Meas. Sci. Technol. 7 (1996) 1687–1706. Printed in the UK REVIEW ARTICLE Mass flow measurement of bulk solids in pneumatic pipelines Yong Yan† School of Science and Technology, University of Teesside, Middlesbrough, Cleveland, TS1 3BA, UK Received 30 May 1996, in final form 1 August 1996, accepted for publication 16 September 1996 Abstract. Many types of techniques for metering the mass flow rate of bulk solids in a pneumatic pipeline have been proposed and developed during the past 20 years. This paper presents a detailed and comprehensive review of the techniques and the current state of knowledge and experience. The techniques are classified under three main categories: direct measurement of solids mass flow rates, measurement of volumetric concentrations of solids and measurement of solids velocity. Future developments and possible trends in this field are also included. Yong Yan has worked extensively on the development of electrodynamic, capacitive and radiometric sensors for mass flow measurement of pneumatically conveyed solids. He started his academic career as a Lecturer at Tsinghua University, People’s Republic of China, where he earned a BEng in Instrumentation and Control Engineering and a MSc in Advanced Instrumentation. He entered the UK as a Research Assistant in 1989 and three years later he received a PhD from the University of Teesside. After a period of postdoctoral research in pulverized fuel flow metering, he was appointed Lecturer in the Division of Measurement and Control at Teesside in 1993 and later became Leader of the Measurement Science and Technology Research Unit. He recently joined the School of Engineering at the University of Greenwich as a Senior Lecturer in Electrical/Electronic Engineering and is currently leading a team of researchers developing new instrumentation and control systems for various industrial applications. 1. Introduction 1.1. Background Pneumatic transportation of bulk solids is becoming increasingly important in many industries. The solid materials that can be transported by pneumatic means range from adipic acid to zircon. Although all pneumatic † Present address: School of Engineering, University of Greenwich, Wellington Street, Woolwich, London SE18 6PF, UK. 0957-0233/96/121687+20$19.50 c 1996 IOP Publishing Ltd injection systems exhibit technical similarities irrespective of solid materials, the pulverized coal used in electrical power generation and steel production serves here as an important example to illustrate the significance of mass flow measurement of bulk solids in pneumatic pipelines. Figure 1 shows a block diagram of a typical pneumatic conveyor used in coal-fired power stations. Coal is supplied from the bunker into a pulverizing mill and then conveyed towards the furnace via a number of separate tuyeres (typically 2–4). A coal-fired power station can have up to ten such conveyors feeding pulverized coal with a total number of tuyeres as many as 20 per furnace. The mass flow rate and velocity of pulverized coal in each tuyere are crucial parameters influencing both the operation of pneumatic conveyors and the combustion efficiency of furnaces. Firstly, the coal velocity needs to be maintained around the minimum safe value to achieve optimum conveying conditions. Excessively high particle velocity will cause high energy consumption, severe pipeline wear and particle degradation (this latter can influence the effective calorific value of coal). In contrast, insufficient velocity will cause particle stratification in the pipeline and even pipeline blockage which can result in an explosion. Moreover, the velocity at which pulverized coal is injected into the furnace is an important factor affecting directly the uniformity of combustion flames and ultimately the combustion efficiency. Secondly, the presence of unburned carbon in the ash is a major source leading to low combustion efficiency of coal-fired boilers. To keep this loss to a minimum, the optimal mixing of coal and air fed into a furnace should be achieved by maintaining the continuous, consistent and accurate mass flow rate of coal in each tuyere. 1687 Yong Yan Thirdly, uneven distribution of pulverized coal between tuyeres can cause a series of combustion problems such as flame blow-out, excessive NOx formation, combustion oscillations and slagging problems. A uniform distribution of coal to each burner can be achieved by measuring and controlling the coal flow rate in each tuyere. The exact requirement for flow rate measurement of bulk solids in pneumatic pipelines may differ more or less from the aforementioned example, depending upon the purpose of the injection system. However, online continuous measurement and subsequent closed-loop control of flows is a common requirement in all areas of industry in order to improve plant control and operational efficiency. whereby horizontal conveying of solid particles near the bottom of the pipe can be much slower than that of those above. 1.2.3. Variable particle size. Particle sizes range from a few micrometres, such as pulverized coal in electrical power generation and wheat flour in food processing industry, to a few centimetres like mineral ore as experienced in mining industry. For a given injection system, the range of particle size may be fixed, but variations in particle size can occur. For instance, the particle size distribution of pulverized coal depends dominantly on the current operational performance of the pulverizing mill (figure 1). 1.2. The nature of solids flow in pneumatic pipelines Particulate fluid in a pneumatic pipeline is essentially a solids–gas two-phase mixture. However, it is the mass flow rate of the solids phase that is of primary interest to operators of a pneumatic conveyor. From a flow measurement point of view, pneumatically conveyed solids can be regarded as single-phase solids flow. Problems encountered in solids flow measurement are not normally associated with gas or liquid flows. Looking at the history of gas or liquid flow measurement, the most successful flow sensors have been restrictive, with the exception of the electromagnetic flow meter which is non-restrictive but measures only electrically conductive liquids. Since restrictive methods are unacceptable under pneumatic conveying conditions due to the highly abrasive nature of fast moving particles, reliable non-restrictive flow measurement techniques have to be sought. Before looking at specific techniques for the measurement, it is worth considering the nature of solids flow medium in pneumatic pipelines which may affect the validity of data from metering instruments. Many variables exist which may affect the response of solids flow instruments in ill-defined ways. Examples of the primary variables are outlined below; all other parameters have been addressed and discussed in an earlier paper [1]. 1.2.1. Inhomogeneous solids distributions. The distribution of solids in a pneumatic pipeline can be highly inhomogeneous, depending upon the pipeline orientation, measurement position, phase loading, conveying air velocity and properties of the solid material including particle size, moisture content, cohesiveness and adhesiveness. Some representative examples of solids distribution are shown in figure 2. A particularly difficult type of inhomogeneous distribution is the ‘roping’ flow regime, in which most of the moving particles are concentrated into a small portion of the cross sectional area of the pipe for various reasons [1]. 1.2.2. Irregular velocity profiles. In association with the inhomogeneous solids distribution, a spread of velocities over the pipe cross section exists; some examples are illustrated in figure 2. This irregularity in velocity profile is often pronounced when phase loadings become greater, 1688 1.2.4. Moisture content. Particulate materials can contain 1–30% moisture, depending upon material source, storage conditions and process requirements. Materials containing a high moisture content can only be conveyed at temperatures which ensure that the water is in the form of superheated steam such as occurs at coal-fired power stations where the feed can be a low-quality fuel derived from coal washing. In general, a solids flow instrument is designed to meter the flow rate of ‘try’ material excluding moisture. This means that solids flow sensors should be insensitive to moisture in the material. 1.2.5. Other parameters. In addition to the aforementioned four variables, other parameters can also affect the instrument performance, including material type and the degree of accretion of fine particles on the inner wall of the measurement section. Some of these parameters may be impossible to control or predict, and certainly show wide variation between different conveyors and types of solid materials. To reduce effects of inhomogeneous solids distribution and irregular velocity profiles on the measurement, the flow sensor should generally be mounted on a vertical pipeline to meter upwards flow. This is practical because pneumatic conveyors are often used to transport bulk solids from a lower level to an upper level over a distance where a vertical pipe section can be made available. However, horizontal installations are sometimes unavoidable due to limited space in plant configuration, or the measurement section has to be on a horizontal pipeline as required by the process control. In principle, a solids flow instrument should ideally provide a true indication of solids flow rate, regardless of the orientation of measurement section, inhomogeneties in solids distribution, irregularities in velocity profile, and variations in particle size, moisture content and other parameters. 1.3. The scope and organization of this review This paper reviews all proposed methodologies with particular reference to their sensing principles, individual characteristics and the current state of development. Signal conditioning and signal processing techniques are also briefly described where appropriate. Mass flow of solids in pipelines Figure 1. A typical pneumatic conveyor in a coal-fired power station. methodologies of metering the flow rate of solids in a pneumatic pipeline can be divided into two main categories, direct and inferential. A direct solids flow meter has a sensing element that responds directly to the mass flow rate of solids through the instrument. An inferential solids flowmeter determines both the instantaneous volumetric concentration of solids and the instantaneous velocity of solids over the pipe cross section, from which the mass flow rate of solids can be deduced according to the following equation: (1) Ms (t) = ρs AVs (t)βs (t) Figure 2. Typical solids distribution and velocity profiles over the pipe cross section. Flow instruments for metering bulk solids in pneumatic pipelines may be characterized as either restrictive or nonrestrictive. Restrictive flow sensors, based on an orifice plate, a Venturi tube, a Coriolis tube, an impact plate or a turbine, suffer wear problems due to the abrasive nature of fast moving particles and diversion and constriction of flow lines [1]. Although substantial research has demonstrated that these instruments are applicable to mass flow measurements of gravity-fed solids in many cases, they are unsuitable for metering of bulk solids in pneumatic pipelines. This review includes only the techniques which are non-restrictive in nature. All restrictive technologies including mechanical and thermodynamic methods will be presented in a later paper. As with single-phase liquid or gas flow measurement, where βs (t) and Vs (t) are the instantaneous and cross sectionally averaged volumetric concentration and solids velocity respectively, ρs the true density of the solids material and A the cross sectional area of the pipe. The measurements of Vs (t) and βs (t) may involve either the same or two different sensing techniques. Knowledge both of the solids velocity and of the volumetric concentration in a pneumatic pipeline is more valuable to the process operators than just the mass flow rate. In fact, an inferential solids flow instrument can provide users with three parameters, the velocity, concentration and mass flow rate of solids. In some cases the measurement of solids velocity is important because it can lead to considerable savings in energy, reduced pipeline wear and breakage of bulk material. Although review work has previously been conducted on various relevant topics in this field [1–3], many new developments and measurement techniques have emerged in recent years. This review, covering both current and most recent work over a wide range of topics, is presented in a systematic way under three main categories: direct measurement of solids mass flow rates, measurement of volumetric concentrations of solids and measurement of solids velocity. 2. Direct measurement of solids mass flow rates 2.1. Thermal methods This technique relies on the measurement of a rise in fluid temperature as a result of a constant heat input. This type of solids flowmeter operates on the basic equation Ms = H Cp 1T (2) 1689 Yong Yan Figure 3. A heat transfer mass flow meter. Figure 4. An electrostatic mass flow meter with a charging chamber. where Ms is the mass flow rate of solids, H the rate of heat input, Cp the specific heat of the fluid at constant pressure and 1T the change in temperature measured upstream and downstream of the heated sensing section. A prototype meter operating on this principle is shown in figure 3 [4]. The sensing section comprises a thin walled (2.1 mm thick) electrically heated tube and the temperatures of two points (Tc1 and Tc2 ) symmetrically positioned upstream and downstream of the meter on the tube surface are measured by iron–constantan thermocouples. The mass flow rate is then inferred from equation (2), where 1T = Tc1 − Tc2 . For 100 µm aluminium oxide powder of mass flow rate 0–1000 kg h−1 and phase density 0–47.2 kg solids per kg air, the measurement error is within about ±10% of full scale. The measured time constant of the system was of the order of 100 s, indicating that the instrument is unsuitable for applications in which fast dynamic responses are essential. It has been concluded that this system is more applicable to dense phase solids flow in relatively smaller pipes. Other researchers have found that one major disadvantage of this type of solids flow meter is its poor repeatability [5]. produces a current which compensates for the current leaking from the charging chamber through the middle insulating gasket. The thermister also compensates for temperature variations in the transducer. This system was first used for continuous tests of a coal-dust mass flow at a coal-fired power station with particles not exceeding 600 µm in diameter. The flow ranged from 0.1 to 5 kg s−1 and it is claimed that the basic relative error did not exceed ±1.5%. The latest evaluation of the same type of flowmeter [7] with two grades of coal and two kinds of granularity shows that, when the ambient temperature is within 20–100 ◦ C and the moisture content 1–11%, the basic relative error is less than ±1% in a 40 mm pipe and ±2% in a 100 mm pipe. In-depth studies on this system have demonstrated that the geometrical dimensions of the chambers (particularly the ratio of the axial length to pipe diameter) and the value of the high-voltage source have direct effects on the system’s overall sensitivity and linearity. One factor, which must be considered when the system is used for metering pulverized coal, is that the value of the voltage source should not be excessively high to avoid spontaneous ignition or even explosion. 2.2. Active charging and detecting methods 2.3. Passive charge detecting methods Two types of solids flow meters based on electrostatic charge detection have been developed. In the first method particles are charged by an external voltage source (‘active mode’), whereas the second method detects the natural charge on the particles (‘passive mode’) which will be described in section 2.3. The active charging and detection flow meter [6] is shown schematically in figure 4. The instrument comprises two chambers, an electrostatic charging chamber and a measuring chamber. The solids material passes through the charging chamber where the particles are charged by a high voltage source, typically 100–500 V. The particles then flow through the measuring chamber and induce electrical charge on the chamber wall which produces an electrical current (I0 ) directly proportional to the mass flow rate of the solids, Ms = cI0 (3) Electric charge transfer can take place when a particle strikes a metallic pipe wall. Passive charge detecting methods are based on the fact that the amount of charge transferred from particles in dilute suspension flow to a metallic pipe is proportional to the mass flow rate of solids. The charge transferred from the particles to an earthed pipe can be measured as an electric current using a suitable electronic device such as an electrometer. It is known, however, that the measured current depends not only on the mass flow rate of solids but also on the initial charge of particles and electrostatic properties of the pipe wall material. To eliminate the effect of initial charge on the measurement, two measuring chambers with different materials have to be utilized. Figure 5 shows a schematic diagram of this kind of twin-chamber system, the details of which have been presented in [8]. The two detecting chambers together with three thin insulators are covered within a long electric shield. The two chambers are made from stainless steel with inner surfaces coated with titanium (work function 2.92 eV) and platinum (work function 5.64 eV) respectively. The currents I0 and where c is a constant obtained by calibrating the instrument with the solids material to be measured. A thermister (Rt ) powered from a controlled and stabilized DC voltage source 1690 Mass flow of solids in pipelines 3.1.1. Capacitive sensors. The simplest capacitor consists of two parallel metal plates separated by a dielectric material. The capacitance of this capacitor is given by C = ε0 εAp /d Figure 5. An electrostatic mass flow meter with twin detecting chambers. I1 generated from the two detecting chambers are measured using an electrometer via a multiplexer under computer control. The mass flow rate of solids is then determined from the equation Ms = aI0 − bI1 (4) where a and b are constants depending upon the electrostatic properties of the particles and detecting chambers. In practice, both a and b are determined experimentally prior to on-line measurement. Experimental tests of this system with fly ash and alumina powder showed that the relative error between the measured flow rate Ms and the reference reading obtained by direct weighing is within ±10%. It should be noted that an important assumption is made in design and testing of the above system, namely that the particles are fully dispersed in the pipe so that each primary particle impacts upon the inside walls of the detecting chambers. 3. Measurement of volumetric concentrations of solids The volumetric concentration of solids (βs ) represents the total quantity of moving solids within a pneumatic pipeline and is defined as the cross sectional area occupied by the moving solids (As ) normalized with respect to the pipe cross sectional area (A), βs = As /A. (5) In this review, sensing techniques for solids concentration measurement are classified into four main groups according to the sensing strategies employed, including electrical, attenuation, resonance and tomograph methods. 3.1. Electrical methods Dielectric and electrostatic properties of particulate materials have been utilized to develop simple and lowcost sensors for solids concentration measurement. The well-known capacitive and electrodynamic sensors are two typical examples. (6) where ε0 is the permittivity of free space (8.85 pF m−1 ), ε is the relative permittivity of the dielectric material, Ap is the area of overlap of the plates and d their separation. The capacitance method of solids concentration measurement is based upon the simple principle that the presence of solids within a capacitive sensing volume will increase the measured capacitance (C) because of the increased relative permittivity (ε) when referenced to gas (near ε0 ). This change in the capacitance can be converted into a suitable electrical signal (current, voltage or frequency) by incorporating the ‘capacitor’ within a suitable electronic circuit such as an AC deflection bridge. Considerable research into the design and applications of capacitance sensors for solids flow metering has been performed in recent years. The sensing electrodes can be in the form of plates, multi-plates, ring-type, quarter-ring, or even pin-type and can be mounted internally or externally to an insulated section of the pipeline. The measured capacitance may be related either directly or indirectly to the solids concentration in the measurement section. The first method, which is the most commonly used capacitance technique and has a wide range of applications, is termed steady-state capacitance measurement and the second is dynamic capacitance measurement. Figure 6(a) shows an example of a steady-state capacitance sensor [9]. The major advantage of a capacitive sensor is its low cost. Problems associated with the steady-state capacitance method include baseline drift, low overall sensitivity, building up of solids in the sensing volume, temperature effects, variations in permittivity of the conveyed medium and moisture content, and flow-regimedependence. The baseline drift and low-sensitivity problems can be minimized in the capacitance measuring circuit [10]. The use of a dynamic measurement technique can obviate the effect of solid deposition in the test section [3]. The variation of the total capacitance with temperature can be reduced by using a temperature compensation technique [11]. To eliminate effectively the dependence of sensor output on the permittivity of the conveyed medium, the sensor should be calibrated off-line using static models made of the same solids material with known moisture content prior to on-line measurement [1]. The flow-regime-dependence of the sensing electrodes may be minimized by the use of multiple electrodes whose outputs are switched electronically to produce a rotating field across the pipe cross section. Figure 6(b) shows a rotating-field capacitance sensor which has been developed commercially [12]. Vertical installation of the sensing head may also reduce the effect of flow-regime-dependence, insofar as a more uniform solids distribution exists within vertical pipelines. An 80 mm bore capacitive sensing head has been tested in a vertical pipeline [13] using 4 mm PVC cubes for solids volumetric concentrations in the range 0.05–0.3%. It has been claimed that the results show a good agreement 1691 Yong Yan Figure 6. (a) A simple capacitance sensor for solids concentration measurement. (b) A capacitance sensor with a rotating field. between calculated and measured values, but further tests with other materials such as coal powders and the detailed performance of the system have not been reported. A capacitive mass flow meter manufactured by Auburn International has been evaluated on a 25 mm test rig circulating pulverized fuel [12]. Results indicate that variations in the particle-size distribution of coal between 75 µm −70% + 30% and with moisture content up to 4.6% have no apparent effect on the flow meter performance. It has been verified that the product of the particle velocity and the solids concentration signal is a linear function of the mass flow rate measured by load cells. Deviation of data points from the linear regression line is around ±5 to ±15% over the velocity range 5–20 m s−1 . Solids flow meters employing capacitive concentration sensors, so called ‘Granucor’ measuring systems [14], have been marketed by Endress and Hauser for a number of years, but they are stated to be applicable only to flow conditions under which the solids:air mass ratio is greater than 5:1 kg per kg (approximately 0.5% concentration of coal particles) with a homogeneous solids distribution over the pipe cross section and pipe diameters no greater than 200 mm. 3.1.2. Electrodynamic sensors. In description of this sensing technology the term electrodynamic is synonymous with other well-known words like triboelectric or electrostatic. Particles in pneumatic pipelines carry a certain amount of net electrostatic charge due to collisions between particles, impacts between particles and pipe wall, and friction between particles and air stream, with charge densities in the range 10−7 –10−3 C kg−1 . The charge on the particles can be detected by a screened and insulated electrode in conjunction with a suitable charge detection circuit. The cross section of the electrostatic sensor based on a ring-type electrode is shown in figure 7. Since the electrodes in electrodynamic sensors are very similar to those in capacitance sensors, the two sensing techniques appear to be the same. In fact, an electrodynamic sensor detects the natural electrostatic charge on the moving particles, whereas a capacitance-based sensor uses the dielectric properties of the solids. The charge detection circuit can be designed such that it measures the magnitude either of the varying or of the alternating components of the charge signal between 1692 Figure 7. Ring-type electrodynamic sensor for solids concentration measurement. a defined bandwidth (the so-called AC method) or the overall magnitude of the signal without concern for the frequency (DC method). Extensive experimental work has shown that the AC method gives more repeatable results than the DC methods. The major advantage of applying an electrodynamic sensor for concentration metering is its high sensitivity. Yan et al [15] conducted an experiment in which a single particle of diameter 0.5 mm dropped through the centre of a 53 mm bore electrodynamic sensing head (figure 7) generated a useful signal. Such an extreme flow condition is equivalent to a volumetric concentration of solids no greater than 0.01%. The major problem in applying this sensing technique (both AC and DC methods) lies in relating the solids concentration to be measured to the magnitude of the charge signal, which depends upon the physical properties of the particles (size, shape, distribution, conductivity, permittivity, chemical composition, moisture content and so on) and conveying conditions (pipe size, pipe wall roughness, line temperature and so on). The concentration and velocity of solids are also known to be factors contributing to the magnitude of the charge signal. It is, therefore, extremely difficult to interpret measurement results except when all the above parameters are well defined and constant. An instrument based this principle may be calibrated with solids materials of known particle properties under steady flow conditions, but substantial errors could be expected especially when the actual particle properties and the flow conditions vary with time unpredictably. Some of the factors influencing the concentration measurement can be minimized by optimizing the design both of the sensor and of the charge detection electronics. For instance, the effect of Mass flow of solids in pipelines an inhomogeneous distribution of solids over the pipe cross section can be reduced to an acceptable level by adequately increasing the axial length of the electrode [15]. Despite the aforementioned drawbacks, this sensing technique offers the most inexpensive and the simplest means of measuring solids flows in pipes. Because electrodynamic sensors respond only to moving solids in the pipe, the measured concentration data enjoy a large degree of immunity from the effects of solids accretion which adversely affect other technologies. 3.2. Attenuation and scattering methods In general, the attenuation of a monochromatic electromagnetic wave or a sound wave being transmitted through a particulate medium is assumed to obey the Lambert–Beer law (7) I = I0 e−µx where I0 and I are the intensities of the incident and transmitted electromagnetic or sound waves respectively, x is the effective thickness of the medium traversed along the wave and µ is a constant (the linear attenuation coefficient). Based on this fundamental principle, several sensing techniques for metering solids concentration have been developed by transmitting one or a number of electromagnetic or sound waves through a pneumatic pipeline, where the solids concentration is inferred from the measured attenuation by the fluid medium. Electromagnetic waves for concentration measurement can be visible light, a laser beam, microwaves, γ -rays or x-rays, whereas a sound wave can be produced using a suitable acoustic or ultrasonic sensor for the same purpose. Closely related to the attenuation methods, a scattering configuration of an electromagnetic wave can also be applied to determine solids concentration by measuring the radiation scattered by particles. 3.2.1. Optical sensors. Light attenuation/scattering methods have been used to determine average solids concentrations in gas–solids flows. Derived from the Lambert–Beer law, the Mie theory forms the theoretical basis for this type of sensor. It states that the intensity of light transmitted through a dilute gas–solids mixture should be exponentially related to the solids concentration in the light beam. The sensor employs either a laser or flash light as a light source and a photomultiplier tube or a photoresistive cell via an optical fibre as a light detector. Efforts have to be made to ensure minimum thermal drift in the detector and to eliminate stray light around the detector. Several systems based on the light attenuation or scattering techniques have been developed in recent years [16–18]. A typical system operating on this principle [16] is shown schematically in figure 8. Measurements were performed in an 80 mm vertical pipeline for a mean particle size range of 247–832 µm and for solids concentrations in the range 0.15–7%. Test results show that this system is suitable for determining average solids concentrations below 7%. One of the major advantages of optical sensors is that variations in chemical composition and moisture content Figure 8. An optical solids concentration monitor. have virtually no effect on the system output, assuming that all particles being measured are opaque. However, the measurement depends significantly on the particle size with smaller particles producing higher attenuation of the light beam for the same concentration of solids. It is also found that the measured results diverge significantly from theoretical predictions (Mie theory); therefore, the instrument has to be calibrated with particles of known sizes. This technique is certainly inapplicable to densephase gas–solids flows. Optical sensors suffer from contamination and misalignment of entry and exit windows which can cause false signals, resulting in erroneous readings. Sophisticated systems fitted with air purges can operate accurately and reliably, although they may require regular maintenance. A developed form of optical sensor which, instead of measuring the intensity of light transmitted across the pipeline, monitors the dynamic components in the received signal, thus overcoming the problems of misalignment and windows coating [19]. 3.2.2. Microwave sensors. Microwave attenuation techniques have been used to measure the solids concentration in pneumatic conveying [20]. Figure 9(a) shows the fundamental principle of this type of instrument, although the actual sensing configuration may vary. Solids particles within the pipeline absorb microwave energy and increase the attenuation between the microwave source and the detector. For a fixed microwave path length, the greater the solids concentration the larger will be the attenuation. It should be noted, in figure 9(a), that a diagonal sensing arrangement is to enhance the microwave attenuation by the dilute flow medium [1]. Field trials of this type of system have been carried out on a 356 mm test rig circulating pulverized coal and the results demonstrate that the attenuation varies significantly with changes in moisture content, particle size and coal grade. Moreover, deposition of coal particles on the microwave windows causes a dramatic increasing in the attenuation, resulting in spurious readings. Other types of microwave instruments for solids concentration measurement have been developed and evaluated in recent years. A real-time solids concentration monitor based on the Doppler shift in frequency of microwave radiation is developed by Hrin and Tuma [21]. 1693 Yong Yan particulate concentration in effluent gases at high pressures and temperatures such as can be found at the output of a coal gasifier feeding a gas turbine. However, the monitor is insensitive to particulates of size greater than 1 µm. Based on the same principle, Endress and Hauser produced a device called ‘Granuflow’ [22] that can measure the concentration of solids in a pipe where the solids are moving at a constant velocity. A low-power, continuous signal emitted from a Gunn diode in the device is reflected from moving particles in the pipe, allowing it to differentiate between moving and non-moving particles in the pipe (figure 9(c)). The device must be mounted in such a way that the microwave energy will reach the moving target material and illuminate the entire width of the fluid. Accuracies of 5–10% have been stated to be achievable with proper process installation and calibration, although the measurements are known to be affected by particle size, chemical composition of solids and solids velocity. Figure 9. Microwave solids concentration monitors: (a) attenuation, (b) Rayleigh scattering and (c) ‘Granuflow’. The system utilizes a microwave cavity and measures the Rayleigh scattering of incident microwave radiation. The radiation scattered by the solids is measured by ‘beating’ it with the non-scattered radiation which provides an indication of the solids concentration. A high-Q multimode microwave cavity is formed in a section of the flue conduit and this produces a high-intensity microwave field from a low-power microwave source (figure 9(b)). Solids moving through the cavity scatter the microwave radiation with a Doppler-shifted frequency spectrum determined by the velocity distribution of the solids and the direction of propagation of the microwave radiation inside the cavity. Measuring only the intensity of reflected Doppler-shifted energy, the monitor is thus immune from the effect of solids deposition in the sensing zone. Preliminary experimental work on this technique was performed on a brass cylindrical pipe of 30 cm diameter with a velocity of air flow in the cavity of 2.5 m s−1 . The system is intended for monitoring 1694 3.2.3. Radiometric sensors. Radiometric flow instruments use ionizing radiation in the form of γ -rays or xrays to interrogate the flow medium. The line attenuation of a radiation beam transmitted through a medium depends predominantly on the total effective mass per unit area of the material traversed along the beam trajectory and is independent from the particle distribution within the beam. There is also no strong moisture-specific effect over most of the available spectral range [23]. In principle, the radiometric attenuation technique offers a promising method of measuring solids concentrations in pneumatic pipelines. Although this type of instrumentation system can be expensive and sometimes administratively inconvenient, it may offer an absolute reference against which solids flow instruments based on low-cost sensors could be calibrated or validated, off-line or during actual plant operation. The most difficult problem in developing such a system is that it must be able to accommodate inhomogeneous solids dispersal over the pipe cross section. This means that a measurement based on a single line-attenuation geometry would be unrepresentative of the total solids concentration. A scanning densitometer can be a practical and economical tool for measurements on steady-state systems, but would be unsuitable for on-line transient measurements. The multi-path configuration, which has been used for void concentration measurement in liquid–gas flow where welldeveloped flow regimes can be identified, is only likely to be satisfactory if a very large number of beam trajectories simultaneously interrogate the whole pipe cross section. A prototype instrument employing a broad single-beam interrogation geometry and a single-element detector with uniform sensitivity profile has been developed by Yan et al [24]. Figure 10 is a schematic diagram of the instrument, in which a low-photon-energy γ -ray point source (Am241) is used. This instrument in conjunction with other devices has been tested in a 53 mm bore horizontal pipeline on a pneumatic conveyor circulating pulverized coal and cement with 2–5% moisture [15]. The results show that the prototype instrument is capable of measuring total solids concentrations (both moving and deposited particles) within the concentration ranges 0.2–1.6% for coal and 0.2–8% for Mass flow of solids in pipelines Figure 10. A radiometric system for solids concentration measurement. cement respectively. A later version of the system based on the same geometry (figure 10) has also been developed by the research team [25], in which a miniature, high-stability, air-cooled x-ray generator produces a soft radiation field whilst a multi-element photodiode array is used as the receptor. The soft x-ray field yields much higher radiation attenuation than does the γ -ray version, making the system applicable to dilute-phase conveying systems. A scattering configuration might also be considered in which the cross sectional solids distribution is mapped by measuring radiation scattered through a range of different angles determined by spectral analysis of the radiation using an energy-sensitive detector [26]. However, such a configuration may be susceptible to measurement error arising from photon counting statistics, especially under dilute-phase conditions, unless a very high-intensity radiation source is used. 3.2.4. Acoustic/ultrasonic sensors. Solids concentrations can be roughly inferred by monitoring the aerodynamic sound generated by the turbulent nature of the solids flow [5]. The sensing element may be an ordinary microphone strapped to the outer surface of a pipeline. It has been concluded that this technique may be well suited for determining flow or no-flow conditions, but not for absolute measurement of solids concentrations because the sound level depends also on the velocity and size of the particles. To achieve reliable measurement, other approaches using acoustic sensors have been proposed. One approach is to measure the attenuation of the incident beam when sound energy is transmitted across the pipe, as shown in figure 11. The main difficulty encountered is that of obtaining an efficient energy coupling into the gas flow from the acoustic transmitter. To achieve this, source frequencies of the order of 100 kHz have to be used. A theoretical study of this approach has been reported [27]. It was concluded that such a system is feasible but problems could be expected if the particles had a wide range of sizes and concentrations. For example with small, light particles the optimum transmission frequency is about 30 kHz, but with large particles the optimum frequency rises to about Figure 11. An acoustic sensor for concentration measurement. 400 kHz. Moreover, several transducers may have to be used in order to interrogate the entire cross section of the pipe. An alternative approach to obtaining the solids concentration is to use an acoustic resonance technique (section 3.3.3). 3.3. Resonance methods Physical resonance can take place within a particulate material under certain circumstances for which stimulation or injection of external energy is often required. Several sensing techniques based on the resonance principle have been investigated using magnetic, microwave and acoustic sensors. 3.3.1. Magnetic resonance sensors. If an electromagnetic field of suitable frequency is applied to a material which possesses a net magnetic moment, the atomic nuclei within this material may absorb energy from the field at their Lamor frequencies and this phenomenon is known as nuclear magnetic resonance (NMR). The electron magnetic resonance (EMR) is closely related to NMR. NMR senses the nuclei of a selected species contained within a material whereas EMR senses the free or unpaired electrons present. 1695 Yong Yan Figure 12. A magnetic resonance system. Figure 13. A microwave mass flow meter. Both methods require that the sample material be exposed to a relatively static magnetic field. Detection in both cases is accomplished by sensing the effects of interactions between an applied electromagnetic field and the magnetic moments of sub-atomic particles of interest. The magnitude of the NMR response is proportional to the number of appropriate nuclei per unit volume (that of EMR is proportional to the number of unpaired electrons). Thus, the NMR and EMR measurements provide the basis for determining the pertinent flow parameters and physical properties of solids, including concentration, velocity, mass flow rate and moisture content. In development of laboratory instruments, magnetic field intensities are generally made as high as feasible to enhance the sensitivity both of EMR and of NMR. However, in the case of a solids flowmeter, magnetic fields of lower intensities should be employed to minimize the size, weight and magnet power requirements. A combination of NMR and EMR techniques with field intensities as low as 714 G has been applied to measuring the hydrogen, carbon and moisture contents of pneumatically conveyed coal in a 10 mm pipeline [28, 29]. The system, as shown in figure 12, can handle dense phase flows with solids concentrations 5–15%, velocities 8–30 m s−1 and moisture contents from 4–8%. It has been reported that 1–2% repeatability of the system has been achieved depending upon the parameter selected for readout. Other researchers have found that the maximum flow rate through the NMR meter is limited by the relaxation time of the nuclei [5]. This limitation will depend on the instrument designed and on the fluid being metered. Moreover, the pipeline material within the sensing head needs to be non-metallic to permit penetration by the electromagnetic field. 3.3.2. Microwave resonance sensors. The other type of microwave instrument for solids concentration metering also uses a microwave cavity, but measures the resonant frequency shift. A section of cylindrical dielectric material covered with conducting metal pipe can form a microwave cavity resonator (figure 13). If this cavity is connected to a microwave system by means of a small aperture, it is found that the cavity abstracts appreciable power from the microwave system at certain distinct frequencies known as resonant frequencies. The shift in a resonant frequency 1696 from the ‘empty’ cavity to when solids are present is proportional to the solids concentration within the cavity. Based on this principle and coupled with the microwave velocity measurement (section 4.3.2), a microwave powder mass flow meter has been developed by Kobyashi and Miyahara [30]. The results of the test wherein the developed flow meter was used as a branch pipe of 19.4 mm in a pneumatic conveying system of fine coal (90% < 75 µm, moisture content 0.4–0.6%) gave a reported accuracy of ±7% for the full scale of 800 kg h−1 . However, the resonant frequency shift technique suffers from the shortcoming that the frequency shift may be positive or negative depending on the dielectric properties of the solids. In addition, an important drawback to the use of a microwave cavity is that it is extremely sensitive to changes in moisture content and temperature. Thus, special precautions must be taken to stabilize the temperature and to correct for variations in moisture content [5]. 3.3.3. Acoustic resonance sensors. The speed of sound propagation through a fluid depends upon the density of the fluid. When solids are added to a gas to form a dispersed two-phase fluid, the sound speed in the mixture is lower than that in the gas phase alone. In appropriate geometries acoustic resonance occurs with the resonant frequencies being directly proportional to the sound speed. A measurement of acoustic resonant frequencies infers a measure of sound speed and hence solids concentration. A sample application of pulverized coal conveyed by air in power plant coal piping was examined by Vetter and Culick [31] with a first-order perturbation–iteration acoustic analysis. The system studied is shown schematically in figure 14. It was concluded that the average solids concentration can be determined by measuring the frequency of a single transverse acoustic resonance. However, analytical results showed that the system output is substantially related to particle size and sensitive only to particles smaller than 100 µm. No test results of this system have been reported. 3.4. Tomography methods Process tomography is a measurement technique representing a further approach to determining the concentration of Mass flow of solids in pipelines Figure 14. An acoustic resonance meter for solids concentration measurement. Figure 15. A capacitance flow imaging system. solids in pneumatic conveyance. Originally motivated by medical computed tomography (CT) scanners, a great deal of effort has been made over the last few years to obtain cross sectional distributions of components in multicomponent mixtures over the pipe cross section in the form of a visual image. Results have been obtained of welldefined flow regimes in gas–liquid mixtures (such as gas– water two-phase flow or oil–gas–water three-phase flow) by tomographic reconstruction from sensor data recorded in different orientations around the pipe. However, attempts to extend this technique to solids–gas flows over a wide range of operating conditions have met with only limited success due to poor resolution and low sensitivity of the sensors. It is relatively easy to acquire and display an image of the flow on a computer screen, but deriving quantitative information from the image is an intractable task. To date, on-line real-time measurement of solids concentration from reconstructed images has not yet been achieved. individual particles suspended in the air, it can detect solids deposition. Whether this type of system is suitable for reliable on-line measurement of solids concentration remains to be determined. 3.4.1. Capacitance sensors. Figure 15 shows a schematic diagram of the essential components required in a tomography system based on capacitance sensors [32]. In this particular system, eight electrodes positioned on a section of non-conductive pipe are connected to a data acquisition system for sensor excitation and successive impedance measurements. Measurements of capacitance are made between any two of the electrodes in all possible combinations using a charge–discharge method. The resultant two-dimensional images reflect the distribution of solids across the pipe section. Initial trials of the system have been carried out on a test plant circulating polypropylene, acetal and seal salt under dilute- and dense-phase conveying conditions [33]. The trials were intended to investigate attribution of particles in a pneumatic conveyor and the dynamic behaviour of conveying processes by observing and estimating the cross sectional area occupied by a settled layer of solids at different locations on a horizontal pipeline. However, no image was obtained when the particles were fully dispersed under the operational conditions because the particles were too small to be resolved. It seems that the air trapped in the settled layer of particles was indeterminable with this technique, resulting in considerable errors in the investigation. The authors concluded that, although the capacitance tomograph technique is unable to detect 3.4.2. Optical sensors. Optical fibre sensors for process tomography have recently been studied [34] by researchers striving to measure the flow of pneumatically conveyed particles. A total of 32 1 mm wide light beams, produced via optical fibres from a dichroic halogen bulb with an integral reflector and a bi-convex lens, interrogate the measurement section of an 80 mm bore pipe both in the horizontal and in the vertical direction (figure 16). The fibre sensors are evenly spaced 5 mm apart from each other. This means that approximately 40% of the cross section (20% in each dimension) is directly interrogated, the remaining 60% not being in a direct path between a source and its receiver, though there may be some cross sensitivity due to the optical aperture of the transmitting fibre and light scattering by the particles. The light transmits continuously and any particle passing though the volume interrogated by a fibre sensor is detected as a variation in the level of illumination of the sensor. Preliminary experiments with a single-fibre sensor using a laboratory-scale gravity-flow rig showed that the resulting voltage in the form of a time-averaged signal is directly proportional to the quantity of particles (600 µm silica) in the beam and that the cross sensitivity between adjacent receivers is insignificant. Substantial further work is required to achieve the concentration measurement of inhomogeneously distributed solids either directly from the Figure 16. Optical fibre sensors for solids flow imaging. 1697 Yong Yan Figure 17. γ -Ray sensors for solids flow imaging. voltage signals or indirectly from the reconstructed flow images. 3.4.3. Radiometric sensors. Based on the same parallelbeam approach, radiometric sensors may also be applied to obtain images of particle flows as illustrated in figure 17 [35]. In this case, the radiometric sensors are advantageous insofar as the line attenuation of a narrow radiation beam depends predominantly on the total effective mass per unit area of material traversed along the beam trajectory and is independent of the solids distribution along the beam line. A low-energy radioactive strip source (such as Am241) in conjunction with a strategically designed collimator generates a set of parallel radiation beams interrogating the entire pipe cross section. A multi-element photodiode array (such as CdZnTe photodiodes) detecting the transmitted beams provides attenuation measurements. The outputs of detector arrays in the two dimensions provide two sets of ‘projected’ data for flow imaging and solids concentration measurement. The collimating system consists of two separate collimators which are located adjacent to the source and detector respectively and aligned precisely. To avoid ‘dead gaps’ between the active radiation beams, a second collimating layer is introduced into the sensing system. The two collimating layers are identical in structure but cross each other along the pipe axis, so that no part of the pipe cross section is left not interrogated by the radiation. The resolution of the sensors depends on the size of each element of the linear detector arrays, which can now be made less than 0.9 mm pitch. Such arrays are available commercially. The sensing system together with a new signal processing unit is currently being developed. 4. Measurement of solids velocity Since there is a spread of velocities within a pneumatic pipeline wherever the flow sensor is mounted, the term solids velocity here stands for the average of all instantaneous streamline velocities over the entire pipe cross section. Although it is now possible to quantify the spread of velocities within a conveyor using techniques such as optical particle image velocimetry (PIV), it is the instantaneous, cross sectionally averaged solids 1698 velocity that should be measured from the process control standpoint. Non-restrictive measurements of solids velocity can be realized using Doppler, cross correlation or spatial filtering methods. These are reviewed in the following sections. It should be noted that the NMR techniques described in section 3.3.1 may also be applied for solids velocity measurement, provided that a suitable signal processing device is available. For example, the velocity of coal particles can be measured by hydrogen NMR (from the phase of the received signal) and it was claimed that excellent results have been achieved with the hydrogen transient NMR under dilute-phase conditions. Solids velocity can also be measured by injecting recognizable tracers, such as radioactive, magnetizable or phosphorescent tracer particles, into the flow stream [2]. The injected tracer methods may be successful in some cases, but they are generally unsuitable for applications to industrial processes for safety reasons and thus excluded in this review. 4.1. Doppler methods When electromagnetic energy from a source (ft ) is transmitted into a two-phase solids–gas flow, some of the energy (fr ) will be reflected by the solids particles to a receiving detector. According to the well-known Dopplershift principle, the difference in frequency between the transmitted and received signals is directly proportional to the solids velocity (Vs ): f r − ft = 2Vs ft cos θ c (8) where c is the velocity of the electromagnetic energy and θ is the viewing angle of the transmitted energy to the flow. Equation (8) indicates that the solids velocity can be determined by measuring the Doppler frequency (fr − ft ). A particle velocimetry system based on the Doppler-shift principle can be constructed using either a laser or microwaves as an energy source. 4.1.1. Laser Doppler sensors. The development and application of laser Doppler velocimetry (LDV) systems have been the major interest of many companies and academic institutions during the last 20 years. The primary reason for this is that the technique is capable of achieving non-intrusive point velocity measurements of good spatial resolution and extremely high accuracy without calibration. The LDV can be applied to measure solids velocity in two typical operational modes, namely, the reference beam mode and differential Doppler mode [36]. In the reference beam mode (figure 18(a)), the laser light scattered from particles passing through the small illuminated region is mixed with a reference beam and then focused through a pinhole onto a photoelectric detector that detects the frequency shift between transmitted and reflected light. The differential Doppler mode (figure 18(b)) utilizes the scattered light from two focused incident beams which converge on the moving particles from different directions. As the particles pass through the crossover region of the Mass flow of solids in pipelines Figure 18. LDV (a) in the reference beam mode and (b) in the differential Doppler mode. beams, they scatter light from each beam resulting in two Doppler-shifted frequencies. The difference between the two frequencies, detected by a photodetector, is related to the velocity of the solids particles. The LDA has the potential to measure a wide range of solids velocities in the range 0.1 mm s−1 to 100 m s−1 on an on-line continuous basis. However, this technique is only applicable to dilute-phase flow conditions with the maximum solids concentration determined by the power of the laser source and the sensitivity of the signal processing equipment used in detecting the Doppler-shifted frequency signal in the presence of considerable noise. The LDV system developed by Birchenough and Mason [36] was operational for solids concentrations somewhat below 0.1%, when alumina powder was the test material. The authors claimed that the system might be suitable for flows of solids concentrations up to 0.4% with special equipment arrangement. Experimental work on a pneumatic conveyor circulating pulverized coal has shown that the maximum measurable solids concentration was no greater than 0.1% [37]. The LDV system is surely unsuitable for densephase conditions because a basic optical path is required to achieve a reasonably good signal-to-noise ratio. Whichever of the two operational modes is used, the LDV is only applicable to local or point velocity measurements, making it well suited for quantifying the spread of velocities within a conveyor for research purposes rather than for routine measurement of the cross sectionally averaged solids velocity. 4.1.2. Microwave Doppler sensors. A microwave Doppler solids velocimeter can be configured either in bistatic mode or in monostatic mode. Figure 19(a) shows the sensing arrangement in the bistatic mode, whereby the transmitter and receiver together with two separate antennae are utilized for transmission and reception of the microwave signals via the ‘transparent’ windows. The sensing volume is defined by the overlapping beams of the transmitting and Figure 19. Microwave Doppler velocimeters (a) in bistatic mode and (b) in monostatic mode. receiving antennae. However, the monostatic mode, shown in figure 19(b), uses a microwave transceiver rather than a separate transmitter and receiver. Isolation between the transmitted and received signals is provided by a ferrite circulator, so only one antenna is needed in this mode. Such a sensing arrangement has the advantages of lower cost and easier installation than the bistatic mode, although the latter gives a more clearly defined sensing volume. The two sensing modes were studied experimentally by Hamid and Stuchly [38] using wheat and rapeseed as test materials. The average Doppler frequency measured was in the region 2–12 Hz, corresponding to solids velocities 0.04– 2 m s−1 . The monostatic configuration has been applied in a commercial microwave solids flowmeter (figure 9(c)). In contrast to LDV sensors, the microwave Doppler sensors have very poor spatial resolution due to the large sensing volume. Irrespective of whether the bistatic or monostatic mode is used, the detected Doppler signal is composed of many different frequencies as a result of a large number of particles travelling at different velocities and viewed at different angles. Although the mean Doppler frequency will bear some relationship to the cross sectionally averaged velocity of solids, it is illogical to assume that this relationship is linear. However, low cost, simplicity and easy installation of microwave Doppler velocimeters make their applications for routine use in hostile environments very attractive. Unlike the LDV, the window material does not need to be optically transparent. 4.2. Cross correlation methods The speed of a car can be determined from the known distance between two traffic lights over the transit time measured by a stopwatch. The same concept is applied here to measure the velocity of solids in a pneumatic pipeline. 1699 Yong Yan Figure 20. A capacitance solids mass flow meter. Figure 21. Cross correlation solids velocimeter using electrodynamic sensors. Two identical sensors are installed an axial distance (L) apart from each other. The transit time (τm ) taken by the particles moving from the upstream sensor to the downstream sensor is measured by cross correlating the two signals using a dedicated signal processor or a correlator. The solids velocity (Vs ) is then calculated from the known sensor spacing L and the transit time τm : Vs = L/τm . (9) The two flow signals are obviously not identical, the downstream signal being a time-delayed but corrupted version of the other, but the correlator should be capable of picking up the similarities between the two signals in a statistical manner. Suitable sensors together with a reliable correlator are essential in applying cross correlation principles to solids velocity measurements. It should be stressed that the spread of velocities within a conveyor can result in an ill-defined correlogram peak. Although some sort of average transit time can be determined from such a peak, a built-in systematic error in the velocity measurement will be inevitable, depending upon the sensing configurations and correlation processing algorithms used. 4.2.1. Capacitive sensors. Cross correlation particle velocimetry systems based on capacitive sensors have been researched in recent years and a substantial literature on this subject is available. However, experiments or field trials of most systems reported so far have been confined to small-scale test rigs. Figure 20 shows a capacitive solids flow meter developed by Amano et al [11]. This system has been tested on a 25 mm vertical tuyere in a pulverized fuel injection system. The achieved measurement error is stated to be within ±5% over the velocity range 5–20 m s−1 . The authors noticed that the measurement accuracy depends upon solids distribution and velocity profiles due to the intrinsically non-uniform spatial sensitivity of the sensors. A flow stabilizer (similar to a disperser in gas– liquid flow) was installed upstream of the flow meter to homogenize the flow. This approach, however, introduces a significant pressure drop, making the measurement of a restrictive nature. A similar capacitive flow meter of 25 mm diameter has also been tested in vertically upwards flows using 450 µm spherical SiO2 sand [39]. The solids velocities measured by cross correlation have 1700 been compared with those measured by high speed cinematography. Considerable divergences between the two velocities were noticed at higher velocities (15– 25 m s−1 ). 4.2.2. Electrodynamic sensors. The electrostatic charge carried on the particles can be detected by suitable sensors, which derive electrical signals by sensing the random changes in induced charge arising from the passage of the flowing particles. ‘Quarter-ring’ electrodes have been used for velocity measurement of gravity-fed sand and plastic granules over the velocity range 4.5–6.5 m s−1 [40]. ‘Ring’-shaped electrodes have an advantage over other forms in that they are able to average the flow velocity circumferentially and have a higher sensitivity than the ‘quarter-ring’ types of the same axial length. The ring-type electrodes along with signal processing elements are shown schematically in figure 21. Experimental evaluation of a 53 mm version of this type of instrument was detailed in a paper by Yan et al [15], including the electrodynamic sensing mechanism, the spatial sensitivity and spatial filtering properties of the sensor together with their relationships to measurement accuracy and the effects of solids velocity profiles. Off-line test results obtained using gravity-fed sand and glass beads showed that the system repeatability is within ±0.5% over a velocity range 2–4 m s−1 for solids concentrations no greater than 0.2%. Results obtained on a calibrated pneumatic conveyor circulating pulverized fuel and cement demonstrate that the system is capable of achieving repeatability better than ±2% and linearity within ±2% within the velocity range 20– 40 m s−1 for solids concentrations between 0.01–0.44%. 4.2.3. Acoustic sensors. Metering of solids–gas flows using acoustic sensors is very difficult because of the high acoustic attenuation in gases and high particle impingement noise. An acoustic cross correlation flow meter for measuring particle velocities in pneumatic conveying has been studied by Sheen and Raptis [41] in terms of the concept that the particle impact noise modulates the applied acoustic fields in the duct wall. Figure 22 is a block diagram illustrating the sensing arrangement and associated signal processing elements. The correlator used was a commercial spectrum analyser. The success of such a system relies on the insertion of an acoustic de-coupler between the two Mass flow of solids in pipelines Figure 24. A radiometric cross correlation solids velocimeter. Figure 22. An acoustic cross correlation flow meter for solids–gas flow. Figure 23. An optical cross correlation solids velocimeter. sensing ducts on which transducers are installed. The decoupler used was a rubber hose connecting the sensing ducts with a gap in between which served to eliminate acoustic cross talk. Preliminary tests with a 50.8 mm bore prototype version were performed on a test facility circulating limestone. The measured particle velocities were scattered by up to 30% of their linear relationship. No further work has been reported on spatial properties of the sensing fields and the measurement accuracy in relation to spread of velocities and inhomogeneous solids distribution. 4.2.4. Optical sensors. Matsumoto et al [42] utilized two optical beams of 3 mm diameter spaced 10 mm apart in conjunction with a pair of photodiodes to measure the velocity of solids in a fully developed region of gaseous suspension in a 20 mm vertical pipeline. The cross correlation function was obtained using a FFT signal analyser. Results obtained with glass and copper particles of 200–3000 µm show good agreement between the measured correlation velocity and the reference velocity obtained by the photographic method. To overcome the problems of irregular solids velocity profile and inhomogeneous solids distribution, multiple optical beams may have to be employed. A total of 16 parallel optical beams have been used to measure particle velocities along eight chord lines as shown in figure 23 [43]. The particle velocity along an individual chord line is determined by cross correlating the two signals derived from a pair of axially spaced phototransisitors along that chord line. The cross sectionally averaged solids velocity is then obtained by combining all chord-line velocities measured. Experiments with the preliminary system were conducted for upwards flows on a vertical transparent channel of inner diameter 80 mm with a sensor spacing 8 mm over the velocity range 2–6 m s−1 under dilute-phase flow conditions (solids concentration 0.2–2%). It was claimed that the measured average solids velocities were within ±10% of calculated values. However, all experiments were performed under carefully maintained flow conditions, for example relatively large and heavy particles were used in order to avoid the impact of the particles with the pipe wall for better signals. It was also somewhat impractical to use an ordinary PC to achieve on-line computation of multichannel correlation functions. Apart from a number of hand-sketched correlation functions, no actual correlograms directly from the ‘correlator’ were given. 4.2.5. Radiometric sensors. A 25 mm prototype cross correlation solids flow meter utilizing radiometric techniques was designed and built by Linn and Sample [44] for measuring gas–solids two-phase flows in coal gasification systems. This system used two single γ -ray beams from 3 GBq Cs-137 sources (600 keV) with beam diameters of approximately half of the pipe diameter (figure 24). Preliminary tests were performed and the results showed that the velocity uncertainty was around ±20% for pulverized fuel flow with velocities 1–5 m s−1 and solids concentrations 8–10%. However, the authors did not discuss how the instrument performance might depend on inhomogeneous solids distribution over the pipe cross section. Two divergent soft radiation fields produced from a miniature x-ray generator together with a pair of multi-element photodiode arrays were used to derive solids velocities by cross correlation of signals from corresponding pairs of elements [25]. The velocity data were compared with results from an electrodynamic sensing 1701 Yong Yan head (figure 21) on gravity-flow streams of aluminium powder within the velocity range 1–9 m s−1 . An ideal version of the system is being implemented using parallel radiation beams and new photodiode arrays in conjunction with multi-tasking correlation signal processing techniques [35]. 4.3. Spatial filtering methods In principle, any flow sensor exhibits some form of spatial filter effect on the original flow signal due to the finite physical size and geometrical shape of the sensing volume. A spatial filter operating on a particular sensing technique can be strategically designed so that the frequency of the filter is directly proportional to the solids velocity. The resulting signal of the filter either approximates to a bandwidth-limited white noise or contains a periodic component buried in a strong background noise, depending upon the mechanism and structure of the sensor. The spread of velocities within a conveyor poses most of the measurement problems when applying spatial filtering methods. Autocorrelation function analysis is often a useful tool in determining the bandwidth of the filtered signal or the frequency of the periodic component, from which the solids velocity is then inferred. Spatial filtering methods of solids velocity measurement using capacitive/electrodynamic, optical and microwave sensors combined with autocorrelation signal processing techniques have been reported in recent years. 4.3.1. Capacitive/electrodynamic sensors. Previous research [45] has demonstrated that the electrode in a capacitance/electrodynamic sensor acts as a low-pass filter to the flow signal (figures 6(a) and 7). Both theoretical analysis and experimental tests have shown that the bandwidth of the filtered signal (B) is related to the axial width of the electrode (W ) and the mean solids velocity (Vs ) via the relationship B = Kb Vs /W (10) where Kb is a dimensionless proportionality relating to the solids distribution over the pipe cross section, velocity profile, particle size and inhomogeneities in the fluid. For a given sensor under steady flow conditions, Kb is a constant and can be determined experimentally. Since the solids flow signal is close to a bandwidth-limited white noise, B can be estimated from the first zero crossing point of the auto-correlation function of the signal. Results obtained on a pneumatic conveyor showed that the measured solids velocity using the spatial filtering technique is within ±8% of the reference velocity determined by an electrodynamic cross correlation velocimeter (figure 21). It was also found that the particle size is a critical factor affecting the frequency of the filter, so the instrument must be calibrated using materials of known sizes prior to application. Because of the dependence of Kb on a number of other variables, this technique is only applicable under steady-flow conditions. 1702 Figure 25. A microwave resonator for solids velocity measurement. 4.3.2. Microwave sensors. The microwave resonator described in section 3.3.2 can also provide measurement of solids velocity by spatial filtering. The sensing head is designed in such a way that the axial length of the resonator (Lm ) is several times the wavelength of the standing wave (λ). This relationship is illustrated in figure 25. When particles move in the axial direction, the resonant frequency shift changes periodically and the changing frequency (f0 ) of the frequency shift is proportional to the solids velocity, namely (11) f0 = Vs /λ. Because the spatially filtered signal for velocity measurement is superimposed on the slowly varying frequency shift for concentration measurement, a frequency-tracking circuit together with a high-pass electronic filter have been devised to extract velocity data from the signal [30]. Again, the first zero crossing point of the auto-correlation function was used to determine the frequency of the periodic component in the signal. Four prototypes (Lm = 9λ as shown in figure 25) were tested on 19.4 mm bore vertical pipelines of a coal injection system over the velocity range 9–15 m s−1 . Regrettably, no detailed velocity data and independent measurement error were reported. 4.3.3. Optical sensors. Optical spatial filters were originally proposed for aircraft speed detection in the 1960s [46]. To apply the same principle for solids velocity measurement, an optical spatial filter comprising a number of identical sensors needed to be designed and constructed. Figure 26 is an example of this type of system [18]. Parallel light beams emitted from a set of evenly spaced infrared LEDs pass through a section of transparent pipe and then reach an array of photodiodes of the same wavelength as the LEDs. The signals derived from the photo-diodes were combined in a differential mode, thus allowing the frequency of the spatial filter (f0 ) to be related to the solids velocity by Vs (12) f0 = 2L0 where L0 is the spacing between an adjacent pair of the LEDs as shown in figure 26. It is obvious that the system performance depends on several factors including the number of beams, beam width and spacing, and the signal processing method. A 40.7 mm bore version of the system (L0 = 10 mm) has been tested on a free-fall solids flow rig for the velocity Mass flow of solids in pipelines This differential sensing approach may be adaptable to new direct solids flow meters based on other sensing techniques in future. However, the use of a direct approach often results in a long axial length of the sensing head, making it unsuitable for application to large-scale pneumatic conveying systems as used in electrical power generation, with pipe sizes ranging typically from 300 to 600 mm in diameter. 5.2. Concentration measurement Figure 26. Optical sensors for solids velocity measurement. range 2–5 m s−1 . The test particles were glass, ceramic and plastic balls or styroform pellets ranging from 2.5 to 10 mm in size. The results obtained by spatial filtering methods showed a good linearity with a maximum relative error ±9%, when the actual solids velocity was acquired using a high-speed video camera. It should be emphasized that a frozen pattern flow was assumed in the original design of such a system; that is, particles in the flow stream travel without distortion between the sensors, resulting in a combined signal in a periodic form. Whether the system could still operate under turbulent powder flow conditions was not known and no details of the signal processing method were reported. 5. The current state and future developments It is evident that mass flow measurement of pneumatically conveyed solids has been the interest of many industrial organizations and academic institutions all over the world. In addition to thermal, electrical and acoustical methods, almost all regions across the electromagnetic spectrum, from γ -rays to microwaves, have been applied to develop suitable sensors for this application. Each sensing principle employed is intended to utilize one of the physical properties of solids. Some are based upon natural behaviours of the solids flow such as thermal and electrical ones, whereas others need stimulation or injection of energy such as resonance and attenuation. Although many sensors and instruments based upon different principles have been proposed, few are currently operating in industry. Apart from the intrinsic complexity of the measurement problem, over-simplified assumptions made by the researchers about the nature of the solids flow medium and plant operating conditions are the major reasons for this underdevelopment. 5.1. Direct measurement The direct approach of solids mass flow measurement is most straightforward and requires simple signal processing elements. A common feature amongst the three direct solids flow meters reviewed in section 2 is that the mass flow signal is acquired by setting a pair of sensing/detection elements in a differential mode along the pipe axis. The inferential approach has been adapted by most researchers and various degrees of progress have been made using a wide range of sensing techniques both in solids concentration and in solids velocity measurements. Dilutephase operation presents the most difficult problem in nonrestrictive measurement of solids concentration. Relatively large measurement errors are expected for most of the proposed systems under dilute-phase conveying conditions. As two exceptional cases amongst all the proposed sensing techniques, the passive charge detection (electrodynamic) and optical attenuation methods are inherently suitable for dilute-phase applications. For all other techniques special measures have to be taken to achieve reasonably high sensitivity in the measurement. An inhomogeneous solids distribution causes further operational errors since, in general, any non-restrictive sensor exhibits some form of non-uniform spatial sensitivity. Effects of the solids distribution on capacitive, electrostatic, microwave and radiometric sensors have been studied [15, 24, 30, 47] and a systematic evaluation of the four types of sensors in terms of sensing field homogeneity has also been performed [48]. Any remaining nonuniformity in the sensing field can affect the measurements both of concentration and of velocity in a very complex way [1]. Despite substantial time and effort having been spent on this topic, approximately 10% non-uniformity error still exists under optimum design conditions using any of the single-sensor-based methods. It has also been found that the achievement of a more uniform sensing field often entails a larger physical size of the sensing head and hence of the pipe wall thickness or longer axial length. It seems that considerable research on this topic is still required to develop more uniform sensing fields. Combination of multiple sensors in conjunction with new signal processing techniques may offer a better solution to this problem and new sensor designs will continue to emerge in the next few years. As an alternative to the requirement for a homogeneous sensing field, imaging-based microsensing approaches may also provide a promising solution. Sensing fields for solids flow measurement can be classified into either ‘hard’ or ‘soft’ according to the directional definability of the field. Soft fields produced by capacitive, electrodynamic, microwave and acoustic sensors have poor geometrical resolutions, giving poor sensing field homogeneity, whereas hard fields realized using parallel radiometric or optical beams are the most suitable for the micro-sensing approach (figures 16 and 17). Variations in particle size influence the operation of most sensors except those based on capacitance and 1703 Yong Yan radiometric principles. As a result, calibrations of concentration measurement with a particular sized powder are not necessarily transportable to particles having a different size distribution. Since variations in particle size (and shape) can be regarded as a kind of inhomogeneous distribution of solids over the pipe cross section, the use of a homogeneous sensing field or a micro-sensing approach may be the most effective way to overcome this problem. Most sensors including all those based on electrical and resonance methods depend upon the chemical properties of solids. For instance, the dielectric constants of moisture and some constituents of coal ash (silicon dioxide and sulphur trioxide) are very large in comparison with that of carbon, so that variations in moisture content and coal grade can induce significant errors in concentration measurement using capacitive sensors. It is estimated that a variation of 1% in the moisture content of coal can introduce 25–30% variation of the concentration measurement [11]. For this reason, solids flow meters operating on electrical, resonance and attenuation principles must be calibrated using the specific type of solids to be measured on plants [1]. Optical sensors are inherently insensitive to chemical properties of solids and therefore superior to other types in this context. Particle accumulation in the sensing zone due to static electrical adherence of particles and other factors presents a serious problem in many applications [49]. Little attention has been paid to this problem in the past. Particle accumulation can cause large errors in concentration measurement. In this sense, the electrodynamic and dynamic capacitance methods have advantages over the other types. In the micro-sensing approach, the stationary particles are not accounted for, because the measured stream line velocity would be zero, making no contribution to the overall mass flow rate to be measured. Optical sensors are certainly unsuitable for this application, because the accumulated solids (or even fine dust) can block the transmission of optical beams through the pipe cross section. It is suggested, in the use of optical methods, that a special measure should be taken to keep the sensing surface clean. This may be realized by purging the air flow or introducing a special wiping mechanism into the sensing head. 5.3. Velocity measurement Non-restrictive measurement of solids velocity is relatively easier than the concentration measurement. Doppler methods can yield either very accurate results at a welldefined point in the flow stream using laser sensors or crude estimation of the solids velocity across the pipeline using microwave sensors. Cross correlation velocity measurement provides a means better than the Doppler approach. Electrical sensors (electrodynamic and capacitive) are advantageous over attenuation sensors (acoustic, optical and radiometric) in terms of cost, maintenance and applicability. Problems are expected in applying capacitance and attenuation sensors for dilute-phase system due to low signal-to-noise ratios. In-depth theoretical studies and field trials have shown that the electrodynamic cross correlation solids 1704 velocimeter is superior to virtually all other types of known cross correlation flow instruments. This is particularly true in dilute-phase applications. A reliable correlation signal processor is absolutely essential in applying the cross correlation principle for solids velocity measurement. Although there are commercial correlators available for various applications, new correlators designed for flow metering will appear on the market within the next few years. These new correlators should be capable of processing multi-channel flow signals in a multi-tasking mode on a near real-time basis. In addition to new developments in sensor designs and signal processing techniques, the exact physical meaning of the measured correlation velocity (equation (9)) in relation to the sensing mechanism, spatial sensitivity, solids distribution and spread of velocities within a conveyer will continue to be the subject of many researchers in this field. Computer modelling, computerized fluid dynamics, process tomography and PIV may provide useful tools in achieving the goal. The spatial filtering method as a relatively new technique will be a major counterpart of the cross correlation approach. In general, sensors operating on spatial filtering methods are more complicated than cross correlation methods in terms of design and construction, but the signal processing devices required are relatively simpler. However, strategically designed spatial filters allow much more information to be extracted from a solids flow, so more accurate results can be achieved when they are combined with powerful signal processing techniques. It is likely that the majority of sensing techniques previously utilized in concentration measurement will be applied in conjunction with the spatial filtering technique to obtain more satisfactory results in the velocity measurement. 6. Conclusions This paper has attempted to bring together a wide selection of non-restrictive measurement techniques applicable to solids flows in pneumatic pipelines and to review our state of knowledge in the wake of more than 20 years of development. The author is aware that there are inventions and publications which have not been included for various reasons. The trend towards relating process demand changes directly to the solids mass flow rate in pneumatic pipelines is increasing in all areas of industry in order to achieve efficient use of energy and raw materials. Many processes now require the accuracy of solids mass flow to be as high as 1% of a set point and this requirement can only be satisfied by reliable on-line continuous measurement and subsequent closed-loop control of solids flows in pneumatic pipelines. Although some progress has previously been made on various topics, considerable research and development work is required to advance the technologies, particularly in sensor design, signal processing, development of reliable test facilities and establishment of measurement standards. The existing nonrestrictive solids flow instruments, especially those based on the inexpensive electrical sensors whether direct or Mass flow of solids in pipelines inferential, will be refined to achieve better performance and wider applicability. With advances in new materials, electronic components and signal processing techniques, new design ideas and new devices will continue to emerge in the near future. It will be extremely interesting to watch further developments and applications in the next decade. Acknowledgments The author wishes to thank all the publishers and authors who willingly gave permission to reproduce or adapt their figures. The author’s close colleague, Dr B Byrne of the University of Teesside, is particularly thanked for helpful suggestions and discussions in preparing the manuscript. References [1] Yan Y 1995 Flow rate measurement of bulk solids in pneumatic pipelines–problems and solutions Bulk Solids Handling 15 447–56 [2] Thorn R, Beck M S and Green R G 1982 Non-intrusive methods of velocity measurement in pneumatic conveying J. Phys. E: Sci. Instrum. 15 1131–9 [3] Beck M S, Green R G and Thorn R 1987 Non-intrusive measurement of solids mass in pneumatic conveying J. Phys. E: Sci. Instrum. 20 835–40 [4] Moriyama T, Fujii S, Abe K and Kobayashi M 1986 Mass flowmeter using heat transfer for dense phase solid gas two-phase flow Proc. Int. Conf. on Fluid Control and Measurement, Tokyo, Japan, September 1985 (Oxford: Pergamon) [5] Slight H A 1991 Meas. Control 24 214 [6] Izakov F Y, Zubtsov P A, Malyshev G N and Ivin L A 1979 Flowmeter for loose materials Meas. Techn. 10 50–1 [7] Zhang X L 1988 The development and test of FG-1 type flowmeter for pulverized materials Proc. First Int. Conf. on Measurement and Control of Granular Materials, Shenyang, Peoples’ Republic China pp 90–4 [8] Masuda H, Matsusaka S and Nagatani S 1994 Measurement of powder flow rate in gas–solids pipe flow based on the static electrification of particles Adv. Powder Technol. 5 241–54 [9] Hammer E A and Green R G 1983 The spatial filtering effect of capacitance transducer electrodes J. Phys. E: Sci. Instrum. 16 438–43 [10] Huang S M, Stott A L, Green R G and Beck M S 1988 Electronic transducers for industrial measurement of low value capacitances J. Phys. E: Sci. Instrum. 21 242–50 [11] Amano H, Sakakibara M, Doi Y, Morimatu Y, Kawahara H and Matsushita S 1985 Two-phase flow meter for pulverized coal/air flow Proc. Int. Conf. on Fluid Control and Measurement, Tokyo, Japan, September 1985 (Oxford: Pergamon) [12] Mathur M P and Klinzing G E 1984 Flow measurement in pneumatic transport of pulverized coal Powder Technol. 40 309–21 [13] Beck M S, Green R G, Plaskowski A B and Stott A L 1990 Capacitance measurement applied to a pneumatic conveyor with very low solids loading Meas. Sci. Technol. 1 561–4 [14] 1994 ‘Granucor’: Dynamic Bulk Flow Measurement, Endress and Hauser, Technical Information, T107/00/e [15] Yan Y, Byrne B, Woodhead S and Coulthard J 1995 Velocity measurement of pneumatically conveyed solids using electrodynamic sensors Meas. Sci. Technol. 6 515–737 [16] Lu Q, Wang S M and Zhang X F 1988 An experimental study of measurement of particle volume concentration by light-scattering technique Proc. First Int. Conf. on Measurement and Control of Granular Materials, Shenyang, People’s Republic China pp 114–17 [17] Payne A, Werner M, Plank D and Crowe C T 1982 Mass flow measurement of gas–solids suspensions using venturimeters with light attenuation Proc. Symp. on Instrumentation and Control of Fossil Energy Processes, ANL-82-62, Houston pp 77–85 [18] Ichikura Y and Watanable K 1994 Measurement of particle flow by optical spatial filtering Proc. 1994 IEEE Instrumentation and Measurement Technology Conf. part 1 of 3, pp 340–3 [19] PCME Ltd 1995 Product data sheets, issue 05/95 [20] Howard A V 1980 Development of techniques for the measurement of concentration and mass flow rate of pneumatically conveyed coal dust, Central Electricity Generating Board, Scientific Services Department, Harrogate [21] Hrin G P and Tuma D T 1977 Doppler microwave cavity monitor for particulate loading IEEE Trans. Instrum. Meas. 26 13–7 [22] 1995 ‘Granuflow’, GMR 130, Mass Flow Indicator, Endress and Hauser, Technical Information, TI 008G/01/e [23] Yan Y, Byrne B and Coulthard J 1993 Radiation attenuation of pulverised fuel in pneumatic conveying systems Trans. Inst. Meas. Control 15 98–103 [24] Yan Y, Byrne B and Coulthard J 1994 Radiometric determination of dilute inhomogeneous solids loading in pneumatic conveying systems Meas. Sci. Technol. 5 110–19 [25] Mennell J, Byrne B and Yan Y 1996 Non-invasive two-phase flow measurement using soft x-ray attenuation Advances in Sensors for Fluid Flow Measurement, IEE Colloquium, London, April 1996 [26] Kondic N N and Lassahn G D 1980 Nonintrusive density distribution measurement in dynamic high-temperature systems ISA Trans. 19 11–19 [27] Raptis A C, Fitzgerald J W, Popper G F, Doolittle R and Sheen S H 1978 A feasibility study of acoustical/ultrasonic flowmeters for solids/gas systems, Argonne National Laboratory Report, ANL-FE-49622-TMO1 [28] King J D, Rollwitz W L and Santos A D L 1982 Magnetic resonance coal flowmeter and analyser Proc. Symp. Instrumentation and Control of Fossil Energy Processes, ANL-82-62, Houston (Argonne National Laboratory) pp 30–40 [29] King J D and Rollwitz W L 1983 Magnetic resonance measurement of flowing coal ISA Trans. 22 69–76 [30] Kobyashi S and Miyahara S 1984 Development of microwave powder flowmeter Instrumentation (Japan) 27 68–73 [31] Vetter A A and Culick F E C 1987 Acoustical resonance measurement of particle loading in gas-solids flow J. Eng. Gas Turbines Power 109 331–5 [32] Huang S M, Plaskowski A B, Xie C K and Beck M S 1989 Tomographic imaging of two-component flow using capacitance sensors J. Phys. E: Sci. Instrum. 22 173–7 [33] McKee S L, Dyakowski T, Williams R A, Bell T A and Allen T 1995 Solids flow imaging and attrition studies in a pneumatic conveyor Powder Technol. 82 105–13 [34] Green R G, Horbury N M, Abdul Rahim R, Dickin F J, Naylor B D and Pridmore T P 1995 Optical fiber sensors for process tomography Meas. Sci. Technol. 6 1699–704 [35] Yan Y, Byrne B and Coulthard J 1995 Design of a true mass flowmeter for non-restrictive measurement of bulk solids Proc. 2nd Int. Conf. on Multiphase Flow vol 2 pp 47–51 1705 Yong Yan [36] Birchenough A and Mason J S 1976 Local particle velocity measurement with a laser anemometer in an upward flowing gas–solids suspension Powder Technol. 14 139–52 [37] Woodhead S, Pittman A N and Ashenden S J 1995 Laser Doppler velocimetry measurements of particle velocity profiles in gas–solids two-phase flows Proc. IEEE Int. Conf. on Integrating Intelligent Instrumentation and Control pp 770–3 [38] Hamid A and Stuchly S S 1975 Microwave Doppler-effect flow monitor IEEE Trans. Indust. Electron. Control Instrum. 22 224–8 [39] Irons G A and Chang J S 1983 Particle fraction and velocity measurements in gas–powder streams by capacitance transducers Int. J. Multiphase Flow 9 289–97 [40] Xie C G, Stott A L, Huang S M, Plaskowski A and Beck M S 1989 Mass flow measurement of solids using electrodynamic and capacitance transducers J. Phys. E: Sci. Instrum. 22 712–19 [41] Sheen S H and Raptis A C 1985 Active ultrasonic cross-correlation flowmeters for mixed-phase pipe flows ISA Trans. 24 53–8 [42] Matsumoto S, Harakawa H, Suzuki M and Ohtani S 1986 Solids particle velocity in vertical gaseous suspension flows Int. J. Multiphase Flow 12 445–58 1706 [43] Dyakowski T and Williams R A 1993 Measurement of particle velocity distribution in a vertical channel Powder Technol. 77 135–42 [44] Linn J M and Sample D G 1982 Radiometric mass flow measurement of solids/gas stream using radiometric technique Proc. Symp. Instrumentation and Control of Fossil Energy Processes, SAND-82-2327, ANL-82-62, Houston (Argonne National Laboratory) pp 12–20 [45] Yan Y 1996 Spatial filtering methods of particle velocity measurement in pneumatic suspension using a single capacitance sensor 5th World Congress for Particle Technology, San Diego, USA, 14–18 July 1996 [46] Kobayashi A 1980 Spatial filter and the application [I] J. SICE 21 409–17 [47] Xie C G, Stott A L, Plaskowski A and Beck M S 1990 Design of capacitance electrodes for concentration measurement of two-phase flow Meas. Sci. Technol. 1 65–78 [48] Yan Y, Byrne B and Coulthard J 1995 Sensing field homogeneity in mass flow rate measurement of pneumatically conveyed solids Flow Meas. Instrum. 6 115–19 [49] Yan Y and Byrne B 1996 Measurement of solids deposition in pneumatic conveying Powder Technol. accepted for publication