Business challenge

With fierce competition for capital allocation and low oil prices, oil companies must reduce non-productive time to keep operations profitable.

Transformation

In a first-of-a-kind trial, TOTAL is developing real-time Predictive Drilling Analytics capabilities with IBM® Analytics solutions—ushering in a new era of ultra-efficient, precision drilling.

Results

Empowers

engineers to predict complications that can cause NPT with greater accuracy

Millions of dollars

saved by understanding hidden risks and optimizing well delivery

Helps

a new generation of engineers follow in the footsteps of those before them

Business challenge story

Delivering a technical miracle

As crude oil prices fluctuate, and shale gas and renewable power encroach on market share in the energy sector, oil companies are eager to find new ways to reduce costs and increase margins. Increasing the productivity and profitability of their wells is crucial to meet these challenges.

To drive profitability, oil companies must reduce the costs of drilling wells, and aim to keep their wells running productively and efficiently for longer.

Maintaining consistent drilling and production on any oil well is a highly technical endeavor. The drilling and completion process is incredibly expensive, and the costs need to be recouped through production. Moreover, downhole conditions (temperatures and pressures) within a well increase the likelihood of encountering technical difficulties that can slow down the drilling process.

Joy Oyovwevotu, Lead Drilling Engineer at TOTAL, elaborates: “Time spent drilling and completing a well is very expensive and can be divided broadly into productive and non-productive time [NPT]. Some NPT can be avoided with sufficient warning and a good understanding of downhole environment and interactions.”

“Stuck pipe” is a significant cause of NPT that could be preventable and occurs when a drill string or casing pipe can no longer be moved in a well as a result of mechanical issues, or due to difference between wellbore and formation pressure holding the string or casing against formation. Whilst some occurrences of stuck pipe are recoverable, others are not.

In some cases, stuck pipe may lead to the abandonment of a well-bore, which means re-drilling formations already drilled before the stuck pipe event, at a cost of millions. In other cases, the situation may be recoverable, but the cost implications can still be huge, impacting the profitability of the well. Stuck pipe can also lead to changes in the well architecture that can complicate how a well is completed and disrupt production from the well as well as increase maintenance later in the well’s lifecycle.

Drilling engineers at TOTAL use their vast industry expertise to anticipate and mitigate the risks of stuck pipe alongside other engineering and geological challenges. However, even with extensive experience, engineers still run into drilling difficulties due to the challenging drilling conditions.

Oyovwevotu says: “To mitigate the risks in delivering a well, our engineers plan and monitor drilling operations using highly detailed physics-based models. However, these models are limited and can take a long time to produce and interpret, so the information they produce is often out of sync with a well’s actual condition.

Physics based modelling can be time-consuming and difficult to deliver at a granular level in real-time whilst drilling often relies on long term trends that can mask smaller events that may lead to stuck-pipe and hence can offer limited insight into conditions that lead up to stuck-pipe.   

As drilling continues, the conditions of the well change continuously both as a function of the rock being drilled at that moment and the borehole that has already been drilled. Modelling this with pure physics-based models is something the industry has attempted for many years with limited success. 

“To address this challenge, we set out to invest in a hybrid model that could combine data from the physical model of a well with a statistical model that can ingest large volumes of real time data about the condition of a particular well," says Oyovwevotu. "If successful, this hybrid approach could help us gain clearer insights into the current state of a well and how it may change in the future."

Additionally, with an aging workforce, it is more important than ever for TOTAL to bring new engineers up to speed and equip them with the skills and knowledge they need to manage successful drilling operations.

Oyovwevotu comments: “Drilling rigs are highly complex pieces of equipment, and it is impossible for the human eye to monitor all of the rig’s sensors and interpret them while drilling in real time. As a result, much of a driller’s knowledge about drilling is not based purely on sensor readings, but also on their personal experience and a feel for interpreting contextual conditions and information. This kind of experiential knowledge is tacit, almost impossible to record or pass on to the next generation—which is where the predictive drilling solution comes in.”

By using technology to monitor and interpret sensor readings in real time and provide decision support for its drilling engineers, TOTAL would be able to reduce NPT and empower future generations of hydrocarbon explorers to operate successful, productive rigs.

When you consider that NPT can cost oil companies millions of dollars in lost revenue, our Predictive Drilling Analytics solution could dramatically increase the profitability of extracting oil by helping to avoid NPT in the drilling phase.

Joy Oyovwevotu, Lead Drilling Engineer, TOTAL

Transformation Story

Drilling down into data

To revolutionize the efficiency of drilling, a team of TOTAL engineers joined forces with IBM technology experts to develop a Predictive Drilling Analytics application.

TOTAL’s development project is unlike any other, bringing together the combined power of statistical modeling and traditional physics-based models to help engineers predict events that can cause NPT.

Oyovwevotu explains: “In this project we built a model that amalgamates statistical data with the knowledge and experience of veteran engineers and data from physics-based modeling. By capturing the expertise of engineers in the model, we aim to build upon their experience and fill in gaps in our knowledge and understanding using statistical modeling.”

TOTAL selected IBM as its technology partner for this project, and set to work building statistical models of its drilling operation utilizing tools from the IBM Watson® Studio (previously known as IBM Data Science Experience) platform, which is hosted on the IBM Cloud™.

Oyovwevotu says: “Our engineers were understandably skeptical about this project, as they have built up extensive experience managing the risk of drilling wells for decades and were not sure if Artificial Intelligence (AI) can help them anticipate and manage downhole problems better than they could themselves. Because of this, we were cautious to ensure that we reduced the risk of the project not working. One of the ways we did this was to adopt an iterated, agile approach to development.”

TOTAL’s choice of technology for its Predictive Drilling Analytics project also helped keep risk to a minimum. Oyovwevotu adds: “One of the fantastic things about the IBM Cloud is that we didn’t have to make a large up-front investment in infrastructure, we could simply spin up a development environment that matched our needs. This helped us to keep costs low while providing excellent scalability.”

Each well and well-section produces subtly different but numerically large amounts of fast-moving data, so TOTAL required infrastructure that was not only flexible and high-performing, but also cost-effective. With the IBM Cloud, TOTAL scaled capacity to match its data and processing needs—eliminating the expense of underutilized infrastructure and delivering excellent performance.

“Developing in the IBM Cloud also helped us to kick-start the Predictive Drilling Analytics project,” continues Oyovwevotu. “One of our initial challenges was how we could monitor a well without changing anything on the rig, especially in the context of this first-of-a-kind project. The IBM Cloud solved this problem, because it removed the need for us to install IT infrastructure on the rig itself. We could get our project up and running easily with just the data itself, an internet connection and a web browser.”

Oyovwevotu reflects on TOTAL’s partnership with IBM: “Working with IBM on this project was a great experience. Not only did the IBM team provide cutting-edge technology and analytics expertise, they also recognized the impact that Predictive Drilling Analytics could have on the oil industry and brought a genuine ambition to make a real difference with the project.”

We look forward to developing our Predictive Drilling Analytics solution even further with IBM and we are eager to help the industry harness the transformative potential that this technology has to offer.

Joy Oyovwevotu, Lead Drilling Engineer, TOTAL

Results story

Beckoning a new era of precision drilling

To date, TOTAL has successfully tested its Predictive Drilling Analytics solution on two different wells. Already, the company can see that its solution holds the potential to revolutionize the way wells are drilled, by enabling oil companies to achieve millions of dollars in efficiency savings.

Oyovwevotu says: “In the first trial we only had a very small window of opportunity to prove whether the Predictive Drilling Analytics solution actually worked. Even on the first well we were able to establish evidence that the model can help engineers predict stuck pipe and other events that cause non-productive time.

“What’s more, the second well on which we tested the Predictive Drilling Analytics application was quite different. We were initially unsure how well the model would work with a different well design, but we were pleased to see that the model’s machine learning algorithms could easily and successfully adapt and generate positive results.”

IBM has agreed to continue training and refining the Predictive Drilling Analytics solution. Once the model is in production, TOTAL anticipates that it could help oil companies significantly reduce NPT.

“If engineers can predict and avert events that lead to NPT, oil companies can keep their well costs down in this new challenging era. When you consider that NPT can increase well cost by millions of dollars, our Predictive Drilling Analytics solution could dramatically increase the profitability of extracting oil by helping to avoid NPT in the drilling phase.”

Equally, as the expertise of industry veterans is built into the Predictive Drilling Analytics solution, it can help address the challenge of an aging workforce. This empowers new entrants to the sector to draw upon the collective wisdom and historical insights of their predecessors, and to continue their legacy of drilling safe, efficient and profitable wells.

Oyovwevotu concludes: “We look forward to developing our Predictive Drilling Analytics solution even further with IBM and we are eager to harness the transformative potential that this technology has to offer.”

About TOTAL

TOTAL is the world’s fourth-largest oil and gas company, and a major stakeholder in renewable energy. Headquartered in France, TOTAL operates in more than 130 countries and employs 98,000 people worldwide. In August 2017, TOTAL acquired Maersk Oil, a company owned by the A.P. Moeller Maersk Group.  

Solution components

Take the next step

IBM offers one of the world's deepest and broadest data science platforms that allows teams to explore, build and put their data science practice into production faster. For more information about how IBM’s data science solutions  help  transform industries and professions with data, visit ibm.com/analytics/data-science. Follow us on Twitter at @IBMDataScience, on our blog at ibmbigdatahub.com and join the conversation #IBMDataScience.

View more client stories or learn more about IBM Data Science