Patent Application
20180245441
The present disclosure generally relates to methods and systems for hydraulic fracturing, and in particular, to mapping the natural fracture network in shale.
Shale is a sedimentary rock formed from the compaction of natural materials, such as mud and silt, under pressure over many years. During the formation of certain types of shale, such as black shale, the earth warms the mud and silt to thereby cause organic matter therein to decompose over time into oil and gas. In some cases, the oil and gas migrates to pockets between the shale to form reservoirs. In others, the oil or gas remains in the shale, specifically, in small pores within the shale.
To extract the oil and gas from the shale, oil companies have employed techniques such as hydraulic fracturing. In this regard, a well site first is drilled at a location suspected of being within a shale reservoir. A slurry of water, proppant, and other additives are pressurized and introduced into the shale reservoir to thereby release the hydrocarbon, if any, that may have been generated or trapped therein. The pressurized slurry forces the hydrocarbon (by fracturing the rock) to travel through pipes or other delivery mechanisms to the earth's surface for collection at the well site. To fully exploit the potential yield of a shale reservoir, the hydraulic fracturing operation is performed on a first length of the shale reservoir (commonly referred to as a “stage”), then performed on a second adjacent length or stage of the shale reservoir and so on. In some cases, a full length of the shale reservoir is divided into ten to thirty (or more) stages.
The Natural Fracture Network (NFN) of the shale reservoir is a significant contributor to well productivity in shale wells. As such, opening existing networks of natural fractures before (or along with) creating new fractures in shale improves well productivity.
As shale well productivity is at least partially based on the NFN of a shale reservoir, there is a need for a method of mapping the NFN of a shale reservoir. The following methods and computer-readable medium storing the methods, address the aforementioned need.
According to an aspect of the present disclosure, systems and methods are provided that address the above mentioned needs.
In an embodiment, a method of generating a map of a Natural Fracture Network (“NFN”) of a shale reservoir is provided. The method includes defining a set of rules to output an NFN density value, the set of rules based on a plurality of parameters including a well volume/completion parameter, a well stimulation parameter, and a well production parameter, each parameter divided into at least two fuzzy clusters, wherein the set of rules includes a plurality of combinations made up of the plurality of parameters and the fuzzy cluster of each parameter, each combination being assigned an NFN density value. Historical data related to a plurality of wells drilled in the shale reservoir is analyzed, where the data includes volume/completion data, stimulation data, and production data. Corresponding values are assigned to the historical data based on the set of rules to thereby determine the NFN density value of each of well. The NFN density value of each well is mapped on a graph to thereby identify the NFN of the shale reservoir, and the graph is displayed as the map.
In an aspect of the embodiment, each of the well volume/completion parameter, the well stimulation parameter, and the well production parameter are divided into three fuzzy clusters.
In another aspect of the embodiment, each combination is assigned a truth value, and the NFN value is based, in part, on the truth value.
In still another aspect of the embodiment, the well volume/completion parameter is generated from one or more of a well spacing along a lateral length of the hydraulic fracturing job, a bulk volume of the hydraulic fracturing job, initial water saturation to calculate a hydrocarbon pore volume of the hydraulic fracturing job, an initial pressure used in the hydraulic fracturing job, a TOC, or a total number of stages included in the hydraulic fracturing job.
In another aspect of the embodiment, the well stimulation parameter is generated from one or more of an amount of proppant used in the hydraulic fracturing job or an amount of fluid used in the hydraulic fracturing job.
In another aspect of the embodiment, the well production parameter is generated from one or more of a well-head pressure used during a hydraulic fracturing job, an initial reservoir pressure used during the hydraulic fracturing job, or a 180 days cumulative production amount obtained from the hydraulic fracturing job.
In another aspect of the embodiment, the shale reservoir includes a developed portion and an undeveloped portion, the developed portion includes the plurality of wells, and the method includes imposing a Cartesian (or other type of) grid on the generated map, selecting seismic attributes from a seismic map of the shale reservoir, using the seismic attributes and NFN density values of the developed portion to design, train, calibrate, and validate a neural network model, deploying neural network model in a forecast mode, setting the seismic attributes of the undeveloped portion to generate NFN density values for the shale reservoir including the undeveloped portion, and generating the map using the NFN density values generated for the undeveloped portion.
According to another embodiment, a non-transitory computer-readable medium is provided storing instructions, which when executed, cause a processor to perform a method of generating a map of a Natural Fracture Network (“NFN”) of a shale reservoir. The method includes defining a set of rules to output an NFN value, the set of rules based on a plurality of parameters including a well volume/completion parameter, a well stimulation parameter, and a well production parameter, each parameter divided into at least two fuzzy clusters, wherein the set of rules includes a plurality of combinations made up of the plurality of parameters and the fuzzy cluster of each parameter, each combination being assigned an NFN density value, analyzing historical data related to a plurality of wells drilled in the shale reservoir, the data including volume/completion data, stimulation data, and production data, assigning corresponding values to the historical data based on the set of rules to thereby determine the NFN density value of each of well, mapping the NFN density value of each well on a graph to thereby identify the NFN of the shale reservoir, and displaying the graph as the map.
In an aspect of the embodiment, each of the well volume/completion parameter, the well stimulation parameter, and the well production parameter are divided into three fuzzy clusters.
In another aspect of the embodiment, each combination is assigned a truth value, and the NFN value is based, in part, on the truth value.
In still another aspect of the embodiment, the well volume/completion parameter is generated from one or more of a well spacing along a lateral length of the hydraulic fracturing job, a bulk volume of the hydraulic fracturing job, initial water saturation to calculate a hydrocarbon pore volume of the hydraulic fracturing job, an initial pressure used in the hydraulic fracturing job, a TOC, or a total number of stages included in the hydraulic fracturing job.
In still another aspect of the embodiment, the well stimulation parameter is generated from one or more of an amount of proppant used in the hydraulic fracturing job or an amount of fluid used in the hydraulic fracturing job.
In yet another aspect of the embodiment, the well production parameter is generated from one or more of a well-head pressure used during a hydraulic fracturing job, an initial reservoir pressure used during the hydraulic fracturing job, or a 180 days cumulative production amount obtained from the hydraulic fracturing job.
In still yet another aspect of the embodiment, the shale reservoir includes a developed portion and an undeveloped portion, the developed portion includes the plurality of wells, and the method further comprises imposing a Cartesian (or other type of) grid on the generated map, selecting seismic attributes from a seismic map of the shale reservoir, using the seismic attributes and NFN density values of the developed portion to design, train, calibrate, and validate a neural network model, deploying neural network model in a forecast mode, setting the seismic attributes of the undeveloped portion to generate NFN density values for the shale reservoir including the undeveloped portion, and generating the map using the NFN density values generated for the undeveloped portion.
The above and other aspects, features, and advantages of the present disclosure will become more apparent in light of the following detailed description when taken in conjunction with the accompanying drawings, in which:
Network in shale formations in the linear model), according to an embodiment;
As noted briefly above, the Natural Fracture Network (NFN) in shale is a significant contributor to well productivity in shale wells. The most significant contribution of inducing hydraulic fractures for the extraction of hydrocarbons is to open the existing network of natural fractures before (or along with) creating new fractures in shale. To determine the location of NFN in shale, engineers and scientist have used seismic surveys and stochastic modeling as inspiration. For example, in numerical reservoir simulation, the NFN is modeled using a stochastic approach. In particular, using information that sometimes may be available from wellbore image logs, or using information from outcrops, important parameters that are needed to randomly generate NFN models are assumed such as fracture density, size, and orientation, etc. Further, core analysis and seismic survey have also been used to define natural fracture distribution in formations and to help model the geometry of their network in shale. However, it is impossible to map the NFN in shale (or any other formation, for that matter) through direct measurements.
The presence and the extent of the Natural Fracture Network determine the productivity of a shale well. Parameters that impact the productivity of shale wells are:
Systems and methods are now provided that can be used to map the NFN in shale. The system and methods take into account the characteristics of the reservoir rock (shale) to determine the degree by which a combination of the completion characteristics and stimulation/hydraulic fracturing characteristics results in a shale well productivity. The systems and methods of the present disclosure employ the characteristics of the reservoir rock (shale) include parameters that can be and are measured, such as well logs, porosity, initial water saturation, formation thickness, total organic carbon (TOC), and the like, and that cannot be directly measured. If, when comparing similar parameters for two wells, the parameters appear to be substantially similar except one well displays a higher productivity than the other and a distinction cannot be identified with all measured reservoir characteristics (for example, total vertical depth drilled (TVD), porosity, initial water saturation, formation thickness, neutron porosity, gamma ray measurement, deep resistivity, density, TOC, and volume of clay (VCL)), then a determination is made that NFN may be affecting the well productivity. In an embodiment of the present invention, field measurements are used in order to generate a map of the “Effective” Natural Fracture Network, which may be used to explain an observed discrepancy between two wells.
Returning to step S102, the plurality of parameters for the set of rules are based on the sub-parameters of each parameter. Each parameter is divided into classes or clusters. The clusters used in this analysis may be “fuzzy (soft) clusters or crisp (hard) clusters” because they do not have crisp boundaries and may overlap each other. In dividing the “well productivity” into fuzzy clusters, each sub-parameter is considered. For example, the sub-parameters taken into account for “well-productivity” include, but are not limited to, the well-head pressure applied during the frac job, the initial reservoir pressure applied during the frac job (may be presented by TVD as a proxy), and the average cumulative production (BOE) of the frac job over 180 days. When the three sub-parameters are taken together in consideration for the “well-productivity” parameter, a pressure-corrected production indicator may then be provided.
With reference to
The sub-parameters considered for “stimulation” include but are not limited to the amount of fluid in barrels per foot of lateral length and the amount of proppant in pounds per foot of lateral length used in the frac job. These sub-parameters are used to represent the quality of the hydraulic fracturing. As such, other parameters that may better represent the quality of the frac job can also be included in this parameter category. In an example using the amount of fluid in barrels per foot of lateral length and the amount of proppant in pounds per foot of lateral length, a supervised fuzzy cluster analysis is used to divide the “stimulation” parameter into three overlapping clusters small, average and large hydraulic jobs. The centers of each cluster are determined and imposed on the algorithm (i.e. supervised) in order to complete the supervised fuzzy cluster analysis. For example, with reference to the graph depicted in
To verify the results, bar graphs are used, as depicted in
The “volume/completion” parameter represents the quality and the size of the shale reservoir, along with the completion practices incorporated for the frac job. Sub-parameters for this category include, but are not limited to, well spacing, the lateral length and formation thickness of the well for calculating bulk volume, porosity and initial water saturation for calculating the Hydrocarbon Pore Volume (HCPV) of the shale reservoir, the initial pressure of the frac job, TOC, and the total number of stages employed for the frac job. In an example, the “volume/completion” parameter, which is represented by eight sub-parameters, namely, well spacing, lateral length, formation thickness, porosity, initial water saturation, TOC, initial pressure, and number of stages, is analyzed by using five of the eight sub-parameters (namely, well spacing, lateral length, formation thickness, porosity, and initial water saturation) to calculate the hydrocarbon pore volume (HCPV) identified as one integrated sub-parameter. The other three sub-parameters (TOC, initial pressure, and number of stages) are then entered into the analysis and interacted with the HCPV integrated sub-parameter to form the clusters representing the volume/completion. Using supervised fuzzy cluster analysis the each subparameter of the volume/completion parameter is divided into three overlapping clusters of small, medium, and large volume/completions.
The cluster centers for small, medium, and large volume/completions are determined by using a process which includes cross plotting all the involved sub-parameters against one another and identifying the location of each cluster center. An example of this process is shown in the graphs depicted in
Once the process of determining the location of cluster centers is completed, the values of each cluster center are plotted in bar graphs to demonstrate that the multi-dimension attributed to each cluster center is internally consistent. For example, the values of HCPV (
The plurality of parameters, namely, “well-productivity,” “stimulation,” and “volume/completion,” are then compared against clusters into which the Natural Fracture Network (NFN) are divided. In an embodiment, the presence and extent of the NFN is divided into three clusters of Minimal, Average, and Extensive. The NFN classes are qualitatively correlated to classes of well productivity based on the assumption that all other parameters are similar. In this regard, in an embodiment, an assumption is made that with all other parameters being equal, higher well productivity is expected in the presence of extensive NFN, or lower well productivity is expected in the presence of minimal NFN.
If a qualitatively linear relationship is further assumed between presence and the extent of the Natural Fracture Network and well productivity (all other parameters being similar), then a qualitative plot, such as one shown in
Each statement regarding the relationship between well productivity and NFN is further qualified (granulized) with a “truth value”. Given that the relationships that we have established are qualitative in nature, and the term “similar” is used for hydrocarbon volume, completion, and hydraulic fracturing, the “truth value” can further qualify the statements and the rules that are made during the next step. The “truth value” includes “Very True”, that has a higher truth value than “True”, which has a higher truth value than “Fairly True”.
The relationship between the presence and the extent of the NFN and well productivity can be either linear (line A as shown in
Once the plurality of parameters is identified, the fuzzy clusters of each parameter and the NFN and well productivity relationship model are determined, then a set of rules are defined to govern the interaction between all of the parameters. As noted above, in an embodiment, the rules may be based on fuzzy logic. Here, Fuzzy Set Theory is used to define the set of rules; as such, the set of rules are fuzzy rules. Fuzzy rules have two distinct characteristics:
Based on the set of rules, an NFN distribution map may be generated using the following steps:
As noted briefly above with respect to method 100, after the rules are defined, historical data related to a well is analyzed at step S104. In an example in which “Well #23” in a Marcellus shale field is analyzed, fuzzy membership values for Volume/Completion, Stimulation, and Productivity are identified. With reference to
Once the fuzzy membership values are obtained, corresponding fuzzy rules are identified which are used to assign corresponding values to the historical data to determine the NFN density value of the well. In particular, taking the fuzzy membership values of “Well #23” identified above into consideration and with reference to
The results of executing all the applicable rules in parallel is the generation of a set of fuzzy membership values for the NFN, as shown in the far right graphs depicted in
Using Artificial Intelligence, a distribution of the Natural Fracture Network of a shale asset in Eagle Ford was mapped.
The heterogeneous nature of the distribution of the Natural Fracture Network that is shown in
To demonstrate the accuracy of the results generated by the methods described above, a few examples are provided. In a first example shown below in Table 1, two wells (Well #BJU3 and Well #RF101) having similar volume/completion and similar lateral lengths are compared. Well #BJU3 was stimulated with a proppant of 802 lbs/ft and Well #RF101 was stimulated with a proppant of 466 lbs/ft, and were thus different frac job sizes. However, Well #BJU3, which was treated with a larger frac job, demonstrated less productivity Well #RF101, which was treated with smaller frac job. After applying the algorithm described above, the discrepancy in well productivity can be explained by attributing the difference to the presence and the extent of the Natural Fracture Network throughout this asset. In particular, the algorithm assigns Well #BJU3 with a NFN density value of 65.76 and Well #RF101 with a NFN density value of 84.08.
Table 1 also shows that the well with better productivity is located in a part of the field with higher NFN density (NFN Density score of 84) while the well with lower productivity, although it has been stimulated with a smaller frac job is located in the part of the filed that has lower NFN density (NFN Density score of 66). In other words, when field measurements that are related to the well (for example, how the well is drilled, how much hydrocarbon volume is accessible to it, how it was completed, how big of a frac job was performed and what choke sizes were used to produce it) cannot explain well productivity, then the only characteristic left to attribute to the well productivity is the presence and the extent of the NFN.
After compiling the NFN density values for every well in an asset and applying the above-described reasoning to every other square foot of the asset based on the field measurements of the asset, the map shown in
Once the Natural Fracture Network is mapped throughout the part of the field where wells have been completed and are under production, the map can be extended to the rest of the field of the asset (play). In this regard, a neural network is designed where the inputs to the network are a set of seismic attributes from the portion of the field of the asset where the NFN map has been generated and the output of the network is the NFN density value generated using the process described above.
First, NFN mapping is performed to generate the NFN Density values for a developed portion of a field at step S2302. In this regard, method 100 is employed to generate the NFN density values. Using the developed portion of the field, NFN mapping is performed to generate the NFN Density values for the developed portion of the field.
Then, a reasonably fine Cartesian (or other type of) grid is then imposed on top of the generated map at step S2304. The Cartesian (or other type of) grid is typically one that is already in existence, since it is necessary when generating the distributed NFN Density. The grid is extended to the entire play, including the undeveloped portions of the shale play.
Next, the seismic survey for the entire shale play is then used and the seismic attributes are then selected at step S2306. The seismic attributes are generated without limitation for the entire shale play, including the developed and the undeveloped portions of the play.
Using the seismic attributes for the developed portion of the field and the NFN density for the same portion of the field, a robust neural network model is designed, trained, calibrated, and validated that correlates the seismic attributes to the NFN density at step S2308.
The model is deployed in a forecast mode, and the set of seismic attributes for the rest of the field (play) is used to generate the Natural Fracture Network for the entire field (play), including an undeveloped portion of the field, at step S2310. For example, the trained neural network may be deployed in a predictive (forecast) mode. The seismic attributes (for example, those that were eventually used to train the neural network in the previous step) are used for each Cartesian (or other type of) grid of the undeveloped portion of the field and fed to the trained neural network. The output of the neural network is the NFN density for the given Cartesian (or other type of) grid.
A determination is made as to whether additional undeveloped portions of the field are to be analyzed at step S2312, and if so, method 2300 reiterates at step S2310. If not, the NFN density for the entire undeveloped portion of the shale field is generated and mapped for the entire undeveloped portion of the field at step S2314 and method 2300 ends.
The embodiments disclosed herein are examples of the disclosure and may be embodied in various forms. For instance, although certain embodiments herein are described as separate embodiments, each of the embodiments herein may be combined with one or more of the other embodiments herein. Specific structural and functional details disclosed herein are not to be interpreted as limiting, but as a basis for the claims and as a representative basis for teaching one skilled in the art to variously employ the present disclosure in virtually any appropriately detailed structure. Like reference numerals may refer to similar or identical elements throughout the description of the figures.
The embodiments disclosed herein are examples of the disclosure and may be embodied in various forms. For instance, although certain embodiments herein are described as separate embodiments, each of the embodiments herein may be combined with one or more of the other embodiments herein. Specific structural and functional details disclosed herein are not to be interpreted as limiting, but as a basis for the claims and as a representative basis for teaching one skilled in the art to variously employ the present disclosure in virtually any appropriately detailed structure. Like reference numerals may refer to similar or identical elements throughout the description of the figures.
The phrases “in an embodiment,” “in embodiments,” “in some embodiments,” or “in other embodiments” may each refer to one or more of the same or different embodiments in accordance with the present disclosure. A phrase in the form “A or B” means “(A), (B), or (A and B).” A phrase in the form “at least one of A, B, or C” means “(A); (B); (C); (A and B); (A and C); (B and C); or (A, B, and C).”
The methods described herein may be executed using a computing device, a simplified block diagram of which is shown in
Any of the herein described methods, programs, algorithms or codes may be converted to, or expressed in, a programming language or computer program. The terms “programming language” and “computer program,” as used herein, each include any language used to specify instructions to a computer, and include (but is not limited to) the following languages and their derivatives: Assembler, Basic, Batch files, BCPL, C, C+, C++, Delphi, Fortran, Java, JavaScript, machine code, operating system command languages, Pascal, Perl, PL1, scripting languages, Visual Basic, metalanguages which themselves specify programs, and all first, second, third, fourth, fifth, or further generation computer languages. Also included are database and other data schemas, and any other meta-languages. No distinction is made between languages which are interpreted, compiled, or use both compiled and interpreted approaches. No distinction is made between compiled and source versions of a program. Thus, reference to a program, where the programming language could exist in more than one state (such as source, compiled, object, or linked) is a reference to any and all such states. Reference to a program may encompass the actual instructions and/or the intent of those instructions.
Any of the herein described methods, programs, algorithms or codes may be contained on one or more machine-readable media or memory 1904. The term “memory” may include a mechanism that provides (e.g., stores and/or transmits) information in a form readable by a machine such a processor, computer, or a digital processing device. For example, the memory 1904 may include a read only memory (ROM), random access memory (RAM), magnetic disk storage media, optical storage media, flash memory devices, or any other volatile or non-volatile memory storage device. Code or instructions contained thereon can be represented by carrier wave signals, infrared signals, digital signals, and by other like signals. The memory 1904 may store an application 1914 and/or data 1916, similar to data and values discussed above with respect to method 100. The application 1914 may, when executed by the controller 1902, cause the method 100 to be performed and to cause the display 1906 to display the data, values, and/or outcomes of the method 100. The network interface 1908 may be configured to connect to a network such as a local area network (LAN) consisting of a wired network and/or a wireless network, a wide area network (WAN), a wireless mobile network, a Bluetooth network, and/or the internet. Input device 1910 may be any device by means of which a user may interact with computing device 1900, such as, for example, a mouse, keyboard, foot pedal, touch screen, and/or voice interface. Output module 1912 may include any connectivity port or bus, such as, for example, parallel ports, serial ports, universal serial busses (USB), or any other similar connectivity port known to those skilled in the art.
It should be understood that the foregoing description is only illustrative of the present disclosure. Various alternatives and modifications can be devised by those skilled in the art without departing from the disclosure. Accordingly, the present disclosure is intended to embrace all such alternatives, modifications and variances. The embodiments described with reference to the attached drawing figures are presented only to demonstrate certain examples of the disclosure. Other elements, steps, methods, and techniques that are insubstantially different from those described above and/or in the appended claims are also intended to be within the scope of the disclosure.
The systems described herein may also utilize one or more controllers to receive various information and transform the received information to generate an output. The controller may include any type of computing device, computational circuit, or any type of processor or processing circuit capable of executing a series of instructions that are stored in a memory. The controller may include multiple processors and/or multicore central processing units (CPUs) and may include any type of processor, such as a microprocessor, digital signal processor, microcontroller, programmable logic device (PLD), field programmable gate array (FPGA), or the like. The controller may also include a memory to store data and/or instructions that, when executed by the one or more processors, causes the one or more processors to perform one or more methods and/or algorithms.
Any of the herein described methods, programs, algorithms or codes may be converted to, or expressed in, a programming language or computer program. The terms “programming language” and “computer program,” as used herein, each include any language used to specify instructions to a computer, and include (but is not limited to) the following languages and their derivatives: Assembler, Basic, Batch files, BCPL, C, C+, C++, Delphi, Fortran, Java, JavaScript, machine code, operating system command languages, Pascal, Perl, PL1, scripting languages, Visual Basic, metalanguages which themselves specify programs, and all first, second, third, fourth, fifth, or further generation computer languages. Also included are database and other data schemas, and any other meta-languages. No distinction is made between languages which are interpreted, compiled, or use both compiled and interpreted approaches. No distinction is made between compiled and source versions of a program. Thus, reference to a program, where the programming language could exist in more than one state (such as source, compiled, object, or linked) is a reference to any and all such states. Reference to a program may encompass the actual instructions and/or the intent of those instructions.
It should be understood that the foregoing description is only illustrative of the present disclosure. Various alternatives and modifications can be devised by those skilled in the art without departing from the disclosure. Accordingly, the present disclosure is intended to embrace all such alternatives, modifications and variances. The embodiments described with reference to the attached drawing figures are presented only to demonstrate certain examples of the disclosure. Other elements, steps, methods, and techniques that are insubstantially different from those described above and/or in the appended claims are also intended to be within the scope of the disclosure.
