842
T H E ] O U R N A L OF I N D U S T R I A L A N D ENGINEERIiVG C H E r l f J S T R Y .
tein than cold alcohol, there does not appear t o be any advantage in its use. 3 . Alcohol varying in strength from 4 j per cent. t o 5 5 per cent. b y weight extracts more protein than alcohol of any other strength, hence it is recommended t h a t j o per cent. b y weight alcohol be used for the determination of gliadin in wheat flour, and that the use of 7 0 per cent. alcohol, whether b y weight or b y volume, be discontinued. DIVISIONO F AGRICULTURAL CHEMISTRY A N D S O I L S , MINNESOTA EXPERIMENT STATION, ST. P A U L .
THE DETECTION OF BENZOIC ACID IN COFFEE EXTRACT. By HERMANc I.YTHGOE A N D CL.4RENCE E Recerved July 21, 1911
MARSH
I n testing a sample of coffee extract for benzoic acid by extracting with ether and testing the ether extract in the usual way with ferric chloride, a precipitate was obtained corresponding to ferric benzoate except in color, but on subliming this precipitate the crystals did not have the characteristic appearance of benzoic acid. A portion of the original sample was then acidified with phosphoric acid and subjected t o distillation with steam. The distillate was made alkaline with sodium carbonate, evaporated t o about 5 0 cc., acidified, extracted with ether, and the ether extract was extracted with ammonia. The ammoniacal solution was then evaporated until free from ammonia and ferric chloride was added which produced a precipitate with the same characteristics as in the previous instance. A sample of coffee extract was then made from pure coffee, and upon repeating these tests this extract was found t o act the same as the commercial extract. A large number of coffee extracts were made from different varieties of coffee and in all cases they reacted in the same manner. The ammonium salt of this substance which was extracted with ether was found t o give precipitates with salts of manganese, nickel, magnesium, calcium, and barium, as well as salts of iron and copper, while benzoates will produce precipitates only with salts of iron and copper, and from these differences the following
Nov.,
1911
method for the detection of benzoic acid in coffee extract has been devised: Make the solution acid and extract several times with ether. Wash the combined ether extracts with water and extract with ammonia. Evaporate the ammoniacal extract to a small volume, adding ammonia from time to time t o prevent it from becoming acid, and add a solution of manganese sulphate. Filter through a small filter, wash with as little water as possible and add ferric chloride t o the.filtrate when a dark greenish precipitate will occur if benzoic acid is present. Evaporate to dryness in the casserole in which the precipitation was made, and sublime b y placing an inverted funnel covered with a filter paper in the dish and heating over an asbestos gauze. Remove the funnel, and determine the melting point of some of the crystals which, if benzoic acid, should be 12I .4 ' C. The rest of the crystals may be dissolved in ammonia, the excess of ammonia evaporated a n d ferric chloride added, when the characteristic fleshcolored precipitate will occur if benzoic acid is present. For quantitative purposes the method of Edmund Clark1 was employed with good results as the natural reacting substance has but little influence. Determinations made on pure extracts b y this method gave from 0.01 t o 0.04 per cent. benzoic acid and a correction can be readily made if desired. The accompanying table gives the analyses of a sample of coffee extract made in the laboratory and of two commercial extracts, one of which contained benzoic acid and glycerine, being very deficient in coffee: ANALYSESOF COFFEE EXTRACTS.
Made in laboratory.. 1 . 0 5 9 1 3 . 5 6 2 . 5 0 0 . 2 2 Commercial. .. . . . . 1,057 13.40 2.29 0.17 Commercial.. . ... . . 1 , 0 7 0 2 1 . 4 2 0 93 0 . 5 8
.
0.36 0.42 0.15
0.67 0.70 0 14
. .. . .. . . 0.19
DEPARTMENT OF FOOD AND DRUGINSPECTION. IvASSACHUSETTS S T A T E BOARDOF HEALTH, BOSTON.
PLANTS AND MACHINERY ment of different gases demonstrating the practicability of the method for commercial purposes. The reasoning on which the operation of the meter Received September 5 , 1911. is based is this: The specific heat of most gases is The application of the electric meter t o commercial measurement of gases first suggested itself t o Pro- a quantity already accurately determined and i t is fessor Carl C. Thomas while carrying on extensive known t h a t this value changes but slightly through experiments on the specific heat of superheated steam wide ranges of temperature and pressure. If a n at Sibley College and the University of Wisconsin. electric heater be placed in a pipe line through which These investigations required the use of some form there is a constant flow of gas, and the heater give of heating device. The difficulty of direct quantitive off heat a t a constant rate, the gas temperature is measurements of heat except b y electrical measuring raised a certain fixed amount. Any change' in gas instruments finally required the adoption of electric flow under these conditions will mean a change in heaters. The great convenience of these heaters its temperature increase, an increase in gas flow in this work led t o a series of tests on the measure*Science, Aug 20, 1909 p 253 THE THOMAS GAS METER. B y H . N. PACKARD.
Nov., 1911
T H E JOCRA\."lL OF 1,VDCSTRIAL A S D ENGIiTEERING CHEMISTR I?.
causing a decrease in temperature difference while a smaller gas flow causes a n increase in temperature difference; that is, the temperature change of the gas with constant heater input is inversely proportional t o the rate of flow. The natural development with these facts in mind was the use of a graphical temperature recorder to show the change of temperature difference as the gas flow varied, and while a constant quantity of heat mas being added. While this method gave a measure of the gas flow i t was not entirely satisfactory from a commercial standpoint as the temperature charts could not be readily integrated to give readings directly in cubic feet. The next step from constant heat input with variable temperature .range was to constant temperature difference with
Fig. 1.-Sectional
Let Q
E 3412
T S
843
cubic feet of gas per hour a t some standard condition. = energy in K. W. = B. T. U. equivalent of I K. LIT. hour. = temperature range through which gas is heated. = specific heat a t constant pressure of the gas, per cubic foot. =
Then Q
=
3412E/TS.
C O N S T R U C T I O X O F T H E METER.
The meter in commercial form consists essentially of two parts: first, a means for adding heat; second, thermometers for regulating the temperature rise due to the addition of this heat. For the heater an electric resistance, B, Fig. I , is used of such a form
Tiem of meter casing showing heater (B)and thermometers ( E )
variable heat input. In this form a satisfactory commercial meter was developed. I n the commercial meter b y means of an automatic control the input to the heater is made to vary with the quantity of gas flowing in such a way as to keep the temperature range through which the gas is heated a constant quantity. That the watts input to the heater gives a direct measure of the weight of gas flowing will be shown in the following example: A commonly accepted value for the specific heat of air at constant pressure is 0 . 2 3 7 5 B. T. U. per pound. Expressing this value in watts, we have directly the watts required to raise the temperature of one pound of air one degree F. As this is not the most convenient form for commercial work where the unit desired is the cubic foot a t some standard condition, we may use the specific heat per cubic foot a t that condition instead of the specific heat per pound and obtain the watts required per cubic foot to raise the air temperature one degree. As the watts used by the heater are in direct proportion t o the gas flow, the wattmeter may be calibrated t o read directly in cubic feet of gas a t standard conditions. To sum up, the meter measures the gas in units of weight, not volume. Thus the pressure and temperature of the gas measured have no influence on results. If the weight per standard cubic foot of gas is known, the meter can be calibrated t o give readings directly in this form. The action of the meter may be expressed by the following simple formula :
as to distribute the heat evenly throughout the gas. I t s current consumption is measured by graphical and recording wattmeters. The thermometers E , Fig. I , are of the electric resistance type built in the form of a screen of small enough mesh to give an almost perfect average of the gas temperature over the whole section of a pipe, no matter how unevenly the gas flow is distributed. They are placed one on each side of the heater, measuring the difference in temperature between the incoming and outgoing gas. The thermometers and the heater are contained in a casing that is slipped into a section of the pipe line. The heater has an automatic controlling rheostat that allows a sufficient current to flow through the heater t o maintain a constant temperature difference between the two thermometers of about two degrees F. On a switchboard a t any convenient distance from the pipe line (Fig. 2 ) are mounted the wattmeters and the automatic heater control device. The heater rheostat control is operated through a Wheatstone bridge galvanometer (Fig. 3 ) , two legs of the bridge being made up of the two thermometers, the other two being standard coils with zero temperature coefficient. The thermometer resistances T, and T, (Fig. 3 ) are adjusted to make them as nearly equal as possible when a t the same temperature, the final adjustment being made by a small rheostat in the bridge circuit for exact balance, with gas flowing through the meter. The temperature coefficient of the thermometer wire being accurately known, a small resistance equivalent
Nov..,
1911
T H E J O U R N A L OF I h ' D U S T R I A L A N D EiVGIXEERIh'G C I I E A V I I S T R Y .
device the proper amount of energy is allowed t o be dissipated in the heater t o maintain a tw'o-degree temperature difference between thermometers T, and T,. Any change in the bridge galvanometer needle from its balanced position, due t o change of the proper temperature range between T, and T,, causes a corresponding change in the heater energy. C O M M E R C I A L REQUIREMENTS O F A METER.
'
For a meter t o be successful commercially there are several essential requirements. Accuracy, large range, ' freedom from interference with operation b y deposits from the gas and a reasonable cost of operation are perhaps the most important. Most forms of commercial meters will work within z per cent. of correct values when properly adjusted and calibrated and when within the range of measurement for which they are designed. While this is true when first installed, after a long period of operation calibration is necessary t o prove accuracy. With most meters this means a removal from the line with a consequent interruption of service. With the Thomas meter this checking for error may be accomplished in a few moment's' time without disturbing it a t all. The operation is as follows: the heater circuit is broken and the temperature difference coil in the Wheatstone bridge is shorted out. This should leave the bridge'in perfect balance, as the two thermometers if properly made have exactly the same resistance a t the same temperature. If not in balance, b y means of the bridge adjusting rheostat already described, this balance may be restored. Now t h a t the thermometers are known t o be of the same resistance the only chance of error is in the temperature resistance coil and the electrical measuring instruments. The resistance coil is so made t h a t a n appreciable change in the resistance is impossible. The wattmeters can be calibrated b y placing instruments of known accuracy in series with them. With most meters deposits from the gas are likely t o affect their accuracy somewhat. With the electric meter the only effect is t o change the speed of the thermometers, the increased mass making them slower t o respond t o temperature fluctuations. A deposit does not seriously affect the accuracy, only making the instrument slow t o register rapid change of flow. The Thomas meter normally covers a range of 15 t o I accurately, but b y a slight adjustment this range can be increased t o about 60 t o I . To do this two temperature difference coils are used. For high range work the gas is heated only two degrees while for the low range a 5 - or Io-degree temperature difference is maintained. This requires t h a t the wattmeter readings for the low range be divided b y the ratio of the two temperature differences t o give the true readings. The only maintenance charge against the meter is the cost of the operating current. A kilowatt hour will measure about 75,000 cubic feet of free gas, thus making the current cost less than one-fourth of one per cent. of the value of the gas measured, ,an extremely small item when it is considered t h a t
,
845
the accuracy usually expected of commercial meters is not better than one per cent. Except for oiling bearings and changing charts on the graphical meter, no attendance is required. All results are direct in standard cubic feet, with no laborious computations or corrections for temperature and pressure. E R R O R S D U E TO V A R I A T I O N
O F GAS COMPOSITION.
In the formula Q = 3412 E/ST, E and T are fixed for a given gas flow, leaving the only theoretical source of inaccuracy in the value of S. The conditions affecting the accuraoy of S, the specific heat a t constant pressure, are chiefly variation of moisture content, and variation of the composition of the gas t o be measured. The moisture correction would be maximum with a saturated gas. Taking the case of air a t 60 degrees F., 30 inches barometric pressure, the weight per cubic feet of the saturated air is 533 grains. The weight of the vapor contained in it is 5.6 grains or 1.05 per cent. of the total. As the specific heat of the aqueous vapor is about twice that of air, the specific heat of saturated air is about one per cent. higher than t h a t of dry air for these conditions. In the measurement of gases such wide variations in moisture content are seldom met with and in most instances can be entirely neglected without appreciable error. I n any case, if desired, a correction may be made for moisture content. The second source of error, t h a t due to variation of specific heat with variable gas composition, actually works out t o give very small errors as the constituents of the gas liable t o variation are such as to have b u t small effect on the specific heat, as the following example will show. A typical city gas is made u p of a mixture of coal and water gas, the proportions of this mixture varying through wide ranges. Below are given sample analyses of the two gases, the specific heats of each computed from the analyseq, and the specific heats of various mixtures of the gases. Water gas 3.8 12.3 0.6 30.4 33.3 14.4 5.2
co*. . . . . . . . . . . . .! . . . . . . . . . . . CnHzn..
.....................
0 ...........................
co .......................... H . . . ........................ CHI ......................... N ...........................
Coal gas. 2.2 3.3 0.5 6.9 50.6 34.2 2.3
From these analyses the following values of specific heat are computed for different mixtures. Water gas. Per cent. 100 90 80
io 60 50 40 30 20 10 0
Coal gas. Per cent. 0 10 20 30 40 50 60
io .
80 90 100
Specific heat. 0.02094 0.02096 0.02098 0.02100 0.02102 0.02104 0.02106 0.02108 0.02110 0.02212 0,02114
The above values of specific heat are in B . T.U. per degree F. for a cubic foot of gas at 14.804 pounds absolute pressure and 60 degrees Fahrenheit.
846
THE JOCR.Y.4L OF I i Y D U S T R I A L AIVD EA\:GI.\;EERIi\7G
As may be seen from this table the maximum error when changing from I O O per cent. coal gas t o I O O per cent. water gas is slightly less than one per cent. The average error would be much less than this. Corrections can be made as desired if frequent analyses are obtained, but in general an average gas composition can be determined for which the specific heat can be computed, giving results over a period of time that are extremely close t o the true value. Excess of moisture in the form of a finely divided spray causes some inaccuracy in results. This may be overcome by use of a separator or a heater to bring the gas above the saturation point, or better still installing the meter in a part of the line where such troubles cannot occur.
L A B O R A T O R Y TESTS.
A number of tests were carried out recently a t the University of Wisconsin with three types of meters in series on a pipe line, measuring air. The meters used were the Venturi meter, Pitot tubes and a Thomas meter of a crude form having a hand-controlled heater. Simultaneous readings were taken for each meter with different rates of air flow. These three instruments, whose design depends on as many different theories, gave results checking each other within small limits, proving the accuracy of the reasoning involved in each case. The electric meter gave results directly in standard units, and with the same degree of accuracy even in a crude form t h a t required laborious computations and extremely careful manipulation with the other two types. COMMERCIAL T E S T S . I . Artificial Gas a%d Air.-A test made on this meter during the development work on artificial gas and air gives a n interesting check on the correctness of its design. The meter was operated in series
CHEAUISTRY.
Nov., 191I
with a large wet meter of known accuracy on both gas and air, and curves plotted from the test results between cubic feet of gas flow per hour and degrees rise in temperature.
It was desired t o compute the specific heat of a gas by means of a test based on the following reasoning: Let G = cubic feet of gas per hour. E = energy in kilowatts. 3412 = B. T. U. equivalent of I K. W. hour. S = specific heat per cubic foot. Then GST = heat energy,equivalent to E or. GST = 3412 E. GT/E = 341z/S = a constant K, this depending on the value of S.
I n Fig. 4 the curves are plotted with values of G, (cubic feet of gas per hour) as ordinates and T / E (degrees rise per kilowatt expended in the heater) as abscissas. Since G X T/E = a constant, these curves are rectangular hyperbolas. That these curves are asymptotic t o the co-ordinate axes is evident from the fact that an infinitely large mass of gas would have its temperature increased an infinitely small amount b y any finite amount of heat, while a t the other extreme a finite amount of heat will cause an infinite rise in temperature to an infinitely small mass of gas. I n the commercial meter the value of T is kept constant. Then K / T will be a constant. Let its value be C. This gives G = K E / T = CE. From the gas curve (Fig. 4) which was made with illuminating gas a t a n average temperature of 59 degrees F. and an average absolute pressure of 6 inches of water plus 29.8 inches of mercury, the value of the constant K is found t o be 170,000. But K = 3412/S = 1 7 0 , 0 0 0 . Then S = 3412/17o,ooo = 0.0201. Reducing this value t o standard conditions of 32 degrees F. and 29.9 inches mercury it becomes 0 . 0 2 1 0 , which is t o be compared with the calculated
specific heat of 0 . 0 2 1 1 . The analysis (if this y a s and the computation ior its specific heat. is yivcn below. \vciy~>t \-'>I. EX/ It.
cop. . . . . . . . . . . . .
ciii,. . . . . . . . . . . . . ox. . . . . . . . . . . . . . .
o
04 o.ii
o ooi
co.. . . . . . . . . . . .
0 33,
Clil.. ........... tir. . . . . . . . . . . . . N~. . . . . . . . . . . . . . .
0,iiili 0.303 o OXR
i'er ru i t lh.
o.iimi 0.0741 o 08163 0 074"i
T
