Three stages could similarly be used, but with a much smaller saving, as an inspection of the area outside the "85 degree horse-power curve" will show, so that the added complication of another stage would probably be a poor investment, especially as other considerations, to be mentioned presently, contribute to reduce the relative economy and desirability of this step.

The air leaving the refrigerator chamber is at the temperature of 25° F., and it is plain on the briefest consideration that this may be used for cooling the incoming air, with a saving of power to the refrigerating machine.

What the amount of this cooling will be is, of course, not a question of entropy, but simply one of heat. It is not, however, one of easy algebraic solution, since the total heat of the vapor is a very complicated function of the temperature, that of the air being, for narrow ranges, a simple one of direct proportion.

It is for this purpose that the curves of heat are added below those of entropy on the diagram.

The problem is: Given 75 lbs. of dry1 air at 25° F., allowing a temperature-difference of 10° F., and using the counter-current system of cooling, to what temperature will 75 lbs. of air at 85°, saturated with vapor at 70° F., be cooled?

Neglecting the insignificant superheat of the vapor, the air must be cooled from 85° F. to 70° F. before the moisture will be a factor, and the specific heat of air being constant this will obviously warm the dry air up from 25° F. to 40° F.

As 10° F. difference of temperature to give the necessary thermal head is assumed, the dry air will be discharged at 75° F., and the heat absorbed by it, from 40° F. to 75° F., is . obviously that available for cooling the saturated air below 70° F.

Taking from the curve "heat of dry air" that due to 35° F. difference of temperature, we measure it off (towards the axis) from the 70° F. point on the curve of total heat; it comes to 1992 British thermal units, the line through which intersects the total heat curve at 58° F.

This transaction is shown by the lines A1B1C1 on Fig. 244. From this it is obvious that the line of maximum entropy for this case is that through the temperature 58° F., which is simply transferred vertically over to the total entropy curve. This line is shown at VP in Fig. 244, the whole rectangle being VPHG, whose area is 1.11 x 56 = 6.22 square inches = 6.22 horse-power, as compared with 11.56 horsepower for the same work without regeneration.

With this improvement can now be combined that of stage refrigeration, by which is obtained the two rectangles PQRS and TSHG, whose areas are 3.72 x 0.66 = 2.45 square inches, and 5.6 x 0.45 = 2.52 square inches, corresponding to a total of 4.97 horse-power per 75 lbs. air when refrigerated in stages and with the aid of regeneration.

From this it will be seen that great economies are effected by the use of the direct-expansion system, by refrigerating in two stages and by regenerating the "cold" in the air which has passed through the system, for the purpose of cooling down the incoming air.

There is also another possibility in the direction of economy which received consideration in the original development of this process by Mr. Gayley and was brought out in the discussion of my paper, that is, the utilization of the reduction of the volume of the air due to its being compressed for delivery to the furnace. If the blast pressure is fifteen pounds obviously the volume of the blast at atmospheric temperature is reduced to fifty per cent. of its original amount, and obviously also since the quantity of vapor present per cu. ft. is a function of temperature only, this means that if the air were saturated before compression half the moisture would be squeezed out of it by compression, if its temperature were reduced to that of the atmosphere after leaving the blowing engine. This means plainly two things. First, less moisture to be condensed; second, the moisture present will be condensed to any desired limit with a higher dew-point. This also means a saving in power.

1 The trivial amount of water-vapor still present is neglected, being only 0.4 per cent. of the total weight.

In the early plants where the air was cooled by the introduction of vast masses of pipe coils into its path it was virtually impossible to have chambers which would withstand the blast pressure necessary, and of a size to contain this mass of coils, without, prohibitive expense and perhaps danger, therefore in these the refrigeration is always done before the air enters the blowing engine, but in later years the system of cooling the air by direct contact with a "rain" cooled by the refrigerating coils has been introduced and this eliminates the necessity for coils in the refrigerating chambers, thus reducing the latter to relatively small dimensions, and this makes possible the use of the "post-compression" system of refrigeration.

This system has been worked out and applied by the Carrier Air Conditioning Company of America, which also uses regeneration and refrigeration in stages. A plant designed by them is shown by Figs. 245, 246 and 247. Fig. 245 shows a side elevation of the ammonia compressor (motor-driven), and the ammonia condenser, also of the brine coolers, while behind, rising above the shed roof of the compressor house, are shown the outlines of the cooling chambers. Fig. 246 shows a plan of the same equipment. C, D, E and F are the various cooling chambers which can best be described by reference to Fig. 247, but before passing to this it is well to note that the cooling chambers are so arranged in reference to the blast main A that the blast can be thrown through them by closing the intermediate valve B, or by closing the cut-off valves in the connecting necks and opening the intermediate valves, the cooling chambers can be cut off and the blast can go to the stoves direct, thus giving an opportunity to throw the refrigerating plant out of service without shutting down the furnace.