The load in mobile networks is subject to variations during the day, due to user mobility and varying network average usage. Therefore, the traditional or Distributed Radio Access Network (D-RAN) architecture, where the BaseBand processing Units (BBUs) are assigned statically to a number of cells, is sub optimal, comparing to a novel, cloud based architecture called Cloud Radio Access Network (C-RAN). In C-RAN a group of cells shares processing resources, and hence benefit from statistical multiplexing gain is expected. In this paper, the energy and cost savings in C-RAN are evaluated numerically using OPNET Modeler. A real case scenario is built upon the mobile traffic forecast for year 2017, a number of recommendations on traffic models and a proposed C-RAN implementation. The results achieved show that the maximum statistical multiplexing gain for user plane traffic in C-RAN architecture is 4 compared to a traditional and D-RAN architecture.