Joule-Thomson Process in the Liquefaction of Helium

Many different devices and cycle arrangements can be used in part.ially cooling gas for helium liquefaction . With the exception of the Simon freeexpansion type liquefier , howeverwhich is suitable only for small-yield, batch productionall uch arrangements have a final stage at the lowest temperatures, in which a so-called Joule-Thomson proces takes place. This process may be considered to include cooling of the high-pres ure gas by the unliq uefied re turn gas in a "J 0 ule-Thom son hea t exchanger", as well as cooling of the high-pre ure ga by the Joule-Thomson effect, Lhat is, i enthalpic expan ion through a "Joule-Thom son valve." If the initial state of the high-pressure gas at the top of the J-T hea t exchanger (which is most conveniently defined by its pre m e and temperature) is fixed within a certain region, a definiLe ma s fraction a of the gas passing through Lhe J-T valve will be liquefied. To determine the yield a of a liquefLCr, the specific enthalpies at cOl'Lain points in the J-T proce must be known as functions of P and T. The mo t readily available data on the specific enthalpy, h , of hclium is from the monograph of Keesom [1],2 or Lhe temperature-entropy diagram of Kec om published by the International Institute of RefriveraLion [2] . The errors and inconsistencies of these sources below about 20° K , however , have been pointed out in a series of papers by Zelmanov [3, 4,5,6] . In addition, Zelmanov [6] has recalculated and redrawn both T-s and hP diagram s which, as indicated later, appear to be of good accuracy and consistency, The Zelmanov Ts diagram is reproduced by Zemansky [7]; and the hP diagram , by Collins [8]; who, however , also reproduces the Kcesom T-s diagram . Because of the crowded naLure of the isobars in the region of interest for the J-T process on a T-s diagram , the hP diagram is more convenient for liquefier calculations. At the beginning of this study there was no easy way to judge the relative value of various conflicting set of data. I n order to determ.ine, at least, the consistency of such data, a new dim ensionless , normalized correlating function we?, T ) was introduced. With thi function , the consistency of Zelmanov's work was evident, and both the lack of self-consistency of the Ts diagram s of Keesom , and