Built on Several New Tools, Marathon’s Digital Oilfield is Maturing
Marathon Oil says its shale fields are producing more oil and gas with less hands-on work from company personnel thanks to a growing arsenal of digital technologies and workflows.
Marathon Oil has recently rolled out several new digital tools as part of its ongoing march toward oilfield automation.
The Houston operator is scaling those technologies across its four US onshore business units and, in the process, fundamentally changing the way its employees work. The multiyear undertaking hasn’t been easy, but the company now has the metrics and case studies to show that progress is being made.
Marathon launched its digital and automation journey in the Eagle Ford Shale of South Texas in 2016 before expanding it to the Bakken Shale of North Dakota, SCOOP and STACK plays of Oklahoma, and the New Mexico portion of the Permian Basin, opening control centers for operations in all four locations.
Rick O’Brien, Marathon Oil digital oilfield project manager, likened implementation in each business unit to starting over again completely, with the each unit “more leery” of the changes than the last. It didn’t help that the Eagle Ford business was seen by other units as sort of a favorite son within the company, “getting all the money and all the resources” while the others felt they were “starving,” he said.
One way the company has cultivated acceptance of the new technologies and the resulting new workflows has been by fostering collaboration among early adopters between the units. O’Brien, who spoke at the recent Data-Driven Drilling and Production Conference in Houston, said he has spent some 300 days on the road over the past 2 years aiding implementation in the four basins. The long trek thus far has resulted in more than 300 reported users of digital oilfield tools across the assets.
“One thing we’ve learned is that field people like to be trained by field people more than they like to be trained by outsiders,” O’Brien said, noting that personnel have tended to shy away from well-produced, readily available video tutorials.
The organization has further nurtured familiarity with the technologies and workflows by conducting cross-training exercises where field operators rotate into the control centers, learn how they function, and offer feedback on the new processes and tools. “It's really proved to be valuable from both sides,” Sarah Renko, control center supervisor for Marathon’s Oklahoma unit, said during an automation-focused panel discussion. “We're learning from them. They're learning from us.”
The method has also helped with training in the Permian, where the average professional experience of the operations team is around 18 months, O’Brien said. The unit interfaced a couple of its best field operators with personnel in the newly minted control center, making them available for ongoing collaboration and training.
Meanwhile, keeping management on board—and maintaining investment in—the digital and automation journey has required continually selling and reselling initiatives. The implementation team collects as many success stories as it can and periodically presents them to management, reminding it of what can be accomplished with digitalization and automation.
These wins, according to calculations from Marathon, include:
- Gaining at least $75 million in revenue from cutting deferred production.
- Increasing its Eagle Ford well count by 25% while reducing pumper head count by 30%.
- Decreasing planned well visits in the Bakken by 50%.
- Reducing well pad visits in Oklahoma by 78% through a retrofit automation project covering 44 locations.
Perhaps the best outcome, O’Brien said, is that all of Marathon’s new wells—flowing or on gas lift or rod pump—now receive a standard automation kit. Marathon in 2016 began installing automated chokes, valves, and supporting instruments on its Eagle Ford wells and subsequently did the same in Oklahoma. It’s now “just getting started” on its Permian wells, he said.
In the Eagle Ford, the automated equipment is ensuring the safe and steady flow of hydrocarbons between Marathon’s wells and central processing facilities (CPF). When a gas compressor at a CPF goes down, for example, there’s no way to get low-pressure gas to a pipeline or back to the well, causing pressure to build back through the system, typically necessitating a shutdown of the well. But the company’s automated equipment regulates pressure at the separators at both the CPF and well site.
“We managed to regulate pressure that could have otherwise shut the well down,” said Doug Johnson, an independent consultant for automation, optimization, and analytics strategies who’s working with Marathon. “It happens all within the SCADA [supervisory control and data acquisition] system, and it's managed without having to send anybody out there to restart wells.”
Automating flow from the choke to the CPF has resulted in an average 2.75-hr reduction in startup time for a network of wells tied in to Marathon’s Blackjack facility, removing 2,640 bbl/year of deferred production. Across four CPFs, the company has eliminated 57,750 bbl/year of deferred production, amounting to some $4 million in savings.
Bringing it all Together via Task Manager
Renko is among those charged with merging Marathon’s new digital tools into a single, cohesive function across it business units. In addressing this challenge, she said, the organization started by breaking down what happens following planned and unplanned events in the field, and then developed workflows with its new tools to streamline the post-event processes as much as possible.
Indispensable to the new workflows has been an application called Task Manager, which has given Marathon the ability to convert multisource events into tasks. This has dramatically simplified daily work for field operators, Renko said, giving them a single tool with which they interact. Most recently, Task Manager was introduced to the company’s Permian unit.
One workflow involves prioritizing dispatches after a well goes down. Currently, alarms from the well are sent to the control center, where potential issues are validated to determine what, if anything, needs action. High-priority alarms are auto-dispatched. The field operator receives dispatches via mobile device or email through Task Manager, and then he or she acknowledges the task, indicates whether they’re acting on it, and documents the actions they’ve taken.
Documented information is sent to the organization’s dashboard for post-event analysis. Learnings from past well events can be referenced and solutions can be repeated.
In Oklahoma, Marathon has reduced the number of alarms requiring response from the field operator by 85%. “Historically, the way we were dispatching alarms, they were going out chronologically. Whereas now they're being dispatched by priority and potential impact, which I think really has made a big difference in the field,” Renko said.
Knowing that not every well deserves a visit each day, the tool is further reducing trips to the pad by using predetermined criteria based on a risk matrix to prioritize site visits. This has halved site visits in the Bakken from 30 wells/route to around 15 wells/route, she said. Marathon has also applied gas measurement software called FlowCal to detect flow anomalies, with dispatches sent through Task Manager.
“There are many other workflows we think that you can apply these tools to,” Renko added.
What’s Next for Marathon?
A major near-term priority for Marathon is to reduce the number of offset frac shut-ins, which are the company’s “single largest cost of production” in the Eagle Ford, with losses of more than $100 million annually in the play alone, O’Brien said. “It’s not that we’re fracing into our wells—we’re so scared of fracing into one that we turn a bunch of wells off.”
“If you can start making that completions group—and the engineers—trust that I can use my automated equipment to monitor those pressures going up on a well that’s near a completion, I can shut that well with my autochoke, then move three stages down and release the autochoke, and that well is back to performing in maybe 6–8 hours vs. 2–3 weeks,” he explained.
Marathon also is partnering with Austin, Texas-based machine learning and artificial intelligence (AI) firm Hypergiant Industries to “take all the knowledge that a really, really strong production supervisor has and turn it into an AI tool that will think on behalf of that person and that asset,” O’Brien said. The idea behind the digital production supervisor tool, like Marathon’s other tools, is to cull information from multiple sources into a platform that makes decision-making easier in an effort “to move beyond dashboards,” he said.
The digital production supervisor tool will warn when something unusual has occurred and use context inputted from actual supervisors to provide more information about those anomalies. With years of learnings inputted and logged, it will allow new supervisors to hit the ground running when they join the team.
If the tool indicates that tank management has been the reason for multiple pad visits, for example, and the person making the visits can calculate the cumulative cost of those visits, he or she can then make a quantitative business case to his management that installing a tank radar gauge would save money. Or, if the tool shows that a person visiting a pad is routinely taking longer than the standard visit, it might indicate insufficient training.
“We're taking a lot of the qualitative job of the supervisor and trying to give them some quantitative information so they can very quickly hone in on where the problem areas are vs. having to sort through a lot of generalities—without having to physically be on the well site,” he explained.
Many of these tools were developed in-house at Marathon, providing the flexibility to customize. However, doing it within the organization comes with numerous challenges and eats up a lot of resources. As it tries to leverage its inflow of data, O’Brien noted, the company is looking to balance its project load with additional partnerships like the one with Hypergiant.