure 6 7 Predicted Maximum 10 minute Odour Units (Based on December 2013 Test) Levelton File # R (Richmond, BC): 2014 Modelling Update Page 17

26 Notes: Contours are 1, 5, 10, 20, 40, and 60 OUs Complaints received by Metro Vancouver from August, 2013 to August, 2014 indicated by blue x s Figure 6 8 Predicted Maximum 10 minute Odour Units (Larger View) (Based on December 2013 Test) Levelton File # R (Richmond, BC): 2014 Modelling Update Page 18

27 Notes: Contours are 0.5, 1, 5, 10, and 25 OUs Complaints received by Metro Vancouver from August, 2013 to August, 2014 indicated by blue x s Figure 6 9 Predicted Maximum 1 hour Odour Units (Based on April 2014 Test) Levelton File # R (Richmond, BC): 2014 Modelling Update Page 19

28 Notes: Contours are 1, 5, 10, and 20 OUs Complaints received by Metro Vancouver from August, 2013 to August, 2014 indicated by blue x s Figure 6 10 Predicted Maximum 1 hour Odour Units (Larger View) (Based on April 2014 Test) Levelton File # R (Richmond, BC): 2014 Modelling Update Page 20

29 Notes: Contours are 0.5, 1, 5, 10, and 25 OUs Complaints received by Metro Vancouver from August, 2013 to August, 2014 indicated by blue x s Figure 6 11 Predicted Maximum 10 minute Odour Units (Based on April 2014 Test) Levelton File # R (Richmond, BC): 2014 Modelling Update Page 21

30 Notes: Contours are 1, 5, 10, and 20 OUs Complaints received by Metro Vancouver from August, 2013 to August, 2014 indicated by blue x s Figure 6 12 Predicted Maximum 10 minute Odour Units (Larger View) (Based on April 2014 Test) Levelton File # R (Richmond, BC): 2014 Modelling Update Page 22

31 Figure 6 13 shows the location of the top 50 predicted odour units for the December 2013 test. All of these occur along the southern fence line of the Harvest Power s facility. Figure 6 13 Location of the Top 50 Predicted Odour Units, December 2013 Test (Blue Crosses) Table 6 2 lists all of the public odour complaints received by Metro Vancouver that are either attributed to or suspected to have come from Harvest Power based on the worst case emissions characteristics (December 2013 Testing). The table shows the distance from the Harvest Power facility and the model predicted 1 hour and 10 minute odour units at that location. In addition, the last column are the results of an analysis was conducted by Scott Kerr of Harvest Power determining the likelihood of the odour complaint due to Harvest Power s operations, based on the wind direction and activities at Harvest Power at the time leading up to the complaint. Highlighted results indicate that the maximum predicted 1 hour or 10 minute odour units at the representative receptor exceed one odour unit. Levelton File # R (Richmond, BC): 2014 Modelling Update Page 23

32 Table 6 2 Complaint Received Date Maximum Predicted Odour Units at August 8, 2013 through August 14, 2014 Complaint Locations, based on December 2013 Test Approximate Address of Complaint Distance from Harvest Power (km) Maximum Predicted 1 hour Odour Units at Representative Receptor Maximum Predicted 10 minute Odour Units at Representative Receptor Favourable Winds and Activities from Harvest Power? 08/06/ Victory St, Burnaby Yes 08/10/ Fedoruk Rd, Richmond Yes 08/15/ No 7 Rd, Richmond Yes 08/15/ Minoru Blvd, Richmond Yes 08/20/ Minoru Blvd, Richmond No 08/20/ Blundell Rd, Richmond Yes 08/21/ Greenlees Road No 08/23/ Minoru Blvd, Richmond Yes 08/24/13 No 4 & Stevestone, Richmond No 08/24/ Minoru Blvd, Richmond No 08/25/ Minoru Blvd, Richmond Yes 08/27/ No 3 Rd, Richmond No 09/04/ Flury Dr, Richmond No 09/07/ Minoru Blvd, Richmond No 09/08/ Minoru Blvd, Richmond No 09/12/ Minoru Blvd, Richmond No 09/19/ Minoru Blvd, Richmond No 10/30/ Flury Dr, Richmond Unknown 10/31/ No. 6 Rd, Richmond Yes 11/04/ No. 5 Rd, Richmond Yes 11/04/ Balsam St, Vancouver No Levelton File # R Page 24

33 Complaint Received Date Approximate Address of Complaint Distance from Harvest Power (km) Maximum Predicted 1 hour Odour Units at Representative Receptor Maximum Predicted 10 minute Odour Units at Representative Receptor Favourable Winds and Activities from Harvest Power? 11/04/ Gibbons Dr, Richmond Yes 11/08/ Block Blundell Road, Richmond Yes 11/12/ Flury Drive, Richmond Yes 11/21/ Princess Street, Richmond No 11/21/ Moffat Rd, Richmond Yes 11/23/ Andrews Rd, Richmond No 11/27/ Lucas Rd, Richmond Yes 11/27/ Westminster Hwy, Richmond Yes 11/28/ Maple Rd, Richmond No 11/28/ Glendower Dr, Richmond No 11/28/ Lynas St, Richmond No 11/28/ Jacombes Rd, Richmond Yes 11/28/ Minoru Blvd, Richmond No 11/28/ Minoru Blvd, Richmond No 12/01/ Blk Blundell, Richmond No 12/05/ Gibbons Dr, Richmond No 12/09/ Blk Rosecroft Crs, Richmond Unknown 12/10/ Blk Andrews Rd, Richmond No 12/10/ Princess St, Richmond No 12/11/13 Lansdowne Mall, Richmond No 12/11/ Blk Gibbons Dr, Richmond No 12/12/13 Williams & No. 3 rd, Richmond Yes 12/16/ Block Housman Street, Richmond Yes Levelton File # R Page 25

34 Complaint Received Date Approximate Address of Complaint Distance from Harvest Power (km) Maximum Predicted 1 hour Odour Units at Representative Receptor Maximum Predicted 10 minute Odour Units at Representative Receptor Favourable Winds and Activities from Harvest Power? 12/17/ Princess St, Richmond Yes 12/18/ Nelson St, Richmond Yes 12/18/ No. 6 Rd, Richmond Yes 12/19/ Princess St, Richmond No 12/20/ No. 6 Rd, Richmond Yes 12/22/ Trites Rd, Richmond No 12/27/ Trites Rd, Richmond No 01/14/ Princess St, Richmond No 01/15/14 Richmond Auto Mall, Richmond Yes 01/21/ Princess St, Richmond No 01/27/ Martynuik Pl, Richmond No 01/28/ Princess St, Richmond No 02/08/ Martynuik Pl, Richmond Yes 02/14/ Block Fedoruk Road, Richmond Yes 02/28/ Algonquin Dr, Richmond No 03/03/ Westminster Hwy, Richmond Yes 04/14/ Cordiale Dr, Vancouver Yes 04/30/ Blk Steveson Hwy, Richmond No 04/30/14 Jacombs & Cambie, Richmond No 05/02/ Balsam St, Vancouver Yes 05/15/ No. 6 Rd, Richmond Yes 05/21/ Glendower Dr., Richmond No 05/23/ Blk Balsam, Vancouver Yes Levelton File # R Page 26

35 Complaint Received Date Approximate Address of Complaint Distance from Harvest Power (km) Maximum Predicted 1 hour Odour Units at Representative Receptor Maximum Predicted 10 minute Odour Units at Representative Receptor Favourable Winds and Activities from Harvest Power? 05/26/ Glendower Dr, Richmond No 05/26/ Minoru Blvd, Richmond No 05/30/ th Ave, Vancouver No 06/05/ Blundell Rd, Richmond No 06/07/ Blk Rosecroft Cres, Richmond No 06/08/ Glendower Dr, Richmond No 06/12/ Jacombs St, Richmond Yes 06/12/ No. 6 Rd, Richmond No 06/20/14 Rhodes & 40th St, Vancouver Yes 07/02/ Minoru Blvd, Richmond Yes 07/02/ Balsam St, Vancouver Yes 07/02/ Block 31 Ave E, Vancouver Yes 07/03/ Balsam St, Vancouver Yes 07/10/ Minoru Blvd, Richmond No 07/11/ Minoru Blvd, Richmond Yes 07/11/ Rosecroft Cres, Richmond No 07/12/ Minoru Blvd, Richmond Yes 07/13/ Ave E, Vancouver Yes 07/13/ Jasper Cres, Vancouver Yes 07/14/ Balsam St, Vancouver Yes 07/15/ Cordiale Dr, Vancouver Yes 07/15/ Cordiale Dr, Vancouver Yes 07/16/ Minoru Blvd, Richmond Yes Levelton File # R Page 27

36 Complaint Received Date Approximate Address of Complaint Distance from Harvest Power (km) Maximum Predicted 1 hour Odour Units at Representative Receptor Maximum Predicted 10 minute Odour Units at Representative Receptor Favourable Winds and Activities from Harvest Power? 07/18/ Cordiale Dr, Vancouver Yes 07/22/ Fedoruk Rd, Richmond Yes 07/28/ General Currie Rd, Richmond No 07/30/ E 40th Ave, Vancouver Yes 08/04/ Balsam, Vancouver Unknown 08/11/ Minoru, Richmond Unknown 08/13/14 No 7 and Westminster, Richmond Unknown 08/14/ Chandlery Pl, Vancouver Unknown Levelton File # R Page 28

37 7 DISCUSSION 7.1 MODEL EVALUATION In order to have a greater understanding of which sources may be contributing to potential odour issues, a source apportionment analysis was completed. Four receptors were chosen based on the location of the maximum predicted odour units and the three closest complaint locations from the December 2013 test (Figure 7 1). At each of the chosen location, the time at which the maximum 1 hour odour units occurred were identified. For this time, each source s predicted odour units at that receptor were selected and compiled into Figure 7 2, which shows the relative contribution of each source to the maximum value at that location. Figure 7 1 Locations Evaluated for Source Apportionment (December 2013 Testing) Levelton File # R (Richmond, BC): 2014 Modelling Update Page 29

38 Figure 7 2 Relative Source Contribution at the Locations Shown in Figure 7 1, December 2013 Testing. Each Location is Based on the Predicted 1 hour Maximum Odour Unit. From Figure 7 2 it is clear that the biofilters make up a majority of the odour units at each receptor location with the Energy Garden Biofilter being the largest contributor. It was noted by Harvest Power that the Energy Garden Biofilter was having operational issues during this time, resulting in higher odour than usual. Upgrades to the Energy Garden Biofilter have been undertaken as described in Harvest Power s Report on EG Biofilter (June, 2014). Figure 6 13 shows the location of the top 50 predicted 1 hour odour units from the December 2013 testing. For each of these predicted concentrations, the associated meteorology from the model was extracted. These maximums tend to occur in the morning before solar insolation has time to mix the atmosphere. In each of the 50 cases, the mixing heights are below 60 meters and the winds are light from the north and north north east. A majority of the hours have neutral stability classes. The aggregated wind rose from these hours is shown in Figure 7 3. Levelton File # R (Richmond, BC): 2014 Modelling Update Page 30

39 Figure 7 3 Wind rose for the Top 50 1 hour Odour Units (Based on December 2013 Testing) 7.1 MODEL LIMITATIONS The results presented in this assessment provide an estimate based on the test data provided by Levelton, Envirochem, and Harvest Power. Harvest Power operates a large facility which could have other potential sources of odour. It is difficult to monitor emissions from all points at the Harvest Power facility due to its size, and therefore a snapshot of emissions was captured from each testing period based on their evaluation of the emission sources. It is possible that these emissions may change over time, from season to season or as on site processes change. The emissions used in the dispersion model are considered to be emitted continuously throughout the year which is not necessarily the case depending on the source. For the purpose of conservatism in conduction the dispersion modelling, continuous emissions were considered. Levelton File # R (Richmond, BC): 2014 Modelling Update Page 31

40 8 SUMMARY AND CONCLUSIONS Levelton Consultants Ltd. was engaged by Harvest Power to conduct updated air dispersion modelling of odours in support of their Metro Vancouver Air Quality Management Permit (Permit GVA1054) for their facility located in Richmond, BC. This assessment focused on the air quality impacts from odour as a result of on site operations. Air dispersion modelling was conducted using the CALPUFF model suite considering a meteorological data set of one year in duration. On the basis of the modeled data from sampling in July / August 2013, December 2013, and April 2014, the predicted odours generated by this facility are contained to the area in Richmond within approximately 7.0 kilometers surrounding the facility based on a 10 minute averaging period with the worst case emissions scenario (December 2013). Upgrades have been made by Harvest Power in 2014 to the Energy Garden Biofilter to improve its efficiency and performance in removing odour. As shown by the worst case test results from December 2013 the Energy Garden Biofilter had the largest contribution to the maximum results predicted. Assuming all other sources are operating in a similar manner, the upgrades to the Energy Garden Biofilter should decrease the geographic extent and magnitude of the odour impacts. 9 REFERENCES BC Ministry of Environment (Environmental Protection Division, Environmental Quality Branch, Air Protection Section), Guidelines for Air Quality Dispersion Modelling in British Columbia Envirochem, Emissions and Odour Measurements, Harvest Power. Prepared by Harvest Power, Inc. and Envirochem Services Inc. December, 2013 Envirochem, April 2014 Sampling of Biofilter Emissions at Harvest Power s Richmond, BC Facility. Prepared by Envirochem Services Inc. for Harvest Power, Inc. July 31 st, Harvest Power Inc., Report on EG Biofilter. Prepared by Harvest Power Inc. June, Levelton, 2013a. Detailed Model Plan for Harvest Power s Richmond Facility. Prepared by Levelton Consultants Ltd. and submitted to Metro Vancouver on behalf of Harvest Power Canada Ltd. January 30 th, Levelton, 2013b. Air Quality Assessment for Harvest Power (Richmond, BC): Odour and Volatile Organic Compounds. Prepared by Levelton Consultants Ltd. and submitted to Metro Vancouver on behalf of Harvest Power Canada Ltd. October 22 nd, Levelton, 2013c. Biofilter Emission Test Report Harvest Power Canada Ltd. 3 rd Quarter Prepared by Levelton Consultants Ltd. and submitted to Metro Vancouver on behalf of Harvest Power Canada Ltd. October 23 rd, Ontario Ministry of Environment, Air Dispersion Modelling Guideline for Ontario (Version 2.0). /std01_ pdf Levelton File # R (Richmond, BC): 2014 Modelling Update Page 32

41 Opus DaytonKnight, Fraser Richmond Soil and Fibre Emissions Characterization Report. May 29, Scire, Joseph S. et al., 2000a. A User s Guide for the CALMET Meteorological Model (Version 5). Earth Tech Inc. January, Scire, Joseph S. et al., 2000b. A User s Guide for the CALPUFF Dispersion Model (Version 5). Earth Tech Inc. January, Scire, Joseph S. et al., CALPUFF Modeling System Version 6 User Instructions. April, US EPA (1995). SCREEN3 Model User s Guide. United States Environmental Protection Agency, Office of Air Quality Planning and Standards, Emissions Monitoring and Analysis Division. Research Triangle Park, North Carolina. September Levelton File # R (Richmond, BC): 2014 Modelling Update Page 33

42 APPENDIX A MODELLING METHODOLOGY Page A 1

43 A.1 MODEL SELECTION CALPUFF is a suite of numerical models (CALMET, CALPUFF, and CALPOST) that are used in series to determine the impact of emissions in the vicinity of a source or group of sources. Detailed threedimensional meteorological fields are produced by the diagnostic computer model CALMET, based on inputs such as: surface, marine and upper air meteorological data, digital land use data and terrain data, and prognostic meteorological data. The three dimensional fields produced by CALMET are used by CALPUFF, a three dimensional, multi species, non steady state Gaussian puff dispersion model that can simulate the effects of time and space varying meteorological conditions on pollutant transport. Finally CALPOST, a statistical processing program, is used to summarize and tabulate the pollutant concentrations calculated by CALPUFF. WRF Three dimensional prognostic meteorological data from the Weather Research and Forecasting (WRF) Nonhydrostatic Mesoscale Model (NMM) was used as an initial guess field for the CALMET model. WRF NMM prognostic data used for this dispersion modelling analysis was run by SENES Consultants Limited (SENES) in August SENES ran WRF NMM in analysis mode, using historical data snapshots from the National Centers for Environmental Prediction (NCEP) North American Mesoscale (NAM) Model as initial and boundary conditions. This historical data includes all available observations, such as satellite, radar, balloon borne, surface, and tower observations. WRF NMM was run on a 250 km by 250 km domain encompassing the Lower Fraser Valley Region of British Columbia. A 12 km by 12 km horizontal resolution with 6 hour time steps was used to define the initial and boundary conditions, while the main run was with a 3 km by 3 km horizontal resolution at an hourly time resolution for all of 2008 through CALMET CALMET Version (121203), an updated version of the United States Environmental Protection Agency (US EPA) approved CALMET Version (130731), was run to calculate meteorological fields for the modelled time period from January 1, 2012 through December 31, Three dimensional prognostic meteorological data from WRF NMM was used in order to improve the performance of the CALMET model. In addition, meteorological input data was also used from 14 surface stations within the CALMET domain. The meteorological data and CALMET output for this modelling period were assessed following the Quality Assurance and Quality Control (QA/QC) procedures outlined in Section A 2, CALMET Quality Assurance and Control. A description of the CALMET methods and data sets follows. A CALMET Modelling Domain The Universal Transverse Mercator (UTM, NAD 83) coordinate system was used for this model application. The CALMET domain is a 32 km by 32 km area, as shown in Figure A 1. The WRF domain incorporated into the CALMET modelling extends across and beyond the entire Lower Fraser Valley Region (UTM NAD83 Zone km to km Easting and km to km Northing). The CALMET model was run with a 250 m grid resolution. The modelling domain and grid resolution were chosen to encompass the main topographical features for generating the CALMET three dimensional diagnostic meteorological fields. In the vertical axis, nine atmospheric layers were chosen, the height of which are given in Table A 1. Page A 2

44 Note: Figure A 1 Location of Harvest Power is indicated by the orange rectangle Map Displaying the CALMET and CALPUFF Modelling Domain Page A 3

45 Table A 1 Heights of CALMET Model Layers Vertical Layer Number Height at Top of Layer (m) A Terrain Elevation and Land Use Data Digital terrain and land use data covering the model domain was included in the CALMET input data set. Digital terrain files with a 1:50,000 scale were used to generate inputs for each CALMET grid point. Land use characteristics for each grid cell based on LandData BC data sets were used. The BC land use class codes were translated into the land use class codes used by CALMET according to the procedures in the BC Air Quality Modelling Guidelines (AQMG) (BC MOE, 2008). Plots of the digital terrain and land use data in the CALMET domain are shown in Figure A 2 and Figure A 3 below. Page A 4

46 Note: Figure A 2 Location of Harvest Power is indicated by the orange star Terrain Data Used in CALMET Page A 5

47 Note: Figure A 3 Location of Harvest Power is indicated by the orange star Land Use Data Used in CALMET A Meteorological Data Surface meteorological stations that record hourly data include those operated by the Meteorological Service of Canada (MSC) and Metro Vancouver. Data from fourteen surface stations, listed in Table A 2, were used as input to the CALMET model. Upper air data was not used as the prognostic data contains the necessary upper air information within the CALMET domain and no upper air station are located in or near the CALMET modelling domain. CALMET requires a measured data value for every hour from at least one meteorological station in order to simulate the three dimensional fields. Missing data procedures were implemented, when required, as per the AQMG. As a supplement to the observational data, three dimensional meteorological fields from the WRF prognostic model were used. The WRF prognostic data was used as input into CALMET as the initial guess field. The "initial guess" wind field is calculated by interpolating the winds to the fine CALMET Page A 6

48 scale and then adjusting them for terrain and land use effects. The wind fields are then adjusted based on the observed meteorological fields from the fourteen surface stations. Table A 2 Surface Meteorological Stations Used for CALMET Input Surface Meteorological Station T02 Vancouver Kitsilano T04 Burnaby Kensington Park T09 Port Moody T13 North Delta T14 Burnaby Mountain T17 Richmond South T18 Burnaby South T22 Burnaby Burmount T23 Burnaby Capitol Hill T24 Burnaby North T31 Richmond Airport T38 Annacis Island T39 Tsawwassen Vancouver Airport (YVR) Operated By Metro Vancouver Metro Vancouver Metro Vancouver Metro Vancouver Metro Vancouver Metro Vancouver Metro Vancouver Metro Vancouver Metro Vancouver Metro Vancouver Metro Vancouver Metro Vancouver Metro Vancouver MSC A CALMET Model Options The CALMET model has a number of user specified input switches and options that determine how the model handles terrain effects, interpolation of observational input data, etc. The differences in the modelled and measured meteorological fields were examined, and this analysis was used to determine which model options were appropriate for modelling period. Table A 3 outlines the CALMET options used in modelling. The current recommended AQMG default parameters were used whenever applicable. Table A 3 Selected CALMET Model Options CALMET Model Option Parameter Option Selected AQMG Default Wind field model selection variable IWFCOD 1 (Yes) Compute Froude number adjustment effects? IFRADJ 1 (Yes) Compute kinematic effects? IKINE 0 (No) Use O Brien procedure for adjustment of the vertical velocity? IOBR 0 (No) Page A 7

49 CALMET Model Option Parameter Option Selected AQMG Default Compute slope flows? ISLOPE 1 (Yes) Extrapolate surface wind observations to upper layers? IEXTRP 4 (Similarity theory used except at layer 1 data at upper air stations are ignored) Extrapolate calm winds aloft? ICALM 0 (No) Layer dependent biases Minimum distance between upper air station and surface station for which extrapolation of surface winds will be allowed Gridded prognostic wind field model output fields Use varying radius of influence? Maximum radius of influence over land of the surface layer BIAS RMIN2 IPROG LVARY 1, 0, 0, 0, 0, 0, 0, 0, 0 1 (Set to 1 for IEXTRP = +/ 4) 14 (Yes, use wind fields from MM5/3D.dat file as initial guess field) F (No, if stations outside RMAX1 are definitely not wanted) No default RMAX1 5 km No default Maximum radius of influence over land aloft RMAX2 20 km No default Maximum radius of influence over water RMAX2 20 km No default Minimum radius of influence used in the wind field interpolation RMIN 0.1 Radius of influence of terrain features TERRAD 10 km No default Distance from a surface station at which the station observations and 1 st guess field are equally weighted Distance from an upper air station at which the observations and 1 st guess field are equally weighted R1 3 km No default R2 5 km No default Relative weighting of the prognostic wind field data RPROG 0 No default Maximum acceptable divergence in the divergence minimum procedure. Maximum number of iterations in the divergence minimum procedure. DIVLIM 5*10 6 NITER 50 Number of passes in the smoothing procedure NSMTH 2, 4, 4, 4, 4, 4, 4, 4, 4 Page A 8

50 CALMET Model Option Maximum number of stations used in each layer for the interpolation of data to a grid point Parameter Option Selected AQMG Default NINTR2 99 Critical Froude number CRITFN 1 Empirical factor controlling the influence of kinematic effects Multiplicative scaling factor for extrapolation of surface observations to upper layers Number of barriers to interpolation of the wind fields X and Y coordinates of barriers Diagnostic module surface temperature option Diagnostic module surface Meteorological station to use Diagnostic module domain averaged lapse rate option Diagnostic module upper air station to use for lapse rate to use Depth through which the domain scale lapse rate is computed ALPHA 0.1 FEXTR2 Unused NBAR Unused XBBAR, YBBAR, XEBAR, YEBAR IDIOPT1 ISURFT IDIOPT2 Unused 0 (Compute internally from hourly surface observations or prognostic fields) 6 (Richmond South 17) 0 (Compute internally from (at least) twice daily upper air observations or prognostic fields IUPT Unused ZUPT 200 Initial guess field wind components IDIOPT3 0 Upper air station to use for domain scale winds IUPWND Unused Bottom and top of layer through which the initial guess winds are computed ZUPWND 1,1000 CALPUFF CALPUFF Version 6.42 (110325), an updated version of the US EPA approved CALPUFF Version (130731), was run for the modelled time period from January 1, 2012 through December 31, The CALPUFF model was used to simulate dispersion of emissions from the various sources at Harvest Power, based on the meteorological wind fields developed by CALMET. Page A 9

51 A CALPUFF Model Options Table A 4 outlines dispersion options used in the CALPUFF modelling. Unless otherwise stated in Table A 4, the applicable default regulatory options currently recommended in the AQMG were used. Table A 4 Selected CALPUFF Model Options Option Parameter Option Selected AQMG Default Vertical distribution used in the near field MGAUSS 1 (Gaussian) Terrain adjustment method MCTADJ 3 (Partial plume path adjustment) Subgrid Scale complex terrain flag MCTSG 0 (Not Modelled) Near field puffs modelled as elongated? MSLUG 0 (No) Transitional Plume Rise modelled? MTRANS 1 (Yes) Method used to simulate building downwash? MBDW 2 (Prime) Stack tip downwash? MTIP 1 (Yes) Vertical wind shear modelled above stack top? MSHEAR 0 (No) Puff splitting allowed? MSPLIT 0 (No) Chemical Transformation Scheme? MCHEM 0 (Not Modelled) Aqueous phase transformation flag (only used in MCHEM =1 or 3) MAQCHEM Unused Wet removal modelled? MWET 0 (No) Dry deposition modelled? MDRY 0 (No) Method used to compute dispersion coefficients MDISP 2 (Dispersion coefficients from internally calculated sigma v, sigma w using micrometeorological variables (u*, w*, L, etc.) Sigma measurements used? MTURBVW Unused Back up method used to compute dispersion when measured turbulence data are missing MDISP2 Unused PG sigma y,z adjusted for roughness MROUGH 0 (Yes) Partial plume penetration of elevated inversion? Strength of temperature inversion provided in PROFILE.DAT extended records? MPARTL 1 (Yes) MTINV 0 (No) Page A 10

52 Option Probability Distribution Function used for dispersion under convective conditions? Parameter Option Selected AQMG Default MPDF 1 (Yes) Sub grid TIBL module used for shore line? MSGTIBL Unused Boundary conditions (concentration) modelled? MBCON 0 (No) Configure for FOG Model output? MFOG 0 (No) Test options specified to see if they conform to regulatory values? MREG 0 (No) A Model Domain and Receptors The CALMET and CALPUFF modelling domain are shown below in Figure A 4. Within the CALPUFF domain, a nested sampling grid of receptors was created with the following spatial distribution which is also shown in Figure A 4: 20 m spacing along the plant boundary; 50 m spacing within 500 meters of the centre of the facility; 250 m spacing within 2 km of the centre the facility; 500 m spacing within 5 km of the centre of the facility; and, 1000 m spacing to beyond 5 km of the centre of the facility. In addition to the nested grids described above, Metro Vancouver requested a higher resolution receptor grid over portions of Metro Vancouver. These areas are considered to be areas with a high population density and were used in the odour dispersion modelling to assess the potential for odour impacts. Page A 11

53 Note: Figure A 4 Location of Harvest Power is indicated by the orange star Receptor Grid for Harvest Power Modelling A.2 CALMET QUALITY ASSURANCE AND CONTROL The BC Ministry of Environment provides recommendations and guidance on QA/QC for CALMET generated data (BC MOE, 2008). Three locations within the CALMET modelling domain were chosen in order to show the appropriateness of the WRF and CALMET modelling data. These three locations are: the Harvest Power site, Vancouver International Airport (MSC Station), and Richmond South (Metro Vancouver Station T17). The meteorological fields generated by the WRF and CALMET models were compared to the observed meteorological data for the January 1, 2012 through December 31, 2012 modelling period, in order to determine the suitability of the WRF and CALMET data for the dispersion modelling. Page A 12

54 Figure A 5 Locations used for the CALMET and WRF QA/QC The metrological parameters of interest for QA/QC analysis conducted are: temperature, wind speed, wind direction, wind fields, mixing heights, and stability classes. Figures which analyze these parameters are presented in the following sections. CALMET parameters were determined by the meteorological data at the nearest CALMET grid cell, while WRF parameters were extracted using bilinear interpolation and the direct profile method near surface using the CALPUFF post processor METSERIES v1.9.0 (121203). Page A 13

55 Vancouver International Airport A Temperature Figure A 6 shows the average monthly surface temperature and Figure A 7 shows the average hourly temperature (binned into intervals) for the available surface data, extracted CALMET output, and extracted WRF output at the YVR station. In addition, the 30 year climate normals from are also plotted in Figure A 6 below. Both plots show an excellent agreement between the predicted and observed values. Figure A 6 Annual Average Temperature Variation at YVR Page A 14

56 Figure A 7 Diurnal Variation of Temperature at YVR A Wind Speed The frequency distribution of measured surface winds from the surface station, the predicted values from the extracted CALMET point, and the predicted values from the extracted WRF point are shown below in Figure A 8. Figure A 9 and Figure A 10 below show the monthly and diurnal variations, respectively, for the YVR surface station, the extracted CALMET point, and the extracted WRF point. The CALMET output follows a similar trend to that of the surface station and WRF data, with the CALMET wind speeds generally having a greater frequency of lower wind speeds than those observed at the surface stations and extracted from the WRF data. Page A 15

57 Figure A 8 Wind Speed Frequency at YVR Figure A 9 Average Monthly Wind Speed at YVR Page A 16

58 Figure A 10 Diurnal Variation of Wind Speed at YVR A Wind Roses Figure A 11 shows a wind rose for YVR based on observational data from January 1, 2012 through December 31, Figure A 12 and Figure A 13 show wind roses for the same modelling period based on the extracted CALMET and WRF data, respectively. All three wind roses show good agreement with each other, although it is noted that the predominant winds are from the west northwest for the WRF data only. For the observational data and CALMET extracted data, the predominant winds for the area are from the east. Page A 17

59 Figure A 11 YVR Observed Wind Rose Page A 18

60 Figure A 12 YVR CALMET Extracted Wind Rose Page A 19

61 Figure A 13 YVR WRF NMM Extracted Wind Rose A Radar Charts Figure A 14 and Figure A 15 below show radar charts comparing the observational, CALMET extracted, and WRF extracted metrological data. Figure A 14 shows the average wind speed as binned into 16 cardinal directions, while Figure A 15 indicates the frequency of occurrence as binned into 16 cardinal directions. The results indicate that the average wind speed and frequency of occurrence are similar comparing the different sources of meteorological data. Page A 20

62 Figure A 14 YVR Radar Chart Showing Average Wind Speed for Each Direction Figure A 15 YVR Radar Chart Showing Frequency of Occurrence for Each Direction Page A 21

63 Richmond South (T17) A Temperature Figure A 16 shows the average monthly surface temperature and Figure A 17 shows the average hourly temperature (binned into intervals) for the available surface data, extracted CALMET output, and extracted WRF output at the Richmond South (T17) Station. Both plots show an excellent agreement between the predicted and observed values. Figure A 16 Annual Average Temperature Variation at Richmond South (T17) Page A 22

64 Figure A 17 Diurnal Variation of Temperature at Richmond South (T17) A Wind Speed The frequency distribution of measured surface winds from the surface station, the predicted values from the extracted CALMET point, and the predicted values from the extracted WRF point are shown below in Figure A 18, Figure A 19 and Figure A 20 below show monthly and diurnal variations respectively for the Richmond South (T17) surface station, the extracted CALMET point, and the extracted WRF point. The surface data, extracted CALMET data, and extracted WRF data all follow a similar trend, however it is noted that the CALMET wind speeds are generally lower than the WRF wind speeds, and the observed wind speeds are generally lower than the CALMET wind speeds. Page A 23

65 Figure A 18 Wind Speed Frequency at Richmond South (T17) Figure A 19 Average Monthly Wind Speed at Richmond South (T17) Page A 24

66 Figure A 20 Diurnal Variation of Wind Speed at Richmond South (T17) A Wind Roses Figure A 21 shows a wind rose for Richmond South (T17) based on observational data from January 1, 2012 through December 31, Figure A 22 and Figure A 23 show wind roses for the same modelling period based on the extracted CALMET and WRF data, respectively. All three wind roses show good agreement with each other, and the predominant winds are from the east and west northwest. Page A 25

67 Figure A 21 Richmond South (T17) Observed Wind Rose Page A 26

68 Figure A 22 Richmond South (T17) CALMET Extracted Wind Rose Page A 27

69 Figure A 23 Richmond South (T17) WRF NMM Extracted Wind Rose A Radar Charts Figure A 24 and Figure A 25 below show radar charts comparing the observational, CALMET extracted, and WRF extracted metrological data. Figure A 24 shows the average wind speed as binned into 16 cardinal directions, while Figure A 25 indicates the frequency of occurrence as binned into 16 cardinal directions. The results indicate that the average wind speed and frequency of occurrence are similar comparing the different sources of meteorological data. Page A 28

70 Figure A 24 Richmond South (T17) Radar Chart Showing Average Wind Speed for Each Direction Figure A 25 Richmond South (T17) Radar Chart Showing Frequency of Occurrence for Each Direction Page A 29

71 Harvest Power Site A Temperature Figure A 26 shows the average monthly surface temperature and Figure A 27 shows the average hourly temperature (binned into intervals) for the extracted CALMET output, and extracted WRF output at the Harvest Power Site. Both plots show an excellent agreement between the predicted values. Figure A 26 Annual Average Temperature Variation at the Harvest Power Site Page A 30

72 Figure A 27 Diurnal Variation of Temperature at the Harvest Power Site A Wind Speed The frequency distribution the predicted values from the extracted CALMET point and the predicted values from the extracted WRF point are shown below in Figure A 28. Figure A 29and Figure A 30 below show monthly and diurnal variations respectively for the extracted CALMET point and extracted WRF point. The extracted CALMET data and extracted WRF data follow a similar trend, however it is noted that the CALMET wind speeds are generally lower than the WRF wind speeds. Page A 31

73 Figure A 28 Wind Speed Frequency at the Harvest Power Site Figure A 29 Average Monthly Wind Speed at Richmond South (T17) Page A 32

74 Figure A 30 Diurnal Variation of Wind Speed at Richmond South (T17) A Wind Roses Figure A 31 and Figure A 32 show wind roses from January 1, 2012 through December 31, 2012 for the Harvest Power site extracted CALMET and WRF data, respectively. Both wind roses show good agreement with each other, and the predominant winds are from the west northwest and east. Page A 33

75 Figure A 31 Harvest Power Site CALMET Extracted Wind Rose Page A 34

76 Figure A 32 Harvest Power Site WRF NMM Extracted Wind Rose A Radar Charts Figure A 33 and Figure A 34 show radar charts comparing the CALMET extracted and WRF extracted metrological data. Figure A 33 shows the average wind speed as binned into 16 cardinal directions, while Figure A 34 indicates the frequency of occurrence as binned into 16 cardinal directions. The results indicate that the average wind speed and frequency of occurrence are similar comparing the different sources of meteorological data. Page A 35

77 Figure A 33 Harvest Power Site Radar Chart Showing Average Wind Speed for Each Direction Figure A 34 Harvest Power Site Radar Chart Showing Frequency of Occurrence for Each Direction Page A 36

78 Wind Fields Representative wind fields for two 24 hour periods are presented in this section. The 24 hour periods were chosen based on having light winds and stable conditions, with one of the periods during the summer season and the other during the winter season. Wind fields are presented at the surface, midlevel, and upper level layers for both the CALMET and WRF data, however these wind fields cannot be directly compared as the CALMET layers are based upon a specified elevation, while the WRF layers are based on sigma levels resulting in varying elevations within the CALMET domain. Note: Figure A 35 Location of Harvest Power is indicated by the orange star Example Wind Field Generated by CALMET for the 10 m Level for the Hour Ending at 05:00 on February 3, 2012 Page A 37

79 Note: Figure A 36 Location of Harvest Power is indicated by the orange star Example Wind Field Generated by CALMET for the 170 m Level for the Hour Ending at 05:00 on February 3, 2012 Page A 38

80 Note: Figure A 37 Location of Harvest Power is indicated by the orange star Example Wind Field Generated by CALMET for the 3000 m Level for the Hour Ending at 05:00 on February 3, 2012 Page A 39

81 Note: Figure A 38 Location of Harvest Power is indicated by the orange star Example Wind Field Generated by CALMET for the 10 m Level for the Hour Ending at 02:00 on August 11, 2012 Page A 40

82 Note: Figure A 39 Location of Harvest Power is indicated by the orange star Example Wind Field Generated by CALMET for the 170 m Level for the Hour Ending at 02:00 on August 11, 2012 Page A 41

83 Note: Figure A 40 Location of Harvest Power is indicated by the orange star Example Wind Field Generated by CALMET for the 3000 m Level for the Hour Ending at 02:00 on August 11, 2012 Page A 42

84 Note: Figure A 41 Location of Harvest Power is indicated by the orange star Example Wind Field Generated by WRF for the Surface Sigma Level for the Hour Ending at 05:00 on February 3, 2012 Page A 43

85 Note: Figure A 42 Location of Harvest Power is indicated by the orange star Example Wind Field Generated by WRF for the Mid Sigma Level for the Hour Ending at 05:00 on February 3, 2012 Page A 44

86 Note: Figure A 43 Location of Harvest Power is indicated by the orange star Example Wind Field Generated by WRF for the Upper Sigma Level for the Hour Ending at 05:00 on February 3, 2012 Page A 45

87 Note: Figure A 44 Location of Harvest Power is indicated by the orange star Example Wind Field Generated by WRF for the Surface Sigma Level for the Hour Ending at 02:00 on August 11, 2012 Page A 46

88 Note: Figure A 45 Location of Harvest Power is indicated by the orange star Example Wind Field Generated by WRF for the Mid Sigma Level for the Hour Ending at 02:00 on August 11, 2012 Page A 47

89 Note: Figure A 46 Location of Harvest Power is indicated by the orange star Example Wind Field Generated by WRF for the Upper Sigma Level for the Hour Ending at 02:00 on August 11, 2012 Page A 48

90 Stability Classes The frequency distribution of predicted stability for Harvest Power was compared with the calculated stability for Vancouver Airport (YVR) using Turner s method. YVR was used as the other surface stations used in the modelling did not have the required parameters to determine stability via Turner s method (AQMG preferred method to calculate stability from observations). As seen in Figure A 47, the extracted CALMET stability classes show a reasonable agreement with calculated stability classes for the Vancouver Airport, with the modelled data displaying a higher percentage of stable conditions (namely stability class 6) and the observation data displaying a higher percentage of neutral stability conditions (stability class 4). Figure A 47 Frequency Distribution of Stability Classes Calculated for Vancouver Airport (YVR) and the CALMET Extracted Points Mixing Heights Predicted mixing heights statistics from CALMET are shown in Figure A 48 through Figure A 50. The mixing heights at YVR are much lower than the mixing heights at Richmond South (T17) and the Harvest Power Site locations. However, this is likely due to the proximity of YVR to the Georgia Straight water body where the mixed layer would be lower in height. Figure A 51 and Figure A 52 show the mixing height variation for a 24 hour period, with the expected trend of higher mixing heights throughout the daytime compared to the nighttime. Page A 49

91 Figure A 48 CALMET Extracted Monthly Mixing Height Variation Page A 50

92 Figure A 49 CALMET Extracted Mixing Height Diurnal Variation Page A 51

93 Figure A 50 CALMET Extracted Mixing Height Frequency Distribution Page A 52