NOTRE DAME DE PARIS (Paris, France)

The experimental campaign aimed at determining the wind-induced pressure field acting on the Cathedral of Notre Dame de Paris. The research interest stems from the limited consolidated knowledge on wind actions for large Gothic cathedrals, whose complex geometry includes potentially vulnerable elements such as large rose windows, towers, and extensive roof surfaces. Following the 2019 fire and the subsequent reconstruction, the need for reliable, physically grounded wind pressure data became even more relevant. For this reason, an extensive test program was designed to provide a three-dimensional, high-resolution characterization of pressures over the main architectural components of the Cathedral.

Scaling criteria and reproduced atmospheric boundary layer

The geometric scale was selected through a compromise among: (i) maximising Reynolds number (and manufacturability at sufficiently large model scale), (ii) maintaining a low blockage ratio, and (iii) ensuring a boundary-layer profile in the wind tunnel consistent with the target terrain exposure. A length-scale ratio of 1:200 was adopted.

The oncoming turbulent boundary layer was generated using a fetch of wooden floor panels equipped with randomly distributed square prisms, extending up to the test section. In addition, castellated barriers were installed near the tunnel inlet (on both floor and ceiling) to reach the target turbulence levels. The tunnel floor was locally raised to account for the elevation difference between the cathedral ground level and the Seine River water level. The resulting flow, evaluated at the beginning of the turning table (where the model and the surroundings are installed), showed good agreement with the target mean velocity profile. Turbulence intensity was slightly lower than the target but acceptable for the purposes of the study. The longitudinal integral length scale, inherently difficult to reproduce at large geometric scales, matched the target reasonably well up to roof height and was underestimated above; this was considered satisfactory because the steeple load was not a primary deliverable and the instantaneous integral loading on towers was not the central focus.

Physical model of the Cathedral

A detailed 3D numerical model provided by the Client was first analysed to define pressure-tap density and measurement requirements. The digital model was then extensively edited for prototyping, including segmentation into multiple printable parts, which were fabricated via 3D printing and assembled for wind tunnel testing.

The assembled model was equipped with >1200 pressure taps. Seven miniaturised 32-port pressure scanners were used, enabling 222 simultaneous pressure measurements. Because the total number of taps exceeded the available simultaneous channels, the campaign was organised into multiple measurement configurations (see below).

Surrounding area: two experimental set-ups

To separate “site-specific” effects from more general behaviours, the tests were performed in two distinct layouts:

• “No surrounding”: the Cathedral model was placed on the turning table and embedded in a layout of generic roughness elements, designed to sustain the reference Eurocode terrain category IV boundary layer across the turning table. This set-up supports broader, transferable conclusions for large Gothic cathedrals under urban exposure.
• “Surrounding”: the Cathedral was tested within a site-specific model of the Paris urban context, reproducing the built environment within approximately 220 m radius from the Cathedral centre. This configuration quantifies the aerodynamic influence of the real surroundings (shielding, channelling, wake effects) on local and global pressure patterns.

Pressure-tap configurations tested

Because only 222 channels could be acquired simultaneously, the >1200 taps were measured through five main configurations, with targeted sub-configurations introduced where a higher spatial resolution was required.

Configuration 1 — Roof (baseline roof mapping)

Primary mapping of roof pressures to describe suction/overpressure patterns on the main roof surfaces.

Configuration 1.A — Roof, longitudinal refinement (main roof)

Additional taps added in the lower portion of the roof to improve load reconstruction along the main (longitudinal) roof development.

Configuration 1.B — Roof, transept refinement (transversal roof)

Additional taps focused on the transept roof portion to better resolve the local pressure field and gradients.

Configuration 2 — Lateral strip (generic Gothic-side representation)

High-resolution mapping on a narrow lateral portion, intended to be representative of a typical large Gothic cathedral side elevation.

Configuration 3 — Transept and rose windows (detail mapping)

Detailed pressure field over the transept, with particular emphasis on the large rose windows and adjacent façade regions.

Configuration 4 — Main façade and towers

Pressure mapping on the principal façade and towers for wind directions normal and oblique to the façade.

Configuration 5 — Lateral sides and apsis (global mapping)

Global pressure distribution on both lateral sides and the apsis (split into two sub-configurations – 5.A / 5.B – during the measurement phase to increase spatial resolution, i.e. more taps covered overall while respecting the simultaneous-channel limit).

Reference system and interpretation rules

For consistent interpretation of the results, the campaign adopted a unique reference system:

0° wind direction corresponds to mean flow normal to the main façade.

Wind direction angles correspond to a clockwise rotation of the mean flow. “Left” and “right” sides are defined for an observer looking at the main façade; for the apsis results only, the observer viewpoint is taken from behind the apsis. When results are shown for both sides simultaneously, the viewpoint is taken from the left side, with the opposite side represented as if “seen through” the building.

Outcome and relevance

The campaign highlighted that the pressure field around a large Gothic cathedral is strongly three-dimensional, highly non-uniform, and not reliably predictable using simplified code-based approaches alone. The dual set-up strategy (“surrounding” vs “no surrounding”) enabled both: (i) quantifying the role of the specific Paris urban context for Notre Dame, and (ii) extracting more general indications for similar cathedral typologies under urban boundary-layer wind.

Flow visualisation and dissemination activities

In addition to pressure measurements, the experimental campaign also included qualitative flow visualisation tests using smoke, aimed at improving the physical interpretation of the measured pressure fields, with particular attention to the behaviour of the flow in the vicinity of the main façade and its towers. These visualisations proved especially useful in understanding the complex three-dimensional flow structures developing along the windward roof and façade regions.

The observations highlighted how the interaction between the steep roof geometry and the articulated vertical development of the Cathedral façades induces strong flow acceleration and significant vertical wind components. In particular, the combined effect of the roof inclination and the superposition of multiple wall orders along the flanks promotes local flow separation and recirculation phenomena near architectural details such as balustrades and tower bases. The smoke visualisations provided a clear physical explanation of these mechanisms, complementing the pressure data and supporting a coherent interpretation of the aerodynamic behaviour of the structure.