Proc. of National Welding Seminar (NWS-2010), Dec. 2010, Visakhapatnam, India
A NEW METHODOLOGY FOR QUALIFICATION OF WELDING PROCEDURE FOR CIRCUMFERENTIAL SHELL WELDS OF STEAM GENERATORS OF PFBR B.P.C. Raoa, C. Babu Raoa, S. Thirunavukkarasua, T. Jayakumara, Baldev Raja, Aravinda Paib, T.K. Mitrab and Pandurang Jadhavc *a
Metallurgy and Materials Group, Indira Gandhi Centre for Atomic Research Kalpakkam, TN- 603 102, E-mail: Email address:
[email protected] *b BHAVINI, Kalpakkam, TN- 603 102 Email address:
[email protected] *c Heavy Engineering Division, L&T, Powai, Mumbai- 400 072 Email address:
[email protected]
Abstract The inner shell of steam generators of prototype fast breeder reactor (PFBR) made in two halves, is TIG welded axially throughout the length and circumferentially at four locations. During the circumferential welding, defects such as arc strike on the tube, spatter of weld metal and fusion of weld root with tubes might form due to the narrow gap between the shell and the peripheral tubes. Presence of such defects, especially, fusion is harmful. This gap between the shell and peripheral tubes is inaccessible for any visual examination. This paper presents a nondestructive testing based methodology for qualification of welding through of arc strike, spatter and fusion type of welding defects formed during circumferential shell welding. The methodology uses eddy current testing from the tube side and ultrasonic testing from the outer surface of the inner shell. Weld parameters have been varied and formation as well as detection of low, medium and high degrees of arc strike, spatter and fusion type weld defects has been investigated. The methodology has been validated and successfully implemented at manufacturing site for structural integrity assessment of four steam generators and for qualification of the welding procedure. Keywords: Nondestructive testing, steam generator, weld qualification, ultrasonic, eddy current.
Introduction Steam generators (SG) are one of the most critical components of sodium cooled fast breeder reactors of Indian nuclear power program. The steam generator of 500 MWe Prototype Fast Breeder Reactor (PFBR) which is under advanced stage of construction at Kalpakkam, is as shown in Figure 1 a vertical (height 25 m) counter flow type heat exchanger with secondary sodium (5500C) in the shell side and water on the tube side. The structural material of the tubes is modified 9Cr-1Mo steel (Grade 91). Any leakage in the steam generator tube (outer diameter 17.2 mm and wall thickness 2.3 mm) would be catastrophic as the sodium–water reaction is violent and exothermic. Hence, utmost care is taken during the fabrication stages to impart stringent quality assurance of the steam generator tubes using various nondestructive evaluation (NDE) techniques [1]. There are a total of 8 SGs in PFBR. In each SG, there are 557 tubes with a thermal expansion bend of 375 mm radius (developed length 1075 mm). The tube bundles are inside inner and outer shells with a diameter of 855 mm and 1237 mm respectively. The gap between the inner shell and the peripheral tubes is approximately 3 to 5 mm. The inner shell sectors are welded longitudinally and circumferentially. The circumferential welding, being the final welding, is important due to the narrow gap between the peripheral tubes and the shell. During this welding, arc strike on the tube wall, spatter of weld metal onto the SG tubes and fusion of weld root with tubes might form as a result of the narrow gap. Presence of such a defect, especially fusion, may create localized hot or cold spot
during the operation of the SG. This may lead to stress corrosion cracking and hence, disturb the structural integrity of the SG. Preventive techniques e.g. use of copper backing fixture to avoid arc strike, restricting the welding current, indoctrinating the welding personnel etc. are adopted to minimize formation of welding related defects. However, possibility of formation of the above explained defects should be ruled out. Hence, it is important to qualify the welding procedure such that arc strike, spatter and fusion type of welding defects, especially fusion type defect, do not form during circumferential welding. There is neither direct line of sight nor access to the welded regions to carry out visual examination. In this paper, a new methodology has been developed for qualification of welding procedure through nondestructive detection of arc strike, spatter and fusion type of welding defects and also for ensuring the structural integrity of steam generators of PFBR. The details of the circumferential welding, proposed methodology and its application to four SGs are discussed n various sections of the paper.
Fig.1 Construction details and manufacturing of steam generator of PFBR.
Circumferential welding of SG shells The inner shell of the SG is made in two halves (modified 9Cr-1Mo steel) and welded axially throughout the length and along the circumference at four locations. These welds are made by TIG welding process. The groove used for the circumferential weld is of ‘V’ type with 700 angle having a root gap width of 2 to 3.5 mm. The filler material used is also modified 9Cr-1Mo steel. A preheat temperature of 200-2500C is maintained before welding to ensure that the component is free from moisture. An interpass temperature of 200-2500C is also maintained during the welding. A post heat of 2000C is maintained for 2 hours after completion of the welding to reduce the cooling rate and avoid weld cracking. The circumferential welding is manually done except at a few locations which are qualified for automated welding as per the specification document evolved based on ASME Section IX [2]. Table 1 gives the welding parameters used for circumferential welding of shells. Table 1: Circumferential shell welding parameters. Welding parameter Preheat temperature Welding current Voltage Torch speed Interpass temperature Post heat
Data 200-2500C 80-200 A 10-18 V 75 mm/min (minimum) 200-2500C 2000C for 2 hours before cooling from preheat temperature
Experimental In order to check the formation of welding defects and develop procedure to detect them nondestructively, simulating the actual shell welding conditions, welding has been carried out on specimens. The specimens consist of a 2000 mm long SG tube (outer diameter 17.2 mm) and a shell (thickness 12 mm and arc length 30 mm which is equivalent to 70 sector of the SG shell). The shell is made of 4 numbers of 500 mm long sectors joined by circumferential welding at three locations. SG shell sector is made by joining two numbers of 1000 mm long sectors for creating fusion. The tube is placed exactly in the middle of the shell sector and welded at the extreme ends maintaining a shell-totube distance of 3 mm throughout the length. In the actual steam generator, several support structures made of Inconel 718 are placed at approximate intervals of 1000 mm. In order to simulate this condition, a piece of support structure material is inserted between the tube and the shell sector. In order to fabricate the specimens with varying degrees of arc strikes, spatters and fusion, the welding current and voltage have been varied systematically as detailed in Table 1. In one specimen, no weld defect is introduced and this specimen is used as reference. While welding current is varied for introducing arc strikes, both welding current and voltage are varied to get different sizes of globules. Figure 2 shows the photograph of all the specimens viz. without any defects (Specimen No: 1), with arc strike (Specimen No: 2), spatter (Specimen No: 3), low fusion (LF, Specimen No: 4), medium fusion (MF, Specimen No: 5) and high fusion (HF, Specimen No: 6). Figure 3 illustrates the close up photographs of the weld regions showing e.g. low arc strike (LA), medium spatter (MS), and high fusion. As can be seen from Figure 3, physical contact is found to exist between the tube and the shell wall for fusion type defect specimens. The diameter of the spatter globules are measured and found to be in the range of 0-0.5 mm, 0.5-1 mm and 1-2.5 mm as given in Table 2. LA
LS MS
MA
SS HA HS Circumferential shell weld
500 mm
Spatter
Arc strike
Defect-free
Circumferential shell weld
HF
SS MF
1000 mm
LF
1000 mm
Fig. 2 Specimens without defects and with low, medium and high arc strike, spatter and fusion type welding defects in shells of SG.
LA
MS
HF
Fig. 3 Close up photographs of low arc strike (LA), medium spatter (MS) and high fusion (HF) in specimen. Table 2: Calibration specimen details and the variability in the welding parameters. Specimen No.
Type of defect
Variability in the welding parameters/situation
1
Without defects
Nil
2
Low arc strike
Welding current = 60 A
Medium arc strike
Welding current =100 A
High arc strike
Welding current =120 A
Low spatter
Spatter globule size = 0-0.5 mm
Medium spatter
Spatter globule size = 0.5-1.0 mm
High spatter
Spatter globule size =1.0-2.0 mm
4
Low fusion
0.5-1 mm root gap length
5
Medium fusion
1-2.5 mm root gap length
6
High fusion
2.5-4.0 mm root gap length
3
NDE Techniques and Methodology Nondestructive evaluation techniques for detection of welding defects being studied in this paper should be chosen such that they are sensitive, reliable, implementable in the shop floor and fast [3]. In the present case, there is neither direct line of sight nor access to the weld regions to carry out visual examination. Use of radiography is not possible due to inaccessibility for film positioning. Ultrasonic techniques are widely used for non-destructive examination of welds [3]. Ultrasonic examination from tube side is difficult. It is possible to detect fusion type defect using ultrasonic technique due to the physical contact between tube and shell wall by the molten weld metal. Ultrasonic testing can be carried out from the outer surface of the inner shell side using normal beam probe for detection of defects. The other possible technique for detection of defects is remote field eddy current (RFEC) technique which is also specified for in-service inspection of the steam generators from inside the tubes [4].
Preliminary analysis and NDE studies on simulated specimens reveal that the RFEC technique is capable of detecting arc strike, spatter and fusion types of welding defects while ultrasonic technique is able to reliably detect and size fusion type welding defects. Based on this observation, a new methodology is proposed for automated detection of all types of defects in the peripheral tubes and to ascertain fusion type of defects from the shell side. Basic principles of the RFEC and ultrasonic techniques and details about the methodology are discussed in this section. Ultrasonic technique Ultrasonic technique uses high frequency (1 MHz-10 MHz) sound energy to detect defects and to determine thickness of materials. This technique is widely used for examination of welds [3]. In this technique, ultrasonic wave propagating through a medium gets reflected when it encounters another medium with different elastic property. The transit time of the ultrasonic wave propagation (A-scan ultrasonic signal) in a material provides information on the presence of defects and other discontinuities in the material. In the present study, the fusion between the shell and tube wall is to be detected using the ultrasonic technique. When there is no fusion, the entire ultrasonic beam is reflected at the shell wall resulting in a single back wall echo. When there is fusion, the ultrasonic beam propagates to the peripheral tube in contact with the inner shell and a part of the beam is reflected at the shell wall and another part is reflected at the tube wall as shown in Fig. 4. The A-Scan ultrasonic signals, thus, obtained is expected to show two echoes.
SG shell Inner Shell wall
UT beam reflecting surfaces
SG tube
Fig.4 Ultrasonic beam profile in a shell weld fused with SG tube and its corresponding BScan ultrasonic image when scanned across the weld. Remote field eddy current technique The RFEC technique is mainly applicable to examination of metallic tubes with probing from tube inside. This is a low-frequency technique (50 Hz-200 kHz) and uses separate exciter and receiver coils that are kept inside the tube, as schematically shown in Fig. 5. The separation distance is usually 2–3 times the inner diameter of the tube. Sinusoidal excitation usually in the range of 50 Hz to 5 kHz, is applied to the exciter coil to generate time-varying magnetic fields [5]. The eddy currents induced in the tube wall, as a result, generate secondary magnetic fields. The primary and secondary fields travel axially and radially. In such a situation, there exist two regions of importance: 1) region inside the tube wall where the field due to the exciter coil decays exponentially along the axial direction, called the direct field region and 2) region inside the tube wall where the concentrated flux in the tube wall enters back into the tube at a distance where the direct field component is less, called the remote field region.
RFEC Probe Receiver
Exciter
Mod. 9Cr - 1Mo SG tube Ø12.6mm
8mm dia. flexible nylon hose
RFEC Instrument Connecting wires 30mm
2.3mm
Fig. 5 Basic principle of RFEC technique, expansion bend region of steam generator tube and schematic of RFEC probe used for testing the tubes. The magnetic flux in the remote field region is attenuated and lagged in phase. When a receiver coil is placed in the remote field region, the back entered magnetic flux induces a voltage which carries information on the condition of the tube wall. The in-phase and quadrature components of the voltage of the receiver coil, called RFEC signal, are measured and correlated with the presence of welding defects such as arc strike, spatter and fusion. A typical RFEC signal for a wall loss defect exhibits a characteristic double peak response; one when the exciter coil moves over the defect and the other when the receiver coil moves over the defect. A characteristic two peak signal is formed for a defect, one when the exciter coil passes through the defect and the other when receiver coil passes. RFEC testing of the peripheral tubes has been carried out using the procedure developed for in-service inspection of SG tubes [4]. The optimized inter-coil spacing of the RFEC probe is 30 mm and the excitation frequency is 1100 Hz. In order to negotiate the thermal expansion bend regions of the SG, a stiff and yet flexible nylon probe holder has been designed. The RFEC probe is attached to the nylon hose with 4 connecting wires drawn through the hose to the RFEC instrument as shown in Fig. 5. In the RFEC instrument, the signals are rotated such that the signals due to support structure are seen predominantly in the vertical component of the complex voltage plane. The horizontal and vertical component signals are recorded during probe scanning inside the tube. A peak finding algorithm based on a sliding window which identifies local minima or maxima in the presence of slowly varying baseline by comparing the data points within the window, has been used for automated detection of signals of defects after suitable thresholding. These data has been subjected to classification and analysis. Proposed methodology Figure 6 illustrates the flow chart of the new methodology proposed for automated detection of three types of weld defects. The methodology consists of the following steps: (1) RFEC testing of peripheral tubes and storing of the horizontal and vertical RFEC signals. (2) Automated detection of RFEC signals of defects based on peak finding algorithm and thresholding. (3) Classification of arc strike & spatter, fusion type of weld defects and signals due to support structures using K-means clustering. (4) Identification of cluster corresponding to probable fusion type indications. (5) Ultrasonic testing of these regions from shell outer surface for confirmation of fusion.
RFEC testing of tubes Thresholding and Automated peak finding of indications
Cluster III
Cluster I
K-means clustering
Support structure
Arc strike/Spatter
Cluster II (Fusion like indication) Ultrasonic testing of identified region
Two back wall echoes
No
No Fusion
Yes Fusion
Fig. 6 Flow chart of the new methodology for detection of welding defects.
The K-means clustering is an unsupervised clustering scheme which partition ‘N’ observations into ‘K’ clusters based on the Euclidean distance [6]. Given a set of observations (x1, x2… xn) where each observation is a d-dimensional real vector, the K-means clustering algorithm aims to partition this set into K-partitions (K
