Urban digital twins promise smarter cities, but municipalities are struggling to deliver

Urban digital twins and 3D city models have been heralded as transformative tools for climate adaptation, citizen engagement and data-driven decision-making, but new research from Denmark and Sweden shows that the grand promises rarely survive contact with everyday municipal planning. A multi-sited ethnographic study, published in Computational Urban Science, followed municipalities in the two countries as they attempted to integrate 3D models into their planning processes, and found a persistent gap between the hype surrounding the technology and the practical realities of fragmented data, limited budgets and overstretched staff.

The expectations have been building for years. As early as 2017, digital twins appeared among Gartner’s top ten technology trends, with predictions that the technology would reach mainstream adoption within a decade. Other forecasts suggested that by 2025 more than 500 cities worldwide would have developed an urban digital twin. Yet development and implementation have remained relatively modest, according to the study’s authors, Emil Nørager Kruse, Thomas Theis Nielsen and Maja de Neergaard. Most examples have been confined to small-scale, experimental settings within laboratory environments and are rarely accessible to the wider public, reflecting the technical and organisational complexity involved.

To understand how the promises translate into practice, the researchers spent 24 months combining interviews, participant observation and workshop participation. They followed three municipalities, two Swedish and one Danish, ranging from 13,700 to 106,150 inhabitants, and conducted 14 semi-structured interviews with public and private actors, including representatives from five technology companies, eight municipalities and one Danish governmental data agency working with Denmark’s national digital twin. They also took part in four workshops with tech vendors, municipal practitioners and citizens, gathering insights from GIS employees, architects and urban planners.

The technology itself carries a long lineage. The idea behind digital twins can be traced back to the 1970s, most notably to NASA engineers during the Apollo 13 mission, who used a mirrored physical system on Earth to simulate and test solutions. The concept was formalised in 2002 by Michael Grieves together with NASA technologist John Vickers within product lifecycle management, defined through three core components: physical products in real space, virtual representations in digital space, and the data connections linking the two. It first entered urban planning in the mid-2010s, building on decades of urban simulation modelling and enabled by advances in GIS, LiDAR scanning and crowdsourced platforms.

Technically, the distinction between a 3D city model and a digital twin remains blurred. Most commonly, 3D city models are associated with static visual representations of the built environment, while digital twins are described as dynamic, data-integrative and predictive, extending beyond geometry to include semantic data, real-time sensor inputs, temporal updating and simulation capabilities, with an assumed feedback loop between the physical city and its digital counterpart. Standards such as CityGML, an open standard from the Open Geospatial Consortium, attempt to create a common semantic and geometric backbone, and Version 3.0 introduces support for time-dependent properties.

The study found that technology vendors actively market 3D models and digital twins as vital steering devices for future urban governance, positioning them as instruments for addressing complex societal challenges. Yet the vendors themselves acknowledged the conceptual fuzziness. One consultant called twinning ‘a trendy buzzword-like term’, while another noted that putting the word in a grant application improves the chances of securing funding. Companies frame their role as ‘help to self-help’, delivering platforms that municipalities should adopt and use themselves, though the researchers note this reflects a business logic in which municipalities remain dependent on external providers for infrastructure, maintenance and updates.

The practical barriers were substantial. Municipalities navigated a fragmented landscape of software solutions, data models and platforms, often resulting in compatibility issues. Models were slow to load and difficult to navigate, requiring specialised technical knowledge; in one observed incident, a municipality’s model took more than five minutes to appear during a remote presentation, with layers appearing and disappearing as they recalibrated. Heavy models demanded computing power that small municipalities lacked, with one GIS employee asking how a system would respond when asked to perform an operation on 100 gigabytes of data. Data quality posed further problems, with buildings appearing with incorrect heights, missing extensions or outdated geometries, and flood simulations that failed to match the intended future scenarios.

Licensing arrangements added another layer of difficulty. In several cases, municipalities discovered that purchased licenses only allowed limited interaction, such as adding text or 2D content, while editing 3D objects or running simulations required costly upgrades, sometimes only after contracts had been signed. One digitalisation employee described the situation as ‘very unfortunate’. These dependencies left municipalities reliant on external consultants, hindering the development of internal competencies, and in one case a municipality discovered after signing that its 3D company had previously gone bankrupt.

Organisational capacity proved equally limiting. In Sweden, a consultant estimated that only about 10 of the country’s 290 municipalities use 3D models, with only the four largest cities having dedicated 3D staff. In Denmark, a 2024 nationwide investigation found that only 19 of 98 municipalities have a 3D model, of which only eight update it annually. Responsibility is often concentrated in a single GIS employee, sometimes on a temporary contract, and knowledge remains confined to technical specialists rather than being embedded across departments. Projects driven by enthusiasts tend to disappear when those employees leave, and models developed for specific initiatives are rarely sustained beyond the project lifecycle.

The researchers conclude that the development of urban digital twins is an open-ended process of experimentation, adaptation and continuous becoming rather than a linear transition towards fully integrated systems. Municipalities selectively incorporate elements associated with digital twins, such as additional data, sensors and simulations, into existing 3D models, and most practitioners use them primarily for visualisation and communication. Even citizens’ encounters with the models complicate the promise of greater accessibility: at a public library workshop in Sweden, a recurring response was that the model was so different from Google Maps that navigating the city in it felt unintuitive. As the field turns towards integrating AI agents into digital twin platforms, the authors argue that foundational challenges of data, capacity and governance must first be addressed, and that any digital twin will necessarily remain partial, situated and incomplete.

Subject of Research: The practical challenges municipalities face when integrating 3D city models and urban digital twins into planning

Article Title: From epistemic promises to real world practices: the ongoing becoming of 3D models and urban digital twins in municipal planning

Article References: Kruse, E. N., Nielsen, T. T., & de Neergaard, M. (2026). From epistemic promises to real world practices: the ongoing becoming of 3D models and urban digital twins in municipal planning. Computational Urban Science, 6(1), Article 62. https://doi.org/10.1007/s43762-026-00299-w

Image Credits: AI Generated

DOI: 10.1007/s43762-026-00299-w

Keywords: urban digital twins, 3D city models, municipal planning, Denmark, Sweden, data infrastructure, GIS, technology vendors, climate adaptation, socio-technical systems, smart cities, agentic AI