• Key drivers for smart-grid electronics are shaping up to be security, antitampering measures, and accuracy.
Freescale e-metering reference design details, www.freescale.com/metering
Texas Instruments e-metering processors and applications, tinyurl.com/yj2va6v
Whirlpool Corp., www.whirlpool.com
A quick primer on other developments in the smart grid, tinyurl.com/y8a42fs
Late last month, appliance maker Whirlpool, Benton Harbor, Mich., got a big check from Uncle Sam. It wasn’t a tax refund. Rather, it was part of $34 billion in federal funds designed to accelerate deployment of smart electric grids.
As part of the deal, Whirlpool will get $19 million to design and deploy a million networked clothes dryers that can respond to fluctuations in grid demand and pricing. It plans to make all its appliances capable of connecting with and responding to smart-grid networks by 2015, as long as suitable standards and policies are in place.
Equipment coming out of this effort will make use of embedded electronics that can sense electrical loads, keep track of demand curves, and dial back power needs if grid conditions get overheated. The development that makes such measures possible is the smart electric meter, or e-meter for short. Proposals funded in the stimulus package include plans to deploy about 18 million e-meters, some of which will act as home-communications hubs.
Appliances and other devices hooked to smart grids must contain circuitry that carries out e-meter functions, on top of managing the control tasks associated with normal operations. It is interesting to see how the architecture of e-meter-equipped devices is likely to change as the grid gets smarter. Warwick Sterling, Whirlpool global director of energy and sustainability, explains that the appliances most easily updated with e-meters are those that are already pretty smart. “Our premium dryers have controls that can handle metering functions with just a change in software. But the majority of dryers don’t have controls sophisticated enough to do this in software. They’ll require hardware modifications on their control boards,” he says.
Sterling says e-metering and smart-grid features will demand changes in two key areas, communication/networking facilities, and in the components that use most of the energy. The first smart-grid dryers will have discrete communication modules that plug into the main controller boards. Communication facilities will be in the form of modules, rather than built onto the controller boards themselves, because standards for grid communication are still in flux. Moreover, different communication standards are favored in different parts of the world. So makers of smart-grid-connecting devices will likely hedge their bets via the plug-in route.
The effect of the smart grid on appliance components that consume a lot of energy is more involved. Some parts of the devices may need a redesign to give more options in how much energy gets consumed. In the case of dryers, for example, “Many of our dryers have only one heating element. We might modify the heating elements so they can be turned on in stages, rather than just be on/off,” says Sterling. “It’s possible we’ll go to two or three-stage elements or a variable-heating element. Alternatively, we might incorporate the option of kicking the heater on initially, then cutting back and letting the clothes tumble. This would extend the cycle, but we’ve found you use less energy this way and clothes don’t wrinkle.”
And the use of energy-saving features such as staged heaters will be an option that is strictly up to the user. “We think the consumer has to be able to override any energy-saving features,” says Sterling. “If the dryer offers to delay a cycle because it takes place during a peak usage period for the grid, the consumer who wants dry clothes immediately would be able to set the controls to ignore this option.”
Whirlpool figures such scheduling flexibility could do more than just lower utility bills. “You could tie the operation of appliances to the times when the grid is using sources of renewable energy,” says Sterling. “If the wind kicks up at night, for example, it might be a good time to run the dishwasher. Plus, sometimes you can schedule tasks for times of off-peak demand and the consumer won’t notice. Refrigerator defrost cycles fall into this category.”
Obviously, consumers don’t want an additional part-time job setting and resetting the energy-demand schedules for their appliances. Whirlpool envisions a common set of controls compatible with essentially all home appliances so consumers needn’t continuously climb a learning curve. Toward this goal, today appliance makers are working with utilities and NIST (the National Institute of Standards and Technology) to develop a common set of interface standards.
Smart-grid features will come at a cost to consumers — Whirlpool figures they add as much as a 10% premium to the appliance cost, though there will likely be savings for consumers as well. “Depending on how pricing at peak demand goes, the consumer might save $30 to $50 per year using smart-grid features,” says Sterling
Of course, e-meter functions won’t just be found in individual appliances. The Automated Metering Infrastructure (AMI) for smart grids incorporates the idea of utility meters at every home and office building to collect information about the load demand there. These would pass such information back to data aggregator nodes on the grid that would analyze conditions in real time to figure out, among other things, the location of demand hot spots. Aggregators would also predict short-term usage based on historical patterns. These devices would, in turn, interact with network sensors called phasor-measurement units. PMUs constantly sample the voltage, current, and frequency of power flowing on the grid. Using embedded GPS modules to synchronize their measurements, PMUs and data aggregators will deliver a real-time analysis of electric grid conditions.
Meter making — Out with the old, in with the new
Two coils act on the metallic disk. One coil is connected so it produces a magnetic flux proportional to the line voltage, the other generates a magnetic flux proportional to the current flow. A lag coil delays the field of the voltage coil by 90° to produce eddy currents in the disk to exert a force proportional to the product of the instantaneous current and voltage. A permanent magnet exerts an opposing force proportional to the speed of rotation of the disc. These two opposing forces act to keep the disk rotating at a speed proportional to the power used. The disk turns a register mechanism which integrates the speed of the disk over time by counting revolutions to move dials that read out the total energy used over time.
Solid-state meters take a different approach. The power sensor typically consists of either an extremely low-resistance resistor in series with the line or a current transformer. As indicated in the accompanying drawing of a typical e-meter circuit from Freescale, the sensed signals go to a processor that combines analog/digital conversion with a real-time clock and display driver. The processor performs the integration operation that computes dissipated power. Besides handling e-metering, the processor might double as a controller when installed in household goods such as dryers and washers. Communication protocols for the smart grid have yet to be standardized. So at least the first round of smart-grid devices will employ a separate communication module, sending smart-grid data either using wireless protocols or a power-line modem.
It looks as though the embedded processors used to devise PMUs and grid-data aggregators will be close relatives of those providing e-meter functions in appliances and utility meters. “You typically don’t need much more than an 8-bit processor for stand-alone metering,” says Jeff Bock, senior manager of industrial global marketing at Freescale, Austin, Tex. “But designs are moving to 32-bit systems because of the communication involved.” He says the first generation of smart-grid appliances will likely employ 8-bit controllers and separate communication modules containing 32-bit machines. Eventually, the two will merge onto one control board.
Developers of AMR gear tend to be concerned about building in more accuracy for readings than has been available with conventional electromechanical devices. The international standard has become class 0.5 — for readings of 99.5% accuracy — compared to old electromechanical units giving class 2 or 3 at best. “Energy is money, so the more quickly and accurately you calculate energy use, the greater the meter’s return on investment,” says Bock. Limitations in analog integration have been one factor preventing accuracies this high. Freescale recently developed a processor targeting smart-grid applications which overcomes the problem by integrating the signal chain on the processor chip. “We think with optimization we can get to class 0.2 or even better,” says Bock.
E-metering designs must also incorporate measures to prevent tampering. “As much as 50% of the energy generated in India is lost to theft,” explains Bock. The need is also acute in countries such as China, where many energy users are on a pay-as-you-go billing scheme. Historically, the main safeguard for meter tampering has been mechanical interlocks that shut off power if someone opens the meter. Accelerometers have been built-in to detect when meters are jarred or tipped upside down.
Newer metering schemes add more-sophisticated measures. One involves a real-time clock to note the time and date of any tampering. Any time the backup-battery connection for the e-meter electronics gets cut — one of the first things energy thieves do, says Bock — the system records evidence of the tampering in a way that can’t be reset or altered.
Makers of processors for e-metering applications also expect to see the industry transitioning toward metering functions that can be altered remotely by beamed-in commands, prompting them to build in software that will ease the transition to this operating mode.
And as has become common practice, embedded-systems designers must devise new ways of minimizing the power their electronics dissipate. “We are being pushed to the point where standby current for appliances such as washing machines must be less than a watt. That is demanding design changes in areas we never considered before,” says Bock.
Another concern for e-metering applications centers on security measures to prevent inappropriate use of real-time data collected for characterizing the grid. “You don’t want your neighbor monitoring your energy consumption,” says Jennifer Barry, product-marketing engineer at Texas Instruments Inc., Dallas, Tex. “For TI’s MSP430 metering MCUs, security, wireless capabilities, and handling the numerous communication protocols are the top issues right now.”
TI sees many developers adding smart grid-type features initially with submeters, basically devices that function like a smart extension cord. Household items would typically plug into the submeter, which would, in turn, both monitor their energy use and turn them on and off. Submetering makes sense because, “You don’t want to force consumers to change everything to implement smart-grid features,” says Barry.
TI also says developers working in e-metering applications seem to have a few common difficulties to overcome. “The main challenges for designing these systems are on the software side. There are several different software stacks and protocols involved that have to be verified. But the key right now is in bringing the cost down of the items surrounding the processor. The low-power aspect of the system is going to be important because you want to make sure the electronics still run in the event of power outages,” says Barry.