V O L U M E 2 4 , NO. 7, J U L Y 1 9 5 2
1087
cylinder carries a carbon dioxide inlet tube connected to the aerator stone at the bottom of the cylinder, the filling tube, and the outlet connection to the absorber. Carbon Dioxide Absorber. The construction details of the absorber are shown in Figure 1. The absorber is filled with potassium hydroxide solution, and is identical t o the one used in the Koppers-Hinckley method ( 1 , 3 ) for the determination of butadiene purity. Equipment of this type is stocked by several laboratory supply houses. Gas Buret. The gas buret is water-jacketed and graduated in 0.1 nil. a t the top. REAGENTS
Potassium Hydroxide Solution. Prepare a solution of approximately 46% by weight of potassium hydroxide. The potassium hydroxide solution may be used for approximately 2 hours of continuous stripping. Carbon Dioxide. The carbon dioxide should be of high purity. The criterion of adequate carbon dioxide purity, as well as of complete removal of air from the apparatus, is that, in 15 minutes of carbon dioxide flow through the apparatus, not more than 0.1 ml. of gas be collected in the buret.
Table I. Hydrocarbon Butadiene Butadiene Butadiene Butadiene Butadiene Methane
Synthetic Sample Analyses of Light Hydrocarbons i n Water Present, P.P.M. 582 640 571 640 27 19
Found, P.P.M. 561 622 554
622 26
18
Recovery. % 96 97 97 97 93 95
PROCEDURE
Lubricate the tapered joint of the sample stripper and all stopcocks of the apparatus with a water-insoluble lubricant. Silicone grease is ideal for the pur ose. Fasten the two members of the tapered joint securely withielical springs or strong rubber bands. Introduce sufficient mercury through the absorber stopcock to counterbalance the head of potassium hydroxide solution to be added to the absorber and buret. The mercury leg acts as a seal t o prevent the potassium hydroxide solution from reaching the absorber stopcock. The absorber and buret are filled with potassium hydroxide solution by leaving the buret open as the reagent is added through the leveling bulb. Pass the stream of carbon dioxide through the sample stripper and out the absorber stopcock to the air. To save time in flushing, increase the flow from the generator. Flush the sample inlet line with carbon dioxide by opening the stripper stopcock. Do not allow the carbon dioxide to pass into the sample stripper xith the outlet stopcocks closed, because the pressure may be sufficient to part the tapered joint. After the purging of the apparatus is complete, pour the water
sample into the graduated filling tube of the sample strip er Stop the carbon dioxide flow and open the absorber stopcocg to the air. Immediately introduce a sample of the desired size into the scrubber through the stripper stopcock. A practical volume of water sample is 1000 ml., but a greater volume may be desired for increased accuracy and the recovery of sufficient hydrocarbon for further analysis. After closing the sample inlet stopcock following the introduction of water sample, turn the absorber stopcock to connect the stripper to the absorber. Start the flow of carbon dioxide and continue until all the hydrocarbon has been stripped from the sample. The removal of all hydrocarbon may be considered complete when successive readings a t 15-minute intervals show not more than 0.1-ml. increase in the gas volume. When the stripping is complete, stop the flow of carbon dioxide. adjust the leveling bulb, and determine the volume of gas collected in the buret. In calculating the weight of light hydrocarbon present, make appropriate corrections for water vapor and the impurity from the carbon dioxide. Normally the probable composition of the gas collected will be known, and the conversion from gas volume to weight can be made. When this is not the case, sufficient gas may be collected in a freeze-down bulb or bomb through the gas buret stopcock to establish its composition by Orsat, fractional distillation, spectrometry, or a specific gravity measurement. The results are reported in parts per million from the following equation : Light hydrocarbon (in p,p.m.) =
weight of light hydrocarbon X 1 0 6 weight of sample
RESULTS AND DJSCUSSION
Table I shoirs analytical data for synthetic samples of butadiene in water and methane in water. The data indicate good recovery of butadiene and methane. An accuracy approaching 1 p.p.m. is indicated a t low concentrations of hydrocarbon gases. While the work of this laboratory has been confined to C, and lighter hydrocarbons, it is believed the method could be modified to determine heavier hydrocarbons in water. It would be necessary to measure both the vapor and liquid phase of the separated hydrocarbon in the gas buret, and it might be desirable to heat the sample to speed the stripping of the heavier hydrocarbons. LITERATURE CITED
(1) Am. doc. Testing Materials, “Standards on Petroleum Products and Lubricants,” Part 5, D 973-481‘, 1949. (2) hlusante, A. F. S., ANAL.CHEM.,2 3 , 1 3 7 4 (1961). ( 3 ) Office of Rubber Reserve, Reconstruction Finance Corp., Buta-
diene Analytical Committee, “Butadiene Laboratory Manual,” Method L.M. 2 . 1 . 1 . 1 0 , 1946. (4) Webber, L. A., Proc. Am. Petroleum Inst., 111,28th Annual Meetr ing, 28, 19 (1948).
RECEIVED for review March 17, 1952.
Accepted June 5 , 1952.
A Differential Refractometer for Process Control E. C. MILLER, F. W. CRAWFORD, AND B. J. SIMMONS Research Division, Phillips Petroleum Co., Bartlesville, Okla.
I
N MODERN petroleum plants many processes require continuous analytical data for the proper adjustment of the process controls. The rapid response of continuous process systems to operating variables, the large throughput capacities of these systems, the ever-increasing product quality specifications, and the economy of operating a t a uniform quality level require that these analytical data be applied to process control with minimum time lag. These requirements are being met by automatic analyzer-controllers. In any particular application satisfactory control depends not only upon the sensitivity of the analyzer as compared to allowable fluctuations of the measured variable but also upon the reliability
and safety of the device under the conditions prevailing a t the installation site. Several continuously recording refractometers have been described in the literature. Barstow ( 1 ) has described a recording refractometer of the dipping type, which is now commercially available. Several recording refractometers (2, 3) are based upon the variation of reflectivity with index of refraction a t a glassliquid interface. These devices all compare the index of refraction of a liquid to that of glass. Thomas ( 4 ) and Zaukelies ( 5 ) have described laboratory-type continuously recording differential refractometers which compare the index of refraction of a fluid to that of a standard fluid without a direct comparison to glaes. To
ANALYTICAL CHEMISTRY
1088
In present-day petroleum processing plants automatic aiialyl,er-controllers are being used to applj the results of continuous analysis directly to process control. -4differential refractometer has been developed for such applications, designed to satisfy safetj requirements and give reliable indications of changes in index of refraction to 0,0001 unit, or better, in outdoor process plant locations. Adequate sensitivity and good stability are achieved without therrnostating by using adequate heat exchange and pressure equalization between the flowing procesb sample and the reference sample. This del ice nul> also he rised to compare the index of refraction of t w o process streams.
the authors’ knowledge, no iecoidiiig differential iefractonieter designed for plant usage has been described in the literature. I n many process control applications the primary interest is in measuring the change in index of refraction relative t o that of some specification material. The differential refractometer does this directly and is very insensitive to the conditions of temperature, pressure, and spectral purity usually associated \$ ith the measurement of small changes in index of refraction.
about the equiconiposition point, with temperature and pressure changes, in a manner dependent upon the temperature and pressure coefficients of refractive index of the material in the cell. I n addition, a t different temperatures and pressures a given concentration of impurity in the sample cell compartment can be expected t o produce slightly different readings because of differences in temperature and pressure coefficients of refractive index for the impure and reference materials.
GENERAL DESCRIPTIO\
Figure 1 is a schematic repiesentation of the reflactometel optical-servo system. The incandescent source is located a t tlie focal point of lenr L1, so that parallel light traveises the double-prism cell Cl-C, Light emerging from the cell traverses lens L1 and after reflection from mirroi N forms an image of the source on the barrier-layel photocells, R1and R2. The difference in output of the tTvo photocells feeds a servc-amplifier to energize the servo-motor. The niotoi rotates mirror M , by means oi a cam, to position the bource image symmetrically on the photocells. This forms a nullbalance system in which the rotational position of the cam is a ineasure of the angular deviation i n i p i teci to the light heam tiaversing the cell.
1
SAM
I I
I
t I
I
1”igure 2. SOURCE
‘
-
I/+
I
‘INSULATION
I
I1
I
I I
I 1
-
Schematic Temperature and Pressure Equalization System
The instiunient is insensitive t o sample color a t the equicomposition point, as in this case there is no dispersion of the light. However, the efkcts due t o sample color and dispersion may need to be considered when the quantity of impurities becomes large, if the dispersion of the impurities is very different from that of the referencc sample. For the highest accuracy the color of the sample may become important, even if the dispersion characteristics of the materials in cell compartment,s C1 and C2 are alike. This arises because the light is dispersed on leaving the cell if the light path deviates from the equicomposition axial path. The authors have no experience with this equipment on highly colored streams. However, the refractometer is insensitive t’o accumulations of dirt in the sample cell. One of the advantages of the differential refractometer relative to refractometers which use a glass reference, is that the errors This these effects can be made negligible for most applications. from can be done simply by using for the reference saniple a sample of the material actually present at the sample point when specification product is beingmade. Thisassures that sample and reference material have similar refractive index coefficients and dispersive characteristics. Thus the accuracy is best when near-specification material is being produced and this accuracy may be maintained over a wide range of temperatures and pressures such as are normally encountered a t plant locations. Figure 2 is a schematic representation of the temperature and pressure equalization system.
1 AMPLIFIER
gigure 1.
Schematic Optical-Servo S>stern
Khen the same material fills Iioth cell coinpartnients (,”I and C,, light n.ill he transmitted without deviation and there will be no dispersion, provided that the temperature and pressure in C1 and C2 are the same. Furthermore, this situation n 3 l be maintained a t all temperatures and pressures. Thus a constant instrument reading at the equicornposition point can tie ohtained without thermostating and without pressure control. The condition thst must be fulfilled is that the differences in temperature and pressure between Cl and Cyz tie mitintuiueti helow the tolerance levels set by the acmmyy desired.
If the refractive index of the sample in C”? is changed by the addition of an impurity, this produces a deviation in the path of the light through the cell. The servo system responds so as t o lialance the light on the phot’ocells, this balance corresponding to a new rotational position of the cam. The change in rotational position of the cam is a measure of the amount and nature of the impurity. For the highest accuracy the instrument must be used at the same temperature and pressure a t lvhicli it was calibrated. The scale of index difference us. cam rotation \vi11 expand and contract
This representation is for a locked-in reference sample in CI : ~ n da Honing sample in C‘?. A thermally insulated box surround-
V O L U M E 24. NO. 7, J U L Y 1 9 5 2
1089
ing a portion of the refractometer contains, among other things, the cell CI-C9 and two heat reservoir-exehangers, H , and Ha. Thus these components will experience a reduced rate of temperature change relative to ambient fluctuations. The heat exchangers are liquid-filled tanks containing a length of tubing through which the sample prisms.
The rotational position of t,he cam is mechanically indicated on the face of the instrument and electrically drives a remotely looitted recorder-cont,roller. The sliding contact of a Helipot voltage divider is directly coupled to the cam shaft., whose rot;it,ion is limited to 315 degrees by a spring-loaded stop. A constant voltage is applied to the divider, so that the voltage between one end of the divider and the sliding contact is a meamre of the cam rotational position. This latter voltage is used to drive the voltagesensitive recorder-controller, Tho cell used in this equipment is constructed of stainle88 steel vith quar1,z \xiridom (two end windows and onc diagonal window) sealed by plastic O-rings. The plaiie of the diagonal u-indow is st 26 degrees with respect to the axis of the cell with an effective aperture of 0.5 inch. This aell has heen hydrostatically tested to 500 pounds per square inch and has been used an process streams where the cell onemtine D T ~ S G U ~ Chave S been a8 hich 2,s 140 pounds per ,r&mz inch. For index of refraction spans of less than 0.01 unit it is necessary t,o use an illuminated slit as the ~ouroeto prevent exoessive drift due to change in lamp filament position. The use of R narrow slit whose image is but slightly wider than t,he gap between the photocells minmiaes the effects due to variation in 8ource intensity, uiiequal photocell sensitivities, and changes in photocell sensitivity. Projection of an image of the lamp (Geneml Electric lamp No. 1493) filament on the collimating lens. L,,maintains a sufficient, level of illumination in the slit image a t the photocells (I,, focal length 6.75 inches, L focal length 19 inches). The components showii in Figure 3, except for the motor, lamp, and indicator, are within the thermally insulated chamber of Figure 2. Cell compart,ment, CZ forms one wall of tho heat rsrhanger, &. Figuro 4 shows the "cuplosion-r~~ist;nt" enclosure for t.hc refractometer. ~~
Figure 3.
Schematic 0ptieal-Meeh;mical SysLeni
The sample upon entering the insulated chamber is brought to the approximate temperature of the components in the chamber by H I . The sample then traverses &, which is thermally connected to CtrCs. Thus the ternmrature of the flowinz samde
satisfactdry for m&t amlicatians'whe~ea &sitivity
of 0,0001
~-
m i s t have sufficient expansive"capacity to ebmpensate jar the thermal expansion of thc reference sample without exceeding the pressure difference tolerance permissible between CI and Cg. Figure 3 is a schematic representation of the opticsl-mechanical system of the refractomeber. In this particular arrangement the retrodirective reflector system shown is used instead of the single rotating mirror. M of
of refmotion man8 of about 0.005 unit. Thn retrodirective SYS-
drstance betGeen the apex of The m i h r system and the axis of rotation of the cam follower.
Figure 4.
Partially Assembled Explosion-Resisting E"OlOSU~C3
This enclosure consists OS an aluminum faceplate and a drawn steel bell with a welded steel bolt flange. The hell weighs about 60 pounds and is not diffiault to handle with the help of the guide rails. The entire ins1,iument is mount,ed on the facedate in such it manner that all components are readily accessi6le when the bell is removed. The- indicator, sample lines, and electrical connections are brought, through the faceplate. The enclosure is Drovided with seals for the electrical connectiona and is designed to meet explosion-proofing specifications for N a t i o i d Elect,rical Code Class I, Group D , locations. It has been t&cd hydrostatically to 500 pounds per square inch and has been sub-
1090
ANALYTICAL CHEMISTRY
jeoted to explo8ion teste using mixtures of hydrocarbons and air. Mixtures inside the chamber were exploded without igniting similar mixtures surrounding the enclosure.
Figure 5 is a top view of the instrument with the explosionresistant hell removed The components are separated into two com artmcnts IIY n vertical plate attached to the faceplate. AttacRed to o w sidr of this plate are a Brown Electronik Continuous balance unit (the servo-amplifierand servo-motor), the telemeter u w e r s u u ~ l j ,
i r e insulated with a 1-inch layer of cork. The light sot& housing shown with the insulation removed is attached to the faceplate. The heat reservoir-exchanger, H I of Figure 2, is a liquid-filled tank attached to and covering most of the hottam of the haseplate. The pressure equalizer is also attached to the bottom of the baseplate, immediately below the source housing and photocells. APPLICATIONS
The installation used for the initial field tests of this type of equipment is shown in Figure 6. The instrument was located a t the base of an 7u-butane-isohntanefractionating column and was protected from the weather by only a meter house. The recorder was located in the control roam. Such outdoor locations as this .suggest the extremes in working conditions to he anticipated for .an$y~er-controllers. I n this application the operating pressure of %he refractometer cell was approximately equal to that of the column to ensure a liquid phase sample. Currently, automatic differential refractometer control of a .distillation column separating hexanes is heiug used as regular plant operating procedure. In this partioulm application a full acale index of refraction span of 0.01 unit is being used. Tests utilizing the equipment described above on a flowing iscpentane sample a t 50 pounds per S q U m inch show a stability of the equicomposition point of +O.OMM2 refractive index unit when the refractometer is subjected to Budden ambient tempers, ture changes of 30' F. When a gas sample is used, the sensitivity and stability in a laboratory environment are about 0.oaOW1 unit. CONCLUSIONS
I U V I U I U b
usually associated with index of refraction mewrements are minimized by using adequate heat capacity, adequate heat exchange between the flowing sample and a stationary reference sample, the differential type of orll, and the pressure equalizer. This equipment should also be useful for comparing the index of refraction o i two separate streams when the difference in index of refraction between two sample points is indicative of process performance.
(2) Jones, H. E., krhmsn, L. E., and Stshly, E. E., ANAL. CHEM.. 21, 1470-4 (1949). (3) Xarrer. E., and Orr. R. S., J . Optical Soc. Am.. 36, 42-6 (1'346).
Thomaa, G.R.,O'Koneki, C. T.. and A :urd. C. D., ANAL.CHEM. 22, 12213 (1950). ( 5 ) Zaukelies, D.. and Frost, A. A,, Ibid., 21, 743-5 (1949). (4)
Experience with refractometers of the type described here has shown that adequate sensitivity and reliability for many applics, tions may t,e achieved without thermostating. TI?e difficulties
Ultravinlclt
Figure 6 . Initial Field Test Setup for Refractometer
RE