Effect of fuel composition and combustion conditions on phosphorus behavior during combustion of biomass

Effect of fuel composition and combustion conditions on phosphorus behavior during

combustion of biomass

Joel Falk

Energy Engineering

Department of Engineering Sciences and Mathematics Division of Energy Science

ISSN 1402-1757 ISBN 978-91-7790-234-8 (print)

ISBN 978-91-7790-235-5 (pdf) Luleå University of Technology 2018

LICENTIATE T H E S I S

Joel Falk Effect of fuel composition and combustion conditions on phosphorus behavior during combustion of biomass

Luleå University of Technology

Department of Engineering Sciences and Mathematics Division of Energy Science

Luleå University of Technology

Department of Engineering Sciences and Mathematics Division of Energy Science

Effect of fuel composition and

combustion conditions on phosphorus behavior during combustion of biomass

Joel Falk

Printed by Luleå University of Technology, Graphic Production 2018 ISSN 1402-1757

ISBN 978-91-7790-234-8 (print) ISBN 978-91-7790-235-5 (pdf) Luleå 2018

www.ltu.se

2

Abstract

Due to concerns for climate change and future supply of phosphorus fertilizer within agriculture, there has been an increased interest in the combustion of phosphorus containing waste residues and opportunity biomass fuels. Previous research has shown that during combustion, phosphorus has large impact on ash transformation reactions and may decrease or increase ash-related problems such as slag formation and bed agglomeration. This is a serious concern if new types of biomass are to be added for heat and power production. Additionally, plant studies and leaching tests of P-rich biomass ash indicate that the plant availability of phosphorus varies greatly with its association in the ash. As such, the ash transformation behavior of phosphorus is of great importance for the success of such ventures. While several studies have been made on the behavior of phosphorus during combustion, no comprehensive study has been made evaluating the effect of fuel composition and combustion conditions.

In this work, the behavior of phosphorus was determined for a wide range of fuels and

combustion conditions. More specifically, the objective was to determine (i) the effect of fuel ash composition and combustion technologies on the fate of phosphorus during combustion, (ii) investigate potential difference in the behavior of phosphorus during combustion of sewage sludge and plant based biomass and (iii) the effect of phosphorus on slag formation and bed agglomeration for the co-combustion of a wide range of plant based biomasses.

The investigation was carried out by comparing experimental data gathered from the combustion of 26 different biomass fuels or fuel blends in a bench scale bubbling fluidized bed (5 kW, 18 experiments), an underfed pellet burner (20 kW, 10 experiments) and a swirling powder burner (150 kW, 7 experiments). This included chemical characterization of bed ash, bottom ash and fly ash fractions by X-ray diffraction (XRD), scanning electron microscopy coupled with energy dispersive spectroscopy (SEM-EDS) in addition to qualitative measures of slagging- and bed agglomeration tendencies.

It was found that phosphorus, irrespective of combustion technology and fuel composition, was mainly found in bed-, and bottom ash fractions and/or coarse fly ash fractions (>1µm). Based on the crystalline phase composition of the phosphates found in bed-, bottom- and coarse fly ash samples, phosphate speciation was correlated to the molar ratio between P, Ca and Mg for all three combustion technologies. Based on these results, it would be possible to control the behavior of phosphorus during combustion and the plant availability of phosphates in biomass ash by designing fuel blends based on their fuel ash composition.

In fluidized bed combustion, it was found that for similar combustion conditions and fuel ash compositions (with respect to K, Ca and P), the speciation of phosphorus in coarse ash fractions was significantly different from experiments with plant based biomass compared to sewage sludge.

Unlike ash from plant based biomass, the crystalline phase composition of ash from sewage sludge did not change with the relative concentration of K, Ca and P in the fuel. The results suggest that the reaction pathway of phosphorus during combustion of sewage sludge is different to plant based biomass due to difference in the association of phosphorus in the fuel.

The effect of phosphorus on slag formation and bed agglomeration in biomass combustion was mainly related to the relative fuel ash concentration of K, Ca, Mg, Si and P. In fluidized bed combustion, P contributes to the formation of agglomerates through the melt induced mechanism, through complex interaction with K, Ca, Mg and Si. Similarly, in fixed bed combustion the

3

composition of slag indicated that slag formation involves the formation of P and Si rich ash melt with a varying content of K, Ca and Mg. In both cases, the severity of problems was related to the melting behavior of the (CaO,MgO)-K2O-(SiO2,P2O5) multicomponent system.

4

Acknowledgements

First and foremost, I would like to thank my principal supervisor Marcus Öhman and my co- supervisors Nils Skoglund, Damiano Varagnolo, and Dan Boström for all the help they have given me over the years.

More specifically, I would like to thank Marcus and Nils for their role in the everyday supervision in addition to discussing and reviewing my manuscripts. You are more patient and understanding than I deserve and I really appreciate working with you.

Damiano, even though our work together was cut short for external reasons, your way of working has been an inspiration for me and the time we spend coding is one of the most enjoyable periods of my PhD studies thus far. I hope you do well in your future endeavors in Norway.

I would also like to give an additional thanks to Nils Skoglund, Alejandro Grimm, and Gunnar Eriksson who contributed most of the experimental data that was used in this licentiate thesis. The years and years of experimental work you put in has given me the opportunity to widen the scope of my work far beyond that of most PhD students.

I would also like to acknowledge the financial support from the Swedish Research Council for Environment, Agricultural Sciences and Spatial Planning (Formas), project 942 - 2015-619, the Swedish Research Council, project 2016-04380 and the Swedish Energy Agency.

To all my colleagues at Energy Engineering, I really enjoy the friendly atmosphere we have at our workplace and the numerous fika/lunch discussions and activities we have had over the years.

I would also like to thank my friend, office-mate and senior ash-kid Hamid Sefidari for the help you have given me over the countless hours we have spent together these last few years.

Finally, I would like to thank Susanne and the rest of my close-family for their support during these years of study.

5

List of appended papers

This licentiate thesis is based on the following papers and manuscripts:

The author participated in planning the contents of the paper, compiled and evaluated the results in addition to a significant contribution in writing the paper. Joel Falk and Nils Skoglund contributed equally to this work.

Nils Skoglund, Joel Falk, Alejandro Grimm, Dan Boström , Marcus Öhman. Fate of phosphorus in combustion of biomass. To be submitted to Nature Energy

The author participated in planning the paper, compiled and evaluated the results and wrote the manuscript.

Joel Falk, Nils Skoglund, Alejandro Grimm, Dan Boström, Marcus Öhman. Difference in phosphate speciation between sewage sludge and biomass ash from fluidized bed combustion. In conference proceedings, 27^th Impacts of fuel quality on power production and the environment;

Lake Louise, Alberta, Canada, 24^th - 28^th of September, 2018

The author participated in planning the paper, compiled and evaluated the results and wrote the manuscript.

Joel Falk, Marcus Öhman, Alejandro Grimm, Dan Boström, Nils Skoglund. Effect of phosphorus on slag formation and bed agglomeration in combustion of biomass. Manuscript

6

Contents

Abstract ... 2

Acknowledgements ... 4

List of appended papers ... 5

1. Introduction ... 7

1.1 Current and future biomass use in Sweden ... 7

1.2 Biomass and its main ash forming elements... 9

1.3 Measures for prevention of ash related problems and phosphorus recovery ... 9

1.3.1 Bed agglomeration and slag formation ... 10

1.3.2 Plant availability of phosphorus in ash ... 11

1.4 Objective of thesis ... 12

2. Method ... 13

2.1 Fuels ... 13

2.2 Experimental procedure and sampling ... 16

2.3 Experimental design... 18

2.3.1 Paper 1 ... 18

2.3.2 Paper 2 ... 19

2.3.3 Paper 3 ... 19

2.4 Derived results and fuel molar ratios ... 19

3. Results and discussion ... 20

3.1 The effect of fuel ash composition and combustion technologies on the fate of phosphorus in combustion of biomass (Paper 1) ... 20

3.2 The effect of fuel association on the fate of phosphorus during combustion of biomass (paper 1, 2) 24 3.3 The effect of phosphorus on slag formation and bed agglomeration in combustion of biomass (paper 3) ... 28

4. Conclusion ... 32

5. Future work ... 34

6. References ... 34

7

1. Introduction

There is a general scientific consensus that the global temperature has been rising since the late 1800s as a result of anthropogenic greenhouse gas emissions such as CO2. It is also understood that this increase in temperature may have serious impacts on the global climate¹. An international environmental treaty (UNFCCC) was adopted in 1992 among the United Nations in an effort to stabilize the greenhouse gas concentration in the atmosphere. In 1997, it was extended in the Kyoto protocol which committed its members to reduce their greenhouse gas emissions until 2012. Eventually, this lead to the Paris agreement in 2015 which had the long term goal among its members of limiting the increase in global average temperature to 1.5 °C above pre-industrial levels.

Sustainable biomass² is considered as renewable fuel by the EU and UN. As such, it is feasible to attain a substantial net reduction in CO2 emissions by substituting fuels with a large carbon footprint such as coal and oil with biomass. When combusted, the inorganic macronutrients (i.e.

the ash forming elements) in biomass are enriched in the solid residues that remain. In some cases, the ash forming elements form problematic compounds that can cause significant ash-related problems such as slag formation or bed agglomeration that may significantly lower the efficiency and availability of the boiler. In addition, as the quality of global reserves of macronutrients such as phosphorus is uncertain, this has made biomass ash a potential future resource for the production of sustainable fertilizers. This section serves to give an overview of these topics and also explain related issues that motivate the objective of this thesis.

1.1 Current and future biomass use in Sweden

Since the 1970s, the total energy usage in Sweden has increased while the usage of fossil fuels has steadily decreased (Figure 1)³. This has mainly occurred through an increased usage of nuclear power and biomass with some contribution by wind- and hydropower. The majority of fossil fuels that were still in use in 2016 (159 TWh) were in the transport sector, mainly in the form of gasoline or diesel. The usage of fossil fuels in the industrial, housing and service sectors is relatively low and decreasing with some exceptions such as the iron and steel industries.

Biomass contributes the major share of renewable energy to the Swedish energy system and supplied approximately a quarter of the total primary resources or 139 TWh in 2016. By category, the majority of biomass originates from forest based biomass, mainly in the form of black liquor, refined woody biomass (wood pellets and briquettes) or unrefined woody biomass (wood chips, sawdust, bark) (Figure 2). These fuels are predominantly utilized through combustion within the industry (56 TWh) and housing sector (40 TWh) while the use within the transport sector is still relatively small (17 TWh). Sweden is currently on target to achieve the goals set by the EU for 2020, but still has quite some way to go to reach the 2030 and 2045 targets, particularly for the transport sector⁴. Therefore, significant changes to the Swedish energy system can be expected in the future. A recent techno-economical study investigated potential changes on the demand for forest based biomass for energy purposes in Sweden⁵. Based on their assessment, the demand for woody based biomass might increase by 10-50 TWh in 2030 and 10-60 TWh in 2050. This demand could potentially be met by an increased use of biomass from the forestry or agriculture sector such as logging residues or straw. These biomass streams alone have an estimated potential of 3.3-11.5 TWh⁶ and 16.1-43.1 TWh⁷ respectively and are not extensively used today.

8

Figure 1 Total energy usage in Sweden by category of fuel since between 1970-2016⁸.

Figure 2. Usage of biomass fuels by category in Sweden between 2005-2016⁸. 0

100 200 300 400 500 600 700

TWh

Windpower Hydropower Primary heat Nuclear fuel Other fuels Natural and city gas

Petroleum products Coal and coke Biomass

0 20 000 40 000 60 000 80 000 100 000 120 000 140 000 160 000

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016

GWh

Biogenic household waste Biogas

Other liquid biofuels vegetable or animal oil Tall- or pitch oil Biodiesel Bioethanol Other solid biomass Black liquor

Unrefined woody biomass Refined woody biomass

9

A limiting factor for the utilization of such fuels is that many existing boilers might not be easily converted to handle a large variety of biomass fuels. Forest residues and agricultural residues tend to have a more complex and problematic composition of ash forming elements than the more commonly used forest based biomasses. Depending on the ash characteristics of the biomass, the inorganic content of the fuel may cause significant operational issues as a result of ash related problems. For this reason, ash related problems such as slag formation, bed agglomeration, fouling, and high temperature corrosion has been extensively studied in the past^9,10. In less severe cases, this can result in more frequent maintenance shut-downs but can in more severe cases lead to

immediate shutdown of the plant to remove obstructing ash material from the boiler or to replace leaking boiler tubes.

Such operational issues can often be alleviated by changing plant parameters such as reducing combustion and steam temperatures or by retrofitting the ash handling systems. Yet, such strategies come at the cost of reducing the overall efficiency of the boiler or large capital costs. A preferred option would be to find other types of strategies to reduce or avoid the ash related problems. One possibility is to find non-problematic biomass blends by mixing different fuels or additives using a trial-and-error based approach. However, such solutions are typically plant specific, time consuming and might potentially cause severe problems if the wrong mixture is used. A better option would be to find more general fuel mixing strategies by developing our understanding of the ash transformation chemistry that determines whether ash related problems occur or not.

1.2 Biomass and its main ash forming elements

Biomass is relatively heterogeneous in terms of composition and can vary significantly depending on its origin, species or even within specific parts¹¹ of the same species. Ash forming elements in biomass fuels can broadly be classified as salts, elements organically bound in the biomass, included or adventitious minerals¹². The adventitious minerals differ from the rest and may include sand minerals and clays which have no biological function but is rather a result of fuel handling during storage or harvest. These are often considered to be less reactive than other inorganic components. Typical guiding values show that the bulk composition of biomass is quite similar and consists of C (46-63 wt%), O (25-45 wt%), and H (5.5-7,5 wt%) on a dry, ash free basis¹³. They also contain varying amounts of metals and non-metals (i.e. ash forming elements) that normally constitute 0.3-7 wt% of the biomass on a dry basis. Based on their abundance and relative importance for the ash transformation reactions, the main ash forming elements for biomass can be simplified to K, Na, Ca, Mg, Al, Si, P, S, and Cl¹⁴.

1.3 Measures for prevention of ash related problems and phosphorus recovery

A commonly used method for dealing with operational issues related to ash behavior is to change the fuel ash composition of the fuel mixture by co-combustion or the usage of additives.

For instance, peat has been extensively used together with different types of forest based biomasses to reduce bed agglomeration, slagging and fines emission^15–17. The positive effect can be related to the relative fuel concentration of certain elements such as Al to Si and/or Ca to Si. As such, a similar positive effect can also be seen by the addition of lime (CaO) or kaolin

additives(Al2Si2O5(OH)4) to name a few¹⁸. Some studies have shown that ash related problems can also be improved by the addition of P containing fuels or additives^19–21.

10

From another perspective, recent years have seen a large increase in interest in various techniques for large scale recovery of phosphorus from various waste streams, including combustion and subsequent recovery of phosphorus from ash,²² due to the importance of P in fertilizers. Currently, agriculture in the world is dependent on the usage of mineral fertilizer sources but reintroducing phosphorus from human or animal based waste stream could go a long way towards making the current system more sustainable. This indicates a possibility to combine the production of renewable heat and power with the production of ash fertilizer by co- combustion of challenging P-poor biomass with P-rich biomass.

Currently, a few challenges remain before such a system could be realized on a larger scale. The effect of phosphorus on ash related problems such as bed agglomeration or slagging is not complete, and may in some cases may exacerbate such problems²³. Additionally, there is currently a lack of knowledge on how to produce and recover combustion ashes with suitable fertilizer qualities for direct application on arable land.

1.3.1 Bed agglomeration and slag formation

Fluidized bed combustion is a common combustion technology in Sweden but also globally, and the effect of ash composition on bed agglomeration has been extensively research for combustion and co-combustion of a number of different biomass fuels with several types of additives^17,24–33. In general terms, bed agglomeration can be described as the formation of larger agglomerates of ash and bed material in the combustion zone of a fluidized bed. Often, this can be dealt by replacing the agglomerated bed material with new sand but can in severe cases lead to complete defluidization of the bed, causing an immediate shutdown of the plant.

The mechanisms for agglomerate formation can be divided into two main categories: Layer (or coating) induced mechanisms and a melt induced mechanism³⁴. Two layer induced mechanisms have been proposed in the literature for quartz bed materials as well as two melt induced

mechanisms23,28,30,35. The layer induced mechanisms are initiated by the formation of a low melting alkali silicates on the surface of the quartz bed grains by gaseous or liquid potassium compounds.

This leads to continuous inner reaction layers that may melt at the process temperature, resulting in viscous flow sintering and agglomeration. This is typical for biomass fuels that are rich in alkali, but poor in P and Si. In fuels richer in Ca such as woody biomass, the formation of K-rich inner reaction layers is followed by the buildup of a Ca-rich outer layer. These layers may also melt and cause viscous flow sintering and agglomeration, but at significantly higher temperatures than previous mechanism due to the presence of high temperature melting Ca-silicates.

The melt induced agglomeration mechanisms can be described as direct adhesion of bed grains by partially molten ash particles with a composition similar to the fuel ash. This mechanism is particularly severe for fuels with a low amount of alkaline earth metals and a high amount of alkali metals and Si or P such as wheat straw or dried distillers grain with solubles. In layer induced agglomeration mechanisms, chemical reactions between fuel ash and bed material plays a central role in the formation of agglomerates and as a result is more dependent on the choice of bed material than is the case for melt induced agglomeration³⁶. Previous studies have shown that P has a dominant role on the formation of agglomerates in fluidized bed combustion^23,37. When added to P-poor biomass through combustion or P additive in sufficient amounts, it leads to the formation of P-rich coating layers and bed ash particles in the bed. These layers may melt causing melt induced agglomeration similar to Si-rich fuels like straw. Additionally, the melting behavior of

11

these layers was found to be heavily dependent on the concentration of alkaline earth metals in the fuel.

The knowledge surrounding slag formation in grate fired/fixed bed boilers of P-poor fuels is quite well developed^18,38 while less is known of P-rich fuels^37,39. Slag formation in grate fired system can be described as fused or sintered ash in and around the grate/bed. The severity of slag formation depend not only on the total amount of molten ash that forms, but also on its viscosity as the ash melt need to be sufficiently viscous to form agglomerates⁹. In P-poor biomass, slag formation is initiated by the reaction of K-containing species with Si, forming a highly viscous silicate melt on the grate⁴⁰. Subsequently, less reactive ash species such as CaO or MgO can be dissolved into the melt. In P-poor fuels, the slagging tendencies of a fuel can generally be described by the total amount of ash and the relative concentration of K2O+Na2O, CaO+MgO and SiO241 with the concentration of Si being a major factor. Based on the studies on

sintering/slagging of P-rich fuels, P has a similar role in slag formation as Si and may form melts together with K, Ca and Mg with slagging severity ranging from low to high³⁹. Additionally, It has been shown that the Ca-additives reduces the slagging tendencies of P-rich fuels, possibly by the formation of high temperature melting Ca-phosphates^39,42. Severe slagging with P-rich fuels has generally been observed with fuels low in Ca and rich in K, P and Mg or K and P^37,39,42. 1.3.2 Plant availability of phosphorus in ash

Studies investigating the plant availability of phosphorus in biomass ash through chemical fractionation or plant studies have been met with varying results. Some indicating low to medium fertilizer value^43,44, while others have suggested good fertilizer effect^45,46. This is a result of the chemical association of nutrients in the ash which has been shown to have a large impact on the nutrient availability of phosphorus^47–49. Ashes exhibiting low fertilizer values may require costly thermal^50–52 or wet chemical methods^53,54 before they can be used, which would add significant cost to the process. Therefore, it’s necessary to understand what factors affect the association of phosphorus in biomass ash to reliably produce ashes with good fertilizer qualities. Consequently, it is also necessary to know the fertilizer qualities of the phosphates found in biomass ash to know what phases to aim for.

Building on the work of Boström et.al.¹⁴, Skoglund⁵⁵ described the behavior of phosphorus in combustion using a Lewis acid/base approach from the initial release from the biomass towards more stable compounds, including intermediate phases for a K, Ca and P system. The approach uses fundamental chemical thermodynamic and high-temperature chemistry concepts to indicate likely initial reaction pathways for the major ash-forming elements in biomass combustion. Based on this approach, meta- and pyro-phosphates should generally form when there is a lack of basic ash compounds such as K, Na, Ca or Mg. When there is a surplus of basic ash compounds, the chemical association of phosphates will tend towards the most stable orthophosphate compounds at present combustion conditions. It has been seen that phosphates in many cases react towards Ca-orthophosphates such as hydroxyapatite and whitlockite given a sufficient surplus of Ca^14,56. However, a recent study by Stemann et. al.⁵⁰ showed that that it is possible to shift the speciation away from Ca-orthophosphates using a method similar to the Rhenania process. By thermally treating Ca-orthophosphate rich sewage sludge ash together with sodium salts (Na2CO3, Na2SO4) at 1000 °C for 30 minutes, the speciation of phosphates was changed to buchwaldite, a mixed Ca- Na-orthophosphates. Additionally, the general reaction pathway of P, K and Ca starts out with the assumption that phosphorus is initially released as gaseous P2O5. This might not always be the case at it has been shown that the association of inorganic elements in the fuel can significantly change

12

the release behavior^57,58. Depending on the fuel, the association of phosphorus in biomass can vary significantly such as Ca-phosphates in meat and bone meal⁵⁹, organically bound P in biomass⁶⁰, or as Fe- or Al-phosphates in sewage sludge⁶¹. If any of these phosphates are not initially released as P2O5 during combustion, a different behavior is to be expected. A recent study also showed that the thermochemical conversion method (pyrolysis, gasification or combustion) and combustion technology should be considered as it was shown to affect the plant availability of P in sewage sludge ashes⁶².

A large variety of phosphate have thus far been identified in biomass ash but can roughly be classified as orthophosphates, pyrophosphates and metaphosphates. These can in turn be associated with K, Na, Ca, Mg, Fe and potentially Al, although the last one is still up for debate⁶³. It has been seen that orthophosphates associated with Ca, Fe or Al have relatively poor fertilizing performance according to the P solubility in neutral ammonium citrate, an indicator for plant availability⁶⁴. However some orthophosphate have shown good fertilizer quality by the same measure such as buchwaldite (CaNaPO4)50. Several pyrophosphates such as CaK2P2O7 has also been suggested to have promising fertilizer qualities^65,66.

1.4 Objective of thesis

While a number of studies have investigated the slagging and agglomeration tendencies of some P-rich biomasses, no comprehensive study on the effect of phosphorus on slag formation and bed agglomeration has been made for a wide range of fuel ash compositions. Additionally, while there exists a significant amount of information regarding the ash transformation behavior of

phosphorus, no comprehensive study has yet evaluated the effects of fuel composition and combustion technology on the chemical speciation of P in biomass ash.

Thus, the objective of this work was to make a comprehensive study on the effect of fuel composition and combustion conditions on the behavior of phosphorus during combustion. More specifically, the objective was to determine:

(i) The effect of fuel ash composition and combustion technologies on the fate of phosphorus in combustion of biomass

(ii) The effect of fuel association on the fate of phosphorus in combustion of biomass (iii) The effect of phosphorus on slag formation and bed agglomeration in combustion of biomass

13

2. Method

The objectives were investigated by a systematic review of data gathered from the combustion of 26 different biomass fuels or fuel blends in three different combustion technologies, i.e. bubbling fluidized bed (5 kW), fixed bed (20 kW), powder burner (150 kW). Samples of bottom ash, coarse fly ash and fine fly ash were gathered and analyzed for elemental composition and crystalline phase composition in addition to qualitative measures of bed agglomeration- and slagging tendencies.

2.1 Fuels

An overview of the different fuels or fuel blends that were used in this work can be seen in Table 1 including the combustion technology they were used in and earlier literature references.

Note that biomass marked in the additive or co-combustion column were only used for co- combustion and was not mono-combusted. These fuels or fuel blends cover a wide range of main ash forming (Table 2) elements but also differs in how phosphorus is associated in the fuel. The fuel association of P in plant based biomass is expected to be primarily organically bonded phosphates (e.g. inositol phosphates, DNA, ATP) that is with or without metallic cationic neighbors. During the initial stages of combustion, these are likely released as volatile P2O514. Phosphoric acid (H3PO4) was used as a model compound to add phosphorus to fuels without changing the overall ash composition of the blend. Although it is inorganic rather than organically bound, it should have similar behavior as the plant based biomass as it decomposes into P2O5, its anhydride, and steam when reaching a sufficiently high temperature. Phosphorus in sewage sludge that was used is likely inorganically bound to iron⁶¹ as a result of chemical precipitation method (FeSO4) but may also be bonded to Ca, Al in addition to some organically bound phosphates.

14

Table 1. Overview of fuels, fuel blends, additives, their abbreviations, in what combustion technology they were used as well as what paper they were used and earlier literature references.

P- poor forest based biomass Abbreviation

Fluidized bed

Fixed bed

Powder burner

Additive # or co- combustion

Paper and references

Bark 1 Ba1 X X 1,321,37,55,67,68

Bark 2 Ba2 X

Logging Residues LR X X 1-320,23,37,55,68,69

Willow Wi X 1,3^23,37,55

Wood Powder 1 WP1 X

Wood Powder 2 WP2 X

P-poor agricultural biomass

Wheat Straw 1 WS1 X X 1-319,20,23,37,55,68,69

Wheat Straw 2 WS2 X

Wheat Straw 3 WS3 X 1⁷⁰

P-rich agricultural biomass

Rapeseed Meal RM X 1,3⁶⁷

RM 80%, WP1 20% RP* X X 1,321,37,55,67,68

Dried Distiller’s Grain with solubles DG X X X 1-323,37,68,69

Waste based biomass

Sewage Sludge SS X

Additives

Phosphoric Acid (H3PO4) PA X

Forest/agricultural blends

Ba1 90 %, RM 10% Ba1RM10 X X 1,321,37,55,67,68

Ba1 70%, RM 30% Ba1RM30 X X 1,321,37,55,67,68

50% Ba2, 50% WS3 Ba2WS350 X 1⁷⁰

25% Ba2, 75% WS3 Ba2WS375 X 1⁷⁰

LR 95%, DG 5% LRDG5 X 2

LR 60%, DG 40% LRDG40 X X 1-323,37,55,68,69

LR 99.93%, PA 0.07% LRPA0.1 X 3^23,37,55

LR 99.76%, PA 0.24 % LRPA0.2 X 2,320,23,37,55

Wi 50%, DG 50% WiDG50 X 1,3^23,37,55

50% WP2, 50% WS3 WP2WS350 X 1⁷⁰

25% WP2, 75% WS3 WP2WS375 X 1⁷⁰

WS1 50%, DG 50% WS1DG50 X X 1-337,55,68,69

WS1 98.86%, PA 1.14% WS1PA1 X X 1-3^20,37,55

WS2 97.88%, PA 2.12% WS2PA2 X 1,3^20,37,55

WS2 97.02%, PA 2.98% WS2PA3 X 1,3^20,37,55

WS2 95.5%, PA 4.5% WS2PA4.5 X 1,3^20,37,55

Waste blends

LR 97%, SS 3% LRSS3 X 2

WS1 91%, SS 9% WS1SS9 X 2¹⁹

# Biomasses marked in the additive or co-combustion column were only combusted in biomass blends and never mono-combusted.

* RM was mixed with ground stem wood (ash 0.3 wt% d.s.) to improve pellet quality without significantly changing the ash composition

Based on the fuel ash composition, the fuels can be categorized into four groups: P-poor forest based biomass, P-poor agricultural biomass, P-rich agricultural biomass, and waste based biomass.

The P-poor forest based biomass has the lowest ash content of the groups (0.3-3.7 wt%) with an

15

ash is dominated by Ca, Si, and K with the exception of Wi (low Si). The P-poor agricultural biomass is represented by three different wheat straws with an ash content that varies between 4.8 to 6.2 wt% and a composition dominated by Si, K, and Ca. The P-rich agricultural biomasses are residuals from ethanol and biodiesel production of cereal crops. The overall ash content is quite high (4.4-7.4 wt%) with an ash dominated by K, P, S, Mg, and Ca. The waste residue is a digested sewage sludge and is representative of a typical Swedish sewage sludge⁷¹. The ash content in this fuel is very high (41.7 wt%) and is dominated by Fe, P, Si, Al, Ca and S.

Table 2. Composition of main ash forming elements for fuels or fuel blends used in this study. Colors represent a low (green), medium (yellow) and high (red) relative molar fraction of the total ash forming elements for each fuel.

Fuel/fuel blend

Wt% mmol × kg-1 d.b.

Ash K Na Ca Mg Fe Al Si P S Cl

Ba1 3.70 49 14 185 26 8 32 178 12 12 6

LR 2.40 43 6 127 25 4 13 103 15 13 0

Wi 2.10 64 5 125 18 2 6 31 19 12 0

WS1 5.70 320 13 100 41 1 2 285 42 59 73

WS3 4.80 140 10 91 34 1 2 293 11 41 118

RM 7.40 338 6 180 220 6 5 32 406 284 8

RP 5.98 271 5 148 185 5 4 26 325 227 7

DG 4.40 271 43 27 114 2 0 36 266 321 62

Ba1RM10 4.07 77 13 185 46 7 30 163 51 40 6 Ba1RM30 4.81 135 11 184 84 7 24 134 130 94 6

Ba2WS350 4.25 83 9 152 32 2 10 217 9 28 62

Ba2WS375 4.53 112 9 121 33 1 6 255 10 34 90

LRDG5 2.50 55 8 122 30 4 13 100 27 28 3

LRDG40 3.20 135 21 87 61 3 8 76 115 136 25

LRPA0.1 2.47 43 6 127 25 4 13 103 22 13 0

LRPA0.2 2.63 43 6 127 25 4 13 103 39 13 0

WiDG50 3.25 168 24 76 66 2 3 33 143 167 31

WP2WS350 2.55 72 5 54 19 0 1 149 5 20 61

WP2WS375 3.68 106 7 73 26 0 1 221 8 30 90

WS1DG50 5.05 295 28 64 78 1 1 160 154 190 68

WS1PA1 6.78 316 13 99 41 1 2 282 158 59 73

WS2PA2 8.19 225 13 112 32 3 8 523 248 34 66 WS2PA3 8.99 223 13 111 32 3 8 518 335 33 66 WS2PA4.5 10.42 220 12 110 31 3 8 510 490 33 65

LRSS3 3.58 45 8 142 31 46 39 140 55 25 0

WS1SS9 8.94 300 19 147 56 125 80 378 161 92 68

In addition, these fuels were co-pelletized at different ratios to cover intermediate compositions between the fuel groups. Phosphoric acid was added to logging residues and wheat straw 1 and 2 at different levels to evaluate the overall effect of adding P to P-poor forest or agricultural biomass.

As a whole, the fuels and fuel blends cover a wide range of composition with respects to the phosphate and silicate system as is illustrated in Figure 3.

16

Figure 3. Quasi-ternary composition diagram [mol %] for the (Ca+Mg)–(K+Na)–P system showing the range of ash composition of fuels used in this work. Data labels indicate Si fraction [mol %] of the ash composition with respect to the main ash forming elements (K, Na, Ca, Mg, Fe, Al, Si, P, S, Cl).

2.2 Experimental procedure and sampling

This section briefly describes the combustion conditions in each combustion technology; the experimental procedure and sampling; and the chemical characterization of the ash samples. For a more detailed account, see appended papers or references in Table 1.

As shown in Table 3, the combustion technologies vary considerably in maximum temperature, heating rate, and fuel separation which are all important physical factors that can affect the stability of ash species but may also affect if reactions can reach thermodynamic equilibrium. The fluidized bed is characterized by a low combustion temperature and medium heating rates due to the large amount of sand in the bed which increases heat transfer and functions as a heat buffer, preventing high peak temperatures. The fuel is distributed evenly throughout the sand bed which reduces the contact between ash elements but also enables interaction between ash and sand. The peak

34%

30%

29%

11%

5%

12%

30%

50%

40%

2%2%

3%

21%

26% 17%

36%

38%

26%

11%

29% 28%

5%

38%

39%

15% 27%

41%

39%

34%

27%

26%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

P K+Na

Ca+Mg

P-poor forest based biomass P-poor agricultural biomass P-rich agricultural biomass Waste based biomass forest/agricultural blends Waste blends

17

combustion temperatures in fixed bed combustion technologies are significantly higher compare to that of the fluidized beds but the heating rate is slower. The fuel particles form a bed on the grate which allows for a high contact between ash elements as the fuel burns out. The powder burner has a high maximum combustion temperature and heating rate which increases the volatility of ash elements. The horizontal setup used in this study leads to a more significant fractionation of ash elements as heavier fuel and ash particles are quickly removed from the flue gas stream by gravity.

As a result, these conditions favor fast gas phase reactions while solid or liquid phase reactions are suppressed, promoting non-equilibrium conditions¹⁴.

Table 3 Differences in relevant experimental conditions across the different combustion technologies.

Combustion technology Max Temp [°C] Heating

rate Fuel

separation Flue gas O2

Bubbling fluidized bed Low (~800 °C) Medium Medium 9±1

Fixed bed Medium (~1200 °C) Low Low 9.5±0.5

Horizontal Powder burner High (>1400 °C) High High 5±1.5

On average, the combustion experiments carried out in fluidized bed, fixed bed, and powder burner lasted 8, 24, and 6 hours respectively. In some cases, the experiment had to be cut short due to severe slagging or bed-defluidization. During the experiments, an air-cooled deposition probe was inserted into the fluidized bed and powder burner and maintained at ~450 °C and 550

°C respectively to mimic superheater surfaces. The probe was inserted once combustion conditions had reached steady state and was removed towards the end of the experiment. Upon removal, the deposit was further divided into a wind and lee side deposit before analysis. Fly ash particles in the 0.03-10 μm range was sampled isokinetic from the flue gas channel by using a preheated 13-step Dekati low-pressure impactor. The impactor fractionates the fly ash particle based on aerodynamic diameter between 13 separate stages. Before analysis, the 13 stages were consolidated into two fractions: fine PM1 fly ash (stage 1-7) and a coarse (~1-10 μm) fly ash (stage 8-13). Towards the end of the fluidized bed experiments, a bed sample was taken followed by a controlled fluidized bed agglomeration test⁷². In this test, fuel flow is stopped and combustion atmosphere is maintained by a propane gas flame that is lit underneath the bed. The temperature of the bed is then slowly raised by preheating the primary air flow and freeboard using electrical heaters until the bed collapses (bed defluidization) or the maximum reactor temperature is reached (1050°C). Once the reactor had cooled, an additional bed sample was gathered.

Several ash samples were gathered once the reactors had cooled down. From the fluidized bed, cyclone ash (cut off size 10 µm) was collected and from the fixed bed and powder burner experiments, bottom ash was gathered. The bottom ash from the fixed bed combustion experiments were further separated into slag fraction and a non-sintered bottom ash fraction by passing it through a sieve. Slag particles that did not pass a mesh (3.15 mm) were considered slag and the rest as bottom ash. The slagging tendencies of the fuel was evaluated by calculating the wt% of total bottom ash that formed slag and a visual sintering category that qualitatively estimated the degree of sintering⁴⁰. A detailed account of all ash fractions that were subject to chemical characterization are seen in Figure 4.

18

Figure 4 Overview, classification and abbreviation of the analyzed ash fractions.

The elemental composition of ash fractions was determined by scanning electron microscope (SEM) combined with an energy-dispersive X-ray spectroscopy detector (EDS) and crystalline phases were identified with powder X-ray diffraction (XRD). The diffractograms produced were further subjected to Rietveld refinement for a semi-quantitative analysis of the crystalline phased present in the ash fractions.

2.3 Experimental design

For a more detailed account on what fuels or fuel blends were included in each paper, see Table 1.

2.3.1 Paper 1

The effect of fuel ash composition and combustion technologies on the fate of phosphorus during combustion of biomass was investigated by a systematic review of experimental data from the combustion a large variety of biomass fuels or fuel blends across three different combustion technologies (BFB, FXB, PB). The fuels included three biomass types with distinctly different ash composition including P-poor forest based biomass, P-poor agricultural biomass, P-rich

agricultural biomass and a P-additive to cover a wide range of fuel ash compositions in terms of Ca+Mg, K+Na, P, and Si. The most important differences between the combustion technologies

19

are described in section 2.2. The main experimental data included chemical characterization of bottom ash fractions, coarse fly ash fractions and fine fly ash fractions by SEM-EDS and XRD coupled with Rietveld calculations.

2.3.2 Paper 2

In paper 2, a similar approach was used as in paper 1. However, only experiments from the bubbling fluidized bed was included. The effect of fuel association on the fate of phosphorus was investigated by comparing the chemical association of phosphorus in the ash when burning fuels with similar fuel ash composition in terms of Ca+Mg, K+Na, P, but with fuels of different phosphate association (phosphoric acid, P-rich agricultural residues or sewage sludge).

2.3.3 Paper 3

The same general approach of paper one was also used in paper 3, but in this case the focus was on bed agglomeration in fluidized beds and slag formation in fixed beds. The effect of P was investigated by comparing the chemical characterization of bed and bottom ash samples with the results from the controlled fluidized bed agglomeration test (initial defluidization temperature), and the two slagging indicators (slag fraction and visual sintering category).

2.4 Derived results and fuel molar ratios

The weight fraction of P that contributed to the formation of submicron particles (PM1) was estimated by the ratio between the average flow of P in PM1 to that introduced with the fuel.

The average flow of P in PM1 was calculated using a combination of impactor data (weight if impactor stages 1-7) and the P concentration (SEM-EDS) of the most loaded impactor stages (typically 4, 5, and occasionally 6) with the assumption that the concentration was representative of sample stage 1-7. Average flow of P into reactor was calculate from average mass flowrate of fuel and P concentration of the fuel.

To enable an easier comparison on the distribution of phosphorus between the various

crystalline phosphate phases, the semi-quantitative XRD results were converted to molar basis and normalized to only include phosphates. Furthermore, they were normalized to consider that phosphate molecules may contain different number of phosphorus atoms per molecule. The final XRD results are presented as a molar % distribution of phosphorus between crystalline phosphate phases.

20

Based on the chemical characterization of ash fractions, it was apparent there was some

correlation between P and the elements Ca and Mg. To help out in the interpretation of results, a molar ratio between P, Ca and Mg was used to sort the experimental data. The phosphorus alkaline earth metal ratio (or PAR), as defined in equation 1, generalizing the ash transformation behavior of phosphorus during combustion of biomass by the relative availability of these ash compounds.

𝑃𝐴𝑅 = 2 ^𝑃

3(𝐶𝑎+𝑀𝑔) Equation 1

3. Results and discussion

3.1 The effect of fuel ash composition and combustion technologies on the fate of phosphorus in combustion of biomass (Paper 1)

The distribution of phosphorus between ash fractions followed a similar trend between the combustion technologies with the highest share of fuel P in bottom ash fractions and coarse fly ash fractions and low to very low amounts in fine fly ash fractions. The exact distribution of P between bottom and coarse fly ash fractions were not quantified due to difficulties of separation bed ash from bed material in fluidized beds, but was observed to differ between the combustion technologies. This was also true for the share of fuel P in fine fly ash (PM1) which differed by more than an order of magnitude between the combustion technologies with the powder burner producing the highest share, followed by fixed bed and fluidized bed (Figure 5).

Figure 5 Relative fraction of fuel phosphorus in submicron particles (wt%). Each point represents one experimental with a specific fuel or fuel blend with colors representing in what combustion technology that was utilized. The X axis indicate the relative concentration of P to Ca and Mg of the fuel. The P-poor forest or agricultural biomass are located to the far left (fuel PAR>0.5), P-rich agricultural biomass is located between 1.5-2.5 and blends are intermediate. The highest values are wheat straw 2 with high additions of phosphoric acid.

0%

2%

4%

6%

8%

10%

12%

0.00 1.00 2.00 3.00 4.00 5.00

Phosphorus PM1 of total fuel P [wt%]

Fuel PAR(1.5 × P ×(Ca+Mg)^-1)

Fluidized bed Fixed bed Powder burner

21

Fine submicron ash particles form as a result of nucleation, condensation, coagulation and agglomeration of gaseous ash species. They often contain a much higher concentration of volatile heavy metals such as Zn and Pb⁷³. These elements are harmful to humans and for this reason, strict regulations are set for their application to farmland. This could severely limit the usage of ash fractions containing a high share of PM1 ash particles as fertilizer. Therefore, a low fraction of P in PM1 is preferable as it indicates that it’s possible to achieve good separation between P and volatile heavy metals during combustion.

In Figure 6, the relative concentration of P to Ca and Mg (or PAR, see equation 1) of the fuel is plotted against the PAR in the ash. It can be seen that for most bottom and fly ash fractions, the ash PAR tend to be equal to or slightly below fuel PAR. This is an indication that P, Ca and Mg are associated in the bottom and coarse fly ash fractions. In addition, this suggest that for fuels with PAR values below one, there will be a sufficient amount of Ca and Mg in bottom and fly ash fractions to exclusively form (Ca,Mg)- orthophosphates such as Ca3(PO4)2. For fuels with fuel PAR above 1, other counter cations such as K⁺ or Na⁺ may fill out the phosphate structure.

Further support for the strong association between P, Ca and Mg in in bottom and coarse fly ash fractions was seen by the systematic review of XRD data. By sorting the XRD results for the bottom/bed ash fractions against the fuel molar ratio PAR (Table 4), some general trends become discernable. In the table, the fuels are sorted with increasing fuel PAR from top to bottom and the ratio between P and Ca or Mg of the identified phosphate phases increasing from left to right. For fuels with a low PAR, the identified phosphates tend to be located to the left. Moving down in the table, the dominant phosphate species tends to move towards the right of the table. In other words, transitioning from a phosphate speciation dominated by Ca5(PO4)3OH and Ca3(PO4), to (Ca,Mg)KPO4, then CaK2P2O7 depending on the availability of Ca and Mg in the fuel. The same trend was seen for the XRD results of coarse fly ash fractions. In addition, the same trend can also be seen across all three combustion technologies.

This suggests that fuel PAR could likely be used as an indicator to estimate what phosphates will form in the bottom and coarse fly ash fractions during combustion of plant based biomass. From the utilization of ash as fertilizer, it would likely be beneficial to aim for a higher fuel PAR as it seemingly allows for the formation of phosphates other than Ca5(PO4)3OH or Ca3(PO4)2 which have previously been shown to have a low fertilizer quality. The inclusion of other cations than Ca such as Mg and K in orthophosphates (PO43-), pyrophosphates (P2O74-) would likely be beneficial as it has been seen that such phosphates have favorable fertilizer qualities^50,65.

22

Figure 6. Comparison of the molar ratio between P to Ca and Mg of the fuel ash and A, bottom/bed ash fractions and B, coarse fly ash fractions (>1 µm). The X axis indicate the relative concentration of P to Ca and Mg of the fuel. The P-poor forest or agricultural biomass are located to the far left (fuel PAR>0.5), P-rich agricultural biomass is located between 1.5-2.5 and blends are intermediate. The highest values are wheat straw 2 with high additions of phosphoric acid.

0 1 2 3 4 5

0 1 2 3 4 5

Ash PAR (1.5 ×P ×(Ca+Mg)-1)

Fuel PAR (1.5 × P ×(Ca+Mg)^-1)

A

Fluidized bed Bed ash Fixed bed Slag Fixed bed Bottom ash Powder burner Bottom ash Fuel PAR

0 1 2 3 4 5

0 1 2 3 4 5

Ash PAR (1.5 ×P ×(Ca+Mg)-1)

Fuel PAR (1.5 × P ×(Ca+Mg)^-1)

B

Fluidized bed fly ash Fixed bed Fly ash Powder burner Fly ash Fuel PAR

23

Table 4. Distribution of phosphorus between crystalline phosphate phases identified by XRD analysis coupled with Rietveld calculations of: A, bed ash particles from fluidized bed; B, bottom ash from fixed bed; C, bottom ash from powder burner and; D slag from fixed bed. Results are sorted on increasing fuel PAR from top to bottom and increasingly Ca- and Mg-rich phosphates from right to left.

a Molar fraction of total crystalline phases that are phosphates ^b on a Si free basis. Asterix denote the relative

concentration of phases in the sample as follows: ****, >50 wt% of sample; ***, 20-50 wt% of sample; **, 5- 20 wt%

of sample; *, <5 wt% of sample.

In addition, this trend in crystalline phase composition is strong support for the general ash transformation pathway suggested by Skoglund⁵⁵ which is further developed in Paper 1. The extended case for the reaction of phosphorus for a system containing P, K and Ca is given below (see reaction 1-6)