Online pulverised-fuel monitoring at Methil power station

To improve combustion efficiency in coal-fired power stations, three critical combustion parameters, relating to the flow of pulverised fuel, need to be known. These are the solids velocity, the solids distribution and the particle size. Three prototype industrial instruments manufactured by ABB Instrumentation, each of which measure these parameters, are operating a t the Scottish Power plant, Methil Power Station. These have been fitted to 14 inch (350 mm) diameter pneurnatic conveyors leading to the burners and are successfully completing an extensive series of trials. The instruments are relatively unaffected by the solids distribution within the pipeline.

byJ. Coulthard, R. Cheng, F! Kane, J. T. Osborne, and R. F? Keech

Introduction To improve the combustion efficiency of pulverised-fuel (PF) fired furnaces in power generation, it is desirable to measure, optimise and then, ideally, control the PF flow velocity and the solids distribution between pneumatic conveyors leading to burners in the furnace.

On leaving the pulverising mill, the solids- air mixture is split into several different pipes each of which feed an individual burner. One mill can feed as many as six burners. lJsually the fuel and air are metered before the mill where the ratio can be accurately set. Ilifferences in routing of the lines injecting PF into the furnaces, and the inevitable phase maldistributions a t the splitting points, result in an uneven feed to the burners. Consequently, the burner’s stoichiometry is disturbed, which leads to increased fuel costs, higher levels of carbon in the ash and excessive specific emissions in the flue gas.

In general, the PF flows in power stations are defined as lean-phase powders with approximately 1.5:1 air:solids ratio by weight resulting in solids volumetric concentration levels of the order of 0.05% a t 100°C. These low concentrations exclude the use of radiological methods of density measurement a t the present time and, in practice, much lower solids concentrations are common.

For control purposes or merely to observe the flow characteristics, it is unnecessary to have an absolute measure of mass flow rate since a relative measure is sufficient. This allows the use of non-restrictive and non- hazardous electrostatic measuring techniques. The advantages of using

electrostatic sensors are high sensitivity, reasonably low cost, hazard free and industrially robust. They are suitable for applications in the hostile environment.

Instrumentation principle

phenomenon in solids pneumatic transport is the electrical charging of conveyed solids.’ The primary sources of electrification are due to the following effects:

A common and well-documented

frictional contact charging 0 charge transfer from one object to

another charging of a conductor by inducement due to the presence of a nearby charge.

Electrical charges are produced on the flowing particles due to friction and to charge transfer caused by particle-wall impact, particle-particle collision and particle-gas friction.

can be considered to originate from two effects:

Electrical charges produced on the sensors

(i) charging due to the particles impacting with the inner surface of the sensor.

(ii) charging due to inducement, when charged particles pass within the vicinity of the sensors requiring no physical contact between the particle and the sensors.

During the passage of charged particles, variations in level of the charge signal on the sensor can be used as an indication of solids concentration.2 It is obvious that this is not an absolute measure of solids concentration

27 POWER ENGINEERING JOURNAL FEBRUARY 1997

1 Flow distribution system

since the charge level is dependent on many properties, but the method has been shown to provide an adequate indication of 'relative solids loading' between pipes carrying the same material.

The charge signal Q can be processed by suitable electronics so that the resulting input signal can be expressed as follows:

dQ d t V, = RI ~

where RI is the amplifier input resistance in parallel with any electrode leakage resistance, which is flow related. Electrostatic sensors detecting the charge signals are used to directly measure two flow parameters: the solids concentration and the solids velocity, the former from a measure of the RMS output signal level and the latter derived from cross-correlation signal processing from two axially-spaced sensors. From the product of these two measurements, a signal related to the mass flow rate of lean-phase conveyed solids can be obtained.

2 Gas-solids flow Plant conditions metering system Fig. 1 shows a diagram of the basic system.

28 POWER ENGINEERING JOURNAL FEBRUARY 1997

Solids from the mill pass through a trifurcating junction into the three pneumatic conveyors in which the instruments were fitted. The pipe runs were comparable so that the solids composition, particle size and moisture content should be reasonably similar in each pipe. The metering system was designed in collaboration with ABB Instrumentation, which manufactured the 14 inch sensors and associated electronics. The metering sections are fitted into the vertical pipeline and have smooth bores with exactly the same diameter as the main flow tubes. To ensure the best possible measurement conditions, the spool pieces were fitted with straight pipe lengths upstream and downstream, in accordance with normal f low metering practice.

Preliminary trials are nearing completion on the 60 M W slurry-burning power station at Methil where the operating conditions are more challenging than those normally found in the more conventional modern coal-fired stations.

The fuel used in this power station is a low-calorific value wet slurry obtained from dumping ponds and from coal washing waste brought in from coal mines. The moisture content of the fuel varies from 18% in the summer to 40% in the winter. The mill inlet temperature is usually around 780°C. This results in the mill outlet temperature being normally cooled t o about 80-1 20°C due t o 'flashing off' of the liquid content of the solids to produce steam. Occasionally, f low temperatures at the measuring point have been observed to reach 220°C under conditions when the solids become obstructed in the hopper, allowing hot gas to pass uncooled through the sensors. All considered, the Methil station proved a hostile environment which had t o be taken into consideration in the sensor design; this is in addition t o the resistance t o abrasion which is also a fundamental requirement.

In spite of the very high moisture content, the electrostatic system still performed extremely well. About 80 tonnes of material pass through each meter each day so that, at the time of writing this article, over 75 000 tonnes of pulverised fuel have passed through the measuring system. The use of the instruments by the plant operators to monitor solids flow conditions is increasing.

In these trials, measurements of solids velocity, relative solids distribution and particle size were taken but, to date, no attempt has been made t o automatically control the flow rates.

The measurement system, shown in Fig. 2, uses a computer to display values of pulverised fuel velocity, solids concentration, relative mass flow rate and particle size in all three pipes in the form of scrolling charts displaying the parameters for the previous 5 minutes. Data logging on the computer hard disc permits examination of the record of all f low parameters over the past few weeks or even months if desired.

Velocity measurement The solids velocity is measured using two

ring-shaped sensors3 detecting variations of induced charge which are cross-correlated using the 'ABB Instrumentation' correlation signal processor. The flow transit time between sensors is measured by determining the centre-of-gravity of the cross-correlation peak. This method is essential if the system is to be used in closed-loop control applications since 'smooth' changes in indicated velocity result from this method of analysis of the cross-correlation function.

The cross-correlation function of the signalsx(t) and y( t ) is given by:

The velocity Vis simply derived from the relation:

(3)

where L is the sensor spacing and T~ is the transit time of the flow between the sensors. The correlation coefficient is given by:

(4)

One typical cross-correlation curve obtained during these on-plant trials is shown in Fig. 3. During normal operating conditions, the corresponding correlation coefficients for all three channels were very similar, ranging from 37% to 43%, and could reach over 60% on occasions. This parameter is likely to be used in future to provide useful data on flow stability.

The combination of the narrow, clearly- 3 Cross-correlation

',":FEt!.' at the defined cross-correlation peaks with high correlation coefficients demonstrates good repeatability in the estimation of flow transit time, and hence measurements of the pulverised fuel velocity should be highly accurate. It was observed that the velocity of pulverised fuel dust in the three pipelines was similar a t any instant in time, varying within the range 14-24m/s depending on plant conditions. The repeatability of the system in measuring solids velocity has been tested using sand flowing vertically under gravity, which showed it to be better than 0.4%.

Confirmation of velocity measurements based on isokinetic sampling shows that the values lie within the limits of the isokinetic measurements which were considered to be accurate to t 2 % a t best. The sensing system alone can be calibrated using a moving surface and this has been shown to have an inherent accuracy of 0.1 YO over a very wide range of velocity.

Fuel concentration test results

using a novel (patent applied for) sensor arrangement and signal processing technique which allows the measurement of solids concentration to be independent of the solids distribution. Under normal conditions, the solids flow rate was fairly constant and the concentration levels in all three pneumatic conveyors showed similar trends, increasing or decreasing together corresponding to variations in the solids feed rate from the mill into the three pneumatic conveyors. The displayed signals can be averaged by using time constants entered from the keyboard and the on-board software can be utilised to provide control signals to the system.

During one test, the damper in one pipe was closed for a few seconds as a check of

The solids concentration was measured

POWER ENGINEERING JOURNAL FEBRUARY 1997 29

4 Variation of PF signal recorded On One conveyor when the damper on the adjacent conveyorwas closed

signals produced by the meter on that and the adjacent pipes. This closure resulted in solids divergence from one pneumatic conveyor to the other two, leading to consequent increases in solids concentration signals which were recorded, and an example is presented in Fig. 4. These results confirmed the ability of the system to track variations in solids flow rates.

results are proceeding based on isokinetic sampling .

Confirmation of velocity and concentration

30

Particle size The particle-size measurement facility has

been tested in the university laboratories and was incorporated onto the plant as an indicator by which the performance of the pulverising mills could be monitored. The technique is based on the principle that, a t a given velocity, a narrow sensor can be used to measure the time occupied by a particle in i ts sensing field and from this the spectrum of the signal will be related to particle size and velocity. Since the velocity is known, the normalised signal bandwidth can be estimated from the width of the cross- correlation function or, better still, from the width of the auto-correlation function. The latter removes loss of resolution due t o spatial filtering:

particle size indication = correlation function with width x solids velocity

This facility is included on the instrument and preliminary indications are that the signals are 'meaningful'. Data on the performance of the system obtained at Teesside University show a linear relationship between particle size and output signal for particles larger than 1 OOpm. The effects of particle-size distribution are presently unknown and

further extensive investigations of this aspect of the instrument will continue into the future but, currently, as an indication of mill performance, the system looks extremely promising.

Conclusions Following the success of these trials it is

proposed to fit similar instruments into a larger coal-burning power station with the aim of closing the loops on velocity and concentration with the aim of controlling and optimising fuel-flow conditions. The operation in such power stations should be less challenging than on the slurry burning facility where the prototypes are being tested.

Acknowledgment The authors wish to acknowledge the

assistance of the UK Department of Trade & Industry in this project.

References 1 CHENG, R.: 'A study of electrostatic pulverised

fuel meters', PhD thesis, University of Teesside, 1996 HIROAKI, M., SHUJI, M., and SHINJI, N.: 'Measurements of powder flow rate in gas- solids flow based on the static electrification of particles', Advanced Powder Techno/. , 1 994,

COULTHARD, J., and CHENG, R.: 'Sensing characteristics of circular sensors', Trans. Inst. M.C., 1997, to be published.

2

pp. 241-254 3

0 IEE: 1997

J Coulthard and R Cheng are with the School of Science &Technology, University of Teesside, MiddlesbroughTSI 3BA, UK P Kaneand J Osborne are with Scottish Power, Methil Power Station, Methil KY8 3RE, UK R P Keech is Development Manager - Flow Products, ABB Instrumentation, Oldsend Lane, Stonehouse GLI 0 3TA, UK

POWER ENGINEERING JOURNAL FEBRUARY 1997