Rationalizing evaluation of bolted joints

Sammanfattning

Bolts in trucks are subjected to cycled loading and are therefore prone to fatigue failure, mostlikely under the bolt head or at the start of the thread. When evaluating a bolt for fatigue,evaluation surfaces must be created manually at the critical positions in a finite element solverwhich is time consuming. The purpose of this project is to reduce the amount of manual inputby developing a method for calculating the stress at the thread start, without an evaluationsurface. Furthermore, bolts with reduced shank have longer fatigue life but the amount ofexperimental test data available is limited. Therefore, in this project, a fatigue test in bendingon bolts with reduced shank will be conducted. By creating an evaluation surface of a bolt in a finite element model, the bending moments fora cross section can be found. These are then used in a polynomial that calculates the axialequivalent stress for the cross section. In a reference method, evaluation surfaces are made bothunderneath the bolt head and at the start of the thread while in the new method developedin this thesis there is no evaluation surface at the start of the thread. Instead, the bendingmoments are extrapolated from the head evaluation surface. By doing a free body diagram ofa bolt, an expression for calculating the bending moment at the start of the thread is derived.The expression states that the bending moment at the thread can be found if the shear forcesand bending moments at the head evaluation as well as the distance between the bolt head andthread start is known. The new evaluation method is implemented in a script. First the distance between the bolthead and start of thread is found by using the head evaluation surface and a node set thatis created automatically when the thread is modeled. Thereafter, the bending moments areextrapolated and the axial equivalent stress is calculated. A finite element model of a boltedjoint is created in Abaqus, and the two different stress evaluation methods are compared. Abenchmark study is performed, where to two methods are used on an existing problem, toinvestigate the differences in a real-life application. A fatigue test in bending is performed on bolts with reduced shank. The test is replicatedin a finite element model, in order to find the axial equivalent stress in the critical areas. Anotherpurpose of the finite element model is to apply the different stress evaluation methods tosee how the results of the extrapolation method are affected by bending. The results show that the difference in axial stress amplitude between the methods is less than6 %, with some exceptions. If contact damping is used in the model the difference increases, dueto a damping force which makes the equilibrium equations invalid. However, a damping forcemakes the model unrealistic, which means that it should be avoided even with the referencemethod. When there is a lot of bending in the bolt, the error can be more significant becausethe evaluation surfaces have different local coordinate systems. By changing the basis of theforces before the extrapolation, the error can be reduced, but it requires an evaluation surfaceat the thread start, which does not exist when using the extrapolation method. However, inmost applications, the angle is small and does not cause an error larger than 6 %. The bending test shows indications that the bolts with reduced shank perform better in fatiguethan regular bolts. For future work, the bolts must be tested in axial fatigue to be able to fitthe polynomial that calculates the axial equivalent stress.