Paper No.

9548
REVIEW OF RECOMMENDED PRACTICES FOR REMOVAL OF HYDROGEN THROUGH
BAKEOUT FOR WELDED REPAIRS

Afolabi Egbewande
Stantec Consulting Ltd.
1200-59th Ave, SE
Calgary, Alberta, T2H Mark M. Sadowski
2M4 Canada Stantec Consulting Ltd.
1200-59th Ave, SE
Stuart Guest Calgary, Alberta, T2H 2M4 Canada
Stantec Consulting Ltd.
1200-59th Ave, SE Richard Tchorzewski
Calgary, Alberta, T2H Stantec Consulting Ltd.
2M4 Canada 1200-59th Ave, SE
Calgary, Alberta, T2H 2M4 Canada
Valer Zapirtan-Lainer
Stantec Consulting Ltd.
1200-59th Ave, SE
Calgary, Alberta, T2H
2M4 Canada
ABSTRACT

Welding in accordance with a fabrication code for new construction may produce hydrogen weld cracking
but it has largely been mitigated using suitably designed welding procedures and advances in steel
processing and ferrous metallurgy. However, when some key elements of a “good” welding practice are
omitted, or in some cases not qualified on full-scale welding mock-ups, weld cracking may occur. Weld
cracking can occur in both new construction and welded repairs of in-service components. In many cases,
failure is attributed to hydrogen introduced into the weld metal and heat affected zone from a combination
of the atmosphere, service conditions, and welding. In some cases, it may be necessary to remove a
significant portion of the hydrogen in the steel to prevent cracking following a welded repair.

The communication presents a summary of hydrogen bakeout history as a means of removing hydrogen
from a component, compiles existing recommendations regarding hydrogen bakeout in codes and
standards, and reviews the results obtained from an industry survey of energy producers in Western
Canada. The paper concludes with a proposed methodology for selecting bakeout parameters based on
the hydrogen concentration derived from sour service conditions reported by one of the survey
respondents.

Keywords: Hydrogen bakeout, diffusible hydrogen concentration, delayed cracking, hydrogen

embrittlement

©2017 by NACE International.

Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 15835 Park Ten Place, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
1
 Definition of Terminology

Bakeout: describes a process by which hydrogen is removed from steels prior to the start of repair
welding operations. In many codes, the terms “preheat”, “post-heat” or “outgassing” may be used as
synonyms but this practice is not adopted in this communication.

Bulk hydrogen: is the residual hydrogen content in the region of the weld. The bulk hydrogen decreases
with time as hydrogen diffuses out of the weld metal and the heat-affected zone.

Diffusible hydrogen: this is the amount of hydrogen present in the base material or weld in an
atomic/ionic form and may diffuse out of the weld region into the atmosphere during weld cooling

Residual hydrogen: is associated with irreversible hydrogen traps in the material such as microvoids,
inclusions and particle matrices (e.g., MnS, Al2O3, TiC, etc). High temperatures, greater than ~ 600 oC,
are required to release residual hydrogen from hydrogen traps.

Local hydrogen: is the diffusible hydrogen trapped in weak trap sites (e.g., interstitial lattice sites,
dislocations, etc) in the material. Diffusible hydrogen removal occurs at temperatures below 300 oC.

Preheat: a standard term used in welding to describe the application of heat to the base material before
welding.

Outgassing: is commonly applied to spacecraft materials and vacuum systems materials and their
treatments where the right choice of materials for high vacuum is of utmost importance.

Weld: encompasses weld metal, heat affected zone, and portions of the surrounding base metal near
the repair/maintenance weld.

Post Weld Heat Treatment (PWHT): Thermal treatments applied after welding to modify the
microstructure and physical properties of the weld region.

Sour Service: Service environments that contain sufficiently high H2S concentrations to cause cracking
and corrosion concerns. The definition of sour service may vary depending on the code or standard.

Introduction

Hydrogen in steels causes welding fabrication problems and in-service material degradation that reduces
the service life of steel components and structures.1–9 Hydrogen related material degradation modes
may be manifested in a variety of forms – blistering, hydrogen induced cracking (HIC), stress oriented
hydrogen induced cracking (SOHIC), high temperature hydrogen attack (HTHA), etc – but are most
commonly described as an unexpected brittle failure under the encompassing term “hydrogen
embrittlement (HE)”. Besides the extreme and aggressive damage mechanisms listed above, hydrogen
embrittlement commonly manifests as cold cracking in welds during/after the initial fabrication or in-
service repairs. Each mode of material degradation requires a tailor-made approach for repair or
maintenance welding that may require either localized material replacement with weld metal, partial
replacement of the equipment material of construction with compatible new materials or replacement of
the entire equipment item or its component. Localized or partial material replacement requires welding
of new material to the in-service (old) material. In these cases, a successful repair may require a
hydrogen bakeout to remove service-related diffusible hydrogen that remains in the material after the
item is taken out of service and prior to repair welding.

©2017 by NACE International.

Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 15835 Park Ten Place, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
2
 It is well understood that the deleterious effects associated with absorbed hydrogen are independent of
the hydrogen source and can be prevented by removing the hydrogen or by keeping the hydrogen
content below some critical level. Since the publication of API(1) PUBL 946 10 in 1981 on hydrogen
outgassing and its subsequent withdrawal without replacement, portions of this work have been
incorporated into codes, standards and recommended practices aimed at preventing HIC during welding
by reducing the steel’s hydrogen concentration prior to welding.11–17 However, these documents have
largely employed non-mandatory language and the guidance addresses a narrow range of service
environments, service temperatures, and operating conditions. For components that fall outside the
ranges addressed in the relevant codes and standards, relatively complicated mathematical analysis
and engineering judgement must be used in the selection of diffusion equations and parameters for
which values are often not available. These variables include the initial lattice hydrogen concentration,
lattice hydrogen concentration threshold for preventing hydrogen cracking, and a few diffusion variables.
This process is often very challenging. The implication of this complexity is that incorrect bakeout
parameters are often selected, leading to cracking of the repair weld cracking which then necessitates
an additional bakeout and repair weld or replacement of the component.

The authors completed a survey of industry practices on hydrogen bakeout, receiving a wide variety of
bakeout temperature and time for similar vessels in identical service conditions. This communication
reports the results of the survey, discusses and critiques industry practices on hydrogen bakeout,
highlights some available industry resources and practices, and provides recommendations on
addressing some of the challenges typically encountered in designing a hydrogen bakeout procedure.
The work will also demonstrate a method for determining a hydrogen bakeout procedure for a sour
service component (partial pressure equal or higher than 0.3 kPa H2S is considered) operating below
200 oC. Maintenance welding of equipment in high temperature hydrogen service (above 200 oC) will
not be discussed. The proposed method could be used to determine unique hydrogen bakeout
procedures for specific service conditions, reducing the likelihood of hydrogen induced cracking and
preventing additional repair work.

Codes Treatment of Hydrogen Bakeout

Select codes, standards and regulatory documents that are publicly available and mention hydrogen
bakeout have been reviewed and the results compiled in Table 1 for illustrative purposes and for basic
reference. The summary in Table 1 is also intended to facilitate evaluation of the results from the
authors questionnaire on hydrogen bakeout practices sent to energy producers in Western Canada and
presented in the following section. A critical review of Table 1 yields a few general observations:

 Some codes and standards provide mandatory guidelines on avoidance of hydrogen cold
cracking of repair welds as an alternative to preheat, post-heat or hydrogen bakeout.15,18 These
codes require controlled cooling after welding in order to allow for hydrogen diffusion out of the
weld before the weld cools to low temperatures where diffusion kinetics are much lower. In
these cases, the operator must follow both the applicable codes and standards and regulatory
requirements for in-service repairs of specific types of equipment or its components.

 Hydrogen bakeout detail is generally described using non-mandatory wording and provides a
guideline for a specific service, bakeout temperatures and the related holding times at
given temperature.11,14 It is up to the operator to decide if a bakeout is required for the specific
repair and to make several critical assumptions and judgements about the hydrogen bakeout
details. Often, the operator has their own in-house bakeout procedures for typical in-service
repairs that considerably reduce the decision time during emergency shutdowns and planned
turnarounds.

(1)
NACE International (NACE), 15835 Park Ten Place, Houston, Texas 77084, USA
©2017 by NACE International.
Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 15835 Park Ten Place, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
3
 Table 1: Summary of Code and Standard Bakeout Parameters
Code / Standard / other Service Evaluation Material* Wall Bakeout Detail
Thickness*
NACE(2) SP0296-2010,
Wet H2S Required Carbon steel N/A  Up to 4 hrs. at 204 °C
Section 4 11.

NACE SP0296-2010,  3 hrs. at 426 oC or, 6 hrs. at 315 oC or, full

Wet H2S Required Carbon steel Plate, 25mm PWHT at holding times shorter than specified
Section 4 11.
for PWHT
 Should be heated for at least 20 min at 204 °C
or higher to drive off any hydrogen in the
ASME(3) B31.12-2014, Pipeline that has metal.
Appendix A, A-11, been used in sour Carbon steel N/A  Heating should be in addition to and
Preheating for pipeline 29 gas service. immediately preceding any preheating
specified in the welding procedure for new
pipeline.

ASME PCC-2-2011  To provide a slower rate of cooling through the

Article 2.8: Alternatives to 204 °C range, allowing more time for
traditional welding preheat, To address the hydrogen that may be present to diffuse away
Article 2.10: In-service potential for from the weld and adjacent base metal to
welding onto carbon steel hydrogen-induced avoid hydrogen cracking.
pressure components or cracking in the Carbon steel N/A  Hydrogen cracking susceptibility should be
pipelines, HAZ due to evaluated by destructive and non- destructive
Article 4.1.1.10: Hydrogen hydrogen in the testing.
Bakeout, & weld environment  Any potential effect the hydrogen bakeout
Article 4.2: Weld Procedure requirement has on the pressure components
Qualification: Testing 15 or pipeline contents shall be evaluated.
To address repair
ASME BPVC, Section XI, or replacement
P-No. 1
Div.1-2007, Article IWA- activities  The completed weld area shall receive a post-
4000, Repair / (carbon
associated with weld hydrogen bakeout by maintaining it at
steel) and P-
Replacement pressure retaining (230 °C to 290 °C) for a minimum of 2 hr. after
Activities.(Rules For In- No. 3 N/A
components and completion of the weld in P-No. 1 materials.
(low alloy
service Inspection Of their supports,
carbon steel)  For P-No. 3 materials, the holding time shall
Nuclear Power Plant and metallic be a minimum of 4 hrs.
materials
Components) 18 portions of Class
CC containments
and their integral  For production welds, the preheat temperature
attachments. should be maintained until final post-weld heat
Control of welding treatment or a post-weld hydrogen bakeout is
for low-alloy steel performed between 200 oC and 400 ºC for a
NRC(4) , Regulatory Guide
components P-No. 3, 4, minimum of 4 hours after which the
1.50, March 2011 Rev.1
during initial & 5A (low component may be slowly cooled to ambient
“Control Of Preheat N/A
fabrication. The alloy steel temperature prior to the performance of the
Temperature For Welding
guide applies to materials) final PWHT.
Of Low-Alloy Steel” 30 light- water-  The post-weld hydrogen bakeout temperature
cooled reactors. and soak time should be based on the
materials being welded, the geometry, and the
welding process used.
 A bakeout before welding may be required
Sour service
and hardness testing of the final weld may
Note: In 2010,
(5) also be required.
ABSA , AB-519, “Pressure Rev. 1, the
Piping Alternative Test document states  For process fluids other than sour service,
Methods Procedure – “Changed from Required N/A N/A process contamination of the pipe material
due to the process fluids and operating
Requirements”. Issued ‘will’ to ‘may’ as in
conditions must be considered by the
2014-01-06, Rev.2 31 some cases a
Designer, to determine if any mitigation
bake-out is not
actions are needed to assure production of a
essential”.
sound weld.
*
N/A implies that the code has no requirements pertaining to this variable

(2)
NACE International (NACE), 15835 Park Ten Place, Houston, Texas 77084, USA
(3)
American Society of Mechanical Engineers (ASME), Two Park Avenue, New York, NY, 10016-5990, USA
(4)
U.S. Nuclear Regulatory Commission (NRC), Washington, DC 20555-415-5575, USA
(5)
Alberta Boilers Safety Association (ABSA), 9410 20th Avenue, NW, Edmonton, AB, Canada, T6N0A4
©2017 by NACE International.
Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 15835 Park Ten Place, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
4
  The authors are not aware of any unified quantitative approach to deciding whether the hydrogen
bakeout as such is needed or not. Most in-service repair welding decisions are based on mostly
qualitative and laboratory/research data that rely on a current understanding of hydrogen
induced cold cracking phenomenon and ways to mitigate or avoid it.

 The code clauses do not adopt a first principles approach to determining hydrogen bakeout
parameters; rather recommendations are based on experience with similar vessels in identical
service conditions. Fully quantitative recommendations must be based first principles and must
explicitly determine:

o The initial diffusible hydrogen content in the wall/weld of components removed

from service or newly fabricated (welded) components
o A bulk hydrogen threshold below which hydrogen cracking would not occur
o A bakeout time that is uniquely determined based on the initial hydrogen
content, hydrogen diffusion parameters and the threshold bulk hydrogen concentration.

The authors believe that in some cases, it is possible to develop a sufficiently sophisticated approach
that uniquely quantifies the hydrogen bakeout parameter on a case-by-case basis. Such an approach is
presented later in this communication.

Industry Survey on Hydrogen Bakeout Practices

The authors, on behalf of an industry interest group, conducted a survey of hydrogen bakeout practices
in the industry. Approximately 60 industry experts were sent a survey questionnaire to obtain industry
data on hydrogen bakeout during in-service repairs by welding. Out of 60 industry experts, 13 (~ 22%)
responses were received. Some respondents graciously provided their hydrogen bakeout manuals
while others only reported specific data points. The combined results of this survey are reported below
in Figure 1. Examining the survey results, the reported industry approaches could be categorized into:

1. “One-size-fits-all” approach:

4 out of 13 respondents disclosed using the same bakeout time at temperature – for all bakeouts. In
these cases, a deviation from this philosophy occurred only where cracking occurred following the
recommended bakeout. In such “cases”, bakeout is repeated for an additional length of time before
welding is repeated. This is similar to the approach in NACE SP0296.11

2. Case-specific approach:

In this approach, bakeout times are carefully chosen to reflect the type of alloy, wall thickness, service
temperature, service environment, etc. The adopted approach could be further divided into:

 Bakeout selection curves: Here bakeout curves were produced from first principles. These
curves usually chart bakeout times at various temperatures for a specific wall thickness.

 Bakeout flowcharts: Here a flowchart style decision matrix is used to determine the appropriate
bakeout parameters. The bakeout parameters cover a range of materials, wall thickness, service
temperature and service environments. Typically, the ranges involved are smaller subsets of
those in the one-size-fits-all approach. Bakeout parameters are generally based on experience
with a given class of equipment or similar ones in identical service conditions.

©2017 by NACE International.

Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 15835 Park Ten Place, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
5
 Figure 1: Summary of industry survey results. Error bars indicate minimum/maximum values
while data labels are the number of data points obtained from survey respondents.

Carefully examining the data in Figure 1, there appears to be a general trend of increasing bakeout
times for thicker components, aligning with common engineering practice. However, it is important to note
the large variations (minimum and maximum) in bakeout time for numerous temperatures and thicknesses.
In these procedures, it is apparent that the respondents had very different bakeout parameters. For
example, bakeout times ranging from 2 to 20 hours were gathered for a temperature of 250 oC and wall
thickness of 38.1-50.8 mm. From the received bakeout parameters, several trends could be observed:

1. There are more missing data points at the lower bakeout temperatures for thicker components,
most likely attributed to the lower hydrogen removal rates which would result in excessively long
bakeout times. Temperatures around 260 oC were generally used for carbon steels.

2. Careful examination of the hydrogen bakeout times below 300 oC showed significant
inconsistencies as the bakeout time did not necessarily reduce with increasing bakeout time or
reducing component wall thickness as expected. Above 300 oC, the bakeout times consistently
reduced with increasing bakeout temperature and reducing wall thickness. Temperatures around
370 oC appeared to be generally used for alloy and stainless steels.

3. The above two trends may also be explained by the fact that:

a. Industry generally considers thermal treatments around 500 oC and above to be a PWHT.
Codes and standards applicable to PWHT require that a welding or metallurgical specialist
consider the effect of the thermal treatment on the materials mechanical properties. This
requirement will generally lead to the use of a more thoroughly-designed bakeout
procedure.

b. Thick wall vessels are most often found in high pressure and/or aggressive service
environments. The inherently high consequence of failure in such service environments
compels the operator to pay careful attention to any thermal treatment applied to the
component.

The service environments reported in the survey could be broadly categorized into:
©2017 by NACE International.
Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 15835 Park Ten Place, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
6
 1. pH in the aqueous solution in contact with steel has been determined, concentrations of active
species in solution are known but no information is available on the partial pressures of species
in the process environment

2. The partial pressure of hydrogen rich gas above the aqueous media (e.g. H2S) and types and
concentration of active species are known

When the results of the Industry Survey are compared to codes and standards requirements, very few
respondents reported having developed procedures that were prepared with more detailed
considerations of the alloy type, wall thickness or service conditions than those in the codes described
in Table 1. The basis adopted by the respondents for producing hydrogen bakeout procedures could
be broadly divided into:

 Experience with similar vessels in identical service environments: These types of

procedures generally align with some of the code specifications highlighted in Table 1. In some
cases, experience with hydrogen bakeout had been used to expand the scope of applicability
of these code requirements beyond those originally specified by the codes. Such codes are
often aided with carefully structured flow charts or decision matrices to help facilitate the most
appropriate bakeout parameters. However, these types of procedures still end up
recommending the same bakeout parameters for a wide range of wall thicknesses and service
conditions. The implication of this is that sometimes the recommended time at temperature is
insufficient and cracking still occurs during/after welding repairs or excessively high bakeout
temperatures and times than required are employed.

 Bailey’s approach: The book “Welding steels without hydrogen cracking”19 has a dedicated
chapter for estimating the removal of hydrogen from a weld by thermal treatments during and
after welding. Several of the survey respondents acknowledged some level of familiarity with
Bailey’s procedure. It recommends a case-specific hydrogen bakeout time at temperature. If
applied correctly, this approach would prevent the specification of excessive bakeout time at
the chosen temperature while reducing the instances where bakeout must be repeated because
of cracking. This work is the closest to a quantitative approach of all the hydrogen bakeout
design tools or codes available discussed and the principles behind this methodology a r e
discussed in the section below.

Bailey’s Approach
The procedure is based on Fick’s second law for diffusion of hydrogen out of metals.19 Hence the
relationship between the lattice hydrogen concentration at any given time during the hydrogen bakeout
process is well described by Fick’s second law. There is a compendium of precise solutions of the general
equation available in literature along with a few assumptions that may help simplify the calculation.20–22
Additionally, Bailey’s approach requires considerations of hydrogen diffusivity coefficients at a given
temperature and material composition, the choice of original hydrogen level prior to thermal treatment,
and the weld geometry effects on diffusion paths from the source of hydrogen. It is understandable
that these types of procedures are not used more widely because it is very difficult to estimate a
reasonable initial lattice concentration or the threshold bulk hydrogen concentration below which HIC
would be avoided. It is worth noting that codes and standards do not provide guidance on how to
determine either of these parameters.

Despite the complexities in determining exact parameters for a specific service condition, Bailey developed
curves that account for a wide variation in values. These curves provide a minimum and maximum bakeout
(diffusion) time for a specific joint geometry, wall thickness, and bakeout temperature. The weld joint
geometries considered by Bailey are butt welds (labelled as plate) and fillet welds (cylindrical). An example
of Bailey’s quantitative curves is shown in Figure 2.

©2017 by NACE International.

Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 15835 Park Ten Place, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
7
 Figure 2: A sample of Bailey’s hydrogen diffusion curves showing the recommended range of
bakeout times for various half plate thicknesses (L) at 250 oC. The book contains curves for
other weld joint geometries and bakeout temperatures. 19

The creation of the hydrogen removal curves is based on experimental observations that the diffusion
of hydrogen out of metals obeys Fick's second law and can thus be described mathematically by
a differential equation that details the relationship between hydrogen concentration in carbon steel (“C”),
time (“t”) and spatial distribution characterized by a Laplace operator (“∇2”):

𝝏𝑪 𝝏𝑪 𝝏𝟐 𝑪 𝝏𝟐 𝑪 𝝏𝟐 𝑪
𝝏𝒕
= 𝑫𝛁 𝟐 𝑪 or − 𝝏𝒕 = 𝑫 ∙ (𝝏𝒙𝟐 + 𝝏𝒚𝟐 + 𝝏𝒛𝟐 ) (1)

There exist precise solutions of the general equation for many different boundary conditions but for
practical problems, approximate solutions are adequate, easier to handle and relate to simpler geometric
forms.20–22 The resulting curves, covering the temperature range 100 oC to 600oC are built on the following
assumptions:

 Initial weld hydrogen concentration (“C”) – the actual value is not known but defined as
100% level, equivalent to the steady-state concentration of hydrogen in the component.

 Time (“t”) – time factor; related in calculations to the percentage of the original hydrogen
concentration left at the center of various geometrical shapes where hydrogen diffusion takes
place as time, “t”, progresses.

 Diffusion coefficient at given temperature (“D”) – “D” can have numerous values, varying up
to 2 order of magnitude (depending on carbon steel composition and condition). Hence, the
calculation was repeated for both the maximum and minimum possible values of D for ferritic
material to produce a band for the hydrogen removal curve at each temperature and thickness,
as shown in Figure 2. The behavior of carbon steel in practice was assumed to fall within this
band so that its limits could be taken to represent the minimum and maximum efficiencies of
hydrogen removal (bakeout time).

©2017 by NACE International.

Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 15835 Park Ten Place, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
8
  Weld joint geometry – Joint geometry is simplified into: 1) plate geometry for butt welds, and 2)
cylindrical geometry for fillet welds. For complex joint configurations, both geometries should be
calculated or an advanced numerical analysis should be carried out. Cylindrical geometries will
produce shorter bakeout times than a plate geometry, attributed to the increased volume of
material that the hydrogen can diffuse through, reducing the local concentration faster.

 Material thickness – if the thickness of two plates being joined in butt weld or in fillet weld
configuration are of similar thickness, the calculations do not present difficulties. Difficulties arise
when the two plates in a butt weld differ greatly in thickness or when a fillet weld cross-section
is large in relation to plate thicknesses. No general rules can be formulated to deal with every
possibility and where such difficulties occur, judgement must be exercised.

There are difficulties with the use of Bailey’s procedure from a practical perspective. These difficulties
limit the widespread use of the procedure and can be grouped into two major categories:

1. Determination of the initial hydrogen concentration is difficult. For materials exposed to wet H2S
environments, the quantitative data derived from generic process conditions and operating
history to estimate the remaining diffusible hydrogen prior to repair welding would be
impractical.
2. Bailey’s procedure assumes that reducing the lattice hydrogen concentration to a suitable lower
amount could prevent cracking. However, there is currently no consensus on what represents
an ideal lower lattice concentration below which hydrogen cracking would not occur. Hydrogen
induced cracking is a function of the steels microstructure, applied stress on the weld, and
hydrogen concentration. It is therefore difficult to determine an exact lower limit for every scenario.

Bailey also provides several rules of thumbs that may be helpful in simplifying the overall hydrogen
bakeout design process. Most of the recommendations and rules of thumb from Bailey’s work has
been incorporated into BS EN 1011-2(6),13 and both references contain examples that might be useful
for the preparation of a unified bakeout approach for North American codes and standards. Pargeter and
Wright23 have provided an update to Bailey’s work that accounts for significant changes to modern steel
compositions and production methods that have resulted in increased strength, improved weldability,
and lower risk of hydrogen cracking. It is very important to note that Pargeter and Wright’s work may
not be directly applicable to older steels which may have had different manufacturing processes or
weld geometries and component designs which have a higher stress acting on the weld region. Some
of the most relevant conclusions of this work include:

 Standard carbon steels (CE(IIW) ≤ 0.45 wt%) that have been in sour service and may have
increased risk of cracking, in many circumstances, can be welded without “bakeout” but with a
preheat of at least 50 °C or 50 °C in excess of that recommended in BS EN 1011-213 for a
comparable hydrogen-free situation. Welding should be conducted with a heat input of ~0.8
kJ/mm and welding consumables with a hydrogen concentration less than 10 mL/100 g of
deposited metal.

 Low-alloy Cr-Mo steels that have been in hydrogen-rich service at temperatures below which high
temperature hydrogen attack (HTHA) could occur, require a hydrogen bakeout combined with
high preheat, low hydrogen consumables and post-heat to further reduce the potential for
hydrogen cracking upon cooling.

(6)
British Standards Institute, 389 Chiswick High Road, London, W4 4AL , UK
CEN-CENELEC Management Centre Avenue Marnix 17 - B-1000 Brussels, Belgium
©2017 by NACE International.
Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 15835 Park Ten Place, Houston, Texas 9 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
9
 Proposed Methodology to Select Bakeout Parameters

Despite the fundamental diffusion work by Bailey, the above-mentioned limitations hinder the widespread
use of the curves by industry for determining a hydrogen bakeout unique to a specific scenario. The
authors are proposing the following methodology to address the difficulties in determining the initial
hydrogen concentration and the lower concentration limit that avoids cracking. The general process is
shown in Figure 3, while the following sections detail the methodology and provide an example based on
a data point provided by a respondent in the industry survey.
Determining Initial Hydrogen Concentration

Determine partial pressure of hydrogen Correlate hydrogen in the rich gas

Determine maximum possible hydrogen Determine maximum hydrogen
in the rich gas of a component, partial pressure to concentrated
concentration in solution concentration in steel
including operating temperature dissolved species

Hydrogen Bakeout Parameter Selection

Determine the lower hydrogen Determine the reduction necessary to Consider alloy metallurgy, joint Obtain minimum and maximum
concentration limit necessary to avoid reduce the initial steel concentration to geometry, and component design in bakeout time from appropriate curve
cracking the lower limit selecting bakeout temperature from Bailey et. al

Figure 3: Proposed workflow for determining the initial hydrogen concentration and the
required bakeout time.

Calculation of Hydrogen Concentration

Recall that the service environments for which data were provided in the survey responses could
generally be divided into two categories. The authors have focused on the second scenario to
approximate the initial concentration of hydrogen in a component operating in a H2S environment.
There are two primary sources of hydrogen that are of concern for repair welds: hydrogen introduced
by the welding electrodes and hydrogen absorbed into the steel from service. It is an industry best
practice to use low-hydrogen welding practices during weld repairs. Hence, the authors have neglected
the welding contribution of hydrogen to the steel plate from the welding electrodes. To demonstrate the
considerations and methodology for approximating hydrogen concentration, several assumptions were
required:

 H2S is the only hydrogen rich gaseous species above the aqueous media in contact with
steel surface

 No other hydrogen rich species exist in the aqueous media

 Partial pressure of the H2S gas above the aqueous media is known

Under such a situation, the following procedure could be used to estimate a reasonable initial lattice
hydrogen concentration in steel. Hydrogen sulfide gas in equilibrium with an aqueous media will dissociate
as follows:

𝐇𝟐 𝐒(𝐚𝐪) + 𝐇𝟐 𝐎(𝐥) ↔ 𝐇𝟑 𝐎+ −
(𝒂𝒒) + 𝐇𝐒(𝒂𝒒) ↔ 𝑺
𝟐−
+ 𝑯+ (2)

When gaseous species are in equilibrium with water, the concentration of the aqueous species that
develop in water has been studied and well understood for various species. For H2S gas in contact with
pure water, for example, Lee and Mather studied the variation of H2S mole fraction in aqueous media
with H2S total pressure and compiled the curves in Figure 4.24 Equations could be fitted to the curves at
each temperature and pressure in order to calculate the H2S concentration in solution.

©2017 by NACE International.

Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 15835 Park Ten Place, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
10
 Figure 4: Variation of mole fraction solubility of 100% H2S in water with temperature at various
pressures 24

The amount of hydrogen in solution (from the dissociation of H2S) could be obtained by combining the
concentration of H2S in solution with the dissociation constant(s) of H2S in water. The dissociation
constant of H2S is well researched and easily obtained from literature. While the authors are aware that
there are various versions of the dissociation constants available in literature, the variance is usually a
product of the experimental design and the user is at liberty to select any other version that fits the service
conditions. In this study, the dissociation constant as defined by Ning et al.25 is adopted:
𝟐𝟎𝟓𝟔𝟓.𝟕𝟑𝟏𝟓
𝟕𝟖𝟐.𝟒𝟑𝟗𝟒𝟓+𝟎.𝟑𝟔𝟏𝟐𝟔𝑻𝑲 −(𝟏.𝟔𝟐𝟕𝟐𝟐 × 𝟏𝟎−𝟒 )𝑻𝟐𝒌 −( ) − (𝟏𝟒𝟐.𝟕𝟒𝟏 𝐥𝐧 𝑻𝒌 )
𝑲𝜶,𝟏 = 𝟏𝟎 𝑻𝒌
(3)

Although there is a second dissociation constant, Ka,2, associated with the further dissociation of HS-, the
second dissociation constant is much smaller than the first dissociation constant, i.e. Ka,2 << Ka,1s, hence,
the contribution of the second dissociation step to the hydrogen concentration in the aqueous media can
be ignored.

If α is the ionized portion of the H2S, then max H in solution can be expressed as:

−𝑲𝜶,𝟏 ± √(𝑲𝟐𝜶,𝟏 + 𝟒.𝑲𝜶,𝟏 .[𝑯𝟐 𝑺])

𝜶= 𝟐
𝑀𝑜𝑙/𝑙𝑖𝑡𝑒𝑟 (5)

In practice, not all the hydrogen in aqueous solution will migrate into the steel component. The authors
elected to assume all generated hydrogen would pass through or become trapped in the steel through a
diffusion controlled process. Under this condition, the midwall hydrogen concentration can be estimated
from Fick’s second law, assuming flat plate geometry, by:

𝐂𝐱− 𝐂𝟎 𝐱
𝐂𝐬 −𝐂𝟎
= 𝟏 − 𝐞𝐫𝐟 (𝟐√𝐃𝐭) (6)

Where x is half the wall thickness, Cx is the hydrogen concentration at midwall, Cs = max (α1, α2), α1 and
α2 are the roots of Equation 5, Co is the initial lattice concentration of hydrogen before the component is
put into service – which can be assumed to be zero.

©2017 by NACE International.

Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 15835 Park Ten Place, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
11
 A benefit of the Bailey’s approach is that for assets that have been in prolonged service (long enough to
x
give 2√Dt ≤ 0.05 ), the erf function in Equation 6 approaches zero, meaning that it is not important to
know the exact diffusivity coefficient for the alloy when calculating the amount of hydrogen in the lattice.
Also the hydrogen concentration at the center of the wall becomes independent of the wall thickness and
is strictly controlled by the surface concentration of diffusible hydrogen. Although, the egress of hydrogen
during the bakeout process is largely controlled by the actual diffusivity coefficient of the alloy, the user
need not be concerned about this either, since Bailey has accounted for the minimum and maximum
possible values of the diffusivity coefficient in constructing the bakeout curves.

To account for the rate of hydrogen adsorption at the surface, which is strongly dependent on surface
conditions/chemistry, several correlations of lattice hydrogen concentrations to the two species in
solution, H+ and H2S, available in literature could be used. For example, Asahi et al.26 studied the
relationship between lattice hydrogen concentration (HFe), hydrogen concentration in aqueous solution
(H+) and H2S concentration in aqueous solution (H2S) and derived the following correlation:

𝑯𝑭𝒆 = (𝟏𝟑 ∙ ([𝑯+ ] ∙ [𝑯𝟐 𝑺])𝟎.𝟐𝟔 ) ∙ 𝟎. 𝟗 𝑚𝐿/100 𝑔 (7)

Using this correlation, a lattice hydrogen concentration and partial pressure can be calculated as shown
in Figure 5. It must be mentioned that local hydrogen concentration may be higher than predicted by α at
high energy sites due to hydrogen trapping. However, even when a higher level of hydrogen may be
liberated from such sites during welding, some hydrogen will also be lost from the liquid weld pool.
Therefore, any underestimation of hydrogen concentration that may occur should be reasonably
balanced by the amount of hydrogen lost to the atmosphere during the welding thermal cycle.

Figure 5: Variation of diffusible hydrogen concentration in a steel lattice with H2S partial
pressure at various service temperatures

Bulk Hydrogen Concentration Threshold for Cracking Prevention

There is no consensus in literature on a hydrogen concentration threshold below which hydrogen

cracking will not occur. Some of the threshold concentrations that have been suggested in various
sources are summarized in Table 2. When considering new construction, the hydrogen introduced into
welds typically comes from the welding electrode or the welding environment. Generally, sufficient control
©2017 by NACE International.
Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 15835 Park Ten Place, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
12
 is maintained over the welding environment to eliminate any significant hydrogen contribution from the
environment. The hydrogen content in welding electrodes typically ranges from about 4 mL/100 g for low
hydrogen electrodes to 60 mL/100 g for cellulosic electrodes. For the most part, crack-free welding is
routinely performed with both types of electrodes. Therefore, it is reasonable to assume that the lower
hydrogen concentration limit that avoids cracking it is likely between 4 to 30 mL/100 g. The authors
recommend using an initial lower hydrogen concentration limit of 10 mL/100 g.

Table 2: Summary of recommended threshold concentration of diffusible hydrogen in various

codes

Source Suggested Threshold (mL/100 g)

19
Bailey et al. 2-5
ASM International 32 5
NACE SP0296-2010(a) 11 1
BS EN 1011-2(b) 13 3-15
Pargeter and Wright 23 12
(a) Number is not explicitly stated but portions of Section 4.3 may be interpreted as suggesting this value
(b) The exact threshold depends on wall thickness, carbon equivalent, and welding heat input

Diffusivity Coefficient

The hydrogen diffusivity coefficient is usually determined by the Arrhenius form D = D0 . exp(−Ediff/
RT), suggesting that the diffusion coefficient of hydrogen primarily varies with temperature and alloy
composition. Boellinghaus et al.27 showed that the hydrogen diffusivity coefficient can vary over two
orders of magnitude. Hence, unless the value of D has been specifically determined for the steel being
considered for bakeout, choosing an appropriate diffusion coefficient can be difficult. Where diffusivity
values have not been determined, the authors suggest selecting a diffusion coefficient from Smithels
et al.28

Example Calculation of a Bakeout Procedure

The procedure highlighted above was compared against the actual bakeout conditions from one of the
survey responses, shown in Table 3. The threshold concentration of diffusible hydrogen, below which
cracking will not occur, is assumed to be 10 mL/100 grams.

Table 3: Summary of a respondent’s operating conditions and successful bakeout procedure

Product Sour gas with 35% H2S
Service Temperature 40°C
Operating Pressure 4,000 kPa
Thickness 84 mm
Joint Geometry Butt
Bakeout Temperature 260°C
Bakeout Time 8 hours

The calculation of the initial hydrogen concentration was performed as follows:

1. Calculate partial pressure of hydrogen sulfide:

a. 0.35 x 4,000 kPa = 1,400 kPa (1.4 MPa)

2. Using an equation fitted to the appropriate curve in Figure 5, determine the concentration of
H2S in water at 40 °C using the partial pressure of H2S:

a. For 1.4 MPa 

XH2S = 0.017 mol fraction or 0.9351 mol/L
©2017 by NACE International.
Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 15835 Park Ten Place, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
13
 3. Assume no other species that affect the solution chemistry are present

4. Assuming the component was operating in continual service and hydrogen diffusion has reached
steady-state, the hydrogen concentration in the steel is equivalent to the hydrogen concentration
in solution (Equation 6).

5. The pH is approximated by considering only the first equilibrium dissociation of H2S.

Calculate the lattice hydrogen concentration (HFe) curve using Equation 7.

a. [H+]= 0.00039

6. Assuming that all the generated hydrogen diffuses through the steel wall:

a. At a partial pressure of 1,400 kPa on Figure 5 (calculated from Step 5 above), the
initial hydrogen content is 13.98 mL/100 g.

Using the now-determined initial hydrogen concentration, and the assumed threshold concentration limit
of 10 mL/100 g, it can b e calculated that a reduction to 72% of the original hydrogen concentration
might be necessary. However, if 20% of the hydrogen content of the weld is lost to the atmosphere
(less than the 50% suggested by BS EN 1011-213), then only an 86% reduction is necessary.
Utilizing the appropriate bakeout temperature curve from Bailey (Figure 2), it is observed that a bakeout
time of 3.5-11 hours is recommended.

The respondent utilized a bakeout time of 8 hours for their successful repair. This time is near the
median value of the minimum and maximum times as recommended by Bailey19 and calculated using the
proposed methodology in this communication. For reference, without the 20% hydrogen loss from the
weld pool, a bakeout time of 8-29 hours would be determined.

It must be mentioned that the concentrations presented in Figure 5 represent the maximum diffusible
hydrogen present if the component was rapidly quenched from an elevated service temperature. These
numbers can only be representative of the actual lattice concentrations for services near the ambient
temperature. For components that operate at high temperatures, the standard practice during
shutdowns is to cool the component slowly which allows a portion of the hydrogen to diffuse out of the
wall prior to bakeout of repair welding. In such situations, the user should consider using a reasonable
fraction of concentrations from Figure 5.

Closing Remarks

A broad literature review of various aspects of hydrogen embrittlement of carbon and low alloy steel
weldments has been carried out to improve the understanding and limitations of current codes and
standard regarding hydrogen bakeout of in-service components. Current codes and standards cover
some practical and theoretical aspects related to minimizing the likelihood of hydrogen embrittlement of
welds. Some codes and standards provide non-mandatory guidelines leave the final bakeout decision
and parameters to the operator. Other codes and standards provide mandatory rules that must be
followed. However, since neither of these approaches truly quantify the bakeout time with adequate
consideration of the alloy type, wall thickness, component geometry and service conditions, suitable
engineering evaluation is generally still required. Many times, difficulties are encountered and some
measure of trial-and error become inevitable.

The industry survey of Western Canadian oil and gas operators demonstrated there is significant variation
in bakeout procedures despite similar service conditions. In some cases, the same bakeout procedure
could be used on components of widely varying service conditions and wall thicknesses, while other
procedures considered the service conditions and thickness.

©2017 by NACE International.

Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 15835 Park Ten Place, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
14
 To reduce the trial-and-error, the authors have proposed a methodology to either validate an existing
bakeout procedure or the procedure can be used to create specific bakeout parameters for a unique
situation. The methodology considers the service environment, alloy chemistry, wall thickness, and
operating conditions. It is the authors’ opinion that if the type of procedure presented here were adopted,
engineering codes could be infused with hydrogen bakeout recommendations based on first principles.

REFERENCES

1. H. Birnbaum, "Hydrogen Effects on Deformation and Fracture: Science and Sociology". Mater.
Res. Soc., 28, 7 (July 2003): pp. 479–485.
2. M. Dadfarnia, P. Novak, D. Ahn, J. Liu, P. Sofronis, D. Johnson, I. Robertson, Recent
Advances in the Study of Structural Materials Compatibility with Hydrogen. Adv. Mater. 22, 10
(March 2010), pp. 1128-1135.
3. A. Barnoush, Hydrogen embrittlement,revisited by in situ electrochemical nanoindentation.
(PhD Thesis, Universität des Saarlandes, 2007).
4. Gangloff, R. in Environment Induced Cracking of Metals (EICM-2), (ed. Shipilov, S.) 1-24
(Elsevier Science, 2006).
5. R. Gangloff, "Science-based Prognosis to Manage Structural Alloy Performance in Hydrogen",
International Hydrogen Conference, pp. 1-21 (Jackson Lake Lodge, WY: ASM, 2008).
6. M. Nagumo, "Hydrogen related failure of steels–a new aspect", Mater. Sci. Technol., 20, 8
(August, 2004), pp. 940-950.
7. Q. Liu, B. Irwanto, A. Atrens, "Influence of hydrogen on the mechanical properties of some
medium-strength Ni-Cr-Mo steels", Mater. Sci. Eng. A, 617, (August 2014), pp. 200-210.
8. Y. Kim, Y., Kim, D. Kim, H. Choe. "Effects of Hydrogen Diffusion on the Mechanical Properties
of Austenite 316L Steel at Ambient Temperature". Mater. Trans. 52, 3 (March 2011), pp. 507-
513.
9. S. Lynch, "Hydrogen embrittlement phenomena and mechanisms", Corros. Rev., 30, 3 (June
2012) 105-123.
10. API 946 (1981). "Effect of Outgassing Cycles on the Hydrogen Content in Petrochemical-
Reactor-Vessel Steels" (Washington, DC: API Publishing Services).
11. NACE SP0296 (2010). "Standard Practice Detection, Repair, and Mitigation of Cracking in
Refinery Equipment in Wet H2S Environments" (Houston, TX: NACE).
12. ASTM B 850 (2015). "Guide for Post-Coating Treatments of Steel for Reducing the Risk of
Hydrogen Embrittlement" ( West Conshohocken, PA: ASTM International).
13. BS EN 1011-2 (2001). "Welding - Recommendations for welding of metallic materials - Part 2:
Arc welding of ferritic steels" ( London: BSI).
14. ASME B31.12 (2011). "Hydrogen Piping and Pipelines" (New York, NY: ASME International) .
15. ASME PCC-2 (2011). "Repair of Pressure Equipment and Piping", (New York, NY: ASME
International).
16. API 582 (2009). "Welding Guidelines for the Chemical, Oil, and Gas Industries" (Washington,
DC: API Publishing Services).
17. ASME Section IX (2007). "Welding, Brazing, and Fusing Qualifications Qualification" (New
York, NY: ASME International).
18. ASME Section XI. "Rules For In-service Inspection Of Nuclear Power Plant Components -
Article IWA-4000" (New York, NY: ASME International).
19. N. Bailey, F. Coe, T. Gooch, P. Hart, N. Jenkins and R. Pargeter, Welding steels without
hydrogen cracking, (Cambridge, UK: Woodhead Publishing Ltd., 2004), pp. 1-150.
20. Barrer, R. M. Diffusion in and through solids, (Cambridge, UK: Cambridge University Press,
©2017 by NACE International.
Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 15835 Park Ten Place, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
15
 1951), pp. 1-164.
21. Iron and Steel Institute Report No. 14, "Some mathematical considerations on the heating and
cooling of steel", (Washington, DC: AISI, 1936).
22. Darken, L.S.; Gurry, R. W. Physical Chemistry of Metals. (New York, NY: McGraw-Hill Book
Co. Inc., 1953), pp. 1-135.
23. P. Pargeter, M. Wright, "Welding of Hydrogen-Charged Steel for Modification or Repair", Weld.
Res. 89, (February 2010), pp. 34s-42s.
24. J. Lee, A. Mather, "Solubility of Hydrogen Sulfide in Water", Berichte der Bunsengesellschaft
für Phys. Chemie, 81, 10 (October 1977) pp. 1020-1023.
25. J. Ning, Y. Zheng, D. Young, B. Brown, S. Nešić, "A Thermodynamic Study of Hydrogen
Sulfide Corrosion of Mild Steel", Corrosion 2013, paper no. 2462 (Houston, TX: NACE, 2013)
26. H. Asahi, M. Ueno, T. Yonezawa, "Prediction of sulfide stress cracking in high-strength t
ubulars" Corrosion, 50, 7 (July 1994), pp. 537-545.
27. A. Boellinghaus, "Scatterband for Hydrogen Diffusion Coefficients in Micro Alloyed Low Carbon
Structural Steels", Weld. World 35, 2 (March 1995), pp. 83-96.
28. W. Gale, and T. Totemeier. Diffusion In Metals: In Smithells Metals Reference Book, 8th
Edition, (New York, NY: ASM International, 2004), pp. 13-1--13-120.
29. ASME B31.12 (2014). "Hydrogen Piping and Pipelines" (New York, NY: ASME International).
30. NRC Regulatory Guide 1.50 (2011). "Control of Preheat Temperature for Welding of Low-alloy
Steel" (Washington, DC: U.S. Nuclear Regulatory Commission).
31. ABSA AB-519 (2014). "Pressure Piping Alternative Test Methods Procedure Requirements"
(Edmonton, AB : ABSA).
32. W. Becker, R. Shipley, "Failures Related to Metalworking - In: Failure Analysis and Prevention",
Volume 11, (Materials Park, OH: ASM International, 2002).

©2017 by NACE International.

Requests for permission to publish this manuscript in any form, in part or in whole, must be in writing to
NACE International, Publications Division, 15835 Park Ten Place, Houston, Texas 77084.
The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association.
16