Enhanced building-specific heat load prediction with limited information through an automated zoning algorithm

Abstract

In urban development, buildings have a key influence on energy use and sustainability. Therefore, accurate building energy demand predictions are required to design climate neutral energy systems. GML models, common for cityscale energy simulations in Europe, often lack detailed building information about architecture and construction. This study addresses the challenge of predicting heat load profiles for buildings with limited available data, especially those characterized by multiple thermal zones and various use types. A Multizone Assignment Algorithm (MZA) is developed to improve load prediction by defining zoning strategies using the concept of Binary Space Partitioning (BSP). The MZA integrates an automated zoning approach, embedded into the TEASER-AixLib framework for seamless energy simulation. By incorporating adjacency relations and interzonal heat transfer, the methodology improves prediction accuracy while requiring minimal manual input. This novel approach accelerates and simplifies district-scale simulations, providing important insights into how different areas within buildings interact and improving the operation of energy systems with limited initial data.