Experimental Study to Verify Elliptical Confidence Limit Method for Bolted Joint Tightening

Abstract

The calibrated wrench method is often used for tightening. When tightening bolted joints, it is important to apply high axial tension. However, since the axial tension is indirectly applied in this method, it varies and has a distribution in the case of tightening carried out in the production line of a factory, for example. However, the calibrated wrench method is still widely used because of the simple tool and easy standardization. In our previous papers, we analyzed and discussed the main points of this research by a theoretical approach as discussed below. Conventionally, this type of distribution has been considered to lie within a rhombus (more precisely, within a rectangular area). However, when considering the tightening torque and axial tension as independent random variables, the distribution becomes elliptical. The same idea applies to the relation between the tightening torque and the equivalent stress for a bolt axis based on shear strain energy theory. On the other hand, regarding the variation in the tightening torque (tightening work coefficient a) actually applied to a bolt, it was reported by Bickford, Kawasaki, and others that it can vary by 15% or more from the target (indicated) tightening torque. However, the torques for wrenches used at actual assembly sites or under lubricated conditions were not reported. Therefore, it is necessary to experimentally verify that the relation between the tightening torque and the axial tension (axial stress) and equivalent stress of a bolt axis is distributed in an ellipse. Furthermore, the screw-thread characteristics (torque coefficient, equivalent stress coefficient, coefficient of friction, etc.) during the tightening process should be clarified by an experimental approach and observation. Thus, in this study, in experiments under dry (as-obtained) and lubricated (Loctite 263) conditions, the tool (preset-type and dial-type torque wrenches) and bolt strength classification (8.8 and 10.9) were changed, and the screw-thread characteristics were observed during actual bolt tightening and the characteristics under different conditions were analyzed. It was clearly shown that the tightening torque and the axial tension (axial stress) of a bolt axis and the equivalent stress vary with an elliptical distribution rather than a rhombic distribution. Finally, the validity of the tightening theory based on the elliptical confidence limit method was also verified experimentally.