Rise of the Machines: How to Use Machine Learning and Artificial Intelligence to Automate Data Analytics09/30/2016
Machine learning can be used to automate more complicated data-processing tasks
This is the second in a three-part series about machine learning’s applications in upstream oil and gas. This blog details how machine learning can be used to automate more robust analytics tasks like production forecasting and well-completion design.
The computer is the greatest tool of our generation. But are you maximizing the utility of your computer processing and software capabilities?
Machine learning and big data analytics are the latest and greatest tools for the upstream oil and gas sector. These tools can be applied to eliminate problems both big and small.
In our first “Rise of the Machines” blog, we discussed how machine learning can be used to eliminate tedious tasks from your workflows. In this blog, we’ll discuss how software can be used to automate more robust analytics tasks such as production forecasting and well-completion design.
Engineering tasks like the examples above have two common characteristics. First, they’re commonplace in all oil and gas operators. And second, a sustainable amount of time is dedicated to completing the tasks. Any task, simple or complicated, with these two characteristics is a perfect candidate to turn over to the machines.
Imagine selecting hundreds of wells, setting a range for the decline curve parameters, and letting the computer do the work. In a matter of just minutes, hundreds of wells can be forecasted – something that would take hours if done by hand. This is the power of using software to automate tasks.
Programs can also be designed to automate completion design. This can be done by meshing regulatory, geologic, and engineering data into one program to ensure that all requirements, both technical and legal, are fulfilled. Automating the first steps of these analytical processes allows staff to focus on interpreting the data.
With any machine learning task, quality control is a must. After a process like automated production forecasting is complete, the forecast must be checked by human eyes. Quality control can become a tedious task if not handled correctly. In the first blog in this series we discussed how machine learning is the ultimate weapon against tedious tasks, and quality control is no exception. A task like automated production forecasting is only a revolutionary tool if quality-control checks are included in the workflow. Using data analytics, computers not only can learn how to perform task automatically, but they also can check their own work and iteratively correct work until results are within a tolerance. Human quality control can never entirely be replaced, but properly designed programs will minimize the necessary amount of human intervention.
Almost all professional fields have seen the use of computers and software optimize the workflows they use. Data science is no exception. Engineers, data scientists, and other technical staffers live to better understand their data and leverage this knowledge to make better decisions. Proper implementation of machine learning and artificial intelligence brings important data anomalies and trends to the forefront so technical staffs can make vital decisions quickly and confidently.
In the third and final installment of the "Rise of the Machines" series, we'll be talking about how machine learning can be used to prevent well failures.