A NEW SOFTWARE TO INCREASE ULTRASOUND SIGNALS RESOLUTION FOR INTERNAL STRESSES MEASUREMENTS IN METALLIC MATERIALS

Transcription

1 2007 International Nuclear Atlantic Conference - INAC 2007 Santos, SP, Brazil, September 29 to October 5, 2007 ASSOCIAÇÃO BRASILEIRA DE ENERGIA NUCLEAR ABEN A NEW SOFTWARE TO INCREASE ULTRASOUND SIGNALS RESOLUTION FOR INTERNAL STRESSES MEASUREMENTS IN METALLIC MATERIALS Marcos S. Farias 1, Paulo V. R. Carvalho 1, Cinthia S. Malta 2, Mauricio A. C. Aghina 1 Jose C. S. Almeida 1, Claudio R. Santanna 1 1 Instituto de Engenharia Nuclear (IEN / CNEN - RJ) Via 5 s/n - Cidade Universitária - Ilha do Fundão Rio de Janeiro, RJ msantana@ien.gov.br paulov@ien.gov.br mag@ien.gov.br jcsa@ien.gov.br santanna@ien.gov.br 2 Universidade Federal do Rio e Janeiro (UFRJ) Cidade Universitária Ilha do Fundão Rio de Janeiro, RJ cinthiavr@yahoo.com.br ABSTRACT The phenomenon of ultrasonic velocity change has been successfully used for the verification of internal stresses in metallic materials at IEN during last years. It is possible to know the ultrasonic velocity change by the measurement of the time elapsed between echoes of ultrasonic signals. Since the ultrasonic time of flight changes on the nanosecond order, it is requested high resolution to the assessment of internal stresses. A method for the measurement of this time using ultrasonic signal processing techniques was developed and tested. The method uses cross correlation and multirate techniques to increase the measure resolution. This paper describes the software developed to improve the application of this method. The software, developed in C++ and Labview, processes the ultrasonic signal with high resolution and performs this task in few milliseconds, allowing the development of portable equipment to assess internal stresses in metallic materials in the field, with direct time of flight measurement. The software developed was implemented in two portable data acquisition systems in use at IEN laboratories and can be adapted for different types of data acquisition systems with few modifications. 1. INTRODUCTION The time of flight measurement of ultrasonic waves has being used in many areas. These measurements are applied in the survey of the elastic constants of metallic and non metallic materials, in the measurement of outflow of liquids, in the ceramic porosity evaluation and in others purposes. The assessment of stresses by ultrasound is based in the ultrasonic signal time of flight change according to the internal stress on materials [1]. Considering that the variations in the time of flight of ultrasonic signals in metallic materials, due to the

2 variations in the material stress, have a nanosecond order, it is requested high resolution to perform this measurement. The method developed for the time of flight measurement uses cross correlation and multirate techniques to increase the measure resolution [2]. In such a way, an acquisition system with a typical resolution of 10 nanosecond order (100 MHz) can have its resolution increased up to 16 times (L16) in the time of flight measurement. This allowed that an assessment of internal stresses in metallic materials can be carried out using an acquisition rate found in common laboratory oscilloscopes [3]. The good results reached in laboratory have excited the interest of different sectors of the national industry. These sectors intend to use this technique to prevent accidents on generating vapor ducts, boilers etc. As consequence of this interest, there is the need of direct field measurements. The ambient conditions, such as temperature, electric oscillation, vibration, transport, amongst others make difficult the use of normal laboratory equipment, as oscilloscopes, computers etc., to configure the system in the field experiments. To carry out essays in the field, we need a portable and robust system in order to carry out its task with the demanded performance requirements. This demanded performance includes, besides of pulses generation and data acquisition of the signal, the processing of this signal with the application of the developed method, that is, interpolation using the multirate technique and cross correlation of the signal. 2. METHOD The time interval elapsed among two echoes of ultrasonic signals is measured calculating the time elapsed among two reference points previously established in these signals. The point of maximum of these echoes can be this reference point. However, the ultrasonic signals are not identical and the selection of the reference point can compromise the precision of the measure process. The figure 1 shows the noises in the maximum of two consecutive echoes of an ultrasonic wave. To measure the delay between two ultrasonic signals, without the use of reference points, we used cross correlation [4]. In this method, the acquired signal, with two echoes, is divided (figure 2). With the application of the cross correlation algorithm, each point of a signal is multiplied and added to all the other points of the other signal, resulting an another function whose maximum is the value of the delay between the two signals [5].

3 Figure 1 Noise in the maximum of an ultrasonic wave Figure 2 (a) Ultrasonic signal with two echoes (b) The first echo (c) Cross correlation of the two signals. With the application of the cross correlation, the error introduced by the reference in the maximum is completely eliminated. Moreover, the ultrasonic wave can generate deformations on the echoes during the propagation. If the propagation time increases, the difference between the two waves also increases, and the deformation of the echo changes. This means that two consecutive echoes are distorted in different regions of the echo. This behavior can be another source of error in the measure of time when using reference points. Problems like these are eliminated with the use of the cross correlation. The resolution of the measure, even after the cross correlation, is determined by the sampling rate of the data acquisition system used. To increase this resolution, with the same data acquisition system, we used the expander sampling rate to subdivide the interval of

4 original sampling inserting null samples, that, after adjusted by filtering, receive coherent values. The multirate technique to increase the sampling rate F to F' value, by an integer factor L (that is, F' = FL), consists of inserting L - 1 null samples between each pair of samples of the original sequence, whose rate we want to modify [2]. The exit sequence of the expander is filtered (Figure 3), performing an interpolation. Expander of sampling rate Filter L h(m) F F = LF F = LF Figura 3 Interpolation process 3. METHOD APPLICATION The aim of the software is to process the data in a very short time interval, so that the result of the time of flight measurement of the ultrasonic wave is presented immediately after the acquisition. In the early experiments we used the text file generated by the oscilloscope, with the data of each acquired signal, and processed this signal with the application of the cross correlation and interpolation. This process was too slow, because there was a need to get data from the hard disk. In the case of the L16 interpolation, the data processing time took some hours. To diminish the execution time of the program, all the data acquisition received from the hardware, as well as the manipulated data during the processing and the table for the filter of the interpolation, are kept in memory. Some execution tests with the compiler C++ Builder [6] had been carried out, creating a DLL (Dynamic-link library) optimized for the L16 interpolation. In the case of the cross correlation, the tests had demonstrated that the execution of the algorithm was faster when used part of one VI (Virtual Instrument) for cross correlation of the Labview compiler [7], associated to a DLL created in C++ Builder.

5 With the optimization applied to the interpolation and cross correlation algorithms, it was possible to carry out these tasks between 40 and 50 miliseconds in a computer with a 2,2 Ghz processor. This time was not verified with specific metric for program execution, being only a proper compiler estimative. With this implementation, the development of portable systems to assess internal stresses in metallic materials in the field became possible. Few modifications in software are needed for the adaptation to the new data acquisition hardware, giving flexibility to the instrument. The only requirement is that the hardware used supply the data for the digital to analogical signal conversion. These data are arranged in an array, the input for the signal processing part of the software. The entire data presentation, the results presentation, and the acquired signals, as well as the storage of the results, were made in Labview. Figure 4 shows one of the software screens with the results of the measures.. Figure 4 Software screen 4. THE INSTRUMENTS DEVELOPED Two portable instruments were developed. The first one (Figure 5) consists of a portable equipment that generates pulses to ultrasonic transducers and makes the acquisition/processing of the echo signals coming from the material This equipment is controlled by software that makes the signal acquisition with 100 Mhz sampling rate, later interpolated for the L16 factor. With this arrangement we obtain a 0,625 picosecond signal resolution. The figure 6 shows the assembly of an experiment. The result of the data processing is presented in the graph. The relation between the applied stress with the time variation obtained in this experiment was compatible with the tests previously carried out using the laboratory equipment (oscilloscopes, pulse generators and computer). However this graph is available at the end of the experiment, without need of further processing.

6 Figure 5 Portable equipment developed 11,64 Tempo(microsegundos) 11,635 11,63 11,625 11,62 11,615 11, , , , , , , , , , , Stress Carga (X1000 Kg) Figure 6 Machine with transducer connected and the graphic time x stress The second system developed was based on a portable oscilloscope with 200 Mhz of maximum sampling rate and an USB interface. The software, in this case, was adapted to control the oscilloscope and to collect the data coming from this oscilloscope. The data obtained were processed using the same method. In order to carry out analyses in the field it is necessary, in this case, to use a portable pulse generator to excite the transducers. 5. CONCLUSIONS The software developed showed that the cross correlation and L16 interpolation method can be applied successfully in different data acquisition systems to increase the time resolution

7 of the acquired signal. This makes possible the assessment of the metallic material stress in a faster way in the field, opening perspective of exploitation of this technique in diverse areas of the industry. The development of two portable systems showed that software can be adapted, with small modifications, to other data acquisition instruments, keeping the main requirement - improve the signal resolution - a key factor to apply the time of flight measurement to the assessment of internal stresses in metallic materials. REFERENCES 1. Hsu, N. N. Acoustical Birefringence and the Use of Ultrasonic Waves for Experimental Stress Analysis. Experimental Mechanics, Vol. 14, No. 5 (1974). 2. Bittencourt, M. S. Q., Pinheiro, M. A. A., Lamy, C. A., Filho, J. C. P. Ultrasonic Time of Flight Measurement to Stress Evaluation. III Pan-American Conference for Nondestructive Testing (2003). 3. Bittencourt, M. S. Q. Desenvolvimento de um Sistema de Medida de Tempo Decorrido da Onda Ultra-sônica e Análise do Estado de Tensões em Materiais Metálicos pela Técnica da Birrefringência Acústica Tese de D.Sc. COPPE/UFRJ (2000). 4. Papoulis, A. Probability, Random Variables and Stochastic Processes. McGraw-Hill (1965). 5. Clarkson, P. M. Optimal and Adaptive Signal Processing. CRC Press (1993). 6. Hollingworth, J., Butterfield, D., Swart, B., Allson J. C++ Builder 5 Development s Guide. Sams (2001). 7. Ritter, D. J. LabView Essential Techniques. McGraw-Hill (2002).