Model predictive control tames oil sand wells

Oct. 15, 2015
Process control engineer at Nexen and Spartan Controls showed how they accomplished managing SAGD well production through advanced process control
About the author
Jim Montague is the Executive Editor at Control and Control Design magazines. Jim has spent the last 13 years as an editor and brings a wealth of automation and controls knowledge to the position. For the past eight years, Jim worked at Reed Business Information as News Editor for Control Engineering magazine. Jim has a BA in English from Carleton College in Northfield, Minnesota, and lives in Skokie, Illinois.

Using steam to melt bitumen hundreds of meters beneath Alberta's arboreal forest, then pumping out the resulting emulsion is no walk in the park. Shifting temperatures and long dead times make controlling these steam-assisted gravity drainage (SAGD) wells extremely difficult. The wells don't respond well to standard proportional-integral-derivative (PID) strategies; multiply this challenge times 100 wells to be controlled simultaneously, and you have a dynamic puzzle of quixotic proportions.

Nevertheless, this was the challenge undertaken recently by Nexen Energy ULC and Spartan Controls at Nexen's Long Lake oil sands facility near Fort McMurray, Alberta, Canada. Nexen is a subsidiary of the China National Offshore Oil Corp. (CNOOC) Ltd., and is based in Calgary, Alberta.

"The oil sands hold reserves of about 170 billion barrels, but much of the McMurray formation in the Athabasca deposit are buried 300-400 meters deep, too deep to mine," said Pranob Banerjee, senior staff APC specialist at Nexen. "In addition, these oil sands are highly viscous due to anaerobic bacterial action.

Closing the production gap

"Before MPC, we had a lot of variability, and after MPC we have a lot less." Nexen's Pranob Bannerjee on the company's use of Emerson's multivariable predictive control solution to extract bitumen from Alberta tar sands.

A SAGD well consists of a 500-1,000-meter long steam injection well running horizontally about 5 meters above a parallel, producing well in the oil sands layer, Banerjee explained. The steam melts the bitumen, forms an emulsion, then drops down into the producing well. It's then pumped out, and then used to make synthetic oil. "The challenge is determining and managing production targets for the emulsion, and bridging the usual 5% gap between our expected and actual production."

Banerjee and Sandeep Kondaveeti, process control engineer at Nexen, and Hailei Jiang, APC engineer at Spartan Controls, showed how they accomplished this feat during "Managing SAGD Well Production through Advanced Process Control" this week at the Emerson Global Users Exchange.

"We target each well daily because we want to get the full potential from them; targets are based on estimated production and simulations," explained Banerjee. "But it's difficult to sustain production targets and manage operating constraints, and it's even harder to meet the individual targets for each of the more than 100 wells at our site. That 5% gap might not seem like much, but it could add up to 4,500 bpd."

Banerjee reported the most important variable in a SAGD application is subcool temperature control, which is the difference between the temperature of the theoretical steam temperature (for the given pressure) and the emulsion temperature.

"If the subcool temperature is too low, then the emulsion will be too viscous, and the emulsion inventory will be very low. However, if the subcool temperature is too high, then the inventory will be high, but the steam may risk breaking the lines," explained Banerjee."

Drag-and-drop MPC

Banerjee added that another major challenge for Nexen was that its operators typically had to handle about 100 wells each, and manage constraints, process variations, alarms and trips and the addition of newer wells. "Our project was how to increase production, better meet our targets, and improve control of our subcool temperatures," he said. "So we talked in house about using advanced process control (APC) and model-predictive control (MPC)."

Consequently, Nexen implemented a DeltaV DCS and accompanying MPC module, MPCPro, from Emerson Process Management. "Basically, our APC/MPC solution takes controlled variables from our wells and plant, and turns them into manipulated variables we can use to improve performance," explained Benerjee. "Before MPC, we had a lot of variability, and after MPC we have a lot less."

Banerjee added that employing the DeltaV MPC solution was more efficient than other advanced control solutions because MPCPro is embedded at the controller level, and because the DeltaV MPC blocks are easy to drag and drop."Typical APC would have taken 10-12 months for us to implement, but we added DeltaV MPC in just eight months, and installed it on all our well pads in just six months. We're also using DeltaV PredictPro for step testing, model identification and controller building."
"We've also improved our overall emulsion production by more than 9%," said Banerjee. "We've significantly improved control. We now have faster realization of planned targets. We increased operations reliability with reduced pump-trips and alarms. It's also easier now for one operator to manage 100-plus wells thanks to improved situational awareness, faster resolution of abnormal situations and increased efficiency."

About the Author

Jim Montague | Executive Editor

Jim Montague is executive editor of Control.