The Hemming Process: A Numerical and Experimental Study

Doctoral thesis, 2003

Finite element simulations of hemming, a method used in the automotive industry to join two sheet metal panels by bending the flange of the outer panel over the inner one, are dealt with in this thesis. This method is used mainly for assembly of closures in automotive bodies (i.e. side doors, hoods, trunk lids and tailgates). Therefore, hemming is considered by many people to be a joining method. Nevertheless, as hemming also has many properties in common with sheet metal forming, some of the knowledge gained in research on sheet metal forming can also be applied to hemming and vice versa. There are some defects that can be associated with the hemming operation. The work presented in this thesis concentrates on predicting the reduction in size of the outer panel during the operation, known as roll-in. This reduction in size has to be compensated for, in the flange die for the outer panel, to obtain an assembled part with the correct dimensions. Today this compensation is based on practical knowledge from simple experiments and on experience gained from similar parts already in production. However, a large amount of both time and money could be saved if the roll-in could be predicted with high accuracy at an early stage of a project. The final goal is to simulate all of the hemming steps of production parts. To make three-dimensional simulations of hemming possible within reasonable simulation times, it is necessary to use shell elements. A problem in hemming simulations is that the radius of curvature of the outer part in the folded area is very small, normally of the same order of magnitude as the sheet thickness. This implies that the conditions for a shell theory to hold are far from being fulfilled. The first half of this thesis is therefore an investigation of the order of magnitude of the errors resulting from the use of shell elements in FE simulation of hemming. To estimate these errors, a two-dimensional FE model of a cross section of a flat part with a straight flange is investigated using both solid and shell elements. The results from these models are then compared with each other and with those from experiments conducted in this project. The solid element model in this study used a new implementation of the Barlat ’91 material model. This implementation assumes plane strain conditions and planar anisotropy; it also uses an iterative, backward-Euler integration and is implemented in the FE code LS-DYNA. The second half of the thesis is devoted to three-dimensional simulations of the hemming of an automotive hood. Two versions of the explicit FE code LS-DYNA and five shell elements were used in these simulations. The influences on roll-in from adhesives, anisotropy, and uniform pre-straining were also studied. Results from a forming simulation were mapped to the flanging and hemming ones in order to study the influence from the stamping of the outer panel on the roll-in.