ATMOSPHERIC IMPACT REPORT In support of

Transcription

1 ATMOSPHERIC IMPACT REPORT In support of Eskom s application for postponement of the Minimum Emission Standards compliance timeframes for the Kriel Power Station Prepared by: Naledzi Environmental Consultants (NEC) 145 Thabo Mbeki Street, Fauna Park, Polokwane, 0700 South Africa Date: November 2018 Report prepared for: Eskom SOC Ltd Version: Draft i

2 This report has been prepared by Naledzi Environmental Consultants (Pty) Ltd in association with umoya- NILU Consulting (Pty) Ltd representing Eskom SOC Ltd. No part of the report may be reproduced in any manner without written permission from Naledzi Environmental Consultants (Pty) Ltd and umoya-nilu Consulting (Pty) Ltd representing Eskom SOC Ltd. Authorship: Naledzi Environmental Consultants (Pty) Ltd (NEC) 145 Thabo Mbeki Street, Fauna Park, Polokwane, 0700 South Africa Lead Author: Sean O Beirne BA Hons (Geography) & MSc (Radar rainfall measurement) Certified Environmental Assessment Practitioner Tel: / sobeirne@tiscali.co.za and: umoya-nilu Consulting (Pty) Ltd P O Box Durban North, 4016 South Africa Co-Authors: M Zunckel A Raghunandan National Diploma (Meteorology); BSc (Meteorology); BSc Hons (Meteorology); MSc and PHD. Professional Natural Scientist: SACNASP: /04 MA (Atmospheric Sciences); BA Hons (Environmental Sciences); BPaed (Education) Tel: / mark@umoya-nilu.co.za Report Date: November 2018 ii

3 EXECUTIVE SUMMARY Eskom s coal-fired Kriel Power Station (hereinafter referred to as Kriel ) in Mpumalanga Province has a total installed capacity of MW. Power generation is a Listed Activity in terms of Section 21 of the National Environmental Management: Air Quality Act, 2004 (Act No. 39 of 2004) (NEMAQA) Kriel is unable to comply with the new plant MES for any of the three pollutants by 2020 due to financial, technical and water limitations. Kriel is scheduled for decommissioning between 2026 and 2029 and the station is thus seeking suspension from compliance with the new plant standard and requesting alternate limits. Kriel will be retrofitted with HFPS so that the power station is meets the existing plant PM MES by 2025 for the south stack only. For the South Stack an alternative limit of 100 mg/nm 3 for PM is requested until station decommissioning. For the North Stack an alternative limit of 125 mg/nm 3 is requested until 2025 and from then an alternative limit of 100 mg/nm 3. The power station is also requesting an alternative NOx emissions limit of mg/nm 3 and an alternative SO2 emissions limit of mg/nm 3 for the remainder of the life of the power station. The purpose of this AIR has been to assess the likely implications of the postponement and the requested alternative emissions limits for human health and the environment. The AIR contains two major parts namely an analysis of the ambient air quality likely to be affected by emissions from the power station and secondly, dispersion modelling of two different emissions scenarios to predict the likely impacts of the power station on that prevailing air quality. These two emissions scenarios are: Current emissions from the power station; and, Compliance with the MES. An analysis of ambient air quality data from the Kriel Village, Kendal, Phola, Komati and Elandsfontein ambient air quality monitoring stations indicates general compliance for the 10-minute and hourly average SO2 National Ambient Air Quality Standard (NAAQS), but non-compliance with the daily average SO2 NAAQS at the Kriel Village (2015), Kendal (2015 and 2017) and Komati monitoring stations (2016) where non-compliance with the annual average NAAQS is also evident. Measured concentrations of NO2 are seen to comply with the hourly average and annual average NAAQS but the data quality is generally poor so that compliance cannot be assured in all circumstances. For PM there is general non-compliance for both size fractions and for all averaging periods with data at some the stations being many times over the limit value. Diurnal hourly averages exhibit pronounced morning and late afternoon peaks for PM10, PM2.5 and NO2, with an approximate midday peak of SO2 indicating the important contribution of ground level sources such as domestic fuel use to the peak values measured. i

4 Summary of compliance with the NAAQS for each of the ambient air quality monitoring stations used in this analysis. Kriel Power Station Averaging period 10 minute SO2 1 hour SO2 Daily SO2 Annual SO2 1 hour NO2 Annual NO2 Daily PM10 Annual PM10 Daily PM2.5 Annual PM2.5 Kriel Village Kendal Phola Komati Elandsfontein 2015 Y Y N Y Y Y N N N* N* 2016 Y Y Y Y Y* Y* N N N N 2017 Y Y Y Y Y* Y* N N N N 2015 Y* Y* N* Y* Y* Y* N* N* N* N* 2016 Y Y Y Y Y* Y* N N N N 2017 Y Y N* Y* Y Y N N N* N* 2015 Y* Y* Y* Y* DD DD Y Y N* N* 2016 Y Y Y Y DD DD N N N N 2017 Y Y Y Y DD DD N N N N 2015 Y* Y Y* Y Y Y N N Y* Y* 2016 Y* Y* N N* Y* Y* N N N Y 2017 Y Y Y* Y Y* Y* N N N N 2015 Y* Y* Y* Y* Y* Y* N* N* N* N* 2016 Y* Y* Y* Y* Y Y N Y N N 2017 Y* Y* Y* Y* Y* Y* N Y N N *Means that the data record is <80% NM Not measured DD Data deficient. Dispersion modelling of the current emissions for SO2, NOx and PM10 from Kriel alone, indicates compliance with the relevant NAAQS for all averaging periods. The net effect of all of the above is that PM is already and unequivocally resulting in unacceptable health risk for a large part of the Highveld. The direct contribution of Kriel alone to that situation is considered to be small even taking into account predicted concentrations of secondary PM2.5. This AIR should be read, however, in conjunction with the Summary AIR that contains the predicted concentrations as a result of the combined emissions from all the power stations. Summary of compliance with the NAAQS for ambient air quality predicted for each of the emissions scenarios modelled for Kriel alone. Averaging period Scenario 1 - Actual Emissions Scenario 2 - New plant MES compliance SO2 (µg/m 3 ) 1-hour Yes Yes 24-hour Yes Yes Annual Yes Yes NO2 (µg/m 3 ) 1-hour Yes Yes Annual Yes Yes PM10 and PM2,5 (µg/m 3 )* 24-hour Yes Yes Annual Yes Yes * includes PM 2.5 predicted for the transformation of SO 2 and NO 2 to particulate form. ii

5 LIST OF ACRONYMS µm 1 µm = 10-6 m AEL Atmospheric Emission License AIR Atmospheric Impact Report APPA Atmospheric Pollution Prevention Act, 1965 (Act No. 45 of 1965) AQMP Air Quality Management Plan BID Background Information Document DEA Department of Environmental Affairs DoE Department of Energy ESP Electrostatic precipitator FFP Fabric Filter Plant FGD Flue gas desulphurisation IRP Integrated Resource Plan LNB Low NOx Burner LPG Liquid Petroleum Gas NAAQS National Ambient Air Quality Standards NEMAQA National Environment Management: Air Quality Act, 2004 (Act No. 39 of 2004) NEMA National Environmental Management Act, 1998 (Act No. 107 of 1998) NO Nitrogen oxide NO2 Nitrogen dioxide NOX Oxides of nitrogen (NOX = NO + NO2) OFA Overfire Air PM Particulate Matter PM10 Particulate Matter with a diameter of less than 10 µm PM2.5 Particulate Matter with a diameter of less than 2.5 µm SO2 TSP WHO Sulphur Dioxide Total Suspended Particulates World Health Organisation iii

6 TABLE OF CONTENTS EXECUTIVE SUMMARY... i LIST OF ACRONYMS... iii TABLE OF CONTENTS... iv TABLES... v FIGURES... vi 1. Enterprise Details Enterprise Details Location and extent of the Plant Atmospheric Emission License and Other Authorisations Minimum Emission Standards National Ambient Air Quality Standards (NAAQS) Nature of the Process Listed Activity or Activities Process Description Atmospheric emissions resulting from power generation Unit Processes Technical Information Raw Materials Used Appliances and Abatement Equipment Control Technology Atmospheric emissions Point source parameters Point source maximum emission rates (normal operating conditions) Point source maximum emission rates (start-up, shut-down, upset and maintenance conditions) Fugitive emissions Emergency Incidents Impact of Enterprise on the Receiving Environment Analysis of emissions Overview Prevailing climatic conditions Current status of ambient air quality Introduction Data quality Ambient air quality monitoring Sulphur dioxide (SO2) Nitrogen dioxide (NO2) Particulate Matter (PM10) Particulate Matter (PM2.5) Source apportionment Dispersion modelling Models used Model parameterisation Model accuracy Comparison between measured and modelled values Modelled ambient concentrations Modelled operational scenarios Combined emissions scenario Annual and 99th percentile concentrations iv

7 5.4.4 Scenario 1 - Current actual emissions Scenario 2 New plant MES compliance Analysis of Emissions Impact on Human Health Potential health effects Analysis Analysis of Emissions Impact on the Environment Complaints Current or planned air quality management interventions Compliance and Enforcement History Additional Information Summary and conclusions Ambient air quality Predicted ambient concentrations References Formal Declarations TABLES Table 1: Enterprise details Table 2: Site information Table 3: Current government authorisations related to air quality Table 4: Minimum Emission Standards for combustion installations (Category 1) using solid fuel for electricity generation (Sub-category 1.1) with a design capacity equal to or greater than 50 MW heat input per unit Table 5: National Ambient Air Quality Standards for SO2, NO2 and PM10 (DEA, 2009) and PM2.5 (DEA, 2012a) Table 6: Activities listed in GN 893 which are triggered by the Kriel Power Station Table 7: Unit processes at Kriel Power Station Table 8: Raw material used at Kriel Power Station Table 9: Production rates at Kriel Power Station Table 10: Energy sources used at Kriel Power Station Table 11 : Appliance and abatement equipment control technology currently used at Kriel Power Station Table 12: Point sources at Kriel Power Station Table 13: Maximum permitted emission rate of pollutants under normal operating conditions at Kriel Power Station Table 14: Start-ups at Kriel Power Station for the period 2016 to Table 15: Emergency incidents as reported by Kriel Power Station for the 2016/17, 2017/18 and 2018/19 financial years (to date) Table 16: Relative positions of the ambient air quality monitoring stations used in this assessment, to the Kriel Power Station Table 17: Summary ambient 10-minute SO2 average concentrations for the Komati, Kriel Village, Elandsfontein, Kendal and Phola ambient air quality monitoring stations. All concentrations are in μg/m Table 18: Summary ambient hourly SO2 average concentrations for the Komati, Kriel Village, Elandsfontein, Kendal and Phola ambient air quality monitoring stations. All concentrations are in μg/m Table 19: Summary ambient 24-hour SO2 average concentrations for the Komati, Kriel Village, Elandsfontein, Kendal and Phola ambient air quality monitoring stations. All concentrations are in μg/m v

8 Table 20: Summary ambient annual SO2 average concentrations for the Komati, Kriel Village, Elandsfontein, Kendal and Phola ambient air quality monitoring stations. All concentrations are in μg/m Table 21: Summary ambient hourly NO2 average concentrations for the Komati, Kriel Village, Elandsfontein, and Kendal ambient air quality monitoring stations. All concentrations are in μg/m Table 22: Summary ambient annual NO2 average concentrations for the Komati, Kriel Village, Elandsfontein and Kendal ambient air quality monitoring stations. All concentrations are in μg/m Table 23: Summary ambient 24-hour PM10 average concentrations for the Komati, Kriel Village, Elandsfontein, Kendal and Phola ambient air quality monitoring stations. All concentrations are in μg/m Table 24: Summary ambient annual PM10 average concentrations for the Komati, Kriel Village, Elandsfontein, Kendal and Phola ambient air quality monitoring stations. All concentrations are in μg/m Table 25: Summary ambient 24-hour PM2.5 average concentrations for the Komati, Kriel Village, Elandsfontein, Kendal and Phola ambient air quality monitoring stations. All concentrations are in μg/m Table 26: Summary ambient annual PM2.5 average concentrations for the Komati, Kriel Village, Elandsfontein, Kendal and Phola ambient air quality monitoring stations. All concentrations are in μg/m Table 27: Parameterisation of key variables for CALMET Table 28: Parameterisation of key variables for CALPUFF Table 29: Current average emissions (tons/annum) and Eskom requested emission limits (tons/annum) for Kriel Power Station Table 30: Maximum predicted annual average concentration and the highest 99 th percentile concentration at the points of maximum ground-level impact for the two scenarios Table 31: Summary of compliance with the NAAQS for each of the ambient air quality monitoring stations used in this analysis Table 32: Summary of compliance with the NAAQS for ambient air quality predicted for each of the emissions scenarios modelled for Kriel alone Table 33: Complaints register for Kriel Power Station FIGURES Figure 1: Relative location of the Kriel Power Station (Google Earth, 2013) Figure 2: Land-use and sensitive receptors within a 30x30 km block of the Kriel Power Station (shown by the white square) Figure 3: A basic atmospheric emissions mass balance for Kriel Power Station showing the key inputs and outputs. Note that all quantities are expressed in tonnes per annum unless otherwise stated and are based on the 2016/2017 financial year Figure 4: Relative location of the different process units at Kriel Power Station Figure 5: Average monthly maximum and minimum temperature, and average monthly rainfall at Loskop Dam from 1961 to Figure 6: Annual windrose for Kriel Village 2010 to Figure 7: Relative positions of the ambient air quality monitoring stations used in this assessment, to the Kriel Power Station Figure 8: Cumulative percentage occurrence of 10-minute average SO2 concentrations measured at the Komati ambient air quality monitoring station. Only values above the 98 th percentile are shown for clarity purposes vi

9 Figure 9: Cumulative percentage occurrence of 10-minute average SO2 concentrations measured at the Kriel Village ambient air quality monitoring station. Only values above the 99 th percentile are shown for clarity purposes Figure 10: Cumulative percentage occurrence of 10-minute average SO2 concentrations measured at the Elandsfontein ambient air quality monitoring station. Only values above the 99 th percentile are shown for clarity purposes Figure 11: Cumulative percentage occurrence of 10-minute average SO2 concentrations measured at the Kendal ambient air quality monitoring station. Only values above the 99 th percentile are shown for clarity purposes Figure 12: Cumulative percentage occurrence of 10-minute average SO2 concentrations measured at the Phola ambient air quality monitoring station. Only values above the 99 th percentile are shown for clarity purposes Figure 13: Cumulative percentage occurrence of hourly average SO2 concentrations measured at the Komati ambient air quality monitoring station. Only values above the 90 th percentile are shown for clarity purposes Figure 14: Cumulative percentage occurrence of hourly average SO2 concentrations measured at the Kriel Village ambient air quality monitoring station. Only values above the 90 th percentile are shown for clarity purposes Figure 15: Cumulative percentage occurrence of hourly average SO2 concentrations measured at the Elandsfontein ambient air quality monitoring station. Only values above the 90 th percentile are shown for clarity purposes Figure 16: Cumulative percentage occurrence of hourly average SO2 concentrations measured at the Kendal ambient air quality monitoring station. Only values above the 90 th percentile are shown for clarity purposes Figure 17: Cumulative percentage occurrence of hourly average SO2 concentrations measured at the Phola ambient air quality monitoring station. Only values above the 90 th percentile are shown for clarity purposes Figure 18: Cumulative percentage occurrence of daily average SO2 concentrations measured at the Komati ambient air quality monitoring station Figure 19: Cumulative percentage occurrence of daily average SO2 concentrations measured at the Kriel Village ambient air quality monitoring station Figure 20: Cumulative percentage occurrence of daily average SO2 concentrations measured at the Elandsfontein ambient air quality monitoring station Figure 21: Cumulative percentage occurrence of daily average SO2 concentrations measured at the Kendal ambient air quality monitoring station Figure 22: Cumulative percentage occurrence of daily average SO2 concentrations measured at the Phola ambient air quality monitoring station Figure 23: Cumulative percentage occurrence of hourly average NO2 concentrations measured at the Komati ambient air quality monitoring station. Only values above the 90 th percentile are shown for clarity purposes Figure 24: Cumulative percentage occurrence of hourly average NO2 concentrations measured at the Kriel Village ambient air quality monitoring station. Only values above the 90 th percentile are shown for clarity purposes Figure 25: Cumulative percentage occurrence of hourly average NO2 concentrations measured at the Elandsfontein ambient air quality monitoring station. Only values above the 90 th percentile are shown for clarity purposes Figure 26: Cumulative percentage occurrence of hourly average NO2 concentrations measured at the Kendal ambient air quality monitoring station. Only values above the 90 th percentile are shown for clarity purposes vii

10 Figure 27: Cumulative percentage occurrence of daily average PM10 concentrations measured at the Komati ambient air quality monitoring station Figure 28: Cumulative percentage occurrence of daily average PM10 concentrations measured at the Kriel Village ambient air quality monitoring station Figure 29: Cumulative percentage occurrence of daily average PM10 concentrations measured at the Elandsfontein ambient air quality monitoring station Figure 30: Cumulative percentage occurrence of daily average PM10 concentrations measured at the Kendal ambient air quality monitoring station Figure 31: Cumulative percentage occurrence of daily average PM10 concentrations measured at the Phola ambient air quality monitoring station Figure 32: Cumulative percentage occurrence of daily average PM2.5 concentrations measured at the Komati ambient air quality monitoring station Figure 33: Cumulative percentage occurrence of daily average PM2.5 concentrations measured at the Kriel Village ambient air quality monitoring station Figure 34: Cumulative percentage occurrence of daily average PM2.5 concentrations measured at the Elandsfontein ambient air quality monitoring station Figure 35: Cumulative percentage occurrence of daily average PM2.5 concentrations measured at the Kendal ambient air quality monitoring station Figure 36: Cumulative percentage occurrence of daily average PM2.5 concentrations measured at the Phola ambient air quality monitoring station Figure 37: Area plot of the range of average diurnal SO2 concentrations for all monitoring stations across the Mpumulanga Highveld ( ) Figure 38: Area plot of the range of average diurnal NO2 concentrations for all monitoring stations across the Mpumulanga Highveld ( ) Figure 39: Area plot of the range of average diurnal PM10 concentrations for all monitoring stations across the Mpumulanga Highveld ( ) Figure 40: Area plot of the range of average diurnal PM2.5 concentrations for all monitoring stations across the Mpumulanga Highveld ( ) Figure 41: Area plot of the range of average diurnal SO2, NO2, PM10 and PM2.5 concentrations for all monitoring stations across the Mpumulanga Highveld ( ) Figure 42: TAPM and CALPUFF modelling domains for Kriel Figure 43: Conceptual illustration of the method used to compare modelled and measured ambient hourly SO2 concentrations Figure 44: Comparison between modelled and measured hourly SO2 concentrations at the Kriel Village ambient air quality monitoring station. An exact correlation would mean 100% Figure 45: Comparison between modelled and measured hourly SO2 concentrations at the Kendal ambient air quality monitoring station. An exact correlation would mean 100% Figure 46: Comparison between modelled and measured hourly SO2 concentrations at the Phola ambient air quality monitoring station. An exact correlation would mean 100% Figure 47: Comparison between modelled and measured hourly SO2 concentrations at the Komati ambient air quality monitoring station. An exact correlation would mean 100% Figure 48: Comparison between modelled and measured hourly SO2 concentrations at the Elandsfontein ambient air quality monitoring station. An exact correlation would mean 100% Figure 49: Predicted annual average SO2 concentrations (µg/m 3 ) resulting from actual emissions from Kriel Power Station (Scenario 1) Figure 50: 99 th percentile concentration of the predicted 24-hour SO2 concentrations for actual emissions from Kriel Power Station (Scenario 1) Figure 51: 99 th percentile of the predicted 1-hour SO2 concentrations resulting from actual emissions from Kriel Power Station (Scenario 1) viii

11 Figure 52: Predicted annual average NO2 concentrations (µg/m 3 ) resulting from actual emissions from Kriel Power Station (Scenario 1) Figure 53: 99 th percentile of the predicted 1-hour NO2 concentrations resulting from actual emissions from Kriel Power Station (Scenario 1) Figure 54: Predicted annual average PM10 concentrations (µg/m 3 ) resulting from actual emissions from Kriel Power Station (Scenario 1) Figure 55: 99 th percentile of the predicted 24-hour PM10 concentrations resulting from actual emissions from Kriel Power Station (Scenario 1) Figure 56: Predicted annual average PM2.5 concentrations (µg/m 3 ) resulting from actual emissions from Kriel Power Station (Scenario 1) Figure 57: 99 th percentile of the predicted 24-hour PM2.5 concentrations resulting from actual emissions from Kriel Power Station (Scenario 1) Figure 58: Predicted annual average secondary particulate concentrations (µg/m 3 ) resulting from actual emissions from Kriel Power Station (Scenario 1) Figure 59: 99 th percentile of the predicted 24-hour secondary particulate concentrations resulting from actual emissions from Kriel Power Station (Scenario 1) Figure 60: Predicted annual average SO2 concentrations (µg/m 3 ) assuming new plant MES from Kriel Power Station (Scenario 2) Figure 61: 99 th percentile concentration of the predicted 24-hour SO2 concentrations assuming new plant MES from Kriel Power Station (Scenario 2) Figure 62: 99 th percentile of the predicted 1-hour SO2 concentrations assuming new plant MES emission from Kriel Power Station (Scenario 2) Figure 63: Predicted annual average NO2 concentrations resulting assuming new plant MES from Kriel Power Station (Scenario 2) Figure 64: 99 th percentile of the predicted 1-hour NO2 concentrations assuming new plant MES from Kriel Power Station (Scenario 2) Figure 65: Predicted annual average PM10 concentrations resulting from new plant MES from Kriel Power Station (Scenario 2) Figure 66: 99 th percentile of the predicted 24-hour PM10 concentrations resulting from new plant MES from Kriel Power Station (Scenario 2) Figure 67: Predicted annual average PM2.5 concentrations (µg/m 3 ) assuming new plant MES from Kriel Power Station (Scenario 2) Figure 68: 99 th percentile of the predicted 24-hour PM2.5 concentrations assuming new plant MES from Kriel Power Station (Scenario 2) Figure 69: Predicted annual average secondary particulate concentrations (µg/m 3 ) assuming new plant MES from Kriel Power Station (Scenario 2) Figure 70: 99 th percentile of the predicted 24-hour secondary particulate concentrations assuming new plant MES from Kriel Power Station (Scenario 2) ix

12 1. Enterprise Details 1.1 Enterprise Details Entity details for Eskom s Kriel Power Station are listed in Table 1. Table 1: Enterprise details Entity Name: Trading as: Type of Enterprise, e.g. Company/Close Corporation/Trust, etc.: Company/Close Corporation/Trust Registration Number (Registration Numbers if Joint Venture): Registered Address: Eskom Holdings SOC Limited Kriel Power Station State owned company 2002/015527/30 Megawatt Park, Maxwell Drive, Sunninghill, Sandton Postal Address: Private Bag X 5009 Kriel 2271 Telephone Number (General): Fax Number (General): Company Website: Industry Type/Nature of Trade: Coal-fired Power Station Electricity Generation Land Use Zoning as per Town Planning Scheme: Agricultural/Heavy industry Land Use Rights if outside Town Planning Scheme: - Responsible Person: Gersh Bonga Emissions Control Officer: Gersh Bonga Telephone Number: Cell Phone Number: Fax Number: Address: BongaMG@eskom.co.za After Hours Contact Details:

13 1.2 Location and extent of the Plant Kriel Power Station (hereinafter referred to as Kriel ) is located in the Mpumalanga Province, approximately 7.5 km west of the town of Kriel. The surrounding land use is zoned as agricultural, comprising low density farmsteads and infrastructure, crops on the arable soils and grazing. It borders the Matla Power Station and the Matla Mine (Exxaro). Site information is provided in Table 2 and the relative location to key landmarks is shown in Figure 1. Figure 1: Relative location of the Kriel Power Station (Google Earth, 2013) Table 2: Site information Physical Address of the Plant (Licensed Premises): Description of Site (Where No Street Address): Coordinates (latitude, longitude) of Approximate Centre of Operations (Decimal Degrees): Kriel Power Station, Ogies Bethal Road 15 km from Kriel Town On Ogies Bethal Road, 15 km from Kriel Town Latitude: ,03 S Longitude: ,95 E Coordinates (UTM) of Approximate Centre of Operations: E S Extent (km²): Elevation Above Mean Sea Level (m) Province: Mpumalanga Province District/Metropolitan Municipality: Nkangala District Municipality Local Municipality: Emalahleni Local Municipality Designated Priority Area (if applicable): Highveld Priority Area 11

14 Receptor Distance(km) Direction Kriel 7.5 E Thubelihle 11.5 ENE Residential area 9.4 NNE Agricultural lands Immediate Surrounding Kriel 7.5 E Figure 2: Land-use and sensitive receptors within a 30x30 km block of the Kriel Power Station (shown by the white square) 1.3 Atmospheric Emission License and Other Authorisations Kriel currently holds a valid Atmospheric Emission Licence (AEL) (Ref no. 17/4/AEL/MP312/11/09) for electricity production, the storage and handling of ore and coal, and the storage of petroleum products in terms of the listed activities promulgated in the Minimum Emission Standards (GNR 893 November 2013) under the National Environmental Management: Air Quality Act, 2004 (Act No. 39 of 2004) [NEMAQA]. The AEL specifies permissible stack emission concentrations for NOx, SO2 and for PM. The licence specifies a number of compliance conditions as well as conditions for emission monitoring and management of abnormal releases. The current governmental authorisations, permits and licenses related to air quality management are provided in Table 3. Table 3: Current government authorisations related to air quality AEL Reference number: Date of AEL: Category of the listed activity* 17/4/AEL/MP312/11/09 20/05/2012 *See Table 6 for more detail Category 1 Category 2 Category 5 12

15 1.3.1 Minimum Emission Standards In terms of NEMAQA, all of Eskom's coal- and liquid fuel-fired power stations are required to meet the Minimum Emission Standards (MES) contained in GNR 893 on 22 November 2013 ("GNR 893"), as amended promulgated in terms of Section 21 of the NEMAQA. GNR 893 does provide for transitional arrangements in respect of the requirement for existing plants to meet the MES and provides that less stringent limits had to be achieved by existing plants by 1 April 2015, and more stringent new plant limits need to be achieved by existing plants by 1 April The MES are listed in Table 4. Table 4: Minimum Emission Standards for combustion installations (Category 1) using solid fuel for electricity generation (Sub-category 1.1) with a design capacity equal to or greater than 50 MW heat input per unit Substance Plant status MES mg/nm 3 under normal conditions of 10% O2, 273 K and kpa Particulate Matter New 50 Existing 100 Sulphur dioxide New Existing Oxides of nitrogen New 750 Existing National Ambient Air Quality Standards (NAAQS) The effects of air pollutants on human health are plentiful with short-term, or acute effects, and chronic, or long-term, effects. Different groups of people are affected differently, depending on their level of sensitivity, with the elderly and young children being more susceptible. Factors that link the concentration of an air pollutant to an observed health effect are the magnitude of the concentration and the duration of the exposure to that particular air pollutant concentration. Criteria pollutants occur throughout urban and industrial environments. Their effects on human health and the environment are well documented (e.g. WHO, 1999; 2003; 2005). South Africa has accordingly established NAAQS for the criteria pollutants, i.e. sulphur dioxide (SO2), nitrogen dioxide (NO2), carbon monoxide (CO), respirable particulate matter (PM10), ozone (O3), lead (Pb), benzene (C6H6) (DEA, 2009) and PM2.5 (DEA, 2012a). The NAAQS for SO2, NO2, PM10 and PM2.5 are listed in Table 5. The NAAQS consist of a limit value and a permitted frequency of exceedances. The limit value is the fixed concentration level aimed at reducing the harmful effects of a pollutant. The permitted frequency of exceedance represents the acceptable number of exceedances of the limit value expressed as the 99 th percentile. Compliance with the ambient standard implies that the frequency of exceedance of the limit value does not exceed the permitted tolerance. Being a health-based standard, ambient concentrations that comply with the standard imply that air quality poses a tolerable risk to human health, while exposure to ambient concentrations that do not comply with the standard, implies that there is an intolerable risk to human health. 13

16 Table 5: National Ambient Air Quality Standards for SO 2, NO 2 and PM 10 (DEA, 2009) and PM 2.5 (DEA, 2012a) Pollutants Averaging period Limit value (µg/m 3 ) Number of permissible exceedances per annum 1 hour SO2 24 hour year 50 0 NO2 1 hour year 40 0 PM10 24-hour 75 4 Calendar year 50 (40) 0 PM hour 40 (25) 4 Calendar year 20 (15) 0 Figures in brackets are due for implementation on 1 January Nature of the Process 2.1 Listed Activity or Activities Table 6: Activities listed in GN 893 which are triggered by the Kriel Power Station. Category of Listed Activity 1: Combustion Installations 2: Petroleum Industry, the production of gaseous and liquid fuels as well as petrochemicals from crude oil, coal, gas or biomass 5: Mineral Processing, Storage and Handling Sub-category of the Listed Activity 1.1: Solid Fuel Combustion Installations 2.4: Storage and Handling of Petroleum Products 5.1 Storage and Handling of Ore and Coal Description of the Listed Activity Solid fuels combustion installations used primarily for steam raising or electricity generation. All installations with design capacity equal to or greater than 50 MW heat input per unit, based on the lower calorific value of the fuel used. All permanent immobile liquid storage facilities at a single site with a combined storage capacity of greater than 1000 cubic meters. Storage and handling of ore and coal not situated on the premises of a mine or works as defined in the Mines Health and Safety Act 29/ Process Description Eskom Holdings SOC Limited is a South African utility that generates, transmits and distributes electricity. The bulk of that electricity is generated by large coal-fired power stations that are situated close to the sources of coal, with most of the stations occurring on the Mpumalanga Highveld area. Kriel is one such station (Figure 1). Kriel has total installed capacity of MW, with an installed capacity of 500MW per unit. At Kriel, just like in all other coal-fired power stations, pulverised coal is combusted in order to heat water in boilers to generate steam at high temperatures (between 500 C and 535 C) and pressures. The steam, in turn, is used to drive the turbines, which are connected, to rotating magnets and electricity is generated. The energy in the fuel (coal) is thus converted to electricity (Figure 3). 14

17 Figure 3: A basic atmospheric emissions mass balance for Kriel Power Station showing the key inputs and outputs. Note that all quantities are expressed in tonnes per annum unless otherwise stated and are based on the 2016/2017 financial year Atmospheric emissions resulting from power generation Emissions from coal combustion include SO2, NOX and particulate matter. SO2 is produced from the combustion of sulphur that is bound in coal. NOX is produced from thermal fixation of atmospheric nitrogen in the combustion flame and from oxidation of nitrogen bound in the coal. The quantity of NOX produced is directly proportional to the temperature of the flame. PM, SO2 and NOX are released to the atmosphere via the power station stacks. The non-combustible portion of the fuel remains as solid waste. The coarser, heavier by-product from the combustion process, is called bottom ash and is extracted from the boiler. The lighter, finer portion is fly ash and, in the absence of abatement, it is emitted as particulates through the stacks. At Kriel, the precipitator plant removes dust in the flue gas stream before the gas passes through the ID Fan and exhausted to atmosphere via stack see fig 2.Each unit is served by a precipitator Plant equipped with precipitator fields that is installed between the air heater outlet and the ID fans. The dust that is removed from the flue gas is collected in hoppers at the base of the filter plant. This dust is transported via a pneumatic conveying system to ash silos from where it is conditioned and conveyed to the wet ash dump. 2.3 Unit Processes A summary of the different unit process is provided in Table 7. The relative location of these is shown in Figure 4. 15

18 Table 7: Unit processes at Kriel Power Station Unit Process Function of Unit Process Batch or Continuous Process Boiler Unit 1 Generation of electricity from coal Continuous Boiler Unit 2 Generation of electricity from coal Continuous Boiler Unit 3 Generation of electricity from coal Continuous Boiler Unit 4 Generation of electricity from coal Continuous Boiler Unit 5 Generation of electricity from coal Continuous Boiler Unit 6 Generation of electricity from coal Continuous Coal stockpile Storage of coal Continuous Fuel oil storage tanks Storage of fuel oil Continuous Boilers 1-6 Figure 4: Relative location of the different process units at Kriel Power Station 16

19 3. Technical Information 3.1 Raw Materials Used The permitted raw materials consumption rate, the permitted production rates and the energy sources at Kriel are listed in Table 8 to Table 10 according to the AEL. Table 8: Raw material used at Kriel Power Station Raw material Maximum permitted consumption rate (Volume) Units (quantity / period) Coal tons/month Fuel oil tons/month Table 9: Production rates at Kriel Power Station Maximum Production capacity permitted (Volume) Units Product/by-product (quantity / period) Electricity MW Table 10: Energy sources used at Kriel Power Station Energy source Sulphur content of fuel (%) Ash content of fuel (%) CV (MJ/kg) Maximum permitted consumption rate (Volume) Units (quantity / period) Coal 0.6 to 1.2% 27 to 32 % tons/month Fuel oil tons/month 3.2 Appliances and Abatement Equipment Control Technology Abatement equipment control technology at Kriel is presented in Table 11. It should be noted that the abatement equipment is only for the control of PM emissions. Neither NOx nor SO2 emissions are controlled directly at the power station. Table 11 : Appliance and abatement equipment control technology currently used at Kriel Power Station. Appliance Name Electrostatic Precipitators (ESPs) SO3 Plant (i.e. flue gas conditioning plant) Appliance Type / Description Electrostatic Precipitator (ESPs) SO3 Injection Appliance Function / Purpose An ESP removes particles from the flue stream using the force of an induced electrostatic charge on the ash particle that is then attracted to and held on a plate. The efficiency of ESPs is dependent on the electrical resistivity of the ash particles (and the particle size). High frequency power supply with further enhance performance. SO3 injection decreases the resistivity of the particles, and significantly improves the performance of the ESP. 17

20 4. Atmospheric emissions 4.1 Point source parameters The physical data for the stacks at Kriel are listed in Table 12. Emission concentrations and emission rates for current production and proposed operational levels are shown in Table 13. The boiler units operate continuously, i.e. 24 hours a day. Table 12: Point sources at Kriel Power Station Point Source Code Stack 1 Stack 2 Source name Boiler unit 1 Boiler unit 2 Boiler unit 3 Boiler unit 4 Boiler unit 5 Boiler unit 6 Latitude (UTM) (m) Longitude (UTM) (m) Height of Release Above Ground (m) Height above nearby building (m) Diameter at Stack Tip / Vent Exit (m) Actual Gas Exit Temp ( 0 C) Actual stack gas volumetric flow (m 3 /hr) Actual Gas Exit Velocity (m/s) Type of emission (continuous/ batch) E S Continuous E S Continuous 4.2 Point source maximum emission rates (normal operating conditions) Table 13: Maximum permitted emission rate of pollutants under normal operating conditions at Kriel Power Station Point source code Sub- Category Pollutant name Maximum emission rate (mg/nm 3 ) Date to be achieved by Averaging period Duration of emissions April 2015 Stack 1 (Units 1-3) & Stack 2 (Units 4-6) 1.1 SO2 NOx PM April March 2025 Daily April April March April 2020 Daily April March 2020 Monthly 50 1 April 2020 Daily Continuous Continuous Continuous

21 4.3 Point source maximum emission rates (start-up, shut-down, upset and maintenance conditions) Kriel Power Station maintains a record of all start-ups that occur, as well as the type of startup. Full details of these for the years 2016 to 2018are provided in Table 14. Table 14: Start-ups at Kriel Power Station for the period 2016 to Period Number of Hot Start- ups Number of Cold Start-ups 2016/ / Fugitive emissions Fugitive emissions at Kriel result from coal storage and handling, and ash handling, which must be controlled through the implementation of dust management plans. Fugitive emission management is guided by the National Dust Control Regulations (GNR November 2013) as promulgated under NEMAQA. Such fugitive emissions are not assessed in this AIR. Kriel s dust management plan is included as Annexure A where dust emission sources and measures that have been put in place to manage these, are presented. Fugitive emissions are extremely difficult to quantify, as they are highly variable in time and space. Fugitive emissions from the ashing facility are highest on the active face (especially in the case of dry ashing) and when wind speeds are high. Fugitive emissions also depend on measures that have been put in place to suppress dust generation, for example vegetation of the ashing facility and sprinklers to suppress dust. The dust fall-out resulting from the fugitive emissions is monitored with dust buckets.

22 4.5 Emergency Incidents A record is maintained of all emergency incidents occurring at Eskom Power Stations reported in terms of Section 30 of the National Environmental Management Act, 1998 (Act No. 107 of 1998) (NEMA). Kriel incurred 10 emergency incidents in the 2016/17 financial year and only 3 in the 2017/18 financial year (Table 15). Table 15: Emergency incidents as reported by Kriel Power Station for the 2016/17, 2017/18 and 2018/19 financial years (to date) No. Stack Date of incident commencement Date of incident end Date when investigation was sent 1 North 22 April April May-16 2 South 22 April April May-16 3 North 19 July July August South 19 July July August South 19 August August September North 21 September September October South 21 September September October 2016 Cause of incident Poor trapping efficiency on unit 3 due to the unavailability of the 3.2 blow tank. The electrostatic precipitator performance dropped to 43.25% resulting in high emissions on the north stack. Unit 1 and 2 handful hopper levels due to the unsteady supply of conveyor air. To manage the situation, the mobile compressor was connected. Compressors were not available, resulting in high hopper levels and high particulate emissions. Additional, electrostatic precipitator fields were underperforming and numerous plate rappers were defective. Blow tank failure on units 1 and 2 from July 2016 due to moisture in the control air system that resulted in sequent failures. Blow tank failure on units 1 and 2 from July 2016 due to moisture in the control air system that resulted in sequent failures Dust handling plant problems on unit 5 which led to high hopper levels. SO3 plant problems on unit 5 exacerbated the problem. Failures in the dust conditioning plant resulted in a fly ash backlog at the conveying system and this eventually resulted in high hopper levels and high emissions. The conditioner cut-off valve failed to close when the conditioner was turned off resulting in uncontrolled fly ash flowing through the said conditioner and spilling over both overland conveyor belts. The spillage resulted in conveyor belts being unable to transport fly ash, hence the backlog on the fly ash transporting system. Failures in the dust conditioning plant resulted in a fly ash backlog at the conveying system and this eventually resulted in high hopper levels and high emissions. The conditioner cut-off valve failed to close when the conditioner was turned off resulting in uncontrolled fly ash flowing through the said conditioner and spilling over both overland conveyor belts. The spillage resulted in conveyor belts being unable to transport fly ash, hence the backlog on the fly ash transporting system.

23 No. Stack Date of incident commencement Date of incident end Date when investigation was sent 8 South 08 January January January South 21 February February March North 24 March March April South 05 October October October South 19 November December December 2017 Cause of incident On 6 January Unit 4 light up was initiated, returning the unit back to service from a 95 day overhaul outage. During the commissioning, testing and optimization of the unit, it was discovered that the SO3 electric heater was failing to start due to a communication signal fault from the programmable logic controller which was not adequately sustaining on the 24 DC relay for the SO3 electric heater to start. This resulted in the PM limit to be exceeded for more than the allowed 72h grace period after a cold start. A gas leak was noted at the unit 6 common injection lance. This required immediate maintenance and the SO3 plant was shut down. This caused PM emission limit exceedances. A Sulphur dosing pump tripped on the common plant injection lance ducting. The affected plant was shut down and fully isolated for repairs the very same day. Online maintenance exceeded 48h. Unit 6 was returned to service after a general overhaul. During commissioning, an electrical issue was noted, resulting in high PM emissions. Additionally other upset conditions were experienced at unit 4 where a feed pump was unavailable. A detailed root cause analysis will be provided. Rapping sequence on the collector plates had been affected by a technical problem. ESP efficiency was reduced by 39%. Re-entrainment of dust into the flue gas contributed to the high emissions. 13 North 22 January January February 2018 Unavailability of SO3 plant on unit 3 and failure of fisher control valve (responsible for supply of auxiliary steam) 21

24 5. Impact of Enterprise on the Receiving Environment 5.1 Analysis of emissions Overview The application for alternate limits does not mean that Komati s SO2 emissions will change from what they are currently and particulate emissions and NOx will improve over the next 5 years. The requested interim emissions limits have been expressed as a ceiling limit to ensure that Eskom can comply with the same under all normal operating circumstances given the variability of emissions from day to day. This assessment is then based on a detailed analysis of the prevailing climate together with an analysis of air quality monitoring data. Thereafter dispersion modelling is used to predict ambient air pollution concentrations in the areas where there are no physical measurements for worst case scenario under the requested interim PM and SO2 emissions. This analysis is presented in the following section Prevailing climatic conditions Temperature and rainfall The climate of a location is affected by its latitude, terrain, and altitude, as well as nearby water bodies and their currents. Climates can be classified according to the average and the typical ranges of different variables, most commonly temperature and precipitation. The Mpumalanga Highveld is located in temperate latitudes between 25 and 26º S and 28 to 29º E, and approximately m above sea level. As a result, it experiences a temperate climate with summer rainfall and dry winters according to the Köppen Climate Classification system. The Mpumalanga Highveld is relatively flat and experiences similar climate throughout. Temperature and rainfall for north-eastern parts of the Mpumalanga Highveld are therefore best illustrated by the long term measurements at the South African Weather Service station at the Loskop Dam. Winters are mild and dry with average maximum temperatures dropping below 25 C in May, June, July, and August but cold at night in June and July when temperatures drop below 7 C. Average summer maximums exceed 27 C from September to March, with extremes reaching more than 30 C particularly from December to January. The area experienced an annual average rainfall of 640 mm with rain occurring almost exclusively in the summer months from October to March, with more than 60% of the rain occurring from November to February (Figure 5). Rainfall seldom occurs in winter between April and September.

25 Temperature (⁰C) Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Average monthly rainfall (mm) Mean maximum temperature Mean minimum temperature Average monthly rainfall Figure 5: Average monthly maximum and minimum temperature, and average monthly rainfall at Loskop Dam from 1961 to 1990 Wind The Mpumalanga Highveld is relatively flat with little influence by topography on the wind flow. Winds at Kriel are best represented by the wind measured at Eskom s monitoring station at Kriel Village, 7km east of the power station. The windrose in Figure 6 illustrates the frequency of hourly wind from the 16 cardinal wind directions, with wind indicated from the direction it blows, i.e. easterly winds blow from the east. It also illustrates the frequency of average hourly wind speed in six wind speed classes. The winds are predominantly northerly and north-westerly winds (Figure 6). Occasional easterly winds occur and are associated with the relative location and strength of the Indian Ocean anticyclone. The winds are generally light with 40% of all winds less than 3 m/s and 60% of all winds less than 6 m/s. Figure 6: Annual windrose for Kriel Village 2010 to

26 5.2 Current status of ambient air quality Introduction In the section that follows an analysis of ambient air quality from various ambient air quality monitoring stations in the vicinity of Kriel is presented. Data from the Elandsfontein, Kendal, Komati, Kriel and Phola ambient air quality monitoring stations is included in the analysis. Ambient data for the three-year period 2015, 2016 and 2017 at the four monitoring stations provide a direct physical measure of ambient air quality in the area and of the sources that influence air quality at the monitoring sites, including emissions from Kriel. The information is framed within the context of the National Ambient Air Quality Standards (NAAQS), and compliance-related conclusions are drawn as a function of those standards. The NAAQS are in turn made up of two components namely a limit value (as defined by a particular threshold concentration) together with the number of times the limit may be may be exceeded. As such, compliance is a function of not where there are exceedances of the limit value but whether the number of exceedances of the limit value is within the allowable number of exceedances. If the allowed frequency of limit-value-exceedances is surpassed, then there is non-compliance with the standard Data quality Data quality is variable and there was poor data recovery for some of the stations in some of the monitoring years. Reasons for poor data recovery can include power and equipment failures, theft and vandalism. Data recovery of 80% and above is generally considered acceptably representative of the monitoring year. Where data recovery is less than 80%, the results must be treated with caution and cannot be viewed as definitive. The data are included simply for the sake of completeness, where data recovery is seen to be less than 80% Ambient air quality monitoring The positions of the ambient air quality monitoring stations used in this assessment are summarised in Table 16 and illustrated in Figure 7. Ambient SO2, NO2 and PM10 concentrations and meteorological parameters are routinely monitored at the stations. In the following sections, the data are presented in frequency distributions that serve to indicate the occurrence of different concentrations measured. In presenting that information it is necessary to detail the data recovery at the station. Table 16: Relative positions of the ambient air quality monitoring stations used in this assessment, to the Kriel Power Station. Site Name From Power Station Distance Direction From Power Station Latitude Longitude Elandsfontein (EL) 26.6 km West-South-West Kriel Village (KV) 9.7 km East-North-East 26 15'4.42"S 29 15'22.88"E Kendal 25.6 km North-West Phola 31.1 km North-North-West Komati 32.5 km North-East Secunda 34.2 km South-South-West 26 32'54.88"S 29 4'48.20"E 24

27 Figure 7: Relative positions of the ambient air quality monitoring stations used in this assessment, to the Kriel Power Station Sulphur dioxide (SO 2) 10-minute averages Cumulative percentage occurrence graphs of 10-minute average SO2 concentrations are shown in Figure 8, Figure 9, Figure 10, Figure 11 and Figure 12, together with a summary table of compliance with the NAAQS in Table 17. It can be seen from the figures and the table that there is compliance with the NAAQS for all the stations. Kendal has a higher SO2 loading than the other stations, which is consistent with its position downwind of the Kendal power station. Data recovery is seen to be generally poor at Elandsfontein, and it is only Kriel Village that has a full complement of data. Where data recovery is seen to be less than 80%, compliance with the NAAQS cannot be assured. 25

28 Figure 8: Cumulative percentage occurrence of 10-minute average SO 2 concentrations measured at the Komati ambient air quality monitoring station. Only values above the 98 th percentile are shown for clarity purposes. Figure 9: Cumulative percentage occurrence of 10-minute average SO 2 concentrations measured at the Kriel Village ambient air quality monitoring station. Only values above the 99 th percentile are shown for clarity purposes. 26

29 Figure 10: Cumulative percentage occurrence of 10-minute average SO 2 concentrations measured at the Elandsfontein ambient air quality monitoring station. Only values above the 99 th percentile are shown for clarity purposes. Figure 11: Cumulative percentage occurrence of 10-minute average SO 2 concentrations measured at the Kendal ambient air quality monitoring station. Only values above the 99 th percentile are shown for clarity purposes. 27

30 Figure 12: Cumulative percentage occurrence of 10-minute average SO 2 concentrations measured at the Phola ambient air quality monitoring station. Only values above the 99 th percentile are shown for clarity purposes. Table 17: Summary ambient 10-minute SO 2 average concentrations for the Komati, Kriel Village, Elandsfontein, Kendal and Phola ambient air quality monitoring stations. All concentrations are in μg/m 3. Ambient air quality No of monitoring station exceedances Allowable Maximum Average % data recovery ,3 40,0 44,8% Komati ,2 55,4 69,9% ,7 44,1 84,8% ,8 41,2 82,3% Kriel ,6 28,9 87,2% ,2 32,1 87,4% ,0 36,3 23,0% Elandsfontein ,2 34,4 74,3% ,2 39,4 56,4% ,1 51,8 37,5% Kendal ,1 33,6 90,4% ,7 36,1 82,1% ,9 26,4 40,0% Phola ,5 24,4 95,6% ,5 26,0 94,2% * Compliance with the NAAQS indicated by green, non-compliance by red. ** Light green data recovery is 80% or higher, light red less than 80%. Hourly averages Hourly average sulphur dioxide concentrations are shown in cumulative percentage occurrence graphs in Figure 13, Figure 14, Figure 15, Figure 16 and Figure 17 together with a summary table of compliance with the NAAQS in Table 18. It can be seen from the graphs and the table that there was NAAQS compliance at all the stations with Kendal and Kriel experiencing the highest SO2 loading. The poor data recovery at Elandsfontein, Komati for 2016, and Kendal and Phola for 2015 means, however, that compliance cannot be assured at these stations and years. 28

31 Figure 13: Cumulative percentage occurrence of hourly average SO 2 concentrations measured at the Komati ambient air quality monitoring station. Only values above the 90 th percentile are shown for clarity purposes. Figure 14: Cumulative percentage occurrence of hourly average SO 2 concentrations measured at the Kriel Village ambient air quality monitoring station. Only values above the 90 th percentile are shown for clarity purposes. 29

32 Figure 15: Cumulative percentage occurrence of hourly average SO 2 concentrations measured at the Elandsfontein ambient air quality monitoring station. Only values above the 90 th percentile are shown for clarity purposes. Figure 16: Cumulative percentage occurrence of hourly average SO 2 concentrations measured at the Kendal ambient air quality monitoring station. Only values above the 90 th percentile are shown for clarity purposes. 30

33 Figure 17: Cumulative percentage occurrence of hourly average SO 2 concentrations measured at the Phola ambient air quality monitoring station. Only values above the 90 th percentile are shown for clarity purposes. Table 18: Summary ambient hourly SO 2 average concentrations for the Komati, Kriel Village, Elandsfontein, Kendal and Phola ambient air quality monitoring stations. All concentrations are in μg/m 3. Ambient air quality monitoring station No of exceedances Allowable Maximum Average % data recovery ,3 36,4 91,1% Komati ,7 55,2 74,3% ,1 44,1 92,0% ,7 41,1 83,5% Kriel ,3 28,9 87,5% ,3 32,1 87,7% ,5 37,6 59,3% Elandsfontein ,6 34,4 75,3% ,3 39,4 56,8% ,0 51,8 36,4% Kendal ,7 33,7 90,8% ,0 36,1 82,3% ,2 26,2 77,4% Phola ,6 24,4 96,4% ,7 26,2 95,2% * Compliance with the NAAQS indicated by green, non-compliance by red. ** Light green data recovery is 80% or higher, light red less than 80%. Daily averages Daily average sulphur dioxide concentrations are shown in a cumulative percentage occurrence graphs in Figure 18, Figure 19, Figure 20, Figure 21 and Figure 22, together with a summary table of compliance with the NAAQS in Table 19. Non-compliance with the NAAQS is evident at Komati (2016), Kriel (2015) and Kendal (2015 and 2017). Such non-compliance with the daily standard is indicative of sustained high loading of SO2 in those areas and it is instructive to note that Kendal had 12 and 13 days where the daily limit value was exceeded (a maximum of 4 is allowed). 31

34 Figure 18: Cumulative percentage occurrence of daily average SO 2 concentrations measured at the Komati ambient air quality monitoring station. Figure 19: Cumulative percentage occurrence of daily average SO 2 concentrations measured at the Kriel Village ambient air quality monitoring station. 32

35 Figure 20: Cumulative percentage occurrence of daily average SO 2 concentrations measured at the Elandsfontein ambient air quality monitoring station. Figure 21: Cumulative percentage occurrence of daily average SO 2 concentrations measured at the Kendal ambient air quality monitoring station. 33

36 Figure 22: Cumulative percentage occurrence of daily average SO 2 concentrations measured at the Phola ambient air quality monitoring station. Table 19: Summary ambient 24-hour SO 2 average concentrations for the Komati, Kriel Village, Elandsfontein, Kendal and Phola ambient air quality monitori ng stations. All concentrations are in μg/m 3. Ambient air quality monitoring station No of exceedances Allowable Maximum Average % data recovery ,0 36,1 94,0% Komati ,5 55,0 76,7% ,1 43,3 95,1% ,0 40,5 87,9% Kriel ,1 28,7 89,0% ,4 31,9 89,9% ,0 37,7 63,3% Elandsfontein ,9 34,7 78,1% ,6 39,8 58,1% ,4 50,3 39,2% Kendal ,6 33,7 93,2% ,5 22,9 68,8% ,3 26,2 41,4% Phola ,2 24,4 98,6% ,8 26,2 97,8% * Compliance with the NAAQS indicated by green, non-compliance by red. ** Light green data recovery is 80% or higher, light red less than 80%. Annual averages Annual average SO2 concentrations are shown in Table 20 for the monitoring stations. The Komati station has non-compliance with the standard for 2015 and Kendal a non-compliance with the standard in These non-compliances with the annual average standard again reflect sustained, elevated concentrations of SO2 in the areas of the monitoring stations. 34

37 Table 20: Summary ambient annual SO 2 average concentrations for the Komati, Kriel Village, Elandsfontein, Kendal and Phola ambient air quality monitoring stations. All concentrations are in μg/m 3. Ambient air quality monitoring station Compliance Maximum Average % data recovery 2015 Yes 134,0 36,1 94,0% Komati 2016 No 172,5 55,0 76,7% 2017 Yes 122,1 43,3 95,1% 2015 Yes 159,0 40,5 87,9% Kriel 2016 Yes 142,1 28,7 89,0% 2017 Yes 207,4 31,9 89,9% 2015 Yes 133,0 37,7 63,3% Elandsfontein 2016 Yes 129,9 34,7 78,1% 2017 Yes 212,6 39,8 58,1% 2015 No 256,4 50,3 39,2% Kendal 2016 Yes 166,6 33,7 93,2% 2017 Yes 119,5 22,9 68,8% 2015 Yes 120,3 26,2 41,4% Phola 2016 Yes 93,2 24,4 98,6% 2017 Yes 149,8 26,2 97,8% Nitrogen dioxide (NO 2) Hourly averages Hourly average nitrogen dioxide concentrations are shown in a cumulative percentage occurrence graphs in Figure 23, Figure 24, Figure 25 and Figure 26 together with a summary table of compliance with the NAAQS in Table 21. It can be seen from the graphs and the table that there is full compliance with the NAAQS with only Kriel in 2017 showing exceedances of the limit value. Where data recoveries are less than 80% it must be recognised that full compliance cannot be assured though. Note that the NO2 data from Phola has not been used as it is deemed completely unrepresentative of the air quality seen to prevail there. Figure 23: Cumulative percentage occurrence of hourly average NO 2 concentrations measured at the Komati ambient air quality monitoring station. Only values above the 90 th percentile are shown for clarity purposes. 35

38 Figure 24: Cumulative percentage occurrence of hourly average NO 2 concentrations measured at the Kriel Village ambient air quality monitoring station. Only values above the 90 th percentile are shown for clarity purposes. Figure 25: Cumulative percentage occurrence of hourly average NO 2 concentrations measured at the Elandsfontein ambient air quality monitoring station. Only values above the 90 th percentile are shown for clarity purposes. 36

39 Figure 26: Cumulative percentage occurrence of hourly average NO 2 concentrations measured at the Kendal ambient air quality monitoring station. Only values above the 90 th percentile are shown for clarity purposes. Table 21: Summary ambient hourly NO 2 average concentrations for the Komati, Kriel Village, Elandsfontein, and Kendal ambient air quality monitoring stations. All concentrations are in μg/m 3. Ambient air quality monitoring station No of exceedances Allowable Maximum Average % data recovery ,6 21,6 97,5% Komati ,9 23,8 74,9% ,0 20,3 71,4% ,1 19,1 86,5% Kriel ,0 18,6 79,9% ,4 24,2 69,3% ,5 15,3 37,3% Elandsfontein ,5 7,3 83,3% ,9 10,4 38,3% ,2 27,2 29,3% Kendal ,9 22,1 66,6% ,1 20,2 81,9% * Compliance with the NAAQS indicated by green, non-compliance by red. ** Light green data recovery is 80% or higher, light red less than 80%. Annual averages Annual average concentrations of NO2 are summarised in Table 22. It can be seen from the table that there is full compliance with the NAAQS for all stations, again recognising that compliance cannot be assured where there is less than 80% data recovery. 37

40 Table 22: Summary ambient annual NO 2 average concentrations for the Komati, Kriel Village, Elandsfontein and Kendal ambient air quality monitoring stations. All concentrations are in μg/m 3. Ambient air quality monitoring station Compliance Maximum Average % data recovery 2015 Yes 119,6 21,6 97,5% Komati 2016 Yes 114,9 23,8 74,9% 2017 Yes 106,0 20,3 71,4% 2015 Yes 121,1 19,1 86,5% Kriel 2016 Yes 148,0 18,6 79,9% 2017 Yes 263,4 24,2 69,3% 2015 Yes 111,5 15,3 37,3% Elandsfontein 2016 Yes 96,5 7,3 83,3% 2017 Yes 112,9 10,4 38,3% 2015 Yes 163,2 27,2 29,3% Kendal 2016 Yes 166,9 22,1 66,6% 2017 Yes 111,1 20,2 81,9% * Compliance with the NAAQS indicated by green, non-compliance by red. ** Light green data recovery is 80% or higher, light red less than 80% Particulate Matter (PM 10) Daily averages Daily average particulate matter (PM10) concentrations are shown in a cumulative percentage occurrence graphs in Figure 27, Figure 28, Figure 29, Figure 30 and Figure 31, together with a summary table of compliance with the NAAQS in Table 23. It can be seen from the graphs and the table that there is wholesale non-compliance with the NAAQS for all stations. For Phola there were 116, 137 and 117 days were seen to exceed the limit value (no more than 4 are allowed) in the three monitoring with similarly high loadings at Komati. The very high PM10 loading at the various stations creates a very high risk of adverse health effects amongst people exposed to the same. Again, even if compliance is implied by the data, compliance cannot be assured where data recovery is seen to be less than 80%compliance. Figure 27: Cumulative percentage occurrence of daily average PM 10 concentrations measured at the Komati ambient air quality monitoring station. 38

41 Figure 28: Cumulative percentage occurrence of daily average PM 10 concentrations measured at the Kriel Village ambient air quality monitoring station. Figure 29: Cumulative percentage occurrence of daily average PM 10 concentrations measured at the Elandsfontein ambient air quality monitoring station. 39

42 Figure 30: Cumulative percentage occurrence of daily average PM 10 concentrations measured at the Kendal ambient air quality monitoring station. Figure 31: Cumulative percentage occurrence of daily average PM 10 concentrations measured at the Phola ambient air quality monitoring station. 40

43 Table 23: Summary ambient 24-hour PM 10 average concentrations for the Komati, Kriel Village, Elandsfontein, Kendal and Phola ambient air quality monitoring stations. All concentrations are in μg/m 3. Ambient air quality monitoring station No of exceedances Allowable Maximum Average % data recovery ,0 63,9 96,4% Komati ,3 55,3 80,5% ,1 70,8 92,6% ,4 42,4 80,3% Kriel ,2 49,5 97,8% ,4 51,0 92,9% ,2 40,9 40,8% Elandsfontein ,0 30,4 87,4% ,0 30,4 91,2% ,3 66,7 36,2% Kendal ,5 58,7 83,0% ,6 61,5 88,5% ,2 92,1 41,4% Phola ,5 76,6 80,0% ,3 65,0 95,6% * Compliance with the NAAQS indicated by green, non-compliance by red. ** Light green data recovery is 80% or higher, light red less than 80%. Annual averages Annual average PM10 concentrations are summarised in Table 24. It can be seen from the table that there is non-compliance with the NAAQS for all years for all stations except for Elandsfontein in 2016 and The multiple non-compliances indicate dangerously high concentrations of PM10 and significant potential health risks for people exposed to these concentrations. Table 24: Summary ambient annual PM 10 average concentrations for the Komati, Kriel Village, Elandsfontein, Kendal and Phola ambient air quality monitoring stations. All concentrations are in μg/m 3. Ambient air quality monitoring station Compliance Maximum Average % data recovery 2015 No 179,0 63,9 96,4% Komati 2016 No 165,3 55,3 80,5% 2017 No 248,1 70,8 92,6% 2015 No 129,4 42,4 80,3% Kriel 2016 No 125,2 49,5 97,8% 2017 No 208,4 51,0 92,9% 2015 No 79,2 40,9 40,8% Elandsfontein 2016 Yes 112,0 30,4 87,4% 2017 Yes 133,0 30,4 91,2% 2015 No 169,3 66,7 36,2% Kendal 2016 No 94,5 58,7 83,0% 2017 No 96,6 61,5 88,5% 2015 No 311,2 92,1 41,4% Phola 2016 No 205,5 76,6 80,0% 2017 No 213,3 65,0 95,6% * Compliance with the NAAQS indicated by green, non-compliance by red. ** Light green data recovery is 80% or higher, light red less than 80%. 41

44 5.2.7 Particulate Matter (PM 2.5) Daily averages Daily average particulate matter (PM2.5) concentrations are shown in a cumulative percentage occurrence graphs in Figure 32, Figure 33, Figure 34, Figure 35 and Figure 36, together with a summary table of compliance with the NAAQS in Table 25. It can be seen from the graphs and the table that there is again wholesale non-compliance for all the monitoring stations where at Kendal for example the limit value was exceeded on no less than 201 days (where no more than 4 is allowed) in All the stations are seen to have similarly high numbers of exceedances of the limit value. The compliance indicated for Komati in 2015 is also in all probability also a non-compliance but is not assured because of the poor data record. Kendal monitoring station is especially noteworthy due to the contribution to PM2.5 of SO2 and it is converted in the atmosphere to sulphate. As has been shown earlier in this report (see 5.2.4) the Kendal monitoring station experiences large concentrations of SO2 by virtue of being directly downwind of the power station. Figure 32: Cumulative percentage occurrence of daily average PM 2.5 concentrations measured at the Komati ambient air quality monitoring station. Figure 33: Cumulative percentage occurrence of daily average PM 2.5 concentrations measured at the Kriel Village ambient air quality monitoring station. 42

45 Figure 34: Cumulative percentage occurrence of daily average PM 2.5 concentrations measured at the Elandsfontein ambient air quality monitoring station. Figure 35: Cumulative percentage occurrence of daily average PM 2.5 concentrations measured at the Kendal ambient air quality monitoring station. 43

46 Figure 36: Cumulative percentage occurrence of daily average PM 2.5 concentrations measured at the Phola ambient air quality monitoring station. Table 25: Summary ambient 24-hour PM 2.5 average concentrations for the Komati, Kriel Village, Elandsfontein, Kendal and Phola ambient air quality monitoring stations. All concentrations are in μg/m 3. Ambient air quality monitoring station No of exceedances Allowable Maximum Average % data recovery ,9 14,9 13,7% Komati ,1 19,6 87,9% ,2 32,5 87,1% ,2 46,1 7,9% Kriel ,7 27,9 94,5% ,3 25,0 86,0% ,9 26,6 62,7% Elandsfontein ,2 23,8 87,4% ,3 21,2 90,4% ,0 48,3 40,8% Kendal ,8 61,4 83,6% ,5 22,9 68,8% ,2 92,1 48,5% Phola ,8 50,9 98,6% ,1 41,1 83,3% * Compliance with the NAAQS indicated by green, non-compliance by red. ** Light green data recovery is 80% or higher, light red less than 80%. Annual averages Annual average PM2.5 concentrations are summarised in Table 26. It can be seen from the table that there is non-compliance with the NAAQS for all years for all the monitoring stations, indicating sustained elevated concentrations of PM2.5 in these areas. For Komati in 2015 noncompliance is also likely and for 2016 the average is 19,6 μg/m 3 compared to the limit value of 20 μg/m 3 which is to all intents and purposes a non-compliance. The concentrations imply very high sustained concentrations of PM2.5 with very high risk of adverse health effects. 44

47 Table 26: Summary ambient annual PM 2.5 average concentrations for the Komati, Kriel Village, Elandsfontein, Kendal and Phola ambient air quality monitoring stations. All concentrations are in μg/m 3. Ambient air quality monitoring station Compliance Maximum Average % data recovery 2015 Yes 25,9 14,9 13,7% Komati 2016 Yes 100,1 19,6 87,9% 2017 No 104,2 32,5 87,1% 2015 No 92,2 46,1 7,9% Kriel 2016 No 119,7 27,9 94,5% 2017 No 63,3 25,0 86,0% 2015 No 61,9 26,6 62,7% Elandsfontein 2016 No 63,2 23,8 87,4% 2017 No 80,3 21,2 90,4% 2015 No 100,0 48,3 40,8% Kendal 2016 No 225,8 61,4 83,6% 2017 No 119,5 22,9 68,8% 2015 No 311,2 92,1 48,5% Phola 2016 No 204,8 50,9 98,6% 2017 No 170,1 41,1 83,3% * Compliance with the NAAQS indicated by green, non-compliance by red. ** Light green data recovery is 80% or higher, light red less than 80% Source apportionment Source apportionment is notoriously difficult, but essential to the analysis presented here. Perhaps the most instructive way of considering source apportionment on the basis of the ambient air quality data, without conducting physical pollutant speciation studies, is through presenting the diurnal variation in pollutant concentrations. This is because the creation of air pollution follows trends in space and time that make it possible to distinguish, by means of insinuation/ implication, between air pollution sources. For example, air pollution stemming from low-level burning practices associated with low-income community activities for heating and cooking, tends to arise in the early mornings and the early evenings, and so it is to be expected that measured peaks in pollutant concentrations during these times could be attributed to sources at a community level. Conversely, it can be assumed that power station emissions are most likely to reach the ground during the middle of the day when the atmosphere is unstable due to increased mixing activities, and so pollution peaks in the daytime can be attributed to industrial sources. As such, diurnal variability in pollutant concentrations is illustrated in Figure 37 - Figure 41, as area plots of average hourly concentrations for each of the monitoring stations for each of the years ( ). For SO2 (Figure 37) a clear midday peak is evident of some 120 ug/m 3 with generally lower concentrations of at or below 50 ug/m 3 from 17:00 through to 06:00, whereafter concentrations are seen to increase again to the midday peaks. For NO2, two peaks are evident, the first at 06:00 and the second at 18:00-19:00 (Figure 38). The lowest concentrations occur between 10:00 and 11:00. A similar pattern is seen for PM10 and PM2.5 where, for both pollutants, a morning peak is evident at between 06:00 and 07:00 and an evening peak at 18:00 (Figure 39and Figure 40). When overlaid, the patterning of the pollutant peaks suggests that the primary sources of SO2 are different from the primary sources of NO2, PM10, and PM2.5 (Figure 41). What is postulated is that the SO2 peak sources are high elevation emissions while the NO2, PM10 and PM2.5 peaks are primarily sourced at ground level. 45

48 250 Concentra on in μg/m Hours of the day Figure 37: Area plot of the range of average diurnal SO 2 concentrations for all monitoring stations across the Mpumulanga Highveld ( ). 250 Concentra on in μg/m Hours of the day Figure 38: Area plot of the range of average diurnal NO 2 concentrations for all monitoring stations across the Mpumulanga Highveld ( ). 46

49 250 Concentra on in μg/m Hours of the day Figure 39: Area plot of the range of average diurnal PM 10 concentrations for all monitoring stations across the Mpumulanga Highveld ( ). 250 Concentra on in μg/m Hours of the day Figure 40: Area plot of the range of average diurnal PM 2.5 concentrations for all monitoring stations across the Mpumulanga Highveld ( ). 47

50 250 PM 10 PM 10 Concentra on in μg/m PM 2.5 NO 2 SO 2 PM 2.5 NO Hours of the day Figure 41: Area plot of the range of average diurnal SO 2, NO 2, PM 10 and PM 2.5 concentrations for all monitoring stations across the Mpumulanga Highveld ( ). The use of domestic fuels for cooking and space heating is a well-known phenomenon in South Africa, most notable in low-income dense settlements, even where electricity may be available. These sources result in emissions of SO2, NO2 and PM10 at ground level, particularly so during the early hours of the morning and the late hours of the afternoon/ early hours of the evening when the need for cooking and space heating peaks. The diurnal patterns described above can be explained as follows: During the night the atmosphere becomes stable with inversions often occurring. When the atmosphere is stable, emissions from elevated sources (e.g. stacks) do not come to ground-level as they are released into a stable atmosphere and simply cannot penetrate down towards the ground. Emissions that occur at ground-level, such as domestic fuel burning and motor vehicle emissions are similarly trapped closer to the ground by the stable atmosphere and cannot disperse. When the sun rises the heating of the earth s surface sees the start of turbulence and mixing in the atmosphere and the dissolution of the surface inversion. The mixing gets deeper and deeper as the day progresses until at some point in the day the elevated source s plume is brought to ground-level. As the elevated source s plume comes to ground, there is a significant increase in the ambient SO2 concentration. As the afternoon wears on, the earth s surface cools and the atmosphere becomes more stable with reduced atmospheric mixing. The stable atmosphere results in the ambient SO2 concentration reducing significantly as, once again, the elevated source s plume is prevented from reaching the ground. The ambient SO2 concentration, as well as the time of day during which peak SO2 concentrations are measured, therefore provide a powerful indicator of the contribution of the elevated sources to ambient air quality. The morning and night-time NO2, PM10 and PM2.5 peaks would then derive from ground level sources, whereas the SO2 peak would imply elevated sources. The diurnal patterning described above is well documented in Venter et al, (2012) and seen to be exhibited in the North West Province too. 48

51 5.3 Dispersion modelling The approach to the dispersion modelling in this assessment is based on the requirements of the DEA guideline for dispersion modelling (DEA, 2014). An overview of the approach for Kriel Power Station is provided here Models used A number of models with different features are available for air dispersion studies. The selection of the most appropriate model for an air quality assessment needs to consider the complexity of the problem and factors such as the nature of the development and its sources, the physical and chemical characteristics of the emitted pollutants and the location of the sources. This assessment is a level 3 assessment, according to the definition on the dispersion modelling guideline (DEA, 2014). The CALPUFF suite of models ( were therefore used. The U.S. EPA Guideline of Air Quality Models also provides for the use of CALPUFF on a case-by-case basis for air quality estimates involving complex meteorological flow conditions, where steady-state straight-line transport assumptions are inappropriate. CALPUFF is a multi-layer, multi-species non-steady-state puff dispersion model that simulates the effects of time- and space-varying meteorological conditions on pollution transport, transformation and removal. CALPUFF can be applied on scales of tens to hundreds of kilometres. It includes algorithms for sub-grid scale effects (such as terrain impingement), as well as, longer range effects (such as pollutant removal due to wet scavenging and dry deposition, chemical transformation, and visibility effects of particulate matter concentrations). The Air Pollution Model (TAPM) (Hurley, 2000; Hurley et al., 2001; Hurley et al., 2002) is used to model surface and upper air metrological data for the study domain. TAPM uses global gridded synoptic-scale meteorological data with observed surface data to simulate surface and upper air meteorology at given locations in the domain, taking the underlying topography and land cover into account. The global gridded data sets that are used are developed from surface and upper air data that are submitted routinely by all meteorological observing stations to the Global Telecommunication System of the World Meteorological Organisation. TAPM has been used successfully in Australia where it was developed (Hurley, 2000; Hurley et al., 2001; Hurley et al., 2002). It is considered an ideal tool for modelling applications where meteorological data does not adequately meet requirements for dispersion modelling. TAPM modelled output data is therefore used as input to the CALMET processor to generate three years (2015, 2016 and 2017) of hourly gridded surface meteorological data for upper air data for the modelling domain input to CALPUFF Model parameterisation TAPM In the southern Mpumalanga Highveld TAPM is set-up in a nested configuration of three domains. The outer TAPM domain is 600 km by 600 km at a 24 km grid resolution, the middle domain is 300 km by 300 km at a 12 km grid resolution and the inner domain is 75 km by 75 km at a 3 km grid resolution (Figure 42). The nesting configuration ensures that topographical effects on meteorology are captured and that meteorology is well resolved and characterised across the boundaries of the inner domain. Twenty-seven vertical levels are 49

52 modelled by TAPM in each nest from 10 m to 5000 m, with a finer resolution in the lowest m. Figure 42: TAPM and CALPUFF modelling domains for Kriel The 3-dimensional TAPM meteorological output on the inner grid include hourly wind speed and direction, temperature, relative humidity, total solar radiation, net radiation, sensible heat flux, evaporative heat flux, convective velocity scale, precipitation, mixing height, friction velocity and Obukhov length. Obukhov length is a proxy for turbulence. The spatially and temporally resolved TAPM surface and upper air meteorological data is used as input to CALPUFF s meteorological pre-processor, CALMET. CALPUFF The CALPUFF grid, which is km 2 is 66 km (west-east) by 66 km (north-south), is a subdomain of the TAPM inner grid and is centred on Kriel (Figure 42). It consists of a uniformly spaced receptor grid with 500 m spacing, giving grid cells (132 X 132 grid cells). The CALPUFF modelling domain is the same as the CALMET modelling domain. The predicted ambient concentrations are presented for a 60 km by 60 km domain, centred on the power station. In deciding in the modelling domain it is always necessary to balance the number of receptors chosen with the spatial area covered by the modelling to ensure that there is adequate resolution with overtaxing the modelling runs. It is generally recognised that the major air quality impacts from power stations are typically about 10 kms from the sources so it is also necessary to balance the near and far field predictions. The 60 km by 60 km modelling domain seeks to achieve that balance. 50

53 The land use data is based on the Global Land Cover Characterisation (GLCC) Version 2 dataset, which has a horizontal grid spacing of 30 arc seconds (~1 km resolution). The digital terrain data is based on the Shuttle Radar Topography Mission (SRTM) 3 Global Coverage Version 2 dataset. It was collected during the Shuttle Radar Topography Mission and has a grid spacing of 3 arc-second (~90 m resolution). Stack emission measurements for total particulates (PM) are available. However, there are no measurements for PM10 and PM2.5 emissions. For the dispersion modelling a conservative approach is adopted and it is assumed that all particulates released are firstly PM10 and secondly PM2.5. NOX emissions are modelled to predict NO2 concentrations, which are compared with the 1- hour and annual NAAQS. Since not all NO converts to NO2, this approach is conservative and should be recognised when comparison is made against the NAAQS. In addition, a default NO2 conversion factor of 0.8 is applied (DEA, 2014). Secondary particulates are formed through chemical reactions in the atmosphere of primary (emitted) pollutants of SO2 and NOx. CALPUFF uses a stoichiometric thermodynamic model to estimate the partitioning of total inorganic nitrate between gas phase nitric acid and particle phase ammonium nitrate, using measured ambient background ammonia and ozone concentrations. The secondary particulates include ammonium nitrate and ammonium sulphate. Ambient concentrations of secondary particulates may be assessed against the NAAQS for PM2.5 as these are typically fine particulates. The parameterisation of key variables that are applied in CALMET and CALPUFF are indicated in Table 27 and Table 28. Table 27: Parameterisation of key variables for CALMET Parameter Model value 12 vertical cell face heights (m) 0, 20, 40, 80, 160, 320, 640, 1000, 1500, 2000, 2500, 3000, 4000 Coriolis parameter (per second) Empirical constants for mixing height equation Neutral, mechanical: 1.41 Convective: 0.15 Stable: 2400 Overwater, mechanical: 0.12 Minimum potential temperature lapse rate (K/m) Depth of layer above convective mixing height 200 through which lapse rate is computed (m) Wind field model Diagnostic wind module Surface wind extrapolation Similarity theory Restrictions on extrapolation of surface data No extrapolation as modelled upper air data field is applied Radius of influence of terrain features (km) 5 Radius of influence of surface stations (km) Not used as continuous surface data field is applied 51

54 Table 28: Parameterisation of key variables for CALPUFF Parameter Model value Chemical transformation Default NO2 conversion factor of 0.8 is applied (DEA, 2014). Wind speed profile Rural Calm conditions Wind speed < 0.5 m/s Plume rise Transitional plume rise, stack tip downwash, and partial plume penetration is modelled Dispersion CALPUFF used in PUFF mode Dispersion option Dispersion coefficients use turbulence computed from micrometeorology Terrain adjustment method Partial plume path adjustment Model accuracy Air quality models attempt to predict ambient concentrations based on known or measured parameters, such as wind speed, temperature profiles, solar radiation and emissions. There are however, variations in the parameters that are not measured, the so-called unknown parameters as well as unresolved details of atmospheric turbulent flow. Variations in these unknown parameters can result in deviations of the predicted concentrations of the same event, even though the known parameters are fixed. There are also reducible uncertainties that result from inaccuracies in the model, errors in input values and errors in the measured concentrations. These might include poor quality or unrepresentative meteorological, geophysical and source emission data, errors in the measured concentrations that are used to compare with model predictions and inadequate model physics and formulation used to predict the concentrations. Reducible uncertainties can be controlled or minimised. This is achieved by making use of the most appropriate input data, preparing the input files correctly, checking and re-checking for errors, correcting for odd model behaviour, ensuring that the errors in the measured data are minimised and applying appropriate model physics. Models recommended in the DEA dispersion modelling guideline (DEA, 2014) have been evaluated using a range of modelling test kits ( It is therefore not mandatory to perform any modelling evaluations. Rather the accuracy of the modelling in this assessment is enhanced by every effort to minimise the reducible uncertainties in input data and model parameterisation. For Kriel the reducible uncertainty in CALMET and CALPUFF is minimised by: Using representative quality controlled observed hourly meteorological data to nudge the meteorological processor to the actual values; Using 3-years of spatially and temporally continuous surface and upper air meteorological data field for the modelling domain; Using actual monthly average emissions for Scenario 1; Appropriate parameterisation of both models (Table 27 and Table 28); Using representative emission data; Using a competent modelling team Comparison between measured and modelled values Earlier in this report mention was made of model accuracy and reducible error. That does not change the fact that there remains an obvious question as to how well the model predicts the concentrations that are measured at the various monitoring stations. Comparing measured and 52

55 modelled concentrations is not straight forward either because the measured concentrations reflect all sources of pollution whereas the model can obviously only predict the ambient concentrations that occur as a function of the emissions included in the model. Past experience has shown that in general terms most of the SO2 derives from the power stations whereas NO2 and especially PM derive from multiple other sources, notwithstanding the contribution of the power stations to secondary aerosol formation. In this section the issue of how well the model outputs replicate what is measured, is presented. In the first instance, only SO2 concentrations are compared as these are deemed to be the only pollutant where the modelled and measured concentrations can be expected to approximate one another. Also, only hourly average concentrations are compared as the findings of the comparison can be extended to the longer averaging periods. Earlier in the report the monitored data were presented as probability distributions (viz. the probability of a given concentration as a function of the measured data). It is not helpful to simply compare a single modelled value with a single measured value because that presents only one part of a more complex relationship. As such the approach that has been used here is to compare the data distributions of the measured values with the data distributions of the modelled values. This comparison can be done statistically using for example the Kolmogorov-Smirnov statistic but there is a specific patterning in the comparisons that needs to be illustrated here namely that in general terms the model does not predict the multiple smaller concentrations that are evident in the measured data. What has been done here therefore, is to compare the two data sets by comparing the respective 10 th, 25 th, 50 th, 75 th and 99 th percentile concentrations in the two data sets (viz. the measured and the modelled) as illustrated conceptually in Figure 43. Figure 43: Conceptual illustration of the method used to compare modelled and measured ambient hourly SO 2 concentrations. Modelled concentrations at the monitoring stations have been compared to measured concentrations for each of the 10 th, 25 th, 50 th, 75 th and 99 th percentile data items. Such comparisons are shown for Kriel Village in Figure 44, Kendal in Figure 45, Phola in Figure 46, Komati in Figure 47 and Elandsfontein in Figure 48. The modelled concentrations derive from 53

56 the combined emissions sources (viz. all coal fired power stations in the domain). It can be seen from the figures that the modelled concentrations are commonly less than the measured concentrations in all but one circumstance (Kriel Village 2016). In general terms it is the smaller concentrations that are not predicted by the model as seen for the 10 th, 25 th and 50 th percentiles where SO2 concentrations of 0 or very close to 0 were predicted in almost all circumstances. This suggests that there is a sustained background concentration of SO2, which may derive from varied other sources including discard coal combustion, industrial processes and even recirculation of SO2 from the power stations. For the higher percentiles (75 th and 99 th ) the modelled concentrations better predict the measured concentrations with Kriel Village seeing a 30% variance (42.1 to 96.4 μg/m 3 ) for the 99 th percentile. For the other stations though the differences are somewhat larger. The average difference in concentrations for Kendal is 299 μg/m 3, Phola is 26.6 μg/m 3, Komati is μg/m 3, and Elandsfontein is 54,5 μg/m 3. Annual average differences are seen to be 15.4, 24.4, 16.6, 31.6 and 21.1 μg/m 3 for the Kriel Village, Kendal, Phola, Komati and Elandsfontein ambient air quality monitoring stations respectively. There will always be an inherent degree of error in the modelling predictions, which plays some role in the differences seen. At the same time other SO2 emission sources (as described above) will play a role in the differences seen between modelled and measured, the quantum of which will vary from place to place across the modelling domain. It is simply not possible to determine the relative role of each source of error but for the purposes of this report it is assumed that the dispersion modelling itself is acceptably accurate and that it is the presence of SO2 from other sources not included in the dispersion modelling, that is the dominant source of error. On the assumption that the model does accurately predict ambient concentrations this analysis suggests that there is a contribution to the measured ambient air quality from sources not included in the model of up to μg/m 3 for hourly averages and up to 31.6 μg/m 3 for annual average concentrations, with the daily average concentrations somewhere between the two. Figure 44: Comparison between modelled and measured hourly SO 2 concentrations at the Kriel Village ambient air quality monitoring station. An exact correlation would mean 100%. 54

57 Figure 45: Comparison between modelled and measured hourly SO 2 concentrations at the Kendal ambient air quality monitoring station. An exact correlation would mean 100%. Figure 46: Comparison between modelled and measured hourly SO 2 concentrations at the Phola ambient air quality monitoring station. An exact correlation would mean 100%. 55

58 Figure 47: Comparison between modelled and measured hourly SO 2 concentrations at the Komati ambient air quality monitoring station. An exact correlation would mean 100%. Figure 48: Comparison between modelled and measured hourly SO 2 concentrations at the Elandsfontein ambient air quality monitoring station. An exact correlation would mean 100%. 5.4 Modelled ambient concentrations Two emissions scenarios have been modelled namely: Actual emissions as a function of the emissions that have occurred from the power station over the last three years ( ); and, Emissions at MES compliance but at full load for the entire three-year modelling period. 56