Systems and methods for forecasting well interference
Disclosed are systems and methods for obtaining input data comprising properties associated with at least one parent well and a child well associated with the at least one parent well, dividing the input data into a training data subset, a validation data subset, and a test data subset, selecting at least one machine learning model using the training data subset, the validation data subset, and the test data subset based on k-fold cross validation, tuning hyper parameters for each of the at least one machine learning model, and generating a learning output using the at least one machine learning model and the hyper parameters for each of the at least one machine learning model, the learning output indicating a test root-mean-square error (RMSE) and a training RMSE.
1. A method comprising:
obtaining, by at least one processor, input data comprising properties associated with at least one parent well and a child well associated with the at least one parent well;
dividing, by the at least one processor, the input data into a training data subset, a validation data subset, and a test data subset;
selecting, by the at least one processor, at least one machine learning model using the training data subset, the validation data subset, and the test data subset based on k-fold cross validation;
tuning, by the at least one processor, hyper parameters for each of the at least one machine learning model; and
generating, by the at least one processor, a learning output using the at least one machine learning model and the hyper parameters for each of the at least one machine learning model, the learning output indicating a test root-mean-square error (RMSE) and a training RMSE, wherein the output is used to predict well interference between the at least one parent well and the child well.
2. The method of claim 1 , wherein the at least one machine learning model comprises XGBoost Regressor, GBoost Regressor, and PolynomialFeatures (2)/Elastic Net.
3. The method of claim 1 , wherein the at least one machine learning model comprises an ensemble model having a weighting ratio of three parts XGBoost Regressor, one part GBoost Regressor, and one part PolynomialFeatures (2)/Elastic Net.
4. The method of claim 1 , further comprising predicting well production for the at least one parent well and the child well using the at least one machine learning model.
5. The method of claim 1 , wherein the properties associated with the at least one parent well and the child well comprise at least one of well production over a period of time, fluid type, fluid volume, additive type, additive concentration, proppant type, proppant size, proppant concentration, proppant mass, and pumping rate.
6. The method of claim 1 , wherein the properties associated with the at least one parent well and the child well comprise well spacing between the at least one parent well and the child well.
7. The method of claim 1 , wherein the properties associated with the at least one parent well and the child well comprise a period of time between when the at least one parent well begins to produce and the child well begins to produce.
8. The method of claim 1 , wherein the child well is a fully-bounded child well.
9. The method of claim 1 , wherein the child well is a half-bounded child well.
10. A system comprising:
at least one processor coupled with at least one computer-readable storage medium having stored therein instructions which, when executed by the at least one processor, causes the system to:
obtain input data comprising properties associated with at least one parent well and a child well associated with the at least one parent well;
divide the input data into a training data subset, a validation data subset, and a test data subset;
select at least one machine learning model using the training data subset, the validation data subset, and the test data subset based on k-fold cross validation;
tune hyper parameters for each of the at least one machine learning model; and
generate a learning output using the at least one machine learning model and the hyper parameters for each of the at least one machine learning model, the learning output indicating a test root-mean-square error (RMSE) and a training RMSE, wherein the output is used to predict well interference between the at least one parent well and the child well.
11. The system of claim 10 , wherein the at least one machine learning model comprises XGBoost Regressor, GBoost Regressor, and PolynomialFeatures (2)/Elastic Net.
12. The system of claim 10 , wherein the at least one machine learning model comprises an ensemble model having a weighting ratio of three parts XGBoost Regressor, one part GBoost Regressor, and one part PolynomialFeatures (2)/Elastic Net.
13. The system of claim 10 , the at least one processor further to predict well production for the at least one parent well and the child well using the at least one machine learning model.
14. The system of claim 10 , wherein the properties associated with the at least one parent well and the child well comprise at least one of well production over a period of time, fluid type, fluid volume, additive type, additive concentration, proppant type, proppant size, proppant concentration, proppant mass, and pumping rate.
15. The system of claim 10 , wherein the properties associated with the at least one parent well and the child well comprise well spacing between the at least one parent well and the child well.
16. The system of claim 10 , wherein the properties associated with the at least one parent well and the child well comprise a period of time between when the at least one parent well begins to produce and the child well begins to produce.
17. The system of claim 10 , wherein the child well is a fully-bounded child well.
18. The system of claim 10 , wherein the child well is a half-bounded child well.
19. A non-transitory computer-readable medium having instructions stored thereon that, when executed by at least one processor, cause the at least one processor to perform operations comprising:
obtaining input data comprising properties associated with at least one parent well and a child well associated with the at least one parent well;
dividing the input data into a training data subset, a validation data subset, and a test data subset;
selecting at least one machine learning model using the training data subset, the validation data subset, and the test data subset based on k-fold cross validation;
tuning hyper parameters for each of the at least one machine learning model; and
generating a learning output using the at least one machine learning model and the hyper parameters for each of the at least one machine learning model, the learning output indicating a test root-mean-square error (RMSE) and a training RMSE, wherein the output is used to predict well interference between the at least one parent well and the child well.
20. The non-transitory computer-readable medium of claim 19 , the operations further comprising predicting well production for the at least one parent well and the child well using the at least one machine learning model.