GB2116699A - Fluid flowmeter - Google Patents

Abstract

The velocity of a fluid flowing in a stream through a pipe (16) for example, is determined from calculations of the phase-angle difference ( theta ) for different frequency values (f) between two spaced apart signals modulated in dependence on the flow velocity by, for example, transmission through the fluid to detector means, such as detectors 14 and 15. The slope of the phase- angle curve is related to the transit time of a particular disturbance in the flow between the two signals, hence the flow velocity. The signals may be optical signals provided by a laser source or an incandescent lamp; the latter having an improved signal to noise ratio than the former and requiring less electronics to perform the calculation of flow velocity. The electronics may include a microprocessor and a fast Fourier transformer. The method may be used to determine average flow velocity in a stream as in a fluid flowmeter, or to determine the flow velocity at a particular point in the flow for flow turbulence studies. <IMAGE>

Description

SPECIFICATION Fluid flowmeters This invention relates to methods of measuring fluid flow, fluid flowmeters and in particular, but not exclusively, to fluid flowmeters of the optical type, noise produced on an optical signal, as a result, for example, of naturally occurring turbulence in a fluid flow stream such as through a pipe or the like, being employed to deduce the flow rate.

The accurate measurement of fluid flow involves numerous difficulties. In particular, an instrument employed to effect such measurement must generally be insensitive to the presence of impurities, such as suspended solids, in the fluid and must also respond to discontinuous changes in flow rate. The instrument must also offer minimal restriction to the fluid flow and, where hazardous fluids are involved, must be electrically isolated from the fluids. Flowmeters which satisfy these requirements are generally costly and require complex electronic circuitry to process the measurement and to inhibit spurious error signals.

According to one aspect of the present invention there is provided a method of determining the flow velocity of a fluid stream, including the steps of modulating two signals, separated in the direction of flow, in dependence on the fluid flow, computing the phase-angle difference between the two modulated signals for different frequencies and calculating the flow velocity therefrom.

According to another aspect of the present invention there is provided a fluid flowmeter including means to provide two signals each carrying modulation representing flow velocity, which signals correspond to positions spaced apart in the direction of flow, detector means to detect the modulated signals, and computation means to determine the phase-angle difference for different frequencies between the detected signals and calculate the average flow velocity therefrom.

Embodiments of the present invention will now be described, by way of example, with reference to the accompanying drawings, in which: Figure lisa schematic diagram of a basic time-of-flight flowmeter; Figure2 is a typical cross-correlation graph; Figure 3 is a typical phase-angle curve; Figure 4 is a schematic diagram of a phase-angle flowmeter according to one embodiment of the present invention; Figure 5shows an idealised phase-angle display of the spectrum analyser in Figure 4 for two flow velocities; Figure 6 is the typical spectrum analysis of signals of the embodiment of Figure 4; Figure 7 is the typical spectrum analysis of signals of a modified arrangement of the Figure 4 embodiment; Figure 8a shows the detector light paths associated with Figure 6 and Figure 8b shows the detector light paths associated with Figure 7;; Figures 9-12 show phase-angle curves for water flow calculated from different numbers of average values of phase-angle; Figure 13 is the typical spectrum analysis of an arrangement using an incandescent light source; Figures 14a and 14b show a plan view and side view of an embodiment of phase-angle flowmeter of the present invention employing an incandescent light source; Figures 15 and 16 indicate a detector arrangement for a phase-angle flowmeter which detects eddies and the deflection of light by an eddy, respectively; Figures 17-20 show phase-angle curves for four different water flow rates; Figures 21 to 24 show phase-angle curves for four different airflow rates; Figure 25 shows a phase-angle curve on an enlarged scale and variations with frequency of the slope thereof; Figure 26 shows a graph of the deviation of computed velocity against centre frequency;; Figure 27 shows plots of frequency span against beam separation for different air flows; Figure 28 shows a graph of flow measurement error against beam separation; Figure 29 shows plots of accuracy versus frequency span for different flows; Figure 30 a schematic circuitry arrangement for a phase-angle flowmeter of the present invention, and Figure 31 shows one method of calculating phaseangle (3 from signals C(t) and D(t).

One type of flowmeter which employs optical signals to deduce the flow rate are the so-called optical time-of-flight flowmeters, such as described, for example, in our co-pending application No.

35900/78 (Serial No. ) (G.D. Pitt-S.l.N.

Gregorig-R.J. Williams 25-3-1). The main attraction of this type of flowmeter is that it does not introduce any obstruction in the pipe through which the fluid is flowing.

Figure 1 shows a schematic diagram of a typical basic time-of-flight meter. Fluid whose flow rate is to be measured flows through a pipe 1 in the direction indicated by arrow 2. Light sources 3 and 4 direct respective optical signals 5 and 6 across the pipe 1 to respective light detectors 7 and 8. Noise is induced on the optical signals during passage through the fluid and electronic means in block 9 is employed to deduce the flow rate from the electrical output signals C(t) and D(t) of the detectors. In the abovementioned co-pending application the time delay between the occurrence of a particular transient at the spaced apart optical signal paths was employed to calculate the flow rate. In the past optical sensing was not pursued extensively other than in purely scientific studies and for laser doppler anemometry (LDA).The advent of new solid state sources and detectors of small size, however, means that the stability and spatial resolution requirements for these techniques may now provide the opportunity for the development of more robust and cheaper instrumentation.

The present invention is particularly concerned with phase-angle opto-electronic flowmeters which can detect, for example, the velocity of water in a velocity range 0.1 to 4 metre/second and have a bandwidth of 5 to 60 kHz. Signal bandwidth is one of the important criteria in the measurement of time-offlight of turbulent eddies between two fixed points using statistical techniques.

According to M. Beck (1981) (J. of Physics (E) 14 7-19) the typical bandwidth of turbulent eddies in a fluid flow is 10 kHz. Ideally flow sensing in a time-of-flight meter should respond to the widest possible range of turbulent frequencies. Flow sensors employing ultrasonics, capacitance and conductivity are limited in their bandwidth ( 1 kHz) and this limitation affects both accuracy and the response time of the system. From this point of view opto-electronic sensors are ideally suited since they detect fluid variations at a microscopic level and a wider bandwidth of information can be expected.

There are many areas in process industries involving aggressive and two-phase fluids such as hydraulically and pneumatically conveyed solid materials (pulp, pulverised fuel, sewage etc.) where the measurement of flow is very difficult and flowmeters are required which do not introduce any obstruction in the pipe. Time-of-flight, ultrasonic and magnetic flow meters find application in such difficult situations. In the case of ultrasonic meters, to obtain an accurate transmission path the transducers have to be in contact with the fluid, whereas time-of-flight meters with ultrasonic transducers can be clamped onto the side of the pipe. Magnetic flow meters can only be used with electrically conducting fluids.

Time-of-flight meters may employ one or other of the following techniques, depending on the nature ofthe flow, namely: time delay; cross correlation; phase-angle.

In a time delay meter the time delay between two corresponding pulses can be measured at spaced detectors 7 and 8 as shown in Figure 1. The similarity between the signals is important for the sensor electronic circuitry to recognise the corresponding pulses.

In its simplest form the pulses can be discrete, caused by light scatter from specific particles, and the sensor electronics detects only those which exceed a given threshold amplitude value. Once these pulses can be separated from the noise then the time delays (t) as they pass the two detectors can be measured. Knowing the separation (1) the velocity vcan be calculated (v = lir). However, generally it is often difficult to consistently separate out from the noise, or lower amplitude effects, definite or discrete events in any fluid flow. In situations where the two signals obtained at the two detectors are in the form of discrete pulses, very often they are of varying frequency, amplitude and shape, thus affecting the accuracy of time delay measurement.

In the cross correlation technique, a correlation function is derived from the signals and the position of the peak of that function indicates the time of flight. Atypical cross correlation graph is shown in Figure 2, in which the cross-correlation function RcD (r) iS plotted against time lag.

Although the feasibility of a cross-correlation technique for flow measurement was established a decade ago, the cost of correlation processing has been high and relatively few correlation meters have appeared on the market. With the advent of microprocessors, however, the cost is expected to fall.

Under ideal conditions, flow turbulence consists of a truly white noise source, and for sensors with infinite bandwidth the cross correlation function would effectively diminish to a discrete impulse at a time delay z. In practice, however, the limited bandwidth of the turbulence and the sensor causes the cross correlation function to degrade as shown in Figure 2.

One of the drawbacks of previous crosscorrelation techniques using acoustic, conductivity or capacitative sensors has been the accuracy, which has normally been quoted to lie between +2 to +3%.

Increasing the integration time will improve accuracy but it would adversely affect the total response of the system.

It has been shown that the standard error in flow measurements E(Q) is given by E(Q) = where E( ) is the standard error and 1 is the distance between detectors 7 and 8. This equation shows that increasing 1 can reduce the error in flow measurements. However, turbulence decays as 1 increases, hence E(t) increases in a non-linear manner. The overall result is that the extremes in separation distance can result in poor flowmeter performance.

The flow velocity can also be calculated from the phase-angle between the detected signals C(t) and D(t) and thus is employed in a phase-angle flowmeter. The cross-power density spectrum is a complex quantity Bco(f) = I CD(f) | ei0ff) where I(2icD(f)i is the amplitude of the cross spectrum and (f) is the phase angle between the signals.The phase angle (3 is a function of the frequency f, and ifO is in degrees and fin Hertz then the transit time Tin seconds will be given by z = (1/2::) ((3/f) The plot of 0 against f is termed the phase-angle curve and the slope of the phase-angle curve is therefore proportional to the transit timer. (Fig. 3).

P.G. Bentley and D.G. Dawson (Trans.Soc.ln- str.Tech. 1966) have reported using this technique for flow measurement. They measured the temperature of water with thermocouples and the signals were reported to be 0.01 - 1"C peak to peak at frequencies of a few cycles per second ( 1-4 Hz).

The phase angle was calculated from Plc(f). Their estimate of overall accuracy for this system was +3 to 5%.

The technique has also been applied to study the surface velocity variations in an open channel water flow. Photodetectors collected the reflected light signals from the surface of the water. From the cross-power density function obtained from a spectrum analyser, polar diagrams were plotted and used to obtain the phase-angle frequency relationship. The phase-angle was plotted over a frequency range of 0-50 Hz and it was reported that this frequency relationship curve indicated the distribution of transit time due to surface velocity variations.

The frequency span over which the phase-angle relationship has been studied as discussed above was low ( 4 to 50 Hz). Opto-electronic sensors with small sample volumes increase this to much higher values, and we have found - 1 - 2 kHz to be the frequency span for water flow over which the phase-angle could be computed. This clearly improves the accuracy of measurement.

In an opto-electronic sensor, as indicated in Figure 1, a beam of light is passed through a fluid flowing in a pipe and transversely to the flow direction. If there are particles in the fluid of a different refractive index to the fluid the light can be obscured, diffracted or scattered thereby, depending on the particle size, and thus provide a "noise" signal, of a given bandwidth at the detector, related to the flow since the light beam will be, for example, variably obscured by the suspended particles. Eddies induced in a turbulent fluid have a lower density and lower refractive index than the bulk fluid have a lower density and lower refractive index than the bulk fluid and thus will deflect a light beam directed across a fluid flow pipe and similarly produce a noise signal at a detector.As will be appreciated optoelectronic flowmeter systems are only applicable to relatively clear fluids. Back-scatter effects can, however, be used for less transparent or opaque fluids, in which case the light source(s) and detectors are arranged on one side of the fluid stream.

A schematic arrangement of optical apparatus for the measurement of flow employing the phaseangle measurement technique and according to an embodiment of the present invention is shown in Figure 4. The light beam from a light source such as coherent He-Ne laser 11 is divided into two parallel beams by means of a beam splitter 12 and a mirror 13. In a particular arrangement the beam splitterwas a Melles Griot type 03 BTF 007 with dimensions of 50 x 50 x 1 mm. At an angle of incidence of 45" the transmittance was 58% with a He-Ne laser wavelength of 650 nm. The mirror 13 and detector 15 may be mounted on translation units to facilitate adjustment of the distance between the beams directed across the flow in pipe 16 to respective dedectors 14 and 15.The detectors may, for exam- ple, comprise silicon PIN photodiodes MRD 500 with external amplifiers, or sacrificing faster response but gaining circuit simplicity silicon PIN photodiodes RS 308-067 which have integral amplifiers.

A spectrum analyser 17, which may comprise a Hewlctt-Packard Model 3582A with a frequency span which could be varied from 0-1 Hz to 0-25 kHz, is used for the measurement of the phase angle between the detected "noise" signals C(t) and D(t), and to measure the frequency power spectra. The analyser 17, in its transfer function mode, digitises the signals and stores them as 512 values. From these, 128 values of phase angle are calculated as a function of frequency. The analyser collects the sample data over a period of time (time record length) which depends upon the frequency span set on the analyser. The time record length varied from 250 second (frequency span 0-1 Hz) to 10 milliseconds (0 - 25 kHz). In order to compute the phaseangles the time record length should be greater than the time of flight.

Figure 5 shows an idealised phase-angle display of the analyser 17. The vertical scale has a range of + 200 with the zero degree line in the centre, thus providing a 20 overlap at both ends. Whenever an ordinate reached a value > - 1800 it was switched over to the top of the scale and therefore a folded curve was displayed. Two plots are illustrated corresponding to velocities vl and v2; V1 being greater than V2.

The spectrum analysis of signals indicated the presence of a high level of noise, as shown in Figure 6. Curve A represents the laser noise. Curve B represents the total noise (laser, vibration and reflection). Curve C represents the signal plus total noise. The results shown in Figure 6 were obtained with the laser beam perpendicular to the flow direction. Provision of antivibration mounts increased the signal to noise ratio (~39 - 102%) at all frequencies, but considerable improvement ( 4 to 7 fold was found when, in addition, the optical bench was mounted at an angle of 15 degrees normal to the pipe (Figure 7). This improvement was due to the reduction in the multiple reflected rays falling on the detector area.Figure 8a shows the case with the light beam perpendicular to the flow, and Figure 8b shows the case with the light beam at 150 to the normal. Element 18 comprises a flowmeterwindow in the pipe, element 19 a detector window and the detection area is indicated by 19a.

Figures 9-12 show phase-angle curves obtained for a water flow of 50 gpm in a two inch pipe employing values of phase-angle calculated by averaging 1,2,8 and 32 sets of 128 values of phase-angle from a sample respectively. Whilst Figures 9 and 10 do not display any clear slope, this is clearly evident in Figures 11 and 12.Alarger number of averages, however, effectively means a slower response of the flowmeter. For the equipment providing the readings of Figures 9 to 12 it is concluded that the best working (minimum) choice would be eight averages in order to obtain a clean phase-angle curve suitable for slope determination by regression analysis. The number of averages could be further reduced for other working systems.

Two types of light sources have been used, a 6 volt incandescent lamp and the HeNe laser referred to above. Figures 9 to 12 were obtained when using the 6 volt incandescent lamp. The minimum number of averages when using laser light was found to be 64 rather than the 8 for the incandescent lamp. The reason for this difference was attributed to the noise of the laser source and the interference speckle effect observed when the laser beam passed through the windows. Figure 13 shows the spectrum analysis of signal and noise when the incandescent lamp was used, curve A being the lamp noise; curve B being the total noise (lamp, vibration, reflection) and curve C being the signal plus total noise. The light beam was at 75" to the direction of flow and antivibration mounts were used.With the incandescent lamp the absolute values of the signals are lower than with the laser; the peak to peak value being 10 mVforthe lamp compared with 150 mVfor the laser, however, the signal to noise levels are clearly improved when an incandescent lamp is used instead of a laser noise. Whereas a lens 20 was used to collimate the beam 21a from incandescent lamp 21 (Figure 14a and b), in practice it was not truly collimated, the angle of divergence being 4.6 degrees. The light beam 21a was not divided into two separate beams as in the case of the laser beam, instead the distance between the two effective beams was determined purely by the distance between the detectors 22 and 23. Figure 14a shows a plan view of the whole arrangement and Figure 14b shows the position of the detectors relative to the light beam 21a.In a particular arrangement the cross-sectional area of the detection volume was calculated to be approximately 0.68 x 0.68 mm covering the diameter of a 2 inch water pipe. The minimum distance between the detection points was calculated to be 3.39 mm, being limited only by the dimensions of the detector packages. For commercial operation an incandescent lamp arrangement is clearly preferable to a laser arrangement, being cheap and requiring less electronic circuitry for computation. The technique is also applicable to back-scatter effect systems, as indicated above.

The above description is concerned with liquid (water) flow, gas (air) flow can, however, be similarly measured. For airflow with an He-Ne laser as the light source and a splitter arrangement as shown in Figure 4 may be used. The two beams 24 and 26 are aligned relative to a detector 26 (Figure 15) such that beam deflection by an eddy 27, as indicated in Figure 16, is measured as distinct from obscuration or light scatter. Figure 16 shows the deflection of the light beams caused by the eddy 27 passing therethrough.

The detector 26 is arranged such that it only half covers the overall beam comprising beams 24 and 26. In the vicinity of point a the eddy 27 will deflect the light beam 24 towards the detector 26, resulting in an increase of the output signal. In the vicinity of b any light passes straight through the eddy 27. In the vicinity of pointcthe beam 25 is deflected away from the detector, resulting in a decrease in the output signal. For such deflection considerable beam definition is required, hence the use of the laser. In order to amplify the differential signals obtained, diode arrays can be employed as the detector.

Figures 17 to 20 show phase angle curves for four different water flow rates, 25 gpm (gallons per minute), 50 gpm, 100 gpm and 120 gpm respectively, in a 2 inch diameter pipe. Figures 21 to 25 show phase angle curves for four different air velocities, 8.45 metres per second, 16.9 mitres/second, 25.35 metres per second and 33.8 metres per second, the air flow being in a 4 inch diameter pipe. The variation in slope for different flows can be clearly seen. The slope of the phase-angle curve for any particular flow was observed to be a function of frequency, as shown in Figure 25. When the velocities were calculated it was found that for lower frequencies the velocities were lower than the average, and at higher frequencies they were higher. This tendency was noticed for both water and airflow.

Water flow experiments were conducted over a range of Reynolds numbers 10,000-150,000. In this range the flow can be expected to be turbulent with eddies of different sizes, travelling at different velocities. Each eddy can be considered to be generating a pulse at the detector. At low flows the eddies will be large, resulting in higher zero frequency output and lower output at higher frequencies. As the flow velocity increases, the eddies become smaller in diameter with less output at zero frequency and higher output at higher frequencies. The variation in the slope of the phase-angle curve (Figure 25) is, therefore, an indication of the velocity variations at the point of detection.

For water flow the velocity has been computed over a frequency span of 250 Hz at centre frequencies of 125,375,625 and 875 Hz. Figure 26 shows the deviation of the computed velocity as a function of centre frequency for a flow rate of 60 gpm (2.06 metres/second). The value of the deviation for a centre frequency of 500 Hz was found to be 0.45% (Figure 26). Therefore the error of such a phase-shift flowmeter may be confined to a very low value, such as + 0.5%, by correct choice of centre frequency and the frequency span.

The phase-angle measurement technique provides information about the variations in the flow velocity at a point. This may be used to construct a flowmeter to measure the average velocity or as an instrument to observe velocity fluctuations such as to probe flow turbulence effects. For a flowmeter the choice of frequency span is important. The frequency span is a function of the flow and the beam separation distance. Figure 27 shows the results of air flow tests to determine the relationship between these parameters. The tests were conducted with a 4 inch air flow pipe, a He-Ne laser light source and an RS photodiode with integral amplifier (type 308-067) detector. For all values indicated in Figure 27 the time record length of the sample data was greater than the time of flight.The maximum frequency over which the phase-angle curve was displayed by the spectrum analyser was roughly taken to be the frequency span. At higher flows and shorter beam separation distances the frequency span was found to be larger. This was considered as due to the turbulence pattern remaining stable.

Tests were conducted with water flow of 50 gpm in a 2 inch pipe with an incandescent lamp source, the same detector as for Figure 27 and a fixed bandwidth (frequency span) of 1 kHz, to determine the effect of beam separation distance on accuracy. Figure 28 shows the variation of flow measurement error with beam separation distance. The optimum distance for this flow and frequency span was 7.5 mm. However, it was found that better accuracies could be obtained at all other distances by changing the frequency span. For distances > 7.9 mm distortion in the curve was noticed in the region 750 - 1000 Hz. This may be due to the breakup of the turbulence pattern as it proceeded along the pipe. Conversely, the dip noticed in the error curve for distances < 3.39 mm could be explained as the effect of the turbulence pattern remaining stable due to the short distance between the beams.

Due to physical limitations, the detector size in particular, the effect of shorter beam separations than 3.39 mm have not yet been studied, and the following accuracy of flow measurement discussion thus employs 3.39 mm as an optimum value for water flow, in a 2 inch pipe. It has been shown, however, that the beams can be split optically and focussed on to the detectors to obtain smaller separations, and so the beam separations quoted here should not be taken as minimum on maximum values. Correspondingly for air flow the shortest presently practicable beam separation is 7.5 mm.

The choice of flow range and the beam separation distance precede the choice of frequency span.

Larger frequency span provides for more complete information about velocity fluctuations and therefore better accuracy.

For a beam separation of 3.39 mm the accuracy for various frequency spans was calculated from experimental results obtained for water flow. The low frequency cut-off point was chosen to be zero. The frequency span was varied in steps of 250 Hz. Figure 29 shows the accuracy versus frequency span for different flows. The curves indicate that accuracies of ~0.5% are feasible with the phase-angle measurement technique. In order to adjust the frequency span according to the flow a tracking filter would be required.

In a phase-angle flowmeter for industrial applications a microprocessor may be employed to calculate the phase-angle and subsequently the flow velocity. The potential system response time to calculate the velocity may be approximately 3 seconds. This assumes an optical detector with a time response of a few microseconds and a 40 kHz bandwidth. At a sampling rate of 100 kHz the time for collecting a sample of 512 values is 5.12 milliseconds. The time to calculate one average of 128 values of phase-angle from the sample is 500 milliseconds, and assuming an average of 4 samples we needed to produce an analysable result, the total time to compute velocity is approximately 3 seconds.

Figure 30 shows a schematic arrangement of circuitry for a phase-angle flowmeter. High pass filters 30 and 31 are employed to reduce low frequency noise on the signals C(t) and D(t) caused by mechanical vibrations. A tracking filter 32 is employed to provide a control signal to low pass filters 33 and 34, an optimum frequency span control signal.The analogue signals are digitised by respective analogue-to-digital converters 35 and 36 and applied to a computation unit 37, which may perform the following functions; compute phase angle (3 between C(t) and D(t) over a fixed number of averages, in accordance with the set frequency span, for different frequencies in block 38; by means of at least squares fit method compute average value of (3 in block 39; and compute the average value of flow velocity from the average value of (3 in block 40. The computation unit may incorporate a microprocessor system utilising a Fast Fourier Transformer, such as that designated 52814 of American Microsystems.

One possible method of computing the phase angle (3for various frequencies from the signals C(t) and D(t) is illustrated in Figure 31. The Fourier transfor- mation Z(f) may be calculated using one or more of the above-mentioned Fast Fourier Transformers.

The specific device mentioned is capable of performing 32 complex (or 64 real point) FFT in 1.5 msec. The remainder of the calculation could be performed by further microprocessor techniques. In particular, digital correlation processing electronics, such as that designated TDC 1023J of TRW Inc., can be used as part of the circuitry. It is considered that this approach will improve the economics compared to the standard correlation packages currently available.

In order that a particular flowmeter could operate over a wide range of particle concentrations, for example up to > 5 ppm, the optical system may be arranged to sense both scattered and obscured light, by using a two-element annular detector, comprising a core detection area and a surrounding annular detection area, such as the IPL 31 from Integrated Photomatrix or a diode array as supplied by Reticon.

Fibre optic detector arrays can also be used.

Semiconductor lasers with short lengths of optical fibres accurately aligned (pig tail) during manufacture have been employed for the light sources. Such semiconductor lasers operate at a typical wavelength of 0.850 nm with a peak radiant power of 75 mW. The core diameter of typical optical fibre is very small (~50 um) and therefore is well suited as an effective small sized source for the present purposes. However, the numerical aperture is --0.24, which necessitates the design of suitable optics to collimate the beam. Single mode optical fibre has a core diameter of the order of 2 um and carries a coherent wavefront.As a source such a fibre should have negligible emission depth, and it is possible to produce highly parallel beams, or focus to extremely small sampling volumes, which are particularly relevant for flow turbulence studies, for which it provides an alternative to LDA (Laser Doppler Anemometry) which requires seeding in gases, using the compressibility effect. By focussing the beams the size of the measuring volume can be reduced to, for example < 0.5 mm diameter and 1 mm long, and the velocity fluctuations at a point can be measured. The point of measurement can also be traversed across the diameter of the pipe, thus obtaining a velocity profile of the flow. At each point of measurement the range of turbulent velocity can be more easily obtained than with LDA systems.By automating the traversing of the point of measurement a velocity profile with the range of turbulence intensities at each point of measurement can be obtained.

Because of its accuracy and its ability to detect velocity variations, the phase-angle measurement technique is well suited to the measurement of flow rate as well as to the study of turbulence. For industrial flow determining applications a system employing a white light source (incandescent lamp) is probably preferable on the grounds of lower cost of the lamp and, in view of the improvement in signal to noise ratio, lower cost ofthe computing facility and resultant faster speed of operation, due to the need to calculate fewer averages. An optical system for use with white light can advantageously produce two parallel light curtains 50 and 51 (Figure 32b) in order to measure the average velocity. The optical system may include a light source 32, lens 53 and 54, and a light detector 55 arranged as shown relative to a flow pipe 56.

Claims (21)

1. A method of determining the flow velocity of a fluid stream, including the steps of modulating two signals, separated in the direction of flow, in dependence on the fluid flow, computing the phase-angle difference between the two modulated signals for different frequencies, and calculating the flow velocity therefrom.

2. A method as claimed in claim 1, wherein the signals are optical beams transmitted through the fluid from one side of the stream to detector means arranged on the opposite side of the stream.

3. A method as claimed in claim 1 or 2, wherein the average flow velocity of the stream is calculated from the phase-angle differences.

4. A method as claimed in claim 2, wherein the beams are focussed such that the velocity fluctuations at a respective point in the stream are measured whereby to probe turbulence effects in the fluid.

5. A method as claimed in claim 3 as appendant to claim 2, wherein the detector means comprise two detectors, spaced apart in the direction of flow, from whose respective outputs are computed the phaseangle difference values.

6. A method as claimed in claim 5, wherein the two optical beams comprise two separate beams obtained from a single laser source.

7. A method as claimed in claim 5, wherein the two optical beams comprise two portions of a single beam obtained from an incandescent lamp.

8. A method as claimed in any one of the preceding claims as appendant to claim 2, wherein during passage through the fluid the optical beams are obscured, diffracted or scattered by particles in the fluid.

9. A method as claimed in claim 2 or any one of claims 3 to 8 as appendant to claim 2, wherein the optical beams are transmitted through the flow at an angle of the order of 15 to the normal to the direction of the flow.

10. A method as claimed in claim 1 or claim 3 as appendant to claim 1, wherein the signals are optical beams which are modulated by backscatterfrom the fluid.

11. A method as claimed in claim 1 or claim 2 or any one of claims 3 to 7 as appendant to claim 2, wherein during passage through the fluid the optical beams are deflected by eddies in the flow.

12. A method as claimed in claim 11 as appendant to claim 2, wherein the two optical beams comprise two separate beams obtained from a single laser source, and wherein the detector means comprises a single detector arranged to cover half of the overall beam comprised by the two separate laser beams when the latter are undeflected by eddies.

13. A method as claimed in any one of the preceding claims wherein the said phase angle differences are computed by means including a microprocessor and a fast Fourier transformer.

14. A method as claimed in claim 13, including the step of tracking the frequency of the detector means output and employing it to set the optimum frequency span and data sampling period for the computation means.

15. A fluid flowmeter including means to provide two signals each carrying modulation representing flow velocity, which signals correspond to positions spaced apart in the direction of flow, detector means to detect the modulated signals, and computation means to determine the phase-angle difference for different frequencies between the detected signals and calculate the average flow velocity therefrom.

16. A meter as claimed in claim 15 wherein the signals are optical signals transmitted through the fluid to the detector means from an optical signal source.

17. A meter as claimed in claim 15 or 16, wherein the computation means includes a microprocessor and a fast Fourier transformer.

18. A meter as claimed in claim 16 or claim 17, as appendant to claim 16 wherein the optical signal source comprises a single laser source.

19. A meter as claimed in claim 16 or claim 17 as appendantto claim 16 wherein the optical signal source comprises an incandescent lamp.

20. A method of detecting the flow velocity of a fluid stream substantially as herein described with reference to Figures 3 to 12, and 21 to 24, with or without reference to Figure 30 or Figure 31; Figures 13, 14 and 17 to 20 with or without reference to Figure 25, Figure 26, Figure 27, Figure 28, Figure 29, Figure 30, Figure 31 or Figure 32; or Figures 3 to 12, 15 and 16 and 21 to 24 with or without reference to Figure 30 or Figure 31, of the accompanying drawings.

21. An opto-electronic fluid flowmeter substantially as herein described with reference to Figures 4 and 8; Figures 14 and 8; Figures 4, 15 and 16 or Figure 32, with or without reference to Figure 30 or Figure 31.