In this thesis, a new method for non-destructive testing (NDT) of cylindrical structures is developed. The aim is the detection of defects as well as the determination of their positions. In order to avoid the time-consuming scanning of the whole structure, guided elastic waves (Chapter 1) are used. The scattered field resulting from the interaction with a possible present defect is measured at different locations. Instead of analyzing these complex time signals to determine the presence and position of a defect, the recorded signals are evaluated with a time reverse numerical simulation (TRNS).

The main idea of this method is the time reversal of the wave propagation phenomena (Chapter 3). If the recorded time histories of a scattered field, generated by a defect, are time reversed and played back in an identical structure, the waves travel back and interfere constructively. This leads to an amplitude maximum at the position of the defect.

Therefore, an experiment is performed in a defective tube and the displacements of the scattered field are measured at several points along the circumference at one end. Since the TRNS method only works if all three displacement components are played back, a three-dimensional vibrometer is built (Chapter 4). Instead of playing back the time histories in the experiment, this step is replaced by a simulation. The main reason is the determination of the position of the maximum amplitude which is easy in a simulation by monitoring the displacement and stress components. Even if no defect is present in the simulated sample, the played back displacement histories interfere, and the maximum amplitude is reached at the exact position where the defect was located during the experiment. A three-dimensional finite-difference code in cylindrical coordinates is used in the present work. In order to determine the exact axial and circumferential position of the defects, a fourth-order algorithm is used (Chapter 2).

Different experiments are performed with axial and circumferential notches as well as a notch that is orientated at a 45 degree angle with respect to the axis of the tube. The defects are at a distance of 0.8 m from the measurement position and are only part-through the thickness of the tubes. The experimental results show that the defects are detected in all cases and that the axial, as well as the circumferential positions and the orientation of the defect could be determined.