Blog

Big Data Analytics in Oil & Gas Industry - Ahmed Khamassi, Former VP Data Science, Equinor

Ahmed Khamassi, Former VP of Data Science of Equinor with 12+ years’ experience in top global companies including J.P. Morgan, Wipro Digital, PayPal, and Google gave an interview in a new BGS Talks episode 

Equinor is the largest operator on the Norwegian continental shelf who aims to be a digital leader within the industry. The company established a Digital Centre of Excellence in 2017 to increase the value generated from data. In 2019, Equinor’s cash flow impact from digital initiatives reached USD 400 million.

Ahmed discussed how he started his career as a data scientist, the role of Big Data in the Oil & Gas industry, Omnia Data Platform, cybersecurity, how data can enhance safety and increase ROI. 
Watch the interview, read the highlights and subscribe to BGS Talks YouTube channel for new episodes! 

Before arriving in the oil and gas industry, you worked in Google, SAS, and other born-digital companies. However, the oil and gas industry is not like that. 

Culturally, I’m very different from a typical oil and gas employee. In the oil and gas industry, the first thing we care about is safety, making sure that things run properly, even though we don't have the most innovative way to do it. If you inject a little piece of code in Google Photos and it creates a little bug the next day, it'll be rolled back and nobody would be hurt. But if you make the wrong decision on a platform then you will have a big impact. It means I need to operate at a different speed to what I'm used to, to accept less change.

Equinor has its own data platform; tell us more about it.

Omnia is the cloud-based data platform where we develop solutions for managing our data. Our main partner is Microsoft Azure. The data type that we consume the most is the sensor data and Omnia platform will allow us to get this data out of the IMS systems and manage it. We also use it for text data, for safety, for subsurface data, and everything else. It's an ever-evolving element because we keep solving problems and we add cases as we progress. It is also a place where we can collaborate. We standardize a lot of what we do; with Omnia we have one place where we have the common pattern for data and where we share them.

As I understand, Omnia is unique because it's too difficult to put all data in one place, however, Equinor accomplished that.

Even though it's not Google, not Facebook in terms of data, Equinor did not make terrible mistakes in the past. Equinor looked after its data. That helped us to speed up. Equinor has also been early in terms of digitalization, thinking about cloud and platform and they experimented a lot even before the whole digitalization initiative was institutionalized. The other thing is we have a Digital Center of Excellence. Almost all of the programs we have that consume a lot of data, happen in one place which is the Digital Center of Excellence. That allowed us to standardize and grow Omnia. Omnia starts to do sensor data, and then we have a product focused on using text to learn about previous safety incidents, so we need to solve the issue of how we deal with text, with the computation on it, with the knowledge graph. That helps us to have one place where we have this text data, this knowledge graph sitting next to the sensor data. We don't merge them today but Omnia gives us a heads up on this merging.

At what stage is it now?

It's growing but it's half-baked and the reason for that is we are focusing on two things primarily. One is automation and the second is scale. We are about to create 9,000 machine learning models for predictive maintenance.

Is it Omnia.Prevent project?

Yes. We will do that across everything on the Norwegian continental shelf. Omnia.Prevent is one of the key value potentials for us as an oil and gas company. Generally, you take a machine and you have sensor data, you know when it fails and try to use machine learning to predict these failures. That is a very long process. Omnia.Prevent flips the problem statement on its head. We have the equipment with sensors and we need to be able to understand when one of these is working normally and when it's not. So we need AI to distinguish between the sensor values that I’m expecting to see and what I'm not expecting. This is where the engineers help. And if I do that in an automated way, then I can have machine learning everywhere.

Follow the link to watch the full interview.


ADDITIONAL OPTIONS FOR YOU:

  • find more interviews at BGS Talks YouTube channel and subscribe for new episodes
  • find out more about Automation and Digitalization Congress AUTOMA at automacongress.com
  • request marketing materials here
  • share the article in social networks using the buttons below and our tags #AUTOMA2021 and #BGStalks

back to the news list