Novel electrical particle emission reduction methods for small-scale biomass combustion

(1)

REPORT SERIES IN AEROSOL SCIENCE N:o 242 (2021)

NOVEL ELECTRICAL PARTICLE EMISSION REDUCTION METHODS FOR SMALL-SCALE BIOMASS COMBUSTION

HEIKKI SUHONEN

Department of Environmental and Biological Sciences Faculty of Science and Forestry

University of Eastern Finland Kuopio, Finland

Academic dissertation

To be presented, with the permission of the Faculty of Science and Forestry of the University of Eastern Finland, for public criticism in auditorium MD100, Mediteknia Building at the University of Eastern Finland, Kuopio, on October 29th, 2021,

at 12 o'clock noon.

Kuopio 2021

(2)

Author’s Address: Fine Particle and Aerosol Technology Laboratory University of Eastern Finland

Department of Environmental and Biological Sciences P.O. Box 1627

FI-70211 Kuopio, Finland email. heikki.suhonen@uef.fi

Supervisors: Associate Professor Olli Sippula, Ph.D.

Department of Environmental and Biological Sciences and Department of Chemistry

University of Eastern Finland

Professor Jorma Jokiniemi, Ph.D.

Department of Environmental and Biological Sciences University of Eastern Finland

Docent Jarkko Tissari, Ph.D.

Department of Environmental and Biological Sciences University of Eastern Finland

Reviewers: Docent Patrik Yrjas, Ph.D.

Laboratory of Molecular Science and Engineering Åbo Akademi University

Associate Professor Topi Rönkkö, Ph.D.

Physics Unit Tampere University

Opponent: Professor Heinz Burtscher, Ph.D.

Institute for Sensors and Electronics

FHNW University of Applied Sciences and Arts

ISBN 978-952-7276-65-5 (printed version) ISSN 0784-3496

Helsinki 2021 Unigrafia Oy

ISBN 978-952-7276-66-2 (pdf version) Helsinki 2021

http://www.FAAR.fi

(3)

Abbreviations

BC Black Carbon

CHX Condensing Heat Exchanger

CO Carbon monoxide

CPC Condensation Particle Counter DMA Differential Mobility Analyzer Dp Diameter of particles

EC Elemental Carbon

ELPI Electrical Low-Pressure Impactor ESP Electrostatic Precipitator

FTIR Fourier Transform Infrared Spectroscopy GMD Geometric Mean Diameter

HiTESC High-Temperature Electric Soot Collector MCP Medium Combustion Plant

NELPI Particle number concentration measured by ELPI NTP Normal Temperature and Pressure

OC Organic Carbon

PAH Polycyclic Aromatic Hydrocarbon PM Particulate Matter

PN Particle Number

PTFE Polytetrafluoroethylene

RBC Residential Biomass Combustion RE Reduction Efficiency

REF Reference heat exchanger SCC Shielded Corona Charger

SMPS Scanning Mobility Particle Sizer TSP Total Suspended Particles

(4)

Acknowledgements

The research presented in this thesis was carried out in the Fine Particle and Aerosol Technology group, Department of Environmental and Biological Sciences, University of Eastern Finland. Financial support for the study was provided by the Doctoral School of the University of Eastern Finland, Doctoral Program in Environmental Physics, Health and Biology.

I wish to extend my deepest gratitude to my principle supervisor Associate Professor Olli Sippula for providing me guidance and support during this work. Your help has been irre- placeable in both planning the methods used in this work, but above all, in writing the scientific publications. I would like to thank my supervisors Professor Jorma Jokiniemi and Docent Jarkko Tissari for giving me the opportunity to be part of the Fine Particle and Aerosol Technology group and providing the facilities for conducting this research.

I would also like to express my gratitude to the official reviewers of this thesis, Docent Patrik Yrjas and Associate Professor Topi Rönkkö, for their valuable comments on the presented work. I am also grateful to Professor Heinz Burtscher, for kindly accepting the invitation to serve as an opponent in the public examination of the thesis.

I want to thank all the co-authors I have had the honor to work with. Special thanks to Miika Kortelainen for supporting me in all my laboratory work, and sorry for the early wake ups and long measurement days. I want to acknowledge Ari Laitinen for sharing the comprehensive expertise related to electrostatics. Thanks to all the wonderful colleagues at the Fine research group for the inspiring conversations during the lunch and coffee breaks.

I want to thank my best friends for providing me something else to think and experience during my free time. I thank my father Martti, my mother Helena and my brother Mikko for their endless support throughout my studies and adventures that led to this point. Final- ly, my warmest gratitude goes to my beloved fiancée Nina, who has solved my hardest MATLAB problems, but also has given me the mental strength to complete this work.

Kuopio, October 2021 Heikki Suhonen

(5)

Novel electrical particle emission reduction methods for small-scale biomass combustion

Heikki Tapio Suhonen

University of Eastern Finland, 2021 Abstract

The particulate matter (PM) emissions of residential biomass combustion cause deteriora- tion of air quality, adverse health effects and climate effects. Black carbon (BC), which is often the main component of particle emissions in incomplete combustion, contributes to climate warming by directly absorbing sunlight in the atmosphere and by changing the albedo of glaciers. Therefore, the European Union is regulating the particle emissions of small-scale combustion. The new regulations create a need for novel technologies for emission reduction.

The aim of this thesis was to investigate and develop novel technologies for the reduction of particle emissions of small-scale biomass combustion. The study focused on character- izing two different methods, which both use electrostatic forces for particle capture. The first method is a diffusion charger, which can be used as a retrofit device in existing boiler systems. The diffusion charger has been introduced in earlier studies, but it was further developed and optimized for small-scale biomass combustion appliances in this doctoral work. The diffusion charger was tested in combination with a condensing heat exchanger, a regular heat exchanger, and a cyclone, which served as collection surfaces for the charged particles. The second method introduced in this thesis is based on the deposition of naturally charged particles released from flames. The natural charge is utilized for particle capture with an electric field that is introduced into the combustion zone. The developed methods were compared to an industrial large-scale electrostatic precipitator (ESP).

The diffusion charger proved to be effective when the collection surface provided sufficient residence time and surface area. Both heat exchangers removed approximately 90%

of fine particles (PM1) from a wood chip grate burner. The combination of a diffusion charger with a cyclone was less effective, providing a 27% PM1 reduction efficiency. The natural charge method decreased PM1 emissions by 45% from a manually fired masonry heater. The method was dependent on the combustion phase in batch-wise combustion, because the production and lifetime of natural charges varies in different phases and temperatures.

The advantages of both developed methods are their simple construction, low space requirement and applicability as retrofit devices in existing combustion appliance designs.

The natural charge method also has minimal power consumption, and does not require separate cleaning mechanisms, because the collected particulate mass is oxidized on the surfaces of a combustion chamber. The reduction in efficiency of this method is a draw- back, but it is still often enough for a combustion appliance to reach the upcoming emission regulations. It is practical to use the diffusion charger with a condensing heat exchanger, because the particles are collected on surfaces rinsed with condensed water, which provides an automatic cleaning mechanism.

Keywords: residential biomass combustion, electrical particle reduction, electrostatic precipitation

(6)

II Suhonen, H., Laitinen, A., Kortelainen, M., Yli-Pirilä, P., Koponen, H., Tiitta, P., Ihalainen, M., Jokiniemi, J., Suvanto, M., Tissari, J., Kinnunen, N., Sippula, O., 2021. High temperature electrical charger to reduce particulate emissions from small biomass-fired boilers. Energies 14, 109. https://doi.org/10.3390/en14010109

III Suhonen, H., Laitinen, A., Kortelainen, M., Koponen, H., Kinnunen, N., Suvanto, M., Tissari, J., Sippula, O., 2021. Novel fine particle reduction method for wood stoves based on high-temperature electric collection of naturally charged soot particles. J. Clean. Prod. 312, 127831. https://doi.org/10.1016/j.jclepro.2021.127831

IV Sippula, O., Huttunen, K., Hokkinen, J., Kärki, S., Suhonen, H., Kajolinna, T., Kortelainen, M., Karhunen, T., Jalava, P., Uski, O., Yli-Pirilä, P., Hirvonen, M.-R., Jokiniemi, J., 2019. Emissions from a fast-pyrolysis bio-oil fired boiler: Compari- son of health-related characteristics of emissions from bio-oil, fossil oil and wood.

Environ. Pollut. 248, 888–897. https://doi.org/10.1016/j.envpol.2019.02.086

(9)

9

1 Introduction

Residential biomass combustion (RBC) gives rise to significant amounts of ambient air particulate matter globally. For example, in Finland, there are over 2 million residential wood combustion appliances, and the amount of energy provided by residential biomass combustion alone is 15 TWh (TSY, 2017). Particulate emissions are often not perceived as a health risk by the public (Cori et al., 2020), but scientific studies have estimated that 1500 people in Finland (Lehtomäki et al., 2018) and 400 000 people in Europe (EEA, 2014) die from PM2.5 (particulate matter under a size of 2.5 μm) emissions yearly. RBC is a major contributor to PM2.5 emissions in Finland (Savolahti et al., 2019).

Biomass is often considered to be carbon neutral, but it has been declared that the role of land usage and the supply chain (harvesting, collecting, processing and transporting) should be acknowledged (Brack, 2017; EEA, 2011; Sterman et al., 2018). Furthermore, RBC produces so-called short-lived climate pollutants such as methane and black carbon.

It has been estimated that RBC’s largest role in global warming, though, is black carbon particles. Black carbon particles contribute to climate warming by absorbing sunlight directly in the atmosphere, and additionally via affecting the albedo of glaciers and other snowback areas (Ramanathan and Carmichael, 2008).

The European Union is pushing emission levels lower with the new Ecodesign directives (European Union, 2015a), starting with new devices in 2022 for solid fuel local space heaters, aka, fireplaces. The new emission limits for domestic boilers and industrial power plants are already in force in many countries (EEA, 2019). Typical stoves and fireplaces used in RBC are not very advanced considering particulate emissions, because of the simplified designs of the devices. Modern boilers, however, have automated fuel feeding and combustion air staging, leading to considerably lower emissions than batch-wise combustion devices (Ozgen et al., 2014). So far, emission after-treatment technologies for residential combustion appliances have not become common in Europe. In North America, small- scale catalytic combustors in wood stoves have been available for a few decades (Hearth, 2013). In Europe, there are a few of examples of commercial electrostatic filtration devices (Carola Clean Air, 2021; Exodraft, 2021; Kutzner-Weber, 2021; OekoSolve, 2021;

Schräder, 2021). It can be expected that in future these kinds of after-treatment technologies will become necessary and in common use for residential-scale biomass combustion units.

Aims / objectives of the thesis:

The general aim of this study was to develop novel electrical filtration technologies especially suited for small-scale biomass combustion appliances.

(10)

10 The more specific objectives were:

 To develop and evaluate a shielded corona charger for small-scale biomass combustion appliances (Papers I and II)

 To investigate the effects of the operating conditions on the performance of a shielded corona charger (Paper II)

 To combine the shielded corona charger with a condensing heat exchanger or a cyclone (Paper I and II)

 To develop a method that utilizes naturally charged particles for particulate emission reduction, (Paper III)

 To determine and compare the particle-size dependent collection efficiency of new and traditional electrical particle filtration systems (Papers I, II, III and IV)

(11)

11

2 Scientific background

2.1 Particulate emissions of small-scale biomass combustion

The nature of particulate emissions (PM) depends on the source of emissions. This chapter covers the background of different biomass fuels and combustion appliances and how they affect PM emissions. Particle formation mechanisms and their chemical composition, size distributions and electrical resistivity, which it is important to consider with electric filtration appliances, are also described. Furthermore, ion formation in flames as a source for electric charge is discussed.

2.1.1 Combustion of biomass

The solid biomass combustion process can be divided into four phases: drying, pyrolysis, flaming and char burning (van Loo and Koppejan, 2008). The combustion procedure can be either continuous or in repeating cycles, as in batch-wise combustion. The combustion phases overlap, especially in batch-wise combustion (Elsasser et al., 2013), e.g. in the manual combustion of wood logs, because the outer layer of the log can be in the char burning phase while the inner parts of the log are still drying. Once part of the fuel material has dried, it begins to heat up, and begins to release gases that are mostly formed by the thermal decomposition of cellulose, hemicellulose and lignin, i.e. pyrolysis. The pyrolysis products then react with oxygen, which generates heat and accelerates both the pyrolysis and oxidation reactions, which in practice can be seen as flaming. Once all the volatile material of the fuel is pyrolyzed, the remaining char oxidizes via heterogeneous reactions (Miller, 2016, sec. 4.1.3).

Ideally, in complete combustion, a hydrocarbonaceous fuel would oxidize into carbon di- oxide and water. However, in small-scale biomass combustion appliances, the combustion processes are typically never complete. As a result of incomplete combustion, carbon monoxide, organic compounds and black carbon particles are formed. Organic compounds exist in both gaseous and solid form, depending on their vapor pressures and temperatures in the flue gas (Elsasser et al., 2013; Fitzpatrick et al., 2008).

2.1.2 Particle emission formation in small-scale biomass combustion

The fine particles in biomass combustion flue gas consist mostly of soot (black carbon, elemental carbon), ash and organic material. In addition, flue gases may contain coarse particles, which are ash or char residues. The geometric mean diameter of particle number emissions is typically below 100 nm in efficient combustion conditions (Lamberg et al., 2011; Leskinen et al., 2014; Nuutinen et al., 2014; et al., 2008a).

(12)

12

Soot particles are formed in an early phase of the combustion process via high-temperature polymerization reactions in the oxygen deficit area of the flame. Polycyclic aromatic hy- drocarbons (PAHs) are considered to be soot precursors (Richter and Howard, 2000), which grow into solid soot particles (Fitzpatrick et al., 2008). Next, the particles increase in size by taking up material from the gas phase (surface growth) and by coagulation (i.e.

particle-particle collisions) (Haynes and Wagner, 1981). In flue gas, soot particles are oxidized at a temperature mainly above 800 °C (Lamberg et al., 2018) when the residence time is high enough. In incomplete combustion, some of the soot particles escape from the combustion chamber before being oxidized.

In biomass combustion, fine fly ash is mainly formed from the volatilization of alkali metals, which also happens in complete combustion (Boman et al., 2004). Volatilized ash- forming compounds form new fine particles by homogeneous nucleation or heterogeneous condensation at temperatures of 950-1050 °C, when the alkali sulfate vapors begin to condense. A second step in ash particle formation is the conversion of alkali hydroxide vapors into solid particles, which occurs around temperatures of 650-750 °C. Third, alkali chlo- ride vapors condense at a temperature of approximately 550 °C. Other alkali species (ni- trates and hydroxides) presumably condense below 600 °C (Sippula et al., 2007).

Organic particles are formed when organic vapors are condensed onto existing particles, or the gases may form new particles by nucleation (Tissari et al., 2008). The gas-particle conversion of organic vapors usually occurs far below a temperature of 500 °C (Sippula et al., 2007)

The particulate chemical composition is important in the design and operation of electrical filtration systems. Soot particles are electrically conductive (Lin et al., 1990), whereas ash particles are electrically resistive (White, 1953). It is important to consider the electrical characteristics of particles when electrostatic filtration is applied. This is discussed in more detail in section 2.2.1. Furthermore, it is important to acknowledge the condensation temperatures of different particle chemical components when designing abatement technologies. For example, condensable organic species may be still present in the gas phase in the particle filtration system, if it is operating at high temperatures.

2.1.3 Natural charge characteristics of freshly formed combustion particles

The process of combustion generates an electrical charge. Oxidation reactions include chemi-ionization and subsequent ion chemistry, which leads to the formation of ions and electrons (Fialkov, 1997). Soot particles, also produced in the early phase of combustion, are likely to collide with these ions and therefore carry a charge. In addition, neutral soot, ash and organic particles can become charged during and after combustion via electron emission and collisions with existing charge carriers (Burtscher, 1992; Fialkov, 1997;

Maricq, 2006). The lifetime of the charged particles is short, because both negative and

(13)

13

positive charges exist and they attract each other, ultimately leading to agglomerated neutral particles (Burtscher, 1992).

Natural charge of combustion particles is not commonly utilized for reduction of particle emissions, although it has been discovered as a possibility many decades ago by Bowser and Weinberg (1974). Modern applications of the natural ionization in flames include electrodynamic combustion control, which has been in the process of commercialization by ClearSign for medium and large scale power plants (Energy XPRT, 2021). Similar studies have been conducted by Barmina et al. (2015) and Zake et al. (2001), where the idea is to affect the ion wind with an external electric field to enhance the combustion characteristics of a flame. Flame ions are also utilized for example in engine control (Badawy, 2013).

2.1.4 Effects of fuel on particle emissions

Regarding particulate matter emissions, the most important features of different solid biomass fuels are the chemical composition and physical attributes of the fuel. Chemical composition includes the volatile matter and ash content of the fuel. The chemical composition of wood is primarily carbon (47-52%), oxygen (38-45%) and hydrogen (6.1-6.3%).

These elements form primarily compounds of cellulose (40-45% dry weight), hemicellulose (20-35%) and lignin (15-30%) (van Loo and Koppejan, 2008). Wood also contains minor fractions of nitrogen (≥0.5%) and ash-forming species (≥0.5%).

Physical attributes of the fuel affecting the particulate matter emissions include moisture, shape and size, porosity (density) and the tendency to be fragmented during combustion (Mitchell et al., 2016). Water content in fuel decreases the combustion temperature and thereby may lead to incomplete combustion. The shape and size of the fuel affects its total surface area, which has an important effect on the pyrolysis rate (Mehrabian et al., 2012).

Thus, logs burn faster when cut into smaller pieces.

Several chemical attributes of fuel affect particulate matter emissions. Mitchell et al.

(2016) found that in batch combustion of woody biomass fuels, the amount of volatile content in the fuel correlates to the total PM emission. Other biomass fuels, such as agri- cultural residues and wood bark, may have much higher ash contents, which has been found to increase particulate emissions (Sippula et al., 2007). Both Feldmeier et al. (2019) and Lamberg et al. (2013) found that in automatic combustion appliances, the fuel potassium content correlates with the amount of particulate emissions.

2.1.5 Combustion appliances and operational practices

Combustion appliances with poorly optimized combustion technologies and uncontrolled combustion processes can lead to high emissions of incomplete combustion. In older, manually operated small-scale combustion appliances, there is only a primary air supply,

(14)

14

which usually directs combustion from under the grate. This can easily result in a high combustion rate and poor mixing in the proximity of the flames, which can generate a lack of oxygen concentration locally (van Loo and Koppejan, 2008). Overall, RBC appliances operate with excess air supply, but together with poor mixing, the excess air causes a rapid decrease in temperature after pyrolysis. The low temperature and local deficiency of oxygen lead to incomplete combustion, and the pyrolyzed gases may “escape” the flames.

Modern RBC appliances have secondary air supplies, which improve mixing inside the combustion zone. The secondary air supply is usually implemented using air channels that provide air on top of the fuel in the combustion chamber. This way, the purpose of the primary air below the grate is to regulate the combustion rate, while the secondary (and possibly tertiary) air enhances the oxidation of the pyrolyzed gases.

Small-scale devices are typically operated with fuel batches or with an automated fuel supply. Batch-wise combustion is problematic emission-wise, because whenever a new batch is added, there is a large amount of material to pyrolyze and the highest temperature, i.e. the embers, is below the fuel batch. Therefore, especially without active control of the combustion air supply, much of the organic gas escapes unoxidized for a few minutes at the beginning of the batch (Kortelainen et al., 2018). Furthermore, in the char burning phase the oxidation happens in the surface of the char, which generates high CO emissions (Tissari et al., 2008). Pellet and wood chip boilers usually have an automated fuel supply and air supply, and they are able to keep emission levels low for the whole combustion cycle (Ozgen et al., 2014).

The operational practices of a small-scale appliance plays a major role in its emissions (Schmidl et al., 2018). Fuel selection, air settings, ignition technique and recharging fre- quency are all dependent on user behavior, especially in batch-wise combustion. The so- called primary measures for emission reduction are based on optimizing the combustion process. The optimization is ineffective if, for example, low-quality fuel is used. With modern combustion appliances, the manufacturer provides user guides which should be followed strictly to keep the emissions low.

2.2 After-treatment technologies for small-scale biomass combus- tion

This section summarizes the most common technologies that are applied in large-scale combustion plants or other industries. All of these technologies could be applied to small- scale combustion, but whether it is reasonable, e.g. considering their size and economic feasibility, is arguable. Currently, there are a few different types of emission after- treatment technologies under development for small-scale biomass combustion, although they are mostly not yet commonly used.

(15)

15

A conclusion on the introduced after-treatment technologies is presented in Table 1. It should be noted that after-treatment technologies are the secondary measure in emission reduction. Primary measures, i.e. optimizing the combustion process, are necessary so that the secondary measures can be designed to be small in size and therefore reasonable in cost (Nussbaumer, 2003). When emissions are brought to low levels with primary measures, then secondary measures can be utilized to further decrease the emission levels.

Table 1. Examples of particulate reduction efficiencies of different after-treatment technologies (Miller, 2015, sec. 3.4)

Particle size (μm)

Control device <1 1-3 3-10 >10

High efficiency ESP

96.5 98.25 99.1 99.5

Small-scale ESP

80*

Fabric Filter 100 99.75 >99.95 >99.95

Venturi Scrub- ber

>70 99.5 >99.8 >99.8

Multicyclone 11 54 85 95

* (Jaworek et al., 2020)

2.2.1 Electrostatic methods 2.2.1.1 Electrostatic precipitator

The electrostatic precipitator (ESP) is a common device in large-scale combustion plants.

The technology was invented as early as in the 19th century and commercialized at the beginning of the 20th century. The principal idea in electrostatic precipitation is to (1) give particles an electrical charge, (2) collect the charged particles with an electric field. Charg- ing and collecting can be two separate stages of the device or can be implemented in the same physical space of the device (Figure 1). The collection plates are cleaned periodical- ly by rapping the plates. Another cleaning solution is a liquid spray, which is used in wet ESPs.

(16)

16

Figure 1. Example of a single-stage electrostatic precipitator. Credit: Evan Mason, CC BY-SA 3.0 via Wikimedia Commons.

Charging of the flue gas particles is achieved with a corona discharge. A corona discharge occurs when a high voltage is conducted to a sharp object, i.e. a needle or a thin wire, forming a dense electric field close to the object. The electric field is so strong that it exceeds the dielectric strength of the surrounding gas in a small area, allowing electrons to emit from the conductor. These electrons start a chain reaction with the surrounding gas;

they collide with atoms of the surrounding gas, ionizing them and thus causing more free electrons to emit. Gradually, an electron avalanche is formed. Depending on the polarity of the corona discharge voltage, which can be either positive or negative, the electrons are launched either away from or toward the electrode, respectively. To simplify, a positive corona discharge emits positive ions that are leftovers from ionization; a negative corona discharge produces negative ions when the electrons collide with atoms or molecules of the flue gas. These ions finally collide with flue gas particles, making them available for electrostatic precipitation (Aro et al., 2003, sec. 3.1). Thus, the charge distribution produced by a corona charger is unipolar, contrary to the bipolar distribution of natural charges discussed in section 2.1.3.

Electrostatic precipitation is the most effective for particles larger than approximately 0.5 μm. For ultrafine particles (0.01-0.1 μm), the collection efficiency is lower in conventional ESP designs (Zukeran et al., 1999). ESP can be designed to be effective for all particle sizes, but the size of the device has to be increased. Furthermore, the corona discharge electrode design has an important role in the efficiency. The charging efficiency can be up to 90-100% for 100 nm particles in proper design (Chen et al., 2018). ESPs can handle

(17)

17

large gas volumes, maintain a low pressure drop and operate in temperature ranges up to 650 °C (Miller, 2015, sec. 3.4.2). ESPs are not expected to affect gaseous emissions. The problem for ESPs is re-entrainment of particles, which is usually solved by chaining multiple ESPs. Another option to solve re-entrainment is to use wet ESP, where the water film on collection electrodes adheres to the particles more effectively. Furthermore, the resistivity of the flue gas particles plays a key role in ESP operation. Electrically conductive particles are prone to re-entrain, as they lose their charge quickly to the collecting electrode. Electrically resistive ash particles may create a problem of back corona (Ni et al., 2015), where the charged particles form an opposing electric field on the collecting electrode, as the resistive particles maintain the electric charge when accumulated on top of each other. The optimal resistivity for dust particles in ESP operation is 10⁴ to 10¹⁰ohm- cm (Miller, 2015, sec. 3.4.2).

ESP can be downscaled to suit small-scale biomass combustion. A traditional design (Fig- ure 1) can be downscaled and inserted between the combustion appliance and chimney.

Another option is to use a tubular-type ESP, where the collection electrode is a tube and in the center of the tube is a corona wire. The tubular design can be inserted into the flue gas channels or on top of the chimney. Both types of ESP have been commercialized (Carola Clean Air, 2021; Exodraft, 2021; Kutzner-Weber, 2021; OekoSolve, 2021; Schräder, 2021). The problem in down-sizing is mostly the cost of the device. A high-voltage power supply is needed to generate the corona discharge and an efficient collecting electric field, which alone creates a significant cost related to the cheapest combustion appliances. To bring the expenses down, the small-scale ESP designs may lack a cleaning system for the accumulated dust on the collecting surfaces. Simplified designs are prone to fouling on top of the high-voltage electrode (Migliavacca et al., 2014; Patiño et al., 2016). Due to the electrically conductive nature of soot particles, this fouling can create an electrical short circuit inside the ESP. Small-scale ESPs operating in real conditions, not in laboratory conditions, reach an approximately 80% reduction efficiency (Jaworek et al., 2020). How- ever, long-term operation of these devices has been found to decrease in efficiency (Obernberger et al., 2012).

2.2.1.2 Sonic jet charger

A sonic jet charger is a variant of the charger inside an ESP. Traditional ESPs, especially single-stage ESPs, are optimized to remove large particles. When the corona discharge is physically in the same space as the collection electrodes, the electric field is strong every- where in that area and the particle charging is dominated by this so-called field charging process. Field charging is effective for charging large particles (> 1 μm), but less effective for ultrafine particles. (Hinds, 1999). Ultrafine particles require large amounts of free ions and long residence times to be effectively charged, which is achieved with diffusion charging. A field charging process is not necessary if there are no large particles in the flue gas.

(18)

18

The corona chargers of traditional ESPs are located in the flue gas and are therefore prone to the corrosive effects of aggressive chemical components. The chemical composition, temperature and particle load of the flue gas can change the corona characteristics and even prevent stable corona formation (Oglesby and Nichols, 1978; Yan et al., 2016). Soot particles are electrically conductive, and therefore fouling of the corona electrode and its insulation surface can create an electrical short circuit from the electrode to the body of the ESP, which disables the corona discharge (Migliavacca et al., 2014; Patiño et al., 2016).

A sonic jet charger is designed to produce a large number of ions in a separate chamber and use pressurized air to carry the ions to the flue gas (Laitinen and Keskinen, 2016).

Therefore, the corona discharge avoids the influence of the flue gas and operates in a clean gas flow. Charging of particles is based on diffusion, as the electric field generated by the high-voltage electrode is situated in the separate chamber. Sonic jet chargers are not widely used, but the advantages compared to traditional ESP make this technology interesting when considering small-scale biomass combustion.

2.2.2 Fabric filters

Fabric filters have become very common in industrial applications since the 1970s to replace electrostatic precipitators in situations where the fly ash is not suitable for an ESP (Miller, 2015, sec. 3.4.3). The filters collect particles into filter bags. An industrial fabric filter consists of several, even thousands of, filter bags. Filter materials are woven fabrics or felts of nylon, PTFE etc., for low-temperature solutions (Sanz et al., 2015). For higher temperatures, glass fibers or metallic materials can be used (Spaite et al., 1961).

The advantages of fabric filters are their ability to handle large flue gas volumes, a very high collection efficiency for all particle sizes and the lower cost than an ESP of similar requirements for collection efficiency. The disadvantages are their large size to prevent pressure drop and inability to withstand high temperature changes because of condensation (Brandelet et al., 2020). Furthermore, there is a possibility of explosion or fire if sparks are present in the vicinity of a baghouse. Maintenance costs for fabric filters are high because the filter materials have to be replaced regularly. Operation of a fabric filter includes regular cleaning of the filter material. As particles are collected inside the filter, the pressure drop increases and it affects the performance of the filter. Cleaning is performed typically by shaking, with a reverse flow or a with an air pulse (Miller, 2015, sec. 3.4.3).

Fabric filters are not considered a viable option as an after-treatment technology for small- scale biomass combustion. Small-scale combustion appliances are usually not operated continuously but ignited and turned off even many times a day. This creates regular condensation issues for the fabric filter (Brandelet et al., 2020). Furthermore, the pressure drop of a fabric filter is problematic for small-scale appliances.

(19)

19

2.2.3 Wet scrubbers and condensing heat exchangers

Wet scrubbers include various designs depending on the intended use. The most common wet scrubbers are designed for desulfurization and are used in processes that emit sulfur oxides, such as coal combustion and waste incineration. The operating principle of a desulfurization wet scrubber is to scrub a liquid that contains a reagent, usually limestone, that reacts with sulfur oxide, and the resulting sludge waste can disposed of, or in more advanced systems, recycled. Another use for wet scrubbers is to recover heat by condensation of water vapor from flue gases (Chen et al., 2012; Mussatti and Hemmer, 2002).

For particle removal, wet scrubbing is based on liquid droplets intercepting the particles from the flue gas flow. Particles are attached to liquid droplets by thermophoresis (Pilat and Prem, 1977), inertial impaction, gravitational settling, Brownian diffusion, diffusiophoresis, electrophoresis and thermophoresis (Miller, 2015, sec. 3.4.4). Collection is highly efficient for particles larger than 1 µm, but the efficiency decreases rapidly for smaller particles (Lee et al., 2013). Thus, a wet scrubber is an efficient and relatively low- cost particulate removal option for processes that produce particles larger than 1 µm. The downsides of wet scrubbers are high pressure drop and the wastewater that, depending on the source of the particles, needs to be processed.

In venturi scrubbers, a high flow velocity is used to collide flue gas particles into liquid droplets. A venturi scrubber can separate smaller particles than a traditional wet scrubber (Miller, 2015, sec. 3.4.4).

A condensing heat exchanger (CHX) is another type of wet scrubber. A CHX cools the flue gas below the dew point, making the flue gas moisture condense on the heat exchange surfaces. A CHX can be optimized to capture fine particles (Grigonyte et al., 2014; Li et al., 2020), when the CHX consists of a battery of small tubes. In this case, particles are separated by diffusiophoresis and thermophoresis. Furthermore, particles are attached to the condenser surfaces efficiently due to the sticky properties of the wet surface (Kleinhans et al., 2017).

Wet scrubbers and condensing heat exchangers are a viable PM reduction option for small and medium-scale biomass boilers. A CHX can be used to replace a traditional heat exchanger in a boiler, providing better heat recovering efficiency together with PM reduction. Wet scrubbers are likely excessively large for stoves and fireplaces, and the necessity of liquid input and wastewater output makes the technology overwhelmingly complex compared to the simplicity of the combustion appliance.

2.2.4 Cyclone separator

Cyclone separators are based on centrifugal force, which separates particles from the flue gas flow. The centrifugal force is achieved with a spinning flow that is created by the me- chanical design of the device (Figure 3). Cyclones are usually used to separate coarse par-

(20)

20

ticles (> 10 µm), as pursuing lower particle sizes would create an overwhelming pressure drop in the system (Miller, 2015, sec. 3.4.1). The separating efficiency of a cyclone is often reported as a d50 value, i.e. the particle size that is separated with 50% efficiency.

Smaller particles are separated with decreasing, larger particles with increasing efficiency (Shepherd and Lapple, 1940). Multicyclones, i.e. devices containing multiple cyclone chambers, can be used to increase collection efficiency.

Cyclones are very robust and reliable devices. They are often used as a pre-separator for other after-treatment technologies. As a standalone device, the cyclone lacks efficiency for the required particle sizes in RBC appliances. The pressure drop is also an important factor to consider in the smallest combustion appliances.

Figure 2. Schematic of a cyclone separator. Credit: Cburnett, CC BY-SA 3.0 via Wikimedia Commons

2.2.5 Catalytic methods

Catalysis is a process where the rate of a chemical reaction is increased by adding a sub- stance known as a catalyst. The catalyst is not consumed in the process and the process is continuous. The catalytic process has a long history and is very widely used in many industries, for example manufacturing of margarine (hydrogenation with a nickel catalyst), gasoline (catalytic cracking) and catalytic converters in combustion engines. Over 90% of the world's manufactured chemicals involve catalysis at one or more stages (Mile, 2000).

(21)

21

Catalysis is based on lowering the activation energy for a chemical process. In wood combustion, the oxidation process, i.e. combustion, happens naturally when there is fuel, oxygen available and the temperature is high enough. However, often the requirement of oxygen availability at a high enough temperature to completely oxidize fuel molecules is not met in small-scale biomass combustion appliances, leading to emission of incomplete combustion. In the presence of the appropriate catalyst, the oxidation process can happen at a much lower temperature because the activation energy is lowered.

Catalytic combustors have been in use in North America for decades (Hearth, 2013).

Commercial catalytic combustors are usually platinum/palladium coated honeycombs that are integrated or retrofitted to the stove. They are placed close to the combustion chamber so that the temperature of the flue gases is high enough for the catalyst to work. There is a bypass for the flue gases that is used during the ignition phase when the temperature is not high enough (Figure 3). For ceramic honeycombs, the catalyst light-off temperature is 260

℃ and for steel honeycombs 204 ℃ (Condar, 2021). Other types of catalysts have been tested in scientific studies (Bensaid et al., 2012; Hukkanen et al., 2012; Nevalainen et al., 2019; Ozil et al., 2009; Pieber et al., 2018; Reichert et al., 2018)

Figure 3. Example of a catalytic stove. Modified from Condar (2021).

The problems with a catalytic element in a wood stove are catalyst deactivation and efficiency. The element should be designed so that it does not introduce excessive pressure loss to the flue gas flow. Otherwise, the element will negatively affect the combustion process and any benefits from the catalytic reactions will be minor compared to the emissions from the combustion chamber. Furthermore, the element should have a high surface area and flue gases sufficient residence time so that the catalytic reactions can occur. To reduce particle emissions, the element should be dense, so that particle capture can happen.

(22)

22

Catalyst deactivation means that the catalyst loses its activity or selectivity over time. De- activation can happen chemically by poisoning, physically by fouling, and thermally by sintering. Deactivation is a major challenge in RBC (Bindig et al., 2012; Carnö et al., 1996) and should be studied in more detail. On a larger scale, for example catalysts for reduction of NOx emissions in biomass combustion power plants with selective catalytic reduction, where NOx is reacted with ammonia to generate nitrogen and water, the domi- nant deactivation mechanism is poisoning due to impurities in the biomass (arsenic, phos- phorous, zinc and alkali metals) and fouling by ash plugging (Larsson et al., 2007). For another commercial catalyst material, in woody biomass combustion, the main catalyst deactivation mechanism is poisoning by potassium (Zheng et al., 2005).

The typical catalyst support material is a monolith made of ceramic or steel. Catalytic combustors used in North America resemble honeycombs, which is one type of monolith.

Monoliths used in vehicle emission control are much denser, containing thousands of par- allel channels or holes, with thin walls between. In small-scale wood combustion, the support structure should be designed free of significant pressure drop, or an exhaust fan is necessary.

2.3 Emission limits of small and medium biomass combustion in Europe

In Europe, several national and international emission limits are either already in force or coming into force in the coming years. This section briefly reviews the most important regulations, focusing on the most recent changes and the particulate emission aspect of those regulations (Figure 4). The European Union has planned emission regulations for different small-scale biomass appliances based on their heat output levels as part of the Ecodesign directive. Currently for residential small-scale combustion, for example, Ger- many, Austria, and Switzerland have the most stringent emission limits in their national regulations (EEA, 2019).

The new medium combustion plants (MCP) directive, which includes boilers in the size range of 1-50 MWth, has been tightening PM emission limits of particularly the smallest boiler units. New boiler installations with 1-5 MWth have had to reach PM emissions of 50 mg/m³ since 2018, whereas existing plants in the same size range have a limit of 300 mg/m³ until 2030 (European Union, 2015b).

It is important to acknowledge the measurement standards used with different regulations, because, for example, batch-wise combustion can generate a wide range of emissions during different stages of combustion, which are all not included in the measurement phase.

The Ecodesign directive will regulate the emissions of manually fired solid-fuel domestic combustion appliances starting at the beginning of 2022 (European Union, 2015a). The directive sets limits based on three different test methods. In the first option, the measurement is carried out while the stove is providing its nominal output (and if appropriate part-

(23)

23

load) and the PM is sampled for 30 min without dilution on a heated filter. The second option includes measurement over the full burn cycle, in which the PM is collected from diluted gas on a filter at ambient temperature, which most resembles the method used in this study. The third option involves a measurement sampled over a 30-minute period and PM collection of diluted PM on the filter at ambient temperature or PM collection by an electrostatic precipitator. The emission limits given by the Ecodesign directive are 40 mg/m³ at 13% flue gas excess oxygen, 5 g/kg and 2.4 g/kg, for methods 1, 2 and 3, respectively.

Figure 4. Particulate matter (PM) emission limits for different combustion plant size ranges in Europe according to the Ecodesign directive.

(24)

24

3 Methods

The studies in the thesis included test measurements with real-life combustion appliances, and the studied reduction methods were operated in situations resembling realistic operation during real-life usage. The studies for Papers I and II were performed at the IL- MARI facility of the University of Eastern Finland. The studies for Paper III were ac- complished at the SIMO measurement container in University of Eastern Finland. For Pa- per IV, the measurements were performed at a Fortum industrial heating plant. All the studies were performed with a similar measurement setup including both online and offline measurement devices. Typically, several online measurement devices were used to deduce the particulate mass and number concentrations in the changing situations when the emission reduction method was turned on and off. Based on the concentrations, a reduction efficiency was calculated for the device. Offline filter collections were used in similar fashion, but longer sampling times were used and the filters were later analyzed to determine the organic and elemental carbon contents of the samples.

Electrical emission reduction methods were designed as part of this study for Papers I, II and III. For Paper IV, performance of a commercial industrial electrostatic precipitator was evaluated for comparison. Furthermore, the condensing heat exchanger in Papers I and II was developed in earlier studies by the FINE group at the University of Eastern Finland.

3.1 Combustion appliances

Combustion appliances of three different size ranges were used in this study. For Papers I and II, a 40 kW stoker burner equipped with a reciprocating step-grate (Ariterm MultiJet) was fired with wood chips. The nominal heat output of 40 kW is suitable for heating a large detached house. The fuel feed rate, combustion air feed for primary and secondary air, and pressure conditions were logic controlled. A flue gas fan was used for pressure control. The combustion appliance consisted of flue gas ducting in which the emission reduction devices could be installed. The setup and operational details are described in Paper I, II and by Leskinen et al. (2014). Three different heat exchanger setups were used with the grate burner; (1) a traditional boiler tube heat exchanger, (2) a condensing heat exchanger (CHX) and (3) stainless steel flue gas channels followed by a cyclone, which is not actually a heat exchanger but in this case the thermally uninsulated flue gas channels acted as a heat disperser to acquire a suitable operation temperature for the electrical charging and cyclone separation of particles. Different setups were used to evaluate their applicability as a collection surface for electrically charged flue gas particles.

The smallest appliance in this study was a log wood fired masonry heater in Paper III.

The heater is a modern prototype model with a secondary air supply. A mounting hole for the emission reduction method was drilled through the walls of the heater. The combustion procedure and sizes of the batches are described in Paper III.

(25)

25

Paper IV presents an industrial scale bio-oil operated boiler. Pyrolysis oil was used as fuel in a 49 MW district heating plant located in Espoo, Finland. The plant was equipped with a 2-field electrostatic precipitator (Ahlstom ELPAC 2.1).

3.2 Shielded corona charger

The shielded corona charger (SCC) is a modification of a sonic jet charger described in section 2.2.1.2. It includes a corona charger that is protected by a ceramic cover, and a shield air flow that is purged through the ceramic cover to further protect the corona discharge from the effects of the flue gas flow. The difference to a sonic jet charger is that the shield air flow is not sonic because larger geometries, and in some cases, multiple corona needles, are used. In this case, achieving a near-sonic-speed purge gas flow would require either a small orifice or an excessive amount of pressurized air to produce the shield air flow. The SCC was developed by Tassu ESP (Mikkeli, Finland) and the design was further optimized for the applications in this study.

Paper I describes an SCC design that has multiple corona needles along opposing sides of a round ceramic tube. A total of 12 needles peep through the ceramic cover and a small gap for the shield air surrounds each needle. A condensing heat exchanger was used as a collection surface and the SCC was installed upflow of the CHX into the ducting of the combustion appliance. A high voltage of 20 kV was fed to the high-voltage electrode leading to the corona needles.

For Paper II, the design of the SCC (Figure 5) was further optimized to aim the charged particles more effectively into the desired collection surfaces. This was achieved with a shorter electrode and a single corona needle. The shorter design creates a smaller area of effective electric field around the electrode itself, which decreases the electrical particle deposition on the surrounding surfaces. A single corona needle requires lesser sheath air flow to maintain operation. The SCC was installed in a new charging chamber that was designed to support its operation prior to the CHX. Furthermore, the SCC was operated in high temperatures close to the burner to study the effect of temperature on corona discharge. A high voltage of 20-40 kV was used in the experiments, depending on the operating placement and surrounding temperature of the SCC.

(26)

26

Figure 5. Design of the shielded corona charger (SCC) (Paper I).

3.3 High-temperature electric soot collector

The high-temperature electric soot collector (HiTESC) was designed as part of this study to create a simple and maintenance-free alternative to traditional electrostatic precipitators.

Unlike ESPs, the HiTESC does not require particle charging. Instead, the natural electric charge, described in section 2.1.3, generated in combustion is utilized in particle collection.

The design of the HiTESC includes a high-voltage electrode covered by a closed dielectric ceramic (Figure 6). The high-voltage electrode generates an electric field that traps the naturally charged particles from the flue gas flow. Particles with the same polarity as the electric field are pushed towards the walls of the fire chamber and particles with the oppo- site polarity are attracted towards the electrode. Trapped particles are oxidized under the influence of the high temperature. The ceramic cover prevents electric arcing and contam- ination of the HiTESC electrode. The device was installed into the fire chamber of the masonry heater approximately 50 cm above the grate.

HiTESC could be considered as a primary measure for emission reduction, as it operates near the flames and may also have an effect on the combustion process, although in this study no other significant changes were observed than the reduced PM concentrations.

Therefore, in this study, HiTESC is considered as a secondary measure to simplify its comparison to the other methods.

(27)

27

Figure 6. Design of the high-temperature electric soot collector (HiTESC) (A) and photograph of the device installed into a fireplace (B) (Paper III).

3.4 Measurement practices and methods

For all papers in this study, the measurement setup consisted of gaseous emission measurement instruments, one or two filter sampling systems, and several online aerosol measurement devices. A combustion appliance was operated in real-life conditions to produce typical emissions for each combination of fuel and combustion appliance. The studied emission control device was installed into the flue gas in its desired position. The performance of the emission control device was evaluated by measuring emissions either by moving the measurement setup upflow and downflow (Paper IV) of the control device or by switching the control device on/off and keeping the measuring setup in the same place (Papers I, II and III). Furthermore, for Paper II, some measurements were performed from between the two emission control devices; the shielded corona charger (SCC) and condensing heat exchanger (CHX). Measuring upflow and downflow simultaneously would be the most accurate way of evaluating the efficiency of a control device, but would require two sets of measuring devices. Furthermore, sampling from very hot flue gas is not directly proportional to its cold equivalent, due to the behavior of condensing vapors (Sippula et al., 2012).

Gaseous emissions were measured using a Fourier Transform Infrared spectroscopy (FTIR) gas analyzer (Gasmet DX-400, Vantaa, Finland) from undiluted flue gas through a heated sampling probe. Data for gaseous emissions was primarily used for ensuring that the combustion process was comparable between different measurement days and setups.

The studied emission reduction methods presumably do not affect the gaseous emissions.

Filter samplings were performed primarily using a PM1 filter holder on diluted flue gas.

Samples were analyzed for gravimetric (PTFE filters) and organic and elemental carbon (quartz fiber filters) content using a thermal-optical carbon analyzer (Sunset Laboratory Inc.). The analyses were performed using the NIOSH5040 protocol (NIOSH, 1999). For Paper II a total suspended particle (TSP) fraction was also collected using isokinetic sampling of undiluted flue gas in a heated quartz fiber filter.

(28)

28

Particulate emissions and their size distributions were measured online from diluted flue gas with an electrical low-pressure impactor (ELPI, Dekati, Kangasala, Finland) and scanning mobility particle sizer (SMPS, TSI Inc., Minneapolis, MN, USA) (CPC model 3776, DMA model 3081, and model 3085). ELPI and SMPS results were also used to determine emissions of fine particle mass (PM) and number (PN). Furthermore, the black carbon (BC) mass concentration was measured using an Aethalometer (AE33-7, Magee Scien- tific) analyzer.

(29)

29

4 Results and discussion

4.1 Emission factors of the studied combustion appliances

The studied combustion appliances were a masonry stove for burning wood logs, an automatic grate burner fired with wood chips, and an industrial boiler operated with pyrolysis bio-oil and equipped with an ESP. The appliances present relatively low, average, and good emission levels, respectively. As seen in Table 2, the normalized emission concentrations from the masonry heater are many times higher than from the other devices. The carbon monoxide emissions of the masonry heater were measured in thousands of mg/m³, and the PM1 mass concentration is over 100 mg/m³ and consists mostly of elemental carbon, i.e. soot. The emission factors of the automatic grate burner and pyrolysis oil boiler are closer to each other. The automatic grate burner emits far more carbon monoxide than the industrial-size pyrolysis oil boiler (60 mg/m³ vs 8 mg/m³), but the PM1 emissions are on the same level (46 mg/m³ and 38 mg/m³). The particle number concentration of the pyrolysis oil boiler emissions exceeds that of the other two appliances, and the geometric mean diameter (GMD) of the particles is only 47 nm. Thus, the pyrolysis oil boiler emits a higher number concentration of particles, but they are so small that the total mass of the particles is still lower than in the other two appliances.

Table 2. Emission factors of the studied combustion appliances. Standard deviations of the averaged results are given in brackets (Papers I, III and IV).

Unit Masonry heater^a Grate burner^b Pyrolysis oil^c CO mg/m³ 2095 (±82) 60.4 (±24.6) 8 (±3)

NELPI 1/cm³ 4.2E+07 (±3.9E+06) 2.4E+07 (±3.9E+06) 1.7E+08 (±7.9E+02) GMDELPI nm 86 (±20) 93.5 (±4.8) 47 (±1)

PM1 mg/m³ 133 (±35) 46.3 (±12.3) 38 (±6.0) OC mg/m³ 19 (±11) 3.3 (±4.5) 0.29 (±0.25)

EC mg/m³ 91 (±23) 4.1 (±4.3) below detection limit

a Concentrations are normalized to 13% flue gas O2 and dry gas at NTP. Results are averaged over batches 2 and 3.

b Concentrations are normalized to 10% flue gas O2 and dry gas at NTP. Results are averaged over a stabilized operation period of the burner.

c Concentrations are normalized to 3% flue gas O2 and dry gas at NTP.

The studied combustion appliances represent typical emission values for each type of device. For example, the PM emissions of manually fired stoves and fireplaces have previ- ously been reported to be in the range of 45-220 mg/m³ (Nuutinen et al., 2014; Ozgen et al., 2014; Rönnbäck et al., 2016; Savolahti et al., 2016). The PM emissions of automatical- ly fired small-scale boilers are typically in the range of 10-46 mg/m³ (Lamberg et al.,

(30)

30

2013; Schmidl et al., 2011), thus the emissions of the combustion appliance used in this study were in the upper range of these devices. Pyrolysis oil PM emissions range from 50 to 550 mg/m³ (Feng et al., 2017; Oasmaa et al., 2015; Tzanetakis et al., 2011; Zadmajid et al., 2017).

4.2 Performance of the studied after-treatment technologies

4.2.1 Effects on particle mass emissions

The particle emissions of the studied after-treatment technologies are presented in Figure 7. The high-temperature electric soot collector (HiTESC) operated in a masonry heater reduced on average 45% of the PM1 concentration. The shielded corona charger (SCC) with a condensing heat exchanger (CHX), a reference heat exchanger (REF) or a cyclone, operated in an automatic grate burner setup, reduced an average of 93%, 88% and 27% of the PM1 concentration, respectively. The industrial electrostatic precipitator (ESP) operated in a pyrolysis oil power plant reached a reduction efficiency of 96%.

Figure 7. Particulate matter emissions of the studied combustion appliances and the effect of after- treatment on those emissions. HiTESC = high-temperature electric soot collector, SCC = shielded corona charger, CHX = condensing heat exchanger, REF = reference heat exchanger, ESP = electrostatic precipitator (Papers I, II, III and IV).

The HiTESC differs from the other studied methods because the reduction efficiency is not as stable, but depends on the combustion phase of the combustion appliance. Utiliza- tion of the phenomena of natural charges requires sufficient temperature and distance from the flames to be effective. In batch-wise combustion, there are phases when the natural charge density is low in the proximity of the soot collector, and therefore the reduction efficiency is low. In an automated combustion appliance, where the combustion process is kept stable, HiTESC should also have a stable reduction efficiency.

The effectiveness of the SCC was also found to be dependent on the surrounding temperature, but for different reasons than with the HiTESC. The SCC had a problem with soot

(31)

31

deposits on the surface of its insulation cover, which decreased the electric breakdown threshold by creating an electrically conductive channel from the corona needle to the walls of the flue gas channel. Therefore, the operating voltage of the SCC had to be reduced, which led to decreasing ion formation and weaker reduction efficiency. It is possible to avoid the problem of soot deposits by operating the SCC above soot oxidation temperature, above 500 °C (Nevalainen et al., 2019). On the other hand, at higher temperatures, the corona discharge is known to become more unstable due to increased ion mobility (Czech et al., 2012), thermal ionization (Fialkov, 1997), and the presence of naturally charged particles (Kim et al., 2005; Sgro et al., 2010). Contrary to the HiTESC applica- tion, with the SCC the natural charge is an unwanted phenomenon, because the electric breakdown threshold through the flue gas is weakened, i.e., the voltage range for stable corona operation is narrow. As a result, the SCC was found to be most effective directly above the soot oxidation temperature.

Further experiments were conducted to measure the effect of the SCC on particle concentrations before the CHX, thus the effect of the CHX was not interfering with the results (Table 3) and the effects of the SCC sheath air can be directly measured. It was found that sheath air protects the corona needle and decreases the temperature around it, enabling its operation in otherwise difficult circumstances. Below soot oxidation temperature (<500

°C), the SCC had only a 50% reduction efficiency with a very high deviation, because the voltage was not stable. By applying sheath air for 20 dm³/min, the corona voltage was stable and reduction efficiency 76%. Applying more sheath air did not further improve the reduction efficiency by any great margin. In high temperatures at 800 °C, the corona was unstable without sheath air and with 20 dm³/min. A sheath air flow of 60 dm³/min was enough to cool down the surrounding temperature, and the operating voltage was stable at a high voltage of 41 kV and the reduction efficiency was 87%. This indicates that the operation of the SCC can be adapted to varying temperature conditions, which may occur e.g. due to varying boiler load, by adjusting the sheath air flow rate. On the other hand, high sheath air flow rates are not preferred, because this decreases the thermal efficiency of the combustion appliance.

Table 3. Shielded corona charger (SCC)-induced reduction efficiencies of particulate matter emissions measured upstream of the heat exchanger in different operating conditions (Paper II).

400 – 500 °C 800 °C No sheath 50 ± 42% @ 40 kV 49% @ 26 kV 20 dm³/min sheath 76% @ 45 kV 46% @ 27 kV 60 dm³/min sheath 78 ± 4% @ 43 kV 87% @ 41 kV

(32)

32 4.2.2 Effects on particle number emissions

The particle number emissions of the studied combustion appliance with and without the respective after-treatment are presented in Table 4. Particles were removed by 31%, 92%

and over 99% by the HiTESC, SCC-CHX combination and industrial ESP, respectively.

For the masonry heater, particle number concentration was 2.9 × 10⁷ even after the applied reduction technology. The pyrolysis oil boiler had very high particle number concentration before the ESP, but after the ESP the count was in the hundreds.

Table 4. Particle number emissions of the studied combustion appliance with and without the respective after-treatment (Papers II, III and IV).

Unit Masonry heater + HiTESC

Grate burner + SCC-CHX

Pyrolysis oil + ESP

Filter OFF 1/cm³ 4.20E+07 (±3.9E+06) 8.79E+06 (±1.0E+06) 1.70E+08 (±7.9E+02) Filter ON 1/cm³ 2.90E+07 (±4.7E+06) 7.22E+05 (±1.6E+05) 310 (±46)

Reduction

Efficiency % 31 92 >99

The previous section discussed that the SCC was made operational at high temperatures by applying sheath air, which decreases the surrounding temperature of the corona charger.

Table 5 shows the reduction efficiencies in the same situations as in Table 3 but calculated based on particle number emissions. At a temperature range of 400-500 °C, the reduction efficiency is on a similar level as based on PM. In contrast, at 800 °C, even a negative reduction efficiency was calculated. The decrease in particle number reduction efficiency is explained by new particle formation; the SCC removes particles that would normally act as seed particles for vaporized alkali metals. Furthermore, applying sheath air creates local low-temperature areas which enable the condensation of these vapors, which then may lead to the formation of new particles by homogeneous nucleation.

Table 5. Shielded corona charger (SCC)-induced reduction efficiencies of particle number emissions measured upstream of the heat exchanger in different operating conditions (Paper II).

400 – 500 °C 800 °C No sheath 56 ± 31% @ 40 kV 25% @ 26 kV 20 dm³/min sheath 85% @ 45 kV -13% @ 27 kV 60 dm³/min sheath 80 ± 8% @ 43 kV 39% @ 41 kV

(33)

33

4.2.3 Effects on particle number-size distributions

Particle number-size distributions can be utilized to define the reduction efficiency of the studied reduction methods in relation to particle size. The mean diameters of the particle number emissions were 86, 93.5 and 47 nm for the masonry heater, grate burner and pyrolysis oil boiler, respectively (Table 2). After the reduction device, the respective GMDs were 73, 49 and 89. Thus, the HiTESC and SCC moved the particle number-size distribution toward a smaller size but the industrial ESP toward a larger size. Typical ESPs have a lower reduction efficiency for particle sizes of 0.2-1 μm (Hinds, 1999), which explains the increase in GMD in the latter case. The HiTESC and SCC have a lower reduction efficiency for particles below 50 nm, but the highest efficiency for particle sizes of 0.05-1 μm, as seen in Figure 8. It is also possible with the HiTESC and SCC that the smallest particles are a result of new particle formation (nucleation) after the effective collection phase in the flue gas, as both of these methods operate at high temperatures. In some cases, the SCC induced a very large nucleation mode which is discussed in the previous section and in Paper II.

Novel electrical particle emission reduction methods for small-scale biomass combustion