WO1993002289A1

WO1993002289A1 - Pump control using calculated downhole dynagraph information

Info

Publication number: WO1993002289A1
Application number: PCT/US1992/006105
Authority: WO
Inventors: G. Wayne Westerman; Richard C. Montgomery
Original assignee: Westerman G Wayne; Montgomery Richard C
Priority date: 1991-07-22
Filing date: 1992-07-22
Publication date: 1993-02-04
Also published as: AU2416092A; WO1993002210A1

Abstract

Apparatus and methods are disclosed for controlling the operation of a rod-pumped oil well using calculated downhole performance of the subterranean pump (6). The surface load and position are measured at equal increments of time. The measured load and position values, in combination with physical parameters of the pumping system, including the size, weight, elastic properties, and length of the rod string (4), are used by one of three preferred mathematical-analysis techniques (figs 9-11) to construct a downhole dynagraph plot of load versus position over time as it exists at the subterranean pump (e.g., 110 or 111). The characteristics of the downhole dynagraph are then analyzed by one of a plurality of algorithms (figs 7 or 8) to detect pump-off or other pump problems.

Description

DOWNHOLE DYNAGRAPH INFORMATION Technical Field This invention relates to a method for determining the operating characteristics of a pumping well and for making control decisions based on those determinations. The invention is directed to a process for automatically determining downhole conditions of a pumping well on a real-time basis from data that are received, measured, and manipulated at the surface of the well. The invention has specific applications in the control of well operations based on calculated changes that occur at the subsurface, or downhole, pump.

Background Art When fluid is pumped out of a well faster than it can be replenished, at some point the downhole pump barrel will fail to fill completely on each stroke. That condition is known as "pump off. " Not only is pump-off inefficient, but also it can lead to damage to the pumping equipment.

A number of known techniques exist to detect pump-off using surface measurements of rod load and rod position during a stroke, which may be plotted on a "dynagraph." See, for example, U.S. Patent No.4,487,061, issued to McTamaney, et al., which describes several such methods, U.S. Patent No. 4,286,925, issued to Standish, U.S. Patent No.4,015,469, issued to Womack, and U.S. Patent No. 3,951,209, issued to Gibbs, all of which are hereby incorporated by reference. Current methods of detecting pump-off require operator competence in analyzing surface dynagraph information to obtain an optimum setting for the detection of pump-off under the conditions applicable to a particular well. Often a significant amount of time and experience is required by the operator to assure that proper control conditions have been set.

The rod string can stretch, causing the movement of the downhole pump, in displacement and phase, to differ from the movement of the driving equipment at the surface. The actual load at the downhole pump also differs from the load measured simultaneously at the surface. Those differences become more significant with well depth. As a result, interpretation of surface data is difficult, and methods of detecting pump-off based on surface measurements may be inaccurate. Moreover, control devices that operate based on surface data tend for those same reasons to lack consistency. Methods are also known to infer mathematically downhole load and position data from surface measurements. For example, see U.S. Patent No.3,343,409, issued to Gibbs, and U.S. Patent No. 3,527,094, issued to Yew, et al, which are hereby incorporated by reference. Such downhole-analysis methods are often mathematically complex and have required significant computer time to infer downhole conditions. Thus, while such techniques are useful for analyzing well performance, they have been unsuitable heretofore for use as the basis for an automatic, real-time pump-off controller.

It would be advantageous to provide a system that could control a well using downhole data instead of surface data. One advantage of using downhole data in pump control is to permit more rapid and accurate adjustment of the pump controller by personnel with a lower level of dynagraph-analysis capabilities. It would also be advantageous to control the well on a "real-time" basis, whereby data are analyzed on-site as the well operates. Prior systems have been unable to transform surface data into downhole data fast enough to perform the calculations for each stroke simultaneously with the operation of the well.

Another advantage of such a system is uniformity. The shape of the surface dynagraph is affected by such factors as length, size, and thickness of the rod string, type of material used in the rods, well depth, stroke length, pump size, pumping speed, degree of slip of the prime mover, pumping friction, presence or absence of a tubing anchor, andpumping-unit geometry. Accordingly, differently configured systems often have widely varying dynagraph shapes even though they are experiencing the same condition, such as a stuck pump or a leaking valve. Such variation results in a vast number of possible dynagraph shapes, as reflected in API Bulletin 11L2, for example, which presents over 1100 dynagraph shapes for full- pump conditions in a single pumping unit geometry. Also, the presence of pump-off will often have a major effect on the shape of the surface dynagraph. Such variation causes difficulty in correctly adjusting the operation of a pump controller. By contrast, the shape of the downhole dynagraph, calculated at the downhole pump, is more regular and consistent across variations in pumping equipment and conditions. Consequently, the tasks of recognizing pump faults or alarm conditions, such as pump-off, and of properly adjusting the pump controller are greatly simplified. In particular, the current system eliminates the need for the user to set up the controller for each well, for example by determining "full-pump" and "pumped-of " conditions and saving data collected during those states for later reference by the controller. In addition, the controller can more easily and reliably detect and classify downhole faults using downhole data. Disclosure of Invention It is therefore an object of the invention to provide an automatic, real-time pump controller using downhole-analysis techniques.

It is another object of the invention to use mathematical techniques for inferring downhole data that are sufficiently simple for use in a real-time pump controller.

It is another object of the invention to provide a system for pump control that need not be calibrated for each well on which the controller is used, and which permits calibration with greater ease and with less training. It is another object of the invention to provide a pump controller that uses downhole analysis to detect fault conditions with greater uniformity across wells of different depths or types, operated at different pumping speeds, and containing different rod materials.

It is another object of the invention to provide a pump controller with sufficient processing power to allow the well to be controlled on a real-time basis using inferred downhole data.

It is another object of the invention to provide such a pump controller that is specifically applicable to detecting and controlling the condition of pump-off. The above and other objects are achieved in the present invention by combining calculations of the downhole dynagraph with methods of automatic pump control. In a preferred form, load and position data are measured at frequent time intervals at the surface by transducers. The data are conditioned and converted for use by a computer controller. Data stored in the controller describe the physical characteristics of the rod string in the well. The controller includes a processor programmed to implement a wave-form transformation to create, on a real-time basis, an inferred downhole dynagraph, describing the load and position at the pump, based on the surface load and position data and the known physical parameters or configuration of the well. A computer program determines whether a fault, such as a pumped-off condition, exists by evaluating the downhole dynagraph data. Upon recognition of a fault, such as pump-off, control of the pump motor may be transferred or the pump maybe stopped for a time.

The system is flexible enough to implement several transformation algorithms or fault-recognition methods. The system could choose among the several alternatives manually or automatically, or the system could implement multiple algorithms simultaneously and reach an average or consensus result. Other aspects of the invention will be appreciated by those skilled in the art after a reading of the detailed disclosure of the present invention, below.

Brief Description of Drawings The novel features of this invention are described with particularity in the claims. The invention, together with its objects and advantages, are better understood after referring to the following description and accompanying figures. Throughout the figures, the same reference numerals refer to the same elements.

FIG. 1 shows a cross-section of the pumping equipment, both at the surface and downhole. FIG.2 shows a close-up view of a polished-rod load transducer.

FIG.3 shows a close-up view of the downhole pump, shown in a pumped-off condition.

FIG.4 shows a conceptual block diagram of the computer hardware and software used in the pump controller, together with the computer's input and output signals.

FIG.5 shows a general logic flowchart for a preferred embodiment of the invention.

FIG. 6 shows example surface and calculated downhole dynagraphs for both "full pump" and "pumped-off" conditions, with indications of how the downhole cards are used in several preferred embodiments to detect pump-off.

FIG. 7 shows a high-level software flowchart for one embodiment of the pump-off-detection algorithms described in Fig. 6, the full stroke integration method.

FIG. 8 shows a high-level software flowchart for another embodiment of the pump-off-detection algorithms described in Fig. 6, the downstroke integration method.

FIG. 9 is a software flowchart for a preferred embodiment of the downhole- conversion algorithm using a discrete Fourier-series solution.

FIG. 10 is a software flowchart for a preferred embodiment of the downhole- conversion algorithm using the fast Fourier algorithm. FIG. 11 is a software flowchart for a preferred embodiment of the downhole- conversion algorithm using a finite differences solution.

Best Modes for Carrying out the Invention Shown in Fig. 1 is an overview of a typical oil well and pumping unit 2. It is common practice to employ a series of interconnected rods, called sucker rods, comprising the rod string 4, for coupling pumping unit 2 to subsurface pump 6. The uppermost rod, generally referred to as polished rod 8, passes through stuffing box 10, allowing the rod string to move up and down in the well without leaking well fluid. Referring to Fig. 2, rod string 4 is suspended from pumping unit bridle 12 on carrier bar 14 by means of polished rod clamp 16.

Referring to Fig.3, rod string 4 connects the pumping unit to plunger 18 of the pump, which is moved up and down in barrel 20 by the reciprocating motion of rod string 4. On the upstroke, the fluid (shaded) in tubing 22 is raised by the pump, and all of the fluid load is supported by plunger 18 and travelling valve 24. On the downstroke, plunger 18 moves downward into pump barrel 20 filled with liquid. The pressure of the fluid in barrel 20 causes the ball of travelling valve 24 to open and allows plunger 18 to travel downward through the liquid in pump barrel 20. On the downstroke, therefore, the fluid load is transferred from plunger 18 and travelling valve 24 to standing valve 26 and tubing 22.

When the hydrostatic head of the fluid level in the annuls between tubing 22 and well casing 28 is reduced to below the critical pump-intake pressure, the subsurface pump will cavitate due to incomplete filling, creating a condition commonly called "pump-off." Due to incomplete filling of the pump, vapor is present in the upper portion of pump barrel 20, the condition illustrated by the shading in Fig. 3. The pressure in the vapor space below travelling valve 24 is insufficient to cause the valve to open, and the load does not transfer from rod string 4 to tubing 22 until pump plunger 18 strikes the subsurface vapor-liquid interface in pump barrel 20 violently, causing a condition called "fluid pound." The rapid transfer of energy that occurs during fluid pound may cause damage to rod string 4, the subsurface pump, or the surface driving equipment. It is therefore desirable to detect accurately the occurrence of pump-off-induced fluid pound and to stop operation of the well until fluid can rise in the annuls between casing 28 and tubing 22.

To obtain the necessary data to make the calculations for the downhole dynagraph, it is necessary to obtain measurements of surface load and position over time. A number of known techniques for making such measurements may be used for that application. For example, U.S. Patent No. 4,143,546, issued to Wiener, and U.S. Patent No. 3,457,781, issued to Elliott, each disclose at least one type of load- measurement technique and one type of position-measurement technique, and are incorporated herein by reference.

For example, referring back to Figs. 1 and 2 of the present application, a load transducer 30 maybe inserted between carrier bar 14 and polished-rod claim 16, which thereby carries all of the rod load. The electrical output from such a polished- rod-mounted load transducer 30 is directly proportional to the load on polished rod 8. See, e.g.. U.S. Patent No.4,363,605, issued to Mills, incorporated herein by reference.

Alternatively, load on polished rod 8 can be measured by welding a load transducer 32 to the top flange of the walking beam 34. See, e.g._f U.S. Patent No. 3,817,094, issued to Montgomery, et al., incorporated herein by reference. Polished rod 8 imposes a load on the walking beam — through carrier bar 14, polished rod claim 16, and bridle 12 — which causes the walking beam to bend slightly, thus elongating the top flange of the walking beam 34. Such a beam-mounted load transducer 32 measures the elongation in the top flange of the walking beam 34, which is proportional to the load on polished rod 8. It should be noted that the walking beam is also subjected to deformation due to differential heating of the top and bottom flanges of the walking beam 34 and 36. When the temperature of the top flange of the walking beam 34 is higher than the temperature of the bottom flange of the walking beam 36, elongation of the top flange of the walking beam 34 occurs. To use a beam-mounted load transducer 32 for quantitative calculation of the downhole dynagraph, it is desirable to compensate for the error caused by the differential- temperature effect on the walking beam.

Besides measuring load, to make the calculations necessary to determine the shape of the downhole dynagraph, it is necessary to measure the coincident position of polished rod 8. The angle of the walking beam 38 maybe measured by connecting the body of a position transducer or potentiometer 40 to the static structure of the pumping unit, for example to sampsonpost 42. Potentiometer 40 contains a shaft connected parallel to the walking beam through a shaft extension 44 and a chain 46. As the walking beam rotates through its pumping arc, the shaft extension 44, and thus the internal wiper in potentiometer 40, is rotated through the same angle 38 as the walking beam. The position of the potentiometer wiper is thus proportional to the position of polished rod 8.

Alternatively, an inclinometer 48 may be mounted on the top flange of the walking beam 34, which produces an electronic signal proportional to the angle 38 of the walking beam and thus to the position of polished rod 8. See, e.g._f U.S. Patent No. 4,561,299, issued to Orlando, et al., incorporated herein by reference.

Yet another means of detei_____ning position is to mount a position switch 50 to the static structure of the pumping unit in such a way as to allow detection of the passage of crank arm 52, giving an inferred indication of the position of polished rod 8 at one point in each pumping stroke. Because the motion of polished rod 8 is generally sinusoidal, and the period of the stroke is known, a good representation of the polished rod position is possible through analysis of the period of the stroke, the geometry of the pumping unit and the slip in the prime mover, motor 54.

The units selected to measure load and position are connected via transducer cables 56 to pump-off controller 58. Pump-off controller 58 analyzes the load and position data and determines pump-off. Upon detection of a pump-off condition, pump-off controller 58 issues an electronic signal to motor starter 60, which may be integrated with pumping unit 2. Upon receipt of such a signal, motor starter 60 may turn off motor 54. Alternatively, pumping unit 2 may be configured to switch control of motor 54 to a timer or other control mechanism (not shown).

Fig. 4 shows a system block diagram of the hardware and software within controller 58, including its inputs and outputs. Reference is also made to the high- level general logic flowchart shown in Fig. 5. Load-transducer signals 62 and position-transducer signals 64 pass to a signal-conditioning module 66, which in turn passes the signals to analog-to-digital converter 68 within controller 58, which translates the data into digital form for use by the computer. Controller 58 may be controlled by a user entering commands at an external keypad 70.

Control processor 72 supervises collection and storage of digitized load and position data from analog-to-digital converter 68. Control processor 72 also provides control of input and output functions 84. Control processor 72 can access data from data storage 88, which may include operating information, data presenting historic dynamometer information, and operating parameters input from keypad 70. The major function of control processor 72 is to provide a control signal to output control 90 to stop motor 54 (or otherwise transfer control) when a condition, such as pump- off, exists, and to send signals instructing a re-start of motor 54 when a user- programmed down-time has elapsed (or when control is otherwise returned to the pump-off controller). Control processor 72 maybe an off-the-shelf micro-controller integrated circuit such as those manufactured by Intel Corp. in the 8051 family. Math processor 74 uses load, time, and position data collected by control processor 72, together with information regarding the rod string and system damping contained in input data files 76 (stored in random access memory), to perform the downhole conversion according to a program 78 (stored in read-only memory). Math processor 74 may also perform calculations necessary to run fault- control algorithm 82. Math processor 74 may be an off-the-shelf integrated circuit microprocessor such as those manufactured by the Intel Corp. in the 80x86 series (8086, 80286, 80386, 80486). For additional speed in solving the wave equation for making the downhole conversion, a math coprocessor, such as those manufactured by the Intel Corp. in the 80x87 series may also be used.

Math processor 74 creates an output data file 80, also stored in random access memory, containing the converted transducer data intended to represent the downhole dynagraph. Math processor 74 then evaluates data from output data file 80 pursuant to a fault-detection control program 82, also stored in read-only memory. When a fault, such as pump-off, occurs, math processor 74 passes a violation signal to control processor 72, which increments a violation counter. For each stroke that a fault violation is not found, a reset command is passed to control processor 72 to reset the violation counter.

When control processor 72 determines that the violation counter has exceeded a user preset limit, it sends a signal to output control 92, which signals motor 54 to transfer operation. An optional communications port 94 may permit control processor 72 to inform the user of the fault in another fashion, for example by radio transmission to a central location. Control processor 72 also has access to control programs for input/output 84 and display control programs 86, which permits it to indicate the detection of a fault on an external display 90. In addition, control processor 72 maybe instructed by the user, through instructions entered on keypad 70, to store the results in data storage 88, which may be random access memory or permanent memory such as tape or disk storage, upon recognition of a fault.

Fig. 6 illustrates example shapes of the surface and downhole dynagraphs. At the surface, the dynagraph for a pump operating under normal conditions may appear as a highly irregular shape 96. By contrast, the calculated downhole dynagraph would look more like a regular, rectangular figure 110. The irregularity in the surface dynagraph results from the stretching of the rod string and delays in sensing changes in load in the downhole pump. In each of the dynagraphs illustrated in Fig. 6, the vertical axis indicates load, and the horizontal axis indicates position (up or down) relative to the casing. As the pump proceeds through its cycle, the load-position point moves in a clockwise direction around the dynagraph. For a well that is experiencing a fault condition such as pump-off, the downhole dynagraph changes its appearance 111. For example, as shown in Fig.3, a normally operating pump immediately transfers load on the downstroke from travelling valve 24 to standing valve 26. In a pumped-off situation, however, the pump moves downwards rapidly through vapor before contacting the fluid-vapor interface, which delays travelling valve 24 from opening. That change in position without decrease in load is indicated in Fig. 6 on downhole dynagraph 111 as a horizontal line moving to the left from near the top right corner of the figure. For a pumped-off well, the surface dynagraph also changes 102, although the change may be more difficult to recognize.

That easy-to-see alteration of downhole dynagraph 111 suggests tests for detecting faults such as pump-off. Two preferred techniques for detecting pump-off using downhole data are given below, illustrated in Fig. 6, and described further in the flowcharts of Figs. 7 and 8. Each method relies on calculating the area within all or a portion of the downhole dynagraph card, which corresponds to work done by the pump. A pumped-off well performs less work, as seen by comparing the areas within the figures on the top row of Fig. 6 with the corresponding areas within the figures on the bottom row. In the first method, the calculated area for each stroke is compared to the area within a calculated downhole dynagraph card 110, taken when the pump is expected to be filled, i.e., after the well is operational for long enough to begin pumping all fluid and no gas. If the calculated area drops below a preset percentage of the "full pump" dynagraph's area, as in card 111, the well is declared pumped-off.

Another possible test for pump-off, shown in the right-hand column of Fig. 6 and in Fig. 8, is to calculate the area within the lower portion of the downhole dynamometer card 116, which corresponds to the pump work done on the downstroke. An integration reference line 114 may be calculated by dividing the difference between measured maximum load and measured minimum load by a factor, either preset or manually input. In Figs. 6 and 8, that factor has been set at two. The area between reference line 114 and the bottom of the dynagraph is calculated for each stroke and the result is compared to a preset value. The preset value may be a percentage of the area of the full-pump dynagraph's downstroke work or an absolute, manually entered value. When the area is determined to be less than the preset value (compare dynagraph 115 to dynagraph 116), pump-off is declared.

Other fault-detection algorithms can be utilized to achieve similar results. By using downhole-analysis techniques to detect faults, the system can be calibrated easily and is more stable across wells with difference characteristics. The fault- recognition algorithm can be more simple inform than those developed for use on surface data.

To achieve that result, however, it is necessary to calculate the characteristics of the downhole dynagraph. A number of techniques are known and used in the oil industry to calculate the characteristics of the downhole dynagraph from surface load and position measurements. See U.S. Patent No.3,343,409, issued to Gibbs and incorporated above, and the article "The Design of Sucker Rod Pump Systems," by James W. Jennings, SPE Paper NMTECH 890012, also incorporated herein by reference. The downhole dynagraph may be obtained by mathematically manipulating the load, position, and time data measured at the surface of the well to determine the forces and displacements that would have been required at the pump to produce the measured surface data. The calculation assumes that the sucker rod string can be modelled as a long, slender, vibrating rod. The rod has a known weight and is subject to tensional forces. As the rod is reciprocated, the rod accelerates, but because it is elastic, the acceleration is not constant along the rod's length. Some energy is lost by viscous damping effects. Those characteristics maybe captured in a one-dimensional damped wave equation, shown in Gibbs' U.S. Patent No. 3,343,409 as equation (1). That equation is a second-order differential equation, which can be split into two first-order differential equations, one equation representing load and the other, position. When accompanied by knowledge of boundary conditions, the equations can be solved simultaneously. See Master's Thesis by D.J. Schafer (Texas A&M May 1987), particularly pages 5-12 and 18-21; Master's Thesis by T.A. Everitt (Texas A&M Dec. 1987), particularly pages 13-30; and Master's Thesis by M.J. Bastian (Texas A&M Dec. 1989), particularly pages 12-29, all of which are incorporated herein by reference.

Known numerical-analysis methods, including Fourier transforms and finite difference analysis, can be used to solve the differential equations to calculate the force-displacement curve at the pump. Software flowcharts illustrating those methods are shown in Figs. 9, 10, and 11. The shape of the downhole dynagraph can be calculated using those techniques or other wave-form-analysis techniques. Referring to Fig.9, measured surface load and position data points are transformed by the use of a standard discrete Fourier expansion into phase and amplitude terms. Meanwhile, stored data regarding the top-most discrete length, or "taper," of the rod string are analyzed by a damping formula to create coefficients for the Fourier series. The data include such information as the thickness, material composition, and length of the taper. The damping formula may be derived experimentally, such as explained in Gibbs' U.S. Patent No. 3,343,409 (although Gibbs assumes that the data are consistent throughout the rod string). Applying the coefficients to the transformed data results in the phase and amplitude of the set of points representing the load and position of the first taper as a function of time. The process is then iterated for subsequent, lower tapers, using the (possibly different) data stored for each taper. The result calculated for the bottom-most taper is the phase and amplitude of the set of points representing the load and position of the downhole pump over time. The program then converts the data back into load and position points at specific time intervals by performing an inverse Fourier transform. The results are then stored for analysis by the fault- detection algorithm.

Fig. 10 shows the so-called fast Fourier transform, a known algorithm for performing discrete Fourier analysis more efficiently. The number of load and position points are first counted and tested to see if they are equal to a power of two. If not, additional points are interpolated to make the total number of points equal to the next higher power of two. The fast Fourier transform algorithm permits a faster calculation by restricting the number of data points to factors of two. The analysis then proceeds, in the same fashion as described above, to work down through the tapers to the pump.

The finite differences analysis, illustrated in Fig.11, uses an iterative process. At the beginning of this procedure, the position data are differentiated to determine velocity as a function of time, and at the end of the procedure, the calculated velocity data are converted back to position data by integration. Load data are not similarly converted.

The finite differences analysis proceeds from the observation that it is possible to calculate the velocity and load at any segment in the rod string from knowledge of velocity and load of the immediately preceding segment and the characteristics of the current segment. Therefore, the rod string is divided into a number of segments of equal, arbitrary length, for example 150 meters per segment. The program then calculates the average physical properties for the segment, for each segment that contains rods of different sizes or materials, using the ratio of the length of each different rod to the overall length of the segment.

The method starts at the first segment and, using the average characteristics of that segment and the surface load and velocity data, calculates the velocity that would prevail at the bottom of the segment as a function of time. From that velocity data and the surface load data, it is possible to calculate the load at the bottom of the segment over time, taking into account the weight of the rod string below the segment, the fluid weight, and the acceleration, which alters the load. The process is then repeated for the next segment. The calculated velocity and load at the bottom of the first segment is used, together with the average characteristics of the second segment, to calculate the velocity and load functions at the bottom of the second segment. The process is iterated until the calculation is made for the last, bottom segment, which results in the downhole data.

To summarize, a preferred embodiment of the invention might operate as follows. Referring to Fig. 1, as pumping unit 2 operates, rod-mounted load transducer 30 and position transducer 40 collect data representing the load and position of polished rod 8, which is located at the surface. The data pass through transducer cables 56 to on-site pump controller 58, where they are conditioned and converted to digital form. Referring now to Figs. 4 and 5, control processor 72 collects a stroke's worth of such data and passes them to math processor 74 for analysis. Using pre-stored information about the characteristics of the various tapers of rod string 4, math processor 74 utilizes wave-form downhole-calculation program 78 to convert the surface load and position data to a stroke's worth of data representing the load and position at the pump, which are then stored in an output data file 80. Downhole-calculation program 78 might operate according to the finite differences method illustrated in Fig. 11.

Returning to Figs.4 and 5, math processor 74 evaluates a stroke's worth of downhole data stored in data file 80 using control calculation 82, which tests for the existence of a fault, such as pump-off. That test might be the downstroke integration method illustrated in the right column of Fig.6 and in Fig.8, in which the portion of the card below an integration reference (load) limit 114 is integrated. If the integral exceeds a preset value, the well is considered not pumped off, and a reset signal is sent to control processor 72. Otherwise, control processor 72 increments a fault counter, which, when it exceeds a user-selected limit, causes control processor 72 to signal a fault, as shown in Fig.5. The user might opt to have control processor 72, upon recognition of a fault, issue a command to motor controller 60 (in Fig. 1) to shut down motor 54 for a preset time.

Although the preferred embodiments have been described above, other types of fault-testing or numerical conversion algorithms can be utilized. Furthermore, numerous forms of programming can be used to carry out the specific command and control of the computers. For example, it is envisioned that state-machine programming could be employed. Also, the algorithms for fault recognition or downhole conversion could be implemented in signal processors or other dedicated integrated circuits, as well as computer programs. In addition, the functions of control processor 72 and math processor 74 might be distributed among one, two, or more integrated circuits or other hardware or software devices. Thus, it will be understood by those skilled in the art that numerous alternate forms and embodiments of the invention can be devised without departing from its spirit and scope.

Claims

It is claimed:

1. A system for detecting faults in a well with a lifting apparatus that includes a prime mover coupled to a subterranean pump by a sucker-rod string comprising: (a) first sensing means for generating a load signal representative of load on the sucker rod at the surface of the well;

(b) second sensing means for generating a position signal representative of the position of the sucker rod at the surface of the well;

(c) circuit means coupled to the first and second sensing means for conditioning and converting the load and position signals into digital data;

(d) memory means coupled to the circuit means for storing the digital data representative of surface load and position and for storing data defining the physical characteristics of the pumping system;

(e) computer means coupled to the memory means and circuit means and including a conversion program for using the stored surface load and position data to mathematically infer a downhole dynagraph representative of load and position at the downhole pump;

(f) control means coupled to the computer means and including a fault-detection program for evaluating the downhole dynagraph inferred by the computer means to detect a fault condition in the downhole portion of the lifting apparatus and for issuing a control instruction in response to recognition of a fault condition; and

(g) means for applying the control instruction to alter the operation of the lifting apparatus in response to recognition of a fault condition. 2. The system of claim 1 wherein the computer means and conversion program utilizes mathematical wave-form analysis techniques to infer the downhole dynagraph.

3. The system of claim 1 wherein the computer means includes a first processing means for controlling the data and a second mathematical processing means for calculating the downhole load and position.

4. The system of claim 1 wherein the fault-detection program includes means for detecting pump-off.

5. The system of claim 1 wherein the control means includes means for detecting pump-off by comparing an area contained within at least a portion of the downhole dynagraph inferred by the computer means with an area limit set to represent the minimum acceptable level. 6. A method of detecting pump-off in a well containing a prime mover coupled to a subterranean pump by a sucker-rod string comprising the steps of:

(a) measuring the surface load on the sucker rod at the surface of the well at a series of time intervals and generating signals representative thereof;

(b) measuring the surface position of the sucker rod at the surface of the well at a series of time intervals and generating signals representative thereof;

(c) conditioning and converting the signals representative of the surface load and position of the sucker rod to generate digital surface load- position data pairs;

(d) processing the digital surface load-position data to create a downhole dynagraph representative of the downhole load at corresponding positions of the subterranean pump; (e) comparing the downhole dynagraph to data stored in the memory of a computer to detect a pumped-off condition; and

(f) controlling the operation of the pump upon recognition of a pumped-off condition.

7. The method of claim 6 wherein the processing step uses mathematical wave-form analysis techniques.

8. The method of claim 6 wherein the step of detecting a pumped-off condition includes integrating a portion of the downhole dynagraph to represent the work done over at least a portion of a stroke and comparing the result of the integration with a work limit set to represent the minimum acceptable level. 9. A method for detecting faults in a sucker-rod pumping system having surface equipment including a motor and a rod string coupled to and reciprocated by the motor, and downhole equipment including a fluid pump coupled to and driven by the rod string, the method comprising:

(a) using a load sensor to generate an analog signal that varies in relation to a varying surface load on the rod string as the rod string reciprocates;

(b) using a position sensor to generate an analog signal that varies in relation to a varying surface position of the rod string as the rod string reciprocates; (c) conditioning the load and position signals to convert and synchronize the signals into digital load-position data pairs; (d) storing in memory accessible by a computer controller:

(i) data defining the physical characteristics of the pumping system;

(ϋ) data representative of pre-defined fault conditions; (iii) at least one conversion program capable of performing a downhole conversion of the surface load and position data based on the data defining the physical characteristics of the pumping system to generate a corresponding downhole dynagraph defining the load at corresponding positions at the downhole pump, and (iv) at least one fault-detection program capable of analyzing the downhole dynagraph and comparing it to the stored data representative of predefined fault conditions to detect faults; (e) as the sucker-rod system is operating, storing the digital surface load and position pairs in memory accessible by the controller; (f) implementing the conversion program and using the data stored in memory defining the physical characteristics of the pumping system to perform a downhole conversion of the digital surface load and position pairs to generate a downhole dynagraph representing the load at corresponding positions of the downhole pump; (g) storing the output file defining the downhole dynagraph in memory accessible by the controller;

(h) implementing the fault-detection program to analyze the downhole dynagraph and to compare it to the stored data representative of pre-defined fault conditions to detect and classify a current fault condition in the downhole equipment;

(i) generating a control signal in response to a detected fault condition in the downhole equipment; and

(j) applying the control signal to alter the operation of the surface equipment in response to recognition of a fault condition. 11. The method of claim 9 wherein the step of implementing the fault-detection program to analyze the output file comprises the steps of:

(a) calculating an area contained within at least a portion of the dynagraph; and

(b) comparing the area to a limit set to represent the minimum acceptable level. 12. The method of claim 9 wherein the steps of implementing the conversion program and the fault detection program include using a math processor to assist in performing the downhole conversion and to analyze the downhole dynagraph. 13. An apparatus for detecting faults in a sucker-rod pumping system, the system having surface equipment including a rod string coupled to and reciprocated by a driving motor, and downhole equipment including a fluid pump coupled to and driven by the rod string, the apparatus comprising:

(a) load sensing means coupled to the surface equipment of the system for generating an analog signal that varies in relation to the varying surface load on the rod string as the rod string reciprocates;

(b) position sensing means coupled to the surface equipment of the system for generating an analog signal that varies in relation to the varying surface position of the rod string as the rod string reciprocates; (c) signal conditioning means coupled to the output of the load and position sensing means for synchronizing the load and position signals and for converting the synchronized signals into digital and load-position data pairs;

(d) memory means for storing data, including data defining the physical characteristics of the pumping system and data representative of predefined fault conditions;

(e) program storage means for storing programs, including at least one conversion program capable of performing a downhole conversion of the surface load and position data based on the data defining the physical characteristics of the pumping system, and at least one fault-detection program;

(f) computer means coupled to the signal conditioning means, the memory means, and the program storage means for implementing the conversion program and using the data stored in memory defining the physical characteristics of the pumping system to perform a conversion of the digital surface load-position data to generate an output file defining a downhole dynagraph of the load at corresponding positions of the downhole pump;

(g) the computer means further including means for implementing the fault-detection program to analyze the output file defining the downhole dynagraph and to compare it to the stored data representative of predefined fault conditions to detect a current fault condition in the downhole equipment;

(h) control means for issuing a control signal in response to a detected fault condition in the downhole equipment; and (i) means for applying the control signal to alter the operation of the surface equipment.

14. The system of claim 1 wherein the means for applying the control instruction includes means for stopping the prime mover in response to recognition of a fault condition. 15. The method of claim 9 wherein the step of applying the control signal includes stopping the prime mover in response to recognition of a fault condition.

16. The system of claim 14 further comprising timer means for restarting the prime mover after a predetermined time period has elapsed.

17. The method of claim 15 further comprising the step of restarting the prime mover a predetermined time after recognition of a fault condition.

18. The system of claim 1 wherein the means for applying the control instruction includes means for switching, upon recognition of a fault condition, to a second control means for controlling operation of the lifting apparatus.

19. The method of claim 9 wherein the step of applying the control signal includes switching control of the lifting apparatus to a second controller upon recognition of a fault condition.

20. The apparatus of claim 9 wherein the steps of applying the control instruction includes altering operation of the surface equipment in response to receipt of a plurality of control instructions indicating detection of a plurality of consecutive fault conditions.