How to speed the way to autonomous production

Posted by Chris Quigley on Aug 28, 2019 12:08:00 PM

Autonomous speed for oil and gas production operations

The benefits of making oil and gas production autonomous are many. Production increases based on the ability to detect and remedy problems in seconds rather than hours is topmost, followed by safety, efficiencies when physical repairs are needed and more precision in PM schedules are among the others.

We’re not there yet—some would point out that the Oil Patch, due to its boom-and-bust nature, lags behind other industries in this area—but we’ve made big strides since the 2015-16 bust. Necessity being the mother of invention.

There’s still much to be done to get there, and a lot of it involves rethinking the process of solving production issues.

Training artificial intelligence (AI) and Machine Learning (ML) is sort of like parenting a toddler—it takes interaction and some thinking.

For AI/ML, production engineers are part of the parenting process, in conjunction with automation companies like us. In ages past, the one concern of a production engineer lay in how to resolve a pump or formation issue on a well-by-well basis. Principles could be learned that would help in the future, but the immediate plan was to fix well X.

Now, there is indeed the issue of how to fix each well, but we can add a layer to that—how can we ‘teach’ the AI system to detect the problem and to know what adjustments to make to fix it, or to at least send an alarm.

Autonomous speed for oil and gas production operationsThe first step may simply be to collect enough historical data to recognize patterns arising before there is an incident. Certain combinations of changes in pressure, flow rates, pump speed, power use and more can predict a wide range of issues—and the earlier the better.

Even if we begin simply with an alarm, engineers can then quickly use their training and experience to create a list of remediation options, which can in turn be added to the AI’s algorithm.

Some would ask if this is like working yourself out of a job, and we say No. Engineers will always be needed for several reasons. First, with new production procedures, formations, chemicals and other developments there will always be new issues to solve.

In fact, engineers will be more free to be creative because their time will not be absorbed by drop-everything emergencies.

Second, most companies today are shorthanded anyway—letting each engineer take on more wells than ever before is already happening. Automation will just create more efficiency.

Third, on the field side, PM will still be needed, things will still break, and equipment will still need to be moved or replaced. What there won’t be is wasted trips to check on a perfectly operating site, or trips to a broken site in which the pumper has to return to the shop for the right tools and parts at the end of an already long day (we’re talking safety here).

Safer, more efficient, better profits, fewer environmental issues—this is the future to which engineers will be training AI to walk toward in baby steps.

It’s a future worth approaching.

 

More about the Zedi autonomous journey

 

Topics: Oil and Gas, Efficient, Autonomous, Future, Machine Learning, AI, Oil and gas production software, Fast, oil production safety