A multi-point geostatistical modeling method and system

Abstract

The present invention discloses a multi-point geostatistical modeling method, including: constructing a training image and a simulation grid of a simulated work area, and obtaining the sedimentary sequences of all wells in the simulated work area; positioning all sedimentary sequences in the simulation grid, and determining initial parameters; selecting a point to be estimated in a simulation line, determining a data template specification and a data event for the point to be estimated according to the initial parameters, and traversing the training image according to the template specification to obtain a training pattern with the same data structure as the current data event; performing a similarity comparison between the data event and the training pattern, selecting the optimal training pattern and assigning the sedimentary sequence at the corresponding point therein to the point to be estimated; after completing the conditional simulation of all unknown points, generating a multi-point geostatistical model. The simulation results disclosed by the present invention are reliable and have high precision, and can realize sedimentary phase modeling of complex reservoirs.

Description

Technical Field

[0001] The invention relates to the field of oil and gas field development geological technology, and in particular to a multi-point geostatistical modeling method and system. Background Art

[0002] At present, there are many types of oil and gas reservoirs being developed, and the reservoir characteristics are complex. Conventional modeling methods (mainly two-point geostatistical modeling methods based on variograms and target-based geological modeling methods) are difficult to establish models that conform to geological knowledge. Multi-point geostatistical modeling methods use "training images" to obtain geological information, and have become an important means of quantitatively characterizing reservoirs in recent years.

[0003] The existing multi-point geostatistical modeling technology is mainly divided into: probability-based multi-point geostatistical modeling method and similarity-based multi-point geostatistical modeling method. Among them, the probability-based multi-point geostatistical modeling method simulates the center point of the data event by probability, and the modeling effect is significantly improved compared with the traditional modeling method, but it is easy to cause discontinuity of the target body. The multi-point geostatistical modeling method based on similarity compares the similarity between the training pattern and the data event to simulate the center point of the data event or replace the data template as a whole. This type of method uses the data template as a whole for similarity comparison, which improves the simulation effect of the continuity of the target body.

[0004] In multi-point geostatistical phase modeling, how to determine the value of the point to be estimated and reduce the uncertainty of modeling has always been the key to the success of the multi-point geostatistical modeling algorithm. The above-mentioned multi-point geostatistical modeling algorithm based on similarity, whose algorithm framework uses a data template of a certain dimension as the carrier and uses Euclidean distance as the similarity judgment standard, still has deficiencies in the degree of integration of the modeling model with the geological meaning and the sedimentary significance of the reservoir. For the reservoir space with sedimentary phases with clear sedimentary significance and sequence, the existing multi-point geostatistical modeling method is difficult to solve the modeling problem of complex reservoir sedimentary phases.

[0005] Therefore, there is a need in the prior art to provide a new multi-point geostatistical modeling method so that the method can be applicable to reservoir spaces with complex reservoir sedimentary phase characteristics. Summary of the invention

[0006] In order to solve the above technical problems, the present invention provides a multi-point geostatistical modeling method, comprising: constructing a training image and a simulation grid of a simulated work area, and obtaining the sedimentary sequences of all wells in the simulated work area; positioning all sedimentary sequences in the simulation grid, and determining initial parameters, wherein the initial parameters include a template shape, a number of conditional points, and a simulation line composed of all unknown points on the top surface of the simulation grid; selecting a point to be estimated in the simulation line, determining a data template specification and a corresponding data event for the point to be estimated according to the initial parameters, and traversing the training image according to the template specification to obtain a training pattern having the same data structure as the current data event; comparing the data event with the training pattern for similarity, selecting the optimal training pattern and assigning the sedimentary sequences at the corresponding points therein to the point to be estimated; after completing the conditional simulation of all unknown points, generating a multi-point geostatistical model containing reservoir sedimentary phase distribution characteristics.

[0007] Preferably, in the process of determining the data template specifications for the point to be estimated based on the initial parameters, it includes: taking the point to be estimated as the center, scaling the data template according to the template shape, and determining the minimum specification corresponding to the number of known points that can cover the conditional points as the data template specification.

[0008] Preferably, in the step of comparing the similarity between the data event and the training pattern, it includes: setting a corresponding weight for each deposition phase in all deposition sequences; screening out the deposition sequences corresponding to the first type of top surface known points in the data event and the deposition sequences corresponding to the second type of top surface known points in the training pattern that are at the same position as the first type of top surface known points, and based on this, in combination with the weights, calculating the similarity between the deposition sequence corresponding to each first type of top surface known point and the deposition sequence corresponding to the second type of top surface known point at the corresponding position, recorded as the intra-group similarity; after summing up all the intra-group similarities, generating the comparison result of the current training pattern.

[0009] Preferably, the training pattern to which the smallest comparison result belongs is selected as the optimal training pattern.

[0010] Preferably, in the process of calculating the intra-group similarity, it includes: determining the weight of the sedimentary phase at each sequence position in the sedimentary sequence corresponding to the current first-type top surface known point, and the weight of the sedimentary phase at each sequence position in the sedimentary sequence corresponding to the second-type top surface known point, based on which, a dynamic programming method is used to calculate the intra-group similarity.

[0011] Preferably, the method further comprises: determining whether the conditional simulation of all unknown points in the simulation line is completed; if not completed, selecting a new point to be estimated to perform conditional simulation on the current new point to be estimated.

[0012] On the other hand, the present invention also provides a multi-point geostatistical modeling system, comprising: an information acquisition module, which is configured to construct a training image and a simulation grid of a simulated work area, and obtain the sedimentary sequences of all wells in the simulated work area; an initialization configuration module, which is configured to locate all sedimentary sequences in the simulation grid, and determine initial parameters, wherein the initial parameters include a template shape, a number of conditional points, and a simulation line composed of all unknown points on the top surface of the simulation grid; a training pattern generation module, which is configured to select a point to be estimated in the simulation line, determine a data template specification and a corresponding data event for the point to be estimated according to the initial parameters, and traverse the training image according to the template specification to obtain a training pattern with the same data structure as the current data event; a point assignment module to be estimated, which is configured to compare the data event with the training pattern for similarity, select the optimal training pattern and assign the sedimentary sequence at the corresponding point therein to the point to be estimated; and a model generation module, which is configured to generate a multi-point geostatistical model containing reservoir sedimentary phase distribution characteristics after completing the conditional simulation of all unknown points.

[0013] Preferably, the initialization configuration module is further configured to scale the data template according to the template shape with the point to be estimated as the center, and determine the minimum specification corresponding to the number of known points that can cover the conditional points as the data template specification.

[0014] Preferably, the module for assigning values ​​to points to be estimated includes: a similarity comparison submodule, which includes: a sedimentary phase weight setting unit, which is configured to set a corresponding weight for each sedimentary phase in all sedimentary sequences; an intra-group similarity generation unit, which is configured to screen out the sedimentary sequences corresponding to the first type of top surface known points in the data event, and the sedimentary sequences corresponding to the second type of top surface known points in the training pattern that are at the same position as the first type of top surface known points, and based on this, combined with the weight, calculate the similarity between the sedimentary sequence corresponding to each first type of top surface known point and the sedimentary sequence corresponding to the second type of top surface known point at the corresponding position, which is recorded as the intra-group similarity; a comparison result generation unit, which is configured to generate the comparison result of the current training pattern after summing all the intra-group similarities.

[0015] Preferably, the intra-group similarity generating unit is further configured to determine the weight of the sedimentary phase at each sequence position in the sedimentary sequence corresponding to the current first-type top surface known point, and the weight of the sedimentary phase at each sequence position in the sedimentary sequence corresponding to the second-type top surface known point, and based on this, a dynamic programming method is used to calculate the intra-group similarity.

[0016] Compared with the prior art, one or more embodiments of the above scheme may have the following advantages or beneficial effects:

[0017] The present invention discloses a multi-point geostatistical modeling method and system. The method and system integrate sedimentary sequences into the algorithm, so that the sedimentary phase simulation of the point to be estimated has an intrinsic assumption of sedimentation. Therefore, from the principle and design of the algorithm, it can better reflect the actual geological meaning than the existing multi-point geostatistical modeling algorithm, the simulation results are reliable and more accurate, and the sedimentary phase modeling of complex reservoirs can be realized.

[0018] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0020] Figure 1 It is a step diagram of the multi-point geostatistical modeling method of an embodiment of the present application.

[0021] Figure 2 It is a flow chart of the multi-point geostatistical modeling method of an embodiment of the present application.

[0022] Figure 3 It is a schematic diagram of the principle of selecting the data template specification process in the multi-point geostatistical modeling method of the embodiment of the present application.

[0023] Figure 4 It is a schematic diagram of the principle of the intra-group similarity calculation process in the multi-point geostatistical modeling method of the embodiment of the present application.

[0024] Figure 5 It is a schematic diagram of the training image of the first example in the multi-point geostatistical modeling method of an embodiment of the present application.

[0025] Figure 6 It is a schematic diagram of a single-well sedimentary sequence image of the first example in the multi-point geostatistical modeling method of an embodiment of the present application.

[0026] Figure 7 It is a schematic diagram of a multi-point geostatistical model of the first example in the multi-point geostatistical modeling method of an embodiment of the present application.

[0027] Figure 8 It is an overall schematic diagram and fence diagram of the training image of the second example in the multi-point geostatistical modeling method of an embodiment of the present application.

[0028] Fig. 9It is an overall schematic diagram and fence diagram of a multi-point geostatistical model generated by the second example in the multi-point geostatistical modeling method of an embodiment of the present application when the number of conditional points is 3.

[0029] Fig.10 It is an overall schematic diagram and fence diagram of a multi-point geostatistical model generated by the second example in the multi-point geostatistical modeling method of an embodiment of the present application when the number of conditional points is 4.

[0030] Fig.11 It is a module block diagram of the multi-point geostatistical modeling system of an embodiment of the present application. DETAILED DESCRIPTION

[0031] The following will describe the implementation methods of the present invention in detail with reference to the accompanying drawings and embodiments, so that the implementation process of how the present invention applies technical means to solve technical problems and achieve technical effects can be fully understood and implemented accordingly. It should be noted that as long as there is no conflict, the various embodiments of the present invention and the various features in the embodiments can be combined with each other, and the technical solutions formed are all within the protection scope of the present invention.

[0032] At present, there are many types of oil and gas reservoirs being developed, and the reservoir characteristics are complex. Conventional modeling methods (mainly two-point geostatistical modeling methods based on variograms and target-based geological modeling methods) are difficult to establish models that conform to geological knowledge. Multi-point geostatistical modeling methods use "training images" to obtain geological information, and have become an important means of quantitatively characterizing reservoirs in recent years.

[0033] The existing multi-point geostatistical modeling technology is mainly divided into: probability-based multi-point geostatistical modeling method and similarity-based multi-point geostatistical modeling method. Among them, the probability-based multi-point geostatistical modeling method simulates the center point of the data event by probability, and the modeling effect is significantly improved compared with the traditional modeling method, but it is easy to cause discontinuity of the target body. The multi-point geostatistical modeling method based on similarity compares the similarity between the training pattern and the data event to simulate the center point of the data event or replace the data template as a whole. This type of method uses the data template as a whole for similarity comparison, which improves the simulation effect of the continuity of the target body.

[0034] In multi-point geostatistical phase modeling, how to determine the value of the point to be estimated and reduce the uncertainty of modeling has always been the key to the success of the multi-point geostatistical modeling algorithm. The above-mentioned multi-point geostatistical modeling algorithm based on similarity, whose algorithm framework uses a data template of a certain dimension as the carrier and uses Euclidean distance as the similarity judgment standard, still has deficiencies in the degree of integration of the modeling model with the geological meaning and the sedimentary significance of the reservoir. For the reservoir space with sedimentary phases with clear sedimentary significance and sequence, the existing multi-point geostatistical modeling method is difficult to solve the modeling problem of complex reservoir sedimentary phases.

[0035] Therefore, in order to solve the above technical problems, the present invention proposes a multi-point geostatistical modeling method and system based on sedimentary sequences. The method is based on the multi-point geostatistical theory. First, a training image and a simulation grid are established for the reservoir area to be simulated; the sedimentary sequences corresponding to all single wells in the training image are located in the simulation grid to form a simulation line composed of all unknown points on the top surface of the simulation network; in the process of conditional simulation of unknown points, the sedimentary sequences of known points around the unknown points are compared with the sedimentary sequences at corresponding positions in the training graph obtained after traversing the training image to select the optimal training graph; finally, the sedimentary sequence with the same position point as the unknown point in the optimal training graph is matched to the unknown point position in the simulation network to complete the conditional simulation of the sedimentary sequence of the current unknown point, so as to generate a multi-point geostatistical model containing reservoir sedimentary phase distribution characteristics after completing the conditional simulation of all unknown points.

[0036] In this way, the present invention takes into account the sedimentary sequence characteristics and geological significance of the reservoir phase, adopts the similarity comparison method of the sedimentary phase sequence, improves the ability to combine the multi-point geostatistical modeling method with geology, and forms a multi-point geostatistical modeling method that can take into account the sedimentary significance of the reservoir phase, thereby improving the accuracy of multi-point geostatistical phase modeling, and can model the reservoir space with complex reservoir sedimentary phase characteristics.

[0037] Figure 1 It is a step diagram of the multi-point geostatistical modeling method of an embodiment of the present application. Figure 2 This is a flow chart of the multi-point geostatistical modeling method of the present application embodiment. Figure 1 and Figure 2 The multi-point geostatistical modeling method described in the present invention is specifically described.

[0038] Step S110 constructs a training image and a simulation grid of the simulation work area, and obtains the sedimentary sequence of all wells in the simulation work area. It should be noted that in the embodiment of the present invention, the simulation work area is a partial area within the entire oil and gas field reservoir when a certain oil and gas field reservoir needs to be geostatistically modeled. The present invention does not specifically limit the scale of the simulation work area, and those skilled in the art can set it according to actual needs, the geological structure of the oil and gas field reservoir, the accuracy of the geostatistical model and other factors.

[0039] In step S110, it is necessary to construct a training image containing geological sedimentary phase information of the simulated work area. The training image is a gridded image containing reservoir sedimentary laws, geological body geometry and distribution characteristics, and each grid corresponds to the sedimentary data of the actual geological conditions to which the grid belongs. Among them, the sedimentary data at least includes: the location information of the current grid, the sedimentary phase type and the corresponding value. In this way, the training image can record the geological meaning information of the oil and gas field reservoir in an all-round way. Since establishing training images is a relatively independent technology in multi-point geological modeling technology, it will not be described here. Figure 2 and Figure 8 As shown, the longitudinal section image, stereo image and raster image of the training image are shown respectively.

[0040] In addition, the simulation grid constructed in the embodiment of the present invention is constructed by a plurality of grid blocks, and the specifications of the simulation grid (number of grids, unit grid volume and size, overall volume and size, etc.) do not need to be consistent with the scale of the training image. In the embodiment of the present invention, those skilled in the art can determine the specifications of the simulation grid according to the actual working area and the modeling target area.

[0041] Furthermore, in step S110, it is also necessary to determine all the single wells in the simulation work area and obtain the single well deposition sequence of each single well. It should be noted that since the single well in the simulation work area may be a partial area of ​​the whole single well, it is not necessary to obtain the complete deposition sequence of the whole single well. It is only necessary to obtain the deposition sequence corresponding to the depth section of the single well in the simulation work area according to the depth position of the simulation work area. The deposition sequence of a single well is composed of deposition data at different depth positions, and the deposition data at different depth positions are arranged in order from shallow to deep. Figure 6 As shown, the single well sedimentation sequence of each single well in the simulated work area is displayed. A set of single well sedimentation sequences contains sedimentation data corresponding to grids at different depths.

[0042] In this way, after the basic data and model required for the geostatistical model construction process are completed, the process proceeds to step S120. Step S120 locates all sedimentary sequences in the simulation grid and determines initial parameters.

[0043] In step S120, first, the single well deposition sequence corresponding to all single wells in the simulation work area needs to be located in the simulation grid. In the actual application process, since the simulation grid is a partial area in the oil field reservoir, in the construction of the simulation grid, it is necessary to record the position information (depth position and horizontal coordinate position) of each grid block in the current simulation grid in the actual oil and gas field reservoir. At the same time, since in the deposition sequence of all wells in the simulation work area obtained from step S110, the deposition data at different depths of each well are marked with the position information (depth position and horizontal coordinate position) of the position. Therefore, according to the position information of the deposition data and the position information of each grid block in the simulation grid, the deposition data at different depths in the single well deposition sequence of all single wells in the simulation work area are located one by one to the grids at the corresponding positions in the simulation grid. At this time, the grids at different positions in the simulation grid model (of the simulation work area) are written with corresponding deposition data, thereby determining the grids with written deposition data as known grid points, and determining the grids without written deposition data as unknown grid points.

[0044] It should be noted that since the single-well deposition sequence is a sequential combination of deposition data of a well at different depths, after the single-well deposition sequence is assigned to the simulation grid model, it appears as a "columnar" data strip in the simulation grid, and the deposition data at the same position is configured in the depth position corresponding to each grid block.

[0045] Further, after completing the known point allocation operation of the simulation grid, it is necessary to determine the initial (configuration) parameters required for the subsequent unknown point conditional simulation experiment. The initial parameters include, but are not limited to: template shape, number of conditional points, and simulation circuits composed of all unknown points on the top surface of the simulation grid. In an embodiment of the present invention, the template shape can be a cube, a cuboid, or the like of a certain specification.

[0046] The number of conditional points is the number of known points distributed around the unknown point when the data template is used to simulate the unknown point. The larger the number, the greater the amount of calculation required for the unknown point simulation, but the higher the accuracy of the simulation result. Conversely, the smaller the number, the smaller the amount of calculation required for the unknown point simulation, but the accuracy of the simulation result will decrease accordingly.

[0047] The simulation line is set on the top surface of the simulation grid. Specifically, in the actual unknown point simulation process, if a certain unknown point completes its own conditional simulation, then the unknown point will record the corresponding deposition sequence after the simulation, thereby converting it into a known point. In this way, based on the above principle, when planning the simulation line, it is necessary to plan the simulation order of all unknown points on the top surface of the simulation grid model according to the relative position relationship between all unknown points and all known points, forming a simulation line marked with the simulation order of the top surface unknown points, and mark the corresponding simulation sequence number for each unknown point in the line.

[0048] Therefore, after completing the initial parameter configuration, the process proceeds to step S130. Step S130 selects a point to be estimated in the simulated line, determines the data template specification and corresponding data event for the point to be estimated according to the initial parameters, and traverses the training image according to the data template specification to obtain a training pattern with the same data structure as the current data event.

[0049] In step S130, first (step S1301, not shown) it is necessary to select the unprocessed point to be estimated with the smallest simulation sequence number in the simulation line as the object of the current conditional simulation. Then, (step S1302, not shown) determine the data template specifications for the current point to be estimated according to the template shape and the number of conditional points in the initial parameters. Specifically, in step S1302, with the current point to be estimated as the center, the data template is scaled (proportionally) according to the determined template shape, and the specifications of the data template with the smallest specifications are determined as the data template specifications required for the conditional simulation processing of the current point to be estimated when covering the known grid points of the conditional points. Among them, the data template size includes: the number of grids contained in the long side, the number of grids contained in the wide side, and the number of grids contained in the high side (the number of grids on the high side can be 1 or more).

[0050] Figure 3 Schematic diagram of the selection principle of the data template specification process in the multi-point geostatistical modeling method of the embodiment of the present application. Figure 3 As shown, Figure 3 (a) shows an example of the top surface of the simulation grid of the simulation area, that is, the top view of the simulation grid. In the top surface of the simulation grid, there are sedimentary data (4 points) of 4 known wells, and it is assumed that the number of current condition points is 4. Figure 3 The light gray area in (b) shows the top surface size of the 3*3 data template. The data template does not cover the 4 known points of the top surface, and therefore does not meet the current conditional point quantity condition. Figure 3 (c) shows that the size of the top surface of the data template obtained after expanding the data template is 5*5. After the number of conditional points is met, the final size of the top surface of the data template is determined to be 5*5.

[0051] After determining the data template specifications, (step S1303, not shown), continue to determine the data events corresponding to the current data template specifications. The current data event is centered on the current point to be estimated, including: a data structure composed of several vectors formed by the current point to be estimated and the number of known grid points of the surrounding conditional points, and the deposition data of each grid point. Among them, the number of vectors is consistent with the number of conditional points. The data structure records: the geometric position relationship formed between each known grid point and the current point to be estimated (including: the distance between each known grid point to the current point to be estimated, and the angle between adjacent vectors and other information), and the deposition type corresponding to each known grid point. In this way, the determination step of the current data event is completed, and then enters step S1304 (not shown).

[0052] Further, step S1304 will traverse the training image constructed in the entire step S130 according to the data template specifications determined in step S1302, and obtain several training patterns with the same data structure as the current data event. Among them, the specifications of the training pattern are consistent with the data template specifications. Specifically, when searching for training patterns with the same data structure, it is necessary to detect whether there are conditional points around the center of each training pattern that are consistent with the sedimentary phase type and geometric position relationship of each known grid point in the current data structure. If yes, the current training pattern is intercepted as an alternative condition for subsequent conditional simulation. If not, the next training pattern is tested. In this way, after searching all training patterns, enter step S140.

[0053] Step S140 compares the data event determined in the current step S130 with the training pattern for similarity, selects the best training pattern, and allocates the deposition sequence at the corresponding point in the best training pattern to the point to be estimated.

[0054] In step S140, first (step S1401, not shown) the data event determined in step S130 and several training patterns that meet the candidate conditions screened out in step S130 are subjected to similarity comparison processing respectively; (step S1402, not shown) based on the similarity comparison processing result, the optimal training pattern is selected, and the deposition sequence at the corresponding position of the optimal training pattern is allocated to the current point to be estimated.

[0055] The following is a detailed description of the similarity comparison process.

[0056] S1. It is necessary to set a corresponding weight for each sedimentary phase (type) involved in the sedimentary sequence of all known wells obtained in step S110. For example: Figure 4(b) There are four types of sediments in the simulated work area, namely A, B, C, and D. Then, corresponding weights need to be set for these four types of sediments, that is, the weight of sediment A is set to a, the weight of sediment B is set to b, the weight of sediment C is set to c, and the weight of sediment D is set to d.

[0057] Next, after completing the setting of the sedimentary phase weights, S2 selects the sedimentary sequence corresponding to each first-type top surface known point in the top surface of the data template in the current data event (recorded as the first sedimentary sequence), and the sedimentary sequence corresponding to each second-type top surface known point in the top surface of the training pattern that is at the same position as the first-type top surface known point (recorded as the second sedimentary sequence). Based on this, combined with the weight value set in step S1401, the similarity between the first sedimentary sequence corresponding to each first-type top surface known point and the second sedimentary sequence corresponding to the second-type top surface known point at the corresponding position is calculated, thereby obtaining the sedimentary sequence for each group of known points at the same position, and obtaining the corresponding intra-group similarity (calculation result).

[0058] Specifically, in the first step, each known point in the top surface of the current data template is recorded as a first-class top surface known point (with a conditional point number), and the deposition sequence of each first-class top surface known point is extracted. It should be noted that what is extracted here is not only the deposition data corresponding to the first-class top surface known grid point, but the (first) deposition sequence formed by the deposition data at a series of different depth positions below the first-class top surface known grid point as the starting point in the data template. The number of the first deposition sequence here matches the number of grids contained in the high side of the data template specification.

[0059] Then, in the second step, the data points in the top surface of the current training figure that are in the same position as the first type of top surface known points are marked as the second type of top surface known points (with the number of conditional points), and the deposition sequence of each second type of top surface known point is extracted. It should be noted that what is extracted here is not only the deposition data corresponding to the second type of top surface known grid points, but the (second) deposition sequence formed by the deposition data at the starting point and a series of different depth positions below the starting point of the second type of top surface known grid points in the current training figure. The number of the second deposition sequence here matches the number of grids contained in the high side of the current training figure.

[0060] Then, since each first-class top surface known point in the data template has a corresponding second-class top surface known point at the same position of the training pattern, that is, for each first deposition sequence in the data template, there will also be a corresponding second deposition sequence at the same position of the training pattern. Therefore, in the third step, based on the first deposition sequences corresponding to all first-class top surface known points (respectively) and the second deposition sequences corresponding to all second-class top surface known points (respectively), combined with the weights of all depositional phase types involved in the simulation work area, the similarity (i.e., the distance between the first deposition sequence and the second deposition sequence) between the first deposition sequence corresponding to each first-class top surface known point and the second deposition sequence corresponding to the second-class top surface known point at the corresponding (same) position is calculated. In this way, for each group of first deposition sequences and second deposition sequences at the same position, the corresponding similarity calculation result is calculated, which is the intra-group similarity, thereby obtaining the number of conditional points intra-group similarities.

[0061] Furthermore, in an embodiment of the present invention, when calculating the similarity within each group, it is necessary to first determine the weight of the deposition phase at each sequence position in the current first deposition sequence, and the weight of the deposition phase at each sequence position in the second deposition sequence, and then, based on these weight data, a dynamic programming method is used to calculate the corresponding intra-group similarity data.

[0062] Figure 4 Schematic diagram of the principle of the process of calculating the similarity within a group in the multi-point geostatistical modeling method of the embodiment of the present application. Figure 4 The calculation principle of the intra-group similarity is explained. Figure 4 The left and right sequences of (a) are examples of the first deposition sequence and the second deposition sequence to be compared at the same position, respectively. Figure 4 (b) setting weights of the deposition phases involved in the first deposition sequence and the second deposition sequence, Figure 4 (c) shows the similarity comparison process of two sedimentary sequences.

[0063] In the dynamic programming method, if the first deposition sequence is specified as a sequence in which the data contents at different depth positions are arranged in a first order from shallow to deep, and the second deposition sequence is specified as a sequence in which the data contents at different depth positions are arranged in a second order from shallow to deep. The intra-group similarity is the weighted cost required to convert the data contents at different depth positions in the first deposition sequence into a sequence with the same data content and arrangement order as the second deposition sequence. In actual application, since different conversion methods (conversion paths) are used in the conversion process, for two deposition sequences at the same position, when calculating the distance between the two sequences, several total weighted cost values ​​will be obtained. At this time, the minimum total weighted cost value can be used as the intra-group similarity data.

[0064] For example, if Figure 4 As shown in (c), the first column in the table represents the data content of the first deposition sequence at different depths: A, C, B, and arranged according to positions 1, 2, and 3, and each data content has a corresponding weight a, c, and b; the last column in the table represents the data content of the second deposition sequence at different depths: A, B, C, D, and arranged according to positions 1, 2, 3, and 4, and each data content has a corresponding weight a, b, c, and d. Among them, the theoretical formula of the distance between the first deposition sequence and the second deposition sequence is expressed by the following expression:

[0065] d[i,j]=min{[d[i-1,j]+insertCost(target i )],

[0066] [d[il,j-1]]+substituteCost(source j , target i )],

[0067] [d[i,j-1]+deleteCost(source j )]}

[0068] Where d[i,j] represents the distance between the first deposition sequence j and the second deposition sequence i, insertCost(target i ) represents the weight cost value generated by the insertion action performed during the sequence conversion process, substituteCost(source j ,target i ) represents the weight cost value generated by the replacement action performed during the sequence conversion process, deleteCost(source j ) represents the weight cost value generated by the deletion action performed during the sequence transformation process.

[0069] For example, refer to Figure 4(c), if you want to convert from a sequence A containing only data to a sequence A containing only data, then the weight cost value to be paid is 0. If you want to convert from a sequence C containing only data to a sequence B containing only data, then the weight cost value to be paid is b+c (delete C and insert B). If you want to convert from a sequence A, C, B to a sequence A, B, C, there are two paths. The first path is to delete C first and then insert C. The weight cost value to be paid for this path is 2c. The second path is to delete B first and then insert B. The weight cost value to be paid for this path is 2b. Further, in an embodiment of the present invention, if you want to convert from a sequence A, C, B to a sequence A, B, C, D, there are at least two paths. The first path is to delete C first, then insert C, and then insert D. The weight cost value to be paid for this path is 2c+d; the second path is to delete B first, then insert B, and then insert D. The weight cost value to be paid for this path is 2b+d.

[0070] So, reference Figure 4 (c) The present invention obtains two different total weighted cost values ​​(see the content of the upper right corner grid in Table 4c) when calculating the distance between two deposition sequences at the same position through the above dynamic programming method, which are: 2b+d and 2c+d. In the actual application process, the minimum total weighted cost value is taken as the similarity calculation result of the two deposition sequences at the same position, that is, the intra-group similarity.

[0071] Finally, after completing the calculation of sequence similarity of multiple locations at the same position, the corresponding intra-group similarity data is obtained for each group of two deposition sequences at the same position. Therefore, S3 sums up all the intra-group similarities to generate the similarity comparison result of the current training pattern.

[0072] Furthermore, through the above steps S1 to S3, corresponding similarity comparison results are obtained for each training pattern used as an alternative condition. Therefore, step S1402 will first select the optimal training pattern based on these similarity comparison results. Specifically, in step S1402, the training pattern to which the smallest similarity comparison result belongs is selected as the optimal training pattern. Then, according to the position information of the point to be estimated in the current data template, the grid point with the same position as the current point to be estimated is determined in the optimal training pattern and recorded as the target grid point. Finally, the deposition sequence at the target grid point is extracted from the optimal training pattern and the deposition data therein is assigned one by one to the position of the point to be estimated, so as to assign deposition data to the grid points at different depths of the point to be estimated and below it.

[0073] In this way, through the above steps S130 and S140, the conditional simulation of the current point to be estimated is completed and the deposition sequence is assigned, and then the process proceeds to step S150.

[0074] After completing the conditional simulation of all unknown points, step S150 generates a multi-point geostatistical model containing the reservoir sedimentary phase distribution characteristics. In step S150, first, it is necessary to determine whether the conditional simulation task of all unknown points in the simulation line formed in step S120 has been completed. If there are still unknown points that have not been subjected to conditional simulation, then return to step S130 and select a new point to be estimated with the smallest sequence number that has not been subjected to conditional simulation to perform conditional simulation on the current new point to be estimated. In addition, if there are no unknown points that have not been subjected to conditional simulation in the simulation line, then it means that all unknown points have completed the conditional simulation, and all grid blocks in the current simulation grid model have corresponding sedimentary data, thereby completing the task of building a multi-point geostatistical model.

[0075] At this time, the multi-point geostatistical model constructed by the present invention has complete sedimentary phase distribution characteristic information of the reservoir in the simulated work area. The embodiment of the present invention integrates the sedimentary sequence into the algorithm, so that the simulation of the sedimentary phase of the estimated point has the inherent assumption of sedimentation, rather than just "taking it out of context" of the existing algorithm. Therefore, from the principle and design of the algorithm, the multi-point geostatistical model constructed by the present invention can better reflect the true geological meaning than the existing multi-point geostatistical modeling algorithm.

[0076] The multi-point geostatistical modeling method described in the embodiments of the present invention is applied to the following two embodiments:

[0077] First example

[0078] A training image TI (grid size: 200*1*45) of the simulated work area was established, and a simulation grid (same size as the training image) was created. There were 5 single wells in the simulated work area, and the number of conditional points was set to 4. Conditional simulation was performed on the simulated work area. Figure 5 It is a schematic diagram of the training image of the first example in the multi-point geostatistical modeling method of an embodiment of the present application. Figure 5 The longitudinal profile of the current training image is shown, and areas with different grayscale levels represent different sedimentary facies types. Figure 6 It is a schematic diagram of a single-well sedimentary sequence image of the first example in the multi-point geostatistical modeling method of an embodiment of the present application. Figure 6 The sedimentary characteristics represented by the sedimentary sequences of 5 single wells in the work area are displayed respectively. Figure 7 It is a schematic diagram of a multi-point geostatistical model of the first example in the multi-point geostatistical modeling method of an embodiment of the present application. Figure 7 Longitudinal sections of the current multi-site geostatistical model are shown.

[0079] Second example

[0080] A training image TI (grid size: 69*69*39) of the simulated work area was established, and a simulation grid (same size as the training image) was created. There were 7 wells in the simulated work area, and the number of conditional points was set to 3 and 4, and conditional simulation was performed on the simulated work area. Figure 8 It is an overall schematic diagram and fence diagram of the training image of the second example in the multi-point geostatistical modeling method of an embodiment of the present application. Figure 8 (a) shows the overall schematic diagram of the current training image. Figure 8 (b) shows the raster image of the current training image, where areas with different grayscale levels represent different sedimentary phase types.

[0081] Fig. 9 It is an overall schematic diagram and fence diagram of a multi-point geostatistical model generated by the second example in the multi-point geostatistical modeling method of an embodiment of the present application when the number of conditional points is 3. Fig. 9 (a) shows the overall schematic diagram of the multi-point geostatistical model generated when the number of conditional points is 3. Fig. 9 (b) shows the raster image of the multi-point geostatistical model generated when the number of conditional points is 3. The areas with different grayscales represent different sedimentary facies types.

[0082] Fig.10 It is an overall schematic diagram and fence diagram of a multi-point geostatistical model generated in the second example of the multi-point geostatistical modeling method in an embodiment of the present application when the number of conditional points is 4. Fig.10 (a) shows the overall schematic diagram of the multi-point geostatistical model generated when the number of conditional points is 4. Fig.10 (b) shows a grid diagram of a multi-point geostatistical model generated when the number of conditional points is 4. Regions with different grayscales represent different sedimentary facies types. By comparing the multi-point geostatistical models generated when the number of conditional points is 3 and 4, it can be seen that increasing the number of conditional points can improve the simulation effect of the present invention.

[0083] Since the present invention incorporates the geological significance of the sedimentary sequence in the algorithm's similarity comparison, the new method has greatly improved the accuracy of phase modeling and can better reproduce the geometric shape and distribution characteristics of the geological body displayed in the training image. The application of the present invention for phase modeling is feasible and reliable.

[0084] On the other hand, based on the above-mentioned multi-point geostatistical modeling method, the present invention also proposes a multi-point geostatistical modeling system (hereinafter referred to as "modeling system"). Fig.11 is a block diagram of the multi-point geostatistical modeling system of the embodiment of the present application. Fig.11As shown, the above-mentioned modeling system includes: an information acquisition module 10, an initialization configuration module 20, a training graph generation module 30, a to-be-estimated point assignment module 40 and a model generation module 50.

[0085] Further, the information acquisition module 10 is implemented according to the method described in step S110 above, and is configured to construct a training image and a simulation grid of the simulated work area, and obtain the deposition sequence of all wells in the simulated work area. The initialization configuration module 20 is implemented according to the method described in step S120 above, and is configured to locate all deposition sequences in the simulation grid, and determine the initial parameters. Among them, the initial parameters include the template shape, the number of conditional points, and the simulation line composed of all unknown points on the top surface of the simulation grid. The training pattern generation module 30 is implemented according to the method described in step S130 above, and is configured to select the point to be estimated in the simulation line, determine the data template specifications and corresponding data events for the point to be estimated according to the initial parameters, and traverse the training image according to the template specifications to obtain a training pattern with the same data structure as the current data event. The point assignment module 40 to be estimated is implemented according to the method described in step S140 above, and is configured to compare the current data event with the training pattern for similarity, select the optimal training pattern and assign the deposition sequence at the corresponding point therein to the point to be estimated. The model generation module 50 is implemented according to the method described in step S150 above, and is configured to generate a multi-point geostatistical model containing reservoir sedimentary phase distribution characteristics after completing the conditional simulation of all unknown points.

[0086] Furthermore, the initialization configuration module 20 is further configured to scale the data template according to the template shape with the current point to be estimated as the center, and determine the minimum specification corresponding to the number of known points of conditional points as the data template specification.

[0087] Further, the module 40 for assigning values ​​to points to be estimated includes at least a similarity comparison submodule 41. The similarity comparison submodule 41 includes a sedimentary phase weight setting unit 411, an intra-group similarity generation unit 412, and a comparison result generation unit 413. The sedimentary phase weight setting unit 411 is configured to set a corresponding weight for each sedimentary phase in all sedimentary sequences. The intra-group similarity generation unit 412 is configured to filter out the sedimentary sequence corresponding to the first type of top surface known point in the current data event, and the sedimentary sequence corresponding to the second type of top surface known point in the training pattern at the same position as the first type of top surface known point. Based on this, combined with the weight, the similarity between the sedimentary sequence corresponding to each first type of top surface known point and the sedimentary sequence corresponding to the second type of top surface known point at the corresponding position is calculated, which is recorded as the intra-group similarity. The comparison result generation unit 413 is configured to sum up all the intra-group similarities and generate the comparison result of the current training pattern.

[0088] Furthermore, the intra-group similarity generation unit 412 is further configured to determine the weight of the sedimentary phase at each sequence position in the sedimentary sequence corresponding to the current first-type top surface known point, and the weight of the sedimentary phase at each sequence position in the sedimentary sequence corresponding to the second-type top surface known point. Based on this, a dynamic programming method is used to calculate the intra-group similarity.

[0089] The present invention discloses a multi-point geostatistical modeling method and system. The method and system start from the sedimentary significance of phase modeling, take the multi-point geostatistical theory as a guide, scan the simulation grid and training image through a variable data template, determine the corresponding data event and the training pattern of the corresponding data event; use the dynamic programming method to calculate the similarity between the data event and the corresponding training pattern; then, use the sedimentary sequence at the corresponding point of the most similar training pattern to assign a value to the data event to be estimated point, so as to obtain a simulation realization after traversing the unsimulated grid on the top surface of the model grid once. The present invention integrates the sedimentary sequence into the algorithm, so that the sedimentary phase simulation of the point to be estimated has the inherent assumption of sedimentation, rather than just the "out of context" of the previous algorithm. Therefore, from the principle and design of the algorithm, it can better reflect the actual geological meaning than the existing multi-point geostatistical modeling algorithm, the simulation result is reliable and more accurate, and the sedimentary phase modeling of complex reservoirs can be realized.

[0090] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by any person familiar with the technology within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0091] It should be understood that the embodiments disclosed in the present invention are not limited to the specific structures, processing steps or materials disclosed herein, but should be extended to equivalent substitutions of these features understood by ordinary technicians in the relevant field. It should also be understood that the terms used herein are only used for the purpose of describing specific embodiments and are not meant to be limiting.

[0092] The "one embodiment" or "embodiment" mentioned in the specification means that the specific features, structures or characteristics described in conjunction with the embodiment are included in at least one embodiment of the present invention. Therefore, the phrases "one embodiment" or "embodiment" appearing in various places throughout the specification do not necessarily refer to the same embodiment.

[0093] Although the embodiments disclosed in the present invention are as above, the contents described are only embodiments adopted for facilitating the understanding of the present invention and are not intended to limit the present invention. Any technician in the technical field to which the present invention belongs can make any modifications and changes in the form and details of the implementation without departing from the spirit and scope disclosed in the present invention, but the patent protection scope of the present invention shall still be subject to the scope defined in the attached claims.

Claims

1. A multi-point geostatistical modeling method, characterized in that: include: Constructing a training image and a simulation grid of a simulated work area, and obtaining a sedimentary sequence of all wells in the simulated work area; Positioning all deposition sequences in the simulation grid, thereby determining the grids with deposition data written in as known grid points and the grids without deposition data written in as unknown grid points, and then determining initial parameters, wherein the initial parameters include the data template shape, the number of conditional points, and a simulation circuit composed of all unknown points on the top surface of the simulation grid; Select the point to be estimated in the simulation line, determine the data template specification and corresponding data event for the point to be estimated according to the initial parameters, and traverse the training image according to the template specification to obtain a training pattern with the same data structure as the current data event, which includes: taking the point to be estimated as the center, scaling the data template according to the template shape, and determining the minimum specification corresponding to the conditional point number of known points as the data template specification required for the conditional simulation of the current point to be estimated. The data event of the current point to be estimated includes: a data structure of several vectors formed by the current point to be estimated and its known grid points within the data template specification range and the sedimentary sequence data of each known grid point, and the data structure includes the geometric position relationship between the known grid point and the point to be estimated and the sedimentary phase type of the known grid point; Comparing the data event with the training pattern for similarity, selecting the best training pattern and assigning the deposition sequence at the corresponding point therein to the point to be estimated; After completing the conditional simulation of all unknown points, a multi-point geostatistical model containing the reservoir sedimentary facies distribution characteristics is generated, in which: The step of comparing the data event with the training pattern for similarity, selecting the best training pattern and assigning the deposition sequence at the corresponding point therein to the point to be estimated includes: Set corresponding weights for each sedimentary phase in all sedimentary sequences; Screen out the deposition sequences corresponding to the first type of top surface known points in the data event and the deposition sequences corresponding to the second type of top surface known points in the training pattern that are at the same position as the first type of top surface known points. Based on this, in combination with the weights, calculate the similarity between the deposition sequences corresponding to each first type of top surface known point and the deposition sequences corresponding to the second type of top surface known points at the corresponding positions, which is recorded as the intra-group similarity. The intra-group similarity is the weight cost required to convert the data content at different depth positions in the deposition sequence corresponding to the first type of top surface known points into a sequence having the same data content and arrangement order as the deposition sequence corresponding to the second type of top surface known points, thereby taking the minimum total weight cost value as the similarity calculation result of the current known point; After summing up the similarities within the group of all known points, the comparison result of the current training graph is generated; From the comparison results of all training patterns corresponding to the current point to be estimated, select the training pattern to which the smallest similarity comparison result belongs as the optimal training pattern; According to the position information of the point to be estimated in the current data template, a grid point with the same position as the current point to be estimated is determined in the optimal training pattern and recorded as the target grid point; The deposition sequence at the target grid point is extracted from the optimal training pattern and the deposition data therein is distributed to the position of the point to be estimated, so as to assign deposition data to the point to be estimated and the grid points at different depths below it.

2. The method according to claim 1, characterized in that The process of calculating the intra-group similarity includes: The weight of the sedimentary phase at each sequence position in the sedimentary sequence corresponding to the current first-type top surface known point and the weight of the sedimentary phase at each sequence position in the sedimentary sequence corresponding to the second-type top surface known point are determined. Based on this, a dynamic programming method is used to calculate the intra-group similarity, wherein the distance between the first sedimentary sequence to be compared and the second sedimentary sequence is calculated by using the following expression to obtain the intra-group similarity: d[i,j]=min{[d[i-1,j]+insertCost(target i )],[d[i-1,j-1]]+substituteCost(source j ,target i )],[d[i,j-1]+deleteCost(source j )]} Where d[i, j] represents the distance between the first deposition sequence j and the second deposition sequence i, insertCost(target i ) represents the weight cost value generated by the insertion action performed during the sequence conversion process, substituteCost(source j ,targeti) represents the weight cost value generated by the replacement action executed during the sequence conversion process, deleteCost(source j ) represents the weight cost value generated by the deletion action performed during the sequence transformation process.

3. The method according to claim 1 or 2, characterized in that: The method further includes: determining whether the conditional simulation of all unknown points in the simulation line is completed; if not completed, selecting a new point to be estimated to perform conditional simulation on the current new point to be estimated.

4. A multi-point geostatistical modeling system, characterized in that: include: An information acquisition module configured to construct a training image and a simulation grid of a simulated work area, and to acquire a sedimentation sequence of all wells in the simulated work area; An initialization configuration module is configured to locate all deposition sequences in the simulation grid, thereby determining the grids with deposition data written in as known grid points and the grids without deposition data written in as unknown grid points, and then determining initial parameters, wherein the initial parameters include a data template shape, a number of conditional points, and a simulation circuit composed of all unknown points on the top surface of the simulation grid; A training pattern generation module is configured to select a point to be estimated in the simulated line, determine a data template specification and a corresponding data event for the point to be estimated according to the initial parameters, and traverse the training image according to the template specification to obtain a training pattern having the same data structure as the current data event; A module for assigning values ​​to points to be estimated, configured to compare the data events with the training patterns for similarity, select the best training pattern and assign the deposition sequence at the corresponding point therein to the points to be estimated; The model generation module is configured to generate a multi-point geostatistical model containing reservoir sedimentary facies distribution characteristics after completing the conditional simulation of all unknown points. The initialization configuration module is further configured to scale the data template according to the template shape with the point to be estimated as the center, and determine the minimum specification corresponding to the known points of the conditional point number that can cover the conditional point as the data template specification required for the conditional simulation processing of the current point to be estimated, and the data event of the current point to be estimated includes: a data structure of several vectors formed by the known grid points of the current point to be estimated within the data template specification range and the sedimentary sequence data of each known grid point, and the data structure includes the geometric position relationship between the known grid points and the point to be estimated and the sedimentary phase type of the known grid points; The to-be-estimated point assignment module is further configured as: Set corresponding weights for each sedimentary phase in all sedimentary sequences; Screen out the deposition sequences corresponding to the first type of top surface known points in the data event and the deposition sequences corresponding to the second type of top surface known points in the training pattern that are at the same position as the first type of top surface known points. Based on this, in combination with the weights, calculate the similarity between the deposition sequences corresponding to each first type of top surface known point and the deposition sequences corresponding to the second type of top surface known points at the corresponding positions, which is recorded as the intra-group similarity. The intra-group similarity is the weight cost required to convert the data content at different depth positions in the deposition sequence corresponding to the first type of top surface known points into a sequence having the same data content and arrangement order as the deposition sequence corresponding to the second type of top surface known points, thereby taking the minimum total weight cost value as the similarity calculation result of the current known point; After summing up the similarities within the group of all known points, the comparison result of the current training pattern is generated; From the comparison results of all training patterns corresponding to the current point to be estimated, select the training pattern to which the smallest similarity comparison result belongs as the optimal training pattern; According to the position information of the point to be estimated in the current data template, a grid point with the same position as the current point to be estimated is determined in the optimal training pattern and recorded as the target grid point; The deposition sequence at the target grid point is extracted from the optimal training pattern and the deposition data therein is distributed to the position of the point to be estimated, so as to assign deposition data to the point to be estimated and the grid points at different depths below it.

5. The system according to claim 4, characterized in that The to-be-estimated point assignment module is further configured to determine the weight of the sedimentary phase at each sequence position in the sedimentary sequence corresponding to the current first-type top surface known point, and the weight of the sedimentary phase at each sequence position in the sedimentary sequence corresponding to the second-type top surface known point. Based on this, a dynamic programming method is used to calculate the intra-group similarity, wherein the distance between the first sedimentary sequence to be compared and the second sedimentary sequence is calculated by using the following expression to obtain the intra-group similarity: d[i,j]=min{[d[i-1,j]+insertCost(target i )],[d[i-1,j-1]]+substituteCost(source j ,target i )],[d[i,j-1]+deleteCost(source j )]} Where d[i, j] represents the distance between the first deposition sequence j and the second deposition sequence i, insertCost(target i ) represents the weight cost value generated by the insertion action performed during the sequence conversion process, substituteCost(source j ,target i ) represents the weight cost value generated by the replacement action performed during the sequence conversion process, deleteCost(source j ) represents the weight cost value generated by the deletion action performed during the sequence transformation process.

Images