WO2009068090A1

WO2009068090A1 - Method for detecting a no-flow situation in a vortex or swirl flow meter

Info

Publication number: WO2009068090A1
Application number: PCT/EP2007/062936
Authority: WO
Inventors: Andrea Andenna; Abhisek Ukil; Frank Kassubek
Original assignee: ABB Research Ltd Switzerland
Current assignee: ABB Research Ltd Switzerland
Priority date: 2007-11-28
Filing date: 2007-11-28
Publication date: 2009-06-04
Anticipated expiration: 2010-05-28
Also published as: CN101925801B; CN101925801A; DE112007003712T5; DE112007003712B4

Abstract

A no-flow condition is detected and indicated by spectral analysis of the output signal of a vortex or swirl flow meter. In a normal, positive flow rate condition large magnitude components are, possibly apart from known interferers like the 50Hz component caused by the electrical power supply, clustered around the frequency of a maximum magnitude component related to the flow rate. In a no-flow condition the components of large magnitude are mostly caused by some motor driving a pump or the like and consist of a basic frequency, e.g., 40Hz, as caused by the motor, and its harmonics, e.g., 80Hz and 120Hz. If a set of large magnitude components comprises a component with a frequency larger than the frequency of a maximum magnitude component multiplied by 1.15 or smaller than the same divided by 1.15 this may already indicate a no-flow condition. Depending on the application, more stringent criteria like the presence of frequency quotients close to small integers like 2, 3 and their inverses or quotients can be employed.

Description

D E S C R I P T I O N

METHOD FOR DETECTING A NO-FLOW SITUATION IN A VORTEX

OR SWIRL FLOW METER

Field of the invention

The invention concerns a method for the evaluation of an output signal of a vortex or swirl flow meter. Flow meters of these types are used for measuring fluid flows in tubes for metering and other purposes.

Prior art

Various types of vortex flow meters and swirl flow meters are known, see, e.g., ABB Operating Instruction D184B097U02 'Vortex Flowmeter FV4000-VT4/VR4 Swirl Flowmeter FS4000- ST4/SR4' . Both types of flow meters exploit the fact that in a fluid flow vortices or swirls form at the downstream end of a bluff body with a frequency that is proportional to the flow rate. The vortices or swirls cause pressure variations in time which are picked up by a downstream sensor, usually a piezo pressure sensor or a paddle connected to a piezo sensor. The sensor produces an electrical output signal which reflects the pressure variations or the motions of the paddle caused by them, respectively. Under normal flow conditions the signal contains a periodic component reflecting the frequency of vortex formation or swirl formation, respectively, whose magnitude is considerably larger than the magnitude of components pertaining to different frequencies.

Under the said conditions the output signal of the flow meter can be evaluated in a straightforward manner by subjecting it in each case to a fast Fourier transform over an evaluation interval and then identifying the position of the maximum in the frequency domain. The flow rate then results from multiplication with a constant.

However, in a no-flow condition where the flow meter is completely shut off from the remaining parts of a tubing, e.g., by closed valves, the method explained above will often lead to a spurious result indicating a positive flow rate. In a no-flow condition it is usually some part of the system like a motor driving a pump which produces the periodic component of largest magnitude. Its frequency which is related to the rotation frequency of the motor will be misread as the component caused by vortex formation or swirl formation, respectively, and a positive flow rate indicated which may in turn lead to inappropriate corrective measures.

Summary of the invention

It is an object of the invention to provide a method for the evaluation of an output signal of a vortex or swirl flow meter where a no-flow condition is identified and indicated with a high degree of accuracy. This object is achieved by the features indicated in the characterizing portion of claim 1.

The method according to the invention allows for immediate detection of a no-flow situation. Spurious positive flow results are identified as such and can be corrected. A detected no-flow condition can be rectified immediately where desired. Brief description of the drawings

In the following, the invention will be explained in more detail with reference to the following figures which show only an embodiment.

Fig. 1 schematically shows a longitudinal section through a vortex flow meter suitable for applying the method according to the invention,

Fig. 2a diagrammatically shows the spectrum of the output signal of a vortex flow meter under normal operating conditions, and

Fig. 2b diagrammatically shows the spectrum of the output signal of a vortex flow meter in a no-flow condition .

Description of the preferred embodiments

Fig. 1 shows a vortex flow meter comprising a tube-shaped housing 1 which will normally be connected to a tubing system consisting of a network of tubes and containers with motor-driven pumps, valves, sensors and other components. Centrally within the housing 1 a wedge-shaped bluff body 2 is fixed and somewhat downstream a sensor which comprises a paddle 3 suspended via a piezo element. The latter produces an output signal which is evaluated in an evaluation unit 4 comprising a digital signal processor. The evaluation unit 4 produces a digital flow rate signal and a no-flow flag signal as explained below.

If a fluid flows through the housing 1 vortices having alternating offsets with respect to the axis of the housing and opposite directions of rotation form at the downstream end of the bluff body 2 and are then carried away by the flow. As they pass by the right and left sides of the paddle 3 causing a lowering of the static pressure in each case, the same is alternately deflected to the left and the right, respectively, i.e., subjected to an oscillatory motion whose frequency is proportional to the frequency of the pairs of vortices passing by the paddle 3. This oscillatory motion is reflected in the output signal of the piezo element.

As is well known, the frequency of the vortices carried by the flow is essentially proportional to the flow rate, following the relation

(1) v = f x d/St

where v is the flow velocity, f the vortex frequency, d the width of the bluff body 2 and St the Strouhal number, i.e., the flow velocity is related to the vortex frequency by the so-called k-factor d/St which is a constant that has previously been determined. Under normal operating conditions it is therefore sufficient to extract the periodic component of the output signal which reflects the oscillation of the paddle 3 caused by the passage of the vortices and to determine its frequency. As the component in question usually has the largest magnitude by far of all periodic components contained in the output signal this is straightforward.

The output signal of the sensor, an analog voltage, is sampled at a rate of, e.g., 1OkHz. Parameters determining the processing of this series of digits in the digital signal processor of the evaluation unit 4 depend on the relevant frequency range which varies strongly with the type of application. The example given below is based on liquid flows in tubes with diameters of a few centimetres. The range of relevant frequencies is in this case an interval bounded by approximately 9Hz and 200Hz. With liquid flow in larger diameter tubes, the relevant frequencies are generally lower whereas in the case of gas flows in tubes of comparable dimensions they tend to be much higher.

The sequence of digits is first subjected to a Fourier transformation - a fast Fourier transformation (FFT) carried out by the digital signal processor - over an evaluation interval which is about lsec. The frequency resolution is of the order of 0.1Hz. The Fourier transform is then modified in that a DC component is subtracted. The result is a spectrum consisting of a number of components each represented by one of about 2'000 frequencies equidistantly spaced between 9Hz and 200Hz and a magnitude corresponding to the frequency and represented by a real number.

In the example the magnitude of the component corresponds in each case to the absolute value of the amplitude but it may also be represented by some other number reflecting the value of the amplitude, e.g., its square. Many other details and choices of parameters depend on the type of application and its pertinent range of relevant frequencies.

Fig. 2a shows the spectrum of the output signal, i.e., the magnitudes of the Fourier transform of the same as a function of the frequency under normal flow conditions where the flow-related frequency is 12Hz. The Fourier transform exhibits a pronounced maximum at this frequency which is easily identified whereas the remaining components having relatively large magnitude cluster around this maximum magnitude component frequency. Only one relatively far removed component of rather large magnitude is present, namely the interferer at the 50Hz frequency of the electrical power supply. As it is known beforehand that this interferer will appear it can be removed or masked. If there are other known interferers they can be treated in the same way. From the frequency of the component of largest magnitude the flow rate can be derived using (1) .

In a no-flow situation the method outlined above would yield a spurious positive flow result. This can be seen from the Fig. 2b which shows such a no-flow situation where the housing 1 containing the flow meter is separated from the remainder of the tubing system by closed valves. In this case a rather pronounced maximum at 80Hz which is caused by a motor driving a pump in the tubing system would be misread as indicating a substantial flow rate, in particular as it is comparable in magnitude with the flow-related maximum magnitude component at 12Hz. It has been found, however, and is also apparent from the diagram that in a no-flow condition the periodic components of the output signal and their distribution over the frequency range have specific properties which are reflected in the Fourier transform and can be used to identify this condition. There are several magnitude maxima which are either completely isolated or belong to a small cluster of further components with relatively large magnitudes. The frequencies of the maximum magnitude components in the frequency range are, at least in part, relatively far removed from each other, their quotients being distinctly larger or smaller than 1. This is due to the fact that the dominating periodic components are a component related to the motor frequency - 40Hz in the case shown in Fig. 2b - and its harmonics - which in Fig. 2b appear at 80Hz and 120Hz. Here again, the power supply component at 50Hz with its rather large magnitude can be removed or masked before evaluation. A no-flow condition can therefore be detected by an appropriate evaluation of the output signal, in particular of its Fourier transform, where the presence of features reflecting the said properties is investigated. An undetected misreading of the output signal can thereby be avoided.

As a first step, a set of components of large magnitude is selected from the components of the Fourier transform. In particular these selected components are local maxima in the frequency spectrum. To this end, the components of the

Fourier transform which may have been preprocessed, e.g., by removing known interferers as explained above, are ordered by descending magnitude. A predetermined number which depends on the frequency resolution and other parameters, e.g., fifteen, of the components at the upper end of the ordered sequence are then identified and assigned to the set of large magnitude components. If components have frequencies which are very close to each other, a single component can be formed from them and only this component considered further. This can be done, e.g., by rounding each of the frequencies to its closest integer and retaining only one component for a given rounded frequency while dropping the others. In a normal flow situation this process will usually leave only the maximum magnitude component and possibly further components with frequencies which are close to that of the maximum magnitude component. If only the maximum magnitude component remains the prevailing condition is identified as a normal flow condition with a positive flow rate reflected by the frequency of the said component according to (1) . Frequency and flow rate can then be assessed with greater precision by methods known in the art. Otherwise one of the components from this set of large magnitude components, usually the one with the largest magnitude, is identified as a reference component and its frequency as a reference frequency. Quotients of the remaining frequencies of the components in the set by the reference frequency are then calculated and those frequency quotients compared with a quotient threshold of, e.g., 1.15 and its inverse. In a normal situation with positive flow rate there will be either only one frequency with no frequency quotients to be calculated or the frequency quotients will be close to 1, i.e., either smaller than 1.15 or larger than its inverse (which is 0.87) .

Under a no-flow condition, on the other hand, the set of large magnitude components contains components pertaining to frequencies which correspond to the motor frequency and its harmonics, that is, the set of large magnitude components will contain several, usually five or more components with frequencies which are, at least in part, rather far removed from each other, their quotients being close to small integers or inverses or quotients of small integers.

Although the magnitudes of the components pertaining to the motor frequency and harmonics of the same vary considerably depending on the motor frequency and other parameters, with sometimes the motor frequency component having the largest magnitude, sometimes its second or third harmonic, there will always be at least one component in the set whose frequency differs from the frequency of the maximum magnitude component by at least a factor of 1.2, the frequency quotient of the sixth and fifth harmonics or, taking errors caused by rounding and other effects into account, by at least 1.15, or by its inverse which is 0.87. In order to decide whether there is a no-flow condition or not it is therefore usually sufficient to choose the maximum magnitude component as a reference component and calculate frequency quotients by dividing the frequencies pertaining to the other elements of the set of large magnitude components by the reference frequency pertaining to the reference component and check whether at least a predetermined minimum number, e.g., one, two or three of the frequency quotients is either larger than a quotient threshold of 1.15 or smaller than its inverse. In many applications this may be enough to imply a no-flow condition. This condition can then be indicated by switching a Boolean no-flow flag indicating the respective condition from its default value of 0 - indicating a normal flow condition - to 1 - indicating a no-flow condition. If the no-flow flag equals 1 a spurious positive flow rate result can be corrected to 0 and corrective measures taken where appropriate .

In particular where the set of large magnitude components contains only two or three elements applying additionally after a positive outcome or alternatively a more stringent criterion may be helpful to improve the reliability of the no-flow indication. For this purpose, the frequency quotients are checked for whether they are close to the quotients of a basic frequency and its harmonics. A set of numbers can be calculated or retrieved from a memory which consists, e.g., of integers between 2 and 6 and its inverses or of 2 and 3 and multiples thereof and its inverses, possibly complemented by 2/3, and the frequency quotients compared with them.

A no-flow condition is then indicated only if at least one of the frequency quotients coincides with one of those numbers, i.e., deviates from the same by not more than a predetermined deviation threshold which accounts for limited resolution, rounding errors and other effects and may be, e.g., 2%. The condition can be made more stringent by demanding that more than one frequency quotient coincides with one of the numbers or even specific ones of them.

One or several comparison sets can be formed from the numbers in question and indication of a no-flow situation made dependent on whether every element in a comparison set coincides with one of the frequency quotients. The results can then be logically OR-connected or connected in more complex ways. For example, all subsets of a certain cardinality, say, 1 - as described above - or 2, 3 or 4 of the numbers can be formed and used as comparison sets, i.e., coincidence of all members in each subset with one of the frequency quotients checked for and the results OR- connected. A no-flow condition is indicated if the check yields a positive result in at least one case, that is, if, apart from a basic frequency normally equalling the motor frequency, at least one, two, three or four harmonics, respectively, are present.

Many deviations from the above-described methods are possible within the scope of the invention. In stead of a vortex flow meter as described a swirl flow meter can be employed. In the evaluation of the output signal normal flow conditions and no-flow conditions are distinguished based on the magnitudes and distributions over the frequency range of periodic components of the signal. In a general way, the existence of several maxima which are - at least in part - at some considerable distance from each other in the frequency domain points to a no-flow condition. Depending on the application it may be sufficient to select the largest magnitude components and fuse them into one if they are close together in the frequency domain and count the remaining components. If, in particular after the removal of known interferers, more than one component remains this may already indicate a no-flow condition. The fusing may be done as explained above by dropping components close in frequency to components of larger magnitudes or by summing or integrating magnitudes over certain frequency intervals and assigning the resulting magnitude to the frequency of the component of largest magnitude in the frequency interval or to a weighted mean frequency or median frequency or in other ways .

In other applications more stringent criteria may be more adequate and minimum distances, i.e., quotients of frequencies of large magnitude components required. In particular, such minimum distance from a reference frequency may be required for one or more components in the set of large magnitude components. As there is usually at least a motor frequency and its second harmonic in the said set the minimum distance may in most cases safely be chosen as slightly less than 2. It is also possible to additionally apply an upper limit for the distance. An even stricter criterion - as explained above - involves the comparison of frequency quotients with certain small integers and their quotients which are bound to turn up where harmonics appear. General or specific, stricter criteria may be used - either alone or step-wise - dependent to the application, according to results from experiments and practical experience. List of reference symbols

tube bluff body paddle evaluation unit

Claims

P A T E N T C L A I M S

1. A method for the evaluation of an output signal of a vortex or swirl flow meter, characterized in that the output signal is examined for the presence of a no-flow condition in that components of the output signal which are periodic in time are identified and their magnitudes and frequencies assessed, a set of components of large magnitude, is selected, frequencies of the components of large magnitude are identified, at least one reference frequency is selected from them and frequency quotients of the other identified frequencies by the at least one reference frequency are calculated, and a no-flow condition is detected if at least the condition is met that the set of frequency quotients comprises at least a quotient which is outside an interval between 0.87 to 1.15.

2. The method of claim 1, characterized in that the minimum number of quotients is at least two.

3. The method of claim 1 or 2, characterized in that a component whose magnitude is larger than the magnitudes of the remaining components of large magnitude is identified as a reference component and its frequency identified as the reference frequency, and the frequency quotients are the quotients of the frequencies of the remaining large magnitude components by the reference frequency.

4. The method of one of claims 1 to 3, characterized in that a no-flow condition is indicated only if the set of the frequency quotients comprises a non-empty set of numbers deviating from the numbers in at least one comparison set consisting of a set of integers and quotients of integers by less than a predetermined deviation threshold.

5. The method of claim 4, characterized in that the at least one comparison set is any subset of a predetermined cardinality of a predetermined set of integers and quotients of integers.

6. The method of claim 5, characterized in that the predetermined cardinality is between 1 and 4.

7. The method of one of claim 5 or 6, characterized in that the predetermined set of integers and quotients of integers consists of integers between 1 and 6 and quotients of such integers.

8. The method of one of claims 5 to 7, characterized in that the predetermined set of integers and quotients of integers consists of the integers 2, 3 and multiples thereof and inverses of such integers.

9. The method of one of claims 1 to 8, characterized in that the selection of the set of components of large magnitude involves the step of ordering components according to descending magnitudes and selecting a predetermined number of components at the upper end.

10. The method of one of claims 1 to 9, characterized in that the selection of the set of components of large magnitude involves the step of replacing several components close in frequency by a single component.

11. The method of one of claims 1 to 10, characterized in that the components of the output signal which are periodic in time are in each case determined by

Fourier-transforming the output signal as a function of time over a predetermined evaluation interval.

12. The method of one of claims 1 to 11, characterized in that before or during the selection of a set of components of large magnitude at least one interferer of known frequency is removed.

13. The method of claim 1, characterized in that the set of selected components of large magnitude is the set of selected components of local maxima in the frequency spectrum.