| CPC G01N 33/18 (2013.01) [C02F 1/008 (2013.01); C02F 3/006 (2013.01); G06N 3/08 (2013.01); C02F 2103/10 (2013.01)] | 18 Claims |
|
1. A method for determining a water treatment plan for produced water originating from a subterranean formation, the method comprising:
receiving sample water analysis for the produced water;
receiving one or more key performance indicators (KPIs) that each indicate a selected treatment result for the produced water;
providing the sample water analysis and the KPIs to a machine learning model;
determining a water treatment plan for the produced water using the machine learning model, wherein the water treatment plan comprises one or more additives for the produced water that are to provide the produced water with the KPIs;
testing the water treatment plan on a sample of the produced water;
analyzing the sample of the produced water after the testing to produce test information;
updating the machine learning model with the test information to produce an updated machine learning model;
providing the sample water analysis and the KPIs to the updated machine learning model; and
determining an updated water treatment plan for the produced water with the updated machine learning model; and
determining a confidence score for the water treatment model based on a difference between the sample water analysis and historical water analysis information, wherein the historical water analysis information comprises primary component analysis (PCA) for a plurality of water samples,
wherein testing the water treatment plan on a sample of the produced water comprises performing a bottle test or a field test based on the confidence score.
|