Industry insight: Feeding the machine, AI as the next energy storage disruptor

Industry insight: Feeding the machine, AI as the next energy storage disruptor

Industry insight: Feeding the machine, AI as the next energy storage disruptor 500 358 Energy Storage Journal

 

For some, such as Elon Musk and the late physicist Stephen Hawkins, the risk to mankind in the development of AI is a pronounced and real possibility.

But in the world of energy storage the technology is driving forward the roll out of virtual power plants, where the flow of decentralized, behind-the-meter, power will be controlled by tech companies and no longer traditional system operators.

These energy storage systems of the future will be able to capture data on load, performance and grid dispatch (and much more) on a one-second basis, feeding terabytes of information to machine-learning software and predictive analytics.

And the more data that is captured, the smarter the network becomes.

In an exclusive interview with Energy Storage Journal, Brendan Harney, the director of business development at US information and technology services firm Stem, discusses the markets and how AI has become a market disrupter — only time will tell if its positive.

To read more about Virtual Power Plants email admin@energystoragejournal.com to sign up to Energy Storage Journal and receive your copy of this quarter’s edition, where we take an in depth look at the technology.

 

Where does battery storage and PV sit in the overall US power markets and supply chain — for example, in the UK it’s understood that storage will only play a small part in keeping the lights on as the country moves to a low carbon economy?

We believe that ‘behind the meter” energy storage that is enabled with superior artificial intelligence (AI) will play a large role in America’s energy landscape as regulators increasingly articulate its value proposition against other infrastructure and peaker alternatives.

Customers want it, the technology is dropping in costs and is scalable, and the trends of investment in energy storage point towards enormous growth.

Moreover, as renewable energy continues to drop in price, its penetration into the US power markets will rise, requiring new solutions that help smooth renewables’ intermittency and that create fast-dispatchable, highly localized solutions for the utility when and where they need it most.

The most important point is that storage and solar do not need to be co-sited to provide value to the utility and to the grid; in fact standalone storage can sometimes perform more functions, in the form of distribution congestion relief, local capacity, or other grid services.

For example, Stem dispatched over 600 times in the California wholesale market in 2017 to provide aggregated load reductions during unprecedented heat waves, with hundreds of dispatches on a five-minute response.

Energy storage — if amplified by cutting-edge AI, such as Stem’s Athena product — can reduce energy costs for non-residential customers while newly engaging them in Virtual Power Plant networks that provide grid modernization services, which, in turn, increases grid reliability, eliminating the need for billions in infrastructure spending.

What are the challenges companies face in monetizing storage and renewables in the current market?

A smart developer needs to be able to navigate a number of policy-related issues relating to the tax treatment and charging requirements for solar+storage, the evolution of net energy metering rules and time-of-use rates, and interconnection rules, to succeed.

To do so, they need superior software to direct the associated energy storage system, which is why Stem has made AI a cornerstone of our business model.

In terms of pure cost savings, having the acumen to effectively harness and deploy energy storage in a way that maximizes the economic benefit makes or breaks the deal for the customer.

Athena, our AI, is designed to solve that problem: it can rely on predictive analytics relating to weather, load, and tariff options, and automate the thousands of real-time calculations and decisions that would otherwise necessitate tremendous manpower, and therefore mitigate cost savings.

Will new technologies, economies of scale, and policy affect business models?

Absolutely, the continued improvement in lithium-ion storage and other component learning curves, the reduction of soft and hard costs through economies of scale, and federal and state policy decisions will affect the potential growth of the market.

Each will have an impact on whether customer-sited energy storage will be allowed to offer the 13 different services to customers, utilities, and grid operators that were so well articulated by Rocky Mountain Institute in its 2015 report on the Economics of Battery Energy Storage.

Moreover, US regulators are interested in moving towards new services that give customers increasingly more options and control over their energy decisions, and more ability to participate in grid modernization.

We are building the largest partner ecosystem to extend the ability to create service offerings, whether combining traditional Demand Response and storage DR working with CPower, addressing site-level demand management and reduction of the peak load contribution charge with constellation, our solar storage offerings with multiple solar partners, or in new combined offerings — our AI makes that possible and reliable.

As a result, Stem invests heavily in dialogues with state and federal policymakers to bring our experience to the design of sensible regulations that allow new technologies and business models to flourish for multiple stakeholders’ benefit at lower cost to ratepayers.

How much of a disrupter will AI and data analytics be to the market, and will the affect be positive or negative?

We see AI enabled through customer-sited energy storage as a huge, positive disrupter. Managing the timing of energy use in real time to minimize costs and respond to changing grid needs is extremely complex.

Even with the ability to store energy, there are still thousands of calculations, forecasting simulations, and split-second decisions required to produce meaningful results.

Energy super-intelligence is a new way of looking at storage — getting away from the concept of hardware towards what’s really needed to get the maximum value — an AI-based platform operating a network of energy storage systems that provides services to numerous stakeholders within the space of a day.

For example, Stem has over 860 systems installed or under contract in five US states, Japan, and Ontario, Canada. All our systems capture load, performance, grid dispatch, and other data on a one-second basis, feeding terabytes of information to our cloud, and in turn, feeding our machine-learning software and predictive analytics.

The more data we capture, the smarter our network becomes.

Utilities and grid operators today need these highly flexible, localized resources to manage the increasing intermittency of renewable energy penetration and local grid congestion events, and AI-driven energy storage can offer that data and real-time execution.

That’s what we hear repeatedly from our eight utility customers and grid operator stakeholders across our markets.