· Table of Contents
· Fundamental & Applied Research
Semi-direct measurements of ultrasonic pulse velocity in proposed concrete reference specimensChih-Hung Chiang
Department of Construction Engineering, Chaoyang University of Technology
Wufeng, Taichung, Taiwan 413.
Stress wave techniques such as measurements of ultrasonic pulse velocity are often used to evaluate concrete quality in structures. For proper interpretation of measurement results, the dependence of pulse transit time on the average acoustic impedance and the material homogeneity along the sound path need to be examined. Semi-direct measurement of pulse velocity could be more convenient than direct (through transmission) measurement. It is not necessary to assess both sides of concrete floors or walls. The current study concerns with the construction and application of concrete reference specimens for semi-direct measurement of pulse velocity. A novel measurement scheme is proposed and verified based on statistical analysis. It is very effective for gathering large amount of pulse velocity data from concrete reference specimens. The variability of measurements is comparable with that reported by American Concrete Institute using either break-off or pullout tests.
Keywords: pulse velocity; concrete; reference specimen
The nondestructive testing and condition assessment of residence buildings have brought significant attention in Taiwan since the devastated Chi-Chi earthquake in 1999. The most frequently asked questions by the owners or construction supervisors include the quality of concrete floor and the bond strength between reinforcing steel and concrete at critical positions such as beams and columns. Stress wave techniques are advantageous over other nondestructive methods in terms of ease of use for on-site measurements. Ultrasonic pulse velocity is perhaps the most well developed stress wave technique in the following aspect:
There are three alternatives for obtaining pulse velocity in concrete. Except the most common and perhaps most reliable direct measurement (or through transmission), semi-direct (or diagonal transmission) and indirect (or surface transmission) measurements may also be used with care [2, 3]. That the transducer arrangement requires the access of two facing sides of a concrete component sometimes limits the applicability of direct measurement. For floors and walls, the semi-direct and indirect measurement is more attractive for being less restrictive in transducer arrangement. The sound path length in a semi-direct measurement is approximately equal to the distance of a third side of a triangle formed by the adjacent sides of a concrete floor, as shown in Fig. 1. It is important to address the dependence of pulse transit time on the average acoustic impedance and the material homogeneity along the sound path. This is also one of the key issues that relate the variation in properties (mechanical and non-mechanical) of concrete to its local material homogeneity. The current study concerns with the construction and application of concrete reference specimens for semi-direct measurements of pulse velocity. A novel measurement scheme is proposed and verified based on statistical analysis.
|Fig 1: Transducer arrangement for semi-direct measurement in concrete reference specimens: (a) Front view, the sound path length is indicated by the dashed line. (b) Top view, two different measurements are taken at positions D2 and A12. Each small grid is 5cm×5cm..|
It is very common to calibrate ultrasonic testing apparatus against standard reference specimen before nondestructive inspection of metals. Reference blocks specified by ASTM E127, JIS STB-A1, and CNS STB-A1 are often used to calibrate ultrasonic flaw detectors prior to field application. The calibrated signal intensity and time-position characteristics are critical to proper interpretation of ultrasonic measurement results of metals. However, it is very rare that a reference specimen may be used in a similar fashion for measuring pulse velocity of concrete. There does exist some difficulties for such a specimen since the constituents of concrete are more complex than most metals and alloys. The fabrication processes contribute heavily on the variability in material properties of concrete structures. The most widely used cylindrical concrete specimens are usually for compressive tests instead of nondestructive testing. With regard to pulse velocity through concrete, most standards may simply state that the testing apparatus should be calibrated in order to check "the proper operation of the time-measuring circuit"[3, 4]. STN 73 1371 does mention the reference bar with regard to the precision of test apparatus . It is obvious that the focus is on the precision of transit time rather than the similarity between target and reference specimens.
As previously mentioned, the variability in material properties of concrete depends on the casting and curing processes. In addition to pulse velocity measurement of existing building, there is also a need to examine the quality of precast cellular concrete floor or other similar precast concrete components. ACI 523.2R-96 states the following ,
According to ACI 214, good field practice in making concrete would be indicated by a coefficient of variation in the range of 10 to 15 percent for compressive strength. For precast concrete, a coefficient of variation not greater than 10 percent should be maintained.
The pulse velocity is recognized by ACI C228 to be suitable for estimating relative compressive strength, providing a pre-established graphical correlation between pulse velocity and compressive strength is available [6, 7]. The authors herein argue that a reference specimen is necessary for practical estimation of compressive strength based on pulse velocity of concrete in structures such as precast concrete cellular floor. A proposed specification for a concrete reference specimen is described in the following.
30cm×30cm×15cm specimens were chosen for semi-direct measurements of pulse velocity. Type I cement was used in the concrete mixture. The weight ratio of cement/water/fine aggregate (dry)/coarse aggregate (dry) was 1.00/0.58/2.31/3.16. Specimens were cast and cured for one day in the mold, followed by 28 days in water. Compressive strength was 3000psi (or 20.7MPa or 211 Kg/cm2).
Through transmission pulse velocity measures the average over the entire thickness of a concrete layer such as a floor or a wall. 2% accuracy may be achieved if the specimen is reasonably homogeneous and if care is taken to ensure good contact between transducers and concrete surfaces. On the other hand, the semi-direct measurement measures the average over a sound path, which varies as the transducer arrangement changes. This is illustrated in Fig. 1. Note that the length of sound path is a function of the angle between the top surface of concrete and the line connecting two transducers. The top surface, or Surface T, may be marked by a simple grid system, as shown in Fig. 2. To distinguish various transducer arrangements, each small grid is numbered according to the positions of both transducers. For example, D1 indicates the measurement is taken by placing transducer R1 on top of the top left corner of Surface T. Transducer R2 is placed in direct contact with the front surface of concrete, or Surface F. These two transducers are always aligned so that adjacent measurements are reasonably independent on each other. One remark: the 5-cm interval in Fig. 1 is chosen solely for producing five data points (rather than three or less) along each line of measurements across the specimen. Discussion on possible selection of optimal combination of interval size and points will be given later on.
|Fig 2: Measurement scheme corresponding to arrangements of transducers as illustrated in Fig. 1(b). Each measurement is numbered and listed in the four tables surrounding the center table.|
The pulse velocity is strongly dependent on the positions of semi-direct measurements. If transducer R2 remains at the same position, than the pulse velocity in the center is usually lower than that near the edge of the specimen, as shown in Fig. 3. However, similar trend cannot be found along each line of measurements across the specimen. For example, line 3A is from the top (3329 m/s) to the bottom (3322 m/s). Line 1D is from the left (3864 m/s) to the right (4877 m/s). Every point in one particular line of measurements is taken along parallel sound paths of the same path length. Second order regression analysis was applied to each line and produced mixed results in different specimens.
|Fig 3: Pulse velocity in specimen No. 2 .|
Fig. 4 displays the regression curve of pulse velocity on distance from the top of Surface T. The curve is based on measurements of line 3C of specimen No. 1. Some interesting patterns can be identified as the following. Regression curves with highest value of the square of correlation (r2) often are associated with either maximum velocity in the center position or the opposite. In both cases, higher value of r2 can be obtained by averaging data taken from the same position on surface T and of the same path length such as the one-on-one pair of lines 3A and 3C. It happens because of one single measurement is easily affected by some unknown non-homogeneity or discontinuity along the sound path. This path is inclined from transducer R1 to transducer R2 and thus passes through different layers of concrete. The moisture content in different layers of concrete could change because of continuous hydration process of harden concrete. The average effect of the semi-direct measurement along the sound path is a function of great complexity. For hardened concrete, multiple scattering of micro cracks and unevenly distributed ingredients should all be part of such a function. Discussion of this sort will surpass the scope of the current study given the fact concrete is essentially an air-filled porous material.
Fig 4: Second order regression of (a) measurement line 3C (b) average of measurement lines 3C and 3A in specimen No. 1.
Specimens No. 3 and No. 4 both had an embedded plastic cylinder (y 5cm´15cm).The idea came from CNS STB-A1. The through-transmission pulse velocity of this plastic inclusion was 2420 m/s. The semi-direct measurements of these two specimens were affected significantly by the presence of such a plastic inclusion, as shown in Fig. 5.
|Fig 5: Pulse velocity in specimen No. 3 (with plastic inclusion).|
The quality control of concrete is usually measured by the strength test according ACI 214 . Test data from large concrete project frequently show a pattern similar to a normal distribution . To statistically analyze the data gathered in the current study, two different perspectives need to be brought out first. In-place tests are often performed on existing building for estimating the compressive strength of concrete. In-place strength estimate based on nondestructive methods, including pulse velocity, has been documented . On the other hand, ASTM C597-83 specifically objects the use of pulse velocity through concrete as a measure of strength. It did note that a velocity-strength relationship may "serve as a basis for the estimation of strength by further pulse-velocity tests on that concrete". Clearly the fabrication processes of concrete structures impose significant concern over the reliability of in-place strength estimation based on pulse velocity along. Fortunately, the credibility of pulse velocity has been improved as more and more measurements and analyses were published in recent years. To solve the dilemma between variability in material properties of concrete and the need for in-place (or on site) estimate toward quality control or safety assessment, we need to be more concerned about the use of reference specimens when possible.
The variability of the pulse velocity in the current study can be addressed in terms of its statistical distribution and the coefficient of variation [8, 10]. The former is not as ideally symmetric as the author would like it to be. It is, however, similar to a bell-shape distribution whose median (2981 m/s) is close to mean (3117 m/s). It is certainly possible to improve the symmetry by increasing the number of specimens. The latter is a measure of within-test variability, or Vwt, for evaluation of strength test results of concrete. Vwt can be calculated by 
where is the average range for all tests, is the average or mean value, and d2 is 2.326 (dependent on the sample size). Vwt in this study is comparable with other strength estimate method, such as break-off test and pullout tests. For example, the worst Vwt of average break-off tests is 13.3 and that of pullout tests is 14.8 . The Vwt of pulse velocity ranges from 3.4 to 32.9 in this study.
The statistical methodology used in current study does not include strict factorial experimental design. Two issues have been considered. First, lurking variables had to be identified. Second, whether the material homogeneity can be quantified based on semi-direct measurements was an open question worthy of research. After carefully examining the measurement results, the author conclude that pulse velocity, similar to other in-place strength estimators, can approach a symmetric distribution providing a large amount of data is available. The confounding effects of material homogeneity and sound path on wave propagation in concrete need to be further studied. In addition, the weight of a concrete specimen is a practical issue that is critical to the ease of measurement. Semi-direct measurement may be less accurate than direct measurement (through transmission) when measuring pulse velocity of a single point. It is, however, very suitable for taking large amount of measurements in reasonable short period of time. For precast or cast-in-place concrete floors and walls, semi-direct measurement is also more flexible and assessable than direct measurement. Indirect measurement [2, 6] sacrifices the efficiency by measuring multiple data points for a single velocity value and averages only a surface layer of concrete. The proposed scheme is illustrated by semi-direct measurements on reference specimens. It is very effective for gathering large amount of pulse velocity data. Although ACI reports an ideal test density of a grid from 30 to 60 cm, a much smaller grid should be desirable for establishing the reference for further measurements on the same concrete as part of a long-term condition assessment.
|© AINDT , created by NDT.net|||Home| |Top||