The east coast office of GreenMountain in Cambridge, Massachusetts, has a reasonably impressive array of electrical meters. They're all Elster A3 "Time-Of-Use" meters, meaning that they record not just the amount of electricity used, but also the time it was used. In theory, the utility can then use a rate structure to encourage their customers to use less electricity at peak times. The meters have the typical ANSI optical port, so you can hold an ANSI reader up next to the meter and read data locally. The meters in our building also have Itron's 50ESS ERT radio transmitter, which is a separate module that slides into the side of the stock A3 meter. I'm not sure how NSTAR (the local electric utility for most of the towns around Boston) is actually reading the meters, but they're probably driving around with vans containing short-range radios to read the data. That's about as smart as our grid gets in 2009.
I first worked on smart grid hardware in 2002. (We didn't call it "smart grid" back then.) Now, we'd say that I was working on AMI, "automatic metering infrastructure." At the time, I worked on a line of power distribution units for data centers.
We used an embedded processor with no operating system, just our own firmware written from scratch in C. Early on, we decided to use a proprietary TCP/IP stack; the next closest competitor was an IC called "i2chip," which implemented TCP/IP entirely in hardware and cost $12 in quantity 100. (This is now the Wiznet W3100; I believe in a box somewhere I still have a sample from the first batch of silicon the gentlemen from Wiznet made.)
Our web-based monitoring interface used Javascript in a hidden iframe to refresh the interface automatically without reloading the entire page. We measured current with PCB-mounted current transformers; I remember looking at datasheets for the first high current Hall effect sensor ICs and thinking that our upgrade path had finally been invented.
Less than a decade later, the same system would hardly be recognizable. Our latest remote monitoring deployment (actually for use in the photovoltaics industry, not smart grid) uses an entirely open source stack. We run Python on Linux with the Yahoo User Interface library to make the front end smooth. That all the components are open source is a sign that we're working in a domain where the benefits of interoperability outweigh the costs of open development.
I suspect that the same transition to open technologies developed to interoperability standards will occur in the larger world of smart grid, but it needs help to happen. The development of such standards has been underway for years, but since NIST received a mandate to coordinate the development of smart grid interoperability standards in the Energy Independence and Security Act of 2007, it has accelerated. In a roadmap workshop in late April, 2009, NIST released the first draft of the list of smart grid standards they will recognize in their framework. The revised release is expected at Gridweek this Thursday.
NIST's list gives only a few words to explain what the standards are, so for those of you new to the industry, here's a summary with a little background on each item.
There are 16 standards on the list so far, but the upcoming revision will probably add more. The links that start each section point to the canonical source. You'll notice that the Open Smart Grid subcommittee is a big player.
AMI-SEC: A task force was formed in 2007, but they haven't released a standard yet. The standard will relate to security for advanced metering infrastructure, i.e. how do you prevent miscreants from sniffing your meter data, while still allowing the power company to do it? At first blush, this seems like an inconsequential problem, but it's not. For example, if you wanted to figure out whose house was ripe for burglary because they're on vacation, and you could read data from electric meters, you could do it with a few trips through a neighborhood with a van that records usage data and compares it to the past. Hey, looks like the folks in Suite 100 of 5395 Pearl Parkway aren't home; let's rob them!
ANSI C12.19: This is a standard that describes structure of data that can be transferred by the ANSI optical port I mentioned above. The standard was originally released in 1997; a revised version, ANSI C12.19-2008 was released last year. This standard isn't likely to play a big role in smart grid because optical ports are slow and short-range.
BACnet: BACnet is a standard that specifies a communication protocol for building automation and control. The standard was created by ASHRAE, the HVAC experts, in 1995. The most recent release is ANSI/ASHRAE 135-2008. By now, there are open source implementations of BACnet in at least three languages: C, Python, and C#.
DNP3: A protocol for communication between SCADA equipment at electrical substations. If you're interested, the DNP User's Group has a good primer.
IEC 60870-6/TASE.2: Protocol for communication between electric utilities.
IEC 61850: Covers automation within individual substations.
IEC 61968/61970 These standards define APIs for use by utilities. The former is for automatic meter reading, the latter for energy management systems.
IEC 62351 Parts 1-8: A security standard for power systems, published in eight parts. The first 6 parts have been released; part 7 is expected in 2010; and part 8 is aimed at late 2013.
IEEE C37.118 This standardizes communication about phase information between different parts of the grid. The title is "IEEE Standard for Synchrophasors for Power Systems." The word "syncrophasor" refers to the concept of using absolute time references for phase information, rather than just assuming that two systems are operating at exactly 50.0 or 60.0 Hz. The standard also mandates minimum reporting frequencies. The standard is managed by the IEEE Power System Relaying Committee Working Group H11. NIST would like to see IEEE C37.118 harmonized with IEC 61850, mentioned above, which covers phase information relaying within substations.
IEEE 1547 Standard for connection of distributed resources, meaning small, non-centralized power producers like solar panels or wind turbines (though they need not use renewable sources of energy), to the grid.
IEEE 1686-2007 Security for the ill-named intelligent electronic devices (IED's) in substations. This standard is produced by the IEEE Power Engineering Society Committee on Substations. (Alarmingly, the password used to secure the committee's "Protected Files" is stored in plaintext in the webpage source code, and the files are hidden with only an obfuscated URL. Perhaps even more alarmingly, the only document they've seen fit to secure is an announcement about a Halloween costume party. And they're generating a secuity standard?)
NERC (North American Electric Reliability Corporation) CIP 002-009: Another security standard, CIP 002-009 lays out 8 different topic areas that must be addressed when setting up
network equipment for bulk electric systems. These range from what
hardware can be used, to how to train personnel, to physical security
of the site. For the purpose of the standard, NERC has defined the "Bulk Electric System" to be the electrical generation resources, transmission lines, interconnections with neighboring systems, operated at or above 100kV.
NIST Special Publications SP 800-53 [PDF] and SP 800-82 [PDF]
These NIST special publications are also security standards. The first, SP 800-53, makes recommendations about security controls for computer systems used by federal executive agencies. The second, NIST SP 800-82, is broader, covering industrial control systems. It's not specific to smart grid, though certainly all grid systems use industrial controllers. Further, the standard does list the SCADA controls of the grid as the primary example of an industrial control system that needs security.
Open Automated Demand Response: OpenADR is a communication protocol for transmitting demand response signals from utilities to electric customers' appliances, which presumably will react to use electricity at off-peak times. The refrigerator that Tendril announced today with GE at GridWeek uses price signals to figure out when it should make ice (at night, when it's cheap); based on an earlier press release, I'd bet it's using OpenADR.
OpenHAN: First released in 2008, OpenHAN describes a framework for secure communication over a network with consumers' devices to implement, among other features, load-shedding and metering. The acronym HAN is an unpleasant bastardization of the already-odd "local area network" (LAN) into the truly nonsensical "home area network." (Not as bad as the bizarre SAN, "storage area network.")
ZigBee/HomePlug Smart Energy Profile Zigbee is a standard for wireless communication based on IEEE 802.15.4. Implementors must pay a fee, not just for a copy of the standard, but $3500 annually for a membership in the Zigbee Alliance, plus an additional fee for each Zigbee product released. The smart energy profile allows wireless devices to adjust to price signals and deliver usage statistics, among other features.
Those are the 16 NIST-approved standards we've got so far. We'll see what Commerce Secretary Gary Locke announces on Thursday.
