In this study, we developed and parametrized a friction model for finite element (FE) cutting simulations of AISI4140 steel, combining experimental data and numerical simulations at various scales. Given the severe thermomechanical loads during cutting, parametrization of friction models based on analogous experiments has been proven difficult, such that the cutting process itself is often used for calibration. Instead, our model is based on the real area of contact between rough surfaces and the stress required to shear adhesive micro contacts. We utilized microtextured cutting tools and their negative imprint on chips to orient chip and tool surfaces, enabling the determination of a combined surface roughness. This effective roughness was then applied in contact mechanics calculations using a penetration hardness model informed by indentation hardness measurements. Consistent with Bowden and Tabor theory, we observed that the fractional contact area increased linearly with the applied normal load, and the effective roughness remained insensitive to cutting fluid application. Additionally, we calculated the required shear stress as a function of normal load using DFT-based molecular dynamics simulations for a tribofilm formed at the interface, with its composition inferred from ex-situ XPS depth profiling of the cutting tools. Our friction model demonstrated good agreement with experimental results in two-dimensional FE chip forming simulations of orthogonal cutting processes, evaluated by means of cutting force, passive force, and contact length prediction. This work presents a proof of concept for a physics-based approach to calibrate constitutive models in metal cutting, potentially advancing the use of multiscale and multiphysical simulations in machining. Graphical abstract