Green Soda Project
Xiaofan Fred Jiang and David E. Culler
National Science Foundation
Electrical energy usage in modern commercial buildings today represents a large part of our total energy expenditure and carbon footprint. The computer science building at UC Berkeley (Soda Hall) alone consumes around 600 kWh of electricity per day .
This usage can be roughly broken down into three parts: HVAC, lighting, and plug-loads. While HVAC and lighting represent the larger part of the pie, plug-loads are still significant, especially in this digital age where computers and portable electronics are becoming ubiquitous. Preliminary data has shown that the plug-load usage in Soda Hall is around 20% of the total electrical usage, or 120 kWh, which is certainly significant.
The Green Soda project focuses on this 20%: the plug-loads, which include everything from desktop computers to cellphone chargers, lamps, monitors, servers, network routers, refrigerators, and anything else that comes with a plug.
In the first phase of this project, we will audit Soda Hall using a hundred ACmes deployed throughout the building. ACme  (AC Meter) is a plug-through, IP-based wireless device that provides real time energy usage measurement and control for AC devices. The data from the audit will allow us to generate a time-varying map of plug-load electrical usage and help identify various attributes of AC devices and their contribution to the total. It will also allow us to better quantify the breakdown of total electrical usage in the building. Furthermore, by correlating the individual usage with the type of AC load, we can identify opportunities for saving, both autonomously and human-effort based.
Research has shown that real-time, per-appliance electricity usage feedback can induce behavior changes that lead to 10% to 20% reduction in usage. In the second phase of this project, we will provide real-time feedback of energy usage on a per-appliance level to participants using ACmes.
This part of the project consists of six stages:
(1) Base: We will hand out ACmes without further instructions. Data is collected and stored in a database.
(2) Pull-Stimulation: We will inform the participants of a web site that provides detailed real-time energy measurements of the AC device plugged into ACme. Data is collected as per stage 1.
(3) Push-Stimulation: We will actively inform the participant of the level of their energy usage by using various stimulations, such as blinking the LEDs on ACme in a certain pattern (e.g., red if greater than 500 watts, green if less than 100 watts).
(4) Control: We will introduce various levels of control over the AC devices. For example, we will enable the embedded switch inside ACme to allow the user to remotely turn on and off the AC device. We will inform participants of ways to configure the computer for lower energy consumption.
(5) Automation: We will use active-badge to automate turning on and off appliances. For example, in the simplest case, when a user wearing an active-badge comes into the vicinity of an ACme connected to a lamp, ACme will automatically turn on the lamp, and turn off when the users leaves the radio range. We will consider developing learning algorithms to automate turning on and off the AC appliance by predicting user behavior.
(6) Social Effect: We will develop social networks based on energy usage data. Peers can view and “review” each other's electrical usage and trend.
Figure 1: Green Soda ACme network topology