COMBUSTION-CHAMBER BAFFLE WALL.

A baffle wall, constructed of special fire brick, was built in the combustion chamber. The object of this wall was to divert the gases from their straight course in order to mix the free oxygen more thoroughly with the unburnt volatile matter of the coal. It was also intended to act as a heat regenerator, absorbing heat when the temperature of the gases was high, and giving it out when the temperature was low, thus keeping the temperature above the ignition temperature of the distilled gases.

It was learned by experiment that only large blocks made of the best material could stand the high temperature and the slagging action

FIG. 34.-Gas-mixing wall built in the Heine safety water-tube boiler.

of the gases for any length of time. The baffle wall shown in fig. 34 was built of large fire brick, 18 by 12 by 6 inches, said to be of the best material that could be obtained. The wall was built in three sections. The bottom consisted of seven blocks set on end, forming pillars, on top of which six similar blocks were laid diagonally across. The space between the baffle thus formed and the tile roof was filled with small bricks of good material, so that the spaces between the pillars gave the only passage left for the gases. The object of this construction was to divert the stream of furnace gases, which struck the upper portion of the baffle, and break it into many smaller streams,

thus mixing the distilled gases and the free oxygen. It is probable that eddies caused by the obstacles in the path of the gases greatly aided the mixing. The first baffle wall of this construction lasted just six months, but later ones were not so durable, and such walls were finally abandoned, at least temporarily, and replaced by three small piers set on the bridge wall. The first baffles, however, were far more efficient as smoke preventers and heat regenerators, although they absorbed considerably more draft.

COMBUSTION-CHAMBER TEMPERATURE.

Fig. 35 shows the manner in which the "unaccounted-for" and flue losses of about 260 tests vary to some extent inversely. The encircled

FIG. 35-Relations of combustion-chamber temperature (°F.) to heat balance (with Illinois and Indiana coals): Curve No. 1, per cent of unaccounted-for loss plus per cent of loss up stack; curve No. 2, per cent of loss up stack; curve No. 3, per cent of unaccounted-for loss.

numbers near the points indicate the number of tests falling within each temperature group. The inverse variation of these curves is a consequence of the facts that the flue-gas analyses, though accurate. in themselves, were misleading because of air infiltration, and that the flue temperatures obtained were unreliable because of inherent difficulties; both these causes are discussed in the section devoted to accuracy and reliability of data (pp. 149-151). It is noticeable on this chart that the sum of the flue loss and unaccounted-for loss increases slightly with rise of temperature, notwithstanding the fact that the efficiency of the boiler as a heat absorber increases several per cent with the rise of temperature; thus again the indication is that higher temperatures accompany less complete combustion.

ATTEMPTS TO EXPLAIN UNACCOUNTED-FOR LOSS IN HEAT BALANCE.

The large percentage of "unaccounted-for" loss appearing in the heat balances of the many tests made by the boiler division has long been a cause of discussion.

The settings of the boilers used for testing the various coals have been, even with the utmost diligence, in a very unsatisfactory condition. There were numerous air leaks, so that the furnace gases were always diluted when they reached the base of the stack. It was thought that this dilution of gases introduced error in the calculation of heat loss up the stack. It also seemed that the radiation loss was larger than it should be.

Recently a new setting has been built and completely inclosed by a sheet-iron casing made with air-tight joints, so that the leakage was reduced to a minimum. The walls of the new setting were built of hollow tiles and this with the addition of a sheet-metal casing should have reduced the radiation loss; at least this loss could not be greater than formerly. Still the unaccounted-for loss continues.

A great many classifications have been made on the data and results obtained from our tests. In all cases a few relations con

tinually appear. The important one in this discussion is that low efficiency is always accompanied by the highest per cent of CO in the flue-gas analysis, and the highest efficiency by the lowest per cent of CO. Moreover, the high CO values always go with the highest combustion-chamber temperatures. It is also found that as the combustion-chamber temperature increases the per cent of black smoke increases.

It is not possible to account for very much of the loss in burning coal by the amount of CO found in the gas analysis. Therefore it seems that this appearance of CO in the gas analysis is indicative of bad conditions, such as an irregular fuel bed or the escape of hydrocarbons unburned.

RELATION OF NITROGEN IN FLUE GASES TO EFFICIENCY 72* AND TO SIZE OF COAL, AND OF PER CENT OF CO TO EFFICIENCY 72*.

The curves of fig. 19 (p. 28), based on tests 89 to 401, inclusive, present a combination of two charts. The upper one classifies the tests on a nitrogen basis. This nitrogen is presumably what is left in the flue gases after subtracting from 100 the sum of the percentages by volume of carbon dioxide, oxygen, and carbon monoxide, neglecting any small traces of hydrocarbons which may have been present and were not determined. Curve No. 1 indicates that the code "boiler efficiency" (72*) increases markedly when the nitrogen content rises from 80 to 81 per cent-that is, when Orsat gasanalysis totals decrease from 20 to 19 per cent. The reader can

choose between the many possible conjectures as to the fundamental significance of this curve.

Curve No. 2 indicates that for any size of coal, up to maximum size, two different gas analyses may be obtained. The reader may make whatever conjectures he chooses.

Curve No. 3, in the lower chart, is very significant. It indicates that a volumetric percentage of CO above 0.4 per cent is threatening to efficiency. Inasmuch as these were analyses of flue gases, into which 10 to 50 per cent of air had filtered after the gases of combustion passed through the combustion chamber (p. 64), this CO content represents a much larger value. The significant fact is that an increase of CO content from 0.3 to 1 per cent is of itself sufficient to account for only about one-third of the drop from 65 to 54 per cent. in code "boiler efficiency" (72*). Figs. 13 (p. 22) and 14 (p. 23) show that the percentage of CO rises with the temperature of combustion; further, the efficiency of the boiler as a heat absorber increases slightly with a rise of furnace temperature; we therefore reach the inevitable conclusion that at least two-thirds of the large drop in code "boiler efficiency" (72*) with rise of CO is due to incomplete combustion losses not represented by CO, so that high CO is a decided danger signal. Curve No. 1 of fig. 18 (p. 27) shows exactly the same thing, with grouping on a "boiler efficiency" (72*) basis. With CO rising from 0.3 to 0.6 per cent the "efficiency" drops from 60 to 55 per cent. The same range of CO in fig. 19 (p. 28) gives the same amount (65 to 60 per cent) of efficiency drop-though in a different region-which is explained by the fact that in grouping any set of related occurrences in different ways the same tests will not often fall in successive groups. For instance, only part of the tests falling in the middle group in one classification are apt to fall in the middle group in any other classification.

RELATION OF CO, IN FLUE GASES TO PER CENT OF COMPLETENESS OF

COMBUSTION (E3).

The curve of fig. 31 (p. 51) was obtained by grouping tests according to volumetric CO, content of flue gases and then averaging the per cent of completeness of combustion (E) of each group. These values of E, were obtained by the mathematical calculation explained on page 139, and although many or all of them may be considerably in error, it is likely that the general shape and the amount of drop of this curve are nearly correct.

The curve shows clearly that as the oxygen is decreased simultaneously with a rise of CO, content the completeness of combustion decreases, which is just what we should expect when reducing the proportion of oxygen molecules present, according to the law of mass action discussed on pages 170–172.

Thus even in furnaces as long as those of the boilers used in the work of the boiler division the increasing incompleteness of combustion with rising temperature is practically sufficient to offset an increase of a few per cent in the efficiency of the boiler as a heat absorber after the furnace temperature has reached about 2,400° F.

RELATION OF DIFFERENCES OF DRAFT TO POUNDS OF DRY CHIMNEY GASES PER POUND OF "COMBUSTIBLE."

The curves of fig. 36 were determined by plotting differences between draft in stack and draft over fire with pounds of dry chimney gases per pound of "combustible." Curve No. 1 represents

FIG. 36.-Relation of differences between draft in stack and draft over fire to pounds of dry chimney gases per pound of "combustible:" Curve No. 1, figured from analysis of gas in stack, tests 120-400; curve No. 2, figured from analysis of gas in rear of combustion chamber, tests 318-380.

about 200 tests. The values for pounds of dry chimney gases per pound of "combustible" were figured from stack samples. Curve No. 2 is the average of only a few points. The values for pounds of dry