Buildings consume a large amount of energy and environmental resources. At the same time, current practices for whole-building energy simulation are costly and require skilled labor. As Building Energy Modeling (BEM) and simulations are becoming increasingly important, there is a growing need to make environmental assessments of buildings more efficient and accessible. A building energy model is based on collecting input data from the real, physical world and representing them as a digital energy model. Real-world data is also collected in the field of 3D reconstruction and image analysis, where major developments have been happening in recent years. Current digital photogrammetry software can automatically match photographs taken with a simple smartphone camera and generate a 3D model. This thesis presents methods and techniques that can be used to generate a building energy model from a digital photogrammetry-based 3D model. To accomplish this, a prototype program was developed that uses 3D reconstructed data as geometric modeling inputs for BEM. To validate the prototype, an experiment was conducted where a case-study building was selected. Photographs of the building were taken using a small remotelycontrolled Unmanned Aerial Vehicle (UAV) drone. Then, using photogrammetry software, the photographs were used to automatically generate a textured 3D model. The texture map, which is an image that represents the color information in the 3D model, was semantically annotated to extract building elements. The window annotations were iii used as inputs for the BEM process. In addition, a number of algorithms were applied to automatically convert both the 3D model and the annotated texture map into geometry that is compatible for a building energy model. Through the prototype, pre-defined templates were used with the geometric inputs to generate an EnergyPlus model (as an example building energy model). The feasibility of this experiment was verified by running a successful energy simulation. The results of this thesis contribute towards creating an automated and user-friendly photo-to-BEM method.
