As network data become increasingly available, new opportunities arise to understand dynamic and multilayer network systems in many applied disciplines. Statistical modeling for multilayer networks is currently an active research area that aims to develop methods to carry out inference on such data. Recent contributions focus on latent space representation of the multilayer structure with underlying stochastic processes to account for network dynamics. Existing multilayer models are however typically limited to rather small networks. In this paper we introduce a dynamic multilayer block network model with a latent space represention for blocks rather than nodes. A block structure is natural for many real networks, such as social or transportation networks, where community structure naturally arises. A Gibbs sampler based on P\'olya-Gamma data augmentation is presented for the proposed model. Results from extensive simulations on synthetic data show that the inference algorithm scales well with the size of the network. We present a case study using real data from an airline system, a classic example of hub-and-spoke network.