Journal of Loss Prevention in the Process Industriesrepository.um.edu.my/86008/1/Failure case studies of SA213-T22... · Failure case studies of SA213-T22 steel tubes of boiler through

lable at ScienceDirect

ARTICLE IN PRESS

Journal of Loss Prevention in the Process Industries 22 (2009) 719–726

Contents lists avai

Journal of Loss Prevention in the Process Industries

journal homepage: www.elsevier .com/locate/ j lp

Failure case studies of SA213-T22 steel tubes of boiler throughcomputer simulations

J. Purbolaksono a,*, J. Ahmad b, A. Khinani a, A.A. Ali a, A.Z. Rashid a

a Department of Mechanical Engineering, Universiti Tenaga Nasional, Km 7 Jalan Kajang-Puchong, Kajang 43009, Selangor, Malaysiab Kapar Energy Ventures Sdn Bhd, Jalan Tok Muda, Kapar 42200, Malaysia

a r t i c l e i n f o

Article history:Received 8 April 2009Received in revised form8 June 2009Accepted 9 June 2009

Keywords:SA213-T22 steelCreep ruptureRemnant life assessmentElevated temperatureBoiler tubeFinite element modeling

* Corresponding author. Tel.: þ60 3 89212213; fax:E-mail address: [email protected] (J. Purbolak

0950-4230/$ – see front matter � 2009 Elsevier Ltd.doi:10.1016/j.jlp.2009.06.005

Please cite this article in press as: PurbolaksJournal of Loss Prevention in the Process In

a b s t r a c t

Increased temperature and decreased hardness values of the tube metal and development of oxide scaleon the inner surface of boiler tubes over prolonged period of time are typical problems in power plants.Appropriate life assessments or condition monitoring of boiler tube should be carried out from timeto time. Computer simulations may economically support the post-failure assessment method, i.e. visualinspections, metallurgical examinations and mechanical strength measurements. However, estimationsobtained from the simulations may provide an advanced warning to take preventive actions prior tofailure. In this work two failure cases of the reheater and superheater tubes made of a typical material ofSA213-T22 steel are evaluated. As the oxide scales are increasingly developed on the inner surface, theincreasing of temperature and decreasing of hardness value in tube metal for both cases are determined.The remnant life estimations are then made in the form of creep cumulative damages.

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

The strength of low-alloy steel may change for prolonged periodof service. Estimations of change in temperature, hardness and oxidescale thickness during service may be used to estimate the remnantlife of the component. In particular, estimations of the averagetemperature in tube metal are important in heat recovery steamgenerator (HRSG). It may provide an advanced warning of failure byestimating temperature increase in water-tube boiler.

Port and Herro (1991) reported that almost 90% of failurescaused by long-term overheating occur in superheaters, reheatersand wall tubes. Tubes that are especially subjected to overheatingoften contain significant deposits. The deposits reduce coolant flowand the tubes will experience excessive fire-side heat input. Scalesand other materials on external surfaces will slightly reduce metalstemperatures. The thermal resistance of the tube wall may causea very slight drop in temperature across the wall. When heattransfer through the steam-side surface is considered, the effect ofdeposits is reversed. Steam layers and scales insulate the metalfrom the cooling effects of the steam. It results in reducing of heattransfer into the steam and increasing of metal temperatures.

Starr, Castle, and Walker (2004) also described that oxidation onthe steam side of the tubing can induce premature failures due to

þ60 3 89212116.sono).

All rights reserved.

ono, J., et al., Failure case studdustries (2009), doi:10.1016/j

the insulating effect of the oxide scales raising tube temperatures.In addition, scale spallation could also increase tube temperatures,as spallation debris may collect in the bottom of tubes, blockingsteam flow. Attention is drawn to a potential problem in which thetube temperature and rate of oxidation increase with time as theoxide builds up.

Chaudhuri (2006) described some aspects of metallurgicalassessment of boiler tubes. He discusses some failure problems incarbon steel, reheater and superheater tubes. Ray et al. (2007)reported remaining life assessment and creep analysis of superheaterand reheater tubes made of 2.25Cr–1Mo steel of a thermal powerplant. The tubes had operated for 17 years with average operatingtemperature of 540 �C and having design pressure of 40 MPa. Theremnant life is predicted through dimensional, hardness and tensilemeasurements. Viswanathan, Foulds, and Roberts (1988) performedestimation on the temperature of reheater and superheater tubes infossils boilers. They made correlation between hardness and Larsen–Miller parameter for 1Cr–½Mo, 2¼Cr–1Mo and 9Cr–1Mo steels.

Finite element simulations have been used to investigate failuresin superheater tube (Othman, Purbolaksono, & Ahmad, 2009) andreheater tubes (Purbolaksono, Hong, Nor, Othman, & Ahmad, 2009).Othman et al. (2009) simulated the deformed superheater tubeusing the finite element method. The simulation results have a goodconformity with the finding from the visual site inspection. Pur-bolaksono et al. (2009) reported evaluation on reheater tube failure.The geometry and the scale thickness of the as-received failed tube

ies of SA213-T22 steel tubes of boiler through computer simulations,.jlp.2009.06.005

mailto:[email protected]

www.sciencedirect.com/science/journal/09504230

http://www.elsevier.com/locate/jlp

Nomenclature

Cp specific heat at constant pressure (J kg�1 C�1)D inner diameter of the tube (m)G Gas mass velocityh convection heat transfer coefficient (W m�1 C�1)k thermal conductivity (W m�1 C�1)L length of the tube (m)mo

mass flow rate (kg h�1)N number of tubeNu Nusselt numberP Larsen–Miller parameterPr Prandt numberRe Reynold numberS pitchT temperature (�C)DT Increasing of metal temperature (�C)

t time (h)W gas flow (kg h�1)X scale thickness (m)

Greek symbolsm viscosity (kg m�1 s�1)r density (kg m�3)

Subscriptsaveh average for considering hardnessaves Average for considering scaleg gasi innero outers steamt transverse

J. Purbolaksono et al. / Journal of Loss Prevention in the Process Industries 22 (2009) 719–726720

ARTICLE IN PRESS

were measured and used to generate the finite element models. Thescale thickness inside tube over service time is considered asa linear scale growth. Results obtained from the simulations haveshown agreement with the result from the microscopic examina-tion. Both results showed that the failed reheater tube had over-heating for prolonged period of time.

Nowadays, a large percentage of power and chemical plantsworldwide have been in operation for long durations. There are strongeconomic reasons and technical justifications for continued operationof the power plants. In order to realize the continuing operations inpractice, however, appropriate techniques and methodologies areneeded to evaluate the current condition of the plant components andto estimate their remaining useful lives. The techniques should also bevaluable with respect to relatively younger power plants in thecontext of safety, availability and reliability, operation, maintenanceand inspection practices. An important ingredient in the continuingoperation of power plants is the remaining life assessment technology(Viswanathan, 1989). The remaining life estimations could help insetting up proper inspection schedules and operating procedures inorder to avoid premature retirement of the plants.

An accurate prediction of the temperature distribution in tubemetal of the superheater and reheater will aid the power plantinspectors or engineers in evaluating the remaining life of the boilertubes. A continually increasing scale thickness may occur on the innersurface of superheater and reheater tubes during the service.Presence of oxide scales on the inner surface of boiler tubes maysignificantly contribute to the increased tube metal temperature.Consequently, in the prolonged exposure this phenomenon willworsen situation that leads to potential tube rupture problems. Whena power plant is forced to shut down because of the single componentfailure, the cost of the lost electric-power generation can run toseveral hundred thousand dollars a day. It is essential to perform lifeassessments through the operational condition-based monitoring ofthe power plant regularly than allowing the equipments to fail. In thiswork the life assessments are performed by using finite elementsimulations and utilizing the empirical formulae through iterativeprocedures. The first empirical formula is correlating scale thicknesswith Larsen–Miller parameter (Rehn, Apblett, and Stringer,1981). Thesecond empirical formula is correlating the experimental hardnessdata for 2¼Cr–1Mo steel with Larsen–Miller parameter (Viswana-than et al., 1988). The finite element analysis is carried out usingsoftware package of ANSYS (ANSYS, 2008). Finite element models forheat transfer analyses, that involve forced convections on the innersurface due to the turbulent flow of steam and on the outer surface

Please cite this article in press as: Purbolaksono, J., et al., Failure case studJournal of Loss Prevention in the Process Industries (2009), doi:10.1016/j

due to cross flow of the hot flue gas over bare tubes, are carried out inorder to determine temperature distribution in the tube. An iterativeprocedure is performed to determine the average temperature andhardness of the tube steel over period of time as oxide scale thicknesson the inner surface increases. Two failure cases in reheater (Case 1)and superheater (Case 2) tubes are evaluated through the finiteelement simulations and iterative procedures. The reheater andsuperheater tubes are made of a typical material of SA213-T22 steel.As the oxide scales are increasingly developed on the inner surface,the increasing of temperature and decreasing of hardness intube metal for both cases are determined. The remnant life estima-tions are then made in the form of creep cumulative damages. Besidescomputer simulations utilizing parameters of the operationalcondition may economically support the post-failure assessmentmethod, i.e. visual inspections, metallurgical examinations andmechanical strength measurements, the estimations obtained fromthe simulations may provide an advanced warning to take preventiveactions prior to failure.

2. Heat transfer parameters

Model of the tube section used is 25 mm in length. In modelingof the steady state heat transfer for the problem using ANSYS(ANSYS, 2008), the area of the model is divided into two regions, i.e.scale region and tube region (see Fig. 1). The steam region is takeninto account in determining the convection coefficient of steamfilm for fully developed turbulent flow in a circular tube.

The chemical composition of SA213-T22 steel is listed in Table 1.Steam properties and the thermal conductivities for SA213-T22 andoxide scale (magnetite) are shown in Table 2. The steam-side scaleis usually reported to be duplex (inner spinel (Fe–Cr–Mo)3O4 layerand outer magnetite (Fe3O4) layer) or triplex (inner spinel layer,middle magnetite layer and outer hematite (Fe2O3) layer). In thisstudy material of the scale is treated to be all magnetite.

Phenomenon of heat transfer inside the boiler tube is consideredas forced convection with turbulent flow. Correlation for fully devel-oped turbulent flow in tube is expressed as (Incropera & DeWitt,1996)

Nus ¼ 0:023ðResÞ0:8ðPrsÞ0:4 (1)

where Res is Reynolds number that may be expressed as

Res ¼ 4ms:o

pDims(2)


oxide

scaletube metal

steam

hot gas

hollowradius

remaining tubemetal thickness

initial tube metal thickness

Fig. 1. Model of the superheater and reheater tubes with scale on the inner surface.

Table 2Properties of fluid and solid materials.

Inlet steam properties (Incropera & DeWitt, 1996)Case 1 Case 2Temperature, 576 �C Temperature, 540 �C

Thermal conductivity 0.0636 W/m �C 0.0604 W/m �CSpecific heat 2185 J/kg �C 2161 J/kg �CDynamic viscosity 2.965 e-05 N s/m2 2.834 e-05 N s/m2

Mass flow rate 3600 kg/h 7200 kg/h

Water wall properties (French, 2000)Tube material SA213-T22Thermal conductivity 34.606 W/m �C

Fe3O4 iron oxide (magnetite) (French, 2000)Thermal conductivity 0.592 W/m �C

J. Purbolaksono et al. / Journal of Loss Prevention in the Process Industries 22 (2009) 719–726 721

ARTICLE IN PRESS

in which mso

is mass flow rate of the steam; Di is the inner diameterof the tube; ms is steam viscosity, and Prs is its Prandtl number thatis defined as

Prs ¼msCps

ks(3)

in which Cps and ks are specific heat and thermal conductivity of thesteam, respectively.

Convection coefficient of steam film for fully developed turbu-lent flow in circular tube is expressed as (Incropera & DeWitt,1996):

hs ¼ 0:023ks

DiðResÞ0:8ðPrsÞ0:4 (4)

where ks is steam conductivity.Heat transfer outside the boiler tube is considered as forced

convection due to cross flow of the hot flue gas over bare tubes.A conservative estimated convection coefficient of flue gas hg onouter surface of bare tube in inline and staggered arrangements(see Fig. 2) is given by (Ganapathy, 2003)

hg ¼ 0:3312kg

D0

�Reg�0:6�Prg

�0:33 (5)

where kg is flue gas conductivity; D0 is outer diameter of the tube;Prg is defined as

Prg ¼mgCpg

kg(6)

Table 1Chemical compositions of SA213-T22 steel.

Code C Si Mn P, max S, max Cr Mo

SA213-T22 0.05–0.15 0.5 0.3–0.6 0.025 0.025 1.90–2.60 0.87–1.13


in which Cpg and kg are specific heat and thermal conductivity ofthe flue gas, respectively. The corresponding Reynolds number Reg

may be expressed as

Reg ¼GD0

12mg(7)

where G is gas mass velocity and may be defined as

G ¼ 12Wg

NwLðSt � D0Þ(8)

in which Wg is gas flow; Nw is number of tube wide; St is transversepitch (see Fig. 2), and L is the tube length.

However, the evaluation will be more accurate if the non-luminous coefficients for water vapour and carbon dioxide aretaken into account in determining the gas side convective heattransfer coefficient. Hence, there will be an additional increaseon the overall heat transfer coefficient at flue gas temperature of800–900 �C by 5–6%, assuming that the external or direct radiationfrom the furnace is absent.

3. Estimations of temperatures, scale thickness and hardness

The increasing scale thickness may occur in superheater andreheater tubes during the service. Therefore, estimation must bemade of the average temperature in the oxide scale as a function oftime and scale thickness. In this work in order to perform a scalegrowth prediction, steam-side scale formation for ferritic steel of1–3% chromium correlated with the Larson–Miller parameter asreported by Rehn et al. (1981) is utilized (see Fig. 3). The data ofFig. 3 may be approximated as

log�

X0:0254

�¼ 0:00022P � 7:25 (9)

where X is scale thickness in mm.In the Larson–Miller method, time and temperature are related

by the following equation:

P ¼�

95

T þ 492�ðC þ Log tÞ (10)

where P is the Larson–Miller parameter; T is the absolute temper-ature in degree Celsius; t is the service time in hours; C is a constantequal to 20.

Correlation between hardness (HV) and the Larsen–Millerparameter for 2¼Cr–1Mo steel in the as-quenched condition (seeFig. 4) may be expressed as (Viswanathan et al., 1988)

Hardness ðHVÞ ¼ 961:713� 0:020669P (11)


Fig. 2. Inline and staggered arrangements of the bare tubes.


ARTICLE IN PRESS

The increasing of scale thickness DX in the reheater or super-heater tubes (SA213-T22) may be obtained from Eq. (9) corre-sponding to the given running hours and average temperature inthe oxide scale. The steps of time as shown in Table 3 are used in theiterative procedures. The iterative procedures used to determinescale thickness and hardness of the tube as a function of time andtemperature are as follows:

3.1. Procedure 1

For initial step i ¼ 1, the design temperature for the steam is setto Ts at the inlet of reheater or superheater tube. From thenumerical simulation in the absence of scale (X0), the averagetemperature of Taves_i is the temperature on the inner surface of the

Fig. 3. Steam-side scale formation for ferritic steels of 1–3% chromium correlated withthe Larsen–Miller parameter (Rehn et al., 1981).


tube, the average temperature of Taveh_i is determined from averageof the temperatures on the inner and outer surfaces of the tube. Eqs.(9) and (10) are used to calculate the scale thickness of Xia for theservice hours of 1 h and the scale thickness of Xib for the servicehours of 250 h (see Table 3) using Taves1. Next, by subtracting onefrom the other, the scale increase of DXi (¼Xib � Xia) is determinedand a new scale thickness of Xi (¼X0þDXi) is obtained. Eqs. (10) and(11) with Taveh_i are used to determine the hardness of Hia for theservice hours of 1 h and the hardness of Hib for the service hours of250 h. Hardness Hi is equal to Hia.

Fig. 4. Correlation between hardness and the Larsen–Miller parameter for 1Cr–½Mo,2¼Cr–Mo and 9Cr–1Mo steels (Viswanathan et al., 1988).


Table 3Steps of time used in the iterative procedure.

Step Hours

1 2502 5003 10004 25005 50006 10,0007 20,0008 40,0009 60,00010 80,00011 100,00012 120,000

Table 5Parameters used to determine gas mass velocity G for validation of actual data.

Case 1 Case 2

Gas flow, kg/h 500,000 800,000Number of tube wide 50 50Transverse pitch, m 0.1016 0.1016Tube length, m 8 8


ARTICLE IN PRESS

3.2. Procedure 2

Set i ¼ i þ 1. The average temperatures of Taves_i and Taveh_i arethen determined from the numerical modeling with the new scalethickness on the inner surface. The average temperature of Taves_i

obtained from the average of the temperatures on inner surfacetemperature and at the scale/metal interface is then used tocalculate the incremental thickness of scale from 250 to 500 h usingEqs. (9) and (10). For service hours of 500 h, P is calculated using Eq.(10) and Xib is found from Eq. (9). For service hours of 250 h, P iscalculated using Eq. (10) and Xia is found from Eq. (9). Subtractingone from the other (Xib � Xia) produces the incremental scaleformation from 250 to 500 h, which is added to Xi to give a newscale thickness of Xi. The average temperature of Taveh_i obtainedfrom the average of the temperatures at scale/metal interface andon the outer surface is then used to calculate the hardness of thetube metal for service hours of 250 h and 500 h using Eqs. (10) and(11). For service hours of 500 h, P is calculated using Eq. (10) and Hib

is found from Eq. (11). For service hours of 250 h, P is calculatedusing Eq. (10) and Hia is found from Eq. (11). Hardness for servicehours of 250 h (¼Hi) may be determined from the average of Hia andHib. Repeat Procedure 2 for further predictions up to the requiredhours with the steps of time shown in Table 3.

Since the initial increment of time determines the further esti-mation results, it is proposed to use the steps of time as shown inTable 3. A smaller increment of time might provide a better esti-mation, whereas a bigger increment of time for initial iteration maybe resulting in inaccuracy estimation or less conservative predic-tion. However, for estimations starting from the service hours of20,000 h, an increment of time may be proposed to be taken atevery 20,000 h.

4. Finite element modeling

The geometry of the as-received tubes and heat transferparameters governing the problem are used to generate the finiteelement models. It is important to note that all the geometricalunits used for modeling are in m. Hence, the meshing size control of0.0001 is used to generate the 2D solid triangular elements in orderto allow the model having appropriate size of elements. Theproperties of the elements are then defined as 2D-axisymmetricsolid elements.

Table 4Geometry of the as-received tubes, service time and the inner scale thickness.

Case Innerradius, m

Tubethickness, mm

Servicetime, h

Scalethickness, mm

Date offailure

1 0.0219 3.5 117,522 0.58 10/10/20032 0.0163 4.0 115,494 0.2 04/08/2003


As described in Procedure 1, the finite element model is producedin the absence of oxide scale at the initial step (i ¼ 1). The bulktemperature and convection coefficient hg of the flue gas are appliedon the right edge of the model (Fig. 1). Next, the bulk temperatureand convection coefficient hs of the steam are applied on the leftedge of the model. In the presence of the oxide scale (i>1) the modelwill have two domain areas, i.e. scale and tube metal. In orderto make connectivity of the domain areas at scale/metal interface,a merge-size control of 0.00001 is used. The merge-size controlshould considerably be smaller than the meshing size control. Thebulk temperature and convection coefficient hg of the flue gas areapplied on the right edge of the tube metal region and the bulktemperature and convection coefficient hs of the steam are appliedon the left edge of the oxide scale region. The convection coefficientsof hs and hg are obtained by using Eqs. (4) and (5) respectively.

5. Case studies

The most common approach for calculating the cumulativecreep damage is computing the amount of life expended by usingtime fraction as measures of damage. When the fractional damagesadd up to unity, then the failure is postulated to occur. The mostprominent rule is given as (Robinson, 1938)

X tsi

tri¼ 1 (12)

where tsi is the service time and tri is the time to rupture.The actual data of the available reports on the failed reheater

(Ahmad, 2004) and superheater (Ahmad, 2003) tubes at KaparPower Station Malaysia are used for the case studies. Detaileddescriptions of the failed tubes are as follows:

- Operating steam temperature of the reheater and superheatertubes are 576 �C and 540 �C respectively. The average operatingsteam pressures in Case 1 and Case 2 are 40 bar (4.0 MPa) and100 bar (10 MPa) respectively.

- The average flue gas temperatures were reported at around800 �C for the reheater tube and 850 �C for the superheater tube.

The detailed data taken from the reports are shown in Table 4.Parameters used to determine gas mass velocity G and the esti-mated convection coefficients hs and hg for the internal andexternal surfaces are shown in Tables 5 and 6 respectively. Fig. 5shows the isometric view of a half expansion model from 2Daxisymmetric model and the temperature distribution of Case 2 forthe service hours of 115,494 h. It can be seen that the temperatureon the external surface may be estimated at around 574 �C.

Diagram of Larsen–Miller parameter with stress variation torupture of annealed material 2.25Cr–1Mo steel (ASTM) (Smith,

Table 6The estimated convection coefficients hs and hg for internal and external surfacesrespectively.

Case hs, W/m2 �C hg, W/m2 �C

1 2053.65 126.012 6084.32 129.99


Fig. 5. Temperature (�C) distribution of Case 2 for the service hours of 115,494 h.

Fig. 6. Diagram of Larsen–Miller parameter with stress variation to rupture ofannealed material 2.25Cr–1Mo steel (ASTM) (Smith, 1971) (1 ksi ¼ 6.895 MPa).


ARTICLE IN PRESS

1971) as shown in Fig. 6 is utilized to determine the rupture time. Inorder to obtain conservative estimation, the minimum curvecorrelating stress variation and Larsen–Miller parameter is used.

The Larsen–Miller parameters for Case 1 and Case 2 may beobtained from Fig. 6. It is shown that the parameters for Case 1 andCase 2 are 39,900 and 37,800, respectively. The estimated life forthe tubes in Case 1 and Case 2 are determined in the form of thecumulative creep damage as expressed in Eq. (11). If the total of thefractional damages is greater or equal to 1, the failure is predicted tooccur. It can be seen from Fig. 7, the estimated lives for both casesare around 6–15% higher than that of the actual data. The estimatedresults are reasonably shown to be in agreement with the actualdata. The estimated scale thickness as shown in Table 7 is alsoshown to be in agreement with the actual data (Table 4). Theincreasing of the average temperatures of Case 1 is significantlyhigher than that in Case 2 as shown in Fig. 8. It results in theestimated hardness values for Case 1 showing moderately less thanCase 2. It is believed that the geometry of the tube and mass flowrate of steam influence the phenomena of heat transfer. It isimportant to note that if the applied internal steam pressure ishigher, the mass flow rate of the steam has also to be considerablyhigh. Insufficient mass flow rate of steam may cause a significantincrease of the metal tube temperature and scale growth. Thisfeature may be seen from the conditions of Case 1 and Case 2. It canbe seen from Table 7 and Fig. 8, the increased tube temperatureswould accelerate the developments of the oxide scale on the innersurface. The relation between the hardness values and the averagetemperature of the tube over period of time can be used for futurereferences. The strength of the low-alloy steel will change with theincreasing of temperature over period of time.

The life estimations through computer simulations utilizingparameters of the operational condition may provide an advancedwarning to take preventive actions prior to failure. Poor mass flow rateof steam, higher operational steam temperature and higher convective

Please cite this article in press as: Purbolaksono, J., et al., Failure case studies of SA213-T22 steel tubes of boiler through computer simulations,Journal of Loss Prevention in the Process Industries (2009), doi:10.1016/j.jlp.2009.06.005

0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 20000 40000 60000 80000 100000 120000Service hours

Cu

mu

lative creep

d

am

ag

e

Case 1 - reheater tube

Case 2 - superheater tube

Tubes are predicted to failure

Fig. 7. The estimated cumulative creep damages.

125

145

165

185

205

225

245

265

285

305

325

0 20000 40000 60000 80000 100000 120000Service hours

Ha

rd

ne

ss

, H

V

540

550

560

570

580

590

600

610

620

Te

mp

era

tu

re

, °C

Hardness-Case 2Hardness-Case 1Temperature-Case 2Temperature-Case 1

Fig. 8. The estimated hardness and the average metal tube temperatures over periodof time.


ARTICLE IN PRESS

coefficient and temperature of flue gas would lead to increasing tubemetal temperature and oxide scale thickness. Tube metal temperaturemay increase slowly over many years or rapidly over a few hours.Internal oxide scales usually result in long-term overheating thatgradually increase the tube metal temperature. The increasing tubemetal temperature and oxide scale growth have a reciprocal casual. Asthe tube metal temperature increase, so does the rate of internal scaleformation. As the oxide scale thickness increases, so does the tubemetal temperature. The cycle continues progressively over period oftime. Loss of boiler feedwater or impaired steam flow usually results inrapid overheating and often rapid failure. Localized high temperatureflue gas flow definitely causes localized overheating of the boiler tube.These features potentially lead to rupture of the boiler tubes thatconsequently leads to unscheduled and costly outages.

It is an essential element of condition-based maintenance inwhich the equipment is maintained on the basis of its condition, or itallows actions to be taken to avoid the consequences of failure, beforethe failure occurs. It is typically much more cost-effective thanallowing the components to fail. By selecting a physical measurementwhich indicates that deterioration is occurring, the readings on theparameter need to be taken at regular interval. Since failure occurs toindividual components, the monitoring measurements need to focuson the particular failure modes of the critical component. If a powerplant has been operating with breakdown maintenance or regularplanned maintenance, a change over to condition-based mainte-nance can result in significant improvements in plant availability and

Table 7Estimations for scale thickness for Case 1and Case 2.

Case 1 (reheater tube) Case 2 (superheater tube)

Service hours Scale thickness, mm Service hours Scale thickness, mm

0 0.0000 0 0.0000250 0.0556 250 0.0229500 0.0736 500 0.03011000 0.0965 1000 0.03922500 0.1371 2500 0.05505000 0.1783 5000 0.070710,000 0.2315 10,000 0.090720,000 0.3008 20,000 0.116340,000 0.3916 40,000 0.149160,000 0.4584 60,000 0.172880,000 0.5133 80,000 0.1920100,000 0.5610 100,000 0.2085117,522 0.5982 115,494 0.2200


in reduced cost. Then, life estimations of the boiler tubes through thecomputer simulations may be employed from time to time.

Failure analysis on the water-tube boiler can also be performedby referring to Section 2, Part D of The ASME Boiler and PressureVessel Code (ASME, 1998) for the maximum allowable stress valuesand data from United States Steel Corporation (1972) as shown inTable 8. The maximum allowable stresses and the operational hoopstresses of the tubes (Case 1 and Case 2) over the service hours arethen plotted in Fig. 9. It can be illustrated that the critical conditionsmay be predicted when the stresses in the tube exceeding themaximum allowable stress. It can also be seen from Fig. 9 that thetube can be estimated when the overheating condition occurs.Since the operating hoop stresses are higher than the allowablestresses, tubes in Case 1 and Case 2 are estimated to be critical atthe beginning of service. Even though the operating hoop stressesremain much lower than the ultimate strength shown in Table 8,a careful consideration needs to be taken. Monitoring on thetemperature increase over period of time is particularly importantwhen overheating in the tube is concerned. Thus, combination ofestimating the life expectancy using the cumulative creep damage,considering the maximum allowable stresses and monitoring theincreasing of temperature in tube metal may provide bettermonitoring of the boiler tubes. Appropriate preventive actions maybe taken during the scheduled outage for close inspections andnecessary examinations in order to avoid forced outage due toa single failure of a component.

It is reported that the microscopic examination for Case 1 andCase 2 were also carried out to support the life assessment. Thefinding of Case 1 indicated that the tube metal microstructure hada complete stage of spheroidization where the carbide particles

Table 8The maximum allowable stresses and tensile strengths for seamless tube SA213-T22.

Temperature, �C Max. allow. stress,MPa (ASME, 1998)

Tensile strength, MPa (UnitedStates Steel Corporation, 1972)

537.78 55.16 377.16565.56 39.30 –593.33 26.20 282.69621.11 16.55 –648.89 9.65 –784.00 – 151.69


540

550

560

570

580

590

600

610

620

0 20000 40000 60000 80000 100000 120000Service hours

Averag

e tem

peratu

re, °C

0

10

20

30

40

50

Ho

op

stress, M

Pa

Temperature - Case 2Temperature - Case 1Max. allowable stress (Case 2)Max. allowable stress (Case 1)Operational pressure, 4 MPaOperational pressure, 10 MPa

Fig. 9. Operational hoop and maximum allowable stresses of the failed tubes for Case 1and Case 2 at the corresponding average tube temperatures and service hours.


ARTICLE IN PRESS

have coalesced and dispersed uniformly as shown in Fig. 10. Similarfeature of the microstructure was also reported for Case 2.The change of microstructure confirmed that the failed tube hadoperated at higher temperature or experienced overheating forprolonged period of time. It agrees with the estimation obtainedfrom the simulations.

Abnormal governing parameters clearly cause loss of expected lifedue to microstructural changes and creep failures as result of over-heating. Well-informed data for heat transfer parameters accordingto variations in the operating conditions from the monitoring systemwill improve estimations obtained from the iteration procedure. Inother words, better estimation of the average temperature in the tubemetal could be obtained, provided that all the heat transfer param-eters used are well specified. Early warning of failure may be moni-tored from time to time through simulations utilizing data of theoperational heat transfer parameters taken from the monitoringdevices. Regular measurements need to be taken, e.g. for months oryears, before a critical situation arises. It is expected that moredetailed evaluation can indicate the nature of the problem so thatrectification action can be planned.

Fig. 10. Complete stage of spheroidization with the coalesced carbide particles to formchain of pores (Ahmad, 2004).


6. Conclusions

Finite element simulations incorporated with the iterative proce-dures may be used to estimate the increased temperature anddecreased hardness values of the tube metal and development of oxidescale on the inner surface of boiler tubes over prolonged period of time.Two failure cases in reheater and superheater tubes were evaluated.Estimations obtained from the simulation were shown to be inagreements with the actual data. Combination of estimating the lifeexpectancy using the cumulative creep damage, considering themaximum allowable stresses and monitoring the increasing oftemperature in tube metal may provide an advanced warning to takepreventive actions prior to failure. A reciprocal casual of increasingtube metal temperature and oxide scale growth in a hazardous envi-ronment of boiler operation would lead to tube leakage/rupture, bywhich the safety and cost-effective of the thermal power plant isinfluenced seriously. The ruptured boiler tubes inpower plant generatea serious problem that often leads to unscheduled and costly outages.

Acknowledgements

This work is supported by the Ministry of Science Technologyand Innovation, Malaysia through the research projects of IRPA09-99-03-0033 EA001 and Sciencefund 04-02-03-SF0003. Theauthor wish to thank Universiti Tenaga Nasional and Kapar EnergyVentures Sdn. Bhd for permission of utilizing all the facilities andresources during this study.

References

Ahmad, J. (2003). Technical memorandum of Kapar Power Station Sdn Bhd Malaysia:Report of remnant life study for superheater tubes. TNB Generation Sdn Bhd.

Ahmad, J. (2004). Technical memorandum of Kapar Power Station Sdn Bhd Malaysia:Report of reheater tube failure. TNB Generation Sdn Bhd.

ANSYS. (2008). Multiphysics version 11.0. Southpointe 275 Technology DriveCanonsburg, PA 15317: ANSYS, Inc.

ASME. (1998). ASME international electronic stress table, table 1A: The maximumallowable stress values for ferrous materials, section II, part D of the ASME boilerand pressure vessel code. Copy Right� 1998. ASME International.

Chaudhuri, S. (2006). Some aspects of metallurgical assessment of boiler tubes –basic principles and case studies. Materials Science and Engineering. A – Struc-tural Materials: Properties, Microstructure and Processing, 432, 90–99.

French, D. N. (2000). Metallurgical failures in fossil fired boilers. New York: A Wiley-Interscience Publication, John Wiley and Sons Inc.

Ganapathy, V. (2003). Industrial boilers and heat recovery steam generators: Design,applications, and calculations. New York: Marcel Dekker.

Incropera, F. P., & DeWitt, D. P. (1996). Introduction to heat transfer (3rd ed.). NewYork: John Wiley and Sons Inc.

Othman, H., Purbolaksono, J., & Ahmad, B. (2009). Failure investigation on deformedsuperheater tube. Engineering Failure Analysis, 16(1), 329–339.

Port, R. D., & Herro, H. M. (1991). The NALCO guide to boiler failure analysis. NewYork: Nalco Chemical Company, McGraw-Hill Inc.

Purbolaksono, J., Hong, Y. W., Nor, S. S. M., Othman, H., & Ahmad, B. (2009). Eval-uation on reheater tube failure. Engineering Failure Analysis, 16(1), 533–537.

Ray, A. K., Tiwari, Y. N., Roy, P. K., Chaudhuri, S., Bose, S. C., Ghosh, R. N., et al. (2007).Creep rupture analysis and remaining life assessment of 2.25Cr–1Mo steel tubesfrom a thermal power plant. Materials Science and Engineering. A – StructuralMaterials: Properties, Microstructure and Processing, 454–455(2007), 679–684.

Rehn, I. M., Apblett, W. R., Jr., & Stringer, J. (1981). Controlling steamside oxide exfoliationin utility boiler superheaters and reheaters. Materials Performance, 20(6), 27–31.

Robinson, E. L. (1938). Effect of the temperature variation on the creep strength ofsteels. Transactions of the ASME, 160, 253–259.

Smith, G. V. (1971). Supplemental report on the elevated temperature properties ofchromium molybdenum steels. Metal Properties Council, ASTM data series, DS6S1. American Society for Testing and Materials.

Starr, F., Castle, J., & Walker, R. (2004). Potential problems in the identification of theroot cause of superheater tube failures in 9Cr martensitic alloys. Materials atHigh Temperatures, 21(3), 147–160.

United States Steel Corporation. (1972). Steels for elevated temperature service.United States Steel Corporation.

Viswanathan, R. (1989). Damage mechanics and life assessment of high-temperaturecomponents. Ohio: ASM International.

Viswanathan, R., Foulds, J. R., Roberts, D. A. (1988). Method for estimating thetemperature of reheater and superheater tubes in fossils boilers. In Proceedingof the international conference on life extension and assessment, The Hague.


Documents

Journal of Loss Prevention in the Process Industriesrepository.um.edu.my/86008/1/Failure case studies of SA213-T22... · Failure case studies of SA213-T22 steel tubes of boiler through