An experimental evaluation method for the influence of variable speed fracturing on the reconstruction volume of tight reservoirs

Abstract

The application relates to an experimental evaluation method for the influence of variable-speed fracturing on the reconstruction volume of a dense reservoir, belonging to the technical field of oil exploitation, and comprising the following steps: preparing a standard sample; designing a variable-speed fracturing experiment scheme corresponding to actual working conditions; obtaining energy and breaking point parameters in the experiment process by using an acoustic emission monitoring system; calculating a fracturing reconstruction volume by using a point cloud envelope body model; processing a data set constructed by using the obtained data; constructing a neural network model, inputting the data set, adjusting model parameters by using fitting errors, and obtaining a trained neural network model; predicting rock crack reconstruction volumes of different experiment schemes; and obtaining an optimal fracturing scheme by using orthogonal experiments and neural networks for multiple times. The application can more economically and efficiently evaluate the influence of variable-speed fracturing on rock crack reconstruction volumes, has a contrast with actual fracturing construction, is favorable for pre-designing a fracturing fluid injection mode, and adopts an optimal injection mode to perform fracturing.

Description

Technical Field

[0001] This invention belongs to the field of petroleum extraction technology, specifically relating to an experimental evaluation method for the impact of variable-speed fracturing on the volume of tight reservoir stimulation. Background Technology

[0002] Hydraulic fracturing is an effective method for oil and gas well development, which involves injecting high-pressure fluid into shale reservoirs to create artificially complex fractures. However, in current oil and gas extraction processes, the plastic deformation of the rock significantly influences the fracture size and volume of hydraulic fracturing. The complex nonlinear deformation of the rock under the combined effects of pore fluid pressure, bottom hole fluid pressure, and in-situ stress, along with the mutual interference between hydraulic fracturing fractures, hinders the rock matrix's ability to supply oil into the fractures. This renders traditional hydraulic fracturing methods less efficient.

[0003] Variable-rate fracturing (VRF) is a novel fracturing technology proposed in recent years. VRF involves artificially altering parameters such as injection rate, alternation time difference, alternation amplitude, alternation frequency, and alternation pattern during fracturing operations to create a complex fracture network. The goal of VRF operations is to optimize the injection rate parameters to maximize the fracturing volume.

[0004] Chinese patent CN113445976A discloses a fracturing method for highly plastic formations using alternating fracturing fluid injection. It employs acidic clean fracturing fluid and acidic homoionic slickwater for fracturing, and further optimizes the use of guar gum fracturing fluid, providing a basis for on-site fracturing design and addressing the issues of large embedment in plastic formations and strong compaction of fracture walls. However, it lacks research on the velocity gradient variation patterns at each stage of fracturing and has significant limitations in practical design due to a lack of substantial data support. The conclusions obtained are not precise enough to provide an optimal fracturing scheme.

[0005] Chinese patent CN116362121A provides a method for determining fracture parameters in horizontal well fracturing. It utilizes a neural network to calculate fracture parameters from input static geological data and dynamic fracturing operation data, and then corrects these parameters based on on-site monitoring technology. However, the small volume of stimulation in tight rock formations and the difficulty of on-site monitoring increase the prediction error of conventional fracturing, making it unsuitable for predicting the stimulation volume of variable-speed fracturing using neural networks.

[0006] The paper "Research on the Fracturing Mechanism of Pulsed Circulation Hydraulic Fracturing in Tight Reservoirs" proposes a numerical model for dynamic fracturing in tight rocks using pulsed hydraulic fracturing, analyzing the influence of pulse parameters on the stress, pore pressure, and damage around the perforation. However, in the integrated numerical simulation method for hydraulic fracturing and oil and gas production, each step of fracture propagation involves updating the information transmission and relationship links of the mesh, significantly increasing the simulation time and complexity. Furthermore, fluid-structure interaction increases the nonlinearity of the model, frequently leading to non-convergence and low solution efficiency. Therefore, it requires a significant amount of time, which does not meet the needs of oilfield sites for rapid and accurate construction parameters. Summary of the Invention

[0007] To address the problems of existing technologies, this invention discloses an experimental evaluation method for the impact of variable-speed fracturing on the volume of stiffened tight reservoirs. A training set is established by collecting fracture-induced volume data after laboratory true triaxial fracturing experiments. An artificial neural network is used to evaluate the influence of parameters such as injection rate, alternation time difference, alternation amplitude, and alternation frequency on the fracturing volume, establishing an optimal evaluation model. This invention, by comparing experimental methods with actual field conditions, clearly demonstrates the impact of variable-speed fracturing on the actual fracturing volume and fracture propagation morphology of tight reservoirs, quantitatively characterizing the fracture-induced volume of variable-speed fracturing. This provides a clear optimization scheme for actual construction and is of great significance for improving oil and gas production.

[0008] To achieve the above objectives, the technical solution of the present invention is as follows:

[0009] An experimental evaluation method for the impact of variable-speed fracturing on the volumetric stimulation of tight reservoirs includes the following steps:

[0010] S1. Collect surface outcrop rock samples from the fracturing section and process them into test samples. Prepare the experimental setup, including a servo loading system, an acoustic emission monitoring system, a liquid injection system, and a large-size true triaxial fracturing test machine.

[0011] S2. Design a fracturing test scheme. Based on the similarity criteria between the test and the actual construction site, obtain the fracturing fluid injection rate corresponding to the test and the actual site. Determine different fracturing parameter combination schemes by changing the alternation mode. Change the alternation mode by changing the alternation time difference, alternation amplitude, and alternation frequency of the injected fluid.

[0012] S3. Conduct variable-speed fracturing experiments according to the experimental scheme designed in S2. Use an acoustic emission monitoring system to record the instantaneous energy and overall acoustic emission energy signal changes when changing the injected liquid parameters. Monitor the spatial location, number, and corresponding time of the fracture points. Calculate the spatial modification volume of the rock fractures during this period and determine the fracture morphology change pattern.

[0013] S4. Using the alternating time difference, amplitude, and frequency of fracturing as factors, select different levels to establish an orthogonal experimental table. Conduct the experiment again according to the fracturing combination scheme listed in the orthogonal experimental table, monitor the fracture point information and acoustic emission energy signal of the fracture during the new experiment, and determine the fracture morphology change law and fracturing modification volume.

[0014] S5. Repeat S4 3-4 times with different parameter combinations to obtain the basic dataset, establish a neural network model, and substitute the preprocessed data of each stage obtained from the fracturing experiment into the neural network model for training to obtain a trained neural network model.

[0015] S6. Find the optimization interval, formulate a suitable orthogonal experimental table for further optimization, input each set of data in the newly generated orthogonal experimental table into the neural network model, and predict the corresponding fracturing volume.

[0016] S7. Repeat the process described in S6 multiple times, and use the prediction function of the neural network to continuously obtain a narrowing optimization range through the analysis of acoustic emission signals until the optimal combination of variable speed fracturing parameters is obtained.

[0017] Preferably, in step S2, the similarity ratio coefficient is calculated using the similarity criterion, and the fracturing fluid injection rate corresponding to the experiment and the actual field is determined based on the actual engineering parameters. Different fracturing parameter combination schemes are determined by changing the alternation mode. Changing the alternation mode by changing the alternation time difference, alternation amplitude, and alternation frequency of the injected fluid refers to: based on the fracturing fluid injection rate corresponding to the actual field, the same fracturing fluid injection rate is gradient-increasing or gradient-decreasing to achieve a variable gradient and variable speed fracturing alternation mode; in the fracturing parameter combination scheme, different experimental injection rates are injected alternately, and different alternation time differences are obtained by changing the difference in time between different injection rates within the cycle, and different alternation amplitudes are obtained by changing the rate difference between different experimental injection rates; different alternation frequencies are achieved by changing the number of alternations of the injection rate within the cycle.

[0018] Preferably, in step S3, the SRV spatial modification volume is estimated by constructing a spatial fracture envelope model, and the final fracturing modification volume is obtained by fitting the degree of change in acoustic emission energy during the fracturing process.

[0019] Preferably, in step S3, the spatial data of the fracture point monitored by the acoustic emission monitoring system is sliced ​​to construct the envelope model of each slice region, thereby obtaining the envelope model of the overall fracturing volume of the fracture.

[0020] Preferably, in step S3, the method for calculating the fracturing volume using the envelope model is as follows: draw the planar point cloud outline for each slice area, calculate its surface area, and then select different polyhedral models to calculate the fracturing volume.

[0021] Preferably, in step S3, the fracturing volume of the fracture is obtained through experiments, and the volume is accurately represented by energy parameters. A neural network is then used for prediction to reduce random errors in monitoring and improve the optimization effect of the neural network.

[0022] Preferably, in step S5, the method for establishing the neural network model includes the following steps:

[0023] S51. Preprocess the fracturing volume data in the basic dataset to remove some data with excessive noise.

[0024] S52. Normalize the data and divide it into training and test sets;

[0025] S53. Train the processed data using a neural network model, setting the input layer, output layer, number of hidden neurons, descent gradient, learning rate, and loss function to obtain the basic neural network model.

[0026] S54. Adjust the grid parameters of the neural network model according to the fitting error so that the error between the predicted value and the true value is less than 10%.

[0027] S55. Use the trained neural network model to establish a predictive model for the volume of future crack space modification.

[0028] Preferably, in step S7, a trained neural network model is used to predict the fracturing combinations of different dense rocks. By repeatedly predicting, the optimization range is gradually narrowed down, and finally the corresponding optimal fracturing range is obtained.

[0029] The beneficial effects of the experimental evaluation method for the influence of variable-speed fracturing on the volume of tight reservoir stimulation of this invention are as follows:

[0030] This invention is rationally designed, clearly explained, simple, and efficient. Utilizing the principle of similarity, it achieves a high degree of similarity with actual working conditions. It can be used to determine the corresponding fracturing combination based on the fracturing conditions available in actual working conditions, thereby obtaining the optimal fracturing result. This optimized result can effectively guide fracturing operations, significantly increase the volume of fracturing, and markedly improve the construction effect, thus achieving the best economic benefits. Specifically, it is reflected in:

[0031] 1. After acquiring the basic fracturing experimental dataset, this invention designs a new fracturing combination scheme through orthogonal experiments. It utilizes the powerful distributed processing, storage, and learning capabilities of neural networks to handle the complex nonlinear relationship between the alternating amplitude, frequency, time difference, and fracture modification volume of variable displacement fracturing, and accurately predicts the fracture volume.

[0032] 2. This invention combines an acoustic emission monitoring system to conduct variable-speed fracturing tests. By using variable-speed fracturing, a higher fracturing volume can be obtained, and the fracture morphology and volume differences of different variable-speed schemes can be obtained. The influence of various factors of variable-speed fracturing on the fracture propagation morphology and volume can be accurately analyzed, and the morphology and complexity of the fracture can be quantitatively calculated.

[0033] 3. This invention can calculate the corresponding fracturing combination parameters based on the actual on-site fracturing conditions using similarity criteria, predict the corresponding modification volume, and determine the appropriate fracturing scheme based on the fracturing conditions. The results obtained from the fracturing tests designed based on similarity criteria show a high degree of consistency with the results of actual on-site construction. This invention, by using similarity criteria to guide experimental design, has high research value and credibility.

[0034] Instruction manual illustrations

[0035] Figure 1 , one A technical roadmap for experimental evaluation methods of the impact of variable-speed fracturing on the volume of tight reservoir stimulation.

[0036] Figure 2 A flowchart for predicting the volume of variable displacement fracturing using neural networks.

[0037] Figure 3 A schematic diagram of acoustic emission energy signals and liquid pressure in variable displacement fracturing.

[0038] Figure 4 A schematic diagram of the point cloud envelope model slice partitioning. Detailed Implementation

[0039] The following description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0040] The following embodiments can be understood as illustrating a part of the structure or method of the present invention individually, or as combining the embodiments to explain the broader structure or method of the present invention.

[0041] Example 1: An experimental evaluation method for the impact of variable-speed fracturing on the volumetric stimulation of tight reservoirs, such as... Figure 1-4 As shown, it includes the following steps:

[0042] S1. Collect surface outcrop rock samples from the fractured section and process them into test samples. Prepare the experimental setup, including a servo loading system, an acoustic emission monitoring system, a liquid injection system, and a large-size true triaxial fracturing test machine. This experimental setup is existing technology and there are relevant documents in the patent database, so it will not be described in detail here.

[0043] S2. Design a fracturing test scheme. Based on the similarity criteria between the test and the actual construction site, obtain the fracturing fluid injection rate corresponding to the test and the actual site. Determine different fracturing parameter combination schemes by changing the alternation mode. Change the alternation mode by changing the alternation time difference, alternation amplitude, and alternation frequency of the injected fluid.

[0044] S3. Conduct variable-speed fracturing experiments according to the experimental scheme designed in S2. Use an acoustic emission monitoring system to record the changes in acoustic emission energy signals instantaneously and overall when the injected liquid parameters are changed. Monitor the spatial location, number, and corresponding time of the fracture points. Calculate the spatial modification volume of rock fractures during this period and determine the fracture morphology change law.

[0045] S4. Using the alternating time difference, amplitude, and frequency of fracturing as factors, select different levels to establish an orthogonal experimental table. Conduct the experiment again according to the fracturing combination scheme listed in the orthogonal experimental table, monitor the fracture point information and acoustic emission energy signal of the fracture during the new experiment, and determine the fracture morphology change law and fracturing modification volume.

[0046] S5. Repeat S4 3-4 times with different parameter combinations to obtain the basic dataset, establish a neural network model, and substitute the preprocessed data of each stage obtained from the fracturing experiment into the neural network model for training to obtain a trained neural network model.

[0047] S6. Find the optimization interval, formulate a suitable orthogonal experimental table for further optimization, input each set of data in the newly generated orthogonal experimental table into the neural network model, and predict the corresponding fracturing volume.

[0048] S7. Repeat the process described in S6 multiple times, and use the prediction function of the neural network to continuously obtain a narrowing optimization range through the analysis of acoustic emission signals until the optimal combination of variable speed fracturing parameters is obtained.

[0049] In Example 2, step S2 involves calculating the similarity ratio coefficient using similarity criteria and determining the fracturing fluid injection rate corresponding to the experimental and actual field conditions based on actual engineering parameters. Different fracturing parameter combinations are determined by changing the alternation mode. Changing the alternation mode by altering the alternation time difference, alternation amplitude, and alternation frequency of the injected fluid refers to: based on the fracturing fluid injection rate corresponding to the actual field condition, achieving a variable gradient and variable speed fracturing alternation mode by gradually increasing or decreasing the same fracturing fluid injection rate; in the fracturing parameter combination scheme, different experimental injection rates are injected alternately; different alternation time differences are obtained by changing the difference in time between different injection rates within a cycle; different alternation amplitudes are obtained by changing the rate difference between different experimental injection rates; and different alternation frequencies are achieved by changing the number of alternations of the injection rate within a cycle.

[0050] Example 3: In step S3, the SRV spatial modification volume is estimated by constructing a spatial fracture envelope model, and the final fracturing modification volume is obtained by fitting the degree of change in acoustic emission energy during the fracturing process.

[0051] In Example 4, in step S3, the spatial data of the fracture point monitored by the acoustic emission monitoring system is sliced, and an envelope model of each slice region is constructed to obtain the envelope model of the overall fracturing volume of the fracture.

[0052] Example 5: In step S3, the method for calculating the fracturing volume using the envelope model is as follows: draw the planar point cloud outline for each slice area, calculate its surface area, and then select different polyhedral models to calculate the fracturing volume.

[0053] Example 6: In step S3, the fracturing volume of the fracture is obtained through experiments, and the volume is accurately represented by energy parameters. The volume is then predicted using a neural network to reduce random monitoring errors and improve the optimization effect of the neural network.

[0054] Example 7, in step S5, as follows Figure 2 As shown, the method for building a neural network model includes the following steps:

[0055] S51. Preprocess the fracturing volume data in the basic dataset to remove some data with excessive noise.

[0056] S52. Normalize the data and divide it into training and test sets;

[0057] S53. Train the processed data using a neural network model, setting the input layer, output layer, number of hidden neurons, descent gradient, learning rate, and loss function to obtain the basic neural network model.

[0058] S54. Adjust the grid parameters of the neural network model according to the fitting error so that the error between the predicted value and the true value is less than 10%.

[0059] S55. Use the trained neural network model to establish a predictive model for the volume of future crack space modification.

[0060] In Example 8, in step S7, the trained neural network model is used to predict the fracturing combination of different dense rocks. By repeatedly predicting, the optimization range is gradually narrowed, and finally the corresponding optimal fracturing range is obtained.

[0061] Example 9: Based on the above examples, this example discloses a case of specific implementation of the method of the present invention:

[0062] Step 1: Sample preparation and preliminary setup of testing equipment.

[0063] Core samples taken from the engineering site will be processed into samples with dimensions of 300mm in length, width, and height using specialized core drilling, cutting, and grinding equipment. During processing, the parallelism tolerance of the two end faces of the core sample will be ensured to be no more than 0.3mm to prevent any impact on the test results.

[0064] The system includes an acoustic emission monitoring system, a large-size true triaxial fracturing testing machine, a liquid injection system, and a servo loading system.

[0065] Eight acoustic emission probes were connected to the specimen using a coupling agent.

[0066] It should be noted that the use of a coupling agent at the connection between the probe and the specimen in the acoustic emission monitoring system is to reduce the acoustic impedance difference between the sensor and the detection surface, thereby reducing energy reflection loss at this interface and obtaining more accurate energy and image data.

[0067] Noise detection is performed on the fracturing test. The threshold of the digital filter is set to the lowest level for no-load sampling, so that the acoustic emission monitoring instrument can correctly receive the signal without digital filter processing. The large-size true triaxial autoclave is struck, and the maximum value of the acoustic emission signal at this time is recorded. The signal within this range is the background noise. The background noise is removed, and the maximum noise value is set as the threshold of the collected signal, which is usually set to 25dB-50dB.

[0068] The acoustic emission monitoring system mainly includes sensors, amplifiers, filters, a data acquisition unit, a host computer, and acoustic emission monitoring and processing software. The threshold values ​​set by the acoustic emission monitoring system affect the acquired signal data. The acoustic emission software will only record data when the monitored signal exceeds this threshold, effectively reducing noise interference.

[0069] Step 2: Preloading of the experimental model and calculation of similarity criteria.

[0070] After the rock sample is loaded into a large-size true triaxial fracturing tester, a load of 2 MPa is applied to the sample in the z-direction, and then loads in the x and y directions are applied to fix the sample.

[0071] The injection rate of fracturing fluid in actual construction and the injection rate of fracturing fluid in experiments were calculated using similarity criteria.

[0072] The similarity criteria used in the calculation are as follows:

[0073]

[0074] Where idem is the number of criteria with the same name; K L p is the filtration coefficient. f γ is the pore pressure; Q is the pump flow rate; q is the wellbore pressure of the fracturing fluid; γ is the specific weight; η is the viscosity coefficient; K IC This is the critical value of the stress intensity factor required for crack propagation.

[0075] The similarity factor is calculated using the similarity criterion as follows:

[0076]

[0077] Wherein, the injection rate is equal to the quotient of the pump flow rate and the injection area, calculated by the injection rate similarity factor c. i It is 1.36 × 10 -5 Based on the similarity criteria and actual fracturing operation parameters, the fracturing fluid injection rates were calculated to be 6 ml / min, 8 ml / min, 10 ml / min, and 12 ml / min.

[0078] The results obtained from the fracturing test designed based on the similarity criterion are highly consistent with the results of actual field construction. This invention uses the similarity criterion to guide the design of experiments, which has high research value and credibility.

[0079] Step 3: Design a variable-speed fracturing test scheme by changing the alternation mode and conduct fracturing tests.

[0080] First, fracturing fluid with a viscosity of 1.0 mPa·s was injected at a rate of 8 ml / min. After reaching the fracture pressure, injection continued, and acoustic emission signals and fracture point information were recorded.

[0081] When the fracturing reaches the extension stage, fracturing fluid is continuously injected at a rate of 6 ml / min for 1 min. Then, the injection rate is changed to 8 ml / min for 1 min. The injection rate is then changed again to 10 ml / min for 1 min. The injection rate is then changed again to 12 ml / min for 1 min. The timing of the rate change, the location and shape of the overall fracture point, and the instantaneous and overall energy signals are recorded. This is recorded as the first stage.

[0082] The first stage was repeated twice, and these were recorded as the second and third stages, respectively. A hydraulic fracturing test with a gradient increase in fracturing fluid injection rate was completed.

[0083] By changing the injection rates of 6, 8, 10, and 12 ml / min to 12, 10, 8, and 6 ml / min respectively during the above steps, hydraulic fracturing fracture signals and energy signals with a gradient decrease in fracturing fluid injection rate can be achieved.

[0084] Thus, we obtained hydraulic fracturing acoustic emission fracture signal data and energy signal data with alternating liquid injection rate modes of gradient increase and gradient decrease.

[0085] Change the test mode to conduct fracturing tests with different alternation time differences in the liquid injection rate alternation mode.

[0086] First, a fracturing test was conducted using fracturing fluid with a viscosity of 1.0 mPa·s. The fracturing fluid was injected at a rate of 8 ml / min until the fracturing pressure was reached. The fracturing fluid was then injected at the same rate, and the changes in acoustic emission fracturing signals and energy signals during this process were recorded.

[0087] When the pressure begins to extend, continue injecting fracturing fluid at a rate of 8 ml / min for 100 s. Then change the injection rate to 10 ml / min for 140 s. Observe the changes in fracture signal and energy signal at the moment of changing the injection rate and the changes in fracture signal and total energy signal detected by the acoustic emission system after the speed change. Record this as the first stage.

[0088] Repeat the first stage twice and name them the second and third stages respectively. Complete the fracturing test with an injection rate alternation time difference of 40 s.

[0089] The experiment was repeated. When the pressure began to extend, fracturing fluid was injected at a rate of 8 ml / min for 80 seconds. Then the injection rate was changed to 10 ml / min for 160 seconds. The changes in the fracture signal and energy signal at the moment of changing the injection rate and the changes in the fracture signal and total energy signal detected by the acoustic emission system after the change of rate were observed and recorded as the first stage.

[0090] Repeat the first stage twice and name them the second and third stages respectively. Complete the fracturing test with an 80-second time difference between injection rate alternations.

[0091] Thus, we obtained hydraulic fracturing acoustic emission fracture signal and energy signal data with different alternation time differences in the fracturing fluid injection rate alternation mode.

[0092] The alternating rate time difference described in step 3 is the difference between the injection time of the first injection rate and the second injection rate.

[0093] Change the test mode to conduct fracturing tests with different alternating amplitudes of fracturing fluid injection rate alternation mode.

[0094] First, a fracturing test was conducted using fracturing fluid with a viscosity of 1.0 mPa·s. The fracturing fluid was injected at a rate of 8 ml / min until the fracturing pressure was reached. The fracturing fluid was then injected at the same rate, and the changes in acoustic emission fracturing signals and energy signals during this process were recorded.

[0095] When the pressure begins to extend, continue injecting fracturing fluid at a rate of 8 ml / min for 2 minutes. Then change the injection rate to 10 ml / min for 2 minutes. Observe the changes in fracture signal and energy signal at the moment of changing the injection rate and the changes in fracture signal and total energy signal detected by the acoustic emission system after the speed change. Record this as the first stage.

[0096] Repeat the first stage twice, naming them the second and third stages respectively. The result is as follows: Figure 3 The acoustic emission energy information shown was used to complete the basic variable-speed fracturing test.

[0097] Repeat the above steps by changing the fracturing fluid injection rate, alternating between 6 ml / min and 12 ml / min, and between 6 ml / min and 10 ml / min.

[0098] Thus, we obtained hydraulic fracturing acoustic emission fracture signal and energy signal data with different alternating amplitudes for the fracturing fluid injection rate alternation mode.

[0099] Change the experimental mode to conduct fracturing tests with different alternating frequencies of fracturing fluid injection rate alternation mode.

[0100] First, fracturing is initiated using fracturing fluid with an initial viscosity of 1.0 mPa·s at a rate of 8 ml / min. Once the injection pressure reaches the fracture pressure, injection continues until the extension stage.

[0101] When the pressure begins to extend, continue injecting fracturing fluid at a rate of 8 ml / min for 1 minute. Then change the injection rate to 10 ml / min for 1 minute. Continue changing the injection rate to 8 ml / min. Finally, change the injection rate to 10 ml / min to end one cycle. Observe the changes in fracture signal and energy signal at the moment of changing the injection rate and the changes in fracture signal and total energy signal detected by the acoustic emission system after the speed change. Record this as the first stage.

[0102] Repeat the above steps twice, and record them as the second and third stages respectively.

[0103] Change the single injection time of the fracturing fluid injection rate within one cycle, modify the single injection time of the fracturing fluid to 40s, and continue to inject at alternating rates of 8ml / min and 10ml / min, with each cycle lasting 4 minutes, for a total of three cycles.

[0104] Thus, acoustic emission fracture point data and energy signal data were obtained for fracturing fluid injection rate alternation mode with different alternating frequencies.

[0105] Step 4: Design an orthogonal experimental scheme using the alternation time difference, amplitude, and frequency of the fracturing pump injection fluid as factors, conduct fracturing tests, and calculate the fracturing stimulation volume.

[0106] An orthogonal experimental table was designed using orthogonal experiments. The number of factors was 9: factor 1 was the amplitude of the first cycle speed change with 3 levels; factor 2 was the time difference of the first cycle speed change with 3 levels; factor 3 was the frequency of the first cycle speed change with 3 levels; factor 4 was the amplitude of the second cycle speed change with 3 levels; factor 5 was the time difference of the second cycle speed change with 3 levels; factor 6 was the frequency of the second cycle speed change with 3 levels; factor 7 was the amplitude of the third cycle speed change with 3 levels; factor 8 was the time difference of the third cycle speed change with 3 levels; and factor 9 was the frequency of the third cycle speed change with 3 levels.

[0107] Based on the orthogonal experimental table described in step 4, different factors are used to realize the fracturing scenario of cross-combination of fracturing fluid velocity amplitude, time difference, and frequency.

[0108] Substitute the data into the corresponding relationship of the orthogonal test table described in step 4 to form the fracturing combination scheme of the test group. The three levels of the alternation amplitude of the liquid injection rate correspond to 2ml / min, 4ml / min, and 6ml / min, respectively; the alternation time difference corresponds to 0s, 40s, and 80s, respectively; and the alternation frequency corresponds to 1 time, 2 times, and 3 times, respectively.

[0109] It should be noted that in step 4, the scheme for each cycle is designed according to the combination of orthogonal experimental tables, and there may be differences in the fracturing schemes between cycles.

[0110] Fracturing tests were conducted based on the fracturing combination scheme designed according to the orthogonal experimental table. The number, location, and energy signal of each fracture point detected by the acoustic emission monitoring system in each cycle of velocity change were obtained, and the fracture signal and total energy of the whole process were obtained.

[0111] In this invention, the volume of an envelope is used to represent the volume of the crack space modification. The specific method is as follows:

[0112] The time of the break point and the corresponding energy information of the acoustic emission system are exported, and a point cloud data set is formed according to the location of the break extension.

[0113] The point cloud data is sliced ​​according to the break point number, such as... Figure 4 The diagram shows the division into various point cloud slice groups;

[0114] Determine the outline boundaries of each point cloud slice and calculate the area enclosed by the outline;

[0115] By using out-of-plane points to determine an appropriate polyhedral structure, a point cloud polyhedral model is constructed.

[0116] The volume and corresponding energy signal of each point cloud slice are calculated separately. The volume and energy of each part of the point cloud are linearly fitted to obtain the spatial modification volume of each part after processing. The final point cloud envelope model, as well as the final fracturing modification volume and total energy signal, are accumulated.

[0117] It should be noted that during the fitting process, the acoustic emission energy signal is regarded as a proportionality coefficient. The coefficient of the current stage energy is calculated based on the average energy and the total energy, so as to obtain the final space crack modification volume.

[0118] Step 5: Repeat the experiment with different parameters to obtain a basic dataset and build a neural network prediction model.

[0119] Repeat the experiment twice by changing the injection rate.

[0120] The first and second injection rates were increased by 1 ml / min and decreased by 1 ml / min three times to obtain the fracture point signal and acoustic emission energy signal detected by the acoustic emission system under this injection rate parameter combination, and the final fracturing volume was calculated.

[0121] The data results of the variable displacement fracturing test described in steps 3, 4, and 5 are preprocessed to form a basic dataset.

[0122] In the specific implementation process, the preprocessing includes screening outlier data and normalizing the data to 0-1 using the mapminmax command in MATLAB.

[0123] It should be noted that the purpose of filtering outlier data is mainly to reduce noise in the original data. If the noise in the original data is small, normalization can be performed directly without filtering.

[0124] In step 5, a bp neural network model is established to train a prediction model for the final space crack modification volume calculated based on the data obtained from the acoustic emission system detection.

[0125] The basic dataset is divided into a training set and a test set in a ratio of 4:1. Preset values ​​are set for the input layer, output layer, number of hidden neurons, learning rate, and descent gradient of the neural network. The data from the training set is then substituted into the neural network model for training to predict the volume of the spatial cracks in the training set.

[0126] The test set data is fed into the neural network for prediction. The predicted crack modification volume is compared with the true value. The value of the loss function and the error of the test set are observed to see if they are both within 10%. If they are greater than 10%, the preset value of the neural network model is adjusted until the error of the loss function and the error of the test set are both less than 10%.

[0127] Step 6: Find the optimization interval and formulate a suitable orthogonal experimental table for further optimization.

[0128] In the basic dataset described in step 5, based on the larger range of fracture modification volume, the corresponding range of fracturing parameters for variable displacement amplitude, time difference, and frequency is found.

[0129] In step 6, during the process of finding the optimal range for fracturing parameters, range analysis is used to determine the range. A range analysis table is established, the range values ​​and optimal levels of the factors are calculated, and the optimal combination of factors and optimal levels is determined. The optimal combination of factors and levels is the optimized range of variable displacement fracturing parameters.

[0130] In a further optimized implementation, the variable displacement fracturing parameter range described in step 6 is segmented and regrouped using an orthogonal experimental table. The level number of the alternation amplitude, time difference, and frequency of the liquid injection rate for the three corresponding cycles is 3, thus obtaining a new combination of fracturing parameters.

[0131] The new fracturing fluid parameter combination scheme described in step 6 is input into the neural network model for prediction to obtain the fracture volume and energy signals under the new set of fracturing schemes.

[0132] Range analysis was performed on the newly obtained volume and energy data after neural network prediction to further obtain the optimization range of fracturing parameters.

[0133] Step 7: Repeat step 6 multiple times, using the prediction function of the neural network to continuously obtain a narrowing optimization range through the analysis of acoustic emission signals, until the optimal combination of variable speed fracturing parameters is obtained.

[0134] In summary, this invention can not only be used to determine the optimal combination of displacement parameters for variable displacement fracturing, but also to determine suitable fracturing schemes and predict the expected fracturing oil production results based on the actual conditions of the operating environment. Compared with the prior art, this invention represents a significant advancement.

[0135] The above description is only used to illustrate the technical solution of the present invention, and is not intended to limit the scope of protection of the present invention. Although the present invention has been described in detail with reference to the experimental scheme, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solution of the present invention without departing from the scope of the technical solution of the present invention. Any simple modifications and alterations made based on the essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of the technical solution of the present invention.

Claims

1. An experimental evaluation method for the influence of variable-speed fracturing on the volumetric stimulation of tight reservoirs, characterized in that, Includes the following steps: S1. Collect surface outcrop rock samples from the fractured section and process them into test samples. Prepare experimental equipment, including a servo loading system, acoustic emission monitoring system, liquid injection system, and large-size true triaxial fracturing test machine. S2. Design a fracturing test scheme. Based on the similarity criteria between the experiment and the actual construction site, obtain the fracturing fluid injection rate corresponding to the experiment and the actual site. Determine different fracturing parameter combinations by changing the alternation mode. The alternation mode is changed by changing the alternation time difference, alternation amplitude, and alternation frequency of the injected fluid. Changing the alternation mode by changing the alternation time difference, alternation amplitude, and alternation frequency of the injected fluid means: based on the fracturing fluid injection rate corresponding to the actual site, the same fracturing fluid injection rate is gradually increased or decreased to achieve a variable gradient and variable speed fracturing alternation mode. In the fracturing parameter combination scheme, different experimental injection rates are injected alternately. By changing the difference in time between different injection rates within the cycle, an alternation mode with different alternation time differences is obtained. By changing the rate difference between different experimental injection rates, an alternation mode with different alternation amplitudes is obtained. S3. Conduct variable-speed fracturing experiments according to the experimental scheme designed in S2. Use an acoustic emission monitoring system to record the instantaneous energy and overall acoustic emission energy signal changes when changing the injected liquid parameters. Monitor the spatial location, number, and corresponding time of the fracture points. Calculate the spatial modification volume of rock fractures during this period and determine the fracture morphology change law. The volume of spatial fracturing was estimated by constructing a spatial fracture envelope model, and the final fracturing volume was obtained by fitting the change in acoustic emission energy during the fracturing process. S4. Using the alternation time difference, amplitude, and frequency of the fracturing pump injection fluid as factors, select different levels to establish an orthogonal experimental table. Conduct the experiment again according to the fracturing combination scheme listed in the orthogonal experimental table, monitor the fracture point information and acoustic emission energy signal of the fracture during the new experiment, and determine the fracture morphology change law and fracturing modification volume. S5. Repeat S4 3-4 times with different parameter combinations to obtain the basic dataset, establish a neural network model, and input the preprocessed fracturing volume data of each stage obtained from the fracturing experiment into the neural network model for training to obtain a trained neural network model. S6. Find the optimization interval, formulate a suitable orthogonal experimental table for further optimization, input each set of data in the newly generated orthogonal experimental table into the neural network model, and predict the corresponding fracturing volume. S7. Repeat the process described in S6 multiple times, and use the prediction function of the neural network to continuously obtain a narrowing optimization range by analyzing the acoustic emission signal until the optimal combination of variable speed fracturing parameters is obtained.

2. The experimental evaluation method for the influence of variable-speed fracturing on the volume of tight reservoir stimulation as described in claim 1, characterized in that, In step S2, the similarity ratio coefficient is calculated using the similarity criterion, and the fracturing fluid injection rate corresponding to the experiment and the actual field is determined based on the actual engineering parameters; different fracturing parameter combination schemes are determined by changing the alternation mode, and different alternation frequencies are achieved by changing the number of alternations of the injection rate within the cycle.

3. The experimental evaluation method for the influence of variable-speed fracturing on the volume of tight reservoir stimulation as described in claim 2, characterized in that, In step S3, the spatial data of the fracture point monitored by the acoustic emission monitoring system is sliced, and an envelope model of each slice region is constructed to obtain the envelope model of the overall fracturing volume of the fracture.

4. The experimental evaluation method for the influence of variable-speed fracturing on the volume of tight reservoir stimulation as described in claim 3, characterized in that, In step S3, the method for calculating the fracturing volume using the envelope model is as follows: draw the planar point cloud outline for each slice area, calculate its surface area, and then select different polyhedral models to calculate the fracturing volume.

5. The experimental evaluation method for the influence of variable-speed fracturing on the volume of tight reservoir stimulation as described in claim 4, characterized in that, The volume of fracture fracturing was obtained through experiments, and the volume was accurately represented by energy parameters. A neural network was used for prediction to reduce random errors in monitoring and improve the optimization effect of the neural network.

6. The experimental evaluation method for the influence of variable-speed fracturing on the volume of tight reservoir stimulation as described in claim 5, characterized in that, In step S5, the method for establishing a neural network model includes the following steps: S51. Preprocess the fracturing volume data in the basic dataset to remove some data with excessive noise. S52. Normalize the data and divide it into training and test sets; S53. Train the processed data using a neural network model, setting the input layer, output layer, number of hidden neurons, descent gradient, learning rate, and loss function to obtain the basic neural network model. S54. Adjust the grid parameters of the neural network model according to the fitting error so that the error between the predicted value and the true value is less than 10%. S55. Use the trained neural network model to establish a predictive model for the volume of future crack space modification.

7. The experimental evaluation method for the influence of variable-speed fracturing on the volume of tight reservoir stimulation as described in claim 6, characterized in that, In step S7, the trained neural network model is used to predict the fracturing combinations of different dense rocks. By repeatedly predicting, the optimization range is gradually narrowed down, and finally the corresponding optimal fracturing range is obtained.

Images