Page 1 of 40

Increasing P91 Welding Productivity

with FCAW
William F. Newell, Jr., PE*

ABSTRACT
Benefits associated with using modified 9Cr1Mo (P91) steels in reducing weight, improved
thermal efficiency and saving construction and operating costs are now widely appreciated in
the power generation industry. These advantages can only be obtained if appropriate welding
consumables and processes are available to produce weldments that will complement the
integrity of the completed structures.

At present, shielded metal arc welding (SMAW) and gas tungsten arc welding (GTAW) are
the most commonly used welding processes for fabrication of P91 steels. Because they are
manual processes, productivity is limited. Other processes are available to improve welding
deposition rate and duty cycle. Where use of machine welding is possible, submerged arc
welding (SAW) is generally preferred and perhaps the most productive welding process.
However, for all-position welding and particularly for site welding, the ideal high productivity
process is flux cored arc welding (FCAW). FCAW is well established for welding low
carbon, 1CrMo (P11) and 2CrMo (P22) materials but this is still a relatively new process for
P91 steels. Although a FCAW wire classification is pending (late 2001, early 2002) by The
American Welding Society (AWS) , published performance data for FCAW is scarce.

This paper describes the potential productivity benefits when using FCAW for P91 steels and
offers typical production rates in comparison to other arc welding processes. The suitability
and quality of FCAW consumables and the process is supported by the presentation of the
latest available mechanical testing data, including creep stress-rupture strength, impact and
fracture toughness of the weld metals as it compares with other accepted arc welding
processes. Using the fracture toughness data, a critical crack assessment has also been carried
out to evaluate the acceptability of the FCAW weld metal from a fitness for purpose
standpoint.

*W. F. Newell, Jr. is Co-Founder of Euroweld, Ltd. and President of W. F. Newell & Associates, Inc.
 Page 2 of 40

INTRODUCTION
A variety of welding processes and consumables are available to support fabrication (1,2), but
it is only in the last year that all-position flux cored arc welding consumables have become
available for both site and shop work. FCAW offers significant benefits in terms of both
productivity and welder appeal plus the potential to drastically reduce fabrication times and
costs. In order to achieve the desirable features of such a wire, a particular rutile flux system
has to be used. Most users require reassurance as to the properties of the weld metal and its
fitness for purpose in proposed applications.
The two areas of major concern are:
• Creep properties, in particular creep strength and creep rupture ductility.
• Toughness, where a minimum is specified or there is concern about hydrotesting of
components at high imposed stresses and at ambient temperatures.

The issue of toughness in P91 weld metals has been extensively reviewed in previous papers
(3,4) and it has been argued that toughness is an irrelevant consideration in systems designed
to operate at temperatures in the range of 932-11120F (500-600°C). These temperatures are
far above the range at which any possible risk of fast brittle fracture would be expected.
However, a fitness for purpose approach was adopted for consumables then available and this
has now been extended to cover FCAW weld metal. However, these data are not intended to
cover every design situation or application and a potential user would be well advised to carry
out their own specific analysis. Creep data are also given to provide user/operator confidence
at both ends of the temperature spectrum.

FCAW PRODUCTIVITY & PROCESS COMPARISON

Welding of P91 can be accomplished with FCAW, GTAW, SAW and SMAW processes.
Welding filler metals have been formulated to complement the base metal but do require
multiple refining operations to achieve the low levels of residual elements, especially
phosphorus. A major effort was initiated on a worldwide basis to formulate weld metals that
would exhibit friendly weldability while maintaining the required mechanical properties.
Combustion Engineering formulated and tested nearly 200 different SMAW compositions in
the original test program – only two or three of the compositions exhibited both satisfactory
mechanical properties and welder appeal. (5)

Welding process selection is normally a funciton of thickness, diameter, position and

quantity. Deposition rate is often used as a measure of potential productivity, although many
other factors contribute to operator duty cycle and hence productivity. Table 1 illustrates
typical production rates as a function of welding process. Clearly, the SAW process offers
significant potential where it can be used, with FCAW following next.
 Page 3 of 40

Table 1. Typical Manhours for Pipe to Pipe Welds as a Function of Welding Process (37.5
degree V-groove, Schedule 40). (2,5-8)

Welding Manhours per Pipe Size, Diameter

Process 3-inch 6-inch 12-inch 24-inch
GTAW 0.7 1.9 6.4 27.5
SMAW 0.5 1.0 2.8 12.1
FCAW 0.2 0.3 0.8 3.5
SAW N/A 0.1 0.3 1.4

NOTE: Manhours based on average operating factors and deposition rates. Fit-up and groove
preparation not included. Estimates based on using single process from root to cap.
Gas Metal Arc Welding (GMAW)
The situation with GMAW, particularly with active gas mixtures, is more complex because of
the variable recovery of key elements such as Mn, Si, and Nb/Cb. Modifications to the
compositions of solid or metal cored wires are similar to those applied to covered electrodes
and these are beneficial for microstructural control even though toughness may far exceed
many specification requirements.

It must be strongly noted that qualification and use of solid wire GMAW should be
approached with much caution! The “-B9” composition is lean on deoxidizers, so
important to proper operation and results with GMAW. Because of this, wetting action is
reduced and the preponderance for lack of fusion type defects and oxide inclusion content
affect the ability to perform successful welding. A few fabricators have qualified GMAW,
but few have implemented it into production because of it’s operator specific characteristics
and inability to perform in a reproduceable manner.

Gas Tungsten Arc Welding (GTAW)

Weld metal deposited with GTAW typically exhibits far greater toughness than weld metal
deposited by processes using flux and slag systems. (e.g. FCAW, SMAW or SAW). This
variation is explained by a significant difference in weld metal oxygen content and the
increase in the populations of non-metallic inclusions. GTAW weld metal made with solid
wires typically contains less than 100ppm oxygen compared with 400 - 800ppm for the fluxed
processes. Although much slower than other processes, the GTAW process still provides
weld deposits with the highest integrity. (9)

It is recommended that rod diameter be restricted to 3.2mm (1/8-inch) maximum for manual
GTAW. Insufficient heat is available to implement interbead tempering with the puddle size
associated with the larger diameter rods. (9)

Flux Cored Arc Welding (FCAW)

The operability of the GMAW process using a fabricated wire (FCAW) is strongly influenced
by the type of the shielding gas. In general, a suitably high content of CO2 in the gas is
beneficial. However, higher CO2 levels will normally increase the weld metal oxygen level
which has been shown to be detrimental to impact toughness, as illustrated by Figure 3,
except for those FCAW wires that have been formulated to operate with either Ar-CO2 (80-20
 Page 4 of 40

or 75-25) or 100% CO2 shielding Results with 100% CO2 have actually shown somewhat
better toughness. This is believed to be due the the higher penetration and thus greater
interbead tempering action of previously deposited weld beads. (9)

Flux cored wire, 1.2mm (~0.045-in.) diameter, is capable of a deposition rate which is
competitive or exceeds all other arc welding processes except SAW (2). This advantage is
particularly notable for in-position welding. Compared with solid wire gas metal arc welding
(GMAW), a faster burn-off rate for tubular FCAW is also promoted by higher current density
at the wire tip and I2R resistance heating of the wire extension from the contact tip.
Moreover, the flux cored wire process, which can utilize spray transfer, produces reliable
fusion and penetration in all welding positions. The duty cycle possible with the FCAW
process is also higher than for the GTAW and SMAW processes, which further improves
potential productivity when compared to these processes. The better duty cycle can be
attributed to two main factors: the continuous nature of the process and the all-position
capability of the process without the need for a change in welding parameters. For some
applications, especially numerous short welds, the duty cycle of the FCAW process may also
compete with SAW if the set-up times and positioning of the joints into the flat position
contribute a significant proportion of the time. The ability of FCAW to weld thick section
joints relatively quickly in all positions may allow the FCAW process to compete with SAW
in these situations.

The FCAW process is expected primarily to replace the SMAW process. The GTAW process
will still be required for pipe roots and other small diameter or thin wall pipe, and the SAW
process will be preferred for very thick section welds that can be rotated or manipulated into
the flat position.
The FCAW process is mainly used in the hand held semi-automatic mode, which provides
optimum adaptability and ease of use for both shop and site welding. For joints which lend
themselves to mechanisation the productivity of the FCAW process may be further improved
by the use of suitable automated equipment, Figure 1.
 Page 5 of 40

Figure 1. Live demonstration of automatic orbital welding of a fixed 5G (double up) P91
pipe using FCAW. (Photo courtesy of Euroweld, Ltd. and Liburdi Dimetrics.)

Shielded Metal Arc Welding (SMAW)

Application of the SMAW process is well understood. In most cases, acceptable mechanical
properties are easily achieveable, provided that the electrode formulation is consistant with
the “-B9” composition (Ni + Mn < 1.5) and sufficient welder skill is available.

Greater success has been observed with the “-15” type electrode coatings than other varieties.
Primary factors influencing this success are enhanced compositional control of tramp
elements and better control of bead shape plus interbead tempering during welding as
compared to the iron power type formulations. For example, when using iron powder
additions, deliterious tramp elements or elevated phosphorous can “come along with the ride”.

Submerged Arc Welding (SAW)

Submerged arc welding of P91 materials is readily accomplished by the use of automatic,
machine or semi-automatic processes using both constant current or constant voltage power
sources. When using a semi-automatic apparatus either for semi-automatic use or for an
adaptive machine type use, a big disadvantage is the stiffness of the B9 welding material
when feeding through the sharper turn of the standard semi-automatic heads. Two possible
solutions to this problem are available, either a reduction in the size of the wire or a double
annealing manufacturing process of the wire, which is available in some consumables. When
using the standard machine type head this stiffness translates into a greater need to adjust the
straightening rollers of the SAW apparatus to correct for the helix or cast of the wire. This
also affects contact tube/tip life. (2)
 Page 6 of 40

FCAW FORMULATION
To be useful, a tubular flux cored wire must exhibit excellent operability in all positions for
site welding. In these situations, simple control of the arc, smooth weld metal transfer, flat
bead profiles with minimum spatter and easy slag removal are all essential requirements.
Such a combination of features is imperative for high productivity welding and can only be
achieved using flux cored wire with a rutile (TiO2) based flux system. However, there are
two potential disadvantages associated with this flux system: (9)

• Rutile flux systems have a lower refining capability than classical basic systems resulting
in somewhat higher oxygen content (typically 600ppm for rutile FCAW deposits
compared with 400ppm for submerged arc welds made using basic fluxes (3,4)).
• Rutile flux systems utilize naturally occurring rutile sand as a major ingredient. This is
contaminated to a small degree with niobium and vanadium which in turn results in some
alloy pick up (10), but this is normally hidden because of deliberate additions of these
elements. However, of greater possible importance is the pick up of titanium into the
weld metal from the rutile system. Titanium provides a strong carbide former and yet
more matrix strengthening. FCAW weld metals are generally about 5-10% stronger at
ambient temperature than weld metals from SMAW and SAW processes plus are similar
to those of a GTAW deposit after similar PWHT. The corresponding toughness is
generally lower but creep rupture strength has been seen to be higher than SAW and
SMAW. To mitigate the effects of titanium pick-up (typically 0.02-0.04%Ti), the level
of Nb is deliberately controlled to the minimum consistent with meeting weld metal
specifications. As noted above, a proportion of this Nb is also derived from the rutile
flux system.

WELD METAL COMPOSITION

Various approaches to the weld metal composition have evolved since the original ORNL/CE
projectfunding efforts from EPRI and others. In general terms, those elements which are
beneficial for improving creep performance are detrimental in terms of toughness, for
example: Nb/Cb, V and to a lesser extent N and Si. A balanced composition or alloy that
restricts delta ferrite formation but results in a fully martensitic microstructure helps to
contribute both-optimum toughness and creep performance.

From the above discussion, it can be seen that FCAW can offer not only significant
productivity benefits but also welder-friendly operability, particularly in fabrication positions
where other high productivity processes are not applicable. Nevertheless, it has also been
recognised that these benefits can only be exploited if the deposit composition, hence
microstructure is carefully controlled to achieve a reasonable balance of mechanical
properties - primarily toughness and creep resistance.
 Page 7 of 40

In the design of the flux cored wire, the deposit composition was aimed to be as close as
possible to the requirements of the corresponding SMAW weld metal (e.g. AWS E9015-B9).
The next revision of AWS A5.29 specification for low alloy flux cored wires will include this
grade, and the expected classification for an all-positional wire will be E10XT1-B9. AWS
specifications are ultimately included in Section II Part C of the ASME Boiler and Pressure
Vessel Code.

Table 2 presents the typical all-weld metal composition of a selected FCAW deposit. This
composition is typical of a deposit made using usual Ar-20%CO2 shielding gas. However,
selected FCAW wires are formulated to work with either Ar-15-25%CO2 or 100%CO2
shielding gases. Only minor changes in composition are typically observed.

Table 2. Chemical composition of Supercore F91 weld metal

Elements C Mn Si S P Cr Ni Mo Nb V N
wt% 0.10 0.8 0.3 0.010 0.015 9.5 0.6 1.0 0.03 0.20 0.05

To obtain a proper balance between fracture toughness, creep-rupture strength and resistance
to long term embrittlement, the alloy composition and residual elements must be controlled to
provide a single phase microstructure and avoid delta ferrite. The ORNL/CE programs found
that by keeping the Chrome Equivalent (CE = Cr + 6Si + 4Mo + 1.5W + 11V + 5Cb + 9Ti +
12Al – 40C – 30N – 4Ni – 2Mn – 1Cu), as adjusted per their work, below 10, the tendency to
form delta ferrite is reduced. This number is not absolute, but provides a good guideline since
elevated delta ferrite in this material reduces its toughness. Even materials with a CE between
10 to 12 exhibited adequate toughness provided that the delta ferrite does not exceed five (5)
percent. (10-12)

It is widely recognized that nickel can be useful for improving weld metal toughness. The
addition of a controlled level of nickel is beneficial for two reasons. It lowers the AC1
transformation temperature bringing this closer to the post weld heat treatment (PWHT)
temperature and this improves the response to tempering. It also eliminates the possibility of
residual delta ferrite being present which is undesirable because of its poor creep resistance
and potentially adverse effect on toughness. However, excessive levels of nickel, exceeding
1% are also detrimental. The AC1 may be so low that PWHT at the top end of the
temperature range could cause some austenite to form which in turn transforms to fresh
untempered martensite on cooling. Most fabricators prefer to perform PWHT at least 50oF
(320C) below the transition temperature to allow room for minor temperature excursions.
Excessive nickel also contributes to degradation of creep properties by changing the optimum
long-term evolution of carbide precipitation during service. Nickel is therefore usually
controlled in the range 0.4 - 1.0 wt.% in Europe. The original ORNL work that suggested a
maximum of 0.4 wt.% was thought to perhaps be overly restrictive. Table 6 and Figure 6
compares transition temperature with nickel and manganese content. However, recent work
at ORNL examining the martensite start/finish temperatures has identified a potential and
 Page 8 of 40

significant correlation of these temperatures to nickel content. Where nickel is elevated

toward the upper end of the range, approximately 18% of the austenite is retained at 2040C
(4000F) and the Mf may actually be below room temperature! In most cases, it is
recommended to allow the component to cool to ~1000C (~2000F) before PWHT to minimize
the amount of retained austenite in the weldment. (13) Research continues in this area.

Effect of Ni + Mn on AC1
P91 Weld Metal
1600 870

850
1550

830

AC1, Degrees C
AC1, Degrees F

1500
810

790
1450
1436

770

1400
750

1350 730
0.00 0.50 1.00 1.50 2.00 2.50

Ni + Mn

Figure 2. Effect of Ni+Mn on the AC1.

Table 3. Calculated and measured lower critical temperatures (AC1) as a function of Ni

+ Mn . (14-16)
Ni Mn Ni + Mn AC1b 0F [0C]
0.3 0.51 0.81 1490 [807]*
0.33 0.49 0.82 1499 [816]
0.70 0.37 1.07 1458 [793]
0.71 0.43 1.14 1465 [796]
0.52 0.66 1.18 1454 [789]*
0.70 0.65 1.35 1468 [799]
0.63 0.76 1.39 1436 [778]*
0.79 0.60 1.39 1467 [797]
0.8 1.05 1.85 1416 [769]
1.0 1.3 2.3 1380 [749]
* ORNL Calculation
 Page 9 of 40

Variations in vanadium, carbon and nitrogen have been found to have smaller influences on
toughness. Manganese is typically controlled at a higher level than the parent material to aid
deoxidation and provide a sound weld deposit. However, some users, such as GEC- Alstom,
limit Mn + Ni to 1.5% maximum as a safeguard against austenite reformation at the highest
PWHT temperatures. Even with Ni + Mn at 1.5 wt.%, Figure 2 suggests that the lower
critical temperature could be as low as 7800C (14360F). Standard consumables are normally
manufactured within this limit, but to ensure sufficient manganese for effective deoxidation,
the nickel level is lowered to about 0.5 wt.%. However, the average toughness is usually
somewhat lower. Toughness can be further affected when users specify that both Mn and Ni
in the weld metal must be in accordance with base metal limits. In order to be sure what
effect the PWHT may have on the composition, the ACTUAL composition should be known.
Therefore, welding consumables should be procured with actual chemical compositions; e.g.
Certified Material Test Reports (CMTR’s) if domestic and/or an EN10204 3.1B certificate
from Europe.
What is unknown at this time is the effect of only having a minor portion of a weldment, such
as the root and hot pass, at one composition or nickel level and the remainder of the weldment
completed with another or lower nickel composition.
Silicon is an essential deoxidant required in both parent and weld metal. Combined with
chromium, silicon may also contribute, to a minor degree, to the alloy’s oxidation resistance.
However, although some specifications have effectively the same range as P91 parent
material (0.20 - 0.50%Si), a low level of silicon benefits weld metal toughness. The AWS
specification limit of 0.30% is lower than the parent material and is perhaps too restrictive for
certain consumables, particularly bare wires used with gas shielded processes. Chemical
composition, particularly deoxidation tendency of the wire (Figure 7.), also has a significant
effect on the welding operability of the GMAW process. A silicon level of about 0.35 wt.%
appears to be necessary for satisfactory operating behavior of the metal cored wire and up to
0.50 wt.% for FCAW wires. These silicon levels have not introduced any known problems in
bare wires, SMAW or SAWused in Europe. In fact, silicon levels in bare wire between 0.3
and 0.35 actually aid deoxidation, wetting and manipulation of the puddle for GTAW. The
proposed revision of AWS A5.29 to incorporate “-B9” FCAW, expands the silicon range up
to 0.5 to further enhance deoxidation. (3,4,17,18)

Control of sulfur, phosphorous and residual elements is important. By observing the AWS
"-B9" 0.010 maximum wt.% for sulfur and phosphorus, problems including crater cracking,
maintenance of toughness after PWHT or other undesirable grain boundary effects can be
avoided in SMAW, GTAW and SAW. It has been found that consumables exhibiting
manganese to sulfur ratios greater than 50 provides one "rule of thumb", when combined with
low phosphorous, to avoid crater cracking phenomena. By requiring low phosphorus levels,
other elements in trace quantities that "come along for the ride" and that have been shown to
be detrimental are also reduced. But, some sulfur (~ 0.001 wt.%) and phosphorus (~ 0.002
wt.%) are necessary to promote proper wetting action of the molten puddle where bare wire
and gas shielding is used. Elevated levels of phosphorus up to 0.02 wt. % have been found to
be permissible for FCAW wires because of the interaction of their slag formers and cooling
rates on the deposited metal. Data also suggests that SMAW can also tolerate this level of
phosphorus (approved in European specifications). Where carbon and niobium are both
 Page 10 of 40

toward the upper end of their respective ranges, tolerance for phosphorus, sulfur and other
trace elements is significantly reduced and may result in crater cracking, hot cracking or other
undesirable grain boundary phenomena. As with other chromium-molybdenum consumables
used in critical service, requiring that P91 consumables meet an X Factor < 15, as calculated
with the Bruscato Formula (or "X-Factor"; where, X = 10P+5Sb+4Sn+As), would be prudent
as this approach essentially reduces the presence of other problematic constituants.

Figure 3. Effect of Oxygen Content on Toughness for Selected P91 Weldments. (16)

TENSILE PROPERTIES

Tensile Strength and Hardness (9)

All-weld metal tensile tests were carried out using standard full-sized specimens with a
nominal gauge diameter of 10mm. Typical ambient tensile strength along with the mid-weld
section hardness properties are presented in Table 4.
 Page 11 of 40

Table 4. Ambient temperature tensile/hardness properties of the FCAW weld metals. (9)
PWHT Rp 0.2% Rm A4 Z Hardness (mid-weld section)
°C/h MPa MPa % % HV10
760/2 690 809 20 52 264
760/4 651 777 23 58 250

Data from the above table indicate that the tensile properties of the FCAW weld metal
satisfactorily meet the requirements of the appropriate specifications for the P91 weld metals.
Compared with other processes, the differences are that the tensile strength of weld metal is
slightly higher than that of the SMAW and SAW deposits and very close to that of the GTAW
weld. The elongation is very close to the values achieved by other processes while its
reduction of area is slightly lower.

Hot Tensile Strength (9)

Without doubt, high temperature properties are the most important features for the P91 weld
metals. Although hot tensile tests are not representative of service conditions for P91 steel
components owing to the short duration of the test, they provide a convenient method for
comparison of weld metals with base material data in a short term test. All-weld metal hot
tensile tests were carried out at temperatures of 550, 600 and 650°C. Prior to the tests, the
weld coupon was subject to a PWHT of 760°C×2h+FC. Table 5 lists the test results and
Figure 4 compares these data with SMAW values and base material requirements.

Results indicate that the hot tensile strength of selected FCAW weld metal is comparable with
weld metals from other well established processes and significantly higher than the minimum
requirements for the base material.

Table 5. Hot tensile properties of selected FCAW weld metal. (9)

Test temperature, °C Rp 0.2%, MPa Rm, MPa A4, % Z, %
550 402 495 18.0 71
600 277 405 30.5 83
650 182 292 25.5 87
 Page 12 of 40

Temperature, o F

480 570 660 750 840 930 1020 1110 1200 1290
600
80

500
0.2% Proof Stress, MPa 70

0.2% Proof Stress, ksi

60
400

50
300
40

200 30

20
100 BS1503 Base Material Min.
SMAW (Chromets) 10
FCAW (Supercore F91)
0 0
250 300 350 400 450 500 550 600 650 700

Temperature, o C

Figure 4. Hot tensile property comparison of selected FCAW weld metal with SMAW
deposits and base material minimum requirements. (9)

Creep Considerations (9)

However, encouraging creep rupture results have been observed with reduced niobium levels,
down to 0.025 to 0.03, from the selected FCAW wire. The higher rupture stress of the FCAW
weld metal is believed to be caused by the presence of residual titanium. Since it forms stable
carbonitrides like niobium, a positive influence was anticipated, and the intentional reduction
of Nb was justified. No previous reports of the effects of Ti in P91 weld metal are available,
but a similar level (200ppm Ti) was found to have no effect on 650°C rupture life in a low
nitrogen 11%Cr rotor steel, and contrary to the weld metal here, fracture appearance transition
temperature (FATT) was noticeably reduced. Early results suggest that this approach may
even provide enhanced performance as shown in Figure 5.

Greater possible importance is the pick up of titanium into the weld metal which provides a
further strong carbide former and provide more matrix strengthening. FCAW weld metals are
generally about 5-10% stronger at ambient temperature than weld metals from SMAW and
SAW processes and are similar to those of GTAW deposit after similar PWHT. The
corresponding toughness is generally lower. To mitigate the effects of Ti pick-up (typically
0.02-0.04%Ti), the level of Nb is deliberately controlled to the minimum consistent with
meeting weld metal specifications. A small proportion of this Nb is also derived from the
rutile flux system.
 Page 13 of 40

There is general agreement that the failure mode of weldments is ultimately controlled by
HAZ behavior, but currently there is no consensus as to the optimum choice of weld material
properties to delay such failure and ultimately extend component life. Creep resistance
behavior can be generically categorized as shown in Table 6.

Table 6. Grade 91 Weldment Performance Test Behavior. (14,17,19-21)

Temperature, oC (oF) Performance – Creep Results

550 (~960) Base Metal, Weld and HAZ Have Comparable Properties
600 (~1050) Base Metal and Weld Have Similar Properties; HAZ ~ 20% Less
650 (~1200) Weld and HAZ Have Slightly Inferior Properties

400

+20% 50
565°C/10 5 h
300 P = 29.33
40

-20%
200 600°C/10 5 h 30
P = 30.56
Rupture Stress, MPa

Rupture Stress, Ksi

20

100
90
80
70 P91 base material 10
60 Chromet 9-B9 9
Supercore F91 8
50
25 26 27 28 29 30 31 32

P = K(30+Logt)x10 -3

Figure 5. Larson-Miller plot of all-weld metal stress-rupture test results at 550-660°C for
FCAW and SMAW . (9)

The reported tests have shown that the elevated temperature proof and rupture stress values of
the FCAW multipass weld metal lie within the envelope required for equivalent parent
material. These properties were similar to or even higher than those obtained for the SMAW
weld metals and were considered satisfactory, since failure in transverse tests on weldments
occurs characteristically at the HAZ Type IV zone in the long term unless weld metal creep
strength is severely compromised. (9)
 Page 14 of 40

Creep Stress-Rupture Properties

To provide an assessment of the creep strength of the FCAW weld metal, a series of 3 pairs of
stress-rupture tests were carried out at temperatures initially set at 550°C, 600°C and 650°C,
aiming for rupture at nominally 100h and 1000h at each temperature. These would present a
useful range of values to populate a Larson-Miller curve. For comparison, a similar test
matrix was also included for a representative P91 SMAW electrode (E9015-B9).

All-weld test specimens of 8mm gauge diameter and 40mm gauge length were extracted from
the mid-section of weld coupons after a PWHT the same as used for the CTOD tests. The
testing was conducted under constant load conditions.

The short-term tests were loaded at stresses for expected rupture life of 10-100hrs based on
the median for P91 parent material. These initial tests indicated that the FCAW weld metal
was noticeably stronger than the SMAW, so the stresses were appropriately adjusted for the
longer term tests. The 650°C test was also raised to 660°C to increase the parametric value at
this point. Results for the FCAW weld metal are given in Table 7, and these are presented
with the SMAW and base material data on the Larson-Miller plot, Figure 5.

Table 7. Stress-rupture test results of P91 FCAW weld metal.

Test temperature, °C Stress, MPa Rupture time, h A5, % Z, %
295 308 25.0 76.3
550
265 2782 13.4 42.3
205 287 20.9 61.9
600
175 1361 5.6 15.2
650 140 101 14.0 24.4
660 100 535 7.5 17.9

TOUGHNESS CONSIDERATIONS
Four factors have a significant influence on weld metal toughness:
• Composition
• Post weld heat treatment time and temperature
• Welding process
• Microstructural effects (heat input, bead size, sequence and welding position)
• Toughness Testing Temperature

Mechanical property results vary with welding process. This is especially true for processes
that rely on fluxes and slag systems for alloying and/or shielding gases for protection of the
molten weld pool. Primary factors that can affect weld metal properties include: heat input
 Page 15 of 40

versus amount of weld metal deposited, grain size produced versus bead shape, and influences
from wetting agents, crack inhibitors, deoxidizers and slag formers as they affect gas levels,
microalloying and introduction of tramp or residual elements for the SMAW, FCAW and
SAW processes.

Reductions in mechanical properties, primarily toughness, as a function of welding process

are summarized in Table 8.

Table 8. Toughness Properties as a Function of Welding Process (5)

(GTAW = 100% of achievable properties)
Welding Typical Wetting Agents, Crack
Process Toughness Inhibitors, Deoxidizers or
Reduction Slag Formers

GTAW None Si, Mn, Ti

SAW 25-50% SiO2, MnO3, etc.
SMAW 40-50% CaCO2, etc.
FCAW 50-70% Varies

It has been argued that weld metal toughness is an irrelevant consideration for components
which are designed to operate at temperatures in the range 500-6000C (932-11120F) - far
above the range at which any possible risk of fast brittle fracture could occur. However, there
are situations where components might be pressurized or loaded at ambient temperatures
during hydrostatic testing, construction, or start-up. To handle such conditions, most consider
that the weld metal should have a minimum toughness at + 200C (680F). The American
Welding Society (AWS) filler metal specifications do not specify impact requirements for "-
B9" filler metals, but the non-mandatory appendix to AWS A5.5-96 (covered electrodes)
proposes that a suitable test criterion should be agreed upon between the purchaser and
supplier. Conversely, the recently introduced European specification EN 1599: 1997 requires
a minimum average of 47J (34.7 ft-lbs.) with a minimum single value of 38J at + 200C (28 ft-
lbs @ 680F). These values coincide with some internal corporate specifications from those
whom have imposed toughness criteria and required values in the range 20-50J at 200C (14.8-
36.9 ft-lbs @ 680F) after PWHT. Determination of adequate toughness became an issue,
particularly because of early work with the FCAW process. (4,18)

Toughness Testing Temperature

It must also be noted that the test temperature used when conducting toughness tests can have
a dramatic effect on the results. (3,21) Experimental work has shown that the transition
temperature for this alloy hovers around room temperature. Up to a 30 to 50 percent increase
in toughness can be observed when tests are performed at 22-230C (72-750F) as opposed to
200C (680F)! (2,15)
 Page 16 of 40

Bead Shape & Position Effects

Bead shape and welding position have a significant effect on toughness. Most test data is
reported in the flat welding position (1G). Bead shape plays an important role as shown in
Figure 6. Thin, wide beads that permit some degree of tempering from the heat of welding
induce grain refinement on previously deposited metal. Thicker bead cross-sections minimize
this effect. Loss of toughness is normally observed in the vertical (3G) and overhead (4G)
positions with SMAW– mostly due to the bead shape in the weldment. Even with an elevated
weld metal nickel content (0.76 wt.%) to enhance toughness, the effect of position and
corresponding bead shape and inability to provide tempering or induce grain refinement is
evident and is as shown in Figure 7. However, selected FCAW wires have shown improvent
in toughness in out-of-position welding. Increased interbead tempering is believed to be the
cause. Plus, several manufacturers have formulated their E9015-B9 SMAW electrodes to
provide satisfactory toughness with nickel levels <0.4 wt. %. (9)

Figure 6. Influence of Bead Shape on Weld Metal Toughness for

Typical FCAW and SAW Applications. (9,19,22)
 Page 17 of 40

Figure 7. Effect of Welding Position on Toughness for SMAW. (19)

Charpy Impact Toughness (9)

To cater for fabrications which require hydrotesting, it is generally agreed that the P91 weld
metals should provide a minimum toughness at ambient temperature. Charpy impact tests
were carried out using full-sized 10×10mm specimens notched at the weld centre. Impact
energies of the FCAW weld metal after different PWHT times are given in Table 9. Figure 8
illustrates the effect of PWHT procedure (time (t,h) and temperature (°K)) on the weld
toughness.

Table 9. Typical average Charpy impact toughness of selected FCAW weld metal.
Pre-heat/interpass temperature, °C PWHT procedure Charpy energy @20°C, J
760°C/2h + FC 25
250 760°C/4h + FC 30
760°C/8h + FC 35

Comparing the above data with those achieved by other flux shielded processes, namely
SMAW and SAW (4), the FCAW deposit, as expected, produced somewhat lower impact
 Page 18 of 40

energy values. However, as illustrated by Figure 8, slightly higher PWHT temperature or

longer soaking times are beneficial in improving impact toughness.

Table 10. CTOD and KQ values of selected FCAW weld metal. (9)
PWHT procedure Test temperature, °C CTOD, mm KQ, MPa m
0.021 75.10
20 0.018 61.80
0.030 76.79
760°C×2h+FC
0.029 69.26
0 0.021 55.75
0.025 66.79

50
Metrode tests
Fabricator tests
40
Impact energy @20/22°C, J

30

20

10

0
20 20.5 21 21.5 22

P = °K(logt + 20)10-3
Figure 8. The effect of PWHT on selected FCAW weld metal toughness. (9)

Fracture toughness (CTOD) (9)

Fracture toughness CTOD tests were conducted in accordance with BS7448 (23). In order to
examine a worst-case microstructural condition, the PWHT procedure chosen for the weld
metals was 760°C×2h+FC, which gave the lowest absorbed energy values in Charpy tests, as
shown in Table 10. The dimensions of the CTOD specimens are illustrated in Figure 9. All
 Page 19 of 40

specimens were B×B where B is the plate/weld thickness, and were notched through the
thickness of the weld from the top. Based on a minimum water inlet temperature of 7°C for
hydrotesting, the tests were carried out at two temperatures, namely 20°C and 0°C. The
results in terms of CTOD (δc) and KQ (provisional value of KIC) are shown in Table 11.
The results indicate that the CTOD values for the FCAW weld metal were in the range of
0.018mm to 0.030mm, with small but probably insignificant variation between the values at
20°C and 0°C. These toughness results provide information that enable a maximum tolerable
flaw size to be established for real structural applications.

1.70
7.60

Root radius 0.1 max

60° Nominal 20

130 130 20
80 80

Figure 9. CTOD specimen dimensions. (9)

Tolerable Flaw Size Determination (9)

To answer the above question, calculations of the maximum tolerable flaw sizes were carried
out using TWI’s Crackwise® software (24), which automates engineering critical assessment
procedures set out in BS7910 (25). The model chosen was that used in the previous work (4),
namely a fabricated header of 450mm outside diameter and 50mm wall thickness, as shown in
Figure 10. The design conditions are taken to be 176 bar at 580°C and hydrotest conditions of
1.25 times design pressure at ambient temperature. This ensured that a comparison could be
made between the FCAW and SMAW weld metals under similar conditions.
 Page 20 of 40

To assess the worst case toughness condition, the lowest measured CTOD value, namely δC =
0.018mm at 20°C, was used in the Crackwise® calculations. The results indicate a maximum
tolerable surface flaw size of 125mm in length and 12.5mm in depth for a longitudinal seam
weld, i.e. equal to ¼ of the wall thickness (Figure 10). The corresponding failure assessment
diagram is given in Figure 11 while Figure 12 illustrates the effect of primary membrane
stress on the maximum tolerable flaw depth. In Figure 11, any point which falls inside the
failure assessment line can be considered safe whereas any point outside the line is potentially
unsafe (26). The results indicate a reasonable defect tolerance despite the relatively low
fracture toughness.

Flaw
450
125m
Plan
12.5m

Transverse 125m
12.5m

Longitudinal

Figure 10. Schematic showing header with maximum tolerable surface breaking flaw in
longitudinal seam weld. (9)
2.0
Supercore F91 weld metal Based on CTOD = 0.018mm,
PWHT: 760°Cx2hrs+FC the lowest single value found
Rp 0.2% = 690MPa in tests at 0°C and 20°C
1.5 Rm = 809MPa
Kr

1.0
Unsafe area

0.5
Safe area

0.0
0.0 0.5 1.0 1.5 2.0

Lr

Figure 11. Failure assessment diagram for selected FCAW weld metal. (Lr is a
dimensionless number showing the ratio of applied stress to the yield strength of
the material, while δr is the ratio of applied stress intensity to material toughness,
CTOD (δc) (26).
 Page 21 of 40

20
Based on CTOD = 0.018mm,

Flaw depth (constraint), mm

the lowest single value found
in tests at 0°C and 20°C
15
Unsafe area Supercore F91 weld metal
PWHT: 760°Cx2hrs+FC
Rp0.2 = 690MPa
10 Rm = 809MPa

5 Safe area

0
0 100 200 300 400 500
Primary membrane stress, MPa

Figure 12. Effect of primary membrane stress on maximum tolerable flaw depth
of selected FCAW weld metal. (9)

components owing to the short duration of the test, they provide a convenient method for
comparison of weld metals with base material data in a short term test. All-weld metal hot
tensile tests were carried out at temperatures of 550, 600 and 650°C. Prior to the tests, the
weld coupon was subject to a PWHT of 760°C×2h+FC. Table 5 lists the test results and
Figure 8 compares these data with SMAW values and base material requirements.

From the above presented data, it can be seen that the toughness of the FCAW weld metal is
slightly lower than those from other well-established fluxed processes, namely SMAW and
SAW processes. However, actual welding procedure qualifications by fabricators and
contractors indicated an increase in impact toughness values on test assemblies welded in both
the flat and vertical positions (Figure 8). This may be attributed to a higher degree of
interbead tempering (27). Further, as mentioned earlier, additional toughness increases have
been observed by increasing the temperatures (up to 775°C) and/or soaking times.
Compositional control permits PWHT at these elevated temperatures while leaving sufficient
margin to ensure that the Ac1 is not approached or exceeded.

Nevertheless, the key question is: Are these weld metals still tough enough and fit for
purpose? The productivity benefits, up to 50% reduction in welding time, are such that flux
cored wires are now being used by both pipe fabricators and on-site contractors (27). Because
of the special features associated with this flux cored wire, it is probably unreasonable to
expect the toughness to match that of SMAW deposits. However, in the light of the fitness
for purpose calculations, it would appear that some reduction in toughness can be tolerated.
 Page 22 of 40

PREHEAT & POSTWELD HEAT TREATMENT

Application of elevated preheat and PWHT, including interpass temperature controls, are
absolutely necessary with Grade 91 weldments, regardless of diameter or thickness. The
literature suggests that 200oC (~ 400oF) is adequate for preheating P91 weldments.
Fabricators typically aim for 250oC (~ 500oF) but will go as low as 150oC (~ 300oF), for root
and hot pass layers only, thin-walled components or where GTAW is utilized. (28,29)

Proper tempering of the martensitic microstructure is essential for obtaining reasonable levels
of toughness. In practice this involves selecting both an appropriate temperature and time.
The AWS specification for consumable classification requires PWHT of 730-760’C (1346-
1400 0F) for 1 hour. This time requirement is inadequate for normal fabrication procedures.
Table 11 shows how mechanical properties vary with time at temperature. A minimum of 2-
3h at temperature in the range 750-7600C (1382-1400 0F) is required, or longer for thicker
sections. This temperature-time aspect is recognized by EN 1599 which specifies a PWHT
requirement of 750-770’C for 2-3h for welding consumables. However, it is important to
limit PWHT temperature to avoid the risks of austenite reformation and the transformation to
fresh untempered martensite, particularly in weld metal with elevated nickel. (17,18)

Table 11. Mechanical properties resulting from different times for E9015-B9 PWHT @
7600C (14000F) (22)

Time, UTS, Yield Elong, CVN

Min. ksi 0.2% % ft-bs@680 F

As –Welded 210 - - 3
[46 Rc] [119 @ 10220F]

45 120.3 101.8 17 13

120 104.7 83.5 23 53

[240-260 HV10]

The effect of extended time and PWHT on toughness and hardness are further presented in
Figures 13 and 14. It has been suggested that 760oC (~1400 oF) is the optimum PWHT
temperature, pending heating equipment capability and nickel content of the welding filler
metals. And, even though codes permit less time, PWHT should be conducted for a minimum
of two (2) hours at temperature, even for weld metal testing, to provide sufficient tempering.
 Page 23 of 40

Figure 13. Influence of PWHT Time and Temperature on Toughness (19,20)

Figure 14. Influence of PWHT Time and Temperature on Hardness.(19,20)

 Page 24 of 40

Questions arise over performing PWHT above 14000F on ASME B31.1 work because of the
7040C (13000F) to 7600C (14000F) range specified in Table 132. Interpretation 24-5 offered
the following: (30)

Interpretation: 24-5
Subject: B31.1 Table 132, Postweld Heat Treatment
Date Issued: October 27, 1993
File: B31-93-015
Question: Is it permissible to postweld heat treat SA-182 F91 materials at 14000F-
14500F instead of 13000F-14000F as specified for P-No. 5 in Table 132?
Reply: Yes, provided the lower critical temperature of the SA-182 F91 material is not
exceeded. See Para. 132.2(A).

Figure 15 presents typical preheat, welding/interpass and PWHT schedules for P91 to P91.
This figure illustrates cooling the weldments to room temperature prior to post weld heat
treat to permit complete transformation to a martensitic structure. On large vessels and
heavy-wall piping, this is not always practical. Successful practice over two decades has
shown that satisfactory results can be obtained by allowing the completed weld to cool to
the preheat temperature and holding this temperature continuously up to initiation of
the post weld heat treat. Success is dependent on close attention to maximum interpass
temperatures and time at temperature during the final post weld heat treat. As mentioned
earlier, further investigation is needed to clarify this difference in methodology. Conversely,
it was absolutely necessary to cool to room temperature prior to post weld heat treatment
where X20 materials were involved. Thus, this flexibility offered by P91 materials is much
more attractive to fabricators and installers. (9,18,31)

Also of critical importance on heavy sections, is implementation of a post weld bake

when using flux-type processes. This bake should be at least at the preheat temperature
if the fabrication schedule requires the weldment to cool to room temperature prior to
application of post weld heat treatment. Post bake times and temperatures vary from 4
hours at 2600C (5000F) to only 15 minutes at 3160C (6000F). Success with this much
variation is due to the advances made in consumable formulation and implementation of
proper preheat, welding, storage and handling practices.
 Page 25 of 40

TL > 1450°C

TA 750 +10°C

3 4 5

TMs <350°C
TP/I 1 1 1 ~250°C
2
80–100°C

TRT
tP tW tRM tA

TL = Melting temperature tP = Time of preheating

TA = Tempering temperature tW = Welding time
TMs = Martensite formation temp. tRM = Intermediate time (@80/100°C)
TP/I = Preheat/min.Interpass temperature tA = P.W.H.T. time
TRT = Ambient temperature

1 Temperature range for transformation to Martensite ( >250°C / >480°F )

2 Temperature range for allowing a complete transformation to Martensite ( <100°C / <210°F )
3 First welding pass 4 Second or subsequent passes 5 Last welding pass ( not annealed )
6 ″Temper bead″– effect of the former deposited weld passes/HAZs by appropriate welding
parameters

Figure 15. Typical thermal cycles observed during welding and PWHT. (32)
 Page 26 of 40

DISSIMILAR WELDS
Dissimilar welds involving P91, P11 and P22 and austentic stainless steels are performed on a
routine basis. Welds between the low alloy ferritic steels utilize either P91or materials
matching the lower alloy type base metals. In such weldments, carbon diffusion occurs
during the tempering heat treatment due to the difference in chromium content of the
materials. Carbon will migrate from the lower chromium material to the higher chromium
one. For example: When P22 (E901X-B3/ER90S-B3/EB3) filler metal is used, the
decarburized zone will be in the P22 weld metal with the carburized zone located in the P91
HAZ. If P91 weld metal is used, the carbon depleted zone will be located in the coarse
grained P22 HAZ and the carburized zone in the P91 weld metal. The extent of the
decarburized zone depend on the tempering temperature and time at temperature. The only
way to avoid this condition is to use a nickel-base welding consumable. Decarburization in
these dissimilar weldments typically only affect room temperature toughness properties.
Creep rupture properties are usually not affected. (31)

Transitions or dissimilar welds between P91 and austenitic stainless steels normally uses
nickel-base weld material. The weld metal can be applied to or involves buttering a P91
"pup" piece that can be heat treated and then field installed. In this manner, one weld can be
made and PWHT as a P91/P91 while the other a nickel base to stainless steel. (31)

Extreme care and planning must be observed concerning post weld heat treatment of
dissimilar weldments involving P91. "Pup" pieces or multiple buttering/PWHT operations
may be required when joining P91 to P11 or low carbon steels because tempering
temperatures necessary for P91 may exceed lower transformation temperatures for some of
the lower strength alloys. In these cases, P91 is oftentimes buttered with P22 or -B3 (or
lower) type weld metal, PWHT at P91 temperatures, then the field weld is made with the
lower strength alloy, followed by a subsequent PWHT a temperatures appropriate to the lower
alloy.

Typical dissimilar weld (DSW) combinations and corresponding approaches are presented in
Table12. Weld filler metal selection possiblities for DSW’s are shown in Table 13.
Guideline PWHT temperatures are included in Table 14.

Table 12. Dissimilar Weld Combinations and Approach (29,31)

Combination
Approach
P91 to P22 B3 & PWHT @ 13500F
P91 to P11 Butter P91 w/B2 & PWHT @
13500F; Then, join w/B2 &
PWHT @ 11000F
P91 to SS Butter P91 w/Ni & PWHT @
14000F; Then join to SS w/Ni
 Page 27 of 40

Table 13. Dissimilar Welding Filler Metal Selection (29,31,33,34)

P(T) 11 22 23 91 911 92 SS

11 B2 B2 B2 B2 B2 B2 309

22 B2 B3 B3,G B3 B3 B23 309,Ni

23 B2 B3,G B3,G,Ni G,Ni G G, B9 G

91 B2 B3 G,Ni B9,G B9,G W,G, B9 Ni

911 B2 B3,G G, B9 B9,G G W, G, B9 Ni

92 B2 B3,G G,B9 B9, G G,B9 W, G Ni

SS 309 309,Ni Ni Ni Ni Ni SS,Ni

G = Nonstandard composition
B2 = 1-1/4 Cr ½ Mo
B3 = 2-1/4Cr 1 Mo
B9 = 9 Cr 1 Mo V
W = Tungsten Modified
B = Boron Modified, etc.
Ni = Nickel Base (“A”, “82 or182”)
SS = Stainless, 308H, 309H, 316H, 347H, 16-8-2
Butter or buffer layers may be required.
 Page 28 of 40

Table 14. Recommended PWHT Temperatures (0F) for Dissimilar Welds (29,31,33,34)

P(T) 11 22 23 91 911 92 SS

11 1275 + 25 1350 + 25 1275 + 25 1350 max. 1350 max. 1350 max. 1275 + 25
Butter Butter

22 1275 + 25 1350 + 25 1275 + 25 1350 max. 1350 max. 1350 max. 1350 + 25
Butter

23 1275 + 25 1350 + 25 None 1350 max. 1350 max. 1350 max. None
Butter Butter

91 1275 + 25 1350 + 25 1400 + 25 1400 + 25 1400 + 25 1400 + 25 1400 + 25

Butter Butter

911 1275 + 25 1350 + 25 1400 + 25 1400 + 25 1400 + 25 1400 + 25 1400 + 25

Butter Butter

92 1275 + 25 1350 + 25 1400 + 25 1400 + 25 1400 + 25 1400 + 25 1400 + 25

Butter Butter

SS 1275 + 25 1350 + 25 None 1400 + 25 1400 + 25 1400 + 25 None

Butter Butter

Notes:
1. T23 has shown tendancies toward reheat cracking from PWHT. This alloy was
originally designed to be used without PWHT.
2. T24 typically does not require PWHT. Thus, buttering may be required.
3. Caution must be observed when selecting PWHT temperatures that will not
encroach on the AC1 of the lowere temperature material.

PRACTICAL CONSIDERATIONS (2)

Other than the heat control measures, welding of P91 material can be considered reasonably
close to the welding of carbon steel materials. There is an inherent sluggishness of the weld
puddle, typical of higher alloy weld materials. This sluggishness translates into increased
difficulty in narrow (or semi-narrow) groove applications and can lead to lack of side wall
fusion or slag entrapment problems. The travel speed of the first bead in pipe welding may
have to be slow enough to allow a larger puddle to assist in the side wall fusion which
requires an increased heat input. Once the bevel design is wide enough to allow multiple
beads per layer, the welder or welding operator must be careful in the placement of the
subsequent beads in order to avoid slag entrapment should there be any slag remaining along
the side of the previous bead. A good rule of thumb is to place the wire and therefore the arc
 Page 29 of 40

immediately over the top of the edge of the previous bead so as to concentrate the heat of the
arc at the crevice point of the bead. Crossover/stepover points, as well, must be approached
with greater care than carbon steels. (2)
Welding parameters typical of carbon steels must be reduced, to maintain desirable weld bead
characteristics. Due to the alloy content, Grade 91 (B9) fillers are more sensitive to changes
in volts, amps, electrical stickout and travel speed.

Due to the high chromium level, these materials air harden and exhibit very little
ductility in the as welded condition, therefore the application of elevated preheat,
interpass temperature controls and post weld heat treatment are absolutely
necessary with grade P91 weldments, regardless of diameter or thickness. (5)

Preheat
The literature suggests that 2000C (~4000 F) is adequate for preheating P91 weldments.
Fabricators typically aim for 2000C to 2500C (~4000F to 5000F), but will go as low as 1210C
(~2500 F) for root and hot pass layers, thin walled components or where GTAW is utilized.
Preheat temperature should be considered an interpass minimum, since cooling to room
temperature before the completion of the weld, without proper precautions, is not advisable
when using flux bearing processes. (31)

Interpass Maximum
A typical interpass maximum is 3000C (~6000F), slightly less is acceptable but no more than
3700C (7000F). The interpass maximum helps to prevent the possibility of hot cracking due to
the silicon and niobium content of the weld metal. Also, allowing the weldment to cool to
below the martensitic start temperature (Ms) (typically less than 2000C/4000F, and in some
cases ~1000C (~2000F)) allows at least a portion of the martensitic microstructure to be
tempered by subsequent beads. (13,31)

Post Weld “Bake-Out”

A post weld “bake-out” is of critical importance, especially for heavy sections or where flux-
type processes are used. This involves maintaining the preheat/interpass window for an
extended period of time subsequent to completion of the weld in order to facilitate hydrogen
diffusion from the weldment. Of the many variables involved when attempting to establish
the length of time necessary, are the thickness of the material , length of time the weldment
has been exposed to the heat regime and the extent of “low hydrogen” practices. If the
fabrication schedule requires the weldment to cool to room temperature prior to PWHT, a 2
hour bake-out at the preheating temperature is a good starting point. Shorter times at higher
temperatures are also common; e.g. 15 minutes at 3160C (6000F). (2,31)

Interruption of the Heat Cycle

Interruption of the heat cycle should be avoided if at all possible, but due to production
schedules, this is not always the case. If it becomes necessary, great care should be exercised.
The mass of the weldment should also be considered. Increases in pipe wall thickness
 Page 30 of 40

translates into increases in both the restraint on the weld and the cooling rate from welding
temperatures. Therefore, the weld area is subjected to high residual stresses at a time when it
may have minimum section thickness (or strength) and be less ductile. As the percentage a
weld is completed increases, the more the strength and rigidity of the joint resemble the
completed weld. Given this, any interruption of the welding should be avoided until a
specified minimum amount of the joint has been deposited. Grade 91 weldments should be
completed without interruption, but if interruption is unavoidable, at least 1/4 of the wall
thickness should be deposited and preheat must be maintained until the groove is completed.

Miscellaneous Precautions
To minimize crater cracking or undesirable grain boundary phenomena, low residual element
content (X factor < 15) weld filler metal, strict adherence to the preheat and interpass
temperature requirements, plus use of covered electrodes and fluxes meeting “H4” and/or H5
criteria are advisable.
In all cases, low hydrogen controls must be implemented and maintained during
fabrication operations. Such controls are even more important for procurement, use and
storage of welding consumables. Manufacturers recommendations for storage and
reconditioning must be observed. If electrodes or submerged arc welding flux absorbs
moisture, they should be discarded. Most covered electrodes are available with H4
formulations. SAW fluxes should also be of the low hydrogen variety and meet at least an H5
designation. Figure 16 illustrates exposure criteria for one FCAW product. Note that this
wire meets H4 criteria.

Traditional code required NDE (UT and/or RT) may not identify crater, hot or cold cracking
in P91 weldments. Repetitive patterns of seemingly nonrelevant indications should be
evaluated with more sensitive NDE techniques.

Root passes in piping, tubing or other components require purging with 100 % welding
grade argon or 100% N2 until at least the root and hot pass have been deposited. Both
gases provide adequate shielding. Purge dams and fixturing should be able to accommodate
temperatures up to 300oC (~< 600 oF) to ensure equipment operability and maintenance of a
proper purge, given the elevated preheats.

Heat inputs are typically maintained on the lower end of the process’s usable range with bead
shape being of primary importance. Successful manual procedures rarely exceed 25 to 30
KJ/in (~ 10 to 12 KJ/cm), FCAW and SAW may approach 55 KJ/in.
 Page 31 of 40

15
Relative humidity level:
10%RH
Diffusion hydrogen in weld metal, ml/100g
50%RH
70%RH
90%RH

10

0
0 5 10 15 20 25

Exposure time, hours

Figure 16. Effect of exposure time/condition on weld metal hydrogen content for a selected
FCAW wire. (23)

Extreme care must be observed during fabrication (lifting, handling, fixturing, etc.) to
avoid applying unnecessary bending stresses or loading to weldments that have not seen
PWHT!

Hardness Control
Certain system applications have maximum hardness limitations. This is particularly true in
systems where hydrogen sulfide is present. Maximum values of 235 BHN are not
uncommon. The Grade 91 welding alloys, after typical post weld heat treatment, normally
exceed such values. However, by selective application of elevated PWHT time and
temperature (Figure 14), control of bead shape during welding (Figure 6) to enhance interbead
tempering or softening and specifying weld metal composition limitations (Figure 2, Table 3)
that enable use of elevated PWHT temperatures, lower hardness can be achieved in the weld
deposit.
 Page 32 of 40

WELD METAL AVAILABILITY and PROCUREMENT

Availability
It was not until 1996 that a recognized national code included specifications for the Modified
9Cr-1Mo welding alloys. Subsequent to these specifications, users referred generically to the
material as “505 Modified” or “P/T91”with respect to the standard 9Cr-1Mo composition
which was classified as “505” material. In fact, the “505” requirements were included in the
stainless steel specifications – a situation that is being remedied upon future specification
revisions.
The American Welding Society amended specifications A5.5 (low alloy steel coated
electrodes) and A5.28 (low alloy steel solid wires for gas shielded welding) to accommodate
this alloy and assigned the “-B9” designation to certain 90-series filler metals. Specification
A5.23-97 (low alloy steel wire and flux for SAW) has been revised and includes data for the
“EB9” composition. Table 16 provides summary data on the various weld filler metal
specifications.
Flux core or metal core fabricated wire products do not currently have an AWS classification
and must be ordered to a “G” classification under A5.29 or A5.28, respectively, or by trade
name and a specified chemistry and/or mechanical property criteria.

Procurement
All material should be procured to a recognized specification. Where commercial or technical
criteria prevent direct reference to an existing specification, the nearest specification should
be used and augmented, as necessary, to provide accurate communication between the user
and supplier.

Recognized domestic weld filler metal specifications for P91 welding consumables are
provided in Table 15. Note that specific classifications are not provided for P91 FCAW weld
metal and that such material would have to be procured under a “G” classification; i.e. the
properties and composition are agreed upon between the user and manufacturer. This
situation is due to the fact that FCAW wires have not been commercially available until
recently.

For weld metal, AWS A5.01 “Filler Metal Procurement Guidelines” (7) provides an excellent
means for organizing the procurement of both standard and special welding consumables. For
example, crater cracking and other undesirable grain boundary phenomena can be minimized
by specifying weld metal with low residual element content (X Factor < 15).
Further, ordering covered electrodes (SMAW)to "H4", FCAW and SAW (flux) to “H5”
moisture criteria also reduces the potential for hydrogen or moisture related phenomena.
Consumables must be stored and handled to avoid moisture pick-up. This typically requires
using holding ovens for storing open containers of SMAW electrodes and SAW fluxes.
Where controls are difficult to implement, such as in field operations, consumable packaging
 Page 33 of 40

should be selected to minimize waste or storage issues; e.g. size SMAW or FCAW packaging
for the amount that can be used in one-half shift, etc.
Table 16 illustrates information in the format recommended in A5.01. (7)

Table 15. P91 Weld Metal Specifications (Domestic) (29)

Process AWS Classification (s)
Specification
SMAW A5.5-96 E9015-B9,E9016-B9,E9018-B9,E90XX-G*
SAW A5.23-97 EB9,EG*
GTAW/GMAW A5.28-96 ER90S-B9, ER90S-G*, E90C-G*
FCAW A5.28-96 E90C-G*
FCAW A5.29-XX (Draft) E10XT1-B9 (pending)
All None “505 Modified” –G*

*The “-G” classification is utilized to procure P91 material to an internal corporate specification or to
obtain material that is similar to an AWS classification but has one or more elements outside the listed
AWS composition range. For example, there is no current specification for P91 FCAW wire.
Therefore, the material must be ordered as “505 Modified”, to a corporate specification or
manufacturers brand.

CODE APPROVALS
American Welding Society
Given the availability of commercial productsA5.5 (low alloy SMAW), A5.23 (low alloy
SAW wires & metal core) and A5.28 (low alloy solid and metal core) have all been revised to
include “-B9” classifications. Demand for FCAW has attracted the attention of the AWS
Filler Metal Committee. Efforts are being accelerated (anticipated late 2001/early 2002) to
incorporate criteria for an “E10XT1-B9” classification into A5.29 (low alloy flux core).

ASME (8)
Although approval and incorporation into ASME Section II, Part C, Welding Rods,
Electrodes, and Filler Metals, typically lag AWS approvals, once included, they typically
mirror the AWS filler metal requirements. This is the case for the Grade 91 Consumables.

It should be noted that it is not unusual for Code Cases to be issued to cover new base
materials (ASME Section II, Parts A and B) far ahead of weld filler metal development or
code acceptance. The Grade 91 materials offer a primary example.
 Page 34 of 40

Table 16. Example AWS A5.01 Procurement Specification Guideline Outline. (7)
Guideline
Procurement Specification
(Ref. AWS A5.01-93)
I. General
[Suggested Procurement Detail [Table A2] [Table A1] [Table A4] [Table A3]
Form
A. Quantity 100 lbs. 1,089 lbs. 1,100 lbs. 660 lbs.
B. AWS Specification A5.28 A5.5 A5.23 A5.29
C. AWS Classification ER90S-B9 E9015-B9 EB9 “E101T1-B9”
(-G) {pending}
D. Supplemental Designators N/A H4 N/A H4
E. Diameter 3/32” (2.4mm) 3/32” (2.5mm) 3/32” (2.4mm) 0.045”
(1.2mm)
F. Length 39” (1000mm) 14” (350mm) N/A N/A
G. Unit Package Type and Weight
1. Carton 33 lbs.(15 kg)
2. Can
3. Other 11 lbs. (5kg) 55 lbs. (25 kg) 33 lbs. (15 kg)
Coil Spool

II. Certification and Testing

A. Lot Classification S1 C1 C1 T1
B. Level of Testing H K H K

III. Other Requirements

A. Impring / Marking Heat No. & Lot No. & Label w/Heat Label w/Lot
Classification Classification No. & No. &
Classification Classification
B. Stamping or Tagging Stamp Stamp Label Label
C. Report Tensile/Yield Strength Yes Yes N/A Yes
D. PWHT: 7600C , 2 hrs. min. Yes Yes N/A Yes
E. Report Toughness @ Temp. N/A Yes @ 720F N/A Yes @ 720F
F. X-Factor (Bruscato Number) < 15 < 15 < 15 < 25
G. CMTR or “3.1B” Required Yes Yes Yes Yes
Notes:
a. H4, Optional Diffusible Hydrogen Designator, indicates that the diffusible hydrogen in the deposited
weld metal did not exceed 4.0 mL/100g when tested in accordance with AWS A4.3 and A5.5 specifications.

b. Where [-G] designation is used in the classification, chemical composition of the electrode shall be as
agreed to by purchaser and supplier. In this example, it is as follows: (Some Users Request: “Meets the
mechanical requirements of E9015-B9”, instead of using the “-G” AWS designation.)

C 0.07-0.13% Cr 8.00-9.50% Ti <0.004% Cu <0.20%

Mn 1.25% max. Ni 1.00% max. As <80ppm N2 0.03-0.07%
Si 0.20-0.50% Mo 0.80-1.10% Sb <30ppm Other <0.50% Total
S <0.010% Nb 0.02-0.10% Sn <30ppm
P <0.010% V 0.15-0.25% Al < 0.04% Mn + Ni = 1.5% max.
 Page 35 of 40

CONCLUSIONS
Thousands of tons of P91 materials are in use worldwide. Technical and operational concerns
with FCAW for P91 welding have been addressed and this process is seeing accelerated
acceptance and use.
The two major reasons cited by domestic fabricators and end users for not using the P91
Modified material are being addressed or will be in the future. These reasons include:
• All welds require post weld heat treat, regardless of thickness or diameter
• Installation and heat treatment of dissimilar weldments (e.g., P22 to P(T)91) can be
complex in either shop or field applications.

However, this situation is changing at a rapid rate in the USA. Designs for retrofits and new
installations, particularly cogeneration units, are using P(T)91 as the material of choice. Key
to successful fabrication and installation of this material are the following criteria:

• PWHT should be conducted for a minimum of two (2) hours at temperature, even for weld
metal testing, to provide sufficient tempering. This is consistent with current
manufacturing practice by major fabricators (domestic and European) of utilizing a
minimum of 2 hours at temperature, regardless of thickness. Post bakes are needed when
the weldment will cool to room temperature prior to PWHT. PWHT temperature should
range from 745 to 760 0C (1375 to 14000F), with 760 0C (14000F) perhaps being
optimum. Extra care and planning must be observed where dissimilar weldments with
P91 exist.

• Ordering covered electrodes (SMAW)to "H4", FCAW and SAW (flux) to “H5” moisture
criteria also reduces the potential for hydrogen or moisture related phenomena.
Consumables must be stored and handled to avoid moisture pick-up. This typically
requires using holding ovens for storing open containers of SMAW electrodes and SAW
fluxes. Where controls are difficult to implement, such as in field operations, consumable
packaging should be selected to minimize waste or storage issues; e.g. size SMAW or
FCAW packaging for the amount that can be used in one-half shift, etc.
• Toughness results improve when testing is conducted at 22-23C (72-74F) versus 20C
(68F).
• Lower hardness can be achieved in the weld deposit by selective application of elevated
PWHT time and temperature, control of bead shape during welding to enhance interbead
tempering or softening and specifying weld metal composition limitations that enable use
of elevated PWHT temperatures.
• Where thermal straitening or other bending operations are implemented, caution must be
observed to not encroach on the lower critial (AC1) temperature of the materal. If this is
exceeded, the component must again be normalized and tempered. Some authorities make
these additional heat treatments mandatory for P(T)91.
 Page 36 of 40

• AWS Filler Metal Specifications subcommittees for flux core and metal core low alloy
wires are reviewing data generated by the manufacturers to consider the addition of “B9”
specifications for these products.
• An FCAW wire with satisfactory and reproduceable mechanical properties is available,
including operation with both ArCO2 (80-20, 75-25) mixtures or 100% CO2 shielding gas.
• Inert gas purging gains flexibility because either argon or nitrogen can be used
successfully.
• All welding consumables should be ordered with CMTR’s or EN10204 3.1B’s to ensure
that heat treatment operations do not compromise the integrity of weldments.
• Further investigation should include gathering additional service experience and creep
data plus information and study related to dissimilar weldment design, implementation
and performance.

For a large diameter thick wall fixed pipe joint, reductions in joint completion rate of 25-40%
compared with SMAW can be achieved by employing FCAW.
Mechanical and fracture toughness test results indicate that the FCAW weld metal is slightly
stronger than that produced by SMAW and SAW processes, while the impact toughness is
slightly lower. The calculated maximum tolerable flaw sizes using the lowest CTOD value at
0°C and 20°C are large and very close to the values for SMAW deposits with similar CTOD
values. Any defects of significance should be readily detectable with current NDE
techniques.
At elevated temperatures, the proof stress of the FCAW weld metal was similar to SMAW.
The stress-rupture strength was higher, probably because of the influence of residual titanium
introduced from the rutile flux system. Rupture strength of both FCAW and SMAW weld
metal was within the range expected for parent material. Typical of weld metals, rupture
ductility decreased at shorter durations than parent material. However, given the high creep
strength found, the “type IV” HAZ would still be the weakest region of a weldment in actual
service.
The model and analysis presented in this paper are not intended to cover every design
situation or application. Users are advised to conduct their own specific analysis for critical
applications or systems.
 Page 37 of 40

ACKNOWLEDGEMENTS
This paper would not have been possible without the technical assistance provided Kent
Coleman, David W. Gandy and Shane J. Findlan at EPRI’s Repair and Replacement
Application Center in Charlotte, NC. Additional technical input and experimental results
were also provided by: Mr. Roger A. Swain of Euroweld, Ltd.; Dr. Siegfreid Dittrich (retired,
Thyssen Schweiâtechnik GMBH), Germany; Ian Barnes of I.A. Barnes Co., United Kingdom;
Dr. C. Farrar, Dr. Z. Zhang, Graham Holloway and Adam Marshall of Metrode, Ltd., United
Kingdom; Hermann Zentner, Consultant for Bavaria Schweiâtechnik, Germany; James W.
Hales, Specialty Welding & Machining, Inc.; J. Franklin Turner, Electrode Engineering, Inc.;
J.R. Scott, B.F. Shaw Co.; J. D. Duncan, H. Mantle and W. Spear, Bechtel Corporation; H.
Clark, Fluor-Daniel; and Randy Davis, Consultant.
 Page 38 of 40

REFERENCES
1. Metrode Products Limited, “Welding consumables for P91 steels for the power
generation industry”, Technical Profile, Issue 4, October 2000.
2. W.F. Newell, Jr. and J.R. Scott. “Properties and Fabrication Experience with
Submerged Arc Welding of P91 Piping Systems”, Conference Proceedings, Fourth
International EPRI Conference on Welding and Repair Technology for Power Plants,
Marriott’s Marco Island Resort and Golf Club, Naples, Florida, USA, 7-9 June 2000.
3. Zhang, Z, Marshall, A W and Farrar, J C M. “Present developments in welding
consumables for P(T)91 creep-resisting steels”, Conference Proceedings, International
Conference on Integrity of High-Temperature Welds, Nottingham, UK, 3-4 November
1998, The Institute of Mechanical Engineers, London, pp 77-91.
4. Zhang, Z, Farrar, J C M and Barnes, A M. “Weld metals for P91 – tough enough?”,
Conference Proceedings, Fourth International EPRI Conference on Welding and
Repair Technology for Power Plants, Marriott’s Marco Island Resort and Golf Club,
Naples, Florida, USA, 7-9 June 2000.
5. W.F. Newell, Jr. and D.W. Gandy. “Advances in P(T)91 Welding Using Flux and
Metal Cored Wires”, EPRI Welding and Repair Technology For Power Plants, Third
International EPRI Conference, 9-12 June 98, Scottsdale, AZ.
6. "Weld Cost Analysis Program", Daihen, Inc. 1995.
7. “Filler Metal Procurement Guidelines”, ANSI/AWS A5.01-93, American Welding
Society.
8. “Boiler and Pressure Vessel Code, Section II, Parts A, B & C”, American Society of
Mechanical Engineers, New York.
9. Z. Zhang, A.W. Marshall, G.B. Holloway. “Flux Cored Arc Welding: The High
Productivity Welding Process for P91 Steels”, Metrode Products, Ltd. February 2001.
10. “Thick-SectionWelding of Modified 9Cr-1Mo (P-91) Steel”, ABB Combustion
Engineering Systems, September 1992, Chattanooga, Tennessee.
11. “Thick-Section Welding of Modified 9Cr-1Mo (P-91) Steel; EPRI TR-101394; Project
1403-14, Interim Report, September 1992.
12. “Properties of Modified 9Cr-1Mo Cast Steel”, EPRI TR-106856; Project WO4051-01,
Final Report, September 1996.
13. M. Santella, R. Swindeman. “Some Experimental Data Bearing on the Need to Drop
Preheat Prior to PWHT of Grade 91 Steel”, ASME BPVC Meeting, San Francisco,
February 2001.
14. S. Dittrich, H. Heuser, Hamm, “SchweiBzusatzwerkstoffe fár den 9% Chromstahl P
91” Thyssen Schweisstechnik GMBH, March 1993.
15. Dr. S. Dittrich, Dr. H. Heuser, R. Swain, “Optimized Filler Metals for the Fabrication /
Installation of T(P) 91” January 31, 1994, Harrisburg, North Carolina.
16. M. Santella. Private Communications. February, 2001.
 Page 39 of 40

17. J. C. M. Farrar, Z. Zhang and A.W. Marshall. “Welding Consumables for P(T)-91
Creep Resisting Steels”, Metrode Products Limited, UK. EPRI Welding and Repair
Technology For Power Plants, Third International EPRI Conference, 9-12 June 98,
Scottsdale, AZ.
18. Zhang, Z, Farrar, J C M and Barnes, A M. “Weld Metals for P91 – tough enough?”,
Conference Proceedings, Fourth International EPRI Conference on Welding and
Repair Technology for Power Plants, Marriott’s Marco Island Resort and Golf Club,
Naples, Florida, USA, 7-9 June 2000.
19. “Consumables For The Welding Of 9 Cr - 1 Mo - ¼ V Steels” Including: 1) Welding
of Modified 9% Cr Steel, 2) Optimized Filler Metals for the Fabrication/Installation of
T(P)91, 3) SMAW of P91 Piping with Optimized Filler Metals and 4) TSG Test
Report - Welding of P 91 Material: SMAW, SAW and GTAW. Thyssen Welding,
April 1995, Carol Stream, Illinois.
20. S. Dittrich, D. Gandy, W. Newell, Jr., B. Roberts, R. Swain, J. Turner and W. Zilke.
“Grade 91 Modified Meeting Minutes”, Chattanooga Choo Choo, Chattanooga, TN;
14 May 98
21. C. Coussement, et al. “European State of the Art of Modified 9% Cr Steels: Welding,
Fabrication and Industrial Applications of P91/T91 and New Developments”, Belgian
Welding Institute, et al. EPRI Welding and Repair Technology For Power Plants,
Third International EPRI Conference, 9-12 June 98, Scottsdale, AZ.
22. Z. Zhang, A.W. Marshall, G.B. Holloway, Private Communications/Metrode Files.
23. BS7448-2:1997: “Fracture mechanics tests. Method for determination of KIC, Critical
CTOD and critical J values of metallic materials”, British Standard Institution,
London, 1997.
24. Crackwise® 3 (TWI Structural Integrity Software) – Automation of BS7910:1999
Fracture and Fatigue Assessment Procedures, TWI, Cambridge, UK.
25. BS7910:1999: “Guide on methods for assessing the acceptability of flaws in fusion
welded structures”, British Standard Institution, London, 1991.
26. Hadley, I, Wiesner, C S and Maddox, S J. “PD 6463 becomes BS 7910; what’s new
in fracture and fatigue assessment?”, I Mech E seminar, “Flaw assessment in pressure
equipment and welded structures – PD 6491 to BS 7910”, London, 8 June 1999.
27. Newell, Jr. W F and Swain, R A. Private Communications, Euroweld, Ltd.
28. “Welding of P 91 Material SMAW, SAW and GTAW” Thyssen SchweiBtechnik
GmbH, April 3, 1992.
29. “Recommended Practices for Welding of Chromium-Molybdenum Steel Piping and
Tubing”, ANSI/AWS D10.8-96, American Welding Society.
30. “Power Piping”. ANSI/ASME B31.1
31. G. Guntz, M. Julien, G. Kittmann, F. Pellicani, Apoilly, J.C. Vaillant. “The T91 Book,
Ferritic Tubes and Pipe for High Temperature Use in Boilers”, Vallourec Industries
1994, Revision 2.
 Page 40 of 40

32. Zentner, Hermann. Bavaria Schweiâtechnik Gmbh, Unterschleiâheim, Germany.

33. D. Richardot, J.C. Vaillant, A. Arbab and W. Bendick. “The T92/P92 Book”,
Vallourec & Mannessmann Tubes, 2000.
34. J. Arndt, K. Haarmann, G. Kottmann, J.C. Vaillant, W. Bendick and F. Deshayes.
“The T23/T24 Book, New Grades for Waterwalls and Superheaters”, Vallourec &
Mannessmann Tubes, 1998.
35. D. Canonico, J. Henry. “Planning to Avoid Problems with New Advanced Chrome
Steels for Boilers”, Repair Welding and Serviceability Conference, Pressure Vessel
Research Council, The San Diego La Jolla Marriot Hotel, La Jolla, CA, January 31 &
February 01, 2001.
36. D. A. Canonico, “The Application of Grade 91 Steel for Petroleum Industry Pressure
Vessels”, Second International Conference on Interaction of Steels with Hydrogen in
Petroleum Industry Pressure Vessel and Pipeline Service, October 19-21, 1994,
Vienna, Austria.
37. C. D. Lundin, K. K. Khan, K. A. Al-Ejel, “Modified 9Cr (P91) SMA Weldments
Microstructural Evaluation” Materials Joining Group, July 1994, Knoxville,
Tennessee.
38. Dr. H. Heuser, Dr. G. Wellnitz, “GTA / SA Welding Of The 9% CR T 91 / P 91 Steel”
Annual AWS Convention, March 24, 1992, Chicago, Illinois.
39. “Conference On Understanding Weldment Behavior In Cr-Mo Steels”, April 1991,
Knoxville, Tennessee.
40. "Submerged Arc Welding Handbook", Union Carbide Corporation, 1980.

41. Orr, J and Burton, D. “Experience in data collection and assessment for material
standards”, ECSC Information Day, The Manufacture and Properties of Steel 91 for
the Power Plant and Process Industries, Paper 5.2, Dusseldorf, 1992.
42. Kimura, K, et al. “Microstructral change and degradation behaviour of 9Cr-1Mo-V-Nb
steel in the long term”, Conference Proceedings, Parsons 2000 Conference, Advanced
Materials for 21st Century Turbines and Power Plant, IoM, 2000, pp590-602.
43. Murata, Y, et al. “Effect of trace elements on the creep properties of ferrite steels for
steam turbine rotors”, Conference Proceedings, Advanced Heat Resistant Steels for
Power Generation, San Sebastian, Spain, 27-29 April 1998, IoM.
44. Marshall, A W and Farrar, J C M. “Influence of residuals on properties of austenitic
stainless steel weld metal with particular reference to energy industries”, Conference
Proceedings, Stainless Steel 84, Institute of Metals, 1985, pp271-285.