Experimental techniques to isolate dynamic behavior of bolted connections.

This paper discusses issues that arise in controlling high quality mechanical shock inputs for mock hardware in order to validate a model of a bolted connection. The dynamic response of some mechanical components is strongly dependent upon the behavior of their bolted connections. The bolted connections often provide the only structural load paths into the component and can be highly nonlinear. Accurate analytical modeling of bolted connections is critical to the prediction of component response to dynamic loadings. In particular, it is necessary to understand and correctly model the stiffness of the joint and the energy dissipation (damping) that is a nonlinear function of the forces acting on the joint. Frequency-rich shock inputs composed of several decayed sinusoid components were designed as model validation tests and applied to a test item using an electrodynamic shaker. The test item was designed to isolate the behavior of the joint of interest and responses were dependent on the properties of the joints. The nonlinear stiffness and damping properties of the test item under study presented a challenge in isolating behavior of t4he test hardware from the stiffness, damping and boundary conditions of the shaker. Techniques that yield data to provide a sound basis for model validation comparisons of the bolted joint model are described. INTRODUCTION Bolted joint behavior is known to be nonlinear and has a large effect on the overall energy dissipation of a structure during dynamic response [1,2,3]. An effort has been underway at Sandia National Laboratories to develop computational models for bolted joints [4] and validate these candidate models through experiments. Recent work at Sandia has focused on performing experiments to quantify the energy dissipation and stiffness characteristics of a specific bolted joint in a singular configuration [5]. The intent is to quantify and model the characteristics of the isolated joint at a small scale and then integrate the joint element into a larger model to predict responses of the overall system. The single bolted joint is pictured in Figure 1(a). The nonlinear behavior unique to this bolted joint comes largely from the nature of its inclined interface. Three of these bolted joints provide attachment points for the mock hardware package pictured in Figure 1(b). The bolted joints are the only load path into this hardware. Therefore, the behavior of the bolted joints must be understood and modeled accurately in order to adequately predict the response of subcomponents inside the package to forces applied on the input side of the joints. This hardware was designed to be a good system on which to learn about joint behavior and associated test techniques. Both the single-leg and three-leg configurations can be simply represented as single degree-of-freedom systems with a rigid top mass, equivalent linear stiffness and a nonlinear element to represent energy dissipation and the nonlinear stiffness (Figure 1(c)). Three pairs of mass-mock hardware were fabricated according to the same nominal shop drawings and they were assembled to form a total of nine combinations (combinations of three tops mixed with three bottoms). The range of responses for each test is representative of variation that could be expected due to unit-tounit variability. The purpose of this paper is to discuss issues that arise in controlling high quality mechanical shock inputs to mock hardware for the purposes of model validation. Tests were performed on an Unholz-Dickie T2000 electrodynamic shaker and tests were controlled either by a Spectral Dynamics Jaguar shock and vibration control system or by an in-house control package. The test was first approached as a typical shaker shock test in which it is acceptable to control the chosen input location only to the given shock response spectrum (SRS). However, normal guidelines for shaker shock testing proved inadequate for this work. Several lessons were learned about the intricacies of shaker shock testing and tools that can be used to control the test input in a way that acceptable responses can be gathered for model validation purposes.