Economic Agreements and Energy Management Systems for the Adoption of Rooftop Solar Systems
Many cities have formed policy goals to achieve clean energy by 2030, but the engineering and economic processes to do so remain unclear. The adoption of renewable energy through solar powered microgrids is a rising solution for cities.
This paper investigates the required economic agreements between: property developers and solar developers, and solar developers and utility companies. Furthermore, the control and operation of a 2 player microgrid system is proposed for the local community of Ward 6 - Washington, DC.
Modeling and Control of a connected Photovoltaic Microgrid System using Fuzzy Q-Learning Control
Microgrids are smaller modular power generation units that operate near the end user reducing transmission losses and increasing power reliability. These microgrids are made up of a local energy storage unit, a power generation source, and the ability to connect and disconnect to the main grid.
A microgrid’s purpose is to provide its local community with quality electricity at all times. This paper investigates the control and operation of a single microgrid system for the local community of Ward 6 - Washington, DC. A control policy using Q-Learning methods is explored.
Assessing Sizing & System Reliability of a Solar Wind Energy System using Stochastic Optimization
Introducing a large amount of variable wind and solar generation into an existing electric power system can present significant risk to the reliability of the power grid. The Washington D.C. area has been assessed to analyze the worst case scenario of integrating wind and solar power. This paper explores energy-based probabilistic prediction models to assess the impact of the stochastic characteristics of wind and solar resources on system reliability.
Using the lowest windspeed month, August, the system size is predicted by minimizing the instances peak demand is not met. Reliability is analyzed for both the largest and smallest possible system. Results show it is difficult to meet total peak demand at all times using only local solar and wind power. Energy storage systems, energy management systems, and smart control methods are not considered.
Modeling the California Mosquito Fires using Cellular Automata
Due the chaotic and destructive nature of wildfires they have been the interest of various computational modeling efforts. The California Mosquito Fires are modeled and simulated using cellular automata.
Different cellular automata models are explored through this project - traditional, tracking, and fuzzy tracking. The logic behind fuzzy tracking is recognizing that multiple parameters impact the likelihood of a fire spreading to a certain spot. Fuzzy logic further allows us to reduce the state space while considering all the parameters. Parameters were tuned through stability analysis.
Analyzing Reliability of an off-grid Solar Powered Reverse Osmosis Plant for Yabucoa, Puerto Rico
In reverse osmosis water is pumped through a membrane to remove salt and other impurities. Reverse osmosis is the only desalination process that can be powered by renewable energy as it relies on pressure. Small reverse osmosis plants powered by renewable energy provides easy access to clean water with minimal infrastructure investment.
After Hurricane Maria in 2017, the town of Yabucoa, Puerto Rico was left with no access to clean water for over 6 months. In this study, we explore if small solar powered desalination plants are a reliable option for this town within the constraints of the budget provided by the US Infrastructure Plan for Puerto Rico. The reliability of the system to output enough water is assessed using fuzzy logic methods. The size of the system is found by mapping the fuzzy measure of reliability to the number of panels.
Estimating Wind Speeds via Cloud Cover Imagery from GOES-16
Traditional NOAA methods to estimate wind speeds include analyzing cloud cover movement, cloud tops, and water vapor. Wind speeds can be modeled as a spatio-temporal process. Each wavelength band of the GOES-16 corresponds to a different height in the atmosphere. This project uses Band 11 as there is little atmospheric absorption of energy at this wavelength and provides best representation of cloud top movement in upper troposphere.
As cloud movement is a spatio-temporal process, a motion estimation algorithm was chosen to determine how much and in what direction a pixel moved. Each pixel's surrounding area was searched using 2D Log Search. Sequence of images taken 5 minutes apart by the satellite were analyzed. Each pixel represented 4 km. The motion vector generated by the 2D Log Search algorithm represents the cloud top movement and therefore wind speed and movement vector.