A Data-Driven Multi-scale Digital Twin Framework for Optimizing Energy Efficiency in Public Pedestrian Infrastructure

Abstract

Managing energy systems in public pedestrian spaces such as parks, plazas, and streetscapes is a complex task that entails balancing sustainability objectives with the demands of the people who use these areas. This study presents a novel approach to integrating multi-scale digital twin technology to use various urban data to improve energy efficiency in public pedestrian areas while also promoting usability. The process entails the development of a virtual representation of the National Botanical Garden of Iran, incorporating verified City Information Modeling (CIM) data, real-time Internet of Things (IoT) sensor data, geographical resources, and machine learning algorithms. At the component, facility, and district levels, a modular system architecture links and influences lighting, HVAC, transportation, landscaping, and other systems through two-way dynamic connections. Computational fluid dynamics (CFD) models accurately show how structural and natural materials react to heat and cold in situations involving passive ventilation, active HVAC systems, and renewable energy. We conduct these simulations at the component level. We use facility-level agent-based modeling to simulate pedestrian movement and usage patterns, evaluating the impact of human factors on various designs. The district model, created at a 1:1 scale, provides a visual representation of the linked energy landscape. It allows stakeholders to evaluate data-driven optimizations through immersive virtual reality experiences. We use reinforcement learning to maximize multiple key performance indicators (KPIs) like energy efficiency, safety, accessibility, comfort, and operational carbon footprint. The digital twin system has enabled interactive urban systems to continuously adapt and improve efficiency by utilizing real IoT sensor data. To improve usability, we use simulated outcomes and virtual testing. We measure the effects of optimization before and after using sensor data, utility metering, observational research, and user feedback surveys during different weather seasons. This study offers a versatile data integration and simulation framework that utilizes several modeling methodologies to improve the sustainability and usability of public pedestrian environments. Other urban complexes that require optimization across interacting scales can apply the modular architecture, initially verified at the mixed-use National Botanical Garden of Iran. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.