Sample size effect in ultrasonic testing of geomaterials - numerical and experimental study

Abstract

Nondestructive evaluation of civil structures is of increasing interest to utility owners. Several methods exist to evaluate different properties of concrete, pavement, cemented sands and others. UPVM is the most commonly used ultrasonic technique in civil structures due to its simplicity and ease of use. UPVM is fast and requires minimal skill from operators. It has been used for flaw detection, study of material contents, deduction of general deterioration, determination of elastic properties , measurement of strength, and others. In such applications, accurate measurements of velocity are essential for proper parameter evaluation and thus to increase the validity of conclusions obtained from measurements. Previous research in ultrasonic pulse velocity have found that UPVM are susceptible to specimen size, attenuation and frequency but no clear conclusions have yet to be made on the fundamental reason for the differences.

This work seeks to identify the main factors responsible for velocity differences due to specimen size and measuring frequency in civil engineering materials. The effects are investigated by first performing numerical simulations of concrete specimens of varying sizes, and properties, excited by both a low (55 kHz) and high (850 kHz) frequency input source. Simulations are used to model wave propagation in cylindrical concrete specimen. Transducer sound fields are also numerically studied using known analytical solutions. An experimental program is conducted to study variations in UPVM in 12 mortar and 11 concrete cylindrical specimens of varying widths and heights caused by different measuring frequencies.

Simulations are completed for 12 specimen of different dimensions having heights of 5,10,20 and 30 cm as well as diameters of 10, 20 and 30 cm. Both a low (f = 55 kHz) and high (f = 850 MHz) frequency input source are used on each specimen. Numerical simulations using low frequencies are made for both a damped and undamped series of specimen. Results from low frequency simulations of damped models indicate that wave attenuation can lead to significant errors in first arrivals when complex wave interference is present. Conditions for wave interference at the receiver location are studied and minimum size conditions for both height and width are derived. These conditions guarantee proper pulse separation for UPVM and are dependent on source size, and source pulse width. It is argued that with proper use these conditions will lead to accuracy of measurement better than one quarter of a period of the main excitation frequency when using a full waveform and a skilled operator.

Finally, experiments are performed to assess differences in first arrivals between high and low frequency measurements. Readings are made on 11 mortar and 12 concrete specimen of different heights and widths. Experimentally significant time differences are observed between high and low frequency readings. It is found that differences in first arrivals will increase with specimen length but differences in velocity will decrease with length. Specimens 4 wavelengths in height are deemed sufficient to diminish surface effects to a minimum provided the specimens are healthy (e.g. no internal flaws). Any increase past 4 wavelengths is found to have negligible effects on measured velocity in healthy specimens.