I present a heat conduction model motivated by the Gaspard-Gilbert model. It consists of billiard disks confined by fixed scatterers near lattice points, which exchange energy with their neighbours through some weak potential. The model is studied in the two-stage approach of Gaspard and Gilbert. First we take a weak coupling limit with the appropriate time scaling to obtain a Markov process for the energies, i.e. an interacting particle system. Second, we try to understand the hydrodynamic behaviour of this interacting particle system – which is very similar to the one that Dolgopyat and Liverani obtain as the weak coupling limit of their interacting smooth Hamiltonian systems. The discussion contains many heuristic arguments. I will try to argue why the model is in some sense “better” than the original Gaspard-Gilbert model.