WO1998002734A1

WO1998002734A1 - Distributed process control using imaging spectroscopy

Info

Publication number: WO1998002734A1
Application number: PCT/US1997/012202
Authority: WO
Inventors: Colin J. H. Brenan; Ian Hunter
Original assignee: Mirage Technologies
Priority date: 1996-07-15
Filing date: 1997-07-15
Publication date: 1998-01-22

Abstract

A system for providing real-time control of a process that is monitored at a plurality of process locations by spectrometric techniques. The system utilizes an imaging spectrometer having an optical image input and a spectral image output. A plurality of modules are employed; one module is disposed in each process location. Each module has an illumination source for illuminating such location; a collector for collecting light that has been scattered at such location and directing such light to a collector output; and an optical fiber arranged to provide optical communication from the collector output to a pixel location of the optical image of the imaging spectrometer. In addition an optical transducer is placed in communication with the spectral image output of the imaging spectrometer. The optical transducer provides as an output, spectral data, associated with each pixel location of the optical image input and therefore associated with each process location, for the purpose of process control.

Description

Distributed Process Control Using Imaging Spectroscopy

Technical Field The present invention relates to systems and methods for providing real time on-line control of processes, such as chemical processes, that are monitored using spectrometers.

Background Art Real time on-line process control has become increasingly important in the petroleum refining and blending industries to increase product quality, minimize production costs and fulfill government regulations regarding product composition and pollution emissions. One example of the issues presently faced by these industries are the new EPA guidelines contained in the 1990 Clean Air Act Amendments regarding fuel chemical composition. The EPA is most interested in the percentage of benzene and other aromatics in gasoline in that they stipulate a maximum of 1.6% volume benzene baseline for reaffirmed gasolines. In addition, certain geographic locations must use blends which contain oxygenated additives derived from renewable fuels such as ethanol. This trend in monitoring and tightly controlling the chemical composition of fuels is therefor a present necessity (with the commensurate economic impact) and can only be expected to increase in the future. Typical analysis of petroleum products during the production process involves performing gas chromatography on periodic samples. Although accurate, this process is time consuming, labor intensive and the substantial lag time can result in considerable costs when production errors do occur. A preferred method of analysis would involved real-time and in-line monitoring from a remote location. If accurate, such a monitoring method would allow intelligent process control.

Current spectrochemical analysis techniques combine mid-IR or near-IR absorbance/reflectance spectroscopy with multivariate regression analysis to make limited on-line quantitative analyses of chemical species in fuel. On; application is for the on-line determination of total aromatic concentrations as well as individual aromatic species' concentrations in petroleum fuels during the blending process. Although proven to be a viable on-line detection method for aromatic species, similarities between the near-IR absorbance spectra of many aromatic isomers (i.e., xylene) makes quantisation of individual isomers difficult when more than one isomer is present in significant concentrations.

Recently, Raman spectroscopy has been demonstrated as a viable alternative to mid- or near-IR absorbance/reflectance spectroscopic methods in quantitative assessment of petroleum sample composition. Advances in near-IR laser diode, photosensor and optical filter technologies has resulted in the construction of high optical throughput Raman spectrometers capable of low fluorescent Raman spectral acquisition from petroleum products. When combined with multivariate statistical analysis, these instruments have been shown to measure the gas oil cetane number and cetane index, the percenl fuel composition for liquid fuel mixtures of unleaded gasolines, super-unleaded gasoline and diesel fuels and the concentrations of benzene and other aromatic compounds. Raman spectroscopic analyses of aviation fuel have determined its general hydrocarbon makeup, aromatic components and additives. Summary of the Invention

In accordance with a preferred embodiment of the invention, there is provided a system, for providing real-time control of a process that is monitored at a plurality of process locations by spectrometric techniques. The system utilizes an imaging spectrometer having an optical image input and a spectral image output. A plurality of modules are employed; one module is disposed in each process location. Each module has an illumination source for illuminating such location; a collector for collecting light that has been scattered at such location and directing such light to a collector output; and an optical fiber arranged to provide optical communication from the collector output to a pixel location of the optical image input of the imaging spectrometer. In additicn an optical transducer is placed in communication with the spectral image output of the imaging spectrometer. The optical transducer provides as an output, spectral data, associated with each pixel location of the optical image input and therefore associated with each process location, for the purpose of process control. Related methods are also provided. The spectrometer may, but need not necessarily, be a Raman spectrometer. Indeed, a plurality of types of spectrometry may be employed simultaneously for implementing the process control.

Brief Description of the Drawings The invention will be more readily understood by reference to the following drawings, taken with the accompanying detailed description, in which:

Fig. 1 is a schematic of a preferred embodiment of system in accordance with the present invention; and

Fig. 2 is a schematic showing how a plurality of systems of a type such as shown in Fig. 1 may be used for control of more complex or more distributed processes.

Detailed Description of Specific Embodiments In accordance with a preferred embodiment of the present invention there is provided a system, for providing real-time control of a process that is monitored at a plurality of process locations by spectrometric techniques. The present embodiment, for example may be used, for example, for real-time in - line monitoring of fuel composition based on fiber-optic Raman spectroscopic detection and analysis from multiple points distributed at critical junctures in the process stream. The output of the monitoring system could then be used as one parameter for feedback control of the petroleum refining or blending process. The system design is self-contained and modular for flexible and easy reconfiguration or upgrade to satisfy new measurement requirements as they arise. A real-time computer-based data acquisition system acquires and processes the spectrochemical information before submission in a standardized format to the global process controller. By way of background, illumination of a molecular ensemble with a beam of monoenergetic optical photons having an energy outside any molecular absorption bands results in a large number of elastically (Rayleigh) scattered photons. A small fraction of the incident photons, however, are inelastically (Raman) scattered by inducing transitions between vibrotational molecular energy states in which the molecular polarizability and the final state wavefunction belong to the same symmetry group. Consequently, the scattered light spectrum consists of a dominant Rayleigh line at energy offsets equal to the energy gained or lost by the Raman scattered photons that corresponds to :he energy difference between the initial and final molecular states, raman line intensities are dependent on the magnitude of the derived molecular polarizability tensor and the number of molecules in the initial energy state.

Fundamental molecular parameters, like bond stiffness, as well as other molecular attributes such as molecular symmetry, primary and secondary conformations and side chain reactions, can be deduced from a Raman spe tral measurement while characteristic Raman spectral patterns or "fingerprints" aid in molecular identification. Raman spectral images can be interpreted either as a compositional map, or as a means to view changes in local chemical functionality or reactivity of a particular chemical species. For example, Raman spectral of neat trichloroethylene (TCE) and distilled water show Raman peaks indicative of their different molecular structure. The intense 1601 cm^"1 TCE Raman line arises from C=C double bond stretching modes while the dominant water Raman band centered at 3300 cm^"1 is from the O-H stretch vibration.

Confocal Raman spectroscopic imaging (and Raman spectroscopic imaging in general) is advantageous in many respects and, interestingly, spectroscopic imaging principles can be applied here. Compare the confoc al Raman spectroscopic and reflected intensity images acquired from a mixtL.re of two optically similar yet chemically distinct substances, potassium sulfate (K₂S0₄ and NaHC0₃. The dominant Raman spectral lines used to discriminate K₂S0₄ from NaHC0₃ are 960 cm^"1 and 1040 cm^"1, respectively.

In Fig. 1 are shown the two subsystems utilized in accordance with the embodiment referred to above: one for illumination and collection of the Raman scattered light from the sample (the Illumination/Collection (I/C) modules 11) and the second for Raman spectral analysis, namely, a Raman interferometric spectrometer 12. For the purposes of this figure, it is assumed that the sample resides inside a pipe 13 but in actuality the sample can be located anywhere accessible to an optical fiber bundle. Each subsystem is designed to be self- contained and modular for easy extension of the sensing network and for flexible reconfiguration to meet the changing measurement requirements of the user.

An I/C module 11 contains a miniature distributed Bragg reflector (DBR) near-IR laser diode operating at approximately 850 nm with an output power >50 mW to illuminate through a lens /window the liquid flowing in the pipe. A new technological development in diode lasers, the DBR laser is preferred over the more common index-guided diode laser because a DBR laser exhibits no mode hopping, shows no frequency hysteresis as a function of both temperature and current changes and emits substantially less broadband radiation. A near-IR wavelength laser diode is selected to minimize interfering fluorescent emission from the liquid and to take advantage of the maximal quantum efficiency in this wavelength range of the silicon photosensor that detects the scattered Raman light. Light backscattered from the liquid in pipe 13 is collected and input into a group of fused silica fibers 14 tapped from a larger fiber bundle 15 that acts as an optical "bus" to transmit the light from each module to the imaging Raman spectrometer 12. In the configuration depicted, the fiber bundled output illuminates through a cylindrical lens 16 the spectrometer after removal of the source illumination component by a laser line filter 17. However, a filter could also be located in each I/C module 11 to minimize induced optical fiber fluorescence if that proves to be problematic. The Raman spectrometer 12 shown is one based on ta Michelson-type interferometer. Light from the fiber bundle 15 is amplitude-divided by the beamsplitter 121, reflected from a mirror 122 (at a distance determined by mirror control 124) and recombined by the same beamsplitter 121 to be imaged onto an array 123 of photosensors. Moving one 6

mirror relative to the other introduces an optical phase or time delay between the two beams and results in interference between the two beams after recombination by the beamsplitter. The resulting interference pattern at each point in the image of the fiber bundle is recorded by the photosensor array 123 as a function of mirror position. Application of a suitable transform (Fourier or otherwise) recovers the Raman spectrum of light from each fiber in the bundle. There are several different types of imaging transform spectrometer designs (i.e., Sagnac, etc.) which could be used in a similar manner. Other Raman imaging spectrometers suitable for this application include a tunable optical filter (from the polarization-based dispersive properties of a liquid crystal) and a Fabr -Perot etalon.

When configured in this manner, the Raman imaging spectrometer 12 allows simultaneous multiplex detection of Raman signals from multiple 1/ C modules 11 linked to the spectrometer 12 via the optical fiber bundle 15. To see this, consider the Raman Michelson imaging spectrometer depicted in Fig. 1. Its optical configuration is such that each group of optical fibers 14 carrying Raman scattered light from a given module 11 illuminates a spatially localized group of pixels on the CCD camera after propagation through interferometer. Each module 11, therefore, is assigned a correlate spatial position on the CCD photosensor array 123. Scanning the interferometer over a given number of optical lags records the scattered light autocorrelation function (interferogram) for each module 11. The recorded interferograms are then input to computer 18 and the Raman spectral signal from each module is recovered on taking the digital Fourier transform of each measured autocorrelation function. This spatially multiplexed approach therefore constructs Raman spectral "images" of the process where each "pixel" in the spectral "image" corresponds to a different sensor 11 located at a different point in the process stream.

All signal processing and command and control system functions are implemented with a computer 18 integrated with a suitable data acquisition system. In this embodiment, there may be employed, for example, eight I/C modules 11 per Raman spectrometer 12. As shown in Fig. 2, to handle more spectral measurements of the process, it is possible to utilize a plurality of spectrometer systems 21, each with a spectrometer module 12 and associated I/C modules 11. The spectral output from each spectrometer module 12 is placed on a high speed bus 23 used for communication between computer 22 and each spectrometer system 21. The particular architecture and design for bus 23 and computer 22 are not a part of the present invention.

Claims

What is claimed is:

1. A system, for providing real-time control of a process that is monitored at a plurality of process locations by spectrometric techniques, the system comprising: (a) an imaging spectrometer having an optical image input and a spectral image output;

(b) a plurality of modules, one module disposed in each process location, each module having

(i) an illumination source for illuminating such location; (ii) a collector for collecting light that has been scattered at such location and directing such light to a collector output;

(iϋ) an optical fiber arranged to provide optical communication from the collector output to a pixel location of the optical image input of the imaging spectrometer; (c) an optical transducer, in communication with the spectral image output of the imaging spectrometer, for providing, as an output, spe ctral data, associated with each pixel location of the optical image input and therefore associated with each process location, for the purpose of process control.