Los Alamos National Laboratory is known for interdisciplinary research. Still it's unusual to find a research project in which the Laboratory's Utilities Division leader (Andy Erickson) and a team of high-powered physicists and engineers will use their own workplace, the local Los Alamos County utility, the townspeople, and a foreign collaborator to help solve one of our nation's energy problems. The collaborator, Japan's New Energy and Industrial Technology Development Organization (NEDO), will be demonstrating how its technology could be part of the solution.
The problem—modernizing the nation's overburdened power grid—has captured the imagination of so many because there seems to be a "smart" solution, one that could transform our energy use as profoundly as cell phones and the Internet have transformed our communications. Called the smart grid, the solution involves consumers and smart meters in making the grid more cost effective, more compatible with renewable energy sources, and more resistant to blackouts. It's being tried on various scales across the nation.
The Los Alamos-NEDO trial of a smart grid has a special focus: to help rural communities across the country incorporate renewable energy and smart grid technology in an affordable way. Erickson explains, "Our project is going to include computer simulation tools to help control, in real time, a community-scale smart grid demonstration that gets a significant amount of power from a large solar photovoltaic (PV) array. Based on the data we collect, we will design smart, cost-effective ways for small communities to incorporate renewables on their local grids."
What's Wrong with Today's Grid?
The nation was startled on August 14, 2003. Starting at 4:04 p.m. on a typical warm summer afternoon, the power went out in and around Akron, Ohio, and over the next seven minutes, the blackout rolled across most of the northeastern United States and Canada. At each successive failure, the grid tried to pick up the load, but line after line became overloaded, tripping breakers, shutting down generators, and causing town after town, state after state to go dark. The elevators, air conditioners, computers, electric clocks, commuter trains, airport monitors, street lights, office lights, radios, TVs, appliances, cash registers, the ticker tape at the New York stock exchange, the billboards on Times Square—almost every convenience of modern life—suddenly stopped working, and 50 million people were left to cope. It took days to recover, and business and personal losses totaled around $6 billion.
By volunteering to install smart meters (like the one shown here) in their homes and consider price signals when using electric power, Los Alamos residents can become part of the Los Alamos-NEDO smart grid project.
How could such a massive blackout occur? The reason stems from the properties of the grid. The North American electrical power grid is a system of four loosely connected, very large networks. Together they comprise about 3500 large power plants (mostly fossil fuel), interconnected by 200,000 miles of high-voltage transmission lines connected in turn to millions of miles of low-voltage local distribution networks that deliver electricity on demand to about 131 million separate consumers. This system of centralized supply and distributed demand can produce up to 1 trillion watts of electric power continuously, electrifying virtually every building and facility in the United States and Canada.
The grid is an engineering marvel, but can be a bear to control when things go wrong. Unlike water or natural gas delivery systems, in which large water tanks or pressurized gas containers can store "product" for future delivery, there is little energy storage on today's grid. Thus, electrical energy must be consumed as soon as it's produced; supply and demand must be exquisitely balanced at all times. Because electrical energy travels over the lines at nearly the speed of light, the whole system acts like a single machine with nearly instantaneous feedbacks. That's why a series of accidental faults on the lines or operator errors or unexpected loads at times of peak demand can lead to an event like the 2003 rolling blackout. Even without massive blackouts, power outages and drops in voltage or power quality cause estimated business losses of $100 billion annually.
Because electricity must be consumed in real time and fossil-fuel power generation is much more controllable than consumer demand, demand is the tail that wags the dog in today's grid. Specifically, demand is forecast about 8 hours ahead of generation, affording time for "baseload" generation to be spun up to meet that demand. Typically, the predictions are sufficiently accurate to allow operators to control normal flows of power, and the difference is made up by buying and selling electricity every hour on the hour between utilities.
The system gets stressed during times of peak demand, for example, on hot summer days when all the air conditioners in the South or the West get turned on at nearly the same time. To meet that demand, the grid holds a certain amount of capacity in reserve: either coal-fired generators are operated below maximum power, which leads to high carbon emissions, or in a more expensive option, gas turbines are kept idle most of the time, ready to be spun up rapidly—in 15 minutes from a cold start—and put on line.
Finally, grid stability is seriously incompatible with the intermittent power from renewable energy sources. Solar irradiance and wind are fluctuating sources, and their power output can drop suddenly as clouds obscure the sun or the wind dies down. How do you keep the grid stable when suddenly a wind farm's contribution drops from 1700 megawatts to 300 megawatts in just a few minutes, as happened in Texas in 2008? How do you keep the voltage on the line from dropping, causing motors to burn out as they try to draw more and more current? And how do you manage to match supply to demand at every instant when there are hundreds or thousands of fluctuating solar and wind sources contributing a substantial fraction of the grid's power?
Loren Toole, a principal investigator in the Laboratory's part of the Los Alamos-NEDO smart grid project, recalls what happened in California in the 1980s: "When many wind farms were incorporated into the grid over several years without proper controls and planning, a large fraction of the energy from wind farms had to be dumped before it ever reached the grid." The power was in excess of demand and would have destabilized the grid.
John Arrowsmith (center) at the landfill site for the solar PV arrays with the Laboratory's Venkateswara Dasari and Carolyn Zerkle, both of whom were instrumental in starting the Los Alamos-NEDO project.
"Today, when renewable sources are added to the grid, they are being backed up by adding new fossil-fuel generating capacity. I liken that to driving with one foot on the accelerator and the other halfway on the brake," quips Karl Jonietz, the Laboratory's program manager for the Office of Electricity Delivery and Energy Reliability, which funds several grid-related programs. "That's no way to reduce carbon emissions or make us energy independent."
What Makes a Grid "Smart"?
Clearly, if renewables are to make a significant contribution to the nation's power grid in the next 20 years, the paradigm of grid operation has to change. No longer can demand be the tail that wags the dog. As supply becomes more distributed and less predictable, various means of storing energy have to be added to the grid, and demand has to be responsive to changes in supply. This requires building more intelligence into the grid at every level. And that intelligence can then be used to help solve all the grid's problems. So goes the argument for the smart grid, a set of concepts and technologies that emerged over several decades from interactions among industry, universities, and electric regulatory agencies. Today the Department of Energy is actively encouraging its implementation.
The present grid would become smart if it were given an overlay of information technology that could act, within each area, somewhat like a central nervous system. Digital two-way smart meters (the "nerves") would be located at points throughout the grid down to individual homes. Each meter would measure the state of the grid at its location—the amount of current flowing, the voltage values, and how far AC current deviates from its nominal (standard) value of 60 cycles per second—and send that data (using wireless communication, fiber optics, or other means) to the computerized utility control system (the "brain"). Each control system would input that data to its own predictive computer models that would forecast the future state of the local grid and recommend (to operators), or automatically implement, actions needed to maintain grid stability.
Those actions would include not just modifying power generation but also initiating a new paradigm: "demand response," that is, getting users to change their demand in response to changes in supply. One approach to initiating demand response is the use of real-time price signals. Rates could be reduced during the night when demand is low and raised during times of peak demand or when the grid is stressed because of a failure in the system or a drop in output from renewables. Control systems would send price signals to smart meters, and those meters could be programmed by consumers to turn up the thermostats on hot afternoons when prices are high or turn off certain appliances, or delay the charging of electric cars, in order to reduce demand and help grid stability. In the smart grid revolution, consumers would be more informed and would have more control, either through their own energy management systems or through their local utility. By reducing levels of peak demand, consumers would help reduce the need for excess generating capacity, stresses on transmission lines, losses due to power disruptions, and so on.
The individual pieces of technology—smart meters, smart appliances, communication systems—all exist, but integrating them into a working system, maintaining control over the system, getting the public to participate, and ensuring that the return on investment makes good business sense are objectives that are being explored in towns and cities across the nation.
Los Alamos Gets Involved
The Los Alamos-NEDO smart grid project hopes to answer a particular question: how do you use smart grid technology to incorporate renewables at the community level? Jonietz explains, "Not all homeowners can afford to invest in solar panels, but often it's feasible for local utilities to invest in large solar arrays that serve clusters of homes, small towns, or sections of larger towns. The Laboratory, working with NEDO and the county, will be researching how to make that option work on both a technical level and a business level."
The project will be set up as a research and demonstration project. Jonietz continues, "We will use the Lab's strength in modeling and simulation to come up with a decision-support program and a generic control system that could serve as a blueprint for incorporating renewables and smart grid technology into the nearly 1700 public and cooperatively owned utilities across the United States." That market buys close to 25 percent of the nation's electric power.
The main capital investments for the Los Alamos-NEDO project (see figure below) are two 1-megawatt (MW)solar PV arrays connected to two commercial-size batteries, which are controlled by a micro energy management system (micro EMS). The system will be constructed on a capped landfill near the Laboratory, and its power output will be fed directly into the low-voltage side of the nearby substation, where power coming in on high-voltage transmission lines is transformed to a lower distribution-level voltage. The power will then be distributed to the town through the county utility's distribution feeder lines and to the Laboratory through the Lab's distribution system.
The Los Alamos County Department of Public Utilities, led by John Arrowsmith, will construct one of the 1-MW solar arrays. Erickson, who as Laboratory representative purchases 80 percent of the power coming into the county, will back that investment by guaranteeing to purchase power from the array. The arrangement will support the smart grid project and will help the Laboratory meet the federal requirement that by 2012 all federal sites purchase 7.5 percent of their power from renewable sources.
Left to right: Andy Erickson, Rafael de la Torre (electrical engineer for the county utility), Dr. Satoshi Morozumi, Karl Jonietz, and John Arrowsmith in downtown Los Alamos after finalizing agreements with Japanese industry representatives.
The rest of the installation (solar array, batteries, and micro EMS) will be a demonstration of NEDO-contributed solar and smart grid technology. NEDO fosters greater use of new energy and conservation of energy by promoting research and development collaborations among Japanese industry, universities, and government. NEDO leaders chose Los Alamos to host this utility-scale demonstration because of the county utility's willingness and ability to institute rate incentives without permission from the state regulatory commission. NEDO is also interested in working with the Laboratory on technology standardization and other issues related to the smart grid. In addition, the Japanese see this project as opening doors to a future U.S. market dependent on renewables.
NEDO's Dr. Satoshi Morozumi remarks, "This type of utility-scale system will help get PV systems into the U.S. grid and will get community utilities into a good position relative to the wholesale market. By having their own PV resources, they can choose when to buy power and when to generate their own."
Left to right: The Lab's Andy Erickson, Scott Backhaus, and Loren Toole are honing plans for the Lab's end of the smart grid project.
This sounds good, but it works only if the solar fluctuations on the local array can be balanced. The monthly bill for each local utility has two parts: an energy charge for the total number of kilowatt-hours used and a demand charge for the highest average rate of energy use (number of watts) recorded during any 15-minute period. (The demand charge pays for the excess generating capacity and transmission capacity needed to meet peak demand.) If you're unlucky and the local solar output dips just when consumer demand peaks, the local grid will draw power from the main grid at just the wrong time, and the monthly demand charge will go way up, defeating efforts to reduce costs.
To help avoid those situations, Erickson, wearing his utility hat, is making the Laboratory's infrastructure available as part of the smart grid project. Three on-site generation sources (a diesel generator, a gas turbine generator, and a steam turbine) might be used. Also certain large electrical loads (a wastewater pump at the Laboratory's water treatment plant and the air-conditioning system for the Lab's seven-story NSSB administration building) will be outfitted with smart meters and converted into demand-response loads that will be used to balance fluctuations in solar output and smooth out demand through the day. In addition, Erickson and Arrowsmith are working to bring online a new hydroelectric resource, a low-flow 3-MW turbine at Abiquiu dam, which could be used to smooth fluctuations from the solar array. The county's potable water pumps and storage tanks are also potential demand-response loads.
Misha Chertkov, leader of a separate Laboratory Directed Research and Development (LDRD) theoretical project on smart grid stability comments, "In principle, the county and the Lab can be very creative about controlling these variable sources and commercial-size loads. They could smooth out the fluctuating output from the solar arrays so well that PNM [the utility that controls much of the regional grid in New Mexico] wouldn't be able to detect that the county had put a solar array on the grid. If we can develop and demonstrate such automated control schemes, the solar integration problem will be solved, and other municipalities can copy what's being done in Los Alamos."
The control schemes will be designed from statistical descriptions of the frequency and duration of fluctuations in sunlight and from analyses of the response times and flexibility of the variable sources and demand-response loads being contributed by the Lab and the county. Scott Backhaus, a Laboratory experimental physicist and the second principal investigator on the Los Alamos-NEDO smart grid project, is being supported right now by the DOE's Office of Electricity to perform those analyses and develop control schemes. Backhaus won't be alone. Chertkov, Russell Bent, and other project theoreticians in the LDRD smart grid project are developing general mathematical approaches that likely will prove very helpful.
Control schemes are rules for deciding which power sources or demand-response loads to bring into play to balance supply and demand, given the county's price signals, weather, charge on the batteries, and so on. The rules, rather than being used independently, become one component of a the utility control system: the "brain" in a smart grid.
A second component of the control system is being developed by Toole and his colleagues from an advanced grid simulator called TRANS-EX, created in 2009 to determine the transmission elements that would be needed to gather 20 percent of the grid's total power from wind farms. Whereas Backhaus's control schemes will forecast the power available to meet demand, Toole's simulator will recommend distribution-level actions to operators.
Testing High-Penetration Solar
Even before NEDO chose to site its demonstrations in Los Alamos, Toole used TRANS-EX to create a live computer simulation of a scenario of great interest to NEDO—namely, how the North Mesa neighborhood grid in Los Alamos would respond if the proposed NEDO solar array and battery system were the sole supplier of power (called 100 percent penetration) and suddenly clouds covered the solar array.
The simulation (above) showed how the micro EMS could shift the North Mesa load from the solar array to the battery during a sudden cloud cover. Switching from solar to battery power took just over 2 seconds but involved shedding load by disconnecting about 30 houses for 0.3 second, demonstrating a downside to 100 percent penetration.
This example illustrates the type of information that will come through simulation and live experiments on the Los Alamos-NEDO project. That information can then make its way into a cost-effective control system for balancing solar power fluctuations on a locally run grid.
More Power to the People
Chertkov explains the importance of local control, "In computer science we've learned that it's efficient for different processors to solve different scales of the same problem. Similarly, smart grid control needs to be distributed over many scales. If counties like Los Alamos are nodes in a huge grid, and each node balances its own power fluctuations, then whatever fluctuations remain could be taken care of at a larger scale. But without the smaller-scale control, fluctuations can grow and lead to rolling blackouts. If we solve the control problems, distributed power generation brings so many benefits—a healthier environment, a more-stable grid, more local control, more self-sufficiency, and less risk of rolling blackouts."
"Many locations have the potential to incorporate renewable energy into their mix of power sources," adds Jonietz, "but they lack the tools to do so cost-effectively. If the Los Alamos-NEDO project is successful, the Lab hopes it will result in decision-support software that could be licensed commercially and used daily by small utilities in the West and overseas."
—Necia Grant Cooper
In this issue...
- Dynamic Vision
DARHT FULFILLS ITS DESTINY
- Solar Smart Grid in the Atomic City
TEST BED FOR LOCAL CONTROL OF RENEWABLE ENERGY
EXPOSING AND EXPLOITING THE SECRET LIFE OF SOIL
Clean Air and Abundant Fuel
Shooting Rocks on Mars
Better Fuel Cell Membrane Materials