How To Fix The Ailing Electric Grid

Today’s power grid is broken, but a little bit of teamwork could save it.

How To Fix The Ailing Electric Grid
[Top photo: Flickr user Rennett Stowe]

At the moment, the grid is about where the phone system was in the 1940s. Long-distance phone calls were scheduled two hours or more ahead of time, and operators had to make the connections manually. A long-distance call could take fifteen minutes to set up.


As crazy as this may seem in our day and age, our utility system tends to schedule power generation and transmission twenty-four hours in advance and in fifteen-minute increments to allow time for manual interventions in power plants.

But there’s hope. In the phone system, after World War II, switches were replaced with automated relays, and the area code system we now take for granted was put in place, removing the need for armies of operators. Then, in the 1980s, Cisco invented the software-defined network (SDN) routers that became core to the network. The routers eliminated the need for dedicated lines for each phone call, driving down costs dramatically and opening up the system to cell phones and Internet communications. The same sort of dramatic improvements will come to the electric grid.

The modern electric grid traces its roots to the spring of 1885, when Thomas Edison and George Westinghouse competed to define the standard for bringing the magic of electricity to the country. After settling on alternating current, the United States spent most of the next hundred years building out a power grid providing nearly universal service nationally, relying on 5,800 central station power plants. The build-out of the grid has been called the engineering marvel of the twentieth century, but the basic technology of the grid has changed little since the time of Edison and Westinghouse. The average circuit is forty years old, and some are more than a century old. The grid is showing its age.

Today, many utilities still do not know what’s occurring on their grid or where. The average utility generally learns about problems with its lines by looking for clusters of customers who are calling in to complain, rather than by receiving information on the problems directly. Issues at substations often have to be addressed by sending maintenance workers into the field to flip a switch, rather than being able to have someone in a remote office make the change–or, better yet, have the grid sense the problem and either fix it automatically or route electricity around it.

Generally, only 20 to 40% of the transmission and distribution capacity in the network is in use at a given time, and only about 30 to 40% of the capacity of power plants is being used. Many “peaker” power plants sit idle all but a few dozen hours a year, just so they can be fired up to handle demand during the peak of air-conditioning use on the five hottest days in the summer. These plants, by the way, can cost so much to run that utilities can lose money on peak demand despite charging what consumers see as exorbitant rates.

Power plants turn only a bit more than half the energy in their fuel into electricity, and more than 10% of that electricity is dissipated between the plant and the end user through what are called line losses. Many utilities are still running COBOL code from the 1950s, and some are still rekeying data or transporting tapes to issue paper bills once a month.


Utilities not only have to overcome their massive inefficiencies but have to adapt to the contemporary, rapidly shifting environment. Homeowners are putting solar panels on their roofs, which not only takes customers from the utilities (usually, the most profitable customers) but also requires that utilities figure out how to integrate into the grid the power that the homes sometimes make available. At the moment, in parts of northern Germany, power from solar and wind often becomes available just when the distribution network is maxed out and can’t handle any more power, meaning that the renewable sources have to be shut off.

Engineers are developing advanced building-control technologies, and architects have designed more efficient buildings that consume electricity during off hours–for instance, cities like Austin use cheap electricity at night to freeze water, then, during the day, circulate air over the ice to cool buildings. New air-conditioning systems can similarly use evaporation to cool buildings instead of the compressors that make current air conditioners so noisy. Such compressionless air-conditioning systems may reduce the use of electricity by 50% or more, eliminating much of the afternoon peak that utilities in the United States have typically designed their grids for. Lighting is getting much more efficient, with LED bulbs cutting electricity usage by 85%, compared with incandescents. Utilities can continue to drive increasing electrification (including everything from ports to lawn mowers).

Once electric vehicles deploy in large numbers, utilities will have to get used to the power equivalent of a commercial building or small factory unplugging, moving, and plugging back in somewhere else. Utilities are going to have to develop massive capabilities for integrating not only what they are doing but what all the related players are doing, too.

Plenty of technologies will be able to help with the reinvention. For instance, batteries distributed throughout the grid could reduce costs by allowing stored electricity to be generated in a smooth, steady way, with plants operating at maximum efficiency and with electricity drawn from the batteries as needed–no more need for entire plants to be kept ready all year just for a few hours of work on a few days. Solid-state transformers, which rely on sleek, semiconductor technology and which are currently used in high-speed trains and military airplanes, could provide much higher reliability at much lower cost than conventional transformers, which are bulky, copper-based devices; substations will fit in a small vault rather than occupying a full city block.

Electric vehicles could act as grid storage, drawing down power when demand is low and plants can produce power most efficiently. Sensors in the network will be able to spot problems in the network and fix them or route around them, much as the Internet does today with data traffic. An array of smart grid meters at homes and offices will enable interactions with users–balancing of the grid won’t just happen by generating more power; balancing might also happen by reducing demand for brief stretches. Power will be able to flow in both directions, from where there is more to where it is needed, rather than have a few central plants generate all the power.

But, to get us to a twenty-first-century grid, utilities will need to be superb at system integration. At the moment, the average utility collects about 60 million data points each year–5 million customers and a dozen monthly bills. When smart meters come into widespread use, the average utility may have to handle 5 billion data points every day. The grid will need almost to be redesigned from scratch to get the full benefit of the new types of transformers, the capability to sense problems and solve them automatically, the ability to essentially have little power plants on millions of rooftops as solar prices keep coming down, and so on.


Yet, utilities will have to integrate all the potentially beneficial new technologies in an environment where simply installing a sensor on the network can require shutting down for a disruptively long and expensive stretch. Utilities will need to start by recognizing the extraordinary scope of the challenges and opportunities in front of them. They’ll need to ask themselves tough questions at every step. Has the company evaluated all the opportunities to embed software? Is the company involving suppliers actively in evaluating design alternatives? What will it take to create the tools to evaluate cost, time, and value tradeoffs in design? How well does the new project plug into the global network? And so on.

Flickr user Brian Stocks

Regulators will need to be partners on this journey with the utilities. Although regulators aren’t known for being on the leading edge of innovation, and although most utilities are more than happy to simply collect a set rate of return on their investments, regulators must confront the scope of the transformation that the grid will go through in the next twenty years and start to map out how to get there from here.

To integrate all the aspects of the grid, utilities will need to invest heavily in software and control systems, or they’ll be swamped. Only a very few utilities have a chief technology officer and a software engineering department. Almost none have the kind of software team that is necessary to deliver the integrated hardware-software control routines that will manage a smart, data-intensive grid.

The good news is that many of the capabilities that utilities will need aren’t new, even though they will be very unfamiliar to utilities. While utilities think in terms of millions of interactions with customers a year, not billions a day, cell phone companies know how to manage the volumes that utilities will be facing. Aerospace companies also know how to handle massive amounts of data in a dynamic system. Partnerships are waiting to be made.

Other countries can be good sources for new expertise, too. Renewable generation such as solar or wind power increased from 0.06% of total electric power in the United States in 1996 to 5% in 2013 without any noticeable impact on grid reliability, but there is uncertainty about how to keep increasing the percentage, given the fragmented U.S. regulatory structure and numerous transmission bottlenecks. American utilities could learn from Germany, where utilities are successfully drawing 36% of their power from renewables, on average, and more than 59% at peak times. State Grid in China has developed expertise on bringing power from generating centers in remote deserts to the coastal cities without significant losses of power.

Massive computational models will need to be developed for the new version of the grid to ensure that loads can be balanced, that the network can be protected from external attack, and that future changes to the network can be made as software upgrades, without having whole circuits taken down. As we’ve said, being able to place huge batteries throughout the grid can smooth out the generation of power and make utilities far more efficient–but no one has a basic model for how batteries would change the flow of power on the grid. And that’s just the beginning of the long list of issues that aren’t yet understood. Companies will need to be willing to experiment endlessly to get the full benefits of switching from a model like the phone network of the 1940s, with its armies of operators, to a model like today’s Internet.


Utilities and their suppliers will need to collaborate much more than they do now. Rather than do their own thing so much, every company that contributes to the grid needs to focus on developing standards so that every company, every piece of equipment, and every customer can communicate with every other person or device.

Wireless communication wouldn’t be a tenth as effective as it is now if telecom companies, makers of wireless routers, developers of phones and tablet computers, and others had all pursued their own approaches to transmitting signals. The magic of Wi-Fi is that standards were established early on so that every person and device can talk to every other person and device. Sometimes, standards simply emerge, as when VHS bested Betamax and became the standard for VCRs. Sometimes, the government sets standards, as happened with the Internet. Oftentimes, though, it’s important to have people go through the drudgery of engaging with dozens of other companies and agreeing on standards. It’s no fun, but it’s important–that approach is what gave us Wi-Fi: An industry-wide group of engineers has worked for years to develop the 802.11 standards.

Electric utilities will also need to coordinate with one another and with other types of utilities in planning, capital upgrades, and scheduling. Much of the instability in the U.S. electric grid comes from two types of problems: trees that bring down lines in storms and bottlenecks in transmission or distribution capacity at jurisdictional boundaries. The first problem is relatively simple to solve: Move the lines underground, as Japan and much of Europe have done. The challenge in the United States is not a technical one but a financial one; putting lines underground is expensive.

While living in New York City, Stefan watched his street being dug up by Con Edison, the local electric utility. Then the Department of Transportation repaved the streets, only to have Verizon rip them up again to lay optical fiber and harden its infrastructure after Hurricane Sandy. Moving lines underground wouldn’t be expensive if optical fiber, telephone, power, and cable television could all be laid together in a conduit and the street repaved a single time. New York has tried to achieve this coordination by putting a five-year moratorium on underground permits for tearing up streets that were just repaved, but this approach clearly still falls short of syncing up maintenance and improvement plans across utilities.

The same lack of coordination afflicts all the different bodies that produce and distribute electricity. Currently, each utility develops its own capacity plan, and there are more than 3,000 electric utilities in the United States alone–200 large ones and the rest owned by communities or municipalities. Independent System Operators (ISOs) plan transmission capacity in many markets, but there are dozens of those, as well. In addition, the eastern half of the country, the West, and Texas are not connected; they operate effectively separate grids. Many municipalities run their own power grid, only partly connected to the rest of the state and region.

Balancing the grid becomes much easier and less expensive if coordination happens over a larger area. Differences in wind, sunshine, and temperature can balance out. Simply linking different time zones can help to balance the grid, because daily peaks related to the workday, for instance, come at different times. Our modeling suggests that if you operate a grid that is more than a few hundred miles across and has more than a thousand generation sources, even a mix of renewable energy sources–considered unreliable because the wind doesn’t always blow and the sun doesn’t always shine–begins to look as reliable as nuclear power plants.


Thousands of American utilities, regulated separately by utilities commissions in each of the fifty states, rarely look at whether coordinating with companies beyond their territory would make upgrades less expensive. The system is designed for shipping commodity electrons and ensuring reliability by systematically oversupplying–the so-called reserve margin where between 2% and 6% more power is contracted for than is required by loads in the grid. But the utility business and regulatory models will have to evolve.


About the author

Matt has spent 25 years serving energy clients globally and currently focuses on the roles technology and innovation play in restructuring markets in the energy and industrial sectors. He leads McKinsey’s Americas Petroleum and Electric Power and Natural Gas Practices and helped establish the firm’s Resource Productivity and Clean Technology Practice. He has written extensively on the oil, gas, power, and resource markets and served as senior advisor to the US Secretary of Energy in 2009–10


All week you can attend Innovation Festival keynotes with Robert Downey Jr., Malala Yousafzai, Chip and Joanna Gaines, Janelle Monáe, and more. Claim your free pass now.