Uncategorized

MJ_PTENGenerators2

The RL agents and the older rule-based model were both simulated under ideal operating conditions for validation training. For each simulated day, Patterson-UTI measured the average number of generators online, generator changes per day and “power limitation events,” or instances where the demand within the next five minutes would have exceeded the available power if the rig had followed an agent’s recommendation. Results showed that every RL agent outperformed the rule-based benchmark on at least one metric. The agents marked with a star outperformed the benchmark on all three metrics. Click the image to enlarge. Source: IADC/SPE 230679

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Check This Out
Close
Back to top button