Abstract
When tightening a large number of bolted joints, the calibrated wrench method is used. Since this method is indirect, the axial tension varies greatly in many cases. However, the calibrated wrench method is still widely used because of the simplicity of the tool and easy standardization. When the tightening torque and axial tension are considered to be two independent random variables, the axial tension (stress) is distributed within an elliptical confidence limit. Conventionally, it is thought that the shape of this distribution is a rhombus. Considering the permitted limit for a working load (stress) on a bolted joint, the elliptical variation has a larger margin to the yield point than that of a conventional rhombus. On the basis of this feature, we show in this paper that a higher tightening target torque and a higher axial tension can be set than before. By applying the elliptical confidence limit, one can obtain higher tightening torque and initial axial tension than the conventional values within a smaller range of variations. In this study, in the case of tightening a large number of bolted joints at factories and so forth, tightening reliability is considered as a problem associated with quality or process control and a probabilistic statistical method is investigated. Finally, we carry out analysis to establish the optimum tightening torque for bolted joints.