Drilling advances

The tried-and-true bowtie diagram, loosely defined as a visual representation of the cause-and-effect relationships of ostensibly independent occurrences, has been the industry’s go-to methodology for identifying and mitigating drilling risks. However, what if this largely rigid tool could be given a life of its own, with hot-off-the-sensor data, to enable real-time, hazard-reducing decisions in constantly changing, high-pressure, high-temperature (HPHT) and similarly high-risk applications?

“I’m aware of your Bowtie approach to addressing risk,” Andrew Hartigan, director of Performance Optimization Business for Lone Star Analysis Inc., told the November IADC Drilling Engineering Committee’s (DEC) quarterly technology forum in Houston. “Let’s assume that the thought, the experience and the process that goes into developing a bowtie is flawless, and that information is complete. Our observation, however, is that the information is not updated on a regular basis, from the time it is built until the time decisions are being made as a function of those bowties.”

Therein lies the impetus of Lone Star’s Dynamic Risk Modeling, a probabilistic approach that takes the time-tested Bowtie diagram to a new dimension with continuous real-time risk awareness. “What if you could have that information continuously updated, so your bowties were alive, and not only alive, but connected and integrated? There are times that a barrier or mitigation effect changes on one of your bowties. You address that one, but at the same time that one may be connected to four or five, or as many as 12 others. So, now you have multiple sets of bowties that have just changed. If I have all my bowties digitally connected, where the same barrier in one is the same barrier in another, you would not only have a live representation of your bowtie, you would have a connected, dynamic and interchangeable set of information that was current.”

Hartigan, whose background is in U.S. Naval Aviation, has spent the past two years leading the transfer of the predictive modeling technology developed by the Addison, Texas-based company into the oil and gas sector. “What we have done in support of military, defense and telecommunications is to manage outcomes that are being sought with probabilistic inputs of data. The ability to make decisions, based on real-time data, and turn that current data into predictive analytics, is the secret sauce that makes this work. Otherwise, it’s just a faster bowtie,” he said during the forum, which focused on innovations in drilling safety.

Selective data. Noting that dynamic risk modeling begins “with a solid foundation of bowties,” Hartigan said square one in developing a model is sifting through petabytes of information that can overwhelm even the most data-savvy person. “This different approach is to not look at all the data, but instead focus on the answer you are seeking. I know what information I want, so I'm going to back up and only seek the data that is relevant (to risk),” he said. “A dynamic risk model is valuable, because it is not just taking digital data from databases, but also taking historical data and live data, either from downhole or from systems that are fed into a model."

“The contributors to risk come from different places and organizations, but they were never collated and put in one place for decision-making. That's the value of a dynamic risk tool,” he said. “The real power of this tool is the ability to do ‘what if?’ (scenarios). Before you make a decision you can evaluate the potential outcome.”

Lone Star brought its dynamic risk modeling to the industry’s attention in a proof-of-concept paper, presented at September’s SPE Intelligent Energy International Conference and Exhibition in Aberdeen, Scotland. There, Hartigan, et al, unveiled a synthetic model built around the mitigating factors that would elevate a kick to a blowout in an HPHT drilling environment. The demonstration used widely prevalent cumulative risk data from the Gulf of Mexico’s Macondo disaster, “to show how a real-time representation of current and cumulative risk could have aided decision-makers with critical information and insight.”

“With a probabilistic representation of outputs, I can't tell you exactly what the risk is, but I can tell you the range of possible outcomes,” he told the DEC. To illustrate, he said the nominal risk of a blowout in a HPHT drilling application is .037%, with a likely risk of up to 10%, which can be reduced or increased significantly, depending on the mitigating steps taken during the operation.

Bringing it home. All this, he said, is encapsulated in a software tool that essentially transforms a standard bowtie into a rig-based dynamic risk model, which any operator can build with three days’ training. A pull-down menu provides a fingertip evaluation of pending decisions and their potential impact on risk. “It gives you the ability to enter only the key decisions you know are important to the operation you are performing at the time,” he said.

The model updates every 5 sec, based on temperature and pressure data streaming from downhole sensors. “Pressure and flow differentials are fed into the model and compared to the expected results of a positive pressure test, a negative pressure test or a shoe track conversion to identify whether there is an increased risk. And, this is happening in real time. There's no human in the loop,” he said.

“When you see changes in risk as a result of a potential decision you were going to make, it likely would cause you to slow down and maybe bring some other brains into the conversation.” 

About the Authors

Jim Redden

Contributing Editor

Jim Redden is a Houston-based consultant and a journalism graduate of Marshall University, has more than 40 years of experience as a writer, editor and corporate communicator, primarily on the upstream oil and gas industry.