Electricity makes a long journey from the power plant to your light bulb. This is possible thanks to a modern marvel: the energy grid, which ensures that power consistently flows to every home and business hooked up to it. But some grids, like the network spread across the United States, are old and inefficient, and modernizing them will take significant effort. Meanwhile, renewable energy has become so efficient that it now rivals fossil-fuel–derived processes. Technological innovations have invigorated the potential of energy storage, while data centers—once considered the grand energy drains of the 21st century—have plateaued their power consumption. In short, the power grid of tomorrow stands to benefit from technological advances made today—but it still requires a host of infrastructure improvements to graduate beyond the aging grid of yesteryear.
The power grid is the system that carries electricity from the power plant to your home. It’s broken down into three parts: generation, transmission, and distribution. Power plants generate energy in the form of electricity, which exits via thick transmission lines, typically strung up on huge metal frameworks far above street level. These are often marked “high voltage,” and with good reason: More electricity is lost when it travels at lower voltages, so for most of its journey to your home, it’s kept between 115 and 765 kilovolts. The electricity then runs through several intermediary substations, where transformers step down the power for domestic use. Close to its destination, the electricity is routed through low-voltage overhead or underground lines before arriving at your home or business, where it emerges from outlets at 120 or 240 volts (in the U.S., anyway—domestic voltage varies from 100 to 240 volts, depending on the country).
Energy is inevitably lost over the course of this journey. In fact, 4.7 percent of electricity generated in the U.S. was lost during transmission and distribution in 2015, according to estimates by the Department of Energy’s Energy Information Administration (EIA). But a larger concern is the existing infrastructure—the power lines, stations, and monitoring technology, which is many decades old. A 2016 report from the American Society of Civil Engineers estimated the shortfall in investment for the upkeep of the U.S. grid to be $177 billion through 2025, and $565 billion through 2040.
The U.S. government has been proposing modernization plans for years, but early motions, like the DOE’s 2007 pledge to allocate $51.8 million toward improving the electricity grid, seem overly modest in light of more recent predictions of what these improvements will cost. In 2011, industry experts estimated that the government’s modernization plan would cost up to $476 billion over the next 20 years (to be paid for in part by the power companies, who would pass much of the cost on to consumers), but would provide up to $2 trillion in customer benefits. That investment would ideally create a modern grid, which the Department of Energy’s 2015 Quadrennial Technology Review describes as follows:
“A modern grid must be more flexible, robust, and agile. It must have the ability to dynamically optimize grid operations and resources, rapidly detect and mitigate disturbances, integrate diverse generation sources (on both the supply and demand sides), integrate demand response and energy-efficiency resources, enable consumers to manage their electricity use and participate in markets, and provide strong protection against physical and cyber risks.”
In the first quarter of 2017, 37 U.S. states pursued or enacted grid modernization policies. Most of these are still in the proposal stage, and not everyone is on board, but the U.S. is, indeed, moving toward a modern grid. Here’s what it will look like.
Tomorrow’s grid: smarter and more responsive
Regardless of the budgetary and other challenges it faces, the U.S. energy sector is making progress toward modernizing the power grid. One of the most promising developments is also one of the simplest: demand response. Currently, power plants produce energy around the clock, raising or lowering the amount of electricity they’re pumping into the grid according to estimated demand. But this process is very inexact. By and large, we don’t know exactly how much energy is used, where it’s used, and at what times. But investing in demand response would give power plants—and overseers of the power grid in general—information they currently lack about how homes and businesses use electricity. Demand response turns the one-way power grid into a two-way network, with consumers feeding information back to producers. But this will require significant investment to install metering technology in every home, not just to supply energy producers with up-to-the-moment data, but to share this information with consumers as well.
Read more in issue 4
Energy & Environment
This issue focuses on the impact and intersections of energy, environment, and our daily technologies—from smartphones to rare earth mining, the future of the power grid to what it takes to go carbon neutral.
Part of demand response entails motivating citizens to be conscious of their electricity use, especially compared to the energy demand in their area, and incentivizing them to make energy choices accordingly. According to the EIA, by the end of 2016, nearly half of American electric utility customers had smart meters installed. But many consumers—29 percent, according to a 2015 EIA survey of homeowners—aren’t even aware that their residences are equipped with smart meters. A paltry 8 percent knew they had access to hourly or daily data from their meter, and only 4 percent had accessed or viewed that data.
Smart meters are one element in the so-called “smart grid,” an umbrella term for the myriad improvements that could bring the power grid into the 21st century. More of a catchall term than a concrete list, it nevertheless covers applications of commercial technology to utility infrastructure. Some of these improvements include automated feeder switches, fiber optic and wireless networks, storage, and other new hardware, according to the DOE’s 2015 Quadrennial Technology Review.
Another, larger refinement will help consumers feed nominal energy back into the grid from their own generators. Most of the time, this means using domestically installed solar panels. But it will require a more complex grid that enables two-way electricity transfer. On a larger scale, this includes integrating microgrids, self-sufficient miniature power networks that exist alongside the main grid.
Tomorrow’s grid also has to factor in the growing use of electric vehicles. Their increased adoption will likely result in greater electricity consumption as consumers swap gas stations for charging ports. Widespread electric vehicle use will require a similarly large rollout of charging stations, which the power grid—and its infrastructure—will have to accommodate.
Data centers: stabilized, for now
Demand response turns the one-way power grid into a two-way network, with consumers feeding information back to producers.
Data centers, facilities dedicated to housing and running clusters of data servers, are essential to the internet age. They erupted across the commercial space during the dotcom era, representing a new drain on the power grid: Unlike workforce computers, data servers can’t shut down at the end of the day; they must run constantly to allow anyone to access their data at any time. As more and more data centers were built in the early 2000s, analysts worried about the exponential pace of constructing new data centers—and the energy required to run them. A 2007 EPA report to Congress noted that data centers had doubled their energy consumption since 2000 to 61 billion kilowatt hours (kWh), and projected that the figure would rise to 100 billion kWh by 2011—and would only continue to climb.
That didn’t happen. Instead, data centers became far more efficient, until, in 2016, their consumption largely plateaued at an estimated 73 billion kWh, according to a DOE report released that year.
“That sort of plateau has to do with everything being so grossly inefficient before. There’s all this low-hanging fruit to start making improvements in terms of how data centers are operated and how efficient their heating systems are—how they’re managed, essentially,” said Dr. Arman Shehabi, one of the report’s authors and a research scientist in the Energy Analysis and Environmental Impacts Division of the Energy Technologies Area at the Lawrence Berkeley National Laboratory (LBNL).
The report, which was put together by a coalition of researchers from the LBNL, Stanford University, Carnegie Mellon University, and other institutions, outlined the efficiencies and industry trends in data center energy usage. First, they reevaluated installation rates: Yes, the number of data centers had nearly doubled from 2000 to 2005, but the annual rate of increase fell over the next five years (in part due to the late-aughts recession), leveling off at an estimated 3 percent increase per year from 2016 until around 2020. Furthermore, nearly all of the servers bought since 2010 were destined for so-called “hyperscale” data centers, which are set up and operated for maximum productivity and server utilization.
Smaller data centers—servers stashed in closets or rooms in office buildings under 5,000 square feet—barely apply these efficiency strategies.
Improvements in server utilization have also cut down on energy consumption. Servers were refined to use less power while idle and match the amount of energy consumed to the level of utilization. Centers became more efficient, consolidating the number of servers and eliminating those that that sat idle for too long. But another development dramatically affected server use: virtualization, which located multiple digital “servers” within one physical server to increase its utilization. The industry also refined data center infrastructure, rearranging server layouts to control waste heat and developing new methods to cool the racks of servers, which cut down on the amount of power consumed. While these improvements didn’t lead to a net reduction in energy drain across U.S. data centers, it did flatten their expected yearly increase in energy consumption to the aforementioned 73 billion kWh plateau.
But many of these efficiencies have only been applied to huge data centers run by tech giants like Google and Microsoft, or within companies running server farms to rent out to clients or to perform cloud processing. Smaller data centers—servers stashed in closets or rooms in office buildings under 5,000 square feet—barely apply these efficiency strategies. That’s how small and medium-sized data centers end up consuming 49 percent of the electricity used in U.S. data centers each year, despite owning just 40 percent of the total number of servers, according to a 2014 report by the nonprofit Natural Resources Defence Council (NRDC).
And those efficiency strategies aren’t expected to permanently curb data center power use. They were just the low-hanging fruit, Dr. Shehabi said.
“This shift from less efficient data centers to more efficient data centers has offset the growth—which has been large—in the amount of data center services. It’s not clear if that will continue, [or] for how long that will continue,” said Dr. Shehabi. “The efficiency measures that [are really] driving the overall efficiency improvements are limited, so other measures will have to come out. There’s only so much that can be achieved [with the efficiency measures] that are being done now.”
Companies are spending more on data centers every year: By September 2017, U.S. companies had invested $18.2 billion in data centers that year, which was already double the amount spent in all of 2016, according to a CBRE report. It’s difficult to predict how data centers’ energy consumption will change beyond the next couple of years, Dr. Shehabi said: Just as analysts couldn’t predict the tempering of data center power consumption, it’s possible that new solutions might be uncovered along the way.
But carefully choosing where to locate data centers is a strategy that will continue to keep power consumption down. Since they’re basically just data havens hooked up to the internet, very few need to be adjacent to city centers or located in close geographic proximity to users. Aside from strong online connectivity, data center site prerequisites are limited but specific: Since their servers must always stay on, for instance, it’s risky to put them in areas prone to severe weather and power outages.
But those always-on servers get hot, and considering climate when siting new data centers can mitigate cooling costs. Building a data center in a colder place lets the center draw in cooler outside air, requiring less power to run air conditioners and keep the whole facility cool. In 2011, Facebook announced it would build one in Luleå, a city in northern Sweden, while one of Google’s European data centers exists in Hamina, Finland; several other companies have constructed data centers near the Arctic Circle to save on cooling costs. But that advantage has to be weighed against other expenditures, like the higher electricity prices in remote markets.
Even the most shrewdly planned data centers require a lot of electricity, and many of the biggest tech companies have tried to mitigate the public impact of their operations’ power consumption.
All of this boils down to the simple fact that designing data centers and choosing where to build them involves a lot of factors. Even the most shrewdly planned data centers require a lot of electricity, and many of the biggest tech companies have tried to mitigate the public impact of their operations’ power consumption. Aside from outfitting their campuses with solar panels, a feat more symbolic than effectual in offsetting their energy use, many have taken a more straightforward approach by purchasing renewable energy produced elsewhere: Microsoft, for example, bought its next 15 years of energy from a wind farm in Ireland. And in 2017, Google announced that it expected to reach a major milestone: powering its data centers with 100 percent renewable energy. (Apple claims to have reached that milestone already.) Many big tech companies see the logic—and the good PR—in investing in green energy, whether they’re directly powering their facilities with renewable energy, purchasing energy credits, or a combination thereof. Which brings us to our next point: the effect of renewables on tomorrow’s energy grid.
The renewable revolution
More and more states are beginning to invest in renewable energy. However, progress is slow: Renewables provided only 10 percent of the total energy generated in the U.S. in 2016, according to an EIA report. Fossil fuels remained dominant, providing 70 percent, while nuclear power provided 20 percent.
The Obama administration’s 2015 Clean Power Plan sought to combat climate change by reducing carbon dioxide emissions from electrical power generation nationwide by 32 percent by 2030. To meet the lower emissions limit, states would have likely invested more in renewable energy. But that plan is currently on its way to the scrap heap: The Trump-appointed head of the Environmental Protection Agency, Scott Pruitt, announced in October that the agency plans to kill the plan entirely. It’s unclear when this repeal will happen, or if legal challenges will drag it out.
Some states reaffirmed their commitments to meeting the lowered carbon emissions threshold that the Clean Power Plan had outlined, and overall, the EIA reported that the U.S. produced more wind and solar energy in 2017 than ever before. But renewables pose particular challenges to the grid. While fossil-fuel–based power plants can be positioned near the areas that use the energy they produce, appropriate sites for renewable energy sources are more limited. Hydroelectric plants, for instance, need to be located by bodies of water; wind farms require favorable air currents; and solar facilities benefit from frequently clear skies. Often, these plants are awkwardly located, and are farther away from their destination grids.
The infrastructure of the power grid may not change significantly with the shift toward renewable energy, but Americans will likely have to adapt their energy-producing strategy. Coal and natural gas power plants consume fossil fuels to generate electricity, and can do so at any time. But renewables often farm energy from their environments along set schedules: Solar plants can only collect sunlight during the day, wind turbines only spin when air currents push their blades, and so on. This makes it harder to smartly expand demand response.
Even now, concurrent innovations could mitigate the challenges of renewable energy. Energy storage is still in its developmental stages, but the concept is sound: Gather power when plants are generating electricity, and store it away for when the plants aren’t producing. Tesla has promoted its own energy storage system, the domestic Powerwall and the larger Powerpack, which is aimed at commercial and utility-scale storage, but those are only in their second generation. Puerto Rico could become the first territory to pioneer a combination of solar power and energy storage after Tesla provided aid to the hurricane-devastated island, which led to talks of rebuilding the island’s power grid using Tesla energy technology.
Climate change poses unique challenges to the energy grid. Global warming will have obvious consequences, like increased power usage for air conditioning and location cooling. Less apparent effects include a slight increase in power loss, as warmer transmission lines shed more energy.
Renewable power plants in particular face two serious obstacles. Coastal plants, some of which are just a few feet above sea level, may have to relocate if sea levels rise and permanently flood ocean-adjacent facilities. And a changing climate might shift weather patterns, directing wind currents away from existing wind farms.
“It will affect how you’re siting both your wind and solar [plants]. Climate change could change where the optimum places to put those are. There’s a lot of work going into trying to understand how increases in overall temperatures are going to affect things, especially wind patterns,” Dr. Shehabi said. And then there’s the obvious concern of an overall warmer world: “Where you have increases in overall temperatures, generally that increases your demand for electricity throughout all buildings.”
Assuming climate change continues apace, leading to a projected 4.5-degree Fahrenheit increase in average global temperatures by the end of the century, the Obama-era EPA estimated that consumers could spend 10 percent more annually on heating and cooling—and not just cooling for humans, either. Remember those data centers full of hot servers? Warmer global temperatures means more electricity used to keep those rooms cool. Non-renewable power production facilities that rely on water cooling, like fossil fuel and nuclear plants, might be less efficient.
Should weather patterns shift as the climate warms, extreme weather events could also become more frequent. An MIT study released in November argues that global warming made at least one of the hurricanes that devastated the U.S. Gulf Coast in 2017—Hurricane Harvey, which ravaged Houston in August—more likely to occur. The study suggests that storms of that magnitude are now six times more likely to occur than they were at the turn of the 21st century.
As the Obama-era EPA’s website noted, many aspects of the energy grid’s infrastructure are threatened by weather events that may occur more often due to climate change. Large storms that make landfall in populated areas can damage existing power lines and electricity distribution equipment. Even fossil fuel transportation infrastructure could be affected as rainfall washes out railways or reduces the usability of rivers.
What comes next?
Less apparent effects of global warming include a slight increase in power loss, as warmer transmission lines shed more energy.
Ultimately, the U.S. grid will need to change to address each of these pressures, although the responses will be behavioral and bureaucratic as much as they are technological and infrastructural. As more renewable energy enters the grid, for instance, it will have to adapt to renewables’ more erratic means of generating power.
“We have roughly three dozen ‘balancing authorities’ in the western U.S. who have a hard time exchanging power at short notice. But to accommodate more renewables, we have to be able to shift power quickly across a region to take advantage of surpluses that occur in some places when there are needs elsewhere,” said Benjamin F. Hobbs, chair of environmental management at Johns Hopkins University.
Energy storage could help mitigate this shift, too, although consumers will probably start buying power storage solutions for the same reason they’re adopting electric cars, Hobbs said: for the “cool factor,” not because they’d save money. Policy will also play a role in the growth of energy storage, but financial incentives will have to be weighed against the things consumers can already do to accommodate renewable energy, like changing the timing of their energy use. As it stands, consumer behavior doesn’t really factor into the energy market.
“[Consumers] have remarkable potential to change when and how much they consume, but this is untapped. [But] responsive demand can help compensate for fluctuations in wind and solar output,” Hobbs said. “The day will come when consumers will be rewarded for being flexible, and technologies like Google’s Nest will be cheap and convenient ways to do so. This will be a big change for the grid, and it is coming.”
That change will benefit consumers financially, too. Should enough residences install smart meters, consumers might be more motivated to take action, becoming active participants in the energy market.
“We need to provide more financial reward to the complementary resources that can make up for fluctuations in solar and wind output. This means prices that vary minute to minute rather than hour to hour, so [that] a generator or consumer who can quickly adjust can participate when prices are favorable and go away when they are not,” said Hobbs. “This also means that there have to be communications and actuators that provide the signals on a timely basis and allow market parties to react quickly (based, for example, on set price points).”
The challenges faced by the U.S. power grid have separate but potentially interdependent solutions. Energy storage, combined with ever-cheaper solar power, could be integrated into Puerto Rico’s next grid, for example, which could serve as a model for how to adopt these technologies at a local, regional, or even national scale. In the meantime, empowering consumers to participate in demand response—finally inviting them into the energy market—could alleviate overall demand on the grid. This could help mitigate the shortcomings of renewable energy sources while they are first adopted at a larger scale. Of course, just as the grid will advance in the coming years, so, too, will the impact of climate change.