EP4204664A1 - Computer-implementiertes verfahren zur bestimmung einer betriebseigenschaft einer gestänge-tiefpumpe, analysevorrichtung und pumpsystem hierfür - Google Patents
Computer-implementiertes verfahren zur bestimmung einer betriebseigenschaft einer gestänge-tiefpumpe, analysevorrichtung und pumpsystem hierfürInfo
- Publication number
- EP4204664A1 EP4204664A1 EP21786787.8A EP21786787A EP4204664A1 EP 4204664 A1 EP4204664 A1 EP 4204664A1 EP 21786787 A EP21786787 A EP 21786787A EP 4204664 A1 EP4204664 A1 EP 4204664A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- pump
- load
- operating
- curve points
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Classifications
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH OR ROCK DRILLING; MINING
- E21B—EARTH OR ROCK DRILLING; OBTAINING OIL, GAS, WATER, SOLUBLE OR MELTABLE MATERIALS OR A SLURRY OF MINERALS FROM WELLS
- E21B43/00—Methods or apparatus for obtaining oil, gas, water, soluble or meltable materials or a slurry of minerals from wells
- E21B43/12—Methods or apparatus for controlling the flow of the obtained fluid to or in wells
- E21B43/121—Lifting well fluids
- E21B43/126—Adaptations of down-hole pump systems powered by drives outside the borehole, e.g. by a rotary or oscillating drive
- E21B43/127—Adaptations of walking-beam pump systems
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH OR ROCK DRILLING; MINING
- E21B—EARTH OR ROCK DRILLING; OBTAINING OIL, GAS, WATER, SOLUBLE OR MELTABLE MATERIALS OR A SLURRY OF MINERALS FROM WELLS
- E21B47/00—Survey of boreholes or wells
- E21B47/008—Monitoring of down-hole pump systems, e.g. for the detection of "pumped-off" conditions
- E21B47/009—Monitoring of walking-beam pump systems
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH OR ROCK DRILLING; MINING
- E21B—EARTH OR ROCK DRILLING; OBTAINING OIL, GAS, WATER, SOLUBLE OR MELTABLE MATERIALS OR A SLURRY OF MINERALS FROM WELLS
- E21B2200/00—Special features related to earth drilling for obtaining oil, gas or water
- E21B2200/20—Computer models or simulations, e.g. for reservoirs under production, drill bits
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH OR ROCK DRILLING; MINING
- E21B—EARTH OR ROCK DRILLING; OBTAINING OIL, GAS, WATER, SOLUBLE OR MELTABLE MATERIALS OR A SLURRY OF MINERALS FROM WELLS
- E21B2200/00—Special features related to earth drilling for obtaining oil, gas or water
- E21B2200/22—Fuzzy logic, artificial intelligence, neural networks or the like
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/088—Non-supervised learning, e.g. competitive learning
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
Definitions
- the invention relates to a computer-implemented method for determining an operating characteristic of a rod pump, the pump having a pump head which is connected to a kinematic converter via a rod and the kinematic converter being driven by a motor during operation. is driven and a load-travel diagram with curve points for the feed pump is determined using a detection means by an analysis device and is provided as an operating load-travel diagram with operating curve points.
- the invention also relates to a computer program, an electronically readable data carrier and a data carrier signal.
- the invention relates to an analysis device with a memory and a pumping system for determining an operating characteristic of a deep-acting rod pump.
- Deep pumps or feed pumps are used as conveying devices for extracting liquids stored underground if the reservoir pressure is not sufficient for them to reach the surface independently or in sufficient quantities. Most of them are used to extract oil. Further areas of application are the pumping of brine and healing waters.
- linkage pumps which are also called horsehead pumps, nick donkeys or nickers because of their appearance and their movement.
- the actual pumping mechanism - a piston with non-return valves - is located in a separate pipe string in the borehole near the oil-bearing layer.
- the piston is set in a continuous up and down motion by means of a screwable rod from a pump stand located on the ground surface. moved. This is accomplished by the so-called horse head.
- This consists of a circular arc segment arranged as a balancer at the end, to which a pair of steel cables or chains is clamped at the top.
- the drive is mostly electric. However, if there are sufficient energy-containing gases dissolved in the crude oil, some of these gases can be separated from the conveyed material on site by means of a degasser and fed to a gas engine that drives the pump.
- the working stroke is 1 to 5 m. Two and a half to twelve strokes per minute are normal.
- the rod-type deep pump can be used economically up to pumping depths of around 2500 m. Other pump systems are more suitable for greater depths due to the heavy weight of the liquid column to be lifted.
- the pump type "Mark II" from the Texan manufacturer Lufkin Industries is particularly suitable for high delivery rates from great depths due to its special movement geometry.
- the "Sucker Rod” type of pump features a sucker rod, which is a steel rod typically between 25 and 30 feet in length and threaded at both ends, used in the oil industry to lubricate the surface and downhole components of an in reciprocating pump installed at an oil well.
- An extremely valuable tool for analyzing well performance is a well test rig, which measures polished rod stress versus polished rod position.
- Dynamometers can be used to record bar position and bar load over time.
- the load sensing portion of the dynamometer is attached to the polished rod so that the load can be captured and sent to a recorder.
- a Companion part of the lifting beam mounted dynamometer captures the position of the polished rod and sends it to the same recorder.
- the graph produced is called a dynagraph, or more commonly a dynamometer or dynagraph map, and corresponds to a load-displacement graph.
- Dynamometer maps taken at the surface can seldom be used directly to record the operating conditions of the downhole pump, since they also reflect all the forces (static and dynamic) occurring from the pump to the wellhead. However, if a dynamometer is placed directly above the pump, the recorded map is a true indicator of pump operation.
- Gilbert's Dynagraph (a mechanical dynamometer) succeeded in doing this in the 1930s. Rod loads immediately above the pump, plotted as a function of pump position, give Dynagraph maps a name that distinguishes them from surface maps.
- sensors have been used to record the operating conditions of a rod pump, which record the forces acting or the current position (inclination) of the moving beam (English “beam” or “cranck arm”), for example by means of force sensors, Hall Sensors or proximity sensors. From this, the position of the linkage is calculated. In this case, however, it is complex to calibrate the respective sensors with one another. In addition, imprecise calibration can result in errors that can adversely affect the measurement data evaluation.
- the object of the invention is achieved by a method of the type mentioned above, with the analysis device in a training mode providing at least one model load-displacement diagram with respective model curve points, which is normalized to a predefined reference variable, and for at least a subset of the model curve points, a model based on a Kohonen network with elliptic Fourier descriptors is generated and trained, and in an operating mode the operating curve points are normalized to the reference variable, elliptic Fourier descriptors for the operating curve points are determined, and it is checked whether there is a similarity of the elliptic Fourier descriptors of the operating curve points to the model of the Kohonen network, and if so, the operating characteristic of the pump is determined from this.
- model load-path diagrams can be used by other pumps and, for example, a newly commissioned pump can be used immediately, ie without prior training with your own operating load-path diagrams .
- an elliptic Fourier transform After normalization, an elliptic Fourier transform can be used.
- Deviations of the operational load-displacement diagram from the load-displacement diagram model, ie from the training model, can, for example, be recognized as an undesired operating property of the pump itself.
- a currently detected pumped medium which is conveyed by the pump, can be recognized by comparing it with a corresponding training model that describes precisely this medium.
- the pump medium is usually a mixture of gas, sand/rock particles, water, oil and sometimes also chemical additives.
- the reference variable can thus be a uniform scale range on which the respective diagrams are mapped.
- a development of the invention provides that the Kohonen network has an input level with input variables and a second level with neurons, and a respective input variable is connected to all neurons of the second level via a respective weight function.
- the neurons are arranged in the second level as a virtual, two-dimensional grid.
- the grid can preferably map the geometric position or the location of an image area in the load-path diagram to the assigned neuron in order to create a visual assignment of the individual neurons to positions or areas of a load-path diagram.
- This is advantageous in order to be able to determine a favorable interdependence of the neurons, namely in the form of a direct mapping of the neurons to positions or areas in relevant load-displacement diagrams, which increases the probability of detection when determining the operating characteristics of the pump increased.
- a neuron is trained with an input vector applied to the input variables, and then neighbors of the neuron are determined and the associated weighting functions are adapted for the neighbors and the model is trained again.
- Neighbor determination can be done easily from a visual grid representation of neurons.
- the motor is operated electrically and a detection means is also provided to detect the power consumption of the motor during its operation, from which the operating properties of the delivery pump are determined.
- the recording of the load-displacement diagram can be simplified, at the same time the accuracy in determining the curve points can be increased and the overall accuracy in determining the operating property can be further improved by the combination with the named analysis method.
- the object according to the invention is achieved by a computer program comprising instructions which, when executed by a computer, cause the computer to carry out the method according to the invention.
- the object according to the invention is achieved by an electronically readable data carrier with readable control information stored thereon, which comprises at least the computer program according to the invention and is designed such that when the data carrier is used in a computing device, the method according to the invention is carried out.
- the object according to the invention is achieved by a data carrier signal which transmits the computer program according to the invention.
- the object according to the invention is also achieved by an analysis device with a memory of the type mentioned at the outset, which is set up to To analyze load-displacement diagram with the method according to the invention, and to determine the operating property.
- the analysis device has two modes of operation, namely a training mode and an operating mode.
- the training mode is used to generate and train a load-displacement diagram model of the pump, with separate load-displacement diagram models being able to be trained for a number of operating modes of the pump in order to recognize these modes accordingly .
- the training mode is run through at the start of operation of the pump system in order to make one or more load-displacement diagram models based on "machine learning” (ML) available for the subsequent operating mode, in which the ongoing operation of the pump system can be continuously monitored by the analysis device.
- ML machine learning
- the operating mode is used to compare a currently recorded data set in the form of measuring points recorded by a recording means, for example with a selected load-displacement diagram model and this model if there is agreement with the load-displacement diagram model assigned operating mode.
- a change in the recognized operational property can provide an indication of a further parameter for the operational property.
- a sequence of changes in recognized operating media can provide information as to whether there is material inhomogeneity in the medium.
- the object according to the invention is also achieved by a pump system of the type mentioned at the outset, the pump having a pump head which is connected to a kinematic converter via a linkage, and the kinematic converter is driven by a motor during operation, and a detection means , which is set up to record and provide a load-travel diagram of the pump with curve points, and the analysis device according to the invention with a memory, which is set up to calculate the operating property from the provided operation-load-travel diagram to investigate.
- FIG. 1 shows an exemplary embodiment of a system according to the invention with a rod pump
- FIG. 2 shows an embodiment of a pump head of a rod pump
- FIG. 3 shows an exemplary embodiment of a flowchart of a method for determining a load-path diagram from the power consumption of an electrically operated pump motor
- FIG. 6 Load-displacement diagrams for a pump with different loads and operating modes, 7 shows a time representation of a current profile of an electric drive motor for a rod pump,
- Fig. 29 shows a schematic representation of a neuron
- Fig. 30 is a schematic representation of a Kohonen
- FIG. 1 shows an exemplary embodiment of a pumping system 100 according to the invention with a rod-type deep pump 1 of the sucker-rod pump type.
- the pump system 100 includes a pump head 110 which is connected to a kinematic converter 120 via a linkage 5 , 10 .
- the linkage 5, 10 forms what is known as a "rod cord” and runs through a wellhead 6, to which a flow line 7 for discharging a conveyed medium 14 is connected.
- Adjoining the wellhead 6 is a casing 8 in which runs a tube 9 which carries the rods 5 and 10, respectively.
- the pump head 110 which contains a piston 11 in a barrel 12 , is fastened to the lower end of the linkage 10 . A movement of the piston 11 causes the delivery medium 14 to be pumped out.
- the shell 8 is formed in a borehole 13 .
- the kinematic converter 120 is driven, for example, by a drive machine in the form of an electric motor 3 via a reduction gear 4 .
- the kinematic converter 120 can also include a hydraulic power booster.
- the mechanical connection of the kinematic converter 120 takes place via a moving beam 2, but can vary depending on the type of pump used.
- the kinematic converter 120 converts a rotary movement of the motor 3 into a linear movement of the linkage 5, 10.
- the properties of the kinematic converter 120 can be described, for example, via lever effects and translations, as well as via the electrical drive power and moving masses. It should be noted here that the position of a centrifugal mass along a rotary movement and the corresponding effect of force on the linkage 10 are related in terms of time, which is referred to as the reference phase angle. For a particular pump assembly, a reference phase angle can be determined using the kinematic principles of mechanics, as known to those skilled in the art.
- a detection means 110 is provided, which is set up to detect the current consumption and the operating voltage of the individual phases of the motor 3 during its operation. This can be done, for example, using an ammeter or voltmeter, which is particularly time-sensitive high-resolution discrete measuring points with current or voltage values are recorded.
- the effective power consumption and the apparent power consumption can be determined from the recorded current and operating voltage values.
- an analysis or computing device 140 with a memory 150 is provided, which is set up to carry out the method according to the invention with the aid of the detection means 130 .
- the recording means 130 is set up to record an operating-load-travel diagram of the pump 1 with curve points and to provide it to the computing or analysis device 140 with the memory 150 .
- the analysis device 140 is set up to analyze the provided operating-load-path diagram using the method according to the invention and to determine the operating property from it.
- the method according to the invention can be implemented as a computer program which includes instructions which, when they are executed by a computer 140, cause the latter to carry out the method according to the invention.
- the method according to the invention can be available as an electronically readable data carrier with readable control information stored thereon, which comprises at least the computer program according to the invention and is designed in such a way that they carry out the method according to the invention when using the data carrier in a computing device 140 .
- the method according to the invention can also be available as a data carrier signal which transmits the computer program according to the invention.
- FIG. 2 shows another, more detailed example of a pump head 111 according to the prior art.
- the rod line or the linkage 10 is driven as shown in FIG. 1 and set in an up and down linear motion.
- a cover tube 15 with vertical grooves is arranged in the borehole 13, which guides a rotating tube 18 with spiral grooves inside the cover tube 15 via a holding device 16 and a self-aligning bearing 17.
- a pickup tube 19 is connected via a wing nut 20 to a piston assembly 21 located in a pump liner 22 .
- a calibrated rod 23 is connected to the linkage 10 via a pin 24 and a retainer 25 which drives the piston assembly through linear movement.
- FIG. 3 shows an exemplary embodiment of a flow chart of a method for determining a load-path diagram from the power consumption of an electrically operated pump motor with the following steps: a) Recording the power consumption and the operating voltage of the motor 3 with a sampling frequency over at least one pump cycle, which can be assigned to four operating phases of the deep pump 1, in the form of discrete measuring points with current values, and from this determining the power consumption 72 of the motor 3 Performance values, b) determining a period 85 and a maximum 82 of the power consumption 72 for a pumping cycle, which corresponds to the maximum torque of the deep pump 1, c) determining a reference phase angle for the kinematic converter 120 using the properties of the kinematic converter 120 and the power consumption of the motor 3, which describes the relationship between the maximum 82 of the power consumption and the maximum of the force acting on the linkage of the deep pump 1, d) determining a torque curve from the power consumption of the motor 3 using the properties of the kinematics -Converter 120, e)
- the power values can be determined by the product of the discrete current values and the operating voltage.
- the period duration 85 can be determined, for example, with the aid of an approximated polynomial 80 from the power values of the measuring points.
- the period duration 85 can, for example, also be determined with the aid of a polynomial 80, which takes into account statistical mean values of the power values of the respective measurement points over at least five, preferably at least ten, particularly preferably at least fifty pump cycles for support points of the polynomial.
- a reference value 81 can be determined for the measurement points, at which there is a maximum for the change in the respective power value between two measurement points directly following one another, and the period duration 85 is determined using the reference value 81 .
- the operating properties of the feed pump 1 can be determined using a load-path diagram 30, 50, 54, 57, 60-65, which is calculated from the torque curve determined in step d) using the in step b) determined period and the determined in step c) reference phase angle is determined.
- the reference phase angle can be determined with respect to the absolute maximum of the power values of the measurement points within a pump cycle.
- Fig. 4 to Fig. 6 show examples of load-displacement diagrams, which are often used to determine the operating characteristics of deep-drawer pumps.
- a load-displacement diagram 30 is shown in FIG.
- the x-axis shows the polished bar position 31 and the y-axis shows the polished bar load 32 .
- a lowest point of the pump stroke 33 and a highest point of the pump stroke 34 can be seen.
- a tip of the polished rod 35 (PPRI) is also shown.
- a map 36 of the polished rod for zero pumping speed is drawn in dashed lines. Also, a pump speed polished rod map 37 is greater than zero.
- a minimum polished rod 38 load (MPRL) is shown.
- a gross piston load 39 can also be read off.
- a weight of the rods in the fluid 40 can be determined, as well as forces 41 and 42, and a pump stroke or pump path 43.
- FIG. 5 Shown in FIG. 5 are load-displacement plots 50 of rod load at setpoint as a function of polished rod load 32 versus polished rod position 31 .
- a load-path diagram 51 shows operation at full pump power.
- a load-displacement diagram 52 shows the operation when the conveying medium has been pumped dry.
- a respective target value 53 can be seen.
- load-displacement diagrams 54 are shown with rod load during a change in operation as a function of the load 32 of the polished rod over the respective position 31 of the polished rod, with the respective angles 55, 56 being readable.
- load-displacement diagrams 57 with rod load are shown with the respective mechanical work of the rods.
- FIG. 6 shows load-travel diagrams 60-65 for various operating states.
- Diagram 60 shows load-displacement diagrams for normal operation.
- Diagram 61 shows load-displacement diagrams for a fluid bearing.
- Diagram 62 shows load-displacement diagrams for exposure to gas in the underground storage facility.
- Diagram 63 shows a load-displacement diagram for a stuck piston.
- Diagram 64 shows the load-travel diagram in the event of a leak through a stationary valve.
- a diagram 65 shows a load-travel diagram in the event of a leak through a moving valve.
- the operating characteristics of the pump 1 can be determined by the analysis device 140 from such load-displacement diagrams.
- the analysis device 140 is provided with at least one model load-displacement diagram with respective model curve points in a training mode.
- the model load-displacement diagram is then normalized to a predefined reference value by adjusting and standardizing the value ranges.
- At least two subsets of the model curve points are then detected as a first and at least a second feature based on machine learning.
- a feature can be, for example, a specific curve shape or the position of curve points in the load-displacement diagram, distances or changes in distance between individual curve points in the load-displacement diagram.
- the first and the at least one second feature are used to generate and train at least one random forest model using a kmeans algorithm.
- the analysis device 140 normalizes the operating curve points to the reference variable.
- the analysis device 140 checks whether there is a similarity of at least a subset of the operating curve points to the at least one random forest model.
- At least two random forest models can be formed, which have a low correlation to one another.
- the at least two random forest models with low correlation can be generated by randomly selecting one point from the set of operating curve points and replacing it from the set of operating curve points.
- the at least two low-correlation random forest models may be generated by further considering a subset when splitting a node in a random forest model.
- a sequence of the first and the at least one second feature within a pump cycle during operation of the pump 1 can be taken into account in the respective random forest model when determining the operating characteristics of the pump 1.
- Fig. 7 shows an example of a chronological representation of a power curve of an electric drive motor for a linkage deep pump, which was determined from the power consumption and operating voltage of the motor 3.
- the representation has a time axis 70 and an axis 71 for the amplitude of the current or power consumption.
- a power consumption 72 is shown for which a zero point or zero axis 80 and a polynomial for average power consumption 81 can be determined.
- a maximum value of the average power consumption 82 and zero crossings of the average power consumption 83, 84 can be determined for the polynomial 80.
- a period 85 of the averaged power consumption can be determined for the polynomial 80 .
- phase angle 86 of the average power consumption can be determined, which describes the relationship between the rotational movement of the motor 3 and the linkage 10 of the pump 1.
- a corresponding load-displacement diagram can be determined from the determined values in order to derive the operating properties of the deep-acting rod pump 1 in a simple manner.
- Deep pump 1 can be defined by one or more corresponding load-displacement diagrams in terms of "target values", which are used in a training mode as based on machine learning based model can be generated and trained. It is also possible to access load-displacement diagrams from other pumps.
- a question about the gas content in an oil-water-gas mixture at a well can be answered by creating and training a training model for a known mixture.
- This training model serves as a reference to an operating load-displacement diagram.
- Deviations in the operational load-displacement diagram from the training model can be recognized as an undesirable operational characteristic.
- a load-displacement diagram model with model curve points based on machine learning is generated and trained by the analysis device 140 .
- a reference point check can be performed by the following steps:
- At least two predefined analysis areas which at least partially include the model curve points, can then be determined in the load-displacement diagram model.
- a reference point can then be determined from the model curve points for at least one area of the analysis areas, which, for example, corresponds to the geometric center of gravity of the curve points of the respective area or the area formed by the curve points and the area boundaries, for example the diagram axes.
- the analysis device 140 can check for the operating load-path diagram whether the at least one reference point determined in the training mode is enclosed within the area enclosed by the operating curve points.
- the operating characteristic of the feed pump 1 can be determined from the reference point recognized as "trapped".
- FIG. 8 shows an example of an operating-load-travel diagram DC1 with curve points.
- the areas can, for example, directly adjoin one another, so that there are no areas with curve points enclosed therein that are not assigned to reference points.
- areas can also be excluded, for example to prevent areas with frequently error-prone curve points from being deliberately excluded in order to achieve an improvement in stability when determining the operating properties of the pump.
- area boundaries can also overlap, so that a curve point can be assigned to several areas.
- the measurement curve points and the model curve points between two adjacent points on the respective curves can each have distances which on average are at least 50%, preferably at least 80% and particularly preferably at least 95% of the greatest distance between two adjacent points on the respective curve amounts to. This can result in approximately equal distances between measurement curve points.
- FIG. 9 shows an example of an operating-load-path diagram DC2 with curve points.
- centroid point C21-C24 is drawn in for each of four areas, which corresponds to the geometric focus of the curve points of the respective area or the area formed by the curve points and the area boundaries, for example the diagram axes.
- the areas at least partially overlap one another in order to better define a reference point of the area if, for example, too few curve points are contained in an area.
- At least one area can be defined for which it is provided that no included curve point is taken into account in the subsequent check.
- an area can be excluded from closer examination.
- the recognition of operating characteristics can be carried out in a particularly accurate manner and the recognition rate when determining the operating characteristic can be further improved.
- an iterative test can also be provided by using the method according to the invention, in order to gradually substantiate certain suspicions with regard to a suspected operating characteristic by adapting the criteria with regard to the training model.
- the accuracy requirements can be successively increased by increasing reference points in the respective subsequent training model, or alternative reference points can also be examined in order to draw further conclusions for the one to be examined, for example using a combination of two different training models to draw operating property.
- FIG. 10 shows an example of an operating load-travel diagram DC3 with curve points.
- centroid point C31-C34 is drawn in for each of four areas, which corresponds to the geometric center of gravity of the curve points of the respective area.
- FIG 11 shows an example of an operating-load-path diagram DC4 with curve points.
- FIG. 12 shows an example of an operating load-travel diagram DC5 with curve points.
- centroid point C51-C58 is drawn in for each of eight areas, which corresponds to the geometric center of gravity of the curve points of the respective area.
- FIG. 27 show examples of load-displacement diagrams with different elliptical Fourier descriptors, which each represent harmonics of a total of 15 Fourier coefficient pairs ⁇ i , b i , c i and d i .
- the dashed curves each represent a recorded load-displacement diagram which is to be analyzed and recognized, ie classified, according to the method according to the invention.
- the representation of the dashed curves in the first quadrant only serves to improve clarity; an analysis refers to a normalized representation.
- the solid curves represent a respective harmonic of a load-displacement curve to be examined.
- An nth harmonic is a curve or polynomial of the nth order.
- the values for load and displacement can be mapped to an interval for the displacement between zero and four, and an interval for the load between zero and one.
- FIG. 13 shows an example for the first harmonic, FIG. 14 for the second harmonic, etc. and FIG. 27 for the 15th harmonic of the Fourier coefficient pairs ⁇ i , b i , c i and d i .
- the operating characteristics of the pump can be determined very precisely.
- the examples each show Fourier transformations with curves of the operating load-displacement diagram with iteratively increasing order.
- T is the period, i.e. the sum of all T
- n the number of considered harmonics
- N the total number of all harmonics, a n , b n the elliptic Fourier coefficients of the nth harmonic.
- the Fourier coefficients for the x-projection of the curve, ie the path, can be determined using the following relationship:
- K the total number of all connections, a n , b n the elliptic Fourier coefficients of the nth
- x p the sum of all connections on the x-axis
- p the index in a connection chain
- t p the length of a chain along a path.
- n is the number of harmonics considered
- K the total number of all connections
- c n , d n the elliptic Fourier coefficients of the nth harmonic
- y p the sum of all connections on the y-axis
- p the index in a connection chain
- t p the length of a chain a path.
- FIG. 28 shows an example of a bit mask for filtering load-displacement diagrams, ie a variant for classifying curve shapes in load-displacement diagrams.
- features of a load-displacement diagram can be assigned to corresponding bit patterns of the mask.
- This example shows a rectangular filter mask measuring 80 x 20 bits, with the pattern representing a "healthy" pump.
- a bit vector b 11 ,...,b 1N ,....,b M1 ,...,b MN is used as the input variable, where M is the dimension of the load in rows and N is the dimension of the path (displacement) in columns.
- Fig. 29 shows a schematic representation of a neuron as an element in systems with artificial intelligence.
- Input variables X j which form an input vector X, are passed to a neuron, i.e. a processing element, via weight functions W j , which form a weight vector W. ment (engl. "processing element") supplied, which forms an output variable Y.
- a "Kohonen Feature Mapping Neural Network” is formed from the neurons, which represents a model based on machine learning.
- This model is generated and trained with training data, whereby the training data can also come from other pump systems that are not structurally identical.
- a Kohonen network Similar to a multi-layer perceptron (MLP) network, a Kohonen network also has a multiplicity of neurons PE (processing elements) with a plurality of inputs X j weighted by means of respective weighting functions W j and one Output Y open.
- PE processing elements
- a Kohonen network compared to, for example, an MLP is the architecture, which has an input level LI with N inputs, followed by another level L2 with a plurality of neurons arranged in a two-dimensional lattice.
- the lattice has a length L and a height H, with one neuron being arranged for each lattice position.
- Each input X j is connected to all second level neurons PE ij by means of a weighting function W j .
- FIG. 30 shows a schematic representation of a Kohonen network.
- Input variables X 1 , X 2 , X j to X N are fed via respective weighting functions W ij to a respective neuron PE ij , which forms an output variable.
- each input variable is supplied to each neuron.
- An output variable is determined for each neuron, which represents, for example, a recognized operating characteristic of the pump.
- the neural Kohonen network is trained by modifying the weight functions and other variable rules.
- the aim of a single learning step is to find that neuron PE ij whose weight vector as close as possible to the input vector is located, i.e. the closest match to the input vector is equivalent to.
- the neuron with the best match is labeled with the class number associated with the input vector is associated.
- the tag is updated with the last class number.
- weight vector of the neuron and its neighboring neurons are updated, whereby the remaining weight vectors of the other neurons are not changed.
- This updating means a changed weight value for the respective neurons, with different permutations of weight values being carried out on neighboring neurons in order to find an optimal match.
- Neighbor neurons are adjacent in the lattice, i.e. there are eight neighbors.
- the neighboring neurons come closer to the input vector, while the topology of the input space, i.e. the order of the input variables, is retained.
- the Kohonen network can also be compared to a human visual cortex, in which visible information is processed very efficiently.
- elliptical Fourier descriptors ie Fourier coefficient pairs a 1 , b 1 , c 1 , d 1 ,..., a N , b n , c N , d N , which represent the curve shape of a load-displacement diagram, are used as input variables X 1 , X 2 , X j to X N of the neural network.
- ML machine learning model
- the use of a Kohonen network with elliptical Fourier descriptors are significantly more advantageous than a Kohonen network that is generated using a bit mask, because a significantly larger Kohonen network is formed depending on the size of the bit mask.
- PE PE ij neuron, engl.
- processing element PE, PE ij neuron, engl.
Landscapes
- Geology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Mining & Mineral Resources (AREA)
- Physics & Mathematics (AREA)
- Environmental & Geological Engineering (AREA)
- Fluid Mechanics (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Geochemistry & Mineralogy (AREA)
- Geophysics (AREA)
- Control Of Positive-Displacement Pumps (AREA)
- Investigation Of Foundation Soil And Reinforcement Of Foundation Soil By Compacting Or Drainage (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
Description
Claims
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| EP20202028.5A EP3985224A1 (de) | 2020-10-15 | 2020-10-15 | Computer-implementiertes verfahren zur bestimmung einer betriebseigenschaft einer gestänge-tiefpumpe, analysevorrichtung und pumpsystem hierfür |
| PCT/EP2021/076390 WO2022078736A1 (de) | 2020-10-15 | 2021-09-24 | Computer-implementiertes verfahren zur bestimmung einer betriebseigenschaft einer gestänge-tiefpumpe, analysevorrichtung und pumpsystem hierfür |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| EP4204664A1 true EP4204664A1 (de) | 2023-07-05 |
Family
ID=73037682
Family Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| EP20202028.5A Pending EP3985224A1 (de) | 2020-10-15 | 2020-10-15 | Computer-implementiertes verfahren zur bestimmung einer betriebseigenschaft einer gestänge-tiefpumpe, analysevorrichtung und pumpsystem hierfür |
| EP21786787.8A Withdrawn EP4204664A1 (de) | 2020-10-15 | 2021-09-24 | Computer-implementiertes verfahren zur bestimmung einer betriebseigenschaft einer gestänge-tiefpumpe, analysevorrichtung und pumpsystem hierfür |
Family Applications Before (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| EP20202028.5A Pending EP3985224A1 (de) | 2020-10-15 | 2020-10-15 | Computer-implementiertes verfahren zur bestimmung einer betriebseigenschaft einer gestänge-tiefpumpe, analysevorrichtung und pumpsystem hierfür |
Country Status (4)
| Country | Link |
|---|---|
| US (1) | US20230383641A1 (de) |
| EP (2) | EP3985224A1 (de) |
| CA (1) | CA3198704A1 (de) |
| WO (1) | WO2022078736A1 (de) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP4555194A1 (de) * | 2022-07-15 | 2025-05-21 | Services Pétroliers Schlumberger | Feldausrüstungssystem |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6343656B1 (en) * | 2000-03-23 | 2002-02-05 | Intevep, S.A. | System and method for optimizing production from a rod-pumping system |
| US9574442B1 (en) * | 2011-12-22 | 2017-02-21 | James N. McCoy | Hydrocarbon well performance monitoring system |
| CN113167269B (zh) * | 2018-12-16 | 2024-09-06 | 森西亚有限责任公司 | 泵系统 |
-
2020
- 2020-10-15 EP EP20202028.5A patent/EP3985224A1/de active Pending
-
2021
- 2021-09-24 CA CA3198704A patent/CA3198704A1/en active Pending
- 2021-09-24 WO PCT/EP2021/076390 patent/WO2022078736A1/de not_active Ceased
- 2021-09-24 US US18/030,856 patent/US20230383641A1/en not_active Abandoned
- 2021-09-24 EP EP21786787.8A patent/EP4204664A1/de not_active Withdrawn
Also Published As
| Publication number | Publication date |
|---|---|
| US20230383641A1 (en) | 2023-11-30 |
| CA3198704A1 (en) | 2022-04-21 |
| EP3985224A1 (de) | 2022-04-20 |
| WO2022078736A1 (de) | 2022-04-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3876060B1 (de) | Verfahren und recheneinheit zur ursachenanalyse eines anomalen zustandes einer maschine | |
| DE112008003302T5 (de) | Verfahren und Systeme zur Abschätzung von Bohrlochereignissen | |
| DE112013007388T5 (de) | Optimierte Drehung eines Bohrstrangs im Gleitmodus beim Richtbohren | |
| EP3625629B1 (de) | Vorrichtung und verfahren zur bestimmung des zustands einer spindel einer werkzeugmaschine | |
| DE102016201559A1 (de) | Verfahren und Vorrichtung zum Vermessen eines zu testenden Systems | |
| WO2022078736A1 (de) | Computer-implementiertes verfahren zur bestimmung einer betriebseigenschaft einer gestänge-tiefpumpe, analysevorrichtung und pumpsystem hierfür | |
| WO2022078734A1 (de) | Computer-implementiertes verfahren zur bestimmung einer betriebseigenschaft einer gestänge-tiefpumpe, analysevorrichtung und pumpsystem hierfür | |
| DE102018205561A1 (de) | Vorrichtung zur Klassifizierung von Signalen | |
| DE102022210195A1 (de) | Verfahren zur automatischen Erstellung eines KI-Diagnosemodells zur Diagnose eines anormalen Zustands auf der Grundlage von Geräusch- und Vibrationsdaten, auf die ENAS angewendet wird | |
| Yamaliev et al. | About the deep drilling equipment technical condition recognition method | |
| EP3985222A1 (de) | Computer-implementiertes verfahren zur bestimmung einer betriebseigenschaft einer gestänge-tiefpumpe, analysevorrichtung und pumpsystem hierfür | |
| DE102017126341A1 (de) | Verfahren und Vorrichtung zur Bestimmung eines Verschleißzustands in einer Hydrostatpumpe | |
| WO2022152753A1 (de) | Aufbereitung von betriebsdaten einer mehrheit von parallel geschalteten fördersträngen, wobei jeder strang einen strömungswiderstand aufweist | |
| CN120031176A (zh) | 用于确定钻井作业工况的方法、装置、设备以及存储介质 | |
| DE68925388T2 (de) | Verfahren und Vorrichtung zur Ermittlung einer Pumpeigenschaft oder eines Parameters eines Fluids | |
| DE102018217310A1 (de) | Verfahren zum Auswerten von Zeitseriensignalen | |
| DE102023204278A1 (de) | Verfahren zur Analyse eines Maschinenlernmodells zur Detektion von Anomalien in Sensorsignalen | |
| EP3816444A1 (de) | Verfahren zur bestimmung von betriebseigenschaften einer gestänge-tiefpumpe sowie pumpsystem hierfür | |
| EP3988761A1 (de) | Verfahren, analysevorrichtung und pumpsystem zur erkennung eines ölaustritts bei einer gestänge-tiefpumpe | |
| DE102007007301A1 (de) | Systeme und Verfahren zur Erfassung des Schließens eines Ansaugventils | |
| Han et al. | Multiple fault diagnosis for sucker rod pumping systems based on matter element analysis with F-statistics | |
| DE112022007203T5 (de) | Informationsverarbeitungseinrichtung, Programm und Informationsverarbeitungsverfahren | |
| Bangert | Predictive and Diagnostic Maintenance for Rod Pumps | |
| Krogerus | Feature extraction and self-organizing maps in condition monitoring of hydraulic systems | |
| DE102024200871A1 (de) | Verfahren und Vorrichtung zum Optimieren einer Out-of-Distribution (OOD) Erkennung |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: UNKNOWN |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE |
|
| PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE |
|
| 17P | Request for examination filed |
Effective date: 20230329 |
|
| AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR |
|
| DAV | Request for validation of the european patent (deleted) | ||
| DAX | Request for extension of the european patent (deleted) | ||
| GRAP | Despatch of communication of intention to grant a patent |
Free format text: ORIGINAL CODE: EPIDOSNIGR1 |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: GRANT OF PATENT IS INTENDED |
|
| INTG | Intention to grant announced |
Effective date: 20240328 |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE APPLICATION IS DEEMED TO BE WITHDRAWN |
|
| 18D | Application deemed to be withdrawn |
Effective date: 20240730 |