Although modern technologies such as smartphones with GPS functionality make navigation substantially easier, automobilists still rely on traffic signs to inform them which exit to take or which lane to follow. These traffic signs on highways are of substantial size rendering them, and their support structures, extremely susceptible to aerodynamic load induced fatigue damage. In this thesis traffic sign support structures in the form of steel portal frames consisting of square hollow sections are considered. Antea Group designs such structures using conservative generalized methods prescribed by the Eurocode and ROK norms. In order to investigate whether or not the structures may be optimized with respect to fatigue, a more elaborate load and dynamic response model is formulated. Aerodynamic loads are primarily considered as these are the most significant dynamic loads, which are governing for fatigue damage. The two types of aerodynamic loads that are considered are traffic induced pulse loads generated by large vehicles passing close to the sign, and stochastic turbulent wind flow. The former is in the form of a pulse load, while the latter is in the form of a super-position of sinusoidal waves based on the Solari wind turbulence spectrum. These loads are applied to a dynamic finite element method developed in Python specifically for this thesis, consisting of one dimensional Euler-Bernoulli beam elements. The dynamic response is evaluated using a modal analysis. The resulting dynamic cross-sectional signals are transformed into weld-stress signals using a finite element model in extit{IDEA StatiCa}. Finally the Rainflow Cycle Counting algorithm is used in combination with the Palmgren-Miner rule and SN-curves provided by the Eurocode in order to estimate the fatigue damage of the structure during its lifetime. This methodology is applied to an existing reference structure, and an optimized design. The optimized design is optimized according to the Eurocode with respect to the ultimate and serviceability limit state, but fails in the fatigue limit state. From the results it is established that the calculation model presented in this thesis yields much lower magnitudes of stress range cycles compared to those found in the design norms, leading to significantly reduced fatigue damage. However this model does not take instantaneous wind gust loads into account, reducing its range of applicability. It is suggested to utilize the simulation model for loads that occur more than 10e5 times in a 50 year life time, but utilize the existing design norm models for loads that occur less often. The traffic induced pulse loads are found to be of significant size under the current assumptions, but due to a lack of research in this field the validity of the magnitude of the pulse loads remains uncertain. Ultimately it is show that there is potential for optimization with respect to fatigue, but the simulation model should be used on more structures and should be further evaluated in order to create a more simplified and generalized analytical model that may be applied quickly to any design or even structure.