A method for designing the dosage of flushing isolation fluid based on BP neural network

Abstract

The present invention provides a design method for the dosage of flushing type spacer fluid based on BP neural network, including step (1): determining the main control factors affecting the mixing law of cementing annulus fluid. Step (2): taking the main control factors as independent variables, randomly generating M groups of data, and substituting them into the FLUENT software to calculate the three-phase mixing ratio P of each random array. Quantify the annulus mixing pollution risk under the influence of different factors, randomly generate multiple groups of working condition data according to the on-site working condition parameters and substitute them into FLUENT for calculation. Based on the BP neural network, conduct data training on the calculation results of the random arrays. After continuousizing the data, build a basic database for the design of the spacer fluid dosage, and then combine the results of the flushing experiment. Finally, form a quantitative design program for the spacer fluid dosage. By directly inputting the working condition parameters on the user interface of this program, the dosage of the spacer fluid under the corresponding working conditions can be obtained, which makes up for the deficiencies of the existing technology to a certain extent.

Description

Technical Field

[0001] The invention belongs to the technical field of oil and gas well displacement, and in particular relates to a method for designing the amount of flushing type spacer fluid based on a BP neural network. Background Art

[0002] During the cementing process of oil and gas wells, the isolation fluid, as an indispensable fluid, can isolate the drilling fluid and cement slurry to prevent contact contamination on the one hand; on the other hand, it can also flush the virtual filter cake on the casing wall and the well wall, displace the drilling fluid retained in the annulus, and achieve the purpose of improving the displacement efficiency and the bonding quality of the annulus interface. How to determine the amount of isolation fluid has always been a major problem that plagues field engineering applications. The patent document with application number 201810523524.7 discloses a quantitative method for the reasonable amount of cementing isolation fluid, which uses the Hele-shaw model to calculate the displacement interface length when cement slurry replaces drilling fluid, and then calculates the pre-fluid section length and pre-fluid dosage based on the experimentally measured flushing efficiency. Since the model used in this method is a two-dimensional model, it cannot reflect the three-dimensional flow characteristics of cementing displacement, and fails to analyze the amount of isolation fluid from the perspective of the pollution of the three-phase slurry of drilling fluid, isolation fluid and cement slurry. In addition, field engineers often determine the amount of isolation fluid through interface flushing experiments, that is, using the turbulent flushing time t as the evaluation index (generally t ≥ 10min), and determine the shortest flushing time under the premise of ensuring 90% flushing efficiency. Generally speaking, as the flushing time increases, the flushing efficiency will hardly increase, but will affect the stability of the well wall and even cause engineering accidents such as well collapse. However, the amount of isolation fluid determined by this method has great limitations. It only considers the flushing effect of the isolation fluid, not its isolation effect. The risk of slurry contamination caused by factors such as casing eccentricity, wellbore expansion, and fluid performance still exists.

[0003] In summary, it is necessary to provide a design method for the amount of flushing spacer fluid that meets the actual needs of the project from the perspective of efficient displacement and pollution prevention. The three-dimensional flow process of cement displacement is simulated by numerical simulation, and the error problem caused by the two-dimensional calculation model is avoided. Combined with the results of interface scouring experiments, the amount of spacer fluid under different wellbore conditions and construction parameters is determined based on the BP neural network algorithm on the basis of analyzing the pollution of the three-phase slurry of drilling fluid, spacer fluid and cement slurry. This method has the dual functions of efficient displacement and pollution prevention, and can provide guidance for the design of spacer fluid dosage under different working conditions. According to the search of domestic and foreign literature, no similar quantitative design method for flushing spacer fluid dosage has been given. Summary of the invention

[0004] The object of the present invention is to solve the defects existing in the above-mentioned prior art, and provide a design method for the dosage of flushing-type spacer fluid based on BP neural network. This method quantifies the annulus mixing pollution risk under the influence of different factors, randomly generates multiple groups of working condition data according to the on-site working condition parameters and substitutes them into FLUENT for calculation, trains the data of the calculation results of the random arrays based on BP neural network, builds a basic database for the design of spacer fluid dosage after continuousizing the data, and then combines the results of flushing experiments. Finally, a quantitative design program for the dosage of spacer fluid is formed. By directly inputting the working condition parameters on the user interface of this program, the dosage of spacer fluid under the corresponding working conditions can be obtained, which makes up for the deficiencies of the prior art to a certain extent.

[0005] The present invention adopts the following technical solutions:

[0006] A design method for the dosage of flushing-type spacer fluid based on BP neural network, which successively includes the following steps:

[0007] Step (1): Determine several main control factors (such as eccentricity, hole enlargement, spacer fluid dosage, displacement, fluid properties, etc.) that affect the mixing law of well cementing annulus fluid;

[0008] Step (2): Take the main control factors as independent variables, randomly generate M groups of data, and calculate the three-phase mixing ratio P of each random array by bringing them into the FLUENT software;

[0009] Step (3): Quantify multiple parameters that affect mixing pollution, obtain the pollution risk factor N (N ∈ [0, 1]) to quantitatively evaluate the annulus mud mixing pollution risk. The closer N is to 1, the greater the pollution risk;

[0010] Step (4): According to the calculation results of step (2), use the BP neural network training method to complete the construction of the basic database D for the design of spacer fluid dosage, which is specifically divided into the following steps:

[0011] Step (4-1): Build the mapping function relationship (hidden layer) between the three-phase mixing ratio P (cement slurry, drilling fluid, spacer fluid) and the pollution risk factor N under the influence of each main control factor;

[0012] Step (4-2): Based on the above mapping relationship, establish a GUI instruction interface (input layer and output layer);

[0013] Step (5): Make a drilling fluid filter cake with the help of a high-temperature and high-pressure fluid loss instrument;

[0014] Step (6): Obtain the relationship between the filter cake flushing time and the flushing efficiency through experimental tests, and determine the optimal flushing time t on the premise of ensuring the flushing efficiency;

[0015] Step (7): Calculate the dosage of spacer fluid at the optimal flushing time t, and the calculation formula is

[0016] V = 60Avt(1 - 1)

[0017] Wherein, V is the amount of spacer fluid, with the unit of m 3 ; v is the annular return velocity, with the unit of m / s; t is the flushing time, with the unit of min; A is the annular area, with the unit of m 2 ;

[0018] Step (8): Substitute the amount of spacer fluid V calculated in step (7) into the FLUENT software to calculate the three-phase mixing ratio P*;

[0019] Step (9): Substitute the three-phase mixing ratio P* calculated in step (8) into the mapping function relationship between P and N established in step (4 - 1) to obtain the pollution risk factor N*. Judge N*. If N * ≤0.5, the pollution risk is controllable, and this amount of spacer fluid V meets the requirements and can be output; if N* > 0.5, the pollution risk is relatively large, and this amount of spacer fluid V does not meet the requirements, and it is necessary to return to adjust the amount of spacer fluid until N* ≤ 0.5.

[0020] Advantages of the present invention:

[0021] The advantages of the present invention lie in providing a design method for the amount of spacer fluid based on a BP neural network, comprehensively considering the flushing effect and the isolation effect. Compared with determining the amount of spacer fluid using traditional empirical methods or experimental methods, the present invention uses the BP neural network training method to integrate the basic database and takes into account the flushing efficiency and the mixing pollution risk, and at the same time establishes a design program for the amount of spacer fluid. Starting from the functions of flushing and isolation, it can conveniently and quickly provide accurate guidance for the amount of on-site spacer fluid. Description of the drawings

[0022] Figure 1 It is the block diagram of the design program for the amount of spacer fluid in the present invention;

[0023] Figure 2 、 Figure 3 、 Figure 4 、 Figure 5 They are respectively the relationship diagrams between the filter cake flushing time and the flushing efficiency of each well. Specific embodiments

[0024] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments in the present invention belong to the scope of protection of the present invention.

[0025] Example 1

[0026] As Figure 1 shown, in this example, the liner cementing of Well GS-131X in Block M is used as the reference object. The length of the cemented section of this well is 2,809 m, the annular return velocity v = 1.53 m / s, and the annular area A = 0.01178 m 2 , and the specific implementation steps are as follows:

[0027] Step (1): Determine the four major main control factors affecting the mixing law of the annular fluid during cementing: eccentricity, hole enlargement, displacement of spacer fluid, and displacement rate;

[0028] Step (2): Taking the on-site construction parameters of Block M as a reference, using the above four major main control factors as independent variables, generate 100 sets of random arrays with different eccentricities, hole enlargements, displacements of spacer fluid, and displacement rates, and calculate the corresponding three-phase mixing ratios P of each random array by substituting them into the FLUENT software;

[0029] Step (3): Quantify multiple parameters affecting the mixing pollution, and obtain the pollution risk factor N (N ∈ [0, 1]) to quantitatively evaluate the pollution risk of the annular slurry mixing. The closer N is to 1, the greater the pollution risk;

[0030] Step (4): According to the calculation results of Step (2), use the BP neural network training method to complete the construction of the basic database D for the design of the displacement of spacer fluid, which is specifically divided into the following steps:

[0031] Step (4-1): Establish the mapping function relationship (hidden layer) between the three-phase mixing ratio P and the pollution risk factor N under the influence of each main control factor;

[0032] Step (4-2): Based on the above mapping relationship, establish a GUI instruction interface (input layer and output layer);

[0033] Step (5): Use a high-temperature and high-pressure fluid loss instrument to make a drilling fluid filter cake;

[0034] Step (6): Through experimental tests, obtain the relationship between the filter cake flushing time and the flushing efficiency, as Figure 2 shown, and thus determine the optimal flushing time t = 10 min;

[0035] Step (7): According to Equation (1-1), calculate that when t = 10 min, the displacement of spacer fluid V = 10.8 m 3 ;

[0036] V = 60Avt (1-1)

[0037] In the formula, V is the displacement of spacer fluid, and the unit is m 3; v is the annular return velocity, with the unit of m / s; t is the flushing time, with the unit of min; A is the annular area, with the unit of m 2 ;

[0038] Step (8): Substitute the displacement of the spacer fluid V = 10.8 m 3 into the FLUENT software, and obtain the volume fraction of each phase fluid in the annular flow field through user-defined functions, and calculate the three-phase mixing ratio P * to be 5.8:3.2:1.0;

[0039] Step (9): Substitute P* = 5.8:3.2:1.0 into the basic database D for the design of the displacement of the spacer fluid, and obtain the pollution risk factor N* = 0.68. It is determined that the displacement of the spacer fluid at this flushing time does not meet the requirements, and return to readjust the displacement of the spacer fluid, and repeat Step (8) until N* = 0.42. At this time, output the displacement of the spacer fluid V = 24.7 m 3 .

[0040] Example 2:

[0041] In this example, the liner cementing of Well HS-101 in Block M is used as the reference object. The length of the cemented section of this well is 3398 m, the annular return velocity v = 1.53 m / s, and the annular area A = 0.01178 m 2 , and the specific implementation steps are as follows:

[0042] Step (1): Determine the four major main control factors affecting the mixing law of the annular fluid in cementing: eccentricity, hole enlargement, displacement of the spacer fluid, and displacement rate;

[0043] Step (2): With the on-site construction parameters of Block M as the reference, take the above four major main control factors as independent variables, generate 100 groups of random arrays with different eccentricities, hole enlargements, displacements of the spacer fluid, and displacement rates, and substitute them into the FLUENT software to calculate the three-phase mixing ratio P corresponding to each random array;

[0044] Step (3): Quantify multiple parameters affecting mixing pollution, and obtain the pollution risk factor N (N ∈ [0, 1]) to quantitatively evaluate the pollution risk of the annular slurry mixing. The closer N is to 1, the greater the pollution risk;

[0045] Step (4): According to the calculation results of Step (2), use the BP neural network training method to complete the construction of the basic database D for the design of the displacement of the spacer fluid, which is specifically divided into the following steps:

[0046] Step (4-1): Establish the mapping function relationship (hidden layer) between the three-phase mixing ratio P and the pollution risk factor N under the influence of each main control factor;

[0047] Step (4-2): Based on the above mapping relationship, establish the GUI instruction interface (input layer and output layer);

[0048] Step (5): Fabricate the drilling fluid filter cake with the help of a high-temperature and high-pressure fluid loss instrument;

[0049] Step (6): Through experimental tests, obtain the relationship between the filter cake flushing time and the flushing efficiency, as Figure 3 shown, and thus determine the optimal flushing time t = 13 min;

[0050] Step (7): According to Equation (1-1), calculate that when t = 10 min, the displacement fluid volume V = 14.1 m 3 ;

[0051] V = 60Avt (1-1)

[0052] In the formula, V is the displacement fluid volume, with the unit of m 3 ; v is the annular velocity, with the unit of m / s; t is the flushing time, with the unit of min; A is the annular area, with the unit of m 2 ;

[0053] Step (8): Substitute the displacement fluid volume V = 8.7 m 3 into the FLUENT software, and through the user-defined function, obtain the volume proportion of each phase fluid in the annular flow field, and calculate that the three-phase mixing ratio P* is 5.8:3.0:1.2;

[0054] Step (9): Substitute P* = 5.8:3.0:1.2 into the displacement fluid volume design basic database D, and obtain the pollution risk factor N* = 0.61. Judge that the displacement fluid volume at this flushing time does not meet the requirements, return to readjust the displacement fluid volume, and repeat Step (8) until N* = 0.41. At this time, output the displacement fluid volume V = 21.5 m 3 ;

[0055] Example 3:

[0056] This example takes the liner cementing of Well YA-022 in Block M as the reference object. The cementing section of this well is 2477 m long, the annular velocity v = 1.53 m / s, and the annular area A = 0.02128 m 2 , and the specific implementation steps are as follows:

[0057] Step (1): Determine the four major main control factors affecting the mixing law of the annular fluid in cementing: eccentricity, hole enlargement, displacement fluid volume, and displacement;

[0058] Step (2): Taking the on-site construction parameters of M block as a reference, using the above four major control factors as independent variables, generate 100 sets of random arrays with different eccentricities, hole diameters, spacer fluid volumes, and displacement rates, and substitute them into the FLUENT software to calculate the corresponding three-phase mixing ratio P for each random array;

[0059] Step (3): Quantify multiple parameters affecting mixing pollution, obtain the pollution risk factor N (N ∈ [0, 1]) to quantitatively evaluate the annulus mud mixing pollution risk. The closer N is to 1, the greater the pollution risk;

[0060] Step (4): According to the calculation results of step (2), use the BP neural network training method to complete the construction of the basic database D for spacer fluid volume design, which is specifically divided into the following steps:

[0061] Step (4-1): Establish the mapping function relationship (hidden layer) between the three-phase mixing ratio P and the pollution risk factor N under the influence of each control factor;

[0062] Step (4-2): Based on the above mapping relationship, establish a GUI instruction interface (input layer and output layer);

[0063] Step (5): Use a high-temperature and high-pressure fluid loss instrument to make a drilling fluid filter cake;

[0064] Step (6): Through experimental tests, obtain the relationship between the filter cake flushing time and the flushing efficiency, as Figure 4 shown, and thus determine the optimal flushing time t = 8 min;

[0065] Step (7): According to formula (1-1), calculate that when t = 8 min, the spacer fluid volume V = 15.6 m 3 ;

[0066] V = 60Avt (1-1)

[0067] In the formula, V is the spacer fluid volume, with the unit of m 3 ; v is the annular return velocity, with the unit of m / s; t is the flushing time, with the unit of min; A is the annular area, with the unit of m 2 ;

[0068] Step (8): Substitute the spacer fluid volume V = 15.6 m 3 into the FLUENT software, obtain the volume ratio of each phase fluid in the annular flow field through user-defined functions, and calculate that the three-phase mixing ratio P* is 5.8:3.7:0.5;

[0069] Step (9): Substitute P* = 5.8:3.7:0.5 into the basic database D for the design of the spacer fluid dosage, and obtain the pollution risk factor N* = 0.71. It is judged that the spacer fluid dosage at this flushing time does not meet the requirements. Return to readjust the spacer fluid dosage and repeat Step (8) until N* = 0.43. At this time, output the spacer fluid dosage V = 30m 3 。

[0070] Example 4

[0071] In this example, the liner cementing of Well MX-019 in Block M is used as the reference object. The length of the sealed section of this well is 2556m, the annular return velocity v = 1.53m / s, and the annular area A = 0.01144m 2 , and the specific implementation steps are as follows:

[0072] Step (1): Determine the four major main control factors affecting the mixing law of the annular fluid in cementing: eccentricity, hole enlargement, spacer fluid dosage, and displacement;

[0073] Step (2): Taking the on-site construction parameters of Block M as a reference, using the above four major main control factors as independent variables, generate 100 groups of random arrays with different eccentricities, hole enlargements, spacer fluid dosages, and displacements, and calculate the corresponding three-phase mixing ratios P of each random array by bringing them into the FLUENT software;

[0074] Step (3): Quantify multiple parameters affecting mixing pollution, obtain the pollution risk factor N (N ∈ [0,1]), and quantitatively evaluate the pollution risk of the annular slurry mixing. The closer N is to 1, the greater the risk;

[0075] Step (4): According to the calculation results of Step (2), use the BP neural network training method to complete the construction of the basic database D for the design of the spacer fluid setting amount, which is specifically divided into the following steps:

[0076] Step (4-1): Establish the mapping function relationship (hidden layer) between the three-phase mixing ratio P and the pollution risk factor N under the influence of each main control factor;

[0077] Step (4-2): Based on the above mapping relationship, establish a GUI instruction interface (input layer and output layer);

[0078] Step (5): Make a drilling fluid filter cake with the help of a high-temperature and high-pressure fluid loss instrument;

[0079] Step (6): Obtain the relationship between the filter cake flushing time and the flushing efficiency through experimental tests, as Figure 5 shown, and thus determine the optimal flushing time t = 15min;

[0080] Step (7): Calculate the spacer fluid dosage V = 15.8m at t = 15min according to Equation (1-1)3 ;

[0081] V = 60Avt (1-1)

[0082] where V is the amount of spacer fluid, in m 3 ; v is the annular return velocity, in m / s; t is the flushing time, in min; A is the annular area, in m 2 ;

[0083] Step (8): Substitute the amount of spacer fluid V = 15.8 m 3 into the FLUENT software, and obtain the volume fraction of each phase fluid in the annular flow field through user-defined functions, and calculate that the three-phase mixing ratio P* is 5.8:2.5:1.7;

[0084] Step (9): Substitute P* = 5.8:2.5:1.7 into the basic database D for the design of the amount of spacer fluid, and obtain the pollution risk factor N* = 0.49. Judge that the amount of spacer fluid at this flushing time meets the requirements, so output the amount of spacer fluid V = 15.8 m 3 .

[0085] The above 4 embodiments establish a basic database with the on-site construction parameters of the M block as a reference, obtain the optimal flushing time of 4 wells on-site in this block through flushing experiments, and compare and check with the basic database. After inputting the working condition parameters in the program interface, the recommended amount of spacer fluid is shown in Table 1.

[0086] Table 1 Recommended Amount of Spacer Fluid for Each Well

[0087]

[0088] In summary, considering the influence of factors such as casing eccentricity and drilling fluid performance on the mixing of annular fluids in the 4 embodiments, that is, the greater the eccentricity, the easier the mixing, and the more spacer fluid is required; the smaller the dynamic shear force and plastic viscosity of the drilling fluid, the smaller the design of the amount of spacer fluid. Combining the different lengths of the cementing section and annular volumes of their respective cases, the amounts of spacer fluid are finally determined to be: 24.7 m 3 , 21.5 m 3 , 30 m 3 , 15.8 m 3 .

[0089] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A design method for the dosage of flushing-type spacer fluid based on BP neural network, characterized in that, it includes the following steps: Step (1). Determine the main control factors affecting the mixing law of the cementing annulus fluid; Step (2). Take the main control factors as independent variables, randomly generate M groups of data, and calculate the three-phase mixing ratio P of each random array by bringing them into the FLUENT software; Step (3). Quantify multiple parameters affecting mixing pollution to obtain a pollution risk factor N, N ∈ [0, 1], and quantitatively evaluate the annulus mud mixing pollution risk. The closer N is to 1, the greater the pollution risk; Step (4). According to the calculation results of Step (2), use the BP neural network training method to complete the construction of the basic database D for the design of the spacer fluid dosage, which is specifically divided into the following steps: Step (4-1). Establish the mapping function relationship between the three-phase mixing ratio P and the pollution risk factor N under the influence of each main control factor; Step (4-2). Based on the above mapping relationship, establish a GUI instruction interface; Step (5). Make a drilling fluid filter cake with a high-temperature and high-pressure fluid loss instrument; Step (6). Obtain the relationship between the filter cake flushing time and the flushing efficiency through experimental tests, and determine the optimal flushing time t on the premise of ensuring the flushing efficiency; Step (7). Calculate the dosage of the spacer fluid at the optimal flushing time t, and the calculation formula is V = 60Avt (1-1) Where V is the amount of spacer fluid, with the unit of m 3 ; v is the annular return velocity, with the unit of m / s; t is the flushing time, and the unit is min; A is the annulus area, in m² 2 ; Step (8). Bring the spacer fluid dosage V calculated in Step (7) into the FLUENT software to calculate the three-phase mixing ratio P*; Step (9). Substitute the three-phase blending ratio P* calculated in step (8) into the P-N mapping function relationship established in step (4-1) to obtain the pollution risk factor N*. Determine N*. If N * ≤ 0.5, the pollution risk is controllable, and the displacement fluid volume V meets the requirements, then output; if N* > 0.5, the pollution risk is relatively high, and the displacement fluid volume V does not meet the requirements. It is necessary to return and adjust the displacement fluid volume until N* ≤ 0.

5.

2. The design method for the dosage of flushing-type spacer fluid based on BP neural network according to Claim 1, characterized in that , the main control factors in Step (1) include: eccentricity, hole enlargement, spacer fluid dosage, displacement, and fluid properties.

3. The design method for the dosage of flushing-type spacer fluid based on BP neural network according to Claim 1, characterized in that, the three phases in Step (4-1) refer to cement slurry, drilling fluid, and spacer fluid.

Images