As normal air is about two-thirds to three-fourths saturated and as the temperature in summer may easily rise to 85°, it is very obvious that we have in summer to deal with moisture sometimes reaching a pound or more per thousand cubic feet, and such is the case in many northern iron-producing districts, the average moisture for some months being nearly a pound, while in Alabama the conditions are much worse than this, a pound and a half per thousand cubic feet being not unusual.

Accepting it as proven for the present that the action of moisture on the operation of the blast-furnace is exceedingly detrimental, and realizing that the above figures indicate that we may easily be pumping 50 lbs. or 6 gals. of moisture per minute into a modern furnace with the blast, the question obviously arises, how shall we eliminate it?

The metallurgical world now knows how the question was solved after many years of experiments by James Gayley at the Isabella furnaces in Pittsburg, the moisture being condensed out of the air by artificial refrigeration. Obviously, since the quantity of moisture in saturated air rises with great rapidity with rise of temperature, as shown in Fig. 239, we have only to lower the temperature to cause the precipitation of the moisture. In practice we have to do more than this, since lowering the temperature tends to precipitate the moisture as fog, which, in its form of water mechanically suspended, may easily pass with the rapid current of air through the refrigerating apparatus and thus to the blowing engines and the furnace. Care must therefore be used in the design not only to condense the moisture but to precipitate it out in drops large enough to leave the air. Preferably the drops should form on the refrigerated surfaces and drip from these in such a way that the air current cannot pick the drops up again.

Refrigeration requires power, and what is more serious it requires investment of capital. Therefore it is very important to design the refrigerating plant to use the least amount of power possible since this will in general mean the smallest capital cost also, and the capital cost of this process has been its greatest disadvantage.

The original dry-blast plant, that at the Isabella furnaces of the United States Steel Corporation, was of the brine circulation, single-stage type for the reason that in the vast metallurgical experiment upon which Mr. Gayley was engaged, he felt compelled to take the simplest and easiest means to accomplish the result, irrespective of whether they were the most economical, and on account of the success of this system at the Isabella plant, and of some failures of early attempts to improve upon it, the same system was used for all the early dry-blast plants. Briefly this system is as follows:

The ammonia vapor is compressed by a steam-driven compressor and condensed in a standard ammonia condenser, from which it passes to a brine cooler where it is evaporated, absorbing the necessary heat for evaporation from the brine and thereby cooling the latter. The brine, a solution of salt or calcium chloride with a freezing point many degrees lower than that of water, is then circulated through other coils in a huge chamber through which the air passes on its way to the blowing engine, the air being blown into the chamber by a large fan and removed from it by the suction of the blowing cylinders. Ice forms on the coils, which soon reduces their cooling power and obstructs the passageways for the blast. This necessitates putting out of commission each of the coil chambers in turn every few days while it is thawed off by steam, an expensive operation since not only the steam but the ice, cost coal, and the operation of thawing itself takes some time and labor.

A plant of this type designed by Frank C. Roberts & Co. is shown by Fig. 240. The captions under the different views render them very easy of comprehension.

A careful study of the refrigerating process makes it obvious that this system of refrigeration is capable of great improvement, both from the point of view of first cost and that of power requirements. The general principles controlling this whole matter were pointed out at some length in my paper before mentioned, part of which is reproduced below.

It is well known that the power required for a given quantity of refrigeration per minute, Q, is theoretically proportional to T1-T2/T1 where T1 is the absolute temperature (Fahrenheit + 460 deg.) at which the heat is absorbed by the system and T2 that at which it is rejected, the expression being HP=(778/33,000) Q (T1-T2 /T1) (1) where Q is measured in B.t.u. per minute.

The ammonia-compression refrigerating machine having so largely supplanted all others for land service, we shall consider only that type. Upon it, fortunately, reliable investigations have been made which enable us to establish a relation between its theoretical and its actual performance, and this goes far towards answering our questions.

For a number of tests in each of these series I have calculated the theoretical horse-power by equation (1), and compared it with the actual indicated horse-power of the steam cylinders, the ratio of the latter to the former being called R, it being, in fact, the reciprocal of the efficiency on this basis.

These values of R have been plotted as ordinates with the corresponding values of T1-T2/T1 as abscissae, as shown on Fig. 241.

The values of R plainly decrease with the value of T1-T2/T1 and it is clear that when the temperature range is zero, R should 1, the efficiency being perfect. Accordingly, a straight line has been drawn passing through the point (T1-T2/T1=0,R-1) and this line coincides with the points with surprising accuracy, and may be taken as expressing the law of this relation correctly.