In this work, we consider a heterogeneous network consisting of macro and pico base stations. Our goal is to associate users to one of the base stations,while minimizing the energy consumption of the network. Moreover, the algorithm should consider constraints of a real system and thus, only include the necessary complexity that helps reducing the energy consumption. Therefore, we first formulate the energy minimization problem in three different manners, each one including a different level of modeling details. The solution of each problem is analyzed to identify the important factors to consider when designing an energy‐optimized cell association algorithm. Based on this analysis, we then develop a practical energy‐aware cell association heuristic. Finally, the developed algorithm is evaluated using complex system simulations. Results show that significant energy savings are achievable by the developed algorithm compared with the conventional cell association in cellular networks. Copyright © 2016 John Wiley & Sons, Ltd.