25 April 2019

What’s the role for Big Data and AI in the oil and gas sector?

“Robots will steal HALF of all jobs in the world within 15 years creating unemployment crisis for millions, AI expert warns”, is a real headline from a leading UK national newspaper. This warning says jobs such as truckers, salespeople and couriers will become obsolete as artificial intelligence takes over.

Contrast this with another headline – “Robots and AI will actually create more jobs than they take”, and who knows what to believe.

What we do know is, Industry 4.0 and industrial digitalisation is creating a new way of working across all sectors and it will change the way we work. While the oil and gas industry has been slower than others to embrace big data and artificial intelligence (AI), the pace of adoption in the industry has accelerated and is now an area of innovation for most companies.

The technology available today can provide operational insight at a new level, help operators locate and develop reservoirs, enhance how operators produce and refine crude oil, increase process automation and operational efficiency, and even optimize business activities such as how companies market their products.

Using AI applications in anomaly detection for example, can help to free up time for engineers on site so their time can be invested in more value-added production activities, making operations more efficient. AI can take some of the mundane, but critical tasks, and with the ability to process data at speed, find anomalies and understand a problem will happen, much earlier than any human could ultimately leading to more uptime for customer assets.

But underpinning all of this is data. We need big data to train the AI to know what normal looks like and without it, AI won’t work.

Using customer data to understand normal and adding in historic failure data into an Industrial Internet of Things platform (IIoT), such as Siemens’ MindSphere, can provide the algorithms to train AI in a structured way. If you add to this the knowledge from operators and engineers and suddenly the AI becomes an integral part of decision support within anomaly detection.

By using the IIoT platform, data from multiple assets can be shared so that machine learning is robust, and all customers benefit from the shared knowledge of issues which may not necessarily have happened in their operation. This is important because compared to say trying to get AI to drive a car, there are only a finite number of situations where we can take the data needed to learn from in the oil and gas industry. It also would alert operators working in similar applications and environments of potential issues which need their attention – in theory providing even more notice to take action.

This isn’t pie in the sky. Siemens has real world applications of AI support where issues have been detected before any fault occurred. Our pilot programme centred on electric submersible pumps (ESP), which is used to lift fluids from wellbores.

We operate hundreds of thousands of units across the world, and by using the shared data, have trained AI in anomaly detection. In one pilot use of this applica­tion across a fleet of 30 ESPs, the failure of one was accurately predicted two weeks before it occurred.

This is a step-change in current field monitoring systems. ESP downtime can be costly, but this predictive AI model allowed asset downtime to be reduced, and remedial action to taken. With more time to prepare, a decision can also be taken on whether to manage the issue until the next planned shutdown, or if the failure isn’t avoided, a rig to replace the pump can be mobilised sooner.

The AI works in tandem with a traditional supervisory con­trol and data acquisition (SCADA) system meaning if the AI misses a fault, the SCADA system will still pick it up. But the AI takes the SCADA data and uses it to continuously learn about faults and what normal looks like.

The benefits of such a system are huge. As in the pilot, critical failures can be avoided which could potentially impact the operation of an entire reservoir. This means operational efficiency is improved for ESP availability and most importantly, the reservoirs profitability.

Today, Siemens has more than 200 data scientists working on AI research using decades of oil and gas industry experience. While we’re just scratching the surface of the potential of AI, there are many new applications expected which will improve asset utilisation and profitability across the sector. ESP can also mean extrasensory perception – or a sixth sense – and with AI, that’s what you’re getting to run oil and gas assets.

Find out more at www.siemens.com/digital-oil-gas

Related Tags