Butterfly optimization algorithm (BOA) is a recently introduced nature inspired meta-heuristics that mimics the natural foraging and mating behavior of butterflies. The framework of BOA is based on the fragrance emitted by the butterflies, which helps other butterflies in searching food as well as mating partner. BOA performs both the global as well as local search while looking for the optimal solution for the problem. Despite this BOA sometime stuck in local optima that result in slow or poor convergence. This study embeds the bidirectional search in the structure of BOA to enhance the convergence. The proposal is named as BBOA and experimented on a set of seven unimodal benchmark function along with three real world problems.