Joanna Nicholas, TWI Ltd
Paper presented at Evaluation, Rehabilitation and Repair of Pipelines Conference, 20-21 October 2010, Berlin, Germany
In-service welding is often applied for repair and modification, whether for pipelines or pipework/equipment. TWI has assisted many companies in determining the welding parameters that can be used safely, avoiding burn-through of the carrier pipe and ensuring that fabrication hydrogen cracking does not occur. TWI has a long association with in-service welding procedure development, and also the application of inspection techniques for the geometries of concern.
This paper presents a case study in which TWI's expertise in developing safe in-service welding procedures and phased array inspection methodology for welds were applied to an onshore pipeline in preparation for modification. TWI 's work was based upon knowledge available in advance of the hot tap and the developed welding procedure was qualified at TWI using a 24 inch flow loop which allowed control and monitoring of the flowing liquid. The inspection methodology was developed on the qualification weld, although this was subsequently adapted by the welding contractor due to constraints not known at the time of development.
Joanna Nicholas obtained her degree in Materials Science and Engineering from the University of Wales, University College Swansea in 1999. She joined TWI in the same year, and has been involved in research and failure investigations, mainly with ferritic steels. She has also been involved in various aspects of computer modelling, particularly for research projects.
Joanna has broad knowledge of ferritic steels with particular emphasis on mooring chain, in-service welding and 9% Ni steels. She has carried out numerous investigations in these areas including failure investigation as well as production qualification and research. She is a Member of the Welding Institute and a Chartered Engineer.
In-service welding is routinely carried out for pipelines carrying a variety of fluids. The principal concerns are that of 'burnthrough' and fabrication hydrogen cracking. Burnthrough arises from generating a large enough weld pool such that the underlying ligament cannot contain the pressure of the flowing fluid in the pipe and a leak ensues. Fabrication hydrogen cracking arises from the interaction between stress and hydrogen in the presence of a susceptible microstructure. Welding is a source of hydrogen, the high cooling rates associated with the effects of flowing contents can develop crack-susceptible microstructures and welding residual stresses are present due to the shrinkage of the weld.
Development of appropriate welding and inspection procedures for in-service welding in order to avoid burnthrough and fabrication hydrogen cracking is perhaps the most counter-intuitive part of in-service modification. To avoid burnthrough, the heat going into the carrier pipe must be restricted, but to avoid fabrication hydrogen cracking, the highest possible heating (which retards the cooling rate) is preferred.
The inspection of in-service welds is difficult even under ideal conditions, and work was carried out in the 1990s to determine the best method of volumetric, sub-surface inspection for in-service welds. This paper identified that for such welds, ultrasonic inspection using a range of probe angles from both the pipe and attachment surfaces was the best method of identifying discontinuities. This work was carried out using ferritic weld metal, and so there were no interference issues with dissimilar materials to consider. Since the 1990s there have been developments in ultrasonic inspection and, in particular, the use of phased array probes allows a number of angles to be scanned in a single sweep of the probe. In the earlier program inspection for indications within the weld metal was not considered since the principal concerns were related to pipe-side flaws, or cracking in the pipe or sleeve.
A TWI Member company approached TWI to assist with development of an in-service welding procedure (with a view to hot tapping) for an onshore pipeline carrying crude oil. The line had been successfully hot tapped elsewhere along its length some 15 years previously.
The pipeline in question has been in service since the mid-1970s, and was procured to have a maximum International Institute of Welding (IIW) carbon equivalent of 0.42. At the location of the proposed hot tap, the pipeline is 36in (914mm) outside diameter, with a wall thickness of 9.5mm and the temperature of the line was anticipated to be approximately 10°C.
The previous hot tap information was provided, and used the application of preheat to retard the cooling rate. However the flow rate of oil through the line at the time of the previous hot taps is thought to have been of the order 440 (thousand barrels per day) mbd with the result that preheat maintenance times were very short and welding was particularly difficult. The flow rate at the time of the proposed in-service welding was anticipated to be in the region of 660mbd, which would result in a higher cooling rate. Under such conditions hydrogen cracking in the hardened HAZ microstructures generated in the pipe wall is the principal concern.
The proposed in-service modification involved adding a split-tee connection to the pipeline, (Fig.1). The longitudinal welds were to be made using nominally matching strength ferritic steel filler, and a backing bar. This joint configuration is standard practice for encirclement sleeve longitudinal welds to prevent penetration of the weld into the carrier pipe. As this weld was separated from the carrier pipe, it was not considered as part of the development work, and similar welding and inspection procedures to that used in previous modifications was recommended.
Figure 1 A split-tee (encirclement tee) in-service modification fitting, showing the positions of the welds.
Burnthrough is not generally a problem unless very high heat inputs are used, or the wall thickness of the carrier pipe is less than 5mm. In this case, the minimum wall thickness is expected to be 9.5mm. The critical internal wall temperature for burnthrough is 982°C (1800°F).
In order to establish confidence in this low risk of burnthrough, calculations using the Pipeline Research Council - International (PRCI) Hot Tap model were carried out using a range of parameters to determine the limits to avoid burnthrough. A low heat input (<1kJ/mm) was selected to be confident of avoiding burnthrough.
Additionally, in order to qualify a procedure for in-service welding for all positions, API 1104 indicates that a test rig in which the longitudinal axis of the pipe is held at 45° to horizontal should be constructed. This test rig qualifies welding in all positions and when filled with flowing water should generate more severe thermal cooling conditions than those typically encountered during in-service welding. This gives a lower risk of burnthrough, but higher risk of fabrication hydrogen cracking than would be expected in the field. TWI designed a rig based on the API 1104 recommendations. The plans for the test rig and photographs of the completed rig are shown in Fig.2. In order to comply with the requirements of API 1104, the pipe diameter and wall thickness had to be within a specified range. Thus a minimum pipe outside diameter of 24in (610mm) and maximum wall thickness of 12.7mm was required. To cover the possibility that a material with a maximum carbon equivalent of 0.42 may be present in the pipeline, a 0.42 carbon equivalent material was also required. To simulate the split-tee, half-shells were procured at 25mm thick and with a carbon equivalent of 0.43%. The chemical analyses are reported in Table 1.
a) Plan overview, showing test piece and 45 degree support
b) General view of test rig, with inlet and outlet water hoses visible
Figure 2 Overall plan and assembled test rig.
Table 1. Chemical analysis results
To establish the flow rate required in the test rig to represent field conditions, further calculations were performed using the Pipeline Research Council - International (PRCI) Hot Tap model, comparing the known parameters for the pipeline (crude oil) to the likely temperature and flow rates achievable in the test rig. The calculations indicated that a flow of water at 8°C and 0.02m/s in a 24in, 12.7mm thick pipe was of equivalent severity to the anticipated oil flow conditions. To consider the conservatism involved in the calculations, thermocouples were used to monitor the inside wall temperature on the test rig during all welding operations. The results from the comparison are given in Table 2.
Table 2. Results of calculations. Comparisons of flow
|Pipe OD||Wall thickness||Contents||Flow rate||250-100°C|
Fabrication hydrogen cracking
The main methods available to reduce the risks of hydrogen cracking are to modify the microstructure developed in the HAZ by control of the cooling rate between 800°C and 500°C (t8/5), and to reduce the diffusible hydrogen level in the weld metal. The control of t8/5 is largely dominated by the heat input used, along with some contribution from preheat. The microstructure is also affected by the welding sequence used, and a temper bead sequence would be expected to reduce the hardness of the preceding pass HAZ. The hydrogen level is reduced in two ways, firstly by reducing the level of hydrogen in the consumable used, and secondly by the use of preheat to encourage hydrogen escape from the weld. However, the use of preheat on the operational pipeline is of limited benefit, as earlier experience of in-service welding has indicated that there were difficulties maintaining preheat at an appropriate level for long enough to allow welding to be carried out. Nevertheless, preheat can be applied to the split tee material, to encourage hydrogen escape and slow the cooling for the HAZ in the split tee.
The use of austenitic electrodes to reduce the risk of hydrogen cracking is referenced in standards, as well as being referenced by a number of investigators over the last thirty years[5-7]. The reason for this is principally the greater solubility of hydrogen in an austenitic microstructure than in a ferritic microstructure, coupled with the significantly lower diffusion rate of hydrogen in austenitic materials. The hydrogen in an austenitic weld metal (which is itself relatively insensitive to cracking) will only diffuse slowly into adjacent ferritic microstructure heat affected zones (which may contain microstructures susceptible to cracking). In order to give some further protection against hydrogen cracking, a temper bead technique was suggested, to refine and soften the HAZ, before cracking may occur. By using austenitic electrodes, the need for preheat was eliminated along with the need to maintain preheating between passes.
Calculation and weld procedure development
At each stage of the investigation, the PRCI Hot Tap programme was used to determine the parameters that would equate to the pipeline and the test welds. This model generated a value for the cooling time from 800°C to 500°C for the weld passes, which was used together with the TWI hardness predictor to predict the maximum HAZ hardness. The values predicted were compared to the values determined in the weld procedure development. The results of the calculations are reported in Table 3. The maximum as-welded hardness predicted, based on the combination of outputs from the models are 400HV in the pipe and 457HV in the sleeve. The uncertainty in the predicted hardness values is ±56HV (2 standard deviations).
Table 3. Results of calculations. Hardness prediction
|Pipe HAZ hardness,|
|Sleeve HAZ hardness,|
The values predicted for the 250-100°C heat sink capacity on the test rig were significantly lower than the values actually measured, indicating that the prediction is conservative with respect to test conditions.
The weld procedures proposed were based on the procedures employed in the previous in service modifications. Significant differences for the present study were; the electrodes to be used, the absence of preheat and the need to weld vertically up rather than vertically down.
In order to confirm the validity of the proposed procedure a number of single pass bead on pipe welds and a single buttering layer were deposited on the surface of a length of experimental pipe mounted in the test rig. The water flow rate was 0.1m/s. These welds allowed the maximum hardness and the efficiency of the temper-beading technique to be established. Metallographic sections were taken from these welds to assess the microstructure and hardness of the heat affected zone (HAZ). The HAZ microstructure for the bead on pipe weld is given in Figure 3, and the HAZ microstructure for the buttering layer is given in Figure 4. The maximum hardness measured in the bead on pipe HAZ microstructures was 379HV, which is slightly lower than the predicted values. The tempering achieved by buttering was partially successful, giving a slight modification to the microstructure, but only reducing the hardness to 330HV. Nevertheless, buttering layers were applied to the pipe prior to the attachment weld. For field welding, it was also recommended that buttering layers were applied to the split tee ends, and then postweld heat treated before delivery to site to reduce the hardness of the split tee HAZ before making the attachment weld.
a) Macrophotograph, showing overall microstructural features
b) Detail of the heat affected zone showing a martensitic microstructure
Figure 3. Single pass bead on pipe photographs. Magnification indicated by micron marker.
a) Overall view, showing the limitations of overlay for the temper bead. Scale in mm
b) Detail of HAZ, showing effect of tempering. Magnification indicated by micron marker
Figure 4 Photographs of the buttering trials
The test welds were deposited in accordance with the weld procedure specification (summarised in Table 4), with two welders working simultaneously on opposite sides of the pipe. Welding was carried out intermittently over a five day period, and the welding time was 32 hours. The internal wall temperature immediately below the weld and the water temperature adjacent to the outlet were monitored using thermocouples. The maximum measured internal wall temperature was 122°C. The inner wall temperature measured accounts for the wall thickness and the ability of the flowing contents to remove heat from the wall. For oil, flowing in a thinner walled pipe, the inner wall temperature would be expected to be slightly higher than this value. No significant change in the bulk water temperature was recorded. The test welds were subjected to the relevant qualification tests in API 1104 and exceeded the minimum requirements for all tests. The sections taken as part of the qualification (Figure 5) were subjected to hardness testing, and the maximum values recorded for the HAZ on the pipe (401HV5) and the sleeve (415HV5) were consistent with the predicted hardness, but also highlight that the intended tempering arising from buttering the pipe surface was not very successful. The cooling time from 800°C to 500°C determined was 0.83s.
Table 4. Summary of welding procedure
|Layer||Position||Electrode designation||Electrode size,|
Figure 5 Macrographs of the test welds. Ruler shown
In addition, to compare thermal severity in systems with differing flow rates and flowing contents, a heat capacity measurement was made, in accordance with the method described by Bruce. This method is a standard approach used to determine heat sink for in service welding, endorsed by PRCI. The measurements were repeated for a range of water flow rates from 400litres/min to 1600litres/min (0.025m/s to 0.1m/s), to allow comparison with the values recorded on the pipeline. The results are given in Table 5. The heat sink capacity measurement from the intended hot tap location was 53s, which indicates the increased severity of the flowing water, and the conservatism built into the PRCI Hot Tap model.
Table 5. Heat sink capacity measurements from the TWI test rig
Table 5a. Summary of pipe to weld interface for the NDT test weld
(L x W)mm
size (L x W)mm
|7MHz measured size|
(L x W)mm
|2MHz measured size|
(L x W)mm
Table 5b. Summary of sleeve to weld interface flaw sizes in NDT test weld
(L x W)mm
|7MHz measured size (L x W)mm|
Table 5c. Summary of indications in procedure qualification welds
|Weld||Scan surface||Flaw details (mm)||Comments|
||No significant indications
||Largest of several small indications
||No significant indications - at the fusion line
1. Scan start (0mm) is at 12 o'clock.
2. Scan direction is clockwise looking at the weld from the pipe side (3 o'clock is approximately 518mm measured around the sleeve).
The successful application of welding for in-service modification requires inspection of the welds to ensure compliance with the appropriate standards and to increase confidence in the integrity of the joint. To facilitate development of an appropriate procedure, a bespoke 'test block' with intentional flaws was also made. A number of flaws were introduced adjacent to the pipe and sleeve surfaces during deposition of a trial weld in order to allow development of an ultrasonic inspection procedure based on the use of phased array UT equipment. The flaws were introduced by using TIG welding to produce a 'bridge' over the artificial flaw location. This 'bridge' then allowed a region of simulated lack of fusion to remain after completing the welding in accordance with the WPS. The flaws were intended to be 25 x 5mm (on both the sleeve face and pipe face), 5 x 15mm (on both the sleeve face and pipe face), 10 x 5mm (on both the sleeve face and pipe face) and 5 x 5mm (on the pipe face). The intended dimensions are reported as length x height/width as appropriate.
Following evaluation of the test piece the optimum probe types and test configurations were found to be as detailed in the following summary procedure:
1. From the pipe surface
- Initial flaw detection scans were carried out at full skip range using a 7MHz 32 element 0.7mm pitch, with the probe mounted on an integral 34? wedge. A 40° to 65° azimuthal scan with an increment of 0.3° was employed. The stand-off to the rear of the probe housing was adjusted such that the 40° edge of the beam just covered the weld toe at full skip range. The focal depth was 25mm and the reference gain was set using a 3mm diameter side drilled hole (SDH) at a depth of 30mm. An extra 10dB was added for scanning purposes, which is usual practice for manual UT.
- For measurement of flaw width a local full skip scan was employed using a 2MHz 32 element, 1.5mm pitch probe on a detachable 34° wedge. Due to the large size of the wedge it was necessary to shift the emission point forward by reducing the number of active elements to 25. A 40° to 70° scan in 0.3° intervals was utilised. The adjustment of stand-off, reference target used and additional gain for scanning were all as for the 7MHz initial detection scan.
2. From the sleeve surface:
- Only one scan up to half skip range was required to detect and size lack of fusion flaws occurring on the sleeve fusion face. This scan used the same probe type as used for the initial detection scans from the pipe surface with a swept beam range of 40° to 80° in increments of 0.3°. In this case, the sweep range was split into three sectors, each having a different focal depth as follows:
- 40°-55° focal depth = 20mm 55°-65° = 15mm 65°-80° = 10mm The reference gain was set using a 3mm diameter side drilled hole at a depth of 15mm, with an additional 6dB being added for scanning purposes.
In all cases an A-scan collection interval of 1.5mm was used.
To determine the actual flaw sizes and assess the sizing performance of the above scans the flaws were subject to a 0° phased array scan from the bore of the pipe. This scan used a 7MHz, 128 element, 0.6mm pitch probe focussed at a depth of 12.5mm.
The 7MHz scan from the pipe surface detected a facet of each of the artifical flaws with the exception of the 5x5mm flaw which was not detected. The data enabled an accurate measurement of length, but no assessment of flaw width could be made, due to an insufficient signal to noise ratio at the interface. The 2MHz scan from the pipe surface only detected the 25x5 and 10x5mm flaws. However, the signal to noise ratio was much improved, and, an accurate length and width measurement was possible for each flaw. Table 5 indicates the flaws identified in the procedure development sample. This shows that other than the intended 5x5mm flaw, the 7MHz scan detected and sized the flaws in a comparable manner to those detected from inspection from the inner surface of the pipe. Inspection from the sleeve surface detected all the simulated flaws, allowing measurement of length and height.
The simulated flaws contained in the procedure development weld were very smooth in nature, which meant that they were particularly difficult to detect with oblique incidence beams. In particular, this was a problem when testing from the pipe surface. However, it is likely that real flaws will be less smooth and therefore be more readily detectable.
During the procedure development an assessment was made of the benefits of both weld toe grinding and polishing the scanning surface. It was concluded that toe grinding was detrimental and caused unwanted full skip signals on the images. However, a light sanding with SiC paper of the scanning surface to remove any grinding marks was beneficial and considerably eased the friction experienced during scanning.
During the procedure development a gel couplant was used. It was noticed that surplus couplant in front of the probe caused false indications on the images. To eliminate this effect, it was suggested that water couplant be used where possible.
From the pipe surface the minimum size of planar flaw located at the pipe to buttering layer interface that was reliably detected and sized is 10x5mm (length x width). The sleeve scan identified flaws with a minimum length of 5mm and minimum height of 5mm. In theory, therefore it should have been possible to reliably detect and size a 5x5mm flaw located at the sleeve/weld metal interface from the sleeve surface.
No significant indications were identified on the test weld by phased array inspection on the pipe side of the weld. One intermittent indication, of approximately 10mm length and 7mm height was identified at the fusion boundary from the sleeve side. This was interpreted as lack of fusion. This location was sectioned to identify the nature of the intermittent indication. No flaws could be identified at this location, and the indication was considered to be spurious. In any case, API 1104 indicates that this indication would not be considered to be rejectable, since it was intermittent, the total height was less than half the thickness and the length was less than 25mm. Other ultrasonic indications from the sleeve side inspection were determined to have no measurable through-wall height and no sections were taken.
It should be noted that in the present test welds (25mm leg length) only 84% of the sleeve fusion line and 2x pipe wall thickness (98%) of the pipe fusion line were available for inspection (Figure 6). For in-service welds of leg length 30mm, and pipe wall thickness 9.5mm, only 63% of the pipe fusion line will be available for inspection and for a sleeve thickness of 30mm, only 90% of the fusion line will be inspectable. For larger sleeve thicknesses (for the same leg length), the whole sleeve fusion line can be inspected.
Figure 6. Graphical illustration of extent of coverage of the fusion lines. The root on the pipe side, and the sleeve side toe are not covered by the inspection
The use of phased array probes rather than conventional angled probes was beneficial as the array of probes was programmed to cover a greater range of angles in a single inspection pass, (shorter inspection times), improved sizing capability, pictorial representation of the data, permanent records and post-acquisition data manipulation, such as statistical analysis. However, if the set-up of the probe had not been optimised, a high signal to noise ratio would have been generated, and time was needed to be spent in technique development prior to application. This aspect of the work was extended by further studies carried out by the welding contractor used for the in-service welding.
Following a programme of work, comprising calculation, rig design and manufacture, weld procedure design, welding and weld procedure testing, non-destructive testing and procedure development, metallography and hardness testing, the following conclusions were drawn:
- Austenitic stainless steel consumables can be used to avoid HAZ fabrication hydrogen cracking, even in the presence of susceptible HAZ microstructures.
- The use of a temper bead technique, when applied within tight parameters with respect to bead overlap and temper bead positioning, has been shown to reduce the hardness of the heat affected zone in the pipe. However, it was observed that it was difficult to obtain consistently reduced hardness.
- The use of phased array ultrasonic inspection has demonstrated that indications with a length of more than 5mm can be successfully identified using this technique, for a split tee connection.
- Bruce W A and Kenzie B W: 'Development of optimized nondestructive inspection methods for welds made onto in-service pipelines - Phase III' PRCI Report, 1992
- API 1104: 'Welding of pipelines and related facilities'. Twentieth edition, November 2005.
- Bruce W A, Li V and Gitterberg R: PRCI Hot Tap 4.1: 'PRCI Thermal Analysis Model for Hot Tap Welding by EWI'. PRCI Project PR-185-9632, L51837, May 2002.
- BS EN 1011-2: 'Welding. Recommendations for welding of metallic materials. Arc welding of ferritic steels' BSI, London, 2001.
- Wilson J: 'The development of welding techniques for British fighting vehicles.' Proceedings of Trends in Steels and Consumables for Welding, International Conference, London, 13-16 Nov.1978 Paper 2. pp.473-479. Publ. The Welding Institute.
- Bailey, N: 'Hydrogen cracking and austenitic electrodes' Metal Construction, Vol 10, No. 12. December 1978 pp.580-583.
- Gooch T G: 'Repair welding [of 'hard to weld' ferritic steels] with austenitic stainless steel MMA [manual metal arc] electrodes.' Metal Construction, vol.12, no.11. November 1980. pp.622-627, 629, 631.
- Moreton J, Coe F R and Boniszewski T: 'Hydrogen movement in weld metals' Metal Construction, Vol.3 no.6 , June 1971, pp. 223-228.
- Nicholas J M and Abson D J: 'The prediction of maximum HAZ hardness in C-Mn and low alloy steel arc welds' Paper presented at the 17th International Conference 'Computer Technology in Welding and Engineering' Held at the University of Cranfield, 18-19 June 2008.
- Bruce W A: 'Selecting an appropriate procedure for welding onto in-service pipelines'. International Conference on Pipeline Repairs, Wollongong, Australia, 5-6 March 2001.