<p>For long time,3D tectonic modelling was reserved to analog methods and many practitioners spent a lot of time and energy developing methods and materials to make their naturally 3D "simulations" as cylindrical as possible.</p><p>Fighting with so-called boundary effects, they actually obtained a lot of interesting structures and dynamics related to "border effects" . In the last 5 years, 3D numerical simulations have really emerged thanks to new numerical technics and increase in available 'computational power. The two methods are now competing and sooner or later, with the emergence of exa-scale and quantum technology, it is quite certain that numerical simulations will dominate the field because it is much better suited to tackle multi- physics problems arising in long term tectonics.</p><p>However, before entering an era of mass production, it is interesting to re-think how we introduce 3 dimensionality in numerical models. Numerical models can easily produce perfectly free slip boundary conditions, and it has therefore never been a problem to simulate a perfectly cylindrical situation. Is it useful ? Not really since we can run 2D simulations.</p><p>However, many models introduce the 3 dimensionality by imposing inherited structures in simulations that use perfectly cylindrical boundary conditions. Technically this corresponds to imposing free slip boundaries in the third dimensions. Nobody question it, and in a way, we numerical modellers, are just mimicking traditional analogue model set ups and emphazing on the multi-physics aspect of our simulations.</p><p>Yet, comparing to analogue models, we some time reach different solutions and sometimes, analogue models with their boundary effects produce tectonic structures that are much more realistic than models with perfectly free slip boundaries.</p><p>In this pico presentation, I will show exemples of free slip boundaries that introduce biased in continental break-up propagation models and discuss in which conditions free slips are acceptable and in which conditions are should be carefull in our interpretations of simulation results.</p>
