Rethinking solar potential in dense cities
Bowen Tian defended his PhD thesis at the Department of Built Environment on May 21.
As cities rapidly transition to renewable energy, rooftop photovoltaic (PV) systems have become an essential part of the modern landscape. However, predicting PV power generation in dense urban environments is highly challenging. Complex urban geometries and trees cast unpredictable shadows that can significantly reduce power output. Traditional prediction software often relies on rough estimates and struggles to accurately account for these complex urban realities.
Bowen Tian addresses this gap by developing an innovative high-resolution computer modeling framework that enables highly accurate predictions of solar power generation, even in the most complex city settings.
A three-step innovation
This PhD research is built on a comprehensive three-step approach that combines artificial intelligence with advanced simulation tools.
First, Tian leverages AI to automatically interpret 3D city maps and satellite imagery. This approach eliminates the need for time-consuming manual surveys, rapidly identifying roof geometries, tilt angles, and the precise locations of trees and surrounding buildings.
Second, this enriched dataset is processed using a newly developed solar simulation tool, Pyrano 2.0. Unlike conventional models, Pyrano 2.0 distinguishes between solid structures and semi-transparent elements like tree canopies. This allows for a much more accurate representation of shadow patterns and sunlight exposure.
Finally, the research introduces PYWER, an advanced electrical model that translates sunlight data into actual energy output. This tool operates at the level of individual solar cells, enabling precise calculation of energy losses caused by partial shading across PV systems.
Toward more reliable urban energy planning
By integrating automated data collection with physically grounded simulations, Tian provides a complete and scalable solution for city-wide solar analysis. The framework reduces reliance on generalized assumptions and replaces them with precise, data-driven insights.
This work offers valuable tools for urban planners, grid operators, and policymakers. It supports better decision-making in solar panel placement, improves local grid management, and contributes to accelerating the transition toward sustainable and energy-efficient cities. The research marks an important step forward in making urban solar energy systems more predictable, reliable, and effective.
Title of PhD thesis: Supervisors: Roel Loonen and Roland Valckenborg.