Efficiency of a community microgrid's energy performance depends on the ratio of energy input (on- and offsite energy resources) to energy output (energy consumed and stored). To date, studies on energy performance in community microgrids have been limited to overcoming technical issues and operational efficiency of electrical components at the supply and demand side of a microgrid system. This is despite the fact that numerous energy planning studies have emphasized the impact of urban form on the amount of energy consumed by buildings in communities, and on the feasibility of adopting onsite renewable energy generators. The aim of the proposed research is to add a spatial dimension to measuring energy performance, by developing a data-mining model and real-time predictive software prototype that enhances the spatial design and planning of community microgrids towards a higher energy performance.
This research hypothesizes that each community microgrid has a site-specific optimal configuration of urban physical attributes and geometries, known as urban form, that contributes to its energy performance. A multidimensional vision for designing community microgrids with higher energy performance is considered, leveraging urban form (superstructure) to understand how it impacts the performance of the system's distributed- energy-resources and loads (infrastructure). This vision engages the design sector in the technical conversation of developing community microgrids, leading to energy efficient designs of microgrid-connected communities well before their construction. A new generation of computational modeling and simulation tools are required that address this interaction. The conducted software survey in this research indicates a lack of such computational tools. Thus, the goal of this study is to overcome this gap by developing a software prototype for predicting the energy performance of a given community microgrid design scenario in real-time by the virtue of its urban spatial configuration.
The relationship between community's urban form, net energy consumption, and onsite net PV energy production is assessed herein. San Diego is selected as a case-study due to the availability of energy performance data. Using GIS maps, different urban typologies in San Diego are identified and their rules of urban form are extracted. The identified typologies and urban form configurations are 3D-modeled in Rhinoceros along with the placement of existing PV panels. Using appropriate plugins, the energy performance of these communities is simulated, and the resulted values are compared and calibrated with actual energy data. Based on the extracted rules of different urban typologies, a generation-evaluation algorithm is used to generate multiple configurations of urban form and surfaces for PV energy production. For each configuration, the urban form is quantified by measuring its energy-relevant spatial attributes, and the community's net monthly PV energy production and energy consumption is estimated via the calibrated simulator. A machine-learning algorithm is trained on the aggregated synthetic numerical data of urban form and energy performance to find its underlying relational pattern. The learned model is used as the back-end of the software prototype which inputs 3D spatial design scenarios of communities, and in real-time outputs monthly estimations of net PV energy production and energy consumption as a solar community microgrid.
Ph.D. candidate: Mina Rahimian
Dissertation committee: José Duarte (adviser), Lisa Iulo (adviser), Guido Cervone, Seth Blumsack