Another respect in which the pre-compression and the post-compression system differ is that with the former all the air is delivered to the blowing engine cold, and the reduction in volume resulting from this causes a proportional reduction of the power required by the main blowing engine. On the other hand, the refrigeration imparted to the air at a considerable expense, is lost except for this saving. Now it can easily be proven by thermo-dynamics that if both the refrigerating machine and the blowing engine were perfect the total power required to refrigerate the air down to any desired temperature, and then compress it from that low temperature, would be exactly the same as the power required to compress it directly from atmospheric temperature. It becomes therefore a question as to which machine is the most efficient, a question easy to decide, for we know that good blowing engines are from 85 to 90 per cent. efficient, while even good refrigerating plants are only around 55 per cent. It is obvious, therefore, that it pays to save power in the refrigerating machine even at the expense of the blowing engine. But another point is to be considered here. When the pre-compression system is used the blast containing much of the heat of compression passes at once to the stoves and of course reduces the amount of heat to be imparted by the stoves proportionately, whereas the post-compression system, if applied the same as the original pre-compression systems, would deliver strongly refrigerated air to the stoves with corresponding increase in the heat required of them.

It is partly to overcome this disadvantage that the Carrier Company uses the first exchanger shown in its system, since by the aid of this it can warm up the outgoing air to within a few degrees of the temperature of the incoming air. This puts matters on practically an even basis as regards the heat to be supplied by the stoves.

Using the post-compression system permits giving paramount consideration to the power requirements, and these, as has been pointed out, are greatly reduced by the use of exchangers and the two-stage system. They are also further reduced by the fact that the temperature to which the air must be refrigerated is not nearly so low as under the pre-compression system to reduce the moisture to the same number of pounds per minute. This has a very decided bearing on the power requirements, since it has been shown (see article in Metallurgical and Chemical Engineering, December, 1912, above mentioned) that the power required for any given quantity of refrigeration (Q) is HP = 0.0235 Q (Z + 3.7 Z2) in which Z is equal to T1-T2/T1 and this is obviously less as the temperature to which we refrigerate is raised. Moreover, while the quantity of refrigeration for the condensation of the moisture is the same, that for the cooling of the air is less, because the temperature to which the given weight of air is cooled is not so low in the post-compression system as in the other, it is therefore obvious that the power required must be very materially less with this system than with pre-compression.

A considerable advantage in the use of exchangers inheres in the post-compression system for the reason that the greater density of the air greatly increases the ease with which heat may be imparted to or abstracted from it, and the exchangers can therefore be smaller on this system than they could for an equal degree of efficiency if used on the pre-compression system.

The slightly lowered temperature at which the air is delivered to the stoves constitutes the principal objection to this system when the blast-furnace pressure is twelve pounds or more, but when the pressure is less than this the reduction in volume of the air is very much smaller and the advantage of the system falls off proportionately, so that there may be cases in which it is desirable to use the pre-compression system, but in such cases there is no excuse for not using two-stage cooling and regeneration.

It may be thought that the cost of the regenerator is just that much addition to the total cost, but this is by no means the case. The regenerator does, without consuming any energy, a part of the refrigeration which would otherwise have to be done by the refrigerating machine, and therefore makes a proportionate reduction in the size of refrigerating equipment required, whose cost per unit of work done is of course very much greater than the cost of the regenerator.

By the use of the rain system two great advantages are obtained. First: The avoidance of the heavy cost for brine-circulating coils and the chambers in which to enclose them. Second: The avoidance of the additional drop of temperature required to force the heat from the air into the brine through pipe walls which involves a lower temperature of refrigeration for a given degree of moisture removal. The whole plant is made smaller and more compact and in every way cheaper and better by this system.

If rain be used for carrying the heat from the refrigerating coils to the air, as is most desirable, and if the second stage of the refrigeration takes the temperature below the freezing point as a pre-compression system should probably do, it becomes necessary to use a non-freezing solution for this purpose. Brine, preferably made with calcium chloride, should be used and as this obviously takes up most, if not all of the moisture from the air which it cools, an evaporator must be arranged at some point in the circuit to remove the accumulated water from the system and maintain the strength of the brine. In such a case a heat exchanger should be installed between the refrigerating system and the evaporator so that the "cold" in the incoming and the heat in the outgoing brine may both be conserved.