WO1993020429A1

WO1993020429A1 - Infrared analyzer for determining the concentration of multiple chemical components in a gas or liquid

Info

Publication number: WO1993020429A1
Application number: PCT/US1993/002991
Authority: WO
Inventors: Leslie M. Tack
Original assignee: Altamont Technologies, Inc.
Priority date: 1992-03-30
Filing date: 1993-03-30
Publication date: 1993-10-14
Also published as: AU3971393A

Abstract

An apparatus for determining the concentration of multiple contaminants in a liquid such as the oxygenates MTBE and ethanol in gasoline. The component concentrations are determined by measuring the transmission of infrared light through the liquid. The transmission characteristics (VDref, Vd1, Vd2) are used by a computer (5) to determine absorption according to the Lambert-Beer law. Nonlinear terms are then included by a computer (5) in the calculation of concentrations using the absorbances in a two step process. First the component concentrations are estimated by using only linear components in solving the matrix equation: A = BX. Second, the estimated values for the concentration components are used as initial values in the function (A-BX)T (A-BX) = 0 including nonlinear components and a conjugate gradient technique is applied to find the minimum value for X in the function. The device is portable and outputs results in percent volume enabling us by nontechnical personnel.

Description

INFRARED ANALYZER FOR DETERMINING THE CONCENTRATION OF MULTIPLE CHEMICAL COMPONENTS

IN A GAS OR LIQUID

BACKGROUND

1. Field of the Invention

This invention relates to instruments which determine the chemical concentration of one or more components of a chemical compound. More particularly, the invention relates to an infrared spectroscope applied to determine the concentration of multiple oxygenates in gasoline.

2. Description of Related Art

Legal mandates on the composition of commercially sold liquids such as gasoline have created the need for analyzers that can rapidly determine the chemical components of the liquid in the field as well as the laboratory. For instance, current and impending oxygenated fuel mandates in the U.S. will require oil companies to verify the oxygenate content of gasolines before distribution. State and federal regulations concerning allowable levels of oxygenates in gasoline have created a narrow window for specification. Since exchange of gasoline through a common pipeline is a frequent event for oil companies, a low cost method to verify oxygenate levels is necessary.

Prior methods for quantification of liquids are slow, require a trained operator, and are not amenable to field use. For instance, practical time constraints do not allow the typical turn around times for sending a gasoline sample for gas chromatograph analysis to assure gasoline meets the Federal and State mandates for MTBE content.

More recently, analysis of liquids using infrared devices has shown promise in providing an alternative to gas chromatography with comparable accuracy. Use of infrared spectroscopy to determine the concentration of MTBE in gasoline was reported in 1983 by Fry et al. in Determination of Methyl Tert-Butyl Ether in Gasoline by Infrared Spectrometry. American Chemical Society Anal. Chem V55 N.2 407-8 February, 1983. Improvements upon measurement of MTBE content in gasoline using infrared spectroscopy has also been reported by Brzobohata et al., in Determination of alcohols and ethers in gasolines by IR spectrometry. Ropa Ujlie, 27(5), 307-12, Czechoslovakia, (1985) . Fry et al. and Brzobohata et al. are incorporated herein by reference.

Currently, infrared spectrometry is used to determine the concentration of a component in a liquid by first measuring the transmission of infrared light through the liquid. Using transmission measurements the absorbance is measured by applying the Lambert-Beer law:

(1) Abs=log¹⁰(T₀/T)

where:

Abs is the absorbance at the analyzing wavelength of the liquid being tested; T₀ is the transmission through a reference liquid; and

T is the transmission through the sample being tested. Once measured the absorbance is converted into a volume percent (vol%) measurement using a graph, or the linear equation:

(2) C=B * Abs

where:

C is the component concentration in vol%;

B is a correlation coefficient obtained by linear regression or plotting using absorbance measurements from several liquids for which the concentration of the added component is known; and

Abs is the absorbance of the liquid under test for which the concentration of the added component is unknown.

Fry et al. reported using a process to measure the MTBE content in gasoline in which a linear calibration plot was made using multiple standard samples of MTBE in gasoline over the range of 2 to 20 vol %. Fry et al. used the linear calibration plot to yield the correlation coefficients with which to measure samples of an unknown vol %. Fry et al. claimed a relative error of less than 2 % at an MTBE level of 7.2 vol %.

Brzobohata et al. also reported use of linear calibration curves and the Lambert-Beer law for measurement evaluation of the content of MTBE in gasoline. Brzobohata et al. further used the method for a simultaneous determination of tert-amylmethy1 ether, and methanol in gasoline.

SUMMARY OF THE INVENTION One object of this invention is to provide an instrument which uses infrared spectrometry to measure component concentrations in a liquid so that the device can be small, portable and usable at a field sight where fuel is stored.

It is another object to use a measurement technique which uses nonlinear components to improve measurement accuracy.

It is another object to simultaneously determine the concentration of multiple components in a liquid.

It is another object of this invention to provide an apparatus which is simple enough to be used by nontechnical personnel.

The above objects are met with the present invention of an apparatus for determining the concentration of two or more chemical components in a liquid. The instrument includes an infrared sample analyzer, a computer, keyboard, LCD display, and power supply. The apparatus is portable and provides necessary calculations within the computer to output data in volume percent (vol%) enabling use of the instrument by nontechnical personnel. The apparatus can be operated by injecting a liquid sample into a cell that places the sample in the path of a broadband infrared beam. The broadband beam is divided into separate wavelength components by an optical beamsplitter. Each independent beam is condensed onto a separate infrared detector fitted with a narrow bandpass filter centered on an analyzing wavelength. The filter outputs analog transmission measurements which are converted to digital and read by a computer. The computer calculates absorbances from the transmission data using a modified Lambert-Beer law. From the absorbances, the computer then calculates and outputs the concentration of components in the liquid.

The analyzer is calibrated using a set of standards with contaminants of known composition. Recording the transmission measurements at each analyzing wavelength for each standard provides calibration coefficients necessary for calculating concentrations of unknown composition.

For two components, to determine the component concentrations, the following equations must be solved:

(3A) a, = b_0;0 + b₀₎₁x, + b_0>2x₂ + ^x,² + b^ _j ²; and (3B) a₂ = b_lj0 + b_M j + b₁₎₂x₂ + b_1?3x,² + b x₂ ²

where : aj and a₂ are measured absorbances; b_y are calibration coefficients; and i and X₂ are the concentration components.

If the nonlinear terms in equations (3A) and (3B) are ignored, at component concentrations of 10 vol % or higher error will be on the order of 10 %. Nonlinear terms are therefore included in the solution of equations containing multiple components in a two step process. First, the component concentrations are estimated by using only linear components in solving the matrix equation:

(4) A = BX

where:

A is the n component vector containing measured absorbances;

X is the n component vector containing the concentration solution; and

B is the nxn matrix containing known calibration coefficients. Second, the estimated values for the concentration components are used as initial values in the function

(5) (A - BX)^T (A - BX) = 0

including nonlinear components and the conjugate gradient technique is applied to find the minimum value for X in the function.

BRIEF DESCRIPTION OF THE DRAWINGS Further details of the present invention are explained with the help of the attached drawings in which: Fig. 1 shows a block diagram of the apparatus of the present invention; and

Fig. 2 shows the components of the infrared sample analyzer of Fig. 1.

DETAILED DESCRIPTION

Fig. 1 shows a block diagram of the apparatus of the present invention. The apparatus includes an infrared sample analyzer 1, an Analog to Digital (A/D) Converter 3, a computer 8, a keypad 9, an LCD display 7 and power supply 11. The infrared sample analyzer l can be made from the LAN Liquid Process Stream Analyzer made by General Analysis Corporation, South Norwalk, CT. The computer used is 386 SX single board computer, but a single board computer such as the Z-World Little Giant LGX containing an LCD display and keypad can also be used for development purposes. Standard commercially available components can be used for the remaining elements of the apparatus such as an 8IPC keyboard using alphanumerics, a Metrobyte DAS-8 12-bit A/D converter and a PC style laptop LCD display. Components are chosen so that the device can weigh less than 20 pounds, and have a total of less than 19 inches x 15 inches x 7.5 inches enabling the device to be portable.

The computer 5 is programmed so that the apparatus operates as described below. Initially, the infrared sample analyzer 1 produces voltage signals proportional to the amount of light transmitted through a liquid. The infrared sample analyzer 1, shown in Fig. 1 to be capable of testing two components of a liquid, produces an analog voltage signal reference signal V_mf and two analog voltage signals V_al and V^ proportional to the transmission at two selected analyzing wavelengths, -fe analog voltage signals are digitized by the A/D converter 3 and stored as a digital values V_dref, V_dl, and V^ in computer 5.

The computer 5 calculates absorbances from the transmission measurement signals and uses the absorbance values to calculate the concentration of the two components in the liquid. The computer calculations shield the user from the details of calibration and multicomponent analysis. The user interface consists of keypad 9, LCD display 7, and a port for injecting the liquid (described later) all enabling easy operation by inexperienced and nontechnical personnel. Power for the computer 5 and infrared sample analyzer 1 is provided by power supply 11.

Fig. 2 shows the components housed in the infrared sample analyzer 1 of Fig. 1. A broadband source of infrared energy is produced by a hot wire source 21 and focused by a pair of mirrors 23a and 23b through a sample cell 27. The beam is chopped near its focus point by a beam chopper 25 at approximately 10 Hertz.

The chopped beam is focused into sample cell 27. The beam enters and travels through crystal 29. The beam also travels through the liquid 31 as shown by the large arrows. The liquid is injected into a central tube which branches into input tubes 31a. The liquid can be expelled through output tube 31b. The diverging beam exiting the sample cell 27 is condensed onto a beamsplitter 35 by a set of mirrors 33a and 33b. The beamsplitter 35 will divide and direct the broadband beam to three spatially separated detectors 37a, 37b and 37c. Each detector 37a-37c is fitted with a narrow band interference filter centered on an analyzing wavelength so only a selected analyzing wavelength is detected. The three filters enable three wavelength ranges to be detected simultaneously. To measure the MTBE and ethanol content of gasoline, the detector for MTBE would be centered at 1205 cm^"1 and the detector for ethanol would be centered at 1050 cm^"1. To isolate the filter output including absorption components from the base gasoline or the other components, narrow bandpass filters should be used that have bandwidths of approximately 15 cm^"1 full-width-at half-maximum and over 80 % transmission at the center frequency.

The filters output analog transmission signals which are converted to digital by the A/D Converter 3 and read by a the computer 5.

The computer 5 then begins the determination of concentration by using the voltage transmission measurements V_dref, V_dι, and V_d2 to calculate the absorbance of the liquid at the selected analyzing wavelengths. The absorbance is measured by applying the Lambert-Beer law:

(6) A(λ)=log¹⁰(T_o/T)

where:

A(λ) is the absorption in decibels (dB) of the liquid over wavelength range λ;

T₀ can be defined in one of two ways: (1) T₀ can be the transmission through the liquid over a reference wavelength range λ₀ where transmission is high for all possible samples which can be tested by the analyzer; or (2) T₀ can be the transmission through the cell containing a reference sample of known composition over a wavelength range λ, the reference sample usually having a high transmission over the analyzing wavelength range λ; and

T is the transmission through the liquid over the analyzing wavelength range λ.

Once measured, the analyzer absorption measurements must be converted to a vol% measurement, enabling a layman to operate the analyzer and quickly understand the results. Absorption measurements must therefore be used in a mathematical equation which can be solved by a computer to determine concentration in vol%.

For the simple case of one component in a sample, the relationship between concentration and absorbance can be expressed as:

(7) C=c₀ + CjAbs + CjAbs² + ... + C_πAbs¹¹

where:

C is the concentration in arbitrary units such as vol%, the units of concentration being arbitrary since coefficients in a polynomial expansion will always relate absorbance to concentration; c₀, c_x ... c_n are constant coefficients which are obtained as described below; and Abs is the absorbance of the component of the liquid that absorbs radiation over the analyzing wavelength range λ according to the Lambert-Beer Law.

The coefficients, c₀, c_x ... c_n, are obtained by measuring the absorbance Abs of multiple samples with known concentrations C. For example, if we wanted to determine coefficients, c₀, c_r ... c_n, for ethanol in a mixture of ethanol in gasoline, we would prepare a series of solutions of ethanol and gasoline over the concentration range of ethanol we wanted to quantify. The absorption measurements from the known solutions would then be used in a linear regression technique to determine the coefficients. A computer algorithm for linear regression can be found in Bernardin, C/Math Toolchest. Mix Software , Inc. & Pete Bernardin Software, 1991. Bernardin is incorporated herein by reference. Once the coefficients are known, the concentration of ethanol in compositions which are unknown can be determined by measuring the absorbance and calculating the concentration C using equation (7) . For component concentrations in the 0-30% range by volume, equation (7) up to and including the second order term is usually sufficient.

Analysis to determine the concentration of more than one component in a solution is more complex. F o r instance, to quantify a sample with two components, it is required to know the absorbance at a minimum of two wavelengths. The total net absorbance at each analyzing wavelength is given in terms of a polynomial:

( 8A) a, = b_0>0 + b₀₎₁x, + b₀₎₂X₂ + b₀₎₃x,² +

and ( 8B) a₂ = bι_>0 + b₁₍₁x, + bι_;2x₂ + b_1>3x,² + b,_j4x₂ ²

where : a_! and a₂ are the absorbance at analyzing wavelength's 1 and 2 respectively. It is assumed that components 1 and 2 will dominate the net absorbance at the judiciously chosen wavelengths; by are known calibration coefficients obtained via linear regression of measured absorbances against samples of known composition; and X_j and x₂ are the component solutions Again, only up to second order equations are used, assuming that concentrations are less than 30% by volume.

Since (8A) and (8B) represent a set of nonlinear equations, an exact solution is no longer possible. For component concentrations over 5 % by volume, second order terms in concentration become significant. In fact, for concentrations of 10% by volume, if nonlinear terms are ignored, error may be as high as 10%. Therefore, most applications will require a solution to the nonlinear problem.

A solution to the nonlinear problem is obtained through a two step process. First, initial values for a component vector X are obtained in the following equation by neglecting all terms higher than first order in X.

(9) A = BX

where:

A is the n component vector containing measured absorbances which constrain the solution; X is the n component vector containing the concentration solution which quantifies the sample; and

B is the nxn matrix containing known calibration coefficients for selected sample components.

The solution for X is found by matrix inversion of B:

(10) X = B-^χ Second, a function minimization routine is employed to improve the initial values for X until convergence is reached.

To create a function minimization routine, we define the following function:

(11) (A - BX)^T (A - BX) = 0

Equation (11) can be solved using the initial values for X from equation (10) which are input into a function minimization routine. Minimization routines that can be employed include the least squares technique and the conjugate gradient technique. A computer algorithm for the least squares technique and the conjugate gradient technique can be found in Bernardin, C/Math Toolchest.

The computer program for the conjugate gradient technique requires the least amount of computer memory.

The conjugate gradient technique finds the minimum of a function by following the surface defined by the function along the path of steepest decent.

The two step procedure defined above to solve nonlinear equations will work as long as the linear and nonlinear terms in component concentration contribute to the total absorbance with similar magnitude. In other words, the nonlinear terms should not significantly outweigh the linear terms. Also, the absorbance behavior of different components should be highly correlated for all analyzing wavelengths.

Although the invention has been described above with particularity, this was merely to teach one of ordinary skill in the art how to make and use the invention. Many modifications will fall within the scope of the invention, as that scope is defined by the following claims. For instance, the infrared sample analyzer 1 could be expanded to detect more than two components. More traditional infrared benches such as multiple filter wheel analyzers could also be employed. Additionally, output results could be expressed in weight percent rather than vol % when the density of the sample is also determined.

Although the device is described primarily for use in measuring the oxygenates in gasoline, the apparatus may also be used for determining other components in gasoline or diesel fuels such as alcohols and aromatics. The invention may also be used to identify toxic components of liquids, or to quantify flowing liquids in on-line process control and monitoring applications.

Claims

What Is Claimed Is:

1. An apparatus comprising:

an infrared source transmitting an infrared input light;

a sample cell capable of containing a liquid with a plurality of contaminants, the sample cell receiving the infrared input light, and transmitting the infrared input light .through the liquid to create an output signal;

a processor which converts the output signal into a plurality of concentration values taking into account nonlinear concentration components in a mathematical formula to determine the plurality of concentration values, the mathematical formula being solved using a computation solution; and

means for outputting the plurality of concentration values.

2. The apparatus of claim 1 wherein the means for outputting the plurality of concentration values is an alphanumeric display.

3. The apparatus of claim 2 wherein the alphanumeric display is an LCD display.

4. The apparatus of claim 1 wherein when the plurality of contaminants comprises two contaminants, the mathematical formula comprises: a, = b_{0>1 !} + b_{0j2 2} +

+ b_0ι4X₂ ²; and a₂ = b_wx, + b_lι2x₂ + b_I)3X_! ² + b_lι4x₂ ²

where a and a₂ are measured absorbances obtained from the output signal, by are calibration coefficients, and

X_j and _j are the plurality of concentration values.

5. The apparatus of claim 4 wherein the calibration coefficients are obtained using linear regression applied to measured absorbances of samples of known composition.

6. The apparatus of claim 1 wherein the computation solution is found by

(a) finding an estimated solution for X by solving a formula A = BX using only linear components

where

A is a component vector containing absorbances obtained from the output signal,

B is a matrix containing calibration coefficients, and

X is a component vector containing the plurality of concentration values, and

(b) finding the plurality of concentration values by finding minimum values for X in the function

(A - BX)^T (A - BX) = 0 using nonlinear components and using the estimated solution for X as initial values in a function minimization routine.

7. The apparatus of claim 6 wherein the function minimization routine is a conjugate gradient technique.

8. The apparatus of claim 6 wherein the calibration coefficients are obtained using linear regression applied to measured absorbances of samples of known composition.

9. The apparatus of claim l further comprising an alphanumeric keyboard which controls the processor.

10. The apparatus of claim 1 wherein the frequency of the infrared source is from 2.5 to 25 microns.

11. The apparatus of claim 1 wherein weight of the apparatus is less than 20 pounds.

12. The apparatus of claim 1 wherein the size of the apparatus is less than 20 inches square.

13. An apparatus comprising:

an infrared source transmitting an infrared input light;

a sample cell capable of containing a liquid with a plurality of contaminants, the sample cell receiving the infrared input light, and transmitting the infrared input light through the liquid to create an infrared output signal; a beamsplitter dividing the infrared output signal into a plurality of separate wavelength components;

a processor which converts the plurality of separate wavelength components into a plurality of concentration values taking into account nonlinear concentration components in a process to determine the plurality of concentration values; and

means for outputting the plurality of concentration values.

14. The apparatus of claim 13 wherein a first transmission value is measured from a given one of the plurality of separate wavelength components and a second transmission value is measured from another one of the plurality of separate wavelength components where transmission is high for the plurality of contaminants and an absorbance is calculated using the formula

Abs=^loε10(T₀/T)

where

Abs is the absorbance at the given one of the plurality of separate wavelength components,

T₀ is the second transmission value, and

T is the first transmission value.

15. The apparatus of claim 13 wherein a transmission value is measured from a given one of the separate wavelength components and an absorbance is calculated using the formula Abs-log¹⁰ (T_o/T)

where

T₀ is a transmission value through a sample with a known concentration of contaminants, and

T is the transmission value through the liquid.

16. The apparatus of claim 13 wherein the processor comprises:

(a) means for finding an estimated solution for X by solving a formula A = BX using only linear components

where

A is a component vector containing absorbances obtained from the plurality of separate wavelength components,

B is a matrix containing calibration coefficients, and

X is a component vector containing the concentration solutions, and

(b) means for finding the plurality of concentration values by finding minimum values for X in the function

(A - BX)^T (A - BX) = 0

using nonlinear components and using the estimated solution for X as initial values in a function minimization routine.

17. The apparatus of claim 16 wherein the function minimization routine is a conjugate gradient technique.

18. The apparatus of claim 16 wherein the calibration coefficients are obtained using linear regression applied to measured absorbances of samples of known composition.

19. An apparatus comprising:

an infrared source transmitting an infrared input light;

a sample cell capable of containing a liquid with a plurality of contaminants, the sample cell receiving the infrared input light, and transmitting the infrared input light through the liquid to create an infrared output signal;

a beamsplitter dividing the infrared output signal into a plurality of separate wavelength components;

a plurality of detectors receiving the plurality of separate wavelength components and generating an analog electrical signal proportional to a magnitude of the separate wavelength components received;

an analog to digital converter which converts the analog electrical signal to a digital electrical signal;

a plurality of narrowband filters receiving the digital electrical signal and narrowing wavelength of the separate wavelength components; a computer which receives the digital electrical signal which contains transmittance measurements from the plurality of narrowband filters, converts the transmittance measurements into absorption measurements, and converts the absorption measurements into a plurality of concentration values using a mathematical formula which is solved using a computation solution wherein

(a) an estimated value for X is found by solving the formula A = BX using only linear components

where

A is an component vector containing the absorption measurements,

B is a matrix containing calibration coefficients for selected sample components, and

X is a component vector containing the concentration values, and

(b) the concentration solutions are found by finding minimum values for X in the function

(A - BX)^T (A - BX) = 0

using the estimated solution for X as initial values in a conjugate gradient technique; and

the apparatus further comprising:

an alphanumeric display for outputting the plurality of concentration values; a alphanumeric keyboard for controlling the computer; and

a power supply supplying power to the computer, and the infrared source.

20. The apparatus of claim 19 wherein weight is less than 20 lbs and size is less than 20 inches square enabling the apparatus to be portable.

21. A method for determining a concentration of a plurality of components in a liquid comprising the following steps;

transmitting an infrared input light through a liquid containing a plurality of contaminants to create an infrared output signal;

converting the infrared output signal into a plurality of concentration values taking into account nonlinear concentration components of a mathematical formula which is solved using a computation solution; and

outputting the plurality of concentration values.

22. The method of claim 21 wherein the computation solution comprises the steps of:

(a) finding an estimated solution for X by solving the formula A = BX using only linear components where

A is a component vector containing absorbances obtained from the infrared output signal,

B is a matrix containing calibration coefficients, and

X is a component vector containing the concentration solutions; and

(b) finding the concentration solutions by finding minimum values for X in the function

(A - BX)^T (A - BX) = 0

using the estimated solution for X as initial values in a function minimization routine.

23. The method of claim 22 wherein the function- minimization routine is a conjugate gradient technique.