Full project proposal can be seen here:
https://drive.google.com/file/d/0B58rifcsn9OBclZJTjFkQ2d0UEU/view?usp=sharing
Inspiration
California, as well as many other parts of the world, is rapidly increasing the percentage of renewables on the grid. However, there are significant challenges associated with having a high penetration of variable and intermittent energy generators.
The current approach relies primarily on using natural gas peaker plants; however, this strategy is ultimately incompatible with long-term GHG policy goals, and could result in increased electricity prices.
Another method of reducing variability involves promoting technological and geographic diversity among renewable generators. While effective, this tactic is not sufficient to guarantee that power can be provided reliably. Deployment of energy storage devices has been proposed as an alternative strategy. Although storage is promising, cost and scale remain as hurdles. Fortunately, there exists another approach that is functionally equivalent to deploying massive amounts of storage infrastructure, yet only requires the use of relatively inexpensive electronics and software: automated demand response.
Demand response involves strategically shifting demand for electric power away from times when it stresses the grid and towards times when it can be more easily accommodated. Once primarily performed by large industrial and commercial consumers on hot summer days after a phone call from the local utility, DR now has the potential for adoption on a more distributed scale with the introduction of Internet- connected home and commercial appliances.
Of these, the most promising points of use are thermostatically controlled loads (TCLs) such as HVAC systems, refrigerators/freezers, and hot water heaters.
Our work is also inspired by a research project at UC Berkeley that one of us worked on last semester. This team developed the FlexBox, a wireless sensor network developed by TIER and RAEL to monitor and control freezers and refrigerators, which currently has 30 implementations in Managua, Nicaragua.
What it does
The Netfridge allows refrigerators to provide grid services and minimize carbon emissions using WattTime data and wholesale Locational Marginal Price data. The fridge would also be capable of being switched between perishables, non-perishables, and off modes using a web app.
How I built it
Sensors and relays on the refrigerator are connected to an Arduino microcontroller, which in turn is connected to a Rapberry Pi. The Pi connects to a Heroku-hosted server, which pulls and processes data streams from WattTime and CAISO.
Challenges I ran into
We had some difficulty pulling data from CAISO and WattTime, and were not able to bring the mini-fridge for live sensor data collection. Instead, we used historical data from a fully instrumented refrigerator using a FlexBox, a wireless sensor network developed by TIER and RAEL to monitor and control TCLs in Managua, Nicaragua. The data we used during our pitch was from the first prototype of the FlexBox, which instrumented a refrigerator on the 4th Floor of Sutardja Dai Hall in UC Berkeley.
Accomplishments that I'm proud of
We made great progress on our front-end dashboard and data visualization, and gained experience pulling WattTime API and CAISO LMP data.
What I learned
What's next for NetFridge
We're planning to continue work as part of our Cyber-Physical Systems course.
Log in or sign up for Devpost to join the conversation.