This paper addresses industrial gases supply chains involving multiple products at multiple plants that must be coordinated with multiple depot-truck-routes in order to satisfy customer demands. The full-space optimization problem corresponds to a large-scale mixed-integer linear programming model (MILP). To, solve large-scale industrial problems, this paper proposes a rolling horizon approach with two aggregation strategies for solving the smaller subproblems. The first one relies on the linear programming (LP) relaxation for which the binary variables (complicating variables) of the distribution problem are treated as continuous, while the second one uses a novel tailored model for the distribution side constraints that leads to improved solutions. A real case study of an industrial gases supply chain has been addressed obtaining good results in both objective value and with lower computational effort compared with the full-space solution. The extension to longer time horizons through a receding horizon is also considered.