DEVICE AND METHOD FOR GASIFYING CARBON-CONTAINING FUELS

Abstract

A method for operating a device for gasifying carbon-containing fuels and a corresponding device are provided. The gasification of the carbon-containing fuels provokes a flame. The emission spectrum of the flame is registered and evaluated continuously in real time by a multi-variant method and an evaluation model that is previously recorded.

Description

FIELD OF INVENTION

The invention relates to a device and a method for gasifying carbon-containing fuels.

BACKGROUND OF THE INVENTION

The increasing scarcity of oil as a raw material means that coal gasification is a technology assuming an increasing importance, through which both raw materials for synthesis in the chemical industry and also combustion gas for gas turbines can be obtained from the natural resource of coal which is available in sufficient quantities. As well as the thermal use this step represents a core process of effective use of coal in IGCC power plants.

The power plant concept with CO2 separation which is currently the most developed concept for power plants is Integrated Gasification Combined Cycle (IGCC), in which a gasification of the fuel is situated upstream from the actual combined cycle power plant.

A possible gasification method which can be connected upstream from the combined cycle process is the Siemens Fuel Gasification method (SFG method). This method is suitable for use even of ash-rich, solid, liquid and gaseous input materials. The input material is converted by means of gasification means containing free oxygen in a flame reaction at pressures of up to 10 MPa and at temperatures of up to 1900° C. into CO and H2 (primary components of synthesis gas).

To enable the gasification process, i.e. the substochiometric conversion (λ<1) of the input material to be driven efficiently and also without errors to an energy-rich synthetic gas, important operating parameters must be monitored however. They are for example the detection of the safe combustion of the primary fuel (input material), the flame temperature and variations in the input material composition. However a direct definition of the process parameters is currently only possible with difficulty or is not possible at all.

Described by the typical parameters given above of combustion of the primary fuel (input material), the flame temperature and fluctuations in the input material composition, the previous options are represented as follows:

• A direct monitoring of the ignition of the primary fuel is not currently possible. The safe ignition is displayed and monitored by the growth in the amount of synthetic gas and the rising heat output of the system.
• The flame temperature can currently only be indirectly calculated thermodynamically on the basis of the material and heat flows.
• Fluctuations in the input material composition are only detected indirectly and after a time delay by changes in the heat output.

The strength of the flame emission in a specific wavelength range is measured with flame pyrometers and, assuming a specific emissivity, a conclusion is drawn as to the temperature. This method is sensitive to contamination of the optical window (since the absolute radiation density is used here as a measured value). On the other hand the reliability of the measured values obtained is questionable since the flame concerned is not a grey radiator (as to be assumed for this evaluation) of which the emissivity coefficient is simultaneously unknown. Metals contained in the flame caused overlaid peaks through their flame lights, or can also, through their reabsorption, cause local drops in the spectral radiation density. Both lead via a potential falsification of the power density in the evaluated spectral window to incorrect measurements. And these are not just systematically incorrect (and thus able to be calibrated), since e.g. the power of a peak located in a window created by a specific metal changes in an unpredictable manner with the change of the concentration of the metal in the fuel.

Therefore pyrometers can only be used in lined reactors (in which the wall represents the radiator), as well as only in gas operation, in which a sufficient transparent gas flame is present. The coal flame is not sufficiently transparent because of the carbon particles contained for the radiation of the wall lying behind it.

An attempt can be made to use the strength of the peak by the emission of metals to determine the content of specific metals, which would then deliver a sensible measured value for gasifier operation. The metal content determines the amount of the ash formed. To this end peak heights, surfaces or also specific gradients are evaluated in a specific spectral range. Since however a plurality of overlapping flame emission peaks is present and these overlap greatly, this approach is only able to be carried out to an inadequate extent in practice.

SUMMARY OF INVENTION

The object of the present invention is to specify a method of gasifying carbon-containing fuels with which the said problems can be reduced or resolved. A further object is to specify a corresponding device.

This object is achieved by a method and by a device with the features of independent claims. The dependent claims relate to advantageous embodiments.

In the inventive method for operating a device for gasifying carbon-containing fuels, in which the gasification leads to a flame, the emission spectrum of the flame is recorded, this expediently occurs by access to the flame spectrum being made possible through an optical window in the gasification reactor. The flame spectrum is supplied to a spectrometer for spectral analysis, the spectrum obtained is passed on electronically to an evaluation unit and is evaluated with a multi-variant method with a previously stored evaluation model, especially continuously in real time.

Thus the invention specifies a method which evaluates the emission spectrum of the flame in an innovative and advantageous manner, in order to enable the desired parameters, such as the flame temperature for example, to be continuously monitored. A direct measurement of the parameters by probes introduced into the gasification flame is hardly possible because of the high temperatures, the reactive gases and also a strong tendency for deposits to form.

The spectral analysis can include the range of the electromagnetic spectrum from UV into the mid-IR range. In particular the emission spectrum is recorded in the range from 300 to 2000 nm, especially in the range from 300 to 800 nm.

Preferably the emission lines of the ash components from the emission spectrum are evaluated, especially the emission line of alkali.

In a development of the invention, to form the evaluation model, spectra are recorded for known operating parameters and stored together with the operating parameters in a memory. In such cases especially the spectra with the operating parameters are classified with the methods of multi-variant statistics such as primary component analysis, partial least squares regression, partial least squares discriminant analysis PLSDA, cluster analysis or artificial neural networks and in this way an evaluation model is created which assigns specific operating parameters to a particular spectrum.

Then expediently a spectrum recorded during operation is assigned with the evaluation model to a known spectrum with known operating parameters and in this way the current operating parameters are determined.

In such cases the operating parameters preferably to be taken into account, individually or in combination with one another comprise the following parameters:

• The distinction as to whether operation of the device with gas, for example pressure maintenance, or gasification mode with input material flame is present on the basis of the different fuel, which provokes different spectra, since gas for example only contains very few metals which create spectral lines,
• The totality of anorganic materials, which produce slag on the basis of spectral lines from light-emitting/absorbing metals,
• The temperature of the flame,
• The stochiometry of the combustion.

It is advantageous for the evaluation unit for the measured flame spectrum to be normalized in the evaluation unit before being supplied for spectral analysis, especially to a peak height, the spectral integral, or a signal height at a definable wavelength.

In an advantageous embodiment of the invention a spectrometer is used, which performs a parallel measurement of the spectrum, especially a spectrometer which performs a wavelength dispersion and then maps the result to a parallel measuring line detector in which each pixel then measures a specific wavelength interval. The result achieved by this is that the spectrum is measured at one time. A sequential measurement of different wavelengths can lead, through flickering of the flame, to falsifications of the result.

As an alternative or in addition, to reduce the effects of flame flickering, a measuring time can be selected which is long in relation to the flicker frequency. Similarly to this a number of short measurements can also be taken at time intervals comparable to the time constant of the flame flickering and then averaged before being supplied to the evaluation device.

It is further advantageous for the evaluation unit to determine whether the supplied spectrum lies in the measurement range of the spectrometer in the sense of an overexposure or underexposure and for new spectrometer parameters to be set, especially the measurement time or the number of multiple measurements if this is not the case.

It is likewise advantageous for monitoring of the integrity of the optical path to be performed, by the shape of a measured spectrum being evaluated.

The invention also creates a correspondingly embodied device for gasification of carbon-containing fuels, having:

• Means of recording a spectrum of the flame,
• Means of spectral analysis of the spectrum,
• Means of evaluating the spectral analysis on the basis of an evaluation model,
• Means of storing the evaluation model.

BRIEF DESCRIPTION OF DRAWINGS

Preferred, but in no way restrictive, exemplary embodiments for the invention will now be explained in greater detail on the basis of the figures of the drawing. In this case the features are represented schematically, in which;

FIG. 1 shows a spectrum of a coal flame in the visible spectral range,

FIG. 2 shows a spectrum of a gas flame in the visible spectral range,

FIG. 3 shows a method for evaluating the flame spectrum,

FIG. 4 shows a layout for evaluating the flame spectrum,

DETAILED DESCRIPTION OF INVENTION

FIG. 1 shows a typical spectrum 10 of a coal flame, measured in the UV/VIS range of between around 400 nm and 1000 nm. FIG. 2 shows a similar second spectrum 20 of the gas flame.

An exemplary embodiment for a method for evaluating the flame spectrum is shown in FIG. 3. It is performed in two stages:

1) The formation of the evaluation model: Here spectra 31 are recorded in gasifier mode with known operating parameters and stored together with the operating parameters. Advantageously at this point the entire bandwidth of the operating parameters which are later of interest is measured. After pre-processing 32 of the data, said data is then classified with the method 33 of multi-variant statistics, i.e. an evaluation model 34 is created, which assigns a specific spectrum to specific operating parameters. Examples for these methods 33 are primary component analysis PCA, partial least squares regression PLS, partial least squares discriminant analysis PLSDA, cluster analysis and neuronal networks.

2) Use of the evaluation model. This evaluation model thus determined is then supplied during operation with an unknown spectrum 35, is assigned a known spectrum from the evaluation model and thus the current operating parameters 36 are determined as an output.

In this case the mathematical pre-processing 32 typically comprises a smoothing and/or derivation, normalization, a selection of the spectral range which is to be observed and the discarding of obviously incorrect measurements.

Operating parameters which leave behind their chemical fingerprint in the flame spectrum can be determined with this method. They are for example:

The distinction as to whether an operation of the gasifier with gas (for example pressure maintenance) or the gasification operation with input material flame is present: through the different fuel the spectra are able to be assigned to the two classes, since gas only contains very few metals which emit light.

The totality of anorganic material which produces slag. The slag content is a very important parameter for maintenance: the influencing of the spectrum by light-emitting/absorbing metals represents a quantitative signal and the slag content can be quantitatively derived from the spectra.

Temperature of the carbon flame: the shape of the spectrum is influenced by the excitation varying with the temperature, spectra can be quantitatively assigned to the temperature.

Stochiometry of the combustion: if the stochiometry of the combustion is adhered to precisely, the gas yield is optimized thereby. The stochiometry massively influences the chemical composition and is thus accessible with this method.

A spectroscopic method is involved in which the information content lies in the shape of the curve. Therefore in accordance with one embodiment the measured flame spectrum is normalized in the evaluation unit before being supplied for recognition. Normalization can in such cases for example be undertaken to the peak height, the spectra integral or the signal height at a fixed characteristic wavelength.

The flame in the gasification process is not a stationary process in relation to the time constants of the spectrometer measurement (parts of the second), the typical statistical flame flickering occurs with spectral and absolute intensity variations. If the spectrum is now recorded with a sequentially-measuring spectrometer, the flame flickering leads to a falsification of the recorded spectrum. It is thus advantageous to provide a spectrometer which carries out a parallel measurement of the spectrum. For example the spectrometer is embodied to undertake a wavelength dispersion and then map the result to a parallel-measuring line detector, in which each pixel measures a specific wavelength interval.

In a development of the thinking, effects of flame flickering are eliminated by either selecting a measurement time which is long in relation to the flicker frequency or by recording a number of short measurements at time intervals comparable to the time constant of the flame flickering and averaging them before they are fed into the evaluation unit.

The evaluation unit can assess whether the supplied spectrum lies in the good measurement range of the spectrometer, i.e. whether neither underexposure or overexposure is present. If underexposure or overexposure is present, new parameters are set through commands of the evaluation unit to the spectrometer, e.g. the measurement time, i.e. adapted to the integration time, which is used for recording a spectrum, or the number of multiple measurements.

It can occur that there are deposits on the optical window. On the one hand these cause an attenuation of the light, but can however also cause spectral distortions. This then leads to the shape of the spectrum being distorted, which in the final analysis leads to incorrect measurements. Multi-variant signal processing can recognize however the extent to which a measured spectrum still has similarities to the typical range of known flame spectra. This function is used to recognize a spectral distortion which is too great caused by a change in the optical path/window and if necessary to issue a corresponding error message.

It is advantageous for the absolute intensity not be taken into account in the evaluation method, but instead the curve shape of the flame spectrum. The influence of deposits on the optical window on the measurement is further reduced by this process.

Before the measured spectrum is supplied to the evaluation unit it is advantageous to check whether the measurement is plausible. If it is not, the measurement is discarded. This check especially contains an evaluation as to whether the measured intensity is in the expected range, whether outliers are frequently present and whether there is increased noise present in the measurement data.

FIG. 4 shows a schematic measurement layout 40. The measurement layout 40 comprises a flame detector 41 and an adapter 42. Coupled-out light is conveyed through a fiber-optic cable 43 to a switching cabinet 44 in which an evaluation unit 45 is accommodated.

Claims

1.-16. (canceled)

17. A method for operating a device for gasifying carbon-containing fuels, comprising: recording an emission spectrum of a flame lead by the gasification; andevaluating the emission spectrum in real time with a multi-variant method and with a previously stored evaluation model by an evaluation unit.

18. The method as claimed in claim 17, wherein the emission spectrum is evaluated in a range from ultraviolet radiation to infrared radiation.

19. The method as claimed in claim 17, wherein the emission spectrum is recorded in a range from 300 nm to 2000 nm, or in a range from 300 nm to 800 nm.

20. The method as claimed in claim 17, further comprising determining a flame temperature from the emission spectrum.

21. The method as claimed in claim 17, wherein at least one spectral range in which an emission line of an ash component lies is evaluated from the emission spectrum.

22. The method as claimed in claim 21, wherein the ash component comprises an alkali metal.

23. The method as claimed in claim 17, wherein, the evaluation model is determined by: recording spectra with known operating parameters,storing the spectra with the known operating parameters together in a memory,classifying the spectra with the known operating parameters by statistics of the multi-variant method comprising primary component analysis, partial least squares regression, partial least squares discriminant analysis PLSDA, cluster analysis, or artificial neural networks,wherein the evaluation model assigns the known operating parameters to a specific spectrum.

24. The method as claimed in claim 17, wherein the emission spectrum recorded in operation is assigned by the evaluation model to a known spectrum with known operating parameters for determining current operating parameters.

25. The method as claimed in claim 17, further comprising determining current operating parameters, wherein the current operating parameters comprise: a distinction as to whether the device is being operated with gas to maintain pressure, or in a gasification mode with input material flame based on the fuels that provoke different spectra,a total content of anorganic materials that produce slag based on spectral lines through light-emitting/absorbing metals,a flame temperature, anda combustion stochiometry.

26. The method as claimed in claim 17, wherein the emission spectrum is normalized in the evaluation unit before a spectral analysis to a peak height, a spectral integral, or a signal height for a definable wavelength.

27. The method as claimed in claim 17, wherein the emission spectrum is smoothed in the evaluation unit before a spectral analysis by one or more of the following measures: a distance of outliers according to a threshold method,a Savitzky-Golay method, anda Kalman filter.

28. The method as claimed in claim 17, further comprising performing a parallel measurement of the emission spectrum by a spectrometer, wherein the spectrometer performs a wavelength dispersion and maps the wavelength dispersion to a parallel-measuring line detector in which each pixel measures a specific wavelength interval.

29. The method as claimed in claim 17, wherein effects of flame flickering are reduced by a measurement time or by a number of short measurements, wherein the measurement time is selected that is long in relation to a flicker frequency, and wherein the number of short measurements are recorded at time intervals comparable to a time constant of the flame flickering and are averaged before the evaluating.

30. The method as claimed in claim 17, wherein the evaluation unit determines whether the emission spectrum lies in a range of a measurement of a spectrometer, wherein a spectrometer parameter is reset if an overexposure or an underexposure is present in the measurement, and wherein the spectrometer parameter comprises a measurement time or a number of multiple measurements.

31. The method as claimed in claim 17, further comprising monitoring an integrity of an optical path by a shape of the emission spectrum.

32. A device for gasifying carbon-containing fuels, comprising: a memory for storing an evaluation model; andan evolution unit for performing the method according to claim 17.