Strain-based structural health monitoring: Computing regions for critical crack detection

The development process of aircraft structures requires many fatigue tests. During these tests, engineers have to inspect the aircraft structure continually to avoid final fracture of any component. Structural health monitoring may reduce time and cost associated with these inspections. Evaluating applied strain sensors is one possibility to monitor the structure. As applied strain sensors do not contribute sufficient information in every location for crack monitoring, test engineers have to position them carefully. Only in appropriate positions, strain sensors provide data that may be analyzed to detect crack growth before criticality, including scenarios with multiple cracks and different lengths. This paper presents a method to compute and maximize regions of appropriate positions for strain sensors on a test structure. The resulting detection regions may be used for appropriate positioning and minimizing the number of required strain sensors, while ensuring detection before criticality. The proposed method exploits the fact that growing fatigue cracks change the strain field in the structure. Therefore, the relative change of strain amplitude under periodic loading will serve as detection measure. Once it exceeds a specified critical threshold value that is linked to sensor sensitivity, final fracture is assumed to occur. Appropriate detection regions for strain measurement are computed in three steps: First, the space of crack configurations, each represented by a crack length vector, is probed by a Monte-Carlo sampling scheme. For every configuration, criticality is assessed based on stress intensity factors using a linear Finite Element Analysis. Second, a mathematical substitute model is trained using the sampling data. The result is a fast mapping from the crack configuration space onto the space of stress intensity factors. Third, for all critical crack configurations, the regions of critical detection measures are assembled into a global detection map. This map includes regions where detection is possible before a critical crack configuration occurs. The approach is applied to a demonstrator that resembles the section of an aircraft structure exhibiting two fatigue cracks. An extension to many cracks is discussed.