• Raine Scott

AI Software Can Replace Energy Storage Hardware in Most Cases

The way we’ve used energy has changed significantly in the last 20 years. Previously, we’d flip a switch and a utility’s power plant would slightly ramp up production to meet demand. Today we’re still producing energy from utilities, but we’re also generating and distributing power from other energy sources, such as solar, wind, batteries, and electric vehicles. Because power is no longer coming from a single source, that makes managing and optimizing grid power more complex.

California produces about one-third of its energy from renewable sources, but even at this level, an extremely large amount of the renewable energy produced is curtailed because we don’t have a use for it at the time its produced. (See the chart below.)

As a consequence, utilities have implemented new rate tariffs in the form of fixed demand charges and time-of-use (TOU) rates that encourage energy conservation during peak usage times.

Deciphering all of these price signals can be difficult for businesses. The traditional solution to demand charges and avoiding high TOU rates is to install solar with a very large and expensive Lithium-ion battery. The solar produces energy during peak periods of the day, and the battery can store any extra solar power and discharge it when there’s a sudden demand spike, such as turning up the air conditioner.

While this hardware solution can offset peak demand charges, it’s not a perfect solution. Solar with batteries requires extra permitting, fire suppression systems, an extra inverter, extra wiring, and extra time to install, all of which increase the upfront costs and decrease the solar building’s return on investment. To put the economics in perspective, a study by Energy & Environmental Science claims that if utility-scale battery storage were as cheap as possible, storing 12 hours of energy would cost $2.5 Trillion and raise the price of energy from ~$50 per MWh to over $1,500 per MWh.

A more recent alternative to solar with batteries is solar with artificial intelligence software or “AI.” We refer to DemandEx as a “virtual battery.” While a traditional battery manages demand spikes by absorbing the surges in real-time, our AI software solution predicts spikes before they happen, then avoids or reduces these peaks by preemptively shifting the predicted demand to a less expensive time-of-use period. This AI solution provides similar savings as a battery solution, but at 1/10th – 1/20th the cost.

AI is more cost-effective than energy storage for several reasons. First, the solution mainly consists of software and inexpensive off-the-shelf communications hardware and smart thermostats. Second, there’s no permitting, no fire suppression systems, no extra inverters, or any extra wiring. Third, the installation is simple and takes less than a day, reducing electrician labor costs.

How DemandEx AI Solution Works

DemandEx works by first collecting data. Lots of data. This data includes electric rates, the building’s past and current energy usage, real-time weather data, real-time solar output, and data about the building’s peak energy consumption and usage time. DemandEx uses this data to make predictions about future energy usage, the potential solar offset, and the potential demand peaks. Then DemandEx sends all these predictions through its cost optimizer.

DemandEx's cost optimizer is AI that makes the most cost-effective energy usage decisions 24/7. It reduces demand peaks by controlling the building’s flexible loads, such as heating, ventilation, and cooling (HVAC), electric vehicle chargers, and compressors. By intelligently turning these devices on and off at the best times of day or night, DemandEx reduces the building’s total energy bill without batteries.

In summary, DemandEx’s AI solution provides:

  • Peak savings at less than $50/kWh installed price

  • Works with existing equipment

  • Single-day install

  • Requires no permitting

  • Has no performance degradation or replacement cost

  • Has no fire suppression equipment

If the building manager’s goal is to reduce demand peaks, then the AI solution is clearly more cost-effective than batteries. But what if there are two goals? What if the building manager also wants energy storage for grid-stability and backup power?

While AI can’t store solar power, it can work with batteries to optimize them for resiliency applications. When configured for storage, DemandEx’s AI handles the demand management side and a smaller and less expensive battery is dedicated to resiliency applications, potentially saving tens of thousands of dollars by ‘right-sizing’ the storage system for backup power.

To find out more about DemandEx with and without storage, read our previous blog where we take a deeper dive into how DemandEx provides the demand charge savings equivalent to a battery with just software.

  • Grey LinkedIn Icon
  • Facebook
  • Twitter
  • YouTube


2120 University Avenue Berkeley CA 94704

© 2020 by Extensible Energy.