By:

Edwin Sutrisno

Citizenship: Indonesia

Wenyu Zhao

Citizenship: China

Nicholas Dane Williard

Citizenship: USA

Gilbert Haddad

Citizenship: Lebanon

Neil Holger Whiter Eklund

Citizenship: USA

ELECTRIC SUBMERSIBLE PUMP HEALTH ASSESSMENT

CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] The present application claims priority to U.S. Provisional Application No. 62/253,616, which was filed on November 10, 2015, is entitled "Analytical Approach for Electrical Submersible Pump Health Assessment and Remaining Useful Life Prediction," and is incorporated herein by reference in its entirety for all purposes.

STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT

[0002] Not applicable.

BACKGROUND

[0003] Electric submersible pumps (ESPs) are typically deployed in cased wells to artificially lift substances (e.g., hydrocarbons) from the surrounding formation to the Earth's surface. ESPs are especially useful in applications where the substances do not flow responsive to existing natural forces. If an ESP fails to operate properly, however, significant costs begin to accrue— due, for instance, to production downtime. Accordingly, the ability to predict or at least anticipate an ESP failure by detecting early warning signs provides an operator with the ability to repair or replace the ESP in a timely fashion, thus mitigating downtime costs.

SUMMARY

[0004] At least some embodiments are directed to a method, comprising: obtaining a plurality of parameters from multiple electric submersible pump (ESP) systems, an ESP in each of said ESP systems meeting one or more predetermined health criteria; identifying one or more relationships between the parameters in said plurality of parameters; generating a model based on said identification; obtaining a second plurality of parameters from another ESP system; determining a predicted value using said model and the second plurality of parameters; comparing the predicted value and an actual value from said another ESP system to determine a difference; and displaying a health status of another ESP in said another ESP system on a display based on said difference. One or more such embodiments may be supplemented using one or more of the following concepts, in any order and in any combination: wherein said actual value comprises a measured value; further comprising removing from said plurality of parameters any parameter meeting one or more criteria selected from the group consisting of: a parameter that is determined to be a statistical outlier, a parameter obtained when the corresponding ESP system is in an operating condition that fails to meet an operating state criterion, a parameter obtained at a time when a predetermined, related parameter is unavailable, a parameter obtained when said ESP in the corresponding ESP system is not powered, a parameter obtained when said ESP in the corresponding ESP system is in a warm-up period following a start, and a parameter failing to meet a predetermined quality criterion, wherein said removal is performed after obtaining the plurality of parameters and prior to identifying said one or more relationships; further comprising removing from said plurality of parameters any parameter that is not usable to characterize the health status of an ESP; wherein the multiple ESP systems include at least ten ESP systems; further comprising: performing a mathematical operation on one or more of the plurality of parameters to produce a calculated parameter, and including said calculated parameter in said plurality of parameters when performing said identification; wherein said generation comprises using one or more modeling algorithms; further comprising determining a healthy distribution range corresponding to the predicted value, and further comprising determining said health status based on a comparison of the healthy distribution range and the difference; further comprising repairing or replacing the another ESP based on said health status.

[0005] At least some embodiments are directed to a method that comprises obtaining a plurality of parameters from multiple electric submersible pump (ESP) systems, an ESP in each of said ESP systems meeting one or more predetermined health criteria; identifying one or more relationships between the parameters in said plurality of parameters; generating a model based on said identification; obtaining a second plurality of parameters from another ESP system; determining a predicted value using said model and the second plurality of parameters; obtaining a third plurality of parameters from multiple other ESP systems; determining a set of predicted values using said model and the third plurality of parameters; comparing the predicted value with the set of predicted values to identify one or more differences; and displaying a health status of said another ESP system based on said one or more differences. One or more such embodiments may be supplemented using one or more of the following concepts, in any order and in any combination: wherein said another ESP system and said multiple other ESP systems share a common reservoir; further comprising removing from said plurality of parameters any parameter meeting one or more criteria selected from the group consisting of: a parameter that is determined to be a statistical outlier, a parameter obtained when the corresponding ESP system is in an operating condition that fails to meet an operating state criterion, a parameter obtained at a time when a predetermined, related parameter is unavailable, a parameter obtained when said ESP in the corresponding ESP system is not powered, a parameter obtained when said ESP in the corresponding ESP system is in a warm-up period following a start, and a parameter failing to meet a predetermined quality criterion, wherein said removal is performed after obtaining the plurality of parameters and prior to identifying said one or more relationships; further comprising removing from said plurality of parameters any parameter that is not usable to characterize the health status of an ESP; wherein said identifying one or more relationships comprises identifying one or more weights associated with said one or more relationships.

[0006] At least some embodiments are directed to a system, comprising: an electric submersible pump (ESP) system positioned within a well and configured to lift hydrocarbons from said well to the Earth's surface; one or more sensors coupled to the ESP system and configured to obtain multiple parameters associated with the ESP system; and a processor configured to receive said multiple parameters and to determine a predicted value for the ESP system based on a model relating said multiple parameters to each other, wherein the processor is further configured to determine a difference between the predicted value and a measured value and to indicate a health status of an ESP in the ESP system on a display based on said difference. One or more of these embodiments may be supplemented using one or more of the following concepts, in any order and in any combination: wherein the processor is configured to determine a second predicted value based on a second model, to determine a difference between said second predicted value and a second measured value, and to combine results of both of said differences, and wherein said indication of the health status of the ESP is based on said combination; wherein the processor is configured to generate an alert signal if the health status of the ESP meets a predetermined criterion; wherein the multiple parameters are selected from the group consisting of: a pressure measured in the ESP system; a temperature measured in said ESP system; a vibration measured in the ESP system; a wellhead pressure; a casing pressure; and a tubing pressure; wherein the multiple parameters are selected from the group consisting of: a pressure measured at an intake of the ESP system; a pressure measured at a discharge of the ESP system; a temperature measured within said well; a temperature measured at said intake; and a temperature measured at a motor of the ESP system; wherein the multiple parameters are selected from the group consisting of: an electrical voltage sent downhole to the ESP; an electrical current sent downhole to the ESP; an electrical frequency sent downhole to the ESP; an electrical voltage recorded at a surface power supply system; an electrical current recorded at said surface power supply system; an electrical frequency recorded at said surface power supply system; and a current leaked from a component of the ESP.

BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Embodiments of the disclosure are described with reference to the following figures:

[0008] Figure 1 is a schematic diagram of an electric submersible pump (ESP) system.

[0009] Figure 2 is a schematic diagram of an ESP.

[0010] Figure 3 is a block diagram of networked processing devices that work alone or in tandem to perform some or all of the techniques described in this disclosure.

[0011] Figure 4 is a flow diagram of an ESP system modeling and monitoring process.

[0012] Figure 5 is a flow diagram of a parameter filtering process.

[0013] Figure 6 is a flow diagram of an additional parameter filtering process.

[0014] Figure 7 is a flow diagram of a modeling process.

[0015] Figure 8 is a diagram representing learned relationships between multiple parameters.

[0016] Figure 9 is a flow diagram of an ESP system monitoring process.

[0017] Figure 10 is a flow diagram of an alternative process to evaluate the health of an ESP.

DETAILED DESCRIPTION

[0018] One or more embodiments of the present disclosure are described below. These embodiments are merely examples of the presently disclosed techniques. Additionally, in an effort to provide a concise description of these embodiments, all features of an actual implementation may not be described in the specification. It should be appreciated that in the development of any such implementation, as in any engineering or design project, numerous implementation-specific decisions are made to achieve the developers' specific goals, such as compliance with system- related and business-related constraints, which may vary from one implementation to another. Moreover, it should be appreciated that such development efforts might be complex and time consuming, but would nevertheless be a routine undertaking of design, fabrication, and manufacture for those of ordinary skill having the benefit of this disclosure. [0019] When introducing elements of various embodiments of the present disclosure, the articles "a," "an," and "the" are intended to mean that there are one or more of the elements. The embodiments discussed below are intended to be examples that are illustrative in nature and should not be construed to mean that the specific embodiments described herein are necessarily preferential in nature. Additionally, it should be understood that references to "one embodiment" or "an embodiment" within the present disclosure are not to be interpreted as excluding the existence of additional embodiments that also incorporate the recited features. The drawing figures are not necessarily to scale. Certain features and components disclosed herein may be shown exaggerated in scale or in somewhat schematic form, and some details of conventional elements may not be shown in the interest of clarity and conciseness.

[0020] The terms "including" and "comprising" are used herein, including in the claims, in an open-ended fashion, and thus should be interpreted to mean "including, but not limited to... ." Also, the term "couple" or "couples" is intended to mean either an indirect or direct connection. Thus, if a first component couples or is coupled to a second component, the connection between the components may be through a direct engagement of the two components, or through an indirect connection that is accomplished via other intermediate components, devices and/or connections. If the connection transfers electrical power or signals, the coupling may be through wires or other modes of transmission. In some of the figures, one or more components or aspects of a component may be not displayed or may not have reference numerals identifying the features or components that are identified elsewhere in order to improve clarity and conciseness of the figure.

[0021] Disclosed herein are various techniques to monitor the health of electric submersible pumps (ESPs) and to anticipate or predict failure of the ESPs so that preemptory action can be taken to repair or replace the ESPs prior to failure. Such preemptory action mitigates losses associated with ESP production downtime and damage that cannot be repaired. The techniques generally include using sensors positioned in a healthy, properly-functioning ESP system to collect various parameters pertaining to the ESP system— for instance and without limitation, pressures, temperatures, vibrations, voltages, and currents— and filtering the parameters to remove data points that are flawed or not useful to characterize the health of the ESP. Additional parameters may be then calculated using the set of parameters to expand the set of parameters. After a final set of parameters has been produced, one or more modeling algorithms are applied to the set of parameters to identify relationships between the parameters. For example and without limitation, such algorithms may determine whether one parameter is a function of another parameter; whether a parameter is affected by numerous other parameters; the degree to which each parameter affects another parameter, and how such relationships change over time. These relationships may be used to generate one or more models of the ESP system, with each such model characterizing the behavior of one or more portions of the ESP system.

[0022] The model may be refined over time using multiple sets of parameters, either from the same healthy ESP, one or more other healthy ESPs, or a combination thereof. Because the model is generated and trained using healthy ESPs, they indicate the behavior expected of a healthy ESP. Accordingly, when numerical values are applied to the model, the model produces output parameters that would be expected of a healthy ESP. If the model is being used to test a particular ESP and the ESP behaves in a manner significantly different than predicted by the trained model, that ESP may be defective and may require intervention in the form of repair or replacement. Because such intervention is performed prior to failure of the ESP, production downtime is comparatively reduced and damage to reusable parts is mitigated. Various embodiments of this technique are now described with reference to Figures 1-10.

[0023] Figure 1 is a schematic diagram of an electric submersible pump (ESP) system 100. The ESP system 100 includes a well 103 disposed in a formation; a controller 130 in communication with the well 103 and, more specifically, with equipment within the well 103; an ESP motor controller 150 coupled to the controller 130; a variable speed drive (VSD) unit 170 coupled to the controller 130; and a power supply 105 coupled to the controller 130. The power supply 105 may receive power from a power grid, an onsite generator (e.g., a natural gas-driven turbine), or another appropriate source. The ESP system 100 may communicate with a network 101 of other co-located and/or distributed computers, ESP systems, mobile devices, and the like.

[0024] Disposed within the well 103 is an ESP 110 that includes cables 111 ; a pump 112; gas handling features 113; a pump intake 1 14; a motor 115; and a plurality of sensors 116. The well 103 additionally comprises multiple well sensors 120, some of which may be located in various parts of the well (e.g., disposed within a cement casing or positioned in the wellhead). The sensors 116 may be positioned inside the ESP 1 10, outside the ESP 110, or both and may be configured to sense and measure a variety of parameters. Illustrative, non-limiting parameters that may be sensed and measured by the sensors 116 include: pressures measured in and around the ESP 110, including at pump intake, within the ESP 1 10, and at pump discharge; temperatures measured in and around the ESP 110, including in the annular space between the ESP 110 and the well casing, at pump intake, and at the pump motor; and vibrations detected within or outside the ESP 110 or within or adjacent to one or more of the ESP's components. The scope of disclosure is not limited to these or any other set of parameters. For instance, the sensors 116 may additionally sense and measure electrical voltages, currents, and frequencies sent downhole to the ESP 110; electrical voltages, currents, and frequencies recorded at the output of the drive_or within a power supply system at the wellhead; and current leakages measured within or adjacent to the ESP 110. The scope of this disclosure generally encompasses any and all suitable sensors 1 16 that could conceivably be positioned within, on, or adjacent to the ESP 1 10 for the purpose of sensing and measuring data. Data sensed and measured by the sensors 116 may be stored in any suitable location— for instance, locally within the sensors 116 or in storage at the surface.

[0025] At least some commercially available sensors are representative of some of the sensors 116. One commercially available sensor is the PHOENIX MULTISENSOR® marketed by SCHLUMBERGER LIMITED® (Houston, Texas), which monitors intake and discharge pressures; intake, motor and discharge temperatures; and vibration and current leakage. An ESP monitoring system may include a supervisory control and data acquisition system (SCADA). Commercially available surveillance systems include the ESPWATCHER® and the LIFTWATCHER® surveillance systems marketed by SCHLUMBERGER LIMITED® (Houston, Texas). These commercially available sensors and systems are merely illustrative and do not limit the scope of technologies that may be used to sense and measure various parameters in and around the ESP 1 10.

[0026] The well sensors 120 may be positioned in any suitable location(s) throughout the well 103. For instance and without limitation, well sensors 120 may be disposed within, on, or adjacent to the wellhead; within the casing of the well 103; within the formation surrounding the casing; and on the inner surface of the casing. Well sensors 120 may extend thousands of feet into a well (e.g., 4,000 feet or more) and beyond the position of the ESP 110. The well sensors 120 may sense and measure any and all suitable parameters relating to the well 103. Illustrative, non-limiting examples of such parameters include pressures measured at locations in and around the wellhead including the wellhead pressure, casing pressure, and tubing pressure. Non-limiting examples of some commercially available well sensors 120 include OPTICLINE® sensors or WELLWATCHER BRITEBLUE® sensors marketed by SCHLUMBERGER LFMITED® (Houston, Texas). Such well sensors are fiber-optic based and can provide for real-time sensing of downhole conditions. Parameters sensed and measured by the well sensors 120 may be stored in any suitable location— e.g., locally in the sensors 120 or at the surface.

[0027] Figure 2 shows a simplified schematic of an exemplary and non-limiting ESP 110. In this example, observable parameters such as electro-mechanical data related to the ESP 110 may be acquired during a normal mode of operation. In certain cases, the observable parameters may be obtained in a controlled environment to determine a manifold or envelope of the normal mode of operation of the ESP 110. As shown in Figure 2, the ESP 1 10 includes two motors, lower tandem (LT) motor 202 and upper tandem (UT) motor 204; two protectors, LT protector 206 and UT protector 208; and two pumps, labeled LT pump 210 and UT pump 212, all on a common shaft 214. Although two of each component is depicted in Figure 2, other embodiments are contemplated, including an ESP 1 10 comprising one motor, one protector, and one pump. In one non-limiting example, the observed parameter space comprises a plurality of parameters, such as surface flow rate, pump inlet/discharge pressures, motor temperatures, protector temperatures, motor lead temperatures, vibration along various axes, power consumption, and the like.

[0028] Referring again to Figure 1, the variable speed drive (VSD) unit 170 couples to the motor 115 and drives the motor at a rate dictated by the controller 130. The controller 130, in turn, couples to and receives motor control direction from the ESP motor controller 150. The ESP motor controller 150 may be a commercially available motor controller such as the UNICONN® motor controller marketed by SCHLUMBERGER LFMITED® (Houston, Texas). The UNICONN® motor controller can connect to a SCADA system, the ESPWATCHER® surveillance system, etc. The UNICONN® motor controller can perform some control and data acquisition tasks for ESPs, surface pumps, or other monitored wells. The UNICONN® motor controller can interface with the PHOENIX® monitoring system, for example, to access pressure, temperature, and vibration data and various protection parameters as well as to provide direct current power to downhole sensors. The UNICONN® motor controller can interface with fixed speed drive (FSD) controllers or a VSD unit— for example, the VSD unit 170.

[0029] As shown in Figure 3, the controller 130 comprises any hardware and/or software suitable for controlling one or more components of the ESP system 100. Specifically, in at least some embodiments, the controller 130 comprises a processing device 190 coupled to storage 304 (e.g., random access memory, read-only memory) storing executable code 306 (e.g., software, firmware), display 302, and an input 300 (e.g., keyboard, mouse, touchscreen). The controller 130 couples to one or more remote computers 308 via network 101 (Figure 1), an example of which is shown in Figure 3. In particular, as shown in Figure 3, the computer 308 includes a processing device 310 couples to storage 316 storing executable code 318. The processing device 310 further couples to an input(s) 312 and a display 314. The processing device 310 and related equipment communicate with the processing device 190 via wired and/or wireless connection(s) 320 to facilitate the performance of one or more of the techniques disclosed herein.

[0030] In operation, the controller 130 receives parameters collected by the various well sensors 120 and ESP sensors 116. The controller 130 may receive such parameters directly from the sensors (e.g., through a wired or wireless connection), or the controller 130 may receive such parameters via an indirect route (e.g., through a separate controller located in or near the wellhead that communicates directly with one or more sensors). The controller 130 may collect such parameters as the result of executing code stored within the controller 130 or external to the controller 130 (e.g., in another co-located controller or in a remote computer accessible via the network 101). In addition to collecting sensed parameters from sensors 1 16 and 120, the controller 130 may receive other types of parameters, such as the ESP motor 115 power rating; the number of stages in the pump 112; the physical dimensions of one or more components of the ESP 110; the installation depth of the ESP 1 10; and the petrophysical properties of the well 103, such as gas content, water content, and whether the well 103 is in a shale formation, is hydraulically fractured, or is assisted by an injector well. Additional parameters received by the controller 130 may include whether various valves in the ESP 110 are open, closed, or partially open and to what degree; data regarding the choke opening (e.g., in a choke at the wellhead); number of days that the ESP 110 has been running, both continuously and in total; and operating mode of the ESP 1 10, such as intake pressure mode, speed mode, or discharge pressure mode. Other non-sensed parameters are contemplated and included within the scope of this disclosure. The controller 130 may obtain such non-sensed parameters from a data repository (e.g., a storage coupled to the controller 130, a computer remotely communicating with the controller 130 via the network 101), or the controller 130 may be provided the non-sensed parameters by an operator using appropriate input devices to the controller 130. Upon collecting the parameters, the controller 130 processes the parameters as now described with reference to Figures 4-10. [0031] Figure 4 is a flow diagram of an ESP system modeling and monitoring process 400. The flow diagram in Figure 4 depicts an overall process and Figures 5-9 describe each of the steps depicted in Figure 4 in greater detail. Accordingly, Figures 5-9 are described in parallel with Figure 4. The process 400 begins by filtering the obtained parameters (steps 402 and 404). Because these parameters will be used to generate a model of a healthy, properly-functioning ESP, the parameters themselves should be obtained from one or more healthy, properly-functioning ESPs (e.g., in some embodiments, at least ten ESPs from different ESP systems; in some embodiments, on the order of hundreds of ESPs from different ESP systems). Whether an ESP is healthy and properly functioning may be determined by an operator of the ESP. One purpose of the filtering process in steps 402 and 404 is to remove parameters that are defective or unreliable (step 402) or not useful (step 404). The parameter censoring of step 402 is described in greater detail with regard to process 500 of Figure 5. The process 500 begins by determining whether the parameter being evaluated is a statistical outlier (step 504). Any suitable statistical technique may be used for this step. For instance and without limitation, prior parameter values for the same parameter type may be analyzed to determine the 95% or 98% confidence intervals (based on calculated standard deviations from the mean) for parameter values of that parameter type. If the parameter in question falls outside one of these confidence intervals or some other appropriate range, that parameter may be discarded from the parameter set (step 516). Step 506 comprises determining whether the parameter in question was obtained while the ESP 110 was in an inappropriate operating condition or mode. For instance, if the parameter in question was collected from the ESP when the ESP was operating in a mode such that the parameter would not indicate deteriorating performance by the ESP, that parameter may be removed from the set (step 516).

[0032] Still referring to Figure 5, step 508 comprises determining whether the parameter in question was obtained when a specific, related parameter was unavailable. For instance, if a parameter of a first type has little or no value in observing and predicting ESP behavior and health unless a parameter of a second type is available, and further if the parameter of the second type is unavailable, then the parameter of the first type may be discarded (step 516). Next, the process 500 comprises determining whether the parameter in question was obtained when the ESP 110 was powered off or in a start-up mode after having been powered off. In either of these cases, the parameter in question may be discarded (step 516). Finally, the process 500 comprises determining whether the parameter in question fails to meet a predetermined quality criterion (step 512). The particular set of criteria considered in step 512 is application-specific and may be defined as appropriate. If the parameter in question fails to meet a quality criterion, the parameter in question may be discarded (step 516). The determinations described above with respect to steps 504, 506, 508, 510 and 512 are merely illustrative. These determinations may be modified or removed, and other determinations may be added. Regardless of the specific set of determinations used, if the parameter in question is not discarded as a result of the determinations, that parameter remains in the parameter set (step 514).

[0033] Referring again to Figure 4, the method 400 next comprises filtering the parameter set by removing parameters that are not useful in evaluating or predicting ESP health (step 404). This step is described with respect to the process 600 of Figure 6. The process 600 begins by determining whether the parameter in question is potentially useful to identify a deteriorating ESP (step 602). In most cases, the parameter in question is able to help identify a deteriorating ESP— otherwise, that parameter would not have been collected. Nevertheless, certain types of collected parameters— for instance, parameters collected regarding the marking or lettering printed on ESP equipment— are of little to no value in evaluating ESP health and are thus removed from the set (step 608). The specific types of parameters that should be removed in this instance may be programmed into the controller 130. The process 600 next includes determining whether the parameter in question is useful in modeling other parameters (step 604). Stated another way, the process determines whether the parameter in question affects other parameters and thus is useful in generating one or more parameter models. If not, the process 600 comprises removing that parameter from the set (step 608). Otherwise, the parameter in question remains in the set (step 606).

[0034] Referring again to Figure 4, as a result of completing steps 402 and 404, the obtained parameters are filtered so that the parameters remaining in the parameter set are suitable for use in the remainder of the process 400. Step 406 comprises using the parameter set to generate one or more models characterizing the behavior of the ESP system 100. The modeling process of step 406 and the related step 408 of model fusion are described in greater detail with regard to Figure 7.

[0035] The process 700 of Figure 7 begins by collecting filtered parameters from ESPs that are in various operating modes (step 702). An ESP's behavior may differ significantly depending on the ESP's operating mode. Parameters obtained in different ESP operating modes may have different implications for the health of the ESP. Thus, when building models to characterize an ESP's behavior, it is useful to obtain healthy parameter values from a variety of ESP operating modes so that a healthy parameter value is not regarded as unhealthy merely because it was collected in a different ESP operating mode. The process 700 subsequently comprises mathematically manipulating the set of parameters to determine new parameters (step 704). Stated another way, one or more parameters from the set produced by step 702 may be combined in any appropriate fashion to produce one or more new parameters. The precise manner in which such manipulations may occur is application-specific and is to be determined by, e.g., the ESP operator. A non-limiting, illustrative manipulation may include a mathematical transformation (e.g., time- frequency transformation) using one or more parameters.

[0036] Still referring to Figure 7, the process 700 further includes using one or more modeling algorithms to identify the behavior of various parameters and to identify relationships between the various parameters (step 706). Regression analysis is one well-known technique for identifying such behaviors and relationships between parameters, although other techniques are contemplated and fall within the scope of this disclosure. Figure 8 depicts illustrative relationships identified between eight parameters in a set of parameters. The set of parameters 802 on the left is identical to the set of parameters 802 on the right; the set 802 is depicted twice for clarity. Arrows 804 depict the relationships between parameters. Thus, for instance, a regression analysis may determine that Parameter 1 only affects Parameter 5, that Parameter 4 affects only Parameters 2 and 6, that Parameter 7 is a function of Parameters 5 and 6, etc. In some embodiments, the weights of these relationships also may be identified using linear regression techniques— for instance, it may be determined that although both Parameters 5 and 6 affect Parameter 7, Parameter 5 has a significantly greater impact on Parameter 7 than does Parameter 6.

[0037] After relationships between the parameters in the set have been identified, the process 700 comprises generating one or more models based on the identified behaviors and relationships (step 708). Models are statements or expressions of the identified behaviors or relationships between parameters in the parameter set. In the context of Figure 8, for example, models express the relationships depicted by arrows 804 and, optionally, they express the weights and other information associated with those relationships. Models may be expressed in any suitable form, such as in the form of matrices, equations, functions, computer algorithms, or other mathematical expressions. An anomaly detection algorithm (e.g., k-nearest neighbor; rule-based) may then be used on the resulting model(s) to identify a healthy parameter distribution range so that when the parameters of an ESP being tested are applied to the model, the model's output can be compared to the healthy distribution range to determine whether a potential problem exists (step 710). A "healthy" distribution range may be defined as desired, but, in at least some embodiments, a healthy distribution range may have a 90% confidence interval, a 95% confidence interval, or a 98%) confidence interval. Other confidence intervals are contemplated. Finally, in step 712, multiple models may optionally be combined, or "fused," to reduce the risk associated with relying on the results of a single model. Such combination may be accomplished in any suitable manner. For instance, in some embodiments, logical operators (e.g., AND, OR) may be used to fuse the models, or a variety of suitable statistical techniques (e.g., fuzzy logic; majority voting) may be used to combine the outputs of multiple models.

[0038] At any point in time after a model has been generated in the process 700, that model (or combination of models) may be trained using historical data associated with ESPs with known failure dates. For example, parameters that were collected for numerous old ESPs may be applied to the models to determine whether the models accurately demonstrate ESP deterioration. Similarly, the deterioration in historical ESP parameters may be correlated to the failure date of the historical ESPs. This correlation may be used to train the models so that the models are usable to accurately predict an ESP failure date based on the degree of deterioration suggested by statistical aberrations in the various parameter values. Other techniques for training the models to help predict ESP failure dates based on parameter values are contemplated.

[0039] Referring again to Figure 4, the final step in process 400 is the use of one or more models to monitor one or more ESPs for undesirable behaviors that may indicate impending failure (step 410). Step 410 is described in greater detail with regard to Figure 9, which depicts a process 900 for monitoring an ESP. The process 900 begins by collecting new parameters from an ESP to be monitored (step 902) and optionally calculating one or more additional parameters using one or more of the collected parameters (step 904). Parameters may be collected using any of the techniques described above— for instance, using sensors or receiving parameters as input directly from a human user. Although not specifically depicted in Figure 9, the collected parameters also may be filtered as described above with respect to steps 402 and 404 of Figure 4. The process 900 then comprises applying the resulting set of parameters to one or more models of the ESP to obtain predicted value(s) (step 906). The process 900 then comprises comparing the predicted parameters to the actual parameters of the same type that are measured from the ESP or are otherwise obtained (step 908). For instance, a predicted pressure at ESP intake would be compared to a measured pressure at ESP intake.

[0040] The differences in parameters are then evaluated (step 910). In some embodiments, such evaluation comprises combining the differences (e.g., by summing or averaging the values). In some embodiments, such evaluation comprises keeping the parameters separate. In some embodiments, such evaluation comprises evaluating the parameters in parallel. In some embodiments, the differences (or value(s) computed using the differences) are compared to healthy distribution ranges associated with the predicted values (e.g., 90% confidence intervals, 95% confidence intervals, 98% confidence intervals) to determine the extent to which the differences are aberrant from predicted values. In some embodiments, one or more of these evaluation techniques may be combined, or other evaluation techniques may be used. In all such embodiments, the evaluation process produces an overall health status of the ESP (step 910). If the health status identified in step 910 meets a predetermined criterion (or criteria) for generating an alert signal (step 912), the alert signal is generated (step 914). The predetermined criterion may, for instance, query whether the calculated differences exceed a healthy distribution range (e.g., two standard deviations from the mean). At any step of the method 900, parameters (predicted or actual), the manipulation of parameters, differences between parameters, health statuses, etc. may be provided on a display such as the display 302 or display 314 (Figure 3). In response to viewing the health status or other relevant information, an operator or other entity may take action to repair or replace the ESP or at least one component thereof prior to failure. In addition, during any step of the method 900, parameters (predicted or actual), the manipulation of parameters, differences between parameters, health statuses, etc. may be stored, such as in storage 304, 316, or a combination thereof. Further, during any step of the method 900, such information may be both displayed and stored.

[0041] Figure 10 is a flow diagram of an alternative process 1000 to evaluate the health of an ESP. The alternative process 1000 may be supplemented using one or more of the techniques described above— for instance and without limitation, the data filtering techniques, the identification of weighted relationships, and the generation of models as described above. The process 1000 begins by obtaining a group of parameters from multiple, healthy ESP systems (step 1002). To be considered "healthy," the ESP systems (and, more particularly, the ESPs in those systems) must meet one or more predetermined health criteria. In some embodiments, at least ten such systems are used. In some embodiments, one hundred or more such systems are used. In some embodiments, the ESP systems used are diverse so as to establish a model that accounts for various operating scenarios. The process 1000 continues by identifying relationships between the parameters from the multiple, healthy ESP systems (step 1004). The relationships may be identified using any of the techniques described above, and in some embodiments, weights ascribed to the various relationships also may be identified as described above. The process 1000 continues by generating a model based on the identified relationships (step 1006). The model may be generated using the techniques already explained.

[0042] The process 1000 continues by obtaining a second group of parameters from another ESP system (step 1008) and determining a predicted value using the model and the second group of parameters (step 1010). Similarly, a third group of parameters is obtained from multiple other ESP systems that share a common reservoir with the ESP system mentioned in step 1008 (step 1012). A set of predicted values is obtained using the model and the third group of parameters (step 1014). The process 1000 then comprises comparing the predicted value and the set of predicted values to identify one or more differences (step 1016). A health status of the ESP system of step 1008 is determined based on these differences (for example, as described above) and displayed on any suitable display in the system (step 1018). Alternatively or in addition to such display, the health status may be stored— for instance, in storage 304 and/or 316 (Figure 3).

[0043] The processes described above are illustrative. They may be modified as desired, such as by adding, deleting, rearranging or modifying one or more steps. Although only a few example embodiments have been described in detail above, those skilled in the art will readily appreciate that many modifications are possible in the example embodiments. Features shown in individual embodiments referred to above may be used together in combinations other than those which have been shown and described specifically. Accordingly, all such modifications are intended to be included within the scope of this disclosure as defined in the following claims.

[0044] The embodiments described herein are examples only and are not limiting. Many variations and modifications of the systems, apparatus, and processes described herein are possible and are within the scope of the disclosure. Accordingly, the scope of protection is not limited to the embodiments described herein, but is only limited by the claims that follow, the scope of which shall include all equivalents of the subject matter of the claims.