BACKGROUND OF THE INVENTION

The present invention relates to apparatus for producing a continuous visual image of the physical condtions in the interior of a hot vessel. The technology of interest images hot interior surfaces where the surfaces are obscured by fume and radation emitting/ absorbing gases associated with a process taking place in the vessel. More particularly, the invention relates to imaging the interior of a wood pulp chemical recovery boiler.

^* In the manufacture of wood pulp, raw wood is digested in the presence of inorganic chemicals. In the sulfate or kraft pulping processes the active pulping chemicals are sodum hydroxide and sodum sulfide. An important aspect of kraft pulping is recovery of these inorganic pulping chemicals from the liquid pulping waste stream. . This waste liquid, commonly called black liquor, is an aqueous mixture of lignin extracted from the wood and reacted inorganic pulping chemicals, principally sodum carbonate and sulfate. Economics of the overall pulping process require recovery of the relatively expensive pulping chemicals from the waste black liquor.

Conventionally, black liquor is combusted in a boiler designed to recover the pulping chemicals and extract thermal energy from the organic matter in the black liquor for steam production. The chemical recovery boiler receives black liquor after it has been concentrated in evaporators. The liquor is burned in the furnace and the chemicals are recovered as a molten smelt in the bottom of the boiler from which it dscharges for recycling back into a fresh pulping liquor makeup system. A principal function of the recovery process is the reconversion of sodum sulfate back to active sodum sulfide by the carbon residues in the boiler. Temperfiture and air flows are controlled to maintain a redudng zone in the boiler to change as much of the sodum sulfate as possible into sodum sulfide. The degree of reduction of sulfate to sulfide in the smelt reflects the efficiency of the chemical recovery function and determines the quality of the chemical product from the recovery boiler.

The characteristics of the bed of molten solids and how they affect heat and chemical recovery are not known with certainty. This is largely due to the lack up until now of a means for accurately observing or measuring bed configuration. Fume and gaseous emissions from the smelt bed typically prevent significant continuous visual observation.

The bed itself is generally a very porous lattice containing approximately 5% carbon by weight. The typical smelt bed temperature is about 1000°C with overlying combustion gases at temperatures of 1100°-1300°C. What is known suggests that smelt bed height on the floor of the furnace affects the efficiency of achieving the primary goal of reducing sodum sulfate in sulfide. In general, a medum bed height suggests good chemical reducing environment. On the other hand, a low bed height is thought to be safest, depending upon boiler geometry, providng short cool down times following emergency shutdowns. Also, an unstable bed size is related, in general, to less stable boiler operations, leadng potentially to dangerous, blackouts.

The present knowledge of the effect of bed inventory and dstribution on the operations of the recovery system as a whole is enough to suggest that a constant .bed inventory is essential to maintaining stable boiler operations. To achieve this goal, a clear continuous visual image of the bed would provide much of the information needed to permit improve¬ ment of control of the bed and, hence, improved chemical recovery. The Prior Art One prior method of measuring bed height involves the installa¬ tion of pyrometers at, for example, furnace air port levels. Temperature measurements are then used to infer bed height. This simple technique equates relatively low and high temperatures with the presence or absence, respectively, of the smelt. A problem with this technique is that gas temperatures near the bed can vary greatly with variations in air dstribu¬ tion and cause false readngs unrelated to bed height. Also, such a monitoring system tells little about the dstribution of bed material at the central regions of the bed, important information since past observations suggest that material dstribution is non-uniform. Many chemical recovery boiler operations presently employ closed circuit TV cameras to produce images of the smelt bed. These

cameras should provide all necessary information on bed configuration and height in a form that could be readily interpreted and used by the recovery boiler operator. However, the image produced is intermittent with ind- vidual regions of the bed visible only about 10-20% of the time. An impression of the bed configuration is gained only by observing images for a period of minutes over which various portions fade in and out of view. Most of the time the image is obscured by swirling clouds of dust or incandescent gas. Bright flashes light up small regions of the bed and occasionally silhouette intervening portions of the bed producing for a moment a clear impression of bed configuration. The entire bed is, however, never visible at once and the walls of the furnace are not dscernible. Experience with these video imaging systems has been that because of the intermittent nature of the image, the information provided is generally insufficient to permit operational control decisions. A. number of imaging systems have been developed for viewing interiors of closed, hot vessels other than recovery boilers, to determine some internal characteristic or condtion. Most of these systems are designed for use in blast furnaces, coke ovens and the like.

In a recovery boiler, typical smelt bed temperatures are at about 1000°C, with overlying combustion gases at temperatures of 1100 ^o -1300°C. In blast furnaces, the stock or ore/coke surface temperatures are on the order 150°-300°C and the overlying gases are at temperatures of 90°-130°C. Because of these relatively low temperatures, there is very little visible radation from the stock charge, about 12 orders of magnitude less than the intensity of visible radation from a recovery boiler smelt bed. A number of prior art systems suggest, as does the Ponghis patent, U.S. 3,718,758, the use of an illuminating source in combination with a TV camera, both of which are mounted on a probe and inserted into the vessel. The lack of ambient visible light coupled with an interest in measuring stock surface temperatures led Ponghis to employ optical illumination in the infrared range.

Where temperatures of the surfaces to be imaged are suffi¬ ciently high an infrared scanner may be used if there is insufficient visible light available. The Shimotsuma patent, U.S. 3,588,067, describes just such a system in which an industrial television camera employing an infrared vidcon produces an image correspondng to the height and shape of the

materials in the vessel. The infrared signal also produces temperature dstributions. No mention is made by Shimotsuma of the system's ability to distinguish between the charge and the overlying gases. In light of condtions observed in recovery boiler environments, the lack of mention in Shimotsuma of any necessity to reject interferences from obscuring gases or particles suggests that these condtions are not present in the blast furnace environment.

In the chemical recovery boiler environment, it is not sufficient to use an infrared vidcon equipped camera alone to observe the bed since the vidicon is also sensitive to visible radation. In contrast to the system of interest in Shimotsuma, the size and loadng of fume particles in the recovery boiler are such that visibility is substantially obscured in the visible range virtually all the time. Elemental sodum and possibly some sodum hydroxide are vaporized at the bed surface and react with carbon and oxygen. Sodum carbonate is formed which condenses into liquid droplets of approximately 0.1 to 0.5 micrometer dameter. These constitute the fume present in the lower furnace. As these droplets are swept upward with the rising gas stream, they encounter regions of high sulfur doxide concentration and the carbonate is dsplaced by sulfate so that the composi- tion of the fume in the upper part of the furnace is almost entirely sodum sulfate.

The chemical recovery boiler environment also dffers from blast furnace environments with regard to absorption/emission of radation by the gases present above the material beds. In the recovery boiler, the gases above the smelt bed are significantly hotter than the bed surface. There¬ fore, radation emitted by the gases within the spectral absorption and emission bands assodated with the major gaseous species present will be much more intense than that emitted from the bed. Also, at the tempera¬ tures present in the recovery boiler these gaseous absorption emission bands will be greatly broadened.

In the blast furnace, by comparison, the gases above the stock line are somewhat cooler than the stock surfaces so that the total radation from the gases will be small compared to that of the surfaces. • The absorption/emission bands will be fairly narrow compared with the corre- spondng bands in the recovery boiler. The primary effect of the gases in the blast furnace situation would be to absorb only that portion of the

radiation emitted from the surface which lies within the relatively narrow absorption bands.

Thus, in the blast furnace situation, where there apparently is no significant fume or dust interferences and limited absorption/emission interferences, a broad band system which is sensitive to a wide range of visible and infrared wavelengths would likely work well. However, in the chemical recovery furnace, there are suspended light scattering particles or fume and gaseous absorptions and emissions at high temperature, both of which combine to obscure the surface to be imaged. The broad band infrared system of the prior art does not work much more efficiently than the visible range systems presently in use.

SUMMARY OF THE INVENTION What has been dscovered is that an infrared vidcon camera arrangement for imaging produces images suitable for boiler operation control where the spectral sensitivity of the system to specific IR wave¬ lengths is limited, first of all, to greater than about 1 micrometer by suitable" filtering. Secondly, the spectral sensitivity of the system must also be limited to avoid absorptions and emissions of radation from the major species of gases just above the smelt bed. • The imaging apparatus of this invention for observing infrared emitting surfaces where the surfaces are substantially obscured by fume and dust particles which strongly scatter visible light and by absorptions and emissions of major species of overlying gases includes the following ele¬ ments. An industrial closed circuit video camera, indudng an infrared imaging detector or vidicon tube, is located near but insulated thermally from the surfaces to be detected. A holdng means, mounted on the camera, positions an objective lens for the camera in line of sight with the surfaces to be imaged. An optical filter means limits the spectral sensitivity of the system to wavelengths greater than about 1 micrometer. The optical filter means also selectively rejects wavelengths of strong gaseous absorptions and emissions. Finally, a TV monitor receives the image detected by the vidcon camera and produces a continuous image of the bed surfaces.

The imaging apparatus of the invention produces a substantially clear view of the smelt bed of inorganic chemicals recovered from combus- tion of kraft black liquor derived from the pulping of wood. The optical filter excludng all radation except a narrow band, thus avoidng the

interferences caused by fume particles and gases which overlie the smelt bed surfaces and obscure visibility of those surfaces, is a key element of the invention. One practical filter is centered at 1.68 micrometers and has a band width of 0.07 micrometers. All other narrow band filters which avoid overlying particulate and gaseous interferences are contemplated as within the invention, includng those for other hot surface systems. BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 is a schematic of the imaging apparatus of the invention shown in combination with a chemical recovery boiler smelt bed. Figure 2 predets extinction coefficients (m ^_1 ) as a function of wavelength (micrometers) for emissions of gases and light scattering partic- ulates overlying a smelt bed.

DESCRIPTION OF THE PREFERRED EMBODIMENT

Referring to Figure 1, the apparatus of the invention is depicted schematically. A dosed circuit television camera 10, which indudes an infrared vidcon tube component (not shown in detail), is located adjacent boiler 20 whose interior is to be imaged. A lens tube assembly 11, mounted upon camera 10, extends into the boiler 20 through an aperture 21 in the boiler wall 22. The lens tube contains such objective, collecting and collimating lenses (not shown in detail) as are conventionally necessary to transmit an image to be remotely reproduced from the object to be observed to the infrared vidcon 10. The camera 10 is mounted on a stand 23 which permits horizontal and vertical adjustment to view a substantial protion of the boiler floor 30 and any smelt bed 31 accumulated thereupon. An optical filter 12 is a key element of this invention. ' It is selected to limit the wavelength of light transmitted to the vidcon from the object to be imaged to wavelengths greater than 1 micrometer. This limitation is imposed to avoid image interferences caused by particulate and fume overlying the surface to be imaged. The optical filter 12 further limits the transmitted light from the surfaces to be imaged to a narrow band which avoids the light emissions of the principal spedes of hot gases overlying the surface to imaged. Careful selection of wavelengths transmitted to avoid these gaseous inter¬ ferences is critical to obtaining the dramatically improved images of the surfaces to be reproduced.

Both camera and the lens holder are provided with an endosure 13 designed to protect them from the drty mill environment. A cooling and

purging air system 14 insures that the temperature of the lens and camera do not exceed safe operating levels. The air cooling system also creates a high velocity air purge at the end of the lens to prevent combustion gases, char, or smelt from fouling the optics. It was dscovered that by placing the end of the lens tube 11 approximately 5-10 cm back from the wall tubes 22, fouling of the lens tube was virtually eliminated without sacrificing the field of view.

A cable 17 transmits the image obtained through the lens and vidcon for reproduction upon a television monitor 19. As noted above, a most critical element of this imaging apparatus and its ability to obtain unobscured images lies in the sdection of an optical filter 12 which avoids interferences caused by fume particulate and hot gases overlying the surfaces to be imaged. The selection process requires an analysis of the transport of radation through partide-laden gases, and identification of those properties of the gases and partides which affect this transport process. In its simplest form, the solution to the equation of radative transport may be expressed as:

Kλ) = I_(λ)e ^~Υ + B _λ (T _g )(l-e ^"γil )

where I(λ) is the intensity of radiation at wavelength λ reaching an observer at a distance I from the surface s; I (λ) is the intensity of the radation at wavelength λ emanating from surface s in the drection of the observer; ^B χ(T ) is the black body intensity at the gas temperature, T ; and γ is the extinction coefficient of the particle laden gases between surface s and the observer. This assumes that the extinction coefficient and the gas temperature are constant along the path I.

As can be seen from this expression, when γ is large, the radation emanating from the surface is substantially attenuated and only radation emitted by the gases reaches the observer; when γ is very small, the radation from the surface reaches the observer with little attenuation and the gases provide little interference. It is these regions of low γ, if they exist, where clear images of a surface such as the smelt bed of interest here may be obtained. The extinction coefficient may be expressed as the sum of contributions from the gases and particles;

γ ¹ = γ 'g + γ 'pa + γ 'ps

where γ is the absorption coefficient of the gases, γ is the absorption coefficient of the particles and γ pa is the scattering coefficient of the particles. In order to establish reasonable estimates of the magnitudes of these quantities, it is necessary to examine the condtions which exist in the environment of interest, here the lower furnace region.

Specific information which is required indudes the magnitudes of gas temperature, gas spedes concentrations and optical properties, particle sizes and mass loadng, as wdl as partide optical properties. Virtually no experimental data exists for any of these quantities in the lower part of the furnace. However, reasonable estimates of these quantities can be made which will permit selection of an appropriate filter.

In the absence of rdiable measurement of gas species concentra- tions in the various zones in the recovery furnace, a computer model was designed using thermochemical equilibrium calculations and heat and mass balances to provide at least rough indcations of the probable magnitudes of these quantities and of the gas temperatures in the various zones of the furnace. See Merriam, R. L., "Computer Model of a Kraft Recovery Furnace - Version 1.0," Vol. π, p. 3-7, Arthur D. Little, Inc. (Aug. 1979).

^• The computer model concentrations were combined with data on the spectral properties of the products of the combustion of alcohol, which are

CO, COo, HnO, N _2. and O- _>> and with data on the low temperature absorption spectra of sulfur compounds expected to be generated, to provide rough estimates of the probable contribution these species make to the extinction coefficient as a function of wavdength. Estimates of extinction coeffi¬ cients calculated using the computer model and rdated data, based upon typical concentrations in the lower furnace, are shown in Figure 2 as a function of wavelength. Figure 2 shows that the primary contributions or interferences from gaseous species are due to CO„, H«O, SO« and Na.

A somewhat dfferent analysis is necessary to estimate extinc¬ tion coefficients for the fume particulate interference. The reported measurements of partide emissions from recovery boilers performed both before and after partide removal equipment suggest the presence of three distinct classes of particles in the chemical recovery boiler environment. There are submicron fume partides (0.2-0.5 micrometer dameter)

consisting primarily of sodum sulfate and some sodium carbonate. There are large spheres (10-100 micrometers diameter) of sodium sulfate and carbonate presumably arising from suspension burning of liquor droplets. The third class of particulates includes the very large (up to 2,000 micro- meters diameter) char fragments having high carbon content.

Literature studies show mass mean particle size to be about 1 micrometer. These studes typically involve sampling after the boiler, and it is highly likely that most of large particles in the furnace would have been captured on the boiler tubes. Thus, the mean mass diameter of particles in the furnace will be much larger than 1 micrometer. One model indcates that approximately 35 percent by mass of the particle loading is fume from the bed and 65 percent is char and carryover from liquid suspension burning. This would suggest that the mass mean diameter of particles in the furnace would be heavily weighted toward that of the larger carryover particles. Recent measurements of superheater deposits show a deposition rate consistent with an aerodynamic mean diameter of 50 micrometers, which is supported by scanning electron microscope photographs of spherical particles collected in the upper furnace. Another model shows that about 9 percent of the sodum in the black liquor ends up in particulates which suggests for a typical kraft recovery boiler that the total mass loading of particulates will be about 2.3 g/m .

Analysis of the literature and models permits estimation of particle sizes, mass loadings and compositions as shown the following table.

KRAFT CHEMICAL RECOVERY BOILER

Estimated Particles Sizes, Mass Loadngs and Compositions

Mass Mass Volume Particle Mean Load- Concen- Number Type Diameter ings tration Density Composition

( μ m) (g/m ³ ) (m ³ /m ³ ) (cm ^'3 )

Fume 0. 3 0.8 3.0 x l0 ^"7 2.0 x lO ⁷ Na ₂ SO ₄ + Na ₂ CO, +

To make estimates of the magnitudes of the particle absorption and scattering coefficients, it is necessary to consider both their size and composition. The absorption and scattering coefficients are related to

"efficiency factors" for absorption and scattering described by Mie theory for the interaction of light with small partides. See H.C Van De Hulst, "Light Scattering by Small Particles," John Wiley and Sons, New York (1957) and M. Kerker, "The Scattering of Light and Other Electromagnetic Rada¬ tion," Academic Press, New York (1969). The partide absorption and scattering coeffidents may be expressed in terms of these efficiency factors as follows:

^γ pa = ^{3 2} < ^m > ^{λ) C} v/ ^D 32

γ = 3/2 Q _s (m, λ) C _v /D 32

where Q £L is the mean absorption efficiency for a given collection of particles, Q _s is the mean scattering efficiency, m is the complex refractive index of the particles, λ is the wavelength of light, C is the volume concentration (m of particles per m of gas) and D, ₂ is the Sauter mean diameter of the particle collection which is equal to the total particle volume dvided by the total particle surface area..

The fume particles are composed primarily of sodum sulfate and sodum carbonate which are virtually transparent except in localized spectral regions in the infrared beyond 5 micrometers. Therefore, except in those regions, the imaginary part of their refractive index will be negligible making the absorption efficiency also negligible. In the region beyond

5 micrometers, their small size with respect to the wavelength will result in extremely small absorption efficiencies even in a case where the imaginary part of the refractive index is significant. Thus, in the case of fume particles, γ pa will be negligably small. However, the scattering efficiency exhibits a strong resonance at wavelengths comparable to the particle diameter. Therefore, scattering by these particles will be substantial throughout the visible and out to about 1.0 micrometer in the infrared. This is shown by the large peak in the extinction coefficient curve for particles at about 0.2 micrometer in Figure 2.

The larger carryover and char particles are likely to contain significant amounts of fixed carbon, making them highly absorbing, having large imaginary refractive indexes. This, coupled with their large size, at all wavelengths of interest, results in the scattering and absorption

efficiencies for these particles being both very nearly equal to unity. Because they are so large, however, their number density is very low (2-3/cm ). Thus, the absorption and scattering coefficients due to these particles are small (γ = γ <_ 0.015 m ) for wavelengths less than about 30 micrometers. This effectively sets the lower limit on extinction for the gas partide system throughout the region of interest. Thus, it is apparent that apart from a sodum emission/absorption phenomenon, it is scattering by fume particles which limits visibility in the visible portion of the spectrum. In this region, the extinction coefficient due to fume partides is on the order of 1-5 per meter, which implies that the intensity of radation originating at the bed is reduced by 98 percent in 1-4 meters. This explains why it is not possible to see the far wall of a 10-meter wide furnace utilizing visible light.

Beyond a wavelength of about 4 micrometers in the infrared, the extinction coefficient for particles is predominantly due to large carryover and char partides, and is nearly constant at 0.03 m . Therefore, in the spectral windows between the gaseous absorption/ emission peaks, radation should be able to travel a dstanee of 130 meters before suffering 98 percent attenuation. In these regions, visibility within the furnace should be excellent.

Spectral windows generally suitable for smdt bed imaging identified by the above analysis indude the following: 1.57 to 1.73 micro¬ meters; 2.23 to 2.43 micrometers; 3.25 to 4.05 micrometers; 4.80 to 5.30 micrometers; 6.90 to 7.20 micrometers; 7.60 to 7.80 micrometers; 7.90 to 13.90 micrometers. Other windoes beyond 13.90 micrometers are evident from inspection of Figure 2.

Clearly the best visibility predcted by these calculations occurs beyond about 4 micrometers within the windows in the gaseous absorption/ emission spectrum. However, infrared imaging devices capable of operating at these long wavelengths are quite costly, have rdatively low resolution and often utilize detector elements which require cryogenic cooling, making them inappropriate for permanent unattended operation in a mill environ¬ ment. Furthermore, special infrared transmitting optics are required for wavelengths greater than about 2.5 micrometers. Within the last few years infrared sensitive vidcon tubes com¬ patible with standard dosed circuit TV cameras have become commercially

available. These tubes typically are sensitive to radation out to about 2 micrometers wavelengths.

Inspection of Figure 2 shows that the lowest predeted value of extinction coefficient in the region below 2 micrometers occurs in the window at about 1.7 micrometers between two water vapor bands. At this wavelength the predeted extinction coefficient is approximately 0.2 m . This is an order of magnitude smaller than in the visible region of the spectrum and based on this estimate, light originating at the bed should propagate across the 10 meter-wide furnace with about 86 percent attenua- tion. This implies that the far wall will be visible at this wavelength. Example

The modfied system of this invention is similar to commercial, visible light systems currently used except for the substitution of the infrared vidicon tube or the standard visible light tube and the add tion of a narrow band interference filter. A Motorola high temperature CCTV system, induding an endosed camera and furnace lens tube as described above, was mounted, as shown in Figure 1, with the lens tube inserted through a secondary air port of a Babcock and Wilcox kraft chemical recovery boiler. A Hamamatsu Type N 214, lead sulfide, 2.54 cm (1 inch) diameter infrared imaging detector vidcon tube manufactured by Hamamatsu Company of Hamamatsu City, Japan was substituted for the standard Motorola vidicon tube. A narrow band interference filter centered at 1.684 micrometers and having a band width of 0.07 micrometers was used as the spectral filter. In testing this assembly, it was discovered that focus improved substantially by moving the vidcon tube further away from the lens tube by inserting a spacer between the camera body and the lens tube. It is believed that this increased separation was necessitated by the increase in focal length of the lens elements at the longer wavelengths. Once the focus had been adjusted it was discovered that the visibility at the wavelength selected was much better than with the conventional system. The entire bed within the field of view could be seen clearly and continuously and the height and configuration of the bed could be easily perceived. Surface features of the bed could be distinguished even at the far wall of the furnace. The bed was very low at the time of these tests and the primary and secondary air ports on the far wall could be seen fairly clearly as dark rectangular regions on the generally bright wall surfaces.

It is noted that the resolution of the infrared vidicon system is such that the indvidual primary air ports, approximately 5 cm x 25 cm, could be discerned. The image indicates that under normal operating conditions, the system allows the resolution of features having dimensions of a few centimeters or less at a dstance of 10 meters.

The infrared vidcon also has an intrinsically slower time response than the visible vidcon tube. This may be partially responsible for the fact that even large suspended particles and liquor droplets did not appear in the image. However, the time response is sufficiently short that material falling from the walls was observed as well as large "boulders" rolling across the bed from time to time. Dust raised by the primary air jets and by falling material hitting the bed was observed.

Because the system is relatively insensitive to radation from the gaseous species which appears primarily at filter-excluded wavelengths and because the system operates in infrared range, the radation observed is thermal radation emitted by solid surfaces. Therefore, the observed intensity is related to the temperature of those surfaces. However, the orientation of the surfaces as well as their emissivity will also contribute to observed intensity distribution. The fact that visual judgment of bed contours agrees very well with the direction of travel of _. objects rolling on the bed surface under the influence of gravity supports the view that surface orientation plays a significant role in determining observed intensity distribution. At the same time, the fact that material falling from the walls which has been cooled by contact with the water tubes appears dark when it first hits the surface, but gradually becomes bright and indistinguishable from surrounding bed medum supports the view that the temperature of the material is also of great significance in image intensity.

The ability to view the inside of the furnace clearly provides an unprecedented opportunity to observe internal furnace processes and increase understandng of boiler operating phenomena. This invention is not limited in usefulness to just the kraft pulping chemical recovery boiler system. It is useful in viewing any hot surfaces obscured by overlying hot gases, fumes and particulates where appropriate filters can be selected, accordng to the above-described strategy, to avoid imaging interferences.