RU2427828C1 - Method of measuring flow rate of multiphase fluid by picking up nuclear magnetic resonance (nmr) signal and device for realising said method - Google Patents

Abstract

FIELD: physics. ^ SUBSTANCE: measurement the flow rate of a multiphase fluid by picking up a nuclear magnetic resonance (NMR) signal under the effect of the gradient of a magnetic field directed along the direction of flow, followed by conversion of the measured signal into a frequency spectrum is carried out by exciting nuclei only in a part of the receiver coil - inside the layer directed across the flow and characterised by low thickness compared to the size of the working area of the receiver coil along the direction of flow, where the detection time, length of the receiver coil and flow section of the pipe are such that, during signal detection time at maximum flow rate of fluids, that layer does not fall outside the working area of the said coil. ^ EFFECT: design of a simple and accurate method and device for measuring flow rate of a fluid in a pipe operating in a wide range of velocities and phase composition of the fluid. ^ 8 cl, 10 dwg

Description

The invention relates to the field of flow measurement, in particular to methods for measuring the flow rate and / or flow rate of multiphase fluids, which are a finely dispersed or undispersed mixture of gas and a multicomponent liquid (for example, a mixture of gas, oil and water).

There are flow meters with NMR sensors (application US No. 20070164737, IPC G01V 3/00 .; declared 07.12.2006; publ. 07.19.2007), in which the flow rate is determined by the amplitude of the signal recorded by the NMR sensor resulting from the action on magnetic nuclei substances of a certain sequence of radio frequency pulses. In this case, the dependence of the amplitude of the NMR signal on the flow velocity is based on the effect of the dependence on the flow velocity of the magnitude of the magnetization of the nuclei, determined either by the time they spent in the zone of the external main or prepolarizing magnetic field, or in the area of the transmitter-receiver coil of the NMR.

The main disadvantage of flow meters based on this principle is the dependence of the amplitudes of the NMR signal, and therefore the calculated values of the flow rates and / or flow rate of the fluids, on the characteristics of the spin-lattice relaxation of the nuclei on which the NMR signal is observed. As a result, such flowmeters have limited fields of application, since they require inputting data on the characteristics of the spin-lattice relaxation of the measured media into the calculation algorithm in advance and can hardly be used in cases of multiphase / multicomponent media in which each phase / component will have an individual time relaxation.

Known flowmeters (application US No. 20060020403, G01F 1/00 .; published on January 26, 2006) using a constant gradient of the magnetic field to encode information about the flow velocity in the phase of the NMR signal.

Also known is a flowmeter with an NMR sensor (US Pat. 6452390 USA, IPC G01F 1/716, G01F 1/704, G01V 3/00. No. 09/711914; claimed 15.11.2000; published 17.09.2002), in which the flow rate is determined by phase coding the frequency of the NMR signal due to the formation in the region of the location of the excitation / detection coil of the NMR signal of pulses of a linear magnetic field gradient oriented along the direction of flow.

The main drawback of the principle used in the aforementioned flowmeters is the difficulty of accurately determining the phase of the recorded NMR signal, as well as the ambiguity of its determination for cases when the value 2π is exceeded, which will be realized at high flow rates. In addition, the technical solutions proposed in the analogue do not allow establishing an unambiguous relationship between the value of the detected phase of the signal and the characteristics of the flow velocity for cases of an unknown velocity distribution profile, and, consequently, for cases of different phase flow velocities / components of a multiphase / multicomponent medium.

The closest technical solution is a flow meter (US patent No. 6046587, ЕВВ 47/10, G01F 1/716, 1/74, 1/704, G01R 33/44, 33/60, G01V 3/00, publ. 04.04.2000) in which a nuclear magnetic resonance (NMR) sensor is used to measure the flow velocity, and to determine the fraction of water (or oil) in the flowing liquid, either a second NMR sensor or a second sensor based on the phenomenon of electron paramagnetic resonance (EPR) is used. In this case, as in most flowmeters, the pipe with the flowing liquid first passes through the magnetic field of the pre-polarizing system for preliminary polarization of the fluid cores, and then through the main magnet with a uniform magnetic field, in which the NMR sensor itself, which contains the radio frequency a coil necessary for exciting and recording NMR signals. In fact, the aforementioned patent proposes solutions for several different methods for measuring the flow rate. Some of these methods are not fundamentally different from the solutions proposed in others and, including, in the above-mentioned patents. Thus, it is proposed to evaluate the fluid flow rate from a comparative analysis of the signal amplitudes obtained either from two NMR sensors spaced apart by a certain distance along the flow direction, or from an analysis of the NMR signal amplitude under conditions of a continuous sequence of saturating radio frequency pulses acting on the system. In the first case, the effect of the difference in the polarization values of the nuclei is used, which they reach by the time the liquid passes through the sensors due to the difference in the time of exposure to an external magnetic field. In the second case, the effect of the limited area of the exciting radio-frequency field is used (only inside the radio-frequency coil). This leads to the fact that the NMR signal from the part of the fluid that does not have time to leak out of the coil zone between the pulses is saturated at a sufficiently high repetition rate of the exciting RF pulses, while the other part of the fluid flowing back into the working area of the NMR sensor characterized by a large amplitude signal. Thus, with a successful selection of the parameters of the pulse sequence, the total amplitude of the recorded signal can be directly proportional to the flow rate. In the general case, it is necessary to have a calibration curve of the dependence of the amplitude of the NMR signal on the flow velocity, in which the characteristics of the nuclear magnetic relaxation of the fluid would be taken into account.

Closest to the claimed is the proposal in the prototype, which is aimed at obtaining direct information about the flow rate of each of the components and / or phases of the multicomponent fluid. This goal is achieved by obtaining a distribution function of fluid flow rates based on an analysis of the frequency spectrum of the NMR signal (echo signal). The frequency spectrum itself is determined from the data of the fast Fourier transform of the echo of the NMR signal recorded under the conditions of the presence of an external magnetic field gradient along the direction of the fluid flow after exposure to the spin system of at least two radio frequency pulses with a given time interval. Indeed, in this case, from the frequency spectrum of the spin echo signal obtained after two radio frequency pulses provided that it is detected during the linear gradient of the magnetic field along the direction of flow, the desired velocity spectrum is easily calculated for a known (specified) time of occurrence of the echo signal (time between the first radio frequency pulse and the moment of formation of the echo signal) and the known (specified) value of the magnetic field gradient. If in this case the conditions for the complete polarization of the nuclei in the pre-polarizing magnet were provided, then from the obtained velocity spectrum it is possible to calculate not only data on the distribution of velocities for different phases (gas and liquid), but also on their number. In particular, the flow rates for liquid and gas in the case when the liquid and gas are not mixed can be determined directly from the analysis of the distribution function of the flow velocities, which will be characterized by two maxima in the frequency spectrum, respectively, for liquid and gas.

However, the method described in the prototype also has disadvantages.

Part of the comments relates to those patent proposals which, in one way or another, are related to obtaining information on the flow velocity from the amplitude of the NMR signal based on either the effect of incomplete saturation of the signal due to the flow of liquid through the transmitting-receiving coil, or due to the effect of incomplete polarization of nuclei magnetic field. In both cases, flow rate information is indirect. In addition, with this method of measurement it is very difficult to obtain information about the distribution of speeds. Significant additional uncertainty in the measurement results will be introduced by a possible change in time of the characteristics of nuclear magnetic relaxation of the studied fluid. First of all, this concerns spin-lattice relaxation times, which strongly depend on the phase and component composition of the current fluid, which, in turn, under conditions of varying the composition of the current fluid, will lead to a change in time of both the degree of saturation of the NMR signal with radio frequency pulses and degrees of partial polarization (for the case of incomplete polarization) of nuclei in a prepolarizing magnetic system. It is these effects in this patent, as well as in several other application and patent materials (for example, see application US No. 20070164737, IPC G01V 3/00), it is proposed to use to determine the characteristics of the fluid flow rate. In other words, the discussed method can be successfully implemented only in those cases when the velocity distribution over the flow cross section is predetermined and fixed. In addition, the characteristics of nuclear magnetic relaxation (first of all, the spin-lattice relaxation time) should also be known and unchanged in time. It is obvious that in real conditions these requirements are not feasible. Indeed, the velocity distribution function will depend not only on the flow rate itself (at high speeds, a transition from laminar to turbulent flow is possible), but also on the composition of the liquid, the presence of gas in it, the distribution of gas in the liquid, etc. In addition, such an undesirable effect as the possible deposition of asphalt-resinous and paraffin compounds on the inner surface of the pipe cannot be neglected, which will lead to a change in shape and a decrease in the cross-sectional area of the flow of the current fluid. This will inevitably affect both the distribution of flow velocities and the velocity itself, which, in turn, will lead to errors in determining the velocity values in the discussed method. If we talk about the requirement that the characteristics of the nuclear magnetic relaxation of the current fluid remain unchanged, then the impossibility of this requirement in real conditions is also obvious. The fact is, for example, that the spin-lattice relaxation rate will be different not only for different components (oil, water) of the fluid, but also for different (liquid, gas) phases of it. In cases where the indicated components and phases are spatially separated (have a small contact area), the noted differences in the characteristics of relaxation times can even be used to obtain quantitative characteristics of the flow for each component and phase. However, in most cases this is not so. Gas molecules can be not only distributed in the liquid in the form of bubbles, but also partially dissolved in the liquid, and water can form a water-oil emulsion with oil. In both cases, both due to mutual dissolution and due to the small size of gas bubbles in a liquid or due to the small size of water (or oil) droplets in a dispersion medium, conditions of large values of the specific area S / V (where S is the interface of the phases or components, and V is their volume), which will lead to the possibility of an intense molecular or spin exchange between the spin / molecular subsystems of the phases or components. As a result of this exchange, the nuclear relaxation characteristics will be averaged for a part of the fluid, and the average relaxation rate will depend on many uncontrolled parameters: partial relaxation times of the phases / components mentioned, mobility of the liquid molecules, which depends, inter alia, on the concentration of dissolved gas in the liquid ; the size and shape of the bubbles or drops; interfacial characteristics, etc. Under the conditions of operation of a multiphase flow meter in the oil industry, a mixture of oil, water and gas will act as a multiphase fluid. Obviously, the magnetic relaxation characteristics of the mentioned components of the mixture will depend not only on the field, the specific well, but will also change over time. Moreover, the dependence of the characteristics of the produced water-oil mixture on time is largely random in nature, not amenable to systematization and, consequently, to the meter’s work.

Thus, in the very proposed method of detecting the flow velocity by the dependence of the NMR signal amplitude on the time the nuclei are in the magnetic field, there is also its main disadvantage, since the magnetization of the nuclei reached by the time the signal is recorded, all other things being equal, strongly depends on nuclear relaxation characteristics of the current fluid. This drawback will be especially pronounced when measuring multiphase / multicomponent flows. Indeed, a system of a pre-polarizing magnetic field, by definition, should provide preliminary polarization of the nuclei before they enter the area of the NMR sensor. Moreover, it is desirable that the level of polarization be close to the maximum (equilibrium for the magnitude of the used magnetic field in the region of the NMR sensor) for all components and phases in the stream. Otherwise, different components and phases of the fluid will be polarized to varying degrees depending on the values of the corresponding spin-lattice relaxation times T ₁ : those components and phases that will have the lowest values of T ₁ will be polarized the most. As a result, the relative contribution to the recorded signal from the components and phases of the fluid with large values of the times T ₁ will be underestimated as compared to the case when the pre-polarizing magnetic system ensures equilibrium nuclear polarization for all components and phases of the fluid. This will lead to errors in the calculation of the flow characteristics of the components and phases. Thus, the quality of the entire device ultimately depends on the performance of the pre-polarizing magnet — the accuracy of the measured characteristics. Moreover, this conclusion remains relevant regardless of the principle used in the NMR flowmeter for measuring flow velocity.

The most direct way, ensuring the achievement of a given (equilibrium for a given magnetic field strength) polarization value of the nuclei, involves increasing the residence time of the fluid in the field of this magnet to values of the order of 5T ₁ . In practice, the solution of this problem leads to a significant increase in the size of the pre-polarizing magnet, since the design of this part of the flowmeter is, as a rule, classical. This is a pipe with a fluid / gas flowing in it, placed in the gap between the poles of two magnets, the external fields of which are closed to each other by a massive structure of magnetically conductive material. For example, if you use one of the standard internal diameters of the pipeline 0.07 m and take the spin-lattice relaxation time T ₁ equal to 1 second, then it is easy to find that for a flow rate of 100 m ³ / day the length of the section of the pipeline in a magnetic field must be at least one and a half meters. Therefore, the dimensions of the pre-polarizing magnet should be no less than the specified value. Ultimately, this leads to a significant increase in the size and weight of the entire flow meter.

The following comments relate directly to the proposal of the patent, in which information on the flow rate is proposed to be obtained from the analysis of the frequency spectrum in the spin echo signal.

As the first critical observation, we note that in the proposed device and the scheme of exposure to radio frequency pulses, the amplitude of the spin echo signal will be underestimated, since during the time between the excitation of the NMR signal and the moment of registration of the echo signal, part of the fluid has time to leave the working area of the NMR sensor and does not contribute to the signal echo. Moreover, the higher the flow velocity, the more the amplitude of the echo signal will be underestimated. Since there is always a velocity distribution during fluid flow through the pipe, it is obvious that for the layers with different velocities the marked effect will also be different: it will be maximum for those fluid layers that are characterized by maximum values of the flow velocity. Thus, distortions in the recorded spectrum will manifest themselves, first of all, in the region of high velocities, which determine the main contribution to the flow rate. Therefore, the noted effect can lead to significant errors in determining the characteristics of the fluid flow in general, and the component with high flow rates in particular.

The next problem arises due to the fact that different components and different phases of the fluid, as a rule, will be characterized by different values of spin-spin (T ₂ ) relaxation times and, in addition, by different values of self-diffusion coefficients. As a result, their relative contributions to the echo signal will depend both on the time elapsed from the moment the first exciting RF pulse was applied to the moment the echo was recorded and on the parameters of the magnetic field gradient, which must be used for frequency coding of the velocities of the components and phases of the fluid. Moreover, when registering NMR signals in a fluid flow under the presence of a magnetic field gradient, the effects of the so-called induced diffusion cannot be ignored - the appearance of an additional distribution of molecular displacements to self-diffusion due to the velocity distribution.

Another important problem is that, strictly speaking, the mathematical operation of the Fourier transform is incorrect in cases where the signal is digitized in the time domain under conditions of a variable signal frequency. During the flow of fluid along the direction of the magnetic field gradient, this non-standard condition is realized. Moreover, the precession frequency of the spins of the nuclei present will change in time as quickly as the position of the nuclei along the axis of the magnetic field gradient will change due to the flow velocity. This effect is, in fact, a source of information about the flow rate.

The technical task of the invention is the creation of a simple and accurate method and device for measuring the flow rate of a fluid in a pipeline operating in a wide range of velocities and phase compositions of the fluid.

The technical problem is solved by a method of measuring the flow rate of a multiphase fluid by recording a nuclear magnetic resonance (NMR) signal under the action of a magnetic field gradient directed along the direction of the flow, and subsequent conversion of the measured signal into the frequency spectrum.

What is new is that in order to increase the accuracy of the measurement results and minimize the effect of distorting the signal amplitude caused by the flowing in / out fluid from the working zone of the receiving coil, the initial excitation of the nuclei is carried out only in the part of the receiving coil - in the volume of the layer oriented across the flow and characterized by a small thickness in comparison with the size of the working zone of the receiving coil along the direction of flow, while the moment of time of registration of the signal is chosen so that for a period of time from the moment ervonachalnogo nuclei excitation until the registration signal at a maximum flow rate for all the phases of the fluid layer is located within the working area of the coil, which is determined by the length of the coil.

It is also new that the zone of the initial excitation of nuclei in the form of a layer is formed at the beginning of the receiving coil - from the side of the incoming fluid flow, due to the mutual arrangement of the receiving and gradient (or transmitting) coils.

Also new is the fact that in order to increase the accuracy of measurement results and minimize the effect of incorrectness of the Fourier transform operation under conditions of variable frequency, the time of registration of the echo signal is minimized.

It is also new that with the help of a sequence of radio-frequency pulses two or more echo signals are generated at different time instants, which are recorded in the presence of magnetic field gradients under the most identical conditions, then the Fourier transform is performed for these signals and two or more frequency spectra are obtained, of which, according to the difference in the frequencies of the corresponding spectral lines, that is, according to the dependence of their frequency on the time of the formation of the echo signal, one judges the velocities of the components of the flow.

Also new is the fact that in order to obtain information on the dependences of the frequency and amplitude on time for individual signal components and to use this information in the final calculations of the characteristics of the current fluid, the frequency spectra obtained for different times of the formation of the echo signal are subjected to two-dimensional analysis, from which the dependences frequencies versus time find velocities, by the dependences of the amplitudes on time — additional characteristics of the composition and properties of the fluid, and by extrapolated to zero values amplitudes of - the relative proportions of the fluid phases and components with defined velocity values for them.

To solve the technical problem of the implementation of the proposed method, a device can be used that contains a pipeline for multiphase fluid flow, a housing with a magnetic system for polarizing the nuclei and a sensor operating on the NMR phenomenon, electronic gradient blocks, a transmitter and a receiving system; electronic system for managing, collecting and storing information; a processor for processing incoming information according to given algorithms, pressure and temperature sensors, as well as a pre-polarizing magnetic system.

What is new is that the pre-polarizing magnetic system is made in the form of parallel plates of magnetic material located in cross section inside the casing, simultaneously acting as a pipeline, and oriented along the fluid flow, and the magnetic field lines created by the inner layers are most effectively closed to the poles magnets of adjacent layers, creating a given distribution of the density of lines of force in the gaps between the layers.

Also new is the fact that in order to simplify and facilitate the design of the pre-polarizing magnetic system, the housing is made of a material with high magnetic permeability (for example, iron).

Also new is the fact that in order to increase the efficiency of the external magnetic circuit, the housing is additionally surrounded by wire or sheet material with the corresponding magnetic characteristics.

Figure 1 shows a diagram of the implementation of the method.

Figure 2 shows a diagram of a method using a gradient coil.

Figure 3 shows the frequency spectrum under conditions of a constant signal frequency (f ₀ = 1.59 MHz), where the signal registration time is 40 μs.

Figure 4 - frequency spectrum under conditions of variable signal frequency, where the signal registration time is 40 μs, with the initial frequency value f ₀ = 1.59 MHz.

Figure 5 - frequency spectrum under conditions of variable signal frequency, where the signal registration time is 20 μs, with the initial frequency value f ₀ = 1.59 MHz.

Figure 6 - frequency spectrum under conditions of variable signal frequency, where the signal registration time is 10 μs, with the initial value of the frequency f ₀ = 1.59 MHz.

Figure 7 shows the frequency spectrum under conditions of constant signal frequency (f ₀ = 1.59 MHz), where the signal registration time is 10 μs.

Figure 8 shows the frequency spectrum under conditions of variable signal frequency, where the signal registration time is 40 μs, with the initial frequency value f _t = 4f ₀ = 6.36 MHz.

Figure 9 shows a sequence of radio frequency pulses (rf is the channel of the transmitter), magnetic field gradient pulses g _v and NMR signals (SSI and echo).

Figure 10 shows a view of the spectra P (f, t _e ) obtained as a result of the Fourier transform of the signals echo1, echo2, echo3 and echo4 for the times of their formation t ₁ , t ₂ , t ₃ and t ₄ (see Fig. 9 )

A device for implementing the method comprises a pre-polarizing magnetic system with a housing 1 and magnets 2 for polarizing the nuclei, a main magnetic system 3 with magnets 4 and a sensor (receiving / transmitting coil) 5 (FIGS. 1 and 2) operating on the NMR phenomenon, electronic gradient blocks (not shown), transmitter and receiver system (not shown); electronic control system (not shown), collection and storage of information (not shown); a processor (not shown in FIG.) for processing incoming information according to predetermined algorithms, pressure and temperature sensors (not shown in the drawing), as well as a gradient coil 6 (FIGS. 1 and 2) and a pipeline 7 (FIGS. 1 and 2), passing through the NMR sensor. When this pipeline 7 is made of dielectric. The magnets in the pre-polarizing magnetic system (Fig. 1) are made in the form of parallel plates 2 of magnetic material (magnets - see View B) located in cross section inside the housing 1 and oriented parallel to the fluid flow, and magnetic field lines generated by the inner layers , most effectively close to the poles of the magnets of adjacent layers, creating a given distribution of the density of lines of force in the gaps between the layers. The direction of the magnetic field in the pre-polarizing magnetic system is oriented in the direction coinciding with the direction of the main magnetic field created by the magnets 4 in the main magnetic system 3. The device also contains secondary elements 10 and 11, which are used for tight connection of the pipe 12 with the housing 1 of the pre-polarizing magnetic system and the pipe 7 of a dielectric, which, in the General case, have different cross sections. The sensor 5 (Figs. 1 and 2), made in the form of a receiving coil, is selected such a length l _k , and the pipeline 7 (Fig. 2) with such a cross-section S _k so that during the registration of the signal d _s this layer of the initial excitation of nuclei d _s (Figs. 1 and 2) after magnetization, which is performed in the part of the receiving coil 5, was within the limits of this coil 5.

In order to simplify and facilitate the design of the pre-polarizing magnetic system, the housing 1 of the pre-polarizing magnetic system is made of a material with high magnetic permeability (for example, iron) and acts as an external magnetic circuit. In this case, in order to increase the efficiency of the external magnetic circuit, the housing 1 can additionally be surrounded by wire or sheet material (not shown) with the corresponding magnetic characteristics.

The parallel plates 2 of the pre-polarizing magnetic system are installed in the housing 1 with the possibility of changing their number and the gap between them and / or replacing them with magnetic plates of the same shape, but made of magnetic material, characterized by a higher surface density of magnetic field lines compared to replaced magnetic material plates.

In order to use the method of direct registration of the spectrum of fluid flow rates proposed in the prototype, by applying the fast Fourier transform procedure to the NMR signal, it is necessary:

a) to register the signal in the presence of a gradient g _{v of a} magnetic field directed along the direction of fluid flow;

b) the signal must be recorded after a certain predetermined time t _e after the spin system is excited so that during this time the fluid molecules move along the direction of flow at a distance

where ν is the desired fluid flow rate.

Then, the dependence f (ν) of the frequency of the recorded signal on speed for the case of a linear magnetic field gradient has the form

where γ is the gyromagnetic ratio of resonating nuclei,

H ₀ - the value of the external magnetic field in the area of the NMR sensor,

C is a constant associated with signal frequency transformations in the electronic blocks of the receiving channel of the device and in the processors.

Strictly speaking, to determine the flow velocity, it is necessary to know not the frequency of the NMR signal itself, but its shift (change) in comparison with the frequency f (0) of the signal detected either at zero flow velocity (ν = 0) or at a zero magnetic field gradient ( g _v = 0), or at zero time (t _e = 0). Then, for the frequency shift, we can expect the fulfillment of a simple relation

If the current fluid is characterized by a velocity spectrum or contains phases with significantly different values of flow velocities, then in these cases the corresponding form of the frequency spectrum will be observed, which is uniquely associated with the velocity spectrum. Moreover, if we take into account the effects of nuclear relaxation, then the integrated intensities for the lines in the frequency spectrum will reflect the spin densities of the current components and phases of the fluid.

So, we consider the above from the point of view of the possibility of implementation in practice. As we noted above, the first problem is the distortion of the frequency spectrum of the signal due to the fact that during the time t _e , a shift in the position of the fluid molecules by a distance l _e = t _e · ν must be fixed, which is directly proportional to the flow velocity ν. As a result, for the indicated time, part of the already polarized and capable of giving a signal echo fluid will leave the working area of the NMR sensor and thereby reduce the signal amplitude. The larger the value of ν, the greater the effect of underestimating the signal amplitude, and in the presence of a velocity distribution, the more the shape of the recorded frequency spectrum and, therefore, the shape of the speed spectrum calculated from it will be distorted.

Our proposal (see FIGS. 1 and 2) is to initially excite not only the entire fluid located in the receiving coil region of the NMR sensor with a radio frequency pulse, but only part of it (layer 8). Then, for any given flow velocity ν, the time interval t _e , defined as the time between the excitation of layer 8 and the registration of the echo signal, can be chosen so that, by the time t _{e, the} excited nuclei are in position (layer 9) that does not go beyond the working zones of the receiving coil of the NMR sensor. With the known maximum speed of the phases of the fluid flow, determined from the operating conditions of the pipeline, and the minimum optimal time t _e determined by the characteristics of the electronic components of the device (transmitter, receiver, magnetic field gradient block, etc.) and the requirements for the signal recording time, the parameters are selected the cross section S _k of the pipeline body 7 and the length l _k (Fig. 2) of the receiving coil 5 of the NMR sensor.

It seems most optimal to choose an excitation zone (layer) 8 (FIGS. 1 and 2) with a thickness d _s (FIG. 2) so that it appears in the initial (in the direction of fluid flow) region of the receiving coil 5. Then, with the length of the receiving coil , for example, l _k and flow velocity ν, the position of the echo signal can be selected based on the condition

In particular, from the same inequality it is clearly evident that under the condition l _k = d _{s it is} generally impossible to establish a value for t _e that is not equal to zero.

The task can be achieved in several ways. One of them assumes the presence, for example, of two RF coils: one transmitting (not shown) and the other receiving. In this case, the position of the excitation zone will be determined by the position of the transmitting coil relative to the receiving one.

Since the device already contains a gradient coil 6 (Figs. 1 and 2), another solution based on the effect of selective spin excitation under conditions when the radio frequency excitation pulse and the magnetic field gradient simultaneously act on the spin system is more preferable. Such a combination of effects in NMR imaging is known as selective radio frequency pulse. In this case, the task of forming the zone (layer) 8 (FIGS. 1 and 2) of excitation with a thickness d _s (FIG. 2) so that it appears in the initial (in the direction of fluid flow) region of the receiving coil 5, is solved by a corresponding displacement gradient coils 6 relative to the receiving / transmitting. Then, when choosing the time interval between the moments of excitation and registration of the signal satisfying the condition t _e ≤ (l _k -d _s ) / ν, the excitation zone 8 will move to position 9, but will not go beyond the working area of the coil 5 and, therefore, will be provided registration of an undistorted NMR signal. This will make it possible to obtain undistorted information about the velocity and velocity distribution of the current fluid.

As is known, in order to achieve equilibrium polarization of nuclei, one condition must be met: the current fluid must be in the magnetic field of the pre-polarizing magnet for a time

where T ₁ is the time of spin-lattice relaxation.

For fixed sizes of the pre-polarizing magnet, usually the time t _pol of the fluid in the magnetic field is increased by shaping the pipe in the shape of a coil, as is done, for example, in the prototype. In this case, the residence time of the fluid in the magnetic field will be determined not only by the size (l _II ) of the magnet in the direction of flow, but by the transverse size (l _⊥ ), which for a given external diameter (D) of the pipeline will determine the number (n) of elbows on the coil:

n = l _⊥ / D

Then, for a given average flow velocity ν, the residence time of the fluid in a magnetic field will be determined as t _pol = n · l _II / ν. The disadvantage of this design is the need to create a pre-polarizing magnetic system with a large cross-sectional area of the magnetic field, which, together with the necessary external magnetic circuit, leads to a heavier structure and inefficient use of magnetic material.

To create a more efficient and compact design of the pre-polarizing magnetic system, we propose to arrange the magnetic material in layers inside the housing 1 in the form of plates 2, as shown in Fig. 1, while using an enlarged flow area of the housing 1, which simultaneously serves as a pipeline for the current fluid. In this case, the part of the cross section occupied by the magnetic material itself can be made relatively small by using a thin (approximately 1.6-3.2 mm) magnetic material so that the resulting flow section between the plates remains large enough. The distance between the layers of magnetic material will be determined by the necessary value of the magnetic field strength, however, in any case, it will be significantly greater than the thickness of the magnetic material. According to our calculations, when using high-quality magnetic materials of rare-earth metal plates 2 in the proposed design, it is possible to achieve conditions when the total cross section of the magnetic plates will be no more than 10-20% of the initial internal cross-sectional area of the housing 1. Thus, if for example, the housing 1 is made in the form of a pipe segment, then an increase in its diameter, for example, only 3 times compared with the diameter of the main pipeline 12 (Fig. 1) will lead to a decrease in the flow rate fluid in it no less than 7 times. In this case, the necessary value of the magnetic field strength can be controlled by the number and, therefore, the distance between the layers of sequentially arranged magnets inside the pipeline due to the installation and removal of additional plates. In the proposed design, the fluid flow is carried out in the space between the magnetic plates, the poles of which are oriented so that the magnetic field lines created by the poles of the internal magnetic plates effectively close to opposite poles of the magnets of the neighboring plates, thereby dividing the cross section of the pipeline into areas with sequentially connected magnetic fields and creating a given distribution of the density of lines of force in the gaps between the layers. The effectiveness of the proposed design lies in the fact that an external magnetic circuit is required only for closing the magnetic field lines of the extreme magnetic plates. In this case, the housing 1, made of iron, partially can itself perform the function of an external magnetic circuit. If necessary, the housing 1 in this part may contain an additional external magnetic circuit, for example, in the form of a winding sheet or wire material with corresponding magnetic characteristics. The length and passage section of the pipeline segment (body 1) with the magnetic plates 2 located inside it are chosen so that, for given maximum values of the flow velocity and spin-lattice relaxation times, the conditions for the complete polarization of the nuclei in the pre-polarizing magnet are ensured during the passage of the fluid through it.

It is possible to effectively reduce the time required to achieve the polarization value of the nuclei M ₀ , which is equilibrium for the magnetic field strength H ₀ , by placing the nuclei for some time in a magnetic field with an intensity greater than H ₀ . So, for example, if in a field with intensity H _pol ≈ Н ₀ the polarization value M = 0.9 · M ₀ , equal to 90% of the equilibrium value, is reached in time t≈2.3 · T ₁ , then in a field with intensity H _pol ≈ 2 · N ₀ for the same purpose it will take four times less time. The need for such accelerated polarization of nuclei may arise in cases of high flow rates. The creation in a certain part of the pre-polarizing magnet of an increased magnetic field value compared to H ₀ can be easily achieved by, for example, adding additional plates of magnetic material between existing plates and / or replacing them with magnetic plates of the same shape, but made of magnetic material, characterized by a higher surface density of magnetic lines of force compared with the magnetic material of the replaced plates.

Now consider the problem associated with the fact that, as already mentioned, the mathematical operation of the Fourier transform is incorrect in cases where the signal is digitized in the time domain under conditions of a variable signal frequency. This circumstance is clearly demonstrated in Fig.3-8.

связать с частотой, соответствующей середине (примерно 3 МГц) спектра, показанного на фиг.4.Figure 3 shows the shape of the frequency spectrum obtained as a result of the Fourier transform of a certain model signal recorded in the time domain under conditions of a constant signal frequency f ₀ = 1.59 MHz. The signal recording time was 40 μs. Figure 4 shows the result of the Fourier transform for a signal that differs from the previous one, only in that now the frequency of the signal linearly changed in time (which corresponds to the conditions of a stationary fluid flow along the direction of a linear magnetic field gradient). Hereinafter, the rate of change of the signal frequency was set to be the same and equal to 1.59 MHz for 50 μs. Moreover, at the beginning of the registration of the signal, its frequency corresponded to the value f ₀ = 1.59 MHz (with the exception of the situation shown in Fig. 8). All other conditions were unchanged. From a comparison of the spectra in FIGS. 3 and 4, it is clearly seen how much the result of the Fourier transform is distorted in case of inconstancy of the frequency of the recorded signal. Moreover, the shown distortions of the spectrum cannot be formalized by any simple relations that could, for example, average MHz during the measurement time (40 μs) of the signal frequency

associate with the frequency corresponding to the middle (approximately 3 MHz) of the spectrum shown in figure 4.

The results of the Fourier transform of a signal with a frequency that depends linearly on time are clearly improved if the measurement time is reduced. This is well demonstrated by comparing the spectra shown in FIGS. 4, 5 and 6, which were obtained under conditions of a sequential decrease in the recording time: 40, 20, and 10 μs, respectively.

The spectrum shown in FIG. 7 corresponds to a situation with a time-fixed signal frequency. Both spectra in FIG. 6 and FIG. 7 were obtained under the same conditions with respect to the signal recording time, which was so small (10 μs), that during this time the signal frequency in the case presented in FIG. 6 changed from the original total twenty%.

In order to eliminate errors associated with the incorrect execution of the mathematical operation of the Fourier transform of signals with a time-dependent frequency, our method proposes to perform several measurements of the NMR signal of the current fluid with specified time intervals between them so that in this case all other conditions (measurement duration, amplitude magnetic field gradient, flow rate) signal registration were as similar as possible. The similarity of the above conditions is easy to fulfill for the stationary flow condition, since the amplitude of the magnetic field gradient and the duration of the measurement are set programmatically in the device control unit. The condition of stationarity of the fluid flow in the pipeline in the general case, of course, is not feasible, however, when choosing small values for the time intervals between the mentioned measurements, it will always be possible to talk about the fulfillment of the quasistationary conditions (during the measurement time of the necessary characteristics) of the flow process.

Then, by analyzing the spectra obtained in the above-described way for different signal recording times, it will be easy to find the frequency shift between them, from which, with known parameters of the magnetic field gradient and the time interval between measurements, the flow velocity can be calculated. This is well demonstrated by comparing the spectra shown in FIG. 4 and FIG. 8.

The spectrum shown in Fig. 8 was obtained for a signal that, in contrast to the situation shown in Fig. 4, was shifted upward in frequency by 4.77 MHz. As can be seen from a comparison of the mentioned spectra, their appearance is strongly distorted by the effect of frequency inconsistency, but these distortions are approximately the same, since all conditions for recording the signal were identical and in both cases the signal frequency increases by the same value of 1.27 MHz during the recording time. It is easy to see that in this case the main information encoded in the relative shift of the spectra under discussion relative to each other is not distorted and corresponds to the value 4.77 MHz specified in the model calculations.

So, to implement the above proposal, it is necessary to use such a pulse sequence that would allow at least two echo signals to be generated and thereby make two measurements under the most identical conditions, differing only in time. In this case, for each of these measurements, by means of a fast Fourier transform, frequency spectra are obtained that will be characterized by the same distortions associated with the effect of inconstancy of the signal frequency during the measurement, but by different frequency shifts due to different times of registration of the echo signal. From a comparison of the obtained frequency spectra using the appropriate processing algorithms in the device’s processor, the frequency shift spectrum will be calculated, which, in turn, according to the well-known linear law, is associated with the desired velocity spectrum of the studied fluid.

Figure 9 shows the sequence of radio frequency pulses acting on the spin system (rf is the channel of the transmitter): the first pulse is 90 °, the remaining -180 °. The first 90 ° pulse is selective, since it is formed against the background of the action of the magnetic field gradient g _ν . The shape of the gradient pulse is such that after the formation of the main part with a positive gradient sign, a reverse sign gradient is formed. The duration of the gradient with a negative sign is selected from the condition for compensation of the phase dispersion of the resonating nuclei resulting from the action of the main gradient. In other words, the parameters of the gradient pulses (mentioned and all subsequent ones) are chosen so that their integral values are equal to zero.

If the first gradient pulse is necessary so that the 90 ° radiofrequency pulse has a selective effect on the resonating nuclei in zone 8 (Figs. 1 and 2), then the remaining gradient pulses provide frequency coding of the position of the nuclei at the moments of formation of the echo1 signals (Fig. 9 ), echo2, echo3, etc. The nuclear coordinates at these times will be shifted from the initial ones along the flow direction by an amount proportional to the velocity ν and elapsed from the moment of excitation by the first radio-frequency time pulse - t ₁ , t ₂ , t ₃ , ... for echo1, echo2, echo3, ... signals, respectively.

Each new position of the resonating nuclei at the moments of the echo formation can be detected by digitizing the signal under the conditions of a magnetic field gradient directed along the flow. As a result of the subsequent Fourier transform for each time t _e = t ₁ , t ₂ , t ₃ , ... the echo signal is generated, the frequency spectrum P (f, t _e ) is obtained, which in the stationary mode will be linearly related to the velocity spectrum K (ν) :

The function R (ν, t _e ) in this expression reflects the fact that, with increasing time t _e, the NMR signal is subject to additional attenuation due to spin-spin relaxation, misphasing in an inhomogeneous magnetic field, misphasing due to self-diffusion and induced diffusion. In the general case, these reasons for the decrease in the amplitude of the NMR signal with increasing t _e cannot be factorized, since their characteristics (relaxation times, self-diffusion coefficients, etc.) can differ for fluid components characterized by different velocities ν. The only time at which the unknown function R (ν, t _e ) does not distort the frequency spectrum is the zero time t _e = t ₀ = 0, and only then can the frequency spectrum P (f, t _e ) be considered normalized to unity. Consequently, a complete restoration of the shape of the spectrum K (ν) is possible by a comparative analysis of the frequency spectra R (f, t _e ) obtained for different times t _e and their extrapolation to zero time. All these reasons are described by decreasing functions of the dependence of the amplitude of the echo signal on time. Moreover, exponential functions are often used as such functions. This allows the function R (v, t _e ) to be represented as a first approximation in the form of a certain distribution k (ν) of exponential functions of the type exp (-t _e / T _j (ν)), that is, R (ν, t _e ) ≈k ( ν) · exp (-t _e / T _j (ν)), where T _j (ν) can act as an additional characteristic for the fluid component flowing with velocity ν. For example, this can be the spin-spin relaxation time T _j (ν) = T ₂ (ν) or a factor associated with the coefficient of self diffusion T _j (ν) ∝1 / D (ν). The establishment of such a relationship between the velocity characteristics and relaxation times and / or self-diffusion coefficients of complex fluids allows us to more clearly solve the problem of correlating the obtained experimental data on the velocities in the flow with certain components and phases of the fluid.

Figure 10 shows one of the possible variants of the spectra R (f, t _e ) obtained as a result of the Fourier transform of the signals echo1, echo2, echo3 and echo4 for the times of their formation t ₁ , t ₂ , t ₃ and t ₄ (Fig .10). The time t ₀ (Fig. 10) corresponds to the zero time for which the extrapolation of the values of the distribution density P (f, t _e ) is performed. Figure 10 shows that the spectra P (f, t _e ) with increasing time t _e become wider and shift to higher frequencies. (Note that when the sign of the magnetic field gradients is inverted, the opposite picture will take place: the spectra P (f, t _e ) will shift toward lower frequencies with increasing time t _e ). Moreover, in this example, the velocity distribution can be described bimodal function: it is seen that in general the two components can be distinguished frequency spectrum which conventionally called the components a and b, described by the functions P _a (f, t _e) and P _b ( f, t _e ).

и

частот от времени t_e в спектрах P_a(f,t_e) и P_b(f,t_e) соответственно. Пользуясь простыми соотношениями вида:The dashed lines 1 _a , 3 _a and 1 _b , 3 _b show the trajectories of the extreme (f _min and f _max ) frequencies as functions of time t _e in the spectra P _a (f, t _e ) and P _b (f, t _e ), respectively . Lines 2 _a and 2 _b denote the paths of changes in means

and

frequencies versus time t _e in the spectra of P _a (f, t _e ) and P _b (f, t _e ), respectively. Using simple relationships of the form:

и

для наблюдаемых компонент флюида а и b.easy to calculate average speeds

and

for the observed fluid components a and b.

и

. Таким образом, информация о скорости (фиг.10) вычисляется, по сути, из угла наклона к оси времени прямых, описывающих траекторию изменения частоты от времени t_e. Информация об относительной доле в общем потоке компоненты с определенной таким образом скоростью v находится из вклада этой компоненты в нормированную на единицу функцию P(f,t₀), полученную путем экстраполяции на нулевое время экспериментально измеряемых функций P(f,t_e).On the frequency axis at the time t _e = t ₀ = 0 shows the spectra of P _a (f, t _o) and _{P b (f, t o)} , as a result of extrapolation to zero time experimentally measured distributions P _a (f, t _e) and P _b (f, t _e ). Under the normalization condition P (f, t ₀ ) = P _a (f, t ₀ ) + P _b (f, t ₀ ) per unit, the integral values of P _a (f, t ₀ ) and P _b (f, t ₀ ) will be reflect, respectively, relative fluid fractions characterized by average velocities in the flow

and

. Thus, the information about the speed (figure 10) is calculated, in fact, from the angle of inclination to the time axis of the lines describing the path of the frequency from time t _e . Information on the relative share in the total component flux with the speed v determined in this way is found from the contribution of this component to the function P (f, t ₀ ) normalized to unity, obtained by extrapolating the experimentally measured functions P (f, t _e ) to zero time.

Thus, the method of obtaining information on the flow rates of the components and the phases of the fluid includes measuring at least two echo signals under the influence of a magnetic field gradient and Fourier transforming these measurements into frequency spectra, followed by a two-dimensional analysis of the frequency spectra in order to obtain information about the dependencies the frequencies and amplitudes of the signal components versus time, from which the velocities are found from the dependences of the frequency versus time, the NMR characteristics (relaxation times) from the dependences of the amplitudes versus time, and by extra olirovannym at time zero amplitude values - the relative proportions of the fluid phases and components with defined velocity values. Based on the data obtained from the NMR sensor, including data on the gas / liquid ratio, as well as additional data received from another sensor (for example, an EPR sensor (not shown in FIGS. 1, 2)), the content of oil in the current fluid is calculated according to the given algorithms, the complete distribution function of the velocities in the flow over the components and phases of the fluid. Based on the obtained velocity distribution function, all the necessary flow characteristics are calculated: instantaneous and average for the given period of time flow rates of the components and fluid phases, as well as instantaneous and average values for their relative fractions. To recalculate the proportion of gas in the current fluid to normal conditions, the temperature and pressure sensors (not shown in Figs. 1, 2), which are part of the device, are used.

The proposed method and device for its implementation are simple, accurate and have the ability to work in a wide range of speeds, including maximum and various phase combinations of the method and device for measuring the fluid flow rate in the pipeline.

Claims (8)

1. The method of measuring the flow rate of a multiphase fluid by recording a nuclear magnetic resonance (NMR) signal under the action of a magnetic field gradient directed along the direction of flow, and subsequent conversion of the measured signal into the frequency spectrum, characterized in that, in order to improve the accuracy of the measurement results and minimizing the distorting amplitude of the signal, the effect of flowing in / flowing out fluid from the working area of the receiving coil, the initial excitation of the nuclei is carried out only in the receiving part coils - in the volume of the layer oriented across the flow and characterized by a small thickness compared to the size of the working zone of the receiving coil along the direction of flow, while the recording time, the length of the receiving coil and the bore of the pipeline are selected so that during the registration of the signal at the maximum speed of the flow phases fluid, this layer did not go beyond the working zone of this coil. 2. The method according to claim 1, characterized in that the zone of initial excitation of the nuclei in the form of a layer is formed at the beginning of the receiving coil - from the side of the incoming fluid flow, due to the relative position of the receiving and gradient (or transmitting) coils. 3. The method according to claim 1 or 2, characterized in that, in order to improve the accuracy of the measurement results and minimize the effect of incorrect Fourier transform operations under conditions of variable frequency, the time of registration of the echo signal is minimized. 4. The method according to claim 1 or 2, characterized in that, under the conditions of a magnetic field gradient, two or more echo signals are generated at different time points, which are recorded under the same conditions in order to improve the accuracy of the measurement results, then Fourier is performed for these signals -transformation and judge the velocities of the components of the flow by the difference in the frequency shifts of the spectral lines, that is, by the dependence of the frequency on the time of formation of the echo signal. 5. The method according to claim 1 or 2, characterized in that, in order to obtain information on the dependences of the frequency and amplitude on time for individual components of the signal and use this information in the final calculations of the characteristics of the current fluid, obtained for different times of the formation of the signal echo frequency spectra subjected to a two-dimensional analysis, from which, according to the dependences of the frequency on time, the velocities are found, on the dependences of the amplitudes on time, additional characteristics of the composition and properties of the fluid are found, and from extrapolated to zero time I amplitude value - the relative proportion of components and phases of the fluid at a certain rate values. 6. A device for implementing the method, comprising a pipeline for multiphase fluid flow, a housing with a magnetic system for polarizing nuclei and a sensor operating on the NMR phenomenon, electronic gradient units, a transmitter and a receiving system; electronic system for managing, collecting and storing information; a processor for processing incoming information according to predetermined algorithms, pressure and temperature sensors, as well as a pre-polarizing magnetic system, characterized in that the pre-polarizing magnetic system is made in the form of parallel plates of magnetic material located in a cross section inside the housing, which simultaneously acts as a pipeline, and oriented along the fluid flow, and the magnetic field lines created by the inner layers are most effectively closed at the poles of the magnets of adjacent layers in creating a given distribution of lines of force density in the gaps between the layers. 7. The device according to claim 6, characterized in that, in order to simplify and facilitate the design of the pre-polarizing magnetic system, the housing is made of a material with high magnetic permeability (for example, iron). 8. The device according to claim 6, characterized in that, in order to increase the efficiency of the external magnetic circuit, the housing is additionally surrounded by wire or sheet material with the corresponding magnetic characteristics.

Images