Macroscopic deformations of metals depend on the underlying, often complex, microstructure. This is especially true in the context of metals having a lamellar structure, as three scales (micro-meso-macro) exist. The microscale is built of an arrangement of lamellas. These lamellar packages form colonies or grains, which are visible on the mesoscale and differ in their laminate normal orientation. The macroscale is then an arrangement of colonies. To model and simulate these kind of structures using the Finite-Element method, efficient models and implementations are necessary. To this end, we present a combined analytic and computational homogenization scheme. The analytic homogenization step is built upon laminate theory, see e.g. \cite{Milton}, to describe the deformation behavior of two-phase rank one laminates. This allows us to circumvent the necessity to resolve each lamella in the Finite-Element model and allows us to efficiently compute the deformation mechanisms within a single colony with constant laminate normal orientation. To bridge the gap between meso- and macroscale, we use a computational homogenization scheme. We use the open-source toolbox NEPER to generate representative volume elements of multiple colonies with differing laminate normal orientations. We then prescribe periodic boundary conditions and loading scenarios, to study the effective properties of the lamellar microstructures. This allows a detailed insight into the deformation behavior of the complex microstructure while limiting the computational resources. As a model system, we turn our attention towards the creep behavior of lamellar FeAl alloys, which were previously investigated \cite{Schmitt}. This alloy has a lamellar microstructure composed of the phases FeAl and FeAl2. During solidification, multiple colonies with differing laminate normal orientation form, which can be efficiently investigated using our proposed model. We present the implementation details for the laminate material and periodic boundary conditions and show, that the ansatz is quite successful by comparing experimental results to the results achieved by the combined analytical and computational homogenization setup.