Analysis Methodology Determination For Calculating Load Distributions of Complex Fastened Joints

Abstract

Fastened joints are everywhere on an aircraft, especially on a built-up, riveted airframe. These joints, particularly in the rotorcraft environment in metallic structures, are sources of stress concentrations that lead to fatigue damage over time. Understanding the load distribution and stresses throughout fastened joints is critical to ensure the aircraft is structurally sound and has proper inspection intervals to find incipient fatigue damage or cracks shortly after initiation in structural components prior to a failure. Columbia Helicopters, Inc. (CHI) Operations has seen these failures over more than six decades while operating large, transport category, helicopters. CHI Operations' in-field experience of repairs to various components (skins, webs, and frame caps), while generally benign and captured with existing inspection criteria, can lead to more serious incidents including the loss of an aircraft. With this, CHI has an abiding interest in researching built-up airframe loads and tools for studying fastened joints. In this report, four methods of calculating loads throughout a fastened joint are studied: 1) the classic low-fidelity calculation utilizing Bernoulli-Euler beam methods, 2) a test-supported calculation method created by T. Swift of the Federal Aviation Administration (FAA) for a two layer lap-splice fastened joint, 3) aerospace-industry proven Finite Element Analysis (FEA) processor NASTRAN, and 4) BoltJoint10, an FEA method created for the helicopter industry specifically for fastened joints. A basic lap-splice fastened joint was used as the baseline to determine which analysis methodology will provide accurate load distribution between the two layers. The hand calculated and FEA results were validated to classic Euler-Bernoulli beam hand calculations that have been proven for slender beams. Once a method was settled on, the research continued by evaluating the agreement between BoltJoint10 and NASTRAN results with two fastened joints of increasing complexity. This evaluation determined if the increasing complexity affected the good agreement between the two FEA methodologies which was previously observed with the basic lap-splice model. The conclusion of this research is both the Swift hand calculation method and NASTRAN FEA method are within 10% of the classic hand calculation at the critical fastener, while BoltJoint10 exceeds 20%. In addition, the agreement between BoltJoint10 and NASTRAN FEA methodologies further deviate as the complexity of the fastened joint increases. The author suggests using NASTRAN FEA to determine load distributions throughout a joint.