Smarter Charging

Alec Brooks, February 5, 2008

Ten years ago when I was driving an EV1, one of the best perks was free parking and charging at LAX. I would pull into one of the EV parking places by Terminal One, plug in, and go on my trip. The charger would turn on right away – typically this would be in the middle of the day – and the battery pack would be fully charged even before my plane had left the ground. When I would return a few days later, typically at night, the battery pack would have partially self-discharged and was cold soaked, having sat for days with the charger connected but off. I always thought that surely there would be a smarter way to charge.

My charging setup at home was a little better. A mechanical swimming pool pump timer was installed between the charger and the 240-Volt power in my garage. It was set to turn on when the time-of-use electricity price dropped to the off-peak rate late in the evening. This was a simple setup, but not very flexible. The mechanical timer had to be adjusted twice a year for daylight savings time. With experience, I found that the best time to start charging was not late at night – rather it was in the early morning hours so that the charge would just be finishing as I got in the car to go to work. The battery pack would be fully topped off and not be cold soaked as it would be if charging had completed at around midnight. This seemed like a smarter way to charge, but what would happen if millions of vehicles needed to charge? Should there be a way to coordinate all the vehicles to minimize the impact on the power grid – or maybe even provide a benefit?

The current interest in plug-in vehicles in the United States is driven by three principal considerations: local emissions, energy security, and global warming. In the United States, electricity is largely sourced and produced domestically and is not produced in any significant amount from imported oil. Electricity is produced in a variety of ways, each method with its own set of issues and and impacts. Coal is used for more than half of electricity production in the US. Even though coal-based electricity has relatively high greenhouse gas emissions, plug-in vehicles that are charged with electricity made from coal still have relatively low greenhouse gas emissions. A Toyota RAV4 EV operating on coal-derived electricity would have effective greenhouse gas emissions of about 240 grams/mile, about the same a as a gasoline powered car that gets 46 mpg. But a desirable goal is to get to transportation with zero greenhouse gas emissions. Plug-in vehicles are synergistic with this goal and can actually be an enabler of increased amounts of renewable energy on the electricity grid. Wind energy is especially promising as an energy source for vehicles. One of the negative attributes wind energy is that it is intermittent – wind energy is generated only when the wind is blowing and the amount and timing of wind generation varies from day to day (see the chart below for an example of the variation of wind generation in the PG&E service area for April 2007). Integrating wind energy into the power grid requires having sufficient conventional sources of generation available to be able to compensate for the day to day variations in the availability of wind energy. This adds cost and to some extent limits the maximum percentage of total energy generation that can come from wind energy.

The operation of an electric power grid involves continuously assuring in real time that the total amount of generation matches the total load. If there is a mismatch between generation and load, the frequency of the grid will start to deviate from the nominal value of 60 cycles per second (60 Hz). The power grid in the United States is composed of three main regions: Western, Eastern, and Texas. Within a region, the power grid is interconnected with alternating current (AC) transmission lines. The grid frequency is the same throughout an interconnected region. Regions are further subdivided into control areas. Each control area manages the generation of electricity in their area and interchanges of electricity with other areas through a grid operator (such as the Cal ISO in California). Each grid operator schedules generation in advance to match up with expected loads, and then in real time fine-tunes the level of generation to match to the actual load within its control area. It is a constant balancing act, performed 24/7.

Most uses of electricity are needed pretty much in real time – lights, air conditioning, computers, etc. The electricity has to be generated at the exact moment it is used. But plug-in vehicles are a fundamentally different kind of electrical load. Plug-in vehicles draw energy from the power grid and store it for later use while driving. Plug-in vehicles are typically plugged in for significant periods of time – usually all night – and it doesn’t really matter when the energy is transferred to the vehicle, as long as a certain amount of energy is transferred by a specific time, typically in the morning. This leaves a lot of flexibility as to the timing and exact nature of how plug-in vehicles are recharged. As plug-in vehicles grow to become a very significant fraction of the overall load on the electrical system, this flexibility can be utilized with beneficial results for the vehicle owner and for all electricity customers. What is needed to take advantage of that flexibility is control over the exact timing and rate of a vehicle’s recharging. With communication between grid operators or utilities and plugged-in vehicles, recharging could be controlled to match up to the amount of renewable power being generated at any given time. This turns the typical electricity delivery model upside down for the grid load represented by plug-in vehicles; instead of controlling generation to match the load, the load represented by the sum total of plug-in vehicle battery chargers could be remotely controlled to match the availability of renewable energy generation (such as wind power). (This same approach could also be used to make hydrogen for fuel cell vehicles, but overall this is far less efficient, requiring three to four times as much electricity per mile as compared to plug-in vehicles). This remote control could be implemented through a secure internet connection to vehicles when they are charging – which can be implemented with a variety of available technologies. So with this communication, grid connected vehicles could be controlled to have their total aggregate charging power matched to the availability of wind power each day.

In the longer term, there is the prospect that a large number of plug-in vehicles (in the millions) could enable a significantly higher penetration of intermittent renewable energy sources onto the electric power grid. The current thinking is that intermittent renewable energy sources need to be backed up with other forms of generation in order to provide reliable power to serve the loads on the grid. However, with a large number plug-in vehicles being remotely controlled to match their power draw from the grid to the availability of the intermittent resources, those intermittent resources will not have to be backed-up as much with conventional generation sources. When the wind doesn’t blow as much, vehicles won’t get as much energy. If EV battery packs are big enough, daily charging to full capacity won’t usually be essential, and plug-in hybrids can use their liquid fuel on days when intermittent resources can’t provide all the energy needed. In this way, plug in vehicles can use truly zero-carbon electricity for most miles driven.

One of the disadvantages of intermittent resources like wind and solar is that these sources cannot be controlled to support the frequency stability of the grid. Conventional generating plants have a governor that adjusts power output automatically in response to grid frequency variations from 60 Hz. If the grid frequency drops below the nominal 60 Hz power generation is automatically increased. A typical ‘droop characteristic’ set point for power plant governors is that a 5 percent change in grid frequency would cause a change in output of 100% of a powerplant’s rated output. Of course the grid frequency doesn’t usually vary by as much as 5%. A more typical frequency variation might be about 0.1%; a 0.1 percent droop in frequency to 59.94 Hz would cause the governor on a powerplant to increase power by 0.1/5 or 2 percent of rated output. However, wind and solar energy sources cannot produce more or less power in response to changes in grid frequency; power generated depends on how hard the wind is blowing or how bright the sunlight is. But the same effective governor function can just as well be achieved by modulating loads in response to changes in grid frequency. Plug-in vehicles whose charging power is remotely controlled to match the availability of intermittent resources can further be controlled to provide this governor function. The net result would be that the cumulative recharging power draw of all plug-in vehicles would be varied over a long time constant to match the availability of intermittent resources, and superimposed on top of that would be a shorter-period variation to provide the governor function for those intermittent resources. The figure below shows a notional example of a wind power generation profile and the net recharging power profile of remotely-controlled plug-in vehicles.

Ancillary Services and V2G

Grid operators use a variety of tools to keep the grid operating smoothly. These tools are commonly referred to as “ancillary services”. Some examples of ancillary services are spinning reserves, non-spinning reserves, and regulation. Ancillary services are essentially contracts to provide the ability of the grid operator to vary the amount of power being generated. The various types of ancillary services have different requirements for how quickly the change in power generation must be made. Many grid operators, including the Cal ISO, maintain markets for ancillary services. In California, powerplant operators submit bids for ancillary services on day-ahead and hour-ahead markets. Bids typically include the hour the service will be offered, the amount of the service offered, and the offered price for the service. The Cal ISO evaluates all the bids and determines a market-clearing price for each ancillary service for each hour of the day. The market clearing price is the price point at which the requisite capacity of that particular ancillary service has bid at or below. Each winning bidder is paid the market clearing price for providing the service.

Grid ancillary services are now provided by powerplants, but loads – such as plug-in vehicles – could provide ancillary services just as effectively as powerplants. To the grid operator the effect on the grid looks the same. It is likely that vehicles can provide these services much better than powerplants. Powerplants have limitations on how fast they can change power levels. Vehicles can change power levels virtually instantly. In April 1997, the Federal Energy Regulatory Commission (FERC) formally recognized in FERC order 890 that loads should have equal standing as powerplants in providing ancillary services.

So the potential is there for plug-in vehicles to provide grid ancillary services while they are charging. These services have real monetary value – value that can go toward offsetting some or possibly all of the cost of the electricity itself. The exact value varies with many factors including supply and demand, time of day, season and other factors. The size of the ancillary service market is however limited – for example, all of the regulation ancillary service in California could be performed with about 30,000 vehicles, but the potential for the number of plug-in vehicles in the state is in the millions. But if the amount of intermittent renewable resources grows together with the plug-in vehicles, the overall matching of vehicle charging power to the availability of intermittent resources and providing the governor function as described above will have real value and result in lower costs for electricity used to power these vehicles.

There has been a lot of interest in a related concept called Vehicle-to-Grid, or V2G. The notion with V2G is that vehicles can act as energy storage resources for the grid and feed power from the vehicle back to the grid on demand. The bidirectional power flow offers the potential to provide an expanded level of ancillary services to the grid as compared with a vehicle that only draws power from the grid. The potential is there to provide services whose monetary value substantially exceeds the cost of the electricity used by the vehicle. FERC commissioner Jon Wellinghof is a big support of V2G, and has coined the term “Cash Back Hybrid” to describe a V2G-enabled plug-in hybrid. I was involved in an early V2G demonstration project in 2001 when I was at AC Propulsion (final report here).

Tesla’s V2G Strategy

Test setup with VP10

Tesla’s initial approach to exploring V2G is to focus on ancillary services that can be performed with the vehicle operating as a grid-controlled load, rather than as a system capable of feeding power back to the grid. This approach has many advantages for the initial rollout of V2G. First, it eliminates the interconnect issues around feeding power back to the grid. It is of course technically feasible to make a safe and certified bi-directional charger with the same kind of anti-islanding and other safety features employed in small distributed generation systems. However, state laws and individual utility policies may currently preclude feeding power back to the grid from anything but solar and wind energy systems. Second, battery wear and tear due to bi-directional power cycling is not fully understood and could potentially have a cost impact greater than the benefit produced. More research is needed to quantify this aspect. Third, storing energy in a battery and then discharging it back into the grid results in energy losses due to the conversion of AC to DC in the charger, throughput losses in the battery, and then from DC from the battery back to AC. The cost of the energy needed to make up for the energy losses offsets some of the value created.

Tesla has taken initial steps to explore charging-only V2G in partnership with Pacific Gas and Electric (PG&E news release here). In a project completed in December, 2007, Tesla and PG&E developed a battery charging power profile derived from actual Cal ISO ancillary service dispatch data and set up a wireless connection to Tesla Roadster VP10 to enable remote control of the power drawn by the Roadster’s onboard charger. The tests consisted of sending charging power dispatch commands to the vehicle at 4 second intervals and monitoring the response of the vehicle charger. The system performed as expected with response much faster than any powerplant could achieve. The figures below show the test setup, the profile of charging current vs time and a detail of the vehicle response. Note that the shaded area shown in the charging profile is proportional to the energy drawn from the grid. Possible next steps could include outfitting a few of our customers’ Roadsters with this capability for a field trial. Another area of exploration is to enable a grid frequency-responsive charging rate to perform the governor function mentioned above. An advantage of this approach is that little or no communication is needed to the vehicle; the vehicle charger can directly sense the grid frequency.

Power profiles tested, note zero AC charging current is at the top of the scale.
The shaded area is proportional to the energy transferred.

Vehicle Command and Response


Vehicles that plug in to the power grid for some or all of their energy needs can make valuable contributions to the production, transmission, and distribution of electric power. Plug-in vehicles, both battery electric and plug-in hybrids, will principally be charged at night when there is ample generation capacity. By increasing overall electricity consumption without having to increase the electricity infrastructure, fixed costs will be spread over a wider base, reducing electricity costs to all electricity customers. Plug-in vehicles will also become a new resource to help with operations of the grid. The energy storage capacity in the batteries of a plug in vehicle can be a storage resource to the grid, and vehicle charging rates can be controlled remotely by utilities or a grid operator to perform ancillary services for the grid. Since the load represented by plug-in vehicles could be dispatchable remotely, the penetration of intermittent renewable resources such as wind and solar energy can grow beyond the level that would have been practical without plug-in vehicles. In the future, the current grid power delivery model of dispatching generation to match load can be inverted for a growing fraction of the total load: dispatching load to match available renewable energy generation. Plug-in vehicles will be a key enabling new load that supports a cleaner, more renewable, and lower-carbon grid.


University of Delaware V2G research group
California Independent System Operator