KR101647921B1 - Method for selecting model similar to real gas production and method for predicting gas production from oil and gas reservoir - Google Patents
Method for selecting model similar to real gas production and method for predicting gas production from oil and gas reservoir Download PDFInfo
- Publication number
- KR101647921B1 KR101647921B1 KR1020150043075A KR20150043075A KR101647921B1 KR 101647921 B1 KR101647921 B1 KR 101647921B1 KR 1020150043075 A KR1020150043075 A KR 1020150043075A KR 20150043075 A KR20150043075 A KR 20150043075A KR 101647921 B1 KR101647921 B1 KR 101647921B1
- Authority
- KR
- South Korea
- Prior art keywords
- reservoir
- models
- model
- data
- selecting
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 83
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 55
- 239000012530 fluid Substances 0.000 claims description 13
- 230000035699 permeability Effects 0.000 claims description 13
- 239000011159 matrix material Substances 0.000 claims description 9
- 238000004088 simulation Methods 0.000 claims description 9
- 230000008859 change Effects 0.000 claims description 6
- 238000003064 k means clustering Methods 0.000 claims description 5
- 230000007423 decrease Effects 0.000 claims description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims 2
- 239000007789 gas Substances 0.000 description 44
- 238000010187 selection method Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 238000009792 diffusion process Methods 0.000 description 5
- 230000001186 cumulative effect Effects 0.000 description 4
- 238000005553 drilling Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 230000006399 behavior Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 description 2
- 238000011084 recovery Methods 0.000 description 2
- 239000011435 rock Substances 0.000 description 2
- 230000000638 stimulation Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004138 cluster model Methods 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- HJUFTIJOISQSKQ-UHFFFAOYSA-N fenoxycarb Chemical compound C1=CC(OCCNC(=O)OCC)=CC=C1OC1=CC=CC=C1 HJUFTIJOISQSKQ-UHFFFAOYSA-N 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 239000003345 natural gas Substances 0.000 description 1
- 239000003305 oil spill Substances 0.000 description 1
- 229920000642 polymer Polymers 0.000 description 1
- 239000004576 sand Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 239000011800 void material Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- G06F19/00—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
According to one embodiment of a method for selecting a reservoir model similar to the actual gas production of a shale gas reservoir, (a) data on the reservoir layer, data on the nature of the crack formed in the reservoir, Generating a plurality of reservoir layer models having a plurality of reservoir model models having different initial conditions from each other; (b) calculating a well region (SRV) of each of a plurality of reservoir models according to the fast marching method (FMM); (c) clustering a plurality of reservoir model models into clusters of M (where M is an integer of 2 or more) based on the calculated SRV value; (d) selecting clusters closest to the actual production data of the reservoirs among the M clusters.
Description
The present invention relates to a method and apparatus for predicting the production volume of a shale gas reservoir, and more particularly, to a method for selecting a reservoir model similar to the actual gas production of a shale gas reservoir using a FMM (Fast Marching Method) algorithm and a model selection method. And a method for predicting gas production using a selected reservoir model.
Shale gas is a natural gas produced in shale reservoir and is being actively developed mainly in North America. Shale gas reservoirs produce shale gas through hydraulic fracturing using a horizontal scale because of the very low fluid permeability of the nano-darcy scale. According to the hydraulic fracturing method, a
Because the shale gas field has various production characteristics according to the hydraulic fracturing method, it is necessary to characterize the shale gas reservoir to be suitable for a specific shale gas reservoir in order to predict future production. For this purpose, various shale reservoirs are clustered according to similarity. However, static data such as crack length, fluid permeability, etc. used in the conventional clustering process did not reflect the dynamic connectivity of the reservoir. Therefore, it is necessary to select a reservoir model that has similar behavior to actual production data by suggesting clustering criterion that can reflect dynamic connectivity.
According to an embodiment of the present invention, it is possible to provide a method of characterizing a sail gas storage layer using a fast-marching method (FMM) and a model selection method.
According to the embodiment of the present invention, the similarity distance between the reservoir models can be reflected to reflect the dynamic connectivity of the reservoir. Clustering of the models based on the similarity distance, It is possible to provide a method and apparatus for selecting a reservoir model that can select the closest model.
According to an embodiment of the present invention, there is provided a method for selecting a reservoir model similar to the actual gas production volume of a shale gas reservoir using a computer, comprising the steps of: (a) And a plurality of reservoir models having initial conditions for the fluid in the cracks, wherein the plurality of reservoir models each have different initial conditions from each other; (b) calculating a well region (SRV) of each of a plurality of reservoir models according to the fast marching method (FMM); (c) clustering a plurality of reservoir model models into clusters of M (where M is an integer of 2 or more) based on the calculated SRV value; (d) selecting clusters closest to the actual production data of the reservoirs among the M clusters.
According to an embodiment of the present invention, there is provided a method of predicting a gas production amount of a shale gas reservoir, comprising: selecting a cluster closest to actual production amount data by the reservoir model selection method; And calculating a predicted gas production amount over time of the reservoir layer using the arbitrary reservoir model in the selected closest cluster.
According to an embodiment of the present invention, there is provided a computer-readable recording medium on which a program for executing the method of selecting the shale gas storage layer model or the method of predicting the gas production amount of the shale gas storage layer in a computer is recorded.
The sill gas reservoir layer can be characterized using the fast-marching method (FMM) and the model selection method by the method and apparatus for selecting a reservoir model according to an embodiment of the present invention.
According to the method and apparatus for selecting a reservoir model according to an embodiment of the present invention, it is possible to reflect the dynamic connectivity of a reservoir by applying a similarity distance between reservoir models, clustering models based on the similarity distance, The model selection method has an advantage in that the model that is closest to the actual data can be selected.
1 is a view for explaining a method of producing a gas in a shale gas reservoir,
2 is an exemplary flow chart of a method for predicting gas production of a shale gas reservoir according to an embodiment of the present invention;
FIGS. 3A and 3B are views for explaining modeling of a shale gas reservoir layer,
4 is a view for explaining input data used for modeling a shale gas storage layer,
FIG. 5 is a view for explaining a change in the volume of the discharged oil with time; FIG.
FIG. 6 is a graph showing the time-
Figure 7 is a graph of gas production over time,
Figure 8 is a graph of cumulative gas production over time,
9 is an exemplary flow chart of the step of clustering the reservoir model into clusters,
10 is a diagram for explaining the Housdorff distance,
11 is a diagram for explaining a multi-dimensional scaling algorithm,
Figure 12 is a flow diagram of one exemplary method of selecting a cluster closest to real data,
13 and 14 are diagrams for explaining a step of selecting a cluster closest to actual data,
15 is a flowchart of an exemplary method of selecting a representative model closest to actual data;
FIG. 16 is a block diagram illustrating an exemplary system configuration for predicting gas production of a shale gas reservoir according to an embodiment.
BRIEF DESCRIPTION OF THE DRAWINGS The above and other objects, features, and advantages of the present invention will become more readily apparent from the following description of preferred embodiments with reference to the accompanying drawings. However, the present invention is not limited to the embodiments described herein but may be embodied in other forms. Rather, the embodiments disclosed herein are provided so that the disclosure can be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
In the present specification, when an element is referred to as being on another element, it may be directly formed on another element, or a third element may be interposed therebetween.
Where the terms first, second, etc. are used herein to describe components, these components should not be limited by such terms. These terms have only been used to distinguish one component from another. The embodiments described and exemplified herein also include their complementary embodiments.
In the present specification, the singular form includes plural forms unless otherwise specified in the specification. The terms "comprise" and / or "comprising" used in the specification do not exclude the presence or addition of one or more other elements.
Hereinafter, the present invention will be described in detail with reference to the drawings. Various specific details are set forth in the following description of specific embodiments in order to provide a more detailed description of the invention and to aid in understanding the invention. However, it will be appreciated by those skilled in the art that the present invention may be understood by those skilled in the art without departing from such specific details. In some cases, it is noted that parts of the invention that are not commonly known in the art and are not largely related to the invention are not described in order to avoid confusion in describing the invention.
1. Basic flow chart
2 is an exemplary flow chart of a method for predicting gas production of a shale gas reservoir according to an embodiment of the present invention.
First, in step S10, a plurality of reservoir model is generated. Referring to FIG. 3A, the reservoir model is represented by a set of a plurality of cubic lattices and a hexahedron having a predetermined size. For example, in one embodiment, the reservoir model is represented by a reservoir with dimensions of length x width x height 4,000 ft x 2,000 ft x 150 ft and the reservoir consists of a set of cube with one
Fig. 4 shows input data (initial conditions) used for generating such a reservoir model. In one embodiment, the data relating to the reservoir model includes (i) data on the reservoir layer, (ii) data on the characteristics of the cracks formed in the reservoir, and (iii) data on the fluid in the crack, Variables.
(i) Data on reservoirs: reservoir dimension, grid size, porosity, matrix permeability, enhanced permeability, initial pressure, and temperature and temperature.
(ii) Data on the nature of the crack: It includes at least one of bottom-hole pressure, fracture number, fracture permeability, and fracture half-length.
(iii) data on the fluid: at least one of viscosity, total compressibility, and gas formation volume factor.
The step (S10) of creating a plurality of storage layer models as described above means creating a plurality of storage layer models having the data on the storage layer, the data on the characteristics of the crack, and the data on the fluid as initial conditions as shown in Fig. can do. In this case, each of the generated models has different initial conditions by differentiating at least one of the input data.
The number of reservoir models generated in step S10 in one embodiment is not limited. The following embodiments of the present invention assume that 400 reservoir layer models are created.
Also, a concrete method of generating a plurality of reservoir model is not particularly limited. In one embodiment, the data measured in the shale gas reservoir, which is the actual observation object (i.e., the data in FIG. 4), is used as reference data, and the reservoir models are generated one by one by slightly changing each variable in the reference data. In one embodiment, a sequential Gaussian Simulation (SGS) algorithm may be used to generate a plurality of reservoir models from the reference data.
A part (6) of the plurality of reservoir layer models generated by this step S10 is shown in Fig. 3B as an example. As shown, the number of cracks was set to 5 in all models, and other initial conditions were set differently for each model.
After generating the plurality of reservoir models in step S10, the SRV (Stimulated Reservoir Volume) of each of the plurality of reservoir models is calculated in step S20. In one embodiment, a Fast Marching Method (FMM) algorithm is used to calculate the well region stimulus range (SRV). Fast-marching method (FMM) is a method of predicting the extent of oil spill and ultimate yield, which can take into account the heterogeneity of the reservoir compared to the analytical method and can yield reliable results in a shorter time than other reservoir simulations . However, in the present invention, it is also possible to calculate the well region stimulus range (SRV) using another algorithm other than the fast-marching method. The fast-marching method will be described later with reference to Figs. 5 to 8. Fig.
Referring again to FIG. 2, after calculating the well region excursion range (SRV) for each reservoir model, a plurality of reservoir models are clustering (clustering) into M clusters based on the calculated well region excitation range at step S30 )do. Here, M is the number of clusters by clustering, and may be any integer greater than or equal to 2. In the following embodiment, it is assumed that clustering is performed into seven clusters as one example (that is, M = 7 is assumed).
Any clustering algorithm can be used to clustering multiple reservoir models. In the following embodiment of the present invention, the Hausdorff distance between the reservoir models is calculated and arranged according to the multi-dimensional scaling method, followed by clustering by applying a k-means clustering algorithm , Which will be described later with reference to Figs. 9 to 11.
After the reservoir model is clustered into a predetermined number (M) of clusters, a cluster closest to the actual production data of the reservoir among the clustered clusters is selected in step S40. If there are M clusters each including a plurality of reservoir models in each of the clusters, a representative storage model is selected one by one in each cluster, and M selected representative storage models are compared with actual production data And selects one representative model and selects the cluster to which this representative model belongs. Such an exemplary model selection method will be described later with reference to FIGS. 12 to 15. FIG.
In one embodiment, the clustering of the reservoir model (S30) and the step of selecting one of the clustered clusters (S40) may be repeated a plurality of times. At this time, each time the steps S30 and S40 are repeated, the clustering step (S30) at an arbitrary number of times is a step of selecting a cluster model in the cluster selected in the immediately preceding cluster selection step (S40) .
For example, assuming that 400 reservoir models are created in the first step (S10) and clustered into 7 clusters, seven clusters include about 50 to 60 reservoir models through the first clustering step (S30) . Then, when an arbitrary cluster (for example, including 60 models) is selected in the first cluster selection step S40, the entire reservoir model in this cluster, that is, 60 reservoir models becomes a population, and steps S30 and S40 are performed . Thus, for example, sixty reservoir models will be subdivided into seven clusters by a second clustering step (S30), and each cluster will include eight to nine models, for example. Thereafter, in the second cluster selection step (S40), an arbitrary cluster including, for example, nine models may be selected.
The number of repetitions of the clustering step S30 and the cluster selecting step S40 can be set until the number of the storage layer models in the finally selected cluster becomes equal to or smaller than a predetermined number. Accordingly, it is determined in step S50 of FIG. 2 whether the number of models in the cluster selected in the cluster selection step S40 is n or less (n is an integer of 2 or more as a predetermined number) S30, and S40) are repeated. If the number is smaller than this number, the process proceeds to step S60. In one embodiment, this number can be set to 15 (i.e., n = 15).
If a cluster is finally selected through step S50, simulations are performed using all the reservoir models in the cluster in step S60, and the expected production amount of the reservoir for each model is calculated. As an example, all reservoir models in the last selected cluster can be simulated to predict the range of estimated ultimate recovery (EUR) of this reservoir.
2. Fast Marching method (Fast Marching Method)
Fast-marching method (FMM) is a method of deriving the diffusion equation representing the propagation of pressure in the form of an Eikonal equation to efficiently solve the equation. For example, Xie, J., Yang, C., Gupta, N , King, MJ, and Datta-Gupta, A. "Integration of Shale Gas Production Data and Microseismic for Fracture and Reservoir Properties Using Fast Marching Method," SPE paper 161357 presented at the SPE Eastern Regional Meeting, Lexington, 3-5 October 2012.
First, a diffusion equation representing pressure propagation in a heterogeneous medium is expressed by the form of an iconar equation as shown in
[Equation 1]
In this equation, τ is a diffusive time of flight (DTOF), which is calculated based on the diffusion coefficient value assigned to each lattice. The larger the value of the diffusion diffusivity is, the less time the pressure reaches.
α is a diffusion coefficient and has different values depending on the fluid permeability, porosity, viscosity and compressibility of each lattice, and is expressed by the following equation (2).
&Quot; (2) "
Where k is the permeability, φ is the porosity, μ is the fluid viscosity and tc is the total compressibility and these parameters are entered as initial conditions, .
(1) Time to reach pressure DTOF )
The pressure arrival time (DTOF: tau) is the time it takes for the pressure to reach the predetermined lattice at the initial starting point. According to the fast-marching method (FMM), the above equation (1) is expressed by a quadratic equation according to the finite difference method, and the time of arrival of the pressure (τ) Can be obtained. The pressure reaching time calculated by FMM has a square root unit of time and depends on the properties of the reservoir and fluid such as permeability, compressibility and so on.
The relationship between the pressure arrival time (DTOF) and the actual physical time is as shown in
&Quot; (3) "
Where c is a dimensionless geometric factor with different values depending on the flow behavior. For example, the value of c for a linear flow can be 2, 4 for a 2-dimensional flow (radial flow), or 6 for a 3-dimensional flow (spherical flow). In an embodiment of the present invention, the value of 6 is assumed assuming a three-dimensional flow.
On the other hand, the pressure on a particular lattice means that the lattice has begun to drain out. In other words, it can be seen that the gratings having a pressure arrival time value smaller than the time at an arbitrary time reference are already diverted. Therefore, the drainage volume at a specific time can be expressed by a void volume sum of the gratings whose pressure arrival times are smaller than a specific time, as shown in Equation (4) below.
&Quot; (4) "
FIG. 5 is a view for explaining a change in the endurance volume with time, showing an example of how the endocardial volume increases with time in a three-dimensional reservoir model.
Fig. 5 (a) shows the stocking volume at one month after the start of the first production, and only the lattice near the five cracks was discharged . FIG. 5 (b) shows the volume of the discharged oil after 30 years, and it can be seen that the volume of the discharged oil increases not only around the five cracks but also throughout the reservoir.
(2) Range of Oil Stimulation ( SRV )
FIG. 6 is a graph of the drainage volume according to time calculated based on Equation (4). Generally, there is a large difference between the pressure arrival time of the parent rock and the pressure arrival time of the stimulated area because the fluid permeability of the parent rock of the shale reservoir is much smaller than the fluid permeability of the area stimulated by cracks and the like. Therefore, the value of the drain volume at the point where the slope decreases sharply in the drainage volume graph according to the time shown in FIG. 6 can be regarded as the wellness stimulus range (SRV). In other words, the range of the SRV can be estimated through the interval in which the slope of the endocardial volume is flat over time in the graph. In the case of the graph of FIG. 6, it can be seen that approximately 800 days have elapsed, reaching the oil-field stimulation range of approximately 2.8 MMcf.
Meanwhile, the shale gas production amount can be estimated over time by calculating the exhaust gas volume with time. For example, FIG. 7 is a graph illustrating an example of the gas production amount with time. The cumulative production amount up to an arbitrary time can be calculated from the production amount graph over time as shown in FIG. For example, FIG. 8 is a graph of the cumulative yield of a reservoir over time. According to the graph in FIG. 8, assuming that a given production is produced for about 10 years, the cumulative yield after 3650 days (i.e., 679MMscf) is the estimated ultimate recovery in this production set.
3. Clustering
The clustering step S30 of the model of Fig. 2 will be described with reference to Figs. 9 to 11. Fig.
9 is an exemplary flowchart of a step S30 of clustering the reservoir model into clusters. Referring to the drawing, according to the clustering method according to the illustrated embodiment, a distance matrix is generated by calculating distances related to similarities between the reservoir models in step S310. In one embodiment, Hausdorff distance is used as the distance for this similarity.
Thereafter, in step S320, a plurality of reservoir models are arranged on a two-dimensional plane by applying a multidimensional scaling method to the distance matrix, and then, in step S330, a plurality of reservoir models arranged in a two- Clusters into clusters.
(One) Between models Similarity distance
In one embodiment of the present invention, the Hausdorff distance is used as the similarity distance applied in step S310. The Hausdorff distance is a numerical representation of the similarity between two objects. The closer the two objects are, the smaller the Hausdorff distance (H) is. Methods for obtaining Hausdorff distances are well known and are described, for example, in Dubuisson, M.-P. and Jain, A. K. 1994. A Modified Hausdorff Distance for Object Matching. ICPR 94 (1): 566-568, and the like.
According to this method, for example, d (A, B) and d (B, A) are calculated as shown in Equation (5) below, in order to obtain the house road rope distance between two objects A and B, respectively.
&Quot; (5) "
In the above equation, a is the coordinates of an arbitrary point belonging to A, and b is the coordinates of an arbitrary point belonging to B. d (a, b) is the Euclidean distance between point a and point b. Therefore, d (A, B) means the largest value among the minimum values among the distances from any point belonging to object A to any point of object B, and d (B, A) means any point belonging to object B To the arbitrary point of the object A, among the minimum values among the distances.
Based on Equation (5), the Hausdorff distance H (A, B) means a larger value of d (A, B) and d (B, A), as shown in Equation (6).
&Quot; (6) "
10 shows an example of the Hausdorff distance between the reservoir models. Assuming that there are three arbitrary reservoir models having five cracks as shown in Figs. 10 (a) to 10 (c) Distance.
10 (a), the Hausdorff distance H between the model of FIG. 10 (a) and the model of FIG. 10 (b) is 8.25 , The distance between the model of Fig. 10 (a) and the model of Fig. 10 (c) is 24.00. That is, it can be seen that the degree of similarity between the model of FIG. 10 (a) and the model of FIG. 10 (b) is greater than the degree of similarity between the model of FIG. 10 (a) and the model of FIG. 10 (c).
11 (a) shows that the Hausdorff distance H between two reservoir models as shown in Fig. 10 is calculated for all of a plurality of reservoir models to generate a distance matrix. For example, assuming a total of 400 reservoir models, the distance from the second model (H (1,2)) to the third model (H (1,3),. (H (2,1)) from the first model and the distance (H (2, 1)) from the third model are calculated on the basis of the second model, 3),... The distance (H (2,400)) from the 400th model is calculated, and the distance from each model to the 400th model is calculated in this manner. It is to be understood that the diagonal direction in the matrix of FIG. 11 (a) is 0 since the distance to itself is zero.
(2) Multidimensional Scale method (MDS: multi-dimensional scaling)
After calculating the Hausdorff distance in step S310 and creating the distance matrix, the multidimensional scaling method is applied to the distance matrix in step S320.
The Multidimensional Scale (MDS) is a set of statistical techniques that processes numerical data about objective or subjective relationships among multiple objects and displays them locally in a multidimensional space. In one embodiment of the present invention, models are arranged on a two-dimensional plane as shown in FIG. 11 (b) according to the degree of similarity between the models by applying the multidimensional scaling method to the Hausdorff distance between the reservoir models. In Figure 11 (b), each blue dot represents each model, similar models are close together, and non-similar models are far apart.
(3) k-means clustering
Clustering (clustering) is a technique of classifying models into a predetermined number of groups (clusters) by grouping the models having high similarities among a plurality of models into one group. Many different techniques have been developed for clustering, and it is assumed that k-means clustering is used in one embodiment of the present invention. However, other clustering algorithms may of course be used in alternative embodiments.
The K-average clustering algorithm works by minimizing the variance of the distance difference between each cluster, and is a algorithm for grouping the given data into k clusters. The k-means algorithm is known and is described, for example, in "Caers, J. 2011. Modeling Uncertainty in the Earth Sciences. Chichester, UK: John Wiley & Sons Ltd." And the like. According to this method, a plurality of models can be processed and clustered according to the following procedure.
1) Set k cluster centers
2) Compute each Euclidean distance from each point (models) to k centers.
3) Assign each point (model) to the nearest center
4) Compute the average of the models assigned to the cluster to obtain the new cluster center.
5) Repeat steps 2) to 4) until there is no change in the cluster center.
When such a k-means clustering algorithm is applied to models placed on a plane according to the multidimensional scaling method of Fig. 11 (b), the models can be clustered as shown in Fig. 11 (c). In FIG. 11 (c), it is assumed that 400 reservoir models are clustered into seven clusters, and each cluster is represented by a different color.
4. Model selection method
Referring again to FIG. 2, after the reservoir model is clustered into a predetermined number of clusters (M) in step S30, a cluster closest to the reservoir actual production data among the clusters in step S40 .
The model selection method is a method of selecting clusters that are closest to the actual production data among the clustered models. After selecting the representative models of each cluster, the similarity with the actual production data is measured through the reservoir simulation, and the cluster with the larger degree is selected.
In one embodiment, Bayes' theorem is used to calculate posterior probabilities using prior probabilities and actual observations for model selection. Model selection using Bayes' theorem is a well known technique, for example, Mantilla, C. A. 2010. "Feedback Control of Polymer Flooding Process Considering Geologic Uncertainty." PhD Dissertation, The University of Texas at Austin, Austin, Texas (December 2010). "
FIG. 12 is a flowchart of an exemplary method of selecting a cluster closest to actual data (S40), and FIGS. 13 and 14 are views for explaining a step of selecting a cluster closest to actual data.
Referring to Fig. 12, in order to select clusters closest to the actual data, in step S410, a representative storage layer model is selected from each of a plurality of (M) clusters. Then, in step S420, each representative storage layer model is input to the simulation program and executed, and the production simulation data according to each representative storage layer model is calculated. The simulation program used in this case may be a production amount prediction program using the above-described fast marching method (FMM), or may be a conventional commercial program, for example, a "GEM" simulator of CMG or an "ECLIPSE" simulator of Schlumberger .
FIG. 13B is a production forecast graph of each representative model calculated by this step S420. In the figure, each color graph represents a predictive graph of a representative model of each corresponding color cluster in FIG. 13 (a), and a dotted line represents a 24-month actual production graph of the corresponding reservoir.
Thereafter, in step S430, the actual production amount data is compared with the simulation data, and the representative stock model closest to the actual production amount data is selected. In Fig. 13 (b), it is seen that the graph closest to the actual observation data is a navy blue graph. Step S430 is a step of calculating and determining it by a mathematical algorithm. For example, as shown in Fig. 15, one of the representative models can be selected using the Bayes theorem.
15 is a flowchart of an exemplary method of selecting a representative model closest to actual data. Referring to FIG. 15, in step S431, each prior probability of the representative reservoir model is calculated . Since there is no information about the models before the model selection process starts, all models will have the same prior probability in the first step.
Then, at step S432, a likelihood function for each representative reservoir model is calculated. The likelihood function can be derived based on the result of simulating a representative model of each cluster.
Next, in step S433, the posterior probability of each representative reservoir model when the likelihood function is maximum is calculated. According to Bayes' theorem, the posterior probability can be computed based on the prior probability and the likelihood function. Then, in step S434, the representative storage layer model having the largest posterior probability is selected.
Then, if the representative storage layer model closest to the actual production amount is selected by the step S430 in FIG. 15, the cluster including the selected representative storage layer model is selected as indicated by the step S440 in FIG.
Meanwhile, as described with reference to FIG. 2, the process of selecting one of the plurality of clusters may be repeated a plurality of times. For example, according to the graph of Fig. 13 (b), the navy cluster among the seven clusters of 13 (a) was selected. Thereafter, the blue color cluster is used as a population and step S40 is repeated again. That is, the models of the population are clustered into a plurality of clusters again, and a cluster including the model closest to the actual data value among the clustered clusters is selected
Fig. 14 (a) shows that the storage model belonging to the blue color cluster in Fig. 13 (a) is clustered again into seven clusters, Fig. 14 (b) Is compared with the actual observation graph. In this example, the representative model of the yellow graph is closest to the actual production graph. By repeating the clustering and cluster selection step (steps S30 and S40 in Fig. 2), the representative model closest to the actual observation data can be selected.
FIG. 16 is a block diagram illustrating an exemplary system configuration for predicting gas production of a shale gas reservoir according to an embodiment.
Referring to FIG. 16, a reservoir
The
In this configuration, these various programs and algorithms may be stored in the
As described above, although the present invention has been described with reference to the limited embodiments and drawings, the present invention is not limited to the above embodiments. It will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the present invention as defined by the appended claims. Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined by the equivalents of the claims, as well as the claims.
100: Production forecasting system
110: Processor
120: Memory
130: Storage device
Claims (11)
(a) a plurality of reservoir models having initial conditions of data on the reservoir layer, data on the characteristics of fractures formed in the reservoir layer, and data on the fluid in the cracks, wherein a plurality of reservoir models have different initial conditions ), ≪ / RTI >
(b) calculating a well region (SRV) of each of a plurality of reservoir models according to the fast marching method (FMM);
(c) clustering a plurality of reservoir model models into clusters of M (where M is an integer of 2 or more) based on the calculated SRV value;
(d) selecting a cluster closest to actual production data of the reservoir among the M clusters,
The step (b) of calculating the well region excursion range for each reservoir model,
(b-1) calculating an amount of change in the volume of exhausted water over time from the pressure arrival time calculated by the fast marching method (FMM); And
(b-2) determining a point at which the slope of the amount of change in the volume of exhausted water with respect to time decreases, as a well region.
(e) repeating the steps (c) and (d) a plurality of times for a plurality of reservoir models in the selected cluster,
Wherein each time the steps (c) and (d) are repeated, the step (c) of clustering at an arbitrary number of times is performed by using a reservoir model in the cluster selected in the step (d) Wherein the method comprises the steps of:
Wherein the generating of the plurality of storage layer models comprises generating the plurality of storage layer models using a sequential Gaussian simulation (SGS) algorithm.
Wherein the data relating to the reservoir layer comprises at least one of a reservoir size, a lattice size, a porosity, a matrix permeability, an expanded permeability, an initial pressure, and a temperature,
The data on the properties of the cracks include at least one of a bottom-hole pressure, a number of cracks, a crack permeability, and a fracture half-length,
Wherein the data relating to the fluid includes at least one of viscosity, total compressibility, and gas volume coefficient.
(c-1) calculating a distance according to the geometric similarity between the storage layer models;
(c-2) multi-dimensionally scaling (MDS) the matrix of the calculated distances to place the plurality of reservoir models on a two-dimensional plane; And
(c-3) clustering the plurality of reservoir models into the M clusters.
The distance according to the geometric similarity is calculated by a Hausdorff distance algorithm,
Wherein the step (c-3) of clustering the plurality of reservoir models comprises clustering the reservoir model with a k-means clustering algorithm.
(d-1) selecting one representative storage layer model from each of the M clusters;
(d-2) calculating production simulation data for the representative storage layer model;
(d-3) comparing the actual production amount data and the simulation data among the representative storage layer models, and selecting a representative storage layer model closest to the actual production amount data; And
(d-4) selecting a cluster including the selected representative storage layer model.
(d-3-1) calculating respective prior probabilities of the representative reservoir model;
(d-3-2) calculating a likelihood function for each representative storage layer model;
(d-3-3) calculating a posterior probability of each representative reservoir model when the likelihood function is maximum; And
(d-3-4) selecting a representative reservoir model having the largest posterior probability.
Selecting a cluster closest to the actual production amount data by the method of selecting a reservoir model according to any one of claims 1 to 8; And
Calculating a predicted gas production amount over time of the reservoir layer using an arbitrary reservoir model in the selected closest cluster.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
KR1020150043075A KR101647921B1 (en) | 2015-03-27 | 2015-03-27 | Method for selecting model similar to real gas production and method for predicting gas production from oil and gas reservoir |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
KR1020150043075A KR101647921B1 (en) | 2015-03-27 | 2015-03-27 | Method for selecting model similar to real gas production and method for predicting gas production from oil and gas reservoir |
Publications (1)
Publication Number | Publication Date |
---|---|
KR101647921B1 true KR101647921B1 (en) | 2016-08-12 |
Family
ID=56715051
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
KR1020150043075A KR101647921B1 (en) | 2015-03-27 | 2015-03-27 | Method for selecting model similar to real gas production and method for predicting gas production from oil and gas reservoir |
Country Status (1)
Country | Link |
---|---|
KR (1) | KR101647921B1 (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101838518B1 (en) * | 2017-01-12 | 2018-03-14 | 동아대학교 산학협력단 | A Method for the well placement investigation using Productivity Potential Area Map |
KR101893800B1 (en) | 2017-08-09 | 2018-09-04 | 제주대학교 산학협력단 | Method of sedimentary environment interpretation through electrofacies construction |
KR101904278B1 (en) | 2016-09-05 | 2018-10-04 | 동아대학교 산학협력단 | Method for decline curve analysis according to cumulative production incline rate in unconventional gas field |
KR101975436B1 (en) * | 2018-10-25 | 2019-05-07 | 동아대학교 산학협력단 | Apparatus and method for forecasting production for shale gas well in transient flow using machine learning method |
CN110322363A (en) * | 2018-03-29 | 2019-10-11 | 中国石油化工股份有限公司 | Shale gas reservoir reconstruction volume calculation method and system |
KR20210061716A (en) | 2019-11-20 | 2021-05-28 | 한국지질자원연구원 | Apparatus and method for monitoring operating condition of oil well |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20030014357A (en) * | 2000-02-22 | 2003-02-17 | 슐럼버거 테크놀로지 코포레이션 | Integrated reservoir optimization |
KR100776729B1 (en) * | 2006-08-29 | 2007-11-19 | 울산대학교 산학협력단 | Speaker-independent variable-word keyword spotting system including garbage modeling unit using decision tree-based state clustering and method thereof |
KR20110106698A (en) | 2010-03-23 | 2011-09-29 | 한국과학기술원 | Method and recording medium of a hybrid approach to multiple fluid simulation using volume fraction |
KR20110138088A (en) | 2010-06-18 | 2011-12-26 | 한양대학교 산학협력단 | Method for oil prediction in fractured reservoirs and recording media therefor |
-
2015
- 2015-03-27 KR KR1020150043075A patent/KR101647921B1/en active IP Right Grant
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20030014357A (en) * | 2000-02-22 | 2003-02-17 | 슐럼버거 테크놀로지 코포레이션 | Integrated reservoir optimization |
KR100776729B1 (en) * | 2006-08-29 | 2007-11-19 | 울산대학교 산학협력단 | Speaker-independent variable-word keyword spotting system including garbage modeling unit using decision tree-based state clustering and method thereof |
KR20110106698A (en) | 2010-03-23 | 2011-09-29 | 한국과학기술원 | Method and recording medium of a hybrid approach to multiple fluid simulation using volume fraction |
KR20110138088A (en) | 2010-06-18 | 2011-12-26 | 한양대학교 산학협력단 | Method for oil prediction in fractured reservoirs and recording media therefor |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101904278B1 (en) | 2016-09-05 | 2018-10-04 | 동아대학교 산학협력단 | Method for decline curve analysis according to cumulative production incline rate in unconventional gas field |
KR101838518B1 (en) * | 2017-01-12 | 2018-03-14 | 동아대학교 산학협력단 | A Method for the well placement investigation using Productivity Potential Area Map |
WO2018131992A1 (en) * | 2017-01-12 | 2018-07-19 | 동아대학교 산학협력단 | Method for searching for location of oil well using productivity potential area map |
KR101893800B1 (en) | 2017-08-09 | 2018-09-04 | 제주대학교 산학협력단 | Method of sedimentary environment interpretation through electrofacies construction |
CN110322363A (en) * | 2018-03-29 | 2019-10-11 | 中国石油化工股份有限公司 | Shale gas reservoir reconstruction volume calculation method and system |
KR101975436B1 (en) * | 2018-10-25 | 2019-05-07 | 동아대학교 산학협력단 | Apparatus and method for forecasting production for shale gas well in transient flow using machine learning method |
WO2020085617A1 (en) * | 2018-10-25 | 2020-04-30 | 동아대학교 산학협력단 | Device and method for predicting productivity of shale gas well in transition flow region by using machine learning technique |
KR20210061716A (en) | 2019-11-20 | 2021-05-28 | 한국지질자원연구원 | Apparatus and method for monitoring operating condition of oil well |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
KR101647921B1 (en) | Method for selecting model similar to real gas production and method for predicting gas production from oil and gas reservoir | |
RU2669948C2 (en) | Multistage oil field design optimisation under uncertainty | |
Artus et al. | Optimization of nonconventional wells under uncertainty using statistical proxies | |
US9836561B2 (en) | Optimizing multistage hydraulic fracturing design based on three-dimensional (3D) continuum damage mechanics | |
CA2868756C (en) | System and method for automatic local grid refinement in reservoir simulation systems | |
US10846443B2 (en) | Discrete irregular cellular models for simulating the development of fractured reservoirs | |
US20170227663A1 (en) | Analysis of microseismic supported stimulated reservoir volumes | |
EP2856387B1 (en) | System and method for reservoir simulation optimization | |
CN114595608B (en) | Fracturing construction parameter and working system parameter optimization method and system | |
Rahim et al. | Well placement optimization with geological uncertainty reduction | |
KR101657889B1 (en) | Method for estimating probabilistic EUR of oil and gas reservoir | |
CN115166853A (en) | Shale gas reservoir natural fracture model establishing method and device, electronic equipment and medium | |
Ansari et al. | Response surface method for assessing energy production from geopressured geothermal reservoirs | |
Zhang et al. | Uncertainty Analysis and Assisted History Matching Workflow in Shale Oil Reservoirs | |
Ansari et al. | Statistical modeling of geopressured geothermal reservoirs | |
US20110208499A1 (en) | Method for history matching of a geological model comprising a sub-seismic fault network | |
NO20171058A1 (en) | Integrated a priori uncertainty parameter architecture in simulation model creation | |
EP2678718A2 (en) | Computerized method for the estimation of a value for at least a parameter of a hydrocarbon-producing region, for planning the operation and operating the region | |
CN112610205A (en) | Method for evaluating early productivity of gas well in strong heterogeneous reservoir | |
Ajayi et al. | Quantifying changes in hydraulic fracture properties using a multi-well integrated discrete fracture network (DFN) and reservoir simulation model in an unconventional wolfcamp fm., midland basin, west Texas | |
Becerra et al. | Petroleum reservoir uncertainty mitigation through the integration with production history matching | |
Ansari | Mathematical scaling and statistical modeling of geopressured geothermal reservoirs | |
US20210208304A1 (en) | Systems and Methods for Hydrocarbon Reservoir Three Dimensional Unstructured Grid Generation and Development | |
Jung et al. | Updating uncertainty in the conceptual geological representation of fractured reservoirs using production data | |
Dalaei | Uncertainty Quantification in 3D Reservoir Modeling by Probabilistic Forecasting Approach |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
E701 | Decision to grant or registration of patent right | ||
GRNT | Written decision to grant |