Planning a well configuration using geomechanical parameters

1. A system comprising:

a processing device; and

a non-transitory computer-readable medium including program code that is executable by the processing device for causing the processing device to:

determine local variations of regional stress using formation data detected by sensors and representing a natural fracture network in a region;

generate a quality index for at least one area in the region by combining the local variations of regional stress and petrophysical properties that include a porosity that is computed for the region and a permeability that is computed for the region, wherein the quality index is a sum of the porosity adjusted by a first constant, the permeability adjusted by a second constant, and the local variations of regional stress adjusted by a third constant;

generate a plurality of flow simulations for hydrocarbon recovery of simulated wellbores in the region by:

generating, using the quality index, a plurality of well configuration models for areas in the region;

for each well configuration model of the plurality of well configuration models,

generate a reservoir geomechanical model for a geomechanical parameter that includes propagation characteristics of hydraulic fractures;

determine new hydraulic fractures by simulating propagation through the hydraulic fractures of the reservoir geomechanical model and using geomechanical rules; and

determine a projected hydrocarbon recovery rate by simulating flow with the new hydraulic fractures;

select a well placement plan for a well in the region using a plurality of projected hydrocarbon recovery rates from the plurality of well configuration models; and

output a command that causes an adjustment to a characteristic of a well tool based on the well placement plan that is selected.

2. The system of claim 1 , wherein the non-transitory computer-readable medium includes program code that is executable by the processing device for causing the processing device to select the well placement plan by:

generating a teaching set using the plurality of projected hydrocarbon recovery rates and the plurality of well configuration models;

teaching, using the teaching set, a machine-learning model to output hydrocarbon recovery rates corresponding to well configuration models; and

analyzing the plurality of well configuration models using an optimizer and the machine-learning model to determine the well placement plan among a plurality of well placement plans having a highest projected hydrocarbon recovery rate.

3. The system of claim 1 , wherein the non-transitory computer-readable medium includes program code that is executable by the processing device for causing the processing device to select the well placement plan by:

building a response surface using the plurality of projected hydrocarbon recovery rates and the plurality of well configuration models; and

analyzing the response surface using a minimization algorithm to determine the well placement plan among a plurality of well placement plans having a highest projected hydrocarbon recovery rate.

4. The system of claim 1 , wherein the non-transitory computer-readable medium includes program code that is executable by the processing device for causing the processing device to determine the local variations of regional stress by simulating the local variations of regional stress using a material point method-based algorithm.

5. The system of claim 1 , wherein the non-transitory computer-readable medium includes program code that is executable by the processing device for causing the processing device to generate the plurality of well configuration models by randomly generating the well configuration models, wherein randomly generating the well configuration models includes randomly selecting a well geometry from any possible well geometries.

6. The system of claim 1 , wherein the formation data includes seismic data or image logs that represent the petrophysical properties.

7. The system of claim 1 , wherein the operation of generate the quality index includes, for each area of the at least one area in the region:

QI=α 1 ϕ+α 2 ν+α 3 Δσ

wherein ϕ is the porosity, ν is the permeability, and Δσ is a negative of the local variations of regional stress of the area, and wherein α 1 , α 2 , and α 3 , are constants such that α 1 +α 2 +α 3 =1.

8. A non-transitory computer-readable medium including program code that is executable by a processing device for causing the processing device to:

determine local variations of regional stress using formation data detected by sensors and representing a natural fracture network in a region;

generate a quality index for at least one area in the region by combining the local variations of regional stress and petrophysical properties that include a porosity that is computed for the region and a permeability that is computed for the region, wherein the quality index is a sum of the porosity adjusted by a first constant, the permeability adjusted by a second constant, and the local variations of regional stress adjusted by a third constant;

generate a plurality of flow simulations for hydrocarbon recovery of simulated wellbores in the region by:

generating, using the quality index, a plurality of well configuration models for areas in the region;

for each well configuration model of the plurality of well configuration models,

generate a reservoir geomechanical model for a geomechanical parameter that includes propagation characteristics of hydraulic fractures;

determine new hydraulic fractures by simulating propagation through the hydraulic fractures of the reservoir geomechanical model and using geomechanical rules; and

determine a projected hydrocarbon recovery rate by simulating flow with the new hydraulic fractures;

select a well placement plan for a well in the region using a plurality of projected hydrocarbon recovery rates from the plurality of well configuration models; and

output a command that causes an adjustment to a characteristic of a well tool based on the well placement plan that is selected.

9. The non-transitory computer-readable medium of claim 8 , wherein the non-transitory computer-readable medium includes program code that is executable by the processing device for causing the processing device to select the well placement plan by:

generating a teaching set using the plurality of projected hydrocarbon recovery rates and the plurality of well configuration models;

teaching, using the teaching set, a machine-learning model to output hydrocarbon recovery rates corresponding to well configuration models; and

analyzing the plurality of well configuration models using an optimizer and the machine-learning model to determine the well placement plan among a plurality of well placement plans having a highest projected hydrocarbon recovery rate.

10. The non-transitory computer-readable medium of claim 8 , wherein the non-transitory computer-readable medium includes program code that is executable by the processing device for causing the processing device to select the well placement plan by:

building a response surface using the plurality of projected hydrocarbon recovery rates and the plurality of well configuration models; and

analyzing the response surface using a minimization algorithm to determine the well placement plan among a plurality of well placement plans having a highest projected hydrocarbon recovery rate.

11. The non-transitory computer-readable medium of claim 8 , wherein the non-transitory computer-readable medium includes program code that is executable by the processing device for causing the processing device to determine the local variations of regional stress by simulating the local variations of regional stress using a material point method-based algorithm.

12. The non-transitory computer-readable medium of claim 8 , wherein the non-transitory computer-readable medium includes program code that is executable by the processing device for causing the processing device to generate the plurality of well configuration models by randomly generating the well configuration models, wherein randomly generating the well configuration models includes randomly selecting a well geometry from any possible well geometries.

13. The non-transitory computer-readable medium of claim 8 , wherein the formation data includes seismic data or image logs that represent the petrophysical properties.

14. A computer-implemented method comprising:

determining local variations of regional stress using formation data detected by sensors and representing a natural fracture network in a region;

generating a quality index for at least one area in the region by combining the local variations of regional stress and petrophysical properties that include a porosity that is computed for the region and a permeability that is computed for the region, wherein the quality index is a sum of the porosity adjusted by a first constant, the permeability adjusted by a second constant, and the local variations of regional stress adjusted by a third constant;

generating a plurality of flow simulations for hydrocarbon recovery of simulated wellbores in the region by:

generating, using the quality index, a plurality of well configuration models for areas in the region;

for each well configuration model of the plurality of well configuration models,

generate a reservoir geomechanical model for a geomechanical parameter that includes propagation characteristics of hydraulic fractures;

determine new hydraulic fractures by simulating propagation through the hydraulic fractures of the reservoir geomechanical model and using geomechanical rules; and

determine a projected hydrocarbon recovery rate by simulating flow with the new hydraulic fractures;

selecting a well placement plan for a well in the region using a plurality of projected hydrocarbon recovery rates from the plurality of well configuration models; and

outputting a command that causes an adjustment to a characteristic of a well tool based on the well placement plan that is selected.

15. The computer-implemented method of claim 14 , wherein selecting the well placement plan comprises:

generating a teaching set using the plurality of projected hydrocarbon recovery rates and the plurality of well configuration models;

teaching, using the teaching set, a machine-learning model to output hydrocarbon recovery rates corresponding to well configuration models; and

analyzing the plurality of well configuration models using an optimizer and the machine-learning model to determine the well placement plan among a plurality of well placement plans having a highest projected hydrocarbon recovery rate.

16. The computer-implemented method of claim 14 , wherein selecting the well placement plan comprises:

building a response surface using the plurality of projected hydrocarbon recovery rates and the plurality of well configuration models; and

analyzing the response surface using a minimization algorithm to determine the well placement plan among a plurality of well placement plans having a highest projected hydrocarbon recovery rate.

17. The computer-implemented method of claim 14 , wherein determining the local variations of regional stress includes simulating the local variations of regional stress using a material point method-based algorithm, and wherein generating the plurality of well configuration models includes randomly generating the well configuration models, wherein randomly generating the well configuration models includes randomly selecting a well geometry from any possible well geometries.

18. The computer-implemented method of claim 14 , wherein the formation data includes seismic data or image logs that represent the petrophysical properties.