FI89835C - Method and apparatus for determining the velocity of a gas flowing in a pipe - Google Patents

Description

s ς Λ Γ

Method and apparatus for determining the velocity of the gas flowing in a pipe Förfarande och anordning för bestämning av gaastigheten pä en gas som strömmar i ett rör 5

The invention relates to a method for measuring the flow rate, volume flow and / or mass flow of a gas flowing in a pipe, in which sound sensors are arranged in the flow pipe at a certain longitudinal distance 10 sound and flow rates are determined by the travel times of said sound upstream and downstream in the measurement range, the cross-sectional area of the pipe and the density of the gas to be measured.

The invention further relates to a device for measuring gas flow rate and / or quantities derived therefrom, such as volume flow and / or mass flow, which device comprises a measuring tube in which the flow to be measured flows and which device 20 comprises wideband, low frequency and only plane wavefront sound signals and transmitters power amplifiers, microphones and microphone signal amplifiers as sound detectors, and each device is equipped with calculation devices and software for calculating said flow rate or quantities from the sound detectors, the longitudinal distance of the flow tube, the cross-sectional area of the measuring tube and the flow density, the flow gas density upstream on the distance between the sound detectors.

It is known that the rate of propagation of an acoustic piston mode in a pipe does not depend on the flow profile or other profiles causing local sound velocity variations, but on the average flow rate integrated over the pipe cross-section and the resting B. velocity of the gas composition filled with the pipe. Robertson "Effect of arbitrary temperature and flow profiles on the speed of sound in a pipe" J. Acoust. Soc. Am., Vol 62, No. 4, pp. 813-818, October 2 fc $ ι; 5 i; 1977 and B. Robertson, "Flow and temperature profile independence of flow measurements using long acoustic waves," Transactions of the ASME, Vol 106, pp. 18-20, March 1984]. Accurate, profile-independent flow rate measurement is thus possible by measuring the travel times of the piston mode upstream and downstream 5 within a certain measurement interval, preferably using broadband sound, as disclosed in the applicant's FI patent No. 76885. In order for the higher wave modes interfering with the measurements to be attenuated already in the vicinity of the sound sources, the sound frequency must be sufficiently below a certain cut-off frequency depending on the shape and dimensions of the flow tube and the speed of sound propagation. For a round tube, this cutoff frequency 10 is

The cut-off frequency fe = (c - vwac) / (1.7 * D) (1) where c is the speed of sound in the resting gas, vmax is the maximum rated flow rate and D is the pipe diameter. It is known that flow rates can be calculated from formulas

Flow rate [m / s] v = 0.5 * L * (t / J -1 ^ 1) (2) 20 Volume flow [m ^ / s] Q = v * A (3)

Mass flow [kg / s] M = Q * p (4) where 25 v is the average flow rate L is the distance between the sound sensors in the longitudinal direction of the pipe 11 is the flow time of the sound downstream between L t2 is the flow time of the sound upstream between LQ is the volume flow 30 A is the cross-sectional area of the pipe M is the mass flow p is the gas density I: k 'J · 1 · ό i'j 3

Em. The FI patent discloses the measurement of transit times by correlation technology and the FI patent does not restrict the choice of broadband transmission or correlator type in any way. The subclaims describe the use of a polarity correlator, which can be implemented in full parallel. In practice, it has been found that achieving the required signal-to-noise ratio with the polarity correlation technique in generally noisy measurement conditions requires that audio transmissions be transmitted alternately upstream and downstream and that microphone signals be filtered to pick up only transmitted audio and eliminate interference. The sound transmitted in the form of frequency-scanning and the filter connected to follow it in principle also allow sound to be transmitted in both directions simultaneously, but at different instantaneous frequencies, so that the forward and counter-current signals of both sound sensors can be separated by a total of four sweepable filters. : o 916102.

A disadvantage of the prior art discussed above is that the frequency sweep is in principle non-stationary, in which case the necessary filters are also time-dependent. This causes dispersion, i.e. the delay of the filters is frequency dependent, which further causes an easy error in determining the running time, unless the sweep is selected correctly, the sweep is selected too fast or the start and end of the sweep are omitted from the measurement. These disadvantages can be avoided if stationary, broadband transmissions are used instead of frequency scanning, in which all frequencies ring at a constant amplitude at all times. At the same time, this means that sound is transmitted in both directions at the same time. For example, current DSP technology enables versatile real-time processing of audio frequency signals 25 and offers several possibilities for determining transit time information in a noisy environment.

In order to eliminate the above-mentioned drawbacks and to achieve later goals, the method of the invention is mainly characterized in that said sound is transmitted to the flow tube in the measuring interval simultaneously in both directions as stationary, periodic sequences, downstream sequence Sd (t, T), and downstream where T is the length of the period and t κ 9 9 / '· - 4 is the relative time measured from the beginning of each period, and that said sequences Srf (t, T), Su (t, T) and are orthogonal to each other, i.e. they have no common , non - zero frequency components.

The device according to the invention, in turn, is mainly characterized in that the sound transmitted via said loudspeakers consists of wideband, sequential, simultaneously downstream and countercurrently transmitted sequences T in the downstream sequence Sd (t, T) and the countercurrent sequence SM (t, T), which sequences Sd (t, T), Su (t, T) and are arranged to be orthogonal to each other, i.e. they do not contain common non-zero frequency components.

In the following, the theoretical background of the invention and some application examples of the invention will be described in detail with reference to the figures of the accompanying drawing, in which Fig. 1 is a block diagram and a lower curve of a correlation peak of odd frequencies thought to be measured upstream, Fig. 4 shows an embodiment corresponding to Fig. 3, in which its Hilbert transform is chosen instead of the second correlated sequence, Fig. 5 shows phase-difference curves with even and odd frequencies reflection echoes, and Figure 6 shows the cumulative distributions corresponding to Figure 5.

5 t: c:

Figure 1 shows a general block diagram of a system by which the invention can be advantageously implemented. The flow to be measured, e.g. the natural gas flow, travels at speed v in a flow pipe 1 with sound sources 2a and 2b, preferably speakers generating sound sequences in flow pipe 1. Between sound sources 2a 5 and 2b sound transducers 6a and 6b are arranged. A determine the scaling factor of the flowmeter. Fig. 1 4a and 4b are power amplifiers, 5a and 5b are signal amplifiers, 6 is a real-time processor, preferably a digital signal processor with the necessary analog inputs and outputs, which can implement procedures based on FFT algorithms or FIR filter-10 algorithms, 7 is a system through which communication between the real-time processor 6, the display device 8 and the separate PC workstation 9 is implemented.

In determining the travel time of stationary, wideband sound for flow measurement purposes, it is essential that the sounds transmitted simultaneously from the speakers 2a and 2b in both directions do not interfere with each other and that the measurement is also as insensitive as possible to the noise of the pipe 1. Consider the periodic but otherwise arbitrary sound sequences of the speakers 2a and 2b for a given measurement period T. It is known that such signals consist of discrete, evenly spaced, multiple frequency components AWJ * cos (m * 2 * 7r * t / T + 0 /> J) of the fundamental frequency / = 1 / T, which are orthogonal to each other, i.e. over a period the calculated correlation functions disappear when m and n are different large (Equation 5).

T

2 Λ £ Λ = Am * An * 6mjt * cos (m * 2n * -j) (5)

The main idea of the invention is to transmit downstream and upstream of the speakers 2a and 2b to the measuring tube 1 such wideband and continuous audio sequences, S 1 (t, T) and Su (t, T), which do not contain any common frequency components and are thus orthogonal, i.e. their the correlation function identically disappears: k Q Γ s 6 ~ rc * (0) = JS pj) * Su (t-QJ) * dt SO (6) o Then the signals of the microphones 3a and 3b, let D ^ (t, T) be L the signal of the left end microphone 3a and the signal of the right end microphone 5 3b of the DÄ (t, T) range, except for background noise, can be unambiguously decomposed into mutually orthogonal upstream and downstream components: DL (t, T) = DjJf, T) + DJtJ) DR ( t, T) = D ^ tJ) + D ^ tJ) U) This main idea of the invention makes it possible to solve many problems that degrade the signal-to-noise ratio and leads to a number of different implementation options. Over a period of more than 10 minutes, only the smaller part of the external pipe noise hits these discrete frequency components, the more periods over which the situation is considered. In averaging over N periods, the periodic signal increases in direct proportion to N, while the non-eccentric noise increases in proportion to the square root of N.

15

The phase angles of the discrete frequency components can in principle be chosen freely. In order to minimize the effect of the non-linearity of the speakers 2a and 2b or similar sound sources and for example due to the safety regulations of the natural gas pipelines, momentary sound power peaks should be avoided. This goal is well achieved 20 if said phase angles are chosen randomly. The frequency range used is limited at the upper end by the so-called a cut-off frequency [formula (1)] at which, in addition to the piston mode, the first unattenuated higher mode appears, which has a different propagation speed and thus interferes with the accurate determination of the propagation speed of the piston mode. At the lower end, the mains frequency of the AC network and the frequencies lower than its first supercar 25 should often be cut off.

li W '%:' Ί; r 7 The division of the remaining frequencies between the upstream and downstream audio sequences can be done in different ways. The simplest is to choose even frequencies for one sequence and odd frequencies for another. This selection results in each sequence consisting of two half-cycles, of which the half-cycles of the even frequencies 5 are identical to each other, as shown in the upper graph of Figure 2, and the half-cycles of the odd-numbered frequencies are also identical to each other, as shown in the lower graph of Figure 2. Thus, the upstream and downstream sounds can be simply distinguished from each other by adding an even number of half-cycles together a second time as such and a second time 10 by first changing the sign of every other half-cycle. Another option is to draw the available frequencies irregularly but evenly into the upstream and downstream sequences.

The travel times of the measurement sound in the measurement interval L upstream and downstream can be determined from the same measurement data in several different ways. The basic options are to form a filtered correlation function of the signals of the microphones 3a and 3b as a function of the time difference or to consider the the phase shift of successive frequency components of the signals as a function of frequency. The fast Fourier transforms of digital signal processors with FFT algorithms allow the signal to be processed in time and frequency space 20 on exactly the same devices as needed, interfering only with the real-time program.

Determining the travel time of the measurement sound from the correlation functions is as follows. The most ideal situation for determining the center of the correlation peak is when there is only one peak in the whole correlation function, i.e. the amplitudes of the other possible peaks are so small that their distortion in the region of the main peak is below the allowed level. The origin of the interference peaks may be in the wrong direction of the propagating background noise, from which the correlation filter has left the selected frequency components in the main peak or the sensor connections and connections that cause the 30 interfering echo peaks. If the interference peaks cannot be attenuated small enough, their location and area of influence must be sought to get the main peak out of range. The location of the echoes can be influenced by the connections of the sensors 3a and 3b

b 9 Π 9 S

8 placement and the area of influence so that the frequency dependence of the amplitudes of the frequency components is made uniform while avoiding sharp points of discontinuity. Three different correlation functions can be formed from the signals of the microphones 3a and 3b

n * T

C (0) = -L · f Averm [DL (t, Dl * Averm [DR (t-Q, T,)} * dt n * TJ0

n * T

CW = - ^ f Fd * Averm [DL (t, Dl * Averm [DR (t-Q, T)] * dt n * T o

n * T

/ AverJDu.ftT)] -AverJDJyt-O.T)], dt (8)

n * T o n * T

Ce (0) = J- / Fu * Averm [DL (t, Dl * Averm [DR (tQ, Dl * dt n * T o «* r -LiAw'JPJtm'fimJPJt-VH '* n * T0 5 where Fd and Fu denote filter operators that extract only the downstream or countercurrent portion of the detector signals, Averm means averaging over m periods and averaging n correlation functions by integrating over n periods.The first correlation function is the simplest and contains essentially two peaks, and the other with a negative delay value of 10 corresponding to the flow of sound upstream and downstream.This is also the most sensitive to interference.Each of the following filtered correlation functions contains essentially only one peak, the other corresponding to the flow of sound only downstream and the other only upstream, Figure 3 shows the upstream downstream. a correlation peak consisting of even frequencies and correlates consisting of odd frequencies considered to be measured downstream 15 below The small side peaks are caused by negative reflections from the speaker branches.

Instead of directly correlating the microphone signal of each, upstream or downstream to the selected frequencies, each signal of the microphones 20a and 3b can be correlated separately to a constant phase reference sequence with the desired amplitude distribution, Rrf (t, T) and RM (t, T) derived from the corresponding transmission sequences Sd (t, T) and Su (t, T) of li b '«<' 7- ϋ 9 or from the microphone sequences measured in the zero-flow situation and producing the following correlation functions

T

cjfi, D = fRJM) * DL (t-e, D * dt t m CJQJ) = fR £, T) * DR (t-Q, T) * dt o

T

C J0,7) = fRu (t, 7) * DL (t-QJ) * dt tm C ^, 7) = [Ru (t, T) * DR (tQ, T) * dt o where the time of sound in the measurement interval L downstream is determined as the time difference between the centers of the peaks of the correlation functions £ χ ^ (θ, Τ) and CRd (Θ, Τ) and upstream as the time difference between the centers of the peaks of the correlation functions CLu (6, T) and C ^ U (6, T). For this, either two FIR filter pairs (finite impulse response) can be used, one of which is the downstream reference sequence and the other the countercurrent reference sequence, respectively, or the operation can be performed in frequency space-10 sa as the following multiplications of the components of Fourier transform vectors ^ (ω /) = Λ + Ζ) ι (ω /) (l0b) where * as superscript means complex conjugate, and inversely converting the input vectors into time-space correlation functions. Again, it is possible to make averages similar to those made for the correlation functions.

10 ς Γ ·: <:

Instead of the usual correlation function, which ideally produces completely symmetric peaks, in all the above cases, instead of another sequence to be correlated, a Hilbert transform can be chosen, whereby the correlation peaks become completely antisymmetric. The center of these zero plane 5 intersections may be more accurately determined than the center of the symmetrical peaks. This case is illustrated in Figure 4, which, except for symmetry, otherwise corresponds to the case shown in Figure 3.

Determining the travel time as a frequency dependence of the phase angle is as follows.

10 If the correlation function contains essentially only one peak, the phase angle of its Fourier transform increases or decreases linearly as a function of frequency, the slope being proportional to the peak displacement from the origin, because the time displacement Θ is realized in frequency space by multiplying by exp (2ffi * <a7 * 0). An estimate of the travel time of the measurement sound in the measurement interval L is now obtained by averaging the phase angle differences calculated over 15 consecutive frequency steps of the same parity between 17 ... l2:, * ~,, * Σ argus [D ^ M) iD ^ o),))] 2π /,, - < U) θ- =? Λ- ,, * Σ argutiPi ^ u, .j) * /> * (<*> i + 1) ^ (ω,) * 0Λι (ω /)]

In the formulas (11), Fourier transforms of the signals of the microphones 3a and 3b have always been used as frequency vectors. As discussed above in connection with the correlation functions, also in formula (11) the vector derived from the signal of the second detector 3a, 3b can always be replaced by a Fourier transform of the reference sequences Rd (t, T) and R „(t, T) to give the next four time shifts? dRd, dLu and QRu. Travel times downstream are obtained as the difference between the first two and travel time upstream is obtained as the difference between the latter two. In all the above cases, the successive phase angle differences can alternatively be formed as a cumulative function, and the time shifts can be determined as the slope of this regression line.

li K 9; '7 »11

If there are other peaks in the correlation function, modulation occurs in both the phase difference distribution and the cumulative function. However, the right line is not dropped if the phase angle difference truly remains between -7Γ ... + 7Γ. The risk of exceeding or falling below these limits is reduced if the mean of the phase angle diff-5 rents remains close to zero. For this reason, it is worth comparing the actual travel time with, for example, the travel time of the zero flow situation of the reference pipe 1, in which case only the travel time difference caused by the flow rate v is shown as a small deviation from zero of the average phase difference distribution. Figure 5 shows the phase difference graphs 10 for both even and odd frequencies in the case where the modulation is due to reflection echoes alone. Figure 6 shows the corresponding cumulative distributions.

The applicability of the phase angle method requires that the correlation function have a distinct principal peak and that the phase angle of its Fourier transform at each individual frequency is reasonably close to the value corresponding to the travel time of this peak.

The following claims set forth within the scope of the inventive idea, the various details of the invention may vary and differ from those set forth above by way of example only.

Claims (12)

A method for measuring the flow rate, volume flow and / or mass flow of a gas flowing in a tube, wherein method of placing sound transducers (2a, 3b) in a flow tube (1) at a given longitudinal entry distance in the measuring tube (1) and in which method by means of sound sources (2a, 2b), longitudinal sound advancing downstream and downstream in the flow tube (1) is supplied only in the basic mode in the form of a plane path front to the flow tube (1) from a location outside the measuring distance ( L) which is limited by said transducer (3a, 3b) and the flow quantities are determined from the running times of said sound by and by the current on the measuring distance (L), from the transverse surface (A) of the pipe and from the density (p) of the gas to be measured. hence, said sound is transmitted to the flow tube (1) on the measuring distance (L) in the form of stationary periodic sequences advancing simultaneously in both directions, the sequence Sd (t, T) downstream and the sequence Su (t, T) countercurrent, where T is the length of the period and t is the relative time measured at the beginning of each period, and that said sequences Sd (t, T) and Su (t, T) are mutually orthogonal and thus have no common frequency components at all that deviate from zero. 2. A method according to claim 1, characterized in that the respective orthogonal Sd (t, T) and Su (t, T) sound sequences are designed so that in relation to the fundamental frequency corresponding to their basic period, the sequence Sd (t , T) / Su (t, T) to be transmitted in one direction only uniform and the sequence Su (t, T) / Sd (t, T) transmitted in the other direction only odd frequency components. 3. A method according to claim 1 or 2, characterized in that the duration of said noise on the measuring distance (L) is measured in the countercurrent and countercurrent time as offset values of the midpoint from the origin of the peaks included in the correlation functions formed from the signal DL (t, T) from the left sensor and from the signal DR (t, T) from the right sensor or as a difference between two such offset values, by forming a correlation γ-y; 19 function from the signals as usual from both sensors (3a, 3b), by first forming their averages by counting over several (m) periods T or by still forming the average of the correlation functions also by integrating over several (n) periods , and by thereby determining the maturities 5 as time offsets from the origin of two main peaks of the correlation function • r CjjfB) = f AverJDL (t, T)] * Averm [DR (tQ, T)] * dt n * T JQ 10 . 4. A method according to claim 1 or 2, characterized in that the duration of said noise on the measuring distance (L) is measured in the countercurrent and countercurrent in the space of time as offset values from the origin of the midpoint of the peaks formed by the correlation functions formed by the signal DL (t, T) from the left sensor and the signal DR (t, T) from the higher sensor or as a difference of two such offset values, by forming two different correlation functions from the signals of the two sensors as such, by first forming their means value by counting over several periods (m) or by still integrating the correlation functions over several (n) periods, one Cd (0) filtered with co-current filter Fd and the other Cu (0) with countercurrent filter Fu, and by determining the maturities as time offsets from the origin of the only major peaks of the two correlation functions obtained here κ · Τ CJ.Q) = f Fd * Averm [DL (t, T)) * Avern [ DK (tQ, T)] * dt n + I J0 «r ς, (θ) = / Fu * Averm {DL (t, T)] MverJD / Γ-θ, θ)] * dt ft * l„ 30 0 b 9 L '20 5. A method according to claim 1 or 2, characterized in that the duration of said noise on the measuring distance (L) is measured co-and countercurrent by forming a total of four correlation functions Cm (t), CRd (t), CLu (t) in the time interval. CRu (t) from the signal DL (t, T) of the left sensor and the signal DR (t, T) of the higher sensor, always from a detector signal such as the seed, or by forming the average over several (m) periods and from negate the reference sequence Rd (t, T) or Ru (t, T) in the solid phase read from memory, which reference sequences are derived from corresponding transmission sequences or from detector sequences measured with the current zero, possibly by integrating even more over several periods Η · Τ CJfi) ^ / R / t, n'AvtrJ (DL (te, ni'dt ^ j »* r CjuiÖ) = ^ / RAt> T) * Averm [DR (tQ, m * dt n * T CJO) = - ^ / * Averm [DL (tB, T)] * dt η · Γ CA (0) = f Ru (t, O * Averm [DR (tQ, T)] * dt n * T J0 25 and by determining the maturity downstream as a time difference between the centers of the main peaks of the correlation functions Οω (θ) and CRd (0) and the maturity countercurrent in a similar way as a time difference while the center points of the main peaks of the correlation functions C, u (t) and C-Ru (t). Method according to any one of claims 1-5, characterized in that instead of correlation functions which contain substantially symmetric li 21 21 S, Q P; C • J - V W peaks use their Hilbert transforms as correlation functions and determine the center points of the thus obtained essentially antisymmetric correlation functions most appropriately as the values at the time when the steepest function intersects zero level. 5 7. A method according to claim 1 or 2, characterized in that the duration of the sound co-or countercurrent is determined in the frequency space in such a way that first the Fourier transforms of the sound transducer sequences are formed in the form of complex frequency vectors DLu (<oj), DRd (& >,) and DRu (<oj) and DRu (o>,) 10 and determine the duration of the sound with upstream 0d and countercurrent 0U substantially from the formulas -h 13 2 * i, χ he ,, »-— ττΣ 20 where index 1 applies only even frequency values in the upper equation and only odd in the lower equation. 8. A method according to claim 1 or 2, characterized in that the durations of the sound co-and countercurrent are determined in the frequency space in such a way that first Fourier transforms of the sound transducer sequences are formed in the form of complex frequency vectors ϋω (ω χ), DLu (ti >,), DRd (o> x), and DRu (o> d) and Fourier transforms of the reference sequences derived from the sound transducer sequences measured from the solid phase transmitter sequences or in the zero-current situation in the form of complex frequency vectors Rd (o,) and Ru ( <o,), and determines the duration of the sound 30 upstream 0d and countercurrent 0U as differences where 0 0, 0Rd, 0Lu and 0Ru are essentially determined from the equations 22h 9 ί \ λ κ T * 6U «- ^ - 2ΛΓ« Μ5ΐ / ϊ / ωΜ * Ι) 2 <(ωί.1) · ^ (ωρ * Ι> χ / ω ^)] ^ “'ih 5 th ~ hkt * - -i — Σ cr8us [R £ uUl) * D [M (uh ]) · * Β * (ωι) * DU (ω ^ h "* i 't -r ΘΛ 0Γ ^ Λ. (ΩΜ) * D *» (UbO mR * (ω ^ βΖ) Α. (Ω ^ 10 h ~ hp where index 1 applies only to even frequency values in the TVs are the uppermost formulas and only odd in the two lower equations Apparatus for measuring the flow rate of a gas and / or quantities derived therefrom, such as the volume stream and / or the mass stream, which apparatus comprises a measuring tube (1), wherein the stream to be measured runs and which apparatus comprises loudspeakers (2a, 2b). ) as transmitters for low frequency broadband audio signals advancing only in the form of a ground path front in basic mode and power amplifiers (4a, 4b) for these, microphones (3a, 3b) as sound detectors and amplifiers (5a, 5b) for the microphone signals, and which means is provided with counting devices and programs for calculating said flow quantity or quantities from the longitudinal distance (L) of the flow tube (1) between the sound detectors (3a, 3b), from the transverse surface (A) of the measuring tube, from the density (p) of the flow gas and from the duration of the audio signals co-current (tx) and countercurrent (t2) on the distance (L) between the sound detectors (3a, 3b), the characteristic ad thereof, that the sound transmitted via said speakers (2a, 2b) is composed of broadband sequences of the same period transmitted simultaneously within the period T co-and countercurrent, the co-current sequence Sd (t, T) and the countercurrent sequence Su (t, T), which sequences Sd (t, T) and Su (t, T) are arranged between each other orthogonal and thus contain no common frequency com bons that deviate from noli. 10. Device according to claim 9, characterized in that the sound to be transmitted in one direction contains only even and the sound to be transmitted in the other direction only odd frequency components. Device according to claim 9 or 10, characterized in that the device comprises a real-time signal processor (6), which transmits the output sequences of the sound sources and processes the responses produced by the sound detectors (3a, 3b). Device according to claim 11, characterized in that the device further comprises a system processor (7) which handles the communication between said signal processor (6) in real time, an optional digital display device (8) and an external computer (9). ) used as an operating connection.