In the popular imagination, a truly automated drilling unit is one that is manned by one human and one dog. The human is needed to feed the dog and the dog to keep the human from touching anything. For those earnestly engaged in developing automated drilling processes, however, the guiding image is not the humanfree robot-controlled assembly plant but the airplane autopilot system that advises the pilot and, when appropriate, autonomously flies the airplane.
A concept that has been discussed by the industry for more than a decade because the subsurface is a highly complex environment filled with unknowns, fully automated drilling operations have been slow to emerge. The road to full automation is being paved with very large volumes of data delivered to the surface from newly developed downhole sensors. These new data provide engineers with real-time insight into the state of the well as it is being drilled, but because the volumes arriving at record speed are so large and unwieldy, much of the information goes unused.
“We don’t have a Big Data problem. We have a messy data problem.”
The first order of business, according to researchers, is to manage those data. “We don’t have a Big Data problem,” said Theresa Baumgartner, a researcher at the University of Texas (UT)-based Rig Automation and Performance Improvement (RAPID) consortium. “We have a messy data problem.”
Other barriers to automation are more familiar to developers of oilfield technology, including the industry’s famous resistance to changes that challenge long-held beliefs. For example, automated systems that process more real-time data significantly faster and more accurately than humans can might set drilling parameter limits that are less conservative than those with which drillers are comfortable. Typically, drillers resist or ignore such machine-generated commands and recommendations.
Still, progress toward automation moves forward. Spurred by increasingly remote areas of operation and complex wells, industry members are developing systems to answer operator demands for lower well construction costs and safe and efficient operations in difficult situations.
On the road to automation
Today automated drilling exists primarily as various diverse subsystems that are designed to perform or help humans perform specific repetitive drilling tasks. These subsystems range from those that act as advisers that depend on humans to implement recommendations to autonomous systems that gather and analyze data and then execute the necessary operational commands without human involvement.
Among these repetitive and predictable tasks, none of them impact drilling performance more than fluids management. Despite the proliferation of downhole sensors, drillers continue to rely on input from drilling fluid analyses performed by onsite personnel to help make critical decisions. Traditionally, because much of their time is devoted to managing drilling fluids properties, mud engineers are able to sample and fully analyze drilling fluids only three or four times per day. As a consequence, operators and drillers make drilling and well control decisions based on hydraulics models that were created using data that are often several hours old.
To address this shortcoming, Halliburton researchers have developed the BaraLogix density and rheology unit (DRU), which the company describes as a “fully automated unit that measures the density and rheology of drilling fluids.” Installed near the rig’s mud tanks, the skid-mounted unit incorporates a self-generating nitrogen purge system and is ATEX Zone 1 and IECEx certified.
Depending on the temperature of the mud captured at the supply or return line and the temperature to which it must be heated to meet operator test specifications, the system can be used to perform four or more six-speed rheology tests per hour. In addition, the DRU is able to calculate fluid density as frequently as once per minute.
The data are captured by the Halliburton InSite data acquisition system, which uses the rig’s LAN system to transmit the measurements to drilling engineers in a format that allows them to customize how the data are sorted and viewed. Mud engineers continue to maintain fluid properties based on traditional measurements but are assisted by BaraLogix results. “The mud engineer still performs his tests,” said Jason Bell, product manager with Halliburton’s Baroid business line. “This gives him advanced advisories about changes to make based on test results.”
Automated routine fluid tests that collect and analyze more data more frequently allow drilling engineers to recognize data trends and changes in the drilling fluid and to make appropriate fluid adjustments as they occur. But the greater implications for end users arise from the marriage of the automated system and hydraulics models.
“The real value of this service is the data it provides for hydraulic models,” Bell said. “By transmitting the data into models, we are able to use the model in predictive analytics for hole cleaning, lost circulation, packoff and a host of drilling parameters.”
Reaching the sweet spot
Because the easy reservoirs have been drilled, operators today are reaching target formations through complex and extended-reach wells drilled from a single surface location, so accurate wellbore steering and placement has become a critical function. To ensure wells do not collide and that they land precisely within the target formation, directional drillers guided by real-time downhole sensor data adjust wellbore trajectory using surface-controlled rotary steerable tools. But because trajectory is subject to many downhole variables, directional drillers might not have all the data necessary or might not be able to process and act on those data quickly enough to make adjustments that result in optimal well placement.
Service providers are working on automating the process with rotary steerable systems that monitor all available real-time data and use them to correlate commands given with the bottomhole assembly’s actual response to those commands. The systems use those data to create accurate wellpath projections and to recommend commands that will result in the desired trajectory.
Engineers report they are field-testing a fully automated version of the system that executes downhole steering tool commands autonomously. Also with an eye toward autonomous downhole drilling tools, Baker Hughes has released an automated polycrystalline diamond compact (PDC) depth-of-cut (DOC) control bit that prevents certain drilling dysfunctions by adjusting proactively and autonomously to load increases at the bit.
PDC DOC control bits have been in use since the early 2000s. They are designed to prevent the cutting elements from digging too deeply into the rock and creating loads that lead to torsional vibration or stick/slip events. Like all bits, however, when a DOC bit underperforms, the operator’s only option is to pull the drillstring and replace the bit or to endure less than optimal performance.
By contrast, the Baker Hughes TerrAdapt bit incorporates movable elements that react to the load at the bit. When the bit experiences an increased load that indicates onset of stick/slip, those elements extend outward against the formation to resist the sudden change, limiting the DOC and protecting the cutting structure. During intervals of smooth drilling the elements retract into the bit body and, because the bit can cut more aggressively, ROP is optimized.
The mechanics of the new bit are a cartridge-like component that contains fluid and a restrictor that creates a pressure differential, explained Baker Hughes Product Line Manager Danielle Fusilier. Baker Hughes researchers, who documented the rate of load change associated with the onset of stick/slip, designed the movable elements to move outward in reaction to that rate.
“Typically, if an operator determines stick/slip is occurring, the reaction is to reduce DOC through parameter adjustment,” Fusilier said. “That takes time and may not solve what is happening at the bit. With TerrAdapt we are able to react to vibration issues in real time and prevent further damage. It is a good fit for automated drilling because it is very hard for an operator to know what is happening 10,000 ft [3,048 m] downhole at the bit.”
As Baker released TerrAdapt bits to autonomously adapt to changing rock conditions, Weatherford has developed an automated pressure-control system. Weatherford Global Director of Well Control Technology Robert Ziegler said the system is “a fully integrated real-time multisensor process-control system” that combines the company’s Microflux control system with its OneSync software platform.
The Microflux system measures return flow using a Coriolis flowmeter installed in line with the chokes to detect fluid gains and losses earlier than is possible using traditional methods. This early detection capability minimizes fluid gains or losses. OneSync is the company’s well planning, dynamic drilling simulation and operations software platform.
By using the two in combination, Ziegler said, the driller is able to detect minute losses and influxes and mitigate them without interrupting the drilling process. The combined systems are able to determine the drilling window between pore pressure and fracture initiation pressure and to ensure that the driller remains within those pressure boundaries.
“A very good analogy is the ‘autoland’ system incorporated in every modern transport category aircraft, where automation lands the plane much better than the average pilot with zero visibility, controlling all aircraft systems in real time and showing the pilot via virtual reality where the plane will touch the runway,” Ziegler said.
The common thread that both hinders and aids efforts to create automated systems is the unprecedented volumes of real-time drilling data available to drillers today. Because of the volume and because data are not usually presented in a way that is easily and quickly understood by the user, converting those data into useful information that impacts operational actions and performance is key to automation.
To help manage this embarrassment of riches, researchers at RAPID first analyzed a unique set of actual field drilling data gathered by a consortium member company. Based on their analyses, performed in the consortium’s real-time operations center at the UT School of Engineering, the team developed templates of single-page visuals created automatically from the data. Presented in an easily interpreted format, these visuals enable engineers to make quick decisions that result in improved drilling performance. The contributing company is reportedly working to implement these innovations into its operational workflows.
RAPID researchers, including UT’s Baumgartner, also addressed diagnostic challenges associated with the rate of data transfer between the bit and the surface. To replace the inadequate transfer rates of mud pulse telemetry, some in the industry have begun turning to wired drillpipe, which can deliver up to 57,600 bits/sec compared with the traditional mud pulse technique that typically delivers data at 1 bit/sec to 3 bits/sec. However, because it encourages the industry to add complex downhole sensors that have high sample rates, resulting large volumes of data can make it difficult for end users to detect or identify drilling dysfunctions.
RAPID scientists attacked the problem using field data from multiple operations to develop a value-based approach, which specifies minimum data collection frequencies for each type of drilling dysfunction. “Many different downhole events contribute to movements and vibrations in the drillstring and the bit,” Baumgartner said. “Recognizing distinct patterns and frequency spectra helps us to unravel all these events from a single signal.”
The overall arguments for automated drilling systems, as for all new drilling technologies, are optimized drilling performance, enhanced safety and operating cost reduction. And as complex wells push current drilling performance limitations, some level of automation will almost certainly become standard on high-end drilling rigs. It is much less likely, however, that even the most sophisticated units will be populated by just one human and one dog.