This paper focuses on experimental validation and investigation of the validity of a 3D beam model for representing the assembly of flexible stranded cables in automotive industry. For this purpose, an original test bench has been designed. It allows quasi-static manipulation of cables with a robotic arm and retrieval of the cables 3D centerline position with a system of cameras. Investigation is carried out in three parts. Firstly, the beam parameters are identified with a single buckling test set up on the bench. Along with buckling theory analytical developments, this test leads to homogenized bending and shearing stiffness parameters. The torsional stiffness is estimated from the shearing parameter and cables measurement. Axial stiffness is calculated from the composite theory. Then, other loading experiments are performed. The results of the latter experiments are compared to numerical simulations based on a geometrically exact beam model and using both the identified beam parameters and the initial geometry of the cables as inputs. Lastly, a numerical analysis of the effects on the final geometry of uncertainties, for both identified parameters and initial geometry, is performed. The whole process provides clues on the validity of beam models for representing stranded cables: it gives good results for the prediction of the final geometry and average results for the reaction forces ; torsional stiffness, bending stiffness and initial geometry seem as the most influential 1 parameters of the model ; a buckling test seems to be a viable mean for identifying cable parameters.