In the first part of this featured article, Jane Ren, CEO and a founder of Atomiton, explained how her company’s work with Artificial Intelligence and its application to the Industrial Internet of Things in the oil and gas industry. In this second part, Jane Ren describes scaling AI up to the enterprise level, its contribution to reducing an operation’s environmental impact and her vision for AI’s future in energy generation.
Computing on the edge
Jane Ren explains that one factor that goes hand-in-hand with applying AI to IIoT is to move some processing of gathered data closer to the sensors. Referred to as “edge computing”, the resulting increased speed from running AI algorithms on the edge is significant. She explains, “In our view, edge is what bridges the enterprise and day-to-day operations, although this is actually very complex because you cannot just connect machines directly to an external network.”
“But I think there’s a huge potential, which is where we are focusing our product effort on this year – the edge between the enterprise and the operation – and we will have some good things that will be offered to the market,” she adds.
Operational technology – OT – the field of computing that deals with monitoring and control of devices and processes has been a foundation for the development of the Internet of Things concept. Ren says, “People often talk about IT and OT – about IT and operational technology convergence – and the message I sense is: Where is this heading?”
“Actually, with some of our customers who have started early, this is already an enterprise-level initiative. There are a few applications that extend the benefit of day-to-day operational agility to dynamic business decisions, enterprise-wide. For example, an enterprise that owns oil decides where to put its inventory,” Ren continues.
“One of our customers has up to 20 terminals across different geographies,” she adds. “It’s very important, based on the local demand and the predictability of local demand, to be able to decide inventory allocation.” Ren explains that although the cost of terminals use is a related enterprise level concern, “this ability may be limited because the information from the ground level is not always acceptable for artificial intelligence at the enterprise level.”
Jan Ren asserts that profitability in engineering, procurement and construction (EPC) increases from applying AI at the enterprise level. “There, the enterprise level is about project completion, billing and invoicing. So, at that level, construction data coming from sensors and equipment can now be processed by better algorithms and automatically feed into invoicing to make the revenue cycle turn faster, versus manual management processing the data.”
“And that includes the enterprise level of decision. In short, were already helping pushing this at the enterprise-level with some of our customers, some fairly large oil and gas companies,” she adds.
Environment and cost convergence
From a green perspective, AI can be applied to help reduce an operation’s environmental footprint, while saving costs, Ren explains. “In fact, there is a lot of consciousness or awareness in reducing the energy footprint, the environmental footprint, from our large global oil and gas customers.”
She continues, “It is extremely resource and energy intensive to produce energy. Upstream drilling uses a lot of machine electricity and often a lot of generators. In midstream, to transport and move oil and gas takes energy. And downstream, 50% to 60% of the operating cost of a refinery is consumed energy. At Atomiton we have find ourselves coordinating the processes in this industry –how oil and gas customers reduce energy use to get more efficiently for even better margins.”
“One example is a downstream refinery,” she explains. “One of the customers we work with found that they use a lot of steam. The key to steam use is use it or lose it, and at the end of the day, if you don’t use the steam you’ve generated, the heat may be released into the environment.”
Ren continues, “By using artificial intelligence now, you can actually better predict the steam demand. Do that in line, which means not only that an operator can predict for the next 2 or 3 hours but can adjust steam generation decisions and use less boiler or cogen equipment than they would have by making an intuitive judgment. AI closely monitors the process, which can save energy up to 15 to 20% in terms of steam cost.”
“We also work with midstream oil and gas storage,” says Ren. “This is where a lot of energy is used on loading and unloading into the tanks, as well as pipeline heating. Also, worldwide, there are different utilities’ policies to charge these companies when it comes to the peak usage. That means an operator needs to smooth out usage, which is not easy because at any particular terminal or storage plant there are multiple activities going on at the same time.”
Tracking operations and making cost-saving decisions to avoid waste is a challenge, explains Ren. “It’s really hard for the human mind to coordinate and arrange all these activities based on their priority and time window so they are able to reduce the peak usage. Our software also takes the real-time information and is able to outline the best possible algorithm in line, the ability to shave off the peak usage. So previously we’ve been able to reduce these peaks by as much as 20 to 25%, which resulted in savings that vary depending on a utility’s billing pricing policy.”
When asked about how applying AI to renewables such as wind farms, Jan Ren replies, “We do see that as a convergence for many of our customers going down the line.”
“I think that even though we’re not working yet on the renewables side, in general, the eco-consciousness of the industry is also lining up with the need for generating higher margins. And we see a huge amount of efficiency that’s possible now with artificial intelligence that can run in line.”