Using (BS) sleep mode to save energy has received major attention in recent years. In this work, a centralized approach to identify the combination of active base stations and RAN parameter, namely vertical antenna tilt angle, required to meet a certain traffic demand is developed. The method uses Genetic Algorithm(GA) to reduce the otherwise exponentially increasing search space. We consider Orthogonal Frequency Division Multiple Access based single frequency networks with fractional load modeling, to capture the effect of varying co- channel interference due to lower than peak traffic load condition to evaluate the performance of the proposed scheme. Results on percentage energy savings obtained under different traffic load conditions are presented for dense deployment of base station where highest amount of energy saving is possible. Impact of large and small search space on the energy saving is also shown.