CHARACTERISING FUELS FOR BIOMASS - COAL FIRED COGENERATIONCoal is commonly used as an auxiliary fuel in biomass fired co-generation power stations. Design of the boilers for these stations therefore requires a detailed knowledge of the properties of the biomass, the coal and any interaction between the two. A continuous ash discharge (CAD) stoker is normally used for this combination of fuels. In assessing an acceptable grate rating for biomass fuel firing, the effective moisture of the biomass is the key parameter required whilst the grate rating for coal is a function of its reactivity. The GCV of biomass fuels has been characterised as a formula involving their moisture and ash contents. In the case of bagasse, the values of the constants have been sufficiently defined to avoid the need to measure its GCV empirically. To size the grate for biomass fuel firing the concept of 'effective' moisture has to be introduced to overcome the distortion caused by any variation in ash content. There are sophisticated methods of analysing coal to establish reactivity but the methods are time consuming and only a few laboratories are equipped to use them. A new approach is proposed to use the petrographic composition of the coal to predict reactivity, microscopically examining the coal to determine the make-up of its organic constituents [the 'macerals'] and hence determine its characteristics. The technique is fast and relatively inexpensive and where it has been used recently has yielded commercially viable results. The techniques described allow the grate to be designed for optimum combustion but do not predict how the ash chemistry of the fuels may adversely affect the anticipated results. Furthermore they do not predict any interaction between the fuels which may preclude combination firing. An appropriately designed grate will help resolve differing primary air and ash discharge requirements. The characteristics of the two ashes, however, must be brought to account by means of specialist phase diagrams in order to determine whether the grate and/or furnace will need to be de-rated to prevent clinker and slag formation and boiler fouling. Key words: Combustion, Fuel Characteristics, Biomass, Coal macerals, Stokers CAD Zoned, Cogeneration INTRODUCTION
In common with fossil fuels, biomass fuels have four important properties:
The gross calorific value (GCV) of a biomass fuel as determined in the laboratory with a "bomb" calorimeter, can be written in most cases in the form:
The constant k is usually between 17 500 and 21 000. Fuel moisture varies from 0,05 to 0,60 and the ash fraction from 0,01 to 0,13. The Sugar Milling Research Institute in South Africa (Don et al, 1977) has established that k for bagasse is 19 605. To correct for the fact that the calorific value of brix is less than that of fibre their form of the equation subtracts 3 115 x b from the result where b is the brix fraction of the fuel. Heat liberation rate (GJ/hm2 or kW/m2) is used as a basis for spreader stoker sizing on a biomass fuel. The fuel moisture content, the primary air temperature and the fuel's fouling propensity are the significant factors which determine grate rating. In this paper the furnace plan area at the level at which the biomass fuel is introduced into the furnace is used to calculate grate heat release rate. The work reported here introduces the concept of "effective moisture content" because increasing the ash content of a given fuel effectively increases it's moisture to combustible material ratio. To eliminate this detrimental effect from the calculations all fuel analyses are reduced to an ash content of 2 %. In the cane sugar industry where 53 % moisture fuel having an ash content of 2 % can be burned relatively easily, one having the same moisture content but an ash content of 6 % may be difficult to burn as it has an "effective" moisture of 55,26 %. The first fuel has a moisture to combustible ratio of 1,18:1 and the second a ratio of 1,29:1; i.e. almost 10 % more moisture per unit mass of combustible material whilst fuel moisture has only gone up by 2,26 % points.
Northern Hemisphere coal can often be burned at significantly higher ratings on Continuous Ash Discharge (CAD) stokers than most Southern Hemisphere and Indian (Gondwana Land) coal. For the same grate heat release rates, carbon losses from coals having similar proximate and ultimate analyses derived from Northern Hemisphere seams are usually lower than they are from Gondwana Land seams. In the past, coal was characterised by defining its grading, proximate and ultimate analyses, GCV, ash fusion temperatures and swelling and Hardgrove indices. An understanding of why Gondwana Land coal is so much more difficult to burn and hence why lower stoker ratings are required, can be postulated by bringing its petrography into account. In this paper stoker effective area for coal firing is defined as the width of the stoker measured at the grate level times the distance from the center line of the rear wall tubes to the center line of the stoker front shaft. There are two ways of defining stoker rating for coal burning; as the 'power released per unit of effective stoker area (as for biomass fuels) or as the 'mass of fuel burned per unit of time per unit of effective stoker area (kg/m2h). Where the ash content is relatively low, say less than 10 % and where the coal is reactive, the first gives reasonably coherent results. For higher ash content, low reactivity coals, which need a longer time to burn out, the second gives better results. [The reason for this is straightforward. High ash content coals generate large quantities of ash. To get rid of this ash the stoker must be run at a higher speed. As it takes longer to burn out a high ash content coal, particularly if it is not very reactive, this is contrary to what is required to complete combustion. Rating the stoker on a mass per unit area basis automatically builds in a de-rating factor to accommodate this.] In modern boilers the furnace plan area is closely related to the effective stoker area. Higher stoker ratings, therefore, result in higher furnace gas up-flow velocities. These in turn result in larger particles being carried over into the cooler passes of the boiler where the temperature is too low for combustion to be completed. Small, highly reactive particles, on the other hand, are much easier to burn out in the furnace enclosure. As larger particles are more difficult to burn because their exposed surface area is relatively small, those, which are carried over, have even less chance of burning out than the smaller ones.
Thermo Gravimetric Analysis (TGA) is an experimental tool to establish coal reactivity. TGA reactivity is determined by measuring the rate at which the mass of a burning coal sample reduces with time. Only a few specialist laboratories are equipped to do this type of work. The technique is time consuming and it is rarely possible to use it during the quick turnaround investigative phase of a project. Over the last few years we have attempted to find a more user-friendly method of establishing coal reactivity. Our investigations have led us to the conclusion that relatively inexpensive petrographic analyses are able to provide a practical answer to the problem. TGA is still needed for an in depth study if no process experience with a coal is available. Coal petrography can be described broadly as the microscopic determination of the organic and inorganic constituents of coal and the degree of metamorphosis, or rank, which the coal has obtained. Petrographic constituents can be directly correlated to important properties affecting the combustion behavior of coal. Differences in coal qualities from differing regions and even within single seams can be readily identified. The term 'maceral' is used to represent the different plant tissue from which the coal was originally formed. Vitrinite, liptinite (previously called exinite) and inertinite are the main maceral categories. Vitrinite, for instance, is derived from cell walls, cell contents of precipitated gels preserved under water and is relatively oxygen rich. Algae, spores and waxy leaves for example formed liptinite, the hydrogen-rich maceral type. Inertinite originated from plant tissue that had oxidised, altered, degraded or burnt in the peat stage of coal formation, and is carbon rich. The term microlithotype describes the association of the macerals and mineral matter. The degree of "mixing" of the macerals and the nature of the minerals bound within the coal will influence its mechanical and technological properties. A coal with a higher inertinite content will burn with a longer flame at higher temperatures whilst a vitrinite rich coal will burn more quickly and produce a shorter flame. This often results in boilers designed to burn Northern Hemisphere coal suffering from high steam temperature when burning Gondwana Land coal. Sanyal (1983) observed that steam temperature was influenced by maceral constituents in a pulverised fuel fired boiler burning Polish and South African coals. He attributed this to the longer time needed to burn out the latter coal compared to the former. Earlier, Yakorskii et al (1968) reported that combustibles in ash were largely the result of incompletely burnt inertinite particles. Volatile content is often taken as an indication of coal rank. However, two coals with the same volatile content may perform differently. For instance, one may be an anthracite (high rank, low volatile) as is typically found in South Africa's KwaZulu-Natal coal fields, whereas the second may be a heat affected bituminous coal (low rank, low volatile) associated with some coal seams in the Witbank and Highveld coalfields. Using vitrinite reflectance petrographic techniques (rank analysis) these differences can be readily detected and quantified. Northern Hemisphere coals usually have a high vitrinite content (> 70 %), with the non-vitrinite fraction being predominantly liptinite. The inertinite content is usually < 20 %. Gondwana Land coals on the other hand have a low vitrinite content (< 60 %), low liptinite contents and high inertinite content (> 40 %). Vitrinite and liptinite are both reactive. They enhance the rate of combustion. Inertinite has a low reactivity, which retards combustion. Gondwana Land coals are inherently less reactive than Northern Hemisphere coals and can, therefore, only be burned at efficiencies approaching Northern Hemisphere coals at lower stoker ratings. They also frequently have higher ash contents, which aggravates the problem.
The curve has been tempered by a concept, which we term "ignitability". This concept which was developed by John de KOCK, accommodates the fact that portions of the volatile constituents of some coals are inert and therefore instead of promoting ignition retard it. The ignitability of the volatiles, or volatile energy level, is expressed as it's specific heating value. We have found that if the figure is > 28 MJ/kg the coal will ignite easily. If it is less, ignition difficulties may be experienced. If it is < 18 MJ/kg full-scale tests may be needed to establish a suitable stoker rating.
As the total carbon loss is a function of the amount of coal burned, i.e. the amount of ash produced, the calculation is clearly iterative. We have developed algorithms to facilitate these calculations and a computer program to help with their iterative nature. Typical results for South African and Colombian coals are tabled below:
We believe that the higher than expected Colombian coarse ash loss is due to the fact that in this instance the stoker runs from the front of the furnace to the rear rather than the conventional rear to front. With the arrangement as installed, it is possible to throw coal into the ash pit without giving it a chance to burn. Dual Fuel Operation
Minerals in fuels transform on heating to form ash. The propensity of a fuel to establish problematic formations depends on its elemental volatility and the viscosity of resultant ash. Volatility of the fuel is dependent on the proportion of organically bonded elements and temperature. The viscosity depends on the chemistry of the ash, localised conditions (oxidising/reducing) and temperature. Clinker and Slag Formation Low viscosity ash is molten and will readily coalesce to form a large molten mass or cause the agglomeration of bed particles on the stoker grate. This is called clinker. Unlike pulverised fuel fired boilers, stoker fired boilers have a longer residence time, which favours the formation of clinker. As the molten mass is cooling, mineral phases can crystallise out of the melt, which is the catalyst for the formation of hard clinkers. In contrast, a high viscosity, 'dry' ash does not coalesce readily, thus restricting the formation of clinkers. When a molten ash particle, transported by updrafts collides into the furnace wall, a slag deposit is formed. Alternatively, if the heat transfer surface is covered with a molten sticky slag layer then it is highly probable that 'dry' ash particle will also adhere to the surface. A sintered deposit can also form when particles at high temperature impact on heat transfer surfaces. Sintering is a complicated process, encompassing a variety of phenomena that take place when a powder is transformed into a compact material (Nowok, 1996). Contrary to the objective of the ceramics and glass industries, which is to enhance the strength of materials through sintering, boiler designers try to prevent high strength deposits growing on heat transfer surfaces. Different approaches can be used to predict the clinker/slagging propensity of a fuel. The common approach is to measure ash deformation temperatures under reducing and oxidising conditions or to use the numerous slagging indices which are available and are based on the acid (SiO2, Al2O3) to base composition (CaO, MgO, K2O, Fe2O3 and Na2O) of the ash. The table below schedules values for two of these indices when it is unlikely that clinker/slagging problems may occur, when they may occur and when it is almost certain that they will occur. Index A is the product of the base/acid ratio and the total sulphur content while Index B is the (maximum hemisphere temperature plus four times the minimum initial deformation temperature)/5. These indices should be used as a general guideline as it is common knowledge that indices based on Northern Hemisphere coals are not necessarily applicable to Southern Hemisphere coals. These observed differences could be attributed to differences in coal formation environments and the nature of the mineralogy. For instance, if the major Fe-bearing mineral is Siderite (FeCO3) and not pyrite (FeS2) then slagging index A is not necessarily applicable.
In comparison to coal, biomass is likely to have a higher proportion of alkali-earth elements (K, Ca, Mg and Na) which are organically bound (attached to carbon structures) rather than associated with minerals. On combusting biomass, these elements are volatile and create a fume, which can initiate slag formation. Thus applying a slagging index designed for coal is not necessarily applicable to biomass. An alternative approach to using slagging indices is to use phase diagrams. If applied correctly, phase diagrams can also be used to predict the melting temperatures of the ash much more quickly than laboratory techniques. The compilation of phase diagrams dates back more than 100 years. F.P. Hall and H. Insley published 178 diagrams in the October 1933 issue of the American Ceramic Society Journal These are still used today. The two critical temperatures derived from a phase diagram are the solidus and liquidus temperatures. During the heating of a solid, the solidus (Sh) is the temperature at which the solid will start melting and the liquidus (Lh) is the temperature at which melting is complete. During cooling, the liquidus (Lc) is the temperature at which the first phase will crystallise from the molten melt and the solidus (Sc) is the temperature when crystallisation is complete. Thus, phase diagrams can be used to predict the temperature at which phases will melt during heating (solidus Sh)) or the temperature (liquidus Lc) at which phases will start crystallising from a molten mass during cooling. In terms of predicting slagging propensity of a coal/biomass mixture the solidus (Sh) will be an indication of the temperatures at which 'sticky' fly ash particles are likely to be formed. In terms of forming solid clinkers, the liquidus (Lc) temperature is an indication of when the crystals will start crystallising from the melt and will be the initiation (like a catalyst) of clinker formation. Alternatively, the solidus (Sh) is an indication of the final sintering temperature.
 Fig. 8: CaO-Al2O3-SiO2 phase diagram. Circle indicates liquidus temperature (Lh totally molten) and square indicates the solidus temperature (Sh start melting) for idealised compositions A and B (Phase diagram from Osborn and A.Muan) In addition to predicting melting and clinker formation and sintering temperatures, phase diagrams can be used to predict the composition of the clinkers and possibly obtain an idea of the temperature history to which the clinker was exposed. Depending on the number of elemental components, a binary, ternary or quaternary phase diagram is used. Based on the distribution of the components in the ash analysis, a phase diagram is selected and the ash elemental analysis is normalised and plotted. In a ternary phase diagram (Fig 8), each point of the triangle represents 100% of that component. The liquidus point can be easily defined by the composition of the three ash components selected. The solidus point depends on which of the many compounds present will melt first. The diagram shows that an ash with 45% SiO2, 20% CaO and 35% Al2O3 (composition A) will start melting at 1 170 °C, whereas an ash with 45% SiO2, 19% CaO and Al2O3 36% (composition B) will start melting at 1 345 °C. Thus a 1% difference in the Al2O3 and CaO content results in 175 °C variation in predicted melting points. Interpreting a complex phase diagram such as the SiO2-Al2O2-CaO diagram requires a good understanding of the principals of phase diagrams and what they represent and hence how to move from composition A or B to the respective solidus temperatures. Phase diagram selection is dependent on the dominant elemental components of the ash. Ideally, the components should account for more than 90% of the total elemental composition. For example, the table below shows typical ash analyses for a coal, straw and bagasse. If coal and bagasse were to be combined in variable proportions, then the CaO-Al2O3-SiO2 phase diagram would not be suitable because the Fe2O3 and K2O contents of the ash are too high. A quaternary diagram for CaO-Al2O3-Fe2O3-SiO2 or K2O-Al2O3-Fe2O3-SiO2 might be better. Complex software is available for predicting liquidus and solidus temperatures for these complex quaternary phase diagrams.
Based on the analyses of Table 5 and the hypothetical example quoted above, mixing bagasse (similar to A) with coal would be less likely to cause slagging than mixing straw (similar to B) with coal. This is attributed to the comparatively high proportion of K2O, SO3, CaO and P2O5 and predicted lower solidus temperature (Sh start melting) for the straw. Boiler Fouling The condensation, adherence and later sintering of volatile ash components on to convection heating surfaces causes fouling. The volatile components are usually the alkali metal oxides (K2O and Na2O). There are a number of indices, which point to fouling with biomass fuels. The ratio of the alkali metal oxides to silica ((K2O+Na2O)/SiO2) (Magasiner & De Kock, 1987) and to the gross calorific value ((K2O+Na2O) x % ash/GCV)) (Miles et al, 1995) have been found to be reliable indicators. These two indices supplement each other. Because the latter measures the absolute amount of alkali metal oxides per unit of heat input it may not indicate a potential fouling problem when the total ash content of the fuel is small and the alkali metal oxide to silica ratio is very high. On the other hand the first may overstate the potential problem when this ratio is moderate or perhaps high even though the total amount of the offending materials is too small to affect the commercial operation of the plant. The table below schedules values for these indices when it is unlikely that fouling problems may occur, when they may occur and when it is almost certain that they will occur.
Fouling when burning coal is more complex due to the presence of sulphur, which allows calcium oxides to volatilise. However, coal is assumed to be a fouling fuel and hence coal fired boilers are invariably fitted with sootblowers. This means that the more important parameter in fouling from coal is the mechanical strength of the ash deposit. Recent work undertaken in Europe (Heinzel et al, 2000; Platbrood & Van der Elst, 2000) and published last year in the USA presents a correlation between the elemental signature of the ash and its sintered strength. The relationship is complex. In general it indicates that the alkali metal oxides and calcium oxide increase the compressive strength and that aluminium and iron oxides reduce it. The biggest challenge is to predict when clinkering, slagging and fouling will occur when mixed firing. The only solution is to predict the likely ash chemistry from the expected proportion of each fuel being burnt. The same techniques as have been discussed for the individual fuels can then be applied, a decision being required as to which is most relevant. We are not aware of any solid scientific basis for predicting results confidently at this stage. Previous experience with the fuels is an invaluable aid in minimising risk.
Biomass fuels, and bagasse in particular, can be readily characterised with respect to their combustion performance. Grate rating, the amount of excess air required to complete combustion and combustion efficiency are all a function, interalia, of fuel moisture. Because it is the ratio of moisture to combustible material that determines how well a fuel will burn it is important to determine the "effective" moisture of a fuel by factoring ash content into the equation. Increasing the ash content effectively increases this ratio. Well proven methods such as the SMRI formula can be used to define bagasse calorific value. Similar formulae, without brix corrections, can be used for other fuels. The amount of excess air needed to complete combustion rises rapidly as effective fuel moistures increase above about 50 %. Grate rating can be defined as a function of effective fuel moisture and primary air temperature. This relationship is also dependent on excess air required to complete combustion, furnace temperature, gas up-flow velocity and particle terminal velocity. Biomass fuels, having ash contents of less than 8 %, can generally be burned on stationary grates. If they have higher ash contents it may be better to burn them on CAD stokers. Coal is often used as an auxiliary fuel in biomass fired co-generation power stations. Because of the need to get rid of the large quantities of ash generated by this fuel, CAD stokers are used on this type of installation. Thermo Gravimetric Analysis can be used in the laboratory to determine the relative performance of a particular coal. In practice, the method has its limitations because specialised equipment must be used to carry out the tests. We believe that maceral analysis can be used instead. In this paper we have postulated a relationship between what we define as the "effective re-active maceral" content of a coal and the amount that can be burned at reasonable efficiency per unit area of grate. We have also defined a relationship between this parameter and the amount of excess air needed to complete combustion. These relationships were developed from data collected from a number of geographical areas. Because installation characteristics varied from place to place, the results must be viewed with some degree of caution. They have, however, so far proved reasonably reliable. The analysis will be fine-tuned as more data is collected in the future. The use of phase diagrams to determine the effect which ash chemistry has on the potential of a fuel to form clinker on the grate and slag on the furnace walls is dealt with. This type of analysis is shown to be of particular importance in determining how the ashes of coal and biomass fuels can react detrimentally to enhance the formation of clinker and slag. The use of ash chemistry to determine whether a biomass fuel has the potential to cause boiler fouling is discussed. Two "fouling" indices are identified for this purpose. These indices have proved reasonably reliable practical indicators. When a boiler is fired with such disparate fuels as biomass and coal the boiler designer needs a thorough understanding of the fuel characteristics. It is not only necessary to be able to size the equipment but to be able to predict unburned carbon loss, the amount of excess air required to achieve this loss, and the potential for boiler fouling and clinker and slag formation. Zoning the stoker (Page et al, 2000) along its length provides a useful tool to help optimise combustion and minimise emissions. Acknowledgments The authors thank Nikki Wagner for helping to prepare the notes on petrography and in helping to make this science of use to the combustion engineer. References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
