Controller placement is an important problem in software defined networks. Previously, some works addressed the problem by taking into account primary factors, such as propagation delay of control paths and capacity of controllers. Here, the problem is addressed from a novel standpoint of energy consumption, which is modeled as a binary integer program (BIP). In the BIP model, the energy consumption of the network that serves for the control traffic is minimized under the constraints of the delay of control paths and the load of controllers. In consideration of the high complexity of the BIP in large networks, a genetic heuristic algorithm is designed to find an effective sub-optimal solution. Simulation results show that the energy consumption of the heuristic algorithm is close to that of the BIP solution. No more than 4%, additional links are used by the heuristic algorithm if all the links have the same energy consumption.