Tools, sensors and materials are vital for efficient petroleum retrieval. Also needed are systems that help the well run smoothly from one second to the next. While the average human vastly outperforms the most powerful computers at certain tasks, others are better managed by computers.
Mike Nikolaou, professor of chemical and biomolecular engineering with the Cullen College, is using the superior ability of computers to balance different variables in order to improve the efficiency and safety of petroleum retrieval. His research has proven to be so vital, in fact, that his article “Computer-aided process engineering in oil and gas production,” which was published this year in the journal Computers & Chemical Engineering, was one of the most downloaded articles on the journal’s website.
“Think about juggling. You start with three balls. You cannot do this naturally; you have to train yourself,” he said. “It’s even worse if you have a dozen different plots on a screen that tell you about dozens of different variables. You cannot coordinate those easily if you’re a human. A computer can do that very easily if you tell it what to do in a meaningful way. This is what we’re doing.”
One of Nikolaou’s juggling acts involves maintaining pressure in a well, which must remain within certain boundaries. If the pressure within a borehole falls below or rises above these bounds, a catastrophic accident could result.
A number of different chokes, pumps and valves contribute to a well’s pressure. Typically, said Nikolaou, these are controlled by operators at various locations of an oil rig, or sitting at a control panel. This situation is simply not one that the human brain is wired to manage.
That’s why Nikolaou is developing algorithms for managed pressure drilling, in which a well’s various valves, pumps and chokes are coordinated by a computer system to drill as efficiently as possible while maintaining safety. “It’s an enclosed system to ensure pressure in the borehole stays within the very tight bounds that are necessary for safe operation,” he said.
In a related project, Nikolaou is developing systems to manage the quality and properties of a well’s drilling fluid.
In addition to lubricating and cooling the drill, drilling fluids carry rock cuttings from the bottom of the well to the surface. Before the fluid is sent back downhole, the cuttings must be removed and the fluid’s properties adjusted to meet the specific needs of the well at that time.
Using data from sensors in a solid separations system, Nikolaou’s algorithms will ensure that the drilling fluids have the best possible properties to ensure drilling efficiency and ultimately improved economics.
“Humans are good at certain tasks and computers are good at others, like controlling several variables simultaneously,” said Nikolaou. “These projects are about having computers manage those tasks that they are best at.”
Nikolaou is also undertaking more nuts-and-bolts efforts, including research that addresses environmental concerns.
Much of the opposition to hydraulic fracturing comes from fears of groundwater pollution. Fractures allow gas to flow through the rock, up a well’s metal casing and to the surface. Gas, it is theorized, can also escape through the space between this metal casing and the rock formation, raising water and air pollution concerns.
To prevent these leaks, petroleum companies fill in the space between the well and casing with cement. While the concept is simple, said Nikolaou, the execution is complicated.
“There are multiple pipes of different diameters. There are rock formations of different consistencies and there are significant changes in temperature and pressure from one place to another,” he said. The result: potentially imperfect seals between the rock and casing that may result in gas leaks.
Nikolaou, then, is working on multiple projects with the same goal: developing computer models that will provide instructions for near-perfect cementing throughout a well. Simply feed the properties of a well into the model and it will provide the cement mixture and conditions that will seal the well as completely as possible.
Working with data provided by industry partners, Nikolaou’s models will balance the dozens of variables that can change the cement mixture’s key properties, such as viscosity, strength, adhesion and curing time. Those same industry partners, he noted, are conducting experiments using cementing mixtures the model recommends for specific conditions. This information, he said, allows him to ensure the model works in the real world.
“Right now, cementing is directed by experts and people with experience. It’s not easy to find these experts and, at the same time, humans aren’t very good at managing so many variables simultaneously. If you want the best, safest seal, a computer can help and even beat humans at this if it has been fed enough data to learn.”