When the cable from the mainland that supplies power to Isle au Haut, Maine, was laid down on the seafloor to replace aging diesel generators, the community was told it had a 15- to 20-year lifespan.
That was in 1983.
For the last decade, confronted by the creeping urgency of replacing a power line that could fail any day despite careful maintenance, the community debated how to best keep the lights on in the future.
On the menu were a new cable, diesel generators, fuel cells, wave-energy generators, wind, solar, and batteries.
Cost was the primary concern.
With a full-time resident population of only 70 people or so and a summertime crowd of 200 to 300, Isle au Haut faced a steeper-than-usual capital infrastructure upgrade.
“When we went through the costs for almost any option, the economics were just really nasty,” said Jim Wilson, president of the private, for-profit cooperative Isle au Haut Electric Power Company.
But as the cost of renewable technologies plummeted, certain options began to make more sense.
A solar-and-battery system would run them around $1.8 million.
A new cable: double that.
A diesel system: triple.
So, four years ago, the co-op members voted unanimously to pursue a 300-kilowatt system made up of 900 solar panels, with a 1-megawatt graphene supercapacitor battery to store and supply excess power.
Residents can also add on their own solar panels along with battery systems (and/or water tanks to store excess energy as heat).
In most places, that might be the end of the story.
But there’s a hidden aspect to this solar project that tweaks it into next-wave territory: Behind each of the 140 or so electric meters on the island will be units loaded with machine-learning software developed by Kay Aiken at Maine-based Introspective Systems.
These intelligent control units learn the home’s energy-usage patterns, then coordinate that information with the island’s solar panels and batteries in real time.
This lets each point on the grid decide when it is best—cheapest, most efficient, or at a time preferable to the homeowner—to buy or sell electricity back to the grid, if they have their own panels and batteries.
“We didn’t mean to go to sea,” Wilson said, referring to an anecdote of a small ship that is blown to an unintended but not entirely unwelcome destination.
“We just wanted something to replace the cable.
But then you start looking at how to make it all more efficient, and one thing leads to another.
We’re building a solar project—just not the one we expected three or four years ago.”
Construction was slated to begin in spring 2020; because of the COVID-19 pandemic, that's been delayed, but Wilson still hopes the new system will be up and running by late summer or fall.
Once it’s completed, Isle au Haut will be part of a growing trend in renewable energy generation: autonomously controlled microgrids.
What Exactly Is a Microgrid?
Microgrids aren’t a new idea.
In industry parlance, a microgrid is a small network of electricity users with access to a local source of energy.
The users are all directly connected to the central grid, but during outages, the entire small network can disconnect itself from the central grid, or operate in “island mode,” to continue operation.
Deeply remote communities have relied on microgrids for decades, either because they’re at the end of a long and creaky power-transmission line or out of range of utilities completely.
Hospitals and other emergency and critical facilities also rely on locally produced energy for backup power; they’re often powered by diesel, propane, or another combustible fuel.
But microgrids can also be far more local: When you crank up your generator after a storm outage and run a few extension cables to neighbors, that is, in essence, a microgrid.
What is new is that more and more of these microgrids are being powered by renewable energy methods, thanks to increased affordability and shifting regulations that make them easier to adopt.
And it’s a booming business: Depending on your news source, the microgrid market is expected to be worth $47 billion by 2025, up from $28 billion this year.
In the United States, a massive amount of microgrid movement is happening in California, which has mandated that all of its electricity be from zero-emission sources by 2045.
Californians are also simply concerned that power is available at all: Between catastrophic wildfires, earthquakes, mudslides, and the “public safety power shutoffs” that the state’s largest utility, Pacific Gas & Electric, has recently implemented, people are turning to renewable-energy microgrids in an effort to free themselves of dependence on a large and increasingly erratic energy supply.
Of course, the movement to these microgrids is not just a California thing.
It’s happening in Puerto Rico, where solar-and-battery microgrids offered some respite after back-to-back hurricanes and earthquakes.
And New York state saw the launch of an $11 million prize fund to drive renewable microgrid development after being battered by Superstorm Sandy.
A community microgrid in East Hampton on Long Island supplies 50 percent of the town’s energy needs, keeps the local water plant running and fire stations operational in emergencies, and has allowed the local utility to avoid $300 million in transmission upgrades as a result.
Municipalities across the country are building them to keep critical facilities up and running no matter what, while using green, locally produced energy to power the facility in the meantime.
And microgrids offer energy resources and independence to rural communities and towns in India, Africa, and other developing areas of the world.
Even utilities are exploring ways to get into the game: In Illinois, Commonwealth Edison worked with the Chicago Housing Authority to build the Bronzeville solar-and-battery microgrid for 660 residential units in a low-income neighborhood connected to a nearby university microgrid.
It’s a pilot program to learn how clusters of microgrids can work together when disconnected from the main grid.
Part of the reason for this: Utilities are beginning to recognize they cannot completely guarantee energy supply to their customers, especially in vulnerable areas and at the edges of the grid, said Sascha von Meier, an expert in smart-grid technology and power distribution with University of California-Berkeley and the Lawrence Berkeley National Laboratory.
That’s leading to more of an appetite for investment in grid resilience—and renewable microgrids are playing a big part.
As more and more renewable microgrids come online—from college campuses and corporate headquarters to individual homes—the energy questions begin: How can these resources not only feed into a broader grid? And how can each point in an increasingly complex matrix effectively work together and even share directly with one another—a seemingly simple task the old grid just isn’t equipped to do?
Connecting Resources
In traditional energy-supply systems, control and optimization of power is coordinated among a relatively small number of centralized resources.
Control servers optimize the generation, output, and flow of energy from hundreds or thousands of power plants and send it over the lines accordingly.
But as more people in a given region purchase electric vehicles and embrace renewables, batteries, and smart appliances that vary their power consumption throughout the day, the number of points that are contributing to and drawing from the system begins to climb into the millions.
These “distributed electric resources” (DERs) become a big data problem that can’t be optimized or coordinated in a centralized way.
The solution? Distribute the computing resources, and back them up with nimble algorithms.
Andrey Bernstein, who researches autonomous grid control at the National Renewable Energy Laboratory (NREL) in Golden, Colorado, said that the computing part of this is less about how to handle millions of variables at once so much as it is about breaking them up into manageable chunks, then coordinating communication on nested levels—and making sure that as DERs interact with the grid, they don’t damage the system by accidentally overloading it.
Andrey Bernstein and a colleague (Photo: Dennis Schroeder/NREL) In the lab, NREL-developed algorithms were first tested using a computer-simulated electric feeder line with hundreds of thousands of DERs, then millions.
NREL then partnered with Massachusetts-based Heila Technologies to integrate the software into a small and inexpensive smart control unit that can be installed inside the home or property, behind the electric meter.
Bernstein is currently developing algorithms for optimizing energy distribution from a renewables-powered microgrid to and from the main power grid.
The software, which is being tested in Colorado, is designed to coordinate real-time demand and supply from high numbers of energy-generating and storage devices in homes on a microgrid—solar panels, electric vehicles, smart appliances—by performing the advanced calculations via a small, inexpensive computing controller at each point on the grid.
The first real-world pilot of this technology recently kicked off in the community of Basalt Vista, an affordable housing community in Colorado developed by Habitat for Humanity.
Each home is equipped with a full range of energy-saving appliances and solar panels, making Basalt Vista a carbon-neutral community.
Homes under construction in the Basalt Vista community.
(Photo: Joshua Bauer/NREL) Once the NREL/Heila smart controls are installed at each node within the Basalt Vista neighborhood microgrid, each point can become self-governing but can also interact locally with other nodes to optimize the energy flow to and from each home in the community.
They can determine when to draw power from the local grid (and other users on the grid), when energy from more distant resources is required, when to store energy in batteries or in plugged-in electric vehicles, and learn and anticipate patterns of fluctuations in demand and supply in real time.
Though only four homes are currently participating in the field trial, dozens of homes are slated to join in the near future.
Bernstein said this small demonstration is the first step toward scaling up the technology on larger projects or even with utility partners.
It’s Nice to Share
Another advantage to autonomous control of small-scale, on-site power generation is that it’s a step toward a long-desired wish of many renewables devotees: local power-sharing.
“Right now, neighbors can’t sell power to each other,” said Berkeley’s von Meier.
“So if the power is out, but I have solar on my roof and my neighbor doesn’t, I couldn’t just send some over to my neighbor.”
Recommended by Our Editors
For technical as well as regulatory reasons, she added, energy-sharing is problematic and even physically unsafe without the right equipment.
But it’s a necessary step in upgrading the grid and its resiliency.
Jorge Elizondo, a microgrid engineer and co-founder of Heila Technologies, said that with a controller in each location, energy-sharing becomes more feasible, as does the possibility for an entire neighborhood to serve as an aggregated reserve of power for the main grid: a virtual power plant.
“With a controller in every home, you can coordinate them—the idea is that every house can be an island if it needs to be,” Elizondo said.
“But when the grid is there, you can tie them all together.
And to the grid, they function like a large battery.”
High Standards
One problem, Bernstein noted, is the current lack of standards for how DERs connect to the grid.
So creating control software for them is particularly tricky.
“You can develop a very nice algorithm, but if you need to work hard to adapt it to every device, it’s very challenging,” he said.
“I think eventually there will be standards for DERs, similar to the internet—without them, it will be a mess.”
Standardization is the approach that Gridscape is taking, hoping that a “microgrid in a box” solution will be the key to integrating more of the larger-style municipal and commercial microgrids with the main grid.
The company has worked on 10 projects so far, including several California fire stations, hotels, and affordable housing units, and it expects another two dozen to come online in the next year.
While Gridscape’s kit includes a standard set of batteries, controllers, inverters, and other hardware, the software layer is the secret sauce for building resiliency.
But the company doesn’t think any truly “intelligent” systems will be market-ready anywhere in the short term.
“Anyone who says they have a sophisticated machine-learning system for this problem—well, we’re not quite there yet,” simply because there aren’t enough data points for true learning across the grid to happen, said Alok Singhania, who manages product development for Gridscape.
Its solution relies on statistical and analytical methods to determine optimization to and from microgrids, based on weather forecasts, near-real-time energy usage, market data, utility tariffs and historical data.
Regular automated updates from cloud-based software helps to keep the solutions up-to-date and dynamic, Singhania added.
In smart grids, smart controls at each power-producing node coordinate to provide more efficient energy use. Companies like Sunrun, which markets residential solar-plus-storage products, take a slightly different tack to bring utilities on board as partners: they provide analytics and power management from single-point renewable microgrids themselves.
“Utilities are used to managing large plants, not individual homes,” said Tefford Reed, senior director of advanced products for Sunrun.
He added that the company currently provides monitoring services based on 280,000 customers with their utility partners, and depending on the need from the utility, can also forecast demand or supply from neighborhood-scale virtual power plants.
Power Couples
Ultimately, von Meier said she sees the grid of the future necessarily evolving into a complex hybrid of old and new.
Local power generation will be just as important as the distribution of affordable renewables from far-flung solar and wind farms.
But it makes sense for each component of the grid to be able to decouple and to be independently functional.
And, she added, along with smart controls that can rapidly assess landscape-scale power problems, the human knowledge base of power engineering and distribution will continue to play an important role in integrating ever more renewables, including microgrids.
“Automation systems always run the risk of not having the right data, and being too deterministic in their approach,” she said.
“The data needs to be...
When the cable from the mainland that supplies power to Isle au Haut, Maine, was laid down on the seafloor to replace aging diesel generators, the community was told it had a 15- to 20-year lifespan.
That was in 1983.
For the last decade, confronted by the creeping urgency of replacing a power line that could fail any day despite careful maintenance, the community debated how to best keep the lights on in the future.
On the menu were a new cable, diesel generators, fuel cells, wave-energy generators, wind, solar, and batteries.
Cost was the primary concern.
With a full-time resident population of only 70 people or so and a summertime crowd of 200 to 300, Isle au Haut faced a steeper-than-usual capital infrastructure upgrade.
“When we went through the costs for almost any option, the economics were just really nasty,” said Jim Wilson, president of the private, for-profit cooperative Isle au Haut Electric Power Company.
But as the cost of renewable technologies plummeted, certain options began to make more sense.
A solar-and-battery system would run them around $1.8 million.
A new cable: double that.
A diesel system: triple.
So, four years ago, the co-op members voted unanimously to pursue a 300-kilowatt system made up of 900 solar panels, with a 1-megawatt graphene supercapacitor battery to store and supply excess power.
Residents can also add on their own solar panels along with battery systems (and/or water tanks to store excess energy as heat).
In most places, that might be the end of the story.
But there’s a hidden aspect to this solar project that tweaks it into next-wave territory: Behind each of the 140 or so electric meters on the island will be units loaded with machine-learning software developed by Kay Aiken at Maine-based Introspective Systems.
These intelligent control units learn the home’s energy-usage patterns, then coordinate that information with the island’s solar panels and batteries in real time.
This lets each point on the grid decide when it is best—cheapest, most efficient, or at a time preferable to the homeowner—to buy or sell electricity back to the grid, if they have their own panels and batteries.
“We didn’t mean to go to sea,” Wilson said, referring to an anecdote of a small ship that is blown to an unintended but not entirely unwelcome destination.
“We just wanted something to replace the cable.
But then you start looking at how to make it all more efficient, and one thing leads to another.
We’re building a solar project—just not the one we expected three or four years ago.”
Construction was slated to begin in spring 2020; because of the COVID-19 pandemic, that's been delayed, but Wilson still hopes the new system will be up and running by late summer or fall.
Once it’s completed, Isle au Haut will be part of a growing trend in renewable energy generation: autonomously controlled microgrids.
What Exactly Is a Microgrid?
Microgrids aren’t a new idea.
In industry parlance, a microgrid is a small network of electricity users with access to a local source of energy.
The users are all directly connected to the central grid, but during outages, the entire small network can disconnect itself from the central grid, or operate in “island mode,” to continue operation.
Deeply remote communities have relied on microgrids for decades, either because they’re at the end of a long and creaky power-transmission line or out of range of utilities completely.
Hospitals and other emergency and critical facilities also rely on locally produced energy for backup power; they’re often powered by diesel, propane, or another combustible fuel.
But microgrids can also be far more local: When you crank up your generator after a storm outage and run a few extension cables to neighbors, that is, in essence, a microgrid.
What is new is that more and more of these microgrids are being powered by renewable energy methods, thanks to increased affordability and shifting regulations that make them easier to adopt.
And it’s a booming business: Depending on your news source, the microgrid market is expected to be worth $47 billion by 2025, up from $28 billion this year.
In the United States, a massive amount of microgrid movement is happening in California, which has mandated that all of its electricity be from zero-emission sources by 2045.
Californians are also simply concerned that power is available at all: Between catastrophic wildfires, earthquakes, mudslides, and the “public safety power shutoffs” that the state’s largest utility, Pacific Gas & Electric, has recently implemented, people are turning to renewable-energy microgrids in an effort to free themselves of dependence on a large and increasingly erratic energy supply.
Of course, the movement to these microgrids is not just a California thing.
It’s happening in Puerto Rico, where solar-and-battery microgrids offered some respite after back-to-back hurricanes and earthquakes.
And New York state saw the launch of an $11 million prize fund to drive renewable microgrid development after being battered by Superstorm Sandy.
A community microgrid in East Hampton on Long Island supplies 50 percent of the town’s energy needs, keeps the local water plant running and fire stations operational in emergencies, and has allowed the local utility to avoid $300 million in transmission upgrades as a result.
Municipalities across the country are building them to keep critical facilities up and running no matter what, while using green, locally produced energy to power the facility in the meantime.
And microgrids offer energy resources and independence to rural communities and towns in India, Africa, and other developing areas of the world.
Even utilities are exploring ways to get into the game: In Illinois, Commonwealth Edison worked with the Chicago Housing Authority to build the Bronzeville solar-and-battery microgrid for 660 residential units in a low-income neighborhood connected to a nearby university microgrid.
It’s a pilot program to learn how clusters of microgrids can work together when disconnected from the main grid.
Part of the reason for this: Utilities are beginning to recognize they cannot completely guarantee energy supply to their customers, especially in vulnerable areas and at the edges of the grid, said Sascha von Meier, an expert in smart-grid technology and power distribution with University of California-Berkeley and the Lawrence Berkeley National Laboratory.
That’s leading to more of an appetite for investment in grid resilience—and renewable microgrids are playing a big part.
As more and more renewable microgrids come online—from college campuses and corporate headquarters to individual homes—the energy questions begin: How can these resources not only feed into a broader grid? And how can each point in an increasingly complex matrix effectively work together and even share directly with one another—a seemingly simple task the old grid just isn’t equipped to do?
Connecting Resources
In traditional energy-supply systems, control and optimization of power is coordinated among a relatively small number of centralized resources.
Control servers optimize the generation, output, and flow of energy from hundreds or thousands of power plants and send it over the lines accordingly.
But as more people in a given region purchase electric vehicles and embrace renewables, batteries, and smart appliances that vary their power consumption throughout the day, the number of points that are contributing to and drawing from the system begins to climb into the millions.
These “distributed electric resources” (DERs) become a big data problem that can’t be optimized or coordinated in a centralized way.
The solution? Distribute the computing resources, and back them up with nimble algorithms.
Andrey Bernstein, who researches autonomous grid control at the National Renewable Energy Laboratory (NREL) in Golden, Colorado, said that the computing part of this is less about how to handle millions of variables at once so much as it is about breaking them up into manageable chunks, then coordinating communication on nested levels—and making sure that as DERs interact with the grid, they don’t damage the system by accidentally overloading it.
Andrey Bernstein and a colleague (Photo: Dennis Schroeder/NREL) In the lab, NREL-developed algorithms were first tested using a computer-simulated electric feeder line with hundreds of thousands of DERs, then millions.
NREL then partnered with Massachusetts-based Heila Technologies to integrate the software into a small and inexpensive smart control unit that can be installed inside the home or property, behind the electric meter.
Bernstein is currently developing algorithms for optimizing energy distribution from a renewables-powered microgrid to and from the main power grid.
The software, which is being tested in Colorado, is designed to coordinate real-time demand and supply from high numbers of energy-generating and storage devices in homes on a microgrid—solar panels, electric vehicles, smart appliances—by performing the advanced calculations via a small, inexpensive computing controller at each point on the grid.
The first real-world pilot of this technology recently kicked off in the community of Basalt Vista, an affordable housing community in Colorado developed by Habitat for Humanity.
Each home is equipped with a full range of energy-saving appliances and solar panels, making Basalt Vista a carbon-neutral community.
Homes under construction in the Basalt Vista community.
(Photo: Joshua Bauer/NREL) Once the NREL/Heila smart controls are installed at each node within the Basalt Vista neighborhood microgrid, each point can become self-governing but can also interact locally with other nodes to optimize the energy flow to and from each home in the community.
They can determine when to draw power from the local grid (and other users on the grid), when energy from more distant resources is required, when to store energy in batteries or in plugged-in electric vehicles, and learn and anticipate patterns of fluctuations in demand and supply in real time.
Though only four homes are currently participating in the field trial, dozens of homes are slated to join in the near future.
Bernstein said this small demonstration is the first step toward scaling up the technology on larger projects or even with utility partners.
It’s Nice to Share
Another advantage to autonomous control of small-scale, on-site power generation is that it’s a step toward a long-desired wish of many renewables devotees: local power-sharing.
“Right now, neighbors can’t sell power to each other,” said Berkeley’s von Meier.
“So if the power is out, but I have solar on my roof and my neighbor doesn’t, I couldn’t just send some over to my neighbor.”
Recommended by Our Editors
For technical as well as regulatory reasons, she added, energy-sharing is problematic and even physically unsafe without the right equipment.
But it’s a necessary step in upgrading the grid and its resiliency.
Jorge Elizondo, a microgrid engineer and co-founder of Heila Technologies, said that with a controller in each location, energy-sharing becomes more feasible, as does the possibility for an entire neighborhood to serve as an aggregated reserve of power for the main grid: a virtual power plant.
“With a controller in every home, you can coordinate them—the idea is that every house can be an island if it needs to be,” Elizondo said.
“But when the grid is there, you can tie them all together.
And to the grid, they function like a large battery.”
High Standards
One problem, Bernstein noted, is the current lack of standards for how DERs connect to the grid.
So creating control software for them is particularly tricky.
“You can develop a very nice algorithm, but if you need to work hard to adapt it to every device, it’s very challenging,” he said.
“I think eventually there will be standards for DERs, similar to the internet—without them, it will be a mess.”
Standardization is the approach that Gridscape is taking, hoping that a “microgrid in a box” solution will be the key to integrating more of the larger-style municipal and commercial microgrids with the main grid.
The company has worked on 10 projects so far, including several California fire stations, hotels, and affordable housing units, and it expects another two dozen to come online in the next year.
While Gridscape’s kit includes a standard set of batteries, controllers, inverters, and other hardware, the software layer is the secret sauce for building resiliency.
But the company doesn’t think any truly “intelligent” systems will be market-ready anywhere in the short term.
“Anyone who says they have a sophisticated machine-learning system for this problem—well, we’re not quite there yet,” simply because there aren’t enough data points for true learning across the grid to happen, said Alok Singhania, who manages product development for Gridscape.
Its solution relies on statistical and analytical methods to determine optimization to and from microgrids, based on weather forecasts, near-real-time energy usage, market data, utility tariffs and historical data.
Regular automated updates from cloud-based software helps to keep the solutions up-to-date and dynamic, Singhania added.
In smart grids, smart controls at each power-producing node coordinate to provide more efficient energy use. Companies like Sunrun, which markets residential solar-plus-storage products, take a slightly different tack to bring utilities on board as partners: they provide analytics and power management from single-point renewable microgrids themselves.
“Utilities are used to managing large plants, not individual homes,” said Tefford Reed, senior director of advanced products for Sunrun.
He added that the company currently provides monitoring services based on 280,000 customers with their utility partners, and depending on the need from the utility, can also forecast demand or supply from neighborhood-scale virtual power plants.
Power Couples
Ultimately, von Meier said she sees the grid of the future necessarily evolving into a complex hybrid of old and new.
Local power generation will be just as important as the distribution of affordable renewables from far-flung solar and wind farms.
But it makes sense for each component of the grid to be able to decouple and to be independently functional.
And, she added, along with smart controls that can rapidly assess landscape-scale power problems, the human knowledge base of power engineering and distribution will continue to play an important role in integrating ever more renewables, including microgrids.
“Automation systems always run the risk of not having the right data, and being too deterministic in their approach,” she said.
“The data needs to be...
