Towards a smart grid in the Pacific Northwest


By Rob Pratt

Over the last century, the electric system has delivered nearly all of the power responsible for unprecedented economic growth and prosperity. The system, built service territory by service territory, has consistently delivered reliable load with affordable power. Yet significant improvements are now needed.

The system is under increasing stress – electricity demand is expected to double by 2050, and the issue of carbon management is emerging as a game changer. The capital investment needed to meet these challenges is staggering with the cost of building new transmission often exceeding $0.5 million per kilometre. In the context of these challenges, business as usual is no longer a sustainable model.

There is growing consensus that an important part of the solution lies in deploying “smart grid” technologies. Research indicates that broad use of demand side management could mitigate up to $70 billion of the projected $450 billion investment needed between now and 2020. Beyond these traditional benefits, fully engaging demand creates additional opportunities critical for managing power in a carbon constrained world. Potential ancillary benefits include measuring and validating carbon offsets, demonstrating and verifying efficiency, and helping manage the integration of renewable resources.

Jerry Brous

Jerry Brous, a homeowner in Sequim,
Washington, monitors energy costs
and usages from his home computer

The recently completed Pacific Northwest GridWise® Demonstration Project, led by the Department of Energy’s (DOE) Pacific Northwest National Laboratory (PNNL), offers a comprehensive look at how such a smart grid might be implemented. The project was funded primarily by the DOE, with other support provided by utilities and manufacturers, including Bonneville Power Administration, Portland General Electric, PacifiCorp, Whirlpool Corporation, Clallam County PUD, City of Port Angeles, and Invensys Controls. It examined two separate demand response technologies with potential to mitigate peak demand and increase reliability of the grid.

Olympic Peninsula project
The first portion of the Pacific Northwest GridWise® Demonstration Project, the Olympic Peninsula Project, set out to create and observe a futuristic energy pricing experiment. By inserting intelligence into electric grid components – at the end-use, distribution, transmission and generation levels – significant values of grid transformation were achieved, including a 50% reduction of imported transmission power during peak loading for days and a 15% reduction over the course of one year.

In an effort to manage demand by creating an open market for power, the Olympic Peninsula Project included the following controllable assets that responded to the project’s energy price signals:

  • Residential demand response for electric water and space heating provided by 112 homes using gateways that supported two-way communications. This residential demand response system allowed current market prices to be presented to consumers and allowed users to preprogramme their automatic demand response preferences. The residential participants were evenly divided among three types of utility price contracts (fixed, time-of-use, and real time) and a control group. While all residential electricity was metered, only the appliances in price responsive homes (~75 kW) were controlled by the project.
  • Five 40 hp water pumps, distributed between two municipal water pumping stations, representing a nameplate total load of about 150 kW. The electrical load these pumps placed on the grid was bid into the market incrementally when water reservoir levels were above a designated height.
  • Two distributed diesel generators (175 kW and 600 kW) that served a facility’s electrical load when feeder supply was insufficient. The biddable resource capacity in this case was the removal of the building electric load (~170 kW) from the grid by transferring it to these units. In addition, a small 30 kW microturbine was set up to respond to the two-way market. Unlike the larger generators, it ran in parallel with the power grid. The market prices offered for the supply of these generator units were based on the actual fixed and variable expenses incurred.

The homeowners who participated in the project received new electric meters, thermostats, water heater controllers and other smart appliances connected via Invensys Controls home gateway devices and specially designed software. The software let homeowners customise devices to a desired level of comfort or economy and automatically bid into and responded to a dynamic electricity market in five-minute intervals. During peak, the market price naturally became more expensive, and the software automatically lowered thermostats or shut off the heating elements of water heaters to the pre-set response limits established by individual homeowners.

Participants received constantly updated pricing information via a two-way broadband internet connection. A “virtual” bank account was established to pay a simulated bill for the actual electricity used. Homeowners kept all of the money saved when they operated their household energy use in collaboration with the needs of the grid. With the help of these tools, consumers were able to easily and automatically change how and when they used electricity – for their own financial benefit and the benefit of the grid. On average, customers saved approximately 10% on electricity bills from the year prior.

The Olympic Peninsula Project found a significant number of customers, including residential consumers, will sign up for and respond to a real time price that varies on a five-minute interval when they are provided technology that automates their response and preserves their right to choose their preference for comfort or savings.

Using this design, congestion was successfully managed on the virtual feeder, which acted as if the entire load in the project was on the same feeder. Seasonally, the project imposed a new constraint on the energy that could be imported into the feeder from an external wholesale electricity source. The project then controlled the imported capacity below this constraint for all but one 5-minute interval during the entire project year. On this feeder capacity duration curve, feeder supply was successfully limited to 750 kW. Distributed generators provided additional supply (up to about 350 kW at its peak) when needed.

Water heater

Water heater fitted with a grid
friendly appliance controller

Additionally, The Olympic Peninsula Project controlled both heating and cooling loads. Observation of the project’s residential thermostatically controlled loads for those homes on real time price contracts revealed a significant, surprising shift in energy consumption. Space conditioning loads served by real time price contracts effectively used energy in the very early morning hours when electricity is the least expensive. This effect occurred during both constrained and unconstrained feeder conditions; however, it was more pronounced when the feeder was constrained.

This result is remarkably similar to what one would expect for pre-heating or pre-cooling, but these project thermostats had no explicit prediction capability. It is the diurnal shape of the price signal itself that caused this outcome. The project practiced this form of market-based control in two ways: in commercial buildings, zones within the buildings competed among themselves for conditioned air that could be diverted by thermostatically controlled dampers. These thermostats could respond to, but could not bid into, the regional market. In contrast, residential thermostats in real time contract homes bid directly into and directly affected the regional market.

The Olympic Peninsula Project’s market also deferred and shifted peak load. The magnitude of load reduction under real time price control increased with the peak itself and with the degree to which the feeder import is constrained. The project conservatively estimated that a 5% reduction in peak load was achieved under a 750 kW constraint; easily 20% peak load reduction was obtained under a 500 kW feeder constraint.

The Olympic Peninsula Project also looked at the feasibility of using existing industrial and commercial generators that are normally used for emergency backups as a supplement to the existing grid when power usage is high. Automated two-way communication between the grid and distributed resources enabled energy and demand price signals to deploy dispatchable resources. In this manner, conventionally passive loads and idle distributed generators were transformed into elements of a diverse system of grid resources that provided near real time active grid control and a broad range of economic benefits. The Olympic Peninsula Project successfully managed a “virtual” distribution line, or feeder, and an imposed feeder constraint for an entire year using these technologies.

The Olympic Peninsula Project was particularly successful obtaining useful supply from distributed diesel generators. The project elected to control the generators at their existing emergency transfer switches. The generators and their protected facilities therefore ran separate, or islanded, from the grid. These generators bid the capacity of the commercial building loads they served; the price they offered was based on actual fixed and variable expenses they would incur by turning on and running. These resources were called upon and used many times during the project.

Grid Friendly™ Appliance project

In the second portion of the Pacific Northwest GridWise® Demonstration Project, the Grid Friendly™ Appliance Project, grid friendly appliance (GFA) controllers were embedded in dryers and water heaters in homes in Washington and Oregon. The GFA controller is a small electronic circuit board developed by researchers at PNNL to detect and respond to stress on the electricity grid. When stress is detected, the controller automatically turns off specific functions like the heating element in the dryer. This momentary interruption can reduce electricity consumption enough to stabilise the balance between supply and demand on the grid without the need to turn on inefficient gas turbine generators.

The frequency threshold of the GFA controllers was set at 59.95 Hz – high enough to recognise frequent, shallow frequency excursions of a 60 Hz AC voltage signal available at any residential outlet. Such events occurred once a day on average, each lasting for up to a few minutes. The controller was set to respond within 0.25 seconds after a sudden drop in frequency. These events were reliably detected in the field by GFA controllers and the appliances responded to the signals as designed by shedding portions of their loads.


Appliances fitted with grid
friendly appliance controllers

The under-frequency events observed in the field lasted from several seconds to 10 minutes – short enough that residential customers, when later surveyed, responded that they had neither observed nor been inconvenienced by the curtailments of their appliances. The appliances received virtually the same frequency signal and responded to the signal similarly, despite the distribution of controllers over a wide geographic region. Every appliance responded when the frequency dipped 0.003 Hz or more below the control threshold as measured by a frequency monitor in eastern Washington State.

Any time the controlled appliances were energised, the GFA controllers initialised themselves in their triggered, curtailed states. A short delay therefore occurred before controlled appliances were permitted to operate. Such a cold load pickup capability is obtained at no cost with smart appliance load controllers like the GFA controller. The delay may be designed to ease the introduction of loads onto feeders as they energise.

Among the important attributes of a GFA controller is that it performs its duties autonomously. The only communication that it requires is the AC voltage signal that is available at any appliance’s wall-plug receptacle. For the purposes of this demonstration, however, components of the Invensys Controls GoodWatts™ energy management system monitored the performance of each controller and its appliance and communicated observations of the controller and appliance actions.

The study found that GFA controllers have the technical capacity to act as a shock absorber for the grid and can prevent or reduce the impact of power outages. The appliances responded reliably and participants reported little to no inconvenience. The vast majority of homeowners in the study stated they would be willing to purchase an appliance configured with such grid responsive controls.

The future

The GridWise® Demonstration Project was a unique field demonstration that revealed persistent, real time benefits of GridWise® technologies and market constructs. PNNL will continue to partner with the DOE to extend these tools and concepts for broader implementation around the country, including working directly with industry to help scale up these demonstrations in the real world. PNNL continues to develop the GFA controller with clients DOE, BPA, and independently owned utilities in the Northwest. In these and future projects, PNNL will develop and offer autonomous controller responses for both frequency and voltage, for both contingencies and regulation services. A stretch goal of researchers at PNNL is to apply autonomous controllers to the improvement of electric power quality, including system power factor and harmonic pollution. PNNL has received numerous inquiries from utilities on both the east and west coasts wishing to apply its insights about price responsive resource control to evolving smart grid projects.

Yet even with significant momentum around smart grid technologies, there are still a number of challenges that must be addressed for wide-scale adoption to occur. Deployment will depend on advanced technology, proper customer engagement, and regulatory and policy support.

Utilities seeking to deploy smart grid technology will need to present a strong case to regulators and customers that such an action would be cost effective and mutually beneficial. For decades, utilities have deployed traditional demand response, packaged as interruptible rates, to their customers in return for discounted electricity prices. Industry leaders are beginning to move beyond this model to help demonstrate tangible returnon- investment for all stakeholders.

Unlike critical peak pricing, which is called upon approximately 12 times a year, fully automated demand response can deliver valuable, persistent responses in real time, all the time. The potential benefits include mitigating market volatility; providing cleaner, cheaper and faster regulation services to help renewable resources, improving reliability, increasing resistance to blackouts; deferring expansion costs and responding to weather or natural disasters.

In addition to using demand response to more strategically manage the grid, engaging electricity consumers is essential to the success of smart grid deployment in the residential marketplace. The significance of the consumers in premisebased participation became apparent in January 2008 when the California Energy Commission was subjected to public out lash concerning a proposal to require installation of “smart” thermostats in new buildings.

The proposed thermostats were designed to allow utilities to adjust customers’ preset temperatures during critical peaks. Customers would have been able to override the utilities’ suggested temperatures, but in emergencies the utilities could override customers’ wishes. A large California building association and other groups fueled a public outcry that spread via the internet and talk radio shows. The CEC dropped the proposal entirely from the 2008 edition of the building efficiency standards.

However, in the Olympic Peninsula Project, where consumers exercised complete control of their level of participation, 95% of participants stated they would be likely to very likely to participate in a similar project if offered by their utilities.

Furthermore, policy must provide utilities with the proper incentives to encourage investment in the best technologies. A strong carbon policy will create new incentives and opportunities to deploy smart grid technologies. If carbon prices rise to projected levels, wholesale power cost could double. The ancillary services made possible by optimising grid operation can help mitigate the anticipated financial impact. Leveraging the demand response networks can provide utilities with the required measurement and valuation for carbon offsets with little to no capital costs.

Smart grid technologies can also help utilities reduce carbon output. A carbon policy is projected to lead to increased penetration of wind power onto the grid. The regulation capability of demand response can help ease the operational stresses and expense of intermittent wind resources.

A wave of innovation and opportunities will be unleashed when utilities effectively use technology to engage consumers and when regulatory signals align with industry innovators. This new paradigm will provide enhanced reliability, economic payoffs for utilities and consumers and a smaller environmental footprint.