The relation of mineral composition to slagging, fouling and erosion during and after combustion

Pro~l. EnergyCombust. Sci. 1984. Vol 10, pp. 159-175.

0360 1285'84S0"0C+ 5 0 Pergamon Prc>,~ l t d

Printed in Great Britain.

THE RELATION OF MINERAL COMPOSITION TO SLAGGING, FOULING AND EROSION DURING AND AFTER COMBUSTION WILLIAM T. REID Battelle Columbus Laboratories, Columbus, OH 43201, U.S.A.

Abstract--Inorganic matter in coal has been a major cause of problems in fuel-burning energy-conversion systems since the dawn of the Industrial Revolution. Even in the earliest days, hand-fired boilers were limited in their steaming capacity by the formation of clinkers in the fuel bed. Today, many large coal-burning power plants are restricted in steam generation by accumulations of coal ash on heatreceivingsurfaces. These deposits not only decrease the rate of heat transfer but they also plug gas passages and lead to metal wastage by erosion and corrosion. As a result, coal-burning boiler furnaces must be appreciably larger, and hence more costly, than gas- or oil-fired steam generators, and operating practices are more critical when coal is the source of energy rather than the "cleaner" fluid fuels. The intent here is to describe the characteristics of the mineral matter in coal, to show how the noncombustible material behaves at the high temperatures of boiler furnaces, to explain how coal ash can lead to metal wastage by erosion and corrosion, and to review problems in the collection of flyash attributable to ash composition.

1. CHARACTERISTICSOF COALASH 1.1, Mineral Matter in Coal Mineral matter occurs in coal in two general forms: ( 1) as elements required for the growth of the original plant life from which the coal was formed and (2) as inorganic material, typically sand or silt, or clay, deposited in the accumulating plant debris as discrete particles of foreign material. Inorganic matter also may have been deposited later by mineral-laden water percolating through the coal seam. Simply, these two types of "ash" may be considered as "inherent" and as "extraneous" or "'adventitious" ash. Inherent ash seldom exceeds about 2 % in coal, and generally makes up about 3-20 % of the total mineral matter. 1 It cannot be removed physically as by coal preparation. Iron, calcium, magnesium, phosphorus, potassium and sulfur are commonly present in inherent ash but iron, calcium, magnesium and sulfur invariably are present in much larger quantities in the adventitious ash. Particle size varies widely. Inherent ash elements may be part of the coal maceral. Adventitious ash, sometimes present as partings in the coal bed, can be large enough for removal by float-andsink coal preparation techniques. In earlier days, "'rock", huge chunks of adventitious ash, was removed by hand in coal "breakers". The obvious difference expected in behavior between these two types of ash in boiler furnaces has never been investigated in detail. But the more intimate mixing of the molecularly dispersed ash-forming constituents of inherent ash compared with the massive individual particles of, typically, quartzite, pyrite or calcite in adventitious ash which then must collide in the gas stream or be captured by heatreceiving surfaces to react forming slag, are factors that have not been investigated experimentally.

A recent Russian development2 of a mathematical model of the behavior of mineral matter during combustion is typical of the efforts being made to understand the conversion of mineral matter into objectionable deposits. The eleven mineral species for which disintegration rates were calculated are those commonly present in coal, but the resulting assumed particle trajectories and the prediction of slagging zones and of areas likely to be damaged by erosion are only estimates at best. Bryers 3 has considered in great detail the special problems related to pyrites. Iron sulfide provides about half the iron oxides present in coal ash although this percentage varies widely, particularly between low-rank and high-rank coals. With higher-sulfur coals, the iron oxide contributed by pyrites is a potent flux, lowering the viscosity of slag deposits significantly. This, plus the high density of pyrites and its low grindability, contributes to slagging and fouling problems, particularly in areas in the furnace near the burners. Bryers concluded that the rate of oxidation of pyrites is much lower than the combustion rate of coal. Therefore, since the intermediate compound pyrrhotite (FeS) forms a eutectic with FeO melting at only 940°C, and because this eutectic can exist after the complete combustion of the coal, pyrites can contribute appreciably to slagging. Further, in regions near burners where conditions are not always strongly oxidizing, the low-melting FeO. F'eS eutectic can lead to sulfidation attack of furnace wall tubes. Determining the presence and quantity of minerals in coal is a difficult task. Using the thin-section technique developed by Thiessen of the U.S. Bureau of Mines in the early 1930's with identification in a petrographic microscope, and subsequent manual counting, is a good but tedious method. Electronic particle-counting techniques now available can simplify this task. A computerized counting and mineral

160

W.T. REID

identifying scheme known as CESEMI at Pennsylvania State University showed considerable promise but little use has been made of it. Petrographers have developed a computerized program in BASIC for calculating the composition of a silicate system in terms of standard, or "normative", mineral species.4 This program, called "PETCAL", uses chemical oxide values and ratios to compute probable mineral occurrences that can satisfy a given chemical composition. Although devised for petrographers with their interests in identifying and classifying mineral occurrences, PETCAL also can derive interesting relationships in minerals possibly present in a coal with ash of a known chemical composition. Some problems exist in this application because the ash of some low-rank coals is much lower in silica than naturally occurring rocks, because PETCAL does not deal with sulfur, and because PETCAL differentiates between ferrous and ferric iron whereas coal-ash analyses report only ferric oxide. Nevertheless, this rapid method of estimating the possible occurrence of minerals in coal based on chemical composition of the coal ash may well be worth some attention by fuel technologists. The "trace elements" in coal--essentially the entire periodic table--probably do not affect ash fusibility appreciably, although fluorine and glass-forming elements such as lead and boron may contribute to ash problems. These elements have been ignored by most investigators of ash behavior in furnaces. Minerals in coal--perhaps a hundred or so--are responsible for problems costing the electric-generating utilities many millions of dollars per year. Yet our knowledge of their influence on boiler furnace performance is minimal indeed. The mineralogy of coal ash would seem to be a fruitful field for future research, for it is the minerals in the coal that pass through the flame front that initiate all the eventual problems with slagging and fouling.

1.2. Composition of Coal Ash Seven major constituents make up most coal ash-SiO2, A1203, Fe203, CaO, MgO, N a 2 0 and K20. These are commonly reported as the oxides because

they result from relatively slow oxidation in a laboratory furnace of a sample of coal at 750°C, and because a chemical analysis of coal ash reported on the basis of oxides usually sums approximately to 100 ~o. Most analyses also include P205 and TiO2, although the levels seldom exceed one percent of the ash. Sulfur trioxide usually is reported in an ash analysis. Sulfur in coal ash analyses calls for special attention. Reported as SO 3, in amounts ranging from less than 1 ~o to more than 20 ~ , sulfur retention in laboratoryprepared coal ash is entirely an artifact of the ashing procedure. The level reported is a function of the form of the sulfur and the amount of calcite (CaCO3) in the coal, 5 but it also depends on the heating cycle as the coal is ashed and on the maximum temperature reached. The standard ASTM ashing procedure calls for heating the sample of coal to 500°C in 1 hr, and to 750°C in 2 hr, and then to ignite to constant weight at 750°C. 6 A high-temperature method 7 also is used for determining the total sulfur in coal whereby the ash sample is heated to 1350°C in a stream of oxygen and the resultant acid gases, including SO3, are captured by hydrogen peroxide and then titrated with NaOH. If pyritic sulfur is high, a special ashing procedure may be used whereby the sample of coal is held at 750°C for 4 hr with an ample supply of air over the sample to remove SO3, the intent being to oxidize pyritic sulfur before the calcite is decomposed and converted into CaSO4. Magnesia behaves similarly. Hence the SO 3 reported in an ash analysis depends upon the rate of oxidation of any pyrite in the coal, the calcination of calcite and dolomite, and the capture of SO 3 by the resulting CaO and MgO to form the sulfates. Coal ash prepared at 750°C is not comparable to ash heated in a pulverized-coal flame to temperatures at least as high as 1750°C, or maintained as slag on wall tubes at a typical temperature of 1500°C. Because CaSO 4 and MgSO 4 dissociate to CaO and MgO at temperatures above, roughly, 925°C, there is little or no SO3 present in slags at furnace temperatures. For example, chemical analyses of coal-ash slags seldom show more than 0.1 ~ SO3. Hence the most reasonable procedure is to recalculate coal-ash analyses on a sulfur-free basis in predicting the behavior of ash

TABLE 1. Composition of coal ash 8 Reported composition, percent State AL CO IL IN KY MT ND OH PA WV

County Jefferson Fremont Henry Greene Knott Dawson Mercer Belmont Clarion Harrison

SiO 2

A1203

Fe203

CaO

MgO

Na20

K20

SO 3

45.2 48.9 41.4 46.7 56.3 30.0 15.4 42.4 32.7 26.5

23.5 15.2 20.3 31.6 32.6 25.3 8.0 19.6 20.9 22.0

25.9 6.8 24.3 12.8 4.5 2.9 9.2 27.5 43.3 47.8

1.8 11.3 6.8 1.8 1.7 11.7 23.3 4.8 1.4 1.7

0.9 0.9 1.3 1.2 1.5 4.9 6.7 1.2 0.2 0.4

0.2 0.2 0.5 0.3 0.5 8.1 7.1 0.2 0.1 0.5

2.6 0.2 2.1 3.3 1.6 0.4 0.1 1.6 0.8 0.8

0.6 14.9 2.4 0.4 0.9 12.6 27.4 2.0 1.4 2.8

Mineral composition and combustion at furnace temperatures. This normalized percentage composition poses fewer problems in understanding ash behavior, Such arguments for recalculation to a sulfur-free basis do not apply to flyash, where the particles of ash cool from furnace temperature to about 200°C in flue gas containing, typically, 25-50 ppm SO 3. Under such conditions, SO3 condenses on the surface of the flyash and is a major factor in establishing the electrical resistivity of the flyash, in turn affecting the collectability of the flyash in electrostatic precipitators. The range of composition of coal ash is so wide that there is no "typical" ash analysis. Table 1 illustrates how greatly ash composition can vary. These analyses, as reported by the Bureau of Mines, are not normalized by excluding TiO2, P205 and SO 3. Of these, recalculations on an SO2-free basis has a pronounced affect, as, for example, with the North Dakota lignite from Mercer County where recalculating by excluding SO3 increases the total percentage of CaO and MgO in the ash from 30.0 ~o to 41.3 Vo- This latter figure is the significant one in predicting ash fusibility. Although these examples were picked essentially at random from R.I. 7240, they scarcely hint at the great variability in ash composition that can exist in different coals. Even in any one coal seam, the composition of the ash can vary substantially. And even in a given mine, ash composition is not constant. Hence, the starting point in any evaluation of the mineral matter in coal must include, as a minimum, the chemical composition of the ash.

1.3. Fusibility of Coal Ash The complex array of minerals in coal, dominated by kaolinite and other clay minerals, admixed with fluxes such as pyrites and calcite plus many other reactive species, often begins to sinter at temperatures as low as 650°C. The rate of these solid-solid reactions may be low, but a liquid phase apparently can develop even at 650°C at the interface between some mineral particles. At temperatures above 1000°C, appreciable "melting" can occur but the liquid phase may be highly viscous. At temperatures above 1400°C, reaction between the minerals results in a liquid m e l t - - a "slag"--usually containing no solid phase, and with a viscosity depending upon composition. Reaction rates are temperature dependent. It is the formation of this liquid phase that leads to major operational problems in large boiler furnaces fired with pulverized coal. 1.3.1. Cone fusion Clinker formation in fixed fuel beds was sufficiently troublesome in the early 1900's that the cone fusion test was devised by Fieldner 9 as a method of measuring ash fusibility. This procedure was adopted by ASTM and is now identified as D-1857; it is still the most widely used test method for ash fusibility. Coal ash, prepared by usual ASTM methods involving heating to 750°C, is thoroughly ground, mixed

161

with a small amount of dextrin as a binder, and formed into triangular "cones" 19mm high with a 6.3 mm base. These cones are mounted on a refractory support and heated at 8°C per minute above 800°C in a reducing atmosphere or in air. Temperature is recorded when the tip of the cone first deforms (IT): when the cone has melted into a spherical lump with the height equal to the width (ST): when the height is half the diameter of the base (HT); and when the cone is melted into a layer not more than 1.6 mm high I FT ~. The method is completely empirical, and interpretation of the four values is difficult, based in large part on the low heating rate during the determination. At the IT, enough liquid phase with a moderate viscosity has been formed for some flow to occur, but the remaining skeleton of unmelted solids limits the extent of flow. At the ST and at the HT, observed flow depends upon both the viscosity of the liquid phase and the proportion of the liquid phase to the remaining solid structure of unmelted and unreacted ash particles. And at the FT, viscosity, surface tension, sorption of the melt by the refractory base, and any still unreacted solids influence the temperature where the cone is observed to be "fluid". In an attempt to resolve such questions, the Bureau of Mines determined the cone fusion temperatures of 35 slags after measuring their viscosity-temperature relationship in a high-temperature viscometer. ~° These tests showed that, at the IT, the viscosity ranged between 4 and 10,000 Nsec/m2: at the ST between 0.3 and 10,000 Nsec/m2; and at the ST, between 0.1 and 10 Nsec/m 2. This great variability in the physical state of the ash as it is heated makes it extremely difficult to interpret cone-fusion data. Although many investigators have related cone data to ash composition, the scatter is always great. One of the major boiler manufacturers relies on the melting range between 1T and FT as a guide to slagging behavior in large boiler furnaces, but, generally, the cone fusion temperatures are considered only as a rough guide to ash behavior under furnace conditions. 1.3.2. Flow behavior Even the earliest pulverized-coal-fired boiler furnaces had problems with ash melting in the lower parts of the furnace, so that a "slag screen" was improvised at the bottom of the furnace to cool molten ash particles below the temperature where they would adhere to form objectionable massive deposits of slag, Over the next several years while the utilities were developing expertise in burning pulverized coal and the boiler manufacturers were devising improvements in furnace design, slag continued to be troublesome. Finally, late in the 1920's the first slagtap furnace was converted from a dry-bottom unit at the Huntley Station in Buffalo; a pit was provided at the bottom of the furnace where slag could accumulate, be kept hot, and be removed periodically as molten slag. This innovative development was widely accepted. Over the next few years many slag-tap furnaces were built or converted from dry-bottom

162

W.T. REID

furnaces, but it was soon apparent that all coal ashes did not melt at furnace temperatures to produce a slag fluid enough to be tapped, except perhaps at maximum furnace load. Cone-fusion data were not sufficient to choose coals with the proper ash behavior. In the mid-1930's, the Bureau of Mines undertook a study of ash chemistry as related to slag viscosity, using a high-temperature Margules-type viscometer to determine slag viscosity at temperatures up to 1620oc.11 This study showed that most coal-ash slags were not Newtonian fluids except at the highest temperatures, and that the separation of a solid phase as the slag cooled had a radical effect on the flow characteristics of the slag. ~° The temperature where this transition occurs, called Tcv (the temperature of critical viscosity), is approximately the liquidus temperature of the slag. Determined on the basis of non-Newtonian flow rather than simply the appearance of the first solid phase as the slag cools, Tcv is a useful parameter in determining the thickness of slag on heat-receiving surfaces as well as the flow of slag through a tap-hole from a molten pool. Four variables establish the flow properties of coalash slags: chemical composition, temperature, time and state of oxidation. 1.3.2.1. Chemical composition. This is a complex parameter because at least seven components are involved. Two simplifications have been derived--the "silica percentage ''al and the "base-acid ratio" for calculating viscosity, a2 Obviously, the two are closely interrelated:

Converting from base:acid ratio to silica percentage to utilize the silica-percentage equation can be done by: Silica percentage = 92.33 - (66.67) (base :acid ratio). Calculating Tcv is possible using an equation from BCURA: 13 Tcv = 2 9 9 0 - 1470(A) + 360(A2) - 14.7(B)+0.15(B 2) where A = SiO2/A120 3, B = Fe203 + CaO + MgO, and Tcv is in degrees C. This method was derived with British coals where the SiO2/A120 3 ratio is often lower than for U.S. coals. Such calculations are not reliable when the SiO2/ A120 3 ratio is greater than 3.0. Cone fusion data have been used to estimate Tcv based on the assumption that Tcv occurs at the softening temperature (ST) plus 110°C. Other data 1° shows a direct coincidence between Tcv and ST, but variances occur greater than 100°C. 1.3.2.2. Temperature. The temperature at which viscosity is commonly reported can be taken arbitrarily as 1426°C because most slags behave then as Newtonian fluids. The following equation is useful for calculating viscosity in poises at temperature above Tcv; r/-°1614 = (0.0004519) (Temp, F ) - B

where B is aconstant for a known viscosity calculated for a specified temperature in degrees F.~ 0 = (Sio2 sio= "~ 1.3.2.3. Time. This is a little recognized parameter in + Fe20~+-CaO + MgO J (100) establishing the flow characteristics of coal-ash slags Base: acid ratio at or below Tcv. Because the separation of a solid Fe203 + CaO + MgO + N a 2 0 + K 2 0 phase from a Newtonian slag can occur slowly in a viscous melt, many hours can elapse before equiliSiO2 + AI2Oa + TiO2 brium is established between the liquid and solid Both ignore ratios of the fluxes, mainly Fe2Oa, CaO phases. Contrarywise, this rate of development of the and MgO, and both predict viscosity of complex silisolid phase can occur almost instantly, so that a slag cate melts with at least fair success. Silica percentage may freeze into a solid mass within a few minutes after is particularly useful because it relates viscosity in being cooled 10°C below Tcv. Consequently, the poises at 1426°C to composition on a semi-log plot, behavior of slags at a lower temperature than Tcv is according to the equation: unpredictable presently. It is generally assumed that slags will "freeze" below Tcv so that no flow will occur log q = (0.05784)(silica percentage)- 1.8452, under normal gravitational forces. This point is imwhere r / = viscosity in poises at 1426°C. portant as will be noted later in discussing the charBased on a linear regression analysis of viscosityacteristics of slag deposits on heat-receiving surfaces. temperature-composition data on 30 coal-ash slags, 1.3.2.4. State of oxidation. For slags this is importhe correlation coefficient here is 0.989, attesting to tant because the iron oxides FezO 3 and FeO are the usefulness of this relatively simple calculation. In major fluxes in lowering the viscosity of silicate melts. these slags, the F e 2 0 3 : ( C a O + M g O ) ratio varied Because F e 2 0 3 dissociates at high temperatures, coalfrom 0.25 to 10.7, and the SiO 2 :A120 3 ratio from 1.0 ash slags melted even in air are partially "reduced", to 4.0 without deviating significantly from the mean. with an Fe203 :FeO ratio dependent on composition Hence the flux ratios are not critical in determining ' and temperature. The state of oxidation of slags is Newtonian viscosity. often expressed as the "ferric percentage", numerically The base:acid ratio has not been similarly defined calculated as: in terms of viscosity but curves have been published Fe203 plotting temperatures as a function of base : acid ratio Ferric percentage = × 100 for a given viscosity at a specified state of oxidation. 12 F e : O a + 1.11FeO + 1.43Fe Silica percentage

Mineral composition and combustion with Fe203, FeO and Fe determined by a forms-ofiron analysis of the slag. Under reducing conditions, as in a combustion atmosphere, the ferric percentage commonly will be as low as 10; under oxidizing conditions, as in air, the ferric percentage typically can be about 60 at 1426°C. The effect of this change in state of oxidation is mainly on Tcv, which can be lowered by as much as 300°C in going from a ferric percentage of 50 to 10, with a slag containing, say, 35 ~ equivalent F e 2 0 3. The Newtonian viscosity, measured above Tcv, is essentially unchanged. The probable explanation is that the primary solid phase separating from the melt is hematite (Fe203) in such a slag under oxidizing conditions. Under reducing conditions, with a ferric percentage of 10, fayalite (FeO.SiOz) is the major iron-containing phase and there is less likelihood of the separation of solids as shown by Tcv. Above Tcv, both forms of iron are equally effective in breaking the silicate chains. In summary, slags melted under oxidizing conditions will tend to "freeze" at a much higher temperature than when melted under reducing conditions, but at temperatures above the freezing temperature the Newtonian viscosity is not changed. The extent of this action depends upon the total iron content of the slag. It is pronounced for Eastern bituminous coals with ash often containing 35 ~o or more equivalent Fe203; it is minimal for Western low-rank coals where CaO is lhe major flux rather than Fe20 3.

1.4. Physical Properties The most important physical characteristics of coal ash under furnace conditions are those related to fusibility. At the high temperatures of combustion systems, typically in pulverized-coal-fired boiler furnaces, even the most refractory coal ashes show some development of a liquid phase, Most coal ash "melts" well below about 1500°C, and it is this characteristic that leads to most problems with coal as an energy source. Other physical properties of coal ash involve factors affecting heat transfer, such as thermal conductivity, emissivity, and surface tension. The density and heat capacity of slags have been investigated in connection with heat balances of slag-tap furnaces. 1.4.1. Thermal conductivity Heat flow through slag deposits on heat-receiving surfaces, such as the water-cooled wall tubes of large boiler furnaces, varies not only because of the changing and usually unknown thickness of the slag layer but also because the extent of melting of this layer differs from point to point. Further, with the wall-tube surface temperature not higher than about 400°C and with the outer surface of a thick slag deposit often above 1500°C, the outer layers of slag are more completely melted than the inner ones. A further complication arises because volatile species such as N a 2 0

163

TABLE2. Thermal conductance of ash deposits from combustion systems, (W/m. deg K).I 4 Mean temperature, °C Melting stage of deposit Particulate Sintered Solid Fused

200 0.14 .

400

600

800

1000 1200

0.17 0.24 0.31 0.35 0.36 0.68 1.02 1.38 0.57 0.85 1.18 1.54 . . .

1.97

and K 2 0 tend to migrate through the deposit toward the cooler inner layers, thus affecting the fusibility of the ash. These differences in the degree of fusion affect thermal conductance through the deposit. Table 2 shows the thermal conductance of deposits from a combustion chamber measured over a moderate temperature range as a function of the degree of melting of the deposits. ~4 It is evident that fusion affects conductivity substantially; even the increased sintering between 200°C and 1000°C more than doubles the thermal conductance. Further, "solid" deposits conduct more than four times greater at 1000°C than do particulate deposits. And completely "fused" slag is about 14 times more thermally conductive at 1200°C than is a particulate deposit at 200°C. A further point is that there is nearly a linear increase in thermal conductivity in the temperature range of 400°C-1100°C for all deposits other than particulate ones, where an abrupt increase in conductance takes place at about 1000°C under the conditions of this test procedure. For comparison, the thermal conductivity of a fused deposit at 1200°C is only about one twentieth that of mild steel. Such measurements confirm the expectation that thermal conductance of deposits is related to their density. Unfortunately, this is not particularly helpful, since little is known about the actual density of deposits from point to point from the innermost layer adjacent to a heat-receiving surface to the outermost layer receiving radiation from the flame. Further research is warranted on the thermal conductivity of ash deposits under actual furnace conditions. 1.4.2. Emissivity Coal-ash deposits do not behave as a "gray" body, thus the emissivity varies with wave length. Further, the chemical composition of the deposit has a pronounced effect on emissivity; a glass-like surface such as would be likely with a Newtonian slag at furnace temperatures would be expected to have a lower emissivity than a crystalline surface, as for a highF e 2 0 3 slag below Tcv where hematite is present on the surface. This point seems not to have been recognized generally. Emissivity approaches unity at temperatures below about 400°C, and averages about 0.6 at 4.5 lam at higher temperatures. At 3.5 grn, emissivity is roughly 0.4.15 Large coal-fired furnaces are not black bodies, and emissivity of slag surfaces affects optically determined measurements of slag temperature.

164

W . T . REID

1.4.3. Surface tension Agglomeration of ash particles on the surface of a burning particle of coal establishes at least in part the size of slag droplets downstream of a pulverized-coal flame. Liquid droplets of low-melting mineral matter do not wet a carbon surface, but the ash droplets coalesce eventually as the carbon burns. Surface tension undoubtedly influences such agglomeration, but no real effort seems to have been made on the affect of surface tension on the size of ash droplets downstream of the flame. Hey 13 has quoted an earlier study of the surface tension of glass in dynes/cm at 1400°C, using this equation: Surface Tension (1400°C) = 3.24SIO 2 + 5.85A120 3 + 4.4Fe2 O 3 + 4.92CAO + 5.49MGO + 1.12Na20 + 0.75K20. On this basis, slag from Pittsburgh No. 8 coal would have a surface tension of 410dynes/cm 16 at 1400°C. Unpublished measurements at 1450°C suggest that the surface tension of coal-ash slags probably ranges between 350-400 dynes/cm. 1.4.4. Density Iron oxide content has the greatest influence on the density of coal-ash slags since F e 2 0 3 has a density of 5.24 whereas the other oxides present range from 2.32 (SiO2) to 3.6 (MgO). Density usually is not a property of great interest; it has been used mostly for estimating heat lost in tapped slag based on a known volume of slag. Density also influences the flow of molten slag on vertical surfaces such as furnace wall tubes. 1.4.5. Heat capacity Of little moment except in heat balances, the heat capacity of slags above 15°C range approximately from 200kJ/kg at 1100°C to 400kJ/kg at 1700°C. 13

2. BEHAVIOROF COAL ASH IN PULVERIZED-COAL-FIRED BOILER FURNACES The major part of the coal being burned today raises steam in the pulverized-coal-fired boiler furnaces of the electricity-generating public utilities. Almost two-thirds of the coal mined in the U.S. goes to this market. With the shift away from oil- and gas-fired power plants and the present public outcry against nuclear stations, this use of coal is certain to become increasingly important in the years ahead. The behavior of coal mineral matter in boiler furnaces has been investigated mainly in an empirical fashion. The complexity of the system involving so many variables has limited what might be classed as "fundamental" studies although many investigators have attempted to apply phase diagrams of welldefined few-component systems. Such efforts will certainly continue even though only moderate success has been achieved as yet.

2.1. Slagging The formation of a liquid phase and the eventual complete "melting" of the mineral matter in coal are responsible for the worst of the ash-induced problems. Modern furnaces are essentially water-cooled steel enclosures with a cross-section perhaps 50 ft (15 m) by 80ft (24m) and a height typically of 175ft (53m). With an energy input in the form of pulverized coal of about 6TJ/hr, such a boiler furnace can produce about 2000 metric tons of steam per hour at 540°C, enough to generat about 700 M W of electrical energy. The burners in such furnaces are located about midway in this huge box, either at the corners (tangential firing), or in one wall or two walls (opposed firing). The hot flue gases leave the furnace at the top at about l l00°C through convection tube banks serving as superheaters and reheaters, and sometimes through radiant surfaces. The walls of this furnace are made of mild steel tubing, welded with narrow flanges into vertical panels which connect top and bottom into headers to serve as water and steam manifolds. Preheated water is pumped into the bottom header and steam is collected at the top. Boiler pressure may be, typically, about 17MN/m 2 (2400psi) for subcritical units or 2 4 M N / m 2 (3500psi) for supercritical units. Heat transfer to these wall tubes is roughly half by radiation and half by convection; the maximum permissible heat-transfer rate is about 550 kJ/sec m 2 (1"/5,000 Btu/ft 2. hr), limited by water-side conditions. Steam temperature leaving the superheaters is almost universally 538°C in large utility boiler furnaces. This is a critical temperature as far as the turbine is concerned and it must not be exceeded. Therefore, balancing heat transfer to wall tubes and superheaters is critical.

2.1.1. Accumulation of slag on heat-receiving surfaces If, at a given firing rate, heat transfer in the furnace decreases because of slag on wall tubes, the furnace exit gas temperature rises, increasing the rate of heat transfer in the convective superheater tubes. Steam temperature therefore increases above 538°C as demanded by the turbine. Steam temperature then can be lowered by injecting ultrapure water into the steam line (desuperheating), but the quantity of such water available is limited. Accordingly, when the limits are reached in desuperheating, the only recourse is to lower the firing rate, meaning a decrease in the total amount of steam generated and hence the electrical output of the turbine-generator. Slagging, therefore, limits the load. The rate of change in heat flux with slagging decreases as slag accumulates. In a typical case 17 beyond the first hour, the time for a given decrease in heat transfer increased linearly with time since the tube was deslagged, according to this equation: Hrs ----AT

(0.0092)(Hrs)-0.0032

Mineral composition and combustion where AT is the temperature gradient through the wall tube in degrees F, and Hrs is the time since the tube was deslagged. In this unit, heat transfer normal to the tube surface decreased by nearly half in the first hour after deslagging, and by 80 ~o over one day. 2.1.2. Removal of slag deposits Two techniques are used normally to remove these troublesome deposits of slag: (1) wall blowers using high-velocity jets of steam or air, or even water, to break the bond between slag and wall tube; and (2) decreasing furnace temperature by lowering the firing rate. In extreme cases, hand lancing slag deposits with streams of water has been used, but this method is not widely applied because of concern about potential damage to hot pressurized wall tubes. The effectiveness of wall blowers varies widely. Occasionally slag is bonded so firmly to wall tubes that even the best available wall blowers cannot loosen such deposits. There is no completely satisfactory description of the bonding mechanism between slag and a metal surface at, say, less than 400°C. Slag is essentially a solid at such temperatures. Most explanations of bonding are based on the presence of alkalies, mainly NazSO4, at the interface between slag and metal. Some of this sodium may have migrated through the slag under the thermal gradient established by the slag, but most seems to have come from condensation of alkalies volatilized in the flame at temperatures higher than, typically, 1800°C and condensed on the tube surface at less than 400°C. Lowering load is the commonly used method of deslagging when blowers are ineffective. Load decreases of 1 0 - 2 0 ~ for periods of many hours will often loosen slag deposits so that they break free or are removable by wall blowers. Probably, this is due to a difference in the thermal expansion of slag deposits compared with the furnace walls, but this is based more on supposition than on fact. The fact remains that no completely acceptable explanation exists on why some slags bond so tightly to furnace walls. This is a promising field for investigation.

2.1.3. Effect of additives For many years, efforts have been made to modify slag behavior with "additives", generally accepted as small quantities of chemical compounds added to the coal or blown into the furnace. Additives, mainly MgO, have been used successfully with residual fuel oil to minimize problems with vanadium, but less success has been achieved with additives when burning pulverized coal. A recent review by EPRI of the use of additives by coal-burning utilities 18 showed that the high ash content of coal compared with fuel oil greatly limited the beneficial effects to be expected with economically acceptable quantities of additives. Fluxes based on boron have been used successfully to melt down localized slag accumulations, as near burners. Limestone and dolomite are used frequently with slag-tap furnaces to lower the viscosity of slag to improve slag flow at moderate loads. Magnesia, a

165

highly effective additive with fuel oil, added with the coal or blown into a furnace as a slurry has had wined success in decreasing slagging problems. Generally, the utilities do not use such additives. Copper oxychloride, Cu2(OH)3CI, added to coal at the low rate of 1-15 ppm has attracted much attention recently? 9 Investigated mainly in England, copper oxychloride under some poorly defined conditions has radically improved slag control on wall tubes although such beneficial actions are not always reproducible. Copper normally present in coal ash generally exceeds the amount of copper in the additive. hence it is assumed that copper in the vapor phase, as from the oxychloride, reacts differently than copper already present in a mineral form. Copper oxychloride appears to enhance crystallization in coal-ash slags and to produce micro-bubbles that make the slag more friable than untreated slag. The future status of copper oxychloride as an additive is uncertain. Research on the manner in which copper oxychloride affects slag properties is now going on in England.

2.2. Fouling Coal-ash particles leaving the flame but not captured as slag or bottom ash in the furnace flow with the hot flue gas through the convection passes of the superheaters and reheaters. These particles accumulate as deposits in convective tube banks, interrupting heat transfer, setting up conditions that lead to metal wastage, and blocking the gas passages. 2.2.1. Mechanism (~ldeposition The range of particle size of ash constituents downstream of the flame is exceedingly wide. The smallest particles condensed from volatile species approach molecular dimensions. The biggest particles may be even larger than the pulverized coal (75 ~tm typically) because of the lower grindability of mineral matter and because molten droplets of ash may have impacted other particles to form still larger droplets. Hence velocity and turbulence as well as temperature affect particle size. It is generally thought that the kinetic theory affects motion of the smallest ash fragments, and that the gas laws control motion and collision frequency, collision being particularly important because this leads to larger particles. Brownian motion with particles of, perhaps, 1 gm size, leads to some deviation from the flow pattern of the main gas stream, but the particles still are small enough that gas flow mainly determines their path. Turbulent diffusion of particles in the size range up to about 3 p.m establish particle motion toward surfaces mostly dependent on inertial effects in eddies. And, finally, inertial impaction moves large particles toward surfaces since such particles once moving on a given path continue on that path despite minor changes in gas direction. Much theory has been applied to the problem of ash deposition on surfaces, but few definitive experi-

166

W.T. REID

ments have been conducted and those mostly in England. 2° 2.2.2. Effect on furnace design Retractable "soot" blowers installed to traverse convective tube banks, like wall blowers, use highvelocity jets of air or steam to dislodge ash deposits. Generally, soot blowers are effective, but occasionally deposits accumulate that cannot be removed with soot blowers or by hand lancing through available access doors. In such cases, a forced outage results and the tube banks must be cleaned manually from inside the furnace. To minimize the need for such costly cleaning, furnace designers vary the spacing between tubes, making the space between tubes large as the flue gas enters the tube bank and less as the gases cool flowing through the heat-receiving surfaces. This spacing also depends on ash characteristics, a low-fusion-temperature ash requiring a more generous design. The spacing between tubes varies substantially, from as much as 40cm (18 in) at the entrance to the tube bank at gas temperatures as high as 1200°C (2200°F) to as low as 10cm (4in) at the exit. Despite such design consideration, troublesome coals may still foul convection surfaces excessively, leading to unscheduled outages for cleaning. 2.2.3. Sinterin 9 test Ash characteristics influence tube spacing too. If a liquid phase occurs in coal ash at relatively low temperatures, say, below 900°C, then the particles of ash will adhere ("sinter") sufficiently to agglomerate into effectively solid masses. This sintering can occur because of a low melting point of a large proportion of the ash particles, and it may occur when only the surface of the particles is "sticky". A simple experimental procedure has been used for about the past 25yrs to determine this sintering temperature. It is based largely on the premise that N a 2 0 and K 2 0 volatilized in the flame at temperatures well above, say, 1750°C, then condense on the surface of ash particles as the particles cool passing through the heat exchangers. This relatively thin but rich layer of alkalies on the surface of particles that are predominantly silicates, forms a low-melting surface layer that encourages adhesion to other similarly treated particles. The result is a moderately solid structure at temperatures well below the temperature where the bulk of the ash particle begins to "melt". The test involves ashing a coal in a laboratory combustion rig, and simulating the time and temperature conditions in a pulverized-coal-fired boiler furnace. 21 The resultant "flyash", or flyash from an operating utility furnace, is screened to pass 60 mesh, and heated to 500°C to burn off any remaining carbon. Cylindrical specimens 15 mm in diameter and 19 mm high are formed from this ash in a hand press at 1 M N / m 2, and then heated to a temperature usually between 800°C and 980°C for 15 hr. When cooled, these cylindrical test specimens are crushed in a

compression testing machine. This "crushing" strength is taken as a direct measure of the sintering tendency of the flyash. Although completely empirical, the sintering test differentiates well between fouling and nonfouling coals. It has shown that sintering strength is directly related to the fouling tendency, that sintering increases with an increase in flame temperature, and that an increase in sodium content increases sintering but that potassium has a minor influence. The existing test appears to be a research tool not exploited as well as it deserves. 3. METALWASTAGE Boiler tubes, both wall tubes and tubing in convection banks, are made from iron alloys that develop tightly adhering oxide surfaces. The rate of formation of oxides and possible scaling during service determine the probable life of these pressurized components, nominally expected to be 30 years. Any accelerated oxidation rate, or the mechanical or chemical removal of this oxidized surface leads to excessive loss of metal and to early failure of the component. Hence oxidation resistance and mechanical properties both determine the suitability of an alloy for use in boilers. 3.1. Erosion Mechanical removal of the protective oxide scale on a tube by impinging ash particles can cause severe loss of metal. Superheater, reheater, and economizer elements, fan blades and tubular air heater elements are particularly susceptible to failure by erosion because gas velocity tends to be high in these locations. Many tube failures attributed to erosion have been caused instead by external corrosion. But where solid particles are moving at high velocity over metal surfaces, the failure rate by erosion can be severe. Ash characteristics such as particle size and "abrasiveness", and the angle of impact, all influence erosion, but velocity of the particles is the most important factor in coal-fired boilers. According to Raask, 22 for a given loading of ash particles, the rate of metal loss is proportional on average to particle velocity to the 2.5 power. Experience at CEGB shows that erosion loss is negligible for gas velocity not exceeding 20 m/sec but 30-40 m/sec shortens tube life to less than 50,000 hr. An "abrasive index", numerically the ratio of the erosion wear by ash to the erosion by 100-grn quartz grains under the same test conditions, has been used to predict wear by erosion of mild steel tubes in an ash-laden gas stream. The abrasiveness of coal-ash particles depends on many factors in addition to the quartz content, including the shape of the ash particles, whether "angular" or "spherical", and the "hardness" of the particles. Angle of impact is also important. 23 There is no widely accepted measure of erosion nor any generally used test method. Boiler

Mineral composition and combustion designers apparently set an arbitrarily established upper limit to gas velocities past erosion-prone surfaces to attain reasonable component life. 3.2. External Corrosion Metal wastage by chemical attack was first recognized as a major problem in pulverized-coal-fired boiler furnaces in 1942 when the wall tubes of slagtap furnaces began failing catastrophically. The area where metal loss was most severe was in a pattern related closely to the flame configuration. It was shown, however, that the rate of heat transfer in this region was within normal limits, that the tube metal temperature was not excessive, and that corrosion was occurring only where the wall tubes were covered by slag. 2~ Following an extensive laboratory and field investigation, the metal loss was shown to result from chemical attack by alkali iron trisulfates, typically K3Fe(SO4)3 .2s Although the potassium salt is more stable than the corresponding sodium salt, neither trisulfate can form nor can it exist at a temperature of 600°C except in an atmosphere containing at least 250 ppm SO 3. The reactions leading to metal wastage were taken as: 3Fe + 202 ~ FeO. Fe20 3 2FeO • Fe203 + 9K 2SO4 + 9SO 3 + 1/20 2 6K3Fe(SO4)3 3Fe + 2 0 2 ,~- F e O ' Fe/O 3 whereby the normal oxide film on the metal surface is converted to the trisulfate by reaction with an alkali sulfate and sulfur trioxide; the metal surface subsequently re-oxidizes to establish its normal oxide film at the expense of the parent metal. Hence the metal loss. Although the metal temperature here does not exceed about 370°C, the high heat flux to these wall tubes and the relatively poor contact between deposits and tube metal increases the deposit temperature to as much as 620°C. At this temperature, the trisulfates will not form nor will they be stable unless the SO 3 level is at least 250 ppm. Because only about 1 ~o of the sulfur in coal burned in a pulverized coal flame is oxidized to SO 3, the usual level of SO 3 following such a flame nominally is 25-35 ppm, far too low to account for the presence of the trisulfates. But in the irregular gap betwen tube surface and the overlying slag, penetrated by the products of combustion, the catalytic activity is high and the time is ample to convert a large fraction of the SO z to SO 3. Thus, in this stagnant layer, the SO 3 easily exceeds 250ppm and the trisulfates form at the expense of the oxide film on the metal surface. Another mechanism has been proposed to explain metal wastage by the trisulfates, but related to flyash on convective tube banks rather than to slag on wall tubes. 26 Here the iron oxides and the alkalies in the ash deposit react with catalytically formed SO 3 from

167

the flue gas to form the trisulfates, which then migrate through the thermal gradient to accumulate on the tube surface. The molten trisulfates then attack the tube metal at metal temperature as high as 590°(?. There is no consensus on these two probable explanations of metal loss; there is good agreement on the importance of the trisulfates. Recent efforts to develop alloys for superheater service specifically intended to resist corrosion by molten trisulfates has shown that systems containing aluminum, silicon and manganese are as much as ten times better than commercial stainless steel. Also, a magnesium zirconate coating, applied by plasma spraying to a thickness of no more than 250gm to resist spalling with temperature changes, is resistant to penetration by trisulfates and shows considerable promise in extending the life of superheaters.2 Sulfidation as a cause of metal wastage is a serious problem in gas turbines where metal temperatures are much higher than in boiler furnaces, and where alkali sulfates can be reduced locally to sulfides. In coalfired boilers, sulfides are troublesome only in areas, typically near burners, where raw coal impinges on slag-covered metal surfaces. 4. FLYASH

The mineral matter in coal causes most operational problems within the furnace, but 80-85 °/o of the mineral matter in coal burned in dry-bottom furnaces passes out of the boiler as "flyash". Recent legislation on emissions limits the amount of this ash that can be discharged from the stack. Hence electrostatic precipitators (ESP) are used widely to capture flyash, with wet scrubbers and bag filters as other methods of control. The characteristics of the flyash greatly influence the efficiency with which an ESP can collect ftyash. Particle size is important, but the dominating parameter is the electrical resistivity of the flyash. This property controls the rate at which the electrical charge placed on the flyash particles by the ionizing wires can leak off the flyash caught by the collecting plates. The resistivity should not exceed 10~0 ~ cm for an ESP operating at temperatures less than about 200°C. "Hot" precipitation, where the flyash is above 300°C, is relatively unaffected by resistivity, but these ESP's are less commonly used. The electrical properties of flyash at 200°C are dependent primarily on surface conductance. Leakage of a charge through a mass of flyash particles seems to depend upon a film of SO3 condensed on the surface of the particles. This film, in turn, is established by the SO 3 level in the flue gas and is affected by the presence of SO3-neutralizing substances such as CaO and MgO in the flyash. Sodium in flyash lowers resistivity, apparently because sodium volatilized in the flame and condensed on the particles provides conducting paths for electrical charges. An investigation of Western coals, where the sulfur tends to be low and the alkalies high, has shown that

168

W.T. REID

the sulfur-capturing ash constituents such as CaO and MgO increase electrical resistivity while Naz O and SO a lower it, and that the mole ratio of these two conflicting processes is a good measure of resistivity at 150°C. 2s Tests at 150°C showed that, if the mole ratio of (CaO + MgO)/(Na20 + SOa) is 0.7, the flyash resistivity will be less than 3 x l0 s f~cm; if this mole ratio is greater than 5.1, the resistivity is more than 101°f~cm. Very roughly, the relationship can be shown as: / " ~ C a+OMgO Log resistivity (150°C)= ( 0 . 4 4 ) ~ ~ )

+8.

This is only a guide at best but it can serve in evaluating the potential effectiveness, with different coals, of flue-gas conditioning with added SO 3 to improve ESP performance. 5. COALPREPARATION The mineral composition of coal can be modified to some extent by coal preparation. Although the earliest use of coal beneficiation was in crushing run-of-mine coal and screening for a uniformly sized product, removal of "rock" was about equally important. The early float-and-sink flotation processes were designed to utilize the difference between the low density of coal (1.2-1.7) and the higher density of shale and clay (1.6-2.2) as well as of pyrites (4.8-5.2). Many schemes were devised to lower the "impurities" in coal in this way. Economic advantages have been shown in improving coal quality by beneficiation, with a subsequent lowering of operating and maintenance costs in coal-burning central-station power plants. Most recently, emphasis has been on sulfur removal for minimizing SO 2 emissions, on the basis that removing pyrites from coal is more attractive economically than scrubbing SO 2 from flue gas. An important consequence of coal preparation would be the selective removal of mineral matter. Pyrites, of course, with its high specific gravity is not difficult to remove if the particles of FeS 2 are large enough to have been liberated mechanically from the coal macerals by crushing to the minimum permissible size for float-and-sink processing. About half of the Fe203 in coal ash comes from FeS 2 on the average, hence removal of pyrites can have an appreciable effect on the fusion characteristics of the cleaned coal. Calcite and dolomite, on the other hand, may or may not be removed selectively in coal washing, and the percentage level may actually increase if the float-and-sink process has removed proportionally more shale and clay. Sodium, although blamed for many of the fouling problems with low-rank Western coals, is still an enigma as far as removal is concerned by physical methods. Size as well as density affects the distribution of mineral matter in coal. Although the variance was high, tests in the 1960's showed that NaO and K 2 0 tended to be higher in coal particles larger than about

6mm than in particles smaller than about 0.3 mm; 29 CaO and MgO behaved similarly. On a density basis, not even such general trends could be detected. Washability tests of coals in the past generally have ignored the composition of the coal ash in the fractions separated at different densities. That shortcoming is being recognized, and coal-ash chemical analyses of float-and-sink fractions are becoming more available.

6. SUMMARY Many problems arise from the mineral matter in coal, from clinkering in fixed fuel beds to interference with heat transfer in large modern steam generators; from setting up an environment in boiler furnaces encouraging excessive metal wastage by chemical attack to the loss of metal by erosion; and finally as a major source of air pollution, both gaseous and particulate. Most understanding of mineral-matter behavior during and after combustion is based on empirical experimental work. The mineral systems are so complex, and the variables are so many in large combustion systems, that only limited success has been reached as yet in applying truly scientific methods. As experimental methods improve, and as coal becomes even more important as a source of energy, and as chemists and physicists supplement the efforts of engineers, a much better understanding of coal-ash behavior will result. The surface has barely been scratched at present. 7. REFERENCES l. Mineral Matter and Trace Elements in U.S. Coals, Pennsylvania State University, 184 pp., OCR Contract 140-1-0001-390, July 1972. 2. ZAKHAROV,V. Y., POMERANTSEV,V. V. and RUNDYGIN, Y. A., Diffusion Kinetic Approach to the Problem of Mineral Part of the Solid Fuel, 7 pp, ASME Paper JPGC-Fu-2 (1982). 3. BRYERS,R. W, The Physical and Chemical Characteristics of Pyrites and Their Influence on Fireside Problems, J. Engn# Pwr, 98, 517-527 (1976). 4. BUqGLER, E. C., T~XLER, D. T., KEMP, W. R. and BOI,,raAM,H. F. Jr., PETCAL: A Basic Language Computer Program for Petrologic Calculations, 27 pp, Nevada Bur. Mines and Geology, Report No. 28 (1976). 5. Sulfur in Ash from Coal and Coke, ASTM, Part 26, D-1757-80 (1980). 6. Analysisof Coal and Coke Ash, ASTM, Part 26, D-279569 (1980). 7. Total Sulfur in the Analysis Sample of Coal and Coke, ASTM, Part 26, D-3177-75 (1982). 8. AnERNEXnV,R. F., PEARSON, M. J. and GmSON,F. H., Major Ash Constituents in U.S. Coals, Report Inv. 7240, 9 pp., U.S. Bur. Mines (1969). 9. FmLDNER,A. C., HALL, A. E. and FEILD, A. L., The Fusibility of Coal Ash and the Determination of the Softening Temperature, Bull. U.S. Bur. Mines, 129, 146pp (1918). 10. REID, W. T. and COHEN,P., The Flow Characteristics of Coal-Ash Slags in the Solidification Range, Trans. ASME, 66, 685-690 (1944). 11. NICHOLLS,P. and REID, W. T., Viscosity of Coal-Ash Slags, Trans. ASME, 62, 141-153 (1940). 12. SAG~,W. L. and MCILROV,J. B., Relationship of Coal-

Mineral composition and combustion

13.

14.

15. 16. 17. 18. 19.

Ash Viscosity to Chemical Cofiaposition, Trans. ASME, 82, 145 155 (1960). HoY, H. R., ROBERTS, A. G. and WILKINS, D. M., Behavior of Mineral Matter in Slaggin0 Gasification Processes, Inst. Gas Engineers, Publ. 672, 5, 444-469, June 1965. MULCAHY, M. F. R., BOOW, J. and GOARO, P. R. C., Fireside Deposits and Their Effect on Heat Transfer in a Pulverized-Fuel-Fired Boiler, J. Inst. Fuel, 39, 385-394 (1966). STREET, P. J. and TWAMLEY, C. S., J. Inst. Fuel, 44, 477-478 (1971). REID, W. T., Chem. Coal Utilization, Chap. 21, Wiley, New York (1981). Ref. 16, p. 17. LOCKLIN, D. W. eI al., Elect. Pwr Res. Inst. CS-1318, II!-1 1 (1980). Ref. 18, p. Ill-49.

169

20. Ref. 16, pp. 146-150. 21. ATTIG, R. C. and BARNHART, D, H., Mechanism of Corrosion by Fuel Impurities, Butterworths, London (1963). 22. RAASK,E., Wear, 13, 301 (1969). 23. WRIGHT, I. G. and STRINGER, J., Elect. Pwr Res. lnsl., WS-80-141, 5-1 (1981). 24. REID,W. T., External Corrosion and Deposits, Boilers and Gas Turbines, Elsevier, New York (197l). 25. Ref. 24, pp. 115-143. 26. ANDERSON,C. H. and DIEHL, E. K., ASME Paper 55-./t200 (1955). 27. REHN,I. M., Elect. Pwr Res. Inst. CS-3134, 70 (1983). 28. SELLE, S. J., TUFTE, P. H. and GRONHOVD, G. H., Air Pollution Control Assoc., June, 1972. 29. BORIO, R. W., The Control of High-Temperature Fireside Corrosion in Utility Coal-Fired Boilers, Office of Coal Research, R & D Report No. 41 (l 969 ).