What buildings have to say to smart grids

The challenges of smart energy are well known. How can we as a society based on cheap plentiful always available energy, adapt to shortages, intermittent availability, and a continuing shortage of capacity to move energy from where we make it to where we want to use it. Local shortages and outages will become the norm, although local surpluses might create greater challenges. Most importantly, how can we adapt without abandoning the life-styles that we enjoy, and that we hope our grandchildren can as well.

The national Priority Action Plans (PAPs) for smart grids and smart energy aim to accelerate development of enabling specifications for smart energy. Most look solely to the internal operations of the electrical grid itself. These activities, while important, can only enable the innovations we need; electrical grids themselves will not be the basis the most important changes.

Others PAPs look to extend grid operations into our lives. PAP11, for example, looks to control and track personal vehicle use in service to the grid. We are not willing as a society to assume our position as cogs in the machine. Utilities and policy makers see this as the happy efficiency of well-ordered future; the public instead sees the dystopian factory of Chaplin’s Modern Times. Perhaps we should be glad that while other action plans focus more on openness and change, this activity is being developed under a veil of trade secrets and inside pool that will only speed the early failure of its model.

The newly formed PAP17 launched the ASHRAE SPC (Standard Project Committee) 201, Facility Smart Grid Information Model. SPC 201 offers a consumer-centric model that can support the rapid changes in the ways we use and manage energy. The focus is on how the systems in buildings can interact to create what grid operators call Distributed Energy Resources (DER). Building-based DER addresses the intermittent shortages and outages of the smart grid directly.

Traditional grid-building interactions use direct control. Turn this off, turn this on, to support the needs of the grid. PAP17 assumes the economic communications of price and availability developed by the market-oriented PAPs, and considers what a building needs to communicate internally so it can be mature market participant.

If each building has its own portfolio of DER, sun, wind, perhaps limited pump storage offered for voltage regulation, batteries, ice, load shedding….That building may use a different suite of internal responses each time it sells a response to the grid. Market participation becomes based on reliably producing a change in power use rather than turning on and off a device. If SPC201 fails, it will fail by failing to embrace this economic model, and letting its engineers revert to a model of direct dispatch by the grid.

A distributed energy resource (DER) may be:

1. A private asset of the building, used only for the buildings purposes, perhaps when the grid is unavailable, and not revealed to the grid at all
2. A component of a building’s demand response, so turning off the chiller or turning on the generator are indistinguishable to the grid
3. An intermittent asset of building with availability characteristics which are *may* be revealed to the grid, i.e., the grid operator may contract to know whether it is sun or wind, so the operator may better estimate when it can be relied upon
4. A building asset that happens to be operated by a third party. That party *may* happen to be a traditional player, perhaps one called a “utility”. It could just as easily be an ISO, depending on scale and location. Or it could be some new form of energy service provider
5. Owned by the building but effectively leased to the grid operator, and treated as a forward deployed asset of the grid
6. Owned and operated by a third party and used as a forward deployed asset of the energy services provider

(4, 5, 6) are the ones that look like direct dispatch as we understand it today—but they need not be. Whether the grid sends market signals that are flash traded to negotiate individual contracts for use of DER, or whether that contract is pre-executed, actual use of that DER is best thought of as a call for performance on the contract. Third party service providers will pay better for guaranteed response, and will demand greater penalties for non-performance when a high service level was promised.

During the National Roadmap efforts in 2009, we used the catch-phrase “Every end-node is a microgrid”. A microgrid is responsible for meeting its own needs and purposes by managing its own energy use, generation, storage, recycling, and market operations to deal with surplus or deficit. Note that the market operations are last. Microgrids are defined recursively (as per Galvin): A suite in a building or a production line could be a microgrid. The office park or campus could be a microgrid containing the building microgrids.

At ConnectivityWeek last May, heard a speaker from the DOW described his big goal that every building be able tolerate 8 days of grid down-time with no loss of amenities. Contemplating this requirement suggests what a poor partner the smart grids will be. SPC201 is reaching toward the information sharing that equipment and systems in buildings and homes will need to support us despite smart grids.

If there is a flowering of Green-Tech, it will come from consumer based markets that can tolerate rapid innovation and change. Those markets will require low integration costs based on loose coupling and energy information sharing.

Moving beyond Demand Response (DR) – Pricing Services

Utilities and Regulatory commissions are obsessed with demand response (DR). All want to know how to get more of it. I could, with little effort, attend a national conference on DR every week. A large share of the standards priorities of the National Institute of Standards and Technology (NIST) to support smart grids support DR. And yet, almost everyone recognizes that DR is a short-term solution. Plans are just now underway to move beyond DR.

The most expensive electricity comes from the dirtiest generating facilities used for only a few hours a year. If consumers would use less energy, i.e., reduce demand, in just those few periods, then those expensive dirty plants could be turned off permanently. To do this, electricity suppliers need to anticipate when those moments are coming and take steps to reduce demand. We call this Demand Response.

At its simplest, DR is just turning things off. Rolling black-outs are the simplest form of DR. They make consumers very unhappy. Utilities have worked for years to improve on this model through direct load control. They have been installing remote switches on home heat pumps since the ‘70s. Today, they are developing SEP to control homes device by device using software installed in smart meters. Consumers like it in off-months, when they get a bill reduction and the utilities do nothing. In summer months, when the utilities do something, things get turned off in the home. They make consumers very unhappy.

In the commercial building world, utilities pay per incident. The energy use is greater, the number and complexity of systems on the premises are bigger, and the possible DR per incident is larger. In the most expensive markets, this pays for the custom integrations needed to respond to price signals. This is probably good enough for today’s grid. Tomorrow’s grid will be much less predictable, and the need for more participants will be greater.

Just as there are times when there is a shortage of electricity, there are times when there is a superabundance. Buildings that take responsibility for storing energy in advance are better able to manage demand reductions when asked. If the markets offer fixed prices except for the peaks, then it will be cheaper to ignore storage and DR. Sooner or later Markets must follow availability. The most important feature of smart grids will be to recognize scarcity and abundance faster, and to thereby price better.

In the future, then, a Pricing Service will be the essential load management service that operates the grid. A grid pricing service must be able several questions:

1. What is the price of Electricity now?
2. What will it be in 5 minutes?
3. What was the highest price for electricity in the last day? Month? Year?
4. What was the lowest price for electricity in the last day? Month? Year?
5. What price will electricity have for each hour of the day tomorrow?
6. What was the high price for the day the last time it was this hot?

The answer to each of these questions has another component “How sure are you?” Those prices may be fixed tariffs absolutely locked down. Those prices may be fixed tariffs, “unless a DR event is called.” Those prices may be wild guesses about free markets.

At its core, OpenADR must have price services in the future.

Underpinnings for standardizing Demand Response (DR)

For decades, regulated electricity markets have struggled to deal with volatile energy markets providing to support un-caring customers. Customer’s real-time purchases, called load by the electricity industry, vary throughout the day, and more to the point, co-vary with external events. These issues are not limited to electricity. The “Super-Bowl flush”, which has reached the status of urban legend, names the stresses placed on urban waste water systems as external events synchronize demand.

Public Utilities Commissions defined tariffs to prevent the Super-Bowl flush for decades. Peak use can increase prices for a year. Tiered pricing increases the bill for any amount above a pre-set usage during a month. These approaches pre-date any discussion of smart grid—often they have effects contrary to those desired by the smart grid. The smart grid is smart in that it detects at any time whether there is too little or too much power available, and uses market signals to decrease or increase demand to match supply.

Early efforts at the smart grid focus solely on reduction. When the signal goes out to certain buildings, they reduce their load, or use of electricity, in a pre-defined way. We call this process demand response, and pre-eminent specification for this signal is called Open Automated Demand Response (OpenADR).

Demand response must be more than load reduction. The great wind farms of west Texas, the most successful farms in North America, are, I’m told, able to sell less than 40% of the electricity they generate. The wind generates electricity at times when no one is planning to use it. Even the most inefficient energy storage would be preferable to simply wasting the electricity.

The challenge now is to define signals that common across North America and the world, and that that handle energy surplus as well as deficit. This effort is underway in the OASIS Energy Interoperation Technical Committee.

One problem is that energy market operations have been restricted and confined, both by technology limitations and by public policy decisions. We have discussed DR as a one-way interaction, from utilities to customers. We have tied DR to special tariffs and to direct control systems. Each of these restricts the participants and innovation in DR.

On the Committee, we tried to place electricity in a normal market context. We identified four essential market activities, or services:

1. There is an indication of interest (trying to flush out offers), when a market operator is seeking partners for a demand response or energy source.
2. There is an offer of a service whether megawatts or “nega-watts”
3. There is an execution of a contract (agreement to purchase / supply (b))
4. There is a call for performance of the contract (c) at the price agreed upon.

We are defining these services so they can be combined to meet today’s tariffs. For example, one of today’s tariffs for interruptible power may offer a lower price all year in return for the right to shed load automatically up to six times a year. Under the Energy Interoperation model, this would be standing contract with 6 time-limited pre-executed response contracts. Automated Demand Response is merely a call for performance on existing contracts at the agreed upon price.

Another model, coming into use, is Price-based ADR. If we assume the traditional utility-centric model, we would see the utility publishing an indication of interest in buying DR at a given price. Business and buildings willing to respond would simultaneously offer and execute a contract to shed load. The performance could be called at the same time or later as contracted.

Emergency or "Grid Reliability" events could look left out by this approach. Grid Reliability events require mandatory participation in today’s markets. These could be set up as standing pre-executed options. A grid operator then need merely call for performance as in any other demand-limiting event.

In this way, we can build all the tariffs and markets out of a few low-level services.

Sometime soon, I will write about the requirements for a pricing service.

What is an internet of energy?

In the political world, we often speak as if the smart grid will create and internet of energy. This sounds sexy, but it can be hard to noodle out what it means. I’m pretty sure that it does not mean that we will use smart meters to deliver porn. To find the internet of energy, we must acknowledge straight up the problems with our energy plans.

The internet was built around assumptions, scarcity of bandwidth and fragility of infrastructure, that clearly apply to today’s grid. Long distance transmission was expensive; email used to hop across the country on late-night phone connections to the next state. Every engineering decision was based on occasional connections, local management, and the knowledge that it was risky to rely on anything that was not controlled in-house.

Because we knew so little about what would happen next, we solved little problems. We did not make assumptions about how the next segment would handle our messages, or how reliable they would be. This allowed constant technological churn. Once we had TCP, IP began to drive out local protocols such as IPX and NetBIOS, and quickly supplanted top-down engineered protocols such as GOSIP and SNA.

On the infrastructure side, the churn was just as fast. I remember when X.25 was the future, and supported the first North American installation to supplement the banks of modems at CityNet. I remember when we signed up the Boston Choral Society, and gained users with perfect pitch, who bedeviled tech support by describing modem squawks by note. For ears, I telecommuted via dual ISDN lines back-fed out of Siler City, but tariffed as local connections. Each change of infrastructure was a minor blip for home and office communications.

Even the applications changed, always moving toward the simpler and less architected. Single purpose bulletin boards were replaced by Gopher servers and WAIS. Walled gardens such as AOL and Compuserve began to open up to the wider choices of today’s internet. Waves of push technologies failed and peer-to-peer regularly raised its transgressive disruptive hand against the top-down passive order.

Somehow, by planning for infinite scarcity, in every cell of networking, and in every switch and gateway, we found ourselves with unimaginable surplus, in which computers in our pockets are now network connected with greater bandwidth than used to connect supercomputing centers.

The key decisions of the smart energy are to reduce operating margins, to not build enough transmission and distribution, and to use intermittent power sources such as wind, sun, and tides. We are planning for the grid to provide lower quality service than it has in a hundred years. We have now forced ourselves into the corner in which network communications found themselves in the 70’s. We can only gain the same success by committing to the same principles.

The future of the grid will be based upon intermittently available energy distributed over inadequate and expensive wires. It will be too expensive, both in energy losses and in capacity management, to get our power from far away. We will have to make our energy decisions assuming occasional connections, local management of use, and the certain knowledge that it is risky to rely on anything that we do not manage in-house.

For a while, we will try to solve these problems with central decision-making and a hierarchical organization. Utility-based management of home and business use will make sense to traditional power engineers, just as SNA was briefly the networking strategy natural for mainframe users. This will fail under its own internal contradictions. The DOE envisions homes and businesses able regularly to operate off-grid for a week; it is unlikely that such remote energy management will work when the grid is down.

Each time we plan for unreliability, we can gain another level of reliability, accept another level of innovation. Homes that are indifferent to grid reliability can accept the local installation of self-contained, self maintain pocket power plants. Pocket power plants may be subject to longer outages through poor management, but their customers won’t care. Novel strategies of congestion pricing and load management may provide inconstant power to the neighborhood distribution, but the neighborhood will be relatively insensitive.

At each level, planning for scarcity and unpredictability will add resilience. Resilient systems will be better able to accept diversity; acceptance of diversity is a requirement for allowing innovation. As system that accepts innovation, in ways today’s static grid never will, will accept the creative destruction, the quick success or failure that draws venture capital and engineering ideas together.

The future quality of the grid is lousy; that’s the plan. Embrace its failures and unreliability, because that’s where markets will follow. That is how we will find an internet of energy.