· Table of Contents
· Industrial Plant & Structures
Detecting Inner Quality of Foundation Pile with Ultrasonic CT TechnologyWANG wu-ping, LUO qi-xian, SONG ren-xin, FU xiang
Nanjing Hydraulic Research Institute, Nanjing, Jiangsu Province, China 210029.
Ultrasonic is widely used in detecting inner quality of foundation pile. With ultrasonic CT, we can gain velocity distribution of concrete, identify location and dimension of abnormal sections, deduce the property and intensity of defects. Based on the study on inverse methods for ultrasonic CT, UCT Program(The program for detecting inner quality of foundation pile with ultrasonic CT) is developed. The key technology for application of ultrasonic CT is summarized by analog computation and model tests. A segment of a pile is detected with ultrasonic CT, images of CT including isoline, chromatogram and three dimension distribution of velocity are presented. Series-wound transducer consisting of ten separated transducers is designed in field test. Thus, the efficiency and precision of detection is improved notably.
Keywords : ultrasonic, CT, foundation pile, detecting, images of CT
Bored concrete piles are commonly used in foundations of bridges and multi-story buildings. In order to improve the bearing capacity of a pile,the length and diameter of the pile tend to be designed larger and larger. Low-strain test method isn't suitable for inspecting this kind of pile because the energy of hammer is limited and will be worn off in the course of propagation. Ultrasonic method has high precision. It is not influenced by length, diameter and defects number of a pile. It can also provide intensity of the concrete. Therefore, this method is now used wider and wider in detecting inner quality of foundation piles. Since 1986, we have detected more than 1,500 piles with this method. Recently, We have detected some piles of four bridges( Wuhu Yangzi River Bridge,Nanjing Yangzi River second Bridge, Runyang Yangzi River Bridge and Jiangyin Yangzi River Bridge) at the downstream of Yangzi River. Of all these piles, the maximum length is 100m and the maximum diameter is 3.0m.
Fig.1a is the principle of ultrasonic testing. To detect a pile using ultrasonic method, ultrasonic tubes need to be welded or fastened to a reinforcing cage in advance. According to the diameter of the pile, 2,3 or 4 tubes can be chosen to be distributed which will result in 1,3 or 6 measuring sections correspondingly(fig.1b). Before detectionall tubes have completely to be filled with clean water. With a detection step of 20~50cm, transducers in every two tubes are moved from top to bottom(or from bottom to top) synchronizedly. Ray travel time t and amplitude A of the ultrasonic signals are measured. This is called sweeping inspection. If a received signal is abnormal (such as ray travel time increment, amplitude attenuation, wave aberration), more measurements needs to be done near this point. In order to identify the location and dimension of defects in a pile, cross measurements are required additionally.
|Fig 1a: Principle of ultrasonic testing.||Fig 1b: Layout of ultrasonic tubes.|
Fig.2(a) is the layout of rays in this conventional ultrasonic method. While ticking parallel and cross rays whose travel time and amplitude are abnormal on the figure, the section surrounded by these rays shows anomalies in a pile. But as a received signal represents the average quality of concrete where a ray passes through, it is difficult to obtain the true velocity and deduce the declining degree of the anomalies. In summary, the conventional method depends largely on experience and suffers from low precision, or even causes error in special instance. For example, when anomalies are situated on the bottom of a pile and distribute peripherally, see fig.2(b). Since signals of all rays except 1-1 and 2-2 may be abnormal, it is possible to deduce that the anomalies distribute entirely on the lateral section.
|Fig 2a: Layout of rays in conventional and CT method.|
Ultrasonic CT can solve these problems. As shown in Fig.2(c), the measuring section is divided into cells. When transmitting at a measuring point in one tube, signals at all points in the other tube are received by reception transducers. When using inversion algorithm introduced later, velocity distribution of measuring section can be obtained. Thus, location and dimension of anomalies can be identified. Property of defects and intensity of anomalies can be deduced.
Given that n rays pass through a measuring section which is divided into p ´ q= m cells, as shown in fig.3. Li is the length of ray i (distance between transmission and reception transducer), t i is the time that ray travels along Li. From Randon transform
where Vj(x,y) is the velocity of cell j, fj(x,y) is the slowness of cell j, i.e., the reciprocal of velocity. It is assumed that the cell is small enough, so fj(x,y) of each cell can be considered as constant. Equ.(1) can be written as progression form
where aij is the length of ray i in cell j. In view of mathematics, equ.(2) is a linear equation group
Or it can be expressed as the following system of equations
|t = Af||(4)|
The solution of equ.(4) is to find vector f through ray travel time t to make ||e||2 = ||t-Af||2 to be minimum.
|Fig 3: Cells division of measuring section.|
Owing to four reasons stated below, numerical approximation solution instead of arithmetic linear solution is adopted to solve equ.(4):
In engineering CT, iterative methods, such as Algebraic Reconstruction Technique (ART) Simultaneous Iterative Reconstruction Technique (SIRT) and Damped Little Square QR(DLSQR), are usually used. In addition, an iterative method called Back Projection Technique(BPT)is used frequently to generate initial value of slowness.
3.1 Cells division
Theoretically, the smaller the cell is, the higher the inversion precision will be, but the cost of numerical calculation will also increase rapidly. On the other hand, when considering the characteristic of ultrasonic, the size of a cell shouldn't be too large. 15~30cm can be chosen in engineering use, as anomalies usually have certain size.
3.2 Choice of algorithm
Analog computation indicates that ART has the advantage of calculating rapidly and occupying less memory storage, but it has worse convergence. SIRT has good astringency but calculates slowly. LSQR has much higher precision even if the data have errors. It is necessary to point out that the three iterative methods mentioned above are based on beeline model in which rays are assumed to travel in straight line, not in curve. It is recommended that curve model be used in huge volume concrete structure. ART together with BPT are chosen in UCT program to restructure the velocity images.
3.3 Modification of space between ultrasonic tubes
It can be easily seen that inversion of velocity depends on ray travel time ti and aij. Since ultrasonic tubes are welded or connected one by one in practice, it is difficult to keep one tube parallel to the other. Should we consider the space between tubes on the top as the true space, it would result in lower inversion precision or worse convergence. We had modified space between tubes by curve fitting modification method, but it had less effect. Recently, we have developed PCS program (Program for Computing True Space between Tubes) and solved the problem successfully. In the UCT program, we first use the PCS program to modify travel time so as to eliminate the influence on space between tubes, then conduct inversion and output CT images. Fig.4 is the N-S graph of the UCT program .
|Fig 4: N-S graph of UCT program.|
3.4 Application of series-wound transducers
Ultrasonic CT method needs parallel and cross measurements for many times. The fieldwork is heavy. For example, a segment of a pile(2.6m in length) introduced below was detected with ultrasonic CT, the detection step is 0.2m. 14´14 data needed to be obtained. The transducer in one tube was moved 14 times while the one in the other tube 196 times. If measuring section is enlarged or cell size is divided even smaller, the fieldwork will increase further. In addition, the height of the cell depends on the location of transducer. Since it is difficult to ensure the vertical location of the transducer when it dropped down, the inversion precision will be lowered greatly.
Because of these, we specifically developed series-wound transducers and conversion switcher, which are shown in fig.5. Series-wound transducers consist of ten separated transducers. With switcher being turned, parallel and cross detection data can be obtained easily. The workload of pulling transducers from and fro decreased to one-tenth of the usual workload. Meanwhile, the space of every 2 neighboring transducers is unchanged, which ensures that the detection step equal to the height of the cell.
|Fig 5: Application of series-wound transducer in ultrasonic CT.|
4.1 Test verification
Concrete test sample was constructed. The top of the sample was cuboid and the bottom was bored piles with tubes inside(see Fig.6). A segment of the concrete pipe was mounted around B tube to simulate defects, which was 58cm in length and filled with mixture of mud and cement mortar. The concrete of BC measuring section was detected (between 1-1/ and 12-12/) with ultrasonic CT.
|Fig 6: Graph of concrete test sample.|
Fig.7a is the layout of rays. Fig.7b is the isoline and chromatogram. Fig.7c is the three dimension distribution of velocity. Defects are shown in white in gray diagram (red in color diagram) in Fig.7b. In fig.7c, defects are shown as lower position in three dimension. The depth of the defect was located around 160~210cm and the width 30~64cm. The velocity of medium in this region is 3600~4000m/s,whereas the normal is 4200m/s on average. As the quality of concrete near defect was affected during construction, the dimension of defect shown in fig.7 is much larger than its actual size.
|Fig 7: Velocity CT images of BC measuring section.|
5.2 Application in practice
|Fig 8: CT images of S9 pile.|
Fig.8 is the isoline and chromatogram of S9 pile. From left to right, it shows AB,BC,CA measuring sections. White color shows anomalies, black color shows normal concrete (In color diagram, red represents abnormal and blue represents normal).As shown in Fig.8, concrete between -12.0~-13.0m is abnormal. The velocity varied from 3700m/s to 4200m/s while the velocity of normal concrete is about 4400m/s. By analyzing the data of all three measuring sections, it can be found that anomalies in BC distribute almost horizontally from -12.0m to -12.6m and in AB and CA distribute aslantly from -12.0m to -13.0m.
|© AINDT , created by NDT.net|||Home| |Top||