Beruflich Dokumente
Kultur Dokumente
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report Contents page
Mike Painter, CSIRO Manufacturing Science and Technology Prakash Sabapathy, The University of Adelaide
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report Contents page
Acknowledgement
This is a collaborative project of the Cooperative Research Centre for Welded Structures, under the sponsorship and guidance of the Pipeline Program Management Committee. Therefore it would not have progressed without the guidance, endeavor and financial assistance of the following organisations and people:
Collaborative Partners CSIRO Manufacturing Science & Technology The University of Adelaide BHP Epic Energy AGL Gas The Pipeline Program Management Committee Researchers M.A.Wahab, The University of Adelaide Bing Feng, BHP Prakash Sabapathy,The University of Adelaide Industrial mentors Paul Grace, WTIA Hans Borek, Epic Energy
M.J.Painter, CSIRO Manufacturing Science and Technology P. Sabapathy, The University of Adelaide
CRCWS
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
Collaborative Partners
CSIRO Manufacturing Science & Technology The University of Adelaide BHP Epic Energy AGL Gas The Pipeline Program Management Committee
Researchers
Mike Painter, CSIRO Manufacturing Science & Technology M. A. Wahab, The University of Adelaide Bing Feng, BHP Prakash Sabapathy, The University of Adelaide
Industrial mentors
Paul Grace, WTIA / ex AGL Gas Sydney Hans Borek, Epic Energy
Project Objective
To develop recommended weld procedures for the safe and effective in-service welding of thin-wall, high-strength steel, high-pressure gas pipelines.
Executive Summary
Background
The process of welding onto a live high-pressure pipe is frequently employed for the repair, modification or extension of gas pipelines. This in-service welding has significant economic advantages for the gas transmission and gas distribution industries, since it avoids the costs of disrupting pipeline operation, and it maintains continuity of supply to customers. In-service welding is an essential part of hot-tapping, a technique which allows the creation of a branch connection to a live pipeline. In-service welding is also important for pipeline maintenance, such as the installation of sleeves around damaged sections of pipe. Direct deposition of welds onto live pipes has been suggested as a way of replacing wall thickness lost through corrosion or local damage. If in-service welding is not possible then sections of a pipeline have to be sealed and degassed prior to welding, and then re-purged prior to reinstatement. These are costly, wasteful, and environmentally damaging actions, since there are large gas losses and methane is a green-house gas. TransCanada Pipelines Ltd. has estimated that relative to a cold connection a hot-tap avoids gross revenue losses of approximately 1M$Canadian per hot- tap.
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
heating, and the reduction in pipe-wall-strength during the welding process. If this reduction in strength is too great the pipe wall can burst under the pipes internal pressure. This hazardous event is termed burn-through. Increasing the welding heat input can reduce fast cooling, but this promotes weld penetration and increases the risk of burn-through. Suitable weld procedures must ensure the HAZ hardness is not high enough to cause cracking, whilst heat input and penetration are not so high that the integrity of the pipe wall is jeopardised. In Australia, there is a significant trend towards the use of high yield strength steels for pipeline construction. Future pipelines using X70 and X80 steels could have wall thickness as low as 3mm. Unfortunately, in-service welding is made much more difficult with such thin-walled pipes. Thin pipe walls increase the risk of burn-through during welding, and are more easily cooled by the flowing gas. High strength steels can also be susceptible to the generation of excessive hardness for a given cooling rate. If the economic advantages of in-service welding are to be maintained then technology to support the safe and effective welding of thin-walled high-strength pipelines must be established.
Project Objective
To develop recommended weld procedures for the safe and effective in-service welding of thinwalled, high-strength steel, high-pressure gas pipelines.
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
Project Structure
The important features of the project were: 1. An experimental study of circumferential manual metal arc (MMA) welding in the vertical-up and vertical-down position, with low hydrogen electrodes (E8018G and E7016, 2.5 mm & 3.2 mm diameter). These tests used a water-cooled simulation of in-service welding and a range of typical Australian pipe grades (X42-X80). 2. The extensive development of numerical finite element simulations of in-service welding. These numerical simulations covered 2D and 3D models, and unlike previous research work aimed to develop numerical simulations of burn-through. 3. The transfer of technology to the pipeline industry.
Experimental studies have generally used water flow to simulate the cooling effect of the gas, but there is a general recognition that water gives much higher quench rates than gas flow. It therefore generates conservative welding conditions for a required hardness. That is heat input determined with water-flow will give slower t8/5 with gas flow. The corollary of this however, is that using waterflow simulations may give non-conservative heat inputs for burn-through.
Outcomes from the Experimental Study of Hydrogen Controlled Electrodes. The analysis of welding conditions and welds produced using pipe cooled with a water jacket to generate a rapid quench, gave the following results;
It was possible to generate a weld on 7.8 mm thick X80 pipe with a maximum hardness of 325 HVN with a weld t8/5 cooling time of 3.8 seconds. For the range of pipe grades examined (X42-X80), the Yurioka-1 carbon equivalent (CE) relationship provided the best correlation between composition, hardness and t8/5 cooling time. This correlation gave an absolute error of 5.7% for E5548-G (E8018-G) electrodes, and a
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
3.4% error with E7016. Based on this relationship the X80 grade of pipe used in this work would only require a t8/5 of 0.8 seconds in order to achieve a hardness of >350 HVN. A small beneficial tempering effect was measured from multi-pass [3 passes] welds. This was most noticeable for the X70 grade steel, which with a CE of 0.288 gave a hardness of 381 HVN after a single root pass. After three passes, the maximum hardness was reduced to an acceptable 321 HVN. Only the two X60 steels with CE of 0.38 and 0.41 gave multi-pass hardness >350 HVN. A minor difference in the incidence of weld defects was observed between the two electrode types tested. There was a greater incidence of HAZ cracking with E8018-G in keeping with its reduced heat input. Some HAZ cracks were detected with X80 grade although the maximum hardness was 299-310 HVN. There was no systematic variation in heat input as the weld progressed around the pipe. However, whilst total weld energy remained reasonably constant the natural variation in welding speed, since this is a manually skilled process, caused variation in the heat input. This was significant and amounted to approximately a 20% variation on a nominal value. Penetration into the run-pipe was generally greater when using the E7016 electrode in the vertical-up position rather than with E8018-G in the vertical-down. Penetration into the run-pipe slightly increased with increasing heat input although this effect was largely swamped by a significant variability at a given heat input. At a nominal heat input of about 1 kJ/mm this variability was approximately, 0.2-0.8 mm with E8018-G and 0.5-1.0 mm with E7016.
These experiments also provided, empirical data to relate the deposited weld bead volume, and weld bead shape to heat input for both E8018-G and E7016 electrodes, and preliminary data on measured t8/5 cooling times and measured weld penetrations in order to validate and refine numerical models of MMA in-service welds.
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
Models have mainly considered single root pass welds but some multi-pass welds have been simulated. Various methods of assessing the risk of burn-through have been developed. These were based on: the maximum temperature at the inside surface of the pipe, following Battelle, a thermo-elastic plastic stress analysis, using the thermal field in the pipe wall to calculate the reduction in wall strength.
Thermo-elastic-plastic models of circumferential welds have showed similar deformation patterns to those observed during burn-through, namely a localised bulge in the pipe wall near the weld. Failure can be specified as a bulge that exceeds a limiting height. Plots of internal pressure versus bulge height have shown an effective yield pressure, which can also be used as a failure index. Only a limited numbers of these models were studied because they were very computationally demanding. Estimating the reduction in pipe wall strength in the weld zone has created a novel alternative method of assessing burn-through risk. This method determines the reduction in material strength around the weld based on the predicted temperature field, and the known relationship between material yield strength and temperature. This reduced strength is regarded as equivalent to a local reduction in the thickness of the pipe-wall at constant ambient temperature. Hence, the temperature field around the in-service weld effectively converts to a cavity in the pipe wall. A number of alternate strategies can be considered. The limiting pressure for safe welding can be based on the remaining wall thickness, or based on the effective reduction in cross-sectional area. The risk of burn-through is equivalent to the possibility of this cavity causing rupture at the current operating pressure. This assessment can also be easily carried out utilising the approach specified for the evaluation of corrosion cavities in Australian Standard AS2885. This method has produced excellent results. Although limited by lack of data, comparison between predicted safe welding pressures and published values measured on 5 mm thick pipes has been good. The approach provided a way of assessing burn-through potential which is in agreement with reported behaviour. Longitudinal welds are more prone to burn-through than circumferential ones of the same heat input. Pressure has a significant effect. The width or size of the weld is important as well as penetration. It provides an efficient approach to in-service weld simulation since it does not require a stress analysis and uses only thermal predictions. Unlike Battelles maximum wall temperature approach, it is more realistic since it accounts for weld orientation and internal pipe pressure.
Model Validation
The above numerical simulations were validated by comparing predicted values with: published result for t8/5 cooling times measured on pipes of 4.8 mm wall thickness, measured HAZ and fusion zone geometries from a hot-tap coupon, data from welds carried out on an uncooled, empty pipe, measured values of t8/5 and HAZ hardness obtained from a number of test welds carried out on a flow loop at Duke Energys Gladstone Gate facility.
This last extensive validation used simulated circumferential fillet welds on three materials, 4.8 mm and 5.2 mm thick X70, and 6.4 mm thick Ultrapipe X42, under a range of gas pressures and flows. The t8/5 cooling times were measured, welds were metallographically sectioned and HAZ hardness determined. Predicted values of fusion zone depth, HAZ depth, t8/5 cooling time, and HAZ hardness, compared favourably with measured values. This was an aggressive test of the models validity and accuracy.
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
analysis without significant investment in software and the development of personnel with appropriate expertise. Although access to the modeling capabilities developed within this project will remain, it was felt that results could be put in a more accessible form. One possibility has been developed to a prototype stage. This consisted of using the finite element models to develop a database of predicted values for a range of heat inputs, pipe wall thickness, and heat transfer conditions at the pipe wall. Concentrating on the circumferential fillet weld, it has been established that the t8/5 cooling time is almost independent of pipe diameter provided the heat transfer coefficient at the pipe wall is constant. That is, the heat transfer coefficient determines the weld cooling rate. This effectively means that a single model can provide results for any combination of gas pressure, flow rate and pipe diameter, which gives a constant heat transfer coefficient. There is a smooth variation in calculated t8/5 cooling time, weld penetration, effective cavity sizes etc. as heat input, and pipe wall thickness varies. Hence, it is feasible to interpolate values from a data set with a reasonable degree of accuracy. Using this approach it is possible to develop a very fast computer program which produces estimates of HAZ hardness, and burn-through risk by simply interpolating an established database.
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
Figure 1(d) Drill and cut coupon removed and Slide valve closed
Figure 1.1 An illustration of the hot-tapping process taken from IPSCOs animation of hot-tapping on http:/www.hottap.com.us
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
Figure 1.2 Schematic illustration of burn-through, caused by localised heating and internal gas pressure.
P.D
2.t w
= 0.72. y
where P is the internal pressure, D is the pipe diameter, tw is the pipe wall thickness, and is the minimum specified yield strength.
Hence for a given diameter pipe and gas pressure the tonnage of pipe required for a given distance can be reduced as the materials yield strength is increased. Alternatively using high yield strength pipe permits the transmission of natural gas at higher pressures and flow rates. The Australian pipeline industry recognises the economic advantages of using high strength steels. Unfortunately the use of thin walled, high strength steel pipelines increases the difficulties associated with in-service welding. With the combination of enhanced gas transmission and diminished wall thickness the weld cooling rate for a given weld procedure increases. Such high strength steels have a greater sensitivity to strength reduction during welding and together with the decreased wall thickness are more prone to bulging or burnthrough.
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
10
1.4 Summary
Weld procedure development is particularly difficult for in-service welding. For safety and practicality, experimental test welds can not simply be carried out on live pipelines. Hence, external means of establishing weld procedures have to be used. Traditionally two approaches have been developed. Laboratory simulation of pipe flow conditions (Edison Welding Institute, 1980-1990 (5,6)) or through the use of simple numerical calculations (Battelle Memorial Institute, 1985(7,8)). Approximations in such approaches may lead to inaccuracies and excess conservatism in the choice of weld parameters. Such difficulties will be increased by the Australian pipeline industrys use of higher strength, thin walled pipelines because the existing technology may not apply to such new materials.
To develop recommended welding procedures for the safe and effective in-service welding of thin-wall, high-strength steel, high pressure, gas pipelines.
The technical challenge is to develop methods of establishing welding procedures which produce welds that are free from the risk of cracking, and do not risk bursting the pipe wall during welding: and to confirm their application for the thin walled high strength materials that will be used in future pipeline construction.
References
1. McElligott J. A., Delanty J., & Delanty B. Full Flow High-Pressure Hot Taps: The New Technology and Why Its Indispensable to Industry, Paper Presented at International Pipeline Conference pub. ASME v2, 1988, pp813-820. 2. Venton P., Report Prepared for Pipeline Program of Cooperative Research Centre for Materials Welding & Joining 1996. 3. TransCanada Pipeline, http://www.transcanada.com 4. Bruce W. A., Holdren R.L., Mohr W. C., Kiefner J.F. & Swatzel J.F., Repair of Pipelines by Weld Metal Deposition, Paper presented at PRCI 9th Symposium on Pipeline Research, Houston Texas, September 1996. 5. Bruce W. A. & Threadgill P. L. Welding Onto In-Service Pipelines Welding Design & Fabrication Feb 1991, pp19-24. 6. Cola M.J. & Threadgill P.L., Final Report on Criteria for Hot Tap Welding, American Gas Association, Edison Welding Institute Project J7038, March 1988 7. Kiefner J.F & Fischer R. D. Models Aid Pipeline Repair Welding Procedure Oil & Gas Journal March 1988, pp41-47. 8. Fischer R.D., Kiefner J.F. & Whitacre G.R., User Manual for Model1 & Model 2 Computer Programs for the Predicting Critical Cooling Rates and Temperatures During Repair and Hot Tap Welding on Pressurised Pipelines, Battelle Memorial Institute Report, June 1981.
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
11
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
12
Different joint configurations are used by the pipeline industry which require individual analysis. Australian hot-tap fittings can be broadly classified under three types, the full encirclement, circumferential sleeve fitting, see Figure 2.1(a), the direct-branch with reinforcement-saddle, see Figure 2.1(b), and the direct-branch-to-pipe weld, see Figure 2.1(c). As the pipe thickness decreases the full encirclement sleeve provides the best structural support to both the pipe and attachments. Therefore it is the most common joint configuration. Although longitudinal welds are used to secure the sleeve around the pipe these do not directly contact the pipeline and therefore are not critical. This research program has concentrated on numerical simulations of circumferential sleeve welds and branch connections with reinforcement sleeves.
(a)
(b)
(c)
Figure 2.1 Common in-service welding pipe configurations: (a) full encirclement fitting, longitudinal weld to join sleeves and a circumferential fillet to the run pipe. (b) Reinforcing saddle around branch pipe. (c) Directly welding the branch pipe on to the run pipe.
Numerical models of fusion welding processes always include some empirical factors to ensure that the resulting calculated values agree with those found in practice. This means that such models can not be created without significant experimental input, and their accuracy must be validated. The welding process commonly used for in-service welding in Australia is MMA welding using hydrogen controlled electrodes. Unlike other welding processes, MMA welding requires relatively little equipment (power supply + stick electrode) and is the traditional process for infield pipeline welding. An experimental assessment of the performance of two typical electrode types in common use has been carried out using a range of pipe material grades. Although there is a large body of work on numerical modelling of fusion welding there is little specifically addressing MMA welding. This research has addressed that deficiency. In particular it has developed appropriate modelling strategies, for vertical-up or vertical-down MMA welding positions. The current numerical simulations have been validated by comparing predicted values with: published results for t8/5 cooling times measured on pipes of 4.8 mm wall thickness, with measured HAZ and fusion zone geometries from a hot-tap coupon, with data from welds carried out on an uncooled, empty pipe,
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
13
measured values of t8/5 and HAZ hardness obtained from a number of test welds carried out on a flow loop at Duke Energies Gladstone Gate facility. The results of thermal analysis have been directly compared with micrographs of test welds to ensure the correct calculation of weld penetration and HAZ geometry. The measured cooling rates of test welds have been compared with predicted values. HAZ hardness has been measured for a range of materials and test welds and has allowed further comparisons between predicted hardness and measured values.
This last extensive validation was an aggressive test of the models validity and accuracy under operational conditions on a live pipeline. Although finite element software is readily available, developing and using thermal models to simulate weld processes is a specialised activity. It would be difficult for industry to adopt this approach without significant investment in software and the development of personnel with appropriate expertise. Although access to the modelling capabilities developed within this project will remain, it was felt that results could be put in a more accessible form. One possibility has been developed to a prototype stage. This consists of using the finite element models to develop a database of predicted values for a range of heat inputs, pipe wall thickness, and heat transfer conditions at the pipe wall. Using this approach it is possible to develop a very fast program which produces estimates of HAZ hardness, and burn-through risk by simply interpolating an established database. It is anticipated that through the development of improved numerical simulations of in-service welding, and the establishment of their accuracy and scientific credibility, more efficient weld procedure development will result. Validated numerical models will also allow a safe combination of welding procedure, gas pressures and gas flows to be determined for a given pipe geometry, and this in itself will facilitate more efficient management and control of inservice procedures.
2.3 Outline
Section 3 will discuss the experimental work that has examined in-service welding. The areas of post weld hardness and the possibility of pipe wall failure during in-service welding will be its foci. Section 4 will introduce the concepts related to the numerical simulation of welding processes. Sections 5 & 8 will concentrate on the computer simulation of in-service welding, and on work related to the development of a numerical approach to the prediction of safe welding procedures.
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
14
3. Literature Review:
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
15
through hydrogen embrittlement. They quoted the case of a 200mm diameter 8mm thick, steel pipe with a carbon equivalent (CE) of 0.48. Welds on this pipe gave a HAZ hardness of 415 VHN for a gas flow of 0.518 scm/h and 285 VHN with no gas flow. Whilst their main concern was with the pipes HAZ hardness they also indicated that the composition and cooling rate within the fitting should not be ignored. Cola et al(1) reported on the cracking propensity of welds made with basic and cellulosic electrodes. They considered that with cellulosic electrodes a limiting t8/5 (time to cool from 800C to 500 C) could be established independent of pipe composition. This conclusion was based on data generated from simulated in-service welds using an E6010 electrode at 1kJ/mm heat input. For a range of steels with a CE of 0.3-0.5 cracking was only found for t8/5 cooling times below 5 seconds. When using basic electrodes they recognised that the pipe composition was an important factor, and determined that cracking was only a concern for an HAZ hardness >400 VHN. They also reported that, provided basic electrodes and a sound low-hydrogen welding practice were used, the risk of cracking was not significant for steel having a CE < 0.5. The significance of this relates to the age of the pipe since modern steel compositions generally give a CE < 0.4-0.45. The use of basic electrodes is not just related to their cracking propensity. Whilst they clearly reduce the risk of hydrogen embrittlement, as Phelps et al(7) showed, they also generate significantly lower penetration for a given heat input and hence reduce the risk of burn-through (see Section 9). Welding vertically-down or welding with the electrode DCEN were also reported (7) to give a lower penetration than welding vertically-up, or using electrode DCEP. Boran(15) found that post-weld hardness was influenced by the electrode polarity of the MMA welding process. The polarity influenced the apportionment of heat between the electrode and the workpiece,. DCEN gives the greater fraction of heat in the weld region, which reduces the weld cooling rate and gives the least post weld hardness for a given heat input. Preheating the joint before welding is an obvious, traditional, way of controlling and reducing the cooling rate of during welding. For example, DeHertogh & Illeghems (16) preheated a 323.5 mm 3 diameter 4.4 mm thick, X60 pipe with a gas flow of 25,000 m /hr at 4.8 MPa. They found that the hardness of a weld decreased considerably from 367 to 317 when using an 80C preheat. Using inductive heating they examined preheat levels from 50-200C and reported that minimum hardness was achieved at a 100C preheat, although no explanation of the minimum value was proposed. Cassie et al (17) listed the recommended features of a satisfactory in-service welding procedure as; use of basic low hydrogen electrodes, a preheat of 100C for material with a CE < 0.4 and 150C for material with CE>0.4, and the use of a stringer bead technique. Preheating has an economic penalty of course, but for in-service welding there are other significant difficulties. These relate to the extremely high heat loss generated by the flowing gas. Under such conditions achieving a consistent preheat is difficult. A number of methods were tried, including direct gas flame heating, electrical resistance and inductive heating. Phelps et al (7) reported that none of these was entirely satisfactory. The method they recommended was direct flame heating using a hand-held propane torch. This was used to heat the region ahead of the weld to a maximum temperature of 250C. This preheating step was followed by welding for a short time, until the temperature of the region fell below the desired preheat level. This cyclic process was then repeated. Cooling rates were extremely high so it followed that welding-runs would be short. Since consistency of heat input relies on manual skill of the welder, such an intermittent process can only contribute to the variability in welding speed and in heat input. In-service welds are multi-pass, so the sequence of welding can be controlled to minimise HAZ hardness. Using a stringer bead technique, Figure 3.3(c), in which the current weld tempers the hardening created by the previous one, is recommended by Cassie (17) and Bruce & Threadgill (2). Bruce (18) describes the sequence recommended by British Gas which uses a buttering layer and a temperbead sequence, also shown in Figure 3.3(b). Rietjens (19) made reference to a desirable weld
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
16
preparation for split circumferential sleeves, as shown in Figure 3.3(a) and identified that buttering and temperbead sequences were useful in minimising crack susceptibility. He also advocated using weld metal with low yield strength, in order to relieve residual stress. Cassie et al (17) examined the application of post-weld heating as a means of reducing the hardness at the weld toe. They found that using a gas tungsten arc was effective but they did not consider it a viable field technique, other methods were not reported in detail but were considered ineffective. Variability in the manual welding process was identified as a concern by a number of researchers. Cola et al (1) referred to the inherent variability in the manual process. Cassie (17) identified variations in hardness due to different welders using different welding speeds and hence a varied heat input. Bruce et al (3) recommended the use of controlled deposition rates in order to minimise such variations. There are some reports of a systematic variation in weld properties around circumferential welds. DeHertogh et al(16) indicated an increase in the depth of HAZ as the weld progressed from the top to the bottom-dead-centre of a circumferential MMA weld. They considered that this was a consequence of the general heating of the pipe as welding progressed. For 4.8 mm thick pipe and vertical-down MMAW Phelps et al (7) also found that penetration was significantly greater at the 6 oclock position (1.3-1.4 mm) than at the 3 oclock position (1.0 mm).
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
17
D Where p is pressure, y is minimum yield strength, D is pipe diameter, t is pipe wall thickness, and, c is an assumed thickness reduction, to account for reduced wall strength equivalent to 2.38 mm.
3.3.2 Influence of heat input on burn-through
As the welding heat input is increased the weld penetration, and size of the heated region increases, with a consequent increase in the possibility of burn-through. Cassie(17) investigated the process parameters controlling weld penetration using various out-of-position welds on 6, 9 and 12 mm thick plate. From these results, he determined that welds in a vertical-down position using basic electrodes provided the lowest penetration, and therefore would be the most appropriate for in-service welding. Cassie(17) used simulated in-service welds on pressurised cylinders of 450 mm diameter X52 steel with wall thickness of 3.2, 4.8 and 6.4 mm. He defined a safe limit by specifying the maximum arc current allowable for a given pipe thickness, see Figure 3.5. Safe welding currents could be determined for all pipes with wall thickness above 3.2 mm. However, the final recommendation was that welding should not take place on pipes of less than 4 mm thickness at an internal pressure greater than 7 MPa. Bruce et al(11) also referred to a restriction on welding current for safe in-service welding on pipe walls 3 mm thick. They identified that at the same heat input, welding with a smaller diameter electrode (equivalent to a reduced current) reduced the burn-through risk. The limits suggested are shown diagrammatically in Figure 3.6. This shows an interesting difference in behaviour between 3.2 mm and 4 mm thick pipes. For a 3.2 mm pipe wall the division between safe and unsafe welding was strongly dependent on electrode diameter (arc current), with 2 mm diameter (50 A) electrodes safe, and 2.4 mm diameter (80 A) electrode borderline. With a 4 mm thick pipe-wall the conditions are more dependent on heat input. This apparent difference is unexplained. Bruce et al(11, 12), determined recommendations for safe in-service weld repair as: A maximum internal pressure of 6.7 MPa during welding on a minimum remaining wall thickness of 3.2 mm, Electrode type to be a hydrogen controlled, E7018, with electrode size to be restricted to 2.4 mm diameter, A heat input of 0.51 kJ/mm for the first weld runs.
p=
1.44. y (t c )
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
18
considered burn-through limits in terms of hoop stress, implying that the pipe diameter has an influence.
= 0.2 0.85 /
Where,
t r
t r
0.85 M
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
19
M = 1+
0.4 L Rt
t is the pipe wall thickness L is the axial length of the weld pool perpendicular to the hoop stress r is the maximum depth of the defect R is outside radius of the pipe Using these formulae Bout & Gretskii(22) then estimated the critical length of heated zone (Lcritical) before the pipe would not be able to operate at its maximum load carrying capacity.
For a 3 mm wall thickness, 320 mm diameter pipe with an internal pressure of 4 MPa, burn-through occurred at a heat input of 0.475 kJ/mm with a normal sleeve. With a constraining band the critical heat input was 0.915 kJ/mm. This represents an interesting novel approach. The only concerning feature is that the practical field application quoted are for 9mm thick pipes. Also no consideration was given to the increased cooling rate that would be generated within this joint configuration.
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
20
An important feature of procedure development concerns selecting the minimum suitable heat input. This should be selected to achieve a weld with HAZ hardness below the level likely to cause cracking, but not be so high that the pipe wall may burn-through during welding. The most comprehensive approach is to carry out a conventional weld procedure development using test welds on a special flow loop or pipe by-pass. This provides a controllable segment of pipe under identical conditions to those in-service. However such facilities are not common, and the normal approach is to physically simulate the thermal characteristics of the operational pipeline by using water, air or oil flow through a section of test pipe. Following on from early work by British Gas, EWI developed an experimental method of determining the cooling capacity of a working pipeline(2). This is achieved by a simple procedure suitable for use in the field. An oxy-propane torch is used to heat a 50mm diameter area on the exposed pipe wall. Heating is stopped when the temperature reaches approximately 325C. Using a stopwatch and a thermometer, the temperature drop is monitored as the pipe wall cools. The time taken to cool between 250 and 100C is taken as the cooling capacity of the pipeline. This value is often referred to as the EWI cooling time, Tewi. The experiment is repeated on other spots upstream from the first, and a final cooling time is arrived at by averaging six readings. The philosophy behind the test is to measure a parameter which reflects the cooling capacity of the working pipeline, and to do this in a safe, experimentally simple and practical way. An experimental test bed can then be set up by duplicating the measured Tewi for this test pipe. To use this method for establishing a physical simulation it is not necessary to adhere to the EWI procedure. Simply duplicating the experimental procedure, heat source, spot size, temperature range etc, in the field and in the workshop should be sufficient. The general philosophy is satisfied provided the heat transfer behaviour of the gas or the test fluid is not significantly altered by the wall temperature achieved during welding as compared with that generated during EWI testing. Tests have been reported in which water, water-mist sprays, compressed air or oils are used to achieve the desired cooling capacity. EWIs work also generated much experimental data relating the weld heat input and cooling rate for inservice welds. The data were measured from flow loop tests and from test welds carried out under simulated conditions. The t8/5 weld cooling times were measured for heat inputs between 1 2 kJ/mm using pipe materials ranging in thickness from 4.8-8.0 mm and E6010, E7018 and E8018 electrodes. The cooling capacity was measured using the EWI process and t8/5 values were determined from temperatures measured by thermocouples harpooned into the weld pool. For a given pipe thickness and gas flow an approximately linear relationship between t8/5 and heat input were often found. In addition, EWI established that there was a proportional relationship linking the cooling capacity Tewi, and the t8/5 value obtained for a given heat input. Experimental data were used to form graphs such as Figure 3.10 for a 4.8 mm thick pipe. To use this graph to derive a suitable heat input requires a measurement of the cooling capacity of the operational pipeline. Take the value of 20 seconds for example. This represents a particular line between t8/5 and weld heat input on Figure 3.10. For a given CE, the IIW relationship between hardness and t8/5 can be used to estimate the minimum t8/5 cooling time required to give a hardness level of 350 HVN, (6 seconds in this example). The point on the relevant cooling capacity line at this t8/5 value gives the required minimum heat input (1.1 kJ/mm). The empirical data used to establish these lines are also shown in Figure 3.10, so it can be seen that the data points are relatively sparse and the extrapolations have generally been chosen conservatively. With this approach, the required heat input to achieve a desired hardness can be estimated directly from the measured heat capacity. To make use of the existing EWI data linking t8/5, and EWI cooling time it is necessary to have more concern about using the same procedure as EWI for determining the cooling capacity. The EWI test is not rigorously defined, and there is some concern about possible variability due to the changes in the size of the heated region and the heating rate used (see Appendix 1). EWI carried out some numerical analysis of the test process and concluded that the cooling capacity was not overly sensitive to the chosen spot size or heating rate(1). There are other difficulties here however, since the relationships between heat input and t8/5 are empirical and may vary with electrode type, welding position, joint type
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
21
etc. Similarly, the EWI approach makes no reference to fluid type, joint position, weld preparation, electrode type or preheat. It simply relies on the measured heat capacity and the developed empirical data set. It was pointed out by Cola & Threadgill(1) that the correlation between heat sink capacity and the cooling rate of the weld had only been established over a limited range of field conditions. Certainly the data are sparse and requires considerable extrapolation in some instances. Data are not available for pipes less than 4.0 mm thick. EWI was mainly concerned with determining a weld heat input that would produce a suitable HAZ hardness. The approach is conservative. Empirical relationships are chosen such that t8/5 values are likely to be under estimated, and heat inputs are likely to be over estimated, thus being sure to obtain hardness below the chosen limit. The critical level of hardness is chosen at 350 HVN, which is also considered as a conservative limit for low hydrogen welds(14). Both physical simulation and the EWI test have some problems. There is the additional cost of determining the cooling capacity of the pipeline, for although it is a simple process it may require excavation and incur site costs. To use the EWI methodology requires considerable empirical data to build up a database relating EWI cooling time, t8/5 cooling time and heat-input. The EWI methodology does not make any distinction between the saddle or sleeve configuration, and does not consider the effect of multi-pass welds or tempering effects. It also provides no direct information about burnthrough risk. Simulating the field cooling capacity directly and developing the welding procedure on a sample of the same pipe clearly overcomes the reliance on a database, but in some cases achieving the same cooling capacity can be difficult. Figure 3.11 illustrates the difficulties of selecting a fluid that is capable of duplicating the thermal characteristics of flowing high-pressure gas. In this figure the convective heat transfer coefficients for a selection of test fluids has been estimated using the DittusBohler equation (Holman(24)). The heat transfer coefficient for methane gas flow under typical operational conditions is also included. It seems reasonable to assume that a fluid that gives a similar heat transfer coefficient to that of methane would be suitable as a experimental substitute. Air clearly gives much lower values than methane at the same flow rate. Engine oil likewise gives lower convective coefficients. A low viscosity cooling oil does give heat transfer coefficient similar to that of methane, but it also requires a similar flow rate (1-10m/sec). Such high flow rates are not practical under laboratory testing. Water generally gives higher values and is clearly a more aggressive coolant than a gas flow. Values equivalent to gas flow could be achieved by using a low flow of water, <0.1 m/sec. However, this analysis does not include the boiling or vaporisation of water, which drastically increases heat extraction. The difficulty of generating cooling rates similar to gas flow is also illustrated in Figures 3.12 & 3.13. In these figures, data from EWI are replotted to illustrate how test pipes artificially cooled with water generally have more aggressive cooling than pipe cooled by gas flows. In these graphs, this effect is manifested as a reduced t8/5 cooling time for a given heat input when using water. Oil and air both give less severe cooling characteristics. A comparison between Figure 3.12 for 6.35 mm thick pipe, and Figure 3.13 for 4.8 mm thick pipe demonstrates that the differences between these fluids increases for thin walled pipe. Bruce et al(3) attempted to set up guidelines for the appropriate selection of model fluid, water, oil or air to be used in qualification procedures. Often the rationale is to accept the more aggressive coolant, normally water, since then the experimentally determined critical heat input value is conservative. That is on the operational pipeline it will produce longer cooling times and less HAZ hardness. It should be borne in mind however, that an over conservative heat input with respect to hardness could be non-conservative with respect to burn-through. Unless care is taken to match cooling capacities, Figures 3.12 & 3.13 indicate that this conservatism would increase as the pipe wall thickness is decreased, because there is a greater differential between the cooling effects of water and gas flow. This trend combined with the increasing sensitivity of thin walls to burn-through should be of concern.
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
22
The simulated flow is usually unpressurised, so these tests do not rigorously establish burn-through conditions. If burn-through limits are required, these are commonly determined with a static pressurised test using compressed inert gas. A test vessel is fabricated using a sample length of pipe. This should include an internal cylinder to reduce the volume of compressed gas and increase safety. The static gas has a lower cooling capacity than the gas flow in an operational pipeline so burnthrough limits established on such test are also conservative. That is, for a given heat input wall temperatures during the test are likely to be higher than those achieved on the working pipeline. Hence, under test conditions burn-through is likely to take place at a lower heat input than it would do in practice.
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
23
welding system to accurately control heat input. In addition, he speculated on the possibility or incorporating pre-heat or post heating within such a system.
3.7 References
1. Cola M. J. & Threadgill P. L., Final Report on Criteria for Hot Tap Welding, American Gas Association, Edison Welding Institute Project J7038, March 1988. 2. Bruce W. A. & Threadgill P. L., Welding Onto In-Service Pipelines, Welding Design & Fabrication Feb 1991, pp19-24. 3. Bruce W. A. & Threadgill P. L., Effect of Procedure Qualification Variables for Welding Onto Inservice Pipelines, American Gas Association Report J7141, July 1994. 4. Kiefner J.F & Fischer R. D., Models Aid Pipeline Repair Welding Procedure, Oil & Gas Journal March 1988, pp41-47. 5. Fischer R. D., Kiefner J. F. & Whitacre G. R., User Manual for Model1 & Model 2 Computer Programs for the Predicting Critical Cooling Rates and Temperatures During Repair and Hot Tap Welding on Pressurised Pipelines, Battelle Memorial Institute Report, June 1981. 6. Bruce W. A., Bubenik T. A., Fischer R. D. & Kiefner J.F., Development of Simplified Weld Cooling Rate Models For In-Service Gas Pipelines, Line Pipe Research Proceedings, 8th Symposium, September 1993, Paper 31, pub Arlington VA 22209, 1993, pp31.1-31.22.
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
24
7. Phelps B., Cassie B. A., & Evans N. H., Welding Onto Live Natural Gas Pipelines, Metal Construction, August 1976, pp350-354. 8. Wade J. B., Hot Tapping of Pipelines, Australian Welding Research Association Research Report, Snowy Mountains Corporation 1978. 9. Wade J.B., Effect of Preheat on Hot Tapping Procedures, Australian Welding Research Association Research Report, Snowy Mountains Corporation, September 1978. 10. Wade J.B., Description of Experimental Results on the Effects of Pipeline Damage on Performance and Hot Tapping Techniques, paper presented at Australian Welding Research Associations conference Pipeline Welding in 80s, Melbourne March 1981, Paper 4a. 11. Bruce W. A., Holdren R.L., Mohr W. C., Kiefner J.F. & Swatzel J.F., Repair of Pipelines by Weld Metal Deposition, Paper presented at PRCI 9th Symposium on Pipeline Research, Houston Texas, September 1996. 12. Bruce W. A., Holdren R.L. & Mohr W. C., Repair of Pipelines by Direct Deposition of Weld Metal Further Studies, Final report Edison Welding Institute, EWI Project J7283, November 1996. 13. Bailey N., Welding Procedures for Low Alloy Steels, The Welding Institute Cambridge England 1970. 14. Graville B. A. & Read J. A., Optimization of Fillet Weld Sizes, Welding Journal Research Supplement pp161s-167s. 15. Boran J The Hot-Tapping of Sub Sea Pipelines Welding Review, vol6, no 4 Nov 1987, pp283-284 16. DeHertogh J. & Illeghems H., Welding Natural Gas Filled Pipelines, Metal Construction & British Welding Journal, March 1972, pp224-227. 17. Cassie B. A., The Welding of Hot Tap Connections to High Pressure Gas Pipelines, paper presented Pipeline Industries Guild J. W. Jones Memorial Lecture, October 1974. 18. Bruce W. A., Welding Onto In-Service Pipelines: A Review, paper presented at Pipeline Welding 98, International Symposium on Pipeline Welding, May 1998. 19. Rietjens I. P., Safely Weld and Repair In-Service Pipelines, Pipeline Industry, December 1986, pp26-29. 20. Considerations of Welding Methods Adopted on Pipelines During Operation, IIW Document XI-E477-87, 1987. 21. Hicks D. J. Guideline for Welding on Pressurised pipe, Pipeline & Gas Journal, March 1983, pp17-19 22. Bout V.S. & Gretskii Yu.Ya., Arc Welding Application on Active Pipelines, Pipeline Technology, Volume 1, R. Denys Ed. R. Denys, pub Elsevier Science BV. pp550-558. 23. Australian Standard Pipelines Gas & Liquid Petroleum AS 2885 1987 Clause 7.13.11.
th 24. Holman J.P., 1976, Heat Transfer, 7 edition pub. McGraw-Hill.
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
25
HAZ Cracking
hardness > 350 susceptible microstructure
material composition
Figure 3.1 Factors influencing the possibility of hydrogen assisted cracking during in-service welding.
8 7 T85C Cooling Time (secs) 6 5 4 3 2 1 0 0 0.5 1 1.5 2 2.5 3 3.5 4 Gas Flow (mmscmd) 4.8 mm 6.4 mm 9.3 mm 15.1 mm
Figure 3.2 For a constant heat input of 0.9 kJ/mm this graph shows the reduction of t8/5 cooling time due to increased gas flow in pipes of different wall thickness(3).
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
26
10 8 9 6 2 7 3 4
2t
45 deg . t
5 1
Figure 3.3 Recommended weld bead deposition sequences in order to make most benefit of tempering, (a) general view of recommended joint configuration (b) buttering layers and a temperbead at weld toe (c) stringer bead arrangement.
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
27
Heat Input
Weld Penetration Pipe Wall Temperature
Local Pipe Wall Strength
Welding speed Electrode polarity Electrode type Electrode diameter Welder technique/direction Pipe wall thickness
Gas pressure Yield strength at temperature Weld orientation Local wall support
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
28
180
160 3.2 mm diameter electrode 140 burn-through safe < 0.87 kJ/mm 100 3 4 5 6 7
120
Figure 3.5 Burn through limits expressed as limits to the allowable welding current, for both 3.2 mm and 4 mm diameter electrodes from Cassie(17). Additional points from Bruce et al (11).
1.1
1 4 mm wall thickness 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 50 2.0 mm 60 70 80 2.4 mm 90 100 110 3.2 mm 3.2 mm wall thickness
Figure 3.6 Recommended limits for avoiding burn through during repair welding from Bruce et al(11)
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
29
Welding Direction
Critical Heat Input for Burn-through (kJ/mm) 0.37 0.48 0.48 0.51
14
Wade results, bulge height < 1.0 mm Wade results, bulge height > 1.0 mm
12
10
Figure 3.7 Burn-through limits established by Wade (8) for welding onto a 300 mm diameter, X60 steel pipe of 5 mm wall thickness.
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
30
Figure 3.8 Burn-through limits recommended by Wade(8) based on tests carried out on 300 mm diameter, X60 steel pipe, using test pipes pressurised with non-flowing nitrogen.
Gap ao
Band to support weld zone
fitting
Figure 3.9 Schematic illustration of the joint configuration proposed by Bout & Gretskii(22) to eliminate burn-through by supporting the pipe wall during in-service welding.
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
31
2 1.8 1.6 Heat Input (kJ/mm) 1.4 1.2 1 0.8 0.6 0.4 0.2 0 0 2
Tewi = 40 sec
10
Figure 3.10 A typical diagram derived by EWI(6) to enable the determination of required heat input from a measurement of the pipes heat capacity.
100000
methane 'real' pipe air flow low pressure w ater flow Mobiltherm 603 Engine Oil
10000
1000
100
10
Figure 3.11 The estimated effective heat transfer coefficient at the inside of the pipe wall for different fluids and flow conditions.
CRCWS Project 96:34, In-Service Welding on Gas Pipelines: Final Project Report
32
15
Water Motor Oil
10
7.5
2.5
Figure 3.12 The relationships between t8/5 cooling time and the weld heat input on 6.35 mm pipe containing different fluids, compared with points for gas flow. Data replotted from (3).
25
Water Motor Oil
15
10
Figure 3.13 The relationships between t8/5 cooling time and the weld heat input on 4.8 mm pipe containing different fluids compared with points for gas flow. Data replotted from (3).
100
References
1. Cola M.J. & Threadgill P.L., Final Report on Criteria for Hot Tap Welding, American Gas Association, Edison Welding Institute Project J7038, March 1988 2. Bruce W.A. & Threadgill P.L. Welding Onto In-Service Pipelines Welding Design & Fabrication Feb 1991, pp19-24 3. Bruce W.A. & Threadgill P.L., Effect of Procedure Qualification Variables for Welding Onto Inservice Pipelines, American Gas Association Report J7141, July 1994. 4. Kiefner J.F & Fischer R.D. Models Aid Pipeline Repair Welding Procedure Oil & Gas Journal March 1988, pp41-47 5. Fischer R.D., Kiefner J.F. & Whitacre G.R., User Manual for Model1 & Model 2 Computer Programs for the Predicting Critical Cooling Rates and Temperatures During Repair and Hot Tap Welding on Pressurised Pipelines, Battelle Memorial Institute Report, June 1981 6. Bruce W.A., Bubenik T.A., Fischer R.D. & Kiefner J.F., Development of Simplified Weld Cooling Rate Models For In-Service Gas pipelines 7. Phelps B., Cassie B.A., & Evans N.H., Welding Onto Live Natural Gas Pipelines, Metal Construction, August 1976, pp350-354 8. Wade J.B., Hot Tapping of Pipelines, Australian Welding Research Association Research Report, Snowy Mountains Corporation1978 9. Wade J.B., Effect of Preheat on Hot Tapping Procedures, Australian Welding Research Association Research Report, Snowy Mountains Corporation, September 1978 10. Wade J.B., Description of Experimental Results on the Effects of Pipeline Damage on Performance and Hot Tapping Techniques, paper presented at Australian Welding Research Associations conference Pipeline Welding in 80s, Melbourne March 1981, Paper 4a 11. Bruce W.A., Holdren R.L., Mohr W.C., Kiefner J.F. & Swatzel J.F., Repair of Pipelines by th Weld Metal Deposition, Paper presented at PRCI 9 Symposium on Pipeline Research, Houston Texas, September 1996. 12. Bruce W.A., Holdren R.L. & Mohr W.C., Repair of Pipelines by Direct Deposition of Weld Metal Further Studues, Final report Edison Welding Institute, EWI Project J7283, November 1996 13. Bailey N., Welding Procedures for Low Alloy Steels, The Welding Institute Cambridge England 1970 14. Graville B.A. & Read J.A., Optimization of Fillet Weld Sizes, Welding Journal Research Supplement pp161s-167s 15. Boran 16. DeHertogh J. & Illeghems H., Welding Natural Gas Filled Pipelines, Metal Construction & British Welding Journal, March 1972, pp 224-227
101
17. Cassie B.A., The Welding of Hot tap Connections to High Pressure Gas Pipelines, paper presented Pipeline Industries Guild J.W.Jones Memorial Lecture, October 1974. 18. Bruce W.A., Welding Onto In-Service Pipelines: A Review, paper presented at Pipeline Welding 98, International Symposium on Pipeline Welding, May 1998 19. Rietjens I.P., Safely weld and Repair In-Service pipelines, Pipeline Industry, December 1986, pp26-29 20. Considerations of Welding Methods Adopted on Pipelines During Operation, IIW Document XI-E-477-87, 1987 21. Hicks D.J. Guideline for Welding on Pressurised pipe, Pipeline & Gas Journal, March 1983, pp17-19 22. Bout V.S. & Gretskii Yu.Ya., Arc Welding Application on Active Pipelines, Pipeline Technology, Volume 1, R.Denys Ed. R.Denys, pub Elsevier Science BV. pp550-558 23. Australian Standard Pipelines Gas & Liquid Petroleum AS 2885 1987 Clause 7.13.11.
th 24. Holman J.P., 1976, Heat Transfer, 7 edition pub. McGraw-Hill