Chapter 8 Is Bigger Better?
The only thing more surprising than the partnership between a Japanese tycoon and a Chinese technocrat was the proposal they had joined forces to pitch. The “Global Energy Interconnection,” as they called it, would cost $50 trillion, link every continent with undersea transmission lines, and power the globe with clean energy. Windmills on the North Pole and solar panels in the Sahara Desert would export their power to run the world’s megacities. And oh, by the way, this supergrid would also secure world peace and harmony.
At least one half of the odd couple had made a habit of proposing audacious goals. For Masayoshi Son—the second-richest man in Japan, where as something of a celebrity, he is often called just “Masa”—the supergrid is his latest superlative. Masa has also set his sights on turning Softbank, the conglomerate he leads, into a business empire that will last 300 years, and in 2017, he launched a mammoth, $100 billion fund to invest in the tech sector. He has predicted that by midcentury, computers will surpass human intelligence, achieving an IQ of 10,000.1 His bet on a supergrid fits right in with his other self-described “crazy ideas.”2
Following the 2011 Fukushima disaster, Masa set out to harness renewable energy as an alternative to nuclear energy to power Japan. “I was a total layman in renewable energy at the time of the earthquake,” he admits. Perhaps that inexperience explains a proposal that most experts would dismiss out of hand. But Masa’s initial vision of Japan importing solar and wind power from the Gobi Desert resonated with utility executives in Russia, South Korea, and China. In 2016, the three utilities and Softbank signed a memorandum of understanding to collaborate on a pan-Asian supergrid.3
It was Masa’s Chinese partner, Liu Zhenya—chairman of State Grid Corporation, the national utility and second-biggest company in the world—who masterminded the plan to ultimately go global once an Asian supergrid materialized. Unlike Masa, Liu was not known for articulating bold visions. In fact, he wasn’t known for saying much of anything at all. An engineer by training, Liu had worked in the state power administration his whole career, quietly ascending to the top of State Grid’s ranks.
But his work spoke for itself. Under Liu’s tenure, China constructed a network of ultra-high-voltage-direct-current (UHVDC) transmission lines, a technology that has recently fallen in cost and can transmit power over long distances without excessive losses. China currently has seven of these gargantuan projects, which ship renewable energy from the remote western hinterlands of the country to the metropolitan east coast. It also has twice as many projects on the drawing board. By contrast, the United States is still trying to build its first UHVDC line, a $2.5 billion project to send wind power from Oklahoma to Tennessee.4
Capitalizing on its success at home, State Grid is now offering its expertise abroad. Brazil has hired it to build the longest transmission line in the world, measuring more than 1,500 miles, to deliver hydropower from the Amazon to Rio de Janeiro.5 These efforts represent the first of three stages that Liu envisions to achieve a global supergrid. First, countries will build their own transmission grids—China plans to finish this stage by 2025. Next, countries will link up regionally, in line both with Masa’s early vision of an Asian supergrid and other ambitious proposals, such as a grid to link sunny North Africa with windy Europe. Then, the grand third stage will bring together the regional supergrids to form a single global supergrid, linked by transoceanic undersea cables and electrical superstations where lines would converge (figure 8.1).
A global supergrid. This map combines various proposed regional supergrids (for example, in the Asia-Pacific region, Europe and the Middle East/North Africa, and North America). Black lines represent long-distance HVDC electricity transmission lines.
Source: Reprinted with permission from Gellings (2015).
Now Liu is backing up his vision with action. In 2016, when he reached China’s mandatory retirement age and stepped down as boss of State Grid, he immediately took up the chairmanship of an organization devoted to realizing his Global Energy Interconnection dream (Masa is a vice chairman). He has also written a dense textbook, Global Energy Interconnection, that provides a technical, step-by-step road map, including the specific technologies that will require further research and development (R&D) investment to become commercially viable.6 He sums up by arguing that a global supergrid could “secure a safe, clean, efficient, and sustainable energy supply. With UHV [ultra-high-voltage] grids as its backbone, such a global energy interconnection transmits primarily clean energy.”
Given China’s prowess at building large infrastructure—from bullet trains to planned cities—the supergrid concept is understandably attractive and logical from Liu’s perspective. Indeed, a supergrid is one way to solve the serious problem of renewable energy’s intermittency, which existing, balkanized grids around the world are not equipped to handle well. Although the sun will not always shine over any given location, it will always be out somewhere in the world. A big grid can connect remote renewable resources with urban centers demanding energy. In general, the bigger the grid, the easier it is to match up supply and demand.
Yet a supergrid would create its own host of challenges, which Liu’s tome addresses inadequately despite devoting more than 300 pages to engineering details. Liu appears to find his creative muse in the last chapter, entitled “Global Energy Interconnection Changes the World,” where the previously staid textbook starts to veer off the rails with increasingly wishful subsections entitled “Creating a New Energy Scenario,” “Infusing New Vigor into Economic Growth,” “Creating a Wonderful New Social Life,” and “Turning a New Chapter of Civilization.”
In a similar vein, in a speech at the United Nations, Liu predicted that “the world will turn into a peaceful and harmonious global village with sufficient energy, green lands, and blue sky.”7 But Liu’s predictions about international relations are at odds with how the world actually works. First, getting countries to pony up $50 trillion, as the plan requires, will be a tall order (“politically impossible,” as Masa’s advisors have warned him). Also, many countries might worry about importing power from abroad. Peter Littlewood, professor of physics at the University of Chicago, is skeptical of Liu’s grand vision, invoking Europe’s tumultuous experience depending on Russia for natural gas. In one of his characteristically British understatements, he has remarked, “My sense is that countries will be concerned about being reliant on partners that they may not always be friends with.”8
Despite the political challenges, the practicalities of building a supergrid are worth considering, as are other out-of-the-box ideas for accommodating growing amounts of clean but intermittent sources into energy systems. In any case, the supergrid concept underscores the massive scale that might be needed to enable a global electricity system to rely heavily on fluctuating power sources like solar and wind. Even if technological innovation were to deliver endless rolls of almost-free, ultra-efficient solar photovoltaic (PV) coatings, today’s power grids would still struggle to integrate the resulting gobs of unreliable electricity. Enabling continued expansion of solar PV will also require arresting its value deflation, so its electricity is worth more than the costs of generating it. To meet such demands, countries will have to rethink the entire architecture of their power grids, and do it in time for the world to reach the target of generating a third of its electricity from solar power by midcentury (and, overall, most of its electricity from zero-carbon sources).
Many believe that the key to accommodating solar’s rise and addressing intermittency will be storing excess energy in batteries. But batteries remain expensive—at least for now. Grid expansion is an elegant alternative. Batteries would still have a place but would not be the only, or even the primary, solution (See chapter 9 for more details). If you build a big enough grid, the need to deploy massive amounts of expensive storage falls because the vagaries of weather and volatility of demand average out over large areas. Recognizing the advantage this feature offers, proposals for regional supergrids all over the world, notably in Asia, Europe, and North America, are gaining steam.
The quest to build ever-larger grids is not the only attempt under way to transform the power sector, though. In fact, the exact opposite idea has attracted excitement, particularly from New York, California, and some other U.S. states. Large, interconnected grids are vulnerable to massive failures, as Superstorm Sandy reminded people living in New York and New Jersey in 2012. But decentralized grids that rely more heavily on locally sited power resources—such as rooftop solar panels fuel cells, and batteries—could be more resilient in the face of disasters, natural or otherwise.
Moreover, a decentralized grid could, in theory, be cheaper than a centralized one. Scaling back investments in the sprawling infrastructure of the traditional grid, from power lines to substations, could slash customer bills. And those savings might not disappear if the cost gap between cheap power from large plants and traditionally expensive power from small-scale resources continues to narrow. Finally, if decentralized economic markets, where customers could trade energy services with their neighbors, took off, a decentralized grid could operate much more efficiently than today’s overbuilt, clumsy, and centralized power system.
There are reasons to think that decentralized grids would have more difficulty incorporating solar power than larger grids would, yet counterarguments can also be made. A decentralized grid would rely more heavily on local power sources, such as rooftop solar, which is more expensive than utility-scale solar. And a decentralized grid would be hard-pressed to balance power supply and demand on the scale of individual communities, whereas a centralized grid can more easily average out supply-and-demand volatility over large areas. On the other hand, decentralized economic markets might increase the value of small-scale solar by compensating solar owners for helping to stabilize the grid. This increase in appeal could speed the deployment of distributed solar power installations, especially those of an intermediate size of a few megawatts that are both cost-effective and valuable to the grid. If a decentralized grid were also “smart”—that is, if the grid and the equipment that it powers could talk to each other—then the grid could adjust to customer demand on the fly. It could match demand with the fluctuating supply from intermittent solar power, whether from utility-scale solar plants or distributed installations.
Supergrids and decentralized grids may fall on opposite ends of the size spectrum, but both would require systemic innovation, or novel approaches to fitting all the puzzle pieces of an energy system together. Systemic innovation can involve the use of new technologies; indeed, both new paradigms of the grid will require improved technologies to advance. But systemic innovation is distinct from technological innovation because it also encompasses getting fancy new widgets to work together seamlessly. For example, supergrids will require novel network topologies—that is, designs for laying out new power lines and tying them into existing grids. A decentralized smart grid would add two-way communication between utilities and customers and employ software algorithms to run the grid much more intelligently than today.
Which way is the world headed? And is bigger always better? There are no clear answers yet. The most promising way forward is probably to pursue a hybrid strategy of expanding the grid and localizing it at the same time, all the while making it much smarter than it is today. Such a hybrid grid might rely on a backbone of long-distance transmission lines to link faraway regions and deliver renewable energy from the most sun-drenched, windswept regions. Supplementing these clean energy highways would be decentralized microgrids serving, for example, neighborhoods, military bases, or schools, all networked together into a smart grid. Retiring much of the expensive, overbuilt infrastructure of today’s grid could pay for both supergrids and asset-light microgrids. This hybrid grid model could weather the volatility of renewable energy both through the wide reach of its long-distance transmission backbone and the precise responsiveness of the decentralized smart grid.
Although the supergrid and decentralized grid models are each radical departures in opposite directions from today’s power system, they might actually coexist in a truly advanced hybrid grid. Perhaps such a future is no more improbable than the partnership between Masa and Liu—two men with little in common except dreams of a future powered by clean energy.
Let’s Get Together
A 2015 headline in The Guardian announced, “Wind Power Generates 140 Percent of Denmark’s Electricity Demand.”9 Technically, that was true—it was so windy one summer day that Denmark’s ubiquitous wind farms generated enough power to meet the entire country’s demand, with surplus to sell to its neighbors. The bit about the neighbors might sound like an afterthought. But those neighbors are crucial to Denmark’s ability to supply over a third of its annual electricity from highly variable wind energy.
Denmark’s grid is part of the larger Nordic Synchronized Area, a grid that also encompasses Norway, Sweden, and Germany and delivers nearly thirty times as much power as Denmark alone uses. To that larger grid, Danish wind power is a drop in the bucket. A bucket is indeed an apt metaphor because the Nordic grid features vast hydro reservoirs in Sweden and Norway, which are adept at storing volatile renewable energy by pumping water uphill; the power can be recovered on demand by letting water flow back downhill to run a hydroelectric turbine. Across the whole Nordic grid, wind accounted for less than 15 percent of the electricity produced in 2015, despite the dramatic Guardian headline.10
Denmark’s situation is an argument for designing large, cross-national grids. Such grids can be more efficient than isolated ones because they can draw in energy from whatever sources are cheapest and most available at the time, regardless of whether those sources are nearby. Larger grids can also make much more use of clean but intermittent sources, which reduces waste and increases incentives for expanding the penetration of renewable energy into the electricity sector. The question for planners of such grids is how best to link up the disparate parts.
In the case of the Nordic Grid, its existence enabled much more wind power to be deployed than would have been possible if the constituent countries were disconnected from one another. Indeed, one of the major reasons that Germany is a global leader in renewables, which supplied a third of its power in 2016, is its ability to export surplus power to its neighbors. And to use even more renewable energy, Germany is building new transmission lines to Sweden and Norway to access their extensive hydroelectric storage facilities.
Interconnecting their grids makes it easier for countries to increase their share of volatile renewable energy while still matching up power supply and demand. On the supply side, the territory covered by a larger grid can have more diverse weather patterns than an individual country would. The shared grid makes it possible to exploit these patterns to keep the power supply to customers steady. For example, a large enough grid will rarely experience overcast skies over its entire geographical span, so at least some solar panels will continue to produce power as production from others drops. Similarly, geographically diverse wind turbines will produce less volatile power than if they are regionally concentrated. The wind and the sun are also anticorrelated in many regions. That is, at night when the sun sets, the wind blows harder, so a large grid spanning windy and sunny areas can achieve relatively consistent levels of renewable output. It is true that there are limits to what grid expansion can achieve. Because the sun moves from east to west, grid expansion from north to south does not mitigate power fluctuations from the daily solar cycle. Long-distance east–west transmission lines, however, can help with this daily volatility. Grid expansion also will not help when a solar eclipse strikes—fortunately, those large-scale events are rare and predictable.11
On the demand side, the larger the grid, the easier it is to match up renewable supply with demand centers. Often, the best renewable energy resources are far from major urban areas—think about the sun-soaked Sahara Desert or the windswept Great Plains—but a large grid can link them. In addition, connecting different demand centers has a similar smoothing effect on overall demand as linking geographically diverse sources of various kinds of renewable energy supply. When some area needs a lot of energy, it can obtain it from locales that require less energy at the moment.12
The technology to link faraway regions with transmission lines has actually been around since the first half of the twentieth century. As early as the 1920s, power systems engineers realized that high-voltage direct current (HVDC) power lines were better suited than high-voltage alternating current (HVAC) counterparts for long-distance transmission. This is counterintuitive, given that chapter 5 explained that in the early twentieth century, AC became the global standard of choice for power grids because it was easy to transform the voltage of AC power (and nearly impossible to do so for DC power). At high voltages, less electricity is lost in transmission. AC power grids sprang up to serve burgeoning electricity demand because it made more sense to put power plants relatively far from where the power was consumed, rather than rely on a DC system, which would require colocated generation and consumption. Thomas Edison, who favored using DC, was on the losing side of this argument, even though he publicly electrocuted farm animals with AC to try and turn public opinion against HVAC. Nikola Tesla sealed AC’s victory by powering New York City with an AC transmission line from Niagara Falls.
For transmission over really long distances (i.e., hundreds of miles or more), however, even HVAC just isn’t up to the task. At the high voltages required for those distances (several hundred thousand to over a million volts), HVDC lines are considerably cheaper and lose less energy than their HVAC counterparts. In fact, DC power transmission is always more efficient (i.e., loses less energy) than AC power transmission, and DC power lines require less material than AC power lines, making them cheaper. But the prohibitive cost of converters to increase and decrease the voltage of DC electricity at the beginning and end of a transmission line has traditionally ruled out DC for most grid uses. By comparison, AC voltage transformers are relatively inexpensive.
Still, at the highest voltages and greatest distances, the savings from more efficient and cheaper HVDC power lines more than compensate for the expensive DC converters (figure 8.2). What’s more, HVDC lines require less land when built above ground, and they can even be put underground, whereas long-distance AC lines cannot be put underground because of extremely high energy losses.
Schematic cost comparison of AC versus DC power lines. This schematic graph plots the total cost of AC and DC power lines as a function of their length. AC lines are cheaper over shorter distances because AC terminals, which transform the voltage to higher or lower levels, are cheaper than DC terminals, which convert the voltage to higher or lower levels. But over longer distances, AC lines quickly become more expensive because the cost of the power lines overwhelms the cost of the terminal. This is because long-distance AC lines have high losses and require substantially more material than equivalent-voltage DC lines. Therefore, beyond the breakeven distance, typically several hundred miles, high-voltage DC lines are the cheapest option, even taking into account the expensive terminals.
Recognizing that the voltage converters were the expensive bottleneck raising the cost of otherwise cheaper and superior HVDC technology, engineers in Europe and the United States worked throughout the twentieth century to develop less expensive converters. The first commercial HVDC project connected an island off Sweden with the mainland in 1954. In the 1970s, a better converter technology set off a flurry of new lines in Europe, Asia, and North America.13 (One of these is the Pacific DC intertie, a nearly 1,000-mile HVDC line built in 1970 to connect the Pacific Northwest with Los Angeles. In 2013, when I was an advisor to Mayor Antonio Villaraigosa of Los Angeles, I recommended that the municipal utility invest over $100 million to upgrade the Sylmar converter station serving Los Angeles; those improvements are now ongoing).14
HVDC lines built in the twentieth century were limited to connecting just two faraway points. More recently, however, a new converter technology known as a “voltage source converter” has made it possible to tap into an HVDC line at many points along its journey. This capability raises the prospect of an interconnected, meshed grid of HVDC lines rather than single, one-off projects. Such a grid could allow many more sources of supply and demand to be connected while minimizing transmission losses over long distances. What’s more, voltage source converters would enable this HVDC grid to link to existing AC grids, layering the former right on top of the latter.
For HVDC grids to be widely affordable and effective, more technologies are needed. These include DC circuit breakers that can shut down massive current flows within milliseconds to ensure grid reliability. Recent advances in materials known as “wide-bandgap semiconductors” might enable low-cost, high-performance circuit breakers, voltage converters, and other power electronics for HVDC grids.15 Further on the horizon, replacing today’s copper or aluminum power lines with superconducting cables could boost the power that an HVDC line could carry while plunging the energy losses to virtually zero. Such superconducting lines would need to be cooled to nearly ‒200°C, but the efficiency gains could outweigh the cooling costs.16
Although HVDC technologies are still under development, China is forging ahead, planning to develop everything that it needs along the way. Having already constructed the largest HVDC network in the world, it may spend $4 trillion through 2040 reducing emissions from its power sector and making an even bigger network of UHVDC (which is just HVDC at especially high voltages—around 1 million volts or higher) lines.17 Some of this ambitious buildout might be excessive. Indeed, China has been known to build “white elephant” infrastructure projects, such as roads and railways to sparsely populated areas.18 One study suggests that bolstering the good old AC grid could utilize surplus renewable energy better than building fancy, new long-distance UHVDC lines, which might backfire by providing distant markets for inefficient, inland coal plants.19 Most experts agree, however, that China’s transmission expansion will help reduce the amount of renewable energy that is thrown away because of being generated in remote areas cut off from demand centers.
Building on China’s progress, Liu and Masa envision interconnecting China with other Asian countries to create a regional supergrid. Though expensive, the prospective benefits—such as the ability to send remote, cheap renewable energy to faraway cities—could outweigh the costs. Another advantage of interconnecting grids across national borders is that each country would no longer have to maintain backup generation capacity independently to ensure power reliability. Instead, the countries could pool their reserves, eliminate redundancies, and reduce the total capacity needed. Studies simulating different configurations have found that a Northeast Asian supergrid connecting China, Japan, South Korea, Mongolia, and Eastern Russia could be cost-effective, despite the high cost of building new transmission lines.20 But linking Australia, via underwater cables and overland lines through Southeast Asia, might be a bridge too far for now.21
Another promising regional supergrid proposal would connect Europe with North Africa. This idea, however, has already faced obstacles. In 2009, a consortium of European investors founded DESERTEC, a scheme to supply Europe with solar energy from the sun-drenched deserts of North Africa. But by 2013, the investors backed out and the program collapsed. Still, in 2016, a CSP project in Ouarzazate, Morocco, came online (travel tip: this plant is less than an hour’s drive from the popular tourist village of Ait-Benhaddou and is worth a visit just to gawk at the parabolic mirrors arrayed as far as the eye can see). The project was originally intended to export power to Europe and may help resuscitate the supergrid vision.22
A recent study confirmed the feasibility of connecting Europe’s AC grid with North Africa via HVDC lines to dramatically reduce the carbon emissions of the combined supergrid, but the price tag is over $200 billion.23 So in the near term, proposals to further interconnect Europe’s grid—the largest in the world in terms of connected generation capacity—have the most realistic chance of completion. These include connecting the Nordic grid with Russia, delivering offshore wind power in the North Sea to various countries, and realizing a sixty-year-old plan to link Iceland to the rest of Europe through Scotland.
Across the Atlantic, North America is an excellent candidate for further interconnection, building on the well-integrated U.S.-Canada grid. One study focused on the United States found opportunities for more than 20,000 miles of new HVDC transmission lines to internally link up the entire country to enable a surge in renewable energy.24 And there are further opportunities for continental grid integration. Canadian hydropower exports to the United States could rise tenfold if Canada fully developed its hydropower resources and the two countries built new HVDC connectors.25 The biggest rewards, however, lie to the south. Cross-border electricity trade between the United States and Mexico has historically been limited. Now that Mexico has reformed its energy markets, though, the two countries have discussed integrating their grids with the addition of new transmission capacity.26 This integration would offer the United States access to Mexico’s enormous solar resources in sunny regions, such as Baja California. In 2016, the three North American heads of state agreed on a continental goal to reach 50 percent zero-carbon electricity by 2025.27
When it comes to hammering out the details of grid interconnection across borders, there are opportunities for cooperation (consistent with Liu’s predictions of global harmony)—but also discord. Different countries have different rules and regulations governing their power markets, and harmonizing them across borders could prove tricky. Countries are also liable to squabble over how to allocate the costs of shared transmission projects.
And a whole host of international disagreements could color countries’ willingness to jointly take the plunge and link up their power systems. For all the camaraderie between Masa and Liu, Japan and China are riven by serious political and military tensions, including maritime disputes in the East China Sea. Such tensions might prove the biggest obstacle to countries trusting one another enough to willingly increase their dependence on traded energy.
Therefore, Liu’s eventual goal of going beyond regional supergrids to achieve a global one is pretty implausible, requiring more international trust, far more money, and further advances in such technologies as undersea cables. Nevertheless, that ultimate vision might function as an inspirational, if perhaps impossible, target toward which even partial progress—such as the establishment of regional supergrids—would constitute a big step forward for the world. Indeed, bigger grids will be essential for tapping the world’s best solar and other renewable resources.
Still, the push to create ever-bigger grids has its critics, who contend instead that smaller, decentralized grids deliver a myriad of benefits in comparison with a centralized model. One of the most compelling of those benefits could be superior resilience in the face of natural or manmade disasters.
Bracing for the Next Superstorm
In meteorological terms, Hurricane Sandy in 2012 was a middling storm, a Category 2 runt compared with the Category 5 monsters that the Atlantic Ocean breeds every few years. But with precise aim and impeccable timing, Sandy plowed into New York City right at high tide and became a superstorm, inflicting tens of billions of dollars of damage, the second-highest total in U.S. history. The toll on the power grid was extreme. More than 2 million residents of New York City faced outages as Sandy knocked out a third of the generators serving the city, flooded five major transmission substations, and wrecked power poles and equipment on the distribution grid.28
Thanks to climate change, a higher proportion of future storms will be large. The flood of the century could soon become the flood of the decade.29 At the same time as natural disasters are becoming more deadly, the U.S. electricity grid is aging and growing increasingly more vulnerable to being damaged by storms and other stresses. Even in the absence of aging, the interconnected nature of the grid makes it prone to large-scale disruptions. Being interconnected is great for delivering power across great distances, but it is calamitous when a problem in one section rapidly knocks out many other sections. In 2003, for example, an Ohio power line touched an overgrown tree branch, causing a cascading failure that blacked out vast swaths of the eastern United States and Canada, affecting 50 million people. Added to the mundane risks posed by vegetation are the malevolent threats of foreign adversaries who wish the United States harm and see the grid as a prime target.30
In the aftermath of Superstorm Sandy, New York governor Andrew Cuomo tasked the state’s energy agencies to develop a comprehensive plan to prepare for future disasters. From that mandate evolved “Reforming the Energy Vision (REV),” the brainchild of Governor Cuomo’s energy czar, Richard Kauffman.
REV sought to build resilience into New York’s power grid from the ground up. By redesigning the grid to rely much more on distributed energy resources, the formerly centralized grid would not be vulnerable to failure in one fell swoop. For example, a hospital might be connected to the main grid but also generate and store power with solar panels, fuel cells, and batteries on site. If a disaster were to knock out the central grid, the hospital still would be able to run its critical equipment. Importantly, distributed energy resources would be installed in such a way that they themselves were resistant to failure. By contrast, many diesel generators powering New York City hospitals succumbed to flooding when Sandy struck, triggering evacuations of those hospitals.
Although a need for resilience inspired REV’s decentralized architecture, many other benefits soon emerged. Foremost were the enormous potential economic savings from a power system that required only a skeleton of a central grid and could serve most demand with locally sited distributed energy resources. Richard Kauffman made the case to reform the traditional model of the grid by explaining, “We’ve spent $17 billion on the electric system in the past 10 years, and we’re on track to spend $30 billion in the next 10 years. It doesn’t make the system any smarter or more resilient just to replace what we have.”31
What’s more, REV could also make the power system cleaner, thanks to the emergence of low-cost, clean, and decentralized technologies such as rooftop solar panels, batteries, and energy-efficient appliances. These various benefit streams came together in late 2016 when Con Edison, New York City’s power utility, successfully deferred the construction of a $1 billion substation in Brooklyn by soliciting bids for clean, distributed energy from rooftop solar, batteries, and energy-saving measures.32
I was fortunate to serve as an advisor to Richard and the Governor’s Office during this time, where I helped design the regulations to transition New York from a twentieth-century power system to a modern one. I fondly remember one meeting in particular. The team was reviewing a sweeping new policy that would incentivize New York’s power utilities to support distributed energy resources rather than pour money into expensive transmission lines and substations. Unremarkably titled “Track Two,” this regulatory document was typeset in the stultifying, monospaced font that you would expect to see from a bureaucratic agency.33 But Richard held the sheaf of paper tenderly and announced to the room that this document might be as important to the power industry as the Declaration of Independence had been to the United States.
Indeed, the regulatory reforms that document set in motion are declaring independence from the twentieth-century model of power systems. Under that model, which still characterizes the vast majority of power grids around the world, utilities operate a top-down grid. Centralized power plants produce electricity, which is then transmitted and distributed to customers much as the heart pumps blood through arteries and then capillaries to reach the body’s extremities.
The problem with this power grid is that it is centrally planned, inefficient, and dumb. Utilities try to forecast demand patterns years in advance as they plan which segments of the grid to build out. These predictions are often fraught with errors, however. Utilities then oversize system components so that in the worst case—say, on a hot summer day with very high power demand from air conditioning—a substation or power line can deliver as much electricity as customers instantaneously need, even if that equipment is underutilized most of the year. Finally, most utilities have very little live information about how much power flows over the distribution grid. Until very recently, when some U.S. utilities began to roll out smart meters, they had no idea how much energy each customer consumed until a meter reader paid a monthly visit.
These features make the power grid more expensive than it needs to be, and its sprawling infrastructure has innumerable points of potential failure that threaten the whole system. Utilities, meanwhile, have no incentive to move away from this paradigm. Quite the opposite, in fact—the arcane way in which they make money is by collecting ever-higher customer payments to finance grid infrastructure projects and skimming some profit off the top to return to their investors; this paradigm gives utilities every incentive to continue building out the grid. Now, the industry certainly deserves credit for the miracle of delivering highly reliable power to the entire country. This feat led the National Academy of Engineering to proclaim the U.S. grid the most impressive engineering achievement of the twentieth century (figure 8.3).34 But this century, a revamped grid can do much better.
The twentieth-century paradigm of the electric grid.
Source: Energy Information Administration (EIA).
In that spirit, REV aims to create a more decentralized, efficient, and smart grid. In a departure from existing regulations that compel utilities to support distributed energy resources as a matter of compliance, REV offers utilities new ways to make money. If they can persuade customers to cost-effectively meet local needs with on-site generation and to adopt energy-efficient appliances that can be programmed to reduce demand when the grid is under strain, they can avoid building out expensive new power lines and substations. To align their incentives with those of customers, they should be entitled to a cut of the cost savings to the power system.
Also, rather than a single, centralized wholesale power market that sets the price for power over a broad region, REV aims to set up distributed markets that would allow customers to trade energy services at the local level. It envisions dynamic prices for electricity services that change based on location, date, and time, so that customers receive finely tuned price signals based on the system costs that they incur. In a neighborhood with many rooftop solar panels, for instance, the household with a battery system could take advantage of an attractive price for storage capacity, stowing a noontime glut of solar power to prevent the energy from overloading the local circuitry and sending the stored energy back to the grid later on. And because most customers will not want to actively participate in these complex markets, third-party aggregators could step in to trade on behalf of their clients. This market-based approach would enable the most efficient use of distributed energy resources to meet customer needs.
New York’s REV is not alone in its goals. On the West Coast, California is similarly exploring options to move toward a more decentralized model.35 And, abroad, the United Kingdom has pioneered regulations that eliminated perverse incentives for utilities to invest only in major grid infrastructure projects, rather than meeting customer needs in the most efficient way possible.36 Countless other utilities and regulatory bodies are closely watching what happens in such first-mover jurisdictions as New York to evaluate how they can modernize their power systems as well.
A crucial part of this effort will be using technology to create a smarter grid. Today’s overbuilt grids exist because utilities have had little ability to see the real-time flow of power and the instantaneous demands of customers. But much less infrastructure would be required if an intelligent grid operator—or better, autonomous software programs—could reroute power flows and recruit distributed energy resources to meet local demand on the fly.
Achieving this scenario will require advanced equipment deployed across the distribution grid. Sensors on power lines and at substations could provide real-time intelligence about power flows. Power electronics could dynamically divert power to where it needs to go, rapidly shut down overloaded parts of the system, and restore service nearly immediately. Not only would the grid be less prone to massive outages (because it could section off problems before they spread elsewhere), but it would also be self-healing.37
What does this all have to do with solar? A transition toward a smarter, decentralized grid may well improve the grid’s resilience, reduce power costs, and give customers more choices over their energy. But can it help cost-effectively deploy massive amounts of solar power and mitigate solar’s volatility in the same way as grid expansion can?
At first blush, decentralizing the grid would appear to set back, rather than advance, efforts to integrate more solar power. Relying more heavily on local power sources would limit the size and location of solar installations. Instead of faraway utility-scale solar projects in the sunniest locations, a decentralized grid would recruit locally sited solar panels to power nearby communities. But the smaller a solar project, the more it costs (because bigger projects enjoy economies of scale). So, decentralizing the grid could reduce the economic competitiveness of solar power. And it would be challenging to balance intermittent solar power with local demand.
But distributed markets to facilitate local trade in energy services could increase the market value of distributed solar power. For the grid to maintain a highly reliable power supply, it needs to maintain certain attributes of that power, such as the frequency and voltage, at very precise levels. The equipment that connects distributed solar power to the grid—the inverter—could actually help the local distribution grid accomplish those goals.38 Now that smart inverters have hit the market, they can perform these services and thus help to ensure high-quality power at local scales, earning distributed solar generators revenue in new distributed markets. And, if the value gap between utility-scale and distributed solar power narrows—because distributed solar would become more valuable and offset its high cost—deployment of distributed solar could start to catch up with that of utility-scale solar.
The way to cost-effectively deploy distributed solar power is to aim for the Goldilocks zone that sits between installations that are tiny and those that are massive. Rooftop solar installations are expensive because they are small and highly customized. Utility-scale installations are much cheaper, but they are remotely sited and require expensive transmission lines to deliver their power. In the middle are solar installations sized at just a few megawatts. These are small enough that they can be sited in the often empty surroundings of a grid substation, enabling them to plug right into the grid with limited additional expense. But they are large enough that they enjoy economies of scale that drive down their cost. In addition to being economically favorable, these installations—equipped with smart inverters—are an ideal size to help stabilize the grid.39
Finally, a smarter, market-driven decentralized grid also could make it easier to deal with the unreliability of solar PV installations—be they big, small, or Goldilocks-sized. Such a grid could modulate customer demand so that it matches up with fluctuating solar output. This demand-side solution to the problem of solar’s intermittent supply is what we turn to next.
A Grid Hive Mind
Each January, while the world’s most powerful business and political leaders hobnob at the World Economic Forum, leaders of the North American energy industry meet in Vail, Colorado, for the “Davos of Energy.” Just as Davos has increasingly shined a spotlight on clean energy and climate change, so too has the Vail crowd expanded beyond just the oil and gas industry. On the final day of the 2017 summit, when participants took to the ski slopes, I found myself seated on a chair lift next to Microsoft’s chief sustainability officer, Rob Bernard, with a long ride ahead of us.
We were on the first lift run of the day, at the crack of dawn, and the temperature was barely above zero Fahrenheit. But I forgot about the chill as Rob excitedly shared Microsoft’s idea for how it might be possible to eliminate nearly all the carbon emissions from its data centers by running them only during periods of abundant renewable energy supply on the grid.40 “The opportunity is huge,” he gesticulated with his oversized gloves. One of Microsoft’s data centers in the state of Washington alone uses as much energy as 40,000 homes.41 Many of its operations—such as backing up or indexing its cloud customers’ data—can be flexibly scheduled to increase or decrease the data center’s instantaneous power consumption. And, by using a machine-learning algorithm to predict, based on past records and current weather data, when such renewable generators as solar panels will flood the grid with power, Microsoft can adjust its data centers’ operations to consume that surplus solar power. Rob’s idea was so captivating that I realized that my fingers were freezing only when the end of the lift interrupted our conversation. Graciously, he guided me down the quickest run to the lodge so I could warm up with some hot chocolate.
What Rob was proposing was an example of a general strategy called “demand response,” in which customers adjust their power consumption to help the grid balance supply and demand. Demand response has actually been around for decades, although not in nearly as nimble a form as what Microsoft and others are pursuing today. Currently, utilities have various programs in which they pay customers to turn down the air conditioning on a hot summer day when high demand is straining the grid (to cite just one example). In some cases, utilities can directly control customer appliances to crank up the thermostat a few degrees.42 And some large industrial power customers already regulate their demand at a greater scale. In most of the major U.S. electricity markets, they can sell negative megawatts (“negawatts”) of power savings alongside the megawatts of power supplied by conventional power plants.43
But this is just the beginning for demand response. A wave of start-up companies has persuaded venture capital investors to fund demand-side innovations, early returns from which are much more promising than the flopped investments a decade ago in clean energy supply technologies.44 As smarter grids gain the ability to harness a wide range of distributed energy resources and coordinate them, a much wider range of customer equipment—large and small—could act in concert to modulate demand in a way that matches up with intermittent renewable energy supply.
Many trends are on course to converge at enabling this reality. First, customers are increasingly buying appliances that connect to the Internet—be they smart thermostats, washing machines, lights, or refrigerators. Industrial power customers are also connecting their equipment to the Internet. By 2020, the overall number of devices composing the “Internet of Things” is on track to reach 50 billion—double the number in 2015.45 These devices can be remotely controlled (for example, via a smartphone app), changing their instantaneous energy consumption. The same applies to the power-intensive equipment in large buildings, which increasingly is controlled through smart building energy management control systems.
Second, utilities are rolling out “smart meters” that measure household power consumption on a much more granular basis than every month—some measure it every hour, minute, or even second. These meters can communicate with the grid to help both grid operators and customers find out what the other needs. Over half of American households have smart meters now. Although the torrents of data pouring in from these meters have overwhelmed some utilities, others have started to manage power flows on the distribution grid intelligently to meet local customer needs.46
Third, two-way communications networks and software solutions are emerging that can be overlain on the new hardware to orchestrate effective demand response by managing distributed energy resources in concert. A utility can use these tools to act as the nerve center and centrally control everything on the grid, from power lines and substations to customer appliances.
But decentralized control algorithms are becoming available as well and should mirror the decentralization of the grid’s hardware. In the future, the utility might coordinate only the macroscopic, or large-scale, operation of the grid, a virtual eye in the sky sending broad control signals via the cloud. But numerous local networks may also operate at the grid’s edge, connecting individuals within a local community, such as a neighborhood or a university. Autonomous software algorithms at the local level could speedily coordinate the functioning of distributed energy resources to match up local supply and demand and respond to the coarse signals sent from the central utility. This might resemble an ant colony, in which decision-making is decentralized but manages to advance the goals of the entire collective.47
What could the grid’s new hive mind mean for solar power? One exciting prospect is the emergence of virtual power plants. If distributed energy resources all over the grid could act in concert, their collective effect could be the ideal complement to solar’s intermittency. For example, batteries might absorb energy as surplus solar power floods the grid. Or a combination of thermostats, water heaters, and electrical equipment might instantaneously scale back power demand to compensate for drooping solar output.48 Already, countries around the world are experimenting with this strategy. To cope with worsening blackouts in southern Australia, for example, a utility switched on the world’s largest virtual power plant in Adelaide in 2017, networking together 1,000 home batteries to better balance supply and demand.49 (Tesla also has plans in the region to deploy the world’s largest battery system to help improve the grid’s reliability.)
Thus, a decentralized grid could make it easier to absorb large amounts of intermittent solar power. And combining the best features of centralized and distributed grids could integrate even more solar power. Expanding a centralized grid—possibly culminating in a supergrid—would smooth out solar output through geographical aggregation and also connect faraway solar supplies with the needs of power-hungry cities. Meanwhile, the demand response capability that could emerge from a decentralized grid’s hive mind would complement the supergrid’s activities, adjusting demand in response to variable solar supply.
Back at Microsoft, engineers are working to bridge the gap between the supergrid and distributed-grid strategies by combining virtual power plants with virtual transmission lines. Their idea is to swap out electrical HVDC transmission lines for much cheaper optical fibers that could create a different kind of supergrid. In their vision, a global network of interconnected data centers could shift energy-intensive computation around the world to wherever renewable energy output is strongest at that moment. So when the sun is shining over Europe, U.S. data centers would scale back their power consumption and transmit their data over to Europe to be processed and sent back.50 Thus, demand response on a decentralized grid could give rise to an optical supergrid.
That idea is a little far-fetched—and possibly unworkable, given the latency, or delays, that would accumulate from trying to move data around the world to follow the sun. Also, data centers account for only a small (but rapidly growing) fraction of the world’s power consumption. And yet, the core insight—a hybrid approach that would combine the demand response capability of a decentralized grid with the interconnections of a supergrid—is sound. An electrical, not optical, hybrid grid is not so far off, and it might represent the world’s best shot at integrating massive amounts of solar power.
Best of Both Worlds
“Can you hear me now? Good!” That Verizon representative’s pithy refrain from a memorable ad campaign sums up one of the remarkable successes of the twenty-first century: widespread cell-phone coverage networks. (If the ad instead conjures up frustrating experiences with cell phone carriers, at least remember that a couple of decades ago, there was no service to complain about.) Somehow, cell networks provide reasonably reliable 24/7 service to customers who collectively are forever starting and ending calls, surfing the Web, and moving around unpredictably. Embedded in that success is an important lesson for how to construct a hybrid power grid that copes well with unpredictability.
The way that cell phones access the network of cell towers is complex and dynamic. In the simplest case, a cell phone will access the nearest tower, which offers the strongest signal. If too many users saturate a single tower, your cellphone may be rerouted to a network of towers farther away. Sophisticated protocols hand off cellphone connections between towers when you are on the move.
In a similar way, in one expert’s sketch of a hybrid grid, a honeycomb of micro-grids might connect to one another dynamically in a cellular fashion, rerouting power flows to adapt to congestion and instantaneous customer demand.51 Recall from Chapter 5 that a microgrid networks together a community—perhaps a neighborhood, university, hospital, military base, or large office building—linking distributed energy resources that produce and consume energy. These microgrids would also be connected to a central grid but would each have locally sited control systems and the capability to operate autonomously.52 And they would derive a large fraction of their electricity supply from megawatt-scale distributed solar installations located on site or nearby.
In the case of a disaster, microgrids designed with resilience in mind can disconnect from the main grid and continue to operate. Such independence can also stop a spreading blackout in its tracks. Microgrids can even make it harder for hackers to attack the overall grid by keeping localized intrusions from spreading.53 To avoid having to overbuild power-generation sources and network capacity to ensure that the power supply can meet demand at the local community level even when the demand is unusually high, these modular microgrids could be networked together to connect dynamically with each other and trade energy services.
In this conception, long-distance transmission lines would then form a central grid backbone that connects neighboring regions (and even countries) to create a supergrid. Existing proposals for a supergrid call for adding HVDC interconnections to dramatically expand the grid’s reach and access faraway renewable resources. But those proposals would largely leave the underlying grids interconnected by HVDC untouched. The hybrid grid, by contrast, would reengineer the underlying grids as well.
Much of the existing, overbuilt AC grid would become unnecessary—it would be replaced by bare-bones connections between microgrids. If the microgrids were to run on DC electricity (which is increasingly a good idea given advances in DC power electronics), then the entire hybrid grid would run on DC, from high-voltage, long-distance transmission lines to local, cellular microgrids.54 This would minimize conversion losses from DC to AC, reduce energy consumption thanks to efficient DC appliances, and bring a myriad of other benefits, such as the prospect of underground, long-distance power lines. The grid would thus come full circle from the early days before Edison’s DC technology was superseded by Tesla’s AC grids.
Figure 8.4 portrays an illustrative schematic of what this hybrid grid might look like. It would combine microgrids that efficiently use distributed energy resources to regulate supply and demand at the local level with long-distance transmission links that can shuttle power between faraway regions and access remote but rich renewable resources. And it would replace much of today’s existing, bloated grid that relies on expensive AC transmission and distribution infrastructure, relying on asset-light microgrids at the local scale and HVDC lines at the regional and global scales.
Schematic of what a hybrid grid might look like. The left panel represents a continent-sized supergrid (pictured: the regional supergrid connecting Europe and North Africa proposed by DESERTEC). The symbols represent renewable energy generators, and the lines represent long-distance, HVDC transmission lines. The right panel zooms in to reveal the decentralized microstructure of the hybrid grid. At this scale, the grid consists of networked microgrids with their own distributed energy resources (clockwise from top left are a college campus, a neighborhood, an industrial facility, and a military base). These networked microgrids connect to the superstructure of HVDC lines that connect disparate regions and enable access to faraway renewable resources.
Important note: This schematic is not exhaustive—that is, many other resources, like conventional power plants, can plug into this hybrid grid, though they are not shown owing to space constraints.
Source: DESERTEC map obtained from Wikimedia Commons.
Achieving a hybrid grid would be the pinnacle of systemic innovation. But steep barriers stand in the way. Almost everywhere, clumsy regulations and conservative utilities are unwilling to move beyond the twentieth-century paradigm of the power system—even in such pioneering states as New York and California, progress is slow. There are also considerable technical details to work out—including standards for how smart microgrids can talk to one another and effectively emulate a cellular network.55 Cost is another issue. HVDC lines are expensive, and although a decentralized network of microgrids is a highly efficient setup that avoids the high capital costs of traditional grid equipment, utilities that have already financed the existing grid need to recoup their investments even as a new model of the grid emerges. And of course, there are geopolitical considerations, as the long-distance transmission parts of a hybrid grid might compel countries to depend on unsavory neighbors.
But as more solar power connects to the grid, overwhelming utilities, regulators may be forced to take rapid action. Cost will always obstruct change, but the compelling value proposition of the hybrid grid—saving money by greatly reducing today’s bloated AC infrastructure and investing it instead in highly valuable, long-distance HVDC lines—could attract firms and governments to the new model.
And countries might just be willing to interconnect with their neighbors, friendly or not. Because the hybrid grid would both increase regional connectivity and decentralize control over power supply and demand, communities would become more self-reliant in the event of disruptions to long-distance power flows. Indeed, that is the point of the hybrid grid—to achieve the best of both worlds between decentralizing the grid and creating supergrids. And by dramatically improving the grid’s versatility, the hybrid grid could be the best way to handle an influx of centralized and distributed solar power.
All of this means that perhaps Masa and Liu, the visionaries behind the global supergrid, actually weren’t thinking big enough to envision going beyond even a supergrid to a hybrid grid. Or rather, they weren’t thinking both big and small enough.