In our group at KTH’s Fibre and Polymer Technology Department, we use molecular dynamics (MD) simulations to investigate the transport of nanoparticles and macroparticles through complex materials, with the aim of developing adaptive separation systems. The continuing development of thinner layers and nanostructured materials requires an understanding of processes at atomic and molecular scales, where local interactions govern macroscopic material properties and transport behaviour. We therefore combine atomistic and coarse-grained modelling with experimental measurements. Atomistic simulations provide a detailed description of surface chemistry and particle interactions, while coarse-grained models enable simulations of millions of particles over microsecond time scales. The two levels are connected by using umbrella sampling to calculate potentials of mean force, which are then used to parameterize and optimize tabulated interaction potentials for the coarse-grained model. This multiscale approach combines molecular-level accuracy with the computational efficiency needed to study large systems and long-time transport processes. Accurately representing long-range electrostatics, explicit solvent effects and van der Waals interactions adds substantial complexity and computational cost, making access to high-performance computing resources essential. Advances in protein- and nanoparticle-based therapies, including mRNA vaccines, have created a growing need to separate large biomolecules and particles from complex mixtures. Although membrane filtration offers high throughput, its selectivity remains limited, and current separation technologies cannot easily adapt to particles with different sizes, charges, and interaction properties. Dynamic cellulose nanofibril (CNF) networks offer a promising approach to overcoming these limitations, which is what we investigate in the ERC starting grant DynanoNet. Changing density, surface chemistry, and fibril arrangement can alter the network pore size and transport pathways. It is therefore important to establish how these network parameters govern particle transport and whether they can enhance separation selectivity and resolution, ultimately providing the basis for a dynamically tunable separation platform based on nanofibril networks. Understanding how particles move through these networks is essential for designing adaptive separation materials. However, experimental methods provide limited access to the nanoscale dynamics and interactions that govern particle transport. MD simulations can address this limitation by resolving the structure, dynamics, and interactions within microscopic networks at the molecular level. Using MD simulations, we will investigate transport of of particles to determine how network density, particle size, and particle-fibril interactions govern mobility, retention, and permeation, and thereby identify the system parameters that improve separation efficiency and selectivity. To investigate transport through CNF networks at experimentally relevant length and time scales, we use a coarse-grained bead-spring model implemented in the ESPResSo simulation package. Our simulation model incorporates tens of thousands of coarse-grained beads, enabling meaningful comparison between computational predictions and experimental observations. Furthermore, we plan to perform atomistic and coarse-grained simulations in LAMMPS. This simulation develops a SUPRA model for describing specific interactions between the CNF network and target particles, such as proteins, for which an accurate representation of molecular interactions and surface chemistry is essential.