Although spark-ignited engines have a considerable development history, the relevant flow physics and geometry design implications are still not fully understood. One reason is the lack of experimental and numerical methods with sufficiently high resolution or capabilities of capturing stochastic phenomena which could be used as part of the development cycle. More recently, Large-Eddy simulation (LES) has been identified as a promising technique to establish a better understanding of in-cylinder flow variations. However, simulations of engine configurations are challenging due to resolution as well as modeling requirements and computational cost for these unsteady multi-physics problems. LES on full engine geometries can even be prohibitively expensive. For this reason, the size of the computational LES domain is here reduced to the region of physical interest and boundary conditions are obtained from a RANS simulation of the whole experimental flow domain. This approach required modifications to the compressible in- and outflow boundary conditions of the highly accurate structured LES framework. The extended method allows for oblique in- and outflows with an assumed velocity profile. Results for canonical test cases are presented to verify the approach and show its limitations. The extended numerical framework is used to study port flow under steady boundary conditions in a state-of-the-art gasoline engine geometry using boundary conditions from a RANS simulation for the whole flow domain. Time averaged results are validated against PIV measurement data and port flow performance parameters are evaluated. Additionally, comparison to the classical RANS approach is made. Effects of stochastic flow phenomena are analyzed in detail, which make port flow performance parameters like, e.g. tumble ratio, hard to predict by conventional ensemble-averaged methods. Fluctuations in pressure and velocity fields mostly originate from the complex in-cylinder flow, but propagate upstream to the chosen inflow boundary location.