A risk prediction method and system for simulating oil drilling

Abstract

The present invention is applicable to the technical field of drilling safety prevention and control, and provides a risk prediction method and system for simulating oil drilling, the method comprising the following steps: real-time prediction of the formation pore pressure at a future time t; real-time prediction of the formation fluid concentration flowing into the drilling fluid; real-time correction of the initial density of the drilling fluid according to the formation fluid concentration flowing into the drilling fluid and the concentration of the annular cuttings to obtain the corrected density of the drilling fluid; real-time prediction of the corrected value of the bottom hole pressure at a future time t according to the functional relationship between the future drilling depth and time, the corrected density of the drilling fluid and the predicted value of the annular pressure loss; real-time prediction of the risk of overflow or blowout during the drilling process according to the difference between the corrected value of the bottom hole pressure at a future time t and the formation pore pressure. The present invention can more accurately predict the bottom hole pressure at a certain time in the future, so that the risk of well control accidents such as overflow blowout can be predicted in real time and accurately.

Description

Technical Field

[0001] The present invention belongs to the technical field of drilling safety prevention and control, and in particular relates to a risk prediction method and system for simulating oil drilling. Background Art

[0002] With the continuous deepening of oil and gas exploration and development, drilling operations are facing increasingly complex geological environments and safety risks. Once an oil and gas well accident occurs, it is often accompanied by serious casualties, property losses and environmental pollution. Therefore, it is particularly important to prevent the occurrence of well control accidents such as overflow and blowout.

[0003] Among them, overflow refers to the phenomenon that formation fluids such as oil, gas, and water invade the wellbore, causing the drilling fluid in the wellbore to overflow; blowout refers to the phenomenon that formation fluids flow into the wellbore uncontrollably and spray out of the ground. At present, the density of drilling fluid is generally adjusted to prevent the occurrence of well control accidents such as overflow blowout; however, due to the complexity of the geological environment, in the actual drilling process, as the drilling depth continues to deepen, the risk of well control accidents such as overflow blowout will become more and more difficult to control and predict. Therefore, it is necessary to predict and monitor the risk of well control accidents such as overflow blowout in real time to improve safety in the actual oil drilling process. Summary of the invention

[0004] The purpose of the embodiment of the present invention is to provide a risk prediction method for simulating oil drilling, aiming to solve the above technical problems.

[0005] The embodiment of the present invention is implemented as follows: a risk prediction method for simulating oil drilling, which includes the following steps:

[0006] ‌Explore the internal structure of the formation to be simulated and predicted, obtain the distribution of fluid layers at different formation depths, and obtain the formation pore pressure at different formation depths in real time, and determine the functional relationship between formation pore pressure and formation depth;

[0007] Based on the drilling speed model, the current drilling stroke speed is determined in real time according to the current drilling depth:

[0008] According to the current drilling speed, the functional relationship between future drilling depth and time is predicted;

[0009] According to the functional relationship between future drilling depth and time, and the functional relationship between formation pore pressure and formation depth, the formation pore pressure at the future time t is predicted in real time;

[0010] Based on the functional relationship between future drilling depth and time, and the distribution of fluid layers at different formation depths, the concentration of formation fluid flowing into the drilling fluid is predicted in real time;

[0011] Obtain the drilling fluid annular return velocity and cuttings settling velocity, and calculate the annular cuttings concentration according to the current drilling stroke speed; according to the formation fluid concentration flowing into the drilling fluid and the annular cuttings concentration, correct the initial density of the drilling fluid in real time to obtain the corrected density of the drilling fluid;

[0012] Based on the annular pressure loss calculation model, the annular pressure loss prediction value at the future time t is obtained in real time according to the drilling fluid annular return velocity and the modified density of the drilling fluid;

[0013] According to the functional relationship between the future drilling depth and time, the corrected density of the drilling fluid and the predicted value of the annular pressure loss, the corrected value of the bottom hole pressure at the future time t is predicted in real time;

[0014] Based on the difference between the bottom hole pressure correction value and the formation pore pressure at the future time t, the risk of overflow or blowout during drilling is predicted in real time.

[0015] Further, the drilling speed model is:

[0016] ;

[0017] In the formula, V pc is the current drilling stroke speed; K R is the proposed constant, which is related to the drillability of the rock and the efficiency of the drill bit; W is the drilling pressure, i.e. the force applied to the drill bit, in kN; M is the reference value of drilling pressure, in kN; n is the speed of the drill bit, in rpm; The proposed correction index is related to the wear of the drill bit and the crushing characteristics of the rock; C 2 is the proposed correction factor, which is related to the characteristics of the drill bit; H 0 is the current drilling depth, in m; C p is the proposed drilling weight coefficient, which is related to the relationship between drilling weight and drilling speed; C H is the proposed drill bit coefficient, which is related to the type of drill bit.

[0018] Furthermore, the functional relationship between the future drilling depth and time is:

[0019] ;

[0020] In the formula, t is time, unit is h; H ( t ) is the future drilling depth, in m.

[0021] Further, according to the functional relationship between the future drilling depth and time, and the distribution of fluid layers at different formation depths, the step of real-time prediction of the formation fluid concentration flowing into the drilling fluid specifically includes:

[0022] When the future drilling depth is not greater than the minimum depth of drilling into the fluid layer, the formation fluid concentration is 0;

[0023] When the future drilling depth is greater than the minimum depth of drilling the fluid layer, the concentration of the formation fluid flowing into the drilling fluid is predicted in real time according to the preset hybrid temporal attention network based on physical constraints;

[0024] The hybrid temporal attention network based on physical constraints includes a velocity-concentration coupling physical constraint model; the velocity-concentration coupling physical constraint model is:

[0025] ;

[0026] ;

[0027] In the formula, C b ( t ) is the concentration of formation fluid flowing into the drilling fluid at the future time t; H c is the static liquid column height of the drilling fluid currently containing cuttings; d is the inner diameter of the wellbore; d 1 is the average radius of formation pores; V r ( t ) is the average flow velocity of the invading fluid at the future time t; ρ 0 is the initial density of drilling fluid; ρ b is the average density of the formation fluid; g is the acceleration due to gravity; P p ( t ) is the formation pore pressure at time t in the future; P b is the pump pressure; B is the flow potential energy coefficient; k 1 is the pressure influence coefficient; k 2 is the drilling fluid influence coefficient.

[0028] Further, the calculation formula of the modified density of the drilling fluid is:

[0029] ;

[0030] In the formula, ρ e ( t ) is the modified density of the drilling fluid; ρ a The rock cuttings density ；C a is the concentration of cuttings in the annulus.

[0031] Furthermore, the formation fluid includes one or more of oil, natural gas and water; the method for predicting the average density of the formation fluid specifically includes the following steps:

[0032] Determine the weight of each fluid based on the future drilling depth and the distribution of fluid layers at different formation depths;

[0033] According to the weights of the various fluids, the weighted average of the fluid densities is calculated to obtain the average density of the formation fluid.

[0034] Furthermore, the prediction formula of the bottom hole pressure correction value at the future time t is:

[0035] ;

[0036] In the formula, P W ( t ) is the bottom hole pressure correction value at the future time t; g is the acceleration due to gravity; is the predicted value of annular pressure loss at time t in the future.

[0037] Furthermore, according to the difference between the bottom hole pressure correction value and the formation pore pressure at the future time t, the steps of predicting the risk of overflow or blowout in the drilling process in real time specifically include:

[0038] The corrected pressure difference is calculated based on the difference between the bottom hole pressure correction value and the formation pore pressure at the future time t;

[0039] Calculate the mean and standard deviation of the corrected pressure difference within a certain period of time in the future;

[0040] Based on the cumulative distribution function, the risk probability of overflow or blowout in the future drilling process is calculated according to the mean and standard deviation of the corrected pressure difference.

[0041] Furthermore, the calculation formula for the risk probability of overflow or blowout in the future drilling process is:

[0042] ;

[0043] In the formula, F The risk probability of overflow or blowout during future drilling; X To correct the pressure difference, X=P W ( t ) -Pp ( t ); μ X is the mean value of the corrected pressure difference; σ X is the standard deviation of the corrected pressure difference.

[0044] Another object of the present invention is to provide a risk prediction system for simulated oil drilling, which is used to implement the above-mentioned risk prediction method for simulated oil drilling, and comprises:

[0045] The formation exploration unit is used to explore the internal structure of the formation to be simulated and predicted, obtain the distribution of fluid layers at different formation depths, and obtain the formation pore pressure at different formation depths in real time, and determine the functional relationship between the formation pore pressure and the formation depth;

[0046] The drilling speed determination unit is used to determine the current drilling stroke speed in real time according to the current drilling depth based on the drilling speed model:

[0047] A well depth prediction unit is used to predict the functional relationship between future drilling depth and time based on the current drilling travel speed;

[0048] The formation pore pressure prediction unit is used to predict the formation pore pressure at the future time t in real time according to the functional relationship between the future drilling depth and time, and the functional relationship between the formation pore pressure and the formation depth;

[0049] A formation fluid concentration prediction unit is used to predict the formation fluid concentration flowing into the drilling fluid in real time based on the functional relationship between the future drilling depth and time and the distribution of fluid layers at different formation depths;

[0050] The drilling fluid density correction unit is used to obtain the drilling fluid annular return velocity and the cuttings settling velocity, and calculate the annular cuttings concentration according to the current drilling stroke speed; the initial density of the drilling fluid is corrected in real time according to the formation fluid concentration flowing into the drilling fluid and the annular cuttings concentration to obtain the corrected density of the drilling fluid;

[0051] The annular pressure loss prediction unit is used to predict the annular pressure loss prediction value at the future time t in real time based on the annular pressure loss calculation model and the annular return velocity of the drilling fluid and the corrected density of the drilling fluid;

[0052] The bottom hole pressure prediction unit is used to predict the bottom hole pressure correction value at the future time t in real time according to the functional relationship between the future drilling depth and time, the corrected density of the drilling fluid and the predicted value of the annular pressure loss;

[0053] The risk prediction unit is used to predict the risk of overflow or blowout in the drilling process in real time based on the difference between the bottom hole pressure correction value and the formation pore pressure at the future time t.

[0054] A risk prediction method for simulating oil drilling provided by an embodiment of the present invention can predict the concentration of formation fluid flowing into the drilling fluid in real time, and combined with the concentration of annular cuttings, can make real-time corrections to the initial density of the drilling fluid, and predict the true density of the drilling fluid during the actual drilling process, so as to more accurately predict the bottom hole pressure at a certain moment in the future, thereby being able to predict the risk of well control accidents such as overflow blowouts in real time and accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 A schematic flow chart of a risk prediction method for simulating oil drilling provided in an embodiment of the present invention.

[0056] Figure 2 A flowchart of a method for constructing a hybrid temporal attention network based on physical constraints provided in an embodiment of the present invention.

[0057] Figure 3 A flowchart of a training method for a hybrid temporal attention network based on physical constraints provided in an embodiment of the present invention.

[0058] Figure 4 A schematic flow chart of a method for predicting the average density of formation fluid provided in an embodiment of the present invention.

[0059] Figure 5 A schematic diagram of a flow chart for real-time prediction of the risk of overflow or blowout during drilling provided by an embodiment of the present invention.

[0060] Figure 6 A schematic diagram of a risk prediction system for simulating oil drilling provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0062] As attached Figure 1 As shown, in one embodiment of the present invention, a risk prediction method for simulating oil drilling is provided, characterized in that it includes the following steps:

[0063] S100, exploring the internal structure of the formation to be simulated and predicted, obtaining the distribution of fluid layers at different formation depths, and obtaining the formation pore pressure at different formation depths in real time, and determining the functional relationship between the formation pore pressure and the formation depth;

[0064] S200, based on the drilling speed model, determine the current drilling stroke speed in real time according to the current drilling depth:

[0065] S300, predicting a functional relationship between future drilling depth and time according to the current drilling travel speed;

[0066] S400, based on the functional relationship between the future drilling depth and time, and the functional relationship between the formation pore pressure and the formation depth, real-time prediction is performed to obtain the formation pore pressure at the future time t;

[0067] S500, predicting the concentration of formation fluid flowing into the drilling fluid in real time according to the functional relationship between the future drilling depth and time and the distribution of fluid layers at different formation depths;

[0068] S600, obtaining the drilling fluid annular return velocity and the cuttings settling velocity, and calculating the annular cuttings concentration according to the current drilling stroke speed; correcting the initial density of the drilling fluid in real time according to the formation fluid concentration flowing into the drilling fluid and the annular cuttings concentration to obtain the corrected density of the drilling fluid;

[0069] S700, based on the annular pressure loss calculation model, according to the annular return velocity of the drilling fluid and the corrected density of the drilling fluid, the annular pressure loss prediction value at the future time t is predicted in real time;

[0070] S800, based on the functional relationship between the future drilling depth and time, the corrected density of the drilling fluid and the predicted value of the annular pressure loss, real-time prediction is performed to obtain a corrected value of the bottom hole pressure at the future time t;

[0071] S900, predicting the risk of overflow or blowout during drilling in real time based on the difference between the bottom hole pressure correction value and the formation pore pressure at the future time t.

[0072] In step S100, gravity exploration, magnetic exploration, or seismic exploration can be used to explore the internal structure of the stratum to be simulated and predicted. Different stratum depths correspond to different rock layers and fluid layers; wherein the fluid layer refers to a stratum containing fluids such as oil, natural gas, and water. By obtaining the distribution of fluid layers at different stratum depths, the embodiment of the present invention can predict whether the fluid layer has been drilled according to the future drilling depth. It should be noted that the future drilling depth refers to the well depth after drilling at a certain time in the future.

[0073] In addition, the formation pore pressure at different formation depths can be obtained in real time through existing formation testing tools such as MWD or LWD equipment. By determining the functional relationship between the formation pore pressure of the simulated predicted formation and the formation depth, the formation pore pressure corresponding to the well depth can be predicted according to the future drilling depth.

[0074] In a preferred embodiment of the present invention, the current drilling stroke speed is the mechanical drilling rate (ROP), which is defined as the distance or depth that the drill bit drills in the rock per unit time. The current drilling stroke speed can be calculated by a drilling speed model; the drilling speed model is an estimate of the current drilling stroke speed based on the drilling pressure, the rotation speed of the drill bit and other influencing factors; specifically, the drilling speed model is:

[0075] ;

[0076] In the formula, V pc is the current drilling stroke speed; K R The proposed constant is related to the drillability of the rock and the efficiency of the drill bit and is formulated based on actual conditions; W is the drilling pressure, i.e. the force applied to the drill bit, in kN; M is the reference value of drilling pressure, in kN; n is the speed of the drill bit, in rpm; The proposed correction index is related to the wear of the drill bit and the crushing characteristics of the rock and is formulated according to the actual situation; C 2 is the proposed correction coefficient, which is related to the characteristics of the drill bit and is formulated according to the actual situation; H 0 is the current drilling depth, in m; C p The proposed drilling pressure coefficient is related to the relationship between drilling pressure and drilling speed and is formulated according to actual conditions; C H The proposed drill bit coefficient is related to the type of drill bit and is formulated based on actual conditions.

[0077] In a preferred embodiment of the present invention, the functional relationship between the future drilling depth and time is:

[0078] ;

[0079] In the formula, t is time, unit is h; H ( t ) is the future drilling depth, in m.

[0080] The present invention can predict the well depth after a certain period of drilling in the future by calculating the current drilling stroke speed in real time, thereby facilitating subsequent prediction work.

[0081] In a preferred embodiment of the present invention, in step S400, the future drilling depth obtained by the above method is H ( t) and time, the formation pore pressure at the future time t can be predicted in real time according to the functional relationship between the formation pore pressure and the formation depth determined in step S100. P p ( t ).

[0082] In a preferred embodiment of the present invention, the step of predicting the concentration of the formation fluid flowing into the drilling fluid in real time according to the functional relationship between the future drilling depth and time and the distribution of the fluid layers at different formation depths, i.e., step S500, specifically includes:

[0083] When the future drilling depth is not greater than the minimum depth of drilling into the fluid layer, the formation fluid concentration is 0;

[0084] When the future drilling depth is greater than the minimum depth of drilling the fluid layer, the concentration of the formation fluid flowing into the drilling fluid is predicted in real time according to the preset hybrid temporal attention network based on physical constraints;

[0085] The hybrid temporal attention network based on physical constraints includes a velocity-concentration coupling physical constraint model; the velocity-concentration coupling physical constraint model is:

[0086] ;

[0087] ;

[0088] In the formula, C b ( t ) is the concentration of formation fluid flowing into the drilling fluid at the future time t; H c is the static liquid column height of the drilling fluid currently containing cuttings; d is the inner diameter of the wellbore; d 1 is the average radius of formation pores; V r ( t ) is the average flow velocity of the invading fluid at the future time t; ρ 0 is the initial density of drilling fluid; ρ b is the average density of the formation fluid; g is the acceleration due to gravity; P p ( t ) is the formation pore pressure at time t in the future; P b is the pump pressure; B , k 1 and k 2 is the key parameter of fluid dynamics; specifically, BIt is the flow potential energy coefficient, which reflects the baseline kinetic energy level of fluid motion, affects the basic size of flow velocity, and is related to the characteristics of the fluid itself; k 1 is the pressure influence coefficient, which is used to adjust P p ( t ) on flow velocity, reflecting the sensitivity of flow velocity changes caused by pressure changes, which is related to the relationship between formation pore pressure and drilling fluid pressure; k 2 is the drilling fluid influence coefficient, which is used to adjust the influence of drilling fluid on flow rate and reflects the influence of the static liquid column pressure gradient of drilling fluid on flow rate.

[0089] In addition, the physical constraint-based hybrid temporal attention network also includes a multi-head attention-driven heterogeneous integrated encoding model; B , k 1 and k 2 can be accurately estimated by a heterogeneous ensemble encoding model driven by multi-head attention.

[0090] It should be noted that, in general, fluid layers such as oil, natural gas and water are located at a deeper position in the formation. When drilling begins, the well depth is not enough to reach the fluid layer. At this time, since the fluid layer has not been drilled, no formation fluid flows into the drilling fluid, so the formation fluid concentration is 0. Of course, in this case, there is generally no risk of overflow blowout. As the drilling depth continues to increase, when the drilling depth is further increased in the future, the risk of overflow blowout will increase. H ( t ) When the depth of drilling through the fluid layer is reached, the fluid layer will be drilled through, and it is necessary to consider the situation of fluids such as oil, natural gas and water entering the drilling fluid. Since the density of fluids such as oil, natural gas and water is smaller than the density of drilling fluid, when fluids such as oil, natural gas and water are mixed with drilling fluid, the density will be reduced, thereby reducing the bottom hole pressure, which will easily lead to the risk of overflow blowout.

[0091] In a preferred embodiment of the present invention, the prediction method of the hybrid temporal attention network based on physical constraints comprises the following steps:

[0092] S510, predicting the fluid properties of the fluid layer drilled at time t in the future according to the distribution of fluid layers at different formation depths; the fluid properties include the average density and average viscosity of the formation fluid;

[0093] S520, through the heterogeneous integrated coding model driven by multi-head attention, comprehensively analyze the drilling condition parameters (such as drilling speed, well depth, initial density of drilling fluid, etc.) and formation characteristic parameters (such as formation pore pressure, formation fluid properties, temperature, etc.), and realize the key parameters of fluid dynamics B , k 1 and k An accurate estimate of 2;

[0094] S530, key fluid dynamics parameters to be obtained B , k 1 and k 2 is substituted into the velocity-concentration coupling physical constraint model and the formation fluid concentration flowing into the drilling fluid can be predicted in real time through double equation calculation.

[0095] It should be noted that the heterogeneous integrated coding model driven by multi-head attention is a neural network model driven by a multi-head attention mechanism; the above-mentioned heterogeneous integrated coding model driven by multi-head attention can be constructed using LSTM and Transformer encoders. The model uses a multi-head attention layer to calculate the attention weights between the encoder outputs of the bidirectional LSTM and adaptive attention and the outputs of the hierarchical Transformer temporal encoder. Subsequently, the calculated high-dimensional attention output is further converted into a low-dimensional three-value output through a fully connected layer. These three values ​​represent the flow potential coefficients, respectively. B , Pressure influence coefficient k 1 and drilling fluid influence coefficient k 2. In addition, Figure 2 and Figure 3 As shown, the above-mentioned construction and training method of the hybrid temporal attention network based on physical constraints specifically includes the following steps:

[0096] S521. Obtain data such as drilling stroke speed, drilling depth, properties of drilling fluid (including density and viscosity), fluid properties of fluid layer, formation pore pressure, pump pressure, borehole inner diameter, average radius of formation pores, and concentration of formation fluid in corresponding invading drilling fluid at a certain moment in history, and divide the data set; among them, formation pore pressure can be calculated using logging data (such as acoustic wave time difference) and empirical formulas (such as Eaton formula); average radius of formation pores: can be estimated through core analysis or logging data (such as acoustic wave time difference); drilling stroke speed and pump pressure can be obtained through real-time monitoring during drilling and direct measurement by sensors.

[0097] S522, preprocess the above historical data; specifically, first use Savitzky-Golay filter and other methods to smooth the original data. The data obtained during the drilling process is usually affected by various noises, such as sensor noise, environmental factors, etc. The use of Savitzky-Golay filter can effectively remove these noises, making the data smoother, thereby improving the quality of subsequent feature extraction and model training; then, extract features from the smoothed data. The concentration change of formation fluid in the drilling fluid is usually continuous and time-related, so these local statistical characteristics can be captured by calculating the mean, standard deviation, maximum and minimum values ​​in the rolling window, which helps the model to better understand the time dependence of the data, and calculate the pressure gradient to capture the trend of concentration changes; then, merge the features, and perform feature standardization, perform a standardized robust scale (RobustScaler) on the features, scale the features to a fixed range, so that they have similar scales, so that the model will not be biased towards certain features due to different feature scales during training, thereby improving the robustness and convergence speed of the model; finally, return the standardized features for subsequent model training.

[0098] S523, construct the time series features of the preprocessed data, then perform rotational position encoding, and combine the encoder based on bidirectional LSTM and adaptive attention and the time series encoder based on layered Transformer to build a multi-head attention-driven heterogeneous integrated encoding model. Among them, LSTM focuses on local time series information, while Transformer encoder is good at processing global dependencies. The combination of the two can capture sequence information more comprehensively; then perform denormalization to obtain the key parameters of fluid dynamics B , k 1 and k 2. Substituting into the velocity-concentration coupling physical constraint model, the predicted output (the predicted value of formation fluid concentration in the drilling fluid) can be obtained.

[0099] S524, initialize model parameters, set training hyperparameters, optimizer AdamW, loss function Huber Loss evaluation index Accuracy and other tools needed in the subsequent retraining process. Among them, the AdamW optimizer is an adaptive learning rate method, which can automatically adjust the learning rate according to the importance of the parameters, which is particularly useful for drilling data with complex characteristics. The concentration of drilling fluid is affected by many factors. By using AdamW, these complex nonlinear relationships can be effectively handled. In addition, the Huber Loss loss function used in the embodiment of the present invention has high applicability in the prediction of formation fluid concentration in drilling fluid due to its robustness, balanced error, adaptability and efficiency. It can help the model better deal with the complexity and uncertainty in practical applications, thereby improving the accuracy and reliability of the prediction.

[0100] S525, model training: load batch data, forward propagation, calculate loss function, back propagation, then optimizer update parameters and use Cosine Annealing with Warm Restarts learning rate scheduling method to dynamically adjust learning rate, then batch verification and evaluation, determine whether early stopping is satisfied, if early stopping is satisfied, save the best model; if early stopping is not satisfied, then determine whether the maximum number of iterations is reached; if the maximum number of iterations is reached, save the best model; if the maximum number of iterations is not reached, return to continue training; finally, verify and evaluate the saved best model. In the prediction of formation fluid concentration in drilling fluid, data may have seasonal or periodic changes. Cosine Annealing with Warm Restarts can dynamically adjust the learning rate during training to help the model better capture these change patterns. By periodically restarting the learning rate, this scheduling strategy can help the model escape from the local minimum, which is particularly important for nonlinear and complex data distributions; and in the drilling process, many factors are nonlinear, so this strategy helps to improve the generalization ability of the model.

[0101] In the embodiment of the present invention, the first constraint of the velocity-concentration coupling physical constraint model is the fluid dynamics equation, which describes the motion characteristics of the invading fluid and outputs the fluid velocity. V r ( t ); The second constraint is the concentration conversion equation, which converts the flow rate information into concentration changes and calculates the final formation fluid concentration in the drilling fluid C b ( t). The embodiment of the present invention adds mathematical models to neural networks for deep learning, constructs a hybrid temporal attention network based on physical constraints, simplifies the learning objectives of the neural network (only three key parameters need to be learned), reduces the complexity of directly predicting concentration values, and improves the convergence speed and generalization ability of the model; specifically, the multi-head attention-driven heterogeneous integrated coding model outputs the optimal kinetic parameters by learning historical data B , k 1 and k 2, then by receiving the three kinetic parameters output B , k 1 and k 2. Substitute the velocity-concentration coupling physical constraint model and calculate the predicted value through physical constraints to ensure that the prediction result conforms to the laws of fluid mechanics and is more in line with reality, so as to achieve real-time and accurate prediction of the changes in formation fluid concentration in the drilling fluid and provide a reliable basis for overflow risk warning. In addition, during the model training process, the loss function is used to perform back propagation based on the difference between the predicted value and the actual value. The whole process not only ensures physical rationality, but also has the adaptive ability of machine learning.

[0102] In a preferred embodiment of the present invention, the calculation formula of the corrected density of the drilling fluid is:

[0103] ;

[0104] In the formula, ρ e ( t ) is the modified density of the drilling fluid; ρ a The rock cuttings density ；C a is the concentration of cuttings in the annulus.

[0105] In practical applications, the concentration of cuttings in the annulus C a The calculation formula is as follows:

[0106] ;

[0107] In the formula, v c is the drilling fluid annular return velocity, in m / s; v s is the rock cuttings settling velocity, in m / s; V pc The unit of current drilling stroke speed is m / h; d is the borehole inner diameter, in cm; d 0 is the outer diameter of the drill pipe, in cm.

[0108] like Figure 4As shown, in a preferred embodiment of the present invention, the formation fluid includes one or more of oil, natural gas and water; the method for predicting the average density of the formation fluid specifically includes the following steps:

[0109] S601, determining the weight of each fluid according to the future drilling depth and the distribution of fluid layers at different formation depths;

[0110] S602. Calculate the weighted average of the fluid density according to the weight of each fluid to obtain the average density of the formation fluid.

[0111] In practical applications, the distribution of fluid layers at different formation depths obtained in step S100 can be used to determine the distribution of the fluid layers drilled after drilling at time t in the future. Then, the weight of each fluid flowing into the drilling fluid can be determined based on the proportion of each fluid layer, and the average density of the formation fluid flowing into the drilling fluid can be obtained by the weighted average method.

[0112] In a preferred embodiment of the present invention, the prediction formula of the bottom hole pressure correction value at the future time t is:

[0113] ;

[0114] In the formula, P W ( t ) is the bottom hole pressure correction value at the future time t; g is the acceleration due to gravity; is the predicted value of annular pressure loss at time t in the future.

[0115] In practical applications, the annular pressure loss prediction value The annular pressure loss calculation model can be used for calculation. The working principle of the annular pressure loss calculation model is to comprehensively consider multiple key factors affecting pressure loss to predict the annular pressure loss at a specific time point. The model first incorporates the return velocity of the drilling fluid in the annular space. v c , recognizing that an increase in return velocity will directly lead to an increase in pressure loss; secondly, the model takes into account the corrected density of the drilling fluid ρ e ( t ), because changes in density will change the flow characteristics of the fluid, thus affecting the pressure loss. In addition, the model also takes into account the borehole diameter d and drill pipe outer diameter d 0 and other annular geometric dimensions, these dimensions determine the cross-sectional area of ​​the annulus, which in turn affects the resistance to fluid flow. Specifically, the annular pressure loss calculation model is:

[0116] ;

[0117] In the formula, f The friction coefficient reflects the friction between the fluid and the annulus wall during the flow process. It is usually determined based on experimental data or empirical formulas and is specifically formulated according to actual conditions. l is the drill rod length, in m; it should be noted that, if the parameters appearing in the above formula are not limited in the embodiments of the present invention or obtained by the formula, they can all be considered as known parameters known in the prior art, and can be obtained by actual measurement or formulation according to empirical formulas, and will not be elaborated here.

[0118] like Figure 5 As shown, in a preferred embodiment of the present invention, the step of predicting the risk of overflow or blowout in the drilling process in real time according to the difference between the bottom hole pressure correction value and the formation pore pressure at the future time t, that is, step S900 specifically includes:

[0119] S901, based on the bottom hole pressure correction value at the future time t P W ( t ) and formation pore pressure P p ( t ) to calculate the corrected pressure difference;

[0120] S902, calculating the mean and standard deviation of the corrected pressure difference within a certain period of time in the future;

[0121] S903. Based on the cumulative distribution function, according to the mean and standard deviation of the corrected pressure difference, the risk probability of overflow or blowout occurring in the future drilling process is calculated.

[0122] In a preferred embodiment of the present invention, the calculation formula for the risk probability of overflow or blowout occurring during the future drilling process is:

[0123] ;

[0124] In the formula, F The risk probability of overflow or blowout during future drilling; X To correct the pressure difference, X=P W ( t ) -P p ( t ); μ X is the mean value of the corrected pressure difference; σ X is the standard deviation of the corrected pressure difference.

[0125] In practical applications, the distribution of the corrected differential pressure in a certain period of time in the future can be regarded as a normal distribution. Therefore, the integral of the probability density function (i.e., the cumulative distribution function) can be used to predict the risk probability of overflow or blowout during future drilling. It should be noted that the corrected differential pressure is the bottom hole pressure correction value. P W ( t ) and formation pore pressure P p ( t ), when the bottom hole pressure correction value P W ( t ) is less than the formation pore pressure P p ( t ), that is, when X When it is less than 0, overflow or even blowout accidents are more likely to occur, and the risk is greater; on the contrary, when X When it is greater than or equal to 0, overflow or blowout accidents are unlikely to occur, and the risk is relatively small; therefore, it can be calculated X The integral of the probability density function in the interval (0, +∞) is used to predict the safety probability that no overflow or blowout accidents will occur during future drilling. R On the contrary, it can predict the risk probability of accidents such as overflow or blowout in the future drilling process. F=1-R .

[0126] In summary, the risk prediction method for simulating oil drilling provided by the embodiment of the present invention can predict the concentration of formation fluid flowing into the drilling fluid in real time, and combined with the annulus cuttings concentration, the initial density of the drilling fluid can be corrected in real time, and the true density of the drilling fluid in the actual drilling process can be predicted, so as to more accurately predict the bottom hole pressure at a certain moment in the future, thereby predicting the risk of well control accidents such as overflow blowouts in real time and accurately.

[0127] like Figure 6 As shown, in another embodiment of the present invention, a risk prediction system for simulating oil drilling is also provided, which is used to implement the risk prediction method for simulating oil drilling, and includes:

[0128] The formation exploration unit 10 is used to explore the internal structure of the formation to be simulated and predicted, obtain the distribution of fluid layers at different formation depths, and obtain the formation pore pressure at different formation depths in real time, and determine the functional relationship between the formation pore pressure and the formation depth;

[0129] The drilling speed determination unit 20 is used to determine the current drilling stroke speed in real time according to the current drilling depth based on the drilling speed model:

[0130] The well depth prediction unit 30 is used to predict the functional relationship between the future drilling depth and time according to the current drilling stroke speed;

[0131] The formation pore pressure prediction unit 40 is used to predict the formation pore pressure at a future time t in real time according to the functional relationship between the future drilling depth and time, and the functional relationship between the formation pore pressure and the formation depth;

[0132] The formation fluid concentration prediction unit 50 is used to predict the formation fluid concentration flowing into the drilling fluid in real time according to the functional relationship between the future drilling depth and time and the distribution of fluid layers at different formation depths;

[0133] The drilling fluid density correction unit 60 is used to obtain the drilling fluid annular return velocity and the cuttings settling velocity, and calculate the annular cuttings concentration according to the current drilling stroke speed; the initial density of the drilling fluid is corrected in real time according to the formation fluid concentration flowing into the drilling fluid and the annular cuttings concentration to obtain the corrected density of the drilling fluid;

[0134] The annular pressure loss prediction unit 70 is used to predict the annular pressure loss prediction value at the future time t in real time based on the annular pressure loss calculation model and the drilling fluid annular return velocity and the modified density of the drilling fluid;

[0135] The bottom hole pressure prediction unit 80 is used to predict the bottom hole pressure correction value at the future time t in real time according to the functional relationship between the future drilling depth and time, the corrected density of the drilling fluid and the predicted value of the annular pressure loss;

[0136] The risk prediction unit 90 is used to predict the risk of overflow or blowout in the drilling process in real time according to the difference between the bottom hole pressure correction value and the formation pore pressure at the future time t.

[0137] The risk prediction system for simulating oil drilling provided by the embodiment of the present invention can predict the concentration of formation fluid flowing into the drilling fluid in real time, and combined with the concentration of annular cuttings, it can make real-time corrections to the initial density of the drilling fluid, and predict the true density of the drilling fluid during the actual drilling process, so as to more accurately predict the bottom hole pressure at a certain moment in the future, thereby being able to predict the risk of well control accidents such as overflow blowouts in real time and accurately.

[0138] It should be noted that the above-mentioned units can be implemented in the form of a computer program, which can be run on a computer device. The computer program constituting each module can be stored in the memory of the computer device so that the processor executes each step of the above-mentioned risk prediction method for simulating oil drilling.

[0139] It should be understood that, although each step in the flow chart of each embodiment of the present invention is shown in sequence according to the indication of the arrow, these steps are not necessarily performed in sequence according to the order indicated by the arrow. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0140] A person skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory.

[0141] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A risk prediction method for simulating oil drilling, characterized in that: The following steps are involved: ‌Explore the internal structure of the formation to be simulated and predicted, obtain the distribution of fluid layers at different formation depths, and obtain the formation pore pressure at different formation depths in real time, and determine the functional relationship between formation pore pressure and formation depth; Based on the drilling speed model, the current drilling stroke speed is determined in real time according to the current drilling depth: According to the current drilling speed, the functional relationship between future drilling depth and time is predicted; According to the functional relationship between future drilling depth and time, and the functional relationship between formation pore pressure and formation depth, the formation pore pressure at the future time t is predicted in real time; Based on the functional relationship between future drilling depth and time, and the distribution of fluid layers at different formation depths, the concentration of formation fluid flowing into the drilling fluid is predicted in real time; Obtain the drilling fluid annular return velocity and cuttings settling velocity, and calculate the annular cuttings concentration according to the current drilling stroke speed; according to the formation fluid concentration flowing into the drilling fluid and the annular cuttings concentration, correct the initial density of the drilling fluid in real time to obtain the corrected density of the drilling fluid; Based on the annular pressure loss calculation model, the annular pressure loss prediction value at the future time t is obtained in real time according to the drilling fluid annular return velocity and the modified density of the drilling fluid; According to the functional relationship between the future drilling depth and time, the corrected density of the drilling fluid and the predicted value of the annular pressure loss, the corrected value of the bottom hole pressure at the future time t is predicted in real time; Based on the difference between the bottom hole pressure correction value and the formation pore pressure at the future time t, the risk of overflow or blowout during drilling is predicted in real time.

2. The risk prediction method for simulating oil drilling according to claim 1, characterized in that: The drilling speed model is: ； In the formula, V pc is the current drilling stroke speed; K R is the proposed constant, which is related to the drillability of the rock and the efficiency of the drill bit; W is the drilling pressure, i.e. the force applied to the drill bit, in kN; M is the reference value of drilling pressure, in kN; n is the speed of the drill bit, in rpm; The proposed correction index is related to the wear of the drill bit and the crushing characteristics of the rock; C 2 is the proposed correction factor, which is related to the characteristics of the drill bit; H 0 is the current drilling depth, in m; C p is the proposed drilling weight coefficient, which is related to the relationship between drilling weight and drilling speed; C H is the proposed drill bit coefficient, which is related to the type of drill bit.

3. A risk prediction method for simulating oil drilling according to claim 2, characterized in that: The functional relationship between the future drilling depth and time is: ； In the formula, t is time, unit is h; H ( t ) is the future drilling depth, in m.

4. A risk prediction method for simulating oil drilling according to claim 3, characterized in that: The steps of predicting the concentration of formation fluid flowing into the drilling fluid in real time according to the functional relationship between the future drilling depth and time and the distribution of fluid layers at different formation depths include: When the future drilling depth is not greater than the minimum depth of drilling into the fluid layer, the formation fluid concentration is 0; When the future drilling depth is greater than the minimum depth of drilling the fluid layer, the concentration of the formation fluid flowing into the drilling fluid is predicted in real time according to the preset hybrid temporal attention network based on physical constraints; The hybrid temporal attention network based on physical constraints includes a velocity-concentration coupling physical constraint model; the velocity-concentration coupling physical constraint model is: ； ； In the formula, C b ( t ) is the concentration of formation fluid flowing into the drilling fluid at the future time t; H c is the static liquid column height of the drilling fluid currently containing cuttings; d is the inner diameter of the wellbore; d 1 is the average radius of formation pores; V r ( t ) is the average flow velocity of the invading fluid at the future time t; ρ 0 is the initial density of drilling fluid; ρ b is the average density of the formation fluid; g is the acceleration due to gravity; P p ( t ) is the formation pore pressure at time t in the future; P b is the pump pressure; B is the flow potential energy coefficient; k 1 is the pressure influence coefficient; k 2 is the drilling fluid influence coefficient.

5. A risk prediction method for simulating oil drilling according to claim 4, characterized in that: The calculation formula of the modified density of the drilling fluid is: ； In the formula, ρ e ( t ) is the modified density of the drilling fluid; ρ a The rock cuttings density ；C a is the concentration of cuttings in the annulus.

6. A risk prediction method for simulating oil drilling according to claim 4 or 5, characterized in that: The formation fluid includes one or more of oil, natural gas and water; the method for predicting the average density of the formation fluid specifically includes the following steps: Determine the weight of each fluid based on the future drilling depth and the distribution of fluid layers at different formation depths; According to the weights of the various fluids, the weighted average of the fluid densities is calculated to obtain the average density of the formation fluid.

7. The risk prediction method for simulating oil drilling according to claim 5, characterized in that: The prediction formula of the bottom hole pressure correction value at the future time t is: ； In the formula, P W ( t ) is the bottom hole pressure correction value at the future time t; g is the acceleration due to gravity; is the predicted value of annular pressure loss at time t in the future.

8. The risk prediction method for simulating oil drilling according to claim 7, characterized in that: The steps of predicting the risk of overflow or blowout in the drilling process in real time based on the difference between the bottom hole pressure correction value and the formation pore pressure at the future time t include: The corrected pressure difference is calculated based on the difference between the bottom hole pressure correction value and the formation pore pressure at the future time t; Calculate the mean and standard deviation of the corrected pressure difference within a certain period of time in the future; Based on the cumulative distribution function, the risk probability of overflow or blowout in the future drilling process is calculated according to the mean and standard deviation of the corrected pressure difference.

9. A risk prediction method for simulating oil drilling according to claim 8, characterized in that: The calculation formula for the risk probability of overflow or blowout during the future drilling process is: ； In the formula, F The risk probability of overflow or blowout during future drilling; X To correct the pressure difference, X=P W ( t ) -P p ( t ); μ X is the mean value of the corrected pressure difference; σ X is the standard deviation of the corrected pressure difference.

10. A risk prediction system for simulating oil drilling, used to implement the risk prediction method for simulating oil drilling according to any one of claims 1 to 9, characterized in that: include: The formation exploration unit is used to explore the internal structure of the formation to be simulated and predicted, obtain the distribution of fluid layers at different formation depths, and obtain the formation pore pressure at different formation depths in real time, and determine the functional relationship between the formation pore pressure and the formation depth; The drilling speed determination unit is used to determine the current drilling stroke speed in real time according to the current drilling depth based on the drilling speed model: A well depth prediction unit is used to predict the functional relationship between future drilling depth and time based on the current drilling travel speed; The formation pore pressure prediction unit is used to predict the formation pore pressure at the future time t in real time according to the functional relationship between the future drilling depth and time, and the functional relationship between the formation pore pressure and the formation depth; A formation fluid concentration prediction unit is used to predict the formation fluid concentration flowing into the drilling fluid in real time based on the functional relationship between the future drilling depth and time and the distribution of fluid layers at different formation depths; The drilling fluid density correction unit is used to obtain the drilling fluid annular return velocity and the cuttings settling velocity, and calculate the annular cuttings concentration according to the current drilling stroke speed; the initial density of the drilling fluid is corrected in real time according to the formation fluid concentration flowing into the drilling fluid and the annular cuttings concentration to obtain the corrected density of the drilling fluid; The annular pressure loss prediction unit is used to predict the annular pressure loss prediction value at the future time t in real time based on the annular pressure loss calculation model and the annular return velocity of the drilling fluid and the corrected density of the drilling fluid; The bottom hole pressure prediction unit is used to predict the bottom hole pressure correction value at the future time t in real time according to the functional relationship between the future drilling depth and time, the corrected density of the drilling fluid and the predicted value of the annular pressure loss; The risk prediction unit is used to predict the risk of overflow or blowout in the drilling process in real time based on the difference between the bottom hole pressure correction value and the formation pore pressure at the future time t.

Images