Image-driven heat conductivity coefficient analysis method suitable for natural gas hydrate sediments

Abstract

The invention discloses an image-driven heat conductivity coefficient analysis method suitable for natural gas hydrate sediments, which comprises the following steps of: gridding a representative high-definition image of the natural gas hydrate sediments to obtain a multi-phase volume fraction, a soil particle morphological characteristic parameter and a hydrate distribution mode parameter, initializing all grid units into methane gas heat-conducting properties, and analyzing the methane gas heat-conducting properties; generating a soil particle core unit, expanding the form according to the growth probability, and filling internal holes and edge gaps to form a geometric closed soil particle model; performing transient temperature field simulation on the geometric closed soil particle model, outputting mesoscopic model data, importing the mesoscopic model data into finite element software, calculating heat flux, temperature gradient and effective heat conductivity coefficients through a steady-state heat transfer equation, and iteratively generating a normal distribution statistical result of the effective heat conductivity coefficients through a Monte Carlo method; heat transfer path characterization errors caused by traditional homogeneous phase simplification are eliminated, geometrically closed soil particle model reconstruction is achieved, and heat flux calculation deviation caused by geometric defects is eliminated.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image analysis, in particular to an image-driven thermal conductivity analysis method for natural gas hydrate deposits. BACKGROUND

[0002] The accurate characterization of the thermal conductivity of natural gas hydrate deposits is a key parameter for predicting the evolution of the thermal flow field during their exploitation. Current technical routes mainly fall into two categories: experimental testing methods (such as the transient hot-wire method, thermal probe method) and numerical simulation methods. Among them, the experimental method inverses the thermal conductivity by directly measuring the thermal response of the sediment sample under temperature control, and its reliability has been verified by multiple studies; the numerical simulation method reconstructs the geometry model of the porous medium based on micro-CT scan images, and solves the heat conduction equation by combining the finite element method (FEM) or lattice Boltzmann method (LBM). In recent years, the rise of image-driven technology has further promoted the development of mesoscale thermal conductivity modeling, such as using the four-parameter random growth method (four-parameter growth method) to generate heterogeneous multi-phase structures, or establishing a mapping relationship between image features and thermal conductivity through machine learning, which significantly improves the characterization ability of complex pore structures.

[0003] However, the existing technology still has key deficiencies: on the one hand, as disclosed in the literature with patent publication number CN108169448B, a hydrate in-situ synthesis and comprehensive property testing device can perform comprehensive testing of the thermal conductivity, sound velocity, resistivity, etc. of the sample in a sealed, high-pressure, low-temperature environment, realize in-situ measurement of the sample, and improve the testing accuracy and scope of application. However, this type of solution mainly focuses on the experimental acquisition of sample properties, does not involve automatic identification and geometric reconstruction of occurrence morphology driven by physical fields (such as temperature field), and does not establish a systematic correlation and uncertainty characterization of "microscopic occurrence morphology - effective thermal conductivity", making it difficult to provide reusable methodological support for morphology control and structure optimization. On the other hand, although in CN115468824B, through multi-pipe segment high-pressure in-situ detection and compaction fitting, etc., attempts are made to reduce the error caused by decomposition and contact thermal resistance, but it focuses on the in-situ acquisition of the properties of pure-phase hydrates, and has not yet given a unified methodological framework for the quantitative correlation and uncertainty evaluation of occurrence morphology (bridging, wrapping, suspension, blockiness), interfacial / contact / connectivity structure in multi-phase deposits such as soil particles-water-hydrate, and effective thermal conductivity. Thus, in the aspect of image-driven numerical simulation, the traditional approach often simplifies hydrates as homogeneous single-phase, ignoring the differentiated influence of different occurrence morphologies on the heat transfer path, making it difficult to support accurate prediction and operational structure optimization of the exploitation thermal flow field evolution. SUMMARY

[0004] To solve the above problems, the application provides an image-driven thermal conductivity analysis method for natural gas hydrate deposits, which can solve the problem of distorted thermal conductivity caused by hydrate decomposition due to thermal disturbance in traditional experimental testing, and the problem of insufficient quantification of hydrate occurrence form characteristics in existing numerical models.

[0005] To solve the above problems, the application provides an image-driven thermal conductivity analysis method for natural gas hydrate deposits, which can solve the problem of distorted thermal conductivity caused by hydrate decomposition due to thermal disturbance in traditional experimental testing, and the problem of insufficient quantification of hydrate occurrence form characteristics in existing numerical models. Step 1): grid processing is performed on a representative high-definition image of natural gas hydrate deposits, the high-definition image is converted into matrix data, and multi-phase volume fractions, soil particle morphology characteristic parameters and hydrate distribution mode parameters are identified and quantified, and the overlapping area and geological non-uniformity parameters are processed according to the priority rule; The multi-phase volume fractions include soil particle volume fraction, hydrate volume fraction, methane gas volume fraction and water volume fraction, the soil particle morphology characteristic parameters include core growth probability and four-direction growth probability, and the hydrate distribution mode parameters include volume proportions of four distribution modes of bridging, wrapping, suspension and blockiness. Step 2): all grid cells in step 1) are initialized as methane gas thermal conductivity properties, a soil particle core unit is generated, and the four-direction growth probability is expanded to fill the internal cavities and edge gaps to form a geometrically closed soil particle model; Step 3): the geometrically closed soil particle model in step 2) is simulated for transient temperature field, the hydrate distribution mode is determined, the boundary hydrate and liquid bridge hydrate are distinguished based on the set temperature threshold, the blocky hydrate and the suspended hydrate are randomly generated, the properties of all grid cells are integrated, and the mesoscopic model data containing thermal conductivity are output; Step 4): the mesoscopic model data in step 3) are imported into finite element software, a representative unit body is created and a structured grid is generated, the temperature difference between the two ends of the structured grid and the adiabatic side boundary conditions of the top and bottom are set, and the heat flux, temperature gradient and effective thermal conductivity are obtained by calculating the steady-state heat transfer equation; Step 5): the normal distribution statistical results of the effective thermal conductivity in step 4) are generated by combining the Monte Carlo method, that is, the analysis results of the image-driven thermal conductivity of natural gas hydrate deposits.

[0006] The application has the following beneficial effects after adopting the above technical solutions: 1、The application realizes accurate quantification of hydrate occurrence forms such as bridging type and wrapping type by grid processing of high-definition images and identification of multi-phase parameters, and processes the overlapping area and non-uniformity geological parameters according to the priority rule, thereby eliminating the heat transfer path representation error caused by traditional homogeneous phase simplification, providing high-fidelity input for thermal anisotropy modeling, and improving the subsequent heat conduction simulation accuracy.

[0007] 2、The application solves the problem of "doughnut-shaped" geometric distortion of the traditional four-parameter growth method, improves the reduction degree of the contact area of soil particles-hydrate, increases the coincidence degree of the generated position of the liquid bridge and the actual temperature field distribution, and eliminates the calculation deviation of heat flux caused by geometric defects. BRIEF DESCRIPTION OF DRAWINGS

[0008] In order to more clearly illustrate the technical solutions of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0009] Figure 1 The flow chart of the image-driven thermal conductivity analysis method suitable for natural gas hydrate deposits of the present application; Figure 2 The schematic diagram of the hydrate distribution characteristic region division in natural gas hydrate; Figure 3 The schematic diagram of the parameter identification process of the four-parameter growth method; Figure 4 The field diagram of bridging hydrate; Figure 5 The field diagram of the inclusion type and the suspension type hydrate; Figure 6 The reconstruction flow chart of the improved four-parameter growth method; Figure 7 The generation result schematic diagram of the improved four-parameter growth method under different distribution characteristics; Figure 8 The schematic diagram of the representative unit thermal conductivity calculation model. DETAILED DESCRIPTION

[0010] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings.

[0011] For natural gas hydrate, the volume fraction of the four of soil particles, water, methane gas and hydrate in a unit volume satisfies: (1) Wherein, represents the volume fraction of soil particles in a unit volume; represents the volume fraction of water in a unit volume; φCH4 represents the volume fraction of methane gas in unit volume; φH represents the volume fraction of hydrate in unit volume.

[0012] The saturations of soil particles, water and methane gas satisfy: (2) φCH4 represents the saturation of methane gas in unit volume of pores; φH represents the saturation of hydrate in unit volume of pores; φW represents the saturation of water in unit volume of pores. The relationship between the volume fraction and the saturation is:

[0013] (3) φCH4 represents the saturation of methane gas in unit volume of pores; φH represents the saturation of hydrate in unit volume of pores; φW represents the saturation of water in unit volume of pores.

[0014] With reference to Figure 1 , the present application is applicable to an image-driven thermal conductivity analysis method for natural gas hydrate deposits, comprising the following steps: S1: inputting a representative high-definition image of a natural gas hydrate deposit, the REV image can be obtained by X-ray scanning or optical lens of an in-situ or artificial sample. Based on the high-definition image, the high-definition image is digitally processed, the model grid is obtained through gridding processing, the high-definition image is converted into matrix data, and based on the matrix data, the multiphase volume fraction, the soil particle morphological characteristic parameter and the hydrate distribution mode parameter are identified and quantified, and the overlapping area and the geological heterogeneity parameter are processed according to the priority rule. Through gridding processing, the image containing different colors is converted into matrix data containing soil particles, hydrates, methane gas and water, each unit in the matrix stores a phase state identification value: soil particles = 1, hydrates = 2, methane gas = 3, and water = 4, and the matrix size is consistent with the image resolution.

[0015] The multiphase volume fraction includes soil particle volume fraction, hydrate volume fraction, methane gas volume fraction and water volume fraction. The soil particle morphological characteristic parameter includes core growth probability and directional growth probability. The hydrate distribution mode parameter includes the volume proportions of the four distribution modes of bridging, wrapping, suspension and blockiness, and the volume proportions of the four distribution modes can be identified according to the picture.

[0016] The four distribution modes of natural gas hydrate are shown in Figure 2 Figure 2 a shows the bridging mode, in which hydrates are formed between soil particles, forming a heat transfer channel with good heat transfer efficiency, thereby enhancing the heat transfer performance of the natural gas hydrate deposit. 2b shows the wrapping mode, in which hydrates are formed around soil particles, which generally occurs in strata with good hydrophilicity.​Figure 2 c shows the suspension mode, where hydrates are suspended inside the natural gas hydrate sediments. They have no obvious contact with soil particles and are relatively small in size. Figure 2 Figure d shows the block pattern, where the hydrate is distributed in blocks, and some may come into contact with soil particles, forming heat transfer channels. Figure 2 Image e shows a real-world image of natural gas hydrate distribution. As can be seen from the image, the four distribution patterns of natural gas hydrates overlap. When there is overlap in the identification of distribution patterns, such as... Figure 2 As shown in f, based on the potential contribution of the distribution pattern to the thermal conductivity, the priority design is identified as: bridging, wrapping, suspension, and block.

[0017] The content relationships of the four hydrate distribution patterns per unit volume are as follows: + + + = (4) in, Indicates the content of bridging hydrates; Indicates the content of encapsulated hydrates; Indicates the content of suspended hydrates; Indicates the content of blocky hydrates. This indicates the volume fraction of hydrates per unit volume.

[0018] Water and hydrates have similar thermal conductivity, so they were combined.

[0019] like Figure 3 As shown, Figure 3 (a) is a cross-sectional image of the particle obtained by scanning electron microscopy. Figure 3 (b) is a magnified electron microscope image of a single particle. Figure 3 (c) is a grain-mesh diagram. Figure 3 (d) is the result image of particle identification grid. Figure 3 (e) is a schematic diagram of the recognition algorithm direction, obtaining the characteristic parameters of soil particles, including: the growth probability of solid cores Ps, 13, 24, 57, 68, and the growth probability Ps, 13, Ps, 24, Ps, 57, Ps, 68 in the direction of 13, 24, 57, 68. The core growth probability is calculated by randomly generating recognition points in the grid and counting the ratio of the number of times a random point corresponds to a soil particle feature to the total number of times points are deployed. When a random point falls on a soil particle feature, along the direction of 13, 24, 57, 68, the growth probability of solid cores is obtained. Figure 4 The direction in 'e' expands outwards, and the number of grids expanded is recorded. Then, the statistical growth value of the grids in the corresponding growth direction is divided by the maximum particle size in the figure to obtain the growth probability of soil particles in the corresponding direction.

[0020] Similarly, the characteristic parameters of natural gas can be obtained, including: the growth probability of natural gas core Pg,13,24,57,68 in the direction of Pg,13, Pg,24, Pg,57, Pg,68. It should be pointed out that this is an optional process, which is only applicable to the case of low natural gas content.

[0021] As shown in Figure 4 , the liquid bridge water is determined according to the distribution characteristics of hydrate between soil particles, and the total volume of the liquid bridge water is recorded, as shown in Figure 4 (a). Figure 4 The area of the liquid bridge hydrate in the image is obtained by statistics, and for the case where the liquid bridge water has no boundary, the liquid bridge water is defined as the hydrate area at the shorter position between two particles, as shown in Figure 4 (b).

[0022] As shown in Figure 5 , for the hydrate attached to the surface of the soil particles and in contact with the natural gas at the other end, it is identified as the wrapped hydrate, as shown in Figure 5 (a). Figure 5 (b), if the two ends of the hydrate are in contact with the particles, the area of the wrapped hydrate is the difference between the total area of the hydrate and the area of the liquid bridge, which is identified as the characteristic parameter .

[0023] Hydrate is distributed in blocks in natural gas hydrate deposits, as shown in Figure 3 The method for obtaining the characteristic parameters of soil particles identifies the distribution characteristics of hydrate, including: the growth probability of natural gas core Pg,13,24,57,68 in the direction of Pg,13, Pg,24, Pg,57, Pg,68.

[0024] Hydrate is distributed in small bubbles in natural gas hydrate deposits, which often occurs in an environment with excess water.

[0025] S2. The improved four-parameter growth method (four-parameter growth method) is used to initialize the model grid as the heat conduction property of methane gas, generate the core of soil particles, and extend the shape according to the growth probability to fill the internal voids and edge gaps, forming a geometrically closed soil particle model.

[0026] According to the calculation range set by the artificial, the grid is divided, all grid units are initialized as methane gas heat conduction property, and the methane gas heat conduction coefficient (the heat conduction coefficient is fixed at 0.025 W / m·K) is assigned to each grid unit as the initial background medium.

[0027] According to the identified core growth probability, the corresponding grid is generated, the core grid of soil particles is randomly selected, a uniformly distributed random number matrix (value range [0, 1]) is generated, and the grid with a random number less than the core growth probability is marked as the core unit of the soil particles. When the core growth probability is less than the core growth probability, the final soil particle content will be about 0.3.

[0028] Iterative expansion of soil particle boundary according to direction growth probability: direction growth probability 13 (left lower oblique direction), 24 (right lower oblique direction), 57 (left upper oblique direction), and 68 (right upper oblique direction) four direction growth probability parameters Ps, 13, Ps, 24, Ps, 57, and Ps, 68 iteratively expand the soil particle boundary. When there is geological non-uniformity, the ratio of Ps, 24 to Ps, 13, the ratio of Ps, 57 to Ps, 68, and the average of the two are the vertical / horizontal non-uniformity parameters.

[0029] Scan non-soil particle units. If the adjacent unit in a certain direction is a soil particle and the random number is less than the direction growth probability, the unit is converted into a soil particle unit, and the cycle is executed until the total volume of soil particles accounts for the volume content Vs of soil particles in unit volume. By scanning all grids, pores surrounded by soil particles and pores surrounded by soil particles on three sides are found and filled accordingly to form a geometrically closed soil particle model.

[0030] The results of the core and growth of the soil particles are shown in Figure 6 (a) and 6 (b), it can be found that there are holes in the generated soil particles, Figure 6 the red circle area in (b), and significant gaps exist at the edges, Figure 6 the green circle area in (b), complete single particles rarely appear in nature in this central hole mode. Therefore, the improved four-parameter growth method model increases the judgment of reasonable shape. By scanning all grids, pores surrounded by soil particles and pores surrounded by soil particles on three sides are found and filled accordingly, as shown in Figure 6 the red box in (c) and the green box in 6 (d).

[0031] S3. Based on the transient temperature field simulation, determine the hydrate distribution pattern, set the temperature threshold to distinguish boundary hydrate and liquid bridge hydrate, randomly generate block hydrate and suspended hydrate, and output the mesoscopic model data.

[0032] Import the geometrically closed soil particle model into COMSOL software, set the soil particle area constant temperature to 20℃ and the initial temperature of the geometrically closed soil particle model to 0℃, and set the four surrounding boundaries as thermal insulation conditions. Run the transient heat transfer simulation for 12 seconds and record the final temperature field distribution; the area of natural gas or water is temporarily set to thermal conductivity 2.5 W / m / k and heat capacity 500 J / kg / K.

[0033] Extract the isotherm covered area where the extraction temperature is greater than temperature threshold A = 10℃ and less than temperature threshold B = 12℃, mark the non-solid units in contact with the soil particle units in this area as wrapped hydrate units.

[0034] Extract the isotherm covered area where the extraction temperature is greater than temperature threshold B = 12℃, identify the rectangular area with the smallest width along the connecting line between the centers of the two soil particles, and mark the non-solid units in this rectangular area as bridging hydrate units.

[0035] As shown in Figure 6 (f) and 6(g), the isotherm covered areas of B = 12℃ and A = 10℃ can be used to represent the distribution of boundary hydrate and liquid bridge / boundary hydrate, respectively. Figure 6 (f) and 6(g), the isotherm covered areas of B = 12℃ and A = 10℃ can be used to represent the distribution of boundary hydrate and liquid bridge / boundary hydrate, respectively. Figure 6 (g) contains both liquid bridge and boundary modes, and the effective area of the liquid bridge is the width at the shortest cross section, as shown in Figure 6 (h) the purple area.

[0036] According to the block-shaped volume ratio, the same four-parameter growth rule as the soil particles is adopted: randomly select a core grid, iteratively expand the boundary according to the growth probability of 13, 24, 57, 68 directions, generate an area that strictly avoids soil particle, wrapped hydrate and bridging hydrate units, and finally generate block-shaped hydrate units.

[0037] Randomly select a core position in the remaining non-solid grid, generate a suspended hydrate unit in single grid units, ensure that the unit is not adjacent to the soil particle unit and does not touch the boundary of the block-shaped hydrate unit, and the total volume ratio meets the suspended volume ratio .

[0038] Integrate all grid unit attributes, assign a thermal conductivity of 2.0-10.0 W / m·K to the soil particle unit, a thermal conductivity of 0.6 W / m·K to the wrapped, bridging, block-shaped and suspended hydrate units, and keep the remaining units as methane gas with a thermal conductivity of 0.025 W / m·K. These thermal conductivities can be obtained by field sampling and then tested by a thermal conductivity tester. The output includes the CSV file of the mesoscopic model data containing coordinates, thermal conductivity and five types of unit labels, which are one of the five categories of soil particles, wrapped, bridging, block-shaped and suspended hydrate units. Adding this label here is convenient for later statistics of the volume content of various types of hydrate.

[0039] The distribution results of the method of the present application under typical hydrate occurrence characteristics are shown in Figure 7 The basic parameters of the model are the same, that is, =0.35, =0.1, =0.55, the characteristic parameters of the particles are Ps,c=0.01, Ps,1~8=0.5, and the four typical distribution patterns are as follows: Liquid bridge mode ( =1): such as Figure 7 As shown in (a-1), hydrates are distributed between the particles, connecting them to each other and forming a significant bridging phenomenon, such as... Figure 7 (a-2) is an enlarged view of the liquid bridge mode results. Boundary mode ( =1): such as Figure 7 As shown in (b-1), the hydrates mainly surround the particles, forming a significant encapsulation feature, such as... Figure 7 (b-2) shows an enlarged view of the boundary pattern results, and the particles still maintain an independent relationship with each other.

[0040] Floating mode ( =1): such as Figure 7 As shown in (c-1), the hydrates are small in size and suspended in water. There is no obvious correlation between hydrates and soil particles, as... Figure 7 Enlarged view of the suspension mode result of (c-2).

[0041] Block mode ( =1): such as Figure 7 As shown in (d-1), the hydrates are relatively large in size and randomly distributed in the water, some suspended and some in contact with the particles, as detailed below. Figure 7 (d-2) is an enlarged view of the block pattern result.

[0042] Non-uniform mode ( =0.45): such as Figure 7 As shown in (e-1), the soil particles exhibit significant anisotropy in their distribution, with the long axis of the particles along the lower right or upper left. Details are as follows: Figure 7 (e-2) is an enlarged view of the non-uniform mode result.

[0043] Temperature field such as Figure 7 (a3~e3), Figure 7 (a-3) is the temperature field diagram in liquid bridge mode. Figure 7 (b-3) is the temperature field diagram of the boundary mode. Figure 7 (c-3) is the temperature field diagram in suspension mode. Figure 7 (d-3) is the temperature field diagram of the block model. Figure 7 (e-3) is the temperature field plot for the non-uniform mode. Flux calculation results are as follows: Figure 7 As shown in (a4~e4), Figure 7(a-4) is the heat flux diagram in liquid bridge mode. Figure 7 (b-4) is the boundary mode heat flux diagram. Figure 7 (c-4) is the heat flux diagram in suspension mode. Figure 7 (d-4) is the heat flux diagram for the block model. Figure 7 e-4 non-uniform mode heat flux diagram.

[0044] S4. Import the microscopic model data into the finite element software to construct representative unit bodies and divide them into meshes. Set the temperature difference between the two ends and the boundary conditions on the adiabatic side. Calculate the heat flux and temperature gradient through the steady-state heat transfer equation to derive the effective thermal conductivity. Combine the Monte Carlo method to iteratively generate the normal distribution statistical results of the effective thermal conductivity.

[0045] As shown in Figure 9, the mesoscopic model data is imported into COMSOL software to create representative unit cells with a size of N×N (4cm×4cm as an example), as shown in Figure 9(a). A structured mesh is generated globally over the representative unit cells, and the minimum unit cell side length is set to 40μm, which is 1 / 1000 of the maximum model length. A temperature difference of 10℃ is set at both ends, and adiabatic boundary conditions are set at the top and bottom to prevent heat loss from affecting the results. Thermal conductivity is assigned according to the element type label.

[0046] The temperature field calculation results are shown in Figure 9(b). Due to the non-uniformity of the material, the temperature gradient exhibits an irregular distribution, but overall it maintains a downward trend along the heat flow direction. The heat flux distribution is shown in Figure 9(c), where main heat flux channels of different widths and tortuosities are formed inside the sample.

[0047] Solve the Fourier steady-state heat conduction equation to calculate the volume-average heat flux and the average temperature gradients at the left and right boundaries of the representative unit cell, and output the effective thermal conductivity. Specifically, the calculations of heat flux, temperature difference, and effective thermal conductivity are shown in Equations 5-7, respectively.

[0048] (5); (6); (7); in, This represents the average heat flux within the computational unit. This represents the average temperature difference between the two ends of the calculation unit, i.e., the temperature gradient; Indicates the effective thermal conductivity; These represent the average temperatures at both ends of the calculation unit. This indicates the length of the calculation unit in the direction of heat flow; Indicates heat flow; Indicates the first Heat flow of each unit cell; representing the number of units; representing the area of integration.

[0049] The temperature field of the natural gas hydrate deposit under different distribution characteristics is as shown in ​ (a3~e3) and the flux calculation results are as shown in ​ (a4~e4). Through comparison of the temperature field, it can be found that, due to the influence of the non-uniform distribution of the material, the temperature field distribution also presents a non-uniform trend.

[0050] The liquid bridge distribution natural gas hydrate deposit model, ​ (a-4), the continuity of the heat transfer channel is good, and the length is relatively long, which is mainly due to the contribution of the hydrate bridging soil particles, which improves the heat transfer efficiency. The wrapped distribution natural gas hydrate deposit model, ​ (b-4), the heat transfer channel is relatively thick, which can be attributed to the hydrate wrapped outside the particle that enhances the heat transfer efficiency and produces the effect of particle amplification; the planktonic distribution natural gas hydrate deposit model, the number of heat transfer channels increases significantly, but the width is short; the bulk distribution natural gas hydrate deposit model, the length, width and number of heat transfer channels all increase to a certain extent, but the length increase is less than that of the liquid bridge distribution natural gas hydrate deposit, the width increase is less than that of the wrapped distribution natural gas hydrate deposit, and the number increase is less than that of the planktonic distribution natural gas hydrate deposit, which is mainly due to the existence of the bulk hydrate that forms the characteristics of local liquid bridge, local wrapping and local planktonic distribution. The anisotropic natural gas hydrate deposit model, the heat transfer direction significantly tilts to the right and down.

[0051] 500 times are set as the iteration threshold of the Monte Carlo algorithm, and the Monte Carlo method is executed, and the same input parameters are used in each iteration but the random number seed is updated. Monte Carlo is a representative random method to estimate the physical properties of materials, which can infinitely approximate the true solution through a large number of simulations. Randomness is introduced in the calculation process of the IRMS model of the natural gas hydrate deposit. Although the input characteristic parameters are determined, random numbers are introduced in the distribution position and form of each phase, which can generate many mesostructure models available for finite element analysis. The arithmetic mean and standard deviation of the effective thermal conductivity results of 500 iterations are calculated to verify that the calculation results meet the normal distribution characteristics, and the analysis results of the image driving thermal conductivity are obtained.

[0052] S5. Perform Pearson correlation analysis based on the normal distribution statistical results of the effective thermal conductivity to determine the negative correlation factors and the positive correlation factors, and output the regulation priority decision parameters.

[0053] The complete data set is extracted from the normal distribution statistical results as input parameters, and the Pearson correlation coefficient of each input parameter with the effective thermal conductivity is calculated.

[0054] The complete data set is extracted from the output 500 normal distribution statistical results, including input parameters (methane gas volume fraction, soil particle volume fraction, hydrate volume fraction, suspended type volume ratio, wrapped type volume ratio, block type volume ratio, bridging type volume ratio, soil particle thermal conductivity, vertical / horizontal non-uniformity parameters) and output parameter effective thermal conductivity.

[0055] The Pearson correlation coefficient of each input parameter with the effective thermal conductivity is calculated: first, the covariance of the parameter and the effective thermal conductivity is calculated, and then divided by the product of the standard deviation of the parameter and the standard deviation of the effective thermal conductivity.

[0056] According to the sign of the correlation coefficient, determine the negative correlation factor and the positive correlation factor, and arrange all factors in descending order of the absolute value of the correlation coefficient.

[0057] Specifically, according to the sign of the Pearson correlation coefficient, the negative correlation factors are: methane gas volume fraction, vertical non-uniformity parameter, wrapped type volume ratio, block type volume ratio. The positive correlation factors are: soil particle volume fraction, hydrate volume fraction, suspended type volume ratio, soil particle thermal conductivity, horizontal non-uniformity parameter.

[0058] Arrange all factors in descending order of the absolute value of the Pearson correlation coefficient, and the output order is: methane gas content>soil particle content>hydrate content>suspended model>wrapped model>soil particle thermal conductivity>vertical non-uniformity>block model>bridging model>horizontal non-uniformity. Among them, methane gas content, vertical non-uniformity, wrapped model, block model have negative impact on the thermal conductivity of natural gas hydrate deposit. In addition, by comparing the numerical size, it can be found that the soil particle content and the methane gas content have a greater impact on the heat transfer performance of natural gas hydrate deposit than other factors. Therefore, special attention should be paid to these two factors during the exploitation of natural gas hydrate to ensure the successful implementation of the project.

Claims

1. An image-driven thermal conductivity analysis method suitable for natural gas hydrate deposits, characterized in that... Includes the following steps: Step 1): The representative high-resolution images of natural gas hydrate sediments are processed into grids to convert the high-resolution images into matrix data. The multiphase volume fraction, soil particle morphology parameters and hydrate distribution pattern parameters are identified and quantified. Overlapping areas and geological heterogeneity parameters are processed according to priority rules. The multiphase volume fraction includes soil particle volume fraction, hydrate volume fraction, methane gas fraction and water volume fraction; the soil particle morphology parameters include core growth probability and growth probability in four directions; and the hydrate distribution mode parameters include the volume ratio of four distribution modes: bridging, encapsulation, suspension and blocky. Step 2): Initialize all the mesh cells in Step 1) to have the thermal conductivity of methane gas, generate soil particle core cells, and fill the internal voids and edge gaps according to the growth probability expansion morphology in the four directions, thus forming a geometrically closed soil particle model. Step 3): Perform transient temperature field simulation on the geometrically closed soil particle model in Step 2) to determine the hydrate distribution pattern. Based on the set temperature threshold, distinguish between boundary hydrates and liquid bridge hydrates, randomly generate blocky hydrates and suspended hydrates, integrate all grid element attributes, and output mesoscopic model data including thermal conductivity. Step 4): Import the microscopic model data from Step 3) into the finite element software, create representative unit cells and generate a structured mesh, set the temperature difference between the two ends of the structured mesh and the boundary conditions of the top and bottom adiabatic sides, and calculate the heat flux, temperature gradient and effective thermal conductivity through the steady-state heat transfer equation. Step 5): Combine the Monte Carlo method to iteratively generate the normal distribution statistical results of the effective thermal conductivity in Step 4), that is, the image-driven thermal conductivity analysis results of natural gas hydrate deposits.

2. The image-driven thermal conductivity analysis method according to claim 1, characterized in that: In step 2), the thermal conductivity of methane gas is assigned to each grid cell as the initial background medium; a soil particle core grid is randomly selected, a uniformly distributed random number matrix is ​​generated, and grids with random numbers less than the core growth probability are marked as soil particle core cells. The soil particle boundary is iteratively expanded according to the growth probability in four directions: non-soil particle cells are scanned, and if an adjacent cell in a certain direction is a soil particle and the random number is less than the growth probability in that direction, the cell is converted into a soil particle cell. This process is repeated until the total volume of soil particles reaches the volume content of soil particles per unit volume. By scanning all grids, the pores surrounding the soil particles and the pores surrounded by soil particles on three sides are found and filled accordingly to form a geometrically closed soil particle model.

3. The image-driven thermal conductivity analysis method according to claim 2, characterized in that: Each grid cell is assigned a methane thermal conductivity of 0.025 W / m·K. A soil particle core grid is randomly selected, and a uniformly distributed random number matrix with a value range of [0,1] is generated.

4. The image-driven thermal conductivity analysis method according to claim 1, characterized in that: In step 3), a constant temperature of 20℃ is set for the soil particle region and an initial temperature of 0℃ for the geometrically closed soil particle model. The surrounding boundaries are set to thermal insulation conditions. Isotherm coverage areas with temperatures greater than temperature threshold A=10℃ and less than temperature threshold B=12℃ are extracted. Non-solid units in contact with soil particle units in this area are marked as encapsulated hydrate units. Isotherm coverage areas with temperatures greater than temperature threshold B=12℃ are extracted, and the rectangular area with the smallest width in the direction of the line connecting the centers of two soil particles is identified. Non-solid units in this rectangular area are marked as bridging hydrate units. Isotherm coverage areas equal to temperature thresholds B and A are liquid bridge boundary hydrate and liquid bridge / boundary hydrate, respectively.

5. The image-driven thermal conductivity analysis method according to claim 1, characterized in that: In step 4), the minimum unit body side length is set to 40m, which is 1 / 1000 of the maximum length of the model, and the temperature difference between the two ends is set to 10℃.

6. The image-driven thermal conductivity analysis method according to claim 5, characterized in that: In step 4), the heat flux... Temperature gradient Effective thermal conductivity These represent the average temperatures at both ends of a representative unit cell; This represents the length of the representative unit cell in the direction of heat flow; Indicates heat flow; Indicates the first Heat flow of each unit cell; Indicates the number of unit cells; This represents the region of integration.

7. The image-driven thermal conductivity analysis method according to claim 1, characterized in that: In step 1), the content relationship of the four distribution patterns of hydrates per unit volume—bridging, encapsulation, suspension, and blocky—is as follows: + + + = , Indicates the content of bridging hydrates; Indicates the content of encapsulated hydrates; Indicates the content of suspended hydrates; Indicates the content of blocky hydrates. This indicates the volume fraction of hydrates per unit volume.

8. The image-driven thermal conductivity analysis method according to claim 1, characterized in that: Extract the complete dataset from the normal distribution statistics in step 5) as input parameters, calculate the Pearson correlation coefficient between each input parameter and the effective thermal conductivity, determine the negative and positive correlation factors, and sort all factors in descending order of the absolute value of the correlation coefficient.

9. The image-driven thermal conductivity analysis method according to claim 1, characterized in that: In step 5), set the Monte Carlo algorithm iteration threshold, execute the Monte Carlo method iteration, use the same input parameters for each iteration but update the random number seed, calculate the arithmetic mean and standard deviation of the effective thermal conductivity results of the iteration, and verify that the calculation results conform to the normal distribution characteristics.

10. The image-driven thermal conductivity analysis method according to any one of claims 1-9, characterized in that: The volume fractions per unit volume of the soil particles, water, methane gas, and hydrate contained in natural gas hydrates satisfy the following: in, This represents the volume fraction of soil particles per unit volume. It represents the volume fraction of water per unit volume; This represents the integral number of methane gas per unit volume. This indicates the volume fraction of hydrates per unit volume. The saturation levels of soil particles, water, and methane gas satisfy the following: in, This indicates the methane saturation level in the pores per unit volume. Indicates the saturation of hydrates in pores per unit volume; It indicates the degree of water saturation in the pores per unit volume. The relationship between volume fraction and saturation is as follows: in, It represents the volume fraction of pores per unit volume.

Images