PPCHEM Understanding Variations in Flow-Accelerated Corrosion Wear Rates in HRSG Evaporator Tubes

Understanding Variations in Flow-Accelerated Corrosion

Wear Rates in HRSG Evaporator Tubes
James Malloy, Jan Rusås, and Mark Taylor

ABSTRACT
Very large differences in the rates of tube wall thinning due to flow-accelerated corrosion (FAC), or in some cases fluid
erosion, are observed in the low-pressure (LP) evaporator tubes of certain HRSG designs. The tubes located near the
duct wall and occasionally near the gaps between module bundles have more rapid thinning. Tubes in a given row
nominally should have very similar process conditions, both on the gas-side and on the waterside. Different wall
thinning rates mean that process conditions differ across the tube row.

Computational fluid dynamics (CFD) simulations have been used to investigate the gas-side process conditions in
tube assemblies and to determine the impact of tube location in the row on heat transfer. The impact of differing tube
heat fluxes on waterside process conditions was analyzed by thermohydraulic simulations using a computer model of
the tube assemblies. A correlation between increased thinning rates as a function of tube position and LP evaporator
process conditions was identified.
Author's Copy

INTRODUCTION
Tube thinning caused by flow-accelerated corrosion (FAC) alkalizer is ammonia. In this case the pH is lowered in the
is one of the most frequent causes of failure in the heat liquid phase of two-phase regions because ammonia pref-
exchanger tubes of heat recovery steam generators erentially migrates to the steam phase, i.e. the distribution
(HRSGs), with localized progressive thinning of tube walls ratio of ammonia is greater than 1. This is also true for cer-
eventually leading to rupture and leaks. Any carbon steel tain amines that are used as alternatives to ammonia.
tubes that carry water or wet steam are potentially at risk.
Even when water chemistry and process conditions are
The overall mechanisms of FAC, wherein the protective set by design to minimize the risk of FAC, experience
magnetite layer of the base metal is dissolved at an accel- shows that localized thinning can still be observed in the
erated rate, are relatively well understood [1–6]. Key LPEVAP modules, particularly in the two-phase region
factors controlling magnetite dissolution are temperature, near the outlet header and in the risers to the low-pressure
water chemistry and fluid mass transfer rate. Two strate- (LP) drum.
gies for avoiding or at least greatly reducing the problem in
operating plants are available. The easiest in most cases An analysis of data from the LPEVAP modules in different
is to change the water chemistry by raising the pH and plants pointed out that fluid velocities in the upper section
creating an oxidizing environment. If this is not sufficient or of the tube and the riser can be very high in some HRSG
if other factors (such as the presence of copper-alloy designs [7]. Very high two-phase flow velocities will
components in the steam cycle) prevent this, then it may mechanically damage the magnetite layer and increase
be necessary to replace carbon steel tube sections the FAC rate. At some point mechanical damage to
encountering wear with a more resistant material, namely magnetite will become the driving factor instead of chem-
low-alloy steel containing some chromium. istry or temperature. As velocities increase further, direct
erosion of the base metal will ensue [8].
One area in the HRSG that is prone to FAC is the low-
pressure evaporator (LPEVAP). Fluid temperatures here If the two-phase fluid has a high vapor fraction and a high
are often at or near the peak of magnetite dissolution velocity, the entrained water droplets will erode surfaces.
(150 °C), flow rates can be high, and water chemistry pH This form of erosion is commonly labeled liquid droplet
control is typically all-volatile. The most common volatile impingement (LDI) [9]. Affected areas are pipe bends and

© 2017 by Waesseri GmbH. All rights reserved.

92 PowerPlant Chemistry 2017, 19(2)

 Understanding Variations in Flow-Accelerated Corrosion Wear Rates in HRSG Evaporator Tubes PPCHEM

locations after flow perturbations, or wherever there are pH pH value [– log10 [OH– ]/mol · dm–3]
very high local flow velocities [10]. The droplets contained g oxygen content [pg · kg–1]
in the two-phase mixture do not follow changes in the bulk
flow direction and impinge against the pipe wall in the "line
B and N are factors dependent on temperature and alloy-
of sight" with a force that is dependent on droplet size,
ing elements in the steel, and kc is a geometry factor for the
velocity and angle of impact [11,12].
component type under consideration (elbow, "T" piece,
etc.). The time correction factor, f(t), of the FAC rate equa-
The wear mechanism at work is important in that while
tion is a polynomial function of exposure time that yields
FAC can be controlled by keeping the water chemistry (pH
lower instantaneous wear rates at longer exposures.
and dissolved oxygen, DO) within the appropriate control
band, mechanical erosion cannot. Only a change to flow
For two-phase FAC, the basic condition is annular flow
geometry and/or liquid phase flow velocities would lower
and the reference velocity used is not the velocity of a
the damage rate [3,13–15].
two-phase mixture, but the mean velocity in the film of
water on the wall of the component, WF. A simplified
Earlier work involving field measurements and simulations
equation for this is:
with different LP evaporator designs from operating plants
showed that bulk flow velocities were certainly a factor in m· ' 1 – xst
the risk of thinning occurring. A design limit of no more WF = — ——— (2)
␳w 1 – ␣
than 10 m · s–1 was suggested for the LPEVAP tube flow
velocity [16]. A mechanical erosion (or LDI) component where
was postulated to explain very rapid thinning at some
WF water film mean velocity [m · s–1]
locations.
m· ' mass flux [kg · m–2s–1]
␳w saturated water density [kg · m–3]
Author's Copy

An interesting observation was that in some of the exam-

xst steam quality [–]
ined plants, wear rates were much higher at the edge of
␣ void fraction [–]
the tube modules. Failures occurred in edge tubes at one
of the plants after less than 5 000 hours of operation.
The eventual impact of erosion is separately modeled
using a formulation for LDI developed by Keck [2]. The
This effect has also been noted in other field inspections,
model requires input data for internal pipe diameter, mass
with failures preferentially occurring at these locations.
flow rate, operating temperature and thermodynamic
The mechanism is commonly attributed to "gas bypass-
quality; these inputs are then used to calculate all of the
ing" along the edge of the HRSG duct. Given that water
terms in the equation to find the wear rate caused by LDI.
chemistry and fluid temperatures are essentially identical
The formulation for wear rate yields:
in all tubes of a given row, the phenomenon is assumed
to be due to differing fluid velocities between individual Cm (1 – x) pwFeFhV 4
tubes. m" = ————————— (3)
(Pec)2 Ac

The presence of such an "edge effect" has been examined where

here by simulating gas flows in one of the plants analyzed
previously [7]. The goal was to assess the impact of mod- m" LDI wear rate [m · s–1]
ified gas flow on the FAC wear rates that are predicted m total mass flow of mixture [kg · s–1]
using a formulation developed by Kastner and collabora- C constant [–]
tors at KWU [14] using empirical data. In the KWU-Kastner x thermodynamic quality [–]
model, the pH, oxygen content, liquid velocity, geometri- Fe entrainment fraction [–]
cal factor, total content of chromium and molybdenum in
Fh hitting fraction [–]
steel, and operating temperature are the inputs. The FAC
rate, ⌬␸R, is calculated using the following equation: V characteristic velocity [m · s–1]
P oxide hardness [Pa]
⌬␸R = 6.35 · kc(B · eN·w · [1 – 0.175 · (pH – 7)2] (1) ec critical strain to fracture of the oxide [–]
· 1.8 · e–0.118g + 1) · f(t) Ac characteristic wear area [m2]
where
Suggested values of C, Fh and ec are given in [2]; the
⌬␸R calculated specific rate of [µg · cm–2 per hour]
remaining terms are calculated using plant process data.
material loss
The entrainment fraction, Fe, is the percentage of water in
kc geometry factor [–]
the annular flow which is sheared off the liquid film to form
w flow velocity [m · s–1] droplets, accelerated by the gas.

PowerPlant Chemistry 2017, 19(2) 93

 PPCHEM Understanding Variations in Flow-Accelerated Corrosion Wear Rates in HRSG Evaporator Tubes

CFD SIMULATIONS OF LP EVAPORATOR GAS FLOW end sealing baffle was removed. The upstream gas flow
conditions were identical in both cases.
A computational flow and temperature analysis has been
carried out for the downstream section of economizer and
LPEVAP tubes. A commercial computational fluid dynam- LPEVAP HPECO
ics (CFD) solver [17] was used for the analysis.

The flow model analysis was performed assuming steady

state, ignoring the transient small-scale vortex shedding
from the tubes and fins. A Reynolds-averaged Navier- HRSG
6 m · s–1
Stokes (RANS) equation approach was used for modeling side wall 598 K

the flue-gas flow. The RANS approach is the most practi-

cal choice for a CFD analysis for industrial applications
due to its ability to model turbulent flows with sufficient
accuracy and with reasonable computational resources.
The realizable k-epsilon turbulence model with standard Figure 1:
wall-functions was used for closure of the Reynolds stress Module layout plan view.
terms. Accuracy was improved through the use of higher HPECO high-pressure economizer
order discretization schemes for the advection terms and
the energy equation, the PRESTO1 scheme for the pres- Tube
sure interpolation and a double precision solver. Fin
Inside tube
Only the LPEVAP tubes and the downstream sections of
the economizer were modeled with input gas temperature, Fin Periodic boundary Tube
Author's Copy

flow rate and tube internal transfer coefficients obtained

from a separate thermal process modeling program [18].
Gas and tube metal properties such as specific heat, den- Figure 2:
sity and material conductivity were calculated as tempera- Fin geometry as modeled.
ture dependent.

The module geometry was simplified to allow for a man-

As-Built Duct
ageable calculation effort while retaining the essential
features of the gas flow regime. Vertical tubes were mod- For the as-built case, Figure 3 and Figure 4 show the
eled as one fin pitch with periodic boundary conditions gas temperature and velocity distribution at a plane
between the fins and only one side of the duct was con- between the fins. The flow direction is from right to left.
sidered to a depth of 10 tubes from the edge (Figure 1).

A pure fine hexahedral mesh was used for the computa-

tion with a total number of 8.3E6 computational cells. If
the complete section were calculated with the same mesh
resolution, it would have required 4.8E11 cells, yielding a
prohibitive computational overhead.

The fins were modeled not as spiral fins, but as angled

"slotted" fins as shown in Figure 2. The angled fins effec- Figure 3:
tively mimic the spiral fins while keeping the modeling and As-built case temperature distribution between fins [K].
computational effort manageable.

Two separate scenarios were investigated. One was with

the as-built configuration for the HRSG module, with side
gas baffles and a sealing baffle that butts against the edge
tubes. The second was for the same geometry except the

1
PRESTO is an advanced higher-order pressure interpolation
scheme, which it is advantageous to use in flows with curved Figure 4:
domains, and for natural and forced convection. As-built case velocity magnitude distribution between fins [m · s–1].

94 PowerPlant Chemistry 2017, 19(2)

 Understanding Variations in Flow-Accelerated Corrosion Wear Rates in HRSG Evaporator Tubes PPCHEM

The simulation predicts that gas will stream along the tube side baffle has a major influence; tubes on opposing sides
edge baffle with little impact on tubes located away from of the HRSG are offset. The B side layout shown in
the edge. Figure 5 was modeled, where edge tubes in rows 1 and 3
are close to the side enclosing baffle, and the heat load-
The heat flux into the different tubes was estimated by ings reflect this. The opposing side would presumably
integrating heat flux across the circumference with heat have higher predicted heat loadings in rows 2 and 4.
transfer coefficients from thermal simulation. The bulk Clearly the most significant factor overall in thermal load-
average coefficients were used. This is a reasonably accu- ing is the row number for the tubes that are located away
rate estimation given that the overall heat input along the from the edge.
tube length will be the key factor in outlet process condi-
tions; local variations along the tube can be ignored.
Modified Duct
Table 1 shows the relative heat flux in each tube in com- The second case without the sealing baffle was run
parison to a reference value within the rows and between because many HRSG configurations have this layout, with
the rows. The reference value in the former is the average tubes open to the side duct wall. Figure 6 shows the tem-
heat flux in tubes 9–12 within a row; for the latter it is the perature distribution at a horizontal plane between the
average value in tubes 9–12 of row 4. As would be fins. There is significant bypass flow between the edge
expected based on the temperature profile, the edge tubes and the duct wall. At the LPEVAP the flow is guided
effect is limited to only the three tubes nearest the duct back toward the duct center by the end baffle across the
wall. The relative position of the tube in comparison to the last two rows of tubes.

Row 1 Row 2 Row 3 Row 4 Row 1 Row 2 Row 3 Row 4

Author's Copy

Tube No. Heat Flux Heat Flux Heat Flux Heat Flux Tube No. Heat Flux Heat Flux Heat Flux Heat Flux
[%] [%] [%] [%] [%] [%] [%] [%]
1 24 –4 35 3 1 141 50 69 3
2 –5 –3 –11 –15 2 84 51 12 –15
3 –3 –1 –4 –6 3 88 55 20 –6
4 –1 1 –1 1 4 92 57 24 1
5 –1 1 1 3 5 93 57 27 3
6 0 1 2 2 6 94 57 28 2
7 0 1 2 4 7 94 57 28 4
8 0 1 2 5 8 94 57 28 5
9 0 0 1 0 9 95 56 27 0
10 0 0 0 2 10 94 56 25 2
11 0 0 0 2 11 94 56 25 2
12 0 0 –1 –4 12 94 56 24 –4

Table 1:
As-built design – Relative heat load within row (left) and between rows (right).

A B
Row 1 Row 1 Row 1 Row 1 Row 1 Row 1
Tube 47 Tube 46 Tube 45 Tube 3 Tube 2 Tube 1

Row 2 Row 2 Row 2 Row 2 Row 2 Row 2

Tube 47 Tube 46 Tube 45 Tube 3 Tube 2 Tube 1

Row 3 Row 3 Row 3 Row 3 Row 3 Row 3

Tube 47 Tube 46 Tube 45 Tube 3 Tube 2 Tube 1

Row 4 Row 4 Row 4 Row 4 Row 4 Row 4

Tube 47 Tube 46 Tube 45 Tube 3 Tube 2 Tube 1

Figure 5: Figure 6:
Staggered tube arrangement left and right duct sides in LPEVAP. Modified case temperature distribution between fins [K].

PowerPlant Chemistry 2017, 19(2) 95

 PPCHEM Understanding Variations in Flow-Accelerated Corrosion Wear Rates in HRSG Evaporator Tubes

Figure 7 shows the gas velocity distribution. The flow is

accelerated in the region with the plates blocking the flow,
and again, as with the temperature, the last two edge
tubes in the LPEVAP are most exposed.

Table 2 shows the relative heat flux in each tube in com-

parison to the average within the row and across rows for
the modified case; again the average was taken for tubes
9–12 away from the edge. Tubes up to six positions inward
Figure 7:
are affected; most notable is the very large increase in
Modified case velocity magnitude distribution between fins
heat loading of the edge tubes in rows 3 and 4. The rela- [m · s–1].
tive loadings referenced to the average of tubes 9–12 of
the 4th row are also shown.

Row 1 Row 2 Row 3 Row 4 Row 1 Row 2 Row 3 Row 4

Tube No. Heat Flux Heat Flux Heat Flux Heat Flux Tube No. Heat Flux Heat Flux Heat Flux Heat Flux
[%] [%] [%] [%] [%] [%] [%] [%]
1 5 –18 381 844 1 164 53 545 844
2 –36 –51 272 464 2 60 –8 399 464
3 –28 –52 –7 124 3 81 –10 25 124
4 –23 –46 –28 –25 4 93 1 –3 –25
Author's Copy

5 –16 –30 –21 27 5 111 31 7 27

6 –11 –9 17 7 6 124 70 57 7
7 –5 –3 14 1 7 137 81 53 1
8 –3 5 5 1 8 143 94 40 1
9 1 1 0 –3 9 152 87 35 –3
10 0 1 –1 3 10 150 87 33 3
11 1 3 –1 1 11 152 91 33 1
12 –1 –4 2 –1 12 149 79 36 –1

Table 2:
Modified design – Relative heat load within row (left) and between rows (right).

MEASURED TUBE THINNING

The LPEVAP in the HRSG units under consideration here The as-delivered wall thickness for the LPEVAP tubes is a
produce steam for deaeration only, with low operating key consideration in evaluating any measurements. The
pressures by design. Leaks were detected in the upper nominal minimum was 2.4 mm, yet in as-delivered tubes it
tube bends early in life at the extrados of the elbows was higher. An assessment of probable delivered thick-
where the tubes connect to the upper header; the first five ness was performed on new tubes of the same diameter,
failures were recorded between 3 000 and 4 000 hours of product form and wall thickness as those of the LPEVAP;
operation in the leading tube row. Thinning and failures 54 bent tubes were measured at six points at each end,
were also detected in other rows, although to a lesser for a total of 648 data points. All measurements were
extent. Metallurgical investigations of failures identified between 2.4 and 3.0 mm, with an average of 2.8 mm.
FAC and potentially erosion as contributing factors.
The UT results in four of the HRSG units show thinning in
Field measurement data was obtained using ultrasonic many LPEVAP tubes, with the most severe affecting those
thickness (UT) gauges for the LPEVAP modules in all the directly adjoining the casing in the 1st row facing the gas
HRSGs at the plant. Not all tubes were accessible, never- turbine (GT). There are other reductions in wall thickness
theless a sufficient number were read to provide a good in tubes away from the edge; these are at locations near
basis for comparison with calculations. acoustic baffles within the module. Another zone of

96 PowerPlant Chemistry 2017, 19(2)

 Understanding Variations in Flow-Accelerated Corrosion Wear Rates in HRSG Evaporator Tubes PPCHEM

reduced thickness occurs at the center along baffles assumed to be a linear function of velocity down to zero.
separating modules. The assumption is that gas flow The higher wear at edge tubes is clearly visible in the
perturbations are the cause of the accelerated thinning at measured data for the 1st row tubes (Figure 8).
all these locations.
New thermohydraulic simulations to estimate the impact
Examination of the water chemistry records over the of the "edge effect" were undertaken by diverting an
period of operation indicated that the plant had a feedwa- increasing fraction of the total exhaust flow across four
ter pH averaging about 9 and feedwater DO levels in the tubes out of 94 total. The higher gas streams increased
range of 2–7 µg · kg–1. boiling rates in the four tubes. As the bypass fraction
increased, the tube process parameters changed; the exit
velocity of the steam/water mixture increased as did the
vapor fraction, as would be expected.
PREDICTED THINNING
Earlier process simulations [7] of the HRSG LPEVAP The revised process conditions were input into the
yielded velocities for the two-phase mixture at the tube FAC/LDI calculator. The water chemistry parameters were
outlet of between 28 m · s–1 and 34 m · s–1 in the leading set to values at or near the average over the period of
(hottest) row tubes, depending on the whether the design operation (pH=9 and DO=2 µg · kg–1).
case or the actual minimum observed values were used
for the pressure in the LP system. The predicted wear for bulk tubes (normal gas flow) and
edge tubes with forced bypass is shown in Figure 9. Refer
Calculations [7] using the Keck and Kastner formulations to Table 3 for bypass fraction. Thinning increases as
with the simulated process data gave results that were bypass increases in the first tubes; the effect diminishes
consistent with measurements for tubes away from the rapidly with only a small difference in wear in the 4th row
Author's Copy

edges for pH between 8.8 and 9.2. The Kastner formula- tube compared to the bulk tube population. This is
tion is based on data for flow rates above 5 m · s–1; for because the thermohydraulic simulation of gas bypassing
lower velocities in these calculations the wear rate is assumes sequential flow with tubes in arrayed sequence.

3.0

2.5

29
30 18 19
28 46 29 31
2.0 18 19
45 2
3 3 30 45
Tube Thickness [mm]

2 20 47
46
1 1 20
1.5 47

1.0

Measured WT
0.5 ASME code min
Delivered min
WT predicted at pH 8.8
WT predicted at pH 9.2
0
Tube No.

Figure 8:
Measured and predicted wear for 1st row tubes [7].
WT wall thickness

PowerPlant Chemistry 2017, 19(2) 97

 PPCHEM Understanding Variations in Flow-Accelerated Corrosion Wear Rates in HRSG Evaporator Tubes

Case 1 Case 3
2.4 2.4
2.2 2.2
2.0 2.0
Predicted Wall Thickness [mm ]

Predicted Wall Thickness [mm ]

1.8 1.8
1.6 1.6
1.4 1.4
1 2 3 4 1 2 3 4
Row No. Row No.

Case 7 Bulk Edge Case 9

2.4 2.4
2.2 2.2
2.0 2.0
1.8 1.8
1.6 1.6
1.4 1.4
1 2 3 4 1 2 3 4
Row No. Row No.

Figure 9:
Predicted wear in forced bypass cases.

Case No. 1 3 5 7 9
Bypass fraction [%] 6 16 26 37 50
Author's Copy

–1
Row 1 Exit velocity [m · s ] 27.19 28.72 30.06 31.34 32.84
Tube 1 Steam fraction [%] 5.92 6.3 6.65 6.98 7.39
–1
Row 4 Exit velocity [m · s ] 10.97 12.01 13.03 14.04 15.28
Tube 1 Steam fraction [%] 3.1 3.2 3.31 3.43 3.59

Table 3:
Impact of simulated increased gas flow.

Figures 10 to 13 show a comparison of measured data CONCLUSIONS

and predictions from the FAC/LDI calculations. The data is
The "edge effect" of greater thinning measured by UT in
grouped by row and comprises:
the HRSGs is spread over three end tubes in row 1, which
• UT thickness measurements at three locations on the does not correspond to the heat distribution predicted by
extrados across the upper elbow of LPEVAP tubes after CFD for the as-built case. In the other three tube rows,
3 000 hours of operation. There are readings on 24 of severe thinning is confined essentially to the end tubes,
the 94 tubes in a row, as this is where readings could be which is consistent with the CFD results. The gap between
taken based on accessibility. modules at the center has nearly the same effect on thin-
ning rates and it appears that the modified pitch at the
• Maximum/minimum and average wall thickness from a acoustic baffles also has some smaller effect; neither of
UT survey of similar new tubes to determine the prob - these sections were modeled in CFD. The spread in the
able as-delivered wall thickness range for LPEVAP three UT readings across the tube elbows shows the
tubes in HRSGs. highly localized nature of the wear, the middle reading at
• Maximum/minimum and average predicted tube wall the point of the elbow being typically the lowest.
thickness, with predictions starting from both maximum
and minimum as-delivered initial values. The effect of the asymmetric tube to baffle geometry
(Figure 5) is not reflected in the UT measurements
The B module is on the left of the dotted line and the A whereas CFD predicts that it would be. Actual measured
module on the right. Shaded sections identify the location wear is similar or "mirrored" on both sides of the unit.
of acoustic baffles where the tube pitch is modified; the
dotted black line is the location of the center baffle The FAC calculations using the simulated 40 % increase
between modules. in gas flow along the edge tubes for the "heat distribution

98 PowerPlant Chemistry 2017, 19(2)

 Understanding Variations in Flow-Accelerated Corrosion Wear Rates in HRSG Evaporator Tubes PPCHEM

Row 1
3.0

2.8

2.6

2.4
Tube Thickness [mm]

2.2

2.0
HRSG 1
1.8 HRSG 3
HRSG 4
HRSG 7
1.6
Delivered max/min
Delivered average
1.4
Predicted bulk WT max/min
Predicted bulk WT
1.2 Predicted edge WT (150 % gas flow) max/min
Predicted edge WT (150 % gas flow)
1.0
Tube No.
Author's Copy

Figure 10:
Row 1 UT readings and predicted wear.

Row 2
3.0

2.8

2.6

2.4
Tube Thickness [mm]

2.2

2.0
HRSG 1
1.8 HRSG 3
HRSG 4
HRSG 7
1.6
Delivered max/min
Delivered average
1.4
Predicted bulk WT max/min
Predicted bulk WT
1.2 Predicted edge WT (150 % gas flow) max/min
Predicted edge WT (150 % gas flow)
1.0
Tube No.

Figure 11:
Row 2 UT readings and predicted wear.

PowerPlant Chemistry 2017, 19(2) 99

 PPCHEM Understanding Variations in Flow-Accelerated Corrosion Wear Rates in HRSG Evaporator Tubes

Row 3
3.0

2.8

2.6

2.4
Tube Thickness [mm]

2.2

2.0
HRSG 1
1.8 HRSG 3
HRSG 4
HRSG 7
1.6
Delivered max/min
Delivered average
1.4
Predicted bulk WT max/min
Predicted bulk WT
1.2 Predicted edge WT (150 % gas flow) max/min
Predicted edge WT (150 % gas flow)
1.0
Tube No.
Author's Copy

Figure 12:
Row 3 UT readings and predicted wear.

Row 4
3.0

2.8

2.6

2.4
Tube Thickness [mm]

2.2

2.0
HRSG 1
1.8 HRSG 3
HRSG 4
HRSG 7
1.6
Delivered max/min
Delivered average
1.4
Predicted bulk WT max/min
Predicted bulk WT
1.2 Predicted edge WT (150 % gas flow) max/min
Predicted edge WT (150 % gas flow)
1.0
Tube No.

Figure 13:
Row 4 UT readings and predicted wear.

100 PowerPlant Chemistry 2017, 19(2)

 Understanding Variations in Flow-Accelerated Corrosion Wear Rates in HRSG Evaporator Tubes PPCHEM

as built" tend to underpredict wear in rows 1 and 4 when [6] Uchida, S., Naitoh, M., Okada, H., Uehara, Y.,
compared to measured values. It is clear that the empiri- Koshizuka, S., Svoboda, R., Lister, D. H., PowerPlant
cally derived formulation for FAC and LDI used here does Chemistry 2009, 11(12), 704.
not account for the accelerating thinning rates in higher
[7] Malloy, J., Taylor, M., Fabricius, A., Graham, M.,
velocity two-phase flows of low-pressure systems. An
Moelling, D., PowerPlant Chemistry 2013, 15(4), 264.
additional component that considers accelerated wear,
presumably reflecting more rapid mechanical damage to [8] Koshizuka, S., Naitoh, M., Uchida, S., Okada, H.,
the oxide, is needed. Proc., International Symposium on the Ageing
Management & Maintenance of Nuclear Power
Variations in heat load engendered by the exhaust gas are Plants, 2010 (Tokyo, Japan). Mitsubishi Research
the driving factor in differing wear rates within a given tube Institute, Inc., Tokyo, Japan, 18.
row. The CFD gas flow profiles indicate that the modified
[9] Recommendations for Controlling Cavitation, Flash-
design, which is not uncommon in many HRSGs, would
ing, Liquid Droplet Impingement, and Solid Particle
lead to much higher wear in the edge tubes and immedi-
Erosion in Nuclear Power Plant Piping Systems,
ately adjoining tubes of rows 3 and 4. The UT measure-
2004. Electric Power Research Institute, Palo Alto,
ments indicate that the row 4 edge tubes do see higher
CA, U.S.A., 1011231.
wear in the units investigated. The implication is that the
actual gas flow profile is somewhere between the as-built [10] Barton, N. A., Erosion in Elbows in Hydrocarbon
and modified cases predicted by CFD. A more detailed Production Systems: Review Document, 2003.
CFD model coupled to the thermohydraulic simulations on Available from
an individual tube level would provide further insight into http://www.hse.gov.uk/research/rrpdf/rr115.pdf.
these differences.
[11] Higashi, Y., Narabayashi, T., Shimazu, Y., Tsuji, M.,
Ohmori, S., Mori, M., Tezuka, K., Proc., The Inter-
Author's Copy

national Congress on Advances in Nuclear Power

Plants ICAPP '09, 2009 (Tokyo, Japan). Atomic
ACKNOWLEDGMENT
Energy Society of Japan, Tokyo, Japan, Paper
The authors would like to acknowledge the contribution of # 9495. Available from
José Lorenzana of Loughborough University to this work. http://roko.eng.hokudai.ac.jp/studentadm/docum
ent/ac_activity/Proceedings/2009/ICAPP/9495.pdf.
[12] Recommended Practice – Erosive Wear in Piping
Systems, 2007. Det Norske Veritas, Høvik, Norway,
REFERENCES
DNV RP O501 – Revision 4.2.
[1] Jonas, O., Water & Wastes Digest 2009, January, 22.
[13] Henzel, N., Crosby, D. C., Eley, S. R., Erosion/ Corro-
[2] Keck, R. G., Prediction and Mitigation of Erosive sion in Power Plants Single- and Two-Phase Flow
Corrosive Wear in Steam Extraction Piping Systems, Experience, Prediction, NDE Management, 1986,
1987. Ph.D. Thesis, Massachusetts Institute of 128. Available from
Technology, Cambridge, MA, U.S.A. https://catalyst.library.jhu.edu/catalog/bib_1719794.
[3] Kastner, W., Henzel, N., Stellwag, B., Proc., Third [14] Kastner, W., Riedle, K., VGB Kraftwerkstechnik 1986,
International Topical Meeting on Nuclear Power Plant 86(12), 1171 (in German).
Thermal Hydraulics and Operations, 1988 (Seoul,
[15] Kastner, W., Erve, M., Henzel, N., Stellwag, B.,
Korea).
Nuclear Engineering and Design 1990, 119(2–3), 431.
[4] Zinemanas, D., Herszage, A., Water, Steam, and
[16] Moelling, D., Malloy, J., Graham, M., Taylor, M.,
Aqueous Solutions: Advances in Science and
Fabricius, A., Proc., ASME Power Conference, 2013
Technology for Power Generation. Proc. (on CD),
(Boston, MA, U.S.A.). The American Society of
15th International Conference on the Properties of
Mechanical Engineers, New York, NY, U.S.A.,
Water and Steam (Eds.: R. Span, I. Weber), 2008
POWER2013-98213.
(Berlin, Germany). VDI – The Association of German
Engineers, Düsseldorf, Germany. [17] ANSYS Fluent 14.5, 2012. ANSYS Inc., Cannons-
burg, PA, U.S.A.
[5] Guidelines for Controlling Flow-Accelerated Corro-
sion in Fossil Plants, 1997. Electric Power Research [18] Power Plant Simulator & Designer V15.2, 2015. KED
Institute, Palo Alto, CA, U.S.A., TR-108859. GmbH, Rodenbach, Germany.

PowerPlant Chemistry 2017, 19(2) 101

 PPCHEM Understanding Variations in Flow-Accelerated Corrosion Wear Rates in HRSG Evaporator Tubes

THE AUTHORS
James Malloy (B.S., Physics, M.S., Nuclear Engineering, Jan Rusås (M.S. and Ph.D., Thermo and Fluid Mechanics,
both at Rensselaer Polytechnic Institute, Troy, NY, U.S.A.) Aalborg University, Denmark). After finishing his Ph.D. on
joined Tetra Engineering in 1991, where he is now the the development of a computational fluid dynamics (CFD)
managing director of the European office. model for gas-particle flow linked to pulverized coal dust
combustion in 1996, Jan Rusås worked with thermo and
His past work includes numerous field inspections and fluid mechanics using CFD in the R&D departments of a
consulting missions dealing with HRSG and boiler com- petrochemical company and boiler companies. In 2006 he
ponent failures and performance problems. He has established his own independent consultant company
co-authored several articles and technical papers on specialized in providing consultant work within CFD for
HRSG O&M issues as well as a guidebook on HRSG customers mainly in the power and energy sectors. Apart
inspection. His current work is focused on the practical from this, he also acts as an external supervisor at Danish
and cost-effective application of thermohydraulic, universities for M.S. and Ph.D. projects and is an external
mechanical, and fluid flow simulations to component censor at Aalborg University, in the Department of Thermo
degradation issues encountered in the field. and Fluid Mechanics and Department of Chemical
Process Engineering, and at Aarhus University, in the
Engineering Department.
Mark Taylor (M.S., Mechanical Engineering, University of
Salford, United Kingdom) first worked as an aerospace
engineer on jet engine thrust reversers. In 1996 he moved
to Germany, where he worked freelance. During this time CONTACT
he developed several software tools for different industries.
James W. Malloy
Author's Copy

Principal Engineer
He joined Tetra Engineering (Europe) in 1997 and since
Tetra Engineering (Europe)
then he has developed the bulk of the in-house software
BP 272 Immeuble Petra "A"
and systems, and provided HRSG and boiler inspections,
06905 Sophia Antipolis
failure and performance analyses, and HRSG and piping
France
life assessments. In addition he gives training courses on
HRSG technology. He is a co-author of one book, has E-mail: [email protected]
presented at various power industry conferences, and has
published several articles and technical papers on HRSG
O&M issues.

Chemists and engineers from fossil and nuclear power plants and from industrial power generation, vendors,
OEMs, consultants, E&As, and others in more than 60 countries on all continents read PowerPlant Chemistry®.

You can reach them by advertising in our journal.

PowerPlant Chemistry® is shipped worldwide.

Some examples:

• Argentina • Australia • Austria • Bahrain • Belgium • Brazil • Bulgaria • Canada • Chile •

• China • Colombia • Croatia • Cyprus • Czech Republic • Denmark • Egypt • Finland •
• France • Germany • Great Britain • Greece • Hong Kong • Hungary • India • Indonesia •
• Ireland • Israel • Italy • Jamaica • Japan • Korea • Kuwait • Malaysia • Malta • Mexico •
• Montenegro • Morocco • New Zealand • Norway • Pakistan • Philippines • Poland •
• Portugal • Romania • Russia • Saudi Arabia • Serbia • Slovakia • Slovenia • South Africa •
• Spain • Sweden • Switzerland • Taiwan • Thailand • The Netherlands • Turkey •
• Ukraine • United Arab Emirates • United States •

Visit us at http://www.ppchem.net or write to [email protected]

102 PowerPlant Chemistry 2017, 19(2)