Functional composition is a powerful and convenient approach for the geometric modeling of (boundary-conforming) architected materials. The global (macro) geometry of the structure is represented through a standard spline model, while reference (micro) cells are embedded into the global geometry by means of functional composition. As a result, complicated porous designs can be easily created [1]. On the other hand, their simulation is quite challenging. When micro and macro scales are not clearly separated, homogenization hypotheses are not fulfilled, and full scale resolutions are required. By properly discretizing both the macro and micro shapes, such family of geometrical models are finite element analysis ready. However, they present a huge number of degrees of freedom, what makes them computationally intensive. In order to solve such models fast, and with low hardware requirements, we propose a novel domain decomposition framework. In a first stage, we assemble the finite element operators with a multi-scale procedure [2]. Second, a tailor-made (inexact) FETI-DP solver, combined with reduced order modeling techniques, is applied to the resulting linear system. Thus, instead of solving one single large system, the resolution involves local problems defined over the cells composing the full geometry. During assembly and resolution, we exploit both the presence of two scales and the similarities among cells. This framework allows to simulate very large models, with hundreds (or thousands) of cells, that involve millions of degrees of freedom, without applying homogenization techniques, in a matter of seconds (or very few minutes) using an off-the-shelf laptop. As a result, we end up with a fast analysis method that can be naturally integrated in a design loop. Thus, design optimization examples of architected materials will illustrate the viability, performance, and great flexibility of the developed methods. References [1] Antolin, P., Buffa, A., Cohen, E., Dannenhoffer, J. F., Elber, G., Elgeti, S., Haimes, R. and Riesenfeld, R. Optimizing Micro-Tiles in Micro-Structures as a Design Paradigm. Computer-Aided Design, Vol. 115, pp. 23–33, 2019. [2] Hirschler, T., Antolin, P., and Buffa, A. Fast and multiscale formation of isogeometric matrices of microstructured geometric models. Computational Mechanics, Vol. 69, pp. 439–466, 2022