13
IPEIA 14th Annual Conference in Banff, Alberta. February 4, 2010 High Temperature Hydrogen Attack Resistance Using Autoclave Testing of Scoop Samples SM Tim Munsterman Antonio Seijas Dana G. Williams, P.E. Technology Director – Engineering. Mechanical Engineering Consultant Materials & Corrosion Specialist Lloyd’s Register Capstone, Inc. Marathon Petroleum Co, LLC Houston, Texas USA Ashland, Kentucky USA Abstract A novel approach was used to perform a High Temperature Hydrogen Attack (HTHA) assessment on a SA-302 Grade B (C-Mn-½Mo) Guard Reactor built in 1976. This reactor had been operating above the carbon steel curves given in API RP 941, yet had displayed no evidence of HTHA damage. This assessment evaluated the potential for existing and future damage based on real-time monitoring of coupon swelling during accelerated HTHA testing of Scoop Samples SM removed from the OD of the vessel. Accelerated HTHA testing was performed using specialized autoclaves equipped with displacement sensors to detect incipient damage. Extended exposures were used to determine the rate of through wall damage. The test results showed that the Reactor had better HTHA resistance than given in the API RP 941 C ½ Mo curves. The test conditions were approximately 100 o F above the 100 hour incipient damage curve for C ½Mo steel. The time to incipient HTHA damage was found to be longer than predicted by P w calculations given in the recently published API High Temperature Hydrogen Attack Technical Basis report (September 2008). To help prioritize inspection frequencies and locations, the results were used to show the relative HTHA resistance of the various plate materials. Introduction A Guard Reactor in a refinery Distillate Hydrotreater Unit was constructed in 1976 according to guidelines and requirements of the ASME Code Section VIII, Division 1. The material of construction was a Manganese-Molybdenum alloy steel, material specification SA-302 Grade B. When the vessel was built, the API 941 document gave special consideration for C ½ Mo steels as being better than carbon steel. Industry experience found that while some equipment did possess better-than-carbon steel resistance, some vessels suffered damage at operating conditions significantly below the C ½ Mo curve. It was found that heating conditions during manufacturing and fabrication, especially annealing treatments, had a profoundly adverse affect on the High temperature hydrogen attack (HTHA) resistance of the steel. Some materials were found to be only slightly better than carbon steel. This led to the removal of this curve from the API 941 document; it is currently only included for reference. Industry experience has shown that generally, C ½ Mo steels have minimal risk when operated at 50°F or less above the carbon steel curve. The reactor operated at 755 o F outlet temperature with a hydrogen partial pressure of 490 psia. Since the Guard Reactor was operating more than 200 o F above the carbon steel curve, further assessment was requested. As a part of a “fitness for service” assessment of this Guard Reactor, 12 Scoop Samples were removed from the shell and heads by Lloyd’s Register Capstone, Inc. (LR Capstone) using a proprietary tool called a Scoop Sampler SM . Half of the Scoop Samples were used to determine thermal embrittlement for a Minimum Pressurization Temperature analysis. The

HTHA Autoclave Testing

Embed Size (px)

Citation preview

Page 1: HTHA Autoclave Testing

IPEIA 14th Annual Conference in Banff, Alberta. February 4, 2010

High Temperature Hydrogen Attack Resistance Using Autoclave Testing of

Scoop SamplesSM Tim Munsterman Antonio Seijas Dana G. Williams, P.E.

Technology Director – Engineering.

Mechanical Engineering Consultant

Materials & Corrosion Specialist

Lloyd’s Register Capstone, Inc. Marathon Petroleum Co, LLC Houston, Texas USA Ashland, Kentucky USA

Abstract A novel approach was used to perform a High Temperature Hydrogen Attack (HTHA) assessment on a SA-302 Grade B (C-Mn-½Mo) Guard Reactor built in 1976. This reactor had been operating above the carbon steel curves given in API RP 941, yet had displayed no evidence of HTHA damage. This assessment evaluated the potential for existing and future damage based on real-time monitoring of coupon swelling during accelerated HTHA testing of Scoop SamplesSM removed from the OD of the vessel. Accelerated HTHA testing was performed using specialized autoclaves equipped with displacement sensors to detect incipient damage. Extended exposures were used to determine the rate of through wall damage.

The test results showed that the Reactor had better HTHA resistance than given in the API RP 941 C ½ Mo curves. The test conditions were approximately 100oF above the 100 hour incipient damage curve for C ½Mo steel. The time to incipient HTHA damage was found to be longer than predicted by Pw calculations given in the recently published API High Temperature Hydrogen Attack Technical Basis report (September 2008). To help prioritize inspection frequencies and locations, the results were used to show the relative HTHA resistance of the various plate materials.

Introduction A Guard Reactor in a refinery Distillate Hydrotreater Unit was constructed in 1976 according to guidelines and requirements of the ASME Code Section VIII, Division 1. The material of construction was a Manganese-Molybdenum alloy steel, material specification SA-302 Grade B. When the vessel was built, the API 941 document gave special consideration for C ½ Mo steels as being better than carbon steel.

Industry experience found that while some equipment did possess better-than-carbon steel resistance, some vessels suffered damage at operating conditions significantly below the C ½ Mo curve. It was found that heating conditions during manufacturing and fabrication, especially annealing treatments, had a profoundly adverse affect on the High temperature hydrogen attack (HTHA) resistance of the steel. Some materials were found to be only slightly better than carbon steel. This led to the removal of this curve from the API 941 document; it is currently only included for reference.

Industry experience has shown that generally, C ½ Mo steels have minimal risk when operated at 50°F or less above the carbon steel curve. The reactor operated at 755oF outlet temperature with a hydrogen partial pressure of 490 psia. Since the Guard Reactor was operating more than 200oF above the carbon steel curve, further assessment was requested. As a part of a “fitness for service” assessment of this Guard Reactor, 12 Scoop Samples were removed from the shell and heads by Lloyd’s Register Capstone, Inc. (LR Capstone) using a proprietary tool called a Scoop SamplerSM. Half of the Scoop Samples were used to determine thermal embrittlement for a Minimum Pressurization Temperature analysis. The

Page 2: HTHA Autoclave Testing

IPEIA 14th Annual Conference in Banff, Alberta. February 4, 2010

remaining samples were available for the HTHA assessment study. The HTHA samples were subjected to elevated temperatures and hydrogen pressures to accelerate the testing. Carbon steel samples were included for reference.

Exposure times of as much as 830 hours were used at a test temperature of 1000oF and a hydrogen partial pressure of 1415 psia. The results of the incipient damage testing are the focus of this paper. It has been LR Capstone’s experience that many of the “bad” C ½ Mo steel vessels have developed cracks and have been removed from service. The remaining vessels, therefore, have a greater potential of being “good” C ½ Mo steel. The problem has been in determining the actual HTHA resistance by non-destructive methods. Grain structure evaluations have not worked and carbide extraction methods do not produce quantitative results. Additionally, industry experience has shown that C Mn ½ Mo vessels (A302 steels) have performed better than C ½ Mo (A204 steels) even though API 941 does not have separate curves for these materials.

Therefore, LR Capstone developed a relatively straight forward approach to HTHA resistance evaluation by using methods similar to “remaining life creep” assessment. Specifically, test coupons are removed from the vessel and subjected to accelerated test conditions in autoclaves to determine actual material performance, rather than using lower bound or other assumed properties. These coupons were monitored real-time using displacement sensors to detect a change in swelling rate, thus signaling the onset of incipient damage.

The API 941 incipient damage curves for carbon steel and C ½ Mo steels is shown in Figure 1. The accelerated test conditions used were based on similar test conditions previously performed by LR Capstone. Test temperature was set at 1000°F at a hydrogen pressure of 1415 psia. These test conditions, represented by the black diamond, are shown in Figure 1. Also included in this figure are the 100 hour incipient damage curves for carbon steel and C ½ Mo steel. It can be seen that the test conditions were severe. The black square represents the operating conditions.

Background

High temperature hydrogen attack occurs when carbon and low alloy steels are exposed to temperatures greater than 430ºF and the operating hydrogen partial pressure is greater than 80 psia. At these high temperatures and partial pressures, hydrogen dissociates into atomic hydrogen (H) (See equation #1). With high pressures and operating temperatures, diffusion of small atomic hydrogen is accelerated and readily diffuses through the steel and reacts with the carbides in the microstructure. The reaction given in Equation 2 shows that the atomic hydrogen reacts with the metal carbides and produces methane, a large molecule compared to atomic hydrogen. The gaseous methane pressure forms voids, which result in the formation of micro-fissures along grain boundaries that can link, forming blistering, and/or through wall cracking. When the material is attacked, there is significant deterioration of its mechanical properties in tensile strength and ductility, which could result in brittle failure.

H2 ↔ 2H (hydrogen dissociation) (1)

4H + FeC↔CH4 + Fe (atomic hydrogen reacting with iron carbides) (2)

High temperature hydrogen attack can result in surface decarburization, internal decarburization and fissuring. Surface decarburization occurs at relatively low hydrogen partial pressures and elevated temperatures; therefore, diffusion of hydrogen atoms is relatively slow and only penetrates the surface layer. At the surface, the hydrogen reacts with

Page 3: HTHA Autoclave Testing

IPEIA 14th Annual Conference in Banff, Alberta. February 4, 2010

the carbon from the microstructure and results in a decarburized area. Since the methane is not trapped sub-surface, there is no cracking. The decarburization progresses as carbon migrates from the base metal towards the low carbon zone where hydrogen reacts to form more methane. Since no cracks are formed, the damage is limited to the reduction in strength due to the loss of carbon in the steel at the exposed surface, typically the inner diameter (ID) surface.

Subsurface decarburization and cracking occurs when high partial pressures assist hydrogen penetration into the microstructure where hydrogen and carbon react to form methane, which cannot escape to the surface. The large molecular size of the methane accumulates at the grain boundaries, causing high localized stresses where intergranular cracks or fissures form. The fissures separate the grains and over time, link to form larger cracks, resulting in significant deterioration of the material properties. This form of damage results in a slight increase in thickness or swelling.

The degree of hydrogen attack depends on temperature, hydrogen partial pressure, stress level, surface condition, exposure time, steel composition and structure. Therefore, to be used as an industry standard, G.A Nelson of Shell Oil Development, collected laboratory and plant data from high temperature hydrogen attacks. From the data various curves were developed to represent the effects of hydrogen at high temperatures and pressures. These “Nelson Curves” have gone through periodic revisions as new data has been obtained. Safety factors from approximately 30°F to 50°F have been incorporated, so these curves tend to be conservative. To simulate the Nelson curves, Shewmon developed the Pw equation (Equation 3) for the attack rate, which can be used to describe time/pressure/temperature relations of the general shape of the Nelson curves for C ½ Mo steel.

The Pw parameter is based on the amount of time to incipient damage. The Japan Pressure Vessel Research Council proposed that the critical values of Pw on HTHA of C ½ Mo steel were as follows (as given in the Technical Basis for 941 Report, Appendix F.):

Pw=- 3 ln(PH2)-ln(t)+190000/(8.3145 T) (3)

where PH2 = hydrogen partial pressure in MPa, t = time in hours, T= temperature in °K

Test Procedures

Twelve samples were taken from the external surface of the shell and top head sections of the Guard Reactor, using LR Capstone’s proprietary Scoop SamplerSM. Figure 2 shows locations of the Scoop Samples and Figures 3a and 3b show the Scoop Samples that were removed from the vessel OD surface. All samples were removed from the base metal and weld metal, away from major discontinuities. The samples were taken within limits based on local thin area calculations, thus no repairs were needed.

Test coupons were machined from each of the Scoop Samples to yield two test coupons, one 1.3 inch diameter x 0.3 inch thick coupon for one side exposure and an incipient damage coupon that was 0.25 inch diameter x 0.375 inch long. These coupons were subjected to temperatures of 1000°F at 1415 psia. The incipient damage testing was performed by exposing all sides (full immersion) of the 0.25 inch diameter samples to high pressure hydrogen. The one side exposure samples were used to determine the through-wall-damage rate. Thermocouples were placed on the bottom of the inner (hydrogen) chamber to monitor the coupon temperature during the test. Tests also included carbon steel coupons for reference.

Page 4: HTHA Autoclave Testing

IPEIA 14th Annual Conference in Banff, Alberta. February 4, 2010

Figure 4 shows a simplified drawing of the test chamber arrangement. In order to detect incipient damage in real time, a displacement sensor was used to monitor swelling of the full immersion coupons. Previous research by Shewmon noted that swelling of C ½ Mo material occurred during the test and incipient damage occurred when there was a change in the rate of damage, somewhat analogous to the change from linear secondary creep to more rapid deterioration during tertiary creep.

Testing was performed until the incipient damage coupons displayed a change in growth rate. The test was continued for an additional 20 percent of the incipient time so that detectible damage occurred in the single side exposure coupon.

Results Figures 5 – 7 show the displacement data acquired during three of the autoclave tests. Figure 5 shows the displacement of the carbon steel sample where the growth rate during incubation is low. The transition at incipient damage was clearly seen. Figures 6 and 7 show the results from the testing of two C ½ Mo samples. The C ½ Mo materials displayed growth from the beginning of the tests. The increase in growth rate was used to define the time to incipient damage.

Figures 8A-C show a typical full immersion sample after HTHA testing. Cross section views of this carbon steel sample show the cracking and blistering experienced during testing.

Figures 9A and B show one of the single side exposure test samples after 830 hours of exposure. This coupon was taken from Scoop Sample 6 and displayed the highest HTHA resistance.

Figures 10 and 11 show the difference in HTHA resistance between carbon steel and a C ½ Mo with good resistance. Depth of damage on the carbon steel sample shown in Figure 10 was 10 times the damage of Scoop Sample 5 as seen after 181 hours of autoclave testing.

The results of the calculated Pw factors for incipient attack of carbon steel and the incipient attack of C ½ Mo steels are given in Table 1. Therefore, it is concluded that the Guard Reactor has HTHA resistance as good or better than normalized “good” C ½ Mo steel.

Page 5: HTHA Autoclave Testing

IPEIA 14th Annual Conference in Banff, Alberta. February 4, 2010

HTHA Test Results with Pw Calculations

Test Samples T, °F

P, psia

Hours to Incipient damage

API 941 App. A-1

Critical Pw20 yr

No Attack

Pw Calc.

ProjectedYears to

IncreasedRisk

Carbon Steel (516-70) Sample 1013 1415 10 ----

Scoop Sample# 11 (Upper Course) 1011 1415 110 16.43 110

Scoop Sample# 5 (Lower Course) 1000 1415 151 16.33

Sample# 9 (Top Head) 1000 1415 159 16.28

Scoop Sample# 6 (Lower Course) 1000 1415 232 15.90

C ½ Mo Current Op. Conditions 755 490 17.75 30

The value of using the Pw factor is that it can be used to evaluate different sets of operating conditions and make comparisons based on hours of service. Using the shortest time to incipient damage yields the highest calculated Pw (Pw = 16.43) with operating conditions of 755°F and a hydrogen partial pressure of 490 psia, the reactor has an estimated life of 110 years before reaching an increased susceptibility for HTHA damage. However, to add a margin of safety, the API 941 Appendix A-1 C ½ Mo “No Attack” reference curve was considered as a 20 year line. An upper bound Pw of 17.75 was calculated using the operating pressure at a corresponding temperature on the “No Attack” curve, and setting the time variable to 20 years. This Pw was then used at the operating conditions to determine the time. This gave an estimated time of 30 years before reaching a medium susceptibility to HTHA damage. Inspection plans should be adjusted according to API 581 (or similar) guidelines.

This is contrasted to a typical assessment for a C ½ Mo vessel operating 230oF above the carbon steel curve where a high likelihood would be assessed after just 250 hours of exposure (based on API 581 guidelines found in Appendix I, Table I-3).

Conclusions

The results of the coupon testing using accelerated HTHA exposure techniques indicated that the Guard Reactor has similar or better resistance than “good” C ½ Mo steel, as given by the reference curve in API 941 Appendix A, Figure A-1.

Using Pw calculation techniques, at operating conditions of 755oF and a hydrogen partial pressure of 490 psia, the likelihood will increase to a medium potential for HTHA damage after 30 years of operation. It is recommended to perform HTHA inspections using risk-based guidelines given in API 581 or similar.

From the test results, the upper course and top head displayed lower HTHA resistance. Therefore, it was recommended to perform more thorough inspections of these plates and fewer locations on the lower course. It was also recommended to perform similar testing on the bottom head.

Page 6: HTHA Autoclave Testing

IPEIA 14th Annual Conference in Banff, Alberta. February 4, 2010

400

450

500

550

600

650

700

750

800

850

900

950

1000

1050

1100

0 200 400 600 800 1000 1200 1400 1600

Hydrogen Partial Pressure (psia) from API 941

Tem

pera

ture

(o F)

C- 0.5 MoRef. Curve

C- 0.5 Mo100 Hr.

Incipient Damage

CS Curve 100 Hr.

Incipient Damage

1000F, 1415 psia

Test Condition

CS Curve

755F490 psia

Operating Condition

Figure 1. Graph from API 941 showing High Temperature Hydrogen Attack curves.

Figure 2. Sketch showing locations where Scoop Samples were removed from the heads and shell sections.

Lower Course

Top Head

Bottom Head

Upper Course

Scoop Sample 1

Scoop Sample 4

Scoop Sample 8

Scoop Sample 10

Scoop Sample 3

Scoop Sample 7

Scoop Sample 2

Scoop Sample 9

Scoop Samples 5

and 6

Impact Testing

Scoop Samples 11 and 12

HTHA Testing

Page 7: HTHA Autoclave Testing

IPEIA 14th Annual Conference in Banff, Alberta. February 4, 2010

Figure 3a. Photographs of external surfaces of the Scoop Samples removed from the head and shell section of the Guard Reactor.

Figure 3b. Photographs of cut surfaces of the Scoop Samples removed from the head and shell section of the Guard Reactor.

Page 8: HTHA Autoclave Testing

IPEIA 14th Annual Conference in Banff, Alberta. February 4, 2010

Figure 4. Simplified diagram of the test setup. Two test samples were machined from each Scoop Sample, a small ¼ inch diameter full immersion sample and a one side exposure to simulate vessel conditions. Growth of the full immersion sample was detected by the displacement transducer. Four parallel autoclaves were used, three contained reactor materials, the fourth autoclave contained a carbon steel reference sample.

-0.0050

0.0000

0.0050

0.0100

0.0150

0.0200

0.0250

0.0300

0.0 5.0 10.0 15.0 20.0 25.0 30.0

Elapsed Time (Hours)

Stra

in

10 Hours

Figure 5. Full immersion coupon from Carbon Steel reference material. Displacement sensor data is shown detecting HTHA damage real time during the autoclave testing. The red line is the incipient damage threshold level.

Page 9: HTHA Autoclave Testing

IPEIA 14th Annual Conference in Banff, Alberta. February 4, 2010

-0.00002

0.00000

0.00002

0.00004

0.00006

0.00008

0.00010

0 20 40 60 80 100 120 140 160

Elapsed Hours

Stra

in

110 Hrs

Figure 6. Full immersion coupon from Scoop Sample #11 from the upper course. Displacement sensor data detected HTHA damage real time during the autoclave testing. The red line is the incipient damage threshold level.

-0.00002

0.00000

0.00002

0.00004

0.00006

0.00008

0.00010

0.00012

0.00014

0 20 40 60 80 100 120 140 160 180 200

Elapsed Time (hours)

Stra

in

159 Hrs

Figure 7. Full immersion coupon from Scoop Sample #9 from the top head. Displacement sensor data detected HTHA damage real time during the autoclave testing. The bottom red line is the incipient damage threshold level. The jump in displacement was likely due to blister formation in the sample.

Page 10: HTHA Autoclave Testing

IPEIA 14th Annual Conference in Banff, Alberta. February 4, 2010

Figure 8. Full immersion carbon steel incipient damage coupon after exposure. A. After removal from autoclave. B. Macro Cross Section view after Nital etch. C. Numerous microfissures were visible at 160X, 2% Nital etch.

A

B

C

Page 11: HTHA Autoclave Testing

IPEIA 14th Annual Conference in Banff, Alberta. February 4, 2010

Figure 9A. Scoop Sample 6, Single Side Exposure sample after HTHA testing. A large blister at the center with numerous small blisters can be seen on the exposed surface.

Figure 9B. Cross section macro polish view of the single side exposure coupon from Scoop Sample 6. Depth of damage after 830 hours of testing was 39.5% of the through wall thickness. 12X Nital etch.

Page 12: HTHA Autoclave Testing

IPEIA 14th Annual Conference in Banff, Alberta. February 4, 2010

Figure 10. SA-516 Gr70 (Carbon Steel) One side exposure control sample after 181 hours at 1000°F with a hydrogen partial pressure of 1415 psia. Damage depth was measured at 75% of the sample thickness. Incipient damage occurred at 10 hours of exposure.

Page 13: HTHA Autoclave Testing

IPEIA 14th Annual Conference in Banff, Alberta. February 4, 2010

Figure 11. SEM Photograph of Scoop Sample# 5 (lower course) after 181 hours at 1000°F with a hydrogen partial pressure of 1415 psia. Damage was found to be 7.4% of the sample thickness. Incipient damage was indicated at 151 hours.