1. COMPUTER SCIENCE
COLLOQUIUM
A Ray-Traced Sampling Framework for
Visualizing and Improving Material Detail in
Urban Energy Simulations with QUIC EnvSim
Sarmad Ahsan Siddiqui
Computer Science Graduate Student
ABSTRACT
Materials such as concrete, asphalt and red brick replacing vegetation have given rise to the Urban Heat
Island phenomenon. The Urban Heat Island phenomenon is when the temperature of the region within the
cities is higher compared to the rural surroundings, often resulting in higher energy costs. With
urbanization being very common, this problem needs to be addressed now more than ever. With QUIC
EnvSim, we are able to simulate and better understand energy balance in urban settings. However, QUIC
EnvSim does not represent building material properties in fine-grained detail and is limited by the level of
discretization of the urban domain. Due to the nature of solar energy simulations, the level of details in
the scene can potentially have a large impact on the energy calculations. The hypothesis that this thesis
works on is that adding micro level material details to buildings at a scale smaller than the domain
discretization, will affect the energy balance calculations in the urban environment.
This paper extends on QUIC EnvSim(QES) which is an extendable framework for urban climate
modeling. The newly developed model improves on the visual aspects of the frameworks, making the
scenes more realistic. It provides the option to choose from various types of buildings in the urban
environment. Building type can be either fixed or be randomly assigned based on the settings specified in
the configuration. Apart from improving the visual aspect, the new model also uses this visual
information for more detailed calculation of material properties such as albedo, emissivity, and diffuse
fraction.
The model can be enhanced by defining additional building types, resulting in more complex and detailed
scenes. The system uses OptiX for visualizing the models. The OptiX Ray tracing engine is an accessible,
flexible and reusable platform used for development of ray tracing programs for NVIDIA GPUs. The
model is used to visualize complex scenes such as the University of Minnesota, Duluth and downtown
Salt Lake City, Utah. The results provide better visuals and also provide more accurate estimation of the
energy balances in the urban environments. Simulations using the old and the new system were run over
the course of three days for a street in Gothenburg, Sweden, for which data has been measured
experimentally. The comparison of the results show that adding material details into the scene improved
on the results by taking them closer to the measured values.