Adding filler particles to polymers has influence, among others, on the strength of the material and thus, in turn, affects the load a part can bear until failure. It has been reported from experiments in the context of polymer nanocomposites that this strengthening becomes more pronounced with increasing filler content. In order to understand this observation and to develop targeted strategies to improve the material performance, simulations can supplement experimental studies and reduce the effort and cost required for material development. To this end, however, precise numerical models are required which can appropriately capture the relevant processes across various length and time scales: Particularly in case of polymer nanocomposites, length scales cover the nanometre scale up to the typical dimensions of specimens and entire parts. This, in turn, necessitates an accurate description of the material behaviour at the different levels of resolution, which can be obtained only partly or only indirectly from experiments. This is most obvious for the vicinity of the filler particles, where, within a distance of some nanometres, the properties of the polymer deviate from those of the neat polymer. To obtain property profiles in this polymer-particle interphase required for simulations at higher length scales, we present a methodology integrating molecular dynamics and finite element simulations~\cite{Ries2021,Ries2022}. There, based on molecular simulations, we first characterise the material behaviour separately for the filler and the neat polymer and combine both in subsequent numerical "pseudo" experiments. The mechanical property profiles for the interphase then form the basis for simulations at the level of representative volume elements where direct comparison to experimental findings becomes possible. In the long-term, failure of polymer nanocomposites shall be investigated numerically capturing the peculiarities arising from the interphase at the nanometre scale.