Bolted Joint Analyses for Composite Structures - Current Empirical Methods and Future Scientific Prospects

摘要

One purpose of this paper is to describe and explain some highlights in the history of the analysis and design of bolted or riveted joints in fiber-polymer composite laminates. The second is to project into the future, where it might become practical to replace today's empirical analytical tools, which need copious test data to enable them to be applied, with physics-based composite failure criteria of universal applicability that need very few intrinsic material properties. The paper focuses on the analysis tools and is not a compendium of available test data. The paper begins with a review of the governing geometric parameters, along with their influence on which failure mode will dominate. The concept of optimum joint geometries is introduced, to show how to maximize the gross-section laminate strength and, thereby, to minimize the weight of these structures. It progresses to the two widely used analysis tools, developed years ago at the Long Beach and St. Louis divisions of the former McDonnell Douglas Corporation. The Douglas model is applied via explicit formulae, in conjunction with an empirically established stress-concentration relief factor. The latest forms of these equations are presented here, having changed little since they were first proposed almost 30 years ago. The McDonnell model, identified by the code name BJSFM is also based on algebraic solutions, but is encoded to enable a more thorough assessment to be made of the entire stress field, rather than just the most critical two locations. This code also relies on empirically determined correlation factors, in the form of characteristic offset distances, sometimes mistakenly believed to be true material properties. These same factors can also be applied to modify the predictions of finite-element analyses. The importance of using simple comprehensible models is stressed. The former model can also be used with the non-linear multi-row computer code A4EJ, again with very little need of test data - provided that the fiber pattern does not differ excessively from the quasi-isotropic lay-up. Excessively orthotropic fiber patterns are shown to be unacceptably weak at bolt holes and in need of a disproportionately large number of experimental test data to cover the many additional failure modes that cannot occur for close-to-isotropic fiber patterns. The paper closes with a glimpse of what the new SIFT (Strain-Invariant Failure Theory) for fiber-polymer composites, developed within the Boeing Company and already accepted for use at many locations outside, might do for this technology in terms of finite element analyses that need only five intrinsic material properties for each fiber-polymer combination, regardless of fiber pattern and joint geometry. The opportunities for drastic reductions in the cost of test programs, and accelerated schedules through not having to wait for specific tests for each joint, are clearly very powerful.