The choice between reducing greenhouse gas emissions and maintaining the economy is a false dichotomy, according to research that suggests that moving to green energy by 2035 is both doable and economically beneficial. In other news, a chicken producer is using agrophotovoltaics to improve its feedstock yields, artificial intelligence (AI) could lead to better microgrid performance, and lighting devices that can talk to other building systems can reduce costs and increase energy savings.

90% Green in Another 15

A recent study by researchers at UC Berkeley and GridLab shows that it will be both technically and economically feasible to generate 90 percent of our electricity from carbon-free sources in a mere 15 years. The report also indicates that with the right policies and incentives in place, moving to green power technologies will lead to steady job growth, decreasing electricity costs, a healthier population, and (obviously) reduced carbon emissions.

Report findings at a glance. (Image courtesy of UC Berkeley.)

Citing the falling prices of solar, wind and battery technology, the researchers foresee the wholesale cost of electricity, under a 90 percent-clean plan, being 10 percent lower in 2035 than it is today. While maintaining the status quo would lower wholesale costs a bit more, that doesn’t account for the economic losses incurred by the health hazards associated with burning fossil fuels, or the increased job growth that would result from a more aggressive renewable energy plan. Taking those factors into account, the green plan would save more than a trillion dollars in fuel, health and environmental costs, making it the most economical option.

Achieving the 90 percent renewable goal by 2035 is difficult but doable. (For a little perspective, so was landing a man on the moon, but engineers pulled it off less than a decade after President Kennedy threw down the gauntlet.) The researchers estimate that it will require adding 70 GW of renewable generation (primarily wind and solar, plus a little storage) every year, noting that “new renewable resources can be built cost-effectively in all regions of the country.”

Poultry, Pollinators and Power

Perdue Farms is making a move toward agrophotovoltaics—putting solar arrays on farmland—but with a twist: instead of growing crops in the solar fields, the company is planting pollinator-friendly plants as ground cover. Why? Well, in addition to raising chickens, the company also grows its own feedstock, which relies on bees and other pollinators to produce soybeans, peas and other chicken edibles. Land development and pesticide use have put a double-whammy on these beneficial insects, reducing crop yields and driving up costs. Putting a pollinator-friendly solar array next to the feedstock farm would let the adjacent crops produce more food in the same space. The ground cover won’t grow tall enough to shade the panels, so the area is pretty much maintenance free.

Pollinator-friendly ground cover with a solar array. (Image courtesy of Perdue Farms.)

Steve Levitsky, Perdue’s VP of Sustainability, said, “The land directly under solar panels has traditionally been filled with gravel or grass, and the cost to create and maintain a pollinator habitat is roughly the same, so there’s no financial barrier to incorporating pollinator-friendly ground cover into the facility.”

I’ll let Perdue put a feather in its cap for this effort, but I have to cry “fowl” (ha!) and point out that heavy pesticide use on the aforementioned crops is one reason that pollinators are in danger. While Perdue also raises organic chickens that eat pesticide-free food, their main product line is not organic. Nonetheless, this is a good start. After all, we know how a chicken company crosses the road: one leg at a time.

Networked Lighting Controls for Energy Management

The DesignLights Consortium has released a report on the benefits of networked lighting controls (NLCs) that are “multilingual,” so to speak. Intended for lighting designers, facility managers, utilities, and control system manufacturers, Interoperability for Networked Lighting Controls defines the term “interoperability” as “the ability to exchange actionable information between two or more systems.” For example, any occupancy sensor on a lighting system should be able to send information that’s relevant to controlling an HVAC system, regardless of whether the lighting and HVAC use the same communication protocols. Frequently, lighting and HVAC systems use their own occupancy sensors, leading to redundancy and increased costs. The ability to share data would facilitate simpler, less expensive systems.

Examples of interoperability. (Image courtesy of DesignLights Consortium.)

Interoperability offers the obvious advantages of energy savings through load shedding and demand response, as well as occupant comfort and convenience, but the report also discusses the benefits of an NLC to communicate with an energy dashboard that reports energy usage data to the power company, allowing the building to take advantage of energy-saving incentives offered by the utility.

Machine Learning to Optimize Microgrids

With the growth of renewable energy sources and the inherent security issues associated with a centralized grid, the energy industry is starting to pay more attention to microgrids—self-contained, localized grids that may or may not connect to the main utility grid. Maintaining the delicate balance between supply and demand, when both are highly volatile, requires a pretty sophisticated control system with the ability to predict both the renewable resource and the expected demand at any given time.

Or does it?

A team of scientists from Spain’s Institute for Research in Technology is investigating a machine learning technique called deep reinforcement learning (DRL), which doesn’t rely on predictions based on data that may or may not be reliable. Since every microgrid sees different conditions, it would be costly and time-consuming to create a control algorithm based on a model with questionable validity. Instead, the DRL method allows the system to learn from its experiences, gradually optimizing the system to fit its environment.

Reinforcement learning is a closed-loop control system. (Image courtesy of Energies.)

The researchers ran the algorithm on a simulated system consisting of solar, batteries, and a backup generator and saw favorable results after a two-year system training period. That’s pretty limited, but in a burgeoning field, it shows promise. The paper was published in the open-access journal Energies.

This Week in Green Tech brings you the latest news about renewable energy, electric vehicles, net-zero buildings, energy harvesting, IoT, and more. Subscribe to engineering.com’s “Electronics Design News” section to stay in the loop!