Comment

September 2001

The Generation Gap

by Anne-Marie Borbely

Distributed intelligence, under the guise of the stupid network, will empower end users.

One rainy spring weekend in 1997, David Isenberg, a researcher with AT&T Labs, wrote a sweeping analysis of the future of networked systems. Although The Rise of the Stupid Network was intended for telecommunications, Isenberg's white-collar graffiti quickly threaded its way through the energy industry, which was just beginning to undergo its own painful metamorphosis.

Isenberg noted the assumptions driving investment in the historically intelligent network of telecommunications—centrally controlled, circuit-switched voice streams—were rapidly evolving to decentralized bursts of packet-switched data. The movement of digital information, as mapped out in the Internet's peering architecture, was incubating its own network of "intelligent, user-controlled endpoints where transport responds to the needs of the data, not the design assumptions of the network."

With intelligence embedded in the networks' end nodes rather than central to the system itself, the network would become—well, stupid. No central dispatcher manages the flow of voice, data—or electricity, eventually—in this scenario. Gateways to the system itself act as network hubs, translating and organizing the incoming electron flow as appropriate to the LAN or PC (or refrigerator or compressor) it represents. And companies wedded to centrally switched revenue streams could find themselves the dubious owners of walnut-sized brains in a world of rising temperatures.

The parallels to centrally controlled electricity production, transmission, and distribution were obvious, and Isenberg's examination hit the energy industry like a chainsaw. "I felt I was reading a textbook on guerilla warfare in Virginia while the British were still marching in these neat little rows," recalled Larry Wilson, director of strategic planning for Caterpillar's Electric Power Generation division.

Redesign Energy Landscape

Literally hundreds of new companies have exploded out the portals of electricity deregulation, promising a cacophony of new products and services for the homeowner, the retail store manager, and the facility engineer. Most will fail. A few, of course, will do spectacularly well; two or three dozen more will quietly and comfortably fill their own niche; and a constant stream of newcomers will always scramble out into the sunlight. So far, the story isn't new. But behind this stream of new offerings, The Stupid Network is inexorably reshaping our energy landscape, promising to rewrite our relationship to energy—the oxygen of our twenty-first century economy—and the way we value it, forever.

Call it transactional energy, for lack of a better catchphrase. Or call it smart power or intelligent grids or a dozen other misnomers. It certainly has acquired admirers recently. It represents a growing zeitgeist among energy innovators: the swelling belief that we're headed toward a fully networked energy infrastructure in which energy is packaged according to its location value at the point of consumption (unlike the current single-lane electron highway we've built), and individual transactions will drive offerings, investments, and infrastructure rather than the older centralized system so familiar to our parents.

It is all of these and more. Your household appliances will eventually participate in the electricity distribution network to which they attach. Your office building will someday take orders from the loads it houses and enter into the fray of the "open" marketplace, seeking the lowest-cost provider for its (nonhuman) occupants' needs. And eventually, a much more autonomous, sinusoidal electricity infrastructure will replace the clickity-clacking mechanics of the current system in which human operators drive out to the local substation to reclose a blown switch.

The great challenge before us, however, is the how. We're all fairly sure we know where the edge of the known universe curves. But what steps lie between us—we'll call it Generation Now—and a future so fingertip close? We'll present the invasion battle plans forthwith.

So Twentieth Century

Generation Now: Today, stupid loads such as home appliances, computers, and heavy machinery plug in to a largely illiterate grid managed by skilled system operators. Massive coal-fired or nuclear plants raise or lower output at their own measured rate, sending smaller gas-fired generators scrambling to respond to the electronic exhalations of their gigantic neighbors. Centralized intelligence monitors the system, and operators literally pick up the phone to request a specific generator's response.

And oblivious to it all (by design), the worker bee pulling into her driveway that evening isn't going to wait until 10 p.m. to clean up last night's party because her vacuum cleaner sucks more than 12 amps. No, she'll flip the switch, and her Hoover won't care.

On this flat map a suite of new offerings piggybacks existing communication systems and distribution channels to the energy end user. SmartSynch, a wireless technology vendor, is designing energy information products for large commercial and industrial customers—the bulk of a utility's revenue.

Using SkyTel's public networks system, SmartSynch connects utilities to their customers via the wireless network, providing near real-time load control. The technology, which SmartSynch is co-developing with Florida Power and Light, will roll out by the end of 2001.

Main Street Networks focuses on rural municipal and cooperative utility markets. The company has developed iCenter, a utility customer interface used by 65 municipal and rural utilities that allows consumers to access e-mail and a number of Web sites controlled by Mainstreet and the utility. Current products connect through standard modems and twisted-pair or optical fiber, but Mainstreet is developing a family of devices under a wireless protocol as well. And Mainstreet's focus on municipal and cooperative utilities provides them with a distribution channel into 39 million meters across the U.S.

Low-power radio technology Bluetooth allows a suite of electronic devices to interconnect. Many hail it as the next step toward smart appliances: refrigerators, TVs, air conditioners, and other microprocessor-enhanced household devices that can communicate with a PC or laptop.

Technologies and applications always run the risk of becoming evolutionary dead ends, however, unwittingly steering themselves onto technosuperhighways, their own customer base becoming the stampede that trammels them out of existence. We have a long way yet to go on this theoretical road.

Translate the BABEL

Generation Next: LONTalk, BACnet, CEBus, Modbus, CMIP, DLMS, UCA, PSEM, SNMP . . . the list of communications protocols in use numbs even fluent network engineer. Supervisory control and data acquisition systems and building controls have historically operated on proprietary communication platforms as a means of wedding the customer to the vendor. But the result was more like customer revolt.

Regardless, the landscape is dotted with essentially self-literate legacy systems. For this reason, expect to see new hardware deployed to translate the Babel of building and appliance control protocols into one box, sending a common signal to a service provider. Envenergy and Enflex are two companies currently providing the black box that translates the various protocols into object-oriented frameworks. Envenergy has developed its own Broadway Framework, soon to be a published protocol to which other developers can write.

With the utility as the principal contracting agent during this time, you may also see products such as Battelle's Rooftop Controller, an after-market product attached to rooftop heating, ventilation, and air conditioning (HVAC) systems or other appliances. With off-the-shelf hardware, the controller enables off-site diagnostics or operational control. The device uses extensible markup language to create its own Web page. Combined with communication service packages that emerged in Generation Now, utilities can now enter into new contractual relationships with their customers-all the way down to our office worker bee.

Power Lines Carry New Products

And what about the telecommunications system itself? We've all heard of power line carrier technology, which enables data to transmit over low- or medium-voltage power lines. The HomePlug Powerline Alliance, an industry consortium dedicated to developing a specification for home networking products and services provided over power lines, reported significant strides by companies in the U.S., Germany, and Israel.

Indeed, Phoenix Broadband has been shipping narrowband products that transmit voice and data over 110-volt electrical wiring for more than a decade; the company has more than 7 million nodes installed worldwide. But if the industry moves in this direction, earlier protocols may not necessarily thrive. LONTalk is a power line data transfer protocol, but the technology cannot migrate across transformers.

Another idea will reemerge from its utility predawn slumber: Direct DC systems will deploy within buildings or small campuses. Direct current lost out over alternating current as capitalization of the industry moved to take advantage of the thermal efficiencies inherent in centralized mid-century power plants. But just as technology reshapes end users' options, it also reshapes their requirements.

Microprocessor-embedded appliances-recall the Bluetooth refrigerator-operate on harmonic-free, constant DC, supplied today by utility-sourced AC inverted within the appliance itself. But every energy conversion involves losses-money floating away out the electric air conditioner, the resistant heater, and the power inverter. Companies such as NexTek have developed direct DC links between inherently DC-producing systems (photovoltaics, fuel cells, energy storage) and inherently DC-consuming devices.

The energy end user drives this car.

Another Vote to Open Source . . .

Generation Flex: By the time we reach this point, processors are so small and inexpensive that gateway chip sets for Bluetooth, Echelon, and others embed into the devices themselves. A company to watch in this space is Computrols, a New Orleans-based building control company totally dedicated to open protocols. "Ultimately, it makes more sense to embed the protocol in the controller rather than some middle device which is converting the data," said Rehan Kamal, information technology strategist for the company.

"And the biggest cost of retrofitting a building today is the labor cost—running the wires. But if the building already has CAD 5 installed and you're using TCP/IP, you've slashed 30% of the installation cost alone." But this trend doesn't necessarily bode well for companies that charge a licensing fee for their operating system. As OEMs race down the now familiar slope of Moore's Law, open-sourced code could very well dominate the market.

But that brings us to another pothole on the energy info highway: bandwidth. Not pipeline capacity—at any given moment, less than 3% of the fiber currently installed in the U.S. is actually alight. But the network itself, the protocols used to move data router-by-router across the planet, is under duress.

Lou Schmidt is technology synergist for Cisco Systems and has a long history with both networks and energy systems. For Schmidt, the emerging intelligent energy landscape is analogous to the telecom industry just a few years ago, with similar challenges. "The Internet's design goal was to build a self-healing communication system, immune to mass attack; it was never intended to be so huge. It's actually a series of interconnected networks."

The two original Internet exchanges, now owned by MCI WorldCom—MAE-East in Washington and MAE-West in San Jose, Calif.—serve as access points for the packet-switched traffic of all local and regional Internet service providers. They're part of the Internet backbone (along with "tier 2" metropolitan-area Ethernets, or MAEs, in Chicago, Dallas, Houston, Los Angeles, and New York) of massive LAN switches that move all data packets nationwide.

One of the biggest threats to the continued expansion of the Internet today, noted Schmidt, is the peering protocol. "The [MAE] routing table is getting huge, and it's increasingly difficult to maintain rapid movement."

Which gets us back to the stupid network. In a traditional circuit-switched telephony network, each switching center must ultimately hard wire to all other switching centers for telephone X to reach Y at will.

In a packet-switched network, however, routing tables embedded in each gateway enable a variety of data to "stream" (more like bursts of vintage Boer machine gunfire, actually) in a variety of speeds, levels of redundancy/integrity, or encryption as specified by the receiver. And the data packet en route, not the network, determines the pathway.

"It's when you're addressing the costs of these two systems that the differences become most palpable," said Patrick Hodges, a strategic market analyst. "In 1990, both networks could transmit 100,000 bits per second for a dollar. In 2000, the same 100,000 bits had dropped to 2 cents when moved over a packet-switching network, while centrally switched network costs hadn't really come down at all."

So, energy. We still haven't addressed the problem of a billion machine-to-machine communication lines running right along with the already billowing human-to-human traffic lanes. We haven't outlined the new relationships between energy consumer, provider, and regulator that would fundamentally change the business of electricity. Well, what would representational democracy look like in machine language?

There are two fundamental regulatory changes that must take place—and will eventually happen—to free our recumbent utility industry from its regulated past. First, real-time pricing must be enacted for all energy customers, everywhere. Flat electricity tariffs hide system inefficiencies and force energy customers to unwittingly subsidize products they'd never choose in an openly competitive marketplace. At 6 or 7 cents, kilowatt-hour grid-supplied electricity seems hard to beat at the retail level. But at 4 p.m. every day, that kilowatt can rise as much as tenfold.

Secondly, state utility regulators must redefine "regulated utility." In California, an independent power producer can install energy production or storage facilities and manage the energy requirements for a suite of buildings under private contract. Performance-reporting procedures, liability for loss of service, and other stipulations are contracted also.

But if an electricity distribution line between buildings crosses a public road, your district energy loop is now a utility, subject to the same reporting procedures, liabilities, and oversight at the state level as Southern California Edison. The very regulatory framework designed to protect consumers now snuffs out new business models.

Two-Way Price Game

Generation Apex: The market leap between Generations Flex and Apex marks the edge of the known universe for electricity transactions. With the advent of real-time pricing, in this phase we'll probably see the emergence of a multitiered hierarchy of intelligent devices that negotiate within their own closed-loop auctions for their individual power quality and capacity needs. From that, a single IP address will enable the facility to negotiate with its energy service provider's community of facilities; the ESP will in turn broker a deal with the network system operator.

But by this stage, it's a two-way game. Price or voltage signals may cascade backward down the communication link, causing HVAC systems to cycle without the building's occupants ever noticing a change in temperature. On-site power management systems will include a variety of energy production and storage technologies, all "bidding" into their own auction along with the grid. The lowest-priced provider wins, and power quality management now happens on the customer's side of the meter.

Security gets to be a central question at this point. Mike Cruse, chief technology officer of Envenergy, is quite familiar with it. "We wrote the communication protocol for a Generac system. They had a spare DSL line, which they hadn't before used. Three days after implementation, hackers from three different countries had already tried to break in."

If this sounds like science fiction, AESC, Inc. is co-developing intelligent agents for energy networks with Recticular Systems, Inc.

"We're developing agencies that talk to one another, that represent the interests of the site [generation assets, curtailable load, and the like]," said Jerry Gibson of AESC. "The agents collaborate. We use an option format. Scheduling requires a great deal of local knowledge, understanding the requirements and characteristics of the site's assets. So if you're looking for total system optimization, when you scale everything up via centralized decision making, the situation becomes intractable. The technical problem is the architecture of the decision-making logic itself, not the movement of data through the wires."

In a single unit, the company employs an agent builder tool set and handles operations through rules at a basic level. Java copes with the higher-level functions.

Companies already positioning themselves to perform the stockbroker function include Tridium, Sixth Dimension, and Silicon Energy. Each has significantly different approaches to the marketplace, different software schemes, and some work in more than one phase of this evolution. But they're already here, trying to climb up the flat-earth regulations that currently affect new energy transactions.

Autonomous Agents, Networks

But what lies beyond the farthest shore? How far can we go, rationally, with some semblance of the installed base for electricity distribution?

In Battelle's bid to embrace all dimensions of the energy landscape in a hierarchical dynamic valuation stream, mathematicians working on molecular systems enlisted to consult with electrical and networking engineers. Months of (sometimes heated) debates laid out the framework to approach complexity in energy networks from a totally new perspective. In the predawn of a desert spring morning, they emerged from their last session of petri networks and molecular automata, bleary and inspired. "We couldn't wait to set this in motion," recalled Chassin.

And they're not alone. Santa Fe Institute's Bios Group is wading into the field of autonomous agents and energy networks. MIT, Sandia National Laboratories, and a host of other research organizations are all groping for the next step. How this will all play out is still the Great Unknown. But if I were a gambler, I'd place $100 on the clockmaker. IT

Figures and Graphics

• Breaking communication into packets lets many users share the same data path in the network. Packet switching doesn't require a dedicated line (streaming video is an exception).
  
• Circuit switched is a type of network in which a circuit, or physical path, is obtained for and dedicated to a single connection between two endpoints in the network.
  
• The generation gap (pdf version)
  

Sidebars

• Here there be dragons
  

Behind the Byline