For many physiological phenomena, such as tissue deformation or growth, an adequate perfusion model of the underlying hierarchical vascular network is of great interest. A fully resolved model on multiple spatial scales is still unattainable due to the mostly unknown complex structure of the vascular network. In order to obtain sufficiently accurate results with practical computational resources, a simpler perfusion model is required that still takes into account the hierarchically complex structure over different scales of the tissue and vessels. For this purpose, the vascular network is partitioned into a coupled multi-compartment model where the various hierarchies are represented by the compartments as continua with macroscopic laws. The approach is to describe both the microscale and the hierarchies above as a porous medium using Darcy-type flow models that account for the different spatial scales. Based on synthetic vascular network model data essential parameters are determined for each compartment using averaging procedures. The interaction between different compartments is considered via pressure-dependent mass exchange and applied in an averaged sense, as are the boundary conditions. For the solution of deformation-dependent phenomena, geometry dependent parameters such as the permeability would have to be determined anew in each iteration step. Since this is not in the interest of efficiency, a model for growth-induced deformations was devel- oped that simply updates the model parameters with deformation-dependent quantities only. We illustrate an application to a liver growth model, where the microcirculation and the different hierarchies of the supplying and draining vascular trees are each modeled as an independent compartment, while the upper hierarchies not suitable for homogenization are considered as boundary conditions.