Quantifying particulate emissions from domestic burning in the kwaDela township, Mpumalanga

Quantifying particulate emissions

from domestic burning in the

kwaDela township, Mpumalanga

NC Nkosi

Orcid.org 0000-0002-0123-1498

Dissertation submitted in fulfilment of the requirements

for the degree

Master of Science in Environmental

Sciences

at the North-West University

Supervisor:

Prof SJ Piketh

Co-supervisor:

Dr RP Burger

Graduation ceremony: July 2018

ABSTRACT

Air quality in South African low-income residential areas is poor owing to high gaseous and particulate emissions from a range of sources including domestic burning. These emissions degrade ambient air quality, result in indoor air pollution which has a negative impact on human health. Solid-fuel use and emissions in a household setting are governed by several factors that vary in space and time. This raised a need to collect local reliable solid-fuel use and emissions data. The output is used to understand the relationship between burning behaviour, emissions patterns and ambient air quality. Characterising solid-fuel use and quantifying emissions from domestic burning is the first step to improving air quality in low-income residential areas. Emission factors from domestic burning are sensitive to fuel characteristics, stove-operation behaviour and conditions, therefore, cannot be generalised. The absence of data on fuel use and variability in low-income residential areas introduces uncertainties when quantifying estimates of total emissions from the township. Unreported high uncertainties in emission estimates lead to the development of misinformed emission inventories.

The study aims to quantify emissions of fine particulate matter from domestic burning of coal using field measurements. The first objective of the study is to characterise solid-fuel use and burning-device operation behaviour. The second objective quantifies fine particulate emission factors from domestic burning and characterises the operational behaviour effect of stoves on emission profiles. The last objective estimates the annual, seasonal and daily fine particulate matter emissions from domestic burning in a low-income residential area. Solid-fuel use responses from a survey undertaken in kwaDela (2014) and observation results (2016) were used to identify the dominant solid-fuel and burning-device used; the major household solid-fuel use determinant and describe burning-device operation behaviour. Stove-use monitors (K-type I-Button) data (for 2013 winter and 2014 summer) were used to characterise seasonal burning patterns in the township. To quantify fine particulate PM 2.5 emission factors (g.kg-1_{), isokinetic (2015) and direct (2014) field}

tests were set up to monitor and sample gaseous and fine particulate concentrations during a burning event. A Monte Carlo simulation model was used to compute the residential area’s probable seasonal fine particulate emission estimates from domestic burning. Coal is the predominantly used fuel. Traditional cast-iron coal stoves are the most common solid-fuel burning devices used. Solid-fuel use, stove-operation behaviour and burning patterns vary within the township, however, the variation is insignificant. The majority of the residents have two burning events per winter day and a single event per summer day.

Factors determining the choice of solid-fuel use are: (i) fuel availability (ii) demographics, and (iii) change of seasons.

PM2.5 emission factors ranged from 6.8 g.kg-1 _{to 13.5 g.kg}-1 _{of fuel burned. Fine particulate}

emission patterns change with burning patterns; high emissions are experienced during ignition and refuelling of the burning event. Fine particulate matter emitted from domestic burning per winter day in kwaDela Township ranges between 33.7 kg to 70.1 kg. Daily summer emissions range between 16.6 kg and 35.5 kg

DECLARATION

I, Miss NC Nkosi, declare that the dissertation

“Quantifying particulate matter emissions from domestic burning in

the kwaDela township, Mpumalanga”

is my own work and information acquired from other sources has been correctly referenced. It is being submitted in partial fulfilment of the requirements for the Degree of Magister

Scientiae in Environmental Sciences in the School of Geography and Spatial Sciences at

the Potchefstroom Campus of the North-West University. This dissertation has not been submitted for any degree in the North-West University or in any other institution.

Miss N C Nkosi

This dissertation is dedicated to my family

(Sanny Teigic, Thulane Nicholus, Bongiwe Giveness, Cindy Carol Nkosi), and my son,

CONTENTS

ABSTRACT

I

DECLARATION

IV

CONTENTS

VI

PREFACE: FORMAT OF THE DISSERTATION

XI

ACKNOWLEDGEMENTS

XIV

LIST OF FIGURES

XV

LIST OF TABLES

XVII

ABBREVIATIONS, ACRONYMS AND GLOSSARY

XVIII

CHAPTER 1: INTRODUCTION

1

1.1 INTRODUCTION AND PROBLEM STATEMENT 1

1.2 AIM AND OBJECTIVES 4

1.3 STUDY JUSTIFICATION 4

1.3.1 The reason for focusing on low-income residential areas 4

1.3.2 The selection of kwaDela Township 5

1.3.3 The focus on fine particulate matter 5

1.3.4 The reason for quantifying emissions 6

1.3.5 The use of field measurements 6

CHAPTER 2: LITERATURE REVIEW

8

2.1 FUEL USAGE IN DEVELOPING COUNTRIES LOW-INCOME

RESIDENTIAL AREAS 8

2.2 HOUSEHOLD ENERGY CHOICE DETERMINANTS 10

2.2.1 Fuel type availability and accessibility 11

2.2.2 The effect of climate on fuel use 11

2.2.4 Monetary, time and labour costs associated with fuel 13 2.2.5 Household income allocated for energy use (affordability) 13

2.2.6 Cultural practice and tradition influence on fuel use 14

2.3 DOMESTIC SOLID-FUEL-USE MODELS (ENERGY LADDER AND FUEL

STACKING) 15

2.4 CHARACTERISING PM EMISSIONS FROM DOMESTIC BURNING OF

SOLID FUELS 18

2.5 THE RELATIONSHIP BETWEEN EMISSIONS, FUEL AND STOVE

CHARACTERISTICS 22

2.5.1 Fuel density 22

2.5.2 Fuel particle size 22

2.5.3 Fuel moisture content 23

2.5.4 Fuel chemical composition 24

2.5.5 Impact of stove characteristics on emissions 25

2.5.6 Stove-operation behaviour 26

2.6 IMPACT OF FINE PARTICULATE MATTER EMISSIONS ON HUMAN

HEALTH, VISIBILITY AND CLIMATE 27

2.6.1 Effect of particulate matter on human health 28

2.6.2 Effect of particulate matter on visibility 30

2.6.3 Effect of particulate matter on global climate system 30

2.7 MANAGING PM EMISSIONS IN RESIDENTIAL AREAS 31 2.8 THE NEED FOR EMPIRICAL QUANTIFICATION OF EMISSIONS 35

2.9 THE SCIENTIFIC BASIS FOR THIS STUDY 37

CHAPTER 3: METHODOLOGY AND DATA COLLECTION

39

3.1 STUDY AREA DESCRIPTION 39

3.2 SOLID-FUEL-USE SURVEY STUDY 41

3.2.1 Solid-fuel-use data collection using survey questionnaires 41

3.2.2 Survey quality control procedure 42

3.2.3 Limitations of using a solid-fuel-use questionnaire study 42

3.3 DETERMINING CHARACTERISTICS OF BURNING EVENTS AND SEASONAL BURNING PATTERNS USING TEMPERATURE SENSOR

3.3.1 Seasonal stove-use data collection 42

3.3.2 Burn event count algorithm 43

3.4 SOLID-FUEL-USE OBSERVATIONAL STUDY 44

3.4.1 Observational study quality control 46

3.4.2 Observational study limitations 46

3.5 QUANTIFYING FINE PM EMISSION FACTORS FROM BURNING

SOLID FUELS 47

3.5.1 Particulate matter and gas concentration sampling train main

instrumentation 47

3.5.2 Sampling quality control procedures 52

3.5.3 Limitations of field gaseous and particulate sampling method 54

3.6 ESTIMATING FINE PARTICULATE MATTER EMISSIONS FROM

DOMESTIC BURNING USING A MONTE CARLO SIMULATION 55

3.6.1 Monte Carlo simulation steps for calculating emission estimates 56

3.7 LADDER STEPS TO IMPROVING AIR QUALITY 59

CHAPTER 4: VARIABILITY OF DOMESTIC BURNING HABITS IN

THE SOUTH AFRICAN HIGHVELD: A CASE STUDY IN KWADELA

TOWNSHIP

61

4.1 INTRODUCTION 61

4.2 STUDY AREA, DATA COLLECTION AND METHODOLOGY 64

4.2.1 Description of study area 64

4.2.2 Methodology and data collection 65

4.2.3 Fuel-use survey 65

4.2.4 “First-person” winter observations 66

4.3 RESULTS 67

4.3.1 Types of fuel used in kwaDela Township 67

4.3.2 Sources of solid fuel and distribution in kwaDela Township 68 4.3.3 Relationship between household demographics and solid-fuel use 74 4.3.4 Stove-operation behavioursolid fuel use spatial variability in kwaDela 75

4.3.5 Burning patterns of solid fuel 76

4.3.6 Variability of fuel use in South African townships 77

4.6 CHAPTER SUMMARY 83

CHAPTER 5: STOVE-OPERATION BEHAVIOUR EFFECT ON FINE PM

ON THE EMISSIONS FROM DOMESTIC BURNING IN LOW-INCOME

RESIDENTIAL AREAS: A CASE STUDY IN KWADELA TOWNSHIP

85

5.1 INTRODUCTION 85

5.2 MATERIAL AND METHODS 88

5.2.1 Chimney sampling tests and fine PM emission calculations 88

5.3 RESULTS 92

5.3.1 Direct and isokinetic direct chimney sampling results 92

5.4 DISCUSSION 94

5.5 CHAPTER SUMMARY 97

CHAPTER 6: PROBABILISTIC ESTIMATES OF RESIDENTIAL

SOLID-FUEL-BURNING EMISSIONS IN LOW-INCOME

SETTLEMENTS ON THE SOUTH AFRICAN HIGHVELD

98

6.1

INTRODUCTION

98

6.2

METHODOLOGY

100

6.5 LIMITATIONS OF THE STUDY 107

6.6 CHAPTER SUMMARY 108

CHAPTER 7: SUMMARY, CONCLUSION AND RECOMMENDATIONS

109

7.1 SUMMARY AND KEY FINDINGS 109

7.2 STUDY LIMITATIONS AND RESULTS APPLICATION 111

7.4 CONTRIBUTION OF THE STUDY TOWARDS BROADER KNOWLEDGE 112

7.5 STUDY CONCLUSION AND IMPLICATIONS 113

REFERENCES

116

PREFACE: FORMAT OF THE DISSERTATION

This dissertation is written in an article format and has seven chapters. Chapter 4 and Chapter 5 have been presented at national and international conferences.

STRUCTURE OF THE DOCUMENT Chapter 1: Introduction

The first chapter gives a brief background on domestic burning emissions impact on human health and ambient air quality. It also emphasises the importance of documenting solid-fuel data and estimating emission in air quality management. The chapter further describes the overall aim, objectives, justification and design of the study.

Chapter 2: Literature Review

This chapter outlines a detailed literature review on solid-fuel use at a global, national and local scale. Energy use models are provided on a household scale. Emission of particulate matter, its impact on human health and the environment are also discussed in detail. The last section of the chapter gives insight on air quality legislation and discusses the use of solid fuel and emissions intervention methods implemented to improve air quality in residential areas.

Chapter 3: Methodology and Data Collection

Describes in detail the study area, methods of data collection and analyses with their associated limitations. The chapter describes the function, specification and limitation of the instruments used when sampling particulate matter.

Chapter 4: First Article “Variability of solid-fuel use in the South African Highveld: A

case study in kwaDela Township”.

The chapter comprises the first article that answers Objective 1 and Objective 2 of the study. The article characterises solid-fuel used, stove-operation behaviour and seasonal variability of burning patterns.

Chapter 5: Second Article “Stove-operation behaviour effect on fine PM the emissions

The chapter comprises the second article that answers Objective 3 of the study. The article quantifies emission of fine particulate matter using field measurements and characterises the impact of stove-operation behaviour on the emissions. Moreover, this chapter gives detail of the instruments used, the experimental sampling method used, the specification and limitation of the instrument.

Chapter 6: Third Article “Probabilistic estimates of residential solid-fuel burning

emissions in low-income settlements on the South African Highveld”.

The chapter presents results for Objective 3, which quantify emissions of fine particulate matter for kwaDela Township and identify major uncertainty sources for emission estimates. It also highlights the importance of reporting emission estimates with their associated uncertainties.

Chapter 7: Conclusions, Limitations and Recommendations

The last chapter gives a summary of the conclusions drawn from the results. Additionally, the chapter gives an overview of the limitations of the study, recommendations for future research and proposed solutions to solid-fuel use in low-income residential areas.

Conferences where the conference proceedings were presented or submitted:

- Nkosi, N.C., Piketh, S.J., Burger, R.P., 2016. Quantifying emissions of fine particulate matter from domestic burning of solid fuels in South African townships: A case study in kwaDela Township. In: National Laboratory Association. Centurion, South Africa, 26-27 September 2016.

- Nkosi, N.C., Burger, R.P., Piketh, S.J., 2016. Quantifying fine particulate emissions from domestic burning in low-income settlements: A case study in kwaDela Township. In: National Association for Clean Air. ISBN: 978-0-620-70646-9, Nelspruit, South Africa, 05-07 October 2016.

- Nkosi, N.C., Piketh, S.J., Burger, R.P., Annegarn, H.J., 2017. Variability of domestic burning habits in the South African Highveld: A case study in the kwaDela Township. In: International Domestic Use of Energy. ISBN: 978-1-5090-6433-5. Cape Town, South Africa, 03-05 April 2017.

Ethical Considerations

The study procedure was reviewed and approved by the North West University Ethics Committee for Research on Human Subjects on 12 April 2011, under ethical clearance number NWU-00066-13-A3 (SASOL). The researcher completed an online ethics course (Introduction to Research Ethics, Research Ethics Evaluation, and Informed Consent) registered with the Clinical Trial Centre at the University of Hong Kong.

Prior to observations, the researcher and the research assistants spent a day with the participants to explain in detail the observation process, give assurance of anonymity and request the participants’ verbal consent to take pictures during observations. The researcher explained to the participants that the observation study was voluntary and that they could withdraw at any stage.

ACKNOWLEDGEMENTS

 I would like to thank God for keeping his promise, being a good Father and making all this possible. I would also like to thank my family for their continuing support.

 I am eternally grateful to Prof. Stuart Piketh and Dr Roelof Burger for their guidance, advice and financial support. “Thank you” for giving me this opportunity and for believing in my abilities. No words can describe the gratitude I wish to convey for their continual support and the endless opportunities to which I was exposed during the period of obtaining my Master’s. The lessons I have learned under your supervision have inspired me to grow and I apply them in my everyday life.

 I am exceedingly grateful to Prof. Annegarn and Mr Christopher Pauwl for their guidance, assistance and contribution towards the project.

 Special thanks to “Grow Our Own Timber NWU Potchefstroom” bursary for their financial support for both years of my Master’s study. I would also like to thank SASOL for funding the project.

 I am grateful to the North West University for the opportunity to do my Master’s at Potchefstroom campus.

 I am indebted to Baba Joe Mahlalela, Reighan Du preez, Brigitte Language, Corné

Grové, Madoda Mabuza, Boitumelo Ashley Thlapi, and Lehlohonolo Sello for assisting with data collection.

 I would like to thank the Climate Research Group at North West University for their guidance and continual advice through the project.

 To my friends, “thank you” for your support; for believing in me when my motivation was depleted; for reading my work, and supporting me when presenting my articles. I can honestly say I enjoyed this journey and it was not as difficult as I thought.

 To Jeannette Menasce for proofreading, editing and formatting this document… my

grateful thanks.

 Lastly, I am grateful to the community of kwaDela Township for their pleasant cooperation during data collection and allowing us to invade your homes, we are very grateful.

“Everybody pays for electricity, either with Rands or a couple of years of their life span”.

LIST OF FIGURES

Figure 2.1: Percentage of solid-fuel use for cooking in African countries (WHO,

2010) ... 8 Figure 2.2: Percentage usage of different solid fuels in South Africa obtained from

van den Berg, 2015. ... 9 Figure 2.3: Comparison of the energy ladder and energy stacking model

(van der Kroon et al., 2013) ... 16 Figure 2.4: Emissions of particulate matter from different household burning devices

(Graham & Dutkiewicz, 1993) ... 19 Figure 2.5: Morphology of different particle sizes (Ninomiya et al., 2004) ... 20 Figure 2.6: Particulate matter emission profile per energy produced (mg.MJ-1_{) and}

burning rate (kg.hour-1_{) during a burning event (Mitchell et al., 2016) ... 21}

Figure 2.7: Particle deposition in the human body according to its size diameter

(Londahl et al., 2006) ... 28 Figure 2.8: Human health issues associated with exposure to prolonged high

particulate concentrations (Kim et al., 2015) ... 29 Figure 3.1: A map showing the location of the kwaDela residential area in

Mpumalanga Province ... 39 Figure 3.2: (A): Typical material used for building, (B): Housing insulation material

used, and (C): Domestic waste burning ... 41 Figure 3.3: Stove and kitchen temperature monitors (K-type Ibutton) ... 45 Figure 3.4: DustTrakTM _{II Aerosol Monitor (A TSI DustTrak II Aerosol Monitor 8530) ... 47}

Figure 3.5: Teflon (43 mm diameter) and quartz (47 mm diameter) filters (Source: Photo taken by author at NWU Climatology Research Group Laboratory, 2017) ... 48 Figure 3.6: Horiba gas analysers (PG-350, PG-250) (Source: Photo taken by author

at Gondwana, 2016) ... 49 Figure 3.7: S-type Pitot tube (ISO 3966, 1977) ... 51 Figure 3.8: High-volume air sampler (Source: Photo taken by author at Gondwana,

2016) ... 52 Figure 3.9: Over- and under-sampling (A-B) of PM when sampling under

non-isokinetic conditions (Wilcox, 2012) ... 54 Figure 3.10 Monte-Carlo simulation steps (Adapted from Johnson, 2011) ... 56 Figure 4.1: Common RDP houses in the kwaDela Township ... 65

Figure 4.2: Images of solid-fuel distribution in kwaDela: (A) Local coal

merchandise, (B) Mode of local transport, (C) Wood sold at general

dealer, (D) Wood collected from the field ... 69 Figure 4.3: Damaged traditional cast-iron coal stoves in kwaDela ... 74 Figure 4.4: Number of burning events started per hour of the day in summer and

winter ... 76 Figure 5.1: Diagram showing experiment set up: (A) Chimney sampling train set up

in kwaDela (B) Image of measuring instruments, and (C) Image of

dilution set up ... 89 Figure 5.2: Fine particulate emissions tests results in kwaDela: Tests 1 to 3 ... 93 Figure 6.1: Probability distribution curves for kwaDela’s daily PM emissions in

winter (A) and summer (B). (C) Comparisons of winter and summer

LIST OF TABLES

Table 2.1: Number of residents using coal for domestic purposes at a provincial

scale in South Africa (Qase, 2000) ... 10 Table 3.1: Specifications of the Horiba PG-250 gas analyser ... 50 Table 3.2: Summary of Monte Carlo input descriptive statistics ... 58 Table 4.1: Different energy sources and consumption percentages in kwaDela

Township ... 67 Table 4.2: Comparison of seasonal energy use (solid fuels and electricity) for

different mode of usage ... 68 Table 4.3: Summary of solid-fuel use and cost in kwaDela Township ... 71 Table 4.4: Solid-fuel costs per kg, energy content per joule and residents’

perception ... 73 Table 5.1: Fine particulate matter (PM2.5) filters isokinetic direct chimney sampling

results showing emissions of PM2.5: Tests 4 and 5. (PM=Particulate

Matter) ... 93 Table 5 2: PM2.5 Emission factors from all tests in g.kg-1 ... 94 Table 6.1: Diurnal winter and summer fine particulate emission trends (kilograms

ABBREVIATIONS, ACRONYMS AND GLOSSARY

˚C Degrees Celsius

µg/m³ micrograms per cubic metre

µm micrometre / micron

Al aluminium

BLUD Bottom-Lit Up-Draft

C carbon

Ca calcium

CO carbon monoxide

CO2 carbon dioxide

DEAT Department of Environmental Affairs and Tourism

DME Department of Minerals and Energy

DNEP Department of National Electrification Programme

DOE: INEP Department of Energy’s Integrated National Electrification Programme

EF Emission factor

ESKOM Electricity Supply Commission

Fe iron

FRIDGE Funds for Research into Industrial Development, Growth and Equity

g.kg-1 _{grams per kilogram}

HPA Highveld Priority Area

IEA International Energy Agency

IHME Institute for Health Metrics and Evaluation

IPCC Intergovernmental Panel on Climate Change

K potassium

Mg magnesium

Na sodium

NDIR Non-Dispersive Infrared Absorption

NEMA National Environmental Management Act

NEMAQA National Environmental Management Air Quality Act

NOX nitrous oxide

NOVA Navorsirey Ontwikkcing Virlcoming Aermoede

O2 oxygen gas

P phosphorus

PM Particulate Matter

PM10 Particulate Matter with aerodynamic diameter  10 µm

PM2.5 Particulate Matter with aerodynamic diameter  2.5 µm

RDP Reconstruction and Development Programme

Si silicon

SO2 sulphur dioxide

SO3 sulphur trioxide

STATS SA Statistics South Africa

Ti titanium

TLUD Top-Lit Up-Draft

US EPA United States Environmental Protection Agency

CHAPTER 1: INTRODUCTION

1.1

INTRODUCTION AND PROBLEM STATEMENT

Energy is a basic human need, electricity and other primary energy sources such as categorised as either liquefied petroleum gas, bio-gas coal, wood, or dung are used for lighting, cooking, space heating, electric appliances and entertainment. Approximately 95 % of low-income residents in third-world countries burn solid fuel for cooking and space heating, predominantly in winter (Smith, 1988; Smith et al., 1994). More than 90 % of low-income residents in African countries depend on bio-fuels as their primary source of energy (Ludwig et al., 2003). By 2030, the number of residents burning solid fuels for domestic purposes in Sub-Saharan Africa is predicted to grow to 1 billion people (International Energy Agency) (IEA, 2010). Despite high rates of electrification, residents in rural parts of South Africa continue to use solid fuel. Studies conducted in the Eastern Cape and Limpopo Provinces on coal use reported that 95 % and 40 % households living at QwaQwa and Lebowa depend on coal, respectively (Ward, 1995). Another study by Aitken (2007) in the Eastern Cape, KwaZulu-Natal, and North West Provinces indicated that between 76 % and 98 % of residents continue to use wood for domestic energy. In the Highveld, low-income residential areas use coal as the predominant source of energy. Low-income residents depend mainly on solid fuels because they are readily available, easily accessible, affordable compared with clean, modern fuel and can be used for space heating, cooking and boiling water at the same time (Naidoo, 2014). According to Davis (1998) and Heltberg (2004) energy use varies significantly in time and space depending on factors such as: (i) Income (ii) fuel availability (iii) cultural preferences (iv) social class (v) demographic (vi) seasonality and (vii) location

The variability, complexity and interlinked relationship between the factors implies that factors governing fuel use in other settings may not be generalised. Solid fuels are burned in inefficient stoves, leading to high gaseous and particulate emissions causing indoor air pollution, poor ambient air quality and significant impact on human health. Bi et al. (2006) assert that high emissions from domestic burning of solid fuels result from: (i) poor combustion conditions (ii) poor coal stove-operation behaviour (iii) absence of infrastructure to reduce emissions and (iv) using low-grade fuel.

Particulate matter (PM) is a combination of solid particles and liquids emitted during combustion of fuel. It varies in surface area, chemical composition and size, depending on their origin (Worobiec et al., 2011; Petkova et al., 2013; Ninomiya et al., 2004). Particles with aerodynamic diameter between 2.5 µm and 10 µm are “fine” particles, “submicron” particles are less than 1.0 µm and “coarse” particles have a diameter of more than 10 µm (Zhang et al., 2011). Domestic burning is a major source of fine particulate matter. Particulate matter emitted from domestic combustion is mainly fine particles (PM10 and PM2.5) and consists of chemical compounds such as sulphate, nitrate, and organic carbon particles (Thabethe et al., 2014). Continuous exposure to high particulate matter concentrations has a major impact on human health, decreasing an individual’s life-span by 8.6 months (Orru et al., 2011). The elderly, children and women in low-income residential areas are more susceptible to air pollution diseases when continuously exposed to high concentrations of fine PM (Pope, 2000). South Africa has ranked acute respiratory diseases as second major cause of death in children under the age of 5 years (Thabethe et al., 2014). The concern over fine particulate matter is that it has longer ambient residence time when compared with coarse particulate matter and is small enough to travel through respiratory pathways, penetrating into bronchioles and alveoli of lungs, blocking air sacs causing deficiency of oxygen in the blood (Jimoda, 2012). The use of solid fuel for combustion using inefficient stoves increases the chance of getting acute respiratory diseases by a factor of 2.4 (Rehfuess et al., 2009). Globally, 3 million people die every year due to fine particulate matter-related diseases (Lim et al., 2012; WHO, 2013). For instance, indoor air pollution from solid-biomass use results in about 600,000 premature deaths annually in Africa (IEA, 2004).

The adverse negative impact of fine particulate matter and high consumption rate of solid fuels in low-income residential areas has motivated the development and implementation of the National Air Quality Act (2004) to regulate fine PM emissions and monitor ambient air quality. To effectively manage emissions from domestic burning, high spatial resolution data on solid-fuel consumption in residential areas and emissions from domestic burning are required. One of the major barriers for effective management of air pollution in residential areas is the lack of data regarding solid-fuel consumption and emissions patterns in the domestic sector (DEAT, 1999).

An emission factor (EF) quantifies the total amount of a pollutant emitted per amount of fuel burned (Mitra et al., 2002). Emission factors from domestic burning are highly variable as they are governed by several other factors that vary in space and time. Factors governing

emissions of particulate matter are classified into the following categories by Schmidl et al. (2011)

(i) appliance reliant factors (size of the stove, combustion air supply and regulation, the design and size of the burning chamber)

(ii) fuel reliant factors (type of fuel burned, fuel chemical composition and fuel moisture content), and

(iii) operational factors (lighting method used, and amount of fuel used, refuelling).

Most reported emission factors were measured in developed countries which may lead to an overall under-estimation of PM emissions in developing countries. There is a need for local emission factors to eliminate uncertainties related to data availability and EF variation. The absence of data on fuel use and variability in low-income residential areas introduces uncertainties when quantifying total emissions (Zhao & Frey, 2004). Unreported high uncertainties in emission estimates lead to the development of misinformed emission inventories, emission standards that underestimate risk exposure, ineffective energy policies and solid-fuel use intervention methods (Zhao & Frey, 2004).

Variables that may be controlled in a laboratory setting are (i) fuel load (ii) fuel size, and the ignition method used. The applied method sampling emissions from a field is similar type as that of households use daily. For this reason, laboratory-generated EF may under-estimate fine PM risk exposure in the real-life situations. Rodent et al. (2007; 2009) reported that PM emissions measured from the laboratory experiment were 2 to 4 times smaller than PM EF measured in the field.

Field measurements provide realistic PM EF that cannot be reproduced in a laboratory environment. Laboratory-generated emission factors are suitable for examining how the emissions factors vary with varying burning conditions (Sheng et al., 2013), that is (i) fuel type (ii) fuel moisture, and burning method. In general reported emission factors results are integrated with solid-fuel data to quantify residential areas’ seasonal and annual emission estimates. Emission estimates are subject to uncertainties inherited from systematic, random measurement errors that occurred during data collection and analysis. These errors accumulate causing uncertainties, raising questions about the reliability and accuracy of the reported emission estimate. For emission standards, compliance monitoring and intervention methods to effectively improve air quality, the reported emission estimates must be a good representation of the actual ambient conditions. Hence the new requirement is that uncertainties in emission estimates should be quantified and reported in order to

identify major uncertainties sources and they can beminimised to improve the quality of emission inventories (Zhao & Frey, 2004). The IPCC Good Practice Guidance (2000) and Convention on Long-range Transboundary Air Pollution requested that the uncertainty degree of emissions should be quantified to know the confidence of the reported emission range and to identify major sources of emission uncertainties (Romano et al., 2004). Emission estimates’ systematic, random errors and biases lead to incorrect conclusions about emission trends and sources (Frey et al, 1998). It is, therefore, necessary to understand and report uncertainties of emission data before use (Zheng et al., 2011). A Monte Carlo simulation model can be used to determine the probability distributions and uncertainty range of emission estimates.

1.2

AIM AND OBJECTIVES

This study aims to quantify emissions from domestic solid-fuel burning because this is where the largest uncertainties in ambient air quality modelling lie. Improving emission estimates enables better ambient air quality assessment. This will lead to more accurate health impact assessments that will help decision makers prioritise interventions. This will be achieved by

1. characterising solid-fuel use and burning device operation behaviour

2. quantifying and characterising fine particulate emissions from typical burning events, and

3. estimating the seasonal variability of fine particulate matter emissions from domestic burning in a low-income settlement.

1.3

STUDY JUSTIFICATION

The intention of the study is to quantify emissions of fine particulate matter from residential burning of coal using empirical field measurements.

1.3.1 The reason for focusing on low-income residential areas

Domestic burning is a major source of fine particulate matter in low-income settlements, contributing about 60 % of fine particulate pollution levels at a local scale (Engelbrecht et al., 2002; van der Berg et al., 2015). Gaseous and particulate matter emissions in the domestic sector occur at breathing level, causing indoor air pollution, increasing human risk exposure and degrading overall ambient air quality. Solid-fuel use and emissions from domestic burning are highly variable and, therefore, cannot be generalised. Emissions of fine

particulate matter from domestic-burning devices (for example, coal stoves, coal braziers, wood stoves, liquefied petroleum gas rings, and paraffin wick-stoves) range between 0.0018 g.kg-1 _{and 4.0 g.kg}-1 _{(Graham & Dutkiewicz, 1993). Studies have found that the}

concentration of fine particulate matter in low-income residential areas is 51 % higher than in urban residential areas and 78 % higher when compared with industrial areas (Hersey et al., 2015). High emissions in low-income residential areas are due to the high solid-fuel consumption rate, poor stove-operation behaviour, and lack of infrastructure to reduce emissions (Bi et al., 2006). This makes low-income residential areas one of the priority areas to target for air-pollution mitigation in South Africa.

1.3.2 The selection of kwaDela Township

kwaDela Township is a low-income residential area located in the Mpumalanga Highveld region. In 2007, the Highveld region was declared a High Priority area by the Minister of Environmental Affairs and Tourism due to the high concentration of ambient gases and particulate matter in the area (DEAT, 2007). Monitoring stations in the Highveld reported continuous exceedance of the DEAT fine particulate matter daily limit (180 μg.m-3_{) (Mdluli,}

2007). Mpumalanga Province produces about 95 % of South African coal and is ranked the second-largest coal user for domestic purposes in South Africa (Nuwarinda, 2007). The township is a low-income residential area with an average monthly household income of R1,500.00 and experiences “energy poverty”, with a total population of about 75 % using solid fuels despite 97 % electrification (Wernecke et al., 2015). kwaDela Township was selected because it is far from other important industrial sources therefore one can assume that most of the ambient concentrations are related to residential solid-fuel use.

1.3.3 The focus on fine particulate matter

Particulate matter is a product of incomplete combustion, comprising chemical compounds (such as sulphate, nitrate, and organic carbon) (Thabethe et al., 2014). The extent to which particulate matter can have a significant impact on human health depends on its chemistry and particle-size diameter (Thabethe et al., 2014). Fine particulate matter has an aerodynamic diameter < 2.5 µm, with an ambient residence time of hours to days (Lee et al., 2004). When inhaled it can travel deep into the respiratory system, penetrating the bronchioles and alveoli of lungs, blocking air sacs and causing a deficiency of oxygen in the blood (Jimoda, 2012). Diseases associated with fine particulate matter include asthma, heart attack, stroke and lung cancer (Du et al., 2016). The impact of fine particulate matter on human health is a global concern as it can reduce ~ 8.6 months of an individual’s lifespan and caused a global total of ~ 3.1 million premature deaths in 2010 (Samoli et al., 2008). In

2013, 10,432 deaths were reported because of continuous high particulate matter concentrations from domestic burning in South Africa (IHME, 2015).

1.3.4 The reason for quantifying emissions

Studies reported that when annual fine particulate matter decreased by 2.52 µg.m-3_{, it}

resulted in a 3.5 % reduction in air-pollution-related diseases (Dockery et al., 1993). However, activity rate and emission factor data are required to characterise domestic emissions and solid-fuel-use variability, and to calculate emission estimates to develop emission inventories. A high-quality emission inventory can be used to identify major sources of fine particulate matter, and design emission standards that can be used for compliance monitoring. Knowledge about residential solid-fuel use and patterns of emissions can be used to design applicable intervention methods that can effectively reduce emissions from domestic burning, improving overall ambient air quality in residential areas.

1.3.5 The use of field measurements

Field measurements complement emission patterns, taking into consideration the variation of solid-fuel use and stove-operation behaviour, and give empirical measurements that cannot be reproduced in a laboratory environment (Sheng et al., 2013). However, this does not make laboratory-generated PM EF from fuel combustion insignificant. Laboratory-generated emission factors best examine how the emissions factors vary with varying burning conditions (Sheng et al., 2013), that is (i) fuel type (ii) fuel moisture, and (iii) burning method.

To meet the study objectives, scientific principles governing data collection methods, analysis and presentation of results were implemented. For each objective, methods and instrumentation used – with their limitations and uncertainties – are reported. For example, the methodology to meet the first objective requires the use of a sample population. Therefore, suitable sampling methods were used to ensure “representativeness” of the population. The second objective follows an experimental method where gases and particulate matter are sampled to quantify the particulate matter emission factor. An isokinetic sampling method, in line with EPA standards, was used instead of the conventional method to obtain a good representation of flue gas inside the chimney. Moreover, data correction factors (dilution and DustTrakTM _{II correction) were calculated to}

correct the concentration measurements before use. Collecting data is costly, time-consuming and requires labour. This results in using small sample sizes which may under- or over-represent the population characteristics. To meet the final objective, probabilistic distributions were obtained from literature and expert judgement which may introduce

uncertainties into the results. Details of the scientific principles governing data collection and methods of analysis are documented in Chapter 3.

CHAPTER 2: LITERATURE REVIEW

2.1

FUEL USAGE IN DEVELOPING COUNTRIES LOW-INCOME

RESIDENTIAL AREAS

More than half (52 %) of the global population burning solid fuels is found in developing countries like India, China and countries in Africa (Holdren et al., 2000; Mehta & Pruss-Ustun, 2008). About 80 % of low-income residents in third-world countries burn solid fuel for lighting, cooking and space heating predominantly during winter (Bonjour et al., 2013). This is mainly because they do not have access to electricity and those who do have cannot afford it (Mapako & Pasad, 2005). More than 90 % of Africans residing in low-income settlements depend on bio-fuels as their primary source of energy (Figure 2.1) (Ludwig et al., 2003).

Figure 2.1: Percentage of solid-fuel use for cooking in African countries (WHO, 2010)

At a global scale Sub-Saharan Africa ranked second: 74 % of the population does not have access to electricity and 70 % of the rural residents depend on traditional fuel (wood, coal and kerosene) as their primary source of energy (IEA, 2014). Roughly 77 % of Swaziland’s population residing in rural areas use bio-mass for cooking and space heating (Machisa et al., 2013). About 84 %, 95 % and 69 % of households in Mozambique, Malawi and Nigeria, respectively, rely on bio- and solid fuels for energy needs (Prasad, 2010). Bonjour et al. 2013, reported a 50 % increase in the population using solid fuels in African countries between 1980 and 2010. The number of residents burning solid fuels in Sub-Saharan Africa

is predicted to grow to 1 billion by the year 2030 (IEA, 2010). This assumption is based on the annual population growth, increasing prices of modern fuels and low electrification rate in rural areas (Sulaiman et al., 2017).

According to Doppegieter et al. (1998), 60 % of the South African population use wood and coal as their primary source of energy for domestic use. During the year 2001 about half (between 33 % and 48 %) of the total population was using solid fuel mainly for cooking, ironing and keeping warm especially during the winter season (Norman et al., 2007; Wichmann & Voyi, 2006). Despite the 85 % electrification achievement in South Africa, with a rate of 80 % in urban and 50 % in rural areas, more than 80 % of low-income residents continue to use solid fuels (Mdluli, 2007; Ismail & Khembo, 2015). About 2.5 million of the total population is not connected to the national electricity grid (Ferriel, 2010). Figure 2.2 shows the different types of traditional fuel used in South African provinces.

Figure 2.2: Percentage usage of different solid fuels in South Africa obtained from van den Berg, 2015.

The predominant type of fuel used varies spatially: wood is mostly used in all provinces, coal and dung are predominant in the Highveld region, paraffin is used mainly in the Eastern Cape Province, while liquefied petroleum gas is the dominant fuel in the Northern Cape Province (van den Berg, 2015). The percentage of residents using wood in the rural parts of the North West and kwaZulu-Natal Provinces ranges between 76 % and 98 % (Aitken, 2007). A study conducted in the rural parts of Limpopo Province and the Setswetla informal

settlement in Tembisa Township reported 100 % and 99 % bio-mass and paraffin use, respectively, for cooking and heating (Masekoamang et al., 2014; Kimemia & Annegarn, 2011). A review study (Ward,1995) on fuel use in South Africa reported 95 % and 45 % of coal for domestic purposes in QwaQwa and Lebowa, respectively; more than 80 % used gas in Namaqualand and up to 95 % used dung in Transkei, Lebowa and Ciskei. The Highveld region – which is part of the Gauteng, Mpumalanga and Free State Provinces – is known as a “coal hub” and has the highest number of residents using coal (Table 2.1) (Qase, 2000; Nuwarinda et al., 2007). Coal is readily available at an affordable price; the consequence of low prices is extensive combustion of solid fuels.

Table 2.1: Number of residents using coal for domestic purposes at a provincial scale in South Africa (Qase, 2000)

2.2

HOUSEHOLD ENERGY CHOICE DETERMINANTS

The choice of a fuel that can be used in a household is a complex decision governed by a number of factors that are interlinked with each other and vary in space and time (Thorn, 1994). The factors can be divided into (i) household demographic factors (ii) affordability (iii) fuel-related costs (iv) culture and tradition, and (v) seasonal variability and location.“Demographic factors” include house type (family size, age distribution and level of education). “Household income” includes the total monthly income received from all working members in the house and the percentage of amount allocated for energy use. “Fuel-related

Province

Number of households

using coal

costs” include primary fuel price, price of other available fuels, costs of fuel collection. Recently, “time and labour used for fuel collection” has been added on to fuel-related costs. “Geographical factors” include season and location (rural area or urban). The “level of importance” of each factor varies in space and time. Any of the other factors may be present and/or important in certain areas and may be absent or not that important in other areas or location.

2.2.1 Fuel type availability and accessibility

“Fuel type” and “energy patterns” differ in urban and rural areas (Panchauri & Jain, 2008). Urban residential areas tend to use “modern fuel” (electricity, liquefied petroleum gas), while rural residents tend to use “traditional fuel” (wood and coal). Factors that limit use of traditional fuel in urban areas include: space limitations due to high population density and difficult access to traditional fuel sources (Panchauri & Jain, 2008). Results from a study in Malawi on fuel use showed that an increase in distance by a kilometre from the fuel source to the residential area decreases chances of fuel collection by 3.7 % (Jumbe & Angelsen, 2006). Thus urban sprawl also contributes to the variation of fuel use in rural and urban areas. As a city grows, obtaining traditional fuel becomes difficult and costly due to transport cost and having to buy the fuel since collection of fuel will take more time and human labour (Panchauri & Jain, 2008). Rural residents tend to use traditional fuel mainly because the supply of electricity in rural areas is not usually stable and the costs associated with electrical appliances would be prohibitive if they had to change from using traditional fuels (Xiaohua & Zhenmin, 2003). Even residents who can afford to buy modern appliances and have reticulated electricity but live in rural or remote areas end up using traditional fuels due to the inconsistent and unreliable supply of modern fuels (Leach, 1992). There is not enough space to store traditional fuel in urban areas and collection areas and markets for traditional fuel are usually in rural areas. However, living in an urban area does not necessarily mean that the households use modern fuel due to urban poverty. About 80 % of poor households living in urban areas in China continue to use predominatly traditional fuel (Wang & Feng, 2003).

2.2.2 The effect of climate on fuel use

Change of season has an influence on ambient temperature and vegetation growth. During rainy seasons plants grow, increasing fuel wood availability and agriculatural residue. When climatic conditions are not favourable for plant growth, low-income residents resort to alternatives available: traditional fuel such as kerosene, animal dung or coal. Seasonal fuel shortages drive up fuel costs thus, in seasons where coal or wood is scarce, their prices go up, in turn decreasing demand and consumption (De Cian et al., 2012). During winter, the

high solid-fuel consumption rate increases due to space-heating needs while, in summer, solid fuel is used only for cooking (Naidoo et al., 2014). The number of residents burning solid fuels, and the number and duration of burning events in summer decreases (Carter et al., 2016).

2.2.3 Household demographic factors

“Demographic factors” such as household size, material used used to build the house, family size, age distribution, and level of education of the household influence the type of fuel used. Households in rural areas tend to choose fuels that are cheaper to account for space-heating needs without going over their monthly energy budget. The material used to build the house has an important effect and impact on household indoor temperature. Houses built from wood, mud, or zinc material are usually very cold when it is cold and uncomfortably hot when the weather is warm compared with houses built with bricks. Thus households are most likely to use traditional fuel to keep warm during winter. Chances of using cleaner, efficient fuel increased by 5.67 % for households living in brick/concrete houses (Zhang et al., 2011). It is most likely that there is a need to burn solid fuel in large households compared with small families. A larger family is more likey to use appliances (heaters, traditional cast-iron coal stoves) for space heating, whereas a small family would rather use blankets to keep warm than waste electricty or coal for an individual (Jumbe & Angelsen, 2006). In contrast, Yu (2011) states that large families may have a large enough source of income that they will be able to afford modern fuels, therefore, energy use per individual decreases with increase in number of family members in a household. Households led by educated individuals are most likely to use modern, clean and efficient fuels (Panchauri & Jain, 2008). Education level is correlated positively with high income and knowledge of the negative impact of dirty fuels on human health. Education improves residents’ understanding about the impact of solid-fuel emissions on human health and increases chances of a higher household income which is a stepping stone to cleaner, more efficient energy sources (Boadi & Kuitenen, 2006). Each additional year spent on education or at school increases the chances of using cleaner, more efficient fuel by 0.66 % (Zhang et al., 2011). Studies done in China on fuel use confirmed that there are greater chances of a household using traditional fuel when being led by a household head who has only a foundation level of education (Filippini & Pachauri, 2004). A study in Ghana reported that residents with only a primary or no education are more likely to use traditional fuels (charcoal) when compared with residents that have a secondary or higher education (Boadi & Kuitenen, 2006). However, studies by Wu et al. (2012), found that education has no impact at all on the fuel choice in other places. Instead, fuel choice was driven mainly by other factors like age distribution in the household. Households headed by females are also

more likely to use clean fuels compared with male-headed households (Filippini & Pachauri, 2004). Females are less resistant to change to modern efficient, cleaner fuels when compared with males (Bisu et al., 2016).

2.2.4 Monetary, time and labour costs associated with fuel

There is a definite relationship between costs and price of fuel and choice of energy used in a household. The initial cost for electricity and appliances prevents rural residents from using modern fuels (Panchauri & Jain, 2008). Residents in Tembisa use coal braziers made from cheap local material for cooking and space heating because they are associated with no or less monetary costs compared with electrical appliances (Makonese & Annegarn, 2016). Residents reported receiving brazier-making material at no monetary cost. Braziers are made and sold at lower prices by local residents and sometimes at no labour costs (Makonese & Annegarn, 2016). The price of the fuel and the availability of other fuel(s) determine the type, combination, and how much fuel is used in a houshold. Costs of energy include collection, time and price of the fuel. Costs associated with fuel collection determine the type and amount of fuel used in a household. High prices of modern, efficient, clean fuel drives households to use cheap, dirty fuels (Panchauri & Jiangi, 2008). Prices of the other alternatives affect fuel type and consumption. If the price of electricity or LPG increases, this increases the use of traditional fuel (wood, coal, kerosene). The pressure on other fuels drops off when prices drop. In rural areas where there is enough bio-mass and residents do not have to buy fuel, they are faced only with fuel costs associated with time spent and labour. If a household location is distant from the fuel field, it tends to use less fuel (wood) and look for other options that are affordable and readily available. Residents close to fuel sources (for example, wood fields) are more likely to use more wood because they do not have transport costs since fuel transport is done by human labour and less time is spent on fuel collection (Heltberg, 2004).

2.2.5 Household income allocated for energy use (affordability)

Traditional fuel use tends to increase with a decrease in household monthly income. The less money recieved per month per houshold, the less money they keep aside for their energy budget. This is applicable for households living in both rural and urban areas. Thus households resort to use of traditional fuel that is cheaper and readily available. Rural households generally spend less money on energy compared with urban households, but they spend a higher proportion of their income on energy than urban households. Up to 20 % of rural households’ income is spent on fuel, while urban households spend less than 5 % of their income on fuel (World Bank, 1996a,b; WHO, 2006). The financial status of a household either restricts or allows the households to choose from different energy sources

(Xiaohua & Zhenmin, 2003). The higher the income, the more fuel choices are available to households from which to choose. As household income increases, they are more likely to change from using “traditional” fuels to “modern” fuels because they may want to show off their finacial status by climbing both the “energy ladder” and “social class ladder” (Xiaohua & Zhenmin, 2003). Transition of fuel use is not always evident as household income increases. In rural areas, the transition to modern fuel is not direct and quick; direct transition occurs when the household monthly income increases substantially (Sheng et al., 2015). Conclusions from a study in China were that as household income increases, transition of fuel use may not be evident. It can happen that a certain part of the population may move up the energy ladder but a certain population may choose to use more of the fuel they have been using all along (traditional fuel). Households that recieve less income tend to diversify and use different types of fuel for specific household activities. Electricity may be used for powering appliances like fridges and televisions, whilst coal and wood are used for space heating, cooking, ironing and boiling water. Residents who can afford alternative fuels choose to invest their time and money on education and electricity (Elkhom et al., 2010).

2.2.6 Cultural practice and tradition influence on fuel use

Cultural behaviour and practices play an important role in household decisions regarding fuel use. Cultural factors, such as family diet and cooking habits, have an impact on burning patterns and length of the burning cycles. Traditional meals like tripe, dry beans, samp and cow feet take long to cook because they are cooked at low heat to make sure they do not burn. The longer cooking period is to ensure that they are well cooked and even sometimes overcooked and ensures that the cooked food is tender and tasty. Other residents, especially where the head of the household is an adult or grandmother, prefer to use traditional fuel and stoves because food cooked using traditional fuel and modern fuels does not taste the same. Residents claim that food cooked using traditional fuel has a distinctive taste and tastes even better than when it is cooked using modern fuel. Even households residing in urban areas and receiving a high income choose to use traditional fuel when cooking these types of meals because of the distinctive flavour. Residents in Thailand’s rural, semi-urban and urban areas prefer to cook a specific rice (glutinous rice) using wood or coal and another (non-sticky rice) using electricity (Nansaior et al., 2011). According to Nansaior et al. (2011) coal can preserve the traditional cultural taste residents prefer when cooking a certain dishes. Therefore, despite having access to, and affording, electricity residents will continue to use the preferred fuel for certain food they eat (traditional food). This shows that it will take more than awareness about the health impact of using traditional fuel to change the perception and to adopt modern fuels. Another important cultural aspect that has an influence on energy use in traditional-rooted communities is the Patriachial

system. The system states that only men make important decisions as the head of the house, including decisions regarding energy use. Studies on fuel use concluded that men are resistant to change to using modern fuel when compared with woman. Thus, in such communities, it is most likely that traditional fuels are used predominantly. Even though factors affecting fuel use vary in space and time, the most critical factors determine the choice of household fuel-type are income and location (Elkhom et al., 2010).

2.3

DOMESTIC SOLID-FUEL-USE MODELS (ENERGY LADDER

AND FUEL STACKING)

About 90 % of low-income residents in developing countries rely on solid-fuel use for meeting their energy needs (van der Kroon et al., 2013). The solid fuels are mainly burned in traditional inefficient old stoves which emit pollutants that have a negative impact on human health and ambient air (Davis, 1998). Understanding residents’ energy profiles and fuel switching is required to design efficient energy policies that will assist with low-income residents transitioning to cleaner fuels (van der Kroon et al., 2013). Low-income residents’ standards of living can be improved by encouraging the shift from using traditional fuels to modern fuels, and using efficient, clean stoves (Reddy, 1994). The failure to understand the dynamics of solid-fuel use in multi-purpose traditional burning devices in rural areas may have led to the assumption that electrification will remove the use of solid fuels (Howells et al., 2005). There are four main factors that are taken into consideration when deciding which type of fuel should be used in a household as in the multiple fuel-use model. This includes economic status, fuel availability, type of fuel-burning appliance available, cultural factors and fuel health impact (Heltberg, 2004). Other factors that affect energy-use choices in low-income households include illegal electricity connections. Mostly these factors are not taken into consideration in the models. Moreover, primitive fuels (wood, coal) are collected from the fields or coal dumps or sold informally, therefore, it is difficult to obtain statistics on solid-fuel use. This may lead to an under- or over-estimation of solid-fuel usage in low-income residential areas. Misinformed policies delay the transition to modern fuels in rural and urban areas (Reddy, 2004).

Two models have been put forward to explain residents’ fuel-transitioning processes: the “energy ladder” and “energy stacking”. The theories divide household energy use into “bottom”, “transitioning” and “top” levels. Bottom-level households use dung, agricultural residue and wood. During the transitioning stage, coal, charcoal and kerosene are used while, at the top level, more advanced “cleaner” fuels, such as electricity, LPG and bio-fuels are used (Heltberg, 2004). Figure 2.3 shows how the different types of fuel are clustered in each model and how they change with household economic status.

Figure 2.3: Comparison of the energy ladder and energy stacking model (van der Kroon et al., 2013)

Households in the lower economic brackets use traditional fuels, mainly because they are affordable, available in large volumes, have a multi-purpose function, and use cheap, even unsophisticated, appliances (Naidoo et al., 2014). The traditional energy-ladder theory states that residents will move from using inefficient low-cost, dirty fuels (dung, wood, coal) to clean fuels (electricity) as they move up to higher socio-economic classes (Leach, 1992). The theory suggests that there is a one-way linear transition between the use of traditional and modern fuel-driven households’ economic status (Davis, 1998). The assumption is households will be able to afford and maintain advanced burning appliances as their income increases and their diet changes from eating “traditional” food to “modern” food that cooks faster (Jane Trac et al., 2011). Advanced cooking or heating devices are costly and efficient, require less labour, emit less pollution and cook food faster (Masera et al., 2000). Furthermore, residents aspire to using modern fuel and burning or heating devices to demonstrate improvement in a household’s economic status. The speed and extent at which residents move to modern cleaner fuels is dependent on the household’s income, and the cost and availability of electrical appliances, and the cost, availability and reliability of modern fuels relative to other fuels (Masera et al., 2000). As soon as they can afford to buy electrical appliances and cleaner fuels, they move away from using dirty fuels. Masera et al. (2000) describe how low-income residents’ income is highly variable and uncertain. This acts as a barrier when the residents move to use modern fuels like electricity. Modern fuels require a sustainable income which increases with electricity prices. The price of modern fuel (LPG) can be up to four times more than for solid fuels (Masera et al., 2000). High initial investment to buy and maintain electrical appliances and a well-connected electricity grid that supplies electricity sustainably is a barrier for adopting modern fuels.

Other barriers to complete transition to modern fuels in low-income residential areas are culture, the cooking practices and type of food. “Traditional food” consumed in a community is mainly rooted in their culture (Joon et al., 2009).

A case study in Mexican villages reported that the availability of modern fuels and increase in household monthly income does not necessarily remove the use of primitive fuels (Masera et al., 1997). The findings of the study reported that none of the residents from the villages moved completely away from using solid fuels. At the beginning of the study, residents were using both solid- and modern- fuels simultaneously. After a while, some of the residents stopped using LPG for cooking. Instead, they reverted to using solid fuels. Less than 20 % of the residents moved completely from using wood to LPG. Similar results were reported in a study conducted in Haryana, a rural area in India. About 80 % of the population was using both primitive- (wood, dung, crop residue) and modern- (LPG) fuels at the same time, with primitive fuels as the primary source of fuel (Joon et al., 2009). None of the residents shifted completely to use modern fuel.

Jane Trac (2011) argues that improvement of household’s economic status drives fuel shifting from traditional- to modern- fuel but does influence electricity usage When households use traditional and modern fuel for different at the same time it is referred to as“Energy stacking”. Households are most likely to follow this model as an energy security strategy when transitioning to cleaner fuels. When the family runs out of the modern fuel, they have to wait until they have enough money to refill the LPG cylinder or to buy electricity. The LPG filling station or electricity payment centre might only be in town which requires transportation. During the waiting period (which may take up to days or weeks), residents tend to resort to alternative fuels that are available in their local area and are affordable (Masera et al., 2000).

Households are less likely to shift completely from using traditional fuel and burning devices but rather use both modern- and traditional- fuels to enjoy the advantage of using both at the same time (Masera et al., 2000). Certain fuels are used for specific household activities. For example, for household activities that require a lot of energy (like space heating and cooking) solid fuels are used, while electricity used for lighting and electrical appliances (such as a fridge) (Howell, 2004). Their “traditional” stoves are not abandoned after residents receive “clean” or “modern” stoves as a security measure. This is motivated by the frequent supply and accessibility of solid fuels in the local area and the risks associated with the tendency of the stoves to break or malfunction. Another mitigating factor is the ability of the residents to use their old stoves to avoid the high costs associated with fixing the stoves in cases of malfunction. When an electrical stove breaks, the resident may not

enough cost to fix at the earliest. During the period their modern stoves are sent for (or are awaiting) fixing, residents use their available traditional bigger stoves which use less expensive fuel (Howell, 2004). Most modern stoves do not have enough cooking area for the residents to cook a full meal and boil water at the same time. In most cases, low-income residents have large families, therefore, they use large pots and water boilers. Other residents believe that they have to see a flame for a stove to be able to cook. Therefore, when designing stoves for these communities all these factors should be taken to consideration (Joon et al., 2009).

Foley (1995) argues that it is not the “energy ladder” that determines the household’s energy profile but rather their energy demand. At the early stages of a household’s development, the residents need energy mainly for cooking and space heating. As development progresses, different household appliances that require specific energy types are bought, leading to “fuel stacking”. He argues that solid fuels will remain the primary fuels for basic household activities such as cooking or space heating. Heltberg (2005) further adds that “fuel stacking” is a transition stage that households go through and it may disappear when they are at the last stage of development. However, the case is different in low-income residential areas, solid fuels remain the predominant source of energy even with increasing household income (Patarasuk et al., 2016). Joon et al. (2009) argue that the energy ladder can be used to explain fuel use in urban or semi-urban areas while the multiple fuel-use model can be used in rural areas. Due to high demand and continuing use of solid fuels in low-income residential areas, the use of primitive fuel may not be replaced by modern fuels as anticipated by the “energy ladder” theory (Joon et al., 2009). Therefore, the most applicable intervention methods in low-income residential areas may include building high-quality insulated houses, designing low-smoke stoves with good ventilation, using the top-lit up-draft ignition method (Joon et al., 2009).

2.4

CHARACTERISING PM EMISSIONS FROM DOMESTIC BURNING OF

SOLID FUELS

Domestic burning is a major source of gaseous (carbon monoxide, nitrogen oxides, sulphur) and fine particulate matter, and widely practised in South Africa (Olave et al., 2017; Nuwarinda et al., 2007). Graham and Dutkiewicz (1993) reported fine suspended particles (0.0018 to 4.0 g.kg-1_{) and total suspended particulate matter (0.0019 to 7.9 g.kg}-1_{) from}

burning wood using different household burning devices. The total amount of fine particulate matter emitted from coal-burning stoves was 3.9 g.kg-1 _{of wood burned (Figure 2.4).}

Figure 2.4: Emissions of particulate matter from different household burning devices (Graham & Dutkiewicz, 1993)

Emissions from domestic combustion are accountable for 57 to 75 % of the ambient concentration of fine particulate matter in Soweto with a peak concentration of 110 µg.m-3

(Nuwarinda et al., 2007). Bi et al. (2008) note that high emissions from domestic combustions result from (i) poor combustion conditions (ii) poor stove-operation behaviour (iii) absence of infrastructure to reduce emissions, and (iv) the type and grade of fuel burned. Low-income residents’ burn high-ash coal, emitting large amounts of polycyclic aromatic hydrocarbons and oxy-polycyclic aromatic hydrocarbons which are major components of particulate matter and gaseous precursors for formation of secondary particles (Zhang et al., 2011; Lipsky et al., 2005). Particulate matter (PM) is a combination of solid particles and liquids emitted during combustion of fuel that vary in origin, surface area, chemical composition and sizes (Figure 2.5) (Worobiec et al., 2011; Petkova et al., 2013; Ninomiya et al., 2004).