47

48

Samples of natural gas supplied to larger cities in the United States have been collected and analyzed, and tests have been conducted with reference to the compressibility of natural gas, the conditions under which it exists underground,49 and the variation in composition of natural gas from different sands in the same fields.50

CONSTITUENTS OF NATURAL GAS

Accurate determination of the constituents of natural gas has proved a stumbling block to gas analysts unfamiliar with the work. Technical forms of gas-analysis apparatus and established rules for bringing a gas mixture in contact with the absorbents for different constituents are not always effective. Many samples contain absorbable constituents, such as carbon dioxide and oxygen, in extremely small quantities. The fact that oxygen may be a constant constituent of natural gas as it leaves a well has not been determined absolutely. The authors believe that the traces of oxygen reported in some samples were due to contamination of the samples with air. Phillips 51 detected only minute quantities of oxygen in natural gas from western Pennsylvania after the gas had bubbled continuously for many hours or days through reagents.

Olefin hydrocarbons and carbon monoxide have not been identified in the samples already received at the bureau's Pittsburgh laboratory, if the natural gas had not been mixed with artificial gas. Because higher members of the paraffin series are absorbed by fuming sulphuric acid and cuprous chloride, a natural gas that does not contain olefin hydrocarbons or carbon monoxide but does contain these higher members of the paraffin series, when treated with these solutions will undergo a reduction in volume and lead the analyst to a wrong conclusion.

Natural gas usually contains a large proportion of paraffin hydrocarbons, some samples as much as 99 per cent, so that if the paraffins are determined by explosion methods in which a small quantity (8 or 10 c. c.) of gas is used a slight error of manipulation will be

47 Burrell, G. A., and Oberfell, G. G., Composition of natural gas used in 25 cities, with a discussion of the properties of natural gas: Tech. Paper 109, Bureau of Mines, 1915, 22 pp.

48 Burrell, G. A., and Robertson, I. W., Compressibility of natural gas at high pressures: Tech. Paper 131, Bureau of Mines, 1916, 11 pp. Burrell, G. A., and Robertson, I. W., Compressibility of the constituents of natural gas, with composition of the natural gas used in 31 cities in the United States: Tech. Paper 158, Bureau of Mines, 1917, 16 pp. 49 Burrell, G. A., Condition of natural gas in the earth's strata: Jour. Ind. Eng. Chem., vol. 7, 1915, p. 322.

50 Burrell, G. A., and Oberfell, G. G., Variation in composition of natural gas from different sands in the same field: Jour. Ind. Eng. Chem., vol. 7, 1915, p. 419. Burrell, G. A., and Seibert, F. M., Separation of the constituents in natural gas from which natural gas is condensed: Jour. Am. Chem. Soc., vol. 37, 1915, p. 392.

51 Phillips, F. C., Composition of natural gas; researches upon the phenomena and chemical properties of gases: Am. Chem. Jour., vol. 16, 1894, p. 411.

multiplied 10 or 12 times in calculating results to a percentage basis. Methane is the only hydrocarbon that some natural gases contain; but if a small quantity of sample is taken for the combustion analysis, errors are magnified. The relation between the volume of carbon dioxide and the contraction produced by the combustion may indicate hydrogen, although this gas is absent. Although the error in the observed data may be small, by calculation to a percentage basis it may amount to several per cent of hydrogen. Many published analyses of natural gas are undoubtedly much in error from such causes.

The paraffin hydrocarbons that are gaseous at ordinary temperatures are methane, ethane, propane, and butane. Since the last two are liquefied at ordinary temperatures by pressures below those existing in most producing wells, quantities other than small proportions carried by the permanent gases would hardly be found in natural gases coming from wells under very high pressure. On the other hand, gases drawn from wells under a partial vacuum may contain large amounts of these and higher paraffins.

METHOD OF SAMPLING

Some samples of gas taken by the bureau's investigators are collected in magnesium citrate bottles by air displacement. At the wells or supply pipes a stopper is removed and gas allowed to run into the bottle until air has been entirely displaced. Five or more minutes are usually enough. The stopper is inserted while the bottle is still in place. As a precaution against leakage, melted paraffin is poured over the stopper. The bottles are then carefully packed in suitable wooden boxes and sent to the bureau's Pittsburgh laboratory. There the stoppers are removed under mercury, and the gas is transferred to and analyzed in an apparatus that contains mercury as the confining fluid. Some samples taken at the full pressure of the well have been received in strong iron cylinders. Other samples have been collected by a method devised by G. A. Hulett, former chief chemist of the Bureau of Mines, which is now used for collecting some of the bureau's samples of natural gas. It is as follows:

Use a glass tube about 17 cm. long and 6 cm. wide which has a 15-cm. glass-tube extension at one end with a 6-mm. bore. Leave the tube open. Direct natural gas into the sampling tube through the orifice by means of a long slender brass tube. When the air originally in the tube has been displaced, withdraw the brass tube and close the glass-tube orifice by placing the finger over it. Carry the sampling tube to a small alcohol flame and seal the extension.

SIMPLIFIED APPARATUS USED IN ANALYZING NATURAL GAS

Figure 26 shows the apparatus used to analyze natural gas. Pipette a contains potassium hydroxide solution; pipettes b and d contain alkaline pyrogallate solution. Pipette c is the slow-combustion pipette. The burette has a capacity of 100 c. c. and is graduated to 0.1 c. c.

PROCEDURE OF ANALYSIS

Natural gas is analyzed at the bureau's Pittsburgh laboratory as follows:

Displace oxygen or other gas left in the horizontal capillary train by drawing a few cubic centimeters of nitrogen from pipette d into the burette; then allow this mixture.

to escape into the atmosphere.

Draw, by mercury displacement, about 100 c. c. of the gas sample from the sample container into the burette. Measure the sample in the burette against the pressure in compensating tube f by bringing the mercury in the manometer tube exactly to mark g. Pass the sample first into the potassium hydroxide and then into the alkaline pyrogallate solution to remove carbon dioxide and oxygen, making burette measurements in the same manner as with the original sample.

COMBUSTION ANALYSIS

FIGURE 26.-Laboratory apparatus for analysis of natural gas. For explanation, see text

Discard the residual gas left after the carbon dioxide and oxygen have been determined, and take a fresh part of the sample for the combustion analysis. Clear the capillary connections of combustible gas by dilution with air, and measure about 100 c. c. of oxygen into the burette and pass it into the combustion pipette. Draw about 35 c. c. of the gas sample into the burette from the sample container and measure it. Heat the platinum wire in the combustion pipette to a white heat, and pass the gas sample at the rate of about 10 c. c. per minute into the combustion pipette containing the oxygen. The paraffins burn as fast as they enter, so that an explosion caused by an accumulation of gas and oxygen can not follow. In analyzing natural gas the authors have

obtained the best results by passing the oxygen into the pipette first. When the natural gas is passed in first, the mixture does not always burn as satisfactorily.

After the paraffins have burned, which requires about 4 or 5 minutes, cool the combustion pipette and measure the contraction in volume due to combustion. Determine the carbon dioxide produced by the combustion by absorption in the potassium hydroxide solution. Finally, pass the gas into the alkaline pyrogallate solution to make sure that enough oxygen has been present for complete oxidation of the paraffins. Some samples may contain such a large proportion of the higher-paraffin hydrocarbons that 100 c. c. of oxygen will not be enough for complete oxidation of 35 c. c. of natural gas. For such samples use a smaller quantity of the gas for the combustion. This statement has special reference to those natural gases that are used for gasoline manufacture and contain a large percentage of the higher paraffins.

PRECAUTIONS

Never raise the burette mercury above the upper stopcock. Bring the gas remaining in the capillary tubing at any stage of the analysis in contact with the solution by passing it back and forth several times between the burette and the pipette. For instance, after combustion, some carbon dioxide will remain in the capillary tubing between the combustion pipette and the burette when the gas is drawn back into the burette to record the contraction in volume. After most of the carbon dioxide has been absorbed by the passage of the gas into the potassium hydroxide pipette, sweep out the small quantity of carbon dioxide in the capillary tubing into the potassium hydroxide pipette. Repeat to insure complete removal of the carbon dioxide. If this precaution is not taken, a serious error will result.

QUALITATIVE TESTS FOR CARBON MONOXIDE AND OLEFIN HYDROCARBONS IN NATURAL GAS

The blood test for carbon monoxide and the iodine pentoxide and palladium chloride tests for carbon monoxide and olefin hydrocarbons are used at the bureau's Pittsburgh laboratory in analyzing natural gases.

A dilute solution of palladium chloride undergoes reduction in the presence of olefin hydrocarbons or carbon monoxide.52 The reaction is marked even when traces of these constituents are present in a gas mixture. Palladium separates out as a black cast, or as particles suspended throughout the liquid. The authors found that 0.1 per cent of ethylene in a mixture of ethylene and air could

52 Phillips, F. C., Researches upon the oxidation and chemical properties of gases: Am. Chem. Jour.. vol. 16, 1894, p. 267.

be detected by the appearance of precipitated palladium when 200 c. c. of the mixture was passed through a 0.5 per cent solution of palladium chloride at the rate of 10 c. c. per minute. The reaction becomes more marked when the gas is shaken in a test tube for about 10 minutes with the palladium chloride solution.

Carbon monoxide in a gas mixture in the same proportion and under the same conditions produces a precipitation of palladium from palladium chloride. Carbon monoxide combines with palladium chloride according to the following reaction:

PdCl2+CO+H20=2HC1+CO2+Pd

The reaction of palladium chloride on ethylene does not produce carbon dioxide.

CALCULATIONS FROM COMBUSTION DATA

If the contraction and the volume of carbon dioxide from the combustion of the gas indicate methane only, this constituent is reported; but if the data indicate higher members of the paraffin series, a calculation is made that gives the two predominating constituents. The gaseous hydrocarbons react with oxygen as follows: CH1+202 CO2+2H20; contraction=2XCH1 vol. C2H6+3.502=2CO2+3H2O; contraction=2.5XC2H6 vol. C3H8 +502=3CO2+4H2O; contraction=3XC3Hs vol. C4H10+6.502=4CO2+5H2O; contraction=3.5×C1H10 vol.

But since the experiments of Rayleigh, Leduc, Baumè, Perrot, and others have shown that some gases deviate somewhat from the gas laws,53 corrections must be made for some gas analyses where this deviation is greater than the experimental error.

Below is a table of the theoretical and observed specific gravities of the gases involved in the calculations; it is taken from Landolt and Börnstein.54

Theoretical and observed specific gravities of certain gases at 0° C. and 760

52 Burrell, G. A., and Seibert, F. M., Errors in gas analysis due to the assumption that the molecular volumes of all gases are alike: Tech. Paper 54, Bureau of Mines, 1913, 16 pp.

54 Landolt, H., and Börnstein, R., Physikalisch-chemische Tabellen. 3d ed., 1905, pp. 222-223; Jour. Chem. Phys., vol. 7, 1909, p. 367.