Banff Pressure Industry Conference February 7-9 Banff, Alberta

CHALLENGES OF P1 TO P8 WELDING IN SOUR ENVIRONMENTS

Rory Belanger, P.Eng. Ludwig & Associates Ltd. Frank Langenecker, C.E.T. Ludwig & Associates Ltd.

ABSTRACT The safe use of welded P1 materials (carbon steels) and P8 materials (austenitic stainless steels) in sour environments has been well established and accepted in the oil and gas industry. Generally, the preparation of these welding procedures follow the statutory requirements of ASME IX with supplemental requirements including hardness limits as stipulated in NACE MR0175. This standard provides recommendations for maximum hardness limits for the resulting weldment not greater than 22 HRC. Individually, P1 or P8 material weldments can usually be prepared with hardness levels meeting the NACE MR0175 criteria. However, the joining of P1 to P8 materials does pose problems associated with deposit chemistries due to dilution. These problems are normally mitigated through use of consumables with levels of chromium and nickel which will produce weld metal deposits with the alloy loading necessary to produce a bulk microstructure which is not susceptible to problems associated with transformations. The most common consumable type used in the preparation of these weldments is a type 309L. Unfortunately, the type 309L consumable when used to weld carbon to austenitic stainless steels will produce hard zones between the weld deposit and the carbon steel. These hard zones are essentially unavoidable and for the most part have been ignored by the oil and gas industry as these weldments seldom exhibit susceptibility to failure in sour environments. This paper discusses that hard zone susceptibility in sour environments through the case study review of a failure occurring with such a weldment. Also presented is a primer discussing the use of type 309L stainless steel consumables and the reasons why they have been used historically and metallurgically. Part 1 Procedure Qualification of P1 to P8 Weldments Welding of austenitic 300 series stainless steel grades to carbon or low alloy steels is a common oilfield practice. Successful application has been well documented in numerous services. Applications range from dissimilar metal connections in boilers

to stainless steel threaded couplings welded to carbon steel piping used for water disposal. The problems associated with making these welds are generally well known and steps are taken during material selection and fabrication to minimize service risk. In welding of austenitic stainless steels to carbon or low alloy steels it is customary to use a type 309L stainless-steel electrode. Type 309L stainless steel is sufficiently high in total alloy content to prevent martensite formation when diluted with carbon steel during welding, while at the same time, preserving residual amounts of ferrite to counteract the tendencies for hot cracking during welding. Several studies have been conducted to define the effect of combining alloy additions. One of the most recognized of these are the studies by Schaeffler in the 1960s. The Schaeffler diagram can be used to predict the likely outcome of weld metal constituents when welding stainless steels to carbon steels. For example; a common application calls for the joining of A312 TP 304L austenitic stainless steel pipe to A106 Gr. B carbon steel pipe. For exemplification purposes we have the choice of using either an E308L electrode or an E309L electrode. To determine the most suitable electrode for this application we consult the Schaeffler Diagram (Figure 1).

Nickel equivalent = %Ni+30x%C+0.5x%Mn

40 35 30 25 20 15 10 5 0 Martensite (M) M + F(a ) 0 5 M + F( d ) A+M Austenite (A) A + F( d )

5% 10% 20% 40%

Cr equivalent = 14.05 + 2.96 + 0.01 Cr equivalent = 17.1 For an E308L electrode weldment the nickel equivalent would be: Ni equivalent = % Ni + 30 x % C + 0.5 x % Mn 70% electrode chemistry = 0.7(9 + 30 x 0.04 + 0.5 x 2) 15% A312 TP304L pipe chemistry = 0.15(8 + 30 x 0.04 + 0.5 x 1) 15% A106 Gr. B pipe chemistry = 0.15(0.02 + 30 x 0.15 + 0.5 x 1) Ni equivalent = 0.7(11.2) + 0.15(9.7) + 0.15(4.6) Ni equivalent = 7.8 + 1.45 + 0.69 Ni equivalent = 9.9 For a E309L electrode the chromium equivalent would be: Cr equivalent = % Cr + % Mo + % 1.5 x % Si + 0.5 x % Cb 70% electrode chemistry = 0.7(24 + 0.7 + 1.5 x 0.7 + 0.5 x 0) 15% A312 TP304L pipe chemistry = 0.15(19 + 0.7 + 1.5 x 0 + 0.5 x 0) 15% A106 Gr. B pipe chemistry = 0.15(0.03 + 0.01 + 1.5 x 0.4 + 0.5 x 0) Cr equivalent = 0.7(25.12) + 0.15(19.7) + 0.15(0.64) Cr equivalent = 17.58 + 2.96 + 0.01 Cr equivalent = 20.6 For an E309L electrode weldment the nickel equivalent would be: Ni equivalent = % Ni + 30 x % C + 0.5 x % Mn 70% electrode chemistry = 0.7(12 + 30 x 0.04 + 0.5 x 2) 15% A312 TP304L pipe chemistry = 0.15(8 + 30 x 0.04 + 0.5 x 1) 15% A106 Gr. B pipe chemistry = 0.15(0.02 + 30 x 0.15 + 0.5 x 1) Ni equivalent = 0.7(14.2) + 0.15(9.7) + 0.15(4.6) Ni equivalent = 9.94 + 1.45 + 0.69 Ni equivalent = 12.1 The Cr equivalents and Ni equivalents for the weldmetal using a E308L electrode and E309L electrode are summarized in Table 2. Table 2 Chromium and Nickel Equivalents Electrode Cr Ni E308L 17.1 9.9 E308L 20.6 12.1 When the Cr and Ni equivalents are plotted on the Schaeffler diagram the weld deposit structure for the E308L electrode is austenite, martensite and 2% ferrite. The weld deposit structure for the E309L electrode is austenite and 7% delta ferrite. From this we can determine that the E309L electrode is suitable for the joining of austenitic to carbon steels where the E308L

80%

100% d -Ferrite (F )

d -Ferrite (Fd ) 40

10 15 20 25 30 35 Chromium equivalent = %Cr+%Mo+1.5x%Si+0.5x%Nb

Figure 1 Schaeffler Diagram This diagram has a chromium equivalent on the horizontal axis and a nickel equivalent on the vertical axis. To determine the weld metal constituents we calculate the chromium and nickel equivalents expected for the welded joint. The weld will encompass a mixture of weld electrode and base metal. For the SMAW process an approximate 30% dilution is expected. Therefore the resultant weld metal would be 70% electrode chemistry with 15% A312 TP304L pipe chemistry and 15% A106 Gr. B pipe chemistry (Figure 2.) Chemical constituents relevant to the Schaeffler diagram are given for the pipe and welding electrodes in Table 1.
70% ER309L Deposit 15% Dilution 15% Dilution

SA312 TP304L

SA106 Gr. B

Figure 2 Dilution Effect Table 1 Electrode and Pipe Chemistry Material Cr Ni C Mo Mn *E309L 24 12 0.04 0.7 3 *E308L 19 9 0.04 0.7 2 **TP 304L 19 8 0.04 1 **A106- B 0.03 0.02 0.15 0.01 1 * Electrode ** Pipe material

Si 0.7 0.7 0.4

For an E308L electrode the chromium equivalent would be: Cr equivalent = % Cr + % Mo + % 1.5 x % Si + 0.5 x % Cb 70% electrode chemistry = 0.7(19 + 0.7 + 1.5 x 0.7 + 0.5 x 0) 15% A312 TP304L pipe chemistry = 0.15(19 + 0.7 + 1.5 x 0 + 0.5 x 0) 15% A106 Gr. B pipe chemistry = 0.15(0.03 + 0.01 + 1.5 x 0.4 + 0.5 x 0) Cr equivalent = 0.7(20.07) + 0.15(19.7) + 0.15(0.64)

electrode weld deposit, with a martensitic constituent and low delta ferrite, would exhibit a tendency to crack. When welding austenitic stainless steels to carbon or low alloy steels for sour service applications NACE Standard MR0175 "Sulfide Stress Cracking resistant Metallic Materials for Oilfield Equipment" is generally referred to as a guideline. Although this standard does not specifically refer to carbon steel to austenitic stainless steel weldments, the conclusion is often drawn that since both materials individually are acceptable up to a hardness of 22 HRC, welding of them together under the same acceptance criteria will be acceptable. For welding procedure qualification a cross-section of the weld is taken for hardness measurement. NACE Standard MR0175 does not specify the hardness test method or the testing rigor. The standard states only that the weld, heat affected zone, and parent material will meet the hardness criteria. In this case it is 22 HRC ( Hardness Rockwell, C Scale). Industry has attempted to clarify the hardness test requirement for equipment in severe sour environments by specifying test locations and test methods. An example of a typical Microvickers hardness test survey for welds in sour service is shown in Figure 3. The locations and rigor of testing is clearly shown, giving global weldment hardness suitable for most applications. However, this is where we get into difficulty when applying a comprehensive general specification for determining suitability of dissimilar metal weld for severe sour service application.
.1 2 11 .18 19 . . . . . . 4 . . . 14 . . . . 7 . . . . . . . . 9 17 24 25. 10.

8,000 kPa and was operating at 6,200 kPa at the time of the failure. The exchanger was newly fabricated and operated for about 12 weeks when a crack propagated through the P1 to P8 weldment between the tubesheet and shell and extended about approximately 75% of the circumference. Although operating personnel were in close enough proximity to have witnessed the decompression of the vessel, no personnel injury occurred. VESSEL CONSTRUCTION The gas-gas heat exchanger comprised a 25 mm w.t. x 813 mm OD carbon steel shell which was welded to a 203 mm w.t. stainless steel tubesheet. The shell was manufactured from SA 516 Gr. 70 plate while the tubesheet was SA 240 TP 316L stainless steel plate. The weldment between the shell and the tubesheet was a full penetration groove weld, which was self-backed by virtue of the joint design. The following sketch illustrates the details of the shell to tubesheet construction:
Shell Side Inlet

Inspection Nozzle

Figure 3 - Typical Hardness Survey Typical hardness indentation sizes for Rockwell and Microvickers test methods are shown in Photo 1. From this we can readily see that the sensitivity of the test is dependent on the test method. Testing with Rockwell found a hardness of 86 HRB implying the weldment hardness meets NACE requirements. Testing a very specific location in the Intermediate Mixed Zone found a hardness of 355 HV300 clearly not meeting the NACE requirements of 22 HRC. Part 2 Failure Case Study of a P1 to P8 Weldment Conducted on a Heat Exchanger Between a Carbon Steel Shell and a Type 316L Stainless Steel Tubesheet BACKGROUND The failed component was a gas-gas heat exchanger. The tube and shell sides of the vessel were exposed to hydrocarbons containing approximately 10 % wet hydrogen sulphide at about 65 C. The exchanger had a maximum operating pressure of Figure 4 Exchanger Construction At the 6:00 (o'clock) position, a 141 mm OD x 9.5 mm w.t. nozzle had been welded approximately 33 mm from the tube to tubesheet weldment centerline. The decision was made not to post weld heat treat the heat exchanger due to anticipated problems with differential expansion between the carbon steel shell and the stainless steel heat exchanger tubes. Also considered potentially problematic was the possibility of sensitization of the very thick Type 316L tubesheet. WELDING The shell to tubesheet weldment was performed with a 60 included angle and all welding was conducted with type 309L stainless steel consumables. The root bead was conducted with the gas tungsten arc welding process (GTAW) while the fill and cap was conducted with the shield metal arc welding process

(SMAW). Welding was performed in stringer fashion and completion of the weldment took approximately eight layers. Welding was conducted with a 38C preheat and much of the welding was conducted in position. As previously mentioned, a 6:00 position nozzle immediately adjacent to the tubesheet had been inserted through the shell and welded with a full penetration weldment which was approximately 2/3 biased from the shell ID. Weld qualification comprised a standard ASME IX PQR conducted on 12.7 mm plate material. The material used in the procedure qualification involved the same materials (SA 516 Gr. 70 and type 316L stainless steel) as what to be used in the shell to tubesheet weldment. Because the weldment was to be exposed to sour environments, the PQR was supported by a 25 point 500g Vickers microhardness traverse with limits on hardness being 248 HV500. FAILURE ANALYSIS Visual Examination Visually apparent damage was only observed on one end of the heat exchanger (Photo 2). Cracking was observed to extend along the shell to tubesheet weldment for approximately 270 (Photo 3). The cracking had progressed along the shell side edge of the weld cap and approximately 3 mm of crack displacement was observed between the shell and the weld metal (Photo 4). Examination of the weld cap revealed no obvious weld deficiencies. Eyewitnesses reported the location of the initial breach of the exchanger shell as being near the 6:00 o'clock position. Fracture initiation, through full fractographic analysis of the complete fracture surface, could not be performed as sampling was limited in anticipation of vessel repair. However, it was reasonable to assume that cracking did propagate from the 6:00 position as approximately equal lengths of cracking existed on both sides of the 6:00 position. Material provided for metallurgical examination included a 225 mm x 275 mm section of the shell plate material, which included the nozzle illustrated in Figure 4. The sample provided for examination had been flame cut from the exchanger. The fracture surface of the failed shell to tube weldment was covered with a tightly adherent black corrosion product, while the ID surfaces were covered with a black and rust colored flaking scale (Photo 5). The weld fracture clearly propagated along the shell side fusion line. The shape of each weld pass was well defined on the fracture surface. The weld had been conducted in eight layers and the layers appeared relatively uniform from root to cap. Little variation was observed on the weld fracture surface (Photo 6).

Macroscopic examination was conducted on a slice of the weldment sectioned longitudinally with respect to the vessel axis from the approximate 6:00 position. The heat affect zone (HAZ) adjacent to the fracture was relatively uniform in depth from root to cap (Photo 7). with the last pass of the shell to tubesheet weldment exhibited a slightly darker etching HAZ. The weldment exhibited acceptable profile with the only real weld flaws being one example of porosity and one of entrapped slag. Microscopic Examination Samples of the weld profile were sectioned about the 6:00 o'clock position on the weldment sample. Fracture profile extended from root to cap immediately parallel to the ferritic shell fusion boundary. In all cases, the fracture extended along the Intermediate Mixed Zone (IMZ) between the austenitic filler material and the ferritic shell material (Photo 8). This fusion zone was found to have a width, which ranged from 0.03 to 0.05 mm (Photo 9). Examination of this zone revealed a martensitic/austenitic microstructure when examined at high magnifications (Photo 10). Fractographic examination of the fracture surfaces with SEM revealed both intergranular and transgranular morphologies (Photo 11). Brittle fracture features dominated the fusion zone area of fracture. One small area was observed near the fracture OD where the fracture propagated through weld metal. The fracture features in this area exhibited ductile fracture features (Photo 12). Mechanical Testing Hardness testing conducted on the weldment cross sections revealed for the most part what was considered to be normal hardnesses in the parent metals and heat affected zones. Hardness results from within the weld metal also revealed what was considered to be acceptable hardness levels. However, higher hardness was detected in the IMZ. The following table illustrates the results of hardness testing: Table 3 Hardness Test Results Location Hardness HV300 Parent Shell 159 167 HAZ Shell 207 - 248 IMZ 289 357 Weld Metal 208 - 223 Chemical Analysis Energy dispersive X-ray analysis (EDX) was conducted on a line across the IMZ of the weldment. The following graph plots the alloy composition across the IMZ:

Alloy Composition Across IMZ

25 Alloy Content (% by Mass) Mn Si Cr Ni C Mo 20 15 10 5 0 -15 -10 -5 0 5 10 15

Distance from IMZ Center (microns)

Figure 5 Alloy Composition Across IMZ

Chemical spot testing was conducted on surfaces of the weldment fracture as well as the ID surface of the exchanger shell. The following are the test results recorded:

Figure 6 FEA Model of the Tubesheet to Shell Weldment

Table 4 Chemical Spot Testing RESULTS ION Shell Fracture Sulphides Positive Positive Chlorides Negative Negative Carbonates Negative Negative Figure 7 Stress Intensification at the IMZ Root These high residual stresses were intensified still further with the locating of the 6:00 position nozzle at such a close proximity to the shell to tubesheet weldment as illustrated in Figure 4. The available corrodent to complete the necessary SSC variables was hydrogen sulphide, which was in abundance as evidenced by chemical spot testing. Hydrogen, evolved from the corrosion reaction, in its atomic form, will diffuse into the shell and weld metal and promote cracking in susceptible areas. Further aggravating the problem is residual hydrogen from the original welding of the tubesheet to shell. All of these contributory factors were necessary to have caused the failure and removing any one or more of these may have prevented the failure from occurring. The cause of the formation of the susceptible hard zone was related to the local chemistry. The composition of the type 309L weld metal has been formulated to overload the deposit metal chemistry to compensate for inevitable dilution. As discussed previously, the type 309L deposited metal chemistry will generally produce a microstructure which is metallurgically acceptable. However in the IMZ, a composition gradient will exist. The alloy compositions within the IMZ of the failed P1

DISCUSSION The path of the cracking was found to be along a zone of martensitic transformation products, which existed between the austenitic filler material and the ferritic shell parent material. Based on the fracture morphology and the corrodents available, this specific cracking mechanism was determined to be sulphide stress cracking. Any stress corrosion cracking mechanism requires the presence of three complementing variables. These include a susceptible material, a corrodent, and a sustained tensile stress. In the case of the weldment between the shell and tubesheet, the susceptible material came in the form of a hard martensitic or semi martensitic IMZ between the austenitic filler material and the ferritic shell material. The stresses applied to the weldment included working stresses and those residual from the welding process. Compounding the problem and considered highly contributory was the joint design. The vessel shell was relatively thin and was attached with a full penetration, single sided, and self backed weldment to a very thick tubesheet. These features resulted in a large degree of stress intensification being applied to the weldment at the very sensitive location of the P1 to P8 IMZ root. The following sketch and graphical plot present a finite element analysis (FEA) stress distribution to qualitatively illustrate this intensification at the location from which cracking extended.

to P8 weldment varied significantly as shown in Plot 1. Making some assumptions for the carbon content, the nickel and chromium equivalents were determined and are plotted on the following condensed portion of the Schaeffler diagram.
20 Nickel equivalent 15 A + F(d ) 10 5 0 Martensite (M) M + F(a ) 0 5
ted P Predic hase

A+M

Austenite (A)

that acceptable hardnesses can be achieved with an Inconel 182 consumable. Probably the most appropriate modification, in the case of the shell to tubesheet weldment, would be to change the material used in the shell to an austenitic stainless steel or to change the tubesheet to a carbon steel. The latter of these two options was used in the final repair. REFERENCES: 1. Doody, Thomas, Intermediate Mixed Zones in Dissimilar Metal Welds for Sour Service, Welding Journal, March 1992.

M + F( d ) 10 15 20 Chromium equivalent 25

2.

Figure 8 Condensed Portion of Schaeffler Diagram 3. Its clear from the preceding diagram that, although the widths of the IMZ will vary somewhat, the alloy gradient, which will always exist between austenitic stainless steel consumables and ferritic parent metals, will inevitably produce martensitic microstructures. Lowering the IMZ martensitic hardness could be accomplished with PWHT and this would lower the IMZ susceptibility to SSC. However, reducing the hardness levels to an acceptable range with PWHT would be difficult due to the varied and high alloy compositions of the transformation products. Much of the IMZ will be in the chemical composition of essentially martensitic stainless steel. Martensitic stainless steels require relatively high PWHT temperatures which would be greater than typical PWHT temperatures of carbon steels. The review of the history of this failed vessel, does indicate that an IMZ susceptibility to sulphide stress cracking does exist. Although the presence of this high IMZ hardness is well known, for the most part it is largely ignored either because laboratory hardness testing fails to detect elevated hardness levels (due to the narrow width of the zone) or because of the successful sour performance of P1 to P8 weldments conducted with type 309L consumables in the past. Considering the many thousands of these weldments currently in sour service, the reported numbers of failures must be considered as being minor. What this failure clearly does present however, is that the susceptibility will manifest itself in failure if high levels of stress intensification is forced on the IMZ. Avoidance of such failures would involve removing one of the three contributing factors to a corrosion cracking mechanism. In the case of the shell to tubesheet weldment, removal of stress through post weld heat treat probably would not be a viable option due to the extreme thermo mass of the tubesheet versus the shell. However, assuming that the geometry could not be acceptably altered, the possibility of using nickel based consumables may prove to be acceptable alternative to the use of a type 309L consumable. Laboratory testing has revealed

Omar, A. A., Effects of Welding Parameters on Hard Zone Formation at Dissimilar Welds, Welding Journal, February 1998. Schaeffler, A. L. Selection of Austenitic Electrodes For Welding Dissimilar Metals, Welding Journal, 26(10): 601s to 620s.

Limit of IMZ
Photo 1 Rockwell B hardness impression and microvickers 300g load hardness impression. The Rockwell indent straddles the IMZ. The Microvickers 300g load indent is fully contained in the IMZ. Hardness 86 HRB, 355 HV300 Mag. 80X 2% Nital Etch

Photo 2 Damaged end of heat exchanger.

Photo 4 Cracking extending along the shell side of the tubesheet weldment.

Photo 3 Location of the shell to tubesheet weldment. crack extent illustrated with arrows.

Photo 5 Lower 6:00 inspection nozzle containing the failed weldment removed for laboratory examination.

Photo 6 Lower 6:00 inspection nozzle containing the failed weldment removed for laboratory examination.

IMZ Fracture
Photo 8 Crack propagation along intermediate mixed zone (IMZ) between the ferritic shell and the austenitic weld metal. Mag 100X 2% Nital Etch

Limit of IMZ

Photo 9 IMZ between the ferritic and the austenitic weld metal. Mag 320X 2% Nital Etch

Photo 7 Shell and nozzle side of the tubesheet weldment.

Photo 10 IMZ etched to reveal its martensitic structure. Mag 1700X Marbles Reagent

Photo 12 Ductile fracture surface cracking through austenitic weld metal. Mag 486X SEM Image

Photo 11 Fracture surface of the IMZ cracking exhibiting a mixture of inter and transgranular features. Mag 320X SEM Image