System and method for well interference detection and prediction
Systems and methods for generating an interference prediction for a target well are disclosed herein. A computing system generates a plurality of interference metrics for a plurality of interference events. For each well, the computing system generates a graph based representation of the well and its neighboring wells. The computing system generates a predictive model using a graph-based model by generating a training data set and learning, by the graph-based model, an interference value for each interference event based on the training data set. The computing system receives, from a client device, a request to generate an interference prediction for a target well. The computing system generates, via the predictive model, an interference metric based on the one or more metrics associated with the target well.
1. A method of generating an interference prediction for a target well, comprising: generating, by a computing system, a plurality of interference metrics for a plurality of interference events for a plurality of wells using decline curve analysis based on historical production information for each well, wherein the plurality of interference events comprises a plurality of offset well completion events, and wherein the plurality of interference metrics comprises performance data for the plurality of wells both before and after each corresponding interference event;
for each well, generating, by the computing system a graph-based representation of the well and its neighboring wells;
generating, by the computing system, a predictive model using a graph-based model, comprising:
generating a training data set comprising the plurality of interference metrics for the plurality of wells, well design, completion information, and geological information for each well, and a numerical representation of the graph-based representation of each well; and
learning, by the graph-based model, an interference value for each interference event based on the plurality of interference metrics for the plurality of wells, the well design, the completion information and the geological information for each well, and the numerical representation of the graph-based representation of each well;
receiving, by the computing system from a client device, a request to generate an interference prediction for a target interference event for a target well, the request comprising one or more metrics associated with the target well, the target interference event corresponding to completion of a target offset well; and
generating, by the computing system via the predictive model, an interference metric based on the one or more metrics associated with the target well.
2. The method of claim 1 , wherein the generating the graph-based representation of the well and its neighboring wells, comprising:
generating a node for each well and each neighboring well; and
connecting each node with a directed edge, wherein the directed edge signifies a relationship between a starting node and an ending node.
3. The method of claim 2 , wherein connecting each node with the directed edge comprises:
determining that the starting node and the ending node are connected based on spacing information and time completion.
4. The method of claim 1 , wherein the one or more metrics associated with the target well comprises one or more of a target location of the target well, target geological information associated with the target well, target completion information, target offset well information, target production information, and target completion information associated with each offset well.
5. The method of claim 1 , further comprising:
generating a graphical representation of the target well's relationship with one or more offset wells.
6. The method of claim 5 , further comprising:
converting the graphical representation of the target well's relationship to a matrix-based representation.
7. The method of claim 1 , wherein the graph-based model comprises:
a first hidden layer of neurons configured to generate the target well's relationship to each offset well; and
a second hidden layer of neurons configured to generate each offset well's relationship to its respective neighboring wells.
8. A method of generating an interference prediction model, comprising:
generating, by a computing system, a plurality of interference metrics for a plurality of interference events for a plurality of wells using decline curve analysis based on historical production information for each well, wherein the plurality of interference events comprises a plurality of offset well completion events, and wherein the plurality of interference metrics comprises performance data for the plurality of wells both before and after each corresponding interference event;
for each well, generating, by the computing system a graph-based representation of the well and its neighboring wells;
generating, by the computing system, a predictive model using a graph-based model, comprising:
generating a training data set comprising the plurality of interference metrics for the plurality of wells, the historical production information for each well, and a numerical representation of the graph-based representation of each well; and
learning, by the graph-based model, an interference value for each interference event based on the plurality of interference metrics for the plurality of wells, the historical production information for each well, and the numerical representation of the graph-based representation of each well.
9. The method of claim 8 , wherein the generating the graph-based representation of the well and its neighboring wells, comprising:
generating a node for each well and each neighboring well; and
connecting each node with a directed edge, wherein the directed edge signifies a relationship between a starting node and an ending node.
10. The method of claim 9 , wherein connecting each node with the directed edge comprises:
determining that the starting node and the ending node are connected based on spacing information and time completion.
11. The method of claim 8 , wherein, learning, by the graph-based model, the interference value for each interference event comprises:
comparing each predicted interference value to an actual interference value; and
minimizing an error between the predicted interference value and the actual interference value.
12. The method of claim 8 , wherein the predictive model is trained to predict an interference event for a target well based on one or more metric associated with the target well.
13. The method of claim 8 , further comprising:
generating a graphical representation of the target well's relationship with one or more offset wells; and
converting the graphical representation of the target well's relationship to the numerical representation of the graph-based representation.
14. The method of claim 8 , wherein the graph-based model comprises:
a first hidden layer of neurons configured to generate the target well's relationship to each offset well; and
a second hidden layer of neurons configured to generate each offset well's relationship to its respective neighboring wells.
15. A method of generating an interference prediction for a target well, comprising:
receiving, by a computing system from a client device, a request to generate an interference prediction for an interference event for a target well, the request comprising one or more metrics associated with the target well, the one or more metrics comprising geographical information for the target well, wherein the interference event corresponds to an offset well completion event;
retrieving, by the computing system, offset well information for the target well, wherein each offset well is a neighbor with the target well;
inputting, by the computing system, the one or more metrics and the offset well information into a graph-based model that is trained to generate well interference predictions, the graph-based model trained using a plurality of interference metrics for a plurality of interference events for a plurality of wells using decline curve analysis based on historical production information for each well, wherein the plurality of interference events comprises a plurality of offset well completion events, and wherein the plurality of interference metrics comprises performance data for the plurality of wells both before and after each corresponding interference event;
generating, by the graph-based model, the interference prediction for the target well based on the one or more metrics and the offset well information; and
outputting, by the computing system, the interference prediction for the target well.
16. The method of claim 15 , wherein the one or more metrics further comprises completion information associated with the target well.
17. The method of claim 15 , further comprising:
generating a graphical representation of the target well and each offset well.
18. The method of claim 17 , wherein generating the graphical representation of the target well and each offset well comprises:
generating a node for each well and each offset well; and
connecting each node with a directed edge, wherein the directed edge signifies a relationship between a starting node and an ending node.
19. The method of claim 18 , wherein connecting each node with the directed edge comprises:
determining that the starting node and the ending node are connected based on spacing information and time completion.
20. The method of claim 15 , wherein the graph-based model comprises:
a first hidden layer of neurons configured to generate the target well's relationship to each offset well; and
a second hidden layer of neurons configured to generate each offset well's relationship to its respective neighboring wells.