Neighborhood Digital Twin Energy Models for Demand Load Prediction and Electrification: An Automated Workflow

This work focuses on automating the conversion of three-dimensional neighborhood digital twin point cloud data to a neighborhood energy model (NEM) for demand load prediction, encompassing loads beyond purely electrical, such as heating oil for space heating. The motivation addresses the common challenge of manually collecting and importing data for energy modeling, particularly at larger scales like neighborhoods. The work aims to establish a workflow to convert digital twin point cloud models to NEMs, facilitating electrification scenarios. Point clouds created with open-source sources, discrete methods and machine learning (ML) methods are used to obtain digital twin neighborhood geometric models. Point cloud data processing includes building footprint extraction through image processing, building height determination using convolutional neural network (CNN) and semantic segmentation networks (SSN), and roof shape identification via regional-based neural network (R-CNN). The process involves creating a workflow in the Rhinoceros environment using Grasshopper, focusing on developing a rapid NEM from a point cloud. The proposed workflow is validated by comparing results with two different energy audits in the case study location, demonstrating its accuracy and reliability. The developed workflow is versatile, and adaptable to various projects by changing input data. It converts digital twin geometric models to energy models. The work reduces the effort and time for urban and energy planners to efficiently conduct energy modeling at a neighborhood scale, enabling the evaluation of net-zero potential and sustainable urban planning. It can also help generate scenarios for the electrification of neighborhoods traditionally served by other energy means. This approach saves time and enhances accuracy in predicting demand loads and exploring electrification pathways, ultimately contributing to more effective and sustainable energy management.