Investigation into the Behavior of Bolted Joints
Models to capture the physics of jointed structures have been proposed for over 40 years. These models approximate the behavior of the joint under carefully developed operating conditions. When these conditions change, the model has to be changed. Recent developments in numeric codes like finite elements have created interest in incorporating joint models into the design process but joint models need to represent the joined structure over a broader operating range. This work investigates the dynamic response of a structure with a joint. Isolation of a few dominant effects may give way to a model able to capture a broader operating range. To isolate the effects of the joint two specimens were created. A specimen that is without a joint serves as a control. The second specimen is geometrically similar and contains a double lap joint with a bolt fastener. The differences between the specimens represent the effects of the bolt. Control variables of bolt tension, excitation level and sampling time were chosen. Amplitude response and hysteresis curves were recorded. This data was used to examine the non-linear response of the bolted specimen. Qualitative observations are included. The control specimen shows little effect from non-linear behavior in the frequency response. The bolted specimen shows non-linear behavior in the frequency response. When the joint is introduced to the geometry the system drops in amplitude, drops in resonant frequency, and demonstrates a non-linear softening effect. As the initial bolt tension is reduced the magnitudes of these changes increase. In addition when the system is allowed to dwell with a single sine wave at resonance the amplitude of the response often increases. Hysteresis curves reveal that more than a softening non-linearity affects the response. The curve shows a softening affect when displacing in one direction and a hardening affect when displacing in the opposite direction. This may be affected by the geometry as the control specimen demonstrates a tri-linear stiffness. It is evident that previous joint models do not capture all of the effects observed. Additional research to link the physical cause to the observed affect will aid in adjusting or creating a joint model to be used in numeric codes.
School:Wright State University
School Location:USA - Ohio
Source Type:Master's Thesis
Keywords:bolted joints joint damping structural mechanics of
Date of Publication:01/01/2006