ASM Online Member Community

A 4-inch diameter schedule 40 pipe welded to a bolted flange connection failed in service at the weld. Design limits are specified as 185 PSIG, 375°F. There is no operational data to suggest these parameters were exceeded. Visual inspection showed the crack was at the flange-side toe of the weld, connected there both on the ID and OD, but appeared to be ID-initiated given the visible portion of the ID crack was longer than the OD portion (2-1/4" long on ID vs 1-1/2" long on ID). The materials, as confirmed by handheld XRF, were 316SS for the pipe, 304SS for the flange, and 308SS for the filler. The filler was determined to have 7.5-10% ferrite via a severn gauge. Weld was open-root, multiple pass, and appeared to be TIG based on cap profile (I do not have the weld procedure info).

The crack was broken open in the laboratory. The crack path was confirmed to traverse from toe-to-toe of the ID to OD weld. There was a distinct lip or ledge near the OD where the crack direction abruptly changed direction. The fracture surface exhibited several areas of distinct heat-tint. The shape of all of these tinted regions was more consistent with OD-initiated cracking (somewhat similar in shape/orientation progression marks emanating from the OD toward the ID). The shape of this heat-tint was especially pronounced on the OD ledge.

SEM evaluation showed what I perceived to be transgranular cleavage fracture. There was no overload observed (at least compared to the lab-generated fracture surface, which was a readily identified dimpled surface). Despite most of the fracture surface being transgranular, there were interspersed facets across the fracture that look to me like isolated intergranular fracture faces. SEM-EDS of these faces showed traces of Na and Cl which were present in discrete particles on the surface. There was no difference in the SEM between the fracture surfaces of the hint-tinted areas compared to the non-heat tinted areas (same general concentration of facets, same fracture morphologies, etc). 

Metallographic evaluation showed that, across 4 cross-sections taken, there was no preference for crack propagation with respect to microstructure. Additionally, in all 4 cross-sections, cracks were observed on both sides of the ID weld toe. The cracks always initiated at the toe of the weld on the ID, but did not travel along the fusion line or HAZ. The crack was confirmed to be transgranular in the base material, and did not exhibit any preferential microstructural path when traversing the weld metal. Crack paths were slightly branched, but not extensively. One notable feature from the microstructural evaluation is that there appeared to be some weld repair performed from the ID, evidenced by the shape of the fusion zone. 

I'm having trouble piecing this all together into a coherent picture. The heat tint on the fracture surfaces suggests that the crack at least partially formed during fabrication, but the fracture features between the heat-tinted areas and all other areas are nearly identical. The cracks initiating at the toe of the weld makes sense from a stress-concentration standpoint, probably more-so for SCC than fatigue given they were on the pipe ID and were present on both sides of the weld toe. SCC is further supported by the presence of chlorine...but why don't the cracks show more preference for a certain microstructure? And if some portion of the crack occurred during fabrication, why is the crack path so continuous/smooth (at least until it connects to the OD-initiated "ledge" mentioned earlier)? 

This image dump is not very well organized and I realize it's also captionless, but I'm hoping it's easily correlated enough to what I've typed above and at least a starting point. What other features do I need to look for to properly characterize this failure?

Thank you,

Evan,  I would recommend taking a closer look at the HAZ.  Your fig 3 suggests that the crack tracked grain boundary regions, which would be consistent with sensitization.  Sensitization fractures are common in welded 3xx stainless steels.  The chrome carbides tend to form along grain boundaries and can lower the fracture toughness along the weld zone.  What was the service environment?  If there was any galvanic coupling, that could cause SCC and/or localized hydrogen embrittlement.    Depending on the welding conditions and the post weld heat treatment, the carbides could be quite small.  Examination of the GB regions at high mag might help identify the issue. 

A 4-inch diameter schedule 40 pipe welded to a bolted flange connection failed in service at the weld. Design limits are specified as 185 PSIG, 375°F. There is no operational data to suggest these parameters were exceeded. Visual inspection showed the crack was at the flange-side toe of the weld, connected there both on the ID and OD, but appeared to be ID-initiated given the visible portion of the ID crack was longer than the OD portion (2-1/4" long on ID vs 1-1/2" long on ID). The materials, as confirmed by handheld XRF, were 316SS for the pipe, 304SS for the flange, and 308SS for the filler. The filler was determined to have 7.5-10% ferrite via a severn gauge. Weld was open-root, multiple pass, and appeared to be TIG based on cap profile (I do not have the weld procedure info).

The crack was broken open in the laboratory. The crack path was confirmed to traverse from toe-to-toe of the ID to OD weld. There was a distinct lip or ledge near the OD where the crack direction abruptly changed direction. The fracture surface exhibited several areas of distinct heat-tint. The shape of all of these tinted regions was more consistent with OD-initiated cracking (somewhat similar in shape/orientation progression marks emanating from the OD toward the ID). The shape of this heat-tint was especially pronounced on the OD ledge.

SEM evaluation showed what I perceived to be transgranular cleavage fracture. There was no overload observed (at least compared to the lab-generated fracture surface, which was a readily identified dimpled surface). Despite most of the fracture surface being transgranular, there were interspersed facets across the fracture that look to me like isolated intergranular fracture faces. SEM-EDS of these faces showed traces of Na and Cl which were present in discrete particles on the surface. There was no difference in the SEM between the fracture surfaces of the hint-tinted areas compared to the non-heat tinted areas (same general concentration of facets, same fracture morphologies, etc). 

Metallographic evaluation showed that, across 4 cross-sections taken, there was no preference for crack propagation with respect to microstructure. Additionally, in all 4 cross-sections, cracks were observed on both sides of the ID weld toe. The cracks always initiated at the toe of the weld on the ID, but did not travel along the fusion line or HAZ. The crack was confirmed to be transgranular in the base material, and did not exhibit any preferential microstructural path when traversing the weld metal. Crack paths were slightly branched, but not extensively. One notable feature from the microstructural evaluation is that there appeared to be some weld repair performed from the ID, evidenced by the shape of the fusion zone. 

I'm having trouble piecing this all together into a coherent picture. The heat tint on the fracture surfaces suggests that the crack at least partially formed during fabrication, but the fracture features between the heat-tinted areas and all other areas are nearly identical. The cracks initiating at the toe of the weld makes sense from a stress-concentration standpoint, probably more-so for SCC than fatigue given they were on the pipe ID and were present on both sides of the weld toe. SCC is further supported by the presence of chlorine...but why don't the cracks show more preference for a certain microstructure? And if some portion of the crack occurred during fabrication, why is the crack path so continuous/smooth (at least until it connects to the OD-initiated "ledge" mentioned earlier)? 

This image dump is not very well organized and I realize it's also captionless, but I'm hoping it's easily correlated enough to what I've typed above and at least a starting point. What other features do I need to look for to properly characterize this failure?

Thank you,

Thank you Mark, I appreciate the reply.

This pipe is part of the condensate drain system for an auxiliary steam reboiler. The valve upstream of this piping opens intermittently to allow condensate/steam into the condenser pit of a nuclear power plant. I do not not have good enough drawings or photos to know if there's a possibility of galvanic coupling. I'll have to research hydrogen embrittlement, that has not been on my radar yet. There have been two similar failures of this piping in the past several years, but neither of those were submitted for metallurgical analysis. 

I didn't do a good enough job of labeling my images in my original post. You mentioned figure 3...was that the image with the two micrographs? For what it's worth, all of the fractures that I observed optically were transgranular, both in the base material and in the weld metal, which also matched what I observed in the SEM. I etched all micros with 60% nitric acid in water, 1.0V for ~100 seconds depending on sample size. I've attached a couple additional micrographs from another metallurgical mount. One micrograph shows a crack tip as it progresses through the weld metal and one showing a secondary crack parallel to the primary fracture surface.

I did not try to resolve any carbides yet, and I'm not as familiar with the etching procedure to do so...looks like electrolytic 10% ammonium persulfate may be appropriate? I will try the SEM first. I'm guessing they should be readily identifiable with BSE imaging? What size range would you expect the carbides to be (I have not researched this on my own yet, just curious on your thoughts there)?

Evan,  I would recommend taking a closer look at the HAZ.  Your fig 3 suggests that the crack tracked grain boundary regions, which would be consistent with sensitization.  Sensitization fractures are common in welded 3xx stainless steels.  The chrome carbides tend to form along grain boundaries and can lower the fracture toughness along the weld zone.  What was the service environment?  If there was any galvanic coupling, that could cause SCC and/or localized hydrogen embrittlement.    Depending on the welding conditions and the post weld heat treatment, the carbides could be quite small.  Examination of the GB regions at high mag might help identify the issue. 

A 4-inch diameter schedule 40 pipe welded to a bolted flange connection failed in service at the weld. Design limits are specified as 185 PSIG, 375°F. There is no operational data to suggest these parameters were exceeded. Visual inspection showed the crack was at the flange-side toe of the weld, connected there both on the ID and OD, but appeared to be ID-initiated given the visible portion of the ID crack was longer than the OD portion (2-1/4" long on ID vs 1-1/2" long on ID). The materials, as confirmed by handheld XRF, were 316SS for the pipe, 304SS for the flange, and 308SS for the filler. The filler was determined to have 7.5-10% ferrite via a severn gauge. Weld was open-root, multiple pass, and appeared to be TIG based on cap profile (I do not have the weld procedure info).

The crack was broken open in the laboratory. The crack path was confirmed to traverse from toe-to-toe of the ID to OD weld. There was a distinct lip or ledge near the OD where the crack direction abruptly changed direction. The fracture surface exhibited several areas of distinct heat-tint. The shape of all of these tinted regions was more consistent with OD-initiated cracking (somewhat similar in shape/orientation progression marks emanating from the OD toward the ID). The shape of this heat-tint was especially pronounced on the OD ledge.

SEM evaluation showed what I perceived to be transgranular cleavage fracture. There was no overload observed (at least compared to the lab-generated fracture surface, which was a readily identified dimpled surface). Despite most of the fracture surface being transgranular, there were interspersed facets across the fracture that look to me like isolated intergranular fracture faces. SEM-EDS of these faces showed traces of Na and Cl which were present in discrete particles on the surface. There was no difference in the SEM between the fracture surfaces of the hint-tinted areas compared to the non-heat tinted areas (same general concentration of facets, same fracture morphologies, etc). 

Metallographic evaluation showed that, across 4 cross-sections taken, there was no preference for crack propagation with respect to microstructure. Additionally, in all 4 cross-sections, cracks were observed on both sides of the ID weld toe. The cracks always initiated at the toe of the weld on the ID, but did not travel along the fusion line or HAZ. The crack was confirmed to be transgranular in the base material, and did not exhibit any preferential microstructural path when traversing the weld metal. Crack paths were slightly branched, but not extensively. One notable feature from the microstructural evaluation is that there appeared to be some weld repair performed from the ID, evidenced by the shape of the fusion zone. 

I'm having trouble piecing this all together into a coherent picture. The heat tint on the fracture surfaces suggests that the crack at least partially formed during fabrication, but the fracture features between the heat-tinted areas and all other areas are nearly identical. The cracks initiating at the toe of the weld makes sense from a stress-concentration standpoint, probably more-so for SCC than fatigue given they were on the pipe ID and were present on both sides of the weld toe. SCC is further supported by the presence of chlorine...but why don't the cracks show more preference for a certain microstructure? And if some portion of the crack occurred during fabrication, why is the crack path so continuous/smooth (at least until it connects to the OD-initiated "ledge" mentioned earlier)? 

This image dump is not very well organized and I realize it's also captionless, but I'm hoping it's easily correlated enough to what I've typed above and at least a starting point. What other features do I need to look for to properly characterize this failure?

Thank you,

Evan,  I would recommend taking a closer look at the HAZ.  Your fig 3 suggests that the crack tracked grain boundary regions, which would be consistent with sensitization.  Sensitization fractures are common in welded 3xx stainless steels.  The chrome carbides tend to form along grain boundaries and can lower the fracture toughness along the weld zone.  What was the service environment?  If there was any galvanic coupling, that could cause SCC and/or localized hydrogen embrittlement.    Depending on the welding conditions and the post weld heat treatment, the carbides could be quite small.  Examination of the GB regions at high mag might help identify the issue. 

A 4-inch diameter schedule 40 pipe welded to a bolted flange connection failed in service at the weld. Design limits are specified as 185 PSIG, 375°F. There is no operational data to suggest these parameters were exceeded. Visual inspection showed the crack was at the flange-side toe of the weld, connected there both on the ID and OD, but appeared to be ID-initiated given the visible portion of the ID crack was longer than the OD portion (2-1/4" long on ID vs 1-1/2" long on ID). The materials, as confirmed by handheld XRF, were 316SS for the pipe, 304SS for the flange, and 308SS for the filler. The filler was determined to have 7.5-10% ferrite via a severn gauge. Weld was open-root, multiple pass, and appeared to be TIG based on cap profile (I do not have the weld procedure info).

The crack was broken open in the laboratory. The crack path was confirmed to traverse from toe-to-toe of the ID to OD weld. There was a distinct lip or ledge near the OD where the crack direction abruptly changed direction. The fracture surface exhibited several areas of distinct heat-tint. The shape of all of these tinted regions was more consistent with OD-initiated cracking (somewhat similar in shape/orientation progression marks emanating from the OD toward the ID). The shape of this heat-tint was especially pronounced on the OD ledge.

SEM evaluation showed what I perceived to be transgranular cleavage fracture. There was no overload observed (at least compared to the lab-generated fracture surface, which was a readily identified dimpled surface). Despite most of the fracture surface being transgranular, there were interspersed facets across the fracture that look to me like isolated intergranular fracture faces. SEM-EDS of these faces showed traces of Na and Cl which were present in discrete particles on the surface. There was no difference in the SEM between the fracture surfaces of the hint-tinted areas compared to the non-heat tinted areas (same general concentration of facets, same fracture morphologies, etc). 

Metallographic evaluation showed that, across 4 cross-sections taken, there was no preference for crack propagation with respect to microstructure. Additionally, in all 4 cross-sections, cracks were observed on both sides of the ID weld toe. The cracks always initiated at the toe of the weld on the ID, but did not travel along the fusion line or HAZ. The crack was confirmed to be transgranular in the base material, and did not exhibit any preferential microstructural path when traversing the weld metal. Crack paths were slightly branched, but not extensively. One notable feature from the microstructural evaluation is that there appeared to be some weld repair performed from the ID, evidenced by the shape of the fusion zone. 

I'm having trouble piecing this all together into a coherent picture. The heat tint on the fracture surfaces suggests that the crack at least partially formed during fabrication, but the fracture features between the heat-tinted areas and all other areas are nearly identical. The cracks initiating at the toe of the weld makes sense from a stress-concentration standpoint, probably more-so for SCC than fatigue given they were on the pipe ID and were present on both sides of the weld toe. SCC is further supported by the presence of chlorine...but why don't the cracks show more preference for a certain microstructure? And if some portion of the crack occurred during fabrication, why is the crack path so continuous/smooth (at least until it connects to the OD-initiated "ledge" mentioned earlier)? 

This image dump is not very well organized and I realize it's also captionless, but I'm hoping it's easily correlated enough to what I've typed above and at least a starting point. What other features do I need to look for to properly characterize this failure?

Thank you,

A 4-inch diameter schedule 40 pipe welded to a bolted flange connection failed in service at the weld. Design limits are specified as 185 PSIG, 375°F. There is no operational data to suggest these parameters were exceeded. Visual inspection showed the crack was at the flange-side toe of the weld, connected there both on the ID and OD, but appeared to be ID-initiated given the visible portion of the ID crack was longer than the OD portion (2-1/4" long on ID vs 1-1/2" long on ID). The materials, as confirmed by handheld XRF, were 316SS for the pipe, 304SS for the flange, and 308SS for the filler. The filler was determined to have 7.5-10% ferrite via a severn gauge. Weld was open-root, multiple pass, and appeared to be TIG based on cap profile (I do not have the weld procedure info).

The crack was broken open in the laboratory. The crack path was confirmed to traverse from toe-to-toe of the ID to OD weld. There was a distinct lip or ledge near the OD where the crack direction abruptly changed direction. The fracture surface exhibited several areas of distinct heat-tint. The shape of all of these tinted regions was more consistent with OD-initiated cracking (somewhat similar in shape/orientation progression marks emanating from the OD toward the ID). The shape of this heat-tint was especially pronounced on the OD ledge.

SEM evaluation showed what I perceived to be transgranular cleavage fracture. There was no overload observed (at least compared to the lab-generated fracture surface, which was a readily identified dimpled surface). Despite most of the fracture surface being transgranular, there were interspersed facets across the fracture that look to me like isolated intergranular fracture faces. SEM-EDS of these faces showed traces of Na and Cl which were present in discrete particles on the surface. There was no difference in the SEM between the fracture surfaces of the hint-tinted areas compared to the non-heat tinted areas (same general concentration of facets, same fracture morphologies, etc). 

Metallographic evaluation showed that, across 4 cross-sections taken, there was no preference for crack propagation with respect to microstructure. Additionally, in all 4 cross-sections, cracks were observed on both sides of the ID weld toe. The cracks always initiated at the toe of the weld on the ID, but did not travel along the fusion line or HAZ. The crack was confirmed to be transgranular in the base material, and did not exhibit any preferential microstructural path when traversing the weld metal. Crack paths were slightly branched, but not extensively. One notable feature from the microstructural evaluation is that there appeared to be some weld repair performed from the ID, evidenced by the shape of the fusion zone. 

I'm having trouble piecing this all together into a coherent picture. The heat tint on the fracture surfaces suggests that the crack at least partially formed during fabrication, but the fracture features between the heat-tinted areas and all other areas are nearly identical. The cracks initiating at the toe of the weld makes sense from a stress-concentration standpoint, probably more-so for SCC than fatigue given they were on the pipe ID and were present on both sides of the weld toe. SCC is further supported by the presence of chlorine...but why don't the cracks show more preference for a certain microstructure? And if some portion of the crack occurred during fabrication, why is the crack path so continuous/smooth (at least until it connects to the OD-initiated "ledge" mentioned earlier)? 

This image dump is not very well organized and I realize it's also captionless, but I'm hoping it's easily correlated enough to what I've typed above and at least a starting point. What other features do I need to look for to properly characterize this failure?

Thank you,

Hello Evan,

This an excellent description of the problem. A few things standout:

1. Some areas of the crack show heat tint, indicating that the crack in these regions were open when they were exposed to heat. Tint also indicates that the crack was exposed to oxygen when still hot, giving characteristic tint. TIG process may have some air contamination?
2. Since the cracks are generally transgrannular and show no preference for propagating along a particular phase, I suspect that the crack may be caused by thermal stresses during welding, cooldown and subsequent rewelding at the ID.
3. You may also want to look at the thermal stresses generated during welding and cooldown as a result of different heat capacity of the two sides of the weldment. The flange side is likely to have a higher thermal mass as compared to the pipe side.
4. The sensitization that Mr. Stoudt mentioned may play a role if the cooldown is slow.

You may explore the hypothesis that there were thermal stresses built in due to difference in thermal mass of the two sides of the weldment, particularly if no post weld anneal was carried out. These stresses may cause the cracks. Post rewelding at the ID may cause the heat tint and further exaggerate cracks.

Thanks for sharing!

Ratnesh K. Dwivedi, Ph.D.

RKD ENGG, LLC

www.rkdengg.com

A 4-inch diameter schedule 40 pipe welded to a bolted flange connection failed in service at the weld. Design limits are specified as 185 PSIG, 375°F. There is no operational data to suggest these parameters were exceeded. Visual inspection showed the crack was at the flange-side toe of the weld, connected there both on the ID and OD, but appeared to be ID-initiated given the visible portion of the ID crack was longer than the OD portion (2-1/4" long on ID vs 1-1/2" long on ID). The materials, as confirmed by handheld XRF, were 316SS for the pipe, 304SS for the flange, and 308SS for the filler. The filler was determined to have 7.5-10% ferrite via a severn gauge. Weld was open-root, multiple pass, and appeared to be TIG based on cap profile (I do not have the weld procedure info).

The crack was broken open in the laboratory. The crack path was confirmed to traverse from toe-to-toe of the ID to OD weld. There was a distinct lip or ledge near the OD where the crack direction abruptly changed direction. The fracture surface exhibited several areas of distinct heat-tint. The shape of all of these tinted regions was more consistent with OD-initiated cracking (somewhat similar in shape/orientation progression marks emanating from the OD toward the ID). The shape of this heat-tint was especially pronounced on the OD ledge.

SEM evaluation showed what I perceived to be transgranular cleavage fracture. There was no overload observed (at least compared to the lab-generated fracture surface, which was a readily identified dimpled surface). Despite most of the fracture surface being transgranular, there were interspersed facets across the fracture that look to me like isolated intergranular fracture faces. SEM-EDS of these faces showed traces of Na and Cl which were present in discrete particles on the surface. There was no difference in the SEM between the fracture surfaces of the hint-tinted areas compared to the non-heat tinted areas (same general concentration of facets, same fracture morphologies, etc). 

Metallographic evaluation showed that, across 4 cross-sections taken, there was no preference for crack propagation with respect to microstructure. Additionally, in all 4 cross-sections, cracks were observed on both sides of the ID weld toe. The cracks always initiated at the toe of the weld on the ID, but did not travel along the fusion line or HAZ. The crack was confirmed to be transgranular in the base material, and did not exhibit any preferential microstructural path when traversing the weld metal. Crack paths were slightly branched, but not extensively. One notable feature from the microstructural evaluation is that there appeared to be some weld repair performed from the ID, evidenced by the shape of the fusion zone. 

I'm having trouble piecing this all together into a coherent picture. The heat tint on the fracture surfaces suggests that the crack at least partially formed during fabrication, but the fracture features between the heat-tinted areas and all other areas are nearly identical. The cracks initiating at the toe of the weld makes sense from a stress-concentration standpoint, probably more-so for SCC than fatigue given they were on the pipe ID and were present on both sides of the weld toe. SCC is further supported by the presence of chlorine...but why don't the cracks show more preference for a certain microstructure? And if some portion of the crack occurred during fabrication, why is the crack path so continuous/smooth (at least until it connects to the OD-initiated "ledge" mentioned earlier)? 

This image dump is not very well organized and I realize it's also captionless, but I'm hoping it's easily correlated enough to what I've typed above and at least a starting point. What other features do I need to look for to properly characterize this failure?

Thank you,

Well said. I think the heat tint on the crack surface is the key as it means the crack was already present during welding operations.

Hello Evan,

This an excellent description of the problem. A few things standout:

1. Some areas of the crack show heat tint, indicating that the crack in these regions were open when they were exposed to heat. Tint also indicates that the crack was exposed to oxygen when still hot, giving characteristic tint. TIG process may have some air contamination?
2. Since the cracks are generally transgrannular and show no preference for propagating along a particular phase, I suspect that the crack may be caused by thermal stresses during welding, cooldown and subsequent rewelding at the ID.
3. You may also want to look at the thermal stresses generated during welding and cooldown as a result of different heat capacity of the two sides of the weldment. The flange side is likely to have a higher thermal mass as compared to the pipe side.
4. The sensitization that Mr. Stoudt mentioned may play a role if the cooldown is slow.

You may explore the hypothesis that there were thermal stresses built in due to difference in thermal mass of the two sides of the weldment, particularly if no post weld anneal was carried out. These stresses may cause the cracks. Post rewelding at the ID may cause the heat tint and further exaggerate cracks.

Thanks for sharing!

Ratnesh K. Dwivedi, Ph.D.

RKD ENGG, LLC

www.rkdengg.com

A 4-inch diameter schedule 40 pipe welded to a bolted flange connection failed in service at the weld. Design limits are specified as 185 PSIG, 375°F. There is no operational data to suggest these parameters were exceeded. Visual inspection showed the crack was at the flange-side toe of the weld, connected there both on the ID and OD, but appeared to be ID-initiated given the visible portion of the ID crack was longer than the OD portion (2-1/4" long on ID vs 1-1/2" long on ID). The materials, as confirmed by handheld XRF, were 316SS for the pipe, 304SS for the flange, and 308SS for the filler. The filler was determined to have 7.5-10% ferrite via a severn gauge. Weld was open-root, multiple pass, and appeared to be TIG based on cap profile (I do not have the weld procedure info).

The crack was broken open in the laboratory. The crack path was confirmed to traverse from toe-to-toe of the ID to OD weld. There was a distinct lip or ledge near the OD where the crack direction abruptly changed direction. The fracture surface exhibited several areas of distinct heat-tint. The shape of all of these tinted regions was more consistent with OD-initiated cracking (somewhat similar in shape/orientation progression marks emanating from the OD toward the ID). The shape of this heat-tint was especially pronounced on the OD ledge.

SEM evaluation showed what I perceived to be transgranular cleavage fracture. There was no overload observed (at least compared to the lab-generated fracture surface, which was a readily identified dimpled surface). Despite most of the fracture surface being transgranular, there were interspersed facets across the fracture that look to me like isolated intergranular fracture faces. SEM-EDS of these faces showed traces of Na and Cl which were present in discrete particles on the surface. There was no difference in the SEM between the fracture surfaces of the hint-tinted areas compared to the non-heat tinted areas (same general concentration of facets, same fracture morphologies, etc). 

Metallographic evaluation showed that, across 4 cross-sections taken, there was no preference for crack propagation with respect to microstructure. Additionally, in all 4 cross-sections, cracks were observed on both sides of the ID weld toe. The cracks always initiated at the toe of the weld on the ID, but did not travel along the fusion line or HAZ. The crack was confirmed to be transgranular in the base material, and did not exhibit any preferential microstructural path when traversing the weld metal. Crack paths were slightly branched, but not extensively. One notable feature from the microstructural evaluation is that there appeared to be some weld repair performed from the ID, evidenced by the shape of the fusion zone. 

I'm having trouble piecing this all together into a coherent picture. The heat tint on the fracture surfaces suggests that the crack at least partially formed during fabrication, but the fracture features between the heat-tinted areas and all other areas are nearly identical. The cracks initiating at the toe of the weld makes sense from a stress-concentration standpoint, probably more-so for SCC than fatigue given they were on the pipe ID and were present on both sides of the weld toe. SCC is further supported by the presence of chlorine...but why don't the cracks show more preference for a certain microstructure? And if some portion of the crack occurred during fabrication, why is the crack path so continuous/smooth (at least until it connects to the OD-initiated "ledge" mentioned earlier)? 

This image dump is not very well organized and I realize it's also captionless, but I'm hoping it's easily correlated enough to what I've typed above and at least a starting point. What other features do I need to look for to properly characterize this failure?

Thank you,

Thank you to all who have replied, I appreciate the thoughtful responses. 

David,

When I saw the crack connected at the toes of the weld and the general shape of the crack, I also suspected fatigue, but after seeing the branched crack path combined with cracks on both sides of the ID weld toe, I started to question that. I do see evidence of fatigue in the 10-15% wall thickness closest to the OD. The flange was bolted to a valve body. I do not know how comparatively rigid that valve body was.

John,

I appreciate the etching info. I did not end up going that route, but I looked at the mounts in the SEM in BSE mode. I did not see any chrome carbides along any grain boundaries. I do not have any evidence to suggest this material was sensitized. 

Ratnesh, 

You make a lot of excellent points, and I don't have any information about the welding/PWHT (if any) processes so some of that is hard to answer. Ultimately I've come to agree that there was likely a lot of residual stresses in this weldment.

Aaron,

I used to agree that the heat tint was the key but now I'm starting to second guess that. I realize it can't be ignored, but it's the one piece of evidence that doesn't line up with the rest of the info when I try to piece it all together. For the purpose of this analysis I've noted it in my report but without any fractographic or microstructural evidence to support multiple fracture modes, I don't know what else to make of it.

Ultimately, the conclusion I came to is that this was caused by chloride SCC, for the following reasons. With that said if there's any objection to any of these I'm still interested in hearing it.

• Cracking was longer on the ID than the OD, suggesting ID-initiation
  • Precludes OD-initiated fatigue as the primary mechanism
• Cracking was present on both sides of the weld toe on the ID
  • Confirms ID initiation
  • I've never seen mechanical fatigue propagate on both sides of a weld simultaneously, typically once the crack starts this concentrates stress even more at the primary crack tip (I've also never seen mechanical fatigue initiate on the inside of a pipe but I was trying to keep an open mind)
• Crack progressed ~85% through-wall before an OD-initiated fatigue crack connected and caused final failure
  • OD-fatigue assessed by progression marks emanating from OD surface
• ID initiated crack was primarily transgranular, with minor branching and isolated facets of intergranular cracking, consistent with chloride SCC
  • ASM Handbook, Volume 11, Failure Analysis and Prevention, 2021, page 550
• There is no microstructural or fractographic difference between areas with and without heat tint within the ID-initated portion of the crack.

I know this is a long follow-up but I just wanted to say thanks to everyone for your input. It's good to have so many resources willing to provide some insights and help me look at things a different way.

Well said. I think the heat tint on the crack surface is the key as it means the crack was already present during welding operations.

Hello Evan,

This an excellent description of the problem. A few things standout:

1. Some areas of the crack show heat tint, indicating that the crack in these regions were open when they were exposed to heat. Tint also indicates that the crack was exposed to oxygen when still hot, giving characteristic tint. TIG process may have some air contamination?
2. Since the cracks are generally transgrannular and show no preference for propagating along a particular phase, I suspect that the crack may be caused by thermal stresses during welding, cooldown and subsequent rewelding at the ID.
3. You may also want to look at the thermal stresses generated during welding and cooldown as a result of different heat capacity of the two sides of the weldment. The flange side is likely to have a higher thermal mass as compared to the pipe side.
4. The sensitization that Mr. Stoudt mentioned may play a role if the cooldown is slow.

You may explore the hypothesis that there were thermal stresses built in due to difference in thermal mass of the two sides of the weldment, particularly if no post weld anneal was carried out. These stresses may cause the cracks. Post rewelding at the ID may cause the heat tint and further exaggerate cracks.

Thanks for sharing!

Ratnesh K. Dwivedi, Ph.D.

RKD ENGG, LLC

www.rkdengg.com

A 4-inch diameter schedule 40 pipe welded to a bolted flange connection failed in service at the weld. Design limits are specified as 185 PSIG, 375°F. There is no operational data to suggest these parameters were exceeded. Visual inspection showed the crack was at the flange-side toe of the weld, connected there both on the ID and OD, but appeared to be ID-initiated given the visible portion of the ID crack was longer than the OD portion (2-1/4" long on ID vs 1-1/2" long on ID). The materials, as confirmed by handheld XRF, were 316SS for the pipe, 304SS for the flange, and 308SS for the filler. The filler was determined to have 7.5-10% ferrite via a severn gauge. Weld was open-root, multiple pass, and appeared to be TIG based on cap profile (I do not have the weld procedure info).

The crack was broken open in the laboratory. The crack path was confirmed to traverse from toe-to-toe of the ID to OD weld. There was a distinct lip or ledge near the OD where the crack direction abruptly changed direction. The fracture surface exhibited several areas of distinct heat-tint. The shape of all of these tinted regions was more consistent with OD-initiated cracking (somewhat similar in shape/orientation progression marks emanating from the OD toward the ID). The shape of this heat-tint was especially pronounced on the OD ledge.

SEM evaluation showed what I perceived to be transgranular cleavage fracture. There was no overload observed (at least compared to the lab-generated fracture surface, which was a readily identified dimpled surface). Despite most of the fracture surface being transgranular, there were interspersed facets across the fracture that look to me like isolated intergranular fracture faces. SEM-EDS of these faces showed traces of Na and Cl which were present in discrete particles on the surface. There was no difference in the SEM between the fracture surfaces of the hint-tinted areas compared to the non-heat tinted areas (same general concentration of facets, same fracture morphologies, etc). 

Metallographic evaluation showed that, across 4 cross-sections taken, there was no preference for crack propagation with respect to microstructure. Additionally, in all 4 cross-sections, cracks were observed on both sides of the ID weld toe. The cracks always initiated at the toe of the weld on the ID, but did not travel along the fusion line or HAZ. The crack was confirmed to be transgranular in the base material, and did not exhibit any preferential microstructural path when traversing the weld metal. Crack paths were slightly branched, but not extensively. One notable feature from the microstructural evaluation is that there appeared to be some weld repair performed from the ID, evidenced by the shape of the fusion zone. 

I'm having trouble piecing this all together into a coherent picture. The heat tint on the fracture surfaces suggests that the crack at least partially formed during fabrication, but the fracture features between the heat-tinted areas and all other areas are nearly identical. The cracks initiating at the toe of the weld makes sense from a stress-concentration standpoint, probably more-so for SCC than fatigue given they were on the pipe ID and were present on both sides of the weld toe. SCC is further supported by the presence of chlorine...but why don't the cracks show more preference for a certain microstructure? And if some portion of the crack occurred during fabrication, why is the crack path so continuous/smooth (at least until it connects to the OD-initiated "ledge" mentioned earlier)? 

This image dump is not very well organized and I realize it's also captionless, but I'm hoping it's easily correlated enough to what I've typed above and at least a starting point. What other features do I need to look for to properly characterize this failure?

Thank you,