UTwin: A digital twin of the UT Austin campus

As cities and buildings become more interconnected, there is a need for better data management. Digital twins offer a potential solution, integrating building data, operations, and scenario simulation in one platform. In this paper, we present a preliminary digital twin of the UT Austin campus that mainly focuses on building energy use but also integrates additional geospatial datasets.


INTRODUCTION
The future of the buildings is data-driven.Building information models (BIM) are already widely during construction, encapsulating detailed data about a building's physical properties and functions.As buildings become smarter with more sensors, automated controls, energy generation and storage, there are also new streams of real-time data to be managed and acted upon.With many buildingrelated data sources and models spread across different software platforms, a holistic understanding of a building can be difficult.
While the definition of a digital twin varies among industries and researchers, a digital twin is often considered a digital replica of a physical asset with a real-time information link between the two [4].The ultimate goal for a building-related digital twin appears to be providing a unified platform that satisfies the needs of many different stakeholders.For instance, potential applications of a digital twin include providing parametrically accurate models for maintenance personnel, providing an interface for managing building systems, running energy or scenario simulations, and giving virtual tours to the public.
Here, we demonstrate UTwin, an urban scale digital twin of the UT Austin campus (175 buildings) that integrates a variety of data types (energy, air quality) from different sources (live, past, forecasted).We implement UTwin in OpenCitiesPlanner (OCP).

METHODOLOGY
Building energy data was provided to us from facility operators through Box, a cloud storage provider, and was updated on a daily basis.The provided energy data had already gone through some basic preprocessing such as outlier removal and missing value interpolation.Our task was transferring this data onto the digital twin in the form of colors for the 3D building models and line plots in the embed web application.
The 3D buildings were created with data from OpenStreetMaps (OSM), a crowd-sourced mapping project [3].Where the data was missing footprints or other attributes, the web interface of Open-StreetMaps was used to amend them.Building footprints were then downloaded from OpenStreetMaps using the OSMnx Python package [1].Height was determined using an average floor height of 3 meters.The energy data were used to provide a color attribute to the footprints.The final footprints were exported in the shapefile format and uploaded to the OCP project for use on the map.
To generate the content for popups when a user click on a building on the OCP map, a custom web application was created.GitHub Pages, Danfo.js, and Plotly.jswere used to create and host the embed webpage.Danfo downloads the energy data from a GitHub repository into a user's browser.It also selects and aggregates data based on the selected building and energy type.The resulting data is used by Plotly to create the plots dynamically in the browser.The resulting web application is embed into the OCP map by storing the url, which contains a query parameter to identify each building, in the shapefile mentioned earlier.
UTwin integrates a calibrated campus energy model [2].CitySim was used to simulate building heating and cooling load in the future

RESULTS & DISCUSSION
At the time of this paper, Figure 1 shows the data that was successfully integrated into the digital twin as well as the methods used.The preliminary digital twin was made publicly available1 and a screenshot of the digital twin is shown in Figure 2. The main challenges we faced were data availability and compatibility, which was partly addressed with the technology stack shown above.The integration of multiple data sources offers potential for interdisciplinary research and as we shared the digital twin with the public,