will be considered a s a single phase in the following discussion.

The system is in a single phase a t temperatures and pressures correspondnig to F. A drop in pressure on the system from F to C, will cause a tiny bubble of less- dense phase to appear in the system, ant1 point G is called the "bubble point." A fur ther drop in pressure from G to H will cause vaporization of constituents from the more-dense (or liquid) phase and the enlarge- ment of the less-dense (or vapor) phase. At any tem- perature between A and C, a drop in pressure on the system in the single-phase area above curve A-B-C will cause the formation of the tiny bubble of vapor a s the pressures represented by the curve A-B-C are reached. Accortlingly, this curve is called the "bubble-point curvt:." Other ways of esprcssinp this phenolnenon h a ~ , c beer] to consider the liquid a s saturated a t the pressures and temperatures on line A-B-C, because a slight reduc- tion in pressure will allow the formati011 of a 17apor phase. As the vapor pressure of thc entire systenl as a liquid must equal the total pressure a t tht. bubblc or boiling point, the curve A-B-(: has bee11 called thc "vapor-pressure curve." The us(: of several names to represent the same phenomenon results in confusion, but either of the terms "bubblc point" or "saturation pressure" a t a given ten~psra ture may be paltlcularly descriptive of the phenomchnon. The recommt,nded no- mellclature is to refer to the. system a s a saturated liquid 011 cul A-B-C.

The systeni a t the tempt:rature and pressure of J is ln a single phase. Pressure increase or] the systeni from J to K will cause a small quantity of more-dense, or liquid, phase to condensc in the form of tiny drop- lets. The resemblance of the droplets to dew gave the name of "dew point" to the pressure and temperature represented by K. A further increase in pressure from K to L \rill cause more liquid to be formed, but a t temperatures between C arid D pressure increases be- yond some point L will cause the disappearance of the more-dense, or liquid, phase a t point M. This vaporiza- tion upon pressure increase is opposed to the behavior from (; to H, and so is called "retrograde vaporization." Similarly, the drop in prcJssure on thts system in a single phase a t N to conditions a t M will cause the formation of a small quantity of more-dense, or liquid, phase; and this phenomenon is called "retrograde con- densation" because the change in pressure causing the condellsation is opposite to that usually encountered fur condensation such a s from J to I(. Tho systenl is erl- tirely In the vapor state a t K and a t M, and so both poinrs a re dew points. Curve E-K-D-M-C represents the dew points for thc systc.m or. the saturated-vapor curve.

The measurement of the bubble point for increasing temperatures along curve A-G-B will show a unique pheriomenor~ a t (;. the critical temperatnrt,. Instead of the bubble disappearing by gradual diminution, the meniscus between the vapor and liquid phases disap- pears--creating a single phase. The single-phase forma- tion a t a critical temperature has been observed by Roess * a s a color change. A plot of the density of the system a s vapor a t i ts dew point and liquid a t i ts

bubble point a s a function of temperature u - i l l silow that the saturated liquid density a t t en~pera tu~ .cs just below C approaches the density of the saturated valinr a t temperatures just above C. The temperature and pressure where the dew-point anti bubble-point cu~ .v~ . - meet is called the "critical ten~perature and 1)ressurt.~' of the system.

The system divides into liquid and vapor phases within the bubble-point and dew-point curve, a s indi- cated by the percentage-liquid lines. In describing the system in the single-phase area outside the liquid-vapor region, the term "liquid" o r "vapor" has been used to represent the single phase when it was similar to n liquid or vapor within the two-phase area. Thus, a t con- ditions of F and G, the system is similar; and becausc a t G the system is in the liquid state, the tern1 "liquid" has been applied to the system in the single phase a t F. Likewise the system a t J is similar to the system in the vapor s tate a t K, and has been called "va1)or" or "super-heated vapor" a t J. The similarity between the vapor and liquid in the two-phase area as the critical temperature and pressure a r e approached malies t h t ~ extension of this practice to the single-phase area around the critical temperature and pressure vtxry confusing.

The behavior of the average sub-surface sample of crude oil is represented somewhere in the area of Fig. 1 enclosed by a temperature near C, and a pressure higher than that corresponding to about 30 per cent liquid, If the sample is removed a s a liquid from a reservoir con- taining two phases, i t likely will have its bubble-point pressure a t the reservoir temperature correspond to tho lowest pressure existing on the liquid prior to sampling. As this pressure is often considerably above 1,000 lb., a considerable portion of the natural-gas constituents is present in the sample. A lowering of the pressure on the sample, such a s from G to H in Fig. 1, accompanied by sufficient agitation or time to bring equilibrium con ditions, %-ill result in a liquid a t i ts bubble point and :I vapor a t its dew point a t conditions of H. The pro1)- erties of the liquid a t H will be different from tho properties of the liquid a t G because of the changes 11: composition due to the vaporization. The properties of high-bubble-point-pressure-sub-surface samples, o r so- lutions of large quantities of natural gas in crude oil, have been shown' to depend on composition just a s other hydrocarbon oils depend on composition. Thc properties of the natural-gas constituents art. con siderably different from the high-boiling colistituents. and, therefore, the inclusion of large proportions of these volatile constituents in the liquid phase will lo\\-ci. the density, viscosity, and surface tension of the ligui(l considerably below the corresponding property for th., high-boiling portion.

The attainment of equilibrium conditions betm-eer: liquid and vapor phases requires intimate contact o!, long periods of time."' A drop in pressure below t h r bubble point of a liquid under quiescent conditions ma) not cause vaporization to occur immediately. However, the rising of vapor bubbles through the liquid, once vaporization starts, provides considerable agitation, and

brings the vapor and liquid towards equilibrium. A rise in pressure on a n equilibrium liquid and vapor phase in the sub-surface-sample area of Fig. 1 tends to trans- fe r vapor into the liquid phase. Under quiescent condi- tions the surface of the liquid would interchange con- stituents with the vapor, and possibly reach equilibrium in a few hours. However, the diffusion of the liquid a t the surface into the main body of the liquid is so slow tha t years of time will not always assure equilibrium between all the liquid and all the vapor. This behavior shows the necessity of agitation of vapor and liquid phases to reach equilibrium, especially f o r the cases of t ransferr ing vapor constituents into the liquid phase.

The phase diagram represents conditions fo r the entire system in equilibrium. I t shows the quantities of the two phases present fo r the case of equilibrium or flash vaporization illustrated in Fig. 2. Flash vapori-

-* VAPOR

Flash Vaporization. FIG. 2

zation brings, a t a constant temperature and pressure, all the liquid into equilibrium with all the vapor either in a batch process o r in a continual-flow process.

If the pressure is reduced on a system a t i ts bubble point, vapor is formed. Continuous removal of the vapor a s soon a s i t has formed, with continuous lower- ing of the pressure, is "differential vaporization." This process, illustrated in Fig. 3, occurs fo r a gradually- changing system because the vapor is being removed, and no single-phase diagram could give the phase relations fo r a series of systems. Vaporization of a constituent from a liquid causes the liquid to shrink, and a t rue differential-vaporization process requires a reduction in volume such a s the introduction of a second inert liquid if the vapor space is to remain very small.

Vaporizations of sub-surface samples give off natural gas a s the vapors, and so the nomenclature used is with this fact in mind. The quantity vaporized or g a s liber- ated is measured in terms of gas coming out of solution. The liquid is considered to shrink from the gas evolu- tion rather than having a volume fraction vaporized.

Sampling Procedure The choice of a well will depend on the nature of the

problem being studied, but usually includes considera- tion of previous production history, clean tubing set close t o the producing formation, and surface equip- ment in condition and arrangement suitable fo r the proposed auxiliary measurements.

Sub-surface samples a r e intended normally to ap- proximate the reservoir liquid phase if two phases a r e present in the reservoir and the single phase, if a single phase is present. A liquid in equilibrium with a gas phase in the reservoir might be a t some point on the phase diagram for t h e contents of the reservoir, such a s point H of Fig. 1; but the separation of the liquid from the gas places the liquid on the bubble- point curve of i ts phase diagram. Accordingly, the pres- sure drop which causes the flow from the reservoir to

7 VAPOR

- MERCURY Differential Vaporization.

FIG. 3

the well bore "* '"ill cause a corresponding vaporiza- tion of the liquid, reducing the bubble point of the liquid to the flowing pressure. A single-phase reservoir fluid considerably removed from the two-phase area will have a corresponding degree of freedom in reduction of the bottom-hole pressure without separation of the second phase.

The flowing period prior to sampling should be con- trolled to have a maximum bottom-hole pressure, and corresponding maximum saturation pressure of the liquid phase, without having by-passing of gas and liquid in the flow string. A pressure measurement, prior to sampling, should be made a t the sampling depth a t a steady flow state. Also i t may be advisable to run a bailer to the sampling depth to check tha t wash or formation water is not present a t the contemplated sampling depth. Following these precautions, a well may be sampled a t a slow flow rate, or may be shut in just prior to sampling. Sampling during flowing con- ditions, i n a well having both gas and liquid phases a t the sampling point, may include all o r par t of the gas phase-depending on the dispersion of the two phases.

Static samples a re considered to be liquid a t the point of sampling. The prerequisites for sub-surface samplers a re means for introduction of representative well fluid into the sampler and positive valve mechanisms which enclose the sample, bringing the fluid from the sampling point to the surface without loss. Thermal shrinkage of the liquid in the sampler, a s it leaves the bottom of the well and comes to the surface, will create a vapor phase in the sampler.

The auxiliary data taken with sub-surface samples should be a s complete a s possible if fullest utilization of sub-surface sample data is to be made. Flow tests, well history, and especially gas-oil ratio measurements arc. considered important.

Transferring of Samplen The transferring of samples requires tha t the entire

sample be in a single phase if the sample is to be divided. The vapors formed during the thermal shrink- age must be returned to the liquid phase by pressure in- crease above the bubble-point pressure a t the tempera- ture of the sample. This formation of a single phase usually is accomplished by ~~iercui-y injection and agita- tion to assure equilibrium. Samples may be transported to the laboratory in the sampler or in pressure con- tainers. Transfer from the sampler to containers must be conducted a t pressures sufficient to allow a pressure reduction during flow without reaching the bubble point. Leaks while transporting a re serious; for an aliquot proportion of the two phases is not likely to escape, and thus the composition in the container is changed.

Analyses of Samples The analysis of the sub)-surface samples by loa-

tenlperature fractionation should be included a s one of the tests on the sample; for the analysis and proper equilibrium data l' " will permit the computation of the behavior of the sample under a n y series of vaporiza- tions. The apparatus required is a double-stage distil- lation unit to remove the volatile constituents from the crude, preferably two or three a t a time, and subse- quent separation into pure compounds.

The analytical procedure is similar to tha t used for gas analyses, but especial precautions must be made that the volatile constituents a r e removed from the crude oil, and absolute quantities of the pure consti- tuents must be measured. A sample in the range of 50 to 200 ml. should be analyzed for nitrogen, methane, ethane, propane, isobutane, normal butane, pentanes, and hexanes-with determination of volume, density, and molecular weight '' of the heptanes and heavier fraction. The distillation of hexanes through nonanes on a high-temperature distillation column, with gravities and molecular weights of the cuts, has been found a n improvement over the usage of heptanes plus a s a single fraction. Analyses should be reported a s both mol and weight fractions, or percentages, with inclusion

of the properties of constituents of higher boiling point than pentane or hexane.

The initial vaporization or saturation pressure of a sample a re generally desired. A portion of the sample is compressed to a single phase a t some temperature such a s the reservoir temperature. The volume on this single phase is increased gradually, with measurement of pressure decreases. A s the saturation pressure is reached, the pressure decreases a t a slower rate for given increments of volume increase. A plot of pressure decrease vs. volume increase will show the bubble-point pressure a t the intersection of the two curves. Agitation in some form is necessary for accurate bubble-point determinations. The saturation pressure normally is reported a s pounds per square inch absolute, and must include the temperature of the measurement.

The pressure decrease with increases in volume of the system provides compressibility data on the sample, provided the apparatus containing the sample is calibrated for volume changes with pressure. Reporting of data may be directly a s volume changes with pres- sure based on the volume of the sample a t the bubble- point pressure and temperature. A uniform method of reporting compressibility data is per cent decrease in volume of the sample for 100-1b.-per-sq.-in. increase in pressure based on the volume a t the bubble point, expressed a s applying a t a temperature and over a given pressure interval.

Following the bubble-point measurement, the sample may be vaporized either a s a flash vaporization or by differential vaporization. Two procedures have been used for each method.

The differential vaporization adopted by Lindsy ' deviated from a truly differential process, because the constant volume employed caused a large vapor space to be formed due to the shrinkage of the liquid. This constant-volume differential vaporization may be closer to the t rue conditions in a constant-volume reservoir than a t rue differential vaporization. The customary reporting of results1, h i v e s pressure vs. solubility of gas measured a s cubic feet a t 60 deg. F. and 14.7 lb. per sq. in., absolute, fo r a barrel of the crude oil which remains a t the end of the vaporization process (residual crude oil) but measured a t 60 deg. F. This method presents difficulty if the equilibrium pressure a t the end of the vaporization is not exactly 14.7 lb. per sq. in., absolute, a s the total quantity of vapors is sensitive to small differences in final pressures. However, the difference in solubility between two pressures con- siderably above atmospheric is not dependent on this final pressure, and should check between laboratories even if the total quantity does not. Solubility curves drawn from the measurements a t pressure intervals approach straight lines above 300 to 500 lb. per sq. in., and may be extrapolated short distances, such a s from sub-surface-samplesaturation pressure up to the static reservoir pressures.

The shrinkage or liquid volume fraction vaporized is reported a s per cent of the volume of residual crude oil which was vaporized, and includes the thermal shrinkage of the residual crude oil from the test tem- perature down to 60 deg. F. The reporting of shrinkage is subject to the same difficulties as solubility. Shrink- age from 40 per cent to 30 per cent for a pressure drop from A to B lb. means that the volume of liquid a t pressure A was 1.40 parts; a t pressure B i t was 1.30 parts; and a t the end of the vaporization, with residual crude oil cooled to 60 deg. F., it would be 1.00 part. This use of residual crude oil as a basis appears awk- ward, but allows correlation of data for similar sub- surface samples having different bubble-point pressures.

A variation of the "constant-volume" differential vaporization is used, and will be designated a s the "approximate differential" vaporization." This method eliminates the vapor space developed in the chamber due to shrinkage a t the several pressure decrements used. Mercury injection a t the end of each pressure decrement brings the contents of the vaporization cham- ber to a single phase. The bubble-point pressure is de- termined for the system in a manner similar to the initial value. This method would be a true differential vaporization if very small pressure decrements were used. I t eliminates the computation of the quantity of vapor in the vapor space, which must be added to the evolved gas when computing the solubility data for the constant-volume differential vaporization. There appear to be no more differences between the approxi- mate-differential and the constant-volume-vaporization results than between results of different laboratories and equipment using the constant-volume-differential method. However, future work conducted accurately a t low pressures may require a distinction between the two methods. Flash vapom'zations always tend to bring all the gas

and oil into equilibrium a t a temperature and pressure, but different methods may be employed or measurements made.

The measurement of a unit volume of sub-surface sample a t its bubble-point pressure a t a given tempera- ture, with subsequent measurements of the volume of the two-phase system a t several pressure decrements down to atmosphere, give data for a pressure vs. volume (PV) curve" a t a specified temperature. The relative volumes reported are based on the bubble-point volume.

The reduction in pressure to some lower pressure, often atmospheric, on a given quantity of sub-surface sample a s i t passes through a gas-liquid separator ap- proaches the conditions in the field separators. The quantity of vapor formed is reported a s cubic feet a t 60 deg. F. and 14.7 lb. per sq. in., absolute, per barrel of oil in the liquid phase, but subsequently brought to 60 deg. F. The shrinkage from the volume of the sub- surface sample to the volume of the residual crude oil a t 60 deg. F. may be expressed a s per cent of the volume of the residual oil a s in differential vaporiza- tions, but there appear to be advantages in this case of reporting per cent of the bubble-point liquid volume which was lost by vaporization.

A third variety of tests on the sub-surface sample is to measure the flash vaporizations a t a constant pres- sure on a series of liquids having bubble-point pressures from the sub-surface-sample value down to low pres- sures-the reduction in the bubble-point pressure being accomplished by the approximate differential-vaporiza- tion process.

In all vaporization measurements the residual crude- oil gravity, the gas gravity, and possibly the gas analy- sis should be determined.

Interpretation of Results

The choice of measurements on sub-surface samples is guided by the uses contemplated for the data. Sub- surface-sample data must be accompanied by auxiliary information in order to be utilized to the greatest ad- vantage. Plans are made for having separate discus- sions on the application of sub-surface-sample informa- tion in several directions, but a few examples will be enumerated.

The greatest contribution which sub-surface-sample work has made to the production industry is the devel- opment of our concepts of the behavior of oil and gas in the reservoir. Not until sub-surface samples were obtained did we have a clear idea of the phase relations in the sand-which is necessary in order to make intel- ligent decisions on many daily production problems.

Sub-surface-sample data make possible an alternate method of estimating the oil in place in a constant- volume pool, without knowledge of sand extent.". '' A complete production history of the pool with sub-surface pressures is the auxiliary information needed.

Fractional analyses of sub-surface samples and equi- librium constants are used to predict surface separa- tions of oil and gas a s to gas-oil ratio and natural- gasoline content 'of the gas. Stage separation and natural-gasoline reserves now are understood more completely.

Better identification of horizons is permitted by the combined use of equilibrium data and sub-surface- sample analyses. Because of the high concentration of the more volatile constituents in high-bubble-point pressure samples, an added criterion for identifying a reservoir liquid is available. Sub-surface sample analy- ses and equilibrium data assist in identifying gas zones a s being the equilibrium phase for a reservoir oil or not of the proper composition to be the equilibrium phase.

BIBLIOGRAPHY ' B. E. Lindsy, U . 8. Bur. Mines Rpt . Investigation 321s (1933). 'B. E. Lindsy, Petroleum Engr. 7, 34 (1936). 'D. L. Katr and K. H. Hachmuth. Ind. Eng. Chem. 29. 1072

11QR7b , - - . , . 'B. H. Sage, .J. G. Schaafsma, and W. N. Lacey, Ind. Eng.

Chem. 26, 214 (1934). ' W. B. Kay Ind. Eng. Chem 30 459 (1938). OS. von P i l i t Oil Gas J 35 [10j 54 (1936) ' D. L. ~ a t z , ' Am. ~ n s t . ' Mining Met. ~ n g r s . Tech. Pub. 971,

Oct. (1938). Roes8 J . Inst. Petroleum Tech. 22 665 (1936).

* W ~ ' ~ a c e y Proc API 13 [IV 16 1932). lo K: D: ~orneioy. +. N. Lacey d. E: Scudder, and F. P. Stapp.

Ind. En$: Chem. 25. 1014 (193d). " S. Buckley, Trans. Am. Inst. Mining Met. Engrs. 127,

178 (1937).

"1). I.. K:tte, Trans. Am. Inst. Mining Met. Enyrs. 127, Is!) i ln9r i $A

With regard to the terminations o r "footings" of these several quality lines on either axis of the diagram: It is my understanding of these relations tha t if the point 0 represents absolute zero of pressure and tem- perature, then, physically, all of these lines must foot on the temperature axis rather than on the pressure axis, or a t the origin 0. Fig. 1 (Burnet t ) shows them extending to footings a t temperatures ranging from a lower value R (never zero) for 100 per cent liquid, to a higher value S for 100 per cent vapor, with no mutual intersections except a t the point C.

Actual locations of these curves, of course, depend upon the particular system investigated. F o r some sys- tems the interval R to S may be relatively small; and possibly for some R and S conceivably might coincide, in which case the quality lines would all meet a t t h a t point on the temperature axis a s well a s a t point C. Should a solid phase develop, modifications of Fig. 1 (Burnett) would be required to depict the behavior a t low temperature.

Support for the arrangement indicated in Fig. 1 (Burnet t ) is afforded by the concept of a lower tem- perature limit of vaporization a t zero absolute pressure, the so-called "Nullpunkt der Verdampfung" discussed by Dr. C. von Rechenberg in his book, Einfache und Fractionerte Destillation, 337 to 344, Schimmel and Company, Leipzig (1923).

Referring to Fig. 1 (Burnet t ) , isothermal pressure reduction on the liquid phase from a condition repre- sented by the point F' would not result in evolution of a vapor phase, even if the liquid were exposed to a vacuum. Confirmation of this effect is the use of certain liquids recently developed for "vacuum" monometers.

Isothermal pressure reduction on the liquid phase

from a condition represented by a point F' would result in partial vaporization and a n ultimate non-volatile residue, a situation indicated by the results of isothermal "vacuum" distillation of many hydrocarbon mixtures.

Mr. Katz: With regard to the shape of the quality lines on the pressure-temperature diagram, I do not believe they necessarily should be symmetrical a s indi- cated in Fig. 1 (Burnet t ) . Firs t , there can be no analogy between a pure substance and a complex mixture on quality lines of a pressure-temperature diagram; for pure substances have a single (vapor-pressure) curve when these coordinates a re used. Second, the computed phase diagram for the methane-propane system indi- cated a complete lack of symmetry for the quality lines based on mol per cent. As no experimental data have been reported on the phase behavior of a complex mix- ture adjacent to the critical conditions, I do not wish to imply the lines of Fig. 1 are correct, but I also do not believe they can be shown to be incorrect-especially when considering tha t the per-cent-liquid lines might be either weight, mol, or volume percentages.

With regard to the behavior of the system a t low pres- sures and temperatures (Fig. 1 ) the scales were not in- tended to be absolute scales because of lack of informa- tion a t low pressures and because of the appearance of a solid phase a t low temperatures. The appearance of a solid phase would eliminate any significance of the con- ditions a t F in Fig. 1 (Burnet t ) , a s any mixture con- taining methane surely would contain solids if the tem- perature were so low tha t methane would not vaporize a t low pressures. Fig. 1 was intended to cover only the range of pressures and temperatures a t which a single- or two-fluid (gas and liquid) phase was present.