Table of Contents

• 1.0 Introduction
• 1.1 The Oahu Daily Energy Demand Curve
• 1.2 Smart Grid Technology and the Oahu Daily Energy Demand Curve
• 2.0 From dumb grid, dumb consumers to smart grid, smart consumers
• 2.1 Smart meters
• 2.2 Smart consumers
• 3.0 Smart consumer technology examples
• 3.1 Current Energy
• 3.2 Ecotricity
• 3.3 Google PowerMeter
• 4.0 Research Directions: Toward improved technology for smart consumers
• 4.1 WattDepot
• 4.2 Oscar: Oahu Smart Consumer Analysis and Research
• 4.3 Exploring incentives for energy conservation in office-like environments
• 5.0 Future directions

1.0 Introduction

The smart grid "delivers electricity from suppliers to consumers using digital technology to save energy, reduce cost, and increase reliability and transparency" (Wikipedia: Smart Grid). According to the US DOE Modern Grid Initiative report, smart grid requirements include the ability to: (1) heal itself; (2) motivate consumers to actively participate in operations of the grid; (3) resist attack; (4) provide higher quality power that will save money wasted from outages; (5) accomodate all generation and storage options; (6) enable electricity markets to flourish; and (7) run more efficiently.

This is a very ambitious set of requirements. In this research project, we will focus on (2), (4), and (7): we want to investigate ways to positively influence consumer behavior that lead to less power wasted and increased efficiency. To frame our approach, it's helpful to have a basic understanding of the Oahu "daily energy demand curve".

1.1 The Oahu Daily Energy Demand Curve

The following figure provides a cartoon perspective on the daily energy demand curve for Oahu:

While dramatically oversimplified, this figure nonetheless reveals several interesting features of Oahu energy use. First, the baseline energy usage is around 550 MW. Starting at around 6am each day, usage increases until noon, when it remains approximately level until around 6pm, where there is sharp spike for a couple of hours. Usage then drops down to the baseline level until the next morning. The figure also shows the maximum potential power generating capacity of Oahu as approximately 1700 MW, although the practical maximum is somewhat less since at least one power generating facility can be offline for maintenance at any given time.

Note that the demand curve for any particular day could deviate significantly from this curve for reasons including: (a) time of year: for example, in summer the peak is lower; (b) day of the week: for example, weekends have a different profile from weekdays; and (c) weather: for example, hot weather can increase daytime demand due to increased use of air conditioning.

What is not generally known to consumers is that the cost to produce electricity varies throughout the day, because electrical power is generated by different kinds of plants with differing cost-benefit tradeoffs. The most efficient power generation plants can take a full day to come online and cannot quickly increase or decrease their power output in response to demand. Because of their slow reaction times, these types of plants are used to provide "baseload" power and generally run continuously.

The second type of power generation plant is called "cycling", because they can be turned on and off each day, generally in less than an hour. Cycling plants are less efficient (i.e. use more barrels of oil per MW) than baseload plants, but have the advantage that they can be turned off when not needed.

The third type of power generation plant is called "peak", because they are dedicated to providing power only when spikes in usage occur. Peakers can be turned on quickly, generally in 10 to 20 minutes. Though peak units provide the fastest reaction time to changes in power demand, they are the least efficient and thus most expensive to run.

The following diagram illustates how the daily Oahu power requirements are satisfied through a combination of baseload, cycling, and peak power:

This representation is, again, a simplification. For example, it does not show the contribution of renewable energy sources (solar, wind, wave).

1.2 Smart Grid Technology and the Oahu Daily Energy Demand Curve

One of the primary goals of smart grid technology is to change the daily energy demand curve. The following diagram illustrates aspects of the desired changes:

As the diagram illustrates, introduction of the smart grid might facilitate:

• A lowered minimum and maximum energy usage. This corresponds to an overall general reduction in energy usage.

• A flattened "peak". The new "actual" curve shows a less dramatic "peak" in energy usage at 6pm, but greater energy usage into the evening hours. This illustrates "load shifting", where energy consumption is moved from a high cost region to a lower cost region.

The following diagram shows how this new demand curve reflects upon baseload, cycling, and peak power generation:

Notice that the new demand curve results in reduced requirements for all three types of power: peak, cycling, and baseload.

2.0 From dumb grid, dumb consumers to smart grid, smart consumers

The traditional electrical utility grid, or "dumb" grid, can be characterized as follows:

• Energy producers have only indirect and coarse grained control over energy consumption. Control is indirect in the sense that energy producers "manage" consumption in the form of electrical rates and incentive programs. Control is coarse grained in that rates and incentive programs are changed infrequently.

• Energy consumers have extremely limited insight into the cost and/or availability of electrical power. In a dumb grid, a consumer pays a flat rate for usage, despite the fact that the cost of generating electricity varies throughout the day, as we saw above.

• Billing involves manual, human reading of meters connected to homes and businesses. Energy usage is thus tracked on a monthly or semi-monthly basis.

The "dumb" grid thus forms a kind of "black box" in which the electrical utility is responsible for providing all required electrical power on a moment-to-moment basis, and, apart from rate increases and incentive programs, the utility has very little control over consumer behavior. Conversely, energy consumers have very little insight into the effects of their moment-to-moment energy usage on the grid, making them "dumb" as well. This creates many inefficiencies in the grid, as well as many opportunities for improvement.

2.1 Smart meters

One way to overcome certain types of inefficiencies in the dumb grid is to introduce smart meters, such as the basic model illustrated here:

At a minimum, smart meters provide real-time usage data to both the utilities and the consumer. This can increase billing accuracy and also provide a basis for consumers to monitor their own usage. Unfortunately, consumers must be both motivated and sophisticated to gain insight into their behavior and take useful action based upon this interface.

The next step beyond automated meter reading is to implement automated control over energy consumption. An example of this technology is a smart thermostat under development by UC Berkeley engineers, illustrated here:

With such devices, the utility company can implement demand response, automatically reducing home and business heating and air conditioning power consumption when load levels are high.

2.2 Smart consumers

The evolution of basic smart meters toward controllable meters such as a smart thermostat extends the control of the utility company from just power generation (in the dumb grid) to control over both power generation and power consumption (in the smart grid).

A smart grid does not necessarily lead to smart consumers, however. Put another way, it is entirely possible to create a smart grid where utility companies implement demand response mechanisms, but consumer insight about their electrical consumption is still limited to their monthly bill. In such a situation, the consumers remain just as "dumb" about their usage as they were with the dumb grid.

In contrast, a "smart" consumer is provided with timely, accurate, actionable information about their power usage that enables them to act in harmony with the needs of the grid. While smart grid demand response mechanisms can only target a few types of consumption devices, and require large investment to deploy and manage, smart consumers can change their behavior to increase energy efficiency and reduce cost across all possible consumption devices.

The following table summarizes the four possible combinations of smart/dumb grids and smart/dumb consumers:

The focus of this research is on effective ways to exploit modern information technology practices including open source communities, cloud computing, and social networks to enable "smart consumers" in a cost-effective and secure fashion.

3.0 Smart consumer technology examples

3.1 Current Energy

Laurence Livermore Lab created a site called Current Energy that essentially provided the daily demand curve for various regions. The following screen display shows the energy load for California on June 5, 2002.

Interestingly, on this day the load exceeded the maximum capacity of the grid.

Sarah Darby reported in The Effectiveness of Feedback on Energy Consumption: A Review for DEFRA of the Literature on Metering, Billing and Direct Displays that there is anecdotal evidence that usage of the Current Energy site did lead to some energy conservation on the part of consumers during California's energy crisis of 2001.

3.2 Ecotricity

A UK company called Ecotricity provides a somewhat more actionable view of grid data. Essentially, they analyze the load and provide consumers with a sense for whether or not they should use heavy appliances or not. Unlike the Current Energy site, which tracks production and demand, Ecotricity focuses on carbon intensity. Here is a screen image:

This screenshot was taken during mid-afternoon in Hawaii, which meant it was the middle of the night in Britain; a perfect time to run your clothes dryer.

3.3 Google PowerMeter

Perhaps the highest profile effort in information technology to support smart consumer behavior is Google PowerMeter. Unfortunately, this technology is still under development and there are very few details available. The interface appears to be a Google Gadget, as illustrated here:

In general, data from home meters will be sent directly to Google, which will store this data and then provide analyses based upon it.

They have announced partnerships with several utilities and plan to make further announcements later in 2009.

4.0 Research Directions: Toward improved technology for smart consumers

The previous section introduced projects that illustrate ways that energy data can support smart consumers. However, technology support for smart consumers is still in its infancy and much research needs to be done to understand what kinds of data must be available and how, when, and in what form it should be presented in order to create sustained, positive changes in consumer behavior. A recent article in GreenTechMedia suggests that software built on top of raw energy data might become a significant focus of future investment.

Our research is based upon the following conceptual foundations:

• Smart consumers require utilities to become providers of high quality information as well as providers of high quality power.

• If utilities can make high quality data available in a timely manner, the Internet community can develop tools and techniques for generating actionable information and communicating it to consumers.

• The security risks associated with providing the data necessary for smart consumers are less than the security risks associated with a demand response smart grid, one of which has already been shown to be vulnerable to a computer worm that can hop between smart meters at homes and businesses.

• Cost savings is only one of many potential incentives for smart consumers. Examples of non-monetary incentives include reduction in carbon consumption and out-competing friends, neighbors, and community members in energy-related behaviors.

4.1 WattDepot

Research on smart consumers requires appropriate infrastructure including the ability to acquire, store, and aggregate energy data. We are building an open source, RESTful web service called WattDepot to provide this capability. WattDepot will provide a repository for data generated by energy meters that monitor both energy production and consumption. WattDepot will also provide retrieval and analysis capabilities, including the ability to produce energy profiles that aggregate data from across multiple sources. WattDepot will provide interfaces to its repository that facilitate the creation of high-level user interfaces such as the Google Visualization API.

The following diagram shows how data flows into WattDepot:

The diagram shows three example inputs paths for WattDepot. The first path shows data from a solar photovoltaic panel using a microinverter from Enphase. An Envoy meter from Enphase is recording power production data from the panel. A WattDepot client polls the Envoy meter periodically and then creates Sensor Data resources under the Source corresponding to the meter using HTTP PUTs to the WattDepot server. Similarly, the next two examples show two WattDepot clients polling Obvius Acquisuite meters, one for power consumption data from a building, the other for power generation from a wind turbine.

The following diagram shows how data flows out of WattDepot:

The diagram shows two example output paths for WattDepot. The first example shows a Google Visualization Time Series Chart of instantaneous power readings from a Source over several days. The chart gadget obtains the data to be displayed through HTTP GET requests to the WattDepot server, using the Google Visualization data source API implemented by WattDepot. The second example shows how WattDepot can be used to generate the same kind of stoplight visualization used on the Ecotricity website discussed in the previous section.

WattDepot has been under development since June, 2009 and we expect to have an initial release available for use by October 1, 2009.

As well as providing a repository for real energy data, WattDepot can provide important infrastructure for simulations of energy production and consumption. Our second research project illustrates this.

4.2 Oscar: Oahu Smart Consumer Analysis and Research

What kinds of raw energy data should be provided by Oahu utilities to support smart consumer behavior? How should this energy be analyzed? How can it be made actionable? What user interface (web, mobile, Facebook) or combination of user interfaces would best facilitate smart consumer behavior?

The Oscar (Oahu Smart Consumer Analysis and Research) Project will explore these issues. This project will consist of two parts: design and implementation of a WattDepot simulation dataset, followed by design and implementation of user interfaces that use this data to enable smart consumer behavior.

The design and implementation of a WattDepot simulation dataset involves the creation of a raw dataset that provides plausible sensor data instances for energy-related data over the course of a month on Oahu. For example, the daily energy demand curve discussed above explains how electrical power on Oahu is generated by a variety of power sources over the course of a day, each with their own levels of efficiency. Part of the dataset would involve sensor data that shows when various generators come online and go offline. Meta-data about these generators would include information on their efficiency, such as how many barrels of oil are required by the generator to produce a single MW of power. Sensor data can also indicate the contribution of non-fossil fuel energy sources (such as H-Power). The goal of the simulated dataset is to provide at least the level of information required to duplicate the Ecotricity user interface, in which the current load on the grid can be calculated along with the contribution of various energy sources.

To create the simulation dataset, we will request the cooperation of engineers at HECO to review our assumptions and make sure the generated dataset is plausible. The OscarSimulation page provides more details on the proposed design of this dataset.

We plan to begin work on the simulation dataset immediately and expect it to be available in WattDepot by November 1, 2009.

The second part of the Oscar Project involves the development of user interfaces based upon this dataset. Our initial efforts in this direction will involve class projects in the University of Hawaii software engineering curriculum. Small groups of students will work together to build web applications that make the simulated data available and useful to energy consumers.

These web applications will be available by mid-December, 2009, and will serve as prototypes to facilitate future development of real world interfaces. We also hope through this process to stimulate the interest of these students in energy technology and hope that some of them will continue to work with us in this area after the class is over.

4.3 Exploring incentives for energy conservation in office-like environments

A third research project focuses on what kinds of incentive structures and information technology support can produce and sustain smart consumer behavior. This research will involve a set of energy conservation "competitions" between various floors of Saunders Hall on the UH Manoa campus to see whether, for example, the presence of near real-time feedback about the current electrical usage on the floor motivates members of the floor to conserve energy. We have meters installed in Saunders Hall that will enable us to collect this "local" energy data.

By having competition between different floors, it allows us to apply differing "treatments" to the different floors and thus tease out differences in behavior produced by different kinds of stimulus. For example, one floor could have access to their floor's energy data, while another floor could have access only to the building's energy data, allowing us the possibility of assessing the impact of fine-grained information as a feedback and incentive to behavioral change.

This research will also enable us to investigate the issues involved with energy usage in office-like environments. Many applications of metered energy data appear oriented toward the home environment, where a single person or family has control over all aspects of energy usage, and a single action by an individual (such as turning off an air conditioner) results in an visible change to meter data.

In contrast, "office-like environments" are characterized by a large number of people affect energy usage, and any single action by an individual is less likely to result in a visible change to meter data. In an office-like environment, it is necessary to create not only individual but also collective change to produce significant changes in energy use.

We plan to begin this project in January, 2010.

5.0 Future directions

The preceding sections outline some of the initial research to be done, but we expect all three of these projects to continue for several years at least.

For WattDepot, future directions will include expansion of the API to support new forms of queries, performance optimization, possible deployment within a cloud computing service such as Google App Engine, and support for other energy-related data, such as weather.

For Oscar, we expect that our initial simulation dataset and user interface experiments will yield insights leading to significant changes both in the simulated dataset and the design of the user interface. These insights could lead to a second round of dataset and user interface development. Ultimately, we hope that the Oscar project will provide the insights and experience necessary for actual Oahu power data to made available in real-time.

We hope that our last research project on incentives and office-like environments will grow beyond Saunders. Based upon our initial experiments, we would like to eventually run similar experiments both in other university buildings (such as Holmes Hall or dormitories) as well as in non-university office buildings. Such experiments can yield a diverse range of evidence for developing smart consumers not only at home but in their work place.

We welcome your comments on and corrections to this document. Please email Philip Johnson (johnson@hawaii.edu) with your insights. If you would like to become involved, please do not hesitate to contact us as well.