[E‘iSTANTANEOUS ENDI‘JIDUAL FLOW RATE MERSUREMEN? m A COAL - AER SYSTEM Thesis for the Degree of M. S. MECHEGAN STME UNIVERSITY EDWARD CHARLES HOHMANN, ER. 1967 TH E: LIBRARY D *s ' t' J. .“ I- . L! 5,} l I:- 1m' mw‘” ~ csity ; Stare INSTANTANEOUS INDIVIDUAL FLOW RATE MEASUREMENT IN A COAL-AIR SYSTEM By Edward Charles Hehmann, Jr. A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree 0! MASTER OF SCIENCE Department of Chemical Engineering 1967 A “v f \.". (7 “+37. \.' . ABSTRACT INSTANTANEOUS INDIVIDUAL FLOW RATE MEASUREMENT IN A COAL—AIR SYSTEM This research effort is the initial step in the de— velopment of an integrated control system capable of con- tinuously'maintaining a pulverized fuel energy conversion system in an optimal economic operating configuration. Attention was directed towards the development of an ac- curate mass flow meter for instantaneous individual flow rate measurement in-a pulverized coal-air system. The isokinetic two-phase sampling tube developed was used to measure both coal and air flow rate at points within the twelve inch fuel lines of the Michigan State University Power Plant. Velocity profiles were obtained and are graphically presented. In general, the flow measurement technique used proved to be very good for determining velocity profiles. No suitable alternate method for measuring the profiles in these large conduits was found in a search of the litera- ture on fluidization, two-phase flow, and power plant de— sign and operation. It is anticipated that this isokinetic sampling method can be used to calibrate continuous flow measurement devices to be developed in subsequent portions of this control project. AC KNOWLEDGNUZTNT The author wishes to dedicate this thesis to his parents. Of all his counselors, they have been the most helpful. Location of the UniverSity Power Plant and the as- sistance of Mr. Jesse Campbell and Mr. Wayne‘Iates made it possible to conduct this research at a full-scale plant. Organizational and clerical needs for this project were ably accomplished by the staff of the Division of Engineer— ing Research, Michigan State'University. The Shop Services Supervisor, Mr. Donald Childs, offered valuable suggestions and assistance in the fabrication of equipment. The au- thor is also grateful to the Detroit Edison Company for providing financial support for this research and for the author during the summer of 1967. Especially valuable guidance was supplied by Dr. George Coulman during this research and the author wishes to express his sincere appreciation to him. ii TABLE OF CONTENTS Page ACKNOWLEDGMENT ..................................... ii LIST OF FIGURES .................................... v Sections: 1. INTRODUCTION OOOOOOOOOOOOOOOOOOOO0.0.0.000000.00 1 1.1 Total Project 1.2 Study to Date 1.3 Flow Monitoring ’2‘. BACKGROUND .00....OOOOOOOOOOOOOOOOOOOO00.0000... 6 2.1 Pneumatic Conveying 2.2 Pulverized Coal in Power Plants 2.3 Difficulties in Two-Phase Flow Studies 3. PREVIOUS EXPERIMENTAL INVESTIGATIONS AND THEORY OF SOLID‘GAS FLOW 00.000000000000000...000000000 16 3.1 Velocity Relationships 3.2 Friction Loss Relationships 3.3 Solid and Gas Velocity Profiles 3.u Two-Phase Measurements A. DESCRIPTION OF THE EXPERIMENTAL EQUIPMENT ...... 25 h.l Circular Pitot—Sample Tube h.2 Additional Equipment h.3 Power Plant 5. MPEMNTAL WORK .0..........O................. 37 5.1 Calibration of Equipment in One-Phase 5.2 Flow Profile Measurements 5.3 Total Boiler Input-Output Measurements 5.h Time Constant and Humidity Experiments 6. DISCUSSION .0..0.....................00......0.. hg 6.1 Correlation with Other Two-Phase Measurements 6.2 Flow Stability iii iv Page 6.3 Attainment of Objectives 7. CONCLUSIONS CCOCOCOCCCOOOOOOOCOOOOOCO0....0.0... 52 8. RECOMMENDATIONS FOR FUTURE WORK ................ 5h 8.1 Continuous Flow Measurement 8.2 Coal Heating Value Measurements 8.3 Pulse Phenomena 9. BIBLIOGRAPHY .0.00.0.00...OOOOOOOOOOOOOOOOOOOOOO 57 APPENDIX A "' COAL PROPEFITIES ooooooooooooooooooooooo 60 APPENDIX B - ALTERNATE SOLID-GAS FLOW MEASUREMENT SCHEMES 00000000000000...0.00.00.00.00. 66 10. ll. 12. 13. m. 15. LIST OF FIGURES Schematic Diagram.- Furnace I Michigan State University Power Plant 65 ................... Schematic Diagram - Isokinetic Flow Measurement Equipment ....................... Schematic Diagram.- Isokinetic Sample Tube .. Twelve Inch Fuel Lines Leaving Riffle ....... External View Pulverized Coal Burner ........ Isokinetic Flow Measurement Equipment ....... Isokinetic Sample Tube ...................... Differential Pressure Measurement Equipment . Calibration Curves for Flow Measurement Equipment 000.00.000000000000000000000000.000 EaSt'weSt VBIOCI‘Cy PrOfileS o *. o o o o o o o o o o o o o o o North-South Velocity Profiles ............... Table of Coal Particle Size Distribution in Fuel Line 0.0000.00.0.0.0...OOOOOOOOOOOOOOOOO Profile of Coal Particle Size in Fuel Line .. Typical Coal Particles Magnified 100x ....... Coal Particles Magnified to Show Shape ...... Page 31 32 33 31; 3h 35 35 36 no ’45 62 63 6h 61L 1. INTRODUCTION This research effort is the initial step in the de- velopment of an integrated control system.capable of con— tinuously maintaining a pulverized fuel energy conversion system.in an optimal economic operating configuration. 1.1 Total Project The Detroit Edison Company is sponsoring this re- search on pulverized coal power plants to gain a.more fun- damental understanding of the dynamics of the fuel supply and combustion system. It is anticipated that this under- standing will lead to a functional control system for the integrated power plant. A.pu1verized coal power plant operated by Michigan State University is being used for early work on this pro- ject. The plant, completed in 1965, consists of two 250,000 lb. per hour boilers and has an electrical gener- ating capacity of 25,000 kw. The project consists of four related efforts: (1) development and evaluation of pulver- ized fuel and air flow rate monitoring techniques, (2) de- termination and monitoring of on-line fuel energy value parameters. (3) study of process dynamics and kinetics of the energy conversion operation, and (u) development of dynamic optimizing control algorithm. 1.2 Study To Date The initial effort in this control system.project was directed towards obtaining precise mass flow measure- ments of the coal-air stream, although limited studies were conducted on fuel energy value and process dynamics. Flow measurements were considered most important since a wide search of the literature on fluidization, two-phase flow, and power plant design.and operation yielded income plete theoretical and little practical information. In essence there are three stream properties that must be measured to determine continuously individual flow rates in a solid-gas flow system. These are total flow rate, stream temperature, and effective stream density. However, with such a.metering system, accurate calibration is essential to determine the flow’rates on.an absolute basis. For this reason the development of an accurate cal- ibration device was attempted first. It was decided to construct an isokinetic two-phase sampling tube that could also be used as a circular pitot tube (Figures 3 and 7). Sampling with this tube it was possible to take point samples of both.the solid and air flow rates. Using it as a pitot tube gave an indication of combined flow rates. These capabilities were utilized to determine flow distribution in a vertical section of the fuel supply line for both the coal and air. From.this distribution a rela- tionship was obtained between air and coal flow rates (section 5). The boiler output (pounds steam/hour) was also cor- related with pressure drop across the fuel supply line and a series of point measurements with the sample tube. A study was also made to determine response times of the boiler and associated equipment. 1.3 Flow Monitoring Since continuous monitoring is necessary in a con- trol system, total flow rate, stream temperature, and ef- fective stream.density will be measured with standard in- strumentation that can be calibrated with the aforemen- tioned sample tube. a.venturi meter could be used to determine the total flow rate since solution of the mechanical energy balance will lead to an equation of the form: v»=ce;> (oh-I where V“ = total mass flow rate C = constant AP = pressure drop effective density of solid-gas mixture. Q“ However, the Michigan State University power plant system lacks the room.to insert a.venturi in the pulverized coal feed line. The line runs from the pulverizer to a riffle A (device to split stream) to the burner. In this line (Figures 1 and A) there is a pressure drop of about 8 in- ches of water. It is possible to use the pulverized coal feed line itself as a differential pressure meter, since the solution of the mechanical energy balance for the fuel line will lead to the same equation form as the venturi. Stream.temperature is easily measured by a thermo— couple inserted into the fuel line. A radiation density meter, see Appendix B, would be used to calculate the effective density of the pulverized coal strean without inserting an obstruction into the line and disrupting the flow pattern. Using the principle of conservation of mass one may write: PMs-stig-PEE and ,Dp = density of particles pg = density of air Cp = concentration of particles (1b particles/lb mixture) C = concentration of air (lb air/1b mixture) VH = mass velocity mixture V = mass velocity particles V = mass velocity air T = temperature Then if Vp is known as a simple function of V8 and fog as a function of T: vp = r(vg) and ,08 = ¢ (T) One may determine both VP and V8 as functions of V}! which can be measured as mentioned above vp = g(VH) and v8 = 11w”) The actual relationship between solid particle ve- locity and gas velocity is Shown in Figures 10 and 11. 2. BACKGROUND 2.1 Pneumatic Conveying Pneumatic conveying of solids has been used for quite some time to transport light, bulky materials, such as raw cotton, woodshavings, and wool. However, within the past decade many new installations incorporating pneu- matic transport of pulverized coal, grain, finely divided ore, organic, and inorganic chemicals have been built. Much information is available on the different types of conveying systems, system.selection, equipment, and piping.1’4’27 Hewever, a great deal of this material is not connected by theory or precise empirical relation- ships but rather by I'rule-of-thumb"techniques.20 This is not because of lack of effort or experimental study but because of the exceedingly complex nature of solid-gas flow. There are several modes of flow possible depending upon the important variables grouped below‘:5'39 1. pipe size, roughness, and orientation; 2. solids density, particle size, shape and size distribution; 3. solids-to-gas weight-rate ratio; and A. gas velocity. The solids may be fully suspended and fairly uniformly dis- persed over the pipe cross section at one extreme or they 6 7 may form a layer along the bottom.of the tube, only the surface of which.moves, at the other extreme. Medes of flow between these two extremes can.be easily observed and because of the different paths particles take in each of these modes it is impossible to write one theoretical or empirical equation to describe all modes. Usually the flow is characterized (and specific flow equation chosen) from.the ratio of gas velocity to minimum.carrying velo- city* and from.the solids concentration (dense or diffuse flow). The flow equations, however, are generally rough ,approximations to reality. Pressure drop is of primary concern to a designer of the aforementioned conveying systems both as a function of the tube or dust system.and the flow rate of solid ma- terial. For this reason.most of the research associated with pneumatic conveying has been directed toward pressure drop determination. Some relationships are commonly used and will be presented later. Another area where solid-gas flow occurs is fluid- ized-bed.reactor systems. There are several books and papers in this area that deal with the flow of the cata- 3h'39 lyst (solid) in a gas. In a reaction design there is more interest in the time the particle (catalyst) spends *Velocity required to convey particles in gas. Theoretically it is the terminal falling velocity, though many particles are too large for Stokes Law (1-100 microns) to be strictly applied. 8 in the system than.there is in a bulk conveying system, hence, interest in the instantaneous solid and gas veloci- ties is greater. 2.2 Pulverized Coal Systems In Power Plants The firing of pulverized coal has largely developed as an empirical art with theoretical understanding gen- erally following rather than preceding practical accmm- plishment. The chapter on 'Pulverizers" in Combustion Engineering7 gives an excellent historical background. The initial incentives to burn coal dust were that it was virtually a waste material and by reason of its small size ought to combust very rapidly, whereby permitting a reduc- tion in furnace size. However, in reality general use of pulverized coal and realization of these advantages re- quired development of equipment over many decades. The "direct firing” system is the simplest of the three commong methods of supplying and firing pulverized coal to a furnace. It consists of crusher, classifier, air supply, and tubes to convey the coal directly to the fur- nace. Other systems have long or short time storage hop- pers and suitable collectors between the crusher and fur- nace. Hot air or diluted furnace gas supplied to the pul- verizer furnishes heat for drying the coal and transports the coal to the furnace. This air is known as primary 3;; and is part of the required combustion gig used in the furnace. 9 In the simplest boiler ductwork system.there is a main air duct from.the forced draft fan to the air heater and to the burners. In addition, there is a primary air duct from.the main hot air duct to the pulverizers. From the point of takeoff of the primary air, the air going to the burners is called secondary air in spite of its large volumetric flow rate. A main gas duct carries the flue gas from.the economizer outlet to the air heater to the in- duced draft fan to the stack. If the furnace is pressur- ized (an increasingly common practice) and without an in- duced draft fan, the main gas duct runs directly from.the air heater to the stack. The fineness to which coal Should be pulverized de- pends on.many factors. Type of coal is of prbme import- ance. Some coals are of low volatility or swell when ex- posed to furnace temperatures, while others are quite vola- tile and ignite and burn readily. Large furnaces may op- erate on coarser coal than smaller ones simply because of larger furnace volume, length of flame travel and furnace turbulence. The fineness of pulverized coal is determined by passing the coal through standard sieves (ASTM D 197-30). Since the advent of pulverized coal firing fineness speci- fications have changed considerably. Initially specifi- cations called for 95 percent minus 100 mesh and 85 per- cent minus 200 mesh. Now individual systems and coal types dictate a variation in the minus 200 mesh (7h micron) 10 specification from.65 to 85 percent. The minus 100 mesh requirement is largely replaced by specifying that plus 50 mesh (300 micron) material not exceed 2 percent of the total. It can be seen from the above specifications that it is important to have a large percentage of fine particles, thereby insuring a large surface area, when.combusting pul- verized coal. However, it is equally important not to have oversized (greater than 50 mesh) particles. As little as 5 percent plus 50 mesh.may produce furnace slagging and increased combustion loss (carbon loss) even though com- bustion conditions are excellent for the finer coal and the total surface area.remains unchanged by increasing the minus 200 mesh fraction. Combustion efficiency is normally affected more by larger size composition, whereas grinding power for the pulverizers is more closely related to the finer size per- centages. For these reasons fineness samples are checked periodically in an effort to maintain an acceptable product from the pulverizers with.minimum.power consumption. 0n a.direct fired system.these samples must come from.the fuel line to the furnace. A sampling device, con- sisting of a sampling nozzle with a hose connecting it to a small cyclone collector and sample jar is used. To get a representative sample the nozzle must traverse the whole pipe diameter in two mutually perpendicular paths, both 11 of which are perpendicular to the axis of the fuel line, at a point where the pipe is as free from bends and other disturbances as possible. Pulverized coal feeder and fuel piping determine the maximum primary air temperature that can be used in the system. This may be as high as 500°F, although a low- er temperature is more common. If 130°? air is used to convey and dry the coal, the relative hmnidity may be as high as 70 to 90 percent. It is important to maintain the relative humidity of a system below the saturation limit while drying the coal sufficiently to insure good combus- tion. In direct firing systems higher temperatures may be used since the danger of fires such as may occur in storage bins is less. Common temperatures range from 110° to 200°? depending on the type of coal and moisture content. The pressure in a pulverized coal conveying system depends upon the system resistance resulting from coal and air flow through the fan, ducts, and the pressure in the furnace as well as the source and temperature of the air. Furnace pressures may range from 8 to 20 inches of water. As mentioned earlier the velocity required to convey a particle depends partly upon the density of the particle and its size. The density of coal ranges from 50 to 90 1b/ft3 and the particle size in most systems varies from less thans'm microns to greater than 300 microns. With these wide variations it is difficult to give values of 12 minimum conveying velocity but frompractice10 typical fuel line velocities and densities are: 2.5 Density, lb air/lb coal 1.2-2.5 and over Velocity if horizontal, ft/sec 80-90 75-80 Velocity if vertical, ft/sec 80-90 u5-60 At present the methods of measuring coal flow into a furnace consist entirely of batch methods: usually automatic or manual weighing-dumping equipment incorporat- ed in the fuel handling system. Typical accuracy of the automatic scales is within three to four percent of true total for 10 to 15 dumps. An alternate procedure deter- mines fuel rate from.a unit heat balance and fuel heating value. This is generally considered the more accurate method, although good fuel sampling becomes important. Air flow'measurement is usually accomplished through a velocity head type meter, where a constriction in the pipe or duct causes an increase in velocity and a corresponding decrease in static head which.may be mea— sured. Orifice plates are probably the most common con- striction, although venturi sections are sometimes used if space permits. A pressure differential may also be ob- tained with an airfoil, which has the following advantages over a venturi section: higher differential pressure with lower permanent loss and shorter length. The airfoilah has been tested in an operating power plant and provided good results in a duct 5 feet by 6 feet in cross section. 13 The differential pressure was found to be more exactly proportional to the square of the flow than with the other differential pressure meters in use. Accuracy of flow measurement by pitot tube techniques are limited because of large velocity profiles in the ducts. An air flow measurement that is considered superior to the above methods is accomplished by means of an orsat analysis of the flue gas.7 This method gives an interpre- tation of only the overall air flow, using 002 and 02 con- centrations, and cannot be used for study of air flow dis- tribution within a system. At present there is no method, that enjoys wide use, for obtaining both coal and air flow rates in a con- veying system. Flow modeling of furnaces and ducts can be success- fully used to avoid pitfalls of extrapolation from.smaller 6 to larger units. Combustion Engineering devotes a good portion of Chapter 7 to this problem and presents teoh- niques for showing stream lines using either air with smoke or water with solids. These methods might be of help in developing flow meters for coal-air flow. 2.3 Difficulties in Two-Phase Flow Studies Typical of the difficulties sighted by experiment- ers in the literature is the following statement by Mehtaz6 about solid-gas flow. "Plugging of the pressure 1A taps and tubes to the manometer also prevented accurate measurement of the pressure drop.” In plant experience of the Detroit Edison Company and the Michigan State Univer- sity Power Plant personnel is even.more discouraging. All of the experimental work described in the literature dealt with a less hostile environment than that in an operating power plant. Initial observations by the author showed that even taps as large as‘% inch in diameter became plugged if not cleaned regularly. The coal particles, although small, tend to bridge any openings that do not have a flow through them. To prevent these difficulties the author's sampling equipment and pressure taps were designed to be purged re- gularly. Precautions were also taken to purge the sampling tube, cyclone separator, and rotameter section of the coal collecting apparatus. The system.was purged before each run and this resulted in two advantages: (1) all openings and taps remained unplugged throughout the entire study, (2) all coal was blown out of fine sampling system before each run, thereby accurate coal samples could be taken. It should be noted that the circular pitot-sample tube used was especially designed to Operate without plugging (Fig- ures 3 and 7). Another difficulty that may be encountered in power plant sample taking is agglomeration of the coal in the 15 sample equipment due to the presence of water. The fuel air mixture cools considerably on being removed from the pipe and if the relative humidity is high enough, the gas in the sample tube will be below the dew point, leading to water drop formation. This did not occur in the present study, since the relative humidity in the sample equipment was less than 70%. Insulation on the fuel lines from.the pulverizers to the burners, not a common industrial prac- tice, helped prevent this problem.at the Michigan State Power Plant. 3. PREVIOUS EXPERIMENTAL INVESTIGATIONS AND THEORY OF SOLID-GAS FLOW Progress toward understanding the transport of solid particles by gases has been slow and the few quanti- tative relationships that are available from the litera- ture in this field do not include all relevant variables. These relationships are therefore limited to certain, often undetermined, ranges of the excluded variables. In.the following paragraphs only the experiments and relation- ships most likely to illuminate the problem of pulverized coal measurement are discussed. Hence, dense phase trans- port isignored completely and modifications of the flow equations for vertical tubes are not presented. (Dense phase usually indicates a particle concentration greater than S 1b/ft3, while the dilute phase is of the order of o to S lb/ft3.) 3.1 Velocity Relationships For a certain gas stream traveling at superficial velocity V, a fixed amount of solid, W, can be carried in dispersed flow. The minimum.required gas velocity for this situation is called the saltation velocity. Zenz39 gives a plot which.shows in schematic form the relation- ship between pressure drop per unit length and V for dif- ferent constant solids mass rates. It is interesting to 16 17 note that the lines of constant W are discontinuous at the saltation velocity because solid.materials starts to col- 1ect on the bottom of the tube, 1.6., saltation takes place. Some experimenters have failed to realize that thissaltation was taking place during what they thought were steady-state periods. Both the saltation velocity (for the gas stream) and the critical conveying velocity (for the particles) can.be calculated by empirical correlations given by several au- thors.27’39 The main use of these velocities is in design, where it is necessary that the flow rates exceed the ori— tical values. Unfortunately, the accuracy of the equa- tions is not great enough to permit their use in other applications. If the experimental data available on powders of uniform size (i.e., glass balls) is said to be incomplete, the data on.mixed particle size powders, such as coal dust, is virtually nonexistent. However, the available in- formation indicates that using techniques such as geometric mean particle size and an additional parameter for degree of size distribution will permit the use of empirical equations based on uniformly sized particles for mixed particle size calculations. Stockel3h took a completely theoretical approach to the high speed flow behavior of gas fluidized solids. 18 He derived equations describing flow in a conduit of chang- ing area, using a simple model restricted to one-dimension- al, frictionless, steady flow in the absence of gravity. When he solved the flow equations they gave good approxi- mations to the experimental results he had obtained. 3.2 Friction Loss Relationships The friction losses in a pneumatic conveying system are also of interest, since they result in a pressure drop. 27 An interesting approach described by Molynux and others37 relates the friction losses of a gas-solid mixture to that of the gas alone by a factor, B. B is always greater than one and is calculated from graphs of dimen- sionless variables obtained from.experimenta1 data. A typical value of B given for coal dust is 1.16. Breaking the problem of friction loss down into smaller parts, each perhaps more manageable theoretically, has been the most recent approach to pressure drop calcula- tions. Hinkle, in his Ph.D. thesis at Georgia Institute of Technology, has presented an extensive study of the various factors using data from.his experiments in 2— and 3—inch glass pipes. The six forces contributing to pres- sure drop in dilute conveying are: (l) The friction of the fluid against the pipe wall. (2) The force required to move the solids' mass 19 through the pipe. (3) In vertical pipes the force required to support the weight of the solids. (h) The forces required to support and accelerate the fluid. (5) The forces required to support and accelerate the solid material. (6) The friction between the pipe and the solids. Summing these factors Hinkle obtained the following rela- tionship: AP _sfs_2__’°s_q_ V“) total +f v v horiZe ~ G G PG where f conventional Fanning fluid-to-pipe friction factor; f fluid-to-solids friction loss in conveying p or f = 3E0 GD DT __(V_e___v " __V_p>2 p 29p DP For the above formula to be useful one needs to know the velocity of the particles. From.high-speed photo- graphy Hinkle determined particle velocity and developed the empirical relationship: V _. 0.3 0.5 V8 - 1-o.17913p I‘P This represented observed particle velocities to within 20 i 5 percent, but the more common assumption that vp = vg - v‘ , predicted the particle velocities with no more than i 20 percent error. (V: is a constant representing the lag between gas flow rate and solid flow rate. It may be taken as the saltation velocity.) Experiments by Mehta26 showed that particle veloci- ty varied linearly with air velocity for 364micron par- ticles. (Smaller than Hinkle's 0.01h- to 0.33-inch par- ticles.) However, Mehta's 97—micron particles tended to reach.a maximum velocity in the range investigated. It is difficult to compare the results of Mehta and Hinkle since the particle diameters differ by two orders of magnitude, but Hinkle actually measured the velocity of individual particles while Mehta used two fast acting gate valves in a section of pipe to "capture" the particles in that sec- tion at a given time. 3.3 Solid and Gas Velocity Profiles Of all the literature the author searched, only the two articles discussed in this section provided an indica- tion of what velocity profiles were to be expected in a two-phase flow system.very roughly approximating the power plant fuel line. Other investigations described in the literature used high speed photography, pressure head loss 21 methods, or the "capture technique"% to study two-phase flow. For obvious reasons these techniques do not yield any information on flow profile in a large conduit. Doig8 used 300 and 750 micron glass spheres for the solid phase, somewhat larger than pulverized coal. His conduit was h3-mm.glass tubing ho feet long and suspended in a vertical position. He was able to Obtain.measurements of the air-phase velocity profile with a specially designed cylindrical pitot tube that could be moved across the con— duit perpendicular to the flow. Flow velocities were ap- proximately 20 ft/sec (close to choking velocity), but even at this speed conventioned pitot tubes were rapidly damaged by the particles. Doig made no attempt to measure the solid velocity profile; however, he discusses at length the change in air velocity profile due to the presence of the solids. At low average ratios of particle to air concen- tration the dispersed phase caused the air velocities re— lative to the single phase profiles to increase in the core and decrease at the wall. At high average concentration ratios there is a decrease in core velocity, tending toward plug flow, While the air velocities increase toward the wall. 8003.3 studied particle concentration, mass flow *This method consists of rapidly closing both ends of a section of the test pipe while flow is taking place. The section is then removed and the amount of solid ”cap- tured" is determined. 22 rate, and solid particle velocity of glass beads (50-mi- cron) and magnesia (35-micron). These are about the same size as the coal dust particles of interest. He used a fiber—optic probe to measure concentration and an electro- static probe to study flow rate of the solids. The air flow rate was about 130 ft/sec. Conclusions reached were that concentration of solids increases towards the wall of the pipe while the solids' velocity decreases to perhaps 20 percent of the core velocity at the wall. Also the fluid seemed to follow the 1/7 turbulent velocity law while the particle velocity varied nearly linearly with distance from.the wall to about the 2/3 power. 300 agrees with Mbhtaiin that the solid velocity varies approximately lin- early with the gas velocity. It seems that for this size particles VP 2 1/2 Vs‘ 3.u Two-Phase Measurement Most two-phase flow measurements rely on convention- al methods, such as Doig's, to determine air flow. Howe ever, methods for measuring mass flow rates directly re- quire something different. Most particle flow rates are based on a certain fraction of the particles that interact, proportional to their velocity, with a measuring device. This could consist of electrostatic charge transfer to an electrostatic probe or a condenser, or it might be momentum transfer to a.mechanioal target. ‘With a stretch of the 23 imagination, one might also include change in inductance of a coil to flow of a.material through the core to be in this class. The chief difficulty with these methods or methods that sample the flow is that the particles measured may not be representative of the whole flow. Lieberman23 discusses this problem of representa- tive sampling of fine particles. He also discusses the hm- portant matter of distortion or modification of the par- ticles or flow during sampling periods. Isokinetic* sam- pling is very important and Lieberman presents a graph of the relative error that results in nonisokinetic sampling for different ratios of air speed in conduit to air speed in sampling tube. Particle size is a parameter on this graph, but for larger particle sizes the error in particle concentration can be as large as h0%. Indirect methods for particle flow rate measurement usually consist of some sort of attenuation process of a transmitted signal, be it sound, microwave or light, or other electromagnetic radiation. Nuclear radiation density methods also fit into this class. The mass concentration can be determined and then if particle velocity can.be re- lated to gas velocity the mass flow rate can be calculated. (See appendix B) Perhaps the most novel method noted in the *Velocity in the sampling tube is equal to the velo- city in the flow conduit from.which the sample is taken. 2h literature for measuring two-phase flow involves a nozzle 1 with a sudden expansion section. Ackerman used a nozzle of this design to develop a Solid-Liquid Flow Meter. He solved the momentum and energy relationships simultaneous- ly and was able to determine total discharge as well as solids discharge from the nozzle. 3.36 Some recent foreign articles appear from their abstracts to contain information that might be useful in developing a solid-gas flow meter. Barth3 claims to have developed a new measuring nozzle, which from the diagram in his article appears similar to the meter developed for liquid-solid flow by Ackerman. h. DESCRIPTION OF THE EXPERIMENTAL EQUIPMENT This research was somewhat unique for a.master's thesis project in.that the experimental work was carried out in a full scale Operating power plant. Advantages of this large scale research were: flow, temperature, pres- sure, and particle conditions conformed exactly to those of an actual plant. It would have been difficult to achieve this in a laboratory. Stochastic factors present in the operating power plant also graphically displayed equipment malfunctions that would not have been discovered in the laboratory. Chief disadvantage encountered was that the operating personnel had to meet the power load and routine maintenance requirements before experimenta- tion could be conducted. h.1 Circular Pitot-Sample Tube The circular pitot-sample tube is shown in Figures 3 and 7. This tube is of the author's design and was an attempt to construct a device that could characterize two- phase flowvmuch like a pitot tube characterizes single phase flow. The device constructed by the author should work on both gas-solid and liquid-solid flow, although it was only tested on the former. The pitot-sample tube consists of an aluminum.tube 25 26 one and one-eighth inch in diameter and approximately twenty inches long with a rectangular 7/8 x 5/8 inch notch in one and (see Figure 3). An aluminum asymmetrically tapered plug fits in this end and seals the tube except for a rectangular hole. Two pressure taps are also in— stalled in the tube assembly. One on the outside measures static pressure in the flow conductor and is perpendicular to the large rectangular hole which faces into the flow. The other tap is located inside the tube and measures the pressure rise within the tube. (This is the stagnation pressure if the open end of the tube is blocked). In operation the pitot-sample tube can function in either of two possible configurations, with the end oppo- site the rectangular hole open or closed. W1th.the open end of the tube blocked, it responds like a pitot tube with a somewhat lower coefficient (0.67). (The coefficient proved to be constant over a wide flow range much like a regular pitot tube; see Figure 9). For two phase operation it was found that although the pres- sure in the tube rose immediately to a constant value, the large opening and the volume of the tube prevented plugging until reliable pressure readings could be obtained. It was then a simple matter to blow compressed air through the tube removing the solid material, thereby making continued measurement possible. With the open end of the tube connected to separa- 27 tion equipment as described below, it was possible to as- pirate the separation equipment and tube so that the pres- sure inside the tube was equal to the pressure in the flow conduit as measured by the outside tap. In this condition none of the kinetic energy of the fluid flowing in the conduit was being converted to potential energy in the form of pressure difference; hence, the fluid flowed into the tube as fast as it flowed in the conduit. This proce- dure is called, isokinetic sampling. With this mode of op- eration it was possible to sample a portion of the two- phase flow in a large conduit without disturbing the flow patterns or causing a segregation of the phases. Separa- tion equipment described below was then used to segregate and measure the individual phases. 14.2 Additional Equipment Along with the pitot-sample tube it was necessary to have equipment to separate and measure the individual phases. Figure 2 is a schematic diagram of the separation apparatus used and Figure 6 is a photograph of it. The pitot-sample tube is connected with 1 inch I.D. tygon tub- ing to a short section of copper tubing extending from a cyclone collector. The effluent gas from the cyclone then runs through a cotton filter and to a rotameter which mea- sures the air flow. Coal flow is determined by weight collected per unit time from the cyclone. An air ejector 28 follows the rotameter and is used to control the pressure in the tube so that the flow is isokinetic. Several purge points were included as shown and by proper closure of valves the system could be freed of coal after a measurement period. All fittings and tubing trans- porting the coal-air mixture were sandblasted to remove roughness and prevent accumulation of coal. Thermometer and pressure taps were also located as shown. The wet and dry bulb thermometers after the cotton filter made it pos- sible to determine relative humidity at any point in the system where the temperature was known. This was done by using a standard humidity chart and making the assumption that no water escaped from.the system. A pressure test was conducted on the whole sample apparatus and leaks were stopped so that the equipment would hold air at 10 psi for an extended period of time. This was considered adequate proof that only the air enter- ing the sample tube would pass through the retameter. Coal samples were weighed on a two pan balance with an accuracy of 0.1 g. and coal volume measurements were made with graduated cylinders. Pressure drop across a section of burner fuel line (used in total input-output study) was determined with a Leeds and Northrup Flow Transmitter. Full scale deflection of this particular transmitter was five inches of water and it gave trouble free performance. Figure 8 is a photograph 29 of the pressure sensing cell and recorder. h.3 Power Plant Michigan State University Power Plant 65 is a steam and electrical generating plant. Two 250,000 lb. per hour, two drum, single pass boilers and tangent-tube fur- naces are coupled to two fully—condensing, single-automa— tic extraction steam.turbines, with direct connected gen- erators and exciters. Each boiler-turbine generator is capable of operating entirely independent of the other boiler-turbine generator. This was helpful in the present study in that one of the boilers could be essentially iso- lated from.the power system by setting it at a fixed out- put and meeting load fluctuations with the other boiler. The plant is operated from.a central control room from which start-up, shut-down, and load variation adjust- ments can.be controlled. Coal is delivered to the plant by railroad and stockpiled on the site. A series of belt conveyors trans— port the coal to bunkers with a 1,600 ton capacity. At full load consumption of 2h tons per hour this is almost a three day supply. Therefore, it should be noted for future study that the weater effect on input coal (see section 8) is governed by a large time constant. The coal flows by gravity from.the bunkers to automatic scales which are set for two hundred pound dumps. Following the 3O scales are feeders that provide a constant flow to the Riley Pulverizers. From each pulverizer the coal travels 20 vertical feet to the furnace floor where the flow is split by a riffle (Figure 1;) and then enters the furnace through two burners (Figure 5). 12” coal pipe upper burners (supplied by other pulv.) ~‘ ports _ 0 lower 0 burners 56 I T 5: I‘— S'H +— riffle 6' (splits flow into 4' two 12" pipes) N * kg? “7, sample ' " shufi'iiagzw ‘ u v.0" " ' ‘ l 4 Furnace Floor -—I 18" coal pipe 24' ; pulverizer floor “:1.-er“7 .‘ - a we. fpn" “M35 ' ' :‘o .1. ' - ID 3“, '1') It‘“ e ,0..fl?1.‘ l Al Front Elevation Figure l. Furnace I - coal pipe, Michigan State University Power Plant 65. 31 cotton rota-star filter 9"91’ not &:dry bulb ITFtIIIE'i Pa" in :‘ couture pip- ———/""1 -¢—— 1 -q—- I oJoctar v: v. «12le Figure 2. Inchinntio Flaw ‘lbnaurulnnt Equip-Int flow direction stream tap sample tube tap ”[llflll/jlllllj [ll/[Illjllli 25 ‘J Section A-A ta. 3 are of A l 8-inch copper End View tubing T . - i stream H L tap l 5/3H . sample J * tube tap F 7/8".‘ . (taps are inside tube) TOp .View Figure 3. Isokinetic Sample Tube. 33 . ..... . o...3 i o ufil.s-‘\‘-u . . _.. . (iii; . .3 iii$...v....... ..... . OI I o... l‘ . . . . . 2.2.5.4 o34.-.»1¢.9.§1d..0116;3o.33 . . o o O C C O O . a uIOOOOQ'OQOIOQC‘IOI ”brow-on“. . .. CC...” C" . . ..oo 0 I .II . . . .. . . ..... . ....... .u‘. o oounjocu-hfiuu .mrtwrtzinfiw%XEM£mw .i&kfififi€§ififififiiiéfiéisififififla Twelve Inch Fuel Lines Leaving Riffle FIGURE h Leaving Riffle Twelve Inch Fuel Lines 5 FIGURE 35 Flow Measurement Equipment Isokinetic FIGURE 6 A2 3’ g \ Sample Tube tic Isokine FIGURE 7 36 .ItOicOIOc 'I “nopoool.’l¢.. cooucirav IN.‘ V 001......EO 'III" 0 O .V 9.0.0! d O. o. v... I'DovofCo. 0‘- ‘v 1.. .40.)... 00h... o. n. .v..oo..’- I ‘ I I pr... . -.c.AI-¢..O .0071 cc... . ... ... blelsIvlogoOIc ‘1‘: 9.. u . . ccocoenno. D! I COMIJII .. . ....o .lVOoII I. II C 0...... . o.. ..o nun. so, o\‘oa0p as... ... . ... .p».......o.uoo-..c . p -.oouo -.. .. . . ..oaln ......o.0 0.. .u.u . .- oc. . .. .. . ell no u ‘0‘ . ~|ooon O u .. .. .. I: e - U~Ooooo II on n u no. I! ._. no .. .0 oh. ,0 a. "Docl .~NoIv-u. _..-coo. .- . . . ... . ...0¢¢.. o..-.... o. v...... ..... ...,...........ooio o....:............ .0. t. .c a u . .. . ...... o . . . . . .o...o.a . \o...”'. o. o . a a . s o - . ... . . .. . . . .. . ... .. On... .. .00: to... v . . . . . n‘u , . .. n . u...a .o. . g 5: uunoo..o- nu. O o o I Inc. ..|u .0. .1. a I I v '9 l a ’I .0. a u . ....... ..-..u.o‘.c..noc. ... .5..-.Ioo-I. 00....04aonaklt I. u....-; .....va......v ‘.- .01. 0.... Inst. ‘0'... a. . ...... .oo. .- ... . ...‘t.DI- :.- a.b.)l Cl Cl 0.. .u - -\..IOI a... .C-nclliio I... O. 0”... cl. u ..., -.I .v.-I.uo.a.. . .a. i..oc-o-- 0....I'o“ 0.1 tan-cluolOOICOOI‘IIICQ'0(00~IUQ ~\ 0. .. u.o.o. _... - . .. s 4- I . 0.0.0 5!... it. o .. n. n . ..- .. ...01I\0a|oo cu ..o.‘o| u.uoo; C....n I . e. I... OIOO-l '00,. . .v. IDO‘ 4.9- 1'. 0| .0. ..... . ...\.a ....c...«-.... v..- vo..OI.DoI-‘“0A. no.u...VaoosIvsca..oIoIboo...-4I-‘Q‘ 0.. .o... i .. ......o....- I. vo..-I.¢I’|05!C-ovousoo no. con-v.00. .uoo . I I ‘ .. .u. on o.n.".-o-.pubic-II.bus-oou-FCIICOO¢C.n a... 03-0....0 1- 6 .. a. . . u . ac... o . .\. VIC. o.o§a\001ea . .1 .su. . . ou....... .0..v usual. .0;- n ..¢U . l v. v... . . . . ..I.o \ . s n .uu‘s.n. . u I I s O o a etc-cosrov 0 v .... ... .c .. ... ..... . ... o. A o. o . nace.. . .. .o. .... . o...... oo (nace.. I I .. a .z. .zc. . .ov.. u.- . o. ..Ou.-. . .v...a. . I. o u. in. . .c . I. C I .I. nu. .. a In I . ,0.- . ~_ ..... n \ n I .n.‘ ...... - ... n .. I u 0 O I I .. . ..... o 4. O I .. ..... . .. Q. . . billet-IA I .. ..... t 9 coo-c... n uh . u . a .u ........ n ....... ...... ......... ...... u .i.. . .no ...cnc..0|~0.0.l. . .....-o.- Ins. , . .. a._..u...2aw:s:rv:ww~atu. .- . a '.-u. u; ,0.... _.. I . t O;{. I . I . .. I it...lo.D\.~ ..I..l....l..:... .0.‘ .0'.II.JQ‘II.I'{I‘WH o . . .... ... ..o . .......... ...~COD - on . ,v. .0!..l.l-.-. ‘ ...nI|u .QO - . .. . pi 0.1 a DC . .3... . . fl . .,. . .... .. . I.o.o.n(.a.o,ir.. .. . ~.I......coou..o . o .0- n | - . . ... . . o A . .oa....,-... .uv..- rival... .oc.u.-\ 0...»...9 o . .. \.-.OI Olav-Ole: ..,..co..-.-’. ~.u-a.o.n-}q¢!lai .DIQ.‘-I\(-Inp..un.u- .oflv..’|do-u{lllu .. 0 U 00.... ...u ...v~l.' llvn.‘nn. . c ..I In . .. .u o . . toaoo‘.u..- .. .u.....clll 00000 II. is . c". . - . ,. . I. c.. n. o I . . c . .n u .. . . o. u .01.. . u .a. . O O‘IcI-PIIOCIoI'D I . I. a on. I ....-... .o ..o..-..' a n o. Icil .. . . .a . . n ‘o. l .. . ..o.v... . IKOC . . .l I. O. . ‘lullll I .....l . .01 ot . . . a. . . . ...- .. c.. ............. . .- ‘D n .9. 0. ... ttttt I. I... . o .. . .o .0. 0 on. on ..- to. o .. o I Q. . I ~ . o . ..... .. . ... -.. ........... .... v .n. . -.o n u I. .0 n v at... ---------- no-I . . o . .- .u.. a .. ...... .. cu. .o .u. . ,... . o . co uuuuu at no. - .0...¢ I . .. . . . . in u ............. .o. ..... a. ..n....- ...Ol.~ . . .n. . . ..... o . o..l . at ..... u... .. . . .c. u... .A-a a u~¢uo upon-:o-.ounn- . ctctlllli‘ .c...... no .. ....... ca r...a¢-.o.~.avneu.no. '- ... . u. s. . ..-. I. ..... a .9 v 0.1-o.oooIc-o.p .. .. ... . s o . In. nnnnnn unstocaa . ...... ..... .. .. ..... h. .-.o.u- ....... .aa 1. .i. . a. o l ................. 00.... . I. ...... o a u. ........... o: cannon-Coo-aa .. n n I ,. ....... u in .a-u,a.nn o-l.no.~ov...ncon.n I. -,.. ... . I u . as - you... ............... ...-II- .- n.. a I ...... I ..... to. at! In Da. v.— I. . s t ...I.. . . w a -. .uonon - . o In.tn o..c-Ic u nu}. - luv. u. q I l 0.01. .an .oI.v..|ao-..nn on: t.n..oo.c. i!.\ u oooooooooooo .-. .o..n..-.u..o.~ ..... I~Oo..-..v 1.000. o..,n..-o. - n 0.4.. d QV..I.O¢.0.0 ; .10 ct.... .OII oooooooo ton. 3!! aria-snanac-c. .vav‘ 3.... In. IIOooO-rcunovoon.n....a.. .uv (Dan.floacclo.oo ...-o r ."ll'o.-I.~I¢;'Ono ,..).7. Diffe rential Pressure Measurem ent Equ' 1pment FIGURE 8 5. EXPERIMENTAL WORK After the initial design of the separation equip- ment (Figure 2), the major pieces of hardware were assem- bled in the power plant and trial samples taken. These trials, although qualitative in nature, suggested design modifications. One of these, a cotton filter after the cyclone separator, is visible in Figure 6. All the modi- fications considered necessary were made and the equipment was then constructed in a permanent form on a two foot by three foot chasses. This chassis was sturdily constructed and.mounted on casters for easy movement about the power plmt o 5.1 Calibration of Equipment in One-Phase The separation equipment was tested for leaks with compressed air and after some seals were replaced the equipment was able to hold a pressure of 10 psi for sever- al hours. This is considerably above the operating pres- sure, 8 inches of water, and insured that none of the sampled gas would leak before being measured. Calibration of the sample tube with the and closed, a pitot tube, was done for two reasons. First, it was nec- cessary to learn something about the aerodynamic character- istics of the tube. Did it perform, as intended, like the standard sharp edged pitot tube? Secondly, if the tube 37 38 did behave as a pitot, it would be interesting to try and measure the gas velocity in this independent manner. For this velocity measurement a calibration would be neces- sary. The calibration was performed at the end of a duct 10 inches in diameter and 15 feet long. A blower connect- ed to a variable speed direct current motor provided air flow in the duct and made it possible to vary the velocity continuously to 300ft/min. An accurate velocity measure- ment was made with a calibrated pitot tube that was per- manently mounted in the duct. Hessure differential measured with a pitot tube can be related to fluid velocity by Bernoulli's equation as follows : v --= c<3_g° AP)% where V = velocity = coefficient that is constant over a relatively large AP range gc = acceleration due to gravity 9 = density of fluid AP = measured pressure differential Rewriting the above equation and including some constant factors for units conversion yields: efi 39 where = velocity, ft/min K = coefficient in Figure 9 AP = pressure differential, inches water 9 == density of fluid, lb/ft3 Inspection of Figure 9 shows that in fact the sam- ple tube does behave like a pitot tube, with K becoming constant at higher velocities. This indicates that the turbulence created by the cylindrical sample tube at its opening has negligible effect on the pressure measure- ments. Limited experiments were conducted to determine the effect of tube orientation on.measured pressure differen- tial. It appears that a change of ten degrees from the perpendicular to the flow direction changes the measured pressure differential by only 2%. This is relatively small change compared to that of a normal pitot tube. As mentioned earlier a rotameter was used to mea- sure the air flow rate in the isokinetic sampling equip- ment after the coal had been.removed‘by a cyclone separa- tor. The flow rate through a rotameter can be described as: 2 ! Wr( (’13—?) P % CDf< Ff where 2 ll mass flow rate Pitot Tube Coefficient (see text) Rotame ter Coefficient (see text) no 911.0 930 - 920 . 910 p 900» A 890 l 1 J l J 1 l l 1 J O .1 .2 .3 .ll .5 .6 .7 .8 .9 1.0 Pitot Tube Pressure Differential (inches water) 5.0 300'- 2.0 *- 1.a: 1 i I l l l 1 l l J 7 8 9 10 Rotemeter Reading Calibration Curves For Flow Measurement Equipnent Figure 9. 11 12 13 1h. 15 16 17 18 ’41 C = coefficient dependent on height of float in rotameter Df diameter float Wr weight float ff = density float f = density fluid The density of the steel float used in this rotameter is 100 lb/ft3 and that of the fluid (air) never exceeds 0.08 1b/ft3. Hence, the above equation can be reduced to: _l. W’ = Kifz also w = q f where w = mass flow rate, 1b/min K = coefficient dependent on height of float in rotameter, Figure 9 = density fluid, lb/ft3 q = volumetric flow rate, ft3/min Calibration of the rotameter in the same variable air speed duct used before yielded a linear relationship between K.and the rotameter reading, Figure 9. The cali- bration.was performed'by operating the isokinetic sample equipment exactly as one would in a two-phase flow situa- tion. Thus, the linear profile of Figure 9 not only indi- cates that the rotameter was functioning as designed but also that it is possible to Operate the sample tube iso- i2 kinetically by aspirating the tube and equipment so that the pressure inside the tube is equal to that outside. 5.2 Flow Profile Measurements All flow measurements were taken on the 12 inch up- per burner mel line of Furnace I Michigan State Univer- sity Power. Plant. The physical layout of the line is as shown in Figure l. The sample ports were located five pipe diameters above the riffle and four pipe diameters below a liSo bend. The ports were positioned at right an- gles to each other and it was therefore possible to obtain East-west (Figure 10) and North-South (Figure 11) pro- files. Experimental measurements of velocity profiles were made by putting the pulverizer in the manual operating position and leaving it at a fixed set point. This per- mitted the author to make point flow measurements in the steady-state stream. Each point measurement takes about six minutes. It was quite easy to maintain the sample tube in isokinetic operation by adjusting the air supply to the aspirator until an inclined manometer indicated zero dif- ferential pressure across the tube. Once the air supply was adjusted only small periodic changes were required to maintain zero pressure differential. Thus, coal flow measurements, obtained from amount of coal collected per unit time, could be made withrelatively good precision and #3 ease. On the other hand, air flow rate was determined by reading the height of the float in the rotameter. The float tended to oscillate and it required skill to obtain precise readings. Because of the difficulty encountered in obtaining air flow rates, pitot tube differential pressure measure- ments were taken. These were obtained by clearing the tube with compressed air and then.monitoring the differen- tial pressure across the tube. Typically, the pressure would rise and.reach a constant value for about one minute before the tube became plugged with coal, resulting in an erratic reading. ‘Unfortunately, the pitot tube profile in the pipe (Figures 10 and 11) did not correspond to the air profile obtained from.the rotameter. It appears that the reason for this is that the coal contributed to the dif- ferential pressure reading, making it an uncertain comp bination of two flow rates. Hence, the pitot tube shows little promise as a tweephase measurement instrument. Figures 10 and 11 show’ the velocity profiles ob— tained in the burner line discussed above. The abscissa represents distance along the 12 inch profile while the ordinate is a flow rate measurement. For coal flow the scale on the left side of each figure is used as Shown by an arrow. Air flow rate is read from the top scale on the right side. At conditions within the pipe the air density is 0.0652 lb/ft3 and an air flow rate of 150 lb/min/ft2 .oa enemah 8.23pm Bass?» snowing Genes: 3:. our— 5 e333.” 3.: as... o." o a. e m a n m H as: :5. a a» «A . _ a a q, _ o Ahead: eenonav on. 1 lion ”cassettes ehseeeam .959 ge 1. .. 4 8 “.33 em. I 4 oo u. .4 one E. I i om” oma I . d i on." “Noggin: . 3.3 53 33E mmm 1 n on... .34 can can {sexes->3 33 3E 78 oHH enemam seduces 53 33> fisomcfison a Auofioddvuo e .d _ao a 2...: E a a a a do... swede a." ea 9 o e e m z m N .n fies. 3» . . a; a _ a a _ «l d .1 o . 0 8e I. 0 e $39333 pupae ens-seam a . Bee 18. rq1 . . \ .78 on 33m . . . O 8. r . . . e o O 00 faxeaxsd ovum me I seam doc .24 omd II C a 4 Wu 4 ‘HQ—i nfli «i I)“ 4 q 8 foifiésd easm soda mmw l a: con .. 8a is corresponds to a velocity of 2300 ft/min. Hence, air ve- locity can be obtained from.the figures. Pitot tube pres- sure differentials taken as described above are also given in Figures 10 and 11 on the bottom.sca1e, right side. The differentials were not converted to velocities by using Figure 9 since the figure is for air alone and the coal particle flow apparently influenced.the pressures. 5.3 Total Boiler Input-Output Measurements To correlate the velocity profiles shown in Figures 10 and 11 with total coal flow rate or steam output of a boiler is a difficult task. Several of the references listed in the bibliography9 indicate that large errors can be expected whenever point measurements of even an accu- rately known profile are integrated to yield total flow rate. However, a trial calculation was performed to ob- tain an approximate total flow rate in the burner line based on the profiles presented in Figures 10 and 11. A graphical integration of these profiles indicate that the average coal flow rate is about 3.0 times the center line flow rate. During a given one-half hour period the center line velocity was measured to be 32.2 lb/min/ft2 and in- formation from.the automatic dump scales indicated a flow rate of 168 lb/min (t 15%) to the pulverizer. This would be 8h 1b/min to each.burner line. The average velocity u? for the 12" diameter pipe is calculated to be 76 lb/min, based on the center line velocity. This exceptionally close agreement seems to verify the profiles presented. To correlate the boiler output in pounds steam per hour with the center line velocity'would.require addition- al.measurements over a longer time period. However, the author believes this is quite possible and that very close correlation can be obtained. 5.h Time Constant and Humidity Experiments In addition to the major research for this thesis some supplmental work was done to provide a starting point for subsequent studies dealing with other phases of the Detroit Edison Company power plant control system.project. Gross observations involving response time were made in an effort to learn about the time constants asso- ciated with the power plant. A 10 percent step increase in.the boiler output at a steam.rate of 220,000 lb/hr caused the steam temperature to immediately drop hoF and it took five minutes for the boiler to again reach steady- state. The power plant is also sensitive to weather con- ditions. A summer rain was noted to drop the outside air temperature 5°F in a few minutes with a corresponding de- crease in steam temperature of 8°F in 15 minutes. These are Just examples of the important role played by the dy- namic response characteristics of a power plant. Simp- kin32 has simulated components and portions of these sys- )48 tems by synthesizing transfer functions from.step response data. His models are more convenient to study than an actual power plant because timescaling accelerates test- ing and normal plant operation is not interrupted. Moisture in the coal entering a power plant can greatly affect its operation. During winter it is a come men practice to mix some of the hot flue gas with the air going to the pulverizers to maintain the fuel line temper- ature. This extra heat introduced is largely needed to vaporize water that enters with the coal. It would be in- teresting to monitor fuel line humidity year around and compare it with apparent heating value of the coal used. Pulverizer outlet temperature was constant at 175°F all during this research and the air-coal mixture at the sample ports measured luSoF. Humidity was determined in in the sample apparatus with wet and dry bulb thermometers after the coal was separated out and the air temperature reduced to approthately lOSOF. A psychrometric chart35 can be used to relate the humidity at 105°? to that in the fuel line provided the air is not at its dew point (indi- cating moisture has condensed out in the sample equip- ment). ‘Relative humidity was found to range from.50 to 75 percent in the sample apparatus, corresponding to 15 and 25 percent at the thoF temperature of the fuel line. 6. DISCUSSION It is tmportant to remember that the velocity pro— files and other measurements presented in this thesis were taken in a particular operating power plant. For this rea- son the actual values reported are probably not universal but indicate trends that are more widely applicable. Also the measurement techniques should be applicable to two- phase flow situations in general. 6.1 Correlation with Other Two-Phase Measurements The literature on two-phase flow discusses relative flow rates of the solid and gas phases on an average bas- is.33 However, from.Figures 10 and 11 it can be seen that the ratio of solid to gas velocity changes continuously across the 12 inch line studied. It is worth noting that the solids flow rate is lower in the center core (about 6 inches in diameter). Comparison of Figures 10 and 11 shows that the coal and air point flow rates are a function of both radial and angular position. Since there are only two sample ports on the fuel line (90° apart) the angular dependence was not accurately determined and is not presented graphically. However, one can see that the flow rate next to the wall in Figure 11 is approximately one-half that of Figure 10. The most probable reason for this difference is the riffle M9 50 located five pipe diameters downstream from the sample ports. Inspection of construction diagrams for the riffle show that it consists of parallel plates aligned in an east-west direction. These plates, which split the flow to the two burner lines, should not distort an east-west profile. This is the case of Figure 10, but a north-south profile (Figure 11) will show distortion if the flow is not evenly distributed by the plates. Uneven distribution by the riffle is characteristically a problem.and Figure 11 seems to confirm the difficulty. 6.2 Flow Stability Measurements of velocity were made with the pulver- izer held at a fixed set point as described in section 5.1. This was done to insure constant coal and air flow in the fuel line so that the point measurements could be made on a steady-state velocity profile. However, because the rif- fle was located only.five pipe diameters from the sample port the author was concerned that the flow’might not be stable. If it tended to oscillate between several differ- ent profiles the individual point measurements might be taken from.different ones. Fortunately, periodic replica- tions of some of the point measurements indicated that this was not the case. The flow was stable. 6.3 Attainment of Objectives This research.has accomplished all the objectives 51 outlined in the introduction. An isokinetic sample tube and associated equipment has been developed. This equip- ment was used to measure velocity profiles in an operating power plant and establish a relationship between air and coal flow rate. 7. CONCLUSIONS The isokinetic two-phase sample tube and associated equipment has proven useful in determining heretofor un- known velocity profiles. These profiles can yield inforh- mation about pulverized coal and air flow distribution fol- lowing riffles and'bends in large burner fuel lines. The sample tube actually measures an average velo- city over a relatively small area. It is used to provide information on gross movements of each of the two-phases and will not determine point measurements directly next to the pipe wall. However, it performs quite well in large diameter lines and is trouble-free in operation. The most significant outcome of this research is discovery of the large mass flow rates of coal towards the wall of the fuel lines. It appears that the core (center 6" diameter section of 12" diameter line) carries only a small portion of the total coal flow while in the outer annulus the flow rates are high. This effect causes a classification of the coal particles, with the smaller ones moving to the center of the .line (appendix A). The air velocity profile at the point of measure- ment was discovered to be linear, rising toward the south- east quadrant of the pipe. The reason for this tilted profile in a turbulent flow situation is a riffle located 52 53 only five pipe diameters downstream from the only avail— able sample point. This difficult situation is typical of the problems encountered in an operating power plant. It appears that this isokinetic sampling equip- ment can be used as a calibration device for a continuous solid-gas flow meter as described in the introduction to this thesis. For calibration of a radiation density device it will be important to know about the non-uniform density in the pipe. 8. RECOMMENDATIONS FOR FUTURE WORK One of the advantages of this flow measurement re- search is that it was conducted at an operating plant. In the plant stochastic factors sometimes led to situations that would not normally have developed in the laboratory. Most of these occurrences suggested interesting follow-up investigations, although not all of them.would be perti- nent to the present project. For brevity only three sug- gested investigations are outlined below. _ p —‘.L-l-"‘J*m. ' "‘~ ' ’3 d L.‘ 8.1 Continuous Flow Measurement It appears possible to continue the development of a two-phase mass flow meter as outlined in section 1 of this thesis. The major piece of new equipment required is a radiation density meter and it should be possible to select one of the commercially available types discussed in Appendix B. Section 5 contains data that will be nec- essary in specifying radioactive source intensity and de- tector type. The apparent collection of larger size particles at the pipe wall is a phenomena that should be studied to insure accurate flow'rate measurements with the density by radiation technique. However, it is anticipated that this effect can be accounted for by proper calibration with the isokinetic sampling tube. Variable water content Sh 55 of the coal in the fuel line may also be a problem since the water will absorb an undetermined amount of radiation. Calibration at different coal moisture levels will indi- cate the magnitude of this effect. 8.2 Coal Heating Value Measurements The determination and monitoring of on-line fuel energy value is a complex problem. Initial observations indicate that for a plant operating on coal from a single source the moisture content of the coal may be the largest factor influencing heating value. A test of this could be made by comparing the apparent heating value of a given shipment of coal both before and after a rain. The rate of change of the coal heating value is also an important consideration in the development of a control system. In a separate study for the Detroit Edison Company, samples of pulverized coal were taken from a fuel line at intervals of three minutes for over an hour. The heating value and moisture content of these samples is now being determined and this data Should provide an indica- tion of the sampling interval necessary in a fuel control system. 8.3 Pulse Phenomena During this research it was noted that fluctuations occurred in differential-pressure measurements across the fuel lines carrying pulverized coal and air to the burners. S6 A search of the literature indicated that Knudsen and Olsen19 had observed similar fluctuations in pressure mea- surements across fluidized beds. Following a method es- tablished by Sutherland they were able to obtain a direct indication of particle size in a fluidized bed by'measur- ing the intensity and frequency of the fluctuations. Their method consists of numerically expressing, in arbitrary units, the sinuosity or length of the differen- tial-pressure trace over a specified time. An average particle size is then obtained from.this measurement by an initial calibration with particles of a known size. Knud- sen and Olsen give a method for continuously recording the length of trace per unit time. It may be possible to obtain similar measurements of average particle size in the pulverized coal-air lines of a power plant. These measurements would at least give a qualitative indication of any changes in particle size. 9. BIBLIOGRAPHY l. Ackerman, N. L., and Nigomthai, 0., "Development of solid-liquid flowmeter, ASCE-Proc. 90 (J. Hydraulics Div.). Paper 3829, 121-129 (March 1961;). 2. Anderson, R. J“; and Russel, T. N., "Designing for twggphase flow, Chemical Engineering (December 6, 19 . 3. Barth,‘w., Nagel, R., and van'Waverin, K., Chemic- Ingenieur-Technick, 22, 9, 599-602 (September 1957). h. Bhattacharya, A., and Roy, A. N., "Flow of solid- liquid suspensions in vertical columns," Ind. Eng. Chem, El, 268 (1955). 5. Chemical Engineers Handbook, ed. John H. Perry, hth Edition, McGraw-Hill (1963). 6. Combustion Engineering, ed. Otto deLorenze, 1st Edi— tion, Combustion Engineering Co., Inc. (1951). 7. Combustion Engineering, ed. Glenn R. Fryling, Revised Edition, Combustion Engineering Co., Inc. (1966). 8. Doig, I. D., and Roper, G. H., "Air Velocity Profiles in the Presence of Concurrently Transported Par- ticles," Ind. Eng. Chem. Fundamentals'é, 2 (May 1967). 9. Flow Measurement Symposium, American Society of Mech- anggal Engineers, Pittsburgh, Pennsylvania (September 19 . 10. Fluidization in Practice-A Symposium, American Insti- tute of Chemical Engineers, Brooklyn, New York (February 1955). 11. Fluid Meters—-Their Theory and Application, ASME Re- search Publication, 5th Edition (1959). 12. Foxboro Company, "Target Flow Meters,” personal cor- respondence, Foxboro Company, Foxboro, Massachusetts. 13. Gasterstadt, H. V. D., Forschungsarbeiten, 265 (l92h). 57 1h. 15. 16. 17. 18. 19. 25. 26. 27. 58 Harris, W. F., "Pneumatic conveying-special problems ofégensity," Mechanical Handling, 521-527 (November 19 ). Hull, D. E., "Total Count Technigue in the Refinery," Ind. Eng. Chem. 59 (February 195 ) Hull, D. E., "Radioactive Studies of Flow Patterns in the Industrial Plant," Int. J. Applied Radiation and Isotopes 1 (1957). "Improved system for pneumatic conveying,” Mechanical Handling 26, 298 (May 1959). Industrial Nucleonics, "AccuRay Density Gauges Series 300,” personal correspondence, Industrial Nucleonics Corporation, Columbus, Ohio. Knudsen, I. E., and Olsen, W. F., "Direct Indication of Particle Size in Fluidized Beds," Chemical Engi- neering (April 10, 1967). Kraus, Milton, "Pneumatic conveying," Chemical 1- nearing, 167 (April 12, 1965) and 1A9 (May 10, 19 S). Kienast, I. F., ”Cemputation of the pneumatic feed process in vertical standpipes," Technische Hochschule-Wissenschaftlicke Zert Dresden, 8, 61-7h (1958-1959)- Leva, Max, Fluidization, McGraw-Hill (1959). Lieberman, A., ”Fine-Particle Technology in the Chemi- cal Process Industries-Part 3," Chemical Engineering (April 2h, 1967). McChesney, I. G., "Air Metering for Combustion Con- trol," Combustion 2;, uh-h7 (August 1951). McConnell, J. A., and Smuck, W. W., A.I.Ch.E. Meeting, Houston, Texas (February 1967). Mehta, N. C., Smith, J. M., and Smith, E. W}, "Pres- sure drop in air-solid flow systems," Ind. Eng. Cheme 42. 986-992 (1957). Molyneux, F., "Pneumatic conveying fundamentals," Fluid Handling 137, 152-15h (June 1961) and 138, 192— 19h (July 1961). 28. 29. 30. 31. 320 33. 3LL~ 35. 36. 37. 38. 39. S9 Nuclear—Chicago, "Technical Specifications Models 58h6 and 5901," personal correspondence, Nuclear-Chicago Corporation, Des Plaines, Illinois. Ohmart Corporation, "Beta Gaging Systems," personal correspondence, Ohmart Corporation, Cincinnati, Ohio. Riley Stoker Corporation, "Riley Pulverizers," per- sonal correspondence, Riley Stoker Corporation, Wor— cester, Massachusetts. Semco Systems, "Rotary Airlock Valves," personal cor— respondence, Systems Engineering and Manufacturing Company, Houston, Texas. Simpkin, L. J., and Wooden, L. F., "Use of Analog Computers in Investigations of Power Plant Automatic Control Systems," Engineering Research Department, Detroit Edison Company, Detroit, Michigan. 800, S. L., and Trezek, G. J., "Concentration and mass flow distributions in a solid-gas suspension," Ind. Eng. Chem. Fundamentals 3, 98-106 (May 196h). Stockel, I. H., "High-speed flow of fluidized solids in changing area," Fluidization-Chemical Engineering Progress Symposium.Series, 58, 38 (1962). Taso, G. I., and Wheelock, T. D., "Drying Theory and Calculations," Chemical Engineering (June 19, 1967). Uematsu, T., Morikawa, I., and Cho, E., Japan Soc. Mech. Eng. Report on Pneumatic Conveying in Vertical Pipe, 8, 31 (1966). Vogt, E. G., and White, R. R., "Friction in the flow of suspensions," Ind. Eng. Chem., 41, 268 (1955). Wen, G., and Simone, H. P., "Flow characteristics in horizontal fluidized solids transport," A.I.Ch.E. Journal, 5, 2 (1959). Zenz, F. A., and 0thmer, D. F., Fluidization and Fluid Particle Systems, Chapman and Hall (1961). APPENDIX A COAL PROPERTIES In any particular solid-gas flow system.the solid particles have a unique combination of physical properties that markedly influence the flow. The values of these properties can be more difficult to determine than.might be expected. It is the purpose of this appendix to out- line the information gathered on the two most important physical properties for the pulverized coal studied. A.1 Coal Fineness The accepted method of determining coal fineness is governed by ASTM Standard D197-30. A given sample is first separated by sieves and then the weight fraction of the total sample that collects on each sieve is determined and reported as in Figure 12. The size of the sieves used is somewhat variable, although the combination of a 200, 100, and 50 mesh sieve is common. From Figure 12 one can see that the coal samples taken from the University Power Plant contain a large por- tion of particles smaller than.hh.micron (-325 mesh). To characterize the samples adequately it would be desirable to classify these smaller particles. Unfortunately, standard sieving methods cannot be used for particles smaller than hh.microns. However, photographs (Figure 1h) 60 61 were taken to give a qualitative understanding of the par- ticle size range involved and with this information it should be possible to select appropriate separation equip— ment. Figure 13 is a graph of the particle size distribu- tion (from Figure 12) across a vertical section of the 12 inch burner fuel line. It should be noted from this fig- ure that the larger size particles tend to collect at the pipe wall while the smaller ones move toward the center. This may be an important effect and is certainly connected with the profiles for coal and air mass flow rate shown in Figures 10 and 11. Standard sampling techniques do not consist of point samples as discussed here but rather are collections of equal weight samples from equal increments of annular areas. The above classification effect coupled with the known velocity distributions would tend to make this standard sampling technique biased toward the larger particles. Perhaps, the easiest method for accurately monitoring characteristic coal particle size in a plant would be that outlined in section 8.3 of this thesis. A.2 Coal Density Bulk coal density was determined for this research by weighing a.sample and then.measuring its volume in a graduated cylinder. This method is commonly used in indus- try and yielded precise results. SAMPLE LOCATION Sieve Size Particle Size (microns) Inches from East Wall Line Mill Output (June Mill Output U'l-P‘wmi—l \ MHWflWMHWNP [—l Poem-«10‘ unwatwawmw fl; 30) (July) TABLE OF COAL PARTICLE SIZE DISTRIBUTION IN FUEL LINE +60 +250 NNNNNN NNNNNN N 62 WEIGHT PERCENT GIVEN SIZE -60 +100 -250 +1u9 in 12 O‘O‘O‘CD firowr Figure 12 -100 +200 -lug +74 28 18 16 1h in 12 12 1h 12 12 1h 22 12 12 -200 +325 at an 32 20 16 20 22 -325 a Total 100 100 100 100 100 100 100 100 100 100 100 100 100 63 en." 056E 83 was 5 38 333.3.” .98 no 388m 330:3 098 a.“ aoflduom dug a? to: .3. 3 e o a. o m e m N .n «use _ d _ . _ d _ _ A _ . o 3mg .33 code 3.3m gen. downed; ance.: «0.33.39 flowed—n l3! 0 nodoupadn echoes 1.3+ on 1:... 4 L 2. 64 Typical Coal Particles Magnified lOOX FIGURE 14 Coal Particles Magnified To Show Shape FIGURE 15 65 Coarse wet coal entering the pulverizers had an ap— parent density of A5 lb/ft3. This density was for the slightly packed coal. On the other hand, pulverized coal samples from the 175°F fuel line taken after the riffle had a bulk density of 38 lb/ft3. These results seemingly contradict the fact that there is more void space in the coarse coal sample. There are perhaps two factors that account for this paradox and they are discussed below. The coal in the fuel line is much drier than that in the bunkers. This is particularly true with the very fine particle sizes since the rate of drying is directly proportional to the three-halves power of the particle 35 diameter. Any loss in moisture will cause a correspond— ing drop in density since the specific gravity of the coal is less than 1. Another factor that may be more important in lower— ing density is swell of the coal particles due to contact with.hot air in the mill and fuel line. Figure 15 is a photograph of magnified coal particles that was taken in an effort to identify any swelling. Although it is diffi- cult to identify swelling, Figure 15 clearly shows that the surface of the coal particles are not smooth and would not pack as well as might be suspected. APPENDIX B ALTERNATE SOLID-GAS FLOW MEASUREMENT SCHEMES The author believes that with the isokinetic sample tube calibration technique described in this thesis, it is possible to adapt commercially available equipment to the problem of continuously measuring coal-air flow in power plants. The instrument systems on which vendor literature ‘was obtained, although widely varied, universally were de- veloped as modifications of flow or density meters origi- nally designed for measurement of incompressible (solid and/or liquid) media. The use of these methods to measure ~solid-gas flow will require special attention to undesir- able characteristics which.might make them difficult to use in the application under consideration. The systems are most easily grouped in two categories: mechanical and nuclear. B.1 Mechanical Methods The Foxboro Company of Foxboro, Massachusetts, mar- kets a Force-Balance Target type flow meter. It consists of a circular square-edged target suspended concentrical- ly, essentially forming an annular orifice, in the pipe carrying the flow. The target is connected via a rod to a "force-balance topworks” outside of the pipe which bal- ances the force on the target and transmits it as either a 66 6? pneumatic or an electrical output. The instrument also in- cludes a silicone oil-filled mechanical damping device. It is available in four sizes that fit up to a h-inch pipe. The instrument is adapted to different flow rate ranges by changing target size. B is defined as the "target-to-line ratio." (It is unclear whether this is the ratio of the diameters or the areas.) Foxboro indicates that the flow rate equals a con— stant (K) times the square root of the farce on the tar— get. A graph is presented for K as a function of the pipe “Reynolds number, B, and pipe diameter for gas or vapor flow. Inspection of these graphs for the h-inch size shows that K varies widely over the whole range of Re for a B of 0.5, while for a B of 0.7, K is constant above Re = 2 x 103. The bulletin "Special Coal-Air Target Flow Trans- mitter"12 says nothing about actual experience in.measur- ing coal flow and indicates that K should probably be de- termined experimentally for best results in each system. It seems this is essential. For the 12-inch pneumatic tubes under consideration very large forces and high pres- sure drops would result if the target-to-line ratios dis- cussed in the bulletins12 were used. However, reduction of target size would produce greater variations in K and lead to poorer measurements because of the flow distribu- tions discussed earlier. These variations of I would also 68 make the output difficult to linearize. Another mechanical technique that might be suggest- ed for in plant coal flow measurement is feeder drum re- volution counting. The feeder is located in the raw coal line between the scales (or down spout from.the bunker) and the pulverizer. As coal enters the feeder it drops in- to semi-circular pockets on the rotating feeder drum.and as the drum rotates each pocket releases its coal into the pulverizer. Riley30 and Semco31 both indicate that their rotary airlock valves or feeders are a convenient means of "accurately and uniformly” feeding coal to a pulverizer although they give no quantitative indication of what ac- curacy is to be expected. A count of the number of revo- lutions made by the feeder drum per unit time plus know- ledge of the pocket capacity would determine coal flow rate. Air flow would be determined by pressure drop in the burner fuel line.' (Air flow rate could also be measured in the air supply duct to the mill, but most of these ducts have‘been noted to leak considerable quantities of air.) Chief disadvantage with this method of coal measure- ment lies in operation.of the mill or pulverizer. nest mills do not operate on a fixed amount of coal and accumu- lation or depletion is constantly taking place so that a material balance cannot be used to calculate flow in the fuel line out of the mill. Lastly, some mechanical devices have been developed , .7 «Ill \1 I 69 that could be useful if a small by pass line (say one inch) were taken off the main fuel line as a representa- tive. Fluidynamics Inc. of Arcadia, Galifornia markets what they call a "new primary flow element" which measures the differential pressure caused by centrifugal accelera- tion as fluids pass through a circular loop inserted in the flow path. Of course, the difficulty would be to get a representative sample of the two-phase flow. B.2, Nuclear Methods Nuclear radiation methods in fluid flow studies can be subdivided into three groupings: flow rate by radio- active tracers, density by transmittance, and density by backscatter. Hullls’16 has written several articles on applica— tions of radioactive tracers to determine flow rate in in- dustrial plants. In the flow rate by dilution technique a tracer gas, such as Radon, is added to the test stream and samples are removed at various points and tested for concentration of the tracer, thus yielding total flow rate. This is probably the most expensive technique as far as radioactive material is concerned and might be dan- gerous in large pipelines where a substantial amount of tracer‘was needed. The total count method is used to de- termine flow rates and amount of backmixing in flow sys- tems. A given amount of radioactive material is added and a detector determines the time required for the total 7O amount of tracer to pass a given point. The peak timing technique is considered by Hull to be the best for turbu- lent flow rate measurement. A quick pulse of radioactive material is added to the flow and the time required for the peak of the pulse to reach a point downstream deter- mines flow rate. All the radioactive tracer techniques have two disadvantages in the present power plant situa- tion: (1) they are not continuous and could be used only for periodic measurements, (2) considerable care would have to be taken to prevent radiation hazards. Two companies make equipment to determine density by transmittance. In both these systems a source is lo- cated on one side of the process pipe while a detector is located on the other. The energy from a radioactive source that is transmitted to a detector is: I :: ‘flft where /a==coefficient of absorption P = density 1" = thickness of material Thus, absorption of radiation is dependent upon the coef- ficient of absorption as well as density. For the present application it would be safe to assume that the coefficient of absorption does not change, while density of the coal- air mixture does. Ohmart Corporation, Cincinnati, Ohio, makes Beta radiation density equipment which.may be used to determine 71 the density and hence the coal concentration in a given section of pipe. The equipment costs $3500 and includes source, special windows for pipe, and low voltage radia- tion cell (Ohmart Cell).29 Ohmart supplied this type of equipment to Georgia Power Company who was interested in using nuclear radia- tion absorption techniques and differential pressure mea- surement across a venturi section to measure coal flow and air flow rates in a pneumatic conveying system, Although the results of Georgia Power Company's work are unknown, the literature search.showed that the flow velocities of coal and air are not equal. Thus, knowing concentration and total bulk flow rate does not determine the individual flow rates. However, the coal and air flow rates may dif- fer by a constant factor (1/2 has been suggested) over a reasonable range of flow rates, in which case concentra- tion and bulk flow rate could be used to determine indivi- dual rates. 18 Industrial Nucleonics also makes a radiation den- sity device. Accu-Ray uses gamma.radiation and does not require special windows in the pipeline, although shield- ing is heavier on the source and other equipment. The price is also $3500 and the basic equipment setup is simi- lar to Ohmart's. Both equipment manufacturers advertise their gauges for use on pipes up to 1h inches. Industrial Nucleonics 72 has a special model for up to 2A inch pipes, although for larger vessels or pipes they recommend measuring across a cord. The reason for this is evident from.the above equa- tion on transmittance; a doubling of the pipe diameter requires a 6.7 fold increase in source level to maintain radiation level constant at the detector. For this reason, it might be better to use backscattering techniques as described below for large process pipe applications. This technique would lead to both a smaller radioactive source and shielding requirement. Sun Oil Company25 is using gamma radiation in an unconventional way. When gamma photons are directed into any substance, a certain percentage are'scattered back to- ward the source at a reduced energy level, in an amount proportional to the density of the substance but independ- ent of the composition. Sun has put this principle into use with an instrument designed and built by Densitronics Inc., Columbus, Ohio. The device consists of a 6% lb. handheld probe containing a shielded gamma source (cesium- 137) and a backscatter detector. An additional hi lb. in- strument pack contains auxiliaries. The company‘s use of the backseatter technique of- fers a number of advantages over the conventional techs niques described above: (1) density can.be determined simply by holding the calibrated device against a vessel wall, (2) vessel or pipe diameter is not a limiting 73 factor, (3) radiation is low, therefore shielding require- ments are also low, and (A) one man can.make the measure- ments. The company has had great success with backseatter- ing for finding liquid levels (i.e., abrupt change in den- sity) in reactors, accumulators, and fractionators with metal wall thicknesses as great as 7/8 in. The device has also operated successfully on equipment with metal walls up to % in. thick plus two inches of insulation. Nuclear-Chicago Corporation28 also makes a back- scatter device, which is similar to the Densitronics hand- held unit but weighs 35 lb. The Nuclear-Chicago density gauge comes in two models: one for measuring asphalt den- sity on a road surface, the other for density and moisture determination of compacted soils. The moisture density gauge uses a radiumrberyllium source that emits both neutrons (for moisture determination) and gamma rays. It might be possible to adapt one of these models to density measurement in a pipe, although the low density of the coal-air’mixture would not require as strong a gamma ray source 0 S m R A R nl. l— V. .h s R E w N U E T A“ mull