energies
Article
Failure Analysis of a Centrifugal Compressor Impeller Made of
17-4PH Steel in the Moist Hydrogen Sulfide Environment
Jakub Łagodziński * , Zbigniew Kozanecki and Eliza Tkacz
Institute of Turbomachinery, Lodz University of Technology, 90-924 Lodz, Poland;
zbigniew.kozanecki@p.lodz.pl (Z.K.); eliza.tkacz@p.lodz.pl (E.T.)
* Correspondence: jakub.lagodzinski@p.lodz.pl
Abstract: The impeller under consideration is a rotor wheel of a centrifugal compressor, made of
17-4PH martensitic precipitated hardening stainless steel. Its total operating lifetime was estimated to
80,000 h, but it fractured beforehand, 4 days after the scheduled overhaul without any observed presymptoms. The important causes of the failure were its working conditions: a moist H2 S environment,
the applied shaft fitting stress, and the martensitic structure of its material. In the article the postfailure analysis of an impeller is described, explaining the root cause of the impeller failure by means
of both experimental and Finite Element Methods (FEM). Sulfide Stress Cracking (SSC) was found to
be the primary failure mechanism for the impeller under investigation. Details of the investigations
and possible corrective measures to improve performance impellers used under harsh environment
conditions are discussed.
Keywords: failure analysis; compressor impeller fracture; sulfide stress cracking
Citation: Łagodziński, J.; Kozanecki,
Z.; Tkacz, E. Failure Analysis of a
Centrifugal Compressor Impeller
Made of 17-4PH Steel in the Moist
Hydrogen Sulfide Environment.
Energies 2022, 15, 4183. https://
doi.org/10.3390/en15124183
Academic Editor: Valery Okulov
Received: 4 December 2021
Accepted: 1 June 2022
Published: 7 June 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affiliations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
1. Introduction
In the petrochemical industry, the compressors used for hydrocracking, hydrorefining
and reforming work in very aggressive environments. In these harsh conditions, alloy
steels cannot operate. In particular, in the hydrocracking process, thus in the presence of
water vapor with hydrogen disulfide, compressor impellers are exposed to severe stress
corrosion conditions [1–3]. That is why precipitation hardening stainless steels, such as
17-4PH and 14-5PH, are chosen for modern hydrogen centrifugal compressors, instead of
using alloy steels [4]. Stainless steels with optimized chemical compositions and after heat
treatments have more than good corrosion resistance but also outstanding strength and
weldability [5].
Properties of materials and manufacturing techniques are constantly improved, but
despite the know-how growth failures occasionally happen. The root cause seems to be
exceptionally complex stress conditions and an aggressive working environment [3,6,7].
The impeller described in the paper is of closed type, made of the 17-4PH martensitic
precipitation hardening stainless steel, and is used for hydrorefining. The mechanism of
its failure is a subject of this investigation, including optical examination, material testing,
theoretical calculations and numerical simulations. Special attention has been given to the
material properties, the corrosive environment and the complex stress state.
The considered impeller worked as a first stage wheel of a six-stage compressor. Its
failure occurred 4 days after the scheduled overhaul without any observed pre-symptoms.
The total operating lifetime of the wheel was estimated to be 80,000 h.
The main objective of this paper is to identify the root cause of the failure. First,
the impeller was sent to the Strength of Materials Laboratory, where surface hardness
tests were conducted. Second, a numerical analysis based on Finite Element Method was
done to assess the influence of each of stress generating features: shrink fit, centrifugal
force and keyways. Third, we conducted an investigation of the hypothesis that the
Energies 2022, 15, 4183. https://doi.org/10.3390/en15124183
https://www.mdpi.com/journal/energies
caused the Sulfide Stress Cracking phenomenon to occur. The ultimate objective was to
give post failure recommendations for machine improvement, so prevention countermeasures against failure were suggested.
2. Optical Analysis of the Fracture Surface
of 11
There were six impellers in the compressor and the fracture occurred in the first2one,
2 of 11
as shown in Figure 1. Their blades were welded using a technique of slot-welding. Then
the impellers were mounted on the shaft with a shrink fit and two keyways.
acid An
environmental
impact on
in presence
highinitiated
temperatures
optical examination
ofthe
the17-4PH
fracturestainless
surface steel
revealed
that theof
crack,
from
environmental
impact
on
the 17-4PH
steel
presence
ofobjective
high
temperatures
the propagated
Sulfide
Stress
Cracking
phenomenon
occur.inThe
was
give
acaused
keyway,
diagonally
to
the stainless
externaltodiameter
of ultimate
the impeller
wheel.
In to
Figure
caused
the
Sulfide
Stress
Cracking
phenomenon
to occur.
ultimate
objective
was
to
post
failure
recommendations
for
machine
improvement,
soThe
prevention
countermeasures
2,
in cut
view,
one can
see clear
chevron
marks
pointing
towards
the origin
of failure.
This
give post
failure
recommendations
machine
sohole
prevention
against
failure
were
suggested.
examination
revealed
that the fracturefor
occurred
at improvement,
first in the center
and thencounterspread
measures
against
failure
were
suggested.
out to the blades and the cover. The Strength of Materials Laboratory, after their exami2. Optical
Analysisa of
the Fracture
nation,
identified
brittle
fracture Surface
without visible prior plastic deformation and con2.
Optical
Analysis
of
the
Fracture
Surface
There
six impellers
thethe
compressor
and
occurred in
firsthardone,
firmed
thatwere
the keyway
cornerin
was
initial point
ofthe
thefracture
crack. Moreover,
thethe
local
as shown
inwere
Figure
Their blades
were
welded
using
a fracture
technique
Then
There
six 1.
impellers
in
compressor
and
the
in higher
the firstvalue
one,
ness
of
fracture
was
estimated
at the
336–342
of the HV10
scale,
whichoccurred
isofa slot-welding.
much
the
impellers
were
mounted
onofthe
shaft
with
awheel.
shrink
and two keyways.
as shown
in Figure
1. Their
blades
welded
using fit
a technique
of slot-welding. Then
than
measured
in other
points
thewere
damaged
the impellers were mounted on the shaft with a shrink fit and two keyways.
An optical examination of the fracture surface revealed that the crack, initiated from
a keyway, propagated diagonally to the external diameter of the impeller wheel. In Figure
2, in cut view, one can see clear chevron marks pointing towards the origin of failure. This
examination revealed that the fracture occurred at first in the center hole and then spread
out to the blades and the cover. The Strength of Materials Laboratory, after their examination, identified a brittle fracture without visible prior plastic deformation and confirmed that the keyway corner was the initial point of the crack. Moreover, the local hardness of fracture was estimated at 336–342 of the HV10 scale, which is a much higher value
than measured in other points of the damaged wheel.
Energies 2022, 15, 4183
Energies 2022, 15, x FOR PEER REVIEW
Figure
Figure 1.
1. Failed
Failed impeller
impeller of
of the
the first
first stage.
stage.
An optical examination of the fracture surface revealed that the crack, initiated from a
keyway, propagated diagonally to the external diameter of the impeller wheel. In Figure 2,
in cut view, one can see clear chevron marks pointing towards the origin of failure. This
examination revealed that the fracture occurred at first in the center hole and then spread out
to the blades and the cover. The Strength of Materials Laboratory, after their examination,
identified a brittle fracture without visible prior plastic deformation and confirmed that the
keyway corner was the initial point of the crack. Moreover, the local hardness of fracture
was estimated at 336–342 of the HV10 scale, which is a much higher value than measured
in
other
of the damaged
Figure
1. points
Failed impeller
of the firstwheel.
stage.
Figure 2. Fractured surface of the compressor impeller wheel.
Some important conclusions can be drawn from this fracture surface analysis (Figure
3), namely the chevron pattern and the way this pattern converges in the suspected point
of fracture initiation. There are known other cases of the compressor impeller failure with
similar fracture morphology, where the main cause of damage was Sulfide Stress Cracking [3,7,8]. Based on these facts and the local hardening of the section area, it can be
Figure 2.
2. Fractured
surface of
of the
the compressor
compressor impeller
impeller wheel.
wheel.
Figure
Fractured surface
Some important
importantconclusions
conclusionscan
canbe
bedrawn
drawnfrom
fromthis
this
fracture
surface
analysis
(Figure
Some
fracture
surface
analysis
(Figure
3),
3), namely
chevron
pattern
and
the
waythis
thispattern
patternconverges
convergesin
inthe
the suspected
suspected point
point
namely
thethe
chevron
pattern
and
the
way
of
fracture initiation.
initiation.There
There
known
other
cases
of compressor
the compressor
impeller
failure
of fracture
areare
known
other
cases
of the
impeller
failure
with
with
similar
fracture
morphology,
the cause
main of
cause
of damage
was Sulfide
Stress
similar
fracture
morphology,
wherewhere
the main
damage
was Sulfide
Stress Cracking [3,7,8]. Based on these facts and the local hardening of the section area, it can be
Energies 2022, 15, x FOR PEER REVIEW
Energies 2022, 15, 4183
3 of 11
3 of 11
Energies 2022, 15, x FOR PEER REVIEW
3 of 11
concluded that the fracture was caused by excessive fragility resulting from operation in
Cracking [3,7,8]. Based on these facts and the local hardening of the section area, it can be
the aggressive
environment.
concluded
concluded that
that the
thefracture
fracturewas
wascaused
causedby
byexcessive
excessivefragility
fragility resulting
resultingfrom
fromoperation
operationin
in
the
aggressive
environment.
the aggressive environment.
Figure 3. Fracture surface of the damaged impeller.
Figure3.3.Fracture
Fracturesurface
surfaceof
ofthe
thedamaged
damagedimpeller.
impeller.
Figure
3. Stress Analysis
3.3. Stress
StressAnalysis
Analysis
To confirm the given hypothesis and understand the mechanism of the failure, the
confirm
the
given
hypothesis
understand
thecomplex
mechanism
of theoffailure,
the
To
confirm
the
given
hypothesis
understand
the mechanism
the
failure,
stressTo
levels
and
distribution
need and
to and
be
found.
For
geometries,
the
usestress
ofthe
threelevels
and
distribution
need need
to betofound.
ForFor
complex
geometries,
the
use
of
threestress
levels
and
distribution
be
found.
complex
geometries,
the
use
of
threedimensional Finite Element Method analysis is a common engineering practice. In order
dimensional
Element
is is
a common
engineering
practice.
In order
to
dimensionalFinite
ElementMethod
Methodanalysis
a common
engineering
practice.
order
to
model theFinite
geometry,
a real
spareanalysis
impeller
has been scanned
using
visibleInlight
band
model
the
geometry,
a
real
spare
impeller
has
been
scanned
using
visible
light
band
method.
to model the geometry, a real spare impeller has been scanned using visible light band
method.
Thanks
to
3D
scanning,
the
object
was
transferred
to
the
computer
environment
Thanks
3D scanning,
the object the
wasobject
transferred
to the computer
in a short
method.toThanks
to 3D scanning,
was transferred
to the environment
computer environment
in
a
short
time
and
with
good
accuracy.
The
obtained
geometry
is
shown
in
Figure
4. One
time
and with
accuracy.
The obtained
geometrygeometry
is shownisinshown
Figurein4.Figure
One cannot
in a short
time good
and with
good accuracy.
The obtained
4. One
cannot
forget
that
the
impeller
was
mounted
on
the
shaft
with
two
keyways.
However,
forget
the impeller
was mounted
on the shaft
withshaft
two with
keyways.
However,However,
the exact
cannotthat
forget
that the impeller
was mounted
on the
two keyways.
the
radius
ofcorners
keyways
corners
could
not
found
from
scanning
radius
of keyways
could
not could
be
found
from
the 3D
scanning
technique,
so technique,
it was
the exact
exact
radius of
keyways
corners
not
bebe
found
from
thethe
3D3D
scanning
technique,
so
it
was
assumed
to
be
0.5
mm.
assumed
to
be
0.5
mm.
so it was assumed to be 0.5 mm.
Figure 4. 3D model of the geometry of the impeller under analysis.
Figure
ofof
thethe
impeller
under
analysis.
Figure 4.
4. 3D
3Dmodel
modelofofthe
thegeometry
geometry
impeller
under
analysis.
Energies 2022, 15, x FOR PEER REVIEW
Energies 2022, 15, 4183
4 of 1
4 of 11
The material properties assigned for the 3D geometry were those of precipitation
3
hardening
stainless
steels assigned
17-4PH, for
namely:
= 7850 kg/m
density, ν = 0.3—
The material
properties
the 3Dd geometry
were—material
those of precipitation
5
3
Poisson’s ratio
andsteels
E = 217-4PH,
× 10 MPa—Young’s
modulus.
hardening
stainless
namely: d = 7850
kg/m —material density, ν = 0.3—
Poisson’s
and boundary
E = 2 × 105conditions
MPa—Young’s
modulus. to those of the shrink fit and nomina
The ratio
applied
corresponded
The applied
boundary
conditions
corresponded
to those
of the
shrink fit
and nominal
rotational
speed.
The design
shrink
fit of this rotor
was
relatively
high—approximately
rotational
speed.
The
design
shrink
fit
of
this
rotor
was
relatively
high—approximately
2‰
2‰ with the shaft diameter of ϕ175 mm. In absolute numbers, the exact value
of the dif
with the shaft diameter of φ175 mm. In absolute numbers, the exact value of the difference
ference between the shaft and sleeve diameters was equal to 0.34 mm. The given nomina
between the shaft and sleeve diameters was equal to 0.34 mm. The given nominal speed of
speed of the compressor was equal to 9300 rpm.
the compressor was equal to 9300 rpm.
meshof
ofthe
themodel
modelisispresented
presentedininFigure
Figure5.5.ItItconsisted
consistedofof185,403
185,403nodes
nodesand
and 62,670
AAmesh
elements.
The
mesh
was
mainly
an
unstructured
grid
with
high
enough
mesh
density in
62,670 elements. The mesh was mainly an unstructured grid with high enough mesh
chamfers
and corners
to reveal
stressstress
concentration.
density
in chamfers
and corners
to expected
reveal expected
concentration.
Figure
thethe
impeller
model
withwith
a shaft
section.
Figure5.5.Mesh
Meshofof
impeller
model
a shaft
section.
To estimate the stress distribution in the impeller, the superposition method was used.
To estimate the stress distribution in the impeller, the superposition method was
The multistep numerical analysis was performed, including following consecutive steps:
used. The multistep numerical analysis was performed, including following consecutive
1.
Estimation of stresses induced by the shrink fit—no keyways taken into consideration;
steps:
2.
Estimation of stresses caused by the centrifugal force;
1. Estimation
Estimationofof
stresses
induced
byshrink
the shrink
fit—notaken
keyways
taken into considera
3.
stresses
induced
by the
fit—keyways
into account;
4.
Superposition
of all the conditions—stresses induced by the centrifugal force, keyways
tion;
the shrink
2. and
Estimation
offit.
stresses caused by the centrifugal force;
the first step
the shrink
fit by
wasthe
taken
into fit—keyways
account. This load
condition
was
3. In
Estimation
of only
stresses
induced
shrink
taken
into account;
applied
to
the
model
by
means
of
the
diameter
difference
between
the
shaft
and
the
sleeve.
4. Superposition of all the conditions—stresses induced by the centrifugal force, key
In Figure
the equivalent
ways6 and
the shrinkstress
fit. distribution is shown. Its maximal value in the inner
chamfer, where the stress concentration occurs, was equal to 336 MPa. In the fitting surface
In the first
step
only
thearound
shrink318
fitMPa.
was taken into account. This load condition was
the equivalent
stress
value
was
applied
to
the
model
by
means
of
the
diameter
theapplied
shaft and
the sleeve
In the second step, both the shrink fit and thedifference
centrifugalbetween
force were
to the
In Figure
the hoop
equivalent
stress distribution
shown. Its
maximal value
in 7.
the inne
model.
The6 von
stress distribution
for theseisconditions
is presented
in Figure
Its
exact maximal
value
wasconcentration
406 MPa, which
corresponds
to 59%
of MPa.
the yield
stress
chamfer,
where the
stress
occurs,
was equal
to 336
In the
fitting sur
(σ
689 equivalent
MPa). In thestress
fittingvalue
surface
though,
the318
hoop
normal stress value was around
g = the
face
was
around
MPa.
366 MPa,
slightly
greater
than
in
the
first
step
without
centrifugal
forceforce
actingwere
on theapplied
rotor. to the
In the second step, both the shrink fit and the centrifugal
model. The von hoop stress distribution for these conditions is presented in Figure 7. Its
exact maximal value was 406 MPa, which corresponds to 59% of the yield stress (σg = 689
MPa). In the fitting surface though, the hoop normal stress value was around 366 MPa
slightly greater than in the first step without centrifugal force acting on the rotor.
Energies 2022, 15, 4183
and not
precisely
so the
the simulation
result presented
is more of qualitative
Finally,
themeasured,
last step of
was here
superposition
of the totaland
lo
quantitative
importance.
fit, centrifugal force and keyways. The hoop stress distribution for this com
Finally, the last step of the simulation was superposition of the total loading: sh
presented in Figure 9. The results show that the stress level in the fitting flan
fit, centrifugal force and keyways. The hoop stress distribution for this complex sta
to 394 MPa, which corresponds to 57% of the yield stress for the 17-4PH
ma
5 of 11
presented in Figure 9. The results show that the stress level in the fitting flange was eq
portant
stress
concentration
still
being
predicted
in for
keyway
corners.
to
394 MPa,
which
corresponds to
57%
of the
yield stress
the 17-4PH
material, an
portant stress concentration still being predicted in keyway corners.
Figure
stress
distribution
theinimpeller,
caused
by the
Figure
6.6.Hoop
stress
distribution
in theinimpeller,
caused
by
the shrink
fit. shrink
Figure
6.Hoop
Hoop
stress
distribution
the
impeller,
caused
by thefit.
shrink fit.
Figure
Hoop
stress
distribution
induced
the centrifugal
and fit.
the shrink fit.
Figure 7.7.Hoop
stress
distribution
induced
by the by
centrifugal
force andforce
the shrink
In the third case the shrink fit was taken into account, the same as in the first step but
Figure
Hoop
stress
distribution
induced
by the
and the shrink fit
this
time 7.
two
keyways
were
extruded from
the model.
The centrifugal
hoop normal force
stress distribution
for this simulation is presented in Figure 8, where the stress level value reached 322 MPa
on the fitting flanges. Special attention should be paid to the keyway corners where the
stress concentration occurs, indicating the highest stress levels. However, its value is
strongly dependent on mesh density and keyways corners radius, the latter only assumed
and not precisely measured, so the result presented here is more of qualitative and not
quantitative importance.
Finally, the last step of the simulation was superposition of the total loading: shrink
fit, centrifugal force and keyways. The hoop stress distribution for this complex state
is presented in Figure 9. The results show that the stress level in the fitting flange was
equal to 394 MPa, which corresponds to 57% of the yield stress for the 17-4PH material,
an important stress concentration still being predicted in keyway corners.
Energies 2022, 15, 4183
Energies 2022, 15, x FOR PEER REVIEW
6 of 11
6 of 11
Figure
8. Hoop
stress
distribution caused
byby
thethe
shrink
fit andfit
theand
presence
of the keyways.
FigureFigure
8. Hoop
stress
distribution
caused
shrink
the presence
of the keyways.
8. Hoop stress distribution caused by the shrink fit and the presence of the keyways.
Figure 9. Hoop stress distribution caused by the centrifugal force, the shrink fit and the presence of
the keyways.
To conclude the numerical analysis results, the method of assembling the impeller
on the shaft is probably overdone, meaning that the use of both the tight shrink fit and
two keyways is excessive and introduces significant stress levels, especially stress concenFigureFigure
9. Hoop
stress
distribution
caused
centrifugal
the fit
shrink
fitpresence
and the presence
9. in
Hoop
stress
distribution
caused byby
thethe
centrifugal
force,force,
the shrink
andin
the
tration
keyways
slots. The impeller
design
should have
been
optimized
terms of of
the keyways.
the
keyways.
stress
levels and the shrink fit tightness, as well as keyways design and number being
parameters of the optimization.
To conclude the numerical analysis results, the method of assembling the impeller on
To conclude
thestep
numerical
analysis
theinfluence
methodof of
assembling
The analysis
by step allowed
us to results,
evaluate the
each
loading statethe impel
the shaft is probably overdone, meaning that the use of both the tight shrink fit and two
stressislevels.
The obtained
numerical
results confirm
undoubtedly
the hypothesis
on theonshaft
probably
overdone,
meaning
that the
use of both
the tightthat
shrink fit a
keyways is excessive and introduces significant stress levels, especially stress concentration
two keyways
excessive
and introduces
significant
stress levels,
especially
stress conce
in keywaysisslots.
The impeller
design should
have been optimized
in terms
of stress levels
andinthe
shrink fit slots.
tightness,
well as keyways
design
and number
beingoptimized
parameters of
tration
keyways
Theas impeller
design
should
have been
in terms
the
optimization.
stress levels and the shrink fit tightness, as well as keyways design and number bei
parameters of the optimization.
The analysis step by step allowed us to evaluate the influence of each loading sta
on stress levels. The obtained numerical results confirm undoubtedly the hypothesis th
Energies 2022, 15, 4183
7 of 11
The analysis step by step allowed us to evaluate the influence of each loading state
on stress levels. The obtained numerical results confirm undoubtedly the hypothesis that
the stress concentration in the keyway corner was the cause and the starting point of the
failure mechanism.
4. Influence of the Working Fluid on the Failure
The machine under analysis is a six-stage compressor operating in the hydrocracking
installation. We have limited choice of materials to manufacture parts for turbomachines
working in aggressive environment, namely destructive fluids. According to standards,
ISO 15156-1 or the corresponding American NACE MR0175 [9], 17-4PH steel material
is commonly applied in many aggressive environments due to its high yield stress and
corrosion resistant properties. Nevertheless, there are publications where it has been
shown that this steel is susceptible to Sulfide Stress Cracking under specific environment
conditions [10,11].
Sulfide Stress Cracking is a phenomenon requiring three factors to occur, namely:
a corrosive medium, a susceptible material and a constantly present significant stress state.
The damage caused by this process is mostly sudden and unpredictable. The failure can
occur after hours, months or even years of normal exploitation.
The problem of stress concentration was described in chapter “Stress analysis”. In this
chapter, we take a closer look at the working fluid influence on the failure.
The components of the working medium were mainly hydrogen (90.2% vol.), methane
(4.2% vol.), hydrogen disulfide (3.7% vol.) and other hydrocarbons (the remaining part).
Also, from the process point of view, a certain amount of water vapor is acceptable in
the working medium (up to 0.12% of the molar fraction). All these data indicate that the
compressor works with a relatively acid medium, since the presence of H2 S causes a low
pH. The acid environment, under some conditions, can induce hydrogen absorption by the
metal and thus, as a result, the phenomenon referred to as hydrogen embrittlement.
Not only the composition of the working fluid but also its pressure has a significant
role in the Sulfide Stress Cracking phenomenon. It is even more likely to occur when the
working medium is at a pressure of 0.448 MPa or higher and when the partial pressure of
H2 S in the mixture is higher than 0.545 kPa. In the compressor under consideration, both
conditions were fulfilled.
Thus, according to ISO 15156-1, chapter 7.2.1.2, both parameters (the fluid acidity
and its pressure) need to be taken into account to evaluate a threat of failure due to the
Sulfide Stress Cracking (SSC). This is achieved first by finding the pH level based on partial
pressures of H2 S and carbon dioxide (CO2 ) in the working medium. In our case, there
was no CO2 fraction in the medium under analysis, only hydrogen disulfide with a partial
pressure of 545 kPa to be considered. For this pressure, the acidity of pH = 3.5 was read
(Figure 10). Second, the compressor operation point can be found by transferring this
information to Figure 11, where four regions are indicated based on partial pressures of
H2 S and pH of the environment.
The chart presented in Figure 11 describes four possible levels of environment aggressiveness, namely:
•
•
•
•
SSC Region 0, where the H2 S partial pressure is lower than 0.3 kPa. The choice of steel
in this region requires no special precautions;
SSC Region 1, where the H2 S partial pressure is higher than 0.3 kPa. There is a minimal
risk of steel cracking;
SSC Region 2, where the H2 S partial pressure is higher than 0.3 kPa. There is a risk of
steel cracking;
SSC Region 3, where the H2 S partial pressure is higher than 0.3 kPa. The steel is
especially susceptible to stress corrosion.
The compressor in the case under consideration was operating in Region 3.
Energies 2022, 15, 4183
Figure
Figure 12
12 presents
presents the
the partial
partial pressure
pressure of
of water
water vapor
vapor and
and saturated
saturated vapor
vapor at
at inlets
inlets of
of the
the
consecutive
consecutive compressor
compressor stages.
stages. It
It can
can be
be observed
observed that
that at
at the
the first
first stage
stage inlet,
inlet, the
the tempertemperature
ature was
was 54
54 °C.
°C. Under
Under these
these conditions,
conditions, the
the partial
partial pressure
pressure of
of water
water vapor
vapor is
is higher
higher than
than
the
the saturated
saturated vapor
vapor pressure.
pressure. This
This indicates
indicates the
the condensation
condensation of
of the
the liquid
liquid H
H22O
O phase
phase and
and
8 of 11
thus,
thus, the
the compressor
compressor first
first stage
stage impeller
impeller might
might be
be in
in contact
contact with
with humid
humid and
and acid
acid hydrohydrogen
gen disulfide.
disulfide.
Figure
pH
vs.
partial
pressure
of
CO
Figure
10.
Environment
pH
vs.vs.
partial
pressure
of H
H
and
CO22[5].
[5].
Figure10.
10.Environment
Environment
pH
partial
pressure
of22SSHand
2 S and CO2 [5].
Figure
of
the
environment,
according
to
pressure
of
Figure
11.
Aggressiveness
ofof
thethe
environment,
according
to [5],
[5],
(X—partial
pressure
of H
H22S,
S, Y—pH
Y—pH
Figure11.
11.Aggressiveness
Aggressiveness
environment,
according
to (X—partial
[5],
(X—partial
pressure
of
H2 S, Y—pH
of
of
the
environment).
ofthe
theenvironment).
environment).
The presence of dry hydrogen disulfide favors development of the Sulfide Stress
Cracking process but there is more. Some investigations [3] suggest that the additional
presence of liquid H2 O can rapidly increase the crack development. The chart shown
in Figure 12 presents the partial pressure of water vapor and saturated vapor at inlets
of the consecutive compressor stages. It can be observed that at the first stage inlet, the
temperature was 54 ◦ C. Under these conditions, the partial pressure of water vapor is
higher than the saturated vapor pressure. This indicates the condensation of the liquid
H2 O phase and thus, the compressor first stage impeller might be in contact with humid
and acid hydrogen disulfide.
Energies 2022, 15, x FOR PEER REVIEW
Energies 2022, 15, 4183
Energies 2022, 15, x FOR PEER REVIEW
9 of 11 9 of 11
9 of 11
84
60
60
84
50
50
partial pressure of water vapor
saturated vapor pressure
40
74 69
[kPa] 30
[kPa] 30
54 64
20
20
10
10
0
79 74
partial pressure
of watervapor
vapor
saturated
pressure
40
54
40
79
59
69
64
59
50
0
60
70
80
90
temperature [°C]
40
50
60
70
80
90
Figure 12. Partial pressure of water vapor in the working medium vs. saturated vapor pressure.
temperature
[°C]
Figure 12. Partial
pressure of
water vapor in the working medium vs. saturated vapor pressure.
5. Influence of the Temperature
Influence of
the Temperature
17-4PH is5.martensitic
precipitated
hardening stainless steel. The final material yield
Figure 12. Partial pressure of water vapor in the working medium vs. saturated vapor pressure.
stress and surface17-4PH
hardness
depends on
the hardening
temperature.
TheThe
ISO
15156-1
is martensitic
precipitated
hardening
stainless steel.
final
material yield
standard limitsstress
the permissible
hardness
to on
33 the
HRC,
which corresponds
328 ISO
HV. 15156-1
and surfacesurface
hardness
depends
hardening
temperature.toThe
5. Influence of the Temperature
standard
limits theofpermissible
hardnessthe
to 33
HRC, which
After the failure,
the Strength
Materials surface
Lab measured
hardness
of thecorresponds
material to 328
17-4PH
martensitic
precipitated
hardening
stainless
material
HV.
After
the failure,
the sleeve
Strength
measured
the
hardness
ofyield
the material
sample
taken is
from
the impeller
inner
toof
beMaterials
equal
to Lab
332steel.
HV. The
Thatfinal
value
approached
stress
and
surface
hardness
depends
on
the
hardening
temperature.
The
ISO
15156-1
sample
taken
from
the
impeller
inner
sleeve
to
be
equal
to
332
HV.
That
value
approached
the limit advised by the ISO 15156 standard.
standard
limits
the
permissible
surface
hardness
to
33
HRC,
which
corresponds
to
328
the
limit
advised
by
the
ISO
15156
standard.
The steel resistance to Sulfide Stress Cracking is highly dependent on the temperature
The
steel
resistance
to
Sulfide
Stress
Cracking
is
highly
dependent
on
the
HV.
After
the
failure,
the
Strength
of
Materials
Lab
measured
the
hardness
of
the
material
of the working medium. The chart presented in Figure 13 shows an influence of thetemperature
theimpeller
working
medium.
The
presented
inHV.
Figure
13value
shows
an influence
sample taken temperature
fromofthe
inner
sleeve
tochart
befailure
equal
to
ThatThe
approached
environment
on the
mean
time
to
of332
a sample.
sample
under of the
environment
temperature
on
the
mean
time
to
failure
of
a
sample.
The
sample
under testthe limit
advised
ISO 15156
standard.
testing
was
madeby
ofthe
high-yield
steel
and subjected to an aggressive H2 S environment.
ing
was
made
of
high-yield
steel
and
subjected
to
an
aggressive
H
2S environment. A conThe steelcan
resistance
tofrom
Sulfide
is highly
dependent
thethe
temperaA conclusion
be drawn
thisStress
chart Cracking
that martensitic
stainless
steels on
have
lowest
clusionmedium.
can
be drawn
from this
chart that
martensitic
stainless
steels
have of
thethe
lowest re◦
◦
ture
of
the
working
The
chart
presented
in
Figure
13
shows
an
influence
resistance to SSC at 25 C. For higher temperatures, up to 80 C, the mean time to failure
sistance to SSC at 25 °C. For higher temperatures, up to 80 °C, the mean time to failure
environment
on the meanlevel,
timeno
to failure
a sample.
The sample under testincreases,
andtemperature
above this temperature
failuresofwere
observed.
increases, and above this temperature level, no failures were observed.
ing was
made of
high-yieldcan
steel
subjected
to anchart
aggressive
H2S14,
environment.
A conA similar
conclusion
beand
drawn
from the
in Figure
which presents
a
clusion between
can be drawn
from this chart
thatminimal
martensitic
stainlessstress
steelslevel
have
lowest rerelation
the temperature
and the
permanent
(asthe
a fraction
of
sistance
SSC as
at the
25 °C.
For higher
temperatures,
to 80 °C, For
the the
mean
time to failure
the
yield to
stress)
condition
promoting
the SSCup
occurrence.
compressor
inlet
◦ C), the
temperature
(54above
stress
level
is equal were
to 55%
of the yield stress.
increases, and
thispermissible
temperature
level,
no failures
observed.
Figure 13. Time to sample failure vs. temperature [11].
Figure13.
13. Time
Timeto
tosample
samplefailure
failurevs.
vs.temperature
temperature[11].
[11].
Figure
Energies 2022, 15, 4183
A similar conclusion can be drawn from the chart in Figure 14, which presents a relation between the temperature and the minimal permanent stress level (as a fraction of
of 11
the yield stress) as the condition promoting the SSC occurrence. For the compressor10inlet
temperature (54 °C), the permissible stress level is equal to 55% of the yield stress.
Figure 14.
14. Permanent
Permanent stress
stress level
level vs.
vs. temperature
temperature as
as the
the condition
Figure
condition favoring
favoring the
the SSC
SSC occurrence
occurrence [11].
[11].
Considering the stress levels presented in Section 2, the impeller is not particularly
exposed
to the risk
failure.
However,
should
be2,noted
that a high
concentration
Considering
theof
stress
levels
presentedit in
Section
the impeller
is not
particularly
of
stress
occurs
in
the
keyway
corners
and
these
points
are
especially
prone
to Sulfide
exposed to the risk of failure. However, it should be noted that a high concentration
of
Stress
Cracking.
stress occurs in the keyway corners and these points are especially prone to Sulfide Stress
Cracking.
6. Conclusions
The conducted failure analysis consisted of the optical analysis, the finite element
6. Conclusions
method stress analysis and the evaluation of the working fluid influence according to the
The conducted failure analysis consisted of the optical analysis, the finite element
ISO 15156 standard, as well as temperature influence on the Sulfide Stress Cracking of
method stress analysis and the evaluation of the working fluid influence according to the
17-4PH martensitic precipitated hardening stainless steel. The analysis allows us to draw
ISO 15156 standard, as well as temperature influence on the Sulfide Stress Cracking of 17several conclusions.
4PH martensitic precipitated hardening stainless steel. The analysis allows us to draw
First, the stress analysis suggests the presence of a high level of stress concentration
several
conclusions.
in the keyway
corners. Its value could not be calculated with a high level of confidence
First,
the stress analysis
suggests
the over
presence
a high
levela of
stress point
concentration
but from conducted
simulations
it was
900 of
MPa.
From
design
of view,
in
the
keyway
corners.
Its
value
could
not
be
calculated
with
a
high
level
ofshaft
confidence
the original shrink fit should be sufficient to transfer the torque from the
to the
but
from The
conducted
simulations
it wasweakens
over 900
From
a design
pointheterogeneity
of view, the
impeller.
presence
of the keyways
theMPa.
shrink
fit and
introduces
original
shrink
fit should in
bethe
sufficient
to transfer
from
to the impeller.
of the stress
distribution
matching
surfacesthe
of torque
the shaft
andthe
theshaft
sleeve.
The presence
of
the
keyways
weakens
the
shrink
fit
and
introduces
heterogeneity
of the
According to the ISO 15156 standard, the steel the impellers were made of is especially
stress
distribution
in
the
matching
surfaces
of
the
shaft
and
the
sleeve.
exposed to Sulfide Stress Cracking induced by the working medium. The damaged first
the ISO in
15156
standard,
steel the impellers
ofpressure
is espestageAccording
wheel wastoworking
an acid,
humidthe
environment
of H2 S, were
at themade
partial
cially
exposed
to
Sulfide
Stress
Cracking
induced
by
the
working
medium.
The
damaged
of 545 kPa, which is known to be an initiator of SSC in 17-4PH steel. A negative impact
first
wheel wasby
working
an temperature
acid, humid in
environment
of ◦H
atwhich
the partial
preswas stage
even reinforced
the inletingas
this stage (54
C,2S,for
the partial
sure
of 545
known
to be
anthe
initiator
of SSC
in 17-4PH
steel. A negative impressure
of kPa,
waterwhich
vaporiswas
higher
than
saturated
vapor
pressure.).
pact was
even reinforced
by the inlet
gastotemperature
in this
stage
(54 °C,
forthe
which
the
The impeller
surface hardness
equal
the ISO 33 HRC
limit
indicates
that
material
partial
pressure
of
water
vapor
was
higher
than
the
saturated
vapor
pressure.).
of the wheel is susceptible to SSC. Increased hardness in the fracture surface (35 HRC)
The impeller
surface hardness
equalhad
to the
ISO 33
HRCplace
limitin
indicates
that the
mateindicates
that the embrittlement
process
already
taken
the impeller
material.
rial ofOn
thethe
wheel
is of
susceptible
SSC. Increased
hardness
in the fracture
(35 HRC)
basis
the abovetostatements,
it can
be concluded
that thesurface
presence
of the
indicates
the embrittlement
had already
taken place in the impeller material.
followingthat
unfavorable
conditionsprocess
was highly
probable:
basis ofenvironment
the above statements,
it can be
concluded
thatatthe
of the
• On
Thethe
aggressive
containing humid
hydrogen
sulfide
thepresence
partial pressure
following
unfavorable
conditions
was
highly
probable:
of 545 kPa;
•
•
The
aggressive
humid
hydrogen
the normal
partial stress;
presPermanent
stressenvironment
concentrationcontaining
in the keyway
corners:
over 900sulfide
MPa ofathoop
The use
of kPa;
17-4PH stainless steel prone to SSC as the impeller material may have
sure
of 545
contributed to the crack failure of the first stage compressor impeller wheel due to the
Sulfide Stress Cracking phenomenon.
To avoid similar compressor failures in the future, it is advised to redesign the way the
impeller is mounted on the shaft or select a different material for the impeller wheel in accordance with the ISO 15156-3 standard. Those will be the subject of further investigations.
Energies 2022, 15, 4183
11 of 11
As an alternative to the above-mentioned suggestions, a surface treatment in a low
temperature plasma nitriding process can be considered [12,13]. This treatment introduces
compressive stresses to the material and, thus, removes one of three necessary factors for
Sulfide Stress Cracking to occur.
Author Contributions: Conceptualization, J.Ł. and Z.K.; methodology, J.Ł.; software, J.Ł. and E.T.;
validation, Z.K. and E.T.; formal analysis, Z.K.; investigation, J.Ł. and E.T.; writing—original draft
preparation, J.Ł.; writing—review and editing, E.T.; supervision, Z.K. All authors have read and
agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Conflicts of Interest: The authors declare no conflict of interest.
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
Fantechi, F.; Innocenti, M. Chloride stress corrosion cracking of precipitation hardening S.S. impellers in centrifugal compressor.
Laboratory investigations and corrective actions. Eng. Fail. Anal. 2001, 8, 477–492. [CrossRef]
Musgrove, G.O.; Matheidas, M.T.; Beck, G.C.; Uptigrove, S.O. Measured Effects of Liquid Distribution on Compressor Performance
during Wet Gas Ingestion. In Proceedings of the 44th Turbomachinery and 31st Pump Symposia, Houston, TX, USA, 14–17
September 2015.
Chu, Q.; Zhang, M.; Li, J. Failure analysis of impeller made of FV520B martensitic precipitated hardening stainless steel. Eng. Fail.
2013, 34, 501–510. [CrossRef]
Dowson, P.; Bauer, D.; Laney, S. Selection of Materials and Material Related Processes for Centrifugal Compressors and Steam
Turbines in the Oil and Petrochemical Industry. In Proceedings of the 37th Turbomachinery Symposium, Houston, TX, USA, 8–11
September 2008.
ANSI/NACE MR0175/ISO 15156 Standard; Petroleum and Natural Gas Industries—Materials for Use in H2S-Containing Environments in Oil and Gas Production. International Organization for Standardization: Geneva, Switzerland, 2015.
Mostofi, J.; Zamanian, M. Stress Cracking of Stainless Steel Safety Gate Valve Stem. Mater. Perform. Mag. 2008, 47, 64–67.
Sun, J.; Chen, S.; Qu, Y.; Li, J. Review on Stress Corrosion and Corrosion Fatigue Failure of Centrifugal Compressor Impeller.
Chin. J. Mech. Eng. 2015, 28, 217–225. [CrossRef]
Nie, D.; Chen, X.; Fan, Z.; Wu, Q. Failure analysis of a slot-welded impeller of recycle hydrogen centrifugal compressor. Eng. Fail.
Anal. 2014, 42, 1–9. [CrossRef]
Gutierrez, I. NACE MR0175 2015 and the Stainless Steel Industry. Stainl. Steel World Mag. 2016, 51–55.
Davis, J.R. Stainless Steels (ASM Specialty Handbook); ASM International: Almere, The Netherlands, 1995; ISBN 0871705036.
Della Roverys Coseglio, M.S. Sulphide stress cracking of 17-4 PH for applications in oilfield components. Mater. Sci. Technol. 2017,
33, 1863–1878. [CrossRef]
Kochmański, P. Azotowanie gazowe stali nierdzewnych utwardzanych wydzieleniowo. Eksploat. Niezawodn. 2003, 2, 41–44.
Surowska, B. Wybrane Zagadnienia z Korozji i Ochrony Przed Korozja;˛ Lublin University of Technology: Lublin, Poland, 2002.