Business challenge

Gazprom Neft PJSC needed to drill about a thousand new oil wells each year while achieving a strong ROI for each project.

Transformation

To boost the profitability of its wells, Gazprom Neft teamed up with IBM to develop a predictive analytics solution that enables engineers to anticipate and avoid incidents that could delay drilling operations.

Results

Protects engineers

working on rigs in tough environmental conditions

Helps avoid 75% of disruptions

that cause most of non-productive time

15% reduction in drilling costs

anticipated when the solution is fully up and running

Business challenge story

Optimizing drilling operations

With the global oil market dogged by low prices, the race is on for energy companies to find more efficient ways to extract natural resources, without reducing margins or increasing operational costs. For Gazprom Neft, one of Russia’s leading oil companies, increasing the efficiency of its drilling operations is a top priority.

Igor Simon, Function Manager at Gazprom Neft explains: “Drilling of highly efficient, productive and profitable oil wells is a complex endeavor, and requires us to try and control a wide variety of often unforeseeable events. No matter how effectively our engineers have prospected a new well using geophysical modeling, the ever-changing conditions inside a well can cause a series of sudden, disruptive incidents that lead to non-productive time.”

Non-productive time (NPT) is a term drilling engineers use to describe periods when progress ceases during drilling processes. There many types of NPT that can occur at a rig, but among the most expensive are those that result from different faults or issues that arise during the drilling. Because the cost of operating wells is high—often running into the millions of dollars—oil companies seek to minimize NPT as much as possible. The inefficiencies can hurt revenues, and some of the technical faults that cause NPT can be hazardous—so preventing such events from occurring is an essential part of keeping rig engineers safe.

“At Gazprom Neft, we drill about 1,000 horizontal high-technology wells each year, and at least 50 wells are drilled simultaneously. If we can improve the efficiency of a well drilling, we can reduce operational costs significantly and lift our bottom line,” says Simon. “We recognized that if we developed the capability to detect events before they occur, we could streamline our drilling operations and avoid incidents that can lead to NPT. To achieve this, we needed to gain real-time insight into the conditions of each drilling well. We decided to find a way to mobilize technology to help us achieve this goal.”

Inspired by recent developments in AI, Gazprom Neft decided to create a prediction system and deploy it in its Drilling Management Center (DMC). The center, tasked with using machine learning and predictive analytics technology to improve drilling operations, assembled an expert team of drilling engineers and data scientists who could help the company develop next-generation well monitoring solutions.

We set out to develop different predictive drilling analytics models that would help us identify and avoid over 75 percent of preventable causes of NPT.

Igor Simon, Function Manager, Gazprom Neft

Transformation story

Taking a collaborative approach

Gazprom Neft engaged IBM® Analytics to help it build a predictive drilling analytics solution that can identify a wide range of NPT events before they occur. Working closely with Gazprom Neft’s expert drilling engineers, the IBM team developed and trained machine learning algorithms to monitor drilling conditions and provide alerts whenever a possible problem was detected.

“We decided to work with IBM because they have a proven track record of developing predictive drilling analytics solutions for the oil industry,” says Simon. “IBM provided the data science expertise that we required to understand the true potential of our data. The collaborative approach that we adopted was so effective that working with the IBM team felt like working with our own internal teams.

“Together, we developed a solution that suited the needs of our drilling engineers. We took care to ensure that the user experience was tailored to the daily work of the staff at the DMC, which reduced the time and resources we needed to spend on user training.”

Using historical data taken from thousands of real-world drilling operations, the IBM team trained machine learning algorithms specifically designed for predictive drilling analytics use cases. To structure the project, Gazprom Neft identified several typical causes of NPT—including differential pressure sticking, mud loss, gas shows and bit balling—and developed predictive models that would help engineers predict each of these issues.

To test the machine learning algorithms, Gazprom Neft ran them through a series of simulated drilling scenarios it chose to reflect the diversity of challenges that occur in different types of wells. Simon explains: “Every well is unique, so it is incredibly important that we validate the accuracy of our solution and enhance its flexibility by testing it on a wide variety of complex, high-tech well conditions.”

Simon adds: “Thanks to the close collaboration between our staff and the IBM team, we are now able to build, test and validate a new machine learning model in just one month. This rapid pace of development will ultimately enable us to move into production on target with our project deadline, so we can start reaping the rewards of data-driven drilling data analysis.”

Results story

Reducing operational cost, improving safety

As Gazprom Neft prepares to go live with its predictive drilling analytics solution, the company expects to see significant reductions in operational costs and hopes to eliminate most preventable causes of NPT.

“We set out to develop different predictive drilling analytics models that would help us identify and avoid over 75 percent of preventable causes of NPT,” says Simon. “Thanks to the support of IBM data scientists, we have already achieved this benchmark in the test phase of our solutions.

“We expect to see a 15 percent reduction in the total costs of our drilling operations once we go live. When you consider that one well can cost up to a million dollars to the company—and that’s a conservative figure—even a small operational efficiency saving can add up to hundreds of thousands of dollars saved on each rig. In tough economic conditions, these savings will free up the resources to help us stay ahead of our competitors.”

In addition to delivering savings, Gazprom Neft’s predictive drilling analytics solution significantly reduces the amount of low-value, manual analytics work that its drilling engineers must undertake, yielding more time for them to focus on further enhancing the company’s data-driven drilling capabilities.

Simon explains: “Because the solution delivers near real-time data on the condition of each of our wells, it will make life easier for engineers working at our DMC. At the same time, the solution will strengthen the safety of our rigs, as it will enable us to predict and prevent hazardous well events such as the buildup of flammable natural gas.”

Looking ahead, Gazprom Neft will explore ways to refine its predictive drilling analytics capabilities by applying the technology to the construction phase of oil wells, too. Simon concludes: “With IBM helping us to achieve safer, more cost-efficient drilling operations we are better placed to weather tough market conditions and retain our place as one of Russia’s leading oil providers.”

business logo

About Gazprom Neft PJSC

Gazprom Neft is a vertically integrated oil company primarily engaged in oil and gas exploration and production, refining, and the production and sale of petroleum products. Gazprom Neft is one of the largest oil producers in Russia. Headquartered in Saint Petersburg, Russia, the business has approximately 50,000 employees.

Take the next step

To learn more about the IBM solution featured in this story, please contact your IBM representative or IBM Business Partner, or visit the following website: