Computer simulations are a standard tool for improving vehicle safety. In these simulations, predictions about dummy responses and injury assessment values can be made. For accurate predictions, the behaviour of the retractor as a major part of the seatbelt system has to be known. Tests are needed to generate this knowledge and incorporate it into a simulation model. Standard sled tests are too expensive and generally have too much deviation to be a useful correlation environment. Component tests are of limited use due of the lack of interaction, or the coupling between the different crash phases. Subcomponent tests are only useful if a robust simulation model already exists. Furthermore, a model structure is needed which reflects all main effects of the retractor in a time independent way.

Thus, there are two needs for an enhanced modelling process:​ A correlation test device as well as a model concept which reflects the interaction in a simple and robust way.

This paper demonstrates a new process on how a retractor model can be correlated in different solvers with a component test, within the typical working points of a retractor. The improved process is based on a new easy test assembly for retractors (ETAR) and on a general model structure (GMS) for the retractor models. The new component test assembly reflects the three phases of pretensioning, coupling and load limiting of a frontal crash without the need of a sled and/or dummy. Furthermore, for the correlation of retractor models in different solver codes, ETAR allows to generate test data in a fast and simple way with low deviation. The GMS implements all the main functionalities of a retractor and due to the GMS, the tuning of the models is easily transformed into other solver codes, commonly used for crash simulations.

Correlations between test and simulation for different load-cases and different retractors in different solvers demonstrates the applicability of ETAR and GMS for an improved retractor modelling.