1. INTRODUCTION
This study aims to quantify some of the environmental impacts associated with the combustion of New Zealand industrial coals. It focuses on carrying out an inventory of selected trace elements with particular emphasis on their partitioning between ash and stack emissions and the identification of some of the accompanying ash chemistry which may determine their partitioning behaviour.
Unlike previous studies where the focus has been on the use of bituminous coals in PF fired power stations the emphasis here is on the use of sub-bituminous coals typical of those used industrially in New Zealand under stoker and fluidised bed combustion regimes.
2. EXPERIMENTAL
A large sample (approximately 3 tonnes) of -30 mim industrial Waikato coal was mixed thoroughly and a representative sample taken for laboratory analysis. The remainder was split into firing charges (each approximately 20 kg) and stored in polythene bags until required. For the major elements, the coal was ashed and the ash analysed according to a standard method. (Australian Standard 1038 1981). Arsenic, boron, cadmium, cobalt, chromium, copper, nickel, lead, selenium, vanadium and zinc were measured by 1CP-MS after microwave digestion of the coal in a nitric/ peroxide/HF solution. Boron was also measured colorometrically using the curcumin method (Australian Standard 1038 1998). For mercury the digestion was carried out using a nitric/sulphuriclhydrochloric mixture in a water bath at approximately WC followed by hydride generation and cold vapour AA.
The coal was burned in both an underfeed stoker and a fluidised bed combustion unit each of approximately 50 kW firing rate.
The fluid bed had a diameter of 250 mm with 4 kg of sized sand used as the fluidising medium. This resulted in a static bed height of approximately 50 mm. The air supply was not staged, and a distributor of the perforated plate type was used. Start up was achieved with a LPG burner in the plenum chamber. Coal ignition was obtained at approximately 500C. Coal feed was by means of an overbed screw, with a coal size of -5 mm. The bed temperature was controlled by running water through a cooling tube in the bed. A temperature controller regulated this water flow to maintain the desired bed temperature within ±10°C.
The underfeed stoker consisted of a small (200 min diameter) retort, with a ceramic hearth in a steel furnace chamber. Coal size was +12,-30mm. No overfire air was used.
For both the fluid bed and underfeed stoker, the furnace exit gases
passed over a convective tube bank which
controlled the stack sampling temperature. Particulate matter was removed
by a high efficiency cyclone. Inlet air flow and stack gas flow were measured
using orifice plates, and thermocouples used for temperature measurement.
Samples (approximately 40 kg) of the Waikato coal were burned at 800, 875 and 950°C in the fluidised bed and at a higher temperature (in excess of 1000°C) in the underfeed stoker. In all test runs a gas sample was taken and passed though an isokinetic quartz sampling train to a quartz disc filter to remove any remaining particulates. The sample then entered a series of bubblers containing solutions to trap the trace elements. For all except mercury a solution of 5% nitric acid in 10% hydrogen peroxide was used (Ward 1993) while for mercury a solution of 4% potassium permanganate in 10% sulphuric acid was used (Nott et al 1994). The trace element concentrations in the impinger solutions were measured by 1CP-MS.
During each run samples of flue gas were withdrawn from a point slightly below the isokinetic quartz line, passed through a drying tower and into a gas chromatography gas analyser where levels of oxygen, carbon monoxide, carbon dioxide, methane and nitrous oxide were measured. Flue gas was also drawn from a third sampling port, diluted with inert gas to lower the dew point to below room temperature, and passed to a pulsed UV fluorescence sulphur dioxide analyser.
After each run samples of bottom, cyclone and fly (from the quartz filter) ash were recovered, weighed and analysed for the major and target trace elements. Total recovery of 95-98% of the inorganic matter originally present was achieved. The sand used as fluidising medium in the fluid bed experiments was also analysed for major and trace elements.
A representative sample of the coal was treated in an oxygen plasma to remove the organic material and generate low temperature ash (LTA) - a product that, as nearly as possible, retains the inorganic material present in the original coal. The LTA was analysed for major inorganic constituents. A sample of LTA was heated at 1°C per minute from ambient to 1100°C under reducing, neutral and oxidising conditions in the sample pan of a differential thermallthermogravimetric analyser (DTA/TGA) in order to determine temperatures where significant ash changes were taking place.
Samples of LTA were also heated at 10°C per minute in a tube furnace under neutral and reducing conditions. They were heated to 300, 600, 800 and 1100°C recovered from the furnace and examined by Fourier Transform infra-red spectroscopy and X-Ray Diffraction in order to identify the mineralogy changes undergone. A sample of LTA was also examined by the high temperature infra-red emission method (Vassallo 1992) to enable in situ monitoring of mineralogical changes as the sample was heated.
3. RESULTS AND DISCUSSION
Some properties of the coal are given in Table 1. Trace elements of interest, because of their known tendency to volatilise include arsenic, boron, selenium and mercury. The partitioning behaviour for underfeed stoker combustion and fluidised bed combustion at each of the three temperatures is given in Table 2-5 together with flue gas composition data in Table 6. For the fluidised bed experiments over 95% of the boron, selenium and mercury present came from the coal while for arsenic 66% came from coal and 34% was introduced by the fluidising sand.
The trace element of most interest in this study is boron as by world standards New Zealand coals contain high concentrations. Boron is reported (Clarke and Sloss 1992) to be highly volatile with up to 80% leaving the combustor in the vapour phase. It is sometimes classified with mercury and the halogens as a Class 111 (vaporisation, noncondensation) element although it is also found among the Class 111 elements - those that are volatilised but condensed downstream and concentrated on the fine grained particulates. A typical case is for a coal with a low boron content (16 ug B/g) giving a vapour phase concentration of 350 ug/m³ representing the release of 24% of the available boron.
In the present study for a coal containing 316 ug B/g the highest vapour phase concentration seen was 698 ug/m³ corresponding to release of only 2% boron. This was for underfeed stoker combustion. At the lower temperatures of the fluidised bed combustor the boron levels in the flue gas were lower corresponding to even smaller proportions of the boron being released in the flue gas.
One explanation is simply that we are failing to trap the volatile boron. For the set of experiments described here the boron mass balance ranges from 64% (excluding that in the very small amount of fly ash) for the underfeed run up to 84% in the fluidised bed experiments. However in previous runs on the underfeed apparatus recovery levels up to 92% were achieved with only 1 % being found in flue gas. (Clemens'1994)
Another possibility is that the boron is not volatilising because the temperatures in the fluid bed (800 to 950°C) and underfeed stoker (difficult to measure, but probably in the 1000 to 1100°C range) are lower than those normally associated with PE firing (150°C) to which all earlier boron studies relate. This is unlikely since in New Zealand subbituminous coals much of the boron is organically associated and may be expected to volatilise in the combustion chamber. Once volatilised in the form of boric acid it will remain gaseous provided it does not come in contact with a reactant or a surface below about 30°C.
Our fluid bed results show that much of the boron is volatilised at the lower combustion temperatures but is adsorbed by, or reacts with, the fine grained cyclone and fly ash particulates. This is in general agreement with partitioning trends found for other volatile elements under atmospheric fluidised bed combustion conditions (CRE 1-987), and pressurised fluidised bed combustion (Mojtahedi 1990).
In the underfeed stoker the majority of boron remains in the bottom\ash even though the temperatures are higher and the amount of fine ash material produced is considerably less. The main reason why boron does not escape the firebox may relate to the larger particle size but there could also be chemical factors now coming ini more decisively.
The DTA/TGA experiments on low temperature showed that some events occurred prior to 30°C with it most significant weight losses and thermal response occurring between 600 and 8O0°C under all condition, further small events continued up until 1100°C The X and FT-ir experiments on the low temperature ash sue, the following chain of events occur. The XRD of the temperature ash is dominated by calcium carbonate calcium with minor contributions from quartz and kaolinite. T presence of infra-red signals at 661 and 605 wavenumber indicate bassinite (the semi-hydrated form of calcium sulphate). After heating to 30°C under oxidise conditions these disappear and signals at 680, 616 and 5 wavenumbers, strongly indicative of the unhydrated form replace them. Before 60°C the kaolinite sign, at 1013 and 1034 wavenumbers are lost which is surprising as it usually persists to higher temperatures. Between 61 and 780°C the calcite breaks down accompanied by the loss of quartz signals and the appearance of calcium oxide signals. This is consistent with calcium oxide from the calcite breakdown reacting with the small amount of quartz give calcium silicates. When the quartz is consume calcium oxide remains. Above 780°C the peak intensities in the XRD diminish, suggesting the formation of poor.. ordered glassy phases arising from the calcium silicates.
may be that the release of boron from the firebox in the underfeed is hindered by the glassy phase formation. It is certainly securely bound in the bottom ash, judging by the small amount of boron released in leaching experiments
In contrast the behaviour of other target elements do not vary greatly between fluidised bed and underfeed conditions.
The behaviour of mercury was much as expected Previous studies of trace element partitioning under atmospheric fluidised bed conditions showed it to be the only element to consistently reach the flue gas in the gas phase (CRE 1987) and in our experiments the majority was recovered in this form. A small amount was associated with the fly ash and cyclone ash, with some degree of enrichment compared to the 78 ppb levels at which it was found in the coal. Its behaviour was not affected by the higher temperatures and larger particle sizes associated with underfeed stoker combustion.
The stack sampling temperature is of considerable importance when it comes to arsenic partitioning studies. It is released from the combustor as AS203 and will remain in the gas phase provided the temperature remains above 193°C. In the present set of experiments the stack sampling temperature in the underfeed run was only 179°C whereas for the fluidised bed runs it was between 203 and 211°C. This may account for the difference between the 11 ug/ dsm³ concentration found in the flue gas from the 950°C fluidised bed run and the 0.82ug/dsm³ found in the underfeed. The likelihood of this was confirmed by another underfeed run with a stack sampling temperature above 193°C. In this case an arsenic concentration of 4.8ug/dsm³ was found in the flue gas.
In all experiments the level of arsenic remaining in the combustor was high for an element expected to be found primarily associated with fine particulates and flue gas. This again is probably a consequence of the high calcite concentrations in this coal. Once the calcite has decomposed it is likely that the resulting CaO will react with arsenic oxide vapours to form calcium arsenate. Similar behaviour has been reported recently for a high calcium sub-bituminous North American coal (Bool 1995) and it is known that addition of limestone to coals undergoing emission tests can result in considerable reductions in arsenic emission levels.
Selenium behaved as expected with substantial enrichment on the cyclone and fly ash samples. None was found in flue gas suggesting the particulates were able to trap and condense all of this element. Some selenium was trapped in the bottom ash of the underfeed experiment - again a probable consequence of the larger particle size and the chemical and physical changes that occur at the higher temperatures associated with this mode of combustion.
Nitrous oxide levels in the flue gas were strongly temperature dependent. As the combustion temperature increased the levels decreased from 26 ppm in the fluidised bed at 800°C to 2 ppm in the underfeed. Even small temperature changes had a significant influence. Raising the fluidised bed to 875°C lowered the level to 19 ppm and at 950°C the level was down to 12 ppm The generally accepted: explanation for this behaviour is that increasing temperatures lead to increased concentrations of hydrogen and hydroxy radicals in the gas phase. Both of these species are known to very effectively reduce N20 to N2.
The S02 emission levels were also strongly temperature dependent although the high calcite concentrations of this coal exerted a major influence. The SO2 emission levels were seen (Table 6) to be well below the maximum possible level (approximately 150 ppm) for this sample under the fluidised bed combustion regime. At 800°C the sulphur retention in the cyclone and bottom ash was 61 % but as the temperature rose this steadily decreased to 22% by 950°C. This behaviour is consistent with the view that once the calcite breaks down (which occurs between 600 and 780°C in this sample) the resulting calcium oxide traps the SO2 as CaSO4. As the temperature rises two factors cut down the efficiency of the sulphur trapping. Firstly the volatiles bum more rapidly and reduce the amount of oxygen available to take part in the CaSO4 formation reaction. Secondly, around 880°C it is believed that- the- CaSO4 is-, reduced. by CO to reform CaO and liberate the SO2 The fluid bed results are in good agreement with this explanation.
Interestingly, the underfeed unit, despite having the highest combustion temperature, retained 23% of the sulphur in its ash - slightly more than that seen for the 950°C fluid bed experiment. This improved retention may be due to the large particle size in the underfeed, as the surface area exposed to the gas stream is less than that in the fluid bed for equivalent conditions. The reduced surface area would inhibit the ability of CO to access CaSO4 for reformation to CaO and SO2 The lack of turbulent mixing between the particles in the underfeed bed and the gas stream would also contribute to this effect, compared with the fluid bed. It is possible that the glassy phase formation thought responsible for boron capture may also assist sulphur retention.
4. CONCLUSIONS
Emission levels of N20 are strongly dependent on the temperature of the firebed, with increasing bed temperature leading to reduced flue gas concentrations. Emission levels are not significantly affected by combustion regime. S02 emission levels are also temperature dependent (they increase with increased temperature) but are significantly influenced by the method of firing with underfeed combustion at high temperature giving similar sulphur retention to fluid bed combustion at 950°C. This may be due to inhibited contact between the coal surface and CO in the gas stream of the underfeed combustor leading to suppressed reduction of CaS04 to CaO and SO2
Of the inorganic elements mercury is the only one emitted in substantial amounts in the flue gas. Emission levels are largely unaffected by changes to the combustion regime. For practical purposes it may be assumed that all of the mercury present in the coal will be emitted in the vapour phase.
Boron, arsenic and selenium show strong tendencies toward condensation on fine ash particles although significant amounts of arsenic are retained in the bottom ash recovered from both underfeed and fluidised bed combustion. This is a consequence of the calcite originally present in the coal at high concentrations breaking down and reacting with volatile arsenic oxides to form arsenates. Boron is retained slightly in the fluid bed but to a significant degree in the underfeed combustor. This may be due to the formation of a glassy phase from the calcium silicates which begin forming at around 780°C and become increasingly prevalent as temperatures in excess of 1000°C are reached. Selenium is not retained in the fluidised bed but is to some extent in the underfeed, possibly for similar reasons as boron.
These findings have important practical implications. It is clear that emissions of volatile inorganics are best controlled by a low flue gas temperature (to encourage their condensation on fine ash particles) and an efficient particulate collection system. Only mercury is likely to require a separate scrubbing system. However, the collection of the volatile inorganics in this way raises important issues relating to the disposal of the fine ash particles.
5.REFERENCES
Australian Standard 1038, Part 14.1 -1981.
Australian Standard 1038, Part 10.3 - 1988.
Bool 111 L E and Helbe J J (1995). A laboratory study of the partitioning of trace elements during pulverised coal combustion. Energy and Fuels, 9, 880-887.
Clarke L B, and Sloss L L (1992) Trace elements - emissions from coal combustion and gasification IEACR/49
Clemens A H, Korolev V, Matheson T W and (the late) Purchase N G (1994). Efficiency and emission studies for domestic and industrial applications in New Zealand. Proc. Clean Air Conference, Perth, Australia 407-415
CRE, Coal Research Establishment (1987) Trace element emissions from fluidised bed combustion units Final Report. Technical Coal Research EUR 11160 EN Commission of the European Communities, Luxembourg.
Mojtahedi W, Nieminen M, Hulkkonen S, and Jalikola A (1990) Partitioning of trace elements in pressurised fluidised bed combustion. Fuel Processing Technology 26, 83-97.
Nott B R, Prestbo E, Huyck K A, Olmez I, De Wees W, and Tawney C W (1994) Evaluation and comparison of methods for mercury measurement in utility gas stacks. The Air and Waste Management Association 87th annual meeting and exhibition, Cincinnati, Ohio, USA. (USEPA Federal Regulations 40. Protection of Environment part 61 App. B Method 101A)
Vassallo A M, Cole-Clarke P A, Pang L S K and Palmisano A J (1992) Infrared Emission Spectroscopy of Coal Minerals and their thermal transformations. Applied Spectroscopy, 46(1) 73-78.
Ward T E (1993) Multiple metals stack emission measurement methodology for stationary sources, current status. PB93 185734, EPA/600/A-931089, US EPA.