A container-based cloud is a new trend in Cloud computing that introduces more granular anagement of applications and reduces overheads of virtual machine (VM). Compared with VM-based clouds, container-based clouds can further improve the energy efficiency with a finer granularity server consolidation in data centers. Current VMbased single level of server consolidation approaches cannot be used in container-based clouds because container-based clouds have two levels of placement: containers to VMs and VMs to Physical machines. Existing research lacks energy model and optimization algorithms that consider the joint allocation of containers and VMs. This work aims to improve energy efficiency in container-based clouds by proposing two bilevel energy models and three Evolutionary Computation (EC)-based optimization algorithms for three placement decision scenarios: initial placement of applications, periodic placement of applications, and dynamic placement of applications. The two novel bilevel energy models and three EC algorithms will contribute to better management of resources for energy efficiency in container-based clouds.