TWINEXUS: Digital Twins for Nexus-based Decision Analysis in Urban Planning

Last updated on Oct 23, 2026
Tutorial 3D Python Tutorial LiDAR Python 3D Twinning Workbench

Instructors: Dr. Pirouz Nourian (coordinator) and Dr. Mafalda Madureira

This course is suitable for all specializations of ITC and students from UT (e.g., Civil Engineering, Computer Science, or Creative Technology). It aims to provide the student with knowledge of different 3D city/building modelling methods (as a base for Digital Twins), based on geospatial information. The students will be given the opportunity to practice with a variety of applications (e.g., GIS, BIM, image-based, gaming, or Virtual reality and Augmented reality) to develop and interpret their own 3D city/building model. To achieve this, theoretical and practical activities where the student can learn different 3D city/building modelling techniques and methods, used in a variety of applications, based on geospatial data are used. The student will have the opportunity to work on a 3D modelling/Digital Twin assignment.

Challange: Planning decisions rarely stay inside the sector they are made in. A road relieves congestion but induces demand and restructures land values; a flood barrier protects one zone but redirects risk to another; a solar farm alters urban heat, which alters cooling demand. These feedback loops are the nexus the course is named after: a property of the urban and environmental system itself, not of any modelling toolchain.

A digital twin is useful here only insofar as it makes such trade-offs quantifiable and communicable to the actors who must decide between them. It does not replace planning judgment; it augments it, by making the consequences of a choice visible before that choice is committed to. In TWINEXUS you specify, assemble, and demonstrate a Planning Support System (PSS) for an environmental planning problem of your own choosing, and you defend it orally: the system must make a trade-off visible and defensible, not issue a verdict.

The backbone: five layers, five lectures: The course deliberately carries a lean conceptual backbone and moves to the studio fast. Each lecture covers one layer of the TWIN-NEXUS architecture; each layer produces one formal object that the next layer consumes.

    1. Problem structuring | What is to be decided, by whom, and judged against which indicators? | Use cases, a DPSIR chain, explicit decision variables
    1. Problem discretization | On which cells and links do the decisions live? | A grid or graph with a justified resolution and topology
    1. Cross-sectoral simulation | How does a decision in one sector propagate into another? | A cellular or agent-based model of at least one cross-sectoral effect
    1. Decision analysis | Which configurations are worth arguing about, and for whom? | A multi-objective formulation and an interpreted trade-off space |
    1. Integration | How does a planner query, contest, and reuse all of the above? | A reproducible, working PSS

Discretization is a methodological commitment, not a preprocessing step. The choice of grid encodes assumptions about resolution, area equivalence, and neighbourhood topology that propagate through every downstream computation; it is also what turns a vague “planning measure” into a measurable decision variable, and what makes the decision space finite, enumerable, and, if you like, playable.