This paper presents a novel and superior Genetic Algorithm (GA) based resolver for Optimal Power flow (OPF) problem. Here, the main contrast to other Genetic Algorithm based approaches is that a novel expert based initial generation of population and adaptive probability approach (variable Cross over probability and mutation probability) is adopted in selection of offspring together with roulette wheel technique which reduces the computation time and increases the quality considerably. Selection and Placement of Shunt Devices are considered as a variable in this novel approach. Here continuous variables like Voltage Profile and discrete variable like transformer tapings are considered while minimizing the Fuel cost. The results obtained on standard IEEE 14 bus and 30 bus systems is compared with simple Genetic Algorithm and Particle Swarm Optimization (PSO) to Optimal Power flow and is found that this approach is more efficient, robust and promising.