As straightforward as SAGD bitumen production may sound—pumping steam through one horizontal wellbore to heat bitumen and pumping it up a second horizontal wellbore to produce it—the relatively new production method is a highly complex affair, with a wide assortment of variables to consider that are unique to the oil and gas industry.
There is, for example, the intricate interplay between the surface steam producing facility (water plant) and the subsurface parallel wellbores (the reservoir). Though separate processes, events in one area greatly impact on the other.
Among the variables at play are how far apart to place the well pairs; how to evenly distribute the steam along the length of the lateral; how much steam to inject into each well pair to achieve optimal steam/oil ratio; when to drill infill wells; what rate to operate pumps; and what solvent to co-inject, when and at what volumes.
On a field-wide basis, there are too many dynamic variables at play for any one reservoir engineer, or even a team of engineers, to solve. But there are advanced analytic and machine learning tools developed in other industries that can be harnessed to perform this task.
Drawing on the capabilities of digital technologies such as big data and analytics, machine learning and the Industrial Internet of Things, GE and SAS have independently developed digital oilfield solutions specific to SAGD production. Developed in close collaboration with oilsands producers, they have begun to accumulate project successes, saving producers on the order of millions of dollars annually, as detailed in case studies contained in the recently released Digital Oilfield Outlook Report: Optimizing operations to unlock the hidden barrels. The third in a series of digital oilfield reports, it delves into production asset optimization and predictive maintenance.
In digitizing a SAGD field for a major operator, SAS fed the massive volumes of data generated by the project into an advanced analytics platform where complex statistical models can determine the optimal steam distribution to increase production. It can identify which wells to starve and which wells to feed excess steam, as steam availability dictates, on a well pair, well pad or field basis.
Advanced analytics also allows operators to calculate when potential problems like process upsets or dramatic changes in operating conditions will occur and to take corrective action to avoid that occurrence. Production losses can be avoided or minimized by predicting, for example, steam breakthrough, or through a better understanding of the condition of screens, liners, electrical submersible pumps and other vital equipment.
SAS estimates that on a 100 well pair project operating 350 days per year, its solution would increase production by 1.4 million barrels per year. Assuming a WTI price of oil of about US$46, it would generate $23.9 million annually in pre-royalty cash flow, according to the Digital Oilfield Outlook Report.
“In a case study for a SAGD producer that had a defined subcool target value range, SAS determined there would be potential production increases of between seven per cent and 17 per cent, depending on the well pair evaluated, while remaining within the desired operating range of the project. In other words, if an operator has a fuller understanding of its project, it could significantly increase production from each well pair while achieving its operational goals,” the report states.
At its Customer Innovation Centre in Calgary, GE created a Thermal Production Optimization software solution powered by its Predix platform designed to leverage SAGD production and operations data to its full potential. It provides access to advanced analytical machine learning, which reduces uncertainty and provides accurate well models.
Using these models, operators can run what-if scenarios and constraint-based optimizations to establish ideal operating parameters on individual wells or across an entire field. They can also build a digital twin of a field that can optimize operations from reservoir to facilities, breaking down silos to reveal significant predictive capabilities.
The optimization solution was applied to a SAGD project to determine the optimum steam allocation across the entire field. Since steam production represents a major cost item, optimizing its use field-wide can have a material impact on production costs and company revenue.
The digital twin created allowed the operator to input any number of variables—such as combinations of various steam rates, pumping rates and solvent co-injection concentrations—and to in a matter of seconds run the optimizer across the entire field.
The models are self-learning, allowing them to retrain in near real time to create simulations and predict outcomes with accuracies greater than 95 per cent for steady state and 90 per cent for non-steady state (such as when an equipment outage affects any part of the field), according to the case study in the report.
“In an early application with a SAGD customer, GE’s Thermal Production Optimization software produced a 1–1.5 per cent improvement to steady state operations and a three to five per cent improvement to non-steady state production across the field. Even a one per cent improvement in production—using existing infrastructure and at no additional cost to the operator—can yield millions of dollars in additional revenue annually,” says the Digital Oilfield Outlook Report.
Produced by JWN Energy with partners GE, SAS, Panoptic Automation Solutions and ABB Group, the Digital Oilfield Outlook Report: Optimizing operations to unlock the hidden barrels details the results of an in-depth industry survey of digital oilfield adoption. It also features exclusive case studies from industry and recommendations for digital oilfield implementation. Free download of the report is available by clicking here.