Application of Infrared Nondispersion Analyzer - Industrial

Modification of Bolometer and Bolometer Circuits of Infrared Analyzer. A. W. Wotring , R. F. Wall , and T. L. Zinn. Analytical Chemistry 1956 28 (9), ...
1 downloads 0 Views 664KB Size
PROCESS INSTRUMENTATION reading after a total of 8 hours. The above implies that rapid small changes and/or large changes over long time periods could be tolerated. The values found by test were 1' C. per 15 minutes for the continuous change and & 2' C. for the rapid fluctuations. In conclusion, a number of tests of potential liquid applications are listed in Table I.

FILTER CELL

Table 1. Compound Measured

COMPENSATOR CELL

Figure 1.

H20 HzO H20 Ha0 Hz0 HCHO CHsCOOH CHsCOOH CnHsCHi C6HfiC1

Negative Filter Type Continuous Infrared Analyzer

For example, in monitoring a benzol flow in the laboratory a t approximately 5% toluol, a sample containing dissolved air was syphoned through the sample cell, This resulted in vapor formation in the cell (at 40' C.) a t flow rates to 20 ml. per minute with sample temperatures as low as 20' C. When the sample was degassed (by boiling) and subject to 1 pound per square inch gage flow rates as low as 5 ml. per minute and sample temperatures up to 50' C. could be used in a cell at 80 ' C. Subsequently, in monitoring 0 to 500 parts per million, water in circulating stream, extensive tests were made to determine more precisely the effects of varying temperature and flows. The effects were, in fact, long time transients. To illustrate: a dry benzol stream a t 30' C. was rapidly (15 minutes) brought to 70" C. Within this interval, the instrument made nearly a 100% scale excursion. With the sample held a t 70' C. the instrument recovered 90% in the ensuing 2 hours and returned to the correct

a

laboratory Tests of Liquid Analyses Vehicle CHsOH CHsCHOHCHi CHXCOCIH~ CsHe CClaFi Hi0 (CHsC0)iO CeHe CeHe oCeHG9

Length Cell, Mm. 10 10 0 2 2 0 20 0 0 0 0 0

0 2 2 2 5 5

Range Tested, %

Sensitivity,

40- 55 0-100 0- 2 0-,5000 0-100a

=tO.O05 10.01 10.01

38- 39 0-100 0- 50 0- 10 0- 10

%

A55

rir0.1n 10.01 fO.O1 f0.2

fO.l

=to,1

Parts per million.

~~

Summary

The negative filter-type, continuous infrared analyzer is applicable to a variety of liquid phase plant stream analyses. In monitoring liquids, variations in stream temperature, pressure, and flow are more critical than for gas streams. literature Cited (1) Patterson, W. A,, Chem. Eng., 59, 132-6 (September 1952). RECEIVED for review September 7, 1923

ACCEPTED May 11, 1954.

Application of Infrared Nondispersion Analyzer Refinery Process Streams L. HOLLANDER, G. A. MARTIN, AND C. W. SKARSTROM Standard Oil Developmenf Co., Linden, N. 1.

This paper deals with the application of commercially available nondispersion infrared analyzers to refinery process streams. The application work involves determining how to sensitize and calibrate a particular type nondispersion infrared instrument for an analysis, specifying the analyzer and associated sampling equipment, installing, operating, and maintaining the equipment. A specific application is presented, involving the measurement of 0 to 10% isobutane in a C4-C6 hydrocarbon stream.

T

HE petroleum industry is actively interested at the present

time in the application of continuous analyzers to refinery stream services. This interest stems from the high return on investment for this class of equipment due to decreased deviation from product quality specifications, increased efficiency of process unit operations, and reduction of storage capacity requirements. It is not uncommon for many of these applications to result in 1000% return on investment per year. Continuous analyzers are differentiated from conventional instruments (flow, temperature, pressure, and level) in that they July 1954

can be sensitized to measure specifically the concentration of a product or product contaminant in a flowing stream. Dielectric constant, pH, infrared absorption, and ultraviolet absorption measurements are a few of the methods that have been used for continuous analysis of refinery streams. Because plant-type infrared analyzers are versatile in that they have been sensitized to detect a single hydrocarbon in many complex hydrocarbon streams, they currently enjoy a favorable position in the continuous analyzer class of equipment. Infrared analyzers are either of the dispersive or nondispersive

INDUSTRIAL AND ENGINEERING CHEMISTRY

1377

ENGINEERING, DESIGN, AND PROCESS DEVELOPMENT type. The dispersive type works with essentially a single wave length in the infrared spectrum, whereas the noridispersive type works with all of the energy from the infrared source. To date, commercially available plant-type dispersive infrared analyzers are essentially the same as their laboratory prototypes which are operated intermittently by specially trained technicians. As a result, their complexity and high service requirements have virtually precluded their use in continuous refinery stream services. For this reason, the present paper deals with only nondispersive infrared analyzers. SAhnaLE

IhTERFERENCE

COMPENSATOR

SOURCE

-

-

FILTER CELL

Figure 1.

DETECTOR

Nondispersive Infrared Analyzer

The application of nondispersive infrared analyzers to new petroleum refinery services involves determining how to fiensitize and calibrate the equipment for a particular analysis, specifying the equipment, installing the analyzer and associated sampling equipment, operating, and maintaining the equipment. This paper describes these phases of infrared application work in a refinery and discusses them as they pertain t o the continuous analysis of isobutane in the presence of C j and other Cq hydrocarbons.

Sensitizing and Calibrating the Equipment There are several types of commercially available nondispersive infrared analyzers. Each type has its advantages, and experience has shown that no one analyzer is best for all applications. The plant-type infrared analyzers dealt with in this paper contain no moving parts ( I , $ ) . They can be schematically represented, as in Figure 1,to consist of an infrared energy source, a sample cell, an interference cell, a filter cell, a compensator cell, light trimmers, and a means of measuring the resultant infrared energy (detector). The trimmers are opaque members that can be adjusted to cast more or less shadow on the detector. The function of each of the cells is covered in the discussion below. The infrared energy from the source takes two paths, differing in that one path passes through the compensator cell and the other passes through the filter cell. I n order t o sensitize the analyzer to one component, it is necessary to place compounds or mixtures of compounds in the sensitizing cells (interference, compensator, and filter) and adjust the trimmers so that the difference in path energies reaching the detector will be proportional to the concentration of the key component in the sample cell, regardless of any changes in the relative concentrations of the background components. Thus, the sensitizing phase of the application work primarily involves determining what compounds to place in the sensitizing cells and what the trimmer settings should be. Because infrared equipment manufacturers either do not have customer application laboratories or have such laboratories but not the facilities to handle the light hydrocarbon applications which are most prevalent in the refineries, most of the sensitizing work has to be done by the users in their laboratories. To date, the largest number of refinery nondispersive infrared analyzer measurements have been made in the gas phase at essentially atmospheric pressure. This is because most current refinery applications involve light hydrocarbon streams which are gaseous a t atmospheric pressure and the measuring temperature (usually 130' F. t o 150' F.),and in general, commercially available sample cells are not debigned t o withstand pressures 1378

higher than about 1 atmosphere. Although the following discussion on sensitizing technique is for gas analysis, the same technique is used in liquid analysis. The first step in sensitizing an infrared analyzer for a particu!sr analysis involves the selection of the proper material t o use as cell windows. Because ordinary window glass absorbs practically all of the energy in the infrared spectrum, ite use is precluded for most infrared applications. Materials such as quartz, calcium fluoride, rock salt, and silver chloride succes-. sively absorb less energy in the infrared spectrum than window glass, and all are useful as cell window material. Consequently,. a viindow material should be selected that transmits in the region where the infrared absorption is most specific t o the key component. A simple ieview of the component spectrograms (infrared absorption-wave-length graphs) will often indicate which cell window material t o use for the particular analysis. Other practical aspects must be considered in choosing window material such as the effects of pressure, moisture, sulfur compounds, and thermal shock. The next step in sensit,izing a nondispersive infrared analyzer for a particular service involves the construction of response patterns. As an example, consider the case when thwsensitizing cells and the sample cell contain a gas, such as nitrogen, which absorbs essentially no infrared energy. A change in the setting of one of the trimmers will cast more or less shadow on one of the infrared energy paths, resulting in a change in the signal output of' the detector. If pure gases which absorb in the infrared are subsequently passed through the sample cell, and detector output signal measurements are made for various trimmer settings, then the resultant data may be related to the nitrogen data and plotted as in Figure 2. Thus, the line for nitrogen is a t 45" in Figure 2, while the lines for absorbing gases A , B , C are at lesser angles. G A S I Y SAMPLE CELL 08

2

,NITROGEN.

k-

07

1 ATM.

/

NITROGEN IN A L L SENSITIZING C E L L S

I AIM.

I ATM.

5

I AIM.

04

I

0 0

O C

0

02

03

04

05

06

07

I

08

O P T I C A L T R I M M E R SETTING ( I N TERMS OF DETECTOR SIGNAL FOR N p l

Figure 2.

Infrared Analyzer Response Pattern

This type of plot is called a response pattern. It is constructed by plotting the signal for nitrogen at equal units from the two axes. The absorbing gas signal is plotted on a vertical line through the nitrogen point corresponding to the same trimmer setting. This is repeated a t different settings of the trimmer. In the extreme case with one path blocked 08 entirely, the large signals for absorbing gas and nitrogen measure the transmitted and incident radiation, the ratio of which is the fraction transmitted. This ratio is also the tangent of the angle the gas response line makes with the horizontal axis. Nitrogen, which does not absorb, has a ratio of unity and a response line a t 45'. For a n absorbing gas the ratio is less than unity, and its response line

I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY

Vol. 46, No. 7

PROCESS LNSTRUMENTATION is less than 45" as seer1 in Figure 2. The response pattern displays the relative opacity of absorbing gases from measurements easily made. The illustrative response pattern in Figure 2 shows that the trimmers can be set so that the absorbing gases are indistinguishable from nitrogen as well as from one another. G A S IN S A M P L E C E L L /NITROGEN,

t-

0.8

-I-

GAS X IN FILTER CELL, N t I N INTERFERENCE AND COMPENSATOR CELLS

0.7

/'

/GAS

L 2

I ATM

A, I ATM.

GAS E, I ATM. GAS C, I ATM.

C_ 0 . 5

/GAS

X,

I ATM

a z 0.3

0 0

K L J - L L A

0 0

01

0 2

03

OPTIC 14,

Figure 3.

0 4 0.5 0 6 07 pr tIVlER S E T T I N G GNAL FOR Ne)

0.P

Response Pattern for Infrared Analyzer Sensitized to Gas X

To sensitize the analyzer to gas X in the presence of a mixture of gases A , B, and C, the filter cell is filled with gas X. The resultant response pattern, as shown in Figure 3, is typical for gases which are not spec%roscopicallysimilar to gas X (noninterferers). Again the trimmers can be set to point where gases A , B, and C are indistinguishable from nitrogen as well as from one another. At the same time an appreciable signal is obtained for gas X. The differential signal represented by RS is an indication of the available signal for 0 to 100% gas X measurement. In cases where gases of a mixture have similar absorption spectra, such as hydrocarbon gases of a homologous series, a response pattern will indicate that a point Qr region of minimum interferenee is not possible by trimmer adjustment alone. Such a pattern is shown in Figure 4. Gases B and C yield detector signals as if they were partially like gas X. However, by placing the proper amount and proportion of gases B and C in the compensator cell the response pattern of Figure 4 may be altered to resemble that of Figure 3. Most refinery hydrocarbon applications yield response patterns where the optimum amount and ratio of gases B and C in the compensator cell still does not make the analyzer entirely specific to gas X. Such a response pattern is shown in Figure 5. In this example gas A shows a slight negative interference. This may be minimized by placing gas A in the interference cell or possibly by adding a small amount of gas A to gas X in the filter cell. Up to this point the background gases have been passed through the sample cell individually and a t a pressure of 1 atmosphere. For a particular application, however, gases A , B, and C are in combination and in varying concentrations. Consequently, the next phase of the sensitizing procedure is to obtain calibration curves on the sensitized analyzer by using a t least three sets of samples, each containing varying amounts of gas X in the ranges to be expected in plant operations. The three sets of samples should contain background gases A , B, and C representing high, normal, and low concentrations of the interfering gases expected in plant operations The resultant calibration curves (detector signal versus concentration of gas X ) for the various blends should, for optimum results, be coincidental. The calibration curves show wide diJuly 1954

vergence in some applications, however, and further work is then required. When the calibration curves show wide divergence, the analyzer is said to have poor selectivity. The previous discussion on determining the optimum amounts and ratios of gases to include in the sensitizing cells was directed toward making the equipment selective to gas X. A comparison of Figures 4 and 5 illustrates, however, that the price t o be paid for increased selectivity is a decrease in available detector signal ( R S ) . Similarly in Figure 6, the calibration curves for the various blends can be made to converge (increased selectivity) by decreasing the length of sample cell or by increasing the length of the interference cell, but the sensitivity of the equipment is correspondingly decreased. Only a t this point in the sensitizing and calibration procedure can it be determined if the analysis can be made t o the desired degree of accuracy with commercially available equipment. A decision must often be made either to relax accuracy requirements or to abandon the application. This is a general outline on sensitizing techniques that can be followed for any analysis. In practice, however, previous experience in sensitizing nondispersive infrared analyzers is often drawn on to simplify the procedure. For instance, most refinery applications involve as many as 10 background components in varying concentration ranges. Consequently, a typical blend of all the background components is often placed in the interference cell; gas X is placed in the filter cell, and the response pattern technique is then used to determine which interfering gases and how much of each should be placed in the compensator cell. Past experience regarding known interfering gases for particular analyses often dictates with fair accuracy the gases to place in the compensator cell. GAS IN S A M P L E CELL

N e I N INTERFERENCE

AND COMPENSATOR CELLS

T

A,

I ATM

8. I ATM

/ /

0 6 c

C, I ATM. X . I ATM.

I

0 2

03

04

0 5

0.5

07

08

O P T I C A L TRIMMER S E T T I N G (IN TERMS OF DETECTOR SIGNAL FOR Npl

Figure 4.

Infrared Analyzer Response Pattern for Interfering Gases

This modified technique was actually used in sensitizing a nondispersive infrared analyzer to measure 0 to 10% isobutane in a stream containing the following concentration ranges in background components. Background Component, Mole % ' n-Butane Isobutylene 1-Butene 2-Butene Isopentane n-Pentane

Blend A , Min. Olefins 68.4

2.0 7.9 19.7 1.0 1 .o

Blend B,

Av. Olefins 54.3 3.0 10.7 30.0 1.0 1 .o

Blend C, Max. Olefins 37.5 4.0 16.1 40.4 1.0 1.0

A review of the spectrograms of the components of the stream did not indicate which window material would be best for this isobutane analysis. This was due to the number of background

INDUSTRIAL AND ENGINEERING CHEMISTRY

1379

ENGINEERING, DESIGN, A N D PROCESS DEVELOPMENT gases, their wide concentration ranges, and their high absorption Calcium fluoride windows were selected because of their successful use in another C4 analysis. The rigorous response pattern approach t o the measurement problem was not used t o sensitize the analyzer in the laboratory because of the considerable effort and expense required t o obtain response patterns for the individual pure components. However, reaponse patterns were taken using blends A , B, and C to establish the instrument conditions for maximum selectivity to isobutane. As a result, the analyzer was sensitized to isobutane by GAS IN S A M P L E C E L L

/

GAS X IN FILTER C E L L , NITROGEN IN INTERFERENCE

NITHOGEN, I ATM

I

00 00

01

I

I

l

l

I

l

1

02

0.3

04

05

06

07

OB

resultant signal level is h w (below 0.5% of the detector output signal when one path is blanked off). It has been found in the field and in the laboratory that commercially available nondispersive infrared analyzers do not have the necessary stability or sensitivity for these low signal strength applications. I n general, the infrared analyzer manufacturers are reluctant to perform the necessary improvement work on their equipment because the bulk of their present sales is for higher signal strength applications. Consequently, this improvement work has, of necessity, been taken on by the users in theii laboratories. The stability of one type infrared analyzer, for instance, was improved tenfold with relatively simple modifications of the equipment since adopted by the manufacturer. The experiences gained through this improvement work indicate that the manufactureis should spend more time in the development of their infrared analyzers. Almost every refinery infrared analyzer application requires that the equipment be explosionproof. Considerable difficulty was experienced in obtaining this feature, because the manufacturers often encountered serious stability and circuit design problems when they tried to fit their equipment t o standard explosionproof housings. Most of these problems have been solved.

OPTICAL TRIMMER SETTlhG

//r

( I N T E R M S OF DETECTOR SIGNAL FOR N t l

Ih BLEND 4 OR E

Figure 5. Infrared Analyzer Response Pattern Demonstrating Use of Compensator Cell

placing 1 atmosphere of isobutane in the filter cell, 1 atmosphere of the average olefin background gas (blend B ) in the interference cell, and 413 mm. of ,n-butane and 317 mm. of nitrogen in the compensator cell. All gas pressures were measured a t the thermostatic t,emperature of the analyzer. The optimum amount of n-butane to place in the compensator cell was determined from response patterns for various pressures of n-butane, using blends, A , B, and C. The response pattern for the condition where the compensator cell contained 413 mm. of n-butane resembled Figure 3 very closely. This meant that an instrument condition had been established where variations of the components of the carrier gases (blends A , B, and Cj did not change t.he instrument reading a t bhe zero end of its range (0% isobutane). Calibration was now made using various concentrations of isobutane in blends A ; B, and C to confirm the selectivity of the sensitized analyzer over the 0 t o 10% isobutane range. The results are shorm in Figure 6 for various sample cell lengths. Since the 0 to 10% isobut,ane measurement was dcsired with an accuracy of &0.2% isobutane, the l'/Z-inch sample cell was chosen as adequate for this analysis. Additional experiments on the sensitized analyzer showed that the differential energy of the two paths reaching the detector for full-scale corresponded to 0.2% of the energy in one path.

Specifying Equipment The sensitizing and calibration work was shown to yield information regarding the detector signal available for a particular analysis (Figure 6), and the cell lengths and window material to use. This information should be witten into the specifications for the infrared analyzer, together with its stability requirements, usually expressed ae a per cent of the expected minimum detect,or signal. The discussion on the sensitizing work pointed out that in order t o obtain suitable selectivity for moet refinery applications the

1380

/IN

0

0

1

I

1

2

S C 'SAMPLE CELL I C'INTERFERENCE CELL

BLEND C

1 3

1

4

I 5

6

!

7

j

0

1

9

1

10

MOL % ISOBOTANE

Figure 6.

Effect of Cell Lengths on lnstrument Selectivity

The problem of driit in low signal strength applications can sometimes be solved by including automatic standardization in the analyzer if the drift is not in one direction (stray thermal effects can often be eliminated in this manner). However, leaky c~lls,unstable electronic components in the equipment, or improperly designed detector circuits can cause unidirectional drift. In these cases automatic standardization is not the solution to the stability problem. Consequently, a thorough investigation of the nature of the drift must be made before automatic standardization is specified. At the present time it is necessary to specify each infrared analyzer for each application, but purchase specifications must be relaxed in some cases, and the necessary improvement work must be done by the user. The fact that improvement work must be done by the consumer currently precludes the use of infrared analyzers in many small companies. In the specification for an analyzer to measure 0 to 10% isobutane, it became evident, when the available signal and accuracy requirements for the analysis were compared to the stability specifications cited by infrared analyzer manufacturers, that improvement work was required on any of the commercially available analyzers that were investigated Since the analysis was of particular importance for good process operations, the improve-

INDUSTRIAL A N D ENGINEERING CHEMISTRY

Vol. 46, No. 7

PROCESS INSTRUMENTATION ment work was undertaken by the user. The tenfold improvement in stability cited resulted in an analyzer that was suitable for the service.

Installation The success of any continuous analyzer installation, aside from the suitability of the analyzer itself, depends on good sampling techniques. Obtaining B representative sample with small time lags is considered routine t o the applications engineer and will not be discussed here. However, in using infrared equipment in refinery applications, special steps must be taken: 1. The sample must be dry. 2. The sample must be free from solid particles, even as small as 0.2 micron in size (porous metallic filters have proved very valuable in this regard); 3. The presence of even minute amounts of sulfur compounds has precluded the use of silver chloride cell windows for most refinery services. 4. In many applications the infrared analyzer is sensitive enough to sample flow rate changes requiring that some means of flow control be provided.

For instance, in one application it was found that an 8% change in sample flow rate (normal = 15 cc. per minute) caused a 1% change in recorder reading. Nondispersive infrared analyzers are usually thermostated a t 130" to 150" F. because of instrument stability considerations. Thermostating is done in this temperature range because, it is desirable to maintain the temperature above the maximum expected ambient temperature, and special, more expensive electronic components are required for temperatures above 150" F. In the field, a shed or house is required for these analyzers in order to obtain good temperature control and to work on the equipment in inclement weather. Further, heating facilities are often required for winter operations to achieve good thermostating. These requirements are often costly in terms of per cent of analyzer cost, but they may mean the difference between success and failure for the application.

Operation

It is almost impossible to duplicate plant conditions in the laboratory. Plant operations have shown that spurious signals are always present because of the operation of large electrical equipment which can cause serious stability problems in the infrared equipment, especially for low signal applications. This problem was formerly laid to pickup in the lead lines from the analyzer to the control house recorder, and special techniques of shielding were attempted. Tliis gave only slight stability improvement. As a result of poor stability encountered with the isobutane analyzer and other low signal strength applications, improvement work on one type of infrared analyzer was undertaken. The results indicated that the basic stability problem did not originate in the lead lines but in the detector circuit. Good instrument stability also depends on how well the cell windows can be sealed. Plant operations with the isobutane and other analyzers have shown that greater care must be taken by the infrared analyzer manufacturers in the selection and application of the cell sealing materials. Good initial results have been obtained as a result of general improvement work by using a special high polymer hydrocarbon for the cell sealing material, but more work is indicated. Maintenance The successful installation, operation, and maintenance of this type equipment requires abilities not always available in plant

July 1954

instrument departments. This presents a major problem in the field of continuous analyzer applications. The infrared analyzer is but one example of unconventional equipment which requires high skill in the refinery instrument departments. However, the shortage of technical personnel in industry has resulted in a decrease in the number of technical and semitechnical personnel involved in refinery measurement and control work during the paet 10 yeam, while the demand for such help has shown a broad increase. This problem is now being partially solved by increased attention to this field. In addition to assistance from the manufacturers an average of two man-months of technical work is required before an infrared analyzer installation can be turned over to the instrument department for routine maintenance and service. Moreover, it is necessary to have technical personnel act as consultants for these installations, should the instrument mechanic be faced with maintenance or service problems that he cannot solve. Thus, the successful application of infrared analyzers to refinery streams depends to a large extent on the availability of technical and semitechnical personnel for this work.

Conclusions 1. The response pattern technique of sensitizing nondispersive infrared analyzers permits this task to be performed in an orderly manner. 2. I t would be of considerable advantage to the manufacturers and users of nondispersive infrared analyzers if response patterns for common hydrocarbons were run on the available types of nondispersive analyzers. I n this manner particular analyses could be worked out more quickly and efficiently than i E now possible. This would be comparable to the use of spectrograms for sensitizing dispersive infrared analyzers. 3. -4common basis for expressing the sensitivity of commercially available nondispersive infrared analyzers would be useful. Some manufacturers presently use the sensitivity of carbon dioxide measurement in nitrogen as their standard, but this is unsuitable for the petroleum refining industry. Isobutane in nbutane is suggested as a good standard sensitivity test. 4. Much improvement work is required to increase the sensitivity and stability of plant-type nondispersive infrared analyzers. The manifold improvements that were obtained would indicate that the equipment manufacturers are in general not aware that such possibilities exist or that future refinery applications will require equipment having higher stability and sensitivity. 5. The sensitizing procedure described in this paper pointed out the need for many accurately analyzed samples. The job of obtaining these samples is both expensive and time-consuming. Infrared analyzer manufacturers should increase their efforts to organize customer applications laboratories, so that this versatile continuous analyzer can gain more widespread use in industry.

Acknowledgment The authors wish to acknowledge the work that J. J. Heigl and J. A. Hinlicky, Standard Oil Development Co., have done in establishing the sensitizing techniques outlined in this paper.

Literature Cited (1) Fastie, W. G., and Pfund, A. H., J . Opt. Soe. Amer., 37, 7 6 2

(1947). (2) Patterson, W. A., Chem. E n g . , 59,9,132 (1952).

RECEIVEDfor review September 7, 1953.

INDUSTRIAL AND ENGINEERING CHEMISTRY

ACCEPTEDMarch 13, 1951.

1381