Non-resolved thermal infrared observations enable studies of thermal and physical properties of asteroid surfaces provided the shape and rotational properties of the target are well determined via thermo-physical models. We used calibration-programme Herschel PACS data (70, 100, 160 microns) and state-of-the-art shape models derived from adaptive-optics observations and/or optical light curves to constrain for the first time the thermal inertia of twelve large main-belt asteroids. We also modelled previously well-characterised targets such as (1) Ceres or (4) Vesta as they constitute important benchmarks. Using the scale as a free parameter, most targets required a re-scaling ~5% consistent with what would be expected given the absolute calibration error bars. This constitutes a good cross-validation of the scaled shape models, although some targets required larger re-scaling to reproduce the IR data. We obtained low thermal inertias typical of large main belt asteroids studied before, which continues to give support to the notion that these surfaces are covered by fine-grained insulating regolith. Although the wavelengths at which PACS observed are longwards of the emission peak for main-belt asteroids, they proved to be extremely valuable to constrain size and thermal inertia and not too sensitive to surface roughness. Finally, we also propose a graphical approach to help examine how different values of the exponent used for scaling the thermal inertia as a function of heliocentric distance (i.e. temperature) affect our interpretation of the results.