WO2009074653A1

WO2009074653A1 - Method of characterizing the rate of flow of a two-phase fluid

Info

Publication number: WO2009074653A1
Application number: PCT/EP2008/067339
Authority: WO
Inventors: Eric Hervieu
Original assignee: Commissariat A L'energie Atomique
Priority date: 2007-12-12
Filing date: 2008-12-11
Publication date: 2009-06-18
Also published as: EP2220423A1; FR2925144A1; FR2925144B1

Abstract

The invention relates to a method of determining the rate of flow of a two-phase fluid flowing in a pipe, the method comprising:- constructing (E1) a three-dimensional space (emoy, fmoy, Rdb) in which various two-phase fluid flow speed domains (Dsr, Dsv, Db, Di) are identified, the three dimensions in space being, respectively, the mean spatial position, the mean frequency position and the mean spatial and frequency spread of a flow rate, - calculating (E2), from measurement signals (ms), parameters (emoy p, fmoy p, Rdb p) representative of the two-phase flow rate that is to be determined, and - positioning (E3) the two-phase flow rate that is to be determined in the three-dimensional space (emoy, fmoy, Rdb) on the basis of the calculated parameters.

Description

METHOD FOR CHARACTERIZING DIPHASIC FLUID FLOW REGIME

Technical field and prior art

The invention relates to a method for characterizing a two - phase fluid flow regime and to a method for determining a two - phase fluid flow regime that implements the method of characterizing the invention.

The handling and transport of two-phase fluids is currently a major technology issue in a number of industrial fields: petroleum engineering, chemical engineering, process engineering, energy engineering, and so on.

A two-phase mixture can flow according to several topological organizations, called flow regimes or flow configurations, which are governed by different mechanisms. The macroscopic behavior (pressure losses, heat exchange with the walls, mechanical stress of pipes and structures) of the flow can vary very strongly from one regime to another. From an industrial point of view, it is essential for the safety and longevity of an installation, that it works with the flow configuration for which it was dimensioned. Optimal control of an installation can use an active control procedure. The latter requires - at a minimum - to identify in real time the configuration present in the pipe to be monitored, in order to detect a possible change of regime and to retroact on the piloting of the installation before reaching a damaging configuration.

Traditionally, the identification of the regimes and their transitions is done visually with, in a complementary way, a direct analysis of signals such as, for example, pressure or presence rate signals. The results are represented in the form of a map. In this field, signal processing methods are also an indispensable tool because they extend the ability to observe and analyze the flow. Thus, for example, is it possible to characterize flow regimes by spectral analysis of pressure signals. However, if this methodology makes it possible to characterize the flow regimes, the identification of the boundaries between them is lacking in objectivity. Several attempts have been made to fill this gap. Several studies have also been carried out in the context of parametric methods applied to the estimation of fractal dimensions. Despite the range of possibilities opened up by these techniques, their effectiveness must still be demonstrated on a sufficiently representative database.

In the context of non-parametric approaches, time-frequency and time-scale (wavelet) analysis methods have been applied with great success to a large number of problems. and, in particular, the mechanics of two-phase fluids.

More recently, the use of neuronal methods has been the subject of several attempts, the first of which can be attributed to Mi et al.

(see reference (I)), which led to a network from a few statistical moments of a vacuum rate signal, with the aim of recognizing the different regimes of a vertical flow Crivelaro et al. reference [2]) have attempted to develop a different neural method by grossly exploiting temporal signal measurements.

In summary of the above, it is apparent from the analysis of the prior art that, in an attempt to characterize a two-phase flow, authors have implemented different methods, namely:

1) the statistical or spectral analysis of signals resulting from global measurements (pressure, vacuum rate, vibration) to recognize the flow regime,

2) the time-frequency analysis of signals from local measurements (impedance probe) to detect the transition between two flow regimes, without worrying about what these regimes may be, 3) the neural analysis of time signals raw from local measurements (impedance) to automatically recognize the flow regime,

4) the neuronal analysis of time-frequency signatures of signals from global measurements (pressure), to automatically recognize the flow regime. Type 1 methods suffer from two important shortcomings: they contain a significant amount of subjectivity and do not allow ambiguity to be distinguished between certain regimes whose signatures are similar (because they are global signals that are exploited). As a result, they require user interpretation and therefore do not allow automatic and reliable recognition. The type 2 method is intrinsically more objective. However, it only asserts that the flow is in a transition zone between two established regimes without diagnosing which are these two regimes. The Type 3 method takes advantage of the local flow characteristics, but the use of raw time signals at the input of the neural network considerably slows down the duration of the diagnosis, thereby precluding real-time use.

The type 4 method is based on a spectral basis since it involves a time-frequency analysis (also expensive in computing time) but, again, it is used to process a global signal.

The method of the invention does not have the disadvantages of the methods mentioned above.

SUMMARY OF THE INVENTION Indeed, the invention relates to a method for characterizing a fluid flow regime two-phase dipole which flows in a pipe, characterized in that it comprises the following steps:

Na measures m _D (j = 1, 2, ..., Na) representative of the two-phase flow regime flowing in the pipe, the Na measures being distributed according to a total or partial excursion of the perimeter of the pipe,

For each measurement m, the calculation of a digital signal S ₃ (j = 1, 2,..., Na) consisting of a plurality of time samples,

For each digital signal S _3, calculating a power spectral density PSD ₁₁ (f) of the signal S _3, f is a frequency variable, and the calculation of a given representative of _moy average position e of scheme two-phase flow, of a data of average frequency f _moy representative of the two-phase flow regime and of an average spreading data in space and in frequency R _dB representative of the two-phase flow regime, using the respective formulas following:

R _dB = 20 log (f _σ / e _σ ),

with

and

where P ₃ is the spatial position at which the measurement signal m _D (j = 1, ..., Na) is taken from the perimeter of the pipe.

According to a further feature of the invention, the Na measurement points are distributed substantially uniformly over a total excursion of a perimeter of the pipe.

The invention also relates to a method for determining a two-phase fluid flow regime, characterized in that it comprises:

- a step of characterizing the flow pattern to be determined for forming in accordance with the characterization method of the invention, a mean position data (w _av ^p), a medium selected frequency (f _avg ^p) and a given d mean space and frequency spread (RdB ^P ) of the flow regime to be determined,

a characterization step, according to the characterization method of the invention, of different types of two-phase fluid flow regime (smooth laminate, wave laminate, rough laminate and bubble) to construct a three-dimensional space (e _moy , f _m oy / ^R db) representative of different types of two-phase fluid flow regime, a first dimension (e _avg ) of the three-dimensional space being constructed by a set of mean position data that result from the step of characterizing the different types of flow regime, a second dimension (f _avm ) of the three-dimensional space being constructed by a set of average frequency data that result from the step of characterizing the different types of flow regime, the third dimension (R _db ) of the three-dimensional space being constructed by a set of data of mean space and frequency spread resulting from the step of characterizing the different types of flow regime, and

- a calculation step that positions in the three-dimensional space (E _moy, f _me, Rdb) t average position data (w _av ^p), the average of given frequency (f ^p _avg) and datum d mean space and frequency spread (RdB ^P ) of the flow regime to be determined. The flow regime determination method of the invention implements a signal analysis method for efficiently and quickly recognizing the configuration of a two-phase mixture flowing in a pipe. The flow regime determination method of the invention takes advantage, not only of spatial measurements of the flow, but also of the spectral characteristics of the flow, these spectral characteristics being present, to different degrees, in the measured signals.

Brief description of the figures

Other features and advantages of the invention will become apparent on reading a preferred embodiment with reference to the appended figures, among which: FIGS. 1A-1F represent different types of two-phase flows; FIGS. 2A and 2B show an example of a two-phase fluid flow regime characterization device which implements the method of the invention; FIG. 3 represents, in the case of an intermittent flow of two-phase fluid, an example of time signals which are delivered by the measuring device which forms part of the characterization device represented in FIGS. 2A and 2B; FIGS. 4A-7A and 4B-7B represent, for didactic purposes, signals which illustrate the advantages of implementing the characterization method of the invention; FIG. 8 represents, in a space of characteristic parameters obtained by the characterization method of the invention, the position of different diphasic fluid flow regimes; FIG. 9 represents a schematic diagram of a two-phase fluid flow regime determination method according to the invention; FIG. 10 represents a second example of measuring device used for the implementation of the method of the invention.

DETAILED DESCRIPTION OF EMBODIMENTS OF THE INVENTION In a horizontal pipe, a two-phase liquid / gas mixture, for example a water / air mixture, can flow in various configurations commonly known as flow regimes. These different flow regimes are shown in Figures 1A-1F.

FIG. 1A represents the so-called "smooth stratified flow regime" which is characterized by a continuous and smooth interface between the two phases of mixing. The liquid E, heavier than the gas A, flows down the pipe.

FIG. 1B shows the so-called "stratified wave flow regime" which differs from the previous one by the shape of the interface between water and air which, here, waves in the form of periodic waves Vp. This is due to the speed of the gas which is faster than that of the liquid.

Figure IC shows the regime called "rough stratified flow regime" which differs from the previous by its rough interface between the liquid and the gas. The interface is then composed of Vnp waves that are no longer periodic. This effect is caused by the increase of the velocity of the gas, which destroys the coherence of the waves.

Figure ID represents the regime called "bubble flow regime" which is characterized by a dispersion of the gas phase within the liquid phase. This regime appears for a high flow of the liquid. The high velocity of the liquid causes severe turbulence in the liquid, leading to splitting and dispersing the gas in the form of bubbles b. The gas bubbles b naturally tend to migrate up the pipe under the effect of gravity.

Figure IE represents the so-called "annular flow regime" regime. The annular flow is observed at high gas flow. The gas, predominant in the center of the pipe, projects the liquid L on the walls of the pipe in the form of a film F which flows non-uniformly due to gravity. There may be droplets of water that move at high speed, at the heart of the pipe, into the gas flow.

Figure IF represents the regime called "intermittent flow regime" which is characterized by a periodic alternation of water plugs B and P gas pockets. Its interface is unstable and discontinuous.

FIGS. 2A and 2B show an example of a two-phase fluid flow regime characterization device which implements the method of the invention. With reference to FIG. 2A, the characterization device comprises a device for measurement consisting of a segmented ring conductimetric probe M1, M2, an electronic processing circuit E and a computer K. FIG. 2B shows a cross-sectional view of the measuring device of FIG. 2A at the level of the ring M2.

The segmented rings Ml and M2 are flush with the internal wall of the pipe C in which the two-phase fluid flows. The first ring Ml is an excitation electrode connected to a generator G. The second ring M2 is segmented into Na distinct electrodes, for example sixteen electrodes, which are distributed over the entire internal perimeter of the pipe C. The conductimetric probe measures the azimuthal distribution of the electrical impedance of the two-phase fluid. "Azimuthal distribution" means a distribution according to the circular perimeter of the pipe. The distribution of the electrodes is preferably uniform over the perimeter of the pipe. The conductimetric probe has the advantages of being non-intrusive (the electrodes are flush with the inner wall) and offering a high bandwidth. Each electrode of the ring M2 takes a local measurement representative of the flow regime. Thus, Na measures m _D (j = 1, 2, ..., Na) representative of the two-phase flow regime are taken by the Na electrodes.

The electronic processing circuit E comprises dedicated electronic circuits which process the m _D measurements collected on the electrodes of the ring M2. Each electrode of rank j (j = 1, 2, ..., Na) of the ring M2 is thus connected to a circuit amplification and full wave rectification A ₃ which is itself connected to a low-pass filter F ₃ for eliminating high frequency noise. At the output of the filtering system, are thus delivered Na analog signals Sig _D (t) j (j = 1, 2, ..., Na) whose levels are adjusted during a calibration so that, for example, each analog signal is substantially equal to OV when the probe contains no liquid and substantially equal to 10V when the probe is full of liquid. The analog signals Sig _D (t) j (j = 1, 2, ..., Na) are then transmitted to a computer K, for example a computer.

The computer K comprises an analog / digital converter Ech, a power spectral density calculation module DSP and means for calculating characterization parameters M.

The analog / digital converter Ech comprises sample-and-hold circuits which digitize each analog signal Sig _D (t) while ensuring simultaneous sampling of the Na measurement signals which are taken at the same time by the Na electrodes. Na digital signals S ₃ (j = 1, 2, ..., Na) are thus delivered by the analog-to-digital converter Ech. The signals S ₃ are then transmitted to the power spectral density calculation module DSP which calculates, for each signal S ₃ , the power spectral density DSP ₁₁ (f) associated with the signal S ₃ . The characterization parameter calculation module M then calculates the characterization parameters of the two-phase flow regime, namely: - an average w _av position representative of the two-phase flow regime such that:

an average frequency f _moy representative of the two-phase flow regime such that:

J 'mmooyv

an average spreading data in space and in frequency R _dB representative of the two - phase flow regime such that:

R _dB = 20 log (f _σ with

and

where p _D is the spatial position at which the measurement signal m _D (j = 1, ..., Na) is taken from the perimeter of the pipe.

Advantageously, the characterization parameters e _moy , f _moy and R _dB unequivocally and exhaustively characterize the two-phase flow regime. As will be described in more detail later, with reference to FIGS. 8 and 9, this unambiguous and exhaustive nature of the characterization parameters makes it possible to implement, in a simple and rapid manner, a method capable of determining a flow regime of any kind. two-phase fluid.

FIG. 3 represents, by way of nonlimiting example, for an intermittent flow similar to that represented in FIG. 1F, signals Sig _D (t) delivered by the electronic processing circuit E, in the case where, for example, the measuring device comprises sixteen electrodes. Because of the symmetry of the pipe with respect to a vertical axis parallel to the axis of gravity, only eight signals out of sixteen are used (Sigi-Sigs corresponding to respective electrodes 1-8 of Figure 2B).

This contribution of only half of the measurements to the implementation of the method of the invention is of course possible, more generally, for any type of two-phase flow regime and for any electrode configuration when N electrodes are uniformly distributed over the perimeter of the pipe. FIGS. 4A-7A and 4B-7B represent, for didactic purposes, signals which illustrate the advantages of implementing the method of the invention. Since each electrode has a precise geometric position, it is possible to compare the spatial characteristics of the flow by juxtaposing all the time signals on the same three-dimensional space-time-amplitude diagram. The flow is thus visualized as a function of time and as a function of the different azimuthal positions in the pipe. Such a representation is illustrated for the different flow regimes in FIGS. 4A, 5A, 6A, 7A. Figures 4A-7A correspond to wave laminate flow, rough laminate flow, intermittent flow, and bubble flow, respectively.

As mentioned above, from the measurement signals taken from the electrodes, the method of the invention implements the calculation of the power spectral density for each signal. The use of the power spectral density of the signals collected on the different electrodes advantageously makes it possible to reveal the dominant frequencies of the signals.

FIGS. 4B, 5B, 6B, 7B represent, in a same three-dimensional space-frequency-amplitude reference, the power spectral densities which correspond to the respective signals represented, in the three-dimensional space-time-amplitude reference, on the Figures 4A, 5A, 6A, 7A. The representations of FIGS. 4B, 5B, 6B,

7B make it possible to locate the presence or absence of dominant frequencies in the different flow regimes, as a function of the azimuthal position of the fluid in the pipe.

For example, for the stratified wave flow illustrated in FIG. 4B, a peak located at the frequency of 5 Hertz appears at the No. 5 electrode. Indeed, in the stratified wave flow mode, the greatest fluctuations occur at the water / liquid interface which is in contact with the "equatorial" electrodes and, for this regime, the waves have a periodic structure well marked. The rough stratified flow (see figure

5B) shows a spectral spread at the No. 7 electrode. The energy deployment over a wide frequency range revealed by this flow regime, shows that for this regime, the periodic wave structure has disappeared. Moreover, the fact that the maximum of the power spectral density is obtained on the electrode No. 7 reveals that the water level is lower in the pipe than in wave flow. For the intermittent regime (see figure

6B), we find the value of the alternation frequency pockets-caps, but the form of "dorsal" which appears in the electrode-frequency plane of this representation is significant because it expresses the frequency concentration and the spatial delocalization of the components spectral. So, this scheme is it well characterized by a dominant frequency very marked at a frequency lower than the Hertz, frequency which is present on all the electrodes.

Finally, the bubble flow (see Figure 7B) reveals a spectral spread over a wide frequency range that seems to approach that of the rough stratified flow. However, it is seen that the maximum amplitude is obtained at the electrodes No. 3 and No. 4, which is explained by the fact that the bubbles rather flow in the upper part of the pipe. Thus, it is thanks to the spatial information that one can discriminate the regime of bubbles of the rough regime.

In the context of the invention, a large number of measurements are made for each of the four flow regimes mentioned above. It is then possible to completely delineate the domain of each regime in a three-dimensional space e _av , fmoyr RdB- Figure 8 illustrates this delimitation of domains. The respective domains of the four flow regimes are clearly disjoint: it appears the rough stratified domain Dsr, the stratified wave domain Dsv, the bubble domain Db and the intermittent domain Di. Fig. 9 illustrates a block diagram of the two-phase fluid flow regime determining method of the invention.

Thus, to determine a particular flow regime of two-phase fluid, the method of the invention includes: a step E 1 of characterization, using the method of characterization of the invention, of all the different two-phase fluid flow regimes that are likely to exist, to construct a space of the fluid flow regimes two-phase phase, E2 characterization step, using the characterization method of the invention, the particular flow regime to calculate the characteristic parameters of the particular flow regime, and a calculation step E3 which positions the characteristic parameters from step E2 in the space of two-phase fluid flow regimes constructed at the end of step E1.

From an MS set of measurements made for the set of two-phase flow regimes (typically the regimes mentioned above), the first step E1 of the method consists in constructing the three-dimensional space (e _av , fmoyr R _O B) in which the different diphasic fluid flow regimes are identified by the corresponding domains Dsr, Dsv, Db, Di. Moreover, from spatial measurements ms representative of the particular two-phase flow regime to be determined, the particular average spatial position e _average ^p , the particular average frequency position f _average ^p and the average space and frequency spread R _dB ^p of the particular two-phase flow regime to be determined are calculated in step E2. Step E3 then identifies the flow regime particular I by positioning the characteristic parameters of the particular flow regime delivered at the end of step E2 in the three-dimensional space constructed at the end of step E1. Step E3 is implemented by a calculator, for example a computer. Advantageously, the step E1 of the method of the invention can be carried out once and the data relating to the space of the flow regimes and to the different domains Dsr, Dsv, Db, Di are recorded in a memory, for example a computer memory. When a particular flow regime determination method is implemented, then in step E3, the data thus recorded is used. According to the preferred embodiment of the invention, the step E3 of positioning the flow regime to be determined is implemented by a neural algorithm that can be supervised or not. The three parameters e _average ^p , f _771oy ^p and R _dB ^p are then provided at the input of a network of two-layer perceptrons. At the output of the network, the last layer of perceptrons activates Boolean indicators which correspond to indicators of the various previously identified flow regimes. As known to those skilled in the art, a neural algorithm requires a learning phase. Advantageously, within the framework of the method of the invention, seventy real tests are sufficient to drive the perceptron network, by using a back propagation procedure, with a maximum learning depth of 4000 epochs. The method of the invention described above is based on the exploitation of local and spectral information obtained using measurements from a conductimetric probe. In another embodiment of the invention, the measurements come from a multi-pixel X-ray detector. This other embodiment is represented in FIG. 10. The measuring device comprises:

an X-ray source 1 which generates, in the direction of the duct, a collimated X-ray beam FX in the form of a sheet in fan geometry, and

a multi-pixel detector 2 placed in the beam field on the other side of the pipe. The FX beam illuminates and is attenuated by a cross section of the flow. Each pixel of the detector then measures, at each moment, the attenuation undergone by X-rays, thus revealing the thicknesses of fluid and gas traversed in an azimuthal direction in the pipe. The multi-pixel detector used is, for example, a commercial detector which integrates, over a chosen duration (for example a few tenths of a second), the photonic flux received by each pixel, converts the information thus integrated into an electrical signal and digitizes this electrical signal. This device consequently delivers local time measurement signals from which a two-phase flow regime can be identified according to a procedure identical to the procedure described above. References

[1] Crivelaro K.C. O., Seleghim Jr.P. & Hervieu E., 2002, "Detection of horizontal two-phase flow patterns through a neural network model", Journal of the Brazilian Society of Mechanical Sciences,

Flight. XXIV, N ⁰ I, pp.70-75.

[2] Mi Y., Ishii M. & Tsoukalas L.H., 1998, "Vertical two-phase flow identification using advanced instrumentation and neural networks", Nuclear Engineering and Design, vol. 184, pp. 409-420.

Claims

A method of characterizing a two-phase fluid flow regime flowing in a pipe, characterized in that it comprises the following steps:

- Na measures m _D (j = 1, 2, ..., Na) representative of the two-phase flow regime flowing in the pipe, the Na measurements being distributed according to a total or partial excursion of a perimeter of the conduct,

For each measurement m, the calculation of a digital signal S ₃ (j = 1, 2,..., Na) consisting of a plurality of time samples, for each digital signal S ₃ , the calculation a spectral power density DSP ₁₁ (f) of the signal S ₃ , f being a frequency variable, and

- The calculation of a given representative w _av average position of the two-phase flow regime of a given medium frequency f representative _moy the two-phase flow regime and a means spreading data in frequency space and R _dB representative of the two-phase flow regime, using the following respective formulas:

J 'mmooyv

R _dB = 20 log (f _o / e _o ),

with

and

where _D p is the spatial position at which is removed from the measurement signal m _D (j = l, ..., Na) on the perimeter of the pipe.

2. The method of claim 1, wherein the Na measurement points are distributed substantially uniformly over a total excursion of the perimeter of the pipe.

3. A method for determining a two-phase fluid flow regime, characterized in that it comprises: a step of characterization (E1) of the flow regime to be determined in order to form, according to the method of one of claims 1 or 2, an average position datum (e _average ^p ), a mean frequency data item (f _mo P 'and a spreading datum mean space and frequency (RdB ^P ) of the flow regime to be determined,

a characterization step (E2), in accordance with the method of one of claims 1 or 2, of different types of two-phase fluid flow regime (smooth laminate, wave laminate, rough laminate and bubble) to construct a three-dimensional space (e _moy, f _me, Rdb) representative of the various types of flow regimes of two-phase fluid, a first dimension (w _av) of the three-dimensional space is constructed by a set of position data average which result from the step of characterizing the different types of flow regime, a second dimension (f _avm ) of the three-dimensional space being constructed by a set of average frequency data that result from the step of characterizing the different types of flow regime, the third dimension (Rdb) of the three-dimensional space being constructed by a set of mean spread data in space and frequency e which result from the step of characterization of the different types of flow regime, and

a calculation step (E3) which sets, in the three-dimensional space, the average position datum (e _average ^p ) t the average frequency data item (f _average ^p ) and the average spreading data in space and in frequency (R _d B ^P ) of the flow regime to be determined.

4. The method of claim 3, wherein the computing step that positions in space three-dimensional, the average position data, the average frequency data and the average spreading data in space and in frequency of the flow regime to be determined is implemented by a neural algorithm.

The method of claim 4, wherein the average position data, the average frequency data, and the average space and frequency spread data are input to a multilayer perceptron array, a layer of perceptrons activating Boolean indicators at the output of the network, the Boolean indicators corresponding to indicators of the two-phase fluid flow regime domains.

6. Method according to one of claims 4 or 5, wherein the neural algorithm is supervised.

7. Method according to any one of the preceding claims wherein Na measurements are delivered by a conductimetric probe (Ml, M2) which comprises a segmented ring (M2) consisting of electrodes distributed on the inner wall of the pipe (C). .

The method according to any one of claims 1 to 6, wherein the Na measurements are delivered by a detector (2) which detects attenuation of radiation from an X-ray source which passes through the pipe and the flow.