Urban Digital Twins Levels of Integration: From Conceptualization to Technical Implementation

Urban Digital Twins (UDTs) have emerged as a way to support data-driven urban
planning, monitoring, and decision-making by enabling dynamic and integrated digital
representations of cities and territories.
Yet, despite the growing interest in both academia and practice, the concept still lacks a
clear definition and a shared technical basis. The absence of common standards, together
with the diversity of implementation approaches, has led to different interpretations,
fragmented practices, and difficulties in moving toward operational deployment. Among
the challenges that still hinder the implementation of UDTs in practice, data integration is
considered a major one.
UDTs rely on the integration of heterogeneous data coming from multiple sources, scales,
and domains, including 3D city models, Building Information Models, Internet of Things
data, and simulation outputs. However, current practices show that there is still no
systematic or generic way to integrate these data. In many cases, this results in siloed
implementations and limited interoperability. The challenge is amplified when
considering the dynamic aspects of UDTs, which require continuous updates throughout
their lifecycle, from creation to use and update.
This thesis examines how data integration is understood and implemented in UDTs. It
first looks at the conceptual foundations of the field, highlighting the gap between
theoretical definitions and what is actually developed in practice. It also shows the variety
of integration mechanisms currently adopted. Based on this analysis, the thesis proposes
a formalization of three levels of data integration: Level 1, the conceptual model level,
where integration is achieved through schema extensions; Level 2, the database level,
where data are adapted to fit the data model; and Level 3, the front-end level, where
integration takes place at the visualization or client side. Together, these levels provide a
framework to better understand and compare integration approaches across different
UDT implementations.
Beyond conceptual contributions, the thesis develops and implements a technical
framework to operationalize these integration levels. It proposes City2Twin, a lightweight
and flexible UDT platform based on open standards such as CityJSON and web-based
architectures. The platform is designed to support data integration, visualization, and
analysis. It is tested through several use cases involving BIM models, IoT sensor data, and
simulation outputs, making it possible to compare the advantages and limits of different
integration strategies in terms of flexibility, interoperability, and maintainability.
The findings show that there is no single integration approach that fits all use cases. The
most suitable level of integration depends on the application context, the nature of the
data, and the requirements of the UDT lifecycle. By formalizing these levels and
illustrating how they can be implemented in practice, this thesis contributes to a better understanding of data integration within UDTs and provides guidance for the design of
flexible and interoperable UDT systems.
Overall, this work positions data integration at the core of UDTs and provides a basis for
future developments toward more mature and operational UDT ecosystems.