Project
# | Title | Team Members | TA | Documents | Sponsor |
---|---|---|---|---|---|
21 | Power Demand Response System |
Antonio Martinez William Widjaja |
Evan Widloski | design_document3.pdf design_document4.pdf final_paper1.docx final_paper2.pdf other2.pdf photo1.JPG presentation1.pptx proposal2.pdf |
|
#Problem Capacity factors of power plants are low because power demand is difficult to predict from the lack of granularity on the demand side, which results in high prices and inefficient electricity usage. Demand side power management could greatly reduce the overall load and help form better predictions of power usage. #Solution Overview We propose load control switches, first for refrigeration and heating/cooling systems, combined with software which predicts future load usage. The switch is plugged into a regular 120 V outlet. Then the home appliance refrigerator or heating/cooling system is plugged into the switch. For the focus of this project, we will test only temperature regulation systems which can run in varying cycles. Our device processing subsystem will use a chip microcontroller: either a 32 bit ARM (TM4C123GH6PMI) or Adafruit FT232H. It shares and receives information when connected to wifi or ethernet. The switch will also measure the DC load power consumed and the AC line power rectified. The devices work together to desynchronize the duty cycles of loads within a given area, creating no noticeable difference to the user, but collectively decreasing the total load by a percentage. In a region (regional information which the device will store in memory unless updated from wifi/ethernet), a switch can either run on the “red” duty cycle or the “blue” duty cycle (where “red” could mean on status every even hour and “blue” could mean on status every odd hour for example). Switches are apprised of the presence of other switches in the area every 24 hours and automatically attempt to balance the amount of switches running the “red” or “blue” cycles. The system will also collect load profile data and send that to a server for the software subsystem. The software can then communicate with all devices in a region with this information to run on a more nuanced schedule than the default red/blue. It can also predict the future load based on weather data. Software will be a mix of C and Python. C for the microcontroller, Python for data analysis and visualization of load profiles. Our device can be tested by plugging either real heating/cooling devices in or with a DC electronic load from the ECE power lab. The grid can also be simulated from the power lab. #Solution Components# **Software Subsystem** The software can also periodically send signals back to reconfigure a switch onto a new cycle. This software will also have knowledge of other switches on the network and the regional data of each switch (which might be manually determined for the sake of early prototyping convenience). The sum total of switch locations, weather data and average load profiles will then feed into a “load prediction” model for each switch. - On switch: Cycle control signals. Collective cycle or red/blue cycle control signals. - On switch: Storage and retrieval of load profile information from a database. - On external server: Dynamic prediction algorithm for future load profiles. **Processing Subsystem** - Input from wifi/ethernet. Output to wifi/ethernet **Electrical Subsystem** Electrically, the project hardware is a switch which connects from house circuit to plugged in appliance. Connected at one terminal of the switch, is a multimeter, or some hardware which will measure power consumed. Simpler yet, the average wattage of the plugged in appliance multiplied by the hours “on” would also give the desired information, which would just require occasional polling from the microcontroller about whether an appliance is plugged in or not. This information then runs along a wire to the microcontroller, and that information will either be stored locally and/or interface with wi-fi or ethernet. From that wi-fi or ethernet connection, the information of power consumed by the outlet over time will be uploaded to a database. It could enter as a series of arrays or CSV’s. Regardless, this data can then be analyzed with our software “subsystem” in Python. Separate from the data portion, the microcontroller would also be connected to a small relay to operate the actual switch. - Switch has one electrical AC input from the grid/home outlet and the output to the temperature regulating load. - LED display for basic information, such as cycle type, on/off status #Success Criteria - Close to real time switching response for at least two switches on the red/blue cycle. - Accurate load profiles created - Predicted power consumption based on current weather data is proportionate to the power actually used to maintain the thermal equilibrium **Team members** - William Widjaja (wwidjaj3) - Antonio Martinez (amartnz7) |