Application of numerical modelling to biomass grate furnaces

Abstract

Keywords: Biomass; Grate furnace; Combustion; Numerical modelling

Introduction

Biomass as a renewable energy source is widely applied for heat and power production and its importance will gradually increase in the next decades, because of its sustainability which can be partly CO2 neutral and depletion of fossil fuels. The conversion of solid biomass to useful energy can be achieved through one of two major pathways, (1) biological (fermentation) and (2) thermal. Biological conversion is perhaps the most ancient means of biomass conversion [1]. Thermal conversion of biomass into gases came much later. Commercial usage of small biomass gasifiers began during the Second World War. The four principal routes for thermal conversion are: combustion, gasification, pyrolysis and liquefaction. The last three routes are indirect process where biomass converted to an energy carrier (i.e.char, gas or liquid). Combustion involves the high temperature conversion of biomass in excess air directly into heat. Gasification, in contrast involves a chemical reaction in an oxygen deficient environment to produce gas and char at the first stage, and the

Technical Note – JTEN – 2015 – 39 porpuse to achieve a more efficient reduction of NOx and CO

Grate Firing Technology

emissions. The grate firing technology has the advantage that it is able The secondary air supply systems provide the most flexible to burn fuels with high moisture content (up to 60 wt%), way to retrofit the grate firing systems for a better burnout and different particle sizes (0.005 - >0.1 m), high ash content and lower emissions. The secondary air nozzles can be optimised in blended fuels. However the capability of mixing of woody terms of number, diameter, spacing, location and orientation to biomass and herbaceous fuels are limited. High excess oxygen enhance mixing, to prolong the mean residence time of the flue (5 – 8 vol%) is a major drawback of the system which reduces gas, to distribute the temperature more evenly and reduce the the boiler efficiency and increase the emissions potential. flame temperature peaks. Capacities of grate firing boilers range from 10 kWth to 50 MWth [3]. Modern grate furnaces consist of the following elements: fuel feeding, grate, primary combustion chamber (PCC), secondary combustion chamber (SCC), flue gas recirculation (FGR) and an ash removal system, as it is shown in Figure 1. It is important to keep consistent and uniform fuel feeding and avoid any batch wise feeding that raises emission potential. For biomass fuels that are very heterogeneous in size, a stoker is needed to reduce the tendency that the different sizes separate themselves and spread the biomass evenly on the grate. Typically mechanical stokers are used in biomass grate firing systems and they can be divided into three main categories based on the way the biomass is fed onto the grate: crossfeed, overfeed and underfeed stokers. In the overfeed stoker the fuel is fed onto the top of the bed and moves downward as it is SCC consumed. In the underfeed stoker, the fuel is fed upward at the bottom of the bed. In the crossfeed stoker the fuel cross the horizontal or inclined grate and it burns as it moves along the grate. The grate, on which fuel bed is resting until it burns out, Secondary can be fixed, travelling, reciprocating, rotating or vibrating. The air nozzles main functions of a grate are to provide a uniform consistency of the fuel bed and distribute the primary air entering beneath PCC the fuel bed. An uneven fuel bed inclines the burning toward FGR certain areas which stimulates hot spots and dead air zones within the furnace. In most of the industrial boilers multiple zones of primary air distribution under the grate are used to achieve a complete and stable combustion, a more favourable de-ashing Primary air system temperature distribution and low dust emissions The primary combustion chamber starts from top of the fuel bed until the secondary air nozzles. In modern grate firing boilers the primary combustion chamber has fuel rich condition (air-fuel equivalence ratio λ < 1) in order to reduce nitrogen oxides (fuel NOx) formed during the solid biomass combustion on the grate. Additionally if too much oxygen is available in the Figure 1: Sketch of a biomass grate firing furnace [2] PCC (λ > 1), the volatiles burn in th PCC and release thermal energy at the bottom of the furnace instead of in the upper parts

Numerical Modelling Of Grate Furnaces

where the heat exchanger is located. This premature combustion The legislations on the amount of emitted pollutants and impairs the boiler efficiency and can damage the grate by the plant efficiency of biomass combustion systems are exposing to high temperatures. In order to efficiently control the continually being restricted. As a result the energy producers temperature and, additionally, to improve turbulent mixing in and boiler manufacturers, in the heat and power sector, are the primary combustion chamber, modern grate firing systems forced to improve the combustion efficiency and optimise the are equipped with the flue gas recirculation. The flue gas plant operation continually. Theoretical investigations aided by recirculation (FGR) can be staged, it means entering a fraction numerical modelling are using to understand and describe the of the FGR below the grate and the rest above the grate. The relevant phenomena in the system as a basis for optimisation FGR improves mixing in the PCC by breaking up the strains and new design. Numerical modelling provides much more arising from channelling in the fuel bed. Furthermore it has the detailed information on the underlying processes and freedom 551

Technical Note – JTEN – 2015 – 39 Figure 2. In this model the biomass combustion is modelled by to investigate the effect of geometry, in contrast to the a thermally thick particle model which considers the effect of experimental investigations. The success of the numerical temperature gradients in the particle. Figure 2 illustrates the modelling is highly dependent on the accuracy of the physicalpositions of the three thermal sub-processes, e. i. drying, chemical models and the numerical methods applied. Therefore pyrolysis and char burnout along the particle paths on the grate. it is essential for numerical models to be validated As it can be seen they occur sequentially with an overlap experimentally. In the following sections an overview about the between each other, especially where pyrolysis and char numerical models for relevant phenomena in biomass grate burnout take place. firing systems is given.

FUEL BED Thermal Conversion

The common approach in grate furnace simulations is to separately model the biomass thermal conversion in fuel bed and the gas phase combustion in freeboard, although there is an intensive interaction between them. The 1D empirical models serve to provide the temperatures and species release profiles as boundary conditions for the pursuing simulation of gas phase [4, 5]. The 1D empirical models are not able to sufficiently provide information about combustion on the grate and relevant influencing parameters in different grate systems. Therefore, research work in different groups is focusing on the development of 2D or 3D models for biomass fuel bed combustion. However, there are a few studies, where the combustion model of the fuel bed is incorporated in the entire simulation domain of the biomass grate firing system [6-8]. The review of elaborations on fuel bed modelling published in the literature shows a broad variety of different model approaches to describe packed bed systems [2]. Fundamentally, they are either homogeneous or heterogeneous models. The difference lies in the calculation of the energy equation. In homogeneous models the temperatures of the gas and the solid phase are assumed to be equal, and a single, overall energy balance equation is applied [9-12]. The physical properties which appear as constants in the energy equation are described by their effective values over the entire bed. In heterogeneous models the gas phase and the solid phase have individual energy equations [13-18]. They have different temperatures, and heat and mass transfer between the two phases are described by means of Nusselt and Sherwood correlations. Based on their treatment of the solid phase, heterogeneous models can be classified into continuous models [19-21] and discrete particle models [22, 23]. Continuous heterogeneous models treat both phases as if they were distributed continuously over the whole spatial domain. At each point in space both phases exist with distinguished properties. The common limitation of the continuous packed bed models is that intra-particle effects cannot be sufficiently described. Additionally, it is very difficult to model the shrinkage of the packed bed using continuous models. The discrete particle models enhance packed bed modelling by considering the packed bed as an ensemble of representative particles in which each particle undergoes thermal conversion processes. This enables the inter-particle effects, e.g. momentum and energy exchange, to be fully described. The main drawback of the discrete particle models is the high calculation time. An example of a 3D model for biomass packed bed combustion, which enable a direct link of the bed model with the gas phase combustion models in the freeboard is shown in

Figure 2: Simulation of a 180 kW biomass grate furnace fuel: wood chips; contours of release rates of different species [mg/s] from biomass particles at a vertical cross section in the fuel bed [24].

GAS Phase Combustion

The combustion of volatile species above the fuel bed highly depends on the mixing of reactants in the flow because of the fast combustion reaction rates. Turbulence greatly enhances the heat and mass transfer and promotes mixing of reactants. Therefore it is essential to take into account the effects of turbulence on the reaction rate. The turbulent mixing between the reactants occurs in a rather nonlinear manner and it poses one of the challenges in simulation of turbulent reacting flows. The Eddy Break-Up combustion models proposed by Magnussen and Hjertager [25] are usually applied to model the turbulence chemistry interaction of the gas phase combustion in biomass grate furnaces. They are successfully able to optimise the nozzles for the injection of secondary air and flue gas recirculation to enhance turbulent mixing and flue gas burnout. The Eddy Break-Up models assume the reaction chemistry to be infinitely fast and the combustion process depends on the turbulent mixing. However, in laminar to moderate turbulence zones such as in small-scale biomass combustion plants the Eddy Break-Up combustion models are not valid because they were originally developed for highly turbulent flows. Therefore a hybrid gas phase reaction model was developed which is

Technical Note – JTEN – 2015 – 39 sensitive to local flow conditions [26]. The model reliably As it can be seen in Figure 3 (top) the mixing function distinguishes between the mixing or kinetically dominated indicated that a major part of the reaction zone above the zones. It calculates the effective reaction rate from the laminar packed bed is influenced by streak formation. The NOx finite rate kinetics and the turbulent reaction rate and weights emissions calculated with the hybrid-streak formation model is them depending on the local turbulent Reynolds number. (bottom in Figure 3) in better agreement with measured value Moreover the porous nature of the fuel bed leads to streak [27] due to a more accurate prediction of the species mixing formation influencing gas mixing and combustion above the process above the fuel bed and, therefore, of radicals, that are fuel bed. The concentration of oxidiser and reactants in the gas important for NOx formation processes. released from the biomass fuel bed are strongly spatial For gas phase combustion in biomass grate furnaces usually dependent. Recently some studies [20, 21] dealt with the global reaction mechanisms are applied [4, 5]. The global channelling and its effect on the combustion of biomass fuel reactions describe an overall process by which reactants are bed. Shiehnejadhesar et al. [27] developed a streak model based converted to anticipated products at the end of the process. on the mixing function to describe the mixing process above the Therefore they may not be sufficiently accurate to predict the fuel bed. The mixing function is a combination of the mixing formation and emissions of NOx and unburnt CO for a broad time, the necessary residence time to reach the fully mixed range of operating conditions. Additionally, due to a strong condition, and the flue gas residence time in the region above interaction between the freeboard and the fuel bed, an accurate the fuel bed. The model has been successfully tested with a realmodel for the combustion chemistry in the freeboard is of high scale biomass grate furnace. The selected results of this model relevance. Hence, some studies used detailed and reduced are shown in Figure 3. reaction mechanisms to model the homogeneous combustion in freeboard [8, 24, 28].

Pollutants Formation And Emissions

Biomass combustion is a source of air emissions which destruct the global environment and human health to some extent. The formation of pollutants not only depends on the fuel but also depends on the furnace operating conditions, e.g. temperature, equivalence ratio (λ), flue gas residence time, etc. It is therefore important to know how the pollutants are formed and find a way to lower emissions of pollutants. The combustion process itself always entails pollutants formation and the combustion of renewable fuels, like biomass, has some particular emissions which emanates from the biomass fuel characteristic. The pollutants originate from biomass combustion can be divided into two different groups: gaseous emissions and inhalable particular matter (PM). Emissions in the first group are NOx, SOx, CO, CO2, unburned hydrocarbons (UHC), polycyclic aromatic hydrocarbons (PAH) (partly found on the surface of PM), dioxins and furans (PCDD/PCDF) (partly found on the surface of PM). The second group includes, soot, condensable organic compounds, aerosol and coarse fly ash particles. Hereinafter, the latest efforts of modelling nitric oxide (NOx) formation and emissions as well as the ash related problems are presented. Due to continually stricter emission limits and increasing demand of the combustion of new agricultural and herbaceous biomass fuels with high nitrogen contents, there is a need for efficient design tools for low NOx furnaces. There are three generally accepted mechanisms for NOx formation: thermal NOx, prompt NOx, and fuel NOx [29, 30]. Thermal NOx is formed by the high-temperature reaction of nitrogen with oxygen, by the well-known Zeldovich mechanism. Prompt NOx is formed by the relatively fast reaction between nitrogen, oxygen, and hydrocarbon radicals (e.g. CH and CH2). The products of these reactions could lead to formation of amines and cyano compounds that subsequently react to form NO. Fuel NOx is formed by the direct oxidation of organo-nitrogen compounds contained in the fuel. As a result of the

Figure 3: Simulation of a 180 kW biomass grate furnace, fuel: wood chips; top: Mixing function [-]; bottom: NOx concentration [mg/Nm3 d.b., 13 vol% O2] calculated with considering the effect of streaks arising from the fuel bed [27].

Technical Note – JTEN – 2015 – 39 condensation of ash vapours, deposition of coarse fly ash comparatively low combustion temperatures in biomass grate particles, erosion of the deposit layer by non-sticky particles, firing systems, the fuel NOx is the major source of NOx [31]. aerosol formation and deposition. The model, additionally, can Therefore the most crucial point is the release of the NOx simulate the time dependent ash deposit formation and its precursors from the fuel bed to the freeboard under different influence on heat transfer in the furnace in dependence of fuel environments for different biomass fuels. Klason and Bai [32, type and plant operation conditions. 33] assumed that all the fuel-N released into NH3 and HCN In Figure 4 the simulation results for a 70 kW pellet boiler with 50/50 molar ratio during the devolatilisation and no char-N are shown. As it can be seen the fine particle formation mainly is considered. With these assumptions they simulate NOx initiates ate the entrance of the heat exchanger because formation and emissions of a small-scale wood pellet grate temperatures in the primary combustion chamber is too high for boiler and an industrial grate furnace burning wood chips. Yang ash forming vapours condensation. et al. [34] studied the effect of operating conditions on the performance of a 38 MWe power plant. The biomass fuel was straw and they model the NO emissions. They split the fuel-N into volatiles and char. The volatile-N was assumed to be only NH3 and char-N was directly oxidised to NO. Widmann et al. [35] defined the release profiles for the most relevant N-species (NO, NH3 and HCN) of the NOx precursors during solid biomass combustion on the grate. Validation simulations showed that with this model all relevant trends concerning NOx emissions in dependence of fuel nitrogen content, air staging, flue gas residence times and temperatures could be reproduced. Ash related problems like aerosol formation and ash deposition as well as corrosion are of major relevance in Figure 4: Simulation results regarding aerosol formation in a 70 kW biomass combustion plants especially when firing fuels with fixed bed pellet boiler: left) total particle concentrations [mg/Nm³ high ash content (e.g. waste wood, herbaceous and agricultural d.b., 13 vol% O2]; right: chemical composition of fine particles. [40] fuels). It is well known that the particular matters are formed by two different ash formation procedure [2, 3]. The main rout of

Outlook

particle formation is nucleation and condensation of alkali The continually higher standards for operation of biomass compounds. A considerable fraction of inorganic volatile grate furnaces and increasing demand of new biomass fuels and species such as K, Na, S, Cl, Zn and Pb is released to the gas new combustion technologies indicate that more research is still phase due to the high temperature during char combustion. required. The most significant fields of modelling activities in These ash forming vapours later start to nucleate and form biomass grate firing systems have been summarized in this submicron particles or condensate on the surface of coarser paper. In each category outlined there are still potentials for particles or directly on the heat exchanger surfaces. These further improvements. submicron particles are called aerosols and having diameters Modelling of biomass packed bed combustion is highly between 0.01-1 µm. The second route is the ash fusion and important, because it describes not only the process underlying coagulation of non-vaporised ash residuals such as Si, Ca, Mg, in the packed bed but it also determines the precursors of Fe and Al. It leads to formation of coarse fly ash particles which combustibles, NO x and ash forming volatiles. Despite the entrained to the flue gas from the packed bed. It is worth to note significant efforts have been done in this filed the developed that the aerosols are formed by phase change when flue gas models are not sufficient and need to be strengthened. More temperatures are lower than the saturation point of the ash endeavours are needed to develop adequate models which take forming vapours, while the coarse fly ashes are always in solid into account the channelling effect and uneven consumption of phase. the biomass fuels on the grate which leads to spatial dependent There are some modelling efforts on the aerosol formation porosity in the fuel bed. The heat transfer models in the packed such as work of Jöller et al [36, 37] where the amount and the bed, particularly particle-wall heat transfer, still have some chemical composition of particle emissions in biomass grate space for improvement. Mixing of the biomass fuel on the grate furnaces were investigated. They used a plug flow model and and the effect of the grate movement on it is a challenging topic thermodynamic equilibrium calculations together with a kinetic which needs to be addressed, specially by considering the effect approach to model gaseous sulphate formation. Glarburg et al. of different particle shape and size. [38] developed a reaction mechanism for sulphation of alkali Despite the valuable efforts for modelling the release metals, since the gaseous alkali sulphates may yields aerosols profiles of NOx precursors and ash forming vapours as well as and also contribute to deposition and corrosion. Schulze et al. inorganic elements, they are still preliminary. Therefore it needs [39, 40] developed a model for release of fly ash particles and sophisticated fundamental understanding without introducing ash forming vapours from the fuel bed of biomass fired boilers too many assumptions and simplifications. The deposition and they coupled the model with a commercial software models show an encouraging potential to estimate the (ANSYS FLUENT) to solve the transport and deposition of deposition formation in biomass grate boilers. However, they them on the boiler walls. The model considered the 554

Technical Note – JTEN – 2015 – 39 [7] Colloaz J., Porteiro J., Patino D., Granada E., need to be extended in order to reliably calculate the particle “Numerical modeling of the combustion of densified wood size distribution, composition and stickiness of the coarse fly under fixed-bed conditions” Fuel, vol. 93 pp. 149-159, 2012. ash particles as well as the heterogeneous reactions on the [8] Mehrabian R, Shiehnejadhesar A, Scharler R, entrained particle surfaces. Obernberger I, “Multi-physics modelling of packed bed Gas phase combustion in the freeboard has been biomass combustion” Fuel, vol. 122 pp. 164-178, 2014. extensively investigated and many achievements have been [9] Merrick D., “Mathematical models of the thermal already accomplished. However development of a detailed decomposition of coal. 1. the evolution of volatile matter”, reaction mechanism which is exclusively for biomass fuels to Fuel, vol. 62, no. 5, pp. 534-539, 1983. build up a CFD model is still missing. The already applied gas [10] Van Der Lans R.P., Pedersen L.T., Jensen A., Glarborg phase reaction mechanisms need enhancement concerning the P., Dam-Johansen K., “Modelling and experiments of straw transformation and conversion features of the relevant species combustion in a grate furnace", Biomass and Bioenergy, vol. such as heavy hydrocarbons (tars), NOx and SOx precursors, 19, no. 3, pp. 199-208, 2000. PAH and soot. [11] Vortmeyer D., Schaefer R.J., “Equivalence of one- and It is worth noting that besides the modelling activities, two-phase models for heat transfer processes in packed beds: there is a great need of comprehensive experimental one dimensional theory”, Chemical Engineering Science, vol. investigations on biomass grate furnaces. Lack of reliable 29, no. 2, pp. 485-491, 1974. measurements, particularly in the fuel bed, suffers the packed [12] Finlayson B.A., “Packed bed reactor analysis by bed combustion models for further enhancements. orthogonal collocation”, Chemical Engineering Science, vol. 26, no. 7, pp. 1081-1091, 1971.

Conclusion

[13] Hobbs M.L., Radulovic P.T., Smoot, L.D., “Modeling This paper focuses on the biomass grate firing .technology fixed-bed coal gasifiers”, AIChE Journal, vol. 38, no. 5, pp. and the most relevant modelling activities in this field. It is 681-702, 1992. stated that numerical modelling provides an approximate [14] Bryden K.M., Ragland K.W., “Numerical modeling of overview of biomass grate firing systems. It helps to diagnose a deep, fixed bed combustor," Energy and Fuels, vol. 10, no. 2, and solve operational problems as well as providing assistance pp. 269-275, 1996. when dealing with new designs to reduce emissions and [15] Shin D., Choi S., “The combustion of simulated waste increase the overall efficiency. The numerical modelling should particles in a fixed bed”, Combustion and Flame, vol. 121, no. finally lead to a reduction of design and optimisation procedure 1-2, pp. 167-180, 2000. of new technologies, by providing a deeper understanding of the [16] Blasi C.D., “Dynamic behaviour of stratified fundamental processes in the plant. downdraft gasifiers”, Chemical Engineering Science, vol. 55, Therefore the current models should be improved and no. 15, pp. 2931-2944, 2000. linked to the new innovative numerical models with the final [17] Cooper J., Hallett W.L.H., “A numerical model for objective of a “virtual biomass combustion plant” to simulate packed-bed combustion of char particles”, Chemical all relevant processes. Engineering Science, vol. 55, no. 20, pp. 4451-4460, 2000. [18] Thunman H., Leckner B., “Ignition and propagation of

Share and Cite