In this paper, an effective methodology has been proposed for optimal allocation of renewable generation sources in the distribution system. The objective of this work is to minimize real and reactive power losses in the distribution systems. The output power of solar DGs mainly depends upon solar irradiance level. So, before placement of these sources random nature of solar irradiance should be effectively modeled. A beta probability density function is used to model the solar irradiance and determine the exact output power of these sources. Further, the best location for placement of solar DGs is found out using loss sensitivity factor. The optimal sizing of photo voltaic arrays corresponding to these locations is determined using firefly algorithm (FA). Different time varying load models such as residential, commercial and industrial has been considered for the study along with probabilistic generation pattern. The developed method is tested on IEEE 69 bus test system. The obtained results show that optimal allocation of solar DGs in the distribution system gives a positive impact in terms of achieving better loss reduction and voltage profile enhancement. A comparative study is made for all the load models and the importance of considering different load models is projected.