APRIL, 1939
INDUSTRIAL AND ENGINEERING CHEMISTRY
Literature Cited (1) Axilrod and Kline, J. Research Natl. Bur. Standards, 19, 367 (1937). (2) Crawford and McGrath, U. S. Patent 2,108,044 (Feb. 15, 1938). (3) Du Pont de Nemours, E. I., 8z Co., IND.ENQ.CHEM.,28, 1160 (1936). (4) Elec. Rev., 118, 784 (1936). ( 5 ) Fields, U. S7 Patents 2,057,6734 (Oct. 20, 1936). (6) Gardner, "Physical and Chemical Examination Of Paints, Varnishes, Lacquers and Colors," 8th ed., P. 300, Washington, D. C., Inst. of Paint and Varnish Research, 1937.
387
(7) Gordon, U. S. Patent 2,101,061 (Dec. 7, 1937). (8) Hill, Ibid., 2,045,651 (June 30, 1936). (9) Instruments, 9, 218 (1936). (IO) Kline, Modern Plastics, 15, 47 (April) and 46 (May), 1938. (11) Kline and Axilrod, IND. ENQ.CHEM.,28, 1170 (1936). (12) Kuettel, U. S. Patent 2,063,315 (Dee. 8, 1936). (13) Loder, Ibid., 2,045,660 (June 30, 1936). (14) Macht, Ibid., 2,071,932 (Feb. 23, 1937). (15) St,rain, IND. ENQ.CHEM.,30, 345 (1938). (16) Strain, U.S. Patent 2,030,901 (Feb. 8, 1938). (17) Tatersall. Ibid., 2,071,907 (Feb. 23, 1937).
WATER CONDITIONING IN STEAM GENERATION
.
During the past twenty years boiler pressures have increased from 350 to 2500 pounds per square inch, rates of evaporation per boiler from 150,000 to 1,000,000 pounds per hour, and total steam temperatures from 660" to 950" F. This remarkable development has occurred parallel to, and at least in part as a result of, corresponding progress in the conditioning of boiler water, so that steam could be safely and economically produced continuously at high rates, even from water supplies of inferior quality. From the viewpoint of the chemical engineer, the proper conditioning of water for contemporary boilers is a never-ending series of problems in clarification, filtration, fluid flow, heat transfer, and properties of materials, in the solution of which he must apply all that is known concerning the physical chemistry of aqueous solutions, the phenomena of the colloidal state, and the factors influencing corrosion. The major developments in boiler design, the external softening of feed water, and the internal conditioning of boiler water during the past twenty years are described.
N 1918new boilers just placed in operation a t a pressure of 375 pounds per square inch represented an outstanding advance of about 75 pounds beyond regular practice and were recognized as approaehing the limit for the type of design then standard (1). Engineers were, however, visioning boilers which would produce steam a t 800 pounds pressure, superheated to 800 F. (7). In 1938 a million pounds of steam per hour were produced from a single boiler a t a pressure of 1375 pounds and a temperature of 900"F., and a boiler designed to operate a t 2500 O
EVERETT P. PARTRIDGE Hall Laboratories, Inc., Pittsburgh, Penna
A. C. PURDY Bull & Roberts, New York, N. Y.
pounds per square inch and to produce steam superheated to 950" F. is under consideration. For this rapid change during the last two decades much credit must go to the men who have learned how to burn fuel efficiently a t high rates; to design heat-absorbing surfaces so that the resultant radiant energy would be utilized in the production of steam instead of the destruction of the boiler furnace; to fabricate and assemble the massive and intricate components of these designs; and to control, by sensitive automatic instruments, practically every operation in the making of steam. The combustion engineers, the mechanical engineers, the metallurgical engineers, and the engineers devoted to instrumentation must, however, share some of the credit with the chemical engineer, whose contribution has been the conditioning of water so that it can be evaporated continuously a t a high rate without damage to the boiler. Water conditioning is more than water softening, in the same sense that operating an integrated plant differs from supplying the raw materials for the process. In water conditioning the emphasis is necessarily not merely on the hardness of the water fed to the boiler, but on what is actually taking place a t some particular locality in the boiler or its auxiliaries or in the turbine or process equipment to which it supplies steam. The viewpoint is that of the physical chemist determining the essential factors of a problem and of the chemical engineer applying this knowledge in devising a practical solution. Control of the conditions in the boiler by systematic analysis of samples of the boiler water, which is accustomed routine today, had scarcely been imagined in 1918. In retrospect, it seems as if the developments in steam generation in the following years led inevitably to the necessity for boilerwater conditioning. In turn, the improvement in operation resulting from proper conditioning must have been a factor in stimulating the subsequent great advances in boiler design.
INDUSTRIAL AND ENGINEERING CHEMISTRY
388
This paper will attempt to review the interrelated history of water conditioning and steam boilers during the past twenty years, not only with respect to stationary plants, both utility and industrial, but also in marine and railroad practice.
Changes in Steam Boilers What has happened to boiler design in two decades may be seen by comparison of the representative units in Figure 1, on the one hand, and in Figure 2, on the other. The fivedrum bent-tube unit in Figure 1, typical of 1918, was fired by a stoker and was enclosed between refractory walls. WATER-COOLED FURXACES. Perhaps the best way to characterize the change from 1918 to 1938 is to say that the fire then was under, but now is inside the boiler. As indicated in Figure 2, the furnace of the contemporary boiler is completely lined with heat-absorbing waterwalls. The reason for this fundamental change in design was the fact that, when the burning of powdered coal first began to take the place of chain grates and stokers about 1920, refractory walls did not stand up well. A few water tubes originally introduced to protect the brickwork rapidly multiplied into the complete waterwalls of contemporary design which frequently transfer the major portion of the heat. Thus 55 per cent of the heat
VOL. 31, NO. 4
input to the new boilers at Waterside Station of the Consolidated Edison Company of New York is estimated to pass through the wall tubes (35). A boiler installed in 1936, in the Rouge Plant of the Ford Motor Company, produces a million pounds of steam per hour at 1340 pounds pressure and 910" F. This boiler occupies the floor space in which formerly stood a boiler producing 200,000 pounds of steam per hour a t 240 pounds, which was installed in 1923. I n thirteen years five times the steam has been produced a t five times the pressure in the same floor space. Some idea of the interior of a contemporary boiler furnace may be gained from Figure 3, which shows the view upward from the bottom of one of the boilers now being installed in the L Street Station of the Boston Edison Company. This three-drum bent-tube boiler, which is designed to generate 375,000 pounds of steam per hour a t a drum pressure of 1400 pounds and a final temperature of 910"F., has 5700 square feet of heating surface in the furnace walls as compared to 5635 in the boiler proper. In the early days of pulverized coal, the development of complete waterwalls was retarded by the fear that the process of combustion would be seriously affected by the relatively low wall temperature. Kreisinger has, however, recently
1 Courtesy, Babcoclo & Wilcoz Company
BENT-TUBE BOILER, LARGEST UNITOF FIGURE 1. FIVE-DRUM 1918
Courtesy, Babcock & Wilcoz Company
FIGURE 2. RADIANT-TYPE BOILER, 1938
INDUSTRIAL AND ENGINEERING CHEMISTRY
APRIL, 1939
389
FIGURE 3. LOOKING UPWARD FROM THE BOTTOM O F THE FURNACE IN A BOILER DURING CONSTRUCTION Courtesy, Combustion Engineering Company. Inc.
emphasized that retarded combustion is not actually a problem of the water-cooled furnace (40). Much of the change from 1918 to 1938 is expressed in abbreviated form in Table I, which gives comparative data for the boilers in Figures 1 and 2. Great as is the difference between these central-station units, those for smaller steam output in the average industrial plant have shown an even more marked advance from the hand-fired, horizontal returntubular boiler common in 1918 to types such as the one shown in Figure 4. OF CENTRAL POWERPLANTBOILERS IN TABLEI. COMPARISON
1918 AND 1938
1918 Heating surface of boiler proper, largest unit, sq. ft. Heating surface of boiler, superheater, economizer a n d air heater, sq. ft. Working pressure Ib./sq. in. Max. temp. of su’erheated steam, F. Max. evapn. per goiler unit, lb./hr. Vol. of boiler setting for largest boilers, C,U.
ft.
1938
23,650
53,926
26,800 350 660 147,000
144,402 2,500 950 1,000,000
19,480
42,255
Height, bottom of walls of setting t o center of steam and water drum for 33 largest boilers, ft. Stokers Type of oil and gas firing No Air heaters Yes” Steel economizers Refractory Walls Clinker grinders Ash removal
80 Pulverized coal Yes Yes Water-cooled Slag t a p and rlrv bottom hnttom dry Yes No Steam scrubbers Steel economizers were just coming into use in this period. Q
SLAG-TAP FURNACES. The growing use of pulverized coal after 1920 brought difficulties with slag accumulation when coal with an ash of lower softening point than usual was burned. This defect was turned to advantage in 1926 with the slag-tap furnace in which the low-melting noncombustible portion of a coal with a relatively high content of iron oxide was removed in liquid form. The development of boiler design with respect to the slag problem was recently discussed by Bailey (4), and Caldwell presented the advantageous features of current slag-bottom furnaces (IS).
CIRCULATION OF BOILERWATER. As boilers grew in size and complexity, and rates of heat liberation in boiler furnaces increased, the problem of how to get water to and steam away from heat transfer surfaces required increasing attention. Some of the problems of design of high-pressure, high-capacity units were recently discussed by Van Brunt (74). Adequate circulation by thermal means alone has generally been achieved by the designers, in spite of the almost complete absence of any fundamental data on heat transfer and fluid flow under boiler conditions. I n a sense, however, almost every new large boiler has been a cautious advance into the unknown, and sometimes modifications have been necessary after installation in order to make the water go where it was supposed to go as rapidly as it was supposed to go. With respect to forced circulation, designers in this country have been content to follow European experiment a t a distance, and only within the last few years have trial units been put in the field. STEAM WASHERS. One place where the boiler water is not supposed to go is out with the steam into the superheater and turbine. Dissolved solids deposited in the superheater mean overheating and failure of the metal; in the turbine they mean loss in capacity. The increase in trouble from carry-over which accompanied the development of high-pressure boilers has led to the design of steam-washing devices, in which the incoming feed water scrubs the outgoing steam and thus appreciably reduces the amount of dissolved solids in the moisture entrained in the steam. LARGEUNITSFOR SMALL. The trend away from a large number of small boilers to a small number of large boilers began in the central stations of the utility companies but has spread to industrial power plants. In 1918 when the typical large boiler house contained long rows of boilers, it was stated in a lecture than “one boiler in every five or six should be spare to provide for cleaning and repair and also for repairing the stokers’’ (41). Today boiler plants supplying more steam than their predecessors of 1918 will contain only a few units, and Plant .1 of the Goodyear Tire and Rubber Company at Akron, Ohio, has relied since 1934 upon a single 800-
390
INDUSTRIAL AKND ENGINEERING CHEMISTRY
pound, straight-tube, crowdnun boiler capable of producing 300,000 pounds of steam per hour. FARnICATIoN. The boilers of 1938 have been made possible only by great advances in methods of fabrication. In 1918 the boiler manufacturer had no other way of making a boiler drum than to bend steel plates and fasten them together with rivets. The old lap-riveted joint was frowned npon @),but the riveted butt joint was tbe only better means available. This limited boiler pressures to about 400 pounds. Higher pressures had to await the development of means for making forged drums. Later, welding attained a degree of perfection which won approval of this method of fabrication in 1931. The contemporary boiler, with its one-piece drums anncdcd to remove residual stress, its seamless tubes, and its relatively painstaking method of erection might be said to be a product of the machine shop in comparison to its predecessor from the typical boiler shop of 1918. ENOINEERING. When technically trained men began to take over and critically analyze the operation OF central stations after the World War, a highly competitive professional race began to reduce the heat requirement for generating electric power. The net result is that in twenty years the B. t. u.'s per kilowatt-hour have been more than cut in half to a recently established low of 10,900 set by the Port Washington Station of the Milwankee Electric Railway and Light Company ( I S ) . This improvement reflects an intensive study of heat balances which started about 1921 and led to the progressive rise in steam pressure and superheat temperature; to the general use of economizers and, air preheaters; and to tbc development, on one hand, of the steam reheating cycle and, on the other, of the regenerative feedwater cycle in which steam bled from the turbine at as many as five points is used in successive feed-water heaters. Industrial plants also took advantage of extraction turbines to supply process steam a t various desired temperature levels, and thereby achieved over-all thermal economies,
I
VOL. 31. NO. 4
MARINEBOILERS. I n tho marine field the changes during the past twenty years have reflected those in the stationary field, with the necessary modifications imposed by t.he necessity for conserving space and weight. The traditional Scotch marine boiler had given way to a considerable extent to the straight-tube water-tube and threedrum Yarrow-type boilers before the World War, but i t was not until 1928 that a pressure of 350 pounds was at.t.ainedin American merchant marinc construction. Within the last few years there has been a tendency to modify the conventional boiler design as well as to increase pressure, rating, and steam temperatures. For instance, boilers with five drums, water-cooled furnaces, and economizers, operating at 425 pounds, and boilers with over and under drums, water-conled furnaces, and economizers operating at. 425 pounds pressure, have shown very satisfactory operating econnmies. Recent German, French, and Italian superliners have used the threedrum type of boiler; the & m a Mary entered service with fivedrum boilers. Among the more radical advances in marine boiler installations might be mentioned the Hamburg-American Uckeimarlc with Benson boilers designed for 3200 pounds with 752' F. total temperature, which has been in service since 1930, and the reboilered Italian liner Code Rosso with Loeffler boilers operating a t I900 pounds pressure and 935' F. total temperature (&). There is no doubt that the next few years will see rapid advance in marine boiler design and operation. For instance, merchant vessels are currently baing considered in this country with boilers to operate at 1500 pounds pressure with forced circulation. LOCOMOTIVE BOILERS. The least apparent change in basic boiler design has occurred on the railroads, where the locomotive-type boiler now in use differsfrom that built in 1918 chiefly in improved steel and medbods of fabrication. Riveted construction is still employed, although welded boilers are currently bring tested. Mechanical stokers, superheaters, and feed-water heaters to recover a portion of the heat from the exhaust steam have considerably improved efficiency. Although the steaming capacity of an individual boiler has increased great,ly, the pressure has, in general, gone up only to abont 250 pounds. Recently the Steamotive unit (6),comprising a compact, forced-circulation boiler, turbine-generator, and auxiliary equipment has presented a radical change in design.
A discussion as brief as the preceding can indicate only some of the salient features in the progress of boiler design. For the person desiring additional information without exbnustive treatment, the Boilers and Furnaces Number of Power Plant Enqineering is suggested (21).
Development of Water Conditioning Turning from the general review of what has happened to boilers in the last two decades to what has happened to the water inside of them, a complex picture unfolds. Scale and corrosion were an old story in 1918, although cracking of riveted seams had just been attributed by Parr (47) to high boilerwater alkalinity. How scale actually developed was not known, however, and corrosion was commonly thought to be due to the formation of acids as a result of the hydrolysis of magnesium salts. Carry-over
APRIL, 1939
INDUSTRIAL AND ENGINEERING CHEMISTRY
J regarded as a nuisance when it exceeded one per cent but was was a relatively unimportant problem compared to scale and corrosion.
Conditioning to Prevent Scales of Calcium and Magnesium Salts The scale deposited in the boilers of 1918 and the succeeding ten years consisted largely of calcium sulfate, calcium carbonate, and calcium and magnesium silicates (32, 51). Occasionally magnesium hydroxide and calcium hydroxide were observed. These scales formed continuous insulating layers on the surfaces subjected to the greatest heat input and inevitably caused failure due to overheating of the metal unless they were removed at intervals by mechanical cleaning. As higher boiler pressures and waterwalls came into use, the greatly increased rates of heat transfer caused tubes to fail as LL result of overheating due,to scale deposits so thin that a decade earlier they would have been considered desirable as protection against corrosion. Not infrequently new plants were forced to spend as much time cleaning and replacing tubes as in making steam. Prior to 1924 there were three answers offered to the scale problem-distillation, more complete external softening, or the use of boiler compounds. I n spite of its relatively high cost per unit of water, distillation, which had already been used in marine practice for years, was adopted after 1920 for many new central-station installations where only a few per cent make-up a t the most was required in addition to the return of condensate to the boilers. For plants supplying heating or process steam not returned to the system, zeolite softening was regarded as the best means of preventing scale, but the time-honored process of softening with lime and soda, either in the cold or improved by increasing the temperature to 200’ F. (93” C.) or higher, was widely utilized because of its lower cost. The average boiler plant, however, doctored the raw feed water with compounds running the gamut from straight inorganic chemicals, such as soda ash and sodium silicate, to witches’ broths brewed from garden vegetables, kerosene, and caustic soda. Neither compounds, nor external softening, nor even evaporation of make-up proved a complete answer to the scale problem, for even a zeolite softener may allow enough calcium to enter a boiler to build up anhydrite scale in a waterwall in the course of time; and even in well-run plants, evaporators may foam and condensers leak, with the result that scale-forming constituents may enter the boiler water. Concern over the problem of tube failures from scale, moreover, in some cases obscured other serious problems actually created by the treatment of the feed water. Thus, caustic soda, soda ash, or sodium silicate was frequently fed to waters already high in sodium bicarbonate, with resultant excessive and uncontrolled alkalinity in the boiler water. The same result was occasionally achieved by the incorrect application of zeolite softeners to feed waters high in bicarbonate and by overtreatment beyond reason in lime-soda softeners. I n 1923 a material balance was run for the first time on the water side of a n operating boiler (22). This test might be regarded as a hist,oric event, since it demonstrated incontrovertibly that the calcium sulfate scale found on the heating surfaces a t the end of the 42-day run had not first precipitated in the body of the boiler water but had crystallized out on the steel directly from solution. This result left little place for the deconcentrator; this was a settling tank, with or without a filter, attached to a boiler on the theory that scale was derived from sludge and that if suspended material could be removed continuously from a boiler, scale formation would be prevented. Instead it pointed the way to the rational treatment of boiler scale as a problem in chemical equilibrium.
391
The picture presented by Hall (22,24, 26, 29, 30) of so controlling the composition of the boiler water that the solid phase deposited as a result of continued evaporation would be in the form of a nonadherent sludge instead of a scale, is taken for granted in 1938. I n 1925, however, criticism was about evenly divided between the statements that it was old art to add soda ash t o a boiler and that it was unnecessary to adopt internal conditioning if a “standard” feed water was supplied by proper softening. Some time was required to drive home the basic idea of water conditioning, that only by determining the composition of the boiler water by actual chemical tests on samples removed from the boiler could the optimum conditions be maintained for the prevention of scale, corrosion, and the other troubles to which the steam boiler is heir. The soda ash first used as a conditioning chemical by Hall produced satisfactory results in 150-pound boilers operating on waters low in bicarbonate, but the inevitable decomposition of the desired slight excess in the boiler tended to produce undesirably high alkalinities and rendered the control uncertain. I n spite of higher cost, phosphate (27) soon began to replace carbonate, not only in the high-pressure 400pound boilers but also in the units operated a t pressures of 150 and 200 pounds. With phosphate conditioning, the heat transfer surfaces remained even more free of deposits than had been the case when soda ash was used; the only material found on the steel was a characteristic powdery film of particles of calcium phosphate so minute that they could scarcely be resolved a t a magnification of 450 diameters. It was possible, moreover, to feed mono-, di-, or trisodium phosphate as occasion demanded, in order to control the alkalinity of the boiler water a t the desired low level. The one serious drawback, which had also been experienced with carbonate conditioning, was the accumulation of deposits in feed lines and economizers. By 1930 Hall had discarded the use of orthophosphate in favor of metaphosphate (Si),which until then had failed to find commercial use during nearly a century since it was first studied by Graham in 1833. Metaphosphate had the signal advantage of not causing feed-line deposits when properly introduced; it reverted rapidly to orthophosphate in the boiler, and in so doing it not only removed calcium as calcium phosphate sludge, but also reduced the alkalinity of the boiler water. It was not long before phosphate conditioning drew wider and wider discussion in the technical journals. Tribute to the effectiveness of phosphate in practice was also paid by the manufacturers of many boiler compounds, who shifted the composition of their products to include more or less orthophosphate, frequently mixed with an organic material such as tannin or the alginate extracted from seaweed. Whether or not organic materials have any place in boiler water conditioning has probably been argued ever since the prevention of scale in one of Watt’s boilers was attributed to a lunch of potatoes left to cook and then forgotten by a helper (55). Out of the fantastic claims which have been made for products from almost every species of vegetation, it is difficult to isolate the truth. The general use of tannin not only in boiler compounds but also in connection with controlled conditioning does, however, reflect a useful property of this particular type of material. Tannin alone will not keep boiler heat transfer surfaces free of scale, although in laboratory experiments it exerts a marked effect in reducing scale formation (51, 7 0 ) . However, when used with sufficient phosphate to precipitate the calcium entering the boiler completely, tannin does act as a dispersing agent to keep calcium phosphate sludge from accumulating in regions of limited circulation. Except in the case of natural water supplies very low in hardness, primary softening is desirable both on an economic
392
INDUSTRIAL AND ENGINEERING CHEMISTRY
and an engineering basis to reduce the amount of calcium to be precipitated in the boiler by secondary conditioning with phosphate. Lime-soda and zeolite softeners have, until recently, provided the only means for accomplishing this primary softening. Organic base-exchange materials and primary phosphate softeners have, however, entered the field since 1930. The chief advantage of the organic base exchangers over the zeolites lies in the fact that they may be made to function not only to exchange calcium and magnesium ions for sodium ions, but also, by regeneration with acid, to exchange sodium ions for hydrogen ions (3, 9, 19). To supplement the cation. exchange materials, anion exchangers are now under development which are intended to remove most of the sulfate and chloride ions from boiler feed water. While feed water approximating distilled water is ultimately possible by the combined use of anion and cation exchangers, final conditioning in the boiler will certainly be no less necessary than with evaporated make-up. The outstanding success of secondary conditioning with phosphate in the boiler led naturally to the development of the primary phosphate softener about 1931. By using sodium orthophosphate and caustic soda as precipitants in a hot-process softener, it has been possible to reduce the residual calcium in boiler feed water to a value lower than that obtainable in the best practice with a lime-soda softener. I n some cases, however, it has been necessary to lower the p H by acid treatment following a phosphate softener, in order to maintain the desired high alkalinity in the softener without corresponding excessiveIy high alkalinity in the boiler, and to avoid afterprecipitation of calcium phosphate in the feed lines (58). No substance has yet been found which seems a t all likely to displace phosphate from its position in the final conditioning of boiler water. The suggestion in 1931 that sodium aluminate might be employed to remove calcium and silicate ions from boiler water (14) proved ill-advised, and even the concerns most interested in the development of uses for this product now do not recommend i t for internal treatment but restrict it to its proper application as a coagulant in the preliminary cold softening of feed water. Straub (70) has followed earlier suggestions (17‘) by investigating the possible use of fluoride as a precipitant for calcium, but since his experiments indicate an order of solubility intermediate between calcium sulfate and calcium carbonate, it seems probable that calcium fluoride in actual practice might contribute to scale formation instead of preventing it. The use of soda ash as a conditioning chemical a t pressures up to 2000 pounds, suggested by other experiments of Straub (68),seems equally unlikely to be adopted in practice.
Control of Sodium Aluminum Silicate Scale When the traditional deposits of calcium and magnesium salts characteristic of the boilers of the 1920’s yielded to phosphate conditioning, the scale problem appeared to have been completely conquered. It was not long, however, before the steady increase in boiler pressure and rate of heat input produced a new type of scale, first reported by Powell in 1933 (66). This was essentially a sodium aluminum silicate, corresponding in some cases to the natural mineral analcite (Na20.Al2O3.4SiO2.2H20) although the ratios of the oxides have been found to vary over a rather wide range, as indicated by the analyses in Table 11. Scales of this type are so hard that a turbine cutter will slide over their surface. They are, however, rather brittle. Occasional samples suggest porcelain by their translucence and ring when dropped. Efforts to obviate the formation of sodium aluminum silicate scale by carrying alkalinities higher than normal in the
VOL. 31, NO. 4
boiler water have met with only slight success. Internal treatment by so-called colloidal iron or by iron or zinc salts with the intention of precipitating silica in nonscaling form has generally resulted only in modifying the composition of the scale formed. Two solutions to the problem have, however, come clearly to the fore within the past three years. The first is the substantial elimination of alumina or the nearly complete removal of silica from the feed water; the second, the limitation of the rate of heat input to the boiler surfaces. TABLE 11. VARIATION IN PERCENTAGE COMPOSITION OF SODIUM ALUMINUM SILICATE BOILERSCALES Lab. No. SOa
c0 2
PPOS Si02 Fez03 Ala03
cao
MgO Na2O
B-6203 5.8 Trace
0.9 31.9 3.3 24.5 Trace
0.0 21.7
B-1096 Trace No evidence
8.2 42.4 1.6 19.0 9.2 2.6 10.2
B-4346 0.0 0.0 5.3 52.1 10.0 8.1 5.6 2.1 9.1
Alumina may enter the boiler feed as minute particles of clay from the raw water or incompletely removed flock from a clarification system. In addition, if the pH of lime-softened water is sufficiently high, a potentially dangerous though minute amount of alumina will remain in solution, whereas an alkaline water will dissolve alumina as well as silica during its passage through a zeolite softener. Removal of alumina by proper p H control and efficient settling and filtration is, however, not difficult, although it may require an appreciable investment. Silica, like alumina, may enter the boiler feed either in solution or in the form of suspended particles of clay. Although it is not feasible economically to take out silica as completely as alumina, both ferric hydroxide and aluminum hydroxide are being employed in practice as silica removal agents during the preparation of feed water. The expedient of lowering the rating on a boiler which is forming sodium aluminum silicate scale can scarcely be recommended to the superintendent of a plant forced to carry more than its normal load. T h a t there is a definite relation between rate of evaporation and the formation of deposits of sodium aluminum silicate was, however, one of the lessons of the recent business depression. The explanation goes back to a phenomenon described by one of the writers in 1929 (61, 54)* Briefly, when a bubble of steam forms momentarily at a heat transfer surface, the surface goes dry beneath the bubble, and whatever was present in solution in the water tends to be deposited as a plaque or a ring. As the bubble detaches itself, the water sweeps over the deposited material and tends to redissolve it. If the rate of deposition as a result of bubble formation is less than the rate of resolution, the surface will remain scale-free. However, if the material cannot redissolve as rapidly as it is deposited, it will tend to accumulate as scale. Obviously, however, material crystallized out during local evaporation under a bubble cannot go back into solution when the boiler water washes over it, if the water is already saturated with respect to the various substances comprising the deposit. Each successive evaporation of a film of water as a bubble forms therefore builds up a n increment of scale. This mechanism applies not only to the relatively insoluble substances which form boiler scales, such as calcium sulfate, but also to very soluble ones such as sodium chloride, which is known to “salt up” the heat transfer surfaces of evaporators operating on saturated brine, and has been formed as a scale in experimental apparatus (64). In a boiler the total con-
APRIL, 1939
INDUSTRIAL AND ENGINEERING CHEMISTRY
centration of sodium salts in the water is, however, only a few thousand parts per million (far below saturation), so that the tendency to redissolve any readily soluble substances momentarily deposited by local evaporation is great enough to prevent the inclusion of more than minute amounts in ordinary scales. This theory should be regarded as an extension and elaboration of the original theory of Hall that scale-forming substances were characterized by a decrease in solubility with increase in temperature (82) ; according to this theory, the fact that calcium phosphate does not form a scale is due to its extremely low solubility and to its characteristic tendency to form individual crystals of extremely minute size. As long as bubble evolution at a heat transfer surface deposits, along with the other solids, only a small amount of silica, or if the surface is not a t a sufficiently high temperature to dehydrate the silica rapidly, standard phosphate conditioning will maintain the surface free of scale. When a boiler water containing appreciable silica is evaporated a t a surface which is hotter than normal, a different situation exists, which may be likened to dropping a solution of sodium silicate on a hot stove. A coating of silica is left which cannot readily be redissolved when the boiler water washes over the surface, and this coating tends, moreover, to protect the more soluble constituents. Under the influence of increasing metal temperature, what may initially have been silica gel will tend to dehydrate. Thus, is one waterwall deposit containing more than 75 per cent silica by chemical analysis, examination under the polarizing microscope showed quartz as the chief constituent with cristobalite also present. Such a deposit containing silica rather than silicates, is, however, unusual. Either by formation a t the instant of deposition or by adsorption of constituents from solution after initial deposition of a silica gel, a high-silica deposit found in a phosphate-conditioned boiler will usually show considerable amounts of sodium and aluminum oxides on analysis. Sodium aluminum silicate scale apparently forms only when the temperature of the heat transfer surface and the rate of evaporation a t this surface are high. The necessary limits seem rarely to be exceeded in boilers operating a t pressures below 400 pounds except as a result of a high content of alumina and silica in the boiler water or excessive local evaporation. But the possibility of developing this type of scale should be considered whenever installation of a new boiler at a pressure of 600 pounds or higher is contemplated, so that steps may be taken to exclude alumina and silica from the feed water in so far as this is economically feasible, or to design the boiler so as to minimize the tendency towards localized areas of high heat input or poor circulation.
Conditioning to Prevent Corrosion The only possible claim that could be made in favor of boiler scale was that it sometimes protected the boiler surfaces from corrosion. When central stations began to adopt evaporated make-up about 1920 for the specific purpose of eliminating scale, it was natural that the problem of pitting should immediately claim attention. In a number of cases, “pure” water which had been hailed as the ultimate in boiler feed was found to be well on the way towards causing as much damage as had previously resulted from scale. Severe pitting of drums and tubes occurred, and in a t least one plant headers cracked as a result of what would now be called “stress corrosion.” Varied attempts were made to develop a stable protective coating on the steel or to remove dissolved oxygen from the boiler feed water by chemical or mechanical means. Special paints, the addition of film-forming substances such as chromate and silicate, maintenance of high alkalinity, removal of
393
oxygen b y sodium sulfite, ferrous hydroxide, and organic materials, and efficient mechanical deaeration all claimed attention. Of these, deaeration, the use of chemical scavengers for the residual oxygen after deaerat-ion, and control of alkalinity now form the basis of current conditioning for the prevention of corrosion. I n 1918 the beneficial effect of a,n open heater in removing oxygen from boiler feed mater was not generally recognized, and few operators were willing to “waste” steam by adequately venting a heater: Deaeration as a specific objective was emphasized by McDermet in 1920 (4S), but for years thereafter a residual content of 1 p. p. in. of dissolved oxygen was considered adequately low. Continued emphasis upon oxygen as the agent primarily responsible for pitting and continued improvement in the technique of deaeration by the use of atomizing sprays and countercurrent flow has, however, pushed this limit down, until in 1938 as little as 0.005 p. p. m. of oxygen will be found in the feed water passing to a boiler in a well-operated central station. Early attempts to remove oxygen by chemical means were made shortly after 1920. Sodium sulfite was employed a t the Springdale Station of the West Penn Power Company in 1923, but on water which had not been deaerated it proved too expensive and was discontinued. Later ferrous hydroxide was employed a t the Lakeside Station of the Milwaukee Electric Railway and Light Company. I n 1931 Moberg (46) called attention to sodium sulfite again. Since then the availability of this substance a t a reasonable price has caused a progressive increase in its use. Where large quantities of feed water are handled, economy dictates the removal of the bulk of the oxygen in a deaerator, with subsequent addition of the sodium sulfite as an oxygen scavenger. I n smaller plants sodium sulfite is sometimes employed direct to avoid the capital investment in a deaerator. Conditioning with sulfite is based upon the maintenance of a low concentration, rarely exceeding 30 p. p. m., in the boiler water. I n recent years the question of whether this substance might be unsuitable for use in high-pressure boilers has been raised as a result of the incidental observation (66) that in concentrations of approximately 5000 p. p. m., sulfite is largely reduced to sulfide by iron powder a t 250” C. (482” F.). Later experiments, likewise carried out with concentrations much higher than those present in boiler water, led to the same conclusion (7S), but it has been demonstrated by careful tests on several high-pressure boilers that little, if any, sulfide is produced under operating conditions ($4). Both sodium sulfite and ferrous hydroxide presumably perform their function of removing residual oxygen before the feed water reaches the boiler. Although a reserve supply of the sulfite can be maintained in the boiler water to take care of occasional irregularities in proportioning, this is not true of ferrous hydroxide; the latter is rapidly oxidized by reaction with the water itself a t boiler temperatures to magnetic iron oxide, which is apparently the stable form under boiler conditions. The importance of proper alkalinity in the boiler water was stressed by Hall in 1927 (29)and reiterated in 1930 (26) when he pointed out the necessity of avoiding, on one hand, an alkalinity so low as to be conducive to rapid dissolution of steel and, on the other hand, a n alkalinity so excessive as to promote intergranular cracking. The experiments of Berl (8) a t 310” C. (590” F.) demonstrated that, although a high concentration of sodium hydroxide accelerated the reaction of water with steel, a low concentration retarded it. Therefore, proper conditioning includes, as one important factor, low controlled alkalinity of the boiler water to establish and maintain the oxide coating on the steel in the most protective state, in addition to the removal of dissolved oxygen from the feed water.
394
INDUSTRIAL AND ENGINEERING CHEMISTRY
(hnditioning to Prevent Cracking
VOL. 31. NO. 4
Meanwhile, the question of how sodium sulfate could prevent embrittlement had been argued to practical exhaustion Engineers responsible for boiler operation were talking in without uniform agreement. Many engineers, however, were 1918 about the experilucnts relating to "caustic embrittleinclined to favor the picture that sodium sulfate, present in ment" published the preceding year by Parr (47); in 1938 alkaline boiler water as it concentrated in a capillary seam they are talking mitli iiirdiiriinished interest about the latest between the steel plates in R riveted joint., crystallized from findings of Strauh (71) and of Schroeder (6@, still directed solution to form a mechanical barrier between tlie concenat. determining \Thy boiler steel sometimes fails by int,ergranutrated caustic solution and the steel. If this were so, said Hall (%), t,he usual common-ion effect should exist, and the solubility of sodium siilfato should depend on the concentration of other sodium salts in the boilcr water. From this conviction came a new program of research carried out by tlie U. S. Riireau of Mines in cooperation with the Joint Research Committec on Boiler Feedmater Studies. 13egtm in 1933, this is still roiit,iriuing although the original program of solubility studies (80) was coinplcted in 1936. As is not infreqiiently the case in researcli, the original objective turned out to he less significant than an unexpected development during the work. In 1934 i t was found that caustic soda, long regarded as the primary cause of emhrittlement, did not produce intergranular cracking when pure. A gear later the answer to this riddle had been discovered in silica which; when present only in very small amounts, changed the general surface attack of caustic on steel into selective intergranular corrosion (61, 72). The next upset came when repeated trials in the acrelcrated laboratory tests showed sodium sulfate to have little effect in preventing cracking (6s). Instead, Sehroeder and his associates discovered that lignin derivatives from sulfite waste liquor and certain tannins, such as quebracho and cutch, were the most effective inhibitors of intergranular corrosion by solutions of caustic soda containing silica over the range FIGURE 5 . RESULT ox' A BOILEI~ CXYLOS~ON DUETO CRACK- up to 250" C. (482" F.). Strauh, agreeing that sulfate by ING OF ItiVETEU Sr;axs it,self was incffectivc, stated that if chloride were also present in sufficient amount, embrittlement would be prevented a t lar cracking, and liow this type of apparently brittle failure boiler pressures up to 350 pounds. At higher pressures this may be prevented. That this subject is of more than acacombination failed, but maintedemic interest is emphasized by the photograph in Figure 5 nance of a ratio of &Os to silica of what happened to oiie boiler plant when a riveted seam, of 0.6 apparently afforded protecweakened hy int.ergranular crarking, suddenly let go. tion (71). Both the experiinents of Parr and the experiences of the As the situation now stands, the boiler manufacturers pointed to excessive alkalinity in boiler sit I fa te-al kalinit y ratios recomwater as an important factor in producing the cracks in mended in the Boiler Code are not riveted seams observed prior to 1918. Some uncompromising substantiated by either of two indeoperators retorted, howover, that the cracking x a s due to pendent laboratory investigations. poor fabrication. Out of the debate gradually came two rePractical experience on American sults: Methods of fabrication were improved, and a basis for railroads apparently agrees with the chemical control to prevent cracking was suggested. laboratory results showing the inBy 1923 the practice of punching rivet holes had been effectiveness of sodium sulfat.e in largely supplanted by drilling and reaming, the drift pili was preventing embrittlement. Howin disfavor, and more care was taken i n riveting. In turn, at e\'er, even though adherence to the least one of tlie maniifactiirers had been led to the conclusion A. S. M.E. ratios may not have that the best hope for preventing ernhrittlement lay in the afforded direct protection against maintenance of sodiuni sulfate in the boiler water in a definite embrittlement, indircctly it has unratio to the total alkalinity. This ratio, based upon observadoubtedly been benefieial in helption of a large number of operating boilers, sooii was given ing to limit the alkalinity mainexperimental support by Parr and Straub (49). On the comI.'xxJRE 6. SEmloS tained in boiler waters. No operabmed basis of the field statistics arid tlic Iakioratory data, OF STRAP tor wishes to build up tlie content FROM LONGITWDIthe now-famous sulfatoalkalinity ratios mere witten into NAL S E A ~ IOF D~~~ of dissolved solids unduly; hence, the A. S. M. E. Boiler Code for tile first time in 1926. CRACKEDAS A in order to keep the required conR,esearch on enibrittlenicnt, and argument ahout it conRESULTOF INTEAcentration of sodium sulfate down, aRAKwLAR CoRRotinued. In 1929 an abortive attempt to modify the ratios so the tendency has been to hold alkaSlON that at 600 pounds pressure 8.4 parts of sodium sulfate would linitiesaslowaswascompatihlewith have been required for each part of alkalinity (IO) was the prevention of ordinary corrosion. promptly quashed, not withnot the liclp of those responsible Even before the new information relative to embrittlement for the operation of the higher-pressure boilers of that period, had begun to accumulate, forged drums and then welded who feared the probable penalty of carry-over more than the drums had largely supplanted riveted drums on new boilers. indefinite possibility of cracking. In the course of time, pliosIn such units the tube ends are the most probable locality phate was patented as an inhibitor of omhrittlement (48). for intergranular corrosion. To date, few cases of cracking
INDUSTRIAL AND ENGINEERING CHEMISTRY
APRIL. 1939
FlCURE
395
STAGE OF INTRAO~ANULAR Cnacarrc I X WALLOP1111-ET HOLE 7 (left). EARLY
The fino omoka, which could not be detected by v i s u ~ iexamination, are outlined by t h e iron powder vised in t h e M~gnsRurmethod
INT)I:RGR.&NI:I..UI Cnma m o x c ~PREDOMINAXTLY a~~ FIQURE X (Tight). P w o r o ~ r c ~ oOF
of tube ends have occurred, and no case of embrittlemerit has been reported on a boiler operating above 450 pounds pressure. Proper conditioning to obviate emhrittlement is, for the moment, a matter of choice betveen the old A. S. M. E. ratios, the new ratios described by Straub, and tho organic inhibitors tested by Schroeder. Revision of t,he A. S. M.E. recommendations is anticipated, but in the meantime operators are faced Tyith the dilemma of meeting quasi-legal requirements or of taking the responsibility for disregarding them. Cracks may sometimes develop in riveted seains to the extent shown in the butt strap of Figure 6 before they are discovered, or they may be YO tiny that they escape visual inspection entirely and are revealed only by the Mapaflux test, like those on the internal surface of a rivet hole in Figure 7. In general, however, when cracks are found in riveted seams, the finer cracks will be predorriina~itlyintergranular, as shown in Figure 8. Boiler steel may crack as a result of other influences than selective intergranular attack. Figure 9 illustrates stress corrosion cracks which developed on the inside of the top of a tube at a bend; Figure 10 slioivs the cracks in tlie drum of a high-pressure boiler a t the point where the feed line conveying distilled water was bolted 011 (75). The diagnosis of such cases is perhaps more difficult and certainly no less important than that of cracking in riveted seams.
A
37O-Pol~~u UOILZW
Various proprietary antifoani compounds, many of them containing cast.or oil, were on tlie market in 1924 when Funk published his first paper on foaming (19) and called attention to the lack of any real knowledge concerning the cause of or metliods for preventing it, On the Iiasis of boiling tests a t atmospheric pressure, lie siiggested that high concentrations of dissolved and of suspended solids were important, factors in stahiliainy foam. T h e e years later Joseph and Hancock (86)concluded, hornever, that suspcniied solids had 110 effect on the moisture content 01 steam, Mumford likewise could find no evidence in his plant tests that suspended solids influenced carry-over (46). Subsequent 1aborat.ory investigation has revealed that firrely divided solid matter added to an experimental hoilcr may increase foaming t,emporarilp, but the effectis soon lost (21). In later experiments it was found that in most, cases suspended solids formed by precipitation within the experimental tioiler either had no effect 011 carry-over or actually decreased it (20). '
Conditioning to Prevent Carry-Over When scale and corrosion were major problems in boiler operation, carry-over was a minor one; when the develop ment of coriditioning made it possible to eliminate scale and corrosion, carry-over took their place as a chief source of worry. This was, however, not purely a psychological matter, for tirrbulent discharge irom water nralls and decreased drum diameters in the higher-pressure boilers created new' problems in separating boiler water from steam. In addition to true foaming, it came to be recognized that fine droplets of spray carried along by the steam might introduce enougli material into superheaters and turbines to cause ohjectionahle deposits. Continuous carry-over of droplets of water has been recognized as a mechanical problem and largely solved by the use of various devices located within the boiler drum. Foaming has, however, riglitfully beeii regarded as a problem in water conditioning. The development of foam has been attributed to suspended solids, dissolved solids, alkalinity, and contamination with oil containing saponifiable constituents.
From the standpoint of practical operation, tlie finely divided calcium phosphate resulting from phosphate eonditioning to prevent scale secms much less of a factor in foaming
396
INDUSTRIAL AND ENGINEERING CHEMISTRY
than does the content of dissolved solids, particularly sodium hydroxide. The tests of Mumford (46)illustrate what operators have, in general, experienced; i. e., that carry-over is incrased materially if the alkalinity exceeds some limiting value, which in the case of the boiler tested was from 220 to 250 p. p. m. of sodium hydroxide. Since other sodium cornpounds in higher concentrations tend to promote foaming in the same manner as sodium hydroride, boilers are currently guaranteed on the basis of delivering steam of a specified qiialitg for a total concentration of dissolved solids in the boiler water not exceeding some limiting value, such as 1500 p. p. m. The solids deposited in a superheater as a result of carryover may actually almost plug a tube before failure occurs, as illustrated in Figure 11. Suchdeposits, as well as those in the noncondensing stages of the turbine, reflect tbe composition of tile boiler water, consisting largely of sodium salts together with some calcium phosplrate, as indicated by the analyses in Table 111. In the later stages of a turbine, a t about the p i n t where condensation commences, the character of the deposit changes from a mixture of soluble sodium compounds to an insoluble coating high in silica, represented by the last analysis in Table 111. According to Straub (6.9) the accumulation of soluble salts on turbine blades may be prevented if the ratio of sodium sulfate to sodium hydroxide in the boiler water is greater than 4 or 5 . The essential idea behind this is that, although droplets of water containing sodium hydroxide alone evaporate to sticky liquid particles of Concentrated caustic soda when passing through the superheater, droplets of water containing sodium sulfate alone evaporate to solid particles of dry powder, which presumably do not adhere to the turbine blades. Straub demonstrated experimentally that, by maintaining a high ratio of sodium sulfate to alkalinity in the boiler water, the tendency towards deposition may he decreased. The higher salt concentration may, however, actually increase the amount of carry-over from a boiler.
VOL. 31, NO. 4
are notoriously productive of foaming. Even worse conditions may he caused by saponification in the boiler of the vegetable oils present in the compouiided cylinder oils used on reciprocating engines (83). TABLE111. PERCENTAGE COMI'OSITION OF REPRESENTATIVE DEPOSITB FROM A SUPERBEATER AND FROM DIFFERENT STAGES OF k. TURBINE
NS*SO,
AS. 3 20.6 6.8
N*lCOi
NZCl
sio*
1.2
CdPO'h
LO
MICSiOs
0.5 8.7 8.3
Fe@:
I&nlilon loss
..
+:3 81.3 6.9
0:j 79.3
412
3.2 5.7 0.1 4.2
..
..
Up to the present time no material has yet been discovered which, when added to a boiler, will give even moderately permanent protection against foaming. The temporary effect of castor oil is striking hut is soon lost, owing to saponification which is likely to produce a final condition worse than the original. Treatment of raw water and of condensate returns when foam-inducing constituents are present may be costly, but is at present the most effective means of eliminating trouble. The hope lingers, however, that in spite of previous hck of success a stable and reliable antifoam may some dav be developed.
Water Conditioning for Marine Boilers Marine boilers are subject to the same troubles &s those in stationary plants, so that much in the preceding discussion relative to scale, corrosion, cracking, and carry-over applies directly. Nevertheless, a few points merit specific mention. In the first place, corrosion even more than scale was the original chief problem of the marine boiler operator. By frequent mechanical cleaning, usually every few weeks, i t was possible to keep scale formation below a dangerous level hut boiler metal lost by corrosion could not be replaced. The emphasis therefore has been largely on maintaining sufficient alkalinity to minimize attack of the boiler water on the steel. A fundamental difference between the merchant marine and the Navy is that although the former may count upon carrying bunker water from port to port for make-up to its boilers, the Navy considers i t necessary to distill all make-up from sea water in order to be independent of shore supplies. During and after the World War, the Navy continued to use the standard boiler compound devised primarily for the prevention of corrosion by Lyon (@) and slightly modified from time to time. In 1929 this compound had the conipsition shown in Table IV and was fed to the boilers to maintain an alkalinity 0.2 to 0.5.per cent normal with respect to sodium carbonate (16'). TABLEIV. COMPOSITION OF NAVY STANDARD BOILER COMPOUND
FIGURE10. CRACKS INDRUMOF ~~OO-POUND BOILER AT FEEDLINECONNECTION The problem of carry-over would he comparatively simple if foaming were induced only by inorganic substances in the boiler water. Contamination by organic materials constitutes a more serious problem. Thus certain waters from swamp regions, or from rivers carrying waste liquors from paper mills,
NniC01 NaaPOs.l2H*O NerHPO'
1929 70
10
..
1929
1033 44
Starch
47
watei
..
T*lllliZl
1 2
11
1933
9
..
..
The upward trend of pressure in Navy boilers eventually led to revision of the standard boiler compound. On the basis of extended tests at the Naval Engineering Experiment Station, Solherg and Adams (6'7)arrived at the composition shown in Table IV for 1933 and recommended the use of this compound to maintain boiler-water alkalinities from 0.4 to
APRIL, 1939
INDUSTRIAL AND ENGINEERING CHEMISTRY
0.7 per cent normal, corresponding ti, the limits already found most satisfactory for marine operation with controlled phosphate conditioning. The use of Navy Boiler Compound 1933 under these conditions did not actually mean that adequate phosphate conditioning would be maintained. Even though a soap hardness test was suggested to supplement the alkalinity test, control was based on the latter. This meant that if the distilled make-up were contaminated by sea water from evaporator carry-over or condenser leakage, the phosphate content of the boiler water would be carried unduly high; on the other hand, if the distilled make-up dissolved the cement wash sometimes used to prevent corrosion in the storage tanks, the phosphate would be inadequate to prevent the formation of scale. After the World War, while the Navy was still using the compound developed by Lyon, the U. S.Shipping Board was followinga somewhat related policy. In 1923 the Department of Operations issued instructions in which the use of sal soda or soda ash with kerosene was specified for boiler water treatment to supersede all boiler compounds. Expensive and timeconsuming mechanical cleaning, which had always characterized marine boiler operation, was not obviated by this treatment. In 1930 when the initial marine installations of eontrolled phosphate conditioning were made, a survey indicated that a vessel equipped with three Scotch boilers would normally lose about 12 days of operation per year as a result of lay-up for mechanical boiler cleaning; the expense averaged 5100 per boiler when scale was removed every six months. A vessel equipped with twelve water-tube boilers was spending %1100 for mechanical boiler cleaning a t the completion of each voyage of six weeks; and the Mujeslic, at that. time the largest marine boiler plant in the world with forty-eight water-tube boilers, was cleaning twenty-four boilers per month a t an average cost of $50 per boiler. To minimize cleaning costs and possible boiler failures, many of the newer vessels operating a t higher pressures up to 350 pounds were distilling all make-up at a cost ranging upwards from 50 cents per ton. Since its first applications, phosphate conditioning has demonstrated its ability to maintain scale-free evaporative surfaces in marine boilers with raw water make-up a t a fraetion of the cost of evaporated make-up. However, many designers and operators apparently unfamiliar with the fact that stationary plants are operating satisfactorily on 100 per cent of chemically treated make-up at pressures up to 800 pounds are reluctant to use anything except distillation for the 1to 2 per cent make-up necessary in marine units. The opposed concepts of compound treatment based on alkalinity alone and of conditioning based on independent control, both of phosphate and of alkalinity, have been in eonfliet since 1930. In 1935 the U. S. Shipping Board Bureau, which succeeded the U. 5. Shipping Board, issued its Circular No. 2 on “Feed and Boiler Water Treatment and Testing,” which was essentially a reprint of Chapter 6 of the Xavy Department’s Manual of Engineering Instructions, covering the use of Navy Boiler Compound 1933. Shortly thereafter in Circular No. 3 the bureau called attention, however, to the fact that make-up for Navy boilers was invariably distilled, whereas the use of raw or shore waters was the prevailing practice among merchant ships; it recommended that checks should be made periodically by a competent chemist or chemical engineer on tlie condition of the boiler water and the results of treatment. At present with higher pressures and forced circulation occasioning intense discussion in both the merchant marine and the Navy, American marine practice is in an uncertain stage of transition similar to that which became apparent in the central stations ten to fifteen years ago. It seems probable
397
that one of the features of this devebpment will be a trend towards more thorough conditioning of boiler water.
Water Conditioning for RaiIroads Water treatment has made a significant contribution to the economy and performance of steam locomotives, but water conditioning, in the sense of control of the water within the individual boilers in the manner now general in stationary plants, has scrweely entered the railroad field. Treatment
FIGURE 11. DEFOSIT IN SUPERHEATER TURE
comprises the use of compounds or soda ash added direct to the engine tank, the addition of soda ash to the water in wayside treating tanks, partial softeuing with soda ash in larger units, and complete cold-process lime-soda softening. The development of railroad water practice was well described by Bardwell (6) and by Kiiowles (98). Difficulties due to scale have heen minimized by most of tlie larger railroads at a saving estimated in 1923 by the Water Service Committee of the American FLailway Engineers Association at 13 cents per pound of scale-forming constituents removed from the feed water. Corrosion has been considerably reduced by carrying relatively high alkalinity in the boiler water, but open feed-water heaters, which were proposed for the purpose of deaeration by Koyl in 1925 (SS),are currently regarded as the best protectiou against pitting. Carry-over continues to be a problem iu contemporary locomotive opsration, although more careful control of softening and of treatment with soda ash has improved conditions. Compounds consisting essentially of emulsions of castor oil are prepared by some railroads for their own use as antifoaming agents. Adequate periodic or continuous blowdown, with systematic checking of boiler water concentrations at terminals has, however, been perhaps the most significant factor in reducing carry-over. Difficulties with cracking in riveted seams and evidence that maintenance of the A. S. M. E. sulfate-alkalinity ratios did not obviate this eraeking have recently caused those re-
398
INDUSTRIAL AND ENGINEERING CHEMISTRY
sponsible for railroad water supplies to turn to the organic inhibitors of embrittlement noted previously (63). T h a t further refinement in water treating practice on the railroads might be anticipated was pointed out in 1927 by Knowles (37). Although the large number of boilers represented by railroad locomotives obviously cannot be given the individual attention applied to a large central-station unit, i t seems reasonable to expect that the trend in railroad water supply will be in the direction of more and more controlled conditioning, particularly if high-pressure, forcedcirculation condensing units such as the Steamotive (5) make a place for themselves.
Control of Conditioning Knowledge of the actual conditions in the boiler water and steam in order that they may be properly controlled is the foundation of conditioning, as distinguished from the preparation or treatment of feed water. It is not strange, therefore, that the development of conditioning has been paralleled by a continuous development of methods of analysis. When i t is realized that many of the concentrations which are important in a boiler water lie in a range so low that the average chemical analysis could almost dismiss them with the traditional report of “trace,” and when the necessity for relying in many cases upon the analytical ability of men with little or no chemical training is considered, the problem of developing suitable control methods may be appreciated. When the first conditioning with soda ash was undertaken, the standard alkalinity titration to the end points of phenolphthalein and methyl orange was used, but the resultant gross errors in the calculated concentrations of carbonate ion soon led to the development of the modified Winkler method (g9). Later the evolution method for total carbon dioxide gave a still more certain means of controlling carbonate conditioning (53)* In the meantime, however, phosphate had largely supplanted carbonate as a conditioning radical. After finding that colorimetric methods for phosphate were uncertain in highly colored boiler waters, Hall (28) devised a rapid control method based upon the volume of ammonium phosphomolybdate precipitated under specified conditions. For relatively clear boiler waters, Schroeder (52), working under the auspices of the Joint Research Committee on Boiler Feed Water Studies, found a colorimetric procedure utilizing aminonaphtholsulfonic acid as a reducing agent to be rapid and accurate. During the same investigation Schroeder also modified the method for the determination of hydroxide by substituting strontium for barium chloride as a precipitant for carbonate and phosphate (52); this tended to improve the end point of the Winkler titration by minimizing the precipitation of sulfate. Various modifications of the turbidimetric method for sulfate have been employed to check the A. S. M. E. sulfatealkalinity ratio and to indicate the concentration of the boiler water. Direct titration with tetrahydroxyquinone as an internal indicator, devised by Schroeder (59),has proved rapid and satisfactory on waters fairly free of color and turbidity. The introduction of sodium sulfite as a chemical scavenger for dissolved oxygen necessitated not only the development of methods of testing for this constituent, but also of eliminating its interference in other determinations-notably the standard titration for chloride, the tetrahydroxyquinone method for sulfate, and the estimation of phosphate by means of the precipitate of ammonium phosphomolybdate. Similarly, where a boiler water contained appreciable organic matter, whether introduced as a natural constituent of the feed water or as a dispersing agent for calcium phosphate
VOL. 31, NO. 4
sludge, it has been necessary to develop special procedures to ensure measurement only of excess phosphate available for precipitation without including unavailable colloidally dispersed calcium phosphate. By far the greatest refinements have been introduced in the determination of dissolved oxygen. The basic method of Winkler has been modified by Yoder and Dresher ( 7 7 ) )errors have been canceled by the double titration introduced by Schwartz and Gurney (66))and improvements in the technique of the latter method have been discussed by Daugherty (15). Initial separation of the oxygen from possible interfering substance has even been proposed (76). Early in the development of boiler water conditioning, Hall and Merwin (32) applied the polarizing microscope to the identification of the solid phases deposited from boiler waters. Further use of this valuable tool in connection with boiler problems was described by Powell (56) and by Partridge (50). Measurement of the p H of feed and boiler water, first, by calorimetric and more recently by electrometric methods, has become an important factor in the control of many plants. Steam quality, which was once determined in per cent by the colorimeter, is now measured in parts per million of dissolved salts by means of the electrical conductivity of the condensate. From the pioneel- work of Hecht and McKinney (33) to the contemporary development by Powell (67), electrical conductivity has played an increasingly important part in checking boiler operation with respect to carry-over. Much has necessarily been passed over in this brief review of control methods, but i t is evident that the chemist has made a place for himself in the power plant.
.
“Preventive Medicine” in New Boilers
A fair share of the difficulties encountered in higher pressure boilers installed during the last ten years must be accepted as the inevitable accompaniment of pioneering. Too large a share, however, particularly during the last five years, may be attributed to the highly competitive system under which a new boiler is usually acquired by a plant. When the normal instinct to obtain whatever is needed. for as little money as possible, meets the necessity for entering the lowest bid, the result is likely to be a boiler which might meet the guarantees under the most ideal conditions, but which will almost certainly present serious problems in actual operation. I n the long run, both the boiler manufacturer and the purchaser are likely to lose money which might well have been saved if freedom from operating difficulties had been given as much consideration as initial cost. As things stand, the expert in water conditioning not infrequently is called upon to solve a problem which originates directly in some feature of the boiler design, and which would never have existed but for the great delusion that something can be obtained for nothing. The cost of modifying the boiler is usually many times the increment in price which would have been necessary to obviate trouble when the boiler was originally constructed, so that no changes are made until every expedient has been tried with respect to water conditioning. Sometimes such an expedient will allow a plant to operate, sometimes mechanical changes finally are found unavoidable. I n either case the experience is likely to be costly for everyone concerned. From a practical engineering viewpoint, a boiler should be designed not merely to produce steam from water, but to produce i t a t the minimum long-range cost from the water actually available. This may involve a choice between spending, on the one hand, more money for pretreatment and less money for the boiler or, on the other, putting more money into a boiler which may be expected to operate satisfactorily on a water supply of inferior quality.
APRIL, 1939
INDUSTRIAL AND ENGINEERING CHEMISTRY
One of the outstanding examples of a new boiler plant carefully engineered with respect to the water problem is the recent 800-pound topping installation of the Weirton Steel Company, a t Weirton, W. Va. The two boilers in this plant have operated since June, 1936, on 100 per cent make-up of Ohio River water softened in a hot-process lime-soda softener, with metaphosphate used for conditioning of the boiler water. Although the raw water is high in dissolved solids, variable in composition. and contaminated with industrial wastes, no difficulties due to water have been encountered. This success may reasonably be attributed to the constructive design arrived a t by the cooperation between the company, the boiler manufacturer, and the consultants on water conditioning. This design included such features as large furnace volume with a correspondingly low value for heat release of 25,000 B. t. u. per cubic foot per hour for normal and 31,000 for maximum rating; a normal steam production of 12,500 pound per hour per foot width of boiler and a steam drum of greater diameter than standard.
Acknowledgment The writers are grateful for assistance rendered by The Babcock and Wilcox Company, Combustion Engineering Company, Inc., Cochrane Corporation, Permutit Company, and by A. G . Christie and various members of the staff of Hall Laboratories, Inc.
Literature Cited Anonymous, Power, 47, 108, 127 (1918). Ibzd., 50, 624 (1919). Applebaum, 8. B., and Riley, R., IND.ENG. CHEM.,30, 80-2 (1938). Bailey, E . G., J . A m . Ceram. Soc., 17, 55-67 (1938). Bailey, E. G., Smith, A. R., and Dickey, P. S., Mech. Eng., 58, 771-80 (1936). Bardwell, R. C., J . A m . Water Works Assoc., 22, 335-41 (1930). Berg, E., Power, 48, 834 (1918). Berl, E., and von Taack, F., Arch. Wtlrmewirt., 9, 165-9 (1928). Bird, P. G., Proc. A m . SOC. Testing ikfaterials, Preprint 101 (1938). Boiler Code Comm., Mech. Eng., 51, 388-9 (1929). Boilers and Furnaces Number, Power Plant Eng., 42, 2-86 (1938). Burrell, Harry, IND. ENG.CHEM.,30, No. 3, 358-63 (1938). Caldwell, W. E., Combustion, 10, No. 1, 18-21 (1938). Christman, C. H., Holmes, J. A., and Thompson, H., IND. ENG.CHEM.,23, 637-46, 849-50 (1931). Daugherty, T . H., Proc. A m . SOC.Testing Materials, 37, Part 11, 615-33 (1937). Dinger, H. C., J . A m . Water Works Assoc., 21, 95-6 (1929). Doremus, C. A., U. S. Patent 404,180 (May 28, 1889). Drewry, M. K., Combustion, 9, No. 8 , 18-24 (1938). Foulk, C. W., IND. ENG.CHEM.,16, 1121-5 (1924). Foulk, C. W., and Brill, H. C., Ibid., 27, 1430-5 (1935). Foulk, C. W., and Whirl, S. F., Ibid., 26, 263-7 (1934). Hall, R. E., Ibid., 17, 283-90 (1925). Hall, E. E., J . A m . Water Works Assoc., 21, 79-100 (1929). Hall, R. E., Mech, Eng., 46, 810-17 (1924). Hall, R. E., Natl. Electric Light Assoc., Pub. 051 (June, 1930). Hall, It. E., Proc. Eng. SOC.Western Pa., 41, 347-90 (1925). Hall, R. E., U. S. Patent 1,613,656 (Jan. 11, 1927). Ibid., p. 8, lines 108-121. Hall, R. E., et al., Carnegie Inst. Tech., Mining Met. Invest. Bull. 24 (1927). Hall, €L E.’, Fischer, C., and Smith, G. W., Iron Steel Engr., 1, 312-27 (1924). Hall, FL. E., and Jackson, H. A., U. S. Patent 1,903,041 (March 28, 1933). Hall, R. E., and Merwin, H. E., Trans. A m . Inst. Chem. Engrs., 16, 11, 91-117 (1924). Hecht, M., and McKinney, D. S., Trans. A m . SOC.Mech. Engrs., F u e l s Steam Power. 51. 139-59 (19301. Hitchens, R. M., and Purssell, J. S., Trans. A m . SOC.Mech. Engrs., 60, 469-73 (1938). Johnson, H. il., and Xelting, C. A., Combustion, 9, No. 3, 20-9 (1937). Joseph, A. F., and Hancock, J. S., J. SOC.Chem. Ind., 46, 315-211‘ (1927); 49, 369T (1930).
399
Knowles, C. R., J. Am. Water Works Assoc., 17, 51-7 (1927). Ibid., 23, 481-8 (1931). Koyl, C. H., Ibid., 21, 1013-23 (1929). Kreisinger, H., Trans. Am. SOC. Mech. Engrs., 60, 289-96 (1938). Loiseaux, A. S., Power, 47, 601-2 (1918). Lyon, F., J . A m . SOC.Naval Engrs., 24, 867 (1912). McDermet, J. R., Trans. Am. SOC.Mech. Engrs., 42, 267-94 (1920). Mellanby, A. L., Trans. Inst. Marine Engrs., 50, I1 (1938). Moberg, A. R., Combustion, 3, 36-9 (1931). Mumford, A. R., Trans. A m . SOC.Mech. Engrs., 51,363-73 (1929). Parr, S. W., Univ. Ill. Eng. Expt. Sta., Bull. 94 (1917). Parr, S. W., and Straub, F. G., U. 5. Patent 1,190,403 (May 23, 1933). Parr, S. W., and Straub, F. G., Univ. Ill. Eng. Expt. Sta., Bull. 155 (1926). Partridge, E. P., Proc. A m . SOC.Testing Materials, 37, Part 11, 600-8 (1937). Partridge, E. P., Univ. Mich. Dept. Eng. Research, Eng. Research Bull. 15, 100 (1930). Partridge, E . P., and Schroeder, W. C., Am. Soc. Mech. Engrs., Research Publication, 1933. Partridge, E. P., and Schroeder, W. C., IND. ENQ.CHEM.,Anal. Ed., 4, 271-8 (1932); Collins, L. F., and Schroeder, W. C., Ibid., 4, 278-83 (1932). Partridge, E. P., and White, A. H., IND. ENG.CHEM.,21, 834-8 (1929). Payen, Dingler’s Polytech. J . , 10, 254 (1823). Powell, S. T., Combustion, 5, No. 3, 15-19 (1933). Ibtd., 9, NO. 5, 25-31 (1938). Powell, 5. T., McChesney, I. G., and Henry, F., IND. ENQ. CHEM.,30, 400-6 (1938). Schroeder. W. C.. IND. ENG.CHEM..Anal. Ed.. 5. 389-93 (1933). Schroeder; W. C.; et al., J . A m . Chem. SOC.,57, i539-46 (1935j; 58, 843-9 (1936); 59, 1783-90 (1937). Schroeder, W. C., and Berk, A. A., Am. Inst. Mining Met. Engrs., Tech. Pub. 691 (1936); Combustion, 7, No. 8, 29-33 (1936). Schroeder, W. C., Berk, A. A., and Fellows, C. H., J. Am. Water Works Assoc., 30, 679-94 (1938). Schroeder, W. C., Berk, A. A., and O’Brien, R. A., Trans. Am. SOC.Mech. Engrs., 60,35-42 (1938). Schroeder, W. C., Berk, A. A., and Partridge, E. P., J. Am. Chem. SOC.,59, 1790-5 (1937). Schroeder, W. C., Berk, A. A., and Partridge, E. P., Proc. A m . SOC.Testing Materials, 36, 746-7 (1936). Schwarts, M. C., and Gurney, W. B., Proc. A m . SOC.Testing Materials, 34, Part 11, 796-820 (1934). Solberg, T. A., and Adams, R. C., Jr., Combustion, 5, No. 6,24-9 (1933), discussion pp. 32-7. Straub, F. G., Univ. Ill. Eng. Expt. Sta., Bull. 261 (1933). Ibid., 282 (1936). Ibid., 283 (1936). Straub, F. G., and Bradbury, T. A., Mech. Eng., 60, 371-6 (1938). Straub, F. G., and Bradbury, T. A., Power Plant Eng., 40, 1045 (1936). Taff, W. O., Johnstone, H. F., and Straub, F. G., Trans. Am. SOC.Mech. Eng., 60,261-5 (1938). Van Brunt, J., Combustion, 10, No. 1, 27-34 (1938). White. A. E.. Am. SOC.Mech. Ene.. -4nn. Meetina. - Dee. 5-9. 1938, Preprint 12. White, A. H., Leland, C. H., and Button, D. W., Proc. A m . SOC. Testing Materials, 36, Part 11,697-720 (1936). Yoder, J. D., and Dresher, A. C., Combustzon, 5, No. 10, 18-22 (1934). PREBENTED before the meeting of t h e American Institute of Chemical Engineers, Philadelphia, Pa.
Commercial Mixed Fertilizers-Correction I n the article by F. 0. Lundstrom and A. L. Mehring [IND.ENG.CHEM.,31, 354-61 (1939)], a n unfortunate error occurs in Table I, page 357. T h e grade formulas, states, and dates which head the columns of figures should not have been printed here, since they are obviously repeated from t h e t o p of the first group of samples on page 356 and do not apply t o this second group of samples. T h e first fourteen lines of figures on page 357 belong, of course, t o the group of samples beginning with “3-21-6, Ohio, 1929” and ending with “4-24-12, Ind., 1934.”