|NDT.net - September 2002, Vol. 7 No.09|
The analysis of the Acoustic Emission (AE) produced in Zry-4 CANDU fuel cladding is reported. Tubes with closed ends were submitted to an hydraulic experiment, during which the internal pressure was increased -at room temperature- until they break by explosion.Two sets of samples were tested: one without any manufacture defects and the other with external surface typical defects. Scanning Electron Microscopy (SEM) observation of defects is also presented. It was found that for defect-free samples, most part of AE events appeared when the pressure reaches a value 90% to 100% of the break pressure. For the defective tubes, AE events appeared at lower pressures. The Duration parameter of AE signals was found to be different from defect free samples in those with defects. The fracture surface due to explosion was always ductile, which is in accord to the low production of AE events.
The expression Acoustic Emission (AE) is used to describe the technique and the physical phenomenon in which this technique is based. AE is the result of elastic waves produced by the stresses induced inside the material.
Zirconium alloys have one of the smallest neutron captures by unit length compare with other structural metals commercially available , hence it is used in nuclear power reactors components. Zry-2 and Zry-4 are among the most common, having Sn as mayor alloying element and Fe and Cr as minor ones and 1400 ppm of O . The process to obtain the material is complex. The final step in manufacturing process of the cladding tube is a cold rolling one named: "pilgering process", which gives the material the appropriate finishing . The last anneal temperature is between 475 ÷C and 575 ÷C. The surface finishing of the tubes is done by sand blasting of the interior and polishing with emery paper in the outside.
FAE S.A manufactures the fuel claddings. The whole production is inspected to check wall thickness, diameters and defects. Ultrasound or visual inspections that not fulfill manufacturing specifications determine the rejection of the tube, detecting those external irregularities, which are called in this work "defects".
This paper is a summary of a Magister Thesis and ulterior works [3-4].
2.1. Pressuring equipment
|Fig 1: Scheme of00 hydraulic equipment.|
The dimensions of the tube were 150 mm in length, 13.08 mm of external diameter and 0.38 mm of wall thickness. An important point is the attachment of the tube to the hydraulic equipment to prevent fluid leaks. A screw and a bolt settlement closed one end, the other end was for oil input. A sensor measured the pressure inside the tube and sent it as an external parameter to the AE equipment. The pressure was increase 2204 psi/min till fracture .
2.2. Acoustic Emission equipment
The AE equipment used was an AEDOS by CISE. Its allows the quantitative analysis of the burst. It gives Amplitude, Duration, Risetime, Energy, Events vs. Time and Events vs. external parameter. One channel is used to "hear" the AE and other for the external parameter, in this cases the pressure. The AE channel has a WD-943 wide band sensor by PAC. It was positioned on the bolt that closed the tube, then a 40 dB preamplifier was connected to the AEDOS and then the signal was sent to the amplifier. A Le Croy 9360 oscilloscope for future analysis digitized the events. This was done through the two outputs in the AEDOS. Figure 2 shows the scheme of the system.
|Fig 2: System scheme.|
Concerning the events, the number of valid events has to be distinguishing from the total number of them. Valid events are those that do not saturate the Amplitude and show coherence, that is Risetime less than Duration.
The defects in the external surface are macroscopic and are usually detected by Ultrasound and visual inspection in plant, during the regular control of production.
SEM micrographs of five of the most common defects are shown, the magnification varies according to the characteristics of the defect:
|Fig 3: Type 1 defect, SEM 200 X.||Fig 4: Type 2 defect, SEM 50 X.|
|Fig 5: Type 3 defect, SEM 50 X.|
|Fig 6: Type 4 defect, SEM 200 X.||Fig 7: Type 5 defect, SEM 100 X.|
A total number of 36 explosion experiments at room temperature were performed with samples from tubes of Candu-type Zircaloy, with and without defects. The first 25 tests were used to calibrate the whole system. The actual analysis corresponds to the 11 last tests, six with defects and five without them. The results of the explosion tests are described in the following items.
In all cases the process that ends with the explosion of the tube was the same: at a certain pressure the tube started to bend, a blister was formed in the concave surface where fracture occurred. The fracture was "S" or "C" type, see Figure 8.a) and 8.b). Due to the circunferential stresses fracture started as Mode I. It propagated in axial direction a certain extension, then in both ends a crack appeared along the circunferential direction that is to say Mode II. Depending on the orientation of the crack the fracture ended as an " S" or "C" type.
|a) Type "S" fracture||
b) Type "C" fracture
Fig 8: Fracture types.
Fractures in the 36 essays were always ductile, see Figure 9, and occurred in the central third of the tube length. In samples with defects, fracture never started at any of them.
|Fig 9: SEM fracture surface, 1600 X.|
3.2. Fracture pressure
In spite of the defects all the samples fractured at similar pressures (P). Table I shows the extreme and average values for samples with and without defects.
|TYPE||WITHOUT DEFECTS||WITH DEFECTS|
|Max. Pressure (PSI)||8005||7878|
|Min. Pressure (PSI)||7735||6842|
|Average Pressure (PSI)||7844||7314|
|Standard Deviation (PSI)||126||403|
|Table 1: Pressures of two sets of samples, with and without defects.|
The standard deviation for the samples without defects is quite smaller than the corresponding to the samples with defects. The average pressure is 9 % smaller in the samples with defects. Other clear way to see these differences is to analyze the box and whiskers graph showed in figure 10. There, the Mean Values, Standard Error and Standard Deviation are represented for both sets and the previous result is reinforced.
|Fig 10: Fracture Pressure.|
3.3. Acoustic Emission
3.3.1. Total event number
It was thought in advance that the mean value of the number of events would distinguish the presence or absence of defects in the samples. The results for the Mean and Standard Deviation values of total number of events are indicated in figure 11.
|Fig 11: Total Event Number|
As it can be seen, the Mean Values and Standard Deviations of the total event number for both sets can not be distinguished between them.
3.3.2. Events production
In all cases the maximum amount of events appeared in the interval of 90 % to 100 % of the fracture pressure. In samples with defects the first events appeared earlier, except for those with type 2 defect. It is convenient to say here, that type 2 defect is reported as a change in the surface color instead of a damage in the surface as in the other defects. In figure 12 the Mean value of the Minimum Pressure at which AE started, the Standard Error and Standard Deviation were graphed for both sets. Then it was observed that in tube samples with defects AE events initiated at lower pressures.
|Fig 12: AE Start Pressure.|
3.3.3. AE parameters distributions functions
The Amplitude (A) distribution of AE events in each experiment was studied. It was found that the best fit is a Log-normal function (see the example at figure 13), this is in accord with other works reported [5-6]. For Duration (D) it was found that the best fit is with a Log-normal function except for one test (see the example at figure 14). Concerning Risetime (R), excepting for three tests, it was possible to find a unique fitting distribution, an Exponential one (see the example at figure15). The quality of the fits was checked with Kolmogorov-Smirnov and Chi-Square tests.
|Fig 13: Amplitude Distribution.|
|Fig 14: Duration Distribution.|
|Fig 15: Risetime Distribution.|
3.3.4. Mean AE parameters
Mean values were calculated for the Amplitude (Am) Duration (Dm) and Risetime (Rm) for each set. To distinguish between samples with and without defects these parameters were plotted in a box and whiskers graph showed in figure 16 for Amplitude, in figure 17 for Duration and in figure 18 for Risetime. Values for Am and Dm corresponding to samples without defects are in general higher. There are no important differences concerning Rm. So the Dm. AE parameter showed differences between defective and non-defective claddings during pressurization.
|Fig 16: Mean Amplitudes.|
|Fig 17: Mean Duration.|
|Fig 18: Mean Risetime.|
3.3.5. AE parameters in each test
In this analysis we studied A, D, and R of all AE events in each test, not their mean values. The aim is to find out differences or similarities between defective and no defective sets. Mean values of the two groups had to be compared, which were calculated from no Normal probability distributions, having positive asymmetries. So we used non parametric Kruskall-Wallis test for A, D and R median values. As the P-value obtained for A, D, and R was lower than 0.05, medians equalities were rejected. So we can conclude that using Amedian, Dmedian and Rmedian for each set we could distinguish defective and no defective groups.
Special pressure equipment was designed and constructed to get more than 8800 psi necessary to explode the fuel claddings, monitoring the tests with Acoustic Emission.
Due to the circunferential stresses the cladding fracture started as a Mode I, propagated in axial direction until a certain extension, then at both ends a crack appeared along the circunferential direction, Mode II fracture. Depending on the orientation of the crack the fracture ended as an "S" or "C" type.
In all cases a ductile fracture was expected because material is designed with this purpose. Failures are produced as a consequence of the microvoids nucleation at inclusions and/or second phase’s precipitates. The localized ductile fracture might also occur in a perforation without an important plastic deformation, but this fact was not observed.
In samples with defects, no obvious relation between fracture and defect was found, it is thought that defects should be deeper to show a difference.
All tests were monitored with AE. The AE bursts were also digitized trough a digital oscilloscope for future signals analysis. A total of 36 exploding tests were performed. The first 25 tests were used to calibrate the whole system. The actual analysis corresponds to the 11 last tests, including six surface defective fuel claddings. SEM of each defect-type was showed.
In all the experiments most of the AE events were produced between 90% and 100% of the fracture pressure. In samples with defects, events started at lower pressures than in no-defective samples. It was possible to differentiate by fracture pressure P between defective and non-defective claddings. Then the pressure at fracture could be used to identify the type of group.
AE signals parameters were studied from statistical point of view. It was found that the Amplitude and Duration parameters could be fitted with a Log-normal distribution function, and an Exponential distribution function for Risetime. The mean Duration (Dm) of each AE test during pressurization resulted different for defective and non-defective claddings.
To FAE SA for providing the tubes.
To Alfredo Hey (Atomic Energy National Commission), Carlos Zunino and Roberto Ributtini (FAE SA), for their support and collaborat0ions.
To Rosa Piotrkowski (Universityof of General San Martin) for the discussion of the manuscript.
|© NDT.net - email@example.com|||Top||