Automating the drilling of oil wells
Expert comments on how state-of-the-art technology can improve efficiency but stresses the need for experienced driller involvement
In an era when driverless cars are starting to appear on the roads, what is the scope for applying artificial intelligence and automation to the challenging task of drilling an oil well? During a joint meeting in Anchorage of the Society of Petroleum Engineers and the American Association of Drilling Engineers, Bill Koerderitz, chief technology officer for GK Plus Innovations LLC, talked about what can be learned from recent advances in drilling automation.
Automation, in its essence, uses some highly automated means of displacing human intervention in a process, Koerderitz said.
Expert drillers neededBut Koerderitz stressed the importance of involving expert drillers in the appropriate application of automation technology. He argued for viewing automation as a set of tools rather than a means of replacing human judgment. And the examples he described to illustrate drilling automation success all related to specific components of the drilling process, rather than the complete drilling operation. Moreover, while data gathered from drilling operations indicate that automation can consistently raise drilling efficiency above average, a dedicated, expert drilling crew, intent on achieving high drilling performance, can still compete effectively with an automated system, he said.
Koerderitz said that there can be a number of reasons for automation in drilling, including improving safety, improving efficiency, enhancing drilling crew capabilities and improving drilling performance. It is also possible that automation could reduce the required size of a drilling crew, although Koerderitz expressed some skepticism about this, saying that the impact would more likely be a reduction in the number of people that have to be stationed in the drilling rig itself.
In general for any process, including, for example, the operation of an automobile, there is a spectrum of automation levels, ranging from the automated system suggesting to a human operator alternative ways to perform a task, all the way to the system completely taking a task over, including the decision making for the task. The automated drilling processes which Koerderitz described were towards the top end of that spectrum, with an automated system taking over a set of actions which had previously been controlled by a human operator, but with an operator monitoring what is happening and being able to intervene, especially in the event of safety concerns.
Automation examplesThe first automated process that Koerderitz described involves the control of the weight applied to the drill bit at the bottom of a well. Adjusting that weight to an optimum level using the weight applied to the drill string at the surface optimizes the rate of penetration of the bit through the subsurface rock and leads to better directional control for the well trajectory, he commented.
Performed manually, a driller controls the downhole weight on the bit by frequently watching the weight and making appropriate adjustments to the weight on the drill string at the surface, a technique that requires constant monitoring of downhole weight measurements and making frequent adjustments. But downhole conditions often change rapidly, as the bit moves through the subsurface rocks. An automated system can accommodate these changes by constantly computing and applying the required surface weight, with the driller setting an acceptable range of downhole weights and then just having to monitor what the automated system is doing.
Another area where automation can be applied is in the continuous optimization of the drilling energy efficiency through constant adjustments to the rate of rotation of the drill bit, the weight on the bit and the pressure applied by the mud pump. Essentially, an automated system can use the measured drilling performance to make continuous adjustments to the drilling operation. However, while this type of system typically does elevate the overall drilling performance, the system requires a driller to set bounds for parameters such as the ranges for the mud motor settings and bit rotation speeds. And the driller has to monitor what the automated system is doing, Koerderitz said.
Automation risksAnd there are risks associated with drilling automation.
A particularly worrying concern in the possibility of complacency, with a reduced incentive for a driller to monitor what is happening if the automated system appears to be working well, Koerderitz said. A driller can also experience a sense of a loss of ownership of the drilling process. Furthermore, the effectiveness of the automated process becomes highly dependent on quality of the drilling data being used and on the quality of the control system. The maximum performance also depends on the effectiveness of the automation logic.
Rather than looking for routine applications that might be automated, it can be more effective to seek areas where performance can be improved, and then look for both automated and non-automated ways of achieving performance gains, Koerderitz suggested.
Some misconceptionsAnd there are a number of misconceptions regarding automation in drilling, he said. For example, contrary to some views, automation cannot eliminate the need for an expert driller. On the contrary, the driller should be viewed as an automation enabler on the rig, Koerderitz said. It is also dangerous to assume that the data currently available on a rig is adequate for automation. Similarly, the rig controls typically require modification for an automation application. It is also crucial that the driller understands how the automation works, perhaps through practice on a simulator, so that the driller can make best use of the system.
And rather than arbitrarily trying to automate everything, automation should be designed as a set of tools, simple, modular and focused on assisting the driller. It is unrealistic to expect an automated system to deal with all possible drilling situations, Koerderitz said.