Sunteți pe pagina 1din 9

Engineering FailureAnalysis, Vol 2, No. 4 pp. 247-255.

1995
~ Pergamon Copyright © i995 Elsevier Science Ltd
Printed in Great Britain. All rights reserved
1350-6307/95 $9.50 + 0.00

1350 -6307(95)00024-0

THERMAL FATIGUE FAILURE AND DESIGN MODIFICATION


OF A STEAM EXHAUST SILENCER

SOON-BOK L E E
Department of Mechanical Engineering, Korea Advanced Institute of Science and Technology,
Science Town, Taejon, 305-701, Korea

(Received 2 August 1995)

Abstract--A steam exhaust silencer in a power-plant was subjected to thermal cycling as the
system was started up, operated and shut down. The acoustic splitters in the silencer were
blown out to about 80 m after only 6 months of operation. The cause of the significant failure
was low cycle thermal fatigue and fracture of the connection between the silencer case and the
inside splitter. A failure analysis was performed and a design modification was made to avoid
further failures.

1. I N T R O D U C T I O N
1.1. Background
The piping at a new power-plant was flushed with superheated steam. The exhaust
steam produced a servere acoustic noise. To reduce the noise, heavy-duty silencers
were used. A steam exhaust silencer used in a thermal power-plant failed prematurely
after only 6 months of service. The silencer was operated for several start-up and
shut-down cycles per day. The operating time was about 1 h per cycle. Superheated
steam at 500 ton h -1 was exhausted through the silencer. The outlet temperature of
the superheated steam was 450°C and the pressure was about 0.84MPa. The
maximum allowable noise level was 95 dB at 10 m from the silencer. Figure 1 shows

Fig. 1. Overall view of two silencers installed on the wall of the building.
248 SOON-BOK LEE

the overall view of two silencers installed on the side of the building. The silencer
consisted of a case and three acoustic splitters. The acoustic splitters were concentric
drums and were supported by steel bars welded to the splitter drums and the silencer
case. The drum-shaped case was insulated inside so that the outer surface tempera-
ture of the case was comparable to the ambient temperature. Figure 2 shows a
close-up view of the exhaust side of the silencer. Three concentric acoustic splitters
and the outer case were welded to eight supporting bars. The 100 × 100 m m H - b e a m s
were welded after the accident.

1.2. Observation of the failure


Parts of the silencer were blown out by about 80 m during normal operation. The
silencer was located about 10 m up the window side of the power-plant building.
Many cracks were observed at the front (inlet) side of the silencer as well as in the
outer drum. The cracks were highlighted by fluorescent dye penetrants as shown in
Fig. 3. Figure 4 shows the welding failure on the supporting bar of the splitters and
Fig. 5 shows the thermal deflection of a supporting bar. A failure analysis was
performed and a design modification was performed to avoid further similar failures
of the silencer.

2. M A T E R I A L D A T A

The material of the structural parts of the silencer was SM45C steel (approximately
equivalent to AISI1040 steel). Monotonic and cyclic strain properties of the steel were
obtained from [1] as follows: coefficient of thermal expansion c~ = 10.8 x 10 -6°C -1,
Young's modulus E = 207 GPa, Poisson's ratio v = 0.28, ultimate tensile strength
Su = 621 MPa, yield strength Sy = 345 MPa, cyclic yield strength S'y = 386 MPa, cyclic
hardening exponent n' = 0 . 1 8 , true fracture strain e l = 0 . 9 3 , true fracture stress
of = 1050MPa, fatigue limit Sf = 173 MPa, fatigue strength coefficient or'f= 1540

Fig. 2. Close-up view of the exhaust side of the silencer. Three concentric acoustic splitters
and the outer case are welded to eight supporting bars. The 100 x 100 mm H-beams were
welded after the accident.
Thermal fatigue of a silencer 249

Fig. 3. Front side of the silencer. Cracks are highlighted by fluorescent dye penetrant.

Fig. 4. Weld failure on the supporting bar of the splitters. The index finger points to the
cracks at the welds.

MPa, fatigue ductility coefficient e'~ = 0.61, fatigue strength exponent b = - 0 . 1 4 , and
fatigue ductility exponent c -- - 0 . 5 7 .

3. F A I L U R E A N A L Y S I S
3.1. Applied load on the silencer
The silencer was subjected to fluctuating mechanical stress by flow-induced vibra-
tion thermal cycles and thermal gradients. Vibration loads in the silencer can be
250 SOON-BOK LEE

Fig. 5. Thermal deflection of the supporting bar near a reinforced area.

expected as a result of rapidly fluctuating steam pressures, and turbulent steam flow
at a high flow rate. The silencer was subjected to pressure and thermal cycling during
start-up and shut-down. Transient thermal gradients of a lesser degree occurred
during operation between the silencer case and the acoustic splitter supporting bars.
The outer case of the silencer (2750 mm in diameter) was exposed to air and the
acoustic splitters were subjected to the superheated steam. The outer case and
splitters has a different temperature due to the internal insulation. Drum-shaped
splitters were supported with steel bars attached to the splitters with welds. Eight
supporting bars were welded radially to the concentric splitters. Figure 2 shows the
steam exhaust side of the silencer with supporting bars covered by 100 × 100 mm
H-beams welded after the accident to prevent further blow-out.
The amount of expansion in the supporting steel bar is greater than the radial
expansion of the splitters. Therefore, the supporting bars were under high thermal
stress. This cyclic thermal stress produced low cycle fatigue damage on the supporting
bars between the splitter and the case of the silencer. Once a crack formed in the
supporting bars and grew to a critical length, the splitter became detached from the
silencer and the superheated steam blew the splitter out, causing an accident.

3.2. T h e r m a l stress analysis


3.2.1. T h e r m o e l a s t i c i t y . The thermal strains of unconstrained members of an
isotropic material due to a temperature change A T are [2]
~x, = % = c~ = ,xAT. (1)
The thermal stress for isotropic materials can be written as [2]
vE E E
o~..~ = e + e~. crAT (2)
(1 + v)(1 - 2v) (1 + v) (1 - 2v)
where e = ex.~ + e~.y + ez~. When the x-direction is constrained and the other direc-
tions are free, the elastic thermal stress range with the elastic properties of SM45C
steel becomes
A o = 2.858A T (MPa). (3)
Thermal fatigue of a silencer 251
Figure 6 shows a comparison of the elastically calculated thermal stress range with the
limits imposed by codes.

3.2.2. Thermal expansion calculation. In steady-state operation, the temperature of


the splitter inside the silencer was the same as the superheated steam, i.e. 450 °C.
However, the temperature of the outer case of the silencer might be much lower than
that of the splitter because it is exposed to atmospheric cooling. The temperature
range between the operation period and the rest period for the splitter is bigger than
that of the silencer case. Therefore, the amounts of expansion of the splitter and the
outer case were different from each other. The longitudinal expansion of the steel
bars is AI = l . e = l . a~- A T = 2750(10.8 x 10 -6) × 450 = 13.4 mm. The expansion of
the silencer case is AI = l . e = l . o:. A T = 2750(10.8 x 10 -6) x 100 = 3 mm by setting
A T = 100 °C for the outer shell.

3.3. Low cycle thermal fatigue


Thermal stresses are difficult to avoid and, if severe, can lead to fracture in
relatively few cycles. Cyclic strain is more important than cyclic stress in thermal
fatigue, because non-uniform or differential thermal expansion is the usual source of
cyclic loading [3].
The low cycle fatigue life can be obtained from the total strain range vs life as
follows [1]:
t
I C
Ae = af(Nf)b + e~(Nf) . (4)
E
When the cyclic material data are unavailable, Manson's universal slopes [4], quoted
by Fuchs and Stephens [1], can be used for the strain-life relation to give

As = 3.5Su(Nf) -0"12 + E0"6(Nf)-0'6. (5)


E
Setting A T = 450 °C, the strain range becomes Ae = 0.00486 from Eqn (1). With
the room temperature cyclic properties of SM45C, Eqn (4) provides a fatigue life of
12,160 cycles. However, extended periods of high-temperature operation ensued
between thermal cycles. More rigorous high-temperature life prediction including

1600-

1400

1200
Cl
1000
111
o
z 800
<
f~
(¢)
600 ASME Section V / v . 2
iii
E
i-
u) 400- . . . . . . . . . . . . . . . . . . . .

200-

0 I I I I I I
0 100 200 300 400 500 600
TEMPERATURE DIFFERENCE AT (°C)
Fig. 6. Comparison of elastically calculated thermal stress range with limits imposed by codes.
252 SOON-BOK LEE
creep damage can be evaluated with the damage evaluation procedure outlined in the
physically based phenomenological approach [5]. However, this phenomenological
approach requires extensive material properties of the high-temperature structure,
and also exact life prediction is not necessary for this failure analysis. Therefore, a
simple approach, such as Manson's 10% life rule for high-temperature life prediction,
can be used as follows:

Ae = 3.5 10Nf) ~~: + ~f t~vlvf) . (6)


E
Equation (6) produces a fatigue life for the silencer of 3300 cycles with
A T = 450 °C. The 10% life rule with Eqn (4) produces 1216 cycles, which is very
close to the actual silencer life of 1260 cycles (7 times × 30 days × 6 months).

3.4. Cause of the failure


Since the extension of the supporting bars was larger than that of the outer shell,
the supporting bars were constrained by the outer shell and subjected to thermal
stress. The thermal stress was concentrated at the welded area where the supporting
bar and the shell met. The strength of the weldments was reduced due to the
heat-affected zone during the welding process. The elastically calculated stress ranges
were far above the yield and the limits imposed by codes, demonstrating the
detrimental effect of the welded bars. The thermal stress resulted in plastic
deformation and/or creep deformation, and also produced cracks at the welds as
shown in Fig. 4. Serious bends were observed on the bars as shown in Fig. 5. The
large thermal strains were applied to the supporting bars repeatedly and produced
thermal-fatigue crack initiation and propagation.
Due to the cracks and bends, the supporting bars could not resist the load from the
steam jet. When the cracks grew to a critical length, fast fracture occurred and the
splitter was blown out to about 80 m from the silencer shell.

4. D E S I G N M O D I F I C A T I O N
4.1. Temporary remedv
The silencer was reassembled temporarily with the same design but the outlet of
the silencer was covered with a welded fence as shown in Fig. 2 so that the splitters
could not be blown out by the steam exhaust. However, cracks occurred in the head
portion of the silencer case as shown in Fig. 3. Therefore it was advisable to redesign
the silencer to accept the thermal expansion produced by elevated temperature
operation.

4.2. Redesign with vertical splitters


The drum-type acoustic splitters were changed to series of plate-type acoustic
splitters. Several vertical acoustic splitters were employed which could stand inside
the silencer. Vertical acoustic splitters were designed to slide on the silencer case and
the four corners of the splitter were held by sliding bolts so that the thermal stress
could be reduced by free expansion.
The vertical extension of the splitter for the largest splitter panel will be
Al = l • e = l • ol- A T = 2750(10.8 × 10-6) × 450 = 13.4 ram. The silencer outer shell
will extend about 3.0 mm. Thus, 10.4 m m of clearance should allow freedom from
thermal constraint. In practice, the splitter was designed to expand without constraint
along a 12 m m vertical clearance.
Figure 7 shows a schematic drawing of the modified design of the silencer. Figure 8
shows the installation of the vertical splitters in the silencer case. The overall set-up
INSPECTION HOLE
SHELL ROCK WOOL
~,~H,,~,, ~"~. /, 1 /,~ "K"
~ ,.---~-,
"L" .ADDER

i FLAT BAR
I II~COUSTIC ii
...
!II ACOUSTIC
.. L.,I lI
I DIFFUSER I | ~ SPLITTER ii ill SPLITTER I ol
LL_L_I__!i_ J . . . .4-SADDLE
. L .... k__~QI EL+22612

/ ol

--'L"~E \ I l;I ~--~


-]

L ~850 h 1935 ] ,8,5 .L_ ,,,o _~_] ! ?00 ! 600 i 600 ?oo I
I 2Boo
J
I 8000 I

M16 BOLl

J-p

DETAIL "K" DETAIL "L"


t,J
Fig. 7. Modified design of the silencer. Acoustic splitters were installed vertically inside the silencer shell. Dimensions in mm.
254 SOON-BOK LEE

Fig. 8. Installation of the vertical splitters in the silencer case.

Fig. 9. Overall set-up of the redesigned silencer at the power plant site.

o f the n e w silencer is s h o w n in Fig. 9. F i g u r e 10 shows the s t e a m b l o w i n g t h r o u g h the


two n e w silencers.

5. S U M M A R Y

(1) A s t e a m e x h a u s t silencer in a p o w e r - p l a n t significantly failed. A s p l i t t e r in the


silencer was b l o w n o u t by a b o u t 80 m after o n l y 6 m o n t h s o f o p e r a t i o n . A failure
analysis was p e r f o r m e d a n d a design m o d i f i c a t i o n was m a d e to the silencer.
Thermal fatigue of a silencer 255

Fig. 10. Superheated steam blowing through two redesigned silencers.

(2) High temperatures during operation produced thermal stress which was applied
repeatedly to the splitter, and induced the failure of the silencer by thermal fatigue at
the connection area between the silencer case and the inside splitter.
(3) The concentric drum type acoustic splitters were changed to a series of vertical
plate type acoustic splitters. New acoustic splitters were designed to slip around the
bolts so that the thermal stress could be reduced due to free expansion.

REFERENCES

1. H. O. Fuchs and R. I. Stephens, Metal Fatigue in Engineering, John Wiley, New York (1980).
2. A. P. Boresi, R. J. Schmidt and O. M. Sidebottom, Advanced Mechanics of Materials (5th edn), John
Wiley, New York (1993).
3. D. V. French, Metals Handbook (9th edn), Vol. 11, Failure Analysis and Prevention, American Society
for Metals (1986).
4. S. S. Manson, Exp. Mech. 5, 193 (1965)
5. S. B. Lee and A. K. Miller, Trans. A S M E J. Engng Mater. Technol. 117(3) (1995).