SYSTEMS AND METHODS FOR AUTOMATIC GENERATION OF DRILLING SCHEDULES USING MACHINE LEARNING

Abstract

Systems and methods for automatically generating drilling schedules are disclosed. According to one embodiment, a method of predicting rig movement between wells includes receiving historical well data regarding individual well types and historical rig data regarding individual rigs, and generating a Markov Chain model from the historical well data and the historical rig data. The Markov Chain model includes a plurality of states and a plurality of links between states. Each state of the plurality of states is a well class derived from the historical well data. Each link indicates a number of rigs that traveled between individual well classes. The method further includes determining, using the Markov Chain model, a probability of rigs moving between individual well classes, and predicting movement of individual rigs of a plurality of rigs between future wells based at least in part on the Markov Chain model.

Description

BACKGROUND

Oil and gas companies may annually update a business plan that is used for budgeting forecasts and resources allocation. The business plan may involve two steps. A first step is production of a development-drilling program. This step addresses the question of how many wells are needed in the subsequent years in short-term and long-term bases. A typical development-drilling program includes the number of wells needed and their attributes (e.g, vertical, horizontal, oil producer, gas producer, water injector, etc.). A second step involves identifying the resources required to execute and implement the development drilling program. Under this second step, capital expenditure by the oil and gas company is estimated for future years in short-term and long-term bases and is updated annually.

For reliable budgeting forecasts and resource allocation, a detailed drilling schedule may be generated. A typical drilling schedule will contain forecasts about the start of drilling of every well, the completion of drilling, and the name and type of the drilling rigs. These forecasts have to take many factors into consideration, such as the location of the well (e.g., is it in the onshore or the offshore), the type of the well (e.g., horizontal, vertical, deviated), and other parameters. The generation of drilling schedules is done manually, and it may take hundreds of man-hours to pour over well and rig parameters to optimally match rigs to wells. Thus, significant time and costs are put into the generation of drilling schedules.

SUMMARY

Embodiments of the present disclosure are directed to automatic systems and methods for generation of long-term drilling schedules for oil and gas business planning. Generally, embodiments model the movement of rigs between wells while honoring operational and logistical restrictions to mimic reality. The machine-learning algorithm of the present disclosure learns from historical drilling schedules, and extracts rules and applies them for futures wells.

More particularly, embodiments of the present disclosure focus on the second step of the business plan process outlined above by introducing a new procedure to generate automatic drilling schedules that take into consideration the various attributes of the wells and rigs and the dynamic nature of rig movement between wells, in addition to many rules and logics derived from both historical records and user knowledge. It is assumed that the first step of the business plan process has been fulfilled and the development-drilling program has been completed. It is also a prerequisite to have a priori estimates of drilling times and costs for all the wells in the development-drilling program.

Embodiments are fully automatic and produce a detailed drilling schedule in minutes, reducing the turnaround time by orders of magnitude. In addition, embodiments remove subjectivity of the process by implementing consistent logics and rules capturing the movement of rigs, and rig assignment to wells in a coherent manner using a robust dynamic modelling approach.

According to one embodiment, a method of predicting rig movement between wells includes receiving historical well data regarding individual well types and historical rig data regarding individual rigs, and generating a Markov Chain model from the historical well data and the historical rig data. The Markov Chain model includes a plurality of states and a plurality of links between states. Each state of the plurality of states is a well class derived from the historical well data. Each link indicates a number of rigs that traveled between individual well classes. The method further includes determining, using the Markov Chain model, a probability of rigs moving between individual well classes, and predicting movement of individual rigs of a plurality of rigs between future wells based at least in part on the Markov Chain model.

According to another embodiment, a method of drilling wells includes receiving historical well data regarding individual well types and historical rig data regarding individual rigs, and generating a Markov Chain model from the historical well data and the historical rig data. The Markov Chain model includes a plurality of states and a plurality of links between states. Each state of the plurality of states is a well class derived from the historical well data. Each link indicates a number of rigs that traveled between individual well classes. The method also includes determining, using the Markov Chain model, a probability of rigs moving between individual well classes, predicting movement of individual rigs of a plurality of rigs between future wells based at least in part on the Markov Chain model, generating a drilling schedule for the plurality of rigs based at least in part on the predicted movement of the individual rigs, and drilling the wells using the plurality of rigs according to the drilling schedule.

According to yet another embodiment, a system for predicting rig movement between wells includes one or more processors and a non-transitory computer-readable memory storing instructions that, when executed by the one or more processors, causes the one or more processors to receive historical well data regarding individual well types and historical rig data regarding individual rigs and generate a Markov Chain model from the historical well data and the historical rig data. The Markov Chain model includes a plurality of states and a plurality of links between states. Each state of the plurality of states is a well class derived from the historical well data, and each link indicates a number of rigs that traveled between individual well classes. The instructions further cause the one or more processors to determine, using the Markov Chain model, a probability of rigs moving between individual well classes, and predict movement of individual rigs of a plurality of rigs between future wells based at least in part on the Markov Chain model.

It is to be understood that both the foregoing general description and the following detailed description present embodiments that are intended to provide an overview or framework for understanding the nature and character of the claims. The accompanying drawings are included to provide a further understanding of the disclosure, and are incorporated into and constitute a part of this specification. The drawings illustrate various embodiments and together with the description serve to explain the principles and operation.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 graphically illustrates a Markov Chain model according to one or more embodiments described and illustrated herein;

FIG. 2 graphically illustrates a global Markov Chain model according to one or more embodiments described and illustrated herein;

FIG. 3 illustrates information about a global Markov Chain model according to one or more embodiments described and illustrated herein;

FIG. 4 schematically illustrates the use of priorities to control rig movement according to one or more embodiments described and illustrated herein;

FIG. 5 graphically illustrates initial rig positions according to one or more embodiments described and illustrated herein;

FIG. 6 graphically illustrates rig addition and rig release logic according to one or more embodiments described and illustrated herein;

FIG. 7 illustrates simulation parameters used in an example simulation according to one or more embodiments described and illustrated herein;

FIG. 8A illustrates an example five year drilling schedule plot for a plurality of rigs according to one or more embodiments described and illustrated herein;

FIG. 8B illustrates example rig names on an example drilling schedule according to one or more embodiments described and illustrated herein;

FIG. 8C illustrates example well information on an example drilling schedule plot according to one or more embodiments described and illustrated herein; and

FIG. 9 illustrates and example computing device for generating a drilling schedule according to one or more embodiments described and illustrated herein.

DETAILED DESCRIPTION OF THE DISCLOSURE

Embodiments of the present disclosure are directed to systems and methods for generation of long-term drilling schedules for oil and gas business planning. Generally, embodiments model the movement of rigs between wells while honoring operational and logistical restrictions to mimic reality. The machine-learning algorithm of the present disclosure learns from historical drilling schedules, and extracts rules and applies them for futures wells. Use of the Markov Chain models described herein enable embodiments to automatically generate drilling schedules with significantly reduced time and significantly reduced computer processing power as compared with traditional methods.

Various embodiments of systems and methods for predicting rig movements and for generating drilling schedules are descried in detail herein.

1. Learning Rig Capabilities

First, rig capabilities are learned from historical drilling activities. Rig capabilities refer to the various rig parameters, such as, without limitation, shore type (e.g, onshore or offshore), fluid type (e.g, whether a rig can be assigned to oil wells, water wells, or gas wells, or any combinations of these wells), and rig horsepower (e.g., rigs with relatively low horsepower may not be able to drill horizontal wells). Contracting conditions may also be considered as part of rig capabilities because they influence the assignment of rigs to certain wells. Contracting conditions refer to specific contracts that limit the drilling for a given rig(s) to certain type of well(s) in certain field(s).

In embodiments, rig capabilities and movements are learned by building a Markov Chain (MC) model that tracks the movement of each rig in the history and analyses the type of wells drilled in the process. From this, the algorithm computes transition probabilities that control the assignment of rigs to future wells in prediction. Historical rig data is accessed from one or more databases. This historical rig data includes information about the movements of a plurality of rigs, such as the types of wells and under what conditions the rigs moved from well to well. Historical well data is also accessed to gain an understanding of the well parameters of the wells that the rigs drilled in the past. The historical rig data and the historical well data is used to learn the rig capabilities and movements to further predict the movement of rigs to future wells in the business plan.

FIG. 1 depicts a non-limiting example MC Model 100 for a single rig indicating the number of visits to each event of a well in history. In FIG. 1, the type of wells drilled represent the state, and the number on arrows marks the number of times a rig moves between each state and thus defines the links between states. The MC states may be defined by the user and may contain information about the field, the fluid type (oil or gas), location (onshore or offshore), drilling operation (New well, Re-entry or Workover) and well type (vertical, horizontal, multilateral, etc.).

In the example of FIG. 1, the particular rig returned to wells of a type as defined by state A two times, moved to a well of a type as indicated by state B from a well of a type as defined by state A one time, and moved to a well of a type as indicated by state C from a well of a type as defined by state A one time. As stated above, the states (and thus the types of wells) are definable by the user. The MC models for the individual rigs are generated using the historical rig data and the historical well data.

There may be hundreds of rigs required to effectuate the business plan, with each rig having its own MC model. For these MC models to be used in prediction, embodiments combine all individual MC models for all rigs into a single MC model. The advantage of such approach is realized through increasing the number of data points and obtaining more representative transition probabilities. In this approach, all rigs may share the same MC model.

FIG. 2 illustrates an example global MC model 200 that was built from individual rig MC models, and used in a simulation as described in more detail below. The circles (i.e., the states) indicate well classes, and the number of the class represents a well class with a unique set of well attributes as defined by the user. The well classes of the present example are defined in FIG. 3. The arrows indicate the connection between the different well classes, and the number on the directed arrows indicates the number of times a rig moves between the different classes. The data is based on the movements from the individual rigs, and may be obtained from the historical rig and well data itself or from individual MC models as described above. Embodiments provide the ability to modify the global MC model per simulation run. These MC models control the movement of rigs into future wells. These MC models are also flexible in the sense that they can incorporate subject matter experts' preferences and experiences, which may be important when limited data is available in history about one of the well classes.

The example global MC model 200 shows thirteen red circles used to define well classes, labeled from 1 to 13. FIG. 3 provides information about the global MC model used in the dynamic modeling. The parameters used to define these classes are also displayed indicating five non-limiting parameters: location (onshore/offshore), fluid type (Oil/Gas), drilling operation (New, Re-entry/Workover), internal budget allocation information, and the type of the rig (e.g., drilling rig (DLG), coiled tubing operation (CT), and the like). A software program may present a graphical user interface (GUI) to the user for selection of the parameters to define the individual well classes.

Well Class #1, for instance, represents new gas wells located in the offshore. As mentioned previously, these well classes are defined by the user and can vary in each simulation run. The number on the arrows in FIG. 2 indicates how these well classes are connected. These numbers are obtained from historical records by tracking the movement of the rigs between these well classes. User experience and preference can be reflected into these MC models.

To illustrate how embodiments use these MC models, suppose a rig has just completed drilling a well, which is classified under Well Class #4 (offshore-oil-new), and the rig is ready to move to the next well. The developed solver scans all potential wells and assigns a transition probability to each one depending mainly on the well class of each potential well. For this example, the rig will first choose a well under the same well class, since it shows from history that rigs moved 4602 times between wells under Well Class #4. If the solver could not find a well under this class, the next option is to look for wells under Well Class #6 since it shows the second highest number of visits of 102 and so on. It is noted that the probability of a rig moving from one well class to another well class is calculated by dividing the number of times a rig moved between a particular well class to another individual well class by the total number of movements of rigs from the particular well class. In the example of Well Class #4, the probability of a rig moving to another well of Well Class #4 is 97.6%, the probability of a rig moving from a well of Well Class #4 to a well of Well Class #6 is 2.2% and the probability of a rig moving from a well of Well Class #4 to a well of Well Class #5 is 0.02%.

The global MC model 200 also shows that there is no way for a rig on the onshore location to visit an offshore well. This is demonstrated in FIG. 2 by the clear disconnection of Well Classes 1-6 and 7-13. As previously mentioned, the user can modify and customize these MC models as desired.

As described in more detail below, the global MC model 200 is used to assign rigs to futures wells of the business plan by a simulation used to generate drilling schedule.

2. Specifying Rig Movement Priorities Markov Chain models are not the only attributes that may influence how rigs move during simulation. In some embodiments, the user has the ability to add additional options in controlling the assignment of rigs to wells, which may become important when a rig has equal preference to more than one well.

Accordingly, in some embodiments, another set of conditions and priorities can be specified by the user, which may be beneficial when a rig has multiple wells under the same well class to choose from. FIG. 4 displays a schematic illustration of the ability of the user to add additional condition priorities, and how it affects the assignment of rigs during simulation.

As an example, assume there are four wells: W1 through W4, and two scenarios are made. As used herein, “scenario” refers to parameters that are applied for a simulation. In the first scenario, higher priorities are given to drill wells within the same field, before considering the distances between wells. In this case, a rig starting in W1 will move to W2 before W3 because they are within the same field (i.e., field X), even though the distance between W1 and W2 is longer than the distance between W1 and W3. In the second scenario, the priorities are changed and it is opted to consider shorter distances before changing the field. Now the rig, starting in W1, will move to W3 before W2. As mentioned previously, these priorities may be taken into consideration only for wells that have the same well class and share the same transition probability. In the example shown in FIG. 4, W4 will never be drilled from rigs starting in W1, W2, or W3.

The rig movement priorities may be set up considering many attributes (e.g., distance, field, year, well cost, well drill time . . . etc.). Embodiments are not limited by any particular priority rules.

Thus, the simulation may apply these user-defined priorities in conjunction with the MC model when assigning rigs to wells.

3. Initial Rig Information

In a third component, initial rig information is determined. Initial rig information refers to the number of rigs used as the start of the drilling schedule and their initial locations and capabilities. These parameters represent the initial conditions of the developed dynamic model and influences the generated schedule. In other words, to predict the movement of rigs at the start of a prediction period, the initial location of the rigs at the start of the prediction period should be known.

The drilling schedule that the embodiments of the present disclosure generates may be linked or bridged with the latest executed drilling schedule. To make logical and practical bridging between the two schedules, the logic of embodiments of the present disclosure determines which rigs to carry from history into prediction and after which well should the prediction start for each rig.

To determine which rigs to carry from history into prediction, the wells that are at the end of the current year, which is the year before the business plan, are analyzed. For example, if a rig is for a development drilling, then the rig is picked as development drilling rig. If not, then the rig is excluded.

As stated above, embodiments determine after which well should the prediction start for each rig. Drilling of some of the wells from the current year will be completed in the first year of the business plan, and some of the wells from the first year of the business plan will start drilling in the current year. In addition, in some cases wells from the first year's requirements will be completed in the current year. Therefore, this should be taken into consideration to avoid drilling additional wells or missing required wells.

The logic used in embodiments is based on a set of cutoff dates. Hence, the simulator starts by scanning the latest drilling schedule currently being implemented (i.e., current drilling schedule) and based on the cutoff dates, wells are either fixed in the business plan schedule as starting points or they are dropped. As a result, wells that are spudding beyond the cutoff dates are dropped and wells spudding ahead of the cutoff date are fixed (i.e., included in the simulation that creates the drilling schedule, and kept in the drilling schedule as they are without any alteration to their spud or completion date). However, if a well is spudding beyond the cutoff date and it is part of the business plan requirements, the well may be fixed in the business plan schedule.

The cutoff date is the date at which wells that are in the current schedule being implemented (e.g., the previous drilling schedule) will be used in the simulation to predict the new drilling schedule. Beyond the cutoff date is the start of the simulated prediction. For example, if a well is assigned to a particular rig from the latest drilling schedule (i.e., the current drilling schedule being implemented) but it will start drilling beyond the cutoff date, the well will not be included in the simulator and the simulator will predict a different, better-suited well.

FIG. 5 demonstrates an example of how the logic operates with respect to initial rig positioning. In the illustrated embodiment, the time of the start of the business plan is indicated by line 501. A cutoff period 502 is the period of time of when the cutoff dates will be assigned for a particular type of rig. FIG. 5 illustrates one cutoff period for all five illustrated rigs. The cutoff period can be customized to result in different cutoff dates depending on the type of wells.

The time of prediction (line 503) is the time where the simulation will start scanning wells and look for their end of drilling date to either: 1) assign wells after the current well (start of prediction) or 2) include the next well from the latest drilling schedule to the prediction based on the cutoff date and whether nor not the well will be in the business plan.

The time that the predictions are made (i.e., the time of prediction) is indicated by line 504. The time of prediction is the start of the simulation for a specific rig where the system starts assigning wells from the business plan to the specification rig based on the cutoff date and having the well in the business plan or not.

Rigs (1-4) are picked for development drilling scheduling (labeled BI-60 in FIG. 5) as the last well before the prediction cutoff date (line 503) is under well-type code BI-60. However, Rig 5 will not be considered, since the last well was from a different well type. The start of prediction time for each rig is indicated by the black dots. This means that the simulator will attempt to schedule the individual rigs after the black dots. Wells drilled by the individual rigs to the left of the black dots were previously scheduled in prior business plans or drilling schedules. For Rig 1, prediction will start after (Well A), which is a BI-60 well. At the time of prediction (i.e., the time when the simulation is run), Rig 1 is currently drilling Well W. The next well in the operating schedule is Well A, which is in the business plan for which the simulations are run to develop the drilling schedule. The prediction provided by the simulation will begin at the end of the drilling of Well A.

For Rig 2, it is currently drilling Well X at the time the prediction is run, and the prediction for Rig 2 will start at the spud date of Well B, which is part of the business plan requirements and is completed during the cutoff period 502.

For Rig 3, Well D will be dropped because it is not in the business plan (e.g., it does not meet the requirements of the business plan) and prediction will start after Well C. Thus, Rig 3 may be assigned a different well to drill that meets the business plan requirements.

Regarding Rig 4, it is drilling Well Y at the time the prediction is run. Prediction will start after Well E although Well E is not part of the business plan. However, since prediction occurs after the cutoff period 502 and the date of the start of the business plan (line 501), Well E will be completed and then the prediction for Rig 4 will occur after completion of Well E.

For Rig 5, it is drilling Well Z at the time of prediction. However, Well F, which is the next assigned well according to the prior drilling schedule, is not part of the business plan, and the simulation has no wells in the business plan for Rig 5 to drill. For example, the global MC model indicates a zero probability of assigning a rig-type similar to Rig 5 to a well in the business plan. Therefore, Rig 5 will be excluded from the drilling schedule.

For practicality and to account for the huge variation in well type and drilling durations, the cutoff dates are customizable. Assigning different cutoff periods for different types of wells based on the time it usually takes to drill a well will yield different cutoff dates. For example, it is possible to have different cutoff dates based on fluid type, shore type, filed, well type, etc. As a specific example, gas wells generally take longer to be drilled and completed than oil wells. Therefore, to start the prediction for rigs with different drilling objectives and capabilities, it is practical to have different cutoff dates to account for the time that these rigs take to complete a well, and to eliminate well completion variation prior to the start of the prediction. Moreover, sensitivity analysis can be run in a timely manner to identify the most realistic cutoff dates.

4. Adding/Releasing Rig Logics

In some embodiments, the capability to add future rigs or release existing rigs to and from the drilling schedule is provided. Particularly, this feature defines the implemented logics in adding new rigs when initial rigs cannot drill all needed wells of the business plan in a given year, and in releasing rigs when excessive number of rigs are available. Rig addition and release have slightly different requirements, both of which are described below.

Rig addition is used when the simulation determines that the number of rigs in the business plan is less than the number of rigs required to drill all the wells of the business plan within the predetermined period of time. Referring to FIG. 6, a chart 600 illustrating the scheduling for four rigs is schematically illustrated. Rig 1 is drilling wells of a first well class and complete all of its wells by the end of the business plan period indicated by line 601. However, Rig 2, which is drilling wells of a second well class, cannot complete all of its wells within the business plan period. Thus, another rig should be provided to complete the wells that Rig 2 cannot complete within the business plan time period.

The logic of the simulation gives priority to capitalize on existing resources as much as possible. The logic allows the simulator to scan wells and assign them to the existing rigs in the business plan. However, by the end of the prediction of each year in the plan some of the future wells are not found on the schedule for that year due to resource limitations. In this case, one or more rigs will be added. Referring again to FIG. 6, the simulation determines that a new Rig X must be added to complete the wells that Rig 2 cannot complete. The logic is such that it insures that any new rig starts from the right time in the year so it continues to the end of the year and continue throughout the descendent years. This feature ensures the avoidance of making gaps in the drilling schedule and/or illogical rig addition/releases activities. As shown in FIG. 6, Rig X is contracted to complete its wells at the end of the business plan period. It is noted that, if Rig 3 were drilling wells in the same well class as Rig 2, it would be moved to finish the wells of Rig 2 rather than adding Rig X.

Moreover, the logic allows the ability to set the maximum number of rigs to be added every year. This feature permits for resource management, sensitivities analysis and scenario planning. The rig cap feature is customizable, where new rigs are allowed or prohibited according to fluid type, shore type, field type, etc.

In the case of rig release, the situation is reversed where the number of available rigs is more than the number of rigs needed to drill all the wells of the business plan. In such cases, the logic of the simulation allows for the releasing of rigs. Referring again to FIG. 6, Rig 3 and Rig 4 are drilling wells of the same well class. However, Rig. 3 will be completed with its assigned wells before the end of the business plan period. The simulation does not have any more wells to assign Rig 3 after completion. For example, the global MC model indicates a zero probability of assigning Rig 3 to any wells of the same well class. In this case, Rig. 3 is released after drilling its final well.

However, there are some exceptions. For example, in some cases a rig need to be released toward the end of the year, but, by looking at the next year's requirement, this rig is needed or an equivalent rig will be contracted. In such a case, the logic allows the rig to continue drilling wells from the next year. This logic allows the ability to avoid illogical rig releases since releasing a rig that needs to be replaced in five or six months is impractical.

5. Simulation and Drilling Schedule Output

Once all simulation parameters are specified, simulations are ready to launch. Embodiments enable a detailed drilling scheduling for a three-year business plan to be produced within minutes or even less than a minute.

The simulation is performed using the user-preferences for the MC model, the rig movement priorities, initial rig location information, adding/removing rig parameters, and any other specified user parameters. For each simulation forecast, the output file contains all specifications set by the user, including maximum number of additional rigs that is allowed to be added per year, the rig stopping date in gas and oil wells, in addition to the cutoffs dates for releasing rigs. FIG. 7 shows an example of simulation parameters used in one of the simulations. In FIG. 7, the parameters are user-defined inputs, and embodiments offer the flexibility of varying the parameters in every run. The non-limiting example parameters include simulation start year, simulation end year, global multiplier to well cost, global multiplier to well drill time, maximum number of rigs added for various years of the simulation period, rig starting/stopping month and day in gas and oil wells, released rigs cut-off month, and released rigs cut-off day. It is noted that the global multiplier to well cost and the global multiplier to well drill time are to account for higher cost and time or lower cost and time to increase or decrease the estimated time and cost taken to drill future wells. These parameters may be set to run any correction required in the case of an overestimate of the cost and time where a less than 1 multiplier can be applied or underestimate of the cost and time where a higher than 1 multiplier can be applied.

The released rigs cut-off month and released rigs cut-off day sets the date wherein any rig that is completed with no additional wells assigned before this date is released. For example, any rig that is finished drilling its wells before Jul. 1, 2024 and has no additional wells to drill will be released from the drilling schedule.

When the simulation is run, the initial rig location information is analyzed as stated above. Then, the global MC model is referenced to determine probabilities for moving rigs from well to well. The logic assigns rigs to the wells matching a well class having the highest probability according to the global MC model. Further, user preferences are also accessed to ensure that user-defined preferences and rules are followed when assigning rigs. Rigs are added and dropped from the drilling schedule according to the rig addition/removal logic as described above.

The simulation creates standardized output files for each simulation case. As an example and not a limitation, the standardized output contains the following example files for each simulation case:

• • Drilling schedule plot
  • Markov-Chain plot
  • Rig-year plot
  • Well count plot
  • Detailed Output file summarizing the events of the dynamic modeling
  • Tabulated drilling schedule in Excel format
  • The drilling schedule plot contains the drilling schedule and sequence for all rigs in one view.

The drilling schedule plot may be detailed enough to demonstrate the following non-limiting information:

• • Rig Name
  • Well Name
  • Shore Type
  • Well Type
  • Well Objective (e.g., producer well, injector well, observation well, water supply well, evaluation well, etc.)

FIG. 8A illustrates an example drilling schedule for a five-year business plan. The x-axis is time and the y-axis lists all of the rigs of the business plan. The example drilling schedule includes hundreds of rigs and hundreds of wells. The wells are represented by the colored bars extending on the x-axis. Some drilling schedules may include thousands of rigs and wells.

The bars representing the wells may be color-coded to represent the targeted year for drilling. As a non-limiting example, wells targeted to be drilled in 2021 may be green in color. It is noted that wells having a targeted year may be drilled partially in the previous year and/or the year after the targeted year depending on the schedule. However, the color coding may provide the user with a quick glance at the drilling activity in a given year.

FIG. 8B shows zoomed-in portion 800A of the drilling schedule 800 and illustrates some rig names on the drilling schedule 800. In the illustrated example, each bar includes information about a well or a number of wells for the given targeted year. FIG. 8C shows a zoomed-in portion 800B of the drilling schedule 800 near a transition between the year 2021 and the year 2022. As shown in FIG. 8C, each bar may include multiple wells. Thus, in these cases, a rig will drill multiple wells in one year.

The drilling schedule may then be used to deploy and schedule rigs to extract hydrocarbons from the wells of the business plan.

6. Budgeting Forecasts

In some embodiments, budgeting forecasts are computed using the drilling schedules. These budget forecasts may be used to estimate cost and rig level requirements per year. As a non-limiting example, the following equations maybe used to determine an annual budget:

$\begin{array}{cc}\mathrm{Total}\mathrm{drilling}\mathrm{cost}\mathrm{in}\mathrm{year}\left(j\right)=\sum _{i=1}^n\left(\frac{\mathrm{Partial}\mathrm{drill}\mathrm{time}\mathrm{of}\mathrm{well}\left(i\right)\mathrm{in}\mathrm{year}\left(j\right)}{\mathrm{Total}\mathrm{drill}\mathrm{time}\mathrm{of}\mathrm{well}\left(i\right)}\right)*\mathrm{Total}\mathrm{drill}\mathrm{cost}\mathrm{of}\mathrm{well}\left(i\right)& \left(i\right)\end{array}$

where n is the total number of wells drilled fully or partly in year (j). As shown in FIG. 8, some wells are drilled over the course of two years. Thus, according to the equation above, only the portion drilled in a specific year is applied to that specific year. This budget forecast may be helpful in business planning for future years.

Embodiments of the present disclosure may be implemented by a computing device, and may be embodied as computer-readable instructions stored on a non-transitory memory device. FIG. 9 depicts an example computing device 900 configured to perform the functionalities described herein. The example computing device 900 provides a system for generating drilling schedules and budgeting forecasts, and/or a non-transitory computer usable medium having computer readable program code for generating drilling schedules and budgeting forecasts embodied as hardware, software, and/or firmware, according to embodiments shown and described herein. While in some embodiments, the computing device 900 may be configured as a general purpose computer with the requisite hardware, software, and/or firmware, in some embodiments, the computing device 900 may be configured as a special purpose computer designed specifically for performing the functionality described herein. It should be understood that the software, hardware, and/or firmware components depicted in FIG. 9 may also be provided in other computing devices external to the computing device 900 (e.g., data storage devices, remote server computing devices, and the like).

As also illustrated in FIG. 9, the computing device 900 (or other addition computing devices) may include a processor 930, input/output hardware 932, network interface hardware 934, a data storage component 936 (which may store historical data 938 (e.g., historical well data and historical rig data), business plan rig data 938B, business plan well data 938C, and any other data), and a non-transitory memory component 940. The memory component 940 may be configured as volatile and/or nonvolatile computer readable medium and, as such, may include random access memory (including SRAM, DRAM, and/or other types of random access memory), flash memory, registers, compact discs (CD), digital versatile discs (DVD), and/or other types of storage components. Additionally, the memory component 940 may be configured to store operating logic 942, learning logic 943, and simulation logic 944 (each of which may be embodied as computer readable program code, firmware, or hardware, as an example). A local interface 946 is also included in FIG. 9 and may be implemented as a bus or other interface to facilitate communication among the components of the computing device 900.

The processor 930 may include any processing component configured to receive and execute computer readable code instructions (such as from the data storage component 936 and/or memory component 940). The input/output hardware 932 may include a graphics display device, keyboard, mouse, printer, camera, microphone, speaker, touch-screen, and/or other device for receiving, sending, and/or presenting data. The network interface hardware 934 may include any wired or wireless networking hardware, such as a modem, LAN port, wireless fidelity (Wi-Fi) card, WiMax card, mobile communications hardware, and/or other hardware for communicating with other networks and/or devices, such as to receive the historical data 938A from various sources, for example.

It should be understood that the data storage component 936 may reside local to and/or remote from the computing device 900, and may be configured to store one or more pieces of data for access by the computing device 900 and/or other components. As illustrated in FIG. 9, the data storage component 936 may historical data 938A, which in at least one embodiment includes historical rig data and historical well data. The historical data 938A may be stored in one or more data storage devices. Similarly, business plan rig data 938B and business plan well data 938C may be stored by the data storage component 936 and may include information relating to the rigs (e.g., rig number, rig capabilities, etc.) and the wells (e.g., well class, well type, well parameters, etc.) of the current business plan. In another embodiment, the computing device 900 may be coupled to a remote server or other data storage device that stores the relevant data. Other data to perform the functionalities described herein may also be stored in the data storage component 936 (e.g., initial rig location information and prior drilling schedules).

Included in the memory component 940 may be the operating logic 942, the learning logic 943, and the simulation logic 944. The operating logic 942 may include an operating system and/or other software for managing components of the computing device 900. The operating logic 942 may also include computer readable program code for displaying the graphical user interface used by the user to input parameters and review results of the simulations. Similarly, the learning logic 943 may reside in the memory component 940 and may be configured to facilitate generation of the MC models described above to learn the movements of the rigs of the business plan. The simulation logic 944 may be configured to run the simulations described herein to generate the drilling schedules and/or budgeting forecasts.

The components illustrated in FIG. 9 are merely exemplary and are not intended to limit the scope of this disclosure. More specifically, while the components in FIG. 9 are illustrated as residing within the computing device 900, this is a non-limiting example. In some embodiments, one or more of the components may reside external to the computing device 900.

It should now be understood that embodiments of the present disclosure are directed to systems and methods for automatically generating drilling schedules for a business plan to extract hydrocarbons using machine learning. Embodiments learn the historical movement of rigs between wells by using a Markov Chain model. The Markov Chain model is used to predict the probability of the movement of rigs to drill future wells in a business plan. The probabilities of rig movement, along with user-defined parameters, preferences, and logic are used to automatically generated drilling schedules and/or budgeting forecasts that are used to deploy rigs for completion of the business plan.

Having described the subject matter of the present disclosure in detail and by reference to specific embodiments thereof, it is noted that the various details disclosed herein should not be taken to imply that these details relate to elements that are essential components of the various embodiments described herein, even in cases where a particular element is illustrated in each of the drawings that accompany the present description. Further, it will be apparent that modifications and variations are possible without departing from the scope of the present disclosure, including, but not limited to, embodiments defined in the appended claims. More specifically, although some aspects of the present disclosure are identified herein as preferred or particularly advantageous, it is contemplated that the present disclosure is not necessarily limited to these aspects.

Claims

1. A method of predicting rig movement between wells, the method comprising: receiving historical well data regarding individual well types and historical rig data regarding individual rigs;generating a Markov Chain model from the historical well data and the historical rig data, wherein: the Markov Chain model comprises a plurality of states and a plurality of links between states;each state of the plurality of states is a well class derived from the historical well data; andeach link indicates a number of rigs that traveled between individual well classes;determining, using the Markov Chain model a probability of rigs moving between individual well classes; andpredicting movement of individual rigs of a plurality of rigs between future wells based at least in part on the Markov Chain model.

2. The method of claim 1, further comprising deploying the plurality of rigs according to the predicted rig movement.

3. The method of claim 1, wherein: the Markov Chain model comprises a plurality of well classes corresponding to the plurality of states; andeach well class of the plurality of well classes is defined by one or more well attributes.

4. The method of claim 3, wherein the one or more well attributes comprises one or more of: shore type, fluid type, drilling operation, and well type.

5. The method of claim 1, further comprising receiving the one or more well attributes from a user.

6. The method of claim 1, further comprising receiving one or more rig movement priorities, wherein the predicting movement of individual rigs of the plurality of rigs moving between future wells is further based at least in part on the one or more rig movement priorities.

7. The method of claim 1, further comprising receiving initial rig information for the plurality of rigs, wherein the predicting movement of individual rigs of the plurality of rigs moving between future wells is further based at least in part on the initial rig information.

8. A method of drilling wells, the method comprising: receiving historical well data regarding individual well types and historical rig data regarding individual rigs;generating a Markov Chain model from the historical well data and the historical rig data, wherein: the Markov Chain model comprises a plurality of states and a plurality of links between states;each state of the plurality of states is a well class derived from the historical well data; andeach link indicates a number of rigs that traveled between individual well classes;determining, using the Markov Chain model, a probability of rigs moving between individual well classes; andpredicting movement of individual rigs of a plurality of rigs between future wells based at least in part on the Markov Chain model;generating a drilling schedule for the plurality of rigs based at least in part on the predicted movement of the individual rigs;drilling the wells using the plurality of rigs according to the drilling schedule.

9. The method of claim 8, wherein: the Markov Chain model comprises a plurality of well classes corresponding to the plurality of states; andeach well class of the plurality of well classes is defined by one or more well attributes.

10. The method of claim 9, wherein the one or more well attributes comprises one or more of: shore type, fluid type, drilling operation, and well type.

11. The method of claim 9, further comprising receiving the one or more well attributes from a user.

12. The method of claim 8, further comprising receiving one or more rig movement priorities, wherein the predicting movement of individual rigs of the plurality of rigs moving between future wells is further based at least in part on the one or more rig movement priorities.

13. The method of claim 8, further comprising receiving initial rig information for the plurality of rigs, wherein the predicting movement of individual rigs of the plurality of rigs moving between future wells is further based at least in part on the initial rig information.

14. The method of claim 8, further comprising updating the drilling schedule by adding a new rig to the drilling schedule when at least one future well is not included within the drilling schedule.

15. The method of claim 8, further comprising removing a rig from the drilling schedule after completion of a last well when a probability of moving the rig to a new well is zero based on the Markov Chain model.

16. The method of claim 8, further comprising calculating a budget forecast from the drilling schedule.

17. A system for predicting rig movement between wells, the system comprising: one or more processors; anda non-transitory computer-readable memory storing instructions that, when executed by the one or more processors, causes the one or more processors to: receive historical well data regarding individual well types and historical rig data regarding individual rigs;generate a Markov Chain model from the historical well data and the historical rig data, wherein: the Markov Chain model comprises a plurality of states and a plurality of links between states;each state of the plurality of states is a well class derived from the historical well data; andeach link indicates a number of rigs that traveled between individual well classes;determine, using the Markov Chain model, a probability of rigs moving between individual well classes; andpredict movement of individual rigs of a plurality of rigs between future wells based at least in part on the Markov Chain model.

18. The system of claim 17, further comprising deploying the plurality of rigs according to the predicted rig movement.

19. The system of claim 17, wherein: the Markov Chain model comprises a plurality of well classes corresponding to the plurality of states; andeach well class of the plurality of well classes is defined by one or more well attributes.

20. The system of claim 17, further comprising receiving one or more rig movement priorities, wherein the predicting movement of individual rigs of the plurality of rigs moving between future wells is further based at least in part on the one or more rig movement priorities.