Turbulent flows are known to generate important fluctuations of local velocity and pressure. This fluctuations can have an important mechanical on immersed structures. It is common practice in the engineer community to study the flow around immersed structures using synthetic turbulent flows that do not render the physics of these extreme fluctuations. Furthermore, because these fluctuations are rare, there is no hope in generating them through direct numerical simulations. However, it is possible to bias the statistics of these fluctuations using techniques inherited from statistical physics, recently applied to dynamical systems. The cornerstone of such methods is large deviations theory. The application of such methods to turbulent flows allows one to generate rare fluctuations with a reasonable numerical cost, without having to modify the dynamics of the flow, generated through DNS. This presentation will feature the results of the application of this approach to a simple flow, serving as a proof of concept and thus paving the way to applications to complex systems, closer to industrial concerns.