Here's why energy companies are adopting AI

FlickrPlug and Play Tech Center
Joshua Aviv, founder of SparkCharge, pitches the audience at the Plug and Play Energy Spring Summit 2019.

Big energy companies from oil and gas companies to utilities have a big problem.

They need more and better data to run their businessess more cleanly and efficiently, as well as to deal with the increase in floods, wildfires and extreme heat that's putting pressure on their infrastructure. But they're not generally the fastest players to embrace digital technologies or innovation. 

That problem is an opportunity for the energy program at Silicon Valley-based accelerator Plug and Play. The group connects big global energy companies — from Japanese utility Tokyo Gas to American oil giant Exxon Mobil — with startups that are developing technology that the energy industry can use to save money, decarbonize or build more distributed energy systems.

Last week, Plug and Play held its Spring Summit, featuring 15 entrepreneurs on stage pitching their startup ideas to energy executives and investors in the room. Artificial intelligence (AI) and machine learning, not surprisingly, were featured in more than half of those pitches. Others, such as SparkCharge, which is making a mobile EV charger, were focused on the electrification of transportation.

Many of the companies are using AI to do things such as slash energy use in buildings, monitor powerlines and pipelines, and predict when energy gear might fail. Technology that can help energy companies better predict how their energy infrastructure will perform or when it will need maintenance was a big theme.

"AI and machine learning are tools that can apply across the board" for the energy sector, said Wade Bitaraf, founder and leader of Plug and Play's energy practice. The group looks for startups that can use tools such as AI and machine learning to help energy asset managers and operators make smarter decisions, and reduce cost and risk in maintaining their assets, Bitaraf explained.

Plug and Play plans to open a practice in Houston — home of many oil-and-gas companies — and Bitaraf and his team are already looking for their next cohort of energy startups. No doubt, AI and machine learning technology will continue to be a major theme when it comes to selecting the next group of companies to feature.

Here are three ways startups in the Plug and Play accelerator are using AI and machine learning to try to transform the energy sector:

1. Wildfire powerline and gear monitoring: As the startups were wrapping up their pitches Wednesday, Twitter was ablaze with a report that found that PG&E's gear was indeed responsible for the deadly Camp Fire last year. Yikes. Utilities are desperate for technology that can help them prevent similar occurrences.

For example, a startup called VIA, based in Somerville, Massachusetts, has created a blockchain-based application that it says can help utilities better predict when transformers might be at risk in a disaster. The application makes better use of energy data sources, whether that's data from smart meters or equipment inspections. 

A Korean startup called Alchera uses thermal and standard cameras combined with AI-based image recognition to monitor powerlines and substations in real time. The AI is trained to watch the infrastructure for any abnormal events such as smoke, falling trees or intruders. The company says the tech is already being used by Korean utility KEPCO. 

2. Predict when energy hardware needs maintenance: Energy companies collectively lose $27 billion every year due to reactive operations, says Ensemble Energy CEO and founder Sandeep Gupta. The Palo Alto, California, company combines diverse data sources and machine learning to predict when assets, such as the gearbox of a wind turbine, will need to be maintained or might fail. Ensemble Energy's AI can be used for any type of energy asset, such as a solar system or hydropower plant, Gupta said. 

Another Silicon Valley startup called PreNav uses drones equipped with lidar and cameras to develop super-detailed 3D models of infrastructure, such as a power plant or a dam. Those 3D models (also called a digital twin) can be used to inspect for cracks, corrosion or other changes over time.  

Oil and gas pipeline monitoring is also another hot area for AI. A Tampa, Florida-based startup called mIQroTech builds sensing devices that can be placed every five miles on a pipeline and which collects various data that can help predict if there will be a leak. The company says its models can predict with 96 percent accuracy if there will be a pipeline leak. 

3. Cut plug load power in buildings: No one really seems to pay attention to the energy being used from individual sockets in offices and campuses. But a Philadelphia company called Sapient Industries has developed a smart plug system that's managed collectively by AI, so it can both automatically reduce wasted energy and provide insights into how building device energy is used. Sapient Industries co-founder Sam Parks said during his pitch before the Plug and Play audience that the AI-controlled smart plug devices can cut building energy consumption by 20 percent. 

Reducing energy use in buildings has long been an interesting area for AI. Google acquired the Nest AI-learning thermostat five years ago, and last year Siemens acquired the AI-based Comfy app.