Transient phenomena such as abrupt changes in wind direction occur frequently during the operation of wind farms. The wakes produced behind wind turbines are affected by these situations, modifying the impact on nearby turbines, both in power output and life expectancy of the blades. In this work, a group of 4 turbines in a wind farm is analysed via Computational Fluid Dynamic (CFD) simulations during wind shifts. Large Eddy Simulations (LES) are carried out by means of the open-source software OpenFOAM, including the SOWFA libraries. Once the wakes reach a time average stationary solution the wind direction is modified by means of a pressure gradient forcing, imposing different change rates. An actuator disk with the ability to correct yaw misalignment is used to represent turbines. Besides, neutral Atmospheric Boundary Layer (ABL) flows are considered. Results show that for higher change rates the greatest impact on the downwind turbine is produced when turbines are no longer collinear with wind direction due to the wake curvature. Also, higher change rates produce a higher curvature modifying the impact on power output of downwind turbines. This study provides a deeper understanding of the cause of greater deficits in wind farm output.