The United States Air Force (USAF) has an aging aircraft fleet with high operational demands that requires mission success in a variety of environmental conditions. The USAF has increased its demand for sustainment of aging aircraft (over 25 years) in service, and knowledge of a system's structural integrity is of vital importance in determining the operational status of an aircraft [1]. Structural integrity status is usually obtained through scheduled maintenance inspections, which are time consuming and expensive because they usually require disassembly of a structure so visual or other types of non-destructive inspections can be made. The information gathered on a system's structural integrity through Structural Health Monitoring (SHM) detection methods should result in reduced costly maintenance inspections, enhanced safety, and system failure predictions [2]. This paper focuses on evaluating a "hot-spot" SHM approach on a test article that accurately represents the complex geometry of a relatively inaccessible bulkhead section of an existing aircraft in which fatigue cracks have propagated. The SHM approach being evaluated uses piezoelectric generated Lamb waves, guided elastic waves in a plate, to detect electrical discharge machine (EDM) cuts that simulate real cracks in a test article. Because the damage occurs in a region of restricted geometry, piezoelectric sensor pairs must be placed in close proximity to each other, but on opposing sides of the expected crack. The close proximity of the piezoelectric sensors and restricted geometry create challenges in determining which portion of the collected response signal should be analyzed. The first approach evaluated is a tuning approach where specific excitation frequencies are chosen based on the theoretical arrival times of the S-0 and A(0) Lamb wave modes.. The window lengths are reduced in size to minimize the contribution of reflected energy in the response. The second approach involves evaluating responses over a wide range of excitation frequencies in combination with only evaluating the responses in the dominant mode's theoretical time of arrival window. A simple damage metric is applied to demonstrate crack detection and length estimation.
