To reduce wind-induced vibrations, Tuned Mass Dampers (TMD) are widely used in high-rise buildings. However, traditional TMD system requires large additional damping mass and huge installation space at the top floors of the building. In this paper, a novel distributed-Multiple Tuned Facade Damping (d-MTFD) system is investigated. This system employs the outer skin mass of the Double-Skin Facade (DSF) as the damping mass. In the upper stories, the DSF's outer skin, mounted on a guide rail system, is designed to be parallel moveable. The passive/semi-active d-MTFD system, which is assumed to be installed on a 76-story benchmark building under across-wind excitation, is optimized using multi-objective Genetic Algorithms (GA) for two defined objectives: minimizing the peak top floor acceleration and controlling the maximum peak relative displacement of all the moveable DSF's outer skins. On-off groundhook control and displacement-reducing bang-bang control are used as semi-active control strategies. The rolling friction caused by the guide rail system has a considerable impact on the optimization results. These results are presented in the form of Pareto fronts. Optimized parameters can be selected to yield a good trade-off between both objectives. The optimized passive/semi-active d-MTFD systems can highly improve the structural response. Compared with the optimized passive d-MTFD system, the use of semi-active control can reduce the vibration of the DSF's outer skin significantly. Comparing the selected optimized cases with similar peak top floor acceleration, the peak top floor facade relative displacement using displacement-reducing bang-bang control decreases 58.3%, and the Root Mean Square (RMS) value decreases 61.4%.