Non-invasive device and method for monitoring entrainment of solid particles and / or moisture droplets

Abstract

The invention relates to the field of oil and gas production. A non-invasive device for monitoring the entrainment of solid particles and / or moisture droplets in oil and gas production systems or multiphase fluid flows comprises, connected to one another in series, a measurement module, a computing module, and a calibration and self-test module. The measurement module comprises at least two sensors mounted on a pipe of a production system and configured to be capable of converting, in a synchronized manner, ultrasonic vibrations of the outer surface of said pipe into electrical signals for the computing module. The computing module is equipped with a neural network-based machine learning system for counting and classifying particles and / or moisture droplets on the basis of said electrical signals, classified as background signals, and / or solid particle impact, and / or moisture droplet impact, in a given period of time. This results in more accurate and reliable measurements.

Description

[0001] IPC E21B47 / 10

[0002] G01F 1 / 74

[0003] A non-invasive device and method for recording the removal of solid particles and / or droplet moisture in oil, gas or multiphase fluid flow systems

[0004] Field of technology

[0005] The group of inventions relates to the field of gas and oil production industry and can be used to measure the flow rates and quantities of various components of a suspended gas and multiphase oil-saturated flow under operating conditions of gas and oil wells.

[0006] State of the art

[0007] A prior art method for monitoring changes in the flow rate of solid inclusions and condensed moisture in a gas flow in a pipeline is known, implemented in a device for a similar purpose (RU 2154162 C2, priority date July 27, 1998). The known method involves receiving and converting acoustic signals carrying information about the flow rate of solid inclusions and condensed moisture in the gas flow using a piezoelectric sensor, followed by filtering and detecting the output signals from the piezoelectric sensor and then digitizing them using an analog-to-digital converter.

[0008] A method is known for monitoring changes in the flow rate of solid inclusions and condensed moisture in a gas flow in a pipeline (RU 2389002 C2, priority date 12.05.2008), which consists of receiving and converting acoustic signals proportional to the flow rate of solid inclusions and condensed moisture in a gas flow using a piezoelectric sensor, followed by filtering and detecting the output signals from the piezoelectric sensor and their subsequent digitalization using an analog-to-digital converter.

[0009] A known method for recording the removal of solid fractions in a gas flow is based on the reception of acoustic signals by a piezoelectric sensor (RU 2783082 C1, priority date 02.08.2021) installed on the outer surface of a pipe from collisions of solid fractions with the inner surface, which consists in the fact that high frequencies are extracted from the received signal with a boundary corresponding to ultrasonic waves capable of propagating in the walls of pipes with a given range of wall thicknesses, then quadratic detection is performed, then low-frequency industrial noise and the original carrier frequencies are excluded from the detection result, then the obtained amplitude modulation signal is detected and measured, then the measurement result is declared proportional to the mass velocity of the removal of solid fractions.

[0010] The following devices are also known from the prior art: DSP-AKE-2 from Sigma-Optic JSC (Development of Oil and Gas Fields, Oil Business, 2018, Vol. 16, No. 3, DOI: 10.17122 / ngdelo-2018-3-22-32); DSP Detector from ClampOn (available at the link:

[0011]

[0012] device SAM 400 TC from Roxar (available at: http / www.emerson.co m / en - gb / cataioq / roxar-sam-acoustic-sand-monitor-en-qb).

[0013] An analysis of the current state of the art reveals that existing technical solutions have a number of operational and functional limitations. Specifically, the disadvantages of these devices and methods include their inability to be used in oil wells with liquid fluids, low sensitivity to detecting solid particles with low impact energy, which is typical for particles moving in liquid fluids (oil and water mixtures), and low-velocity solid particles. Other disadvantages of these methods include the need for pre-calibration of the systems for the specific well design (pipe diameter and thickness), product composition, and flow rates. The use of additional frequency filters significantly limits the ability of digital signal filtering to remove industrial noise.

[0014] Disclosure of invention

[0015] The general objective of the group of inventions is to create a new device and method for recording the removal of solid particles and / or droplet moisture in oil, gas or multiphase fluid flow production systems, which are capable of recording solid particles and / or droplet moisture with low impact energy, as well as solid particles and / or droplet moisture moving at low speed.

[0016] The general technical result of the claimed group of inventions is an increase in the accuracy and reliability of measurements while simultaneously increasing ease of use.

[0017] The stated task and the required technical result when using the group of inventions are achieved through a new non-invasive device for recording solid particles and / or droplet moisture in oil, gas or multiphase fluid flow production systems, including a measuring module, a computing module and a calibration and self-testing module connected in series with each other, where the measuring module includes at least two sensors intended for installation on a pipe of the production system and configured to synchronize the conversion of ultrasonic vibrations of the outer surface of the said pipe into electrical signals for the computing module, which includes at least a processor,equipped with a machine learning system based on neural networks for counting and distributing particles and / or droplet moisture based on the said electrical signals into background and / or impact of a solid particle and / or impact of droplet moisture in a given time region,

[0018] In a particular embodiment of the device, the measuring module is equipped with a high-pass filter, an amplifier, a low-pass filter and an analog-to-digital converter connected in series with each other.

[0019] In another particular embodiment of the device, the computing module is provided with memory, a power supply system and a modem.

[0020] In another particular embodiment of the device, the calibration module is provided with a digital analog converter, a buffer, and an emitter connected in series with each other.

[0021] The stated objective and the required technical result when using a group of inventions are achieved through a new method for recording the removal of solid particles and / or droplet moisture in oil, gas or multiphase fluid flow production systems, which includes the following steps:

[0022] recording acoustic signals from collisions of solid particles and / or droplet moisture with the inner surface of a pipe of a production system by means of at least two sensors located on said pipe and configured to synchronize the conversion of ultrasonic vibrations of the outer surface of said pipe into electrical signals;

[0023] amplification of received signals;

[0024] digitalization of signals over a wide frequency range, in which the signal sampling over time has a continuous duration of at least 1 μs and where each sample is subjected to digital processing for subsequent signal counting and distribution by a machine learning system based on neural networks;

[0025] processing of pre-digitized signals by a machine learning system based on neural networks with distribution into background and / or impact of a solid particle and / or impact of droplet moisture and subsequent counting of the number of recorded signals in a given unit of time;

[0026] calibration of the obtained values.

[0027] A distinctive feature of this group of inventions is a novel device design and recording method, which utilize at least two sensors connected by a common circuit board for highly accurate data synchronization, followed by analysis in a specified time domain using neural networks and an adapted machine learning system. The device design and recording method also allow for separate adjustment of the sensor sensitivity. Using highly accurate data synchronization and neural networks for time-domain processing enables the detection of low-impact solid particles and / or droplet moisture, as well as low-velocity solid particles and / or droplet moisture, significantly improving measurement accuracy. This also effectively eliminates background noise (industrial interference and fluid flow noise in pipes), thereby increasing measurement reliability.At the same time, the absence of the need for preliminary calibration for a specific well design (diameter and thickness of the pipe), product composition and volumetric flow rates significantly increases the ease of operation.

[0028] Brief description of the drawings

[0029] The essence of the group of inventions is explained by drawings.

[0030] Fig. 1 shows a block diagram of a device for recording solid particles and / or droplet moisture.

[0031] Fig. 2 shows a longitudinal section of sensor 1.

[0032] Fig. 3 shows a schematic version of the installation of sensors 1 on a pipe.

[0033] Fig. 4 shows a block diagram of a machine learning system 20 for recognizing particles and / or droplet moisture, implemented in a processor 6.

[0034] Fig. 5 shows an example of recording an acoustic signal (left) in a gas flow through a pipe with a diameter of 89 mm, a thickness of 7.34 mm at a gas flow rate of 2 m / sec; a spectrogram (right) found in a sliding window of 1 ms with an overlap of 0.9 ms. Fig. 6 shows an example of recording an acoustic signal (left) in a gas flow through a pipe with a diameter of 89 mm, a thickness of 7.34 mm at a gas flow rate of 1.5 m / sec and a proppant volume flow rate of 16 / 20 5 g / min; a spectrogram (right) found in a sliding window of 1 ms with an overlap of 0.9 ms

[0035] Fig. 7 shows an example of recording an acoustic signal (left) in a gas flow through a pipe with a diameter of 89 mm, a thickness of 7.34 mm at a gas flow rate of 1 m / sec and condensed moisture of 5 ml / sec or 20 l / hour; a spectrogram (right) found in a sliding window of 1 ms with an overlap of 0.9 ms

[0036] Fig. 8 shows an example of monitoring the removal of solid proppant particles at the mouth of a producing gas well during development.

[0037] The following positions are indicated by numbers on the figures:

[0038] 1 - sensor; 2 - high-pass filter; 3 - instrumentation amplifier with adjustable gain; 4 - low-pass filter; 5 - analog-to-digital converter (ADC); 6 - processor; 7 - memory; 8 - power supply system; 9 - modem; 10 - digital-to-analog converter (DAC); 11 - buffer; 12 - emitter; 13 - sensitive elements; 14 - holders; 15 - housing; 16 - clamping nuts; 17 - additional "nose"; 18 - acoustic signals 18 (waveforms); 19 - pre-processing stage; 20 - machine learning system; 21 - stage of classification into classes; 22 - stage of counting the number of recorded impacts per unit of time.

[0039] Implementation of the invention

[0040] Various design and operational features of the device are detailed below with reference to the drawings. It should be understood that the following detailed description, including the drawings and examples of device operation and the recording method based on it, are illustrative and not limiting.

[0041] The proposed recording device includes a series-connected measuring module, a computing module, and a calibration and self-testing module (Fig. 1).

[0042] The measuring module has at least two sensors (converter) 1 for converting ultrasonic mechanical vibrations of the outer surface of the pipe into an electrical signal, a high-pass filter 2, an instrument amplifier 3 with an adjustable gain for amplifying the signal before digitalization, a low-pass filter 4 anti-aliasing and an analog-to-digital converter (ADC) 5 for converting and digitally recording an alternating time signal.

[0043] High-pass filter 2 is designed to suppress low-frequency industrial interference and reduce the intensity of turbulent flow noise inside the pipe, and low-pass filter 4 is an anti-aliasing filter used at the ADC input to improve the quality of signal digitalization, with a cutoff frequency of 5 kHz.

[0044] Each of the two sensors 1 includes a metal housing 15, in which at least three sensitive elements 3 are located, preferably in the form of layered piezoelectric crystals to increase sensitivity (Fig. 2). The sensitive elements 3 used do not have resonant frequencies in the operating frequency range of the sensor 1. The sensitive elements 13 are secured with holders 14, and the clamping nuts 16 ensure reliable contact with the housing 15, on the lower part of which there is an additional "nose" 17 for transmitting acoustic vibrations (Fig. 2). Thus, the sensitive elements 13 are reliably protected from external acoustic and vibrational influences that can penetrate through the housing 15 and / or the housing parts of the sensor 1.

[0045] Furthermore, the sensing elements 13 are electrically isolated from the housing 15 and / or housing components of the sensor 1 and from the sound pickup (not shown in the drawings) to protect against interference. The sensor 1 is additionally equipped with structural fastening elements, such as clamps and separate mounting parts (not shown in the drawings), which ensure its secure and unambiguous attachment to the wellhead pipe. Fig. 3 shows a schematic diagram of the arrangement of the sensors 1 on the pipe.

[0046] The computing module has a processor 6 for controlling the measuring module and processing data, including a machine learning system for recognizing solid particles and / or droplet moisture, which also controls the calibration and self-testing module (Fig. 1). The computing module is additionally equipped with memory 7 for temporarily storing data before processing and for saving the recognition results in order to increase the reliability of the system in the event of a failure of the modem 9 for transmitting data in real time. Power supply for the measuring and computing modules, as well as power supply for the calibration and self-testing module, is provided by power supply system 8, located in the computing module (Fig. 1).

[0047] The calibration and self-test module contains a digital-to-analog converter (DAC) 10, which converts the digital test signal from the computing module into an analog signal after passing through a buffer 11; the analog signal is transmitted to the emitter 12.

[0048] The device works as follows.

[0049] Sensors 1 convert acoustic waves generated in well pipes by the impact of solid particles (sand, proppant, etc.) and / or droplet moisture into an analog electrical signal (AC voltage) proportional to the intensity of the impact of the solid particles and / or droplet moisture on the pipe. A sound pickup (not shown in the drawings) is used to ensure reliable contact of sensors 1 with the pipe surface and to improve the conversion of acoustic waves. In a preferred embodiment, the signal from the sensitive elements 13 of sensors 1 is amplified by a preamplifier (not shown in the drawings) and then fed to an analog filter path (not shown in the drawings). The analog acoustic signal filtering path suppresses noise that distorts the useful signal and isolates signals emanating from the impact of sand particles and / or droplet moisture.

[0050] After digitalization in ADC 5, acoustic signals 18 (waveforms) are fed to processor 6 of computing module for signal counting and distribution by machine learning system based on neural networks (Fig. 4). Each waveform 18 undergoes pre-processing stage 19, during which adaptive filtering of background noise (industrial interference and noise from fluid flow through pipe) and feature extraction for dimensionality reduction (spectral analysis, autoregressive coefficients, reflection coefficients, etc.) are performed in a window from one to several ms. Next, feature matrix is ​​fed to machine learning system 20, which was pre-trained using similar features in labeled database. Events found in waveforms 18 are classified into several classes 21: 0 – background signals, 1 – impact of solid particle, 2 – impact of droplet moisture (only in case of gas flow). Next, the number of recorded beats per unit of time is calculated 22.

[0051] This number is then calibrated in the calibration and self-test module in one of the following ways: either by a known amount of particles and / or droplet moisture in the flow, which is calculated by taking samples on the line or from the separator; or by theoretical calibration to the cross-sectional area of ​​the pipe, taking into account the hydrodynamic properties of the flow.

[0052] In the self-test mode, processor 6 in the computing module processes the data received from the measuring module and compares it with the original test signal to check the sensitivity of the measuring module.

[0053] Below are examples of the implementation of the proposed group of inventions.

[0054] Example 1.

[0055] Figure 5 shows an example of an acoustic signal recorded in a clean flow through a pipe with a diameter of 89 mm and a thickness of 7.34 mm at a gas flow velocity of 2 m / s, as well as the corresponding spectrogram, found in a sliding window of 1 ms with an overlap of 0.9 ms. The recording shows that the flow without particles and droplet moisture generates stationary noise with a frequency below 30 kHz.

[0056] Example 2.

[0057] Figure 6 shows an example of an acoustic signal recorded in a gas flow through a pipe with a diameter of 89 mm and a thickness of 7.34 mm at a gas flow rate of 1.5 m / sec and a volumetric flow rate of 5 g / min of 16 / 20 fraction proppant, as well as the corresponding spectrogram, found in a 1 ms sliding window with an overlap of 0.9 ms. The recording shows that solid particle impacts generate bursts of acoustic energy in the time domain, which appear as broadband responses in the spectral domain.

[0058] Example 3.

[0059] Fig. 7 shows an example of an acoustic signal recorded in a gas flow through a pipe with a diameter of 89 mm and a thickness of 7.34 mm at a gas flow rate of 1 m / sec and a condensed moisture content of 5 ml / sec or 20 l / hour. It also shows the corresponding spectrogram, found within a 1 ms sliding window with an overlap of 0.9 ms. The recording shows that the presence of condensed moisture in the gas flow also generates bursts of acoustic energy in the time domain. The nature of these bursts differs from that of solid particle impacts on the pipe, both in the spectral domain and in other characteristics. Therefore, recognition system 20 enables the identification of different classes of recorded events. Example 4.

[0060] Fig. 8 shows an example of monitoring the removal of solid proppant particles at the mouth of a production gas well during development: 3-8 g / min (16 mm choke, 470 m 3 / day of gas), 8-10 g / min (20 mm 520 m 3 / day of gas), 2-5 g / min (12 mm, 315 m 3 / day of gas). The results of the solid particle detector measurements were compared with daily sampling at the wellhead; the difference in the recorded volume was no more than 20% (the difference may also be caused by non-stationary operating conditions of the wells).

[0061] Below are the technical characteristics of the proposed device and known devices from the prior art.

[0062] Table 1. Comparative analysis of the technical characteristics of the proposed device with the state of the art.

[0063]

[0064] As demonstrated by the presented data, the proposed device, compared to similar devices, exhibits high sensitivity to detecting solid particles with low impact energy, increasing its accuracy. It does not require pre-calibration, improving ease of use. It also lacks additional frequency filters that limit the digital signal filtering capabilities, increasing reliability. Furthermore, the proposed device features predictive analytics and automation. For example, the device can signal the need to warm up and purge wells through a gas fractionation unit (GFU) upon reaching a certain water cut threshold, a signal for a well logging survey (WGS), or a signal indicating the risk of gas hydrate formation.

[0065] All described structural elements, units and blocks of the device can be made from well-known materials, technologies for processing components and equipment.

[0066] This group of inventions is not limited to the described embodiments, but on the contrary, it covers various modifications and variants within the essence and scope of the proposed formula of the group of inventions.

Claims

FORMULA 1. A non-invasive device for recording solid particles and / or droplet moisture in oil, gas or multiphase fluid flow production systems, comprising a measuring module, a computing module and a calibration and self-testing module connected in series with each other, where the measuring module includes at least two sensors (1) intended for installation on a pipe of the production system and configured to synchronize the conversion of ultrasonic vibrations of the outer surface of said pipe into electrical signals for the computing module, which includes at least a processor (6) equipped with a machine learning system based on neural networks for counting and distributing particles and / or droplet moisture based on said electrical signals into background and / or impact of a solid particle and / or impact of droplet moisture in a given time domain.

2. The device according to claim 1, in which the measuring module is equipped with a high-pass filter (2), an amplifier (3), a low-pass filter (4) and an analog-to-digital converter (5) connected in series with each other.

3. The device according to claim 1, wherein the computing module is provided with a memory (7), a power supply system (8) and a modem (9).

4. The device according to claim 1, in which the calibration module is equipped with a digital analog converter (10), a buffer (11) and an emitter (12) connected in series with each other.

5. A method for recording the removal of solid particles and / or droplet moisture in oil, gas or multiphase fluid flow production systems, which includes the following steps: recording acoustic signals from collisions of solid particles and / or droplet moisture with the inner surface of a pipe of a production system by means of at least two sensors located on said pipe and configured to synchronize the conversion of ultrasonic vibrations of the outer surface of said pipe into electrical signals, amplification of received signals, digitalization of signals in a wide frequency range, in which the signal sampling over time has a continuous duration of at least 1 μs and where each sample is subjected to digital processing for subsequent counting and distribution of signals by a machine learning system based on neural networks, processing of pre-digitized signals by a machine learning system based on neural networks with distribution into background and / or impact of a solid particle and / or impact of droplet moisture and subsequent counting of the number of recorded signals in a given unit of time, - calibration of the obtained values.

Images