Spectral analysis of cross-hole sonic logging data for pile integrity assessment

Ensuring the safety of foundations requires advanced non-destructive testing techniques. Cross-hole sonic logging (CSL) is a widely adopted method for evaluating the integrity of deep foundation elements, such as bored piles, barrettes, and diaphragm walls. This method involves analyzing the propagation time and relative energy of ultrasonic pulses transmitted and registered by probes inserted into pre-installed access tubes. However, in certain cases, standard analyses may not effectively distinguish anomalous signals arising from defects and other factors unrelated to concrete quality. Our study explores the potential of an alternative frequency-domain approach for CSL data analysis. We propose new attributes to quantify ultrasonic signal spectra, including spectrum area, normalized spectrum area, weighted mean frequency, and maximum frequency. These attributes are automatically calculated, eliminating the need for additional data processing time and minimizing the risk of human error. The proposed approach was applied to CSL data collected from two bored piles of 46 m length and successfully identified signals classified as anomalous through standard time-domain analysis. Further research is deemed necessary to fully explore the potential of the frequency-domain approach in enhancing the information content and reliability of CSL pile integrity testing.