This is shown by Fig. 243, in which is plotted the ratio of water removed from 75 lbs. air below 40° F. to the power required for its removal. This is based on 70° F. condenser-temperature and single-stage non-regenerative refrigeration.

The saving in fuel in the furnace is directly proportional to the water removed, and the cost of plant and its operation nearly proportional to the power required, so it will readily be seen from Fig. 243 that a limit must soon be reached from the commercial point of view.

In summer all the uniformity possible would be reached by refrigerating to 32° F., and all trouble from ice formation would be avoided by not going below this point.

As the weather became colder and the dew-point fell so as to be sometimes below this point, the diminished refrigeration required, and the lower temperature of cooling water available, would enable the same plant to maintain a lower temperature and still retain the uniformity desired by reducing the temperature of the air below the lowest natural dew-point probable.

This change being entirely under control of the manager would lead to no sudden changes, and would give a coke consumption in summer so nearly the same as that in winter that the difference would probably be inappreciable, since as between a refrigeration temperature of 32° F. and one of 22° F., with a blast temperature of 1000° F. and normal coke practice, the saving for the latter would be about 2 per cent.

Using this chart as an implement of investigation we may now proceed to examine various methods of refrigeration.

In order to represent these methods graphically, and at the same time avoid marring the diagram Fig. 242, for independent use, it is reproduced in Fig. 244 with the rectangles representing the cases to be discussed drawn upon it.

In order to facilitate comparison and reference certain data corresponding to frequent summer conditions, may be assumed, namely: a temperature of 85° F., and a dew-point of 70° F., a condenser-temperature of 85° F., and that the air is to be refrigerated down to 25° F.

Let us also draw for convenience on Fig. 242 the line of 10° F. temperature-difference parallel to the curve of total entropy (VII).

The conditious assumed are then represented on the diagram by the area aJHda.

The rectangle drawn for these conditions is ZJHG, and its area is 5.60 X 1.89 - 11.56 sq. in., corresponding to 11.56 h.p. for drying 75 lbs. of air between the limits and under the conditions assumed.

It is evident that the area below the curve of total entropy and the line of maximum entropy, above the line of minimum entropy, and to the left of the line of condenser-temperature, represents the theoretical minimum work for the performance of that amount of refrigeration. This must be increased in practice by moving the curve of total entropy 10° F. to the left to give the necessary temperature-head. This is the curve of 10° F. temperature-difference on Fig. 244, and is approximately a diagonal line across the diagram, making the area roughly triangular. On the other hand, the energy-diagram of the ammonia refrigerating cycle is, as already stated, a rectangle, and as it must be as large in every direction as the diagram of energy required, it is obvious that there must be a large quantity of waste room in the former diagram and that, accordingly, energy is being wasted.

Horse power required per pound of water removed per minute in refrigerating blast.

Fig. 243. Horse-power required per pound of water removed per minute in refrigerating blast to temperatures below 40° Fahr. (Direct expansion, single-stage, non-regenerative refrigerator, condenser temperature 70° Fahr.).

This is also evident from an inspection of Fig. 239 since this shows that of all the moisture contained in the air at 70° F. half is precipitated when the temperature is lowered to 50° F., and of that removable down to 25° F. almost two-thirds, yet the heat of all this is removed at the temperature of 25° F.

This is precisely as if, in pumping from a shaft 45 feet deep, in which three-fifths of the water entered in the upper 20 feet, we put one pump at the bottom and let all the water run down to that level before pumping it out. It is really much worse than this, because in pumping heat the power required increases much faster than the head, while with water this is not true.

The remedy is as obvious in refrigeration as it is in pumping, to have two pumps, catch the greater part of the heat near the top of its scale and pump it out from there, pumping from the lower level only the heat which comes in at or near that level, a much smaller quantity.

This means the use of two ammonia-compressor cylinders, one working from the 15° F. suction-temperature as before, the other working from a much higher suction-temperature, say 36° F., both, of course, working to the same condenser-temperature.

The rectangle for the latter of these extends from the line of maximum entropy for the case assumed, J K in Fig. 244, down to the line NM through 46° F., on the "total entropy curve," the end lines being drawn as before, the whole rectangle being JKLM.

The rectangle for the first-mentioned cylinder extends from the line MLN down to that of minimum entropy GH, the completed rectangle being MNGH.

The area of the first of these is 3.50 X 1.25 = 4.375 square inches; of the latter 5.60 x 0.64 = 3.575 square inches; total 7.95 square inches, corresponding to 7.95 horse-power per 75 lbs. of air, as compared with 11.56 horse-power for doing the same work in one stage.

This involves also the great advantage that most of the water is removed at a temperature above freezing, so that the quantity of ice which can form is greatly reduced, even in the coils of the second stage.