·Home ·Table of Contents ·Nuclear Industry

Eddy Current Test of Laser welds on AISI 316 LN Cover Plates of ITER Toroidal Field Model Coil R.Vesprini, P. Varone, A. Tatì, M. Ghidi ENEA INN-TEC E.Visca ENEA ERG-FUS Contact

INTRODUCTION

The ITER (International Thermonuclear Experimental Reactor) fusion machine will use superconductor coils for the production of the toroidal magnetic field needed for the plasma containment system. Scaled prototypes of the ITER field coils were manufactured and they will be tested in an 'ad hoc' facility (TOSKA). The toroidal field case contains and supports the superconductor winding packs. After the superconductor cable is put in place the coil case is closed on the two faces by cover plates that are laser-beam welded (Fig. 1). ENEA INN-TEC-AND proposed eddy current testing (ET) as a non destructive testing (NDT) technique to detect the penetration profile of superconductor cover plate laser welds. Ultrasonic testing, for technical reasons, was forbidden. Preliminary laboratory tests performed at ENEA C.R. Casaccia on a sample supplied from the ITER home team showed interesting qualitative results. A first call for a go-no go test on a Dummy section of the torus, at RTM (Vico Canavese) factory confirmed the reliability of the tests carried out by both line and area scanning. Considering the spiral geometry of the welding path, in order to remove any problem of welds tracking, the automatic area scanning was indicated as being more suitable for future inspections. Due to the advanced status of the torus construction, ENEA was called to perform a complete qualitative inspection of the DP1 pancake. The tests were carried out by a computerised x-y scanning system, able to directly give a C-scan and perspective colour images as output, on both DP1 sides. This first intervention made it possible to define test procedures and qualitative criteria adopted for reparation. Subsequently four other pancakes were inspected at RTM. Several lines of welding were found to be exceeding in the qualitative acceptance criteria adopted and so they were consequently repaired. At the end of each pancake inspection a declaration was released when the welding had successfully passed the test indicating a penetration depth greater than 2 mm. A research program for quantitative analysis was launched so as to achieve an accurate measurement of the penetration depth. This paper is composed by two parts. The first part which concerns the acceptance test, contains some examples of the faults, found during the inspection service, in terms of qualitative map which shows the effectiveness of reparation carried out immediately after their detection. The second part contains the results of the quantitative analysis devoted to directly supply in mm the actual penetration depth of the welds.

Fig 1:TOSKA facility. Right: (up) TF model coil pancake, (down) superconductors insertion under cover plates to be welded.

ACCEPTANCE TEST

All the pancakes, for a total length of about 1500 m of welding lines, were inspected by the following equipment configuration and the same test procedures:

Equipment used:
ET instrument Rhomann ELOTEST B2 with an absolute coil probe of 10 mm active diameter; acquisition data I.R.T. "SCAN MASTER" with X/Y encoderised Manual Scanner, able to supply 2D and 3D images of the inspected area.

Procedures
The test system was previously calibrated by using a reference block of AISI 304, which contains machined faults (notches) of different depth, in order to supply readable reference signals from 3 mm up to 1 mm from the block upper surface. The probe zero point was reset every 4 minutes during the scanning operation.

System setting:
Test frequency: 16 KHz; probe lift-off: 1,4 mm; lift-off signal: horizontal; acquisition steps of eddy current (e.c.) signals:1 mm on X axis, 2 mm on Y axis.
Side A and B surfaces were respectively divided into 23 small sectors in order to cover all the surface moving, each time, the scanner on a 56 x 40 cm area. The colour range, for qualitative evaluation of the scanned area images was chosen for positive signal voltage values ("green" about 5 mm,"violet" less than 2 mm).

Acceptability
Because of the large active area of the probe and the proximity of two adjacent welding lines (4 mm) the testing signal is related to the integral contribution of both lines. This means that the actual depth of the single welding line is always slightly under-estimated, thus maintaining the acceptance threshold in conservative conditions. In agreement with ITER team, indications of violet colour, extended for more than 50 mm in length, were considered as defects that must be repaired. Defected lines were marked, on the piece surface, by waterproof ink in order to indicate location and extension of the repair. After rewelding the defected lines were tested again to confirm their acceptability.

Test results
Fig. 2 shows a general view of eddy current inspection with the test apparatus and the NDT operator working on the pancake surface.The inspection was performed, on each side, at the end of the welding process. Faults found were repaired and tested again before turning the pancake on the opposite side.

Fig 2: Eddy current inspection of ITER pancakes at RTM factory

Fig 3: Reconstructed map of a large zone of DP1 pancake side B. Violet colour indicates insufficient penetration depth.

Fig 4: 3D colour map of DP3 side-A, sector-L, showing faults before (a) and after reparation (b).

Fig. 3. is a 2D reconstructed map of a large portion of DP1 pancake side B, sectors-L-M-N-O-P-Q showing a very long fault to be repaired in the last but one external spiral ring.

As an example of repair intervention positively concluded, the testing images concerning the DP3 pancake side-A sector-L are reported in Fig. 4. The 3D-colour map a) shows some faults (zone encircled in white contour) detected before repair. The corresponding image of the partially scanned area after repair is shown in the upper frame b). The reduced height of the signals indicates the acceptable penetration reached.

QUANTITATIVE ANALYSIS

As in all NDT techniques, the purpose of eddy current testing is to locate the position of a crack and to characterise its dimensions, particularly its depth. Inversion methods based on analytical approach or modelling probe and crack interaction, are often scarcely effective when applied to actual testing conditions. For this reason we decided to meet the inversion problem by the experimental approach based on the testing system calibration obtained by analysing the actual probe response on an 'ad hoc' sample containing notches of different depth.

Calibration curves
The quantity of interest in most forms of eddy current inspection is DZ, the change in probe impedance that is induced by a flaw. To provide the quantitative values of depth directly expressed in mm, a new calibration block of AISI 316 (5 mm thick) containing narrow notches simulating small (1 mm), medium (2 ; 3 mm) and large (4 mm) penetration lack of the welds, was used.
Using a probe that has an active diameter of 10 mm, variation of notches width up to 0.6 mm has little influence on eddy current signal. The same probe and parameters setting used in the pancakes inspection were also used for calibration block inspection.
To study the trend of Z amplitude and phase related to the various depth of the faults, the calibration block was inspected on both faces. The probe was initially scanned by passing on the central line of side A (breaking surface faults) and then on the opposite face (side-B). Fig 5 shows the movements of Z spot on the instrument screen. Red curves are related to the notches depth detected from side A and the pink curves from side B; the black curve is common to both the faces as through-wall defect of 5 mm height. Fig. 6 shows the corresponding phase variation of the signals.

Fig 5: Movements of e.c. spot on the instrument screen: red colour side A, pink side B.

Fig 6: Phase (q) variation versus notches depth.

To analyse the behaviour of the signal related to the notches presence, the maximum values of Zx and Zy components, for side A and side B, were then plotted on the impedance plane. In both cases Zy appears as a regular monotone function that can be easily expressed by a mathematical equation that becomes very simple when its domain of existence is not considered around the extremities (no faults and total break conditions). With this limitation the Zy component of impedance can be profitably used to evaluate eddy current response. Fig. 7.a and Fig. 7.b respectively show the interpolation curves of Zy for side A and side B. These are in good agreement with experimental results for depth values from 0.7 to 4.3 mm inclusive. The black straight lines shown on the graphics represent the actual values of depth to be measured. To give a depth value directly in mm, the Zy curves must be converted in to those straight lines.

Fig 7a: Zy interpolation curve for side A scanning.

Fig 7b: Zy interpolation curve for side B scanning.

To recognise if the signal is related to breaking faults on the scanning surface or to breaking faults on the opposite side, the phase (q) of the signal can be used. As you can see in Fig. 6 the phase increases continuously starting from small notches open on the surface to small notches open on the back surface. The separation value of 30° can be used to discriminate the kind of fault (incision or insufficient penetration of the weld).

Correction programme
A software programme was developed to automatically indicate the depth of faults detected during the inspection. For each couple of Zx and Zy data input coming from the inspection system, phase q is calculated. For each Zy value read, the x value must be calculated by one of the following equations (calibration curves):

1)              y₁ = - 0,008 + 0,115x + 0,135x² - 0,008x³
2)              y₂ =   3,500 - 2,116x + 0,450x² - 0,033x³
using the choosing criteria:
• if q< 30° fault open on the surface                                   apply 1)
• if q> 30° fault open on the opposite side                          apply 2)
• change Zy into x value found (output corresponding to the depth in mm).

Graphic results


Fig 8a: Calibration block scanning: Side A qualitative 3D colour image.	Fig 8b: Side B qualitative 3D colour image
Fig 9:Calibration block. Side A: a)qualitative voltage amplitude representation, b)quantitative indication in mm.

Figs. 8.a),b) show the 3D qualitative calibration block colour images related to sides A and B inspection, as supplied by Scan-Master s/w. The side A image was reversed by the operator to intuitively indicate notches open on the scanning surface (side A and side B eddy current signals are both positive).

Figs. 9.a),b) show the comparison between the qualitative 3D calibration block colour image of side A and the quantitative 3D corrected image as supplied by dedicate s/w developed in LabVIEW graphical environment. The image in Fig. 9.b) was automatically reversed by phase analysis and actual values of depth (1,2,3,4) can be directly read in mm on the vertical axis at the lower tips of the colour wave.

Application to actual tests


Fig 10a:	Fig 10b:
Fig 10 : 2D qualitative colour map including transversal and longitudinal profiles (Volts scale) of pancake P1: sector ME, side A: a) Before repair; b) after repair.

As an example, the comparison between qualitative results obtained on a pancake test and their subsequent quantitative measurement is shown here. Figs. 10 a),b) show the 2D qualitative colour map related to the inspection of the pancake P1, sector ME, side A. In Fig. 10. a) the central welding line is clearly identifiable as defected (violet colour). The threshold overcoming is also readable in volts on the orthogonal projective profiles. In the marked position the signal was higher than 4 volts. After being repaired (Fig. 10. b)), the map of the re-welded line clearly shows the reduction of the fault under the acceptance threshold (voltage signal about 2 volts).

Fig 11:Correct image of depth measure: a) welding point of good penetration, b) poor penetration.


The quantitative evaluation of the depth is promptly achieved by using the correction programme, Figs. 11.a),b) show the same welding line of Fig. 10.a) (before repair).
The computer image allows the measurement reading in three different ways: 1) the new colour palette is now related to the depth in mm (the red colour is less than 2 mm); 2) the transversal profile also allows an easy evaluation of the depth in mm; 3) pointing the cursor, the depth of any point appears digitally indicated in mm at the upper right side of the image frame. In picture b), the cursor is pointed on a poor penetration zone that has to be repaired. The value read (1.0 mm) gives the accurate measure of local depth. In picture a) of the same figure, the cursor is pointed on a good penetration zone (read value 2.8 mm). Errors in measurement are mainly related to unwanted inspection condition variations such as lift-off change and thermal drift of the probe/instrumentation system. For this reason probe lift-off and the zero point of the signal must be frequently checked as specified in setting procedures. Many tests performed on reference samples indicate a measure error that does not exceed +/- 0.2 mm.

CONCLUSIONS

This work constitutes a NDT application of eddy current testing techniques, which can be considered alternative to ultrasonics for laser welding penetration measurement on large components in AISI 316 LN. Results obtained on six pancakes demonstrate the validity of this technique to detect the presence of dangerous faults such as penetration depth less than 2 mm and to certificate their disappearance after repairing. The research activity carried out allowed to define the final test procedures and data post-elaboration in order to obtain directly depth measurement in mm at each point of the welded seams. The quantitative analysis of the testing signals gave a positive answer in terms of reliability and accuracy that are usually requested in industrial applications. The results achieved on AISI 316 LN can be easily extended to all non-ferritic metals and to different types of butt-welding processes for plates of limited thickness. A more complete characterisation of the faults and the ET extension to austenitic steels containing d-ferrite in the welded lines, need further investigations.

REFERENCES

1. R. Vesprini, A. Tatì, P. Varone, M. Ghidi
   Eddy Current testing on cover plates laser welding of ITER toroidal field model coil. Final report.
   ENEA R.T. n. TLF027, Dec. 99.
2. B. Wincheski, J.P. Fulton, S. Nath, M. Namkung, J.W. Simpson
   Self-nulling eddy current probe for surface and subsurface flaw detection.
   Material evaluation/January 1994.
3. Arnold H. Kahn
   Impedance changes produced by a crack in a plane surface.
   Reviw of Progress in Quantitative NDE Vol. 1, D.O. Thompson and D.E.Chimenti, (Plenum NewYork, 1982), pag. 369.
4. J. R. Bowler
   Eddy-current interaction with an ideal crack. I. The forward problem.
   J. Appl. Phys. 75. (12), 15 June 1994
5. J. R. Bowler, S. J. Norton, D. J. Harrison
   Eddy-current interaction with an ideal crack. II. The inverse problem.
   J. Appl. Phys. 75. (12), 15 June 1994

© AIPnD , created by NDT.net

|Home|    |Top|