Corrosion, Failure Analysis and Materials Selection Specialists

 

 

 

 

 

Corrosion Testing / Failure Analysis

 

 

 

Corrosion Testing

Failure Analysis

Field Investigations

Litigation

Metallography

Technical Papers

CTL Profile

Pricing & Policies

Contact CTL

Quality Assurance

Return to Failure Analysis Case Histories

Fatigue Cracking of an Inlet Nozzle on a Crude Unit

ENVIRONMENT:

Oil Refinery

EQUIPMENT:

Vapor Inlet Nozzle on a Vacuum Flasher Column in a Crude Unit

MATERIAL:

Type 316 Stainless Steel
SERVICE TIME: Less than 1 year

FAILURE MODE:

Fatigue Cracking Corrosion

 

Background 

The nozzle in question was from a crude unit, and was the west vapor inlet nozzle to the vacuum flasher column.  The weld that failed was a flange-to-pipe stub weld.  The piping connection to this nozzle was from a vacuum column feed heater.  The leak was discovered on startup of the crude unit.

The material for this nozzle was specified as A-240 Type 316 stainless steel, 3/8” wall thickness.  This alloy was confirmed at the site with a portable alloy analyzer. 

The line was installed as a replacement in twenty-one years ago.  A crack developed in the mating flange connection twelve years ago.  That crack was ground out and re-welded and reinforcing gussets added. 

 

Findings

Clearly visible in Figure 1 is one crack running alongside the weld that completely penetrates the wall and extends slightly more than half the length of the sample provided. The crack runs parallel to, but is not touching, the circumferential weld on the nozzle OD surface. 

 

Figure 1. Close-up view of portion of Figure 1 showing end of major crack.  (6X original magnification)

Figure 2. A view of a portion of the upper fracture surface.  (6X original magnification)

 

The part was sectioned.  The section was made in such a way as to allow close examination of the mating surfaces of the through crack fracture surface, as shown in Figure 2.   The fracture was a classic example of fatigue cracking of wrought stainless steel.  The initiation point for this crack was not seen in the sample provided, which meant that the crack started elsewhere on the nozzle and propagated to this point. 

The macroscopic features, coupled with the fact that the crack was longer on the nozzle OD than on the ID, indicated that the crack first occurred on the nozzle OD and propagated both around the nozzle and through the nozzle wall simultaneously. 

Closer examination of the OD surface revealed several additional cracks, some in the weld itself and others adjacent to the weld on both sides. Two cracks following the ripples on the weld bead, as well as cracking alongside the weld on the side away from the large fatigue crack, were documented in Figure 3.  More cracks along weld ripples, most visible where a grinding wheel contacted the surface, were observed elsewhere on the sample. 

 

Figure 3. Close-up view of part of the OD surface, showing cracks in two weld ripples (marked by arrows) and also cracking alongside the weld.  (6X original magnification)

 

The cross-section showed at least three additional cracks – one at the edge of the weld and two within the weld.  The crack at the edge of the weld is shown at higher magnification in Figure 4.  There is much oxidation along the sides of the crack, including at the very crack tip. 

 

Figure 4. Magnified view of a crack at the left edge of the weld. (60X original magnification)

 
Another section revealed an even tighter secondary crack in the part.  This crack, as viewed in the scanning electron microscope, was observed in Figure 5.  As seen in the bottom of the inverted view in Figure 5, there was also a shallow depression or “pit” in the metal surface at this crack location. 

 

Figure 5. An SEM image of a smaller secondary crack.  Note also the depression or “pit” in the metal surface, shown at the bottom in this view. (60X original magnification)

 
An EDS analysis of the base metal and the oxidation product within the crack were made.  The dense product within the surface of the shallow “pit” depression was similarly analyzed.  A black coating was observed on the OD surfaces of the sample, and this, too, was analyzed.  The results were summarized in Table I. 

 

TABLE I – EDS ELEMENTAL ANALYSES 

 (Expressed as approximate weight percents) 

 

Base

Metal

Oxide in

Crack

Product

In “pit”

Surface

Coating

Iron

62

21

71

7

Chromium

17

5

13

2

Nickel

11

3

6

--

Molybdenum

3

--

--

--

Silicon

3

1

1

44

Sulfur

--

2

4

--

Zinc

--

1

1

-

Aluminum

2

-

-

16

Lead

--

--

--

22

 

To further analyze the zinc distribution in the oxides within the crack and “pit”, an EDS map was performed.   The magnification was increased to include only the bulbous oxide at the crack tip, as shown in Figure 6.  A dot map for zinc in this area is shown in Figure 7.  It was seen that zinc was dispersed non-uniformly throughout the oxide products within and adjacent to the crack.  (The individual, widely scattered dots throughout the base metal areas in these two maps were believed to be background from the analysis and not indicative of zinc in the base metal.) 

 

Figure 6. A higher magnification view at the top end of the crack shown in Figure 6.  (Original magnification 220x)

Figure 7. EDS dot map of zinc distribution in Figure 7.

 

A sample of the base metal was mounted, polished, and prepared with two different etchants that selectively reveal certain brittle phases in the grain boundaries.  Etching with Vilella’s reagent would reveal both metal carbides and “sigma” phase in the structure.  As shown in Figure 8 there was significant precipitation in this sample when treated with Vilella’s reagent.  Murikami’s reagent would show metal carbides, but not sigma phase.  As shown in Figure 9, there was clearly no carbide precipitation visible when etched with Murikami’s reagent.  It was clear, therefore, that the grain boundary precipitates in Figure 8 were sigma phase. 

 

Figure 8. The metal structure after etching with Vilella’s reagent.  (250X original magnification)

Figure 9. The same area shown in Figure 10 after re-polishing and etching with Murikami’s reagent. (250X original magnification)

 

Discussion 

The primary cracking failure in this sample was metal fatigue.  No obvious initiation point for the fatigue was visible in this sample.  Apparently the crack began elsewhere on the nozzle and propagated to the portion of the nozzle represented by this sample when the process was shut down. 

The question, then, was whether there were contributing factors in the metal that might have initiated and/or promoted fatigue failure.  At least two such contributing factors were found. 

The numerous secondary cracks in and around the weld appear to be promoted by zinc contamination – a phenomenon referred to as “Type I Embrittlement”.  When unstressed Type 316 stainless steel is exposed to zinc contamination at temperatures above approximately 1050° F, penetration of the metal occurs accompanied by formation of a zinc/nickel intermetallic compound.  With stress present, this penetration occurs in the form of cracks that form perpendicular to the localized stresses. 

The sample had a black coating on its outside surfaces.  This material appeared to be a baked hydrocarbon product, presumably from process leaks near this location.  This coating contained significant levels of metals, including lead, calcium, aluminum and zinc.  It was probably this coating that provided the zinc for reaction with the base metal. 

Metallurgical analysis and selective etching revealed sigma phase precipitation along grain boundaries, but no significant carbide precipitation.  Sigma phase is a brittle metallic phase that, if present, will help to promote fatigue failure along grain boundaries.  Type 316 stainless steel will normally exhibit carbide precipitation in the temperature range of 800 to 1500° F, but those carbides will be re-dissolved into the metal structure at temperatures above approximately 1600° F.  Sigma phase forms in this alloy at temperatures in the range of 1100 to about 1700° F.  The presence of sigma but absence of carbides indicates that this part was heated in the range of approximately 1600 to 1700° F for an extended period. 

Site Index

Site Copyright © 1995 - 2007, All Rights Reserved,

Corrosion Testing Laboratories, Inc.

60 Blue Hen Drive

Newark, Delaware USA 19713

Phone: 1-302-454-8200

Fax: 1-302-454-8204

web@corrosionlab.com