Water injection process parameter and pipeline strength integrated design method and system based on limit service life constraint

Abstract

The application relates to the technical field of injection water system pipeline design, and is an injection process parameter and pipeline strength integrated design method and system based on limit service life constraints, which comprises the following steps: determining the type of an injection pipeline to be designed and corresponding basic data; determining a corrosion rate prediction value and a corresponding design conservative upper limit; setting a target limit service life, and constructing a process parameter feasible region under the life constraint; for a newly-built injection pipeline, strength checking and design thickness determination are carried out in the process parameter feasible region; and for an existing or in-service injection pipeline, an allowed process parameter range is obtained under the constraint of given materials and thickness. The application determines the corrosion rate prediction value and the corresponding design conservative upper limit, combines the target limit service life to construct the process parameter feasible region under the life constraint, makes the life target be implemented into an executable window, and avoids the problem of optimizing the process parameter only according to the production capacity or energy consumption and ignoring the pipeline strength problem possibly caused by corrosion.

Description

Technical Field

[0001] This invention relates to the field of water injection system pipeline design technology, and is an integrated design method and system for water injection process parameters and pipeline strength based on the constraint of ultimate service life. Background Technology

[0002] In oilfield water injection system pipelines, material selection and wall thickness are often based on experience or a simplified annual corrosion rate, using a fixed corrosion allowance C2 (e.g., 1-3 mm as required by GB 50391-2014 "Code for Design of Oilfield Water Injection Engineering"), followed by strength verification according to the code to determine the wall thickness. This approach is simple, but it presents three main problems:

[0003] (1) Process side T, P, v The optimization of parameters χ, which are related to water quality or chemical correction, is mostly focused on production capacity and energy consumption, and is difficult to link with intensity design.

[0004] (2) Corrosion rate C R While the service life τ was used to estimate C2, the ultimate service life (τ) was not included. max The function is transformed into a runnable window of (T,P,v,χ), but lacks boundary basis when running off-design point.

[0005] (3) Uncertainties such as medium fluctuations, flow pattern shear and inhibitor efficiency have not been quantified and propagated, and monitoring data have not been used to form a closed-loop redesign of the process and lifespan.

[0006] The aforementioned issues make it difficult to balance wall thickness, energy consumption, and risk under a given lifespan target. Furthermore, the empirical C2 wall thickness tends to be too thick in the low corrosion range and too thin in the high corrosion range, especially in scenarios with multiple water sources, complex station networks, and fluctuating loads.

[0007] Commonly used methods for optimizing pipeline material selection and wall thickness design include:

[0008] Existing patent document CN107886182B discloses an optimization design method and apparatus for an oilfield gathering and transportation system. The method includes: determining the initial location of a booster station based on the gathering and transportation radius, the number of oil wells under the station's jurisdiction, the station's processing scale, and the total length of pipelines connecting the oil wells under the station's jurisdiction; optimizing the initial location of the booster station based on the total length of pipelines connecting the oil wells under the station's jurisdiction and a weighting coefficient to obtain an optimized location; determining the optimal topology of the series pipeline network based on the optimized location of the booster station and the locations of each oil well to minimize the total length of pipelines connecting the oil wells under the station's jurisdiction; and determining the pipe diameter and wall thickness parameters based on the optimal topology. This method and apparatus provide an objective method and theoretical basis for the site layout of series pipeline network gathering and transportation systems, as well as for the optimization design and planning of pipeline connections and parameters. It avoids the shortcomings of previous subjective determinations based on personal experience and habits, and helps reduce the construction investment and operating costs of the gathering and transportation system. This method does not involve the water injection system temperature T, pressure P, flow velocity v, and water quality or reagent parameters χ, nor does it involve corrosion rate prediction and conservative upper bound conversion for design. Therefore, it cannot combine the feasible domain of process parameters under the constraint of the target ultimate service life to determine the integrated design of water injection process parameters and pipeline strength.

[0009] Existing patent document 2, publication number CN103470234B, discloses a method for optimizing process parameters of field injection for microbial enhanced oil recovery. The method includes the following steps: (1) screening of field injection process parameters; (2) determining the range of values ​​for field injection process parameters; and (3) optimizing field injection process parameters. This invention has the advantages of simple steps, strong logic, strong operability, and high reliability and accuracy of the optimization results. Therefore, it can be widely used in laboratory experiments and field tests of microbial enhanced oil recovery. This method does not involve the water injection system temperature T, pressure P, flow rate v, and water quality or reagent parameters χ, nor does it involve corrosion rate prediction and conservative upper bound conversion for design. Therefore, it cannot combine the feasible domain of process parameters under the constraint of the target limit service life to determine the integrated design of water injection process parameters and pipeline strength. Summary of the Invention

[0010] This invention provides an integrated design method and system for water injection process parameters and pipeline strength based on the constraint of ultimate service life, which overcomes the shortcomings of the prior art. It can effectively solve the problems of process parameters and strength design being separated and life indicators being unable to be converted into an operable process window in the existing integrated design method for water injection process parameters and pipeline strength.

[0011] One of the technical solutions of this invention is achieved through the following measures: an integrated design method for water injection process parameters and pipeline strength based on ultimate service life constraints, comprising:

[0012] Determine the type of water injection pipeline to be designed and obtain the corresponding basic data. The type of water injection pipeline to be designed includes newly built water injection pipelines and existing or in-service water injection pipelines. The basic data for newly built water injection pipelines includes the water injection system temperature. T ,pressure P Flow rate v and water quality or chemical parameters χ Field data or experimental data, basic data corresponding to existing or in-service water injection pipelines, including in-service scenario data and water injection system temperature. T ,pressure P Flow rate v and water quality or chemical parameters χ Field data or experimental data;

[0013] Input water injection system temperature T ,pressure P Flow rate v and water quality or chemical parameters χ Incorporate field or experimental data into the corrosion rate prediction model and determine its corresponding conservative upper bound for design. C R * ( T , P , v , χ The corrosion rate prediction model was constructed using deep learning.

[0014] Set target maximum service life τ max Combined with a conservative upper bound design C R * ( T , P , v , χ The feasible domain of process parameters under the constraint of construction lifetime Ω( T , P , v , χ The lifespan constraint is based on the target maximum service life. τ max And design with a conservative upper bound C R * ( T , P , v , χ )Sure;

[0015] For newly constructed water injection pipelines, within the feasible range of process parameters Ω ( T , P , v , χWithin the specified range, strength verification and design thickness determination are carried out to obtain the allowable range of process parameters for existing or in-service water injection pipelines under the constraints of given materials and thickness.

[0016] The following are further optimizations and / or improvements to the above-mentioned technical solution:

[0017] The process of constructing the above corrosion rate prediction model includes:

[0018] Obtain multiple sets of water injection system temperatures T ,pressure P Flow rate v and water quality or chemical parameters χ A standardized dataset is constructed using historical field data or historical experimental data, along with the corresponding actual values ​​of corrosion rates.

[0019] Determine the initial model and the corresponding physical prior constraints;

[0020] The standardized dataset is divided into K subsets, and the initial model is trained and validated using K-fold cross-validation with time-block or measurement point-layer methods to obtain a corrosion rate prediction model that meets the output conditions.

[0021] The above determination of the initial model and corresponding physical prior constraints includes:

[0022] Determining the initial model based on the generalized additive model:

[0023]

[0024] in, This is a predicted corrosion rate. X =( T , P , v , χ ) is a set of variables; χ These are water quality or reagent parameters; s i ( i ) is a unary term; s i,j ( i , j ) is a binary term;

[0025] The physical prior constraints to be applied are as follows:

[0026]

[0027] in, For coefficients; q inh Inhibitor dosage rate; The efficacy rate of the inhibitor.

[0028] The above can also include setting an extrapolation identification mechanism and a conservative backoff strategy to modify the corrosion rate prediction model and design a conservative upper bound CR*(T,P,v,χ), including:

[0029] An extrapolation identification mechanism is set up based on Mahalanobis distance, kernel density, or nearest neighbor density components. The training data is obtained from a standardized dataset to determine whether the training data is an extrapolation point.

[0030] If it is an extrapolation point, a conservative rollback strategy is executed, which includes increasing the level of conservatism or rolling back to a corrosion rate prediction model containing only monotonic / saturated priors, and applying a safety reduction factor to the candidate upper limit of key operating variables.

[0031] The above uses quantile fitting to construct a conservative upper bound for design. C R * ( T , P , v , χ ).

[0032] The above-mentioned target service life limit τ max Combined with a conservative upper bound design C R * ( T , P , v , χ The feasible domain of process parameters under construction lifetime constraints Ω( T , P , v , χ The lifespan constraint is based on the target maximum service life. τ max And design with a conservative upper bound C R * ( T , P , v , χ ) Determined, including:

[0033] Define the corrosion-side input, search space, and lifetime feasibility criteria, specifically:

[0034] Corrosion-side input: Designed with a conservative upper bound C R * ( T , P , v , χ ) as the corrosion-side input;

[0035] Search space: The domain of candidate process variables. D =[ T min , T max ]×[ P min , P max ]×[ v min , v max ]× χ ;

[0036] Lifespan and thickness-related parameters: Target limit service life set τ max Corrosion allowance C 2 is;

[0037] The lifespan feasibility criteria are as follows;

[0038] ;

[0039] Construct a sample set S in the search space and discretize it;

[0040] Based on the extrapolation identification mechanism and conservative backoff strategy, it is ensured that each sampling point in the sampling set S is a non-extrapolation point, and this is combined with the defined indicator function I( x ), keep all I( x The sampling points with )=1 form a feasible set Ω( x )={ x ∈S|I( x )=1}, where the indicator function I( x As shown below:

[0041] ;

[0042] For feasible set Ω ( x Connectivity extraction and boundary fitting are performed to obtain the feasible set Ω( x The volumetric domain description, the boundary functions / mesh of each connected domain, and the set of boundary curves for typical two-dimensional projections form the corresponding feasible domain Ω( for process parameters). T , P , v , χ );

[0043] Determine the confidence band and multi-level boundaries, and output the feasible domain description and publication format of process parameters.

[0044] The above applies to newly constructed water injection pipelines, within the feasible range of process parameters Ω ( T , P ,v , χ Strength verification and design thickness determination are carried out within the specified range, including:

[0045] Define the input and target, specifically:

[0046] Input: Includes candidate operating points x =( T , P , v , χ )∈Ω, material grade and allowable stress σ a ( T ), welding joint coefficient E w Design coefficient set Y Geometric parameters, set of additional quantities C ;

[0047] Objective: To achieve the desired process parameters within the feasible range Ω ( T , P , v , χ Complete the strength check and determine the design thickness within ) t sd ;

[0048] Determine the required wall thickness for each candidate operating point under internal pressure bearing strength. t s and add additional amount C Obtain the design thickness t sd Then, axial and combined stress checks are performed, and local components and discontinuities are reinforced, including the design thickness. t sd The calculation process is as follows:

[0049]

[0050] in, t s Calculate the thickness of the pipe; t sd Design the thickness of the pipe; P Design pressure for the pipeline; D The outer diameter of the pipe; σ a ( T ) for temperature T Yield strength below; E w For welding joint coefficient; Y Design coefficient; C For additional quantity set; C 1 is for manufacturing negative deviation; C2 represents corrosion allowance; E For coefficients;

[0051] Based on the linkage and screening strategy, within the feasible region Ω of the process parameters ( T , P , v , χ Select the recommended run setpoint and its corresponding design thickness. t sd ;

[0052] Through multi-level feasible regions (Ω) p ,Ω p+ Reliability and safety margin are coupled at the recommended operating setpoint;

[0053] Design thickness t sd Map to standard specifications and verify them.

[0054] The above-mentioned allowable range of process parameters for existing or in-service water injection pipelines, under given material and thickness constraints, includes:

[0055] Without changing the pipeline specifications, calculate the current strength limit and form hard constraints on the operation side;

[0056] Given the input data, calculate the allowed operating window in service, including:

[0057] Determine the input data and objective function, where the input data includes the feasible region Ω of the process parameters. T , P , v , χ The thinnest section of the pipe wall currently measured in the pipe box. t rem Target service life τ max Corrosion life criteria Intensity limit P ≤ P max The objective function is shown below:

[0058]

[0059] Determine the allowed running window in active service, including:

[0060] Fixed strength boundary: for each pipe section i Segmented pressure limit P max (i) As a hard constraint under pressure, it forms a strength domain;

[0061] Lifespan projection: at the upper limit of intensity P =P max (i) Or operating at normal pressure P * On the slice, the boundary query interface is used to obtain the temperature of the water injection system. T ,pressure P and water quality or chemical parameters χ The allowed domain is the lifetime domain;

[0062] Window composition: The intersection of the lifetime domain and the intensity domain is used to obtain the segmented allowable operating window. ;

[0063] Operationalization: Allow segmented running windows Convert to a station control limit table, which includes the upper temperature limit. T max (i) Flow rate limit v max (i) ( T , χ ( ) Upper limit of water quality or chemical parameters χ Pressure limit P max (i) .

[0064] The above also includes access to online / offline monitoring and periodic updates of measured corrosion rate values. C R Design using a conservative upper bound C R * ( T , P , v , χ ) and the feasible range of process parameters Ω ( T , P , v , χ Perform closed-loop consistency verification and recalculation during pipeline operation, including:

[0065] Monitoring data is obtained using online and offline monitoring channels, and operational datasets and in-service parameter tables are constructed.

[0066] Based on the runtime dataset and the in-service parameter table, the following steps are performed in sequence: domain identification, periodic consistency verification, and hierarchical alarms.

[0067] If at least one triggering condition is met, change management is triggered, and process parameter adjustments or redesign and pipe replacement are required. The triggering conditions include: Δlife < 0 for N consecutive evaluation cycles. U σ > Uσ crit Detected by ILI / UT for M consecutive cycles. t min meas Falling threshold; long-term water source drift; model / feasible region version mismatch;

[0068] in , τ max To achieve the target maximum service life, C R * To use a conservative upper bound in the design, To allow for corrosion, Measurement of lifespan exceeding limits;

[0069] in, , For equivalent stress, Y is the allowable stress of the material, and Y is the design factor. As a measure of strength compliance, t min meas This is the minimum remaining wall thickness.

[0070] The second technical solution of the present invention is achieved through the following measures: an integrated design system for water injection process parameters and pipeline strength based on ultimate service life constraints, comprising:

[0071] The data acquisition unit determines the type of water injection pipeline to be designed and acquires the corresponding basic data. The types of water injection pipelines to be designed include newly constructed water injection pipelines and existing or in-service water injection pipelines. The basic data for newly constructed water injection pipelines includes the water injection system temperature. T ,pressure P Flow rate v and water quality or chemical parameters χ Field data or experimental data, basic data corresponding to existing or in-service water injection pipelines, including in-service scenario data and water injection system temperature. T ,pressure P Flow rate v and water quality or chemical parameters χ Field data or experimental data;

[0072] Corrosion response and upper limit determination unit, input water injection system temperature T ,pressure P Flow rate v and water quality or chemical parameters χ The field data or experimental data are fed into the corrosion rate prediction model to obtain the predicted corrosion rate value, and the corresponding conservative upper bound for design is determined. C R * (T , P , v , χ The corrosion rate prediction model was constructed using deep learning.

[0073] Feasible domain generation unit, setting target maximum service life τ max Combined with a conservative upper bound design C R * ( T , P , v , χ The feasible domain of process parameters under construction lifetime constraints Ω( T , P , v , χ The lifespan constraint is based on the target maximum service life. τ max And design with a conservative upper bound C R * ( T , P , v , χ )Sure;

[0074] Strength and thickness design units, for newly built water injection pipelines, within the feasible range of process parameters Ω ( T , P , v , χ Strength verification and design thickness determination are carried out within the specified range.

[0075] The process parameter range determination unit determines the allowable range of process parameters for existing or in-service water injection pipelines, under given material and thickness constraints.

[0076] The following are further optimizations and / or improvements to the above-mentioned technical solution:

[0077] The above also includes a closed-loop verification unit, which connects to online / offline monitoring and periodically updates the measured corrosion rate values. C R Design using a conservative upper bound C R * ( T , P , v , χ ) and the feasible range of process parameters Ω ( T , P , v , χPerform closed-loop consistency verification and recalculation during pipeline operation, including:

[0078] Monitoring data is obtained using online and offline monitoring channels, and operational datasets and in-service parameter tables are constructed.

[0079] Based on the runtime dataset and the in-service parameter table, the following steps are performed in sequence: domain identification, periodic consistency verification, and hierarchical alarms.

[0080] If at least one triggering condition is met, change management is triggered, and process parameter adjustments or redesign and pipe replacement are required. The triggering conditions include: lifespan out-of-bounds measurement Δlife < 0 for N consecutive evaluation cycles. U σ > U σ crit Detected by ILI / UT for M consecutive cycles. t min meas Falling threshold; long-term water source drift; model / feasible region version mismatch;

[0081] in , τ max To achieve the target maximum service life, C R * To use a conservative upper bound in the design, To allow for corrosion, Measurement of lifespan exceeding limits;

[0082] in, , For equivalent stress, Y is the allowable stress of the material, and Y is the design factor. As a measure of strength compliance, t min meas This is the minimum remaining wall thickness.

[0083] This invention introduces deep learning to construct a corrosion rate prediction model to determine the predicted corrosion rate value and to determine its corresponding conservative upper bound for design. C R * ( T , P , v , χ ), and then combined with the target's maximum service life. τ maxBy constructing a feasible region for process parameters under life constraints, the life target can be defined as a clear and actionable window, avoiding the problem of optimizing process parameters solely based on capacity or energy consumption while neglecting pipeline strength issues that may be caused by corrosion. Furthermore, strength verification and design thickness determination are completed within the feasible region of process parameters, reducing the impact of corrosion margins. C 2. Excessive thickness due to oversized components also avoids underestimation of lifespan under highly corrosive conditions; for existing or in-service water injection pipelines, under the constraints of given materials and thickness, the allowable range of process parameters can be accurately obtained, thereby improving the availability and life-cycle consistency of the water injection system. Attached Figure Description

[0084] Appendix Figure 1 A schematic diagram of the integrated design method for water injection process parameters and pipeline strength provided in an embodiment of the present invention.

[0085] Appendix Figure 2 This is a schematic diagram of the corrosion rate prediction model construction method provided in an embodiment of the present invention.

[0086] Appendix Figure 3 This is a schematic diagram of the process parameter feasible domain construction method provided in the embodiments of the present invention.

[0087] Appendix Figure 4 This is a schematic diagram of the strength verification and design thickness determination method provided in an embodiment of the present invention.

[0088] Appendix Figure 5 This is a schematic diagram of an integrated design system for water injection process parameters and pipeline strength, provided as an embodiment of the present invention.

[0089] Appendix Figure 6 This is a schematic diagram of another integrated design system for water injection process parameters and pipeline strength provided in an embodiment of the present invention. Detailed Implementation

[0090] The present invention is not limited to the following embodiments, and the specific implementation can be determined according to the technical solution of the present invention and the actual situation.

[0091] Those skilled in the art will understand that, unless specifically stated otherwise, in the embodiments of the present invention, a "module" or "unit" refers to a computer program or part of a computer program that has a predetermined function and works together with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0092] In addition, in the embodiments of the present invention, "multiple" refers to two or more, and "first" and "second" are used to distinguish descriptions and should not be construed as implying relative importance.

[0093] Based on this, the technical solution of the present invention will be described and explained below with reference to several examples.

[0094] Example 1: As shown in the attached document Figure 1 As shown in the figure, this invention discloses an integrated design method for water injection process parameters and pipeline strength based on ultimate service life constraints, including:

[0095] Step S110: Determine the type of water injection pipeline to be designed and obtain the corresponding basic data. The type of water injection pipeline to be designed includes newly built water injection pipelines and existing or in-service water injection pipelines. The basic data corresponding to newly built water injection pipelines includes the water injection system temperature. T ,pressure P Flow rate v and water quality or chemical parameters χ Field data or experimental data, basic data corresponding to existing or in-service water injection pipelines, including in-service scenario data and water injection system temperature. T ,pressure P Flow rate v and water quality or chemical parameters χ Field data or experimental data;

[0096] In this embodiment, the temperature of the water injection system is obtained. T ,pressure P Flow rate v and water quality or chemical parameters χ Field data or experimental data, including:

[0097] Field data: Real-time acquisition of field data from key sections of the water injection station and pipeline, including water injection system temperature. T ,pressure P Flow rate v and water quality or chemical parameters χ (Including water quality or chemical parameters) χ Including dissolved oxygen (DO) and chloride ions (Cl). - (Including alkalinity / CO2, H2S, bacterial indicators, inhibitor dosage / effectiveness, etc.), and further, material, pipe diameter / wall thickness, welding coefficient, manufacturing deviation and operating condition records can be collected simultaneously.

[0098] Experimental data: Corrosion-related experimental results and corresponding water injection system temperatures obtained in a device simulating water injection media and operating conditions (performing experiments such as those involving padding, electrochemistry, and flow loops). T ,pressure P Flow rate v and water quality or chemical parametersχ .

[0099] Among them, the coupon experiment provides the average corrosion rate and local corrosion morphology after long-term exposure in the simulated medium; electrochemical tests such as polarization curves and electrochemical impedance spectroscopy (EIS) can quickly assess the corrosion tendency, rate and action mechanism of the material and the corrosion inhibitor; the flow loop experiment can simulate the flow state in the pipeline and study the influence of flow velocity and flow state (laminar / turbulent) on corrosion (especially scouring corrosion) and the mass transfer process of the corrosion inhibitor.

[0100] Data on the operational scenarios of existing or in-service water injection pipelines, including:

[0101] (1) Thickness / geometric measurement: online inspection (ultrasonic inspection UT / radiographic inspection RT), excavation, minimum remaining wall thickness at sampling points t min meas Calculate the dimensions of the corrosion defect (depth / length / circumferential angle) and the equivalent remaining wall thickness at the defect. t rem .

[0102] (2) Operational events: pressure rise / start-stop records, pipeline cleaning schedule, historical leakage / repair and reinforcement information.

[0103] (3) Positioning and segmentation: The pipe segment is divided into calculation units (station outlet - valve chamber - junction point), and each unit records independent data. t min meas , E w , σ a ( T ).

[0104] Step S120: Input the water injection system temperature T ,pressure P Flow rate v and water quality or chemical parameters χ The field data or experimental data are fed into the corrosion rate prediction model to obtain the predicted corrosion rate value, and the corresponding conservative upper bound for design is determined. C R * ( T , P , v , χ The corrosion rate prediction model was constructed using deep learning.

[0105] Step S130: Set the target maximum service life. τ max Combined with a conservative upper bound design C R* ( T , P , v , χ The feasible domain of process parameters under construction lifetime constraints Ω( T , P , v , χ The lifespan constraint is based on the target maximum service life. τ max And design with a conservative upper bound C R * ( T , P , v , χ )Sure.

[0106] Step S140, for the newly built water injection pipeline, within the feasible range of process parameters Ω ( T , P , v , χ Within the specified range, strength verification and design thickness determination are carried out to obtain the allowable range of process parameters for existing or in-service water injection pipelines under the constraints of given materials and thickness.

[0107] This invention discloses an integrated design method for water injection process parameters and pipeline strength based on the constraint of ultimate service life. It introduces deep learning to construct a corrosion rate prediction model to determine the predicted corrosion rate value and its corresponding conservative upper bound for design. C R * ( T , P , v , χ ), and then combined with the target's maximum service life. τ max By constructing a feasible region for process parameters under lifespan constraints, the lifespan target can be defined as a clear and actionable window, avoiding the optimization of process parameters solely based on capacity or energy consumption while neglecting pipeline strength issues that may be caused by corrosion. Furthermore, for newly constructed water injection pipelines, strength verification and design thickness determination can be completed within the feasible region of process parameters, reducing excessively thick configurations and avoiding underestimation of lifespan under high corrosion conditions. For existing or in-service water injection pipelines, the allowable range of process parameters can be accurately obtained under given material and thickness constraints.

[0108] Example 2: If the type of water injection pipeline to be designed is a newly built water injection pipeline, the corresponding integrated design method of water injection process parameters and pipeline strength based on the constraint of ultimate service life includes:

[0109] (a) Obtaining the temperature of the water injection system T,pressure P Flow rate v and water quality or chemical parameters χ Field data or experimental data, including:

[0110] Field data: Real-time acquisition of field data from key sections of the water injection station and pipeline, including water injection system temperature. T ,pressure P Flow rate v and water quality or chemical parameters χ (Including water quality or chemical parameters) χ Including dissolved oxygen (DO) and chloride ions (Cl). - (Including alkalinity / CO2, H2S, bacterial indicators, inhibitor dosage / effectiveness, etc.), and further, material, pipe diameter / wall thickness, welding coefficient, manufacturing deviation and operating condition records can be collected simultaneously.

[0111] Experimental data: Corrosion-related experimental results and corresponding water injection system temperatures obtained in a device simulating water injection media and operating conditions (performing experiments such as those involving padding, electrochemistry, and flow loops). T ,pressure P Flow rate v and water quality or chemical parameters χ .

[0112] Among them, the coupon experiment provides the average corrosion rate and local corrosion morphology after long-term exposure in the simulated medium; electrochemical tests such as polarization curves and electrochemical impedance spectroscopy (EIS) can quickly assess the corrosion tendency, rate and action mechanism of the material and the corrosion inhibitor; the flow loop experiment can simulate the flow state in the pipeline and study the influence of flow velocity and flow state (laminar / turbulent) on corrosion (especially scouring corrosion) and the mass transfer process of the corrosion inhibitor.

[0113] (ii) Input water injection system temperature T ,pressure P Flow rate v and water quality or chemical parameters χ The field data or experimental data are fed into the corrosion rate prediction model to obtain the predicted corrosion rate value, and the corresponding conservative upper bound for design is determined. C R * ( T , P , v , χ ).

[0114] In step (II), the process of constructing the corrosion rate prediction model is as follows: Figure 2 As shown, it includes:

[0115] Step S211: Obtain multiple sets of water injection system temperatures. T ,pressureP Flow rate v and water quality or chemical parameters χ Historical field data or historical experimental data, along with the corresponding identifiers of actual corrosion rate values, are used to construct a standardized dataset, including:

[0116] (1) Obtain the temperature of the water injection system T ,pressure P Flow rate v and water quality or chemical parameters χ The historical field data or historical experimental data, along with the corresponding actual values ​​of corrosion rates, are as follows:

[0117] Historical field data: Historical field data collected in real time from key sections of water injection stations and pipelines over a historical period, including water injection system temperature. T ,pressure P Flow rate v and water quality or chemical parameters χ (Including water quality or chemical parameters) χ Including dissolved oxygen (DO) and chloride ions (Cl). - (Including alkalinity / CO2, H2S, bacterial indicators, inhibitor dosage / effectiveness, etc.), and further, material, pipe diameter / wall thickness, welding coefficient, manufacturing deviation and operating condition records can be collected simultaneously.

[0118] Historical experimental data: Corrosion-related experimental results and corresponding water injection system temperatures obtained in devices simulating water injection media and operating conditions (couplings, electrochemical, flow loop). T ,pressure P Flow rate v and water quality or chemical parameters χ .

[0119] Among them, the coupon experiment provides the average corrosion rate and local corrosion morphology after long-term exposure in the simulated medium; electrochemical tests such as polarization curves and electrochemical impedance spectroscopy (EIS) can quickly assess the corrosion tendency, rate and action mechanism of the material and the corrosion inhibitor; the flow loop experiment can simulate the flow state in the pipeline and study the influence of flow velocity and flow state (laminar / turbulent) on corrosion (especially scouring corrosion) and the mass transfer process of the corrosion inhibitor.

[0120] (2) Temperature of the water injection system T ,pressure P Flow rate v and water quality or chemical parameters χ Historical field data or historical experimental data are preprocessed, including anomaly handling, time alignment, and unit unification. Specifically:

[0121] Obvious outliers were removed using 3σ or IQR rules, and a log of the removal was retained. Next, multi-source data were aligned by measurement point and timestamp, unified to the same sampling interval, and time / spatial points were aligned. Measurement units and benchmarks (e.g., flow rate) were standardized. v In m / s, temperature T (Indicated by ℃, ion concentration by mg / L); short-term missing data are interpolated or deleted within a limited window, while long-term missing data retains the missing marker without inference; each data point is accompanied by the source type (field / experiment), measurement point number, instrument calibration date, and sampling frequency.

[0122] Furthermore, if both historical field data and historical experimental data exist, data fusion is performed. Specifically, the data is merged and compiled according to the same variable naming convention, the same time benchmark, and the same unit system. No predictions or corrections are made to the data; only source labeling and traceability are completed to provide clear input boundaries for subsequent steps.

[0123] (3) Put the preprocessed historical field data or historical experimental data into the same set to form a standardized dataset.

[0124] Step S212: Determine the initial model and the corresponding physical prior constraints;

[0125] In this embodiment, the goal of determining the initial model is to establish the corrosion rate and process parameters (water injection system temperature) in the water injection system. T ,pressure P Flow rate v and water quality or chemical parameters χ Functional relationship between ) C R = f ( T , P , v , χ This allows for the acquisition of predicted corrosion rates and the determination of their conservative upper bound for design. C R * .

[0126] When constructing new pipelines with sample sizes typically ranging from hundreds to thousands, a robust and interpretable model with a small sample size can be prioritized as the initial model. This embodiment uses the Generalized Additive Model (GAM) as an example to illustrate the initial model and corresponding physical prior constraints. Specifically:

[0127] The initial model is determined based on the generalized additive model (GAM):

[0128]

[0129] in, This is a predicted corrosion rate. X=( T , P , v , χ ) is a set of variables; χ For water quality or chemical parameters (dissolved oxygen DO, chloride ions Cl) - Alkalinity / CO2, H2S, bacterial indicators, inhibitor dosage rate q inh / efficiency η inh wait); s i ( i ) is a univariate term, that is, the smoothing effect of each independent variable; s i,j ( i , j ) is a binary term used to retain only the interactions necessary for the project.

[0130] The physical prior constraints to be applied are as follows:

[0131] ,

[0132] The aforementioned physical prior constraints correspond to "monotonic splines and bounded splines," meaning that the flow velocity must be considered. v ,temperature T Applying a "non-reduction" constraint to the inhibitor dosage rate q inh Inhibitor efficacy η inh By applying "non-incremental" constraints or saturation constraints, a two-dimensional smooth can be constructed for the inhibitor term to express the coupling of "dosage rate × effectiveness".

[0133] It should also be noted that the initial model in this embodiment can also be constructed using other models, such as Gaussian process regression (GPR), with kernel functions such as ARD-RBF or Matern-ν (ν=3 / 2 or 5 / 2), and independent length scales set for each variable to reflect sensitivity; or gradient boosting tree (GBDT) or neural networks with uncertainty estimation (such as deep ensemble / MC Dropout), for comparison or when the sample size increases significantly.

[0134] Furthermore, it should be noted that for existing or in-service water injection pipelines, an operational time window layer can be introduced. This involves using the most recent M months of operational water quality data as the main window, with earlier data used for prior analysis / smoothing, and outputting quantifiable data. C R * ( T , P , v , χ ).

[0135] Step S213: Divide the standardized dataset into K subsets, and use the K-fold cross-validation method with time-block or measurement point-layer to train and validate the initial model, so as to obtain a corrosion rate prediction model that meets the output conditions.

[0136] In this embodiment, K in the K-fold cross-validation method can be selected as needed; in this embodiment, K can be 5 or 10. The implementation process includes:

[0137] The standardized dataset is divided into K mutually exclusive subsets. Specifically, the time-block K-fold cross-validation method divides the standardized dataset into K subsets in chronological order, while the measurement-point stratified K-fold cross-validation method stratifies the standardized dataset by measurement points to ensure that each subset contains data from different measurement points.

[0138] In each round of training, one subset is selected as the validation set to evaluate the model performance. The remaining K-1 subsets are merged as the training set to train the initial model. This process is repeated K times to ensure that each subset is used as the validation set exactly once.

[0139] If the initial model is determined based on the generalized additive model (GAM), then during training, monotonic / saturated priors can be achieved by selecting and shape-constraining splines using AIC / GCV.

[0140] The verification requirements mentioned above can be set as needed. In this embodiment, they may include, but are not limited to:

[0141] Indicator verification, determining R 2 Indicators such as RMSE and MAE are used, and residual independence and heteroscedasticity are checked.

[0142] Residual verification, using partial dependency curves to verify the rationality of the project (e.g.) v Increase C R Enlarge q inh Increase C R (The price decreases and then tends to plateau).

[0143] Sensitivity perturbation verification involves applying a sensitivity perturbation of ±(10~20)% to the input to verify the physical consistency between the output directionality and amplitude.

[0144] In step (ii), by quantifying the uncertainty, a conservative upper bound for design is constructed based on the predicted corrosion rate. C R * ( T , P , v ,χ Taking the generalized additive model (GAM) as an example, when introducing the GAM to establish the initial model, a conservative upper bound for design is constructed based on quantile fitting. C R * ( T , P , v , χ ),include:

[0145] Pinball loss fitting conditional quantiles Q p ( C R | X )(like p =95 or p + The conditional quantile of (=99) is used as a conservative upper bound for design. C R * ;

[0146]

[0147] in, Q p ( C R | X ( ) represents the measured corrosion rate. C R The upper quantile function; L pin ( p ) represents pinball loss; p The value is the quantile level, ranging from [0.90, 0.99]. i Number the sample; C R This represents the measured corrosion rate (target value). This is a predicted corrosion rate. For indicator functions, when If the value is 1, then the value is 0; otherwise, the value is 0.

[0148] Under pinball loss, the model minimizes L pin (p) Directly approximates the upper quantile of the corrosion rate at a specified confidence level. When p =0.95 corresponds to a conservative upper bound at the design level (i.e., ensuring that the corrosion rate of approximately 95% of samples is below this value); when p + When the value is 0.99, it corresponds to a stricter warning boundary, which is used for early warning or safety verification during operation.

[0149] Therefore, the final measured corrosion rate value is... C R The upper quantile function can be written as the variance predicted by the composite model of cross-validation residuals and measurement errors, forming a conservative upper bound for design. C R * ( T , P , v , χ ), as shown below:

[0150]

[0151] in, For standard normal quantiles (e.g.) p =0.95 =1.645); This refers to the model prediction error; The standard deviation of the cross-validation residuals; This is for measurement error.

[0152] In the case of dual-source data, including both historical field data and historical experimental data, to reflect the difference in uncertainty between the two, confidence-weighted synthesis of the variances of the two parts can be performed. Furthermore, a density penalty coefficient can be added to sparse or extrapolated regions to make the boundary more conservative, avoiding optimistic estimates caused by local overfitting, and forming a conservative upper bound for design. C R * ( T , P , v , χ ), as shown below:

[0153]

[0154] in, Weights for the experimental data; The weights assigned to on-site data reflect the reliability of the data source; ρ(x) As an indicator of local sample density; This is the penalty coefficient for the sparse region; The standard deviation of the experimental data; This represents the standard deviation of the field data.

[0155] Furthermore, an extrapolation identification mechanism and a conservative backoff strategy can be set to modify the corrosion rate prediction model and design a conservative upper bound. C R * ( T , P , v , χSpecifically:

[0156] To prevent the corrosion rate prediction model from giving unreliable predictions due to sparse samples or outside the training domain, this embodiment establishes an extrapolation identification mechanism. This involves determining the Mahalanobis distance, kernel density, or nearest neighbor density of a given training data point in the standardized dataset. Points exceeding a threshold are recorded as extrapolated points; otherwise, they are considered in-domain points. Taking Mahalanobis distance as an example, the specific details are as follows:

[0157] Calculate the Mahalanobis distance of a given data point in the training set;

[0158]

[0159] like This indicates that the data is located in the "high-density region" of the training set, belonging to the domain point, and the corrosion rate prediction model is reliable.

[0160] like This indicates that the data is too far from the center, outside the training domain, and belongs to the extrapolation point, requiring increased conservatism.

[0161] in, d M The Mahalanobis distance is used to determine whether an input point is within the training domain. The mean vector of all training data; S To train the set covariance; These are the quantiles of the chi-square distribution, obtained from the statistical table; d The dimension of the feature space (e.g., the number of variables such as temperature, pressure, flow rate, dissolved oxygen, chloride ions, etc.); α The significance level or the strictness of the extrapolation point determination is typically set to 0.99 or 0.95, and can be set to... Based on this, the sparse region penalty coefficient is set to... .

[0162] For the extrapolation point, a conservative rollback strategy is implemented, which includes increasing the level of conservatism (e.g., ...). p =0.95→ p + =0.99), or revert to a corrosion rate prediction model containing only monotonic / saturated priors (i.e., based on a generalized additive model (GAM)), and adjust key operating variables (such as water injection system temperature). T ,pressure P Flow rate v and water quality or chemical parameters χ Apply a safety reduction factor to the candidate upper bound to ensure that the boundary of the subsequent feasible region is not eroded by extrapolation, such as:

[0163]

[0164] in, The upper limit of safe flow rate; This is the upper limit of the flow rate; For safety reduction factor, ;

[0165] Or use methods such as sparsity penalty to... C R * To be further magnified.

[0166] It should also be noted that the corrosion rate prediction model and the conservative upper bound used in this embodiment are... C R * ( T , P , v , χ After construction is complete, three types of outputs will be provided, including:

[0167] (1) Callable prediction functions or lookup table interfaces, respectively, return the predicted corrosion rate values. With conservative upper limit of design C R * (T,P,v,χ);

[0168] (2) Model diagnosis and confidence parameters (R², RMSE, quantile p or variance term used, extrapolation threshold) are used for design review and verification;

[0169] (3) Complete version and audit information (training time window, variable list, model type and key hyperparameters, shape constraint configuration) are archived in the form of “model version vX.Y” to ensure the traceability of subsequent recalculation and rolling updates during the runtime.

[0170] (III) Setting the target maximum service life τ max Combined with a conservative upper bound design C R * ( T , P , v , χ The feasible domain of process parameters under construction lifetime constraints Ω( T , P , v , [[ID=,107]]χ The lifespan constraint is based on the target maximum service life. τ max And design with a conservative upper bound C R * ( T , P, v , χ Confirmed, as attached Figure 3 As shown, it includes:

[0171] Step S221 defines the corrosion-side input, search space, and lifetime feasibility criteria, specifically:

[0172] Corrosion-side input: Designed with a conservative upper bound C R * ( T , P , v , χ ) as the corrosion-side input;

[0173] Search space: The domain of candidate process variables. D =[ T min , T max ]×[ P min , P max ]×[ v min , v max ]× χ ;

[0174] Lifespan and thickness-related parameters: Target limit service life set τ max Corrosion allowance C 2 is;

[0175] The lifespan feasibility criteria are as follows;

[0176]

[0177] The data points that meet the lifespan feasibility criterion are considered to be lifespan feasible.

[0178] Step S222: Construct a sampling set S in the search space and discretize it, including:

[0179] (1) Construct a sampling set S within the search space;

[0180] In this step, Latin hypercube sampling (LHS) or hypergrid sampling can be preferably used to sample within the search space to obtain a sample set S. It should be noted that the sample set S must satisfy the characteristics of sufficient coverage, uniformity and recalculation.

[0181] Taking Latin hypercube sampling (LHS) as an example, a sampling set S is constructed within the search space, including:

[0182] When the variable dimension is high or the χ² dimension is large, Latin hypercube sampling (LHS) is used (sample size recommended 10). 4 Starting from the order of magnitude, and allocating higher resolution to each dimension according to engineering sensitivity (e.g., for...). v , T (Encryption); Here, the number of χ dimensions is relatively large, which can be determined by setting a threshold. That is, if it is greater than the set threshold, it indicates that the number of χ dimensions is relatively large.

[0183] When the variable dimension is low or the boundary accuracy requirement is high, a hierarchical regular grid is adopted and adaptively refined in the boundary neighborhood; the requirement for high boundary accuracy can be determined by setting a threshold, that is, when it is greater than the set threshold, it indicates that the boundary accuracy requirement is high.

[0184] For discrete / enumerated components (such as different drug formulations), expand each term on χ to form a Cartesian product subdomain for block solution; all sampling points must retain source, random seed and grid step size information for recalculation.

[0185] Step S223, based on the extrapolation identification mechanism and conservative backoff strategy, ensures that each sampling point in the sampling set S is a non-extrapolation point, and combines this with the defined indicator function I( x ), keep all I( x The sampling points with )=1 form a feasible set Ω( x )={ x ∈S|I( x )=1}, where the indicator function I( x As shown below:

[0186]

[0187] It should be noted that, based on the extrapolation identification mechanism and conservative backoff strategy, it is ensured that each sampling point in the sampling set S is a non-extrapolation point, that is, for each sampling point... x =( T , P , v , χ )∈S, based on the extrapolation identification mechanism, it is determined whether it is an extrapolation point. Then, a conservative backoff strategy is used to improve the level of conservatism, and then based on the defined indicator function I( x The extrapolation identification mechanism and conservative backoff strategy are the same as those in step (II), and will not be described again.

[0188] Step S224, for the feasible set Ω ( x Connectivity extraction and boundary fitting are performed to obtain the feasible set Ω( x The volumetric domain description, the boundary functions / mesh of each connected domain, and the set of boundary curves for typical two-dimensional projections form the corresponding feasible domain Ω( for process parameters). T, P , v , χ ).

[0189] Specifically, this includes: marking connected components on the sampling grid according to adjacency relationships (such as K-nearest neighbors or regular grid 6 / 18 / 26-adjacency); eliminating isolated fragmented domains that are too small and unoperable; and using isosurfaces for each connected component. The interpolation / fitting results are used as boundaries (e.g., RBF interpolation, Marching Cubes / feasible isosurface extraction); for commonly used two-dimensional subspaces (e.g. v – T , v – q inh , T – P This generates a cross-sectional curve (the intersection line of the boundary on the cross-section), which facilitates operation settings and illustration.

[0190] Step S225: Determine the confidence band and multi-level boundaries. In this embodiment, multi-level feasible regions can be constructed according to different confidence levels during model construction. Specifically:

[0191] (1) Design boundary Ω p :use p =0.95 (or enterprise standard) C R * ;

[0192] (2) Warning boundary Ω p+ Use higher quantiles, such as p + =0.99 obtained C R * ;

[0193] (3) Operational recommendations: The strip between the boundaries serves as a warning zone, used for subsequent boundary crossing warnings and soft constraint control.

[0194] Step S226: Output the feasible domain description and release format of the process parameters;

[0195] For use in subsequent steps and project delivery, this embodiment generates one or more of the following equivalent descriptions in parallel:

[0196] (1) Lookup table / grid: Publish Boolean mask and boundary grid of Ω at a fixed step size for easy and quick identification;

[0197] (2) Implicit function: The implicit function approximation of the boundary is g(x)≈0, which can be used for numerical optimization and constraint discrimination;

[0198] (3) Inequality envelope: Fit upper and lower envelope curves / surfaces on commonly used two-dimensional / three-dimensional projections (e.g. v ∈[ v min ( T , χ ), v max ( T , χ This facilitates operation and configuration.

[0199] (4) Connected component metadata: For each connected component, provide the volume, principal scale (maximum allowable flow velocity) v max Maximum temperature T max (etc.) and representative center points (which can be used as initial operation setting points).

[0200] In this embodiment, it further includes forming boundary crossing criteria and interfaces, including

[0201] Define lifetime out-of-bounds measurement :

[0202]

[0203] when A value less than 0 indicates a feasible lifetime; for multi-level boundaries, warning thresholds can be further defined. For (by) p + The corresponding upper bound is generated.

[0204] Output the following interface:

[0205] (1) Discriminant function: Input ( T , P , v , χ Returns the feasible / infeasible result and Δ. life ;

[0206] (2) Boundary query: Given any two (or three) variables, return the allowed intervals of the remaining variables (e.g., given any two (or three) variables). T , χ return v max ( T , χ ));

[0207] In this embodiment, further measures include shrinkage and conservatism (operational safety margin).

[0208] To avoid boundary erosion caused by model / monitoring lag, the feasible region Ω of the process parameters can be defined. T ,P , v , χ Perform a shrinking operation, that is, adjust the criteria as follows:

[0209]

[0210] in, δ >0 represents the lifetime safety margin (correlated with the monitoring refresh cycle and the variance level of S2); the contracted domain (the feasible domain of process parameters after applying a safety contraction margin). The operational suggestion window serves as the operational suggestion window, while the feasible domain Ω for process parameters remains the design verification domain; this parameter is determined by enterprise standards or project agreements.

[0211] In this embodiment, further quality control and recalculation requirements are included, namely, recording all sampling, discrimination, fitting, and projection steps: sampling method and random seed, sample size, step size / resolution, extrapolation point ratio, fitting error, boundary smoothing parameters, and their correspondence with the model version. Any input ( τ max , C 2. When the model version or data time window changes, a full-link recalculation or incremental update needs to be triggered, and a new version of the Ω and boundary file needs to be generated for subsequent strength verification.

[0212] (iv) Within the feasible range of process parameters Ω ( T , P , v , χ Strength verification and design thickness determination are carried out within the specified range, as shown in the attached document. Figure 4 As shown, it includes:

[0213] Step S231, determine the input and target, specifically:

[0214] Input: Includes candidate operating points x =( T , P , v , χ )∈Ω, material grade and allowable stress σ a ( T (Consider temperature derating if necessary), weld joint coefficient E w Design coefficient set Y (Safety factor), geometric parameters (outer diameter) D Additional quantity set C (Manufacturing negative deviation) C 1. Corrosion allowance C 2nd grade).

[0215] Objective: To achieve the feasible process parameters within the Ω (T , P , v , χ Complete the strength check and determine the design thickness within ) t sd When necessary, multiple objectives should be weighed to balance thickness with energy consumption / pressure drop risks.

[0216] Step S232: Determine the required wall thickness for each candidate working point under internal pressure bearing strength. t s and add additional amount C Obtain the design thickness t sd Then, axial and combined stress checks are performed, and local components and discontinuities are reinforced, including the design thickness. t sd The calculation process is as follows:

[0217]

[0218] in, t s Calculate the pipe thickness, in mm; t sd Pipe design thickness, mm; P The design pressure for the pipeline is MPa. D The outer diameter of the pipe is in mm. For temperature T Yield strength at the specified value, MPa; E w The coefficient for welded joints is 1 for seamless steel pipes; Y The design factor (safety factor) is set to 0.4. C For the set of additional quantities, mm; C 1 represents a negative manufacturing tolerance, measured in mm. C 2 represents the corrosion allowance, in mm; E For coefficients, when t s < D At 6 o'clock, E The values ​​should be taken from Table 1.

[0219] Table 1 Rules for determining the value of coefficient E

[0220]

[0221] In this step, the axial and combined stress checks are specifically as follows:

[0222] In determining t sdSimultaneously, axial stress is checked (head force generated by internal pressure, thermal stress caused by temperature difference, secondary stress due to pump start-up / stop / terrain constraints, etc.); for components such as elbows, tees, and reducers, hot spot stress is checked according to equivalent cross-sections or component coefficients. If necessary, circumferential-axial strength failure criteria (such as maximum principal stress, Mises equivalent stress) are checked to ensure... σ eq ≤ σ a / Y .

[0223] In this step, local component reinforcement and discontinuity strengthening, specifically for discontinuities such as elbows, tees, flanges, reinforcing rings, and weld seams, involves local thickening or reinforcement verification based on the component's characteristic coefficients. Specifically:

[0224] (1) Elbow: Minimum thickness min is considered based on the bending stress additional coefficient. t sd,elbow ;

[0225] (2) Tee / Branch: The cross-sectional equivalent strength shall be satisfied in accordance with the principle of hole reinforcement;

[0226] (3) Flange connection: Check the additional effects of sealing stress and bolt load on the body.

[0227] Step S233, based on the linkage and screening strategy, within the feasible region Ω of the process parameters ( T , P , v , χ Select the recommended run setpoint and its corresponding design thickness. t sd The linkage and filtering strategies include:

[0228] (1) Minimum thickness optimization: Search for the design thickness within the feasible region Ω(T,P,v,χ) of process parameters. t sd The minimum operating point is used as the recommended operating setpoint;

[0229] (2) Multi-objective trade-offs: Determine the objective function, and use the operating point corresponding to the minimum value of the objective function as the recommended operating setpoint. The objective function is as follows:

[0230]

[0231] in, R op For operational risk metrics (such as the degree of approaching the life boundary, the negative margin of Δlife, etc.); α , β , ζ均 As weight; x This is the operating point.

[0232] Step S234, through multi-level feasible region (Ω) p ,Ω p+ The recommended operating setpoint is coupled with reliability and safety margin, specifically including:

[0233] (1) Determine the design thickness t sd At the design boundary Ω p Inside;

[0234] (2) When the operating condition is close to the warning boundary Ω p+ At that time, regarding the design thickness t sd Apply safety thickening δ t Alternatively, the component partial factor can be increased to create an operational margin to resist boundary erosion caused by monitoring / modeling lag.

[0235] Step S235, design thickness t sd Mapping to standard specifications and reviewing them includes:

[0236] (1) Select the thickness specification after model selection t spec ≥ Design thickness t sd ;

[0237] (2) Combining manufacturing tolerances and inspection rules ( C + Already included C This confirms that the minimum measured thickness still meets the strength and lifespan requirements.

[0238] (3) A one-time comprehensive review of the equivalent thickness or minimum manufacturing thickness requirements of the corresponding components for reducers, elbows, tees, etc., to ensure the consistency of components.

[0239] In this embodiment, after performing the above steps, the corresponding output and interface are as follows:

[0240] The output data includes:

[0241] (1) Design thickness t sd With selected specifications t spec ;

[0242] (2) Recommended operating setpoints and their stress check list (circumferential, axial, equivalent stress, stability / service check results);

[0243] (3) Additional quantity set C (including corrosion allowance) C2) With source;

[0244] (4) If multi-objective optimization is adopted, output the scheme comparison table and weight settings.

[0245] This invention realizes the integrated design of "forward mode" water injection process parameters and pipeline strength for newly constructed water injection pipelines. It introduces deep learning to construct a corrosion rate prediction model to determine the predicted corrosion rate value and its corresponding conservative upper bound for design. C R * ( T , P , v , χ ), and then combined with the target's maximum service life. τ max Under structural lifetime constraints (satisfying) C R * ≤ C 2 / τ max The feasible region of process parameters allows for the realization of life targets as clear, actionable windows, avoiding the pitfalls of optimizing process parameters solely based on capacity or energy consumption while neglecting pipeline strength issues that corrosion may cause. Furthermore, strength verification and design thickness determination are completed within the feasible region of process parameters, reducing the impact of corrosion margins. C 2. Excessive thickness due to oversized components also avoids underestimating lifespan under highly corrosive conditions.

[0246] Example 3: If the type of water injection pipeline to be designed is an existing or in-service water injection pipeline, the corresponding integrated design method of water injection process parameters and pipeline strength based on the constraint of ultimate service life includes:

[0247] (a) Obtaining the temperature of the water injection system T ,pressure P Flow rate v and water quality or chemical parameters χ Field data or experimental data, and in-service scenario data of existing or in-service water injection pipelines, including:

[0248] Field data: Real-time acquisition of field data from key sections of the water injection station and pipeline, including water injection system temperature. T ,pressure P Flow rate v and water quality or chemical parameters χ (Including water quality or chemical parameters) χ Including dissolved oxygen (DO) and chloride ions (Cl). - (Including alkalinity / CO2, H2S, bacterial indicators, inhibitor dosage / effectiveness, etc.), and further, material, pipe diameter / wall thickness, welding coefficient, manufacturing deviation and operating condition records can be collected simultaneously.

[0249] Experimental data: Corrosion-related experimental results and corresponding water injection system temperatures obtained in a device simulating water injection media and operating conditions (performing experiments such as those involving padding, electrochemistry, and flow loops). T ,pressure P Flow rate v and water quality or chemical parameters χ .

[0250] Among them, the coupon experiment provides the average corrosion rate and local corrosion morphology after long-term exposure in the simulated medium; electrochemical tests such as polarization curves and electrochemical impedance spectroscopy (EIS) can quickly assess the corrosion tendency, rate and action mechanism of the material and the corrosion inhibitor; the flow loop experiment can simulate the flow state in the pipeline and study the influence of flow velocity and flow state (laminar / turbulent) on corrosion (especially scouring corrosion) and the mass transfer process of the corrosion inhibitor.

[0251] Data on the operational scenarios of existing or in-service water injection pipelines, including:

[0252] (1) Thickness / geometric measurement: online inspection (ultrasonic inspection UT / radiographic inspection RT), excavation, minimum remaining wall thickness at sampling points t min meas Calculate the dimensions of the corrosion defect (depth / length / circumferential angle) and the equivalent remaining wall thickness at the defect. t rem .

[0253] (2) Operational events: pressure rise / start-stop records, pipeline cleaning schedule, historical leakage / repair and reinforcement information.

[0254] (3) Positioning and segmentation: The pipe segment is divided into calculation units (station outlet - valve chamber - junction point), and each unit records independent data. .

[0255] Furthermore, for data from in-service scenarios, extreme values ​​and spatial interpolation markers can be added to the thickness measurement data (marking only, without extrapolation) to form an in-service parameter table.

[0256] (ii) Input water injection system temperature T ,pressure P Flow rate v and water quality or chemical parameters χ The field data or experimental data are fed into the corrosion rate prediction model to obtain the predicted corrosion rate value, and the corresponding conservative upper bound for design is determined. C R * ( T , P , v , χ ).

[0257] Step (II) involves the construction of the corrosion rate prediction model, which includes:

[0258] Step S311: Obtain the temperature of the water injection system. T ,pressure P Flow rate v and water quality or chemical parameters χ Historical field data or historical experimental data, along with the corresponding identifiers of actual corrosion rate values, are used to construct a standardized dataset, including:

[0259] (1) Obtain the temperature of the water injection system T ,pressure P Flow rate v and water quality or chemical parameters χ The historical field data or historical experimental data, along with the corresponding actual values ​​of corrosion rates, are as follows:

[0260] Historical field data: Historical field data collected in real time from key sections of water injection stations and pipelines over a historical period, including water injection system temperature. T ,pressure P Flow rate v and water quality or chemical parameters χ (Including water quality or chemical parameters) χ Including dissolved oxygen (DO) and chloride ions (Cl). - (Including alkalinity / CO2, H2S, bacterial indicators, inhibitor dosage / effectiveness, etc.), and further, material, pipe diameter / wall thickness, welding coefficient, manufacturing deviation and operating condition records can be collected simultaneously.

[0261] Historical experimental data: Corrosion-related experimental results and corresponding water injection system temperatures obtained in devices simulating water injection media and operating conditions (couplings, electrochemical, flow loop). T ,pressure P Flow rate v and water quality or chemical parameters χ .

[0262] Among them, the coupon experiment provides the average corrosion rate and local corrosion morphology after long-term exposure in the simulated medium; electrochemical tests such as polarization curves and electrochemical impedance spectroscopy (EIS) can quickly assess the corrosion tendency, rate and action mechanism of the material and the corrosion inhibitor; the flow loop experiment can simulate the flow state in the pipeline and study the influence of flow velocity and flow state (laminar / turbulent) on corrosion (especially scouring corrosion) and the mass transfer process of the corrosion inhibitor.

[0263] (2) Temperature of the water injection system T ,pressure P Flow rate v and water quality or chemical parameters χHistorical field data or historical experimental data are preprocessed, including anomaly handling, time alignment, and unit unification. Specifically:

[0264] Obvious outliers were removed using 3σ or IQR rules, and a log of the removal was retained. Next, multi-source data were aligned by measurement point and timestamp, unified to the same sampling interval, and time / spatial points were aligned. Measurement units and benchmarks (e.g., flow rate) were standardized. v In m / s, temperature T (Indicated by ℃, ion concentration by mg / L); short-term missing data are interpolated or deleted within a limited window, while long-term missing data retains the missing marker without inference; each data point is accompanied by the source type (field / experiment), measurement point number, instrument calibration date, and sampling frequency.

[0265] Furthermore, if both historical field data and historical experimental data exist, data fusion is performed. Specifically, the data is merged and compiled according to the same variable naming convention, the same time benchmark, and the same unit system. No predictions or corrections are made to the data; only source labeling and traceability are completed to provide clear input boundaries for subsequent steps.

[0266] (3) Put the preprocessed historical field data or historical experimental data into the same set to form a standardized dataset.

[0267] In this embodiment, when constructing a corrosion rate prediction model for existing or in-service water injection pipelines, a layered in-service time window can be introduced during preprocessing, using the water injection system temperature over the most recent M months. T ,pressure P Flow rate v and water quality or chemical parameters χ The main window is the data, which is used to build a standardized dataset. Early data is used for prior / smoothing.

[0268] Step S312: Determine the initial model and the corresponding physical prior constraints;

[0269] In this embodiment, the goal of determining the initial model is to establish the corrosion rate and process parameters (water injection system temperature) in the water injection system. T ,pressure P Flow rate v and water quality or chemical parameters χ Functional relationship between ) C R = f ( T , P , v , χ This allows for the acquisition of predicted corrosion rates and the determination of their conservative upper bound for design. C R * .

[0270] When constructing new pipelines with sample sizes typically ranging from hundreds to thousands, a robust and interpretable model with a small sample size can be prioritized as the initial model. This embodiment uses the Generalized Additive Model (GAM) as an example to illustrate the initial model and corresponding physical prior constraints. Specifically:

[0271] The initial model is determined based on the generalized additive model (GAM):

[0272]

[0273] in, This is a predicted corrosion rate. X =( T , P , v , χ ) is a set of variables; χ For water quality or chemical parameters (dissolved oxygen DO, chloride ions Cl) - Alkalinity / CO2, H2S, bacterial indicators, inhibitor dosage rate q inh / efficiency η inh wait); s i ( i ) is a univariate term, that is, the smoothing effect of each independent variable; s i,j ( i , j ) is a binary term used to retain only the interactions necessary for the project.

[0274] The physical prior constraints to be applied are as follows:

[0275] ,

[0276] The aforementioned physical prior constraints correspond to "monotonic splines and bounded splines," meaning that the flow velocity must be considered. v ,temperature T Applying a "non-reduction" constraint to the inhibitor dosage rate q inh Inhibitor efficacy η inh By applying "non-incremental" constraints or saturation constraints, a two-dimensional smooth can be constructed for the inhibitor term, expressing the coupling of "dosage rate × effectiveness".

[0277] It should also be noted that the initial model in this embodiment can also be constructed using other models, such as Gaussian process regression (GPR), with kernel functions such as ARD-RBF or Matern-ν (ν=3 / 2 or 5 / 2), and independent length scales set for each variable to reflect sensitivity; or gradient boosting tree (GBDT) or neural networks with uncertainty estimation (such as deep ensemble / MC Dropout), for comparison or when the sample size increases significantly.

[0278] Furthermore, it should be noted that for existing or in-service water injection pipelines, an operational time window layer can be introduced. This involves using the most recent M months of operational water quality data as the main window, with earlier data used for prior analysis / smoothing, and outputting quantifiable data. C R * ( T , P , v , χ ).

[0279] Step S313: Divide the standardized dataset into K subsets, and use the K-fold cross-validation method with time-block or measurement point-layer to train and validate the initial model, so as to obtain a corrosion rate prediction model that meets the output conditions.

[0280] In this embodiment, K in the K-fold cross-validation method can be selected as needed; in this embodiment, K can be 5 or 10. The implementation process includes:

[0281] The standardized dataset is divided into K mutually exclusive subsets. Specifically, the time-block K-fold cross-validation method divides the standardized dataset into K subsets in chronological order, while the measurement-point stratified K-fold cross-validation method stratifies the standardized dataset by measurement points to ensure that each subset contains data from different measurement points.

[0282] In each round of training, one subset is selected as the validation set to evaluate the model performance. The remaining K-1 subsets are merged as the training set to train the initial model. This process is repeated K times to ensure that each subset is used as the validation set exactly once.

[0283] If the initial model is determined based on the generalized additive model (GAM), then during training, monotonic / saturated priors can be achieved by selecting and shape-constraining splines using AIC / GCV.

[0284] The verification requirements mentioned above can be set as needed. In this embodiment, they may include, but are not limited to:

[0285] Indicator verification, determining R 2 Indicators such as RMSE and MAE are used, and residual independence and heteroscedasticity are checked.

[0286] Residual verification, using partial dependency curves to verify the rationality of the project (e.g.) v Increase C R Enlarge q inh Increase C R (The price decreases and then tends to plateau).

[0287] Sensitivity perturbation verification involves applying a sensitivity perturbation of ±(10~20)% to the input to verify the physical consistency between the output directionality and amplitude.

[0288] In this embodiment, when constructing a corrosion rate prediction model for existing or in-service water injection pipelines, different corrosion environments (such as differences in medium composition and flow regime) are considered. Pipe segment classification (unit ID) can be introduced, and pipe segments can be grouped according to unit ID to construct a corrosion rate prediction model for each group.

[0289] In step (ii), by quantifying the uncertainty, a conservative upper bound for design is constructed based on the predicted corrosion rate. C R * ( T , P , v , χ Taking the generalized additive model (GAM) as an example, when introducing the GAM to establish the initial model, a conservative upper bound for design is constructed based on quantile fitting. C R * ( T , P , v , χ ),include:

[0290] Pinball loss fitting of conditional quantiles Q p ( C R | X )(like p =95 or p + The conditional quantile of (=99) is used as a conservative upper bound for design. C R * ;

[0291]

[0292] in, Q p ( C R | X( ) represents the measured corrosion rate. C R The upper quantile function; L pin ( p ) represents pinball loss; p The value is the quantile level, ranging from [0.90, 0.99]. i Number the sample; C R This represents the measured corrosion rate (target value). This is a predicted corrosion rate. For indicator functions, when If the value is 1, then the value is 0; otherwise, the value is 0.

[0293] Under pinball loss, the model minimizes L pin (p) Directly approximates the upper quantile of the corrosion rate at a specified confidence level. When p =0.95 corresponds to a conservative upper bound at the design level (i.e., ensuring that the corrosion rate of approximately 95% of samples is below this value); when p + When the value is 0.99, it corresponds to a stricter warning boundary, which is used for early warning or safety verification during operation.

[0294] Therefore, the final measured corrosion rate value is... C R The upper quantile function can be written as the variance predicted by the composite model of cross-validation residuals and measurement errors, forming a conservative upper bound for design. C R * ( T , P , v , χ ), as shown below:

[0295]

[0296] in, For standard normal quantiles (e.g.) p =0.95 =1.645); This refers to the model prediction error; The standard deviation of the cross-validation residuals; This represents measurement error.

[0297] In the case of dual-source data, including both historical field data and historical experimental data, to reflect the difference in uncertainty between the two, confidence-weighted synthesis of the variances of the two parts can be performed. Furthermore, a density penalty coefficient can be added to sparse or extrapolated regions to make the boundary more conservative, avoiding optimistic estimates caused by local overfitting, and forming a conservative upper bound for design.C R * ( T , P , v , χ ), as shown below:

[0298]

[0299] in, Weights for the experimental data; The weights assigned to on-site data reflect the reliability of the data source; ρ(x) As an indicator of local sample density; This is the penalty coefficient for the sparse region; The standard deviation of the experimental data; This represents the standard deviation of the field data.

[0300] Furthermore, an extrapolation identification mechanism and a conservative backoff strategy can be set to modify the corrosion rate prediction model and design a conservative upper bound. C R * ( T , P , v , χ Specifically:

[0301] To prevent the corrosion rate prediction model from giving unreliable predictions due to sparse samples or outside the training domain, this embodiment establishes an extrapolation identification mechanism. This involves determining the Mahalanobis distance, kernel density, or nearest neighbor density of a given training data point in the training set. Points exceeding a threshold are recorded as extrapolated points; otherwise, they are considered in-domain points. Taking Mahalanobis distance as an example, the specific details are as follows:

[0302] Calculate the Mahalanobis distance of a given data point in the training set;

[0303]

[0304] like This indicates that the data is located in the "high-density region" of the training set, belonging to the domain point, and the corrosion rate prediction model is reliable.

[0305] like This indicates that the data is too far from the center, outside the training domain, and belongs to the extrapolation point, requiring increased conservatism.

[0306] in, d M The Mahalanobis distance is used to determine whether an input point is within the training domain. The mean vector of the training data; S To train the set covariance; These are the quantiles of the chi-square distribution, obtained from the statistical table;d The dimension of the feature space (e.g., the number of variables such as temperature, pressure, flow rate, dissolved oxygen, chloride ions, etc.); The significance level or the strictness of the extrapolation point determination is typically set to 0.99 or 0.95, and can be set to... Based on this, the sparse region penalty coefficient is set to... .

[0307] For the extrapolation point, a conservative rollback strategy is implemented, which includes increasing the level of conservatism (e.g., ...). p =0.95→ p =0.99), or revert to a corrosion rate prediction model containing only monotonic / saturated priors (i.e., based on a generalized additive model (GAM)), and adjust key operating variables (such as v , T Apply a safety reduction factor to the candidate upper bound to ensure that the boundary of the subsequent feasible region is not eroded by extrapolation, such as:

[0308]

[0309] in, The upper limit of safe flow rate; This is the upper limit of the flow rate; For safety reduction factor, ;

[0310] Or use methods such as sparsity penalty to... C R * To be further magnified.

[0311] It should also be noted that the corrosion rate prediction model and the conservative upper bound used in this embodiment are... C R * ( T , P , v , χ After construction is complete, three types of outputs will be provided, including:

[0312] (1) Callable prediction functions or lookup table interfaces, respectively, return the predicted corrosion rate values. With conservative upper limit of design C R * (T,P,v,χ);

[0313] (2) Model diagnosis and confidence parameters (R², RMSE, quantile p or variance term used, extrapolation threshold) are used for design review and verification;

[0314] (3) Complete version and audit information (training time window, variable list, model type and key hyperparameters, shape constraint configuration) are archived in the form of “model version vX.Y” to ensure the traceability of subsequent recalculation and rolling updates during the runtime.

[0315] (III) Setting the target maximum service life τ max Combined with a conservative upper bound design C R * ( T , P , v , χ The feasible domain of process parameters under construction lifetime constraints Ω( T , P , v , χ The lifespan constraint is based on the target maximum service life. τ max And design with a conservative upper bound C R * ( T , P , v , χ ) Determined, including:

[0316] Step S321 defines the corrosion-side input, search space, and lifetime feasibility criteria, specifically:

[0317] Corrosion-side input: Designed with a conservative upper bound C R * ( T , P , v , χ ) as the corrosion-side input;

[0318] Search space: The domain of candidate process variables. D =[ T min , T max ]×[ P min , P max ]×[ v min , v max ]× χ ;

[0319] Lifespan and thickness-related parameters: Target limit service life set [[ID=1"71]]τ max Corrosion allowanceC 2 is;

[0320] The lifespan feasibility criteria are as follows;

[0321]

[0322] The data points that meet the lifespan feasibility criterion are considered to be lifespan feasible.

[0323] Step S322: Construct a sampling set S in the search space and discretize it, including:

[0324] (1) Construct a sampling set S within the search space;

[0325] In this step, Latin hypercube sampling (LHS) or hypergrid sampling can be preferably used to sample within the search space to obtain a sample set S. It should be noted that the sample set S must satisfy the characteristics of sufficient coverage, uniformity and recalculation.

[0326] Taking Latin hypercube sampling (LHS) as an example, a sampling set S is constructed within the search space, including:

[0327] When the variable dimension is high or the χ² dimension is large, Latin hypercube sampling (LHS) is used (sample size recommended 10). 4 Starting from the order of magnitude, and allocating higher resolution to each dimension according to engineering sensitivity (e.g., for...). v , T (Encryption); Here, the number of χ dimensions is relatively large, which can be determined by setting a threshold. That is, if it is greater than the set threshold, it indicates that the number of χ dimensions is relatively large.

[0328] When the variable dimension is low or the boundary accuracy requirement is high, a hierarchical regular grid is adopted and adaptively refined in the boundary neighborhood; the requirement for high boundary accuracy can be determined by setting a threshold, that is, when it is greater than the set threshold, it indicates that the boundary accuracy requirement is high.

[0329] For discrete / enumerated components (such as different drug formulations), expand each term on χ to form a Cartesian product subdomain for block solution; all sampling points must retain source, random seed and grid step size information for recalculation.

[0330] Step S323, based on the extrapolation identification mechanism and conservative backoff strategy, ensures that each sampling point in the sampling set S is a non-extrapolation point, and combines this with the defined indicator function I( x ), keep all I( x The sampling points with )=1 form a feasible set Ω( x )={ x ∈S|I( x )=1}, where the indicator function I( x As shown below:

[0331]

[0332] It should be noted that, based on the extrapolation identification mechanism and conservative backoff strategy, it is ensured that each sampling point in the sampling set S is a non-extrapolation point, that is, for each sampling point... x =( T , P , v , χ )∈S, based on the extrapolation identification mechanism, it is determined whether it is an extrapolation point. Then, a conservative backoff strategy is used to improve the level of conservatism, and then based on the defined indicator function I( x The extrapolation identification mechanism and conservative backoff strategy are the same as those in step (II), and will not be described again.

[0333] Step S324, for the feasible set Ω ( x Connectivity extraction and boundary fitting are performed to obtain the feasible set Ω( x The volumetric domain description, the boundary functions / mesh of each connected domain, and the set of boundary curves for typical two-dimensional projections form the corresponding feasible domain Ω( for process parameters). T , P , v , χ ).

[0334] Specifically, this includes: marking connected components on the sampling grid according to adjacency relationships (such as K-nearest neighbors or regular grid 6 / 18 / 26-adjacency); eliminating isolated fragmented domains that are too small and unoperable; and using isosurfaces for each connected component. The interpolation / fitting results are used as boundaries (e.g., RBF interpolation, Marching Cubes / feasible isosurface extraction); for commonly used two-dimensional subspaces (e.g. v – T , v – q inh , T – P This generates a cross-sectional curve (the intersection line of the boundary on the cross-section), which facilitates operation settings and illustration.

[0335] Step S325: Determine the confidence band and multi-level boundaries. In this embodiment, multi-level feasible regions (Ω) can be constructed according to different confidence levels during model construction. p ,Ω p+ Specifically:

[0336] (1) Design boundary Ω p :use p =0.95 (or enterprise standard) C R * ;

[0337] (2) Warning boundary Ω p+ Use higher quantiles, such as p + =0.99 obtained C R * ;

[0338] (3) Operational recommendations: The strip between the boundaries serves as a warning zone, used for subsequent boundary crossing warnings and soft constraint control.

[0339] Step S326: Output the feasible domain description and deployment format;

[0340] For use in subsequent steps and project delivery, this embodiment generates one or more of the following equivalent descriptions in parallel:

[0341] (1) Lookup table / grid: Publish Boolean mask and boundary grid of Ω at a fixed step size for easy and quick identification;

[0342] (2) Implicit function: The implicit function approximation of the boundary is g(x)≈0, which can be used for numerical optimization and constraint discrimination;

[0343] (3) Inequality envelope: Fit upper and lower envelope curves / surfaces on commonly used two-dimensional / three-dimensional projections (e.g. v ∈[ v min ( T , χ ), v max ( T , χ This facilitates operation and configuration.

[0344] (4) Connected component metadata: For each connected component, provide the volume, principal scale (maximum allowable flow velocity) v max Maximum temperature T max (etc.) and representative center points (which can be used as initial operation setting points).

[0345] In this embodiment, it further includes forming boundary crossing criteria and interfaces, including:

[0346] Define lifetime out-of-bounds metric:

[0347]

[0348] when A value less than 0 indicates a feasible lifetime; for multi-level boundaries, warning thresholds can be further defined. For (by) p+ The corresponding upper bound is generated.

[0349] Output the following interface:

[0350] (1) Discriminant function: Input ( T , P , v , χ Returns the feasible / infeasible result and Δ. life ;

[0351] (2) Boundary query: Given any two (or three) variables, return the allowed intervals of the remaining variables (e.g., given any two (or three) variables). T , χ return v max ( T , χ ));

[0352] In this embodiment, further measures include shrinkage and conservative treatment (operational safety margin).

[0353] To avoid boundary erosion caused by model / monitoring lag, the feasible region Ω of the process parameters can be defined. T , P , v , χ Perform a shrinking operation, that is, adjust the criteria as follows:

[0354]

[0355] in, To provide a safety margin for the service life. >0; the contracted domain The window serves as a running suggestion window, while Ω remains the design verification domain.

[0356] In this embodiment, further quality control and recalculation requirements are included, namely, recording all sampling, discrimination, fitting, and projection steps: sampling method and random seed, sample size, step size / resolution, extrapolation point ratio, fitting error, boundary smoothing parameters, and their correspondence with the model version. Any input ( τ max , C 2. When the model version or data time window changes, a full-link recalculation or incremental update needs to be triggered, and a new version of the Ω and boundary file needs to be generated for subsequent strength verification.

[0357] In this embodiment, for existing or in-service water injection pipelines, the feasible domain Ω of process parameters is obtained. T , P , v , χ After that, prepare a boundary query interface for in-service back-reasoning: givenT , χ beg v max ( T , χ ); given v , χ beg T max ; or in a fixed P Seek below ( T , v , χ Allowed combinations of operations.

[0358] (iv) For existing or in-service water injection pipelines, conduct strength verification and current pressure-bearing capacity assessment. Specifically, without changing the pipeline specifications, calculate the current strength upper limit and establish operational hard constraints, including:

[0359] (1) Upper limit of pressure: based on the most unfavorable cross section ( t min meas or equivalent remaining wall thickness at ILI defects t rem Calculate the upper limit of pressure. P max ( t rem , D , σ a , E w , Y The selection of the approximate formula for thin-walled or thick-walled structures shall be in accordance with the specifications.

[0360] (2) Defect verification (if required): Estimate the bearing capacity at the defect using the equivalent notch or B31G / improved RSTRENG method, and correct the upper limit of bearing capacity. P max .

[0361] (3) Combined stress: Verify the axial / thermal stress and the stability of the external pressure section to obtain the allowable operating pressure range. P min , P max ].

[0362] (4) Segmentation: The above pressure-bearing capacity is output according to the unit ID, forming the segmented pressure-bearing upper limit. P max (i) .

[0363] (v) Determine the input data and solve for the in-service allowed operating window.

[0364] Step S351: Determine the input data and objective function, wherein the input data includes the feasible region Ω of the process parameters. T , P , v , χ The thinnest section of the pipe wall currently measured in the pipe box. t rem Target service life τ max Corrosion life criteria Intensity limit P ≤ P max The objective function is shown below:

[0365]

[0366] Step S352, solve for the in-service allowed operating window, including:

[0367] (1) Fixed strength boundary: for each pipe section i Segmented pressure limit P max (i) As a hard constraint under pressure, it forms a strength domain;

[0368] (2) Lifetime projection: at the upper limit of intensity P = P max (i) Or operating at normal pressure P * On the slice, the boundary query interface of step (iii) is used to obtain the temperature of the water injection system. T ,pressure P and water quality or chemical parameters χ The allowed domain is the lifetime domain;

[0369] (3) Window composition: Take the intersection of the lifetime domain and the intensity domain to obtain the segmented allowable operating window. ;

[0370] (4) Operationalization: Allow segmented running windows Convert to station control limit table: upper temperature limit T max (i) Flow rate limit v max (i) ( T , χ (Water quality or chemical parameter upper limit) χ (e.g., lower limit of inhibitor dosage) q min (i) Pressure limit Pmax (i) .

[0371] The above segmentation allows the running window to... This constitutes an active, running window.

[0372] Furthermore, to mitigate monitoring and model lag, the window can be shrunk, and the criterion can be replaced with... , or to v max Apply a reduction factor (e.g., 0.95×). v max ), forming a running suggestion window Ω δ ′ .

[0373] Example 4: This embodiment of the invention is a further optimization of the above embodiments, and also includes access to online / offline monitoring and periodic updates of the measured corrosion rate values. C R Design using a conservative upper bound C R * ( T , P , v , χ ) and the feasible range of process parameters Ω ( T , P , v , χ Perform closed-loop consistency verification and recalculation during pipeline operation, including:

[0374] Step S410: Using online and offline monitoring channels to obtain monitoring data, construct the operational dataset and in-service parameter table, including:

[0375] (1) Monitoring data is obtained using online and offline monitoring channels and preprocessed to form a runtime dataset. Specifically:

[0376] Monitoring data was obtained using online monitoring channels, including corrosion probes (linear polarization / resistive type) and online water quality data (DO, ). pH / alkalinity, temperature), flow rate / pressure / temperature, inhibitor dosage / efficiency ( q inh , η inh );

[0377] Monitoring data obtained using offline monitoring channels include: clip mass loss, periodic ultrasonic testing (UT) / radiographic testing (RT) thickness measurement, intelligent pipeline cleaning (ILI) metal loss and defect geometry, and bacterial culture / ATP.

[0378] Preprocessing includes: outlier 3σ / IQR removal, uniform granularity (e.g., 1–15 min), and uniform unit (…). v m / s T (℃, ion concentration: mg / L), short missing interpolation / long missing only marking, retain instrument status (calibration date / drift mark).

[0379] (2) Record the pipe section ID (station outlet - valve chamber - junction point) and minimum remaining wall thickness. t min meas Information on components (elbows, tees, valves) and historical repair / reinforcement records are used to form an in-service parameter table.

[0380] Step S420, based on the runtime dataset and the in-service parameter table, sequentially performs domain identification, periodic consistency verification, and tiered alarms, including:

[0381] (1) Domain identification, including:

[0382] Define a lifespan consistency criterion: if Δlife≥0, the lifespan is feasible; otherwise, the lifespan is out of bounds.

[0383]

[0384] Define the strength consistency criterion. Then it is considered qualified;

[0385]

[0386] in, It is the equivalent stress (MPa), calculated from the load (pressure, temperature, external load, etc.); Y is the allowable stress of the material (MPa), usually taken as a percentage of the yield strength or tensile strength; Y is the design factor (safety factor). As a measure of strength compliance; Measurement of lifespan exceeding limits;

[0387] In-domain identification is performed based on the defined lifetime consistency criterion and the defined strength consistency criterion;

[0388] Furthermore, before calculating Δlife, the current operating condition can be determined using Mahalanobis distance / kernel density. x t =( T , P , v , χ If the point is in the training domain, it is an extrapolation point. The level of conservatism needs to be increased (p=0.95→p+=0.99) or it needs to be reverted to the baseline model containing only monotonic / saturated priors, and then Δlife is calculated.

[0389] (2) Periodic consistency check

[0390] In this embodiment, the periodic consistency check is to evaluate the cycle time and representative working conditions. Specifically, it involves generating representative working conditions by summarizing window period data (such as the past 24 hours) according to a set cycle time (hours / day). x t (Mean or upper quantile), and record the health status (missing rate, drift) of each monitoring channel;

[0391] The hierarchical determination is performed according to the hierarchical rules, which are as follows:

[0392] Green (normal): x t ∈Ω δ and U σ ≤ U σ warn (like U σ warn =0.85).

[0393] Yellow (Alert): x t ∈Ω\Ω δ or U σ warn < U σ ≤ U σ crit (like U σ warn =0.90).

[0394] Red (Out of Bounds): Δlife < 0 or U σ > U σ crit .

[0395] Gray (Data Anomaly): Critical channels are missing or drifting, entering conservative mode.

[0396] Among them, Ω δ Ω represents the feasible region of process parameters after applying a safety margin; U σ warn This is the critical threshold. U σ crit This is the warning threshold.

[0397] Furthermore, event-triggered verification can be performed, meaning that when events such as water source switching, pressure increase / stop, pipe cleaning, or abnormal alarms occur, U is immediately recalculated based on the event window data. σ Skip the cycle and wait.

[0398] In step S430, if at least one triggering condition is met, the Change Management (MOC) is triggered, and process parameter adjustments or redesign and pipe replacement are given.

[0399] (1) Triggering conditions, including:

[0400] Δlife < 0 for N consecutive evaluation periods;

[0401] U σ > U σ crit Continues for M cycles;

[0402] ILI / UT detected t min meas The decline exceeds the threshold, among which t min meas Minimum remaining wall thickness;

[0403] Water source drifts over a long period of time;

[0404] Model / feasible domain version mismatch.

[0405] (2) Change Management Organization (MOC), including:

[0406] Yellow (Warning) soft control (does not trigger MOC).

[0407] (1) Adding medication: to enhance q inh Minimum requirement of this variable to the lifetime boundary. q inh min ( T , v Add a safety margin above )

[0408] (2) Rate limiting: Limiting the flow rate to... v ≤0.95 v max ( T , χ );

[0409] (3) Temperature control: Reduce T within the allowable range to increase Δlife;

[0410] (4) Time-sharing staggered peak hours: High-risk periods will be staggered... v , P Lowered.

[0411] Red (out of bounds) hard control (triggers MOC).

[0412] (1) Segmented pressure reduction: Adjust the upper limit of the operating pressure to P ≤ P max safe ;

[0413] (2) Traffic redistribution: reducing high-risk segments across segments. v And compensation will be coordinated by the station network;

[0414] (3) Temporary reinforcement / sleeve (when defects are dominant);

[0415] (4) Planned suspension and replacement (when the process cannot resolve the boundary).

[0416] Furthermore, this embodiment can also be configured for closed-loop recalculation, version binding and release, and human-computer interaction and state machine, specifically:

[0417] (a) Closed-loop recalculation, version binding and release

[0418] Recalculation triggers and content: Rolling recalculation is triggered when the sample size / error threshold is met or when a change management MOC is triggered. Rolling recalculation includes retraining. C R * ( T , P , v , χ ), Reconstruct Ω, Ω δ Review if necessary P max Verification with components.

[0419] Version binding maps the evaluation results, actions, and limit tables to the model / feasible domain / intensity version, generating a "run window version vX.YZ".

[0420] The publishing format outputs limits and interfaces to DCS / reports, including:

[0421] (1) Limit Table: P max , v max ( T , χ ), T max ( v , χ ), q inh min ( T , v), including the "warning zone / hard threshold" label;

[0422] (2) API discrimination: Input ( T , P , v , χ Return Δlife, U σ Domain identifiers and recommended actions;

[0423] (3) Alarm rules: graded threshold, jitter suppression (minimum duration) and score clearing conditions.

[0424] (ii) Human-computer interaction and state machines (optional but recommended)

[0425] Visual elements, Ω, Ω δ Two-dimensional projection ( v – T , v – q inh Current setpoint trajectory, Δlife U σ Indicator table, channel health status.

[0426] The state machine has two branches: [green→yellow→red→MOC→green] and [gray→conservative→green]. The switching conditions are given by S3 / S4. Each state transition is automatically written to the audit log and triggers a limit update.

[0427] This invention addresses newly built or in-service water injection systems by establishing a closed-loop mechanism of "online monitoring → consistency verification → graded handling / change management (MOC) → model / feasibility domain recalculation → versioned archiving" to ensure that the operating parameters (T, P, v, χ) continuously meet the lifespan and strength constraints determined by this invention.

[0428] Example 5: As shown in the attached document Figure 5 As shown, this invention discloses an integrated design system for water injection process parameters and pipeline strength based on ultimate service life constraints, comprising:

[0429] The data acquisition unit determines the type of water injection pipeline to be designed and acquires the corresponding basic data. The types of water injection pipelines to be designed include newly constructed water injection pipelines and existing or in-service water injection pipelines. The basic data for newly constructed water injection pipelines includes the water injection system temperature. T ,pressure P Flow rate v and water quality or chemical parameters χ Field data or experimental data, basic data corresponding to existing or in-service water injection pipelines, including in-service scenario data and water injection system temperature. T ,pressure PFlow rate v and water quality or chemical parameters χ Field data or experimental data;

[0430] Corrosion response and upper limit determination unit, input water injection system temperature T ,pressure P Flow rate v and water quality or chemical parameters χ The field data or experimental data are fed into the corrosion rate prediction model to obtain the predicted corrosion rate value, and the corresponding conservative upper bound for design is determined. C R * ( T , P , v , χ The corrosion rate prediction model was constructed using deep learning.

[0431] Feasible domain generation unit, setting target maximum service life τ max Combined with a conservative upper bound design C R * ( T , P , v , χ The feasible domain of process parameters under construction lifetime constraints Ω( T , P , v , χ The lifespan constraint is based on the target maximum service life. τ max And design with a conservative upper bound C R * ( T , P , v , χ )Sure;

[0432] Strength and thickness design units, for newly built water injection pipelines, within the feasible range of process parameters Ω ( T , P , v , χ Strength verification and design thickness determination are carried out within the specified range.

[0433] The process parameter range determination unit determines the allowable range of process parameters for existing or in-service water injection pipelines, under given material and thickness constraints.

[0434] Example 6: As attached Figure 6As shown, this invention discloses an integrated design system for water injection process parameters and pipeline strength based on ultimate service life constraints, comprising:

[0435] The data acquisition unit determines the type of water injection pipeline to be designed and acquires the corresponding basic data. The types of water injection pipelines to be designed include newly constructed water injection pipelines and existing or in-service water injection pipelines. The basic data for newly constructed water injection pipelines includes the water injection system temperature. T ,pressure P Flow rate v and water quality or chemical parameters χ Field data or experimental data, basic data corresponding to existing or in-service water injection pipelines, including in-service scenario data and water injection system temperature. T ,pressure P Flow rate v and water quality or chemical parameters χ Field data or experimental data;

[0436] Corrosion response and upper limit determination unit, input water injection system temperature T ,pressure P Flow rate v and water quality or chemical parameters χ The field data or experimental data are fed into the corrosion rate prediction model to obtain the predicted corrosion rate value, and the corresponding conservative upper bound for design is determined. C R * ( T , P , v , χ The corrosion rate prediction model was constructed using deep learning.

[0437] Feasible domain generation unit, setting target maximum service life τ max Combined with a conservative upper bound design C R * ( T , P , v , χ The feasible domain of process parameters under construction lifetime constraints Ω( T , P , v , χ The lifespan constraint is based on the target maximum service life. τ max And design with a conservative upper bound C R * ( T , P , v ,χ )Sure;

[0438] Strength and thickness design units, for newly built water injection pipelines, within the feasible range of process parameters Ω ( T , P , v , χ Strength verification and design thickness determination are carried out within the specified range.

[0439] The process parameter range determination unit determines the allowable range of process parameters for existing or in-service water injection pipelines, under given material and thickness constraints.

[0440] The closed-loop verification unit connects to online / offline monitoring and periodically updates the measured corrosion rate values. C R Design using a conservative upper bound C R * ( T , P , v , χ ) and the feasible range of process parameters Ω ( T , P , v , χ Perform closed-loop consistency verification and recalculation during pipeline operation, including:

[0441] Monitoring data is obtained using online and offline monitoring channels, and operational datasets and in-service parameter tables are constructed.

[0442] Based on the runtime dataset and the in-service parameter table, the following steps are performed in sequence: domain identification, periodic consistency verification, and hierarchical alarms.

[0443] If at least one triggering condition is met, change management is triggered, and process parameter adjustments or redesign and pipe replacement are required. The triggering conditions include: lifespan out-of-bounds measurement Δlife < 0 for N consecutive evaluation cycles. U σ > U σ crit Detected by ILI / UT for M consecutive cycles. t min meas Falling threshold; long-term water source drift; model / feasible region version mismatch;

[0444] in , τ max To achieve the target maximum service life, C R * To use a conservative upper bound in the design, To allow for corrosion, Measurement of lifespan exceeding limits;

[0445] in, , For equivalent stress, Y is the allowable stress of the material, and Y is the design factor. U σ crit This is the warning threshold.

[0446] The above content is only a specific embodiment of the present invention, which has strong adaptability and implementation effect. However, the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be covered within the protection scope of the present invention. Therefore, equivalent changes made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1. A method for integrating water injection process parameters and pipeline strength design based on ultimate service life constraints, characterized in that, include: Determine the type of water injection pipeline to be designed and obtain the corresponding basic data. The types of water injection pipelines to be designed include newly built water injection pipelines and existing or in-service water injection pipelines. The basic data for newly built water injection pipelines includes the water injection system temperature. T ,pressure P Flow rate v and water quality or chemical parameters χ Field data or experimental data, basic data corresponding to existing or in-service water injection pipelines, including in-service scenario data and water injection system temperature. T ,pressure P Flow rate v and water quality or chemical parameters χ Field data or experimental data; Input water injection system temperature T ,pressure P Flow rate v and water quality or chemical parameters χ The field data or experimental data are fed into the corrosion rate prediction model to obtain the predicted corrosion rate value, and the corresponding conservative upper bound for design is determined. C R * ( T , P , v , χ The corrosion rate prediction model was constructed using deep learning. Set target maximum service life τ max Combined with a conservative upper bound design C R * ( T , P , v , χ The feasible domain of process parameters under construction lifetime constraints Ω( T , P , v , χ The lifespan constraint is based on the target maximum service life. τ max And design with a conservative upper bound C R * ( T , P , v , χ )Sure; For newly constructed water injection pipelines, within the feasible range of process parameters Ω ( T , P , v , χ Within the specified range, strength verification and design thickness determination are carried out. For existing or in-service water injection pipelines, the allowable range of process parameters is obtained under the constraints of given materials and thickness.

2. The integrated design method for water injection process parameters and pipeline strength based on ultimate service life constraints as described in claim 1, characterized in that, The process of constructing a corrosion rate prediction model includes: Obtain multiple sets of water injection system temperatures T ,pressure P Flow rate v and water quality or chemical parameters χ A standardized dataset is constructed using historical field data or historical experimental data, along with the corresponding actual values ​​of corrosion rates. Determine the initial model and the corresponding physical prior constraints; The standardized dataset is divided into K subsets, and the initial model is trained and validated using K-fold cross-validation with time-block or measurement point-layer methods to obtain a corrosion rate prediction model that meets the output conditions.

3. The integrated design method for water injection process parameters and pipeline strength based on ultimate service life constraints as described in claim 2, characterized in that, Determine the initial model and corresponding physical prior constraints, including: Determining the initial model based on the generalized additive model: in, This is a predicted corrosion rate. X =( T , P , v , χ ) is a set of variables; s i ( i ) is a unary term; s i,j ( i , j ) is a binary term; The physical prior constraints that need to be applied are as follows: in, For coefficients; q inh Inhibitor dosage rate; The efficacy rate of the inhibitor.

4. The integrated design method for water injection process parameters and pipeline strength based on ultimate service life constraints as described in claim 2 or 3, characterized in that, It also incorporates an extrapolation identification mechanism and a conservative backoff strategy to revise the corrosion rate prediction model and design a conservative upper bound. C R * ( T , P , v , χ ),include: An extrapolation identification mechanism is set up based on Mahalanobis distance, kernel density, or nearest neighbor density components. The training data is obtained from a standardized dataset to determine whether the training data is an extrapolation point. If it is an extrapolation point, a conservative rollback strategy is executed, which includes increasing the level of conservatism or rolling back to a corrosion rate prediction model containing only monotonic / saturated priors, and applying a safety reduction factor to the candidate upper limit of key operating variables.

5. The integrated design method for water injection process parameters and pipeline strength based on ultimate service life constraints according to claim 1, 2, or 3, characterized in that, Conservative upper bounds are constructed based on quantile fitting. C R * ( T , P , v , χ ).

6. The integrated design method for water injection process parameters and pipeline strength based on ultimate service life constraints according to claim 1, 2, or 3, characterized in that, Set target maximum service life τ max Combined with a conservative upper bound design C R * ( T , P , v , χ The feasible domain of process parameters under construction lifetime constraints Ω( T , P , v , χ The lifespan constraint is based on the target maximum service life. τ max And design with a conservative upper bound C R * ( T , P , v , χ ) Determined, including: Define the corrosion-side input, search space, and lifetime feasibility criteria, specifically: Corrosion-side input: Designed with a conservative upper bound C R * ( T , P , v , χ ) as the corrosion-side input; Search space: The domain of candidate process variables. D =[ T min , T max ]×[ P min , P max ]×[ v min , v max ]× χ ; Lifespan and thickness-related parameters: Target limit service life set τ max Corrosion allowance C 2; The lifespan feasibility criteria are as follows; ； Construct a sample set S in the search space and discretize it; Based on the extrapolation identification mechanism and conservative backoff strategy, it is ensured that each sampling point in the sampling set S is a non-extrapolation point, and this is combined with the defined indicator function I( x ), keep all I( x The sampling points with )=1 form a feasible set Ω( x )={ x ∈S|I( x )=1}, where the indicator function I( x As shown below: ； For feasible set Ω ( x Connectivity extraction and boundary fitting are performed to obtain the feasible set Ω( x The volumetric domain description, the boundary functions / mesh of each connected domain, and the set of boundary curves for typical two-dimensional projections form the corresponding feasible domain Ω( for process parameters). T , P , v , χ ); Determine the confidence band and multi-level boundaries, and output the feasible domain description and publication format of process parameters.

7. The integrated design method for water injection process parameters and pipeline strength based on ultimate service life constraints according to claim 1, 2, or 3, characterized in that, For newly constructed water injection pipelines, within the feasible range of process parameters Ω ( T , P , v , χ Strength verification and design thickness determination are carried out within the specified range, including: Define the input and target, specifically: Input: Includes candidate operating points x =( T , P , v , χ )∈Ω, material grade and allowable stress σ a ( T ), welding joint coefficient E w Design coefficient set Y Geometric parameters, set of additional quantities C ; Objective: To achieve the feasible process parameters within the Ω ( T , P , v , χ Complete the strength check and determine the design thickness within ) t sd ; Determine the required wall thickness for each candidate operating point under internal pressure bearing strength. t s and add additional amount C Obtain the design thickness t sd Then, axial and combined stress checks are performed, and local components and discontinuities are reinforced, including the design thickness. t sd The calculation process is as follows: in, t s Calculate the thickness of the pipe; t sd Design the thickness of the pipe; Design pressure for the pipeline; D The outer diameter of the pipe; σ a ( T ) for temperature T Yield strength below; E w For welding joint coefficient; Y Design coefficient; C For additional quantity set; C 1 is for manufacturing negative deviation; C 2 represents corrosion allowance; E For coefficients; Based on the linkage and screening strategy, within the feasible region Ω of the process parameters ( T , P , v , χ Select the recommended run setpoint and its corresponding design thickness. t sd ; Through multi-level feasible regions (Ω) p ,Ω p+ Reliability and safety margin are coupled at the recommended operating setpoint; Design thickness t sd Map to standard specifications and verify them; or / and, For existing or in-service water injection pipelines, under given material and thickness constraints, the permissible range of process parameters includes: Without changing the pipeline specifications, calculate the current strength limit and form hard constraints on the operation side; Given the input data, calculate the allowed operating window in service, including: Determine the input data and objective function, where the input data includes the feasible region Ω of the process parameters. T , P , v , χ The equivalent remaining wall thickness at the currently measured defect location in the pipe section. t rem Target service life τ max Corrosion life criteria C R ∗ ≤ C 2 / τ max ,pressure P ≤Intensity Upper Limit P max The objective function is as follows: Determine the allowed running window in active service, including: Fixed strength boundary: for each pipe section i Upper limit of segmented strength P max (i) As a hard constraint under pressure, it forms a strength domain; Lifespan projection: under pressure P = Upper limit of segmented strength P max (i) Or operating at normal pressure P * On the slice, the boundary query interface is used to obtain the temperature of the water injection system. T ,pressure P and water quality or chemical parameters χ The allowed domain is the lifetime domain; Window composition: The intersection of the lifetime domain and the intensity domain is used to obtain the segmented allowable operating window. ; Operationalization: Allow segmented running windows Convert to a station control limit table, which includes the upper temperature limit. T max (i) Flow rate limit v max (i) ( T , χ ( ) Upper limit of water quality or chemical parameters χ max Upper limit of segmented strength P max (i) .

8. The integrated design method for water injection process parameters and pipeline strength based on ultimate service life constraints according to claim 1, 2, or 3, characterized in that, It also includes access to online / offline monitoring and periodic updates of measured corrosion rate values. C R Design using a conservative upper bound C R * ( T , P , v , χ ) and the feasible range of process parameters Ω ( T , P , v , χ Perform closed-loop consistency verification and recalculation during pipeline operation, including: Monitoring data is obtained using online and offline monitoring channels, and operational datasets and in-service parameter tables are constructed. Based on the runtime dataset and the in-service parameter table, the following steps are performed in sequence: domain identification, periodic consistency verification, and hierarchical alarms. If at least one trigger condition is met, change management will be triggered and process parameter adjustments or redesign and pipe replacement will be required. The trigger conditions include: Δlife < 0 for N consecutive evaluation cycles. U σ > U σ crit Detected by ILI / UT for M consecutive cycles. t min meas The threshold for decline is exceeded, where UT refers to periodic ultrasonic detection and ILI refers to intelligent pipeline cleaning. t min meas Minimum remaining wall thickness; long-term water source drift; model / feasible region version mismatch; in , τ max To achieve the target maximum service life, C R * To use a conservative upper bound in the design, To allow for corrosion, Measurement of lifespan exceeding limits; in, , For equivalent stress, Y is the allowable stress of the material, and Y is the design factor. U σ crit This is the warning threshold.

9. A system for integrated design of water injection process parameters and pipeline strength based on ultimate service life constraints, applied to the method described in any one of claims 1 to 8, characterized in that, include: The data acquisition unit determines the type of water injection pipeline to be designed and acquires the corresponding basic data. The types of water injection pipelines to be designed include newly constructed water injection pipelines and existing or in-service water injection pipelines. The basic data corresponding to newly constructed water injection pipelines includes the water injection system temperature. T ,pressure P Flow rate v and water quality or chemical parameters χ Field data or experimental data, basic data corresponding to existing or in-service water injection pipelines, including in-service scenario data and water injection system temperature. T ,pressure P Flow rate v and water quality or chemical parameters χ Field data or experimental data; Corrosion response and upper limit determination unit, input water injection system temperature T ,pressure P Flow rate v and water quality or chemical parameters χ The field data or experimental data are fed into the corrosion rate prediction model to obtain the predicted corrosion rate value, and the corresponding conservative upper bound for design is determined. C R * ( T , P , v , χ The corrosion rate prediction model was constructed using deep learning. Feasible domain generation unit, setting target maximum service life τ max Combined with a conservative upper bound design C R * ( T , P , v , χ The feasible domain of process parameters under construction lifetime constraints Ω( T , P , v , χ The lifespan constraint is based on the target maximum service life. τ max And design with a conservative upper bound C R * ( T , P , v , χ )Sure; Strength and thickness design units, for newly built water injection pipelines, within the feasible range of process parameters Ω ( T , P , v , χ Strength verification and design thickness determination are carried out within the specified range. The process parameter range determination unit determines the allowable range of process parameters for existing or in-service water injection pipelines, under given material and thickness constraints.

10. The integrated design system for water injection process parameters and pipeline strength based on ultimate service life constraints according to claim 9, characterized in that, It also includes a closed-loop verification unit, which connects to online / offline monitoring and periodically updates the measured corrosion rate values. C R Design using a conservative upper bound C R * ( T , P , v , χ ) and the feasible range of process parameters Ω ( T , P , v , χ Perform closed-loop consistency verification and recalculation during pipeline operation, including: Monitoring data is obtained using online and offline monitoring channels, and operational datasets and in-service parameter tables are constructed. Based on the runtime dataset and the in-service parameter table, the following steps are performed in sequence: domain identification, periodic consistency verification, and hierarchical alarms. If at least one triggering condition is met, change management is triggered, and process parameter adjustments or redesign and pipe replacement are required. The triggering conditions include: lifespan out-of-bounds measurement Δlife < 0 for N consecutive evaluation cycles. U σ > U σ crit Detected by ILI / UT for M consecutive cycles. t min meas The threshold for decline is exceeded, where UT refers to periodic ultrasonic detection and ILI refers to intelligent pipeline cleaning. t min meas Minimum remaining wall thickness; long-term water source drift; model / feasible region version mismatch; in , τ max To achieve the target maximum service life, C R * To use a conservative upper bound in the design, To allow for corrosion, Measurement of lifespan exceeding limits; in, , For equivalent stress, Y is the allowable stress of the material, and Y is the design factor. U σ crit This is the warning threshold.

Images