The Implication of Emission Control Systems on the Mass and Energy Balances of the Kraft Pulp Mill Process A Case Study of Södra Cell Mönsterås’s Recovery Boiler Master’s thesis in Innovative and Sustainable Chemical Engineering JESPER JOHANSSON SAMUEL OLSSON DEPARTMENT OF SPACE, EARTH AND ENVIRONMENT CHALMERS UNIVERSITY OF TECHNOLOGY Gothenburg, Sweden 2024 www.chalmers.se www.chalmers.se MASTER’S THESIS SEEX30 2024 The Implication of Emission Control Systems on the Mass and Energy Balances of the Kraft Pulp Mill Process A Case Study of Södra Cell Mönsterås’s Recovery Boiler JESPER JOHANSSON SAMUEL OLSSON Department of Space, Earth and Environment Chalmers University of Technology Gothenburg, Sweden 2024 The Implication of Emission Control Systems on the Mass and Energy Balances of the Kraft Pulp Mill Process A Case Study of Södra Cell Mönsterås’s Recovery Boiler JESPER JOHANSSON & SAMUEL OLSSON © JESPER JOHANSSON & SAMUEL OLSSON, 2024. Supervisors: Jenny Larfeldt, Södra Jakob Johansson, Space, Earth and Environment Rosa Citra Aprilia, Space, Earth and Environment Examiner: Fredrik Normann, Space, Earth and Environment Master’s thesis 2024 Department of Space, Earth and Environment Division of Energy Technology Chalmers University of Technology SE-412 96 Gothenburg Sweden Telephone +46 31 772 1000 Cover: Södra Cell Mönsterås’s kraft pulp mill Typeset in LATEX, template by Kyriaki Antoniadou-Plytaria Gothenburg, Sweden 2024 iii The Implication of Emission Control Systems on the Mass and Energy Balances of the Kraft Pulp Mill Process A Case Study of Södra Cell Mönsterås’s Recovery Boiler Jesper Johansson & Samuel Olsson Department of Space, Earth and Environment Division of Energy Technology Chalmers University of Technology Abstract This master’s thesis concludes on the mass and energy balances of the kraft pulp mill process and investigates the possible implications in connection to new emis- sion control systems. The work is based on a case study of the recovery boiler at Södra Cell in Mönsterås. Key aspects include the energy and element balances oc- curring in the boiler and the emission of NOx and SOx . Three primary emission control measures are evaluated: injection of ammonia solutions (SNCR), injection of ammonia-containing gas (dissolver off gas), and the injection of scrubber effluent. The process model is based on previous work and validated by comprehensive data collection from the reference plant. The model is implemented in the software Aspen Plus. The model is zero-dimensional and equilibrium-based and cannot describe the effects of spatial distribution of air and temperature within the boiler or processes controlled by kinetic-based reactions. The result shows that the implementation of water-diluted ammonia injection with the potential to reduce NOx emissions by 45 % does not significantly affect flue gas temperature. While the use of dissolver off gas is advantageous, due to the utilisation of a waste stream, current ammonia levels from the dissolver off gas is insufficient for achieving significant SNCR effects compared to water-diluted am- monia injection, with a difference of 0.69 mol NH3/s. Recycling scrubber effluent does not significantly alter the sulphidity of the flue gas but does increase reduction efficiency of the smelt, thereby reducing the need for make-up chemicals. The derived model is useful for describing the overall implications of the recovery boiler’s mass and energy balances and could be used to quantify the effects of other process modifications. The work should be complemented with more detailed design tools. Keywords: Södra Mönsterås, kraft pulp mill, recovery boiler, Aspen Plus, emission control, primary measures, NOx , SOx Effekterna av Utsläppskontrollsystem på Mass- och Energibalanserna i Kraftmassabruksprocessen En Fallstudie av Södra Cell Mönsterås Sodapanna Jesper Johansson & Samuel Olsson Institutionen för rymd-, geo- och miljövetenskap Avdelningen för Energiteknik Chalmers Tekniska Högskola Sammanfattning Detta examensarbete sammanfattar mass- och energibalanserna för kraftmassabruk- sprocessen och undersöker de möjliga effekterna i samband med nya utsläppskon- trollsystem. Arbetet är baserat på en fallstudie av sodapannan vid Södra Cell i Mönsterås. Viktiga aspekter är energi- och elementbalanserna i pannan samt ut- släppen av NOx och SOx . Tre primära åtgärder för utsläppskontroll utvärderas: in- jektion av vattenlöst ammoniak (SNCR), injektion av ammoniakhaltig gas (imånga) och injektion av skrubbervätska. Processmodellen är baserad på tidigare arbete och validerad genom omfattande datainsamling från referensanläggningen. Modellen är implementerad i program- varan Aspen Plus. Modellen är nolldimensionell och jämviktsbaserad och kan därmed inte beskriva effekterna av rumslig fördelning av luft och temperatur i pannan eller processer som styrs av kinetikbaserade reaktioner. Resultatet visar att implementeringen av vattenlöst ammoniakinjektion med poten- tial att minska NOx-utsläppen med 45 % inte påverkar rökgastemperaturen nämn- värt. Även om användningen av imångan är fördelaktig, på grund av användningen av en avfallsström, är de nuvarande ammoniaknivåerna från imångan otillräckliga för att uppnå betydande SNCR-effekter jämfört med vattenlöst ammoniakinjektion, med en skillnad på 0.69 mol NH3/s. Återvinning av skrubbervätskan förändrar inte rökgasens sulfiditet nämnvärt i pannan, men ökar reduktionsgraden i smältan och minskar därmed behovet av tillsatskemikalier. Den framtagna modellen är användbar för att beskriva de övergripande effekterna av sodapannans mass- och energibalanser och kan användas för att kvantifiera effek- terna av andra processmodifieringar. Arbetet bör kompletteras med mer detaljerade designverktyg. Nyckelord: Södra Mönsterås, kraftmassabruk, sodapanna, Aspen Plus, utsläppskon- troll, primära åtgärder, NOx , SOx Acknowledgements We extend our gratitude to all those who have supported and contributed to this thesis. A special thank you goes to our examiner, Fredrik Normann, and our supervisor, Jakob Johannson, for their valuable input and thorough assistance throughout this work. We have appreciated our regular meetings that steered us in the right direc- tion as well as engaging discussions. We would also like to thank Rosa Citra Aprilia who supported us in the initial stages of our work and provided valuable information. A heartfelt thank you to Jenny Larfeldt, our supervisor at Södra, whose exceptional guidance and support made this project possible. From providing crucial data to arranging meetings and even booking hotel rooms, your assistance have been very helpful along the way. Additionally, we give our gratitude to Södra as a whole for facilitating our visit to Mönsterås for an entire week. Thanks to Francis Gillet, Mats Häll, and Magnus Tyrberg for extending a warm welcome in Mönsterås and providing us with various data during the project. Finally, a last thank you to Samuel Myrberg for an enjoyable collaborative workflow between our parallel projects and interesting discussions throughout the thesis. Jesper Johansson & Samuel Olsson, Gothenburg, May 2024 Contents List of Figures ix List of Tables xi 1 Introduction 1 1.1 Aim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 2 Theory 3 2.1 The kraft pulp mill and chemical recovery process . . . . . . . . . . . 3 2.2 The recovery boiler . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.2.1 Sulphidity, sodium and sulphur chemistry . . . . . . . . . . . 8 2.2.2 Sulphur oxides and hydrogen chlorine formation routes . . . . 11 2.2.3 Nitrogen oxides formation routes . . . . . . . . . . . . . . . . 14 2.3 Current BAT and primary control measures of NOx and SOx . . . . . 16 2.3.1 Air staging . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.3.2 Ammonia injection . . . . . . . . . . . . . . . . . . . . . . . . 18 2.3.3 Flue gas recirculation, fuel staging and reduced air preheat . . 19 2.3.4 Primary control measures for SOx . . . . . . . . . . . . . . . . 20 2.4 Wet flue gas treatment with NO oxidation . . . . . . . . . . . . . . . 20 3 Method 22 3.1 Collecting plant data . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 3.1.1 Södra’s recovery boiler . . . . . . . . . . . . . . . . . . . . . . 24 3.1.2 Inlet data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 3.1.3 Reference data . . . . . . . . . . . . . . . . . . . . . . . . . . 26 3.2 Process modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 3.2.1 Assumptions and thermodynamic model . . . . . . . . . . . . 28 3.2.2 Design specifications of the model . . . . . . . . . . . . . . . . 29 3.3 Model validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 3.4 Implementation of primary measures . . . . . . . . . . . . . . . . . . 38 3.4.1 Ammonia injection . . . . . . . . . . . . . . . . . . . . . . . . 38 3.4.2 Recycling of scrubber effluent . . . . . . . . . . . . . . . . . . 40 4 Results 41 4.1 Model validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 4.1.1 Mass and energy balances . . . . . . . . . . . . . . . . . . . . 41 4.1.2 Sensitivity analysis . . . . . . . . . . . . . . . . . . . . . . . . 44 vii Contents 4.2 Primary measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 4.2.1 Ammonia injection . . . . . . . . . . . . . . . . . . . . . . . . 47 4.2.2 Scrubber effluent recycling . . . . . . . . . . . . . . . . . . . . 49 5 Discussion 51 5.1 Model validation and limitations . . . . . . . . . . . . . . . . . . . . . 51 5.1.1 Smelt and reduction efficiency . . . . . . . . . . . . . . . . . . 51 5.1.2 Air flows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 5.1.3 Flue gas flow and composition . . . . . . . . . . . . . . . . . . 53 5.1.3.1 Formation of NOx and SOx . . . . . . . . . . . . . . 53 5.1.4 Temperature and energy . . . . . . . . . . . . . . . . . . . . . 54 5.2 Primary measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 5.2.1 Ammonia injection and dissolver off gas . . . . . . . . . . . . 55 5.2.2 Scrubber effluent recycling . . . . . . . . . . . . . . . . . . . . 56 6 Conclusion 58 7 Future work 59 Bibliography 60 A Appendix 1 I B Appendix 2 IX B.1 Dissolver off gas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IX B.2 Scrubber effluent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XI viii List of Figures 2.1 The kraft recovery process. . . . . . . . . . . . . . . . . . . . . . . . . 4 2.2 The main reactions in a recovery boiler at a kraft process [9]. . . . . . 6 2.3 Sulphur and sodium balance in a typical recovery boiler. . . . . . . . 10 2.4 Formation routes for sulphur, sodium and chloride in the flue gas. Cool bed: S/Na2 > 1 [10]. . . . . . . . . . . . . . . . . . . . . . . . . 12 2.5 Formation routes for sulphur, sodium and chloride in the flue gas. Hot bed: S/Na2 < 1 [10]. . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.6 Formation routes of thermal, prompt, and fuel NOx . . . . . . . . . . . 14 2.7 Formation routes of fuel NOx . . . . . . . . . . . . . . . . . . . . . . . 15 2.8 The kraft recovery process with recycling of scrubber effluent. . . . . 21 3.1 Methodology flowchart for this thesis. . . . . . . . . . . . . . . . . . . 23 3.2 Schematic flowchart of the recovery boiler at Södra Cell Mönsterås. . 24 3.3 An overview of the recovery boiler modelling structure. . . . . . . . . 30 3.4 Process flow diagram of the drying section, where the wet black liquor (B) entering the boiler and is completely dried by volatiles (V) trav- eling upwards. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 3.5 Process flow diagram of the pyrolysis section, where the dry black liquor is decomposed into volatiles, inorganic compounds and fixed carbon. The volatiles traveling upwards through the pyrolysis section are combusted in the FLAME unit with secondary air. . . . . . . . . 33 3.6 Process flow diagram of the gasification and reduction section. The primary air provides the optimal amount of oxygen for gasification and reduction, resulting in release of volatiles (V1) and a smelt leaving the boiler. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 3.7 Process flow diagram of the combustion section (above pyrolysis). Flue gases in the volatile stream are combusted with DNCG, tertiary air and quaternary air. . . . . . . . . . . . . . . . . . . . . . . . . . . 35 3.8 Process flow diagram of the steam generation section. The flue gas (red line) exchanges heat to the steam-water system (dashed line) before leaving the boiler. . . . . . . . . . . . . . . . . . . . . . . . . . 36 3.9 Model integration of the ammonia injection. . . . . . . . . . . . . . . 38 4.1 Reduction efficiency in relation to primary air, for different secondary air flows. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 4.2 Flue gas temperature and reduction efficiency in relation to primary air flow (secondary air same as base case: 82.57 Nm3/s). . . . . . . . 45 ix List of Figures 4.3 Deviations in the flue gas composition, starting from 1100°C, with increased temperature, above quaternary air level. . . . . . . . . . . . 46 4.4 CO level in flue gas with increased temperature, above quaternary air level. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 4.5 Ammonia concentration in relation to heat load loss. . . . . . . . . . 48 x List of Tables 2.1 Yearly BAT requirement ranges of emissions for kraft recovery boilers [4]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 3.1 Elemental composition of thick black liquor (dry basis) [22]. . . . . . 25 3.2 Inlet flows and temperatures [20]. . . . . . . . . . . . . . . . . . . . . 25 3.3 CNCG inlet flows and temperature [20]. . . . . . . . . . . . . . . . . 26 3.4 Distribution of temperatures and O2 in the boiler based on design specifications [20]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 3.5 Concentrations and compositions of dissolved green liquor [22]. . . . . 27 3.6 Measured flue gas data carried out by DGE and ILEMA [20]. . . . . . 27 3.7 Yearly emission values from Södra’s recovery boiler compared to BAT requirements [22]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 3.8 Reference data for steam generation [22]. . . . . . . . . . . . . . . . . 28 3.9 Ammonia injection inlet data. . . . . . . . . . . . . . . . . . . . . . . 38 3.10 Dissolver off gas inlet data [20]. . . . . . . . . . . . . . . . . . . . . . 39 3.11 Inlet effluent stream data. . . . . . . . . . . . . . . . . . . . . . . . . 40 4.1 Smelt composition, flow and reduction efficiency compared to the literature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 4.2 Air inlet flows compared to the literature and collected inlet data from the design specifications. . . . . . . . . . . . . . . . . . . . . . . 42 4.3 Flue gas composition and flow compared to the literature, DGE’s and ILEMA’s measurement as reference data. . . . . . . . . . . . . . . . . 42 4.4 Regional temperatures in comparison to reference data from the de- sign specifications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 4.5 Heat exchanger units and excess heat load in flue gas. . . . . . . . . . 43 4.6 Overall energy balance compared to the literature and reference data. 43 4.7 NOx reduction and flue gas heat load losses with water-diluted am- monia injection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 4.8 Initial concentrations of NOx from different reference data and final concentration of NO in dry flue gas, with water-diluted ammonia injection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 4.9 Ammonia dosage in flue gas comparison with adding dissolver off gas (Case 1) and decreased flow of 4th air (Case 2). . . . . . . . . . . . . 49 4.10 Comparison of composition, temperatures and heat load increase in flue gas with added dissolver off gas (Case 1) and decreased flow of 4th air (Case 2). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 xi List of Tables 4.11 Sulphur and sodium balance from base case compared to the case with recycled scrubber effluent. . . . . . . . . . . . . . . . . . . . . . 50 4.12 Flue gas composition, heat load loss and reduction efficiency from base case compared to the case with recycled scrubber effluent. . . . . 50 A.1 Sample composition of black liquor (concentrated product). . . . . . . I A.2 Sample operating parameters for mass and energy balances. . . . . . I A.3 Molar masses needed for calculation. . . . . . . . . . . . . . . . . . . III A.4 Smelt and stoichiometric combustion products. . . . . . . . . . . . . . IV A.5 Air requirement and flue gas flow. . . . . . . . . . . . . . . . . . . . . V A.6 Flue gas composition. . . . . . . . . . . . . . . . . . . . . . . . . . . . VI A.7 Conversion efficiencies and losses. . . . . . . . . . . . . . . . . . . . . VII A.8 Energy balances for a black liquor recovery boiler. . . . . . . . . . . . VIII B.1 Operating parameters for dissolver off gas. . . . . . . . . . . . . . . . IX B.2 Calculations for dissolver off gas and NH3. . . . . . . . . . . . . . . . X B.3 Original scrubber effluent. . . . . . . . . . . . . . . . . . . . . . . . . XI B.4 Converted scrubber effluent. . . . . . . . . . . . . . . . . . . . . . . . XI xii 1 Introduction In response to the escalating environmental concerns, industries around the world are constantly improving their operations and developing new strategies to reduce the emissions of greenhouse gases. Some of the most concerning pollutants from the pulp and paper industries are nitrogen oxides, NOx , and sulphur oxides, SOx , which are formed from combustion at high temperatures. Exposure to both nitrogen and sulphur oxides result in direct negative health effects on living organisms and they contribute to the acidification of forests, soil, and waters. NOx also contributes to acidic rain as well as eutrophication when dissolved in water. Considering a conventional kraft paper pulp mill, utilisation of the recovery boiler and lime kiln are the main sources of such emissions as a result of combustion at high temperatures [1]. There have been many improvements regarding the NOx and SOx reduction for industries throughout the years, but there is still a lot of potential to implement technology for emission control measures to meet the future emission regulations. As part of these control measures, emission reduction commitments for the period 2020-2029 and 2030 and forward have been implemented, increasing the regulations on emissions of NOx and SOx within the EU member states. In Sweden, the emission reduction commitments for the period 2020-2029 have been achieved for both NOx and SOx , as well as the commitment for SOx from 2030 and beyond. However, to comply with the commitments for NOx from 2030 and beyond, a further reduction of 40 % compared to the emissions in 2021 is needed cumulatively across all relevant sectors [2]. Incentives such as NOx fees have also been enforced in different energy sectors as a result of the increasing regulations. In Sweden, the fee is 50 SEK per kg of emitted NOx annually, and the fee is repaid based on how much energy each industry produces. This emission penalty is not applied for boilers in the pulp industry today, but new regulations suggest that the energy credited from recovery boilers should be re-formulated from 100 % to 60 % until 2030 and thereby put a price on emitting NOx [3]. 1 1. Introduction There are requirements for the Best Available Technologies (BAT) covered by EU’s Industrial Emissions Directive (IED). The directive sets requirements for emissions to air from kraft recovery boilers that must be within specific emission ranges an- nually [4]. Presently, many paper pulp mills manage to fulfill the requirements for BAT. However, anticipating future regulations proposed by the EU, concerns re- garding the emission commitments have emerged, resulting in increased interest in emission control. One company that has opted to address these growing concerns is Södra Cell in Mönsterås. Therefore, a case study will be conducted at Södra Cell Mönsterås to investigate possibilities for implementing technologies for emission control in their kraft recovery boiler. 1.1 Aim The aim of this project is to quantify the impact of implementing new emission control technologies on the mass and energy balances of a kraft recovery boiler. The project will also construct the heat and mass balance process model and validate it against industrial data and a literature review. The process model is constructed in the software Aspen Plus based on industrial data from Södra’s pulp mill in Mönsterås, Sweden. The emission control technologies evaluated include water-diluted ammonia injection, dissolver off gas injection, and recycling of scrubber effluent. The results from the model include the impact on the high-temperature chemistry occurring in the recovery boiler, correlations between key performance indicators, process changes, and the formation of NOx and SOx . 2 2 Theory In this chapter, a general explanation of the kraft recovery process is provided, as well as a more in-depth description of the recovery boiler, including descriptions of the chemical processes taking place in the different parts of the unit. The general formation routes for NOx and SOx , along with the sodium and sulphur balance in the recovery boiler are also presented in this chapter, followed by the currently available technology for emission reduction. 2.1 The kraft pulp mill and chemical recovery process The kraft process performed at Södra involves transforming wood into wood pulp. Wood pulp consists mainly of cellulose fibers and is one of the main components in paper production. The process is known for its ability to decompose the wood chips into pulp using a mix of white liquor and steam. White liquor is a solution that consists of sodium hydroxide (NaOH) and sodium sulphide (Na2S) [5]. The process include several steps, preparation of wood chips, cooking, washing, bleaching, and drying. The wood chip preparation is the start of the process where the timber, softwood or hardwood, is debarked and chipped into small pieces. The cooking stage takes place in the pulp digester, where white liquor and steam are added. The purpose is to break the bonds between lignin, hemicellulose, and cel- lulose in the wood and remove the lignin by dissolution. After cooking, the pulp is washed to remove the spent cooking chemicals and lignin. The washing plant also addresses chemical recovery, chemical demand in subsequent bleaching steps, as well as obtaining a clean product. The product is then bleached to further remove color and improve brightness. Various chemicals like chlorine, chlorine dioxide, hydrogen peroxide, or oxygen can be used in this step. When the desired result is obtained the pulp needs to be dried to attain the right moisture content. The final wood pulp product can then be used to manufacture paper and other cellulose-based products [5]. 3 2. Theory Furthermore, one of the key features of the kraft process is the high chemical recovery efficiency. Figure 2.1 illustrates a simplified flowsheet of the kraft recovery process that includes the cooking chemical cycle and the lime cycle. This stage of the pro- cess enables the recycling of pulping chemicals (NaOH and Na2S), which minimise chemical consumption, and also creates energy generation due to the combustion of black liquor in the recovery boiler [6]. Figure 2.1: The kraft recovery process. Figure 2.1 shows that during cooking in the digester, black liquor is formed, which contains dissolved lignin, spent cooking chemicals, and other by-products. When washing the pulp the weak black liquor is separated from the pulp and enters the multi-effect evaporators to remove water content and obtain thick black liquor. The recovery boiler burns the black liquor which generates heat and releases sodium compounds consisting mainly of sodium carbonate (Na2CO3) and sodium sulphide (Na2S) which form the so-called char bed or molten smelt. The smelt is then ex- tracted from the recovery boiler and enters a smelt dissolver tank where the sodium compounds are dissolved in water to form green liquor [7]. The green liquor then enters the causticizing plant. Here, the burned lime (CaO) from the lime kiln reacts with water to form Ca(OH)2, also known as ’slaked lime’. The calcium hydroxide is then mixed with the green liquor (Na2CO3) in causticization vessels and the reaction forms NaOH and CaCO3 as can be seen in Equation (2.1) [8]. Na2CO3 + CaO + H2O → 2 NaOH + CaCO3 (2.1) Calcium carbonate is separated from the white liquor and rewashed before being fed back into the lime kiln for reburning, closing the lime cycle. Na2S as being part of the green liquor goes through the whole causticizing process and ends up in the white liquor together with the sodium hydroxide. The white liquor is then recovered to the digester [8]. 4 2. Theory As mentioned, the recycled lime (CaCO3) from the white liquor preparation enters the rotary lime kiln together with makeup (added to compensate for chemical losses) lime mud. The lime sludge is then heated to a high temperature and reburnt into calcium oxide which can be fed back to the causticizing plant. The reaction that takes place in the lime kiln can be seen in Equation (2.2). CaCO3 → CaO + CO2 (2.2) The causticization process requires approximately 250 kg of calcium oxide per tonne of pulp produced. Therefore, the lime cycle is a very important part of the pulp mill process and is essential to avoid the expensive costs of make-up lime [8]. 2.2 The recovery boiler The purpose of the recovery boiler is to use thick black liquor as a source of fuel to produce superheated steam in the top region and to create smelt in the lower region. The heat released from the combustion of the organic matter of the lignin can be generated as superheated steam with the help of superheaters, boiler banks, and economisers. The superheated steam can further be used to power various processes in the mill, generating electricity through a turbine that produces low- pressure steam for district heating. The thick black liquor, with a dry solid content of 65-85 %, is obtained from the evaporation steps and then sprayed into the lower part of the recovery boiler. The molten smelt, containing mainly Na2CO3, Na2S, and Na2SO4 is formed due to combustion in flight when the black liquor is burned in an oxygen-deficient environment [7]. The air supply to the recovery boiler is regulated at different sections of the furnace through primary, secondary, and tertiary air nozzles, as shown in Figure 2.2. In newer boilers, a quaternary air nozzle is common. Each injection serves a different purpose in the boiler due to the intricate chemical process that takes place. Both the efficiency of the reduction and the steam production are greatly affected by the arrangement of the air injection [9]. 5 2. Theory Figure 2.2: The main reactions in a recovery boiler at a kraft process [9]. The recovery boiler is a very complex furnace with several different physio-chemical reactions happening almost simultaneously, both exothermic and endothermic. To simplify the recovery boiler, it is divided into five different sections: steam produc- tion, combustion, drying, pyrolysis, and gasification followed by reduction. Figure 2.2 provides an overview of the main events and reactions occurring within the recov- ery boiler. Initially, black liquor droplets are sprayed into the boiler, where they are exposed to exhaust gases from the bottom of the boiler, leading to the evaporation of the remaining moisture. Subsequently, the dry black liquor droplets continue downward and mix with sec- ondary air and volatiles from the reduction stage, initiating pyrolysis. During this phase, organic-bound sodium and sulphur are released from the droplets and react with water, while inorganic sulphur also undergoes reactions with water. The solids 6 2. Theory break down to form a mixture of gaseous byproducts such as CO2, CO, SO2, CH4, H2O, total reduced sulphur (TRS) volatiles (CH3SH, CH3SCH3, CH3S2CH3, H2S), and a porous particulate material [9]. The key reactions occurring in the pyrolysis are outlined below: Solids → Volatiles + Inorganic material + Fixed carbon (2.3) Na2O + CO2 → Na2CO3 (2.4) Simultaneously, H2S will react with the Na2CO3 to form: Na2CO3 + H2S → Na2S + CO2 + H2O (2.5) Afterward, the porous particulate material, consisting mainly of Na2CO3, Na2S, Na2SO4, and fixed carbon, reacts with the primary air in the bottom region of the boiler. The main reactions can be seen below. C + 1 2O2 → CO (2.6) C + H2O → CO + H2 (2.7) C + CO2 → 2CO (2.8) CO + 1 2O2 → CO2 (2.9) The combustion that occurs in the reactions above, from the fixed carbon and the O2 from the primary air, is exothermic and provides the energy required for the endothermic reactions in the reduction zone. The aim here is to use the remaining carbon to react with Na2SO4 yielding Na2S. Although additional reactions also take place, but to a lesser degree, as shown below. 4C + Na2SO4 → 4CO + Na2S (2.10) 2C + Na2CO3 → 3CO + 2Na (2.11) Na2S + 2O2 → Na2SO4 (2.12) Since Na2S is the desired product, it is important to prevent its reoxidation back to Na2SO4 in order to obtain a high reduction efficiency. Therefore, to prevent the re- action in Equation 2.12, it is essential to maintain an oxygen-deficient environment. The smelt will then consist of a molten salt mixture, mainly composed of Na2CO3 and Na2S. The reduction efficiency (RE) is calculated using the following equation: 7 2. Theory RE(%) = Na2S (mol%) Na2S (mol%) + Na2SO4 (mol%) (2.13) Moreover, higher concentrations of potassium in the black liquor favor the formation of potassium salts in the smelt, such as K2CO3. Usually, potassium content in the black liquor is significantly lower compared to sodium, thus leading to lower amounts of potassium compounds in the smelt. Potassium behaves similarly as sodium in the recovery boiler due to their equal number of valence electrons, and potassium is, therefore, referred to as a sodium equivalent. The molar mass ratio between the two is 0.59 kg Na per kg K [10]. The gases formed during the pyrolysis stage, as well as the volatiles from the lower region will at the same time combust with the O2 from the secondary and tertiary air to produce energy for the steam generation [10]. Carbon monoxide (CO) is another emission produced during combustion processes. It is typically oxidised to CO2, but a small fraction exits as CO in the flue gas. This can result from poor combustion conditions, such as inadequate burning and mixing. Notably, elevated CO values often correlate with reduced oxygen availability during combustion, since this can minimise NOx emissions [4]. The tertiary air is also used for further combustion in order to reduce overall emis- sions, the most common being NOx , CO, and TRS. A quaternary air nozzle could further help reduce unwanted pollutants. In addition, different operating conditions and purification methods will also affect the result. The flue gas leaving the re- covery boiler mainly consists of N2, CO, CO2, H2O, and H2. Besides these main components, it may contain other species such as NOx , SOx , TRS, Na2SO4, HCl, and CH3SH [10]. A smaller fraction of the sodium and potassium salts that are formed in the lower part of the furnace ends up in the flue gas through dust formation, where potassium compounds are more volatile than sodium compounds due to its lower molar mass. Chlorine enriched in black liquor also contributes to the particulate formation of NaCl and KCl, along with gaseous HCl [10]. The chemical interaction between these species will be explained in Section 2.2.2. 2.2.1 Sulphidity, sodium and sulphur chemistry The term ’sulphidity’ in the kraft process can have different meanings depending on the stage of the process being considered. When considering the sulphidity of the entire kraft or ’sulphate’ pulping process, it can be defined as the percentage molar ratio of Na2S to the active alkali (AA), effective alkali (EA), or titratable alkali (TTA), as shown in the equations below: AA = NaOH + Na2S (2.14) EA = NaOH + 1 2 Na2S (2.15) TTA = NaOH + Na2S + Na2CO3 (2.16) 8 2. Theory The sulphidity equation most commonly used is the one with respect to the active alkali (AA), as shown below: sulphidity(%) = Na2S (mol%) NaOH (mol%) + Na2S (mol%) · 100 (2.17) The digester design establishes target values for sulphidity, AA, and TTA to get a desirable product. The sulphidity levels can differ depending on the type of timber used. Hardwood has a lower sulphidity compared to softwood pulping which can affect the quality of the product. Sulphidity does not contribute to increased pulp yield. However, the presence of hydrogen sulphide (HS– ) ions significantly enhances selectivity and delignification rate during kraft pulping without directly protecting the cellulosic material. In addition, due to its significant impact on delignification rate, HS– ions result in shorter cooking times to achieve the wanted kappa number (an indicator of residual lignin content in the pulp). This shortened cooking time reduces exposure to alkali and/or cooking temperatures, ultimately increasing pulp yield. Additionally, the reduced fiber degradation can be interpreted as improved pulp strength [11]. The sulphidity for the black liquor entering the boiler can be described by the variations in total molar content of sulphur and sodium in the liquor, expressed as the S/Na2 ratio. The same expression can also be used to describe the molar sulphur-to-sodium ratio in the flue gas, which will be further explained in Section 2.2.2. High sulphidity, along with the presence of oxidised sulphur-sodium salts such as Na2SO4 and Na2SO3 in the recovery cycle, contribute to an increase in the S/Na2 ratio for the black liquor. In general, the sulphidity levels and S/Na2 ratios of black liquors are often higher in Scandinavian pulp mills compared to North America. The S/Na2 ratio usually lies between 0.35 to 0.5 in Scandinavia while it ranges between 0.2 to 0.3 in North America [10]. The ideal case would be if all the various sulphur and sodium compounds were converted to Na2S and Na2CO3 in the smelt. However, in practice, the process is more intricate. Figure 2.3 shows a simplified illustration of the sulphur and sodium balance in a typical recovery boiler [10]. 9 2. Theory Figure 2.3: Sulphur and sodium balance in a typical recovery boiler. The composition of sodium and sulphur in black liquor can differ significantly across various processes, or even for the same boiler, and this variation affects the sulphur and sodium balances. Figure 2.3 represents an approximate amount of the sulphur and sodium entering the boiler, as well as the balances occurring inside. Initially, the sulphur and sodium from black liquor enter the recovery boiler, and instead of just reacting to form Na2CO3 and Na2S as the ideal case, they are also transformed into Na2SO4 in the smelt [10]. A significant portion of both sulphur and sodium is transported by the combustion gases upwards in the boiler, primarily as Na2SO4 dust and sulphur-containing gases. The majority of sodium and sulphur dust is captured and removed from the flue gases through an electrostatic precipitator (ESP). The captured dust is then blended with the fresh black liquor and reintroduced into the boiler. Finally, a minor fraction of the sodium and sulphur exits the process as emissions in the flue gases. The primary emission consists of Na2SO4, SO2, H2S and CH3SH. Additionally, a make-up stream of Na2SO4 is introduced into the black liquor [10]. 10 2. Theory The vaporisation of sulphur and sodium from the bed depends heavily on the tem- perature in the furnace. For decreased bed temperatures, the release of gaseous sulphur increases linearly while gaseous sodium decreases at a much higher rate. Maintaining a bed temperature of approximately 1000 °C minimises the release for both gaseous sulphur and sodium [10]. 2.2.2 Sulphur oxides and hydrogen chlorine formation routes The formation of sulphur oxides, SOx , mainly stems from the oxidation of fuel-bound sulphur, with the most prevalent species being SO2 [12]. Other sulphur-related emissions include emissions of TRS, which result from incomplete combustion. In the presence of oxygen, TRS is oxidised to SO2 [4]. The formation of the sulphur gases, SO2 and TRS, begins when sulphur enters the recovery boiler through the black liquor in which it mainly exists as inorganic sulphur compounds, with the dominating forms being sulphide and sulphate. About 30-40 % of the sulphur present in black liquor exists as organic sulphur compounds [13]. The sulphate that ends up in the bed of the furnace is reduced by reducing gases (H2 or CO) or char in the bed to Na2S. The sulphate can also be vaporised in the form of H2S or COS and is carried away by the flue gas. In the lower furnace in which there are reducing conditions, sulphur gases exist mainly as H2S and as it moves toward the upper part of the furnace where there are oxidising conditions it reacts with oxygen to form SO2 [10]. The sulphidity in the flue gas has a major impact on the formation of SO2, and the chemistry in the flue gas and dust is almost entirely controlled by this S/Na2 ratio. Due to the presence of residual SO2 in the flue gas, alkali chloride dust resulting from condensation causes partial conversion into sulphate within the upper section of the boiler [10]. However, it will also form hydrogen chloride, which leaves the boiler as flue gas emission, as seen in the reaction below. 2ACl + SO2 + 1 2O2 + H2O → A2SO4 + 2HCl (2.18) As described, in Section 2.2.1, the release of gaseous sulphur to the flue gas is regu- lated by the temperature in the bed, which subsequently also controls the sulphidity in the flue gas. In addition, the reaction above also implies that the S/Na2 ratio in the flue gases is a critical parameter concerning chlorine chemistry as well [10]. Figure 2.4 and 2.5 illustrates the principal reactions for sulphur, alkali and chloride, considering the sulphidity in the flue gas is higher or lower than one. Here, the sodium compounds represent the overall alkali contribution, including potassium, where the gaseous alkali entering the flue gas is presented as NaOH, originating from the lower part of the furnace. 11 2. Theory Figure 2.4: Formation routes for sulphur, sodium and chloride in the flue gas. Cool bed: S/Na2 > 1 [10]. In Figure 2.4, a substantial part of the SO2 is released to the atmosphere, and another substantial part is further oxidised to sulphate which interacts with sodium compounds to form Na2SO4 as a particulate. The amount of chlorine released in gaseous form is relatively small. Conversely, the alkali chloride released will undergo almost complete transformation into hydrogen chloride. A minor part of the SO2 is oxidised to SO3 of which some is emitted to the atmosphere while some interacts with water and sodium sulphate to form sodium bisulphate dust [10]. 12 2. Theory Figure 2.5: Formation routes for sulphur, sodium and chloride in the flue gas. Hot bed: S/Na2 < 1 [10]. In Figure 2.5, the majority of the SO2 is oxidised to sulphate and interacts with sodium compounds to form NaSO4 with small emissions of SO2 into the atmosphere. Therefore, a large amount of alkali chloride is vaporised and will not undergo sul- phation but instead condenses into alkali chloride dust [10]. To summarise, the SOx emissions from the recovery boiler are increased by increasing the S/Na2 ratio in the black liquor as well as lowering the furnace temperature, resulting in a higher S/Na2 ratio (sulphidity) in the flue gas. Vise versa holds for decreased SOx emissions, whereas the vaporised sodium compounds bind almost all SOx in the flue gas into Na2SO4 and are recycled back through the ESP [10]. For higher sulphidity in the flue gas, one notice to make is that the formation of acidic sulphate (NaHSO4), see Figure 2.4, contributes to higher fouling and corroding effects inside the furnace. On the other hand, for cases with low sulphidity in the flue gas, the presence of alkali chlorides particulate in the dust also increases the fouling and corrosion tendency as the dust is more prone to sticking on the furnace’s walls, see 2.5. This holds especially for the particulate formation of KCl. Alkali chlorides lower the melting temperature range in the dust significantly, resulting in more sticky alkali compounds [10]. Another article suggests that increasing the S/Cl ratio in a combustion system may reduce the formation of alkali chloride deposits, which reduces the corrosion in the furnace. However, the degree of alkali chlorine sulfation is not only dependent on the flue gas composition but also on the temperature, regulating the S/Na2 ratio in the flue gas [14]. 13 2. Theory 2.2.3 Nitrogen oxides formation routes Nitrogen oxide emissions, NOx , commonly refer to the formation of NO, NO2, and N2O during combustion, with NO being the most prevalent. The most significant factors of NOx formation are fuel nitrogen content, oxygen availability, the conver- sion ratio of fuel-bound nitrogen, temperature, and residence time in the combustion zone [15]. There are 3 main routes of nitrogen formation when using air combustion in the recovery boiler, thermal NOx , prompt NOx , and fuel NOx . An illustration of the different formation routes can be seen in Figure 2.6. Figure 2.6: Formation routes of thermal, prompt, and fuel NOx . Thermal NOx is primarily formed at high temperatures, around 1350 °C, and can be explained by the Zeldovich mechanism, in Equation 2.19 and 2.20, where N2 is oxidised by atomic oxygen [16]. N2 + O → NO + N (2.19) N + O2 → NO + O (2.20) Other factors that affect the formation of thermal NOx is the flame length due to its effect on the maximum temperature of the flue gas. The contribution of thermal NOx to total NOx emissions can vary significantly depending on boiler conditions, particularly temperature. It has been observed that at lower temperatures, approx- imately 800-900 °C, thermal NOx constitutes only a minor portion of total NOx emissions, primarily formed from the oxidation of nitrogen in the combustion air [17]. However, as temperatures exceed 1300 °C, thermal NOx can become a signifi- cant contributor to overall NOx emissions. 14 2. Theory Prompt NOx is formed during air combustion when N2 reacts with hydrocarbon rad- icals. The presence of radicals in the flame zone leads to the formation of hydrogen cyanide (HCN), N and NH according to Equation 2.21 and 2.22, which in turn may form prompt NOx [16]. N2 + HC → HCN + N (2.21) N2 + H2C → HCN + NH (2.22) This occurs almost universally in combustion and since this mechanism requires un- burned hydrocarbon fragments it occurs early in the combustion process. However, prompt NOx is a relatively small portion compared to the total amount of NOx emissions [16]. The fuel NOx emissions account for almost all the NOx emissions in the recovery boiler and originate mainly from the nitrogen in the black liquor, even though the dry black liquor content can contain just up to 0.15 wt% of nitrogen [18]. The black liquor nitrogen is devolatilised when sprayed into the recovery boiler. This devolatilisation, pyrolysis, occurs in the oxidising zone of the recovery boiler which also has been referred to as ’combustion in flight’ previously. In the initial devolatilisation, approximately 2/3 of the nitrogen is released as volatile nitrogen compounds, either N2 or NH3. The remaining 1/3 of the total nitrogen is bound in the char carbon matrix, commonly in the form of cyanate (OCN– ) as the sodium compounds in the bed are reduced during the combustion. In general, small portions of nitrogen in the char bed are released forming NO when in contact with oxygen during the reducing phase, but most of this nitrogen exits the recovery boiler along with the green liquor. The cyanate in the green liquor is partly converted into NH3 when leaving the boiler and entering the dissolving tank. NH3 in the oxidising zone can either form N2 or NO. Therefore, the total nitrogen emissions from the recovery boiler are either in the form of NO or N2 where the share of NO released depends on the operation condition in the recovery boiler. These conditions will be explained further in Section 2.3.1. As the NO reaches the atmosphere, the gas is almost fully converted into NO2 [17]. A schematic illustration of the black liquor nitrogen formation routes is shown in Figure 2.7 Figure 2.7: Formation routes of fuel NOx . 15 2. Theory 2.3 Current BAT and primary control measures of NOx and SOx The current Best Available Technologies (BAT) requirements and an overview of the primary control measures for NOx and SOx emission from the recovery boiler will be presented in this section. The control measure techniques are mainly based on the BAT recommendations for power boilers, but the techniques can principally be applied to recovery boilers from kraft pulp mills as well. As explained in the introduction, requirements for BAT set the standard for emis- sions to air from kraft recovery boilers in the EU. These emission reference values are presented as ranges due to varying operating conditions among relevant kraft recovery boilers, resulting in higher or lower overall emissions [4]. These ranges are presented in Table 2.1, both as concentration per normal cubic meters of flue gas and kg per ton air-dried mass (ADt). The concentrations are given at standardised reference O2 content 6 % in flue gas. Table 2.1: Yearly BAT requirement ranges of emissions for kraft recovery boilers [4]. BAT requirements mg/Nm3 at 6 % O2 kg/ADt SOx (as S) 1-100 <0.002 - 0.65 NOx (as NO2) 120 - 250 0.73 - 2.0 TRS or H2S (as S) 0 - 50 0.0007 - 0.40 According to the recommended BAT technologies, for a kraft recovery boiler, emis- sion control within the given ranges is achievable by for example computerised com- bustion control with good mixing of the black liquor and air while also implementing a staged air system [4]. Other techniques like selective catalytic reduction (SCR) have also been proposed as possible NOx reduction measures but are seen as emerg- ing technologies and not the best available technology [4]. Moreover, SCR is not considered a primary measure and will therefore not be discussed here. In recovery boilers, CO concentrations can range from 10 mg CO/Nm3 to 100 mg CO/Nm3 on an annual average, with some mills exhibiting even higher levels. How- ever, high CO concentrations pose safety hazards and can also contribute to cor- rosion of the furnace walls. Currently, there are no BAT requirements specifically addressing CO emissions for recovery boilers. Nevertheless, it remains crucial to maintain a balanced approach, ensuring that both CO and NOx levels are kept within optimal ranges for a safe operation [4]. 16 2. Theory 2.3.1 Air staging Air staging is a technique in which NOx emissions reduction is achieved by intro- ducing two or more combustion zones where the stoichiometric relationship between fuel and air is varied between the zones [12]. For a kraft recovery boiler at stable operating conditions, the black liquor is referred to as the fuel. In a system with four combustion zones, the primary zone would have a substoichiometric relation- ship between fuel and air to limit the conversion of fuel-bound molecular nitrogen to NOx as well as limiting the formation of thermal NOx by lowering the combustion temperature. While in the secondary zone, the relationship between fuel and air is above stoichiometric conditions to achieve burn out of the fuel while also increasing the volume of gas resulting in a lower temperature and limits thermal NOx formation [16]. Ideally, the hydrocarbons in the black liquor are combusted so that all carbon monox- ide is oxidised into carbon dioxide. But, this is not the case for a recovery boiler, where the carbon has to be preserved in the char bed for reduction of Na2SO4. To compensate for the incomplete combustion in the lower region of the recovery boiler, tertiary and occasionally quaternary air supply above the pyrolysis zone is necessary to oxidise the remaining CO [16]. Lowering the amount of excess air by regulating air flows will, on the other hand, reduce NOx emissions since this limits the oxygen available, thereby reducing the amount of fuel-bound nitrogen converted to NOx and to some extent the formation of thermal NOx . Finding the optimal distribution of air through all air stages is therefore crucial to maintain low CO emissions while limiting the formation of NOx [12]. For a conventional kraft recovery boiler, the formation of NOx originates mainly from NH3 in the oxidising zone and less from the char bed, as described in Section 2.2.3. Implementing air staging and changing the size of injected black liquor droplets in the spray nozzle are two aspects that have been further investigated by Åbo Akademi University [17]. By making the droplets finer, a larger proportion of the droplets will have time to burn out before reaching the bed surface, resulting in significantly more formation of NO from the char nitrogen. Thus, less nitrogen will be bound to the smelt. Introducing six air stages with coarser droplets will not affect the amount of char nitrogen formed, but instead decrease the formation of NO greatly as the oxidation of NH3 is limited. The conclusion was that this case in particular formed the least amount of NO. The total formation of NO was significantly higher when combining air staging with finer droplets, but arguably less than for a conventional boiler. Another conclusion was that the NOx formation decreased as the number of air stages increased, given that the supplied air is distributed more evenly in the recovery boiler. Also, having increased temperatures ranging from 800 °C to 1000 °C in the oxidising zones increased the formation of NO significantly [17]. 17 2. Theory 2.3.2 Ammonia injection Another method that has been reported for the reduction of NOx emissions is to inject ammonia diluted in water, into the recovery boiler. By adding NH3 to the exhaust gas as an active reducing agent, the present nitrogen oxides can be converted into N2. 4 NO + 4 NH3 + O2 → 4 N2 + 6 H2O (2.23) For this selective non-catalytic reduction (SNCR) to work properly, the NH3 has to be dosed in an above-stoichiometric ratio in relation to the continuously measured NOx target value. The amount injected should be somewhere between 2.1-3.3 mol NH3 or 1.1-1.6 mol urea per mol NOx . The performance of this process is strongly dependent on the mixing conditions between the ammonia and NOx in the exhaust gas. The reduction reaction is also dependent on an optimal temperature interval in the boiler. Injecting ammonia at temperatures below 900 °C results in lower conver- sion and fewer portions of NH3 will leave the boiler unreacted, so-called ’ammonia slip’. At higher temperatures, above 1000 °C, the oxidation of NH3 into NOx is increased [12]. This SNCR method has successfully been used in power boilers and bark boilers, but the applicability of this in recovery boilers has not been investigated as much. However, one study has shown that injection of ammonia into a recovery boiler with a spruce-based black liquor load of 132 ADt/day gave a reduction of NOx up to 30 %. A solution of 10-11 % ammonia water was injected into the chamber at 910-920 °C. The dosage was 50 l/h which corresponded to approximately 1.5 ml of NH3/Nm3 exhaust gas. At the given dosage, the results showed that no relevant ammonia slip was determined. One difficulty in using this type of SNCR is the varying temperature zones in the recovery boiler and it is still unclear where the injection ports should be placed for the most efficient results, especially when the recovery boiler load is altered. Improvements to the ammonia injection system can be made by having an adequate number of injection ports for increased penetration and mixing conditions with the exhaust gas. Having sufficient residence time for ammonia to react is also a crucial aspect [12]. Tests with SNCR for NOx reduction in a kraft recovery boiler have also been carried out by Metso Power based on a 2000 ADt/day recovery boiler built in the year 2001. The amount injected varied between 0.8-1.7 mol NH3 per mol NOx depending on load ranges over 60 – 100 %. At full operating load, the NH3/NOinitial ratio was 1.3 and there were four injection points just below the nose of the boiler, above the quaternary airports. The results showed a reduction of NOx for all cases, where a 45 % reduction at full load was achieved with limited ammonia slip (17.5 NH3 slip mg/Nm3) and no or only a minor impact of other emissions. It was stated in the study that it was hard to reveal any trends between the NH3/NOinitial ratio and the reduction. One explanation could be that the rate of reduction is heavily dependent on conditions in the injection region. For example, the temperature differed from 830 °C to 925 °C upon measurement at the injection region [19]. 18 2. Theory There are some safety hazards associated with ammonia injection. Managing am- monia slip is complicated due to limited feedback for reagent control, as well as challenges in determining optimal placement. Furthermore, SNCR systems in bark boilers face difficulties with rapid load changes, leading to variable NOx reduction ef- fectiveness and unstable performance under fluctuating operational conditions. An- other potential hazard arises from unreacted ammonia reacting with SO3 to produce ammonium bisulphate, which can precipitate at air heater temperatures, causing fouling and potential plugging [12]. It can also be safety risks when injecting water into a recovery boiler. If the water comes into contact with the smelt, an explosion can occur. This may be due to the rapid conversion of water into vapour with a very large volume or to the de- composition of water into hydrogen and oxygen, which then ignites. It is therefore important that the water is evaporated before it reaches the smelt [20]. 2.3.3 Flue gas recirculation, fuel staging and reduced air preheat Flue gas recirculation is a technique in which NOx formation is reduced by recircu- lating flue gas into the combustion zone. This limits oxygen availability while also reducing the temperature of the flame, which reduces the conversion of fuel-bound NOx and the formation of thermal NOx [12]. Fuel staging is a technique similar to air staging in which the stoichiometric relation- ship between fuel and air is altered. The objective of this technique is, contrary to air staging, not to limit the formation of NOx , but rather to support the conversion of formed NOx back to molecular nitrogen. This is done through the creation of three combustion zones, in the primary zone a majority of the fuel is burnt out with excess air. In the secondary zone, more fuel is added which creates a substoichio- metric atmosphere in which hydrocarbon radicals are formed which in turn react with NOx and reduce the NOx back to molecular nitrogen. The combustion of the fuel is then completed in the third zone by supplying excess air [12]. Other measures that can be taken to reduce NOx emissions are combustion with reduced air preheat temperature, water injection, and installing low NOx burners [12]. Reducing the air preheat or injecting water into the recovery boiler reduces the flame temperature and thereby reduces the amount of thermal NOx formed [12]. Low NOx burners apply the principles of either, air staging, flue gas recirculation, fuel staging, or a combination of the three mentioned techniques to achieve reduced NOx emissions [12]. 19 2. Theory 2.3.4 Primary control measures for SOx The recommended BAT measures for SOx removal from the recovery boiler is to increase the dry solid content and/or decrease the sulphidity of the black liquor. In- creasing the dry solid content of black liquor reduces the SOx emissions by increasing the temperature in the furnace. As described earlier in Section 2.2.2, higher tem- peratures in the bed limit the release of gaseous sulphur and therefore reduces the sulphidity in the flue gas, resulting in lower SO2 emissions. This allows more sulphur to react with sodium in the bed to form Na2CO3. There are, however, some ad- verse effects of increasing the dry solid content as higher temperatures might cause increased emissions of NOx . Decreasing the sulphidity of the black liquor ensures that a majority of sulphur present can react with sodium, limiting the amount of SOx that can be formed [4]. 2.4 Wet flue gas treatment with NO oxidation A study on the coabsorption of SOx and NOx techniques, with NO oxidation from a gas-fired furnace flue gas, was done through technical-scale experiments and sim- ulations. The study indicated that the technique can achieve >90 % NOx removal with > 99 % SOx removal. In this technique, an oxidative agent such as O3, H2O2 or ClO2 is used to oxidise NO to NO2 is injected into the flue gas to make the nitro- gen soluble in water. SO2 is adsorbed spontaneously to water with solubility being limiting. When dissolved, SO2 reacts with water to form bisulphite which forms sul- phite through a dissociative equilibrium reaction which is weighted towards sulphite under alkaline conditions. Sulphite then interacts with the dissolved NO2 through hydralisation which takes place at a comparable rate to that of SO2 absorption to water. This allows for efficient co-removal of NOx and SOx in a wet scrubber like equipment used for SOx removal [21]. This method is heavily dependent on the coexistence of sulphur and nitrogen in the flue gas to achieve high removal efficiency. In the case of combustion of sulphur deficient fuel, Na2SO3 can be added in the scrubber liquid to increase the removal efficiency, however, this is associated with increased operational cost [21]. One po- tential way to reduce the amount additional Na2SO3 to the scrubber liquid could be recycling the scrubber effluent back to the recover boiler. This has been investi- gated in a 25 MWth waste-to-energy power plant. In the study, it was found that recirculating the scrubber effluent would reduce the NOx formation by the addition of nitrites and it is also discussed that recirculated sulphates from the effluent can be reduced to SO2 by alkali sulphate decomposition under the right circumstances [14]. 20 2. Theory However, no increase of SO2 in the flue gas was observed during the experiment which can be attributed to the temperature being too low or the environment not being reducing enough to achieve decomposition at the point of injection. If decom- position of alkali sulphates can be achieved, recirculation of scrubber effluent might be an efficient way to reduce the operational cost of the equipment as it lowers the formation of NOx . This would, in turn, reduce the amount of sulphite needed for absorption while also increasing the SO2 content in the flue gas, thereby decreasing the amount of Na2SO3 needed to be supplied to the absorption liquid. Another ben- efit is that if the scrubber effluent can be recirculated to the recovery boiler, it will decrease the amount of effluent that has to pass through the wastewater treatment [14]. Ideally, implementation of the scrubber would mean that a sulphur cycle would be possible to introduce to the kraft process as depicted in Figure 2.8. Figure 2.8: The kraft recovery process with recycling of scrubber effluent. 21 3 Method The method of this work is illustrated by a methodology flowchart, shown in Figure 3.1. The literature study provides theory of the recovery boiler and its chemistry. Data collection took place at Södra in Mönsterås, involving interviews, plant tours, and data gathering. The recovery boiler is modeled based on the collected data using Aspen Plus V14. The software was chosen for its capability to simulate mass and energy balances under defined conditions and handle unconventional materials like black liquor. Data processing and modelling of the recovery boiler involved iterative adjustments based on data insertion, assumptions, simulation runs, and troubleshooting. Im- provements to the model were performed based on input and output control values. The model was validated by comparing simulated results in Aspen Plus with Södra’s data and the literature through mass and energy balances in Excel, followed by a sensitivity analysis of the finalised base case model. Primary reduction measures were incorporated into the model and were assessed through changes in the mass and energy balances and some key performance indi- cators (KPIs). Despite limitations in reaction kinetics, the model provided insights into the effects of different measures on emissions, aiding the discussion. 22 3. Method Figure 3.1: Methodology flowchart for this thesis. 23 3. Method 3.1 Collecting plant data The data presented in this section were collected from the process at Södra Cell Mönsterås. Both internal and external measurements and reported data were pro- vided under stable and normal operating conditions. Although certain data was sourced from 2022, it remains applicable as there have been no changes in the oper- ation of the boiler since that time. Collecting all relevant data during the visit and getting familiar with the different inlet and outlet streams regarding the recovery boiler was of great importance for modelling accurately. 3.1.1 Södra’s recovery boiler Figure 3.2 presents a simplified flowchart of the recovery boiler at Södra Cell Mön- sterås. It illustrates the pathway of the flue gases through the boiler, followed by entry into the ESP. Here, a portion exits as exhaust gases through the chimney, while the remainder is recirculated as dust. The dust is subsequently blended with weak black liquor before being evaporated with make-up. This results in thick black liquor entering the recovery boiler just above the secondary air. Additionally, the flowchart indicates the entry points for all air injections, as well as the CNCG and DNCG. Furthermore, it shows the locations of feedwater injection and steam exit from the boiler. Figure 3.2: Schematic flowchart of the recovery boiler at Södra Cell Mönsterås. 24 3. Method 3.1.2 Inlet data The pulp produced at the site is mostly spruce and pine-based timber. Therefore, only data regarding softwood-based black liquor was collected. Table 3.1 shows the yearly average elemental composition of the black liquor from laboratory tests, the lower heating value (LHV), and its S/Na2 ratio. The sample was taken just before the black liquor was sprayed into the boiler, which includes the Electrostatic Precipitator (ESP) dust and make-up chemicals to the total composition. Table 3.1: Elemental composition of thick black liquor (dry basis) [22]. C H O N S Na K Cl S/Na2 LHV wt% wt% wt% wt% wt% wt% wt% wt% mol% MJ/kg 31.4 3.71 37.74 0.06 4.59 20.34 2.0 0.16 0.32 12.45 Other inlet data was collected from a previous design specification study performed on Södra’s recovery boiler. Since the design specification study also provided neces- sary outlet data for the boiler, the study was partly used as a reference case when running the Aspen model with the corresponding input data. The study assumed a guaranteed liquor load of 4330 ton dry liquor per day (75 % dry solids content). Table 3.2 shows the inlet rates and temperatures of black liquor feed and air supply entering the boiler, where the diluted non-condensable gas (DNCG) was assumed to have the same composition as air. The flow and temperature of feedwater for steam production in the recovery boiler are also presented in Table 3.2. Table 3.2: Inlet flows and temperatures [20]. Inlet Flows Temp [°C] Black liquor feed [kg/s] 66.82 141 Feedwater for steam [kg/s] 174.0 124 Primrary air [Nm3/s] 47.71 165 Secondary air [Nm3/s] 59.80 165 Tertiary air [Nm3/s] 11.01 30.0 DNCG [Nm3/s] 11.90 90.0 Quaternary air [Nm3/s] 31.98 30.0 Total air [Nm3/s] 162.4 On site, before the black liquor enters the boiler, volatile organic compounds (VOCs) are released from heating and pressurising the black liquor in the evaporator steps. Along with methanol, a series of sulphurous gases are also formed under these conditions, mainly H2S, methyl-mercaptan (MM), dimethyl sulphide (DMS), and dimethyl disulphide (DMDS). These concentrated non-condensable gases (CNCGs) are then injected into the recovery boiler, where the gas is combusted. In Table 3.3, the calculated mass flows of CNCG compounds entering the boiler are listed, based on data from 2022-04-27, when the daily production was 2200 ADt, which is the desired production rate. 25 3. Method Table 3.3: CNCG inlet flows and temperature [20]. Compound Units H2S 7.5 g/s MM (CH4S) 60 g/s DMS (C2H6S) 35 g/s DMDS (C2H6S2) 1.7 g/s Methanol 52 g/s Total flow 0.16 kg/s Temperature 84 ℃ 3.1.3 Reference data Collecting relevant reference data was essential for comparison with the results from the Aspen simulations. Data from the design specification study provided insights into how the temperature was distributed in the boiler and what target temperatures were reasonable for each process operation in Aspen. These data are listed in Table 3.4, where the values were calculated between the air stages inside the recovery boiler. Table 3.4 also shows the compositions of O2 and CO provided by the same study in these regions. Table 3.4: Distribution of temperatures and O2 in the boiler based on design specifications [20]. Region Temp [°C] O2 dry [mol%] Between primary and secondary 1596 1.6 Between secondary and tertiary 1455 0.0 Above tertiary 1301 0.7 Above quaternary 1084 3.1 Nose level 1096 2.5 The monthly average concentrations of sodium compounds in the smelt from Sö- dra’s recovery boiler are presented in Table 3.5 (2024-03). The green liquor sample was dissolved in weak liquor prior to the measurement. Based on the total titrat- able alkali (TTA) and each compound concentration, the weight percentages were calculated, assuming that the compounds presented in the table were the only ones present in the smelt. The reduction efficiency (RE) was approximately 94 % during the measurement period. 26 3. Method Table 3.5: Concentrations and compositions of dissolved green liquor [22]. Smelt compound g/l wt% Na2S 52.0 31.5 Na2CO3 110 66.5 Na2SO4 3.31 2.00 TTA 168 100 Södra’s flue gas composition was measured by two consulting companies, ’DGE’ and ’ILEMA Miljöanalys AB’, in February 2023 and 2019, respectively. DGE conducted their measurements over a single day, while ILEMA’s measurements were taken over three days. However, both were conducted with the measurement point located inside the chimney of the recovery boiler. The load was approximately 4000 tons of dry liquor per day during the measurements. Some of these flue gas species are listed in Table 3.6, along with the measured flue gas flows and temperatures. The flow and average compositions by volume percentage presented in the table were calculated on a wet gas basis. It should also be noted that the CO levels were higher at times during the measurement period, exceeding 1000 ppmv at most. Therefore, the level of CO is presented as an acceptable range due to frequent fluctuations during the measurements. Table 3.6: Measured flue gas data carried out by DGE and ILEMA [20]. Flue gas data DGE ILEMA Units O2 3.10 2.45 vol% H2O 18.4 20.9 vol% CO2 10.3 12.7 vol% CO 100 - 500 150-550 ppmv NOx (as NO2) 70.0 - ppmv SOx (as SO2) < 1.00 - ppmv HCl < 2.50 - ppmv Flue gas flow 241.7 220.1 Nm3/s Flue gas flow dry 197.3 173.9 Nm3/s Flue gas temperature 195.1 200.1 ℃ The reported yearly emission data for sulphur and NOx emissions from the recovery boiler were also provided in Södra’s environmental report (2023), as shown in Table 3.7. The concentration levels are given at a standardised reference O2 content of 6 % in flue gas for softwood, and the values are comparable to the BAT requirements from Table 2.1. 27 3. Method Table 3.7: Yearly emission values from Södra’s recovery boiler compared to BAT requirements [22]. Södra Cell Mönsterås Södra Södra BAT BAT mg/Nm3 kg/ADt mg/Nm3 kg/ADt SOx (as S) 0 0 1-100 <0.002 - 0.65 NOx (as NO2) 141 1.10 120 - 250 0.73 - 2.0 TRS or H2S (as S) - 0.10 0 - 50 0.0007 - 0.40 Relevant data regarding steam generation were provided for the period 2024-03, as shown in Table 3.8. The values were obtained from Södra’s real-time measurement data, where the heat-to-steam and total heat input to the boiler are calculated values based on the mass and energy module of the recovery boiler. Table 3.8: Reference data for steam generation [22]. Steam Generation Units Steam production 174 kg/s Steam temperature 480 ℃ Steam pressure 61.0 bar Heat load to steam 460 MW Total heat load input to boiler 637 MW 3.2 Process modelling The recovery boiler was modeled in a similar way as described in Section 2.2, by dividing it into different parts. Additionally, to model this complex chemical process, some assumptions were made to simplify the recovery boiler. 3.2.1 Assumptions and thermodynamic model The modelling of the recovery boiler was based on certain assumptions. These assumptions, derived from a literature review, are listed below [9]. • The process is kinetic-free and operates at steady-state conditions. • The model is zero-dimensional, with output variables generated based on input variables. • The model only considers chemical and thermodynamic equilibrium. • The recovery boiler operates at 1 bar. • N2 was considered as inert in the RGibbs reactors. • Dust formation in the boiler was not considered. 28 3. Method The thermodynamic model used for this design was the Peng Robinson-Boston Mathias (PR-BM) equation of state, employed to compute the properties of various components throughout the entire process. This combination of the Peng Robinson and Boston Mathias equations makes it suitable for both pyrolysis and gasification of biomass, as it can handle hydrocarbons and light gases [9] [23]. 3.2.2 Design specifications of the model The initial goal of modelling the recovery boiler in Aspen was to match the mass and energy balances with Södra’s industrial data. The recovery boiler functions as a chemical reactor, facilitating the recovery of inorganic compounds as smelt. Due to the intricate nature of black liquor combustion within recovery boilers, the model only incorporated the crucial steps of this process. This section will specify each part of the journey of black liquor and volatiles through the recovery boiler in the model. Figure 3.3 provides an overview of the model structure and outlines the different simulation sections as hierarchy blocks. 29 3. Method Figure 3.3: An overview of the recovery boiler modelling structure. To clarify, the thick black line is the path of the black liquor solids through the recovery boiler going downstream, where part of it ends up in the smelt, and the red lines represent the flow of the volatiles going upstream, exiting the boiler as flue gas. The blue lines are the different air flows into the boiler, with CNCG and DNCG added at corresponding air levels. The dashed blue line represents the inlet of the feedwater entering the steam generation section and exiting as steam (dashed red line). Figure 3.4 to 3.8 will further illustrate in depth how the model was constructed for each hierarchy block. 30 3. Method To facilitate the properties of the black liquor in Aspen, the liquor was defined as a non-conventional solid. The compositions from Table 3.1 were inserted into the ultimate analysis. Since the inorganic compounds could not be inserted into this analysis, Na and K were defined as conventional solids. To ensure a component frac- tion of 1 for the ultimate analysis, the fractions were divided with a normalisation factor of 0.7766, which refers to the difference between the total composition and the fraction of the inorganic compounds. Since the compositions in Table 3.1 are calcu- lated on a dry basis, the moisture content had to be defined in the non-conventional proximate analysis. The first step was to model the drying section, shown in Figure 3.4. The inlet stream of black liquor enters the RStoic unit (DRYING) to simulate the drying process by using stoichiometry reaction modelling. The black liquor particles were then exposed to volatiles from the lower part of the boiler (V7), which caused the rest of the moisture to evaporate. Figure 3.4: Process flow diagram of the drying section, where the wet black liquor (B) entering the boiler and is completely dried by volatiles (V) traveling upwards. Given that black liquor was considered a non-conventional component, Aspen as- sumed its molar mass to be 1 g/mol. With the molar mass of water being 18.02 g/mol, the stoichiometry of the reaction indicated that 1 mol of wet black liquor re- acts to produce 0.055 mol of water. Equation 3.1, therefore, displays the evaporation of moisture that simulated the drying extent in the RStoic. Wet Black Liquor → 0.055 H2O (3.1) 31 3. Method It was assumed in this model that post-drying, the black liquor became entirely dry. After that, the DRY-SEP unit separated the formed water vapor (V8) and the dry black liquor droplets (B2). The SSplit unit (T-EQ1) ensured thermal equilibrium between the dry black liquor, transitioning to the pyrolysis section, and the volatiles originating from the lower part of the boiler (V6). Figure 3.5 illustrates the course of events for the dry black liquor, entering the py- rolysis region. The liquor was mixed with volatiles (V1) from the reduction section through the T-EQ2 before entering the DECOMP unit. Here, the non-conventional components in the liquor were decomposed into a homogeneous mixture of the el- emental substances C, H2, N2, Cl2, S, O2, H2O, and ash. The DECOMP unit is based on the RYield model where the yields had to be specified based on the black liquor ultimate analysis. Since the inorganic solids (compounds of Na and K) were present in the black liquor but not decomposed, the yields based on the ultimate analysis were multiplied by the normalisation factor before being specified in the RYield model. Summing up all yield fractions, including the inorganic fraction, resulted in a total yield of 1. The black liquor then entered the PYRO-REA unit which was based on the RGibbs model that performed the pyrolysis, see Figure 3.5. This model uses Gibbs free energy minimisation method to determine the distribution and composition of the multi-phase mixture, whereas mass and energy balances were coupled based on additional incoming streams of CNCG, volatiles from the reduction section (V2), and secondary air. The purpose of splitting the secondary air (SEC-SPLIT) was to ensure incomplete combustion of carbon in the PYRO-REA unit by limiting the oxygen available. Besides the volatiles released during pyrolysis, described in Section 2.2, the inorganic salts together with the fixed carbon were separated by the PYRO- SEP unit, exiting the pyrolysis zone (B7). The PYRO-SEP unit continued to drive all released volatiles upward in the boiler where the rest of the secondary air was used to oxidise more of the combustion gases through the FLAME unit (RGibbs), especially CO. 32 3. Method Figure 3.5: Process flow diagram of the pyrolysis section, where the dry black liquor is decomposed into volatiles, inorganic compounds and fixed carbon. The volatiles traveling upwards through the pyrolysis section are combusted in the FLAME unit with secondary air. After the pyrolysis of the dry black liquor droplets, the inorganic porous particulate material and the fixed carbon reacted together with the primary air in the RGibbs reactor (GAS & RED). The main gasification reactions that occurred is shown in Equation 2.6 to 2.9. These reactions provided the necessary energy for the endother- mic reactions to take place in the reduction zone, described in Equation 2.10 to 2.12. Achieving high reduction efficiency in the smelt depends on preventing reoxidation of Na2S to Na2SO4. Thus, maintaining an oxygen-deficient environment is crucial for this process. To achieve this, an optimisation tool was used to iterate the necessary amount of primary air required to minimise the deviation between the simulated and targeted reduction efficiency (RE). Equation 2.13 was used to achieve this while the primary air served as the model output, which could be utilised for model validation. The last step was to separate the formed volatiles (V1) and the smelt, mainly composed of Na2CO3, Na2SO4, and Na2S, in the char bed region. 33 3. Method Figure 3.6: Process flow diagram of the gasification and reduction section. The primary air provides the optimal amount of oxygen for gasification and reduction, resulting in release of volatiles (V1) and a smelt leaving the boiler. The gases produced from the gasification in the lower part of the boiler then contin- ued to the pyrolysis section to be combusted with secondary air. Since the burning process may not be complete, extra combustion was needed to burn any remaining carbon monoxide and sulphur gases. Figure 3.7 illustrates the additional tertiary and quaternary air that entered the two RGibbs unit for combustion (COMB1 and COMB2). A stream of dilute non-condensable gases (DNCG) was injected into COMB1 to imitate the real process. The tertiary and quaternary air flow was also optimised for additional combustion to reduce overall emissions, where the quaternary air was further aiding in reducing unwanted pollutants. After the combustion zone, the flue gas continues to travel to- wards the steam generation higher up in the recovery boiler, where the heat released from the flue gas was utilised to produce steam. 34 3. Method Figure 3.7: Process flow diagram of the combustion section (above pyrolysis). Flue gases in the volatile stream are combusted with DNCG, tertiary air and quaternary air. The steam generation occurs in the upper section of the recovery boiler, where the liquid feedwater is transformed into super-heated, high-pressure steam. This convective heat transfer is achieved through the exchange of heat between the flue gases and the water-steam system. Figure 3.8 describes the steam generation process in the model. The dashed blue line represents the feedwater entering the boiler while the dashed red line represents the steam produced. Additionally, the solid red lines indicate the path of the flue gas. 35 3. Method Figure 3.8: Process flow diagram of the steam generation section. The flue gas (red line) exchanges heat to the steam-water system (dashed line) before leaving the boiler. In comparison to real steam generation, where the heat formed is captured by tube wall heat exchangers along the boiler and then used to heat up the steam, this process is a bit simplified but uses the same idea. In order to imitate the heat exchanger tubes, heat from all the different sections was ’taken’ to achieve the same temperatures as the design specification of the boiler (Table 3.4) and was then inserted into a heat exchanger (HX). By doing this, the volatiles were heated to the design specific temperatures before entering the heat exchanger (WALL) which represents all the heat transfer from all tubes. Therefore, as the gases ascend, they undergo heat exchange with the furnace walls, in the WALL unit. Subsequently, the gases then proceed even higher in the furnace, entering the super-heater (SUPER-HE), where the saturated steam is super-heated. Lastly, the gases flow into the economiser (ECONOM), where the liquid feedwater undergoes preheating, before exiting the boiler. Since the model uses equilibrium-based reactors, the amount of CO in the flue gas was below 1 ppmv due to complete combustion and mixing in the RGibbs reactors. However, this is not the case in Södra’s recovery boiler, therefore, to match the level of CO for better validation results, the temperature was increased in the last combustion reactor (COMB2), see Figure 4.4. Since no reactions occurred after the COMB2 reactor, the composition in the flue gas stream was not affected by the temperature decrease in the steam generation section. 36 3. Method 3.3 Model validation To validate the model, mass and energy balances were calculated in Excel according to a literature study [16], based on relevant industrial data collected and used as input for the calculations. The results from the balances were then compared with the simulated result from Aspen as well as the reference data from Södra’s recovery boiler. The main assumptions and considerations made during the mass and energy balance calculations were: • The calculated mass flows were based on 50.115 kg of black liquor solids (BLS), referred to as the total black liquor inlet flow of 66.82 kg/s. • The air flows and flue gas mixture were assumed to be ideal when calculating normal volumetric flows (Nm3/s at 0 °C and 1 atm). • The carbon content in the smelt was 1 wt% of the total smelt. • When calculating the carbon and sulphur losses in the flue gas, mean values of CO and SOx (dry basis) from the DGE measurement (see Table 3.6) were used. The TRS flow in the flue gas stack was obtained from Södra’s environmental report (see Table 3.7). • It was assumed that 10 wt% of the initial sodium would end up in the dust recycle. • The particulate loading was 0.0002 kg/Nm3 dry flue gas in stack. • The steam generation was assumed to have no mass losses (water in was equal to steam out). Blowdown and sootblowing steam flows were also neglected. • Radiation losses were 0.3 % of the total heat input. • Unassociated losses were estimated to be 1.5 % of the total heat input. All comparisons for the validation will be shown in Section 4.1 and the calculations can be seen in Appendix A. A sensitivity analysis of the base case model was performed as part of the validation to verify the stability of the model, as well as finding correlations between key performance indicators such as the composition and temperatures in the upper part of the furnace. Effects on the reduction efficiency and flue gas temperature were also investigated by altering the primary and secondary air flows for the base case model within reasonable operating conditions. The deviations provided from the analysis were plotted and presented in the results. 37 3. Method 3.4 Implementation of primary measures 3.4.1 Ammonia injection Considering the results from the 2000 ADt/day kraft pulp mill study on water- diluted ammonia injection into a recovery boiler (see Section 2.3.2), a similar imple- mentation was investigated in this work. This was done using the given specifications from the study as input data in the model, including the reduction rate of NOx that was 45 %, and calculating the mass flows of water and ammonia based on the initial NOx , see Table 3.9. Table 3.9: Ammonia injection inlet data. Inlet data Units Temperature 20.0 °C NH3/NOinitial 1.30 NH3 0.0167 kg/s NH3 in water 10.5 mol% H2O 0.151 kg/s Figure 3.9 illustrates the selective non-catalytic reduction (SNCR) process occur- ring in the model when inserting the mix of ammonia and water into the recovery boiler. The ammonia mix is inserted in between the quaternary air in the combus- tion section, Figure 3.7, and the steam generation, Figure 3.8, due to the narrow temperature interval described in Section 2.3.2. The stoichiometric reactor (SNCR) in the model will represent the SNCR reaction, Equation 2.23, happening in the boiler. Figure 3.9: Model integration of the ammonia injection. 38 3. Method By implementing the water-diluted ammonia injection into the model for NOx re- duction as an SNCR method, the effects on the operational conditions of the boiler were examined and are presented in the results. Södra was also interested in investigating the possibilities of injecting the so-called ’dissolver off gas’ into the boiler. The idea was that the ammonia-containing gas could possibly provide an SNCR-type effect for NOx reduction when fired into the boiler, at a similar injection point as the water-diluted ammonia injection. Earlier studies done by ÅF at several Swedish mills had shown no increase in NOx emissions for this type of gas firing, but there was also no evidence of an SNCR-effect. As described in Section 2.2.3, the cyanate formed in the green liquor is partly converted to NH3 when entering the dissolver tank. Ammonia is then removed from the dissolver by two identical fans by adding a flow of air. The nitrogen content in the smelt was not measured on site, therefore, the NH3 formed from the green liquor was approximated based on the assumption that half of the cyanate in the green liquor formed NH3 in the dissolver tank. Such high conversion rates in the dissolver were uncommon based on the study by ÅF [24]. Through this approximation, an upper bound of the NH3 availability from the dis- solver off gas on site was set. Some of the dissolving off gas inlet data collected from Södra is shown in Table 3.10. Calculated flows of NH3 in the dissolving off gas and the NH3 concentration in the flue gas when injecting the gas are presented in Appendix B.1. Table 3.10: Dissolver off gas inlet data [20]. Inlet data Units Temperature 341 °C Total mass flow 13.3 kg/s Air flow 11.3 kg/s H2O in air 2.03 kg/s After calculating the approximated amount of NH3 in the dissolver off gas, a similar implementation to the water-diluted ammonia injection was performed for the gas injection in the model, and the operational conditions of the boiler were compared between the two methods. The dissolver off gas was injected into the COMB2 at the quaternary air level, where the exhaust gas temperature is at its lowest before entering the steam generation region. Two different cases were simulated for this injection. In Case 1, the dissolver off gas was simply added together with the initial airflow of quaternary air. For Case 2, the quaternary air flow was decreased and exchanged with the dissolver off gas, since this gas mainly consists of air. Ultimately, results for the flue gas composition, temperature, and heat load between the two cases were compared. 39 3. Method 3.4.2 Recycling of scrubber effluent In a parallel study, the design of a scrubber for wet flue gas treatment of NOx and SOx was performed, also as a case study for Södra’s process. This technique is explained in Section 2.4, where the idea of recycling the scrubber effluent back to the recovery boiler was investigated as a primary measure. For the wet flue gas treatment study, the scrubber was modeled in Aspen Plus using a thermodynamic model called ELEC-NRTL, which is applicable for aqueous and mixed solvents. The effluent stream, therefore, contained ionic compounds that had to be converted into uncharged compounds compatible with the recovery boiler model. See Appendix B.2 for calculations. These compounds are compiled in Table 3.11, followed by the corresponding volumetric flow and temperature of the stream. Table 3.11: Inlet effluent stream data. Effluent Units N2 9.5 g/s CO2 22 g/s Na2SO4 0.17 kg/s H2O 1.2 kg/s Total flow 4.6 m3/h Temperature 58 ℃ The effluent inlet data was inserted at the secondary air level into the model by adding a stream, and relevant changes in the mass and energy balances compared to the base case were observed. Since the Aspen model was limited in simulating dust formation in the boiler, the sulphidity in the flue gas as well as the sulphur and sodium losses to the stack were calculated in Excel for the modified case, similar to Appendix A, Table A.7. To simplify the addition of sodium and sulphur to the system in Excel, the share of recycled dust remained the same, based on the total amount of sodium entering the boiler. In order to approximate the amount of gaseous sulphur in the flue gas, the increase of SOx and TRS was assumed to be linear in the flue gas when adding the effluent. 40 4 Results In this chapter, the results from the Aspen simulations are presented, along with the mass and energy balances, to provide an understanding of the model’s performance, sensitivity, constraints, and key findings. First, the model is validated by comparing data from both literature and Södra. This is followed by a sensitivity analysis of the air staging. Subsequently, the results from the different primary measures are also displayed. 4.1 Model validation 4.1.1 Mass and energy balances The following tables present the most important data for comparison between the simulated results from the base case and the literature-based mass and energy bal- ances, along with some of the reference data collected from Södra. Table 4.1 shows the smelt composition of the inorganic compounds as well as the desired reduction efficiency. Table 4.1: Smelt composition, flow and reduction efficiency compared to the liter- ature. Variable Simulation Literature Units Na2S 23.0 22.4 wt% Na2CO3 66.2 67.0 wt% Na2SO4 2.68 2.60 wt% NaCl 0.562 0.562 wt% K2CO3 7.54 7.52 wt% Mass flow smelt 23.5 23.6 kg/s Reduction efficiency 94.0 94.0 % In Table 4.2, the simulated air flows are presented and compared with the flows from mass balance calculations from the literature. The primary air represents the required air supply to reach the desired reduction efficiency in the lower part of the furnace. It can be seen that the simulated primary air is half the amount compared to the reference data, but that the amount of secondary air compensates for this to achieve similar total amount of air supply. 41 4. Results Table 4.2: Air inlet flows compared to the literature and collected inlet data from the design specifications. Variable [Nm3/s] Simulation Literature Inlet data Primary air 24.95 26.79 47.71 Secondary air 82.57 - 59.81 Teriary air 11.01 - 11.01 Quaternary air 31.98 - 31.98 DNCG 11.90 - 11.90 Total air supply 162.4 163.8 162.4 Table 4.3 presents the simulated composition and total flow of the flue gas. These values are compared with calculated literature data, alongside reference data ob- tained from the DGE and ILEMA measurements. The compositions, presented in volume percentages, were calculated based on a wet gas basis. The fractions of NOx and SOx presented in the table are referred to as NO2 and SO2 respectively. Table 4.3: Flue gas composition and flow compared to the literature, DGE’s and ILEMA’s measurement as reference data. Variable Simulation Literature DGE ILEMA Units O2 2.76 2.82 3.10 2.45 vol% H2O 20.7 20.5 18.4 20.9 vol% CO2 12.8 12.7 10.3 12.7 vol% N2 63.7 63.9 - - vol% NOx 62.4 - 70.0 - ppmv SOx < 1.00 - < 1.00 - ppmv CO 376 - 100 - 500 150-550 ppmv HCl < 1.00 - < 2.50 - ppmv Total flow 201.8 203.8 241.7 220.1 Nm3/s Total flow dry 161.0 160.3 197.3 173.9 Nm3/s In validating the temperature distribution in the boiler, the the reported design specifications served as the primary reference for matching the heat transfer to the tube walls and obtaining the desired steam and flue gas temperatures, as shown in Table 4.4. 42 4. Results Table 4.4: Regional temperatures in comparison to reference data from the design specifications. Variable [°C] Simulation Ref. data Smelt 1003 900-1100 Between primary and secondary 1726 1596 Between secondary and tertiary 1460 1455 Above tertiary 1304 1301 Above quaternary 1101 1084 Flue gas temp 195.1 195.0 Steam temp 480.0 480.0 Table 4.5 shows the simulated heat exchanger loads required to achieve the desired steam production while maintaining temperatures in the range of the design specifi- cation temperatures. To clarify, the excess heat load represents the additional heat that was removed to reach an exhaust temperature of 195 °C for the simulation. Table 4.5: Heat exchanger units and excess heat load in flue gas. Variable [MW] Simulation Heat load to walls 208 Superheater & Economiser 300 Heat load to steam 505 Excess heat load 51.9 In Table 4.6, the overall heat load input to the boiler, losses and heat load to steam are presented and compared with the energy balance from the literature and the reference data from Södra. The total heat load covers not only the radiation and unaccountable losses but also the heat for formation of Na2S in the smelt, the sensible heat in the flue gas in the stack and in the smelt leaving the boiler. See Appendix A, Table A.8. Table 4.6: Overall energy balance compared to the literature and reference data. Variable [MW] Simulation Literature Ref. data Heat load to steam 505 496 460 Total heat load losses - 170 177 Total heat load input - 666 637 43 4. Results 4.1.2 Sensitivity analysis In this section, the results from the sensitivity analysis for investigating the perfor- mance and stability of the model are presented. Considering the performance of the boiler at the lower region, reduction efficiency serves as a critical indicator in particular. Therefore, Figure 4.1 presents a sensi- tivity analysis of primary and secondary air, given their importance in black liquor combustion. As shown in Figure 4.1 each curve corresponds to a fixed secondary air flow. Vari- ation in primary air flow yields new operating conditions and their respective re- duction efficiencies. Despite varying secondary air, all cases show a similar trend, where reduction efficiency inversely correlates with primary air flow. Altering the secondary air did not impact the reduction efficiency significantly, as can be seen in the figure. For a fixed primary air flow of 24.95 Nm3/s, all cases demonstrate a reduction efficiency of 94 % or higher. The green line represents the secondary air supply from the base case of 82.57 Nm3/s. Figure 4.1: Reduction efficiency in relation to primary air, for different secondary air flows. Figure 4.2 is similar to the figure above as it depicts the relationship between re- duction efficiency and primary air. However, it also demonstrate the relation in the flue gas temperature. It is shown that as the primary air is increased, the reduction efficiency decreases while the flue gas temperature rises. 44 4. Results Figure 4.2: Flue gas temperature and reduction efficiency in relation to primary air flow (secondary air same as base case: 82.57 Nm3/s). Considering the evaluation of performance for the upper part of the furnace in the model, Figure 4.3 illustrates how the overall composition of the flue gas is affected by increased temperatures at nose level, after the quaternary air injection. The differential increase or decrease in vol% for each component is shown in the figure, based on the initial flue gas composition at 1100 °C. It can be seen that for a temperature increase to 1600 °C, which was needed to reach a CO level of 376 ppmv in the flue gas for the model, the O2, N2, and H2O content changed by less than 0.02 vol%. The CO2 content decreased less than 0.04 vol% at this temperature increase. Furthermore, the overall flue gas composition at this CO level corresponds to the composition in Table 4.3. Another notice is that the change in composition exhibits an exponential trend with increased flue gas temperature. 45 4. Results Figure 4.3: Deviations in the flue gas composition, starting from 1100°C, with increased temperature, above quaternary air level. The same increase in CO content (376 ppmv) in the flue gas can be observed in Figure 4.4 as the temperature at nose level rises in the model. The CO content is less than 1 ppmv at the initial temperature of 1100 °C. 46 4. Results Figure 4.4: CO level in flue gas with increased temperature, above quaternary air level. 4.2 Primary measures 4.2.1 Ammonia injection Results from the water-diluted ammonia injection to the boiler is shown in Table 4.7, where the reduction of NOx emission was set to 45 % at full operating load. The table provides data on the quantities of NH3 injected, initial NO, and resulting NO after reduction. It also shows the provided heat load loss in the flue gas when injecting from the simulation. The unit mg/Nm3 is per dry flue gas at reference 6 % O2. The change in flue gas composition was negligible due to the small amount injected. However, the flue gas temperature decreased with 1 °C. Table 4.7: NOx reduction and flue gas heat load losses with water-diluted ammonia injection. NH3/NOinitial NH3 NH3 NOinitial NOfinal Losses mol/s mg/Nm3 mol/s mol/s MW 1.3 0.98 74 0.76 0.42 2.5 Comparing the NOx reduction with the yearly specific emission values from the BAT requirements, the values of NOinitial and NOfinal from Table 4.7 were converted to mg/Nm3 at 6 % O2 per dry flue gas, as shown in Table 4.8. In the table, ’Södra’ refers to their Environmental Report from 2023. 47 4. Results Table 4.8: Initial concentrations of NOx from different reference data and final concentration of NO in dry flue gas, with water-diluted ammonia injection. Concentration of NO Södra DGE BAT NOfinal mg/Nm3 141 100 120 - 250 55.0 The optimal concentration of ammonia may differ depending on the recovery boiler operating conditions. Therefore, Figure 4.5 illustrates the relationship between am- monia concentration (water-diluted) and heat load loss. The amount of NH3 is fixed while the amount of water is varied. It can be observed that a lower concentration of ammonia results in a larger loss in heat load, by an exponential increase. Achieving a balance between the concentration of ammonia and the reduction of NOx emissions can, therefore, be a factor to consider. Figure 4.5: Ammonia concentration in relation to heat load loss. Results from the dissolver off gas injection are shown in Table 4.9 and 4.10. Table 4.9 provides a comparison of the calculated ammonia dosages in mg/Nm3 at 6 % O2 per dry flue gas with the addition of dissolver off gas (Case 1) and decreased flow of the quaternary air (Case 2). 48 4. Results Table 4.9: Ammonia dosage in flue gas comparison with adding dissolver off gas (Case 1) and decreased flow of 4th air (Case 2). Dissolver off gas Base Case Case 1 Case 2 Units NH3/NOinitial - 0.389 0.389 NH3 - 0.294 0.294 mol/s NH3 - 20.9 22.1 mg/Nm3 Dis. off gas flow - 11.3 11.3 Nm3/s Quaternary air flow 32.0 32.0 20.7 Nm3/s In addition, Table 4.10 compares the composition, temperatures, and heat load increase in the flue gas based on model simulations, with the same cases as the table above. The vol% is based on wet flue gas. Table 4.10: Comparison of composition, temperatures and heat load increase in flue gas with added dissolver off gas (Case 1) and decreased flow of 4th air (Case 2). Dissolver off gas Base Case Case 1 Case 2 Units O2 2.76 3.45 2.47 vol% H2O 20.7 20.8 21.9 vol% CO2 12.8 12.2 12.8 vol% N2 63.7 63.6 62.7 vol% Temperature 195 205 211 °C Increased heat load 0 22.4 22.5 MW Comparing the ammonia availability in the water-diluted ammonia and the dissolver off gas inlet(Table 4.7 and Table 4.9), the molar flow difference was 0.69 mol/s. 4.2.2 Scrubber effluent recycling Results for the scrubber effluent recycling were provided by the calculations on sulphur and sodium balances from the base case compa