METHOD FOR MODELING A PLURALITY OF HYDROCARBON PRODUCTION WELLS, METHOD FOR PREDICTING AND ESTIMATING A FLOW PARAMETER, WELL PARAMETER AND / OR THE STATUS OF AT LEAST ONE CONTROL POINT, METHOD FOR OPTIMIZING HYDROCARBON PRODUCTION, COMPUTER SYSTEM AND COMPUTER-READABLE STORAGE MEDIUM

Abstract

A METHOD FOR MODELING A PRODUCTION WELL. A method for modeling a plurality of hydrocarbon production wells, wherein each production well is associated with at least one control point in a flow path associated with it. The method comprises: (i) generating a first model capable of describing for any of the first plurality of production wells a relationship between flow parameters, well parameters and / or an associated status of at least one control point, wherein the first model is parameterized by a set of first parameters representative of properties common to all the first plurality of production wells. The model can be applied to estimate well parameters, flow parameters and / or the status of control points. Furthermore, the resulting models can be used to optimize production from the production well.

Description

1 / 81 METHOD FOR MODELING A PLURALITY OF HYDROCARBON PRODUCTION WELLS, METHOD FOR PREDICTING AND ESTIMATING A FLOW PARAMETER, WELL PARAMETER AND / OR THE STATUS OF AT LEAST ONE CONTROL POINT, METHOD FOR OPTIMIZING HYDROCARBON PRODUCTION, COMPUTER SYSTEM AND MEANS OF COMPUTER-READABLE STORAGE

[001] The present invention relates to methods for modeling a hydrocarbon production well (e.g., a gas and / or oil production well). The present invention further extends to corresponding computer systems and computer program products. The resulting models can be applied to estimate and / or predict well parameters, flow parameters and / or the status of control points, such as flow rates, well health indicators, compositional composition of the produced fluid, etc., in a real-time setting and to make future predictions of the production well behavior. Furthermore, the resulting models can be used to optimize production.

[002] In the oil and gas industry, obtaining accurate models of the behavior of production wells is of particular interest. The behavior of production wells can be difficult to measure and / or model accurately, particularly mechanically, and in many cases can vary unpredictably. Additionally, the availability of critical process components changes over time, and thus capacities vary accordingly. Therefore, it is difficult to optimize production configurations for these hydrocarbon production wells and, correspondingly, the production networks in which they are located. Simulations and models can be used in an attempt to predict the behavior of production wells and flow networks in response to changes in process parameters, such as Petition 870230096179, dated 10 / 30 / 2023, page 11 / 106 2 / 81 such as flows, pressures, mixing of different constituents and so on.

[003] Well flow, a primary well feature of interest, is traditionally modeled from conservation laws for mass, momentum, and energy. Such modeling can be considered mechanistic well modeling – that is, based on the real and true physics of well flow. Equation (1) below presents a traditional mechanistic model for use in estimating total flow rate. This model results from an assumption of conservation of momentum and energy. Conservation of mass is also implicitly assumed for steady-state flow, which can be derived as an average (median) of the dominant dynamic behavior of the production well during a configured production period. (1) Q = Kη,ρ,ί,η,ξ)

[004] In the model above equation (1), Q represents the total flow rate of the production well, p represents the well flow rate pressures (collected as a single term), t represents the flow temperatures (collected as a single term), η represents the volumetric (or mass) fraction of the constituents of the produced fluid compared to the total flow, ξ represents the model parameters that are indicative of the physical properties of the system (e.g., fluid properties, geometric properties, and external factors), and i represents the other explanatory variables in the system, including control variables (e.g., the position of a throttling valve in the production well flow path) and measurements of the state of the production system.

[005] In theory, ξ can be an exhaustive list of parameters that specify all the system properties relevant to modeling flow rates. For example, the parameters ξ can describe nano / micro properties of the system (e.g., particle flow paths). Petition 870230096179, dated 10 / 30 / 2023, page 12 / 106 3 / 81 individual factors in the production system), as well as macro system properties (e.g., choke sizing, pipe diameter, fluid viscosities, etc.).

[006] Modeling with such a set of parameters is impractical, however, for two main reasons. First, not all of these parameters can be measured for any system. Second, the large number of parameters, particularly the large number of unobservable parameters, results in an intractable model.

[007] Therefore, in practice, the parameters ξ are decomposed into two different sets of parameters: a first set, α, which represents the parameters of the production system that can be observed and a second set, β, representing the parameters of the system that cannot be observed. The first set of parameters α may include, for example, pipe roughness within the production system, the oil density under standard conditions, etc. Using the observable parameters, a simplified mechanistic model, g, for the total flow of the production system can be produced as shown in equation (2). (2) Q = g(u,p,t,p,a)

[008] The physics of the g-model is simplified compared to the true f-model, since not all parameters of the production system / well are considered (i.e., the β parameters are ignored). Such a simplified mechanistic model will therefore not give a 'true' picture of the production system.

[009] In the past, modelers have attempted to find a close approximation of g (denominated g) through extensive mechanistic modeling. Candidate g-approach models are compared to real well test data or experimental test cycles. In practice, it is necessary to approximate Petition 870230096179, dated 10 / 30 / 2023, page 13 / 106 4 / 81 the g model (i.e., producing a g model) for only certain λ parameter configurations (i.e., for production systems / wells sharing common physical properties and / or characteristics), otherwise the modeling would be too complex to be practically useful. An exemplary λ parameter configuration that can be used as a constraint on the approximate g model is a λ configuration in which the flow regime produced is likely to be a slow flow. By limiting the approximation g model to only certain λ parameter configurations, the modeling of the production system / well can be significantly simplified and data from a diverse range of production wells / systems that fall within the shared λ parameter configuration can be used to produce the approximate g model. Modeling within the shared λ parameter configuration is illustrated by equation (3) below. (3) Q = g(u, p, t, η, a : a € Â),

[010] The model illustrated in equation (3) is generated by testing the approximation model g on data from individual production wells and adjusting the observable parameters α of the model so that the model g is best calibrated to model the specific well of interest. Once the approximation model has been calibrated as such, the model can be used to estimate other production well flow rates.

[011] In the more recent past, data-driven modeling, as opposed to the more traditional mechanistic modeling as described above, has been implemented in modeling the liquid and gas flow rate of a single production well or single-flow network. An example of such a data-driven modeling technique, as it is known in the art, is described in document WO 2019 / 110851.

[012] The models produced in these data-driven techniques of Petition 870230096179, dated 10 / 30 / 2023, page 14 / 106 5 / 81 State-of-the-art models are generated and trained based on data from the single well or single-flow network to be modeled. Thus, because the model is based on real-life data (as opposed to an approximate mechanistic model that ignores certain unobservable parameters of the production well), the model can, in theory, give a truer reflection of production behavior than the approximate, mechanistic models discussed above.

[013] That said, the data-driven models implemented so far have limited applicability.

[014] For example, data-driven models to date are only particularly suitable for modeling the single well / production stream network from which the data used in model generation were collected. As such, state-of-the-art data-driven models have limited applicability in relation to the number of production systems / wells that they can successfully and accurately model.

[015] Furthermore, the amount of data gathered for a single well / flow network will be limited. That is, only the data collected throughout the operational life of that single well / flow network is available for modeling. It will be appreciated that an increased amount of data on which the model is based will result in a more robust model, and likewise a less robust model will be generated when less data is available for its generation. Thus, state-of-the-art models typically have limited robustness.

[016] Additionally, the state-of-the-art models produced will be based only on historical data, recorded during a previous state(s) of the well / network. This data will typically not be indicative of the present / future state of the production well. This is mainly due to the fact that the reservoir drainage (to which the production well is attached) Petition 870230096179, dated 10 / 30 / 2023, page 15 / 106 6 / 81 connected), over time, will result in changes in the behavior of the production well. Such changes in the reservoir resulting in a change in the behavior of the production well can be called the "reservoir effect" in the production well.

[017] Many of the limitations in data-driven modeling techniques used to date are not typically shared by approximate mechanistic modeling techniques, as discussed above. For example, approximate mechanistic models are generated based on a set of observable physical parameters, α, which are (or should be) common to each of the wells in the shared parameter configuration λ under which the mechanistic model is generated. That is, approximate mechanistic models were generated to take into account the behaviors of a plurality of different diverse wells (albeit under a specific parameter configuration λ) and are therefore typically more robust, can realistically model a plurality of different wells / flow networks, and for this reason can also better take into account the reservoir effect.

[018] Improvements in data-driven models are therefore desired.

[019] According to a first aspect of the invention, a method is provided for modeling a first plurality of hydrocarbon production wells, each production well being associated with at least one control point in a flow path associated therewith, the method comprising: (i) generating a first model capable of describing for any one of the first plurality of production wells a relationship between flow parameters, well parameters and / or an associated status of at least one control point, wherein the first model is parameterized by a set of first parameters representative of properties common to all the first plurality of production wells. Petition 870230096179, dated 10 / 30 / 2023, page 16 / 106 7 / 81

[020] Since the model is parameterized by a first set of parameters representing properties common to the entire plurality of production wells, the model can be used to model any of the plurality of production wells with relative accuracy and success. This contrasts with state-of-the-art data-driven models, which, as above, are typically only suitable for modeling a single production network or well (i.e., the production network / well for which the model was designed and from which the data were derived for its generation and formation). For example, with regard to the modeling techniques disclosed in WO 2019 / 110851, there are no shared parameters between the different well models. The parameters within each of the models are very specific, and there is no representation, in these models, of properties common to each of the wells within a plurality.Thus, the models produced in document WO 2019 / 110851 are only suitable for modeling the specific well for which they were produced and have little applicability, in general, to a variety of wells. This contrasts with the first model of the first aspect of the invention, which has applicability to a plurality of production wells.

[021] The first model also improved robustness by virtue of the first set of parameters that are shared and common across the entire first plurality of wells. The optional model generation and training, as discussed in further detail below, can be based on a larger dataset gathered across a wider range of wells. State-of-the-art methods generate their model for the specific well to be modeled and train said model based on data only from the well to be modeled. Thus, biases within the model, both resulting from the nature of the model itself and from the data used to train the state-of-the-art model, more strongly impact the model produced and thus produce Petition 870230096179, dated 10 / 30 / 2023, page 17 / 106 8 / 81 inaccuracies in the resulting model and / or low applicability to wells other than the one it was designed to model.

[022] The set of first parameters that are common to each of the wells can be considered analogous to the configuration of the λ parameter as discussed above in relation to mechanistic modeling techniques. That is, the first parameters are representative of a shared configuration of each of the plurality of production wells (i.e., they are indicative of some physical properties and / or characteristics that are common to each of the production wells). Thus, as observed above, similarly to mechanistic models, the first-aspect model can adequately model any of a plurality of production wells, since it takes into account behaviors and traits common to each of the wells in the first plurality.

[023] However, the number of parameters in the first parameter set is significantly greater than the number of parameters used in mechanistic models. Mechanistic models often comprise between 1 and 4 parameters. In the present invention, the data-driven modeling used allows the first parameter set to optionally comprise more than 1,000 parameters, optionally in the region of 10,000 - 1,000,000. However, sufficient computing power may allow for a greater number of parameters than this. It will be appreciated that this greater number of parameters may allow for improved and more accurate modeling.

[024] The first set of parameters that are described as being common or shared may mean that a given parameter is identical in the first model for every well within the plurality. This is known as rigid sharing. Alternatively, a given parameter in Petition 870230096179, dated 10 / 30 / 2023, page 18 / 106 9 / 81 The first model can be nearly identical (i.e., almost the same) for different wells within the first plurality and still be considered shared. That is, instead of having a given parameter identical for each well, the given parameter can be slightly different between wells within the first model. In this case, the difference between the parameters is penalized to account for their non-identity. This is known as soft sharing. This hard or soft sharing can be applied equally with respect to the parameters of the first model(s), the flow composition model(s), the forecast model(s), etc., as described in further detail below.

[025] The first aspect method may comprise the additional steps of (ii) generating at least one first additional model capable of describing for any one of a different additional plurality of production wells, a relationship between flow parameters, well parameters and / or an associated status of at least one control point, wherein at least one first additional model is parameterized by an additional set of first parameters representative of properties common to all the plurality of production wells; and (iii) combining the first model with at least one first additional model to form a combined model capable of describing for any of the wells in the first plurality and at least one additional plurality of production wells a relationship between flow parameters, well parameters and / or an associated status of at least one control point to which it relates.

[026] The at least first additional model, by itself, shares many of the same advantages as the first model. That is, it allows for successful and accurate modeling of any of the wells in the additional plurality of different production wells, since it contains a set Petition 870230096179, dated 10 / 30 / 2023, page 19 / 106 10 / 81 additional initial parameters that are representative of the behaviors and characteristics common to each of the different additional plurality of production wells.

[027] Therefore, the combination of the first model and at least one additional first model forms a combined model with even greater applicability and / or increased modeling accuracy. The greater applicability to a wider range of wells of this combined model is most notably achieved where some of the wells in the additional plurality of wells are not contained in the first plurality. Thus, the overall combined model is better suited to modeling a larger number of wells than either of the first or additional first models alone.

[028] Greater accuracy is achieved in particular where there is at least some overlap between wells in the additional plurality of wells and in the first plurality of wells. For these overlapping wells, the accuracy of the modeling provided by the combined model is increased, since the combined model is derived from two separate models based on different sets of first parameters that reflect the behaviors of these 'overlapping wells', thus providing a greater overview of the behavior of these wells.

[029] The first aspect method may comprise generating a plurality of first additional models, each first additional model capable of describing for a respective different plurality of production wells a relationship between flow parameters, well parameters and / or an associated status of at least one control point, wherein each first additional model is parameterized by a set of first parameters representative of properties common to the respective plurality of production wells; and combining the first model with the plurality of Petition 870230096179, dated 10 / 30 / 2023, page 20 / 106 11 / 81 first additional models to form a combined model capable of describing for any of the wells, both in the first plurality and in each of at least one additional plurality of production wells, a relationship between flow parameters, well parameters and / or associated status of at least one control point to which it relates.

[030] The plurality of first additional models expands the advantages gained by the single first additional model discussed above. That is, a plurality of first additional models can provide increased applicability and improved modeling accuracy. Thus, improved modeling of a larger number of wells and / or more accurate well modeling can be achieved.

[031] As mentioned above, at least some of the production wells within, or each of, additional production well pluralities may be in the first production well plurality. Furthermore, and / or alternatively, at least some of the production wells within one or more of the additional production well pluralities may be in another of the additional production well pluralities. For these particular overlapping wells, the combined model can provide improved modeling accuracy.

[032] All production wells within at least one, several, or each additional plurality of production wells may be in the first plurality of production wells, and the first plurality of production wells may additionally include additional production wells. As such, at least one, several, or each additional plurality of production wells may be considered as a subset of the first plurality of production wells. Thus, the first additional model(s) may be seen as modeling and taking into account behaviors or characteristics of Petition 870230096179, dated 10 / 30 / 2023, page 21 / 106 12 / 81 wells that may not necessarily be shared throughout the entire first plurality, but may be shared throughout a smaller subset of the first plurality.

[033] For example, the first well plurality might comprise wells spread across several different hydrocarbon assets or reservoirs. The first model can therefore adequately model any behaviors, characteristics, or traits of the wells that are commonly held for wells across each of these different assets / reservoirs. However, some behaviors / traits of the wells may not be common to wells across all of these different assets / reservoirs. For example, some behaviors may be asset-specific. Therefore, an additional first model can be created to model the behaviors of the models in only one specific asset, and once combined with the first model, the combined model can accurately model those wells from the specific reservoir more precisely than either the first model or the additional first model could on its own.This is because the combined model can take into account these behaviors and traits common across various reservoirs / assets, as well as those specific to the reservoir in which the well subset is located.

[034] The above is merely an example of how the first plurality and at least one, several, or each additional plurality of wells can relate to each other. Other divisions / subdivisions / relations of the first plurality and at least one, several, or each additional plurality are envisioned. For example, as a further development of the above example, an additional first model could be introduced that models only a subset of wells in the specific reservoir. Alternatively, Petition 870230096179, dated 10 / 30 / 2023, page 22 / 106 13 / 81 The first plurality of wells may be all the wells in a specific reservoir, and the, or each, additional plurality of production wells may be a subset of the wells in that specific reservoir. Alternatively, the first plurality of wells may be wells across a variety of different assets experiencing some common dynamic behavior, for example, slow flow. The, or each, additional plurality of production wells may then be wells from a specific reservoir or location that exhibit that particular dynamic behavior.

[035] The person skilled in the art would appreciate that there are many additional ways in which the first plurality of production wells and at least one, several, or each different additional plurality of production wells can be split / subdivided. Importantly, when the one or each different additional plurality of production wells is related to a subset of the first plurality of wells, the at least one, several, or each first additional model can be introduced into the combined model to take into account more specific behaviors and traits, while the first model can take into account more generic traits and behaviors. Thus, improved overall modeling accuracy can be achieved.

[036] At least some of the production wells within at least one, several, or each additional plurality of production wells may not be included in the first plurality of production wells. Thus, the first plurality and at least one, several, or each additional plurality of production wells may be completely independent of each other, having no overlap with respect to their wells. Alternatively, there may be only partial overlap between the first and at least one, several, or each additional plurality of wells. In either scenario, the model Petition 870230096179, dated 10 / 30 / 2023, page 23 / 106 The combined 14 / 81 model is based on a larger number of wells than the first model, or at least one, several, or each additional first model by itself. Consequently, the combined model has greater applicability and will have improved accuracy for at least those partially overlapping wells, if such wells exist.

[037] The method may comprise generating a second model that is capable of describing a relationship between flow parameters, well parameters and / or an associated status of at least one control point for only one production well, wherein the second model is parameterized by a set of second parameters representative of specific properties of the production well to which it relates; combining the second model with the first model and, optionally, at least one, several or each additional first model to form a combined model that is capable of describing a relationship between flow parameters, well parameters and / or an associated status of at least one control point for only one production well.

[038] The first model generated in the first-aspect method is advantageous because it allows for robust modeling of a production well that is not overly dependent on the past properties and behaviors of that well, as discussed above. The first additional model(s) are equally advantageous. For this same reason, however, although robust, the first model / first additional model(s) is / are not better positioned to characterize and model the idiosyncratic behaviors that are specific to a particular production well being modeled. That is, while certain behaviors, characteristics, and traits will be shared across a plurality of production wells (and are represented in the first (or first additional) set of parameters), other behaviors, characteristics, and traits will be Petition 870230096179, dated 10 / 30 / 2023, page 24 / 106 15 / 81 unique to a given production well. Thus, useful and / or accurate inferences cannot be drawn about these specific well behaviors / traits / characteristics based on knowledge of other wells.

[039] Therefore, in order to better take into account the physical properties, behaviors and specific characteristics of the well to be modeled, well modeling can incorporate a second model specific to one of the production wells to be modeled and which describes the behaviors that are idiosyncratic to that well by virtue of the set of second parameters. Each set of second parameters reflects behaviors, characteristics and traits unique to the production well to which it relates and, as such, the second model is able to describe well-specific relationships and behaviors for that production well.

[040] As an alternative understanding, since the first model (or first additional model(s)) describes the behaviors common to each of the plurality of production wells (or additional plurality of production well(s)), the first model (first additional model(s)) can be understood to capture the intermediate / middle well within the respective plurality of production wells. Furthermore, since the second model describes the specific behaviors of the production well to which it relates, the second model can be considered to capture the differences in behavior that the specific well has from the intermediate / middle well within the respective plurality of production wells. Consequently, the combination of the first (and / or first model(s)) with the second model allows for an accurate modeling of the specific well because the first model can be customized by the second model to give an accurate representation of those specific wells. Petition 870230096179, dated 10 / 30 / 2023, page 25 / 106 16 / 81

[041] The relationship described by the second model between flow parameters, well parameters, and / or an associated status of at least one control point for the production well to which it relates may not have a tangible physical equivalent in the real world. That is, it may not be possible to equate the relationship described by the second model to a real-world physical relationship. However, regardless of whether the relationship described by the second model can be considered to correspond to a real-world physical relationship or not, the second model remains descriptive of some (perhaps undefined) relationship between flow parameters, well parameters, and / or an associated status of at least one control point for the production well to which it relates.

[042] The set of second parameters can comprise between 1-20 parameters, for example, 10 parameters.

[043] The single well to which the second model relates may be included within the first plurality of production wells, of, the various, or each additional plurality of production wells, and / or any of the pluralities of wells referred to below. As such, the combined model may be customized to specifically model one of the wells within the first and / or the, the various, or each plurality of production wells, or within one of the various pluralities of wells mentioned below.

[044] The single well to which the second model relates may not be included within the first plurality of production wells, the, or each, additional plurality of production wells, and / or any of the pluralities of wells referred to below. In this scenario, the generic behaviors and traits for the first plurality of wells and / or the, or each, additional plurality of production wells may be considered true for a well not included in the first plurality of wells. Petition 870230096179, dated 10 / 30 / 2023, page 26 / 106 17 / 81 production and / or in, or each, additional plurality of production wells. This assumption can be useful when no useful model is available for the generic behaviors / traits to which the second model relates. This assumption can also be a relatively safe assumption to make, particularly when there are a large number of diverse wells within the first plurality or the, of the diverse or of each, other first plurality, or the pluralities of wells referred to below and / or where the well to which the second model relates shares similar behaviors and traits to the wells within the first plurality and / or the, of the diverse or of each, additional first plurality, or the pluralities of wells referred to below.

[045] The method may comprise generating a plurality of second models, each second model capable of describing a relationship between flow parameters, well parameters and / or an associated status of at least one control point for a respective production well, each second model being parameterized by a set of second parameters representative of specific properties of the production well to which it relates; and combining each second model with the first model and, optionally, several or each of the additional first models (or even any of the several models referred to below) to form combined models that are each capable of describing a relationship between flow parameters, well parameters and / or an associated status of at least one control point for the respective production well to which it relates.

[046] Thus, combined models can be provided that adequately take into account the specific and generic properties of a plurality of different individual wells. Petition 870230096179, dated 10 / 30 / 2023, page 27 / 106 18 / 81

[047] Optionally, a second model may be generated for each well discussed above and below.

[048] The plurality of second models can be understood within a second model structure. When one wishes to model for a specific well(s), or only a few selected ones, the second model structure can be reduced by inputting a signal so that only the second model(s) related to the production well(s) of interest remain in the second model structure. This can be achieved, for example, by setting the second parameters in those second models related to other wells not of interest within the plurality to zero. As such, by inputting the well-specific signal, the second model structure is able to describe a relationship between flow parameters, well parameters and / or an associated status of at least one control point only for that production well(s) associated with the well-specific signal.

[049] The concept of using a well-specific signal to maintain only the second model(s) in the second model structure that is / are of interest can be understood as a 'hot well coding'. That is, the second model structure is applied to be specifically capable of modeling 'hot' well(s) (i.e., wells of interest) by virtue of the well-specific signal input. This is advantageous as it provides a computationally inexpensive way to adapt the second model structure, and thus the combined model, to be specifically suited to modeling specific well(s), since the signal reduces the model to include only the second model specific to the behavior and configuration of the well to be modeled.

[050] The second model structure can be a second matrix Petition 870230096179, dated 10 / 30 / 2023, page 28 / 106 19 / 81 model, where each column of the matrix represents a second model (i.e., a vector of the second model). As such, the input signal can be seen to select a vector (or vectors) from the second model structure for use in subsequent steps of the method.

[051] Once contracted (i.e., after selection of the second model(s) of interest), the second model structure can be incorporated as part of the combined model. As such, the resulting combined model is specifically tailored to model the well(s) of interest. By virtue of this combination of the second model structure and the first (and / or first additional) generic well model(s), an improved overall combined model is generated that takes advantage of both the first (first additional) and second models. That is, the resulting model is robust (i.e., it is not strongly influenced by specific historical states of the well being modeled), responds well to the reservoir effect, and can take into account the idiosyncratic behavior of the production well being modeled.

[052] As will be appreciated, the second model structure resulting from the input of the specific well signal may result in a structure comprising one or a plurality of second models. In scenarios where a plurality of second models remains in the adapted second model structure, incorporating the second model structure as part of the combined model may comprise the input of each of the plurality of remaining second models in the second model structure into the respective copies of the combined model / first model.

[053] The specific well signal can be a binary vector. Alternatively, any signal capable of achieving the second 'selection' of the model, as noted above, can be appropriately used as the Petition 870230096179, dated 10 / 30 / 2023, page 29 / 106 20 / 81 specific well signal.

[054] Each second model may consist of the set of second parameters that are representative of specific properties of its related production well. That is, each second model may be merely the second parameters, in vector form or otherwise. The input of the well-specific signal in the second model structure may therefore be merely a simple matrix multiplication, particularly in cases where the well-specific vector is a binary vector.

[055] The incorporation of the second model(s) into the first model / combined model may include the input of data relating to flow parameters, well parameters, and / or an associated status of at least one control point of the production well(s) associated with the second model(s). The second model output(s) may be generated specifically for that production well(s). Each output of the second model may be viewed as a unique fingerprint for the second model(s) and the production well(s) to which the second model(s) relate(s). This output(s) may be used as input(s) for the first model / combined model, so that the first model / combined model is specifically able to describe the behaviors / traits / characteristics of the production well(s) of interest.That is, after the aforementioned input, the first model / combined model is (as is always the case) able to take into account both behaviors / traits / characteristics that are common to each of the wells in the first plurality (and optionally to, the various, or each additional plurality) and, by virtue of the output of the second model, is able to account for the specific behaviors of the well.

[056] The method may involve generating a composition model of Petition 870230096179, dated 10 / 30 / 2023, page 30 / 106 21 / 81 flow that is capable of describing a relationship between the flow composition of the fluid produced from any one of a second plurality of production wells and the flow parameters, well parameters, an associated status of at least one control point and / or time, wherein the flow composition model is parameterized by a first set of flow composition parameters that are representative of the flow composition common to all second pluralities of production wells;and combine the flow composition model with the first model and, optionally, the, several or each additional first model and / or the, several or each, second model to form a combined model that is capable of describing a relationship between flow parameters, well parameters, an associated status of at least one control point and / or time for any of the wells within the second plurality and the first plurality of production wells and, optionally, the, several or each, additional plurality of production wells and / or the, or each, well on which the second model(s) is / are based.

[057] The flow composition model is specifically capable of describing behaviors and traits that are common to the flow composition of the fluid produced from each of the second plurality of wells. As will be appreciated by those skilled in the art, a responsibility of flow composition is of particular importance in the modeling of production wells, since it directly impacts numerous other traits and behaviors and is a key variable in overall production performance. Thus, the generation of the flow composition model and its combination as part of the combined model allows flow composition to be properly accounted for in the modeling of production wells.

[058] As is the case with the first and the, the various or each first Petition 870230096179, dated 10 / 30 / 2023, page 31 / 106 22 / 81 In addition to the additional model discussed above, the flow composition model is generated in such a way that it is able to describe behaviors and traits that are common to the flow composition of a plurality of wells (i.e., the second plurality of wells). This is achieved by virtue of the first set of flow composition parameters that are descriptive of behaviors and traits common to each of the wells within the second plurality. Thus, the flow composition model shares similar advantages to the first model and to, or each, first additional model with respect to applicability and accuracy in relation to flow composition modeling.

[059] The first set of flow composition parameters can comprise between 0-1000 parameters, optionally 0-100. It is also possible that there are more than 1000 parameters in the first set of flow composition parameters.

[060] At least some of the production wells within the second plurality of production wells may be included within the first plurality of production wells, the additional plurality of production wells, several of the additional pluralities of production wells, and / or each additional plurality of production wells. For these particular overlapping wells, the combined model can provide improved modeling accuracy. This is particularly because these are the wells for which flow composition behavior and other generic behaviors are considered in the combined model.

[061] All production wells within the second plurality of production wells may be included in the first plurality of production wells and / or in, in the various or each, additional plurality of production wells.

[062] For example, the wells in the second plurality of wells may be Petition 870230096179, dated 10 / 30 / 2023, page 32 / 106 23 / 81 identical to the wells in the first plurality of production wells and / or in the various or each additional plurality of production wells. Thus, the flow composition model can allow a particularly accurate modeling of all wells within the first plurality and / or the various or each additional plurality of production wells, as it can accurately describe the flow composition behaviors common to each of these wells.

[063] Alternatively, the first plurality of production wells and / or the, the various, or each additional plurality of production wells may additionally include additional production wells. As such, the second plurality of production wells may be considered as a subset of the first plurality of production wells and / or the, the various, or each additional plurality of production wells. Thus, the flow composition model can be seen to model and take into account the flow composition behaviors that are shared across a smaller subset of wells and that are not necessarily shared by each of the first plurality and / or the, the various, or each additional plurality.

[064] As such and similar to what was discussed above in connection with the first additional model(s), since the second plurality of production wells is related to a subset of the first plurality of wells and / or the, several, or each first additional plurality of wells, the flow composition model can be introduced into the combined model to account for flow composition behaviors and traits that are more specific to a certain number of wells, while the first model and / or the, several, or each first additional model can account for more generic traits and behaviors of the wells. Thus, improved overall modeling accuracy can be achieved. Petition 870230096179, dated 10 / 30 / 2023, page 33 / 106 24 / 81

[065] The relationship between the second plurality of production wells and the first plurality of production wells and / or the various or each different additional plurality of production wells may correspond to the relationships established above in relation to the first plurality of production wells and the, or each, additional plurality of production wells.

[066] At least some of the production wells within the second plurality of production wells may not be included in the first plurality of production wells and / or in the, several, or each additional plurality of production wells. Thus, the second plurality, the first plurality, and the, several, or each additional plurality of production wells may be completely independent of each other, having no overlap with respect to their wells. Alternatively, there may be only partial overlap between the second plurality and the first and the, several, or each additional first plurality of production wells. In either scenario, the combined model produced is based on a larger number of wells than either of the individual models alone. Consequently, the combined model has greater applicability and will have improved accuracy for at least those partially overlapping wells, if such wells exist.

[067] The first aspect method may involve generating a plurality of flow composition models (each of which may correspond to the flow composition model discussed above). Each flow composition model may be able to describe a relationship between the flow composition of the fluid produced from any one of a respective second plurality of production wells and the flow parameters, well parameters, an associated status at at least one control point and / or time, where each flow composition model is Petition 870230096179, dated 10 / 30 / 2023, page 34 / 106 25 / 81 parameterized by a first set of flow composition parameters that are representative of the flow composition common to all respective second plurality of production wells to which it relates. The method may further comprise combining each flow composition model with the first model and, optionally, the first additional model, several first additional models, or each first additional model and / or the, or each, second model to form a combined model that is capable of describing a relationship between flow parameters, well parameters, an associated status of at least one control point and / or time for any of the wells within any of the second plurality of production wells and the first plurality of production wells and, optionally, the several, or each of the additional pluralities of production wells and / or the, or each, well on which the second model(s) is / are based.

[068] The first aspect method may comprise generating a well-specific flow composition model that is capable of describing a relationship between the flow composition of the fluid produced from only one production well and flow parameters, well parameters, an associated status of at least one control point and / or time, wherein the well-specific flow composition model is parameterized by a second set of flow composition parameters that are representative of the specific flow composition for the production well to which it relates; combining the well-specific flow composition model with the first model and, optionally, the, the various or each additional first model, the, the various or each second model and / or the, the various or each well composition model to form a combined model that is capable of describing a relationship between flow parameters, well parameters, a Petition 870230096179, dated 10 / 30 / 2023, page 35 / 106 26 / 81 status associated with at least one control point and / or time for only one production well.

[069] The well-specific flow composition model shares similarities with the second model, as it models specific behaviors of only one well (in this case, flow composition behaviors). Thus, combining the well-specific flow composition model into the combined model results in advantages corresponding to those achieved by combining the second model into the combined model. That is, combining the well-specific flow composition model allows flow composition behaviors specific to the well to which it relates to be taken into account in its modeling, which may not be accurately accounted for by any of the other models included within the combined model.

[070] The second set of flow composition parameters may comprise between 0-10 parameters and, optionally, up to 100 parameters or more.

[071] The single well to which the well-specific flow composition model relates may be comprised within the first plurality of production wells, of, the various, or each additional plurality of production wells, of, the various, or each second plurality of production wells and / or any of the pluralities of wells discussed below. Similar to what was discussed above in connection with the second model, greater modeling accuracy can be achieved for this well when comprised within any of the pluralities configured above.

[072] The single well to which the well-specific flow composition model relates may not be included in the first plurality. Petition 870230096179, dated 10 / 30 / 2023, page 36 / 106 27 / 81 of production wells, or each additional plurality of production wells, or each second plurality of production wells, and / or any of the additional pluralities of wells referred to below. Thus, assumptions can be drawn from the generic behaviors of any of these well pluralities and applied to the well to which the well-specific flow composition model relates, while the well-specific composition model can take into account flow composition traits that are unique to that well.

[073] The well to which the specific well model is related may be the same well to which, or at least one of the second model(s), is / are related. Where there is a match in wells, and there is a matching of the specific well model with a second model related to the same well, improved modeling accuracy for this well can be achieved, since specific behaviors of this well model, as taken into account in the second model and as taken into account in the specific well model, are better reflected in the overall combined model.

[074] The method may comprise generating a plurality of well-specific flow composition models, each corresponding to the well-specific flow model described above, but relating to a different respective well. Each well-specific flow composition model may be able to describe a relationship between the flow composition of the fluid produced from only one respective well and flow parameters, well parameters, an associated status of at least one control point, and / or time, with each well-specific model parameterized by a second set of flow composition parameters that are representative of the flow composition that is specific to the single respective production well. Petition 870230096179, dated 10 / 30 / 2023, page 37 / 106 28 / 81 to which it relates. The method may further comprise combining each specific well flow composition model with the first model and, optionally, the, the various or each first additional model, the, the various or each second model and / or the, the various or each flow composition model to form combined models that are each capable of describing a relationship between flow parameters, well parameters, an associated status of at least one control point and / or time, for each respective well.

[075] Thus, combined models can be provided that adequately take into account the specific flow composition behaviors of a plurality of different individual wells.

[076] Optionally, there may be a well-specific flow composition model generated for each well referred to above and below.

[077] The plurality of well-specific flow composition models can be understood within a well-specific flow composition model framework. This framework can closely correspond to the second model framework as configured above. Thus, the well-specific flow composition model framework can leverage the corresponding functionality and resources that the second model framework can, as configured above.

[078] In addition to being able to describe relationships between flow parameters, well parameters, and / or the status of at least one control point, the flow composition model(s) and the well-specific flow composition model(s) are also capable of describing relationships with respect to time. That is, these models can describe how the development of time can impact the development of the flow composition as represented in the flow parameters, well parameters, and / or the status of Petition 870230096179, dated 10 / 30 / 2023, page 38 / 106 29 / 81 at least one control point. Similarly, these models can describe how the development of flow parameters, well parameters, and / or the status of at least one control point resulting from the flow composition is related to time. Thus, the flow composition model(s) and the well-specific flow composition model(s) can allow interpolations and extrapolations in time, thereby providing a description of the flow composition at instances in time where no data may be available (e.g., a past state or an unrecorded future state). This allows modeling and estimations regarding a production well to be made in the future, along with moments in the past where perhaps inadequate data for modeling would otherwise be available.

[079] The method may comprise generating a forecasting model, the forecasting model being able to predict for any one of a third plurality of production wells a change in a flow parameter, well parameter and / or a status of at least one control point based on a hypothetical change in the status of at least one control point, a hypothetical change in a flow parameter and / or a hypothetical change in a well parameter, wherein the forecasting model is parameterized by a set of forecasting parameters that are representative of properties that are common to the third plurality of production wells;and combine the forecast model with the first model and, optionally, the, the various or each additional first model, the, the various or each second model, the, the various or each flow composition model and / or the, the various or each well-specific flow composition model to form a combined model that is capable of predicting a flow parameter, a well parameter and / or the status of at least one control point resulting from a hypothetical change in the status of at least one control point; Petition 870230096179, dated 10 / 30 / 2023, page 39 / 106 30 / 81 control, the hypothetical change in a flow parameter and / or the hypothetical change in the well parameter for any of the wells within the third plurality of production wells and the first plurality of production wells and, optionally, the, the various or each additional plurality of production wells, the, the various or each well on which the second model(s) is / are based, the, the various or each second plurality of production wells and / or the, the various or each well on which the well-specific composition model(s) is / are based.

[080] The forecasting model describes how a hypothetical change (i.e., a proposed or theoretical change) in the status of at least one control point, a well parameter, and / or a flow parameter impacts a flow parameter, well parameter, and / or the status of at least one control point for any of the wells within the third production well plurality. Thus, proposed or theoretical forecasts and / or developments can be determined by incorporating the forecasting model within the combined model. As will be described below, this allows the combined model to be used to determine an optimized state for a production well (i.e., one in which production is optimized).

[081] The set of forecast parameters may comprise between 1,000 - 1,000,000 parameters. Typically, there may be approximately 100,000 parameters included within the forecast parameters.

[082] At least some of the production wells within the third plurality of production wells may be included within the first plurality of production wells, the additional plurality of production wells, several or each additional plurality of production wells, the second plurality of production wells, several and / or each second plurality of production wells. Where there is an overlap of wells, the Petition 870230096179, dated 10 / 30 / 2023, page 40 / 106 31 / 81 The forecast enabled by the forecasting model may be more accurate. This is for reasons corresponding to those discussed above regarding overlapping wells in several other well pluralities.

[083] All production wells within the third plurality of production wells may be included within the first plurality of production wells, the additional plurality of production wells, each or several additional pluralities of production wells, the second plurality of production wells, several and / or each second plurality of production wells.

[084] For example, the wells in the third plurality of wells may be identical to the wells in the first plurality of production wells, a, the various or each additional plurality of production wells and / or o, the various or each second plurality of production wells. Thus, the prediction model may allow particularly accurate modeling of all wells within the first plurality, a, the various or each additional plurality of production wells and / or a, the various or each second plurality of wells, as it may accurately describe the flow composition behaviors common to each of these wells.

[085] The first plurality of production wells, the additional plurality of production wells, the various or each additional plurality of production wells, the second plurality of production wells, the various and / or each second plurality of production wells may additionally include additional production wells not included in the third plurality of production wells. Thus, the prediction model may relate only to a subset of these wells and, therefore, may be seen to predict behaviors or characteristics of wells that may not necessarily be shared across all these wells, but may be Petition 870230096179, dated 10 / 30 / 2023, page 41 / 106 32 / 81 shared across a smaller subset of these pluralities.

[086] At least some of the production wells within the third plurality of production wells may not be included in the first plurality of production wells, the additional plurality of production wells, each additional plurality of production wells, and / or each second plurality of production wells. Thus, the third plurality of production wells may be completely independent of any of the additional pluralities of wells. Alternatively, there may be only partial overlap between the third plurality of wells and any of the additional pluralities of wells. In either scenario, the combined model comprising the forecast model is based on a larger number of wells than either of the individual models alone. Consequently, the combined model has greater applicability and will have improved accuracy for at least those partially overlapping wells, if such wells exist.

[087] The first aspect method may comprise generating a plurality of forecasting models, each corresponding to the forecasting model discussed above. Each forecasting model may be capable of predicting for any one of a respective third plurality of production wells a change in a flow parameter, a well parameter, and / or the status of at least one control point based on a hypothetical change in the status of at least one control point, a hypothetical change in a well parameter, and / or a hypothetical change in a flow parameter, wherein each forecasting model is parameterized by a set of forecasting parameters that are representative of properties that are common to each respective third plurality of production wells. The method may comprise combining each forecasting model with the first model and, optionally, the, or each, first additional model, the, or each, Petition 870230096179, dated 10 / 30 / 2023, page 42 / 106 33 / 81 second model, the, or each, flow composition model and / or the, or each, well-specific flow composition model to form a combined model that is capable of predicting a flow parameter, a well parameter and / or a status of at least one control point resulting from a hypothetical change in the status of at least one control point, the hypothetical change in a well parameter and / or the hypothetical change in a flow parameter for any of the wells within any of the third plurality of production wells and the first plurality of production wells and, optionally, the, or each, additional plurality of production wells, the, or each, well on which the second model(s) is / are based, the, or each, second plurality of production wells and / or the well on which the well-specific composition model(s) is / are based.

[088] The plurality of forecasting models expands the advantages obtained by the single forecasting model discussed above. That is, a plurality of forecasting models can provide greater applicability and forecast accuracy. Thus, it is possible to obtain a forecast for a larger number of wells and / or more accurate forecasts for wells.

[089] The method may comprise the generation of a well-specific forecast model, the well-specific forecast model capable of forecasting for only one production well a change in a flow parameter, a well parameter and / or the status of at least one control point based on a hypothetical change in the status of at least one control point, a hypothetical change in a well parameter and / or a hypothetical change in a flow parameter, wherein the well-specific forecast model is parameterized by a set of well-specific forecast parameters that are representative of specific properties of that production well; and the combination of the well-specific forecast model Petition 870230096179, dated 10 / 30 / 2023, page 43 / 106 34 / 81 well with the first model and, optionally, the, the various or each additional first model, the, the various or each second model, the, the various or each flow composition model, the, the various or each well-specific flow composition model, and / or, various, or each forecast model to form combined models that are each capable of predicting a flow parameter, a well parameter and / or the status of at least one control point resulting from a hypothetical change in the status of at least one control point, the hypothetical change in a well parameter and / or the hypothetical change in a flow parameter for only one production well.

[090] The well-specific forecast model relates to a specific well and describes for that well how a hypothetical change (i.e., a proposed or theoretical change) in the status of at least one control point, a flow parameter, and / or a well parameter impacts a flow parameter, well parameter, and / or the status of at least one control point. Thus, proposed or theoretical forecasts and / or well-specific developments can be determined by incorporating the well-specific model within the combined model. As will be described in more detail below, this allows the combined model to be used to determine an optimized state (i.e., one in which production is optimized).

[091] Where the well-specific forecasting model differs from the (generic) forecasting model is that it considers specific behaviors of the production well to which it relates, rather than generic behaviors shared by a plurality of wells. Thus, the well-specific forecasting model allows for well-specific forecasts relevant to a specific well. This difference between the forecasting model Petition 870230096179, dated 10 / 30 / 2023, page 44 / 106 35 / 81 and the well-specific forecast model can be seen as corresponding to the difference between the flow composition model and the well-specific flow composition model as discussed above.

[092] The set of well-specific forecast parameters can comprise between 0 and 100 parameters. For example, there may be 1 or 10 well-specific forecast parameters.

[093] The single well to which the well-specific forecast model relates may be included within the first plurality of production wells, of, the various or each additional plurality of production wells, of, the various or each second plurality of production wells and / or of, the various or each third plurality of production wells. Similar to the discussion above in connection with the second model and the well-specific flow composition model, greater modeling accuracy can be achieved for this well by virtue of this overlap.

[094] The single well to which the well-specific forecasting model relates may not be included within the first plurality of production wells, or each additional plurality of production wells, or each second plurality of production wells and / or each third plurality of production wells. Thus, assumptions may be drawn from the generic behaviors of any of these well pluralities and applied to the well to which the well-specific forecasting model relates, while the well-specific forecasting model may allow the prediction of traits that are unique to that well.

[095] The well to which the well-specific forecast model relates may be the same well to which, or at least one of the second model(s), relates and / or the same well to which, or at least one of the well-specific flow composition models relates. Where Petition 870230096179, dated 10 / 30 / 2023, page 45 / 106 36 / 81 if there is a match in wells and there is a combination of the well-specific forecast model with the well-specific flow composition model and / or the second model relating to the same well, improved modeling and forecasting accuracy for this well can be achieved.

[096] The method may comprise the generation of a plurality of well-specific forecast models corresponding to the single well-specific forecast model configured above. Each well-specific forecast model may be able to forecast for only one respective production well, a change in a flow parameter, a well parameter and / or the status of at least one control point based on a hypothetical change in the status of at least one control point, a hypothetical change in a well parameter and / or a hypothetical change in a flow parameter, wherein each well-specific forecast model is parameterized by a set of well-specific forecast parameters that are representative of specific properties for the production well to which it relates.The method may also comprise combining each well-specific production model with the first model and, optionally, the various or each additional first model, the various or each second model, the various or each flow composition model, the various or each well-specific flow composition model and / or, various or each forecast model to form combined model(s) that are each capable of predicting a flow parameter, a well parameter and / or the status of at least one control point resulting from the hypothetical change in the status of at least one control point, the hypothetical change in a well parameter and / or the hypothetical change in a flow parameter for each respective production well.

[097] Thus, combined models that can enable predictions Petition 870230096179, dated 10 / 30 / 2023, page 46 / 106 Customized 37 / 81 pipes for a variety of different individual wells can be provided.

[098] Optionally, there may be a well-specific prediction model generated for each well referred to above and below.

[099] The plurality of well-specific forecasting models may be encompassed within a well-specific forecasting model framework. This framework may closely correspond to the second model framework as configured above. Thus, the well-specific forecasting model framework may leverage the corresponding functionality and features that the second model framework optionally does as configured above.

[100] In another aspect of the invention, a method is provided for predicting a flow parameter, well parameter and / or the status of at least one control point for at least one production well, comprising: modeling to produce a combined model incorporating one or more prediction model(s) and / or one or more well-specific prediction model(s) as configured above; and inserting a hypothetical change in the status of at least one control point, a hypothetical change in a well parameter and / or a hypothetical change in a flow parameter associated with at least one production well into the (respective) combined model and thereby obtaining a predicted flow parameter, well parameter and / or status of at least one control point for at least one production well.

[101] As mentioned above, the forecasting model(s) and the well-specific forecasting model(s) generated in the first aspect method can thus be used, as part of their respective combined models, to enable predictions about the performance of Petition 870230096179, dated 10 / 30 / 2023, page 47 / 106 38 / 81 well. Thus, the model produced from the first aspect method can be used as part of the second aspect method to determine how performance will or may have developed and / or to determine how a given change may affect well performance.

[102] The forecast using the combined model may comprise inserting into the forecast model(s) and / or well-specific forecast model(s), prior to their combination as part of the combined model, a hypothetical change in a well parameter, a flow parameter and / or the status at at least one control point in order to determine a change in a flow parameter, well parameter and / or a status of at least one control point. The combination of the forecast model(s) and / or well-specific forecast model(s) into the combined model may then comprise inserting the hypothetical change in a well parameter, a flow parameter and / or the status at at least one control point together with the associated changed flow parameter, well parameter and / or status of at least one control point in the first / combined model in order to provide the forecast.As such, forecasting using the combined model can be bifurcated, where a first set of variables is fed into the forecasting model(s) and / or very specific forecasting model(s) to obtain an output, and then this output (along with the first set of variables) is fed into the first / combined model to obtain the relevant forecast.

[103] The second aspect method may comprise forecasting a flow parameter, a well parameter and / or the status of at least one control point for at least one hydrocarbon production well as configured above; repeating the forecast of a flow parameter, a well parameter and / or the status of at least one control point for at least one hydrocarbon production well as configured above Petition 870230096179, dated 10 / 30 / 2023, page 48 / 106 39 / 81 based on a hypothetical different change in the status of at least one control point, a hypothetical different change in the flow parameter, and / or a hypothetical different change in the well parameter; and determine the status of at least one control point, the flow parameter, and / or the well parameter that is / are optimized and thus allow for optimized hydrocarbon production. As such, the first-aspect method can be used to find an optimized state for the production well (e.g., a state where production rates are maximized). This optimized state can be defined by the status of at least one control point, the well parameters, and / or the flow parameters.

[104] The forecast can be repeated a plurality of times based on a plurality of different hypothetical changes in the status of at least one control point, different hypothetical changes in the flow parameter and / or different hypothetical changes in the well parameter.

[105] An optimization algorithm can be used to determine the status of at least one control point, the flow parameter and / or the well parameter that results in an optimized flow parameter, well parameter and / or status of at least one control point and thus optimized hydrocarbon production.

[106] The forecast and / or optimization set out above can be used as part of a what-if study to determine what effects certain changes may have on production well performance and thus optionally enable optimized performance to be achieved.

[107] The models produced in the first-aspect method can subsequently be used to provide estimates for a production well. This can be achieved by inserting a production well state into the first / combined model produced from the first-aspect method. Petition 870230096179, dated 10 / 30 / 2023, page 49 / 106 40 / 81 to obtain an estimate of a well characteristic for that production well.

[108] Therefore, in another aspect of the invention, a method is provided for estimating a flow parameter, a well parameter and / or the status of at least one control point for at least one hydrocarbon production well, the method comprising: modeling according to any of the statements relating to the first aspect, as set out above; and determining an estimated flow parameter, well parameter and / or status of at least one control point for at least one hydrocarbon production well by inserting into the first model or the (respective) combined model a state of at least one production well, the state comprising a flow parameter, a well parameter and / or an associated status of at least one control point of at least one production well.

[109] Estimates are useful because they allow determinations to be made regarding flow parameters, well parameters, and the status of at least one control point for a production well. These determinations can then be used to make inferences and assessments regarding the production well and its performance – that is, they allow the production well performance to be analyzed.

[110] The state of at least one of the plurality of production wells used in the estimate may be a historical state, a real-time state, or a future state. Future states, in particular, may be derived using the, or each, flow composition model and / or the, or each, well-specific flow composition model, as discussed above, since these models may be descriptive in time and thus allow future states to be determined. Petition 870230096179, dated 10 / 30 / 2023, page 50 / 106 41 / 81

[111] Where the estimate of this aspect and the prediction of the second aspect of the invention differ is that the estimate relates to a state of the production well that has occurred, is occurring, or will occur, or is likely to have occurred, is likely to occur, or is likely to occur. That is, the estimate relates to a state of the well that was, is currently, or will be if the well is left to develop on its own. The prediction of the second aspect relates to hypothetical changes in the state of the well and therefore may and will include states of the well that have not occurred at any point in the life of the production well, nor will they occur in the natural development of the well in its current state.

[112] The estimated / predicted flow parameter, well parameter and / or the estimated status of at least one control point may be a well health indicator, a water cut (WC) of the produced hydrocarbon fluid, a gas-to-oil ratio (GOR) of the produced fluid, a liquid loading risk indicator, a total flow rate of produced fluid (by volume, mass or velocity / flow rate), a gas flow rate, an oil flow rate, a water flow rate, a liquid flow rate, a hydrocarbon flow rate, a carbon dioxide fluid flow rate, a hydrogen sulfide fluid flow rate, a multiphase fluid flow rate, a slag severity, an oil fraction, a gas fraction, a water fraction, a carbon dioxide fraction, a multiphase fluid fraction, a hydrogen sulfide fraction, a gas-to-liquid ratio, density, viscosity, pH, productivity index (PI),BHP and wellhead pressures, post-surface separation rates, separator pressure, other line pressures, flow velocities, or sand production. The estimated / predicted flow parameter, well parameter, and / or the estimated status of at least one control point may, Petition 870230096179, dated 10 / 30 / 2023, page 51 / 106 42 / 81 additionally and / or alternatively be any of these flow parameters, well parameters and / or a status of these control points configured below.

[113] Estimating / predicting a gas flow rate, an oil flow rate, a water flow rate, a carbon dioxide flow rate, or a hydrogen sulfide flow rate may involve modeling using the, several, or each flow composition model and / or several or each well-specific flow composition model. Since the flow composition model(s) and / or well-specific flow composition model(s) describe the flow constituents being produced from the well, these models may be needed to determine the constituent flow rates.

[114] One or more of the models may form part of a statistical approach such that a flow parameter, a well parameter, and / or a status of at least one control point production by one or more models) is generated as a probability distribution with an associated degree of uncertainty. Being able to model and account for the inherent uncertainty within the models by overlaying with a statistical approach is useful because it recognizes that there is both error in the model(s) as they are not perfect reflections of the real-world scenario that they are trying to represent, and since there are inherent errors in the data (e.g., due to recording tolerances or inaccuracies in sensors, meter controls, and the like) on which each model is generated and based. Thus, overlaying a statistical approach provides an understanding of the errors within the model.

[115] At least one control point may be a means / mechanism capable of applying a controlled adjustment to the respective production well, in particular an adjustment to the fluid flow of the production well (for example, the Petition 870230096179, dated 10 / 30 / 2023, page 52 / 106 43 / 81 control points may be able to apply an adjustment to one or more flow parameters. The adjustment can be to any suitable fluid parameter, such as fluid flow and / or pressure. For example, suitable control points may include flow control valves, pumps, compressors, gas lift injectors, expansion devices, and so on. The basic principle of the above methods is compatible with any control that can apply an adjustment within the conduit associated with each of the plurality of production wells. Adjustments need not be solely to flow rate or pressure, but may include other parameters such as a level in a subsea separator and the ESP pump setting.

[116] At least one control point may comprise at least one of: a flow control valve; a pump; a compressor; a gas lift injector; expansion devices; a choke control valve; gas lift valve settings or rates in wells or riser pipelines; ESP (Electric Submersible Pump) settings, effect, speed or pressure lift; downhole branch valve settings, downhole inlet flow control valve settings; or topside and subsea control settings in one or more of: separators, compressors, pumps, scrubbers, condensers / coolers, heaters, extractor columns, mixers, splitters, coolers.

[117] Flow parameters can be properties / characteristics / parameters / behaviors related to the nature of fluid flow, or they can be properties / characteristics / parameters / behaviors related to the nature of the fluid itself. As such, flow parameters can include one or more pressures; flow rate, a gas flow rate, a flow rate of Petition 870230096179, dated 10 / 30 / 2023, page 53 / 106 44 / 81 oil, a water flow rate, a liquid flow rate, a hydrocarbon flow rate, a multiphase flow rate, a flow rate that is the sum of one or more of any of the previous rates (by volume, mass, or flow velocity); an oil fraction, a gas fraction, a carbon dioxide fraction, a multiphase fluid fraction, a hydrogen sulfide fraction, temperatures, a gas-to-liquid ratio, densities, viscosities, molar weights, pH, water cut (WC), productivity index (PI), Gas-Oil Ratio (GOR), BHP and wellhead pressures, rates after topside separation, separator pressure, other line pressures, flow velocities, or sand production. It will be appreciated that the flow parameters of interest would not necessarily include all possible flow parameters associated with a production well.Instead, flow parameters may include a select set of flow parameters that are considered important for the performance of the production well. Flow parameters may be parameters that are impacted, directly or indirectly, by the status of at least one control point and / or well parameters.

[118] Flow parameters can be measured directly, for example, by means of a pressure or temperature sensor, or alternatively they can be measured indirectly, for example, by calculations based on directly measured parameters. Flow parameters can be measurable (i.e., parameters that are readily and commonly measured in connection with production wells by appropriate associated equipment) and / or unmeasurable flow parameters (i.e., those that have no associated recording equipment and / or those that are physically or practically difficult to measure).

[119] Well parameters may include one or more of the following: Petition 870230096179, dated 10 / 30 / 2023, page 54 / 106 45 / 81 depth, length, number and type of joints, inclination, cross-sectional area (e.g., diameter or radius) within / of a production well, wellbore, well branch, pipe, tubing or sections thereof; choke valve Cv-curve; cross-sectional area of ​​the choke valve discharge orifice; heat transfer coefficient (U-value); friction coefficients; material types; insulation types; casing factors; and external temperature profiles. Well parameters may additionally and / or alternatively be one or more of the 'near-well' reservoir parameters. That is, well parameters may include parameters of the reservoir to which the well is attached and which directly impact the performance and behavior of the well.Such reservoir parameters near the well, which can be extracted from production well tests, may include: well productivity index, well casing factor, reservoir permeability, specific reservoir storage, reservoir boundaries.

[120] The first aspect method may further comprise the steps of: (ii) training the first model, or combined model, on data relating to flow parameters, well parameters and / or an associated status of at least one control point of at least two production wells of the first production well plurality; (iii) obtaining an updated set of first parameters from the training of the first model, wherein the updated set of first parameters more precisely parameterizes the properties common to all production wells of the plurality; and (iv) updating the first model, or combined model, based on the updated set of first parameters, wherein the updated first model allows for more precise modeling of any one of the production well plurality. Petition 870230096179, dated 10 / 30 / 2023, pp. 55 / 106 46 / 81

[121] The broad concept of training a model and its associated advantages are well understood in the field of data-driven modeling. That is, generally, a more accurate and improved model can be achieved by virtue of the training step, in particular because the model parameters can be refined and updated during training, so as to provide improved model accuracy through a better fit to the available data.

[122] However, the above detailed training step is unique and advantageous, in that the training is based on data from at least two (i.e., more than one) of the first plurality of production wells. That is, the data are based on at least two independent wells and, as such, any subsequent modeling of a production well performed is based on the first / combined model produced that was trained, at least in part, on data that are independent of and unrelated to the well to be modeled.

[123] In the past, the data modeling training stage was based solely on data recorded from a single well to be modeled. The data used as a basis for training in state-of-the-art modeling techniques may have been data relating exclusively to the well to be modeled or may have been related to a plurality of wells, including the well to be modeled (e.g., where combined data and / or surface data are used in training). In both cases, the state-of-the-art training data are always related to the well to be modeled. In contrast, in the context of the optional training steps of the invention, at least some of the training data are not related to the well to be modeled, but rather to an independent and separate well or wells. Petition 870230096179, dated 10 / 30 / 2023, pp. 56 / 106 47 / 81

[124] This concept of generating and then training the first / combined model based on data from a plurality of different production wells can be considered within the broad concept of 'transfer learning' which, by analogy, can be considered as using 'knowledge' of the behavior of other different and independent production wells to help provide an improved model for modeling a specific production well. In particular, the first / combined model in the first-aspect method may have improved reservoir effect responsibility by virtue of the optional transfer learning involved in its training.This is because typically the at least two production wells within the first production well plurality will comprise wells at various different stages in their operational lives, and therefore the data collected, and thus the model produced, will be better able to explain the reservoir depletion effects that occur during a well's lifespan. Training the first / combined model will also provide better accountability for other physical similarities between the wells in the first plurality, e.g., choke valves, wellbore shapes, etc. The first / combined model may also have improved robustness and be less strongly influenced by the historical data of the well being modeled, since the model is trained on a larger dataset from a plurality of different wells.

[125] To put it another way, by training the first / combined model on data from at least a subset of wells in the first plurality of production wells, the model parameters can be updated to better reflect the true physical properties and characteristics of any production well in the first plurality, since inferences about well behavior (present and future) can be made. Petition 870230096179, dated 10 / 30 / 2023, page 57 / 106 48 / 81 made based on the behavior and corresponding states in other wells within the subset. As such, an updated first / combined model is obtained that better reflects, at least on average, the 'true' behaviors of each of the first plurality of production wells without having deficiencies resulting from the reservoir effect and / or a limited training dataset.

[126] More precisely, the generation and training of the first / combined model based on data from a plurality of different production wells, as discussed in the present invention, can be considered as a form of Multitask Learning (MTL). MTL attempts to leverage data from multiple tasks to improve model performance in all tasks where all or a subset of the tasks are considered related. For example, in the context of the present invention, modeling the flow through a well can be considered as a task. Given data from multiple wells, MTL attempts to simultaneously model all wells. The models are formulated so that a plurality of model parameters are shared across the wells.

[127] To benefit from the advantages of transfer learning / MTL, training should be based on data relating to at least two of the first plurality of production wells. However, the advantages associated with the concept of transfer learning are enhanced when the training of the first / combined model is based on data relating to a larger number of wells and, optionally, the entire first plurality of production wells. A larger number of wells provides a greater amount of data on which the model can be trained, thus providing greater robustness of the first model and better accountability for, for example, reservoir effect, wellbore effect Petition 870230096179, dated 10 / 30 / 2023, pp. 58 / 106 49 / 81 well, throttling valve and other physical similarities between the wells.

[128] As mentioned above, optionally, the first plurality of production wells comprises production wells at various different stages of their operational lives. This is beneficial for the reasons discussed above (i.e., a more eclectic dataset will be used as the basis for training).

[129] The first plurality of production wells may contain production wells that are connected to the same hydrocarbon reservoir to which the well to be modeled is connected. Additionally and / or alternatively, the first plurality of production wells may be connected to one or more different hydrocarbon reservoirs to which the well to be modeled is connected. The various different hydrocarbon reservoirs may be at different stages of their exploration life and / or may have different variations in fluid compositions and constituents. For example, the first plurality of wells may be connected to a reservoir comprising substantially oil, a reservoir comprising substantially hydrocarbon gas, and / or a reservoir anywhere between these two extremes (e.g., a wet gas reservoir).The reservoirs to which the first plurality of production wells may be attached may additionally and / or alternatively comprise a varying degree of water cuts within their produced fluid.

[130] Additionally and / or alternatively, the various or each first additional plurality of production wells, the various or each second plurality of production wells and / or the various or each third plurality of production wells may comprise production wells of the type as described above in connection with the first plurality of wells. Petition 870230096179, dated 10 / 30 / 2023, page 59 / 106 50 / 81 of production.

[131] It is beneficial that the first plurality (and indeed any other plurality) of production wells be connected to a plurality of different hydrocarbon reservoirs, since a larger and more eclectic dataset can be provided, which is beneficial both for the generation of the first / combined model and for the optional training of the first model, which provides a more robust model and more capable of accounting for the dynamic behaviors of a production well, as discussed above.

[132] Training may comprise a plurality of iterative training steps. Each step may be based on a batch of data relating to flow parameters, well parameters and / or an associated status of at least one control point from at least one of a first plurality of production wells. Therefore, in order to train on data from at least two of the first plurality of production wells, the batch(es) used in iterative training must at least (either individually or in combination) be from at least two of the first plurality of production wells.

[133] Each, or several, of the iterative training stages can be based on a different dataset than the other iterative training stages.

[134] The, or at least one, data set and, optionally, several or all data sets, may refer to flow parameters, well parameters and / or an associated status of at least one control point of at least two of the first plurality of production wells and, optionally, more wells.

[135] The, or each, batch of data used in training the model can Petition 870230096179, dated 10 / 30 / 2023, pp. 60 / 106 51 / 81 will be randomly / stochastically selected from the total available data related to flow parameters, well parameters, and / or an associated status of at least one control point from the plurality of production wells.

[136] Training the first model may involve training based on data relating to each well in the first plurality of production wells. When a batch-type approach is implemented in the training steps, this may involve training on a plurality of batches equivalent to the number of wells in the first plurality in a scenario where each is related to flow parameters, well parameters and / or a status of at least one control point of only one of the first plurality of production wells. It will be recognized that fewer iterative batch steps are needed where one or more of the batches are related to data from two or more production wells to train on data from each of the first plurality of wells.

[137] However, it is not necessary that the training of the first / combined model be based on data relating to every well in the plurality, the optional training only needs to be based on at least two of the wells within the first plurality of production wells.

[138] Steps (iii) and (iv) are presented as two separate and sequential steps in the training of the method of the first aspect of the invention. However, in one implementation, steps (iii) and (iv) can be combined into a single step. That is, the steps of obtaining an updated set of first parameters and updating the first / combined model based on the updated set of first parameters can occur in a single stage.

[139] Through the initial generation of the first model, it may be possible Petition 870230096179, dated 10 / 30 / 2023, pp. 61 / 106 52 / 81 generate a set of first parameters that accurately represent the properties common to all the plurality of production wells. In such a scenario, it may not be necessary to change the first parameters of the first model for the first model to accurately model any of the first plurality of production wells. In this event, step (iii) of training the first / combined model may involve obtaining a set of first parameters that are the same as or very comparable to those originally generated in step (i). Once it is confirmed that the first set of parameters resulting from step (iii) is the same as or very comparable to those originally generated in step (i), updating the first parameters in step (iv) can be considered simply as maintaining the first parameters as those that were originally generated.

[140] Training the first / combined model may involve training based on data relating to each well in plurality. When a batch-type approach is implemented in the training steps, this may involve training on a plurality of batches equivalent to the number of wells in a scenario where each is related to flow parameters, well parameters and / or an associated status of at least one control point of only one of the plurality of production wells. It will be recognized that fewer iterative batch steps are needed where one or more of the batches relate to data from two or more production wells to train on data from each of the wells.

[141] Before the training of stage (ii), or before each iterative training stage, the method may comprise inserting the second model or a plurality of second models into the first model / combined model as discussed above. Subsequently, during the training stage (ii) or each iterative training stage, the method may Petition 870230096179, dated 10 / 30 / 2023, pp. 62 / 106 53 / 81 understand obtaining an updated set of second parameters during step (ii), during some and / or during each iterative training step for the second model(s) related to the production well(s), where the updated set of second parameters more precisely parameterizes the specific properties of the production well(s) to which the second model(s) relate(s); and updating the second model structure based on the updated set of second parameters. When iterative training is implemented, the second parameters may not be updated in each of the iterative training steps. They may, for example, be updated only in alternative iterative training steps.Alternatively, additional iterative training steps can be introduced into the iterative training regime where the first parameters are not updated and only the second parameters are updated.

[142] Generating an updated set of second parameters shares many corresponding advantages, as discussed above, with respect to training the first / combined model and generating the first updated parameters. That is, the second parameters can be updated to better reflect the true physical configuration and specific characteristics of the wells to which they relate. As such, an updated second model(s) can be obtained that better reflect the production well(s) and thus allow for improved modeling of said well(s) without having deficiencies.

[143] When one or more second models are introduced into the first / combined model before the optional training step, or before each optional iterative training step, the data used for the training step can only be data relating to the well(s) of Petition 870230096179, dated 10 / 30 / 2023, pp. 63 / 106 54 / 81 production to which the second model(s) relate(s). In this way, the second parameter(s) in the second model(s) is / are adjusted and updated specifically for the well to which they relate, and thus the second model(s) is / are provided with improved specificity for its / their respective well(s).

[144] The first aspect method may comprise introducing at least one additional well into the first production well plurality; retraining the first / combined model on data relating to flow parameters, well parameters and / or an associated status of at least one control point from the at least one additional well; obtaining a re-updated set of first parameters from the retraining of the first / combined model, wherein the re-updated set of first parameters more precisely parameterizes the common properties of the first production well plurality; and updating the first / combined model based on the re-updated set of first parameters.

[145] The at least one additional well may be a well that did not previously exist (i.e., a completely new well) and / or may be an existing well for which data have recently become available.

[146] The at least one additional production well may be multiple production wells. As such, the first / combined model may be retrained on data relating to flow parameters, well parameters and / or an associated status of at least one control point of one, some or all of these multiple production wells.

[147] The method may also comprise the introduction of a second model (optionally as part of the second model framework as discussed above) for at least one additional well; and, before the retraining step, incorporating the second model related to at least one Petition 870230096179, dated 10 / 30 / 2023, pp. 64 / 106 55 / 81 additional well in the first / combined model so that the first / combined model is able to describe a relationship between flow parameters, well parameters and / or an associated status of at least one control point for only the at least one additional well. As such, the first / combined model, prior to the retraining step, is specifically adapted to model at least one additional well. Any updates to the first set of parameters (and optionally, as discussed further below, the second set of parameters) resulting from the retraining step can therefore be ensured to reflect and take into account the behaviors and characteristics of the at least one additional well.

[148] The method may comprise obtaining an updated set of second parameters for the second model relative to at least one additional well from the retraining step of the first / combined model, wherein the updated set of second parameters more precisely parameterizes the specific properties of the at least one additional well; and updating the second model relative to at least one additional well based on the updated set of second parameters relative to at least one additional well. In fact, it is imagined that during retraining it may not be necessary to update the first parameters after the addition of at least one well, since the first parameters may have converged from previous training / retraining steps.Therefore, retraining may involve only an update of the second parameters to account for at least one additional well, where the update of the first parameters can simply be considered as maintaining the first parameters at the value to which they converged.

[149] Similar to the case for training the first / model Petition 870230096179, dated 10 / 30 / 2023, pp. 65 / 106 56 / 81 combined, the retraining of the first / combined model may comprise a plurality of iterative retraining steps. Each step may be based on a different batch of data relating to flow parameters, well parameters, and / or an associated status of at least one control point of at least one additional well.

[150] When iterative retraining is implemented, the second parameters may not be updated in each of the iterative retraining steps. They may, for example, be updated only in alternative iterative training steps. Additionally and / or alternatively, iterative retraining steps may be introduced into the iterative retraining regime in which there is no update of the first parameters, only an update of the second parameters.

[151] In scenarios where at least one additional well is multiple additional wells, a second model for each of the multiple additional wells can be generated / introduced (optionally in the second model structure). This ensures that there are second models that can explain the specific behaviors of each of the multiple additional wells.

[152] Each batch of data used in model retraining can be randomly / stochastically selected from the total available data related to flow parameters, well parameters and / or an associated status of at least one control point of the additional well(s).

[153] A corresponding retraining step of the first / combined model can also be implemented not only when additional wells are newly introduced into the first production well plurality, but also and / or alternatively when new data become available for existing wells within the first plurality. Petition 870230096179, dated 10 / 30 / 2023, pp. 66 / 106 57 / 81 of production wells. That is, the first aspect method may also include obtaining additional data relating to flow parameters, well parameters and / or an associated status of at least one control point from at least one of the first plurality of production wells; retraining the first / combined model on the additional data; obtaining a re-updated set of first parameters from the retraining of the first / combined model, where the re-updated set of first parameters more precisely parameterizes the common properties of the plurality of production wells; and updating the first / combined model based on the re-updated set of first parameters.

[154] Before the retraining stage, the method may involve introducing / incorporating the second model relating to the well for which additional data were obtained into the first / combined model so that the resulting combined model is able to describe a relationship between flow parameters, well parameters and / or an associated status of at least one control point for only at least one well from which additional data were obtained.

[155] The method may also involve obtaining, from the retraining stage, an updated set of second parameters for the second model in relation to at least one well from which additional data were obtained, wherein the updated set of second parameters more accurately parameterizes the specific properties of at least one of the production wells for which additional data were obtained; and updating the second model relative to the well for which additional data were obtained based on the updated set of second parameters. In fact, it is imagined that during retraining it may not be necessary to update the Petition 870230096179, dated 10 / 30 / 2023, pp. 67 / 106 58 / 81 first parameters after additional data have been obtained, since the first parameters may have converged from previous training / retraining steps. As such, retraining may only involve updating the second parameters to account for at least one additional well, where the update of the first parameters can simply be considered as maintaining the first parameters at the value to which they converged.

[156] As is the case for first / combined model training, first / combined model retraining may comprise a plurality of iterative retraining steps. Each step may be based on a different batch of data relating to flow parameters, well parameters and / or an associated status of at least one well from which additional data have been obtained.

[157] When iterative retraining is implemented, the second parameters may not be updated in each of the iterative retraining steps. They may, for example, be updated only in alternative iterative training steps. Additionally and / or alternatively, iterative retraining steps may be introduced into the iterative retraining regime where there is no update of the first parameters, only an update of the second parameters.

[158] Additional data can be obtained for several, or all, of the first plurality of production wells.

[159] When additional data have been obtained from several (or all) of the plurality of wells, the method may comprise, before the retraining step (or each iterative retraining step), inserting the second models relating to the wells for which additional data were obtained into the first / combined model so that the first / combined model is Petition 870230096179, dated 10 / 30 / 2023, pp. 68 / 106 59 / 81 capable of describing a relationship between flow parameters, well parameters and / or an associated status of at least one control point for the wells for which additional data were obtained.

[160] If an iterative approach is adopted for retraining the first model in scenarios where additional data are obtained from several production wells, at least one data set and, optionally, several or all data sets may be related to flow parameters, well parameters and / or an associated status of at least one control point from at least two, and optionally more, of the several wells.

[161] Each batch of data used in model retraining can be selected randomly / stochastically from the total of additional data available.

[162] The optional retraining steps of the first / combined model, as configured above, provide the modeling with good adaptability, so that new wells and / or new data can be accounted for in the existing first / combined model without the need to generate a completely new model. The retraining will take into account the necessary refinements of the first / combined model (due to the updated initial parameters) to incorporate the behavior of the newly added wells / data. Thus, the retrained model can be used to model both existing and new wells in plurality in a relatively efficient and computationally inexpensive way.

[163] The use of retraining to obtain an updated / re-updated set of second parameters provides an adaptability corresponding to the second model(s) as transmitted to the first model by said retraining. By virtue of retraining, the second model(s) can be made to take into account the data Petition 870230096179, dated 10 / 30 / 2023, pp. 69 / 106 60 / 81 additional existing and / or new wells due to a modification of the second parameters.

[164] The optional retraining steps of the first / combined model and the additional optional steps resulting from the first aspect method can be repeated each time an additional well / wells is / are added to the first well plurality and / or new data becomes available from any of the existing wells in the first production well plurality.

[165] As an alternative to retraining the first / combined model, when new wells are added to the first plurality (i.e., when new data become available from a new set of wells not previously available) and / or when new data become available for existing wells within the first plurality, the first / combined model can be generated and retrained based on all available data relating to the first plurality of production wells. That is, the first aspect method can simply be repeated when additional wells are added to the first plurality of wells and / or new data for existing wells in the first plurality of production wells become available.Similarly, the second model(s) may be generated again when new wells are added to the first plurality of production wells (i.e., when new data become available from a new well / new set of wells not previously available) and / or when new data become available for existing wells within the first plurality.

[166] The training and retraining steps (and related concepts of the invention) as described above were described in the context of (re)training the first model / combined model to (re)update the Petition 870230096179, dated 10 / 30 / 2023, pp. 70 / 106 61 / 81 first set of parameters and / or incorporate the second model(s) into the first / combined model and (re)training to (re)update the second set(s) of parameters from the second model(s) into the first / combined model.

[167] However, in addition to, or as an alternative to, these training / retraining steps described above (and related concepts of the invention), the corresponding training and / or retraining steps (and related concepts of the invention) may be implemented in relation to / for (re)updating one or more of: the first set(s) of parameters of the first additional model(s), the first set(s) of flow composition parameters of the flow composition model(s), the second set(s) of flow composition parameters of the well-specific flow composition model(s), the prediction parameters of the prediction model(s), the well-specific prediction parameters of the well-specific prediction models.As noted above, retraining to update any of these parameter sets can occur instead of or in addition to retraining to update the first and / or second parameter sets.

[168] The (re)updating of any of the parameter sets resulting from the (re)training can occur in parallel or separately from the updating of any of the other parameter sets.

[169] Given the correspondence between the first model and the first additional model(s), the forecasting model(s) and the flow composition model(s) as described above, it will be appreciated that any (re)training performed in relation to the parameters relating to any of the first models, the forecasting model(s) and the additional flow composition model(s) can be performed in a closely related manner. Petition 870230096179, dated 10 / 30 / 2023, pp. 71 / 106 62 / 81 corresponding (i.e., mutatis mutandis) to that described above in relation to the first set of parameters of the first model.

[170] Given the correspondence between the second model(s), the well-specific flow composition model(s) and the well-specific prediction model(s) as described above, it will be appreciated that any (re)training performed in relation to the parameters relating to the well-specific flow composition model(s) and / or the well-specific prediction model(s) can be performed in a manner that closely corresponds (i.e., mutatis mutandis) to that described above in relation to the second set(s) of parameters of the second model(s).

[171] Thus, as described above, optional training steps, obtaining an updated set of parameters, and updating the combined model can be implemented in relation to the, several, or each first additional model, the, several, or each, flow composition model, the, several, or each well-specific flow composition model, the, several, or each prediction model, and / or the, several, or each well-specific prediction model. These steps can be performed in a mutatis mutandis manner to the corresponding steps applied to the first model / second model as described above, and can occur in combination (e.g., in parallel) with or as an alternative to each other. Training of any of these models can be performed before their combination as part of the combined model, or it can be performed after the combination in the combined model.

[172] Any updated set of parameters obtained from training can more precisely parameterize the properties relevant to the production well(s) to which the respective model(s) relate(s). Petition 870230096179, dated 10 / 30 / 2023, pp. 72 / 106 63 / 81

[173] Generating an updated set of parameters for either model shares many corresponding advantages, as discussed above, with respect to training the first / combined model and generating the first updated parameters. That is, the parameters can be updated to better reflect the true physical configuration and characteristics of the well / wells. As such, updated models can be obtained that better reflect each of the production wells to which they relate and thus allow for better modeling accuracy.

[174] The data used as the basis for generating or training the models can be measured directly in relation to the status of at least one control point, the flow parameters and / or well parameters. This type of 'raw' data is usually gathered into a real-time database by an operator for a production flow / well network and is stored as an operating log.

[175] The data used as the basis for generating and / or training any of the models may additionally and / or alternatively be data resulting from mining and / or compacting original raw data. The compacted data may be derived from the large volumes of raw data recorded in relation to oil and gas production wells, which are then categorized and compacted based on the categorization of the datasets within time intervals and by the use of statistics. The resulting statistical data may represent certain aspects of the original data in a much more compact form and may also be more readily searchable to identify events or patterns of events. This statistical data may be stored in a compact database, on which the input for training / retraining the first aspect may be based. The statistical data may provide information about the operation and Petition 870230096179, dated 10 / 30 / 2023, pp. 73 / 106 64 / 81 behaviors of the plurality of production wells without the need for all the original raw data. Data compression methods for production well data are described in the Applicant's patent publications WO 2017 / 077095 and WO 2018 / 202796 A1. The methods disclosed in these publications can be used to provide a compressed dataset that forms the basis of the training and / or retraining steps of the present invention.

[176] For example, the first aspect method may comprise: (1) collecting data covering a time period related to flow parameters, well parameters and / or an associated status of at least one control point; (2) identifying multiple time intervals in the data during which at least one control point, flow parameters and / or well parameters can be designated as being in a category selected from multiple categories related to different types of stable production and multiple categories related to different types of transient events, wherein the data therefore include multiple datasets, each framed by one of the multiple time intervals; (3) assigning a category selected from the multiple categories to each of the multiple datasets that are framed by the multiple time intervals;and (4) extract representative statistical data from some or all of the datasets identified in step (2) in order to represent the original data from step (1) in a compact form, including details of the category assigned to each time interval in step (3).

[177] Steps (1) to (4) can be performed before the generation of any of the models and / or the training step. Thus, the generation of any model and / or its training steps can be based on the data in a compact form. Petition 870230096179, dated 10 / 30 / 2023, pp. 74 / 106 65 / 81

[178] In some circumstances, data compression in step (4) is not necessary and, in fact, steady-state intervals can be used directly for model training and / or generation.

[179] The data used in the invention may include data points that are related to only a single well. That is, the data may only be representative of flow parameters, well parameters and / or an associated status of at least one control point from at least one of a first plurality of production wells. Such data may, for example, be collected at a test separator where only the output of one of the plurality of production wells is being fed to said test separator. Alternatively, such data may, for example, be from, or derived from, a flow meter positioned within only one flow path associated with a production well.

[180] Additionally and / or alternatively, the data used in any of the steps of the method may include data points that are related to, or are derived from, data points that are related to multiple wells. As an example, the data may include or be derived from topside data / measurements, where the topside data / measurements are related to several wells. Such data points may include data / measurements collected from flow meters within a flow path containing mixed flow from multiple production wells. These data points may alternatively be from a separator to which the flow from several of the plurality of production wells is directed. As a further example, mass balance equations for mixed flow (based on data relating to several of the plurality of production wells) may be used to create virtual measurements for individual production wells that are not measured.Thus, each data point used can be related to or derived from other points. Petition 870230096179, dated 10 / 30 / 2023, pp. 75 / 106 66 / 81 of data relating to more than one production well. The Applicant's prior patent publication, in document WO 2019 / 110851, further details the use of topside data as the basis for model training, and the use of such data described therein can also be used in the context of the present invention. However, in the context of the present invention, this data can be used in a transfer learning context, rather than for training specific well models, as disclosed in document WO 2019 / 110851.

[181] The generation of any of the models referred to in the present invention may be considered as designing the architecture of a mathematical model and / or a statistical model and / or a data-oriented model and / or a machine learning model and / or a neural network model and / or decision trees and / or support vector machines and / or regression models and / or Bayesian networks and / or genetic algorithms, wherein the design includes, but is not limited to, specifying the number of parameters / variables; specifying the mathematical relationship between random variables and other non-random variables; specifying the relationships and variables / parameters where the relationships can be described as operators, such as algebraic operators, functions, differential operators, and where the variables are abstractions of parameters of the system of interest that can be quantified; and / or specifying the activation functions, connections and weights and / or logical rules.This means that any of the model's parameters / variables can be quantified and, optionally, trained from data from one or more production wells; and / or that any of the models can be used by inputting data from one or more wells to estimate, predict, and / or optimize as configured above.

[182] The combination of any of the models mentioned above Petition 870230096179, dated 10 / 30 / 2023, pp. 76 / 106 67 / 81 may include: specifying the relationship / operators of one or more models to ensure that an output (e.g., in the form of data) from one or more models becomes an appropriate input (e.g., in the form of data) for another or more models. The output of one or more models may be summarized, multiplied, and / or (weighted) averaged with an output from another or more models before input into another or more models.

[183] ​​The aspects of the invention described above will have to be implemented in a type of computer system. That is, the methods described above are necessarily computer-implemented methods.

[184] Thus, in a further aspect of the invention, a computer system is provided for modeling one of a plurality of production wells, for estimating a flow parameter, a well parameter and / or the status of at least one control point for at least one hydrocarbon production well and / or for predicting a flow parameter, a well parameter and / or the status of at least one control point for at least one hydrocarbon production well, wherein the computer system is configured to perform the method of any of the aspects as configured above.

[185] In a further aspect, a computer program product is also provided comprising instructions for execution on a computer system disposed to receive data relating to flow parameters, well parameters and / or an associated status from at least one control point of the plurality of production wells; wherein the instructions, when executed, will configure the computer system to perform a method of any of the aspects configured above.

[186] Certain embodiments of the present invention will now be described, by way of example, and with reference to the accompanying drawings, in which: Petition 870230096179, dated 10 / 30 / 2023, pp. 77 / 106 68 / 81 Figure 1 is a schematic of a generic architecture for modeling the flow rate for one of a plurality of production wells according to an embodiment of the invention; Figure 2 is a schematic of an architecture for throttling flow modeling according to an embodiment of the invention; Figure 3 is a schematic of a wellbore modeling architecture according to one embodiment of the invention; and Figure 4 is a schematic of an alternative generic architecture for modeling according to an embodiment of the invention.

[187] Figure 1 shows a transfer learning architecture having a first model 1 composed of a neural network and a second model structure 3 comprising a plurality of second models 5. In this embodiment, each second model 5 consists of a set of second parameters β in vector form. As such, the second model structure 3 can be considered as a second model matrix.

[188] The first model 1 is capable of modeling the fluid flow rate of any one of a plurality of hydrocarbon production wells and comprises a set of 7 first parameters θ. The first model 1 is initially generated from a desired specification, which includes the variables to be introduced into the model, the desired output variables (in this case, fluid flow rate), the model architecture, and the model / number of model parameters. Once the first model 1 has been generated according to the desired specification, the set of 7 first parameters θ is stochastically generated and inserted into the first model 1 to initialize the first model 1. The set of 7 first parameters θ within the first model are representative of the physical properties and characteristics common to all production wells in the plurality and allow Petition 870230096179, dated 10 / 30 / 2023, pp. 78 / 106 69 / 81 that the model takes such behaviors into account when modeling a specific production well.

[189] Each second 5 model represents one of the plurality of production wells and is capable of describing a relationship between flow parameters, well parameters, and / or an associated status of at least one control point for that production well. As noted above, in this embodiment, each second 5 model consists of a set of second β parameters. The set of second β parameters are specific to the related production well within the plurality and are representative of properties that are specific to that production well. After the initial generation of each of the second 5 models, the second β parameters are stochastically generated to initialize each of the second 5 models.

[190] The second model structure 3 is generated from the plurality of second models 5. This comprises a concatenation of each of the plurality of second models 5.

[191] After the initial generation of the first model 1 and the second model structure 3, the training stage of the first model 1 is initiated. The objective of the training is to update the first parameters θ and the second parameters β within the first model 1 and the second model structure 3, respectively, so that the first parameters θ more precisely parameterize the properties common to all the plurality of production wells and the second parameters β more precisely parameterize the specific properties of each of the plurality of production wells. As a result, the first model 1 will more precisely describe for any of the plurality of production wells a relationship between flow parameters, well parameters and / or an associated status of at least one control point compared to the first model. Petition 870230096179, dated 10 / 30 / 2023, pp. 79 / 106 70 / 81 originally initialized 1 comprising first stochastically generated parameters θ. Similarly, as a result of training, each second model 5 will more accurately describe a relationship between flow parameters, well parameters and / or an associated status of at least one control point for its production well compared to each of the respective second initialized models comprising the stochastically assigned parameters.

[192] Training is achieved by introducing data 9 relating to flow parameters, well parameters and / or an associated status of at least one control point associated with each of the plurality of production wells in the first model 1. In this mode, the data input 9, which supports the training procedure, is data 9 from each (i.e., all) of the plurality of production wells.

[193] In this mode, training the first model 1 initially involves determining a series of training steps that should form the basis of the training procedure before termination (although in other modes an adaptive training regime may be implemented, for example, where a termination condition determines the number of training steps instead of a predetermined number of steps). Once the number of training steps is determined, training begins by stochastically selecting a data set from the total data set 9 available in relation to the plurality of production wells. The data set may contain data from a single well within the plurality, from multiple wells, or it may contain topside data representative of multiple wells. The exact nature of each data set will be determined before the start of training the first model 1 and will depend on the specific iterative training regime to be implemented. Petition 870230096179, dated 10 / 30 / 2023, pp. 80 / 106 71 / 81

[194] After selecting a data batch, a signal 11 is created for the data batch. Signal 11 is specific only to those of the plurality of production wells from which the data batch is derived. The effect of signal 11 is such that, after inputting signal 11 into the second model structure 3, only the second models 5 relating to the wells from which the data batch was collected (i.e., only the second parameters β that relate to the wells from which the data batch was collected) remain within the second model structure 3. As such, after inputting the well-specific signal 11, the second model structure 3 is specifically adapted to model only those of the plurality of production wells to which signal 11 is related.

[195] In the present embodiment, the input signal 11 in the second model structure 3 is in the form of a binary vector. As such, the input operation of signal 11 in the second model structure 3 involves a simple vector-matrix multiplication, where the result is a second fitted and contracted model structure 3 containing only the second models 5 relative to the second models from which the data in the training batch were derived.

[196] Once the second adapted model structure 3 is produced such that only the second parameters β relating to the production wells from which the training data batch originates are included, the second model structure 3 is inserted into the first model 1. In this specific embodiment, this is achieved by producing a plurality of copies of the first model 1 equal to the number of second models 5 in the second model structure 3. Subsequently, each second model 5 of the structure of the second adapted model is fed into its own respective copy of the first model 1 to form a combined model. The resulting combined models will thus be adapted for modeling the specific well to which the second input model 5 is related. Petition 870230096179, dated 10 / 30 / 2023, pp. 81 / 106 72 / 81

[197] At this stage, data 9 from the selected batch are run on the (copies) of the combined model. Only data 9 specifically related to the production well to which the (or each copy of the) adapted combined model is related are entered into the (or each copy of the) combined model.

[198] The input data 9 for each of the combined models, which can also be considered as first adapted models 1, result in an estimated flow rate for the specific production well to which the first adapted model is related. This estimated flow rate is then compared with a flow rate 13 actually measured for that production well at the time the input data were collected. This comparison allows the calculation of a batch loss, which can be considered as an error of each first adapted model 1 in the batch data (i.e., a discrepancy between the estimated and measured flow rate 13). From this batch loss, the gradients of the batch loss with respect to the first θ and second model parameters β can be calculated.These gradients are then used to update the first model parameters θ and second model β to create a first model 1 and a second model structure 3 having reduced batch loss. This updating of the first model θ and second model β parameters to reduce batch loss occurs in parallel across each copy of the first model 1 needed for training at this stage on this batch of data. This training stage is then terminated and the first model 1 and second model structure 3 are updated based on the resulting updated first model parameters θ and second model β.

[199] After the end of this iterative training stage, a new batch of data 9 is stochastically selected and the training stages Petition 870230096179, dated 10 / 30 / 2023, pp. 82 / 106 The resulting 73 / 81, as configured above, are repeated to obtain the updated first model parameters θ and second β. This is iterated for data from each of the plurality of production wells until the predetermined number of training steps has been completed.

[200] Additional specifics of model training are presented in equation (4) below: (4) (θ*. Pi.....Pm) = ar0min(eA.....β= = ^=i Σ / 1 ( h^el(uil,xil))

[201] In equation (4) Uij.Xij represents the input data batch in each iterative step of the method. Here each batch is of size one (i.e., consists of a single data point i for well j), and includes both control variables Uj and state measures x^. άθ,β. represents the first adapted model 1 (i.e., the combined model), which has the structure of the second model 3 relative to the production wells from which the data batch was derived incorporated into it. (θ*. β* Pm) represents the first updated parameters θ and second parameters β obtained from the well training. M represents the number of production wells within the plurality, j represents the index of each well, and Nj represents the data points for each well. The model is trained by solving equation (4) using a stochastic gradient descent (SGD) method as described in general terms above.

[202] After training is complete, an updated first model 1 and a second model structure 3 are obtained, with the updated parameters of the first model θ and the second model β resulting from the iterative training regime. These updated parameters provide both the first model 1 and the second model structure 5 with improved accuracy. Petition 870230096179, dated 10 / 30 / 2023, pp. 83 / 106 74 / 81 in modeling generic well behaviors and specific well behaviors, respectively.

[203] The resulting first trained model 1 and the second model structure 5 can then be used to estimate the flow rate for any of the plurality of production wells. As such, estimates based on a state comprising flow parameters, well parameters and / or an associated status of at least one control point of one of the plurality of production wells can be made for any of the plurality of production wells. This would involve inputting such a state into trained model 1 with the additional input of those second parameters β (i.e., that second model 5) related to the production well for which the estimate is being made. The relevant second parameters β can again be selected from the second model structure 3 by inputting an appropriate well-specific signal into the second model structure 3. Equation (5) sets out an estimate made using the first trained model 1 and the second model structure 3. (5)yfj^^fapXj)

[204] Here, U[j, x^ belongs to the state of the production well for which the estimate is being made, h^p* represents the first trained model 1 (incorporating the first updated parameters θ*) having the second relevant trained model structure 3 (incorporating the second updated parameters β *) input to form a combined model, ey^j represents the estimated flow rate of the production well.

[205] The fact that training in this modality is based on data 9 from each of the wells within the plurality of production wells ensures that, in particular, the first model 1 has improved accountability for, for example, the reservoir effect. It also helps to ensure that the first Petition 870230096179, dated 10 / 30 / 2023, pp. 84 / 106 75 / 81 model 1 is not influenced solely by the limited data from a single well. As such, any estimates made using the first trained model 1 and the second model framework 5 can, by virtue of the training, be assured of improved accuracy with a reduced probability of error resulting from misaccounting, for example, for the reservoir effect and / or a limited training dataset.

[206] Furthermore, not only can the estimates made take into account those properties and behaviors that are common across the entire plurality of production wells without being heavily misguided due to the reservoir effect and / or a limited training dataset by virtue of the first model 1, by virtue of the second refined parameters β within the second model structure 3, the estimates made using the combination of the first trained model 1 and the second model structure 3 can accurately take into account those properties and behaviors specific to each of the plurality of production wells.

[207] Figure 2 is a schematic of a transfer learning architecture designed specifically for modeling throttling flow through throttling valves within the flow paths associated with each of the plurality of production wells. The architecture in Figure 2 can be seen as a more specific example of the architecture underlying the modality in Figure 1 and thus shares many of the same corresponding characteristics. For example, the architecture in Figure 2 comprises a first model 1 in the form of a neural network and a second model structure 3 comprising a plurality of second models 5. Petition 870230096179, dated 10 / 30 / 2023, pp. 85 / 106 76 / 81

[208] As in the embodiment above, the second model structure 3 initially incorporates a second model 5 for each of the plurality of production wells. Each second model 5 comprises a set of second parameters β representative of specific behaviors and properties for each of the plurality of production wells. Then, by inputting a well-specific signal 11 relating to the production wells from which the training data were obtained, a second adapted model structure 3 comprising only the second models 5 relating to the production wells from which the training data were obtained is produced. This is the second model structure 3 shown in Figure 2, with the step of inputting the well-specific signal 11 to contract the second model structure 3 down into its adapted form as described above not being shown in this Figure.

[209] As is also the case with the modality in Figure 1, the first model 1 of the modality in Figure 2 comprises a set of first parameters θ representative of behaviors and properties common to each of the plurality of production wells.

[210] In this embodiment, the second model structure 3 maps the throttling position to throttling conductivity (which can be considered as the resistance to flow through each of the throttling valves). In view of this, the second model structure 3 of the embodiment of Figure 2 differs from that of the embodiment of Figure 1 in that the second models 5 comprise more than just the parameters of the second model β; they additionally comprise an element that allows the input of a position 21 of a throttling valve so that the resistance to flow compared to the position 21 of the throttling valve can be mapped by each of the second models 5. Petition 870230096179, dated 10 / 30 / 2023, pp. 86 / 106 77 / 81 As such, the second model structure 3 of the modality in Figure 2 allows for a simpler interpretation of each of the second models 5, the second parameters β and their output.

[211] The type and sizing of the choke valve may differ from well to well and therefore it is desirable to have a well-specific model 5 that maps the choke position 21 to choke conductivity for each plurality of production wells.

[212] The training and subsequent estimation performed using the model architecture of Figure 2 largely correspond to the training and estimation described above in relation to the modality of Figure 1 and, as such, will not be described again here in detail. Where the training / estimation of the modality of Figure 2 differs, however, is that, in addition to the specific well signal 11, the choke position 21 is entered into the second model structure 3 before each iterative training and / or estimation step. From said input, a mapping of the choke position to the choke conductivity 23 is output from the second model structure 3 and it is this second model output 23 that is entered into the first model 1, along with the data 9, before each iterative training step and / or an estimation of a well feature 13 using the second model architecture.

[213] The model architecture of the modality in Figure 2 can take into account the behaviors and properties that are common to each of the plurality of production wells by virtue of the first model 1 and can, additionally, take into account the throttling conductivity, which is a behavior / property that is specific to each plurality of production wells by virtue of the second model structure 3.

[214] Figure 3 is a diagram of a learning architecture by Petition 870230096179, dated 10 / 30 / 2023, pp. 87 / 106 78 / 81 additional transfer. The transfer learning architecture of Figure 3 is specifically designed for well modeling. The architecture of Figure 3 can be seen as a more specific example of the architecture underlying the modality of Figure 1 and thus shares many of the same corresponding characteristics. For example, the architecture of Figure 2 comprises a first model 1 in the form of a neural network and a second model structure 3 comprising a plurality of second models 5.

[215] As in the above embodiment, the second model structure 3 initially incorporates a second model 5 for each of the plurality of production wells. Each second model 5 comprises a set of second parameters β representative of specific behaviors and properties for each of the plurality of production wells. Then, by inputting a well-specific signal 11 relating to the production wells from which the training data were obtained, a second adapted model structure 3 comprising only the second models 5 relating to the production wells from which the training data were obtained is produced. This is the second model structure 3 shown in Figure 3, with the well-specific signal input step 11 to contract the second model structure 3 down into its adapted form thus not being shown in this Figure.

[216] As is also the case with the modality in Figure 1, the first model 1 comprises a set of first parameters θ representative of behaviors and properties common to each of the plurality of production wells.

[217] In the embodiment of Figure 3, the second model structure 3 consists merely of the second model parameters β, which help to capture the unique relationship for each wellbore between the total flow rate and the Petition 870230096179, dated 10 / 30 / 2023, pp. 88 / 106 79 / 81 data 9 relating to flow parameters, well parameters and / or an associated status of at least one production well control point associated with that well. That is, the parameters of the second model β capture those unique properties for each wellbore and which cannot be generalized across all wells within the first parameters θ.

[218] The training and subsequent estimation performed using the model architecture in Figure 3 largely correspond to the training and estimation described above in relation to the modality in Figure 1 and, as such, will not be described again here in detail.

[219] The model architecture of the modality in Figure 3 can take into account the behaviors and properties that are common to each of the plurality of production wells by virtue of the first model 1 and can, additionally, take into account those that are a unique result of the wellbore to which each production well is connected by virtue of the second model structure 3.

[220] Figure 4 shows an alternative generic architecture for modeling according to alternative modalities. The architecture in Figure 4 shares many similarities with that represented in Figure 1. In particular, the architecture in Figure 4 comprises a first model 1 composed of a neural network and a second model structure 3 comprising a plurality of second models 5. As in Figure 1, each second model 5 consists of a set of second parameters β in vector form. As such, the second model structure 3 can be considered as a second model matrix. The first model 1 and the second model structure 3 of Figure 4 are directly comparable to the corresponding models discussed above in connection with Figure 1 and can be trained and used as a basis for estimation in a manner corresponding to that described above in Petition 870230096179, dated 10 / 30 / 2023, pp. 89 / 106 80 / 81 connection with the architecture of Figure 1.

[221] Where the architecture of Figure 4 differs from that described above in connection with Figure 1, however, is that instead of incorporating each second model 5 into a respective copy of the first model 1 before data input 9 (either during training or estimation as described above in connection with Figure 1), the relevant data 9 are entered into the first model 1 before the second model 5 is entered into the respective first model 1. This is an alternative approach to the modeling architecture discussed above and is a common approach for neural network-based modeling. That is, in the resulting neural network that forms the combined model (i.e., the first adapted model 1) in the context of the Figure 4 modality, the shared (rigid) parameters are part of the first layers of the architecture and the specific parameters are part of the last layer (or layers) of the neural network.

[222] The embodiments described above set forth in detail the aspects of the invention relating to the first model, the second model and their combination with each other. It also defines in detail how the first and second models can be trained and how an estimate can be achieved using the combined model resulting from the first and second models. This description, therefore, gives an appreciation of specific embodiments of the invention, and it will be evident to those skilled in the art how those aspects of the invention that have been described in detail can map to those that are not part of the specific embodiments in the present invention.

[223] For example, from the above discussion in connection with the first model and how it is generated, trained and used as a basis for estimation, the subject matter expert will gain an understanding of how the, or each, first additional model, the, or each, forecasting model and the, or each, model Petition 870230096179, dated 10 / 30 / 2023, pp. 90 / 106 81 / 81 flow composition can be generated, trained, and used as a basis for estimation and / or forecasting given the correspondence between the structure and architecture of these models.

[224] Similarly, from the above discussion in connection with the second model, and how it is generated, trained and used as an estimation basis, the person skilled in the art will gain an understanding of how the, or the, or each, well-specific forecast model, and the, or each, well-specific flow composition model can be generated, trained and used as an estimation and / or forecast basis given the correspondence between the structure and architecture of these models.

[225] The combination of the first and second models, as described above, also provides an understanding of how either of the models of the invention can be combined with each other as part of a combined model for modeling and further use in estimation, forecasting and optimization. Petition 870230096179, dated 10 / 30 / 2023, pp. 91-106

Claims

1 / 19 CLAIMS 1. A method for modeling a hydrocarbon production well from among a first plurality of hydrocarbon production wells, each production well being associated with at least one control point in a flow path associated with it, the method characterized in that it comprises operating a computer system to: receive, for each hydrocarbon production well of the first plurality of hydrocarbon production wells, respective data relating to flow parameters (9) of the respective hydrocarbon production well and / or well parameters (9) of the respective hydrocarbon production well and / or an associated status (9) of at least one control point in a flow path associated with the respective hydrocarbon production well, wherein the wells of the first plurality of wells are not connected to the same hydrocarbon reservoir; and generate, using data-driven modeling,a first data-driven well model (he; he β) that models, for any individual hydrocarbon production well of the first plurality of hydrocarbon production wells: hydrocarbon production well flow parameters, and / or hydrocarbon production well well parameters and / or an associated status of at least one control point in a flow path associated with the hydrocarbon production well, wherein the first data-driven well model is parameterized by a set of first model parameters (7) representative of properties common to all the first plurality of hydrocarbon production wells, and wherein generating the first data-driven well model comprises processing the data received in the computer system using data-driven modeling to determine, for each of the first model parameters,a respective value of the first model parameter depending on the data received for all of the first plurality of hydrocarbon production wells.

2. Method according to claim 1, characterized in that it further comprises: receiving, for each hydrocarbon production well of a further, different plurality of hydrocarbon production wells, respective data relating to flow parameters (9) of the respective hydrocarbon production well, and / or well parameters (9) of the respective hydrocarbon production well, and / or an associated status (9) of at least one control point in a flow path associated with the respective hydrocarbon production well;generate, using data-driven modeling, at least one additional first data-driven well model that models, for any individual hydrocarbon production well of the additional plurality of hydrocarbon production wells: hydrocarbon production well flow parameters, and / or hydrocarbon production well well parameters, and / or an associated status of at least one control point in a flow path associated with the hydrocarbon production well, wherein the at least one additional first data-driven well model is parameterized by an additional set of first model parameters (7) representative of properties common to all the additional plurality of hydrocarbon production wells;and whereby generating the first additional data-driven well model comprises processing the data received in the computer system using data-driven modeling to generate the additional set of first well model parameters in dependence on the data received for all of the additional plurality of production wells; and combining the first data-driven well model (he; he p) with at least one first additional data-driven well model to form a combined model that models, for any individual hydrocarbon production well of the first plurality or the additional plurality of hydrocarbon production wells: hydrocarbon production well flow parameters, and / or hydrocarbon production well well parameters and / or an associated status of at least one control point in a flow path associated with the hydrocarbon production well.

3. Method according to claim 2, characterized in that it further comprises: receiving, for each hydrocarbon production well of a plurality of additional, different pluralities of hydrocarbon production wells, data relating to flow parameters (9) of the respective hydrocarbon production well, and / or well parameters (9) of the respective hydrocarbon production well, and / or an associated status of at least one control point (9) in a flow path associated with the respective hydrocarbon production well; generating, using data-driven modeling, a plurality of additional data-driven first well models, each additional first well model (he;he^) model, for a respective individual hydrocarbon production well of the respective additional plurality of production wells, a relationship between flow parameters, well parameters and / or an associated status of at least one control point, wherein each first additional model is parameterized by a set of first well model parameters (7) representative of properties common to the respective additional plurality of production wells, and wherein generating each first additional data-driven well model comprises processing the data received in the computer system using data-driven modeling to generate the respective additional set of first well model parameters in dependence on the data received for all of the respective additional plurality of production wells;and combine the first data-driven well model with the plurality of additional first data-driven well models to form a combined well model that models, for any individual hydrocarbon production well from the first plurality or any additional plurality of hydrocarbon production wells: hydrocarbon production well flow parameters, hydrocarbon production well well parameters, and / or an associated status of at least one control point in a flow path associated with the hydrocarbon production well.

4. A method according to claim 2 or 3, characterized in that at least some of the production wells within, or each of, the additional plurality of production wells are also in the first plurality of production wells.

5. A method according to claim 4, characterized in that all production wells within, or each, additional plurality of production wells are in the first plurality of production wells, and in that the first plurality of production wells additionally includes additional production wells.

6. Method, according to any one of claims 2 to 4, characterized in that at least some of the production wells within Petition 870260080658, dated 11 / 08 / 2026, page 15 / 30 5 / 19 of, or each of, additional production wells are not included in the first production well plurality.

7. Method, according to any one of claims 1 to 6, characterized in that it comprises: generating, using data-driven modeling, a second data-driven well model (hp) that models: flow parameters (9), and / or well parameters (9) and / or an associated status of at least one control point (9), for only one hydrocarbon production well, wherein the second data-driven model is parameterized by a set of second model parameters (5) that are representative of properties that are specific to the hydrocarbon production well to which it relates; and combining the second data-driven well model with the first data-driven well model (he) to form a combined model (he,p) that models: flow parameters, and / or well parameters, and / or an associated status of at least one control point, for only one hydrocarbon production well.

8. Method according to claim 7, characterized in that the well to which the second data-driven well model (h^) relates is comprised within the first plurality of hydrocarbon production wells and / or the, or each, additional plurality of hydrocarbon production wells.

9. Method, according to claim 7 or 8, characterized in that it comprises: generating, using data-driven modeling, a plurality of second data-driven well models, each second data-driven well model (hp) modeling: flow parameters, and / or parameters of Petition 870260080658, dated 11 / 08 / 2026, page.16 / 30 6 / 19 well, and / or an associated status of at least one control point, for a respective hydrocarbon production well, each second data-driven well model being parameterized by a set of second well parameters ( ) that are representative of properties that are specific to the hydrocarbon production well to which it relates; and combining each second data-driven well model with the first data-driven well model (he) to form combined models (he,p) that each models: flow parameters, and / or well parameters, and / or an associated status of at least one control point, for the respective hydrocarbon production well to which it relates.

10. A method according to any one of claims 1 to 9, characterized in that it comprises generating, using data-driven modeling, a well flow composition model that models, for any individual hydrocarbon production well of a second plurality of hydrocarbon production wells: the flow composition of the fluid produced from any one of a second plurality of hydrocarbon production wells, and the flow parameters, and / or well parameters, and / or an associated status of at least one control point, and / or time, wherein the flow composition model is parameterized by a first set of flow composition parameters that are representative of the flow composition common to all of the second plurality of hydrocarbon production wells;and combine the well flow composition model with the first data-driven well model (he) to form a combined model (he,p) that models: flow parameters, and / or well parameters, and / or an associated status of at least one control point, and / or time.

11. Method, according to any one of claims 1 to 10, Petition 870260080658, dated 11 / 08 / 2026, p.17 / 30 7 / 19 characterized in that it comprises: generating, using data-driven modeling, a well-specific flow composition model that models the flow composition of the fluid produced from only one hydrocarbon production well and: flow parameters, and / or well parameters, and / or an associated status of at least one control point, and / or time, wherein the well-specific flow composition model is parameterized by a second set of flow composition model parameters that are representative of the specific flow composition for the hydrocarbon production well to which it relates; and combining the well-specific flow composition model with the first data-driven well model (he) to form a combined model (he^) that models: flow parameters, and / or well parameters, and / or an associated status of at least one control point and / or time for only one hydrocarbon production well.

12. A method according to claim 11, characterized in that it comprises: generating, using data-driven modeling, a plurality of well-specific flow composition models, each well-specific flow composition model modeling the flow composition of the fluid produced from only one respective well and: flow parameters, and / or well parameters, and / or an associated status of at least one control point, and / or time, each well-specific model being parameterized by a second set of flow composition parameters that are representative of the flow composition that is specific to the single respective hydrocarbon production well to which the well-specific model relates; and Petition 870260080658, dated 11 / 08 / 2026, p.18 / 30 8 / 19 combine each well-specific flow composition model with the first data-driven well model (he) to form combined models (he,p) that each models: flow parameters, and / or well parameters, and / or an associated status of at least one control point and / or time, for each respective well.

13. A method according to any one of claims 1 to 12, characterized in that it comprises: generating, using data-driven modeling, a well prediction model, the well prediction model being capable of predicting for any individual production well among a third plurality of hydrocarbon production wells, a change in: a flow parameter, a well parameter, and / or a status of at least one control point, based on: a hypothetical change in the status of at least one control point, a hypothetical change in a well parameter, and / or a hypothetical change in a flow parameter, wherein the well prediction model is parameterized by a set of prediction parameters that are representative of properties that are common to the third plurality of hydrocarbon production wells;and combine the well prediction model with the first data-driven well model (he) to form a combined model {hQ^) that is capable of predicting for any individual production well within the third plurality of hydrocarbon production wells and the first plurality of hydrocarbon production wells: a flow parameter, and / or a well parameter, and / or the status of at least one control point, resulting from: a hypothetical change in the status of at least one control point, and / or the hypothetical change in a well parameter, and / or the hypothetical change in a flow parameter. Petition 870260080658, dated 11 / 08 / 2026, p. 19 / 30 9 / 19; 14. A method, according to any one of claims 1 to 13, characterized in that it comprises: generating, using data-driven modeling, a well-specific forecast model, the well-specific forecast model being capable of predicting, for only one hydrocarbon production well, a change in: a flow parameter, a well parameter, and / or the status of at least one control point based on: a hypothetical change in the status of at least one control point, and / or a hypothetical change in a well parameter, and / or a hypothetical change in a flow parameter, wherein the well-specific forecast model is parameterized by a set of well-specific forecast parameters that are representative of specific properties of that hydrocarbon production well;and combine the well-specific forecasting model with the first data-driven well model (he) to form combined models (Hθβ) that are each capable of predicting: a flow parameter, and / or a well parameter, and / or the status of at least one control point, resulting from: a hypothetical change in the status of at least one control point, and / or the hypothetical change in a well parameter, and / or the hypothetical change in a flow parameter for only one hydrocarbon production well.

15. Method, according to claim 14, characterized in that it comprises: generating, using data-driven modeling, a plurality of well-specific forecast models, each well-specific forecast model capable of predicting, for only one respective hydrocarbon production well, a change in: a flow parameter, a well parameter, and / or the status of at least one control point, based on: a hypothetical change in the status of at least one control point, and / or a Petition 870260080658, dated 11 / 08 / 2026, p.20 / 30 10 / 19 hypothetical change in a well parameter, and / or a hypothetical change in a flow parameter, wherein each well-specific forecast model is parameterized by a set of well-specific forecast parameters that are representative of properties that are specific to the hydrocarbon production well to which the well-specific forecast model relates; and combine each well-specific forecast model with the first data-driven (he) well model to form combined models (hgp) that are each capable of predicting: a flow parameter, a well parameter, and / or the status of at least one control point, resulting from: the hypothetical change in the status of at least one control point, and / or the hypothetical change in a well parameter, and / or the hypothetical change in a flow parameter for each respective hydrocarbon production well.

16. Method, according to any one of claims 1 to 15, characterized in that one or more of the model(s) (he, h^, he ^) form(s) part of a statistical approach such that: a flow parameter (9), a well parameter (9) and / or a status of at least one control point production (9) by the one or more model(s) is / are produced as a probability distribution with an associated degree of uncertainty.

17. Method, according to any one of claims 1 to 15, characterized in that at least one control point (9) comprises: a flow control valve; a pump; a compressor; a gas lift injector; expansion devices; a throttling control valve; gas lift valve configurations or rates in wells or riser pipelines; ESP (Electric Submersible Pump) configurations, effect, speed or pressure lift; downhole branch valve configurations, downhole inlet flow control valve configurations; or topside and subsea control configurations in one or more: separators, compressors, pumps, scrubbers, condensers / coolers, heaters, extractor columns, mixers, dividers, coolers; and wherein the flow parameters (9) include one or more of the following pressures;flow rate, a gas flow rate, an oil flow rate, a water flow rate, a liquid flow rate, a hydrocarbon flow rate, a nominal flow rate which is the sum of one or more of any of the above rates (by volume, mass or flow velocity); an oil fraction, a gas fraction, a carbon dioxide fraction, a multiphase fluid fraction, a hydrogen sulfide fraction, a multiphase fluid fraction, temperatures, a gas-to-liquid ratio, densities, viscosities, molar weights, pH, water cut (WC), productivity index (PI), Gas-Oil Ratio (GOR), BHP and wellhead pressures, rates after topside separation, separator pressure, other line pressures, flow velocities or sand production;and wherein the well parameters (9) include one or more of the following: depth, length, number and type of joints, inclination, cross-sectional area within / of a production well, wellbore, well branch, pipe, tubing or sections thereof; choke valve Cv-curve; choke valve discharge orifice cross-sectional area; heat transfer coefficient (U-value); friction coefficients; material types; insulation types; coating factors; and external temperature profiles. Petition 870260080658, dated 11 / 08 / 2026, page 22 / 30 12 / 19; 18. A method, according to any one of claims 1 to 15, characterized in that the first data-driven well model is a machine learning well model.

19. Method according to claim 18, characterized in that the set of first well model parameters comprises more than 1000 model parameters, each representative of a property common to all of the first plurality of hydrocarbon production wells.

20. A method, according to any one of claims 1 to 15, characterized in that the first data-driven model comprises a neural network.

21. A method according to any one of claims 1 to 15, characterized in that the hydrocarbon production wells of the first plurality of hydrocarbon production wells are not all connected to the same hydrocarbon reservoir.

22. A method, according to any one of claims 1 to 15, characterized in that it comprises: receiving, for each hydrocarbon production well of the first plurality of hydrocarbon production wells, respective data relating to flow parameters of the respective hydrocarbon production well and / or well parameters of the respective hydrocarbon production well and / or an associated status of at least one control point in a flow path associated with the respective hydrocarbon production well; and training each parameter of the first parameter set from data received from all of the first plurality of hydrocarbon production wells. Petition 870260080658, dated 11 / 08 / 2026, p. 23 / 30 13 / 19 23. A method, according to any one of claims 1 to 15, characterized in that the set of first well model parameters parameterizes the properties common to all of the first plurality of hydrocarbon production wells with a first precision, the method comprises the additional steps of: training the first (he), or combined (he,p), data-driven well model on data relating to flow parameters (9), well parameters (9) and / or an associated status of at least one control point (9), from at least two hydrocarbon production wells; obtaining an updated set of first well model parameters (θ) from the training of the first data-driven well model, wherein the updated set of first well model parameters parameterizes the properties common to all of the first plurality of hydrocarbon production wells with a second precision that is greater than the first precision;and update the first, or combined, data-driven well model based on the updated first well model parameter set, where the updated first data-driven well model allows for more accurate modeling of any of the plurality of hydrocarbon production wells.

24. Method according to claim 23, characterized in that the data-driven well model training step comprises a plurality of iterative training steps.

25. Method, according to claim 23 or 24, characterized in that it comprises introducing at least one additional well into the first plurality of hydrocarbon production wells; retraining the first model / combined data-driven well model on data relating to: Petition 870260080658, dated 11 / 08 / 2026, p.24 / 30 14 / 19 flow parameters (9), and / or well parameters (9), and / or an associated status (9) from at least one control point, from at least one additional well; obtain a first updated well model parameter set from retraining the first model / combined well model driven by data, wherein the first updated parameter set parameterizes the common properties of the first hydrocarbon production well plurality with greater accuracy than the second accuracy; and update the first model / combined well model driven by data based on the first updated well model parameter set.

26. Method, according to any one of claims 23 to 25, characterized in that it comprises obtaining additional data relating to: flow parameters (9), and / or well parameters (9), and / or an associated status of at least one control point (9) from at least one of the first plurality of hydrocarbon production wells; retraining the first data-driven well model (he) / combined model (he^) on the additional data; obtaining a set of first updated well model parameters from the retraining of the first data-driven well model / combined model, wherein the set of first updated well model parameters parameterizes the common properties of the first plurality of hydrocarbon production wells with an accuracy that is greater than the second accuracy;and update the first model / combined data-driven well model based on the updated first well model parameter set.

27. Method, according to any one of claims 23 to 26, characterized in that it comprises: before the aforementioned step of training the first data-driven well model, inserting one or the second data-driven well model (^5), or one or a plurality of second data-driven well models, into the first data-driven well model (he) or combined data-driven well model (Kθ,β), wherein the one or each second data-driven well model models: flow parameters, and / or well parameters, and / or an associated status of at least one control point, for only one hydrocarbon production well to which the second data-driven well model relates,and wherein the one or each second data-driven well model is parameterized by a set of second well model parameters (5) that are representative of properties that are specific to the production well to which the second data-driven well model relates, wherein the set of second well model parameters parameterizes the specific properties of the hydrocarbon production well with a third precision; during the said step of training the first data-driven well model, obtain an updated set of second well model parameters for the second data-driven well model(s) related to the hydrocarbon production well(s),wherein the updated second well model parameter set parameterizes the specific properties of the hydrocarbon production well(s) to which the data-driven second well model(s) relate(s) with an accuracy greater than the third accuracy; and update the data-driven second well model(s) based on the updated second well model parameter set.

28. Method for predicting a flow parameter, and / or well parameter, and / or the status of at least one control point, for at least one hydrocarbon production well, characterized in that it comprises: Petition 870260080658, dated 11 / 08 / 2026, page 26 / 30 16 / 19 modeling defined in any of claims 13 to 15; introducing a hypothetical change in the status of at least one control point, and / or a hypothetical change in a well parameter and / or a hypothetical change in a flow parameter, associated with at least one hydrocarbon production well, in the (respective) combined model and thus obtaining a predicted flow parameter, and / or well parameter, and / or status of at least one control point, for at least one hydrocarbon production well.

29. Method according to claim 28, characterized in that the predicted flow parameter, and / or well parameter and / or status of at least one control point is a well health indicator, a water cut (WC) of the produced hydrocarbon fluid, a gas-to-oil ratio (GOR) of the produced fluid, a liquid loading risk indicator, a total flow rate of produced fluid (by volume, mass or flow velocity), a gas flow rate, an oil flow rate, a water flow rate, a liquid flow rate, a hydrocarbon flow rate, a carbon dioxide fluid flow rate, a hydrogen sulfide fluid flow rate, a multiphase fluid flow rate, a slag severity, an oil fraction, a gas fraction, a water fraction, a carbon dioxide fraction, a multiphase fluid fraction, a hydrogen sulfide fraction, a gas-to-liquid ratio, density,viscosity, pH, productivity index (PI), BHP and wellhead pressures, post-topside separation rates, separator pressure, other line pressures, flow velocities or sand production.

30. Method for optimizing hydrocarbon production from at least one hydrocarbon production well, characterized in that it comprises: Petition 870260080658, dated 11 / 08 / 2026, p.27 / 30 17 / 19 predict a flow parameter, a well parameter and / or the status of at least one control point for at least one hydrocarbon production well defined in claim 28; repeat the prediction defined in claim 28 based on a different hypothetical change in the status of at least one control point, and / or a different hypothetical change in the well parameter and / or a different hypothetical change in the flow parameter; and determine an optimized status of at least one control point, and / or the flow parameter, and / or the well parameter and thus optimize hydrocarbon production; and wherein optionally the prediction defined in claim 28 is repeated a plurality of times based on a plurality of different hypothetical changes in the status of at least one control point, and / or different hypothetical changes in the flow parameter and / or different hypothetical changes in the well parameter.

31. Method of estimating a flow parameter, and / or a well parameter, and / or the status of at least one control point, for a hydrocarbon production well, the method characterized in that it comprises: modeling defined in any one of claims 1 to 15; and determining an estimated flow parameter, and / or well parameter, and / or status of at least one control point for at least one hydrocarbon production well by inserting into the first data-driven well model (he) or the (respective) combined well model (Hθ,β) a state of at least one hydrocarbon production well, the state comprising a flow parameter, and / or a well parameter, and / or an associated status of at least one control point of at least one hydrocarbon production well. Petition 870260080658, dated 11 / 08 / 2026, pp. 28 / 30 18 / 19 32. Method according to claim 31, characterized in that the estimated flow parameter, and / or estimated well parameter and / or status of at least one control point is a well health indicator, a water cut (WC) of the produced hydrocarbon fluid, a gas-to-oil ratio (GOR) of the produced fluid, a liquid loading risk indicator, a total flow rate of produced fluid (by volume, mass or flow velocity), a gas flow rate, an oil flow rate, a water flow rate, a liquid flow rate, a hydrocarbon flow rate, a carbon dioxide fluid flow rate, a hydrogen sulfide fluid flow rate, a multiphase fluid flow rate, a slag severity, an oil fraction, a gas fraction, a water fraction, a carbon dioxide fraction, a multiphase fluid fraction, a hydrogen sulfide fraction, a gas-to-liquid ratio, density,viscosity, pH, productivity index (PI), BHP and wellhead pressures, post-topside separation rates, separator pressure, other line pressures, flow velocities or sand production.

33. Computer system for modeling an individual hydrocarbon production well among a plurality of hydrocarbon production wells, characterized in that the computer system comprises an interface for receiving data relating to: flow parameters (9), and / or well parameters (9), and / or an associated status (9) from at least one control point, from each of a plurality of hydrocarbon production wells, and wherein the computer system comprises stored instructions which, when executed by the computer system, cause the computer system to perform the method defined in any one of claims 1 to 32. Petition 870260080658, dated 11 / 08 / 2026, pp. 29 / 30 19 / 19 34. Non-transient computer-readable storage medium, characterized in that it comprises instructions for execution in a computer system disposed to receive data relating to: flow parameters (9), and / or well parameters (9), and / or an associated status (9) from at least one control point, from each of a plurality of hydrocarbon production wells; wherein the instructions, when executed, will configure the computer system to perform a method defined in any one of claims 1 to 32. Petition 870260080658, dated 11 / 08 / 2026, p. 30 / 30

Images