PRECOOLING IN THE LIQUEFACTION OE’ AIR’ BY W. P. BRADLEY AND G. P. 0. FENWICK
Since 1902, experimental inquiries have been in progress in this laboratory regarding the factors which affect the efficiency of an air liquefier of the single-circuit type. Some of the results have already been published.’ The experiments whose results are embodied in the present paper were completed in June last. It appears to have been generally assumed, doubtless under the leadership of Linde and S ~ h r o t t e r that , ~ the absorption of heat at the valve of an air liquefier is due to the Joule-Thomson effect. P -P (289)’ In the formula adopted by Schrotter, D = 2’ -4 7 . ’ the fall of temperature, D, a t the valve varies inversely as the square of the absolute temperature, T, at which the high pressure air reaches the valve. Of course the same absorption of heat which manifests itself merely as a fall of temperature, in case no liquid is produced, results in an increased yield of liquid when once the liquefaction temperature is attained. It is clear therefore that in this view of the matter extraordinary emphasis is thrown upon the efficacy of precooling as a prerequisite for efficiency and economy in liquefaction. Contribution from the Cryogenic Laboratory of Wesleyan University. and Rowe : Phys. Rev., 19, 330 (1904) ; Bradley and Hale : Ibid., 19,387 (1904). It was not intended to publish these until the general inquiry should he more advanced. I t was learned, however, through the courtesy of Dr. Cottrell, of the University of California, that a similar work had been undertaken by him. Moreover, in the correspondence which followed, it transpired that we had been engaged upon the same specific problem during the last year, and that our results were practically identical. As these seemed of considerable interest, it was agreed to publish them a t once and simultaneously. We are glad to express a t this time our appreciation of the courtesy shown by Dr. Cottrell in connection with this matter. Paper read before the Society of German Engineers, 1895. Abstract in the Scie.ntzfic American Supplement, December P I , 1895.
’ Bradley
276
W. P. Byadley and G. P. 0.Fenwick
Now there are two general methods of lowering the temperature of the air above the expansion valve in a liquefier which is operated under otherwise uniform conditions, One of these may conveniently be called the external, and the other, the internal method. External precooling may occur either entirely outside of the liquefier, as is ordinarily the case, or, less commonly, during the passage of the high pressure air through the interchanger. The latter method has been adopted for instance by Olszewskil who divides his interchanger and inserts a carbon dioxide-ether bath between the halves. The internal method of precooling uses for the purpose some of the liquid air which has already been made and is lying in the reservoir of the liquefier. In connection with the first or external method, a careful analysis of the conditions will be helpful to a clearer understanding of its efficacy. As has been shown,’ the temperature at which air enters the top of the interchanger is an exceedingly potent factor in determining the yield of a liquefier. A little consideration however will show that external precooling can never have the full effect which is due from it. Of course the yield of liquid depends directly upon the temperature of the air just before expansion, and only indirectly upon that of the air as it enters the top of the interchanger. With uniform air supply, the fall of temperature which the air experiences in passing through the interchanger on its way to the expansion valve, depends upon the amount of expanded but unliquefied air which leaves the valve at the temperature of liquefaction, passes upward over the surface Bulletin de l’Acad6niie des Sciences de Cracovie, December, 1902. Zeitschrift fur komprimirte und flussige Gase, 7, I O (1903). Abstract in Chemisches Central-Blatt, 1903, I 543-4. Bradley and Rowe, p. 337, where it will be noted that “initial temperature ’’ refers uniformly to the temperature of the air which is supplied to the interchanger, and not to that of the same air after it has passed through the interchanger and has reached the expansion valve.
P~ecooliitgin the Liquefactioiz of A ~ Y
277
of the interchanger, withdrawing heat therefrom, and finally passes out a t the top of the liquefier. If the interchanger presents a sufficiently large surface, the temperature of this outflowing air becomes practically the same as that of the inflowing current. During equilibrium, the quantity of this outflowing current is constant, and is exactly equal to that of the inflowing air less that of the liquid air which accumulates in the reservoir and so is withdrawn from circulation. This being the case, the temperature of the high pressure air just before expansion is constant. So soon however as the cooling effect of the outflowing current is supplemented by precooling from any external source, the total heat withdrawn from the inflowing current is greater, and its temperature falls accordingly throughout the entire length of the interchanger. In so far, assuming for the sake of simplicity that the specific heat of air is constant for all temperatures, it is clear that the fall in the temperature of the high pressure air a t the top of the interchanger, which is due solely to the external precooling, would shortly be matched by an equal fall a t the bottom of the interchanger, just above the valve. But it is equally clear that the increased yield of liquid thus secured, leaves a smaller amount of unliquefied air a t the service of the interchanger. Thus a withdrawal of heat from the high pressure air outside of the liquefier diminishes to some extent the withdrawal of heat from the same air after it has entered the liquefier, and it comes t o pass as was stated above that the full value of external precooling i s not realized. To be sure, the loss incurred from this cause amounts to only a few percent of the total gain from precooling, and is not a very serious matter in its effect on yield. Neither is the further loss which arises from the fact that the outflowing current leaves the liquefier a t a lower temperature than before, namely, a t the new temperature of the air supply, and consequently carries away per gram less heat than before. The fact of such losses does serve, however, to emphasize the
278
W. P.Byadley and G. P. 0. Fenwick
very extraordinary character of the relation between the temperature of the air supply and the yield of liquid air which was brought out by our earlier experiments. If "initial temperature" in those experiments had been taken a t the valve instead of a t the top of the interchanger, the effect of precooling upon yield would have appeared still greater. It has been assumed in the foregoing discussion that the surface of the interchanger is sufficient to permit of complete thermal equilibrium between the inflowing and outflowing currents of air. Obviously, an interchanger which is ample when the temperature of the air supply is low, may not be when the temperature is higher. In such a case, precooling would mean, among other things, that a condition of imperfect thermal interchange would be replaced by one of perfect functioning in this respect, and so lead to an increase of yield far in excess of that due to the fall of temperaturejev se. Now the observed increase in yield in the experiments cited, from 0.74 liter a t 92' to 1.94 liters a t 30' is in fact partly accounted for in just this way. At 92' the outflow was 5' or 6' colder than the inflow, and of course the yield suffered in consequence. There was still a difference] though much smaller, a t 59'. Indeed the balance is never exact, in practice] a t any temperature. However, a t 30°, and a t temperatures still lower, the difference is so very small and so constant that it is quite negligible. Throwing out the readings above 30' therefore as being certainly influenced by the inadequacy of the interchanger, we note that precooling from 30' to 2' increased the yield from 1.94 t o 2.80 liters, or, reckoned on the basis of one horsepower per hour, from 2 7 4 to 396 cm3, a gain of 122 em3. Schrotter's formula calls for a gain of only j g em3! I n other words, the effect of external precooling is twice as great as it should be on the basis of the Joule-Thomson effect. Clearly there is urgent need of temperature measurements, taken a t the valve, before one can discuss, with satis-
'
PffecooZi?zgin ihe L i p e f a c i i o n of -A&
*
279
faction, such questions as these. We hope soon to be able to furnish them.’ Such extraordinary disparity between the change of initial temperature and the increase of yield which it caused led us at once to the query whether the liquid air itself, as produced by the liquefier, could not be so employed in precooling as to more than replace itself, led us, in short, to experiments on internal precooling. The most obvious device for the attainment of this object is the insertion into the high pressure circuit, between the bottom of the interchanger and the valve, of a suitable length of pipe so coiled as to be immersed in whatever liquid air may be lying in the reservoir. This device was employed both by Dr. Cottrell’ and by the writers, with this difference, that in Prof. Cottrell’s arrangement the extension, or auxiliary coil, was always entirely immersed in liquid air, and the comparison of yield secured with and without its aid was made possible by the use of two expansion valves, which were used alternately. One of these took air directly from the interchanger, in the usual way, without the use of the auxiliary coil, and so without internal precooling, while the other drew its supply of air from the interchanger indirectly through the auxiliary coil. In our experiments, on the other hand, but one valve was used. The appendix, as it may be called for brevity’s sake, wound in the shape of a cylindrical shell with vertical axis, was so disposed that its lower end was a little higher than the bottom of the liquid air reservoir. Thus, after each drawing of liquid air from the reservoir, fresh liquid could accumulate for a short period-a little less than four minutes, before actual immersion began. If drawings were made a t intervals of less than four minutes, therefore, immersion did not occur at all. If they were made at longer intervals, there was immersion of the appendix for a period which increased
* Assisted in part by a grant from the Hodgkins Fund of the Smithsonian Institution. Jour. Phys. Chrm., IO, 267 (1906).
280
W. P. Bradley and G. P. 0.Fenwzck
with the interval itself. Thus with a ten minute interval, which was about as large as was practicable with regard to the dimensions of the reservoir, there would be about four minutes of accumulation each time without immersion, followed by six minutes with more or less of it. At the end of this interval, about half of the appendix, approximately 140 cm of tubing, lay below the normal level which the liquid would have attained if quiescent, while the upper half was almost entirely covered by the froth of the rapid ebullition. Figure I illustrates this arrangement. AA is the lower end of the interchanger, which is enclosed in the casing GG. C is the appendix, enclosed in the reservoir FF, and communicating with the interchanger and with the valve D by the pipes BB and HH respectively. The liquid air is withdrawn through E. Obviously if there were any advantage in this method of precooling, it would be made manifest by a comparison of the yields obtained during accumulation periods of different length. At any rate, there could be no question that notable precooling was both possible and was actually secured by this arrangement. That there is ample room for such precooling was shown in our former paper,l for the temperature of the high pressure air inside the lowest disc of the interchanger cannot be lower than -102O, and beyond any question is considerably higher than that. At the very least, therefore, there is opportunity for from goo to 100' of precooling before the high pressure air ceases to be able to absorb heat from the outflowing current, Under these circumstances, the formula of Schrotter seems to promise that the amount of liquid consumed in precooling would be much more than made good by the resultant increase in production. The promise was not fulfilled. The rate of accumulation was practically constant, whether the appendix was immersed or not. The liquid air used for precooling was exactly replaced. Bradley and Hale, p. 391.
Pvecooling iz the Liquefaction of Aiv
281
Table I gives the yield of liquid air in two runs. In the first, drawings were made at intervals of three, five, six, and ten minutes; in the second, at intervals of four and ten minutes. The duration of each interval was counted from the beginning of one drawing to the beginning of the next. The number of seconds required for the actual discharge of the liquid was practically the same for all the drawings of the same accumulation period. No records, were made, either at the beginning of the runs or after a change in the accumulation period, until equilibrium was fully established, as shown by the concordance of yields obtained.
t
'C
Fig. I
The table also exhibits the efficiency of the liquefier for the same periods. This is obtained by dividing the weight of liquid air produced by that of the air supplied to the liquefier during the same time. The amount of the air supply was measured a t the close of the respective runs in the manner already described.
' Bradley and Rowe, p. 332.
8
P.Bradley and G. P. 0.FenwzcK
282
TABLE I Accuniulation period Minutes.
c
Liquid air C d
Liquid air liters per hour
Efficiency of liquefier Percent
2.54
8.75
2.49
8.58
2.55
8.78
6
IO
5
2IO 220 210 210
'
. .
3
I
165 I 60 I 60 I 65 I 60 165 I 60
2. 5 0
8.61
2.43
7.83
2.43
7.83
IO
The moment when the accumulating liquid in the reservoir begins to envelope the lowest turn of the appendix is
Pvecooling in the Lipztefaction of Aiv
283
marked by two closely connected phenomena. One of these is a very considerable and very sudden increase in the volume of the air which escapes from the top of the liquefier,' and, following this almost immediately, a fall of pressure as indicated on the high pressure gauge of the compressor. The sudden outrush of air is explained partly by the violent boiling of the liquid air when it comes into contact with the comparatively warrn coil. At the same time, however, the high pressure air within the pipe becomes denser through loss of heat, and flows more rapidly through the valve. This makes a heavier demand upon the reserve air in the purifiers, and the pressure of the latter runs down accordingly, unless the expansion valve is at once suitably adjusted. These phenomena always occur at the same moment in the accumulation period, if the liquefier is running steadily, and the new equilibrium which is secured by the adjustment of the valve is not again suddenly disturbed during the period, until the liquid is next withdrawn. The moment, however, when the level of the liquid during withdrawal falls again below the end of the appendix, a correspondingly sudden decrease in the volume of the exhaust occurs, which has to be checked at once by opening the expansion valve to its former position. No one who has had the responsibility of adjusting the valve to these sudden changes in the flow of the high pressure air through it, is left in any doubt as to the thoroughness of the precooling which is secured by the immersion of the appendix. Exactly what the initial temperature thus attained is, it is of course impossible to say, in the absence of direct measureA s the exhaust necessarily meets a certain amount of resistance in the pipe which conducts it back t o the compressor, its pressure is slightly above atmospheric. The resistance is constant if the amount of the air is constant. I n the usual manner, an arrangement was provided whereby the slight pressure necessary to overcome this resistance was measured by a column of mercury. The latter serves thus as a very sensitive indicator of the amount of air which is returning to the compressor and is incidentally a useful guide for adjusting the expansion valve,
W. P.BYadZey aizd G. P. 0. Fenwick
284
ments. That it was as low at any rate as the critical temperature of air, seems to us extremely likely. The interest which attaches to this point is increased by the recent insistence of Pictet,' that high pressure air once cooled to, or below, its critical temperature-which, in other words, has no longer produce any been changed into a liquid-can cooling effect on passing through an expansion valve. Pictet's view will be made clear by the following quotationU2 '' Indem wir die fliissige Luft sich in dem Reservoir ansammeln liessen, welches eine lange Schlange [the interchanger] abschliesst, die bestandig die comprimirte und abgekiihlte Luft zufiihrte, kamen wir dazu, die letzten Windungen mit dieser angesammelten flussigen Luft zu netzen. Wir haben dabei const.atirt, was wir iibrigens vorher gesehen hatten, das namlich die ganze Luft vor der Entspannung sich verfliissigte, der Druck von 80 Atm. auf 60 fiel, dann auf 50, dann auf 40 etc. und die erhaltene fliissige Luft die verdampfte vollig ersetzte, indem sie nur eine schwache Vermehrung zuliess zufolge der Kompressionsarbeit und der Arbeit bei der Entspannung der erhaltenen fliissigen Luft, die von einem hoheren auf atmospherischen Druck iibergeht. Diese Maschine hat in dieser Weise drei "age ohne Unterbrechung funktionirt und lief erte dadurch einen biindigen Beweis fur dieses klassische Theorem, das uns hier beschaftigt, dass namlich die bei der Entspannung eines Gases nach der Beendigung des molecularen Sturmes erhaltene Abkiihlung nur von der ausseren Arbeit der gasformigen, vor der Entspannung abgekiihlten, aber nicht verfliissigten Gasmassen herr iihrt ." In Pictet 's liquefier accumulation practically ceased, for three days of continuous running, from the moment the liquid in the reservoir came into contact with the lower end of the interchanger. In the experiments of Dr. Cottrell and in those Zeitschrift fur komprimirte und flfissige Gase, 7, 1903-4,22, 23, 37, 5 2 ; 8, 1904-51 13.
LOC.cit., 8, 13.
P~ecoolingin the Liquefaction of Air
285
of the writers, accumulation went on at precisely the same rate as before, from what is constructively the same moment. There is one apparent way of explaining this flat contradiction. It will be noticed that Pictet makes no mention of any expansion valve a t the base of his interchanger. In fact, his statement regarding the fall of pressure which occurred a t the moment of immersion seems to indicate that his interchanger ended in an orifice of fixed size. This supposition gains plausibility from the fact that precisely this arrangement was employed in other experiments which are described in the articles referred to. If this is correct, accumulation ceased, not because of precooling, but because the pressure ran down-how far down, he does not say! And the pressure ran down because there was no way of stopping it. To be sure, Pictet expresses just as decisively the further dictum1 that the yield of a liquefier whose interchanger has surface enough for perfect functioning, is independent of the pressure, provided the latter is not less than 25-30 atmospheres-or in other words that the only reason why in practice increase of pressure increases the yield is that the interchanger is not large enough for a complete thermal balance. But we have further shown2 how profoundly the yield is affected by change of pressure, and that too in a liquefier whose imperfection in the matter of thermal interchange never exceeds 2' for pressures between 1500 and 3000 pounds except indeed where the temperature of the air supply is higher than about 30°, as has already been said. That variations in the functioning of the interchanger, which are represented by less than 2' of difference between the temperature of the inflowing and outflowing currents can account for the fact that a yield of 1.6 liters at 1500 pounds rose to 2.84 liters at 2665 pounds, in other words, that the percentage of liquefaction rose from 4.5 to 8.6, is quite out of the question. Even when the appendix is not actually immersed in liquid air, it is exposed more or less directly to the drip of Loc. cit., 7, 40. Bradley and Rowe,p. 339.
286
U.: I? Byadley nizd G. P. 0.Feltwick
liquid from the expansion valve, and to the fog of liquid droplets which lie above the accumulating liquid. Consequently there is always a certain amount of internal precooling which does not occur when the appendix is absent. I n order to see whether this measure of precooling, which could certainly never be accused of extending to the critical temperature of air, has any effect on the yield, runs were made in which the appendix was alternately present and absent. As the amount of air delivered by the compressor is never quite the same in any two runs, however uniform it may remain during the continuance of either one of them, and as the efficiency of the liquefier varies somewhat with the amount of the air supply, as well as because of certain other' circumstances which cannot always be clearly traced, it was desirable to make a considerable number of these runs in order that the comparison should be as decisive as possible. Out of a dozen or more, six are here presented, each of which seems to be entirely normal. In connection with each of the others, some circumstance was noted, during the run itself, whose influence threatened. to obscure the main issue involved. In two, for instance, the liquid air ran cloudy, indicating that the' purifiers were not functioning properly. The trouble was discovered in the KOH cylinder which needed recharging., In another, a small leak occurred in the inner casing of the liquefier, near the lower end of the interchanger. Though it was not serious, the effect, if any, was to diminish the yield of liquid in that particular run. In one of the runs, a coupling in the high pressure circuit developed a small leak, and in still two others near the close of the runs, a readjustment of conditions in the cylinders of the compressor was revealed by the readings of the three lower gauges. None of these circumstances probably had any considerable effect on the final result, but they tended to some extent at least to , mask the issue. Table I1 contains the six selected runs. Incidentally one may observe in these how the efficiency of a liquefier
Pvecoolilzg in the Lipuefaction of Azv
With the appendix No,
I
4 5
Averages :
287
Without the appendix
Yield per hour
Efficiency Percent
No‘
2.50 2.43 2.53
8.61 7.83 8 63
3 6
-
-
2.49
8.36
2
I ~
Yield per hour
Efficiency Percent
2.63 2.56 2.57
8.83 8. I O 8.53
-_
2.59
8.49
288
M P.Byadley and G. P. 0. Fe?zwick
spiral with vertical axis. Into' the outside surface of the turns of the spiral, were drilled minute holes, twenty in number, for the escape of the specifically denser liquid air, and on the inside surface, an equal number of holes having a diameter of 3-3.5 mm, for the escape of the unliquefied air. The outer end of the tube was closed. It will be understood of course that the spiral itself was motionless, and that the centrifugal effect was secured by the spiral motion of the air through it. TABLE I11
l
Naked valve Yield per hour
I
Efficiency
I
2.77 2.55,
-I
&Averages:2.66 1
1
8.48 8.31 .
8.40
Separator
1 Yi2:uEer 1 1 I
1I
2.84 2.63
Efficiency
1
-2.74 ~
8.79 8.65
G
Pvecoo/i~zgin the Liquefaction of Air
289
view that production of cold a t the expansion valve is due to the Joule-Thomson effect. It is for the present at least equally difficult to see how the respective results can be reconciled t o each other. 5 . When the blast from the valve is conducted through a separator, there seems to be an increase in efficiency amounting to 4 percent of that obtained with the naked valve. Wesleyan C'nzzevsLty, .Lftddletown, Conn., January 2 , 1906
