Economy & Finance

Experimental and numerical study of crack initiation and propagation under a 3D thermal fatigue loading in a welded structure

Experimental and numerical study of crack initiation and propagation under a 3D thermal fatigue loading in a welded structure
of 14
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  Experimental and numerical study of crack initiation andpropagation under a 3D thermal fatigue loading in a welded structure O. Ancelet  a,* , S. Chapuliot  a , G. Henaff   b a DEN/DM2S/SEMT/LISN, CEA-Saclay, 91191 Gif-sur-Yvette Cedex, France b Laboratoire de Me´ canique et de Physique des Mate´ riaux, UMR CNRS 6617, 1 Avenue Cle´ ment Ader BP40109, 86961 Futuroscope Cedex, France Received 8 December 2006; received in revised form 12 July 2007; accepted 22 August 2007Available online 31 August 2007 Abstract The incident which has occurred in the Civaux power plant has shown the nocivity of thermal loading, the specificity of this type of loading and the difficulty to take it into account at design level. The objective of this study is to examine the initiation and the prop-agation of crack under thermal loading for the structure with or without weld. In this aim the French Atomic Energy Commission(CEA) developed a new experiment named FAT3D. The various experiments carried out on pipe without weld showed the harmful-ness of a thermal loading, which can initiate a network of cracks and propagate one (or some) cracks through the total thickness of the component. The experiments on welded structure showed a faster initiation of cracks and the importance of the orientation of theweld. Indeed it showed the initiation of two cracks both located the feet of the weld joint as observed in the Civaux power plant.These experimental results associated with a mechanical analysis question the usual criteria of damage based on the variations of the equivalent strain.   2007 Elsevier Ltd. All rights reserved. Keywords:  Thermal fatigue; Crack propagation; Finite elements; Low cycle fatigue; Weld 1. Introduction Due to operating conditions of industrial facilities manystructural materials have to withstand a large variety of loading from mechanical and thermal srcins, which inducea thermal fatigue damage. Thermal fatigue was srcinally aconcern in the field of gas turbines and was studied by Cof-fin and Manson, and more recently in certain componentsof the pressurized water reactors (PWR) like mixing tees of the reactor cooling systems.The examination of the cracked zones observed in thePWR of Civaux in France [1] revealed a fast initiation of a network of small cracks (<3 mm in depth) presentingno preferential orientation (Fig. 1a). Besides it recognisedthe role of weld joints in initiation and propagation. Indeedinitiation of a crack was observed at the feet of the weld,while a network of cracks was present far away from theweld (Fig. 1b). Additionally this crack propagated veryrapidly and grew through the thickness of the pipe [1].From a design point of view, these observations raisedseveral questions. The first question is to determine to whatextent the results of uniaxial mechanical testing can be usedto predict initiation lives in structures submitted to thermalfatigue. The propagation of a crack under a pure thermalloading is also an issue that needs to be addressed. Finallythe influence of a weld joint on this propagation has also tobe examined.In this framework, the role of the CEA Laboratory forStructural Integrity and Standards is mainly to simulateand analyze the initiation and the propagation of cracksunder thermal loading in order to get further insights intothe incident of Civaux. With this aim, the laboratory hasundertaken numerical studies on the mixing zones [2] anddeveloped an experimental thermomechanical test devicenamed FAT3D [3]. The main objectives are: 0142-1123/$ - see front matter    2007 Elsevier Ltd. All rights reserved.doi:10.1016/j.ijfatigue.2007.08.010 * Corresponding author. Tel.: +33 169083883; fax: +33 169088784. E-mail address: (O. Ancelet).  Available online at International Journal of Fatigue 30 (2008) 953–966 International JournalofFatigue    To reproduce the rapid initiation of a crack networkunder thermal loading and to compare the number of cycles to initiation to uniaxial experimental data.   To make cracks propagate under pure thermal loading.   To study the influence of a weld on the thermal fatigueresistance of a pipe.The numerical models used to design tests and analysethe results have been presented in the previous paper [3].In the present paper, the principle of this test is firstrecalled. Experimental results obtained on a structure withor without weld are presented and compared and, finite ele-ment thermal and mechanical analyses are presented.Finally, predictions of crack initiation under thermal load-ing using different criteria are presented and compared todata from uniaxial mechanical experiments. 2. Principle of the FAT3D test The experiments are performed on a pipe under thermalloading in order to produce a 3D stress–strain field.This test was designed according to the followingrequirements:   The loading must initiate cracks within a reasonabletime (maximum 3 months including the time forinspection).   It should generate a significant crack propagation stageunder thermal fatigue.A tube is placed in a furnace (Fig. 2a) in order to main-tain constant temperature around the test-tube. Cold wateris periodically injected on the internal skin of the tube. Thecooled zone is parabolic in shape on the internal skin(Fig. 2b). This thermal shock generates different thermalgradients which can be separated in two categories:   Local gradients consisting in a temperature differencebetween the internal skin and the external skin of thetube.   Global thermal gradients due to a temperature differ-ence between one side of the tube and the other. As inFAT3D experiments the cooling zone is parabolic inshape, a thermal gradient along the  zz  axis and anotheralong the  hh  axis is present.The specimen is a 316L austenitic stainless steel tube of 360 mm height, 166 mm in external diameter and 6.7 mmthickness.Various thermal loading parameters can be modified onone test to another (Fig. 2c):   Duration of the total cycle  t c .   Duration of cooling  t f  .   Thickness of the test-tube  e .   Temperature of the furnace  T  c .A cycle can be divided into two parts: a heating stageand a cooling stage. The heating stage is controlled usingthe parameters  t c  and  T  c . The cooling stage is controlledby  t f   (cold temperature  T  f   is water temperature).In order to minimize the period of heating and thusreduce the duration of the cycle, the high temperatureimposed inside the furnace is  T  c  = 650   C (note that thistemperature does not correspond to the maximum temper-ature of the specimen where cracking is observed). Fig. 1. Effect of thermal fatigue on non welded (a) or welded (b) structure.954  O. Ancelet et al. / International Journal of Fatigue 30 (2008) 953–966   To characterise the thermal loading, several thermaltests have been performed on reference test-tubes equippedwith a large number of thermocouples and a numericalmodel based on the results has been developed in orderto predict the temperature map measured during the ther-mal cycle [3]. 3. Thermal fatigue on base metal test-tube Three tests were performed on test-tube without weld joint with different level of loading (Table 1).For the three tests, the initiation of a crack network isnoticed at the top of the cooling zone. Cracks are perpen-dicular to the boundary of the cooling zone: they are lon-gitudinal at the top of this surface and progressivelyfollow a hoop direction towards the bottom of the zone.As the temperature amplitude is higher at the top of thecooling zone, the longest cracks are localized in the regionwhere thermomechanical loading is the most severe.The tests FAT3D Nos. 4 and 6 were periodicallystopped to examine the propagation of the crack network.It can be noted that cracks rapidly propagate on the sur-face and in depth (Fig. 3), and that the density of crackin the network increases when the load decreases. Indeed,for the first test, a through thickness crack is observed [3].These observations lead to the following conclusions:   The initiation of a network of cracks occurs very early inthe FAT3D experiment.   The first cracks appear at the top of the cooling zone,and are oriented perpendicularly to the border of thethermal spot.   Significant propagation of cracks in surface and in depthis observed; in some case a through the thickness crackis observed. water evacuation furnacetest-tubeparabolic shape cooling zone a zz b cold water injectiont c water injection pipe timet f  c Fig. 2. Principle of the FAT3D experiment.Table 1Synthesis of the FAT3D experimentsFAT3D No. 4 FAT3DNo. 5FAT3DNo. 6Duration of the cycle  t c  (s) 190 130 91Duration of the cooling  t f   (s) 15 15 11 D T   in external skin (  C) 360 290 220Number of initiation cycles 3500–12,000 16,500– 30,09314,000– 23,000Surface length of the first cracksobserved (mm)30 20 6Total number of cycles 17,532 30,093 48,147Final surface length at the end of the experiment (mm)50 15 20 and25Crack depth (mm) Throughthickness crack6.7>4 >3.2Mean length between twoadjacent cracks (mm)3.2 1.7 1.5 O. Ancelet et al. / International Journal of Fatigue 30 (2008) 953–966   955    A higher density of cracks in the network is obtained athigh thermal load level. 4. Thermal fatigue on welded test-tube Two additional tests were performed on pipes contain-ing a shaved weld joint using the same loading condi-tions as during the FAT3D No. 6 test. Fig. 4 describesthe different geometry used for test FAT3D Nos. 7 and8. For these two tests, different orientations of the weld joint were considered: the first orientation of the jointis circumferential (Fig. 4a) while the second one is longi-tudinal (Fig. 4b). Fig. 4c gives the details of the welding procedure.Table 2 sums up the main results of three thermal fati-gue tests without weld (FAT3D No. 6), with circumferen-tial weld (FAT3D No. 7), with longitudinal weld(FAT3D No. 8). It can be seen that the initiation is reducedroughly by a factor of 2 in welded test-tube. The differenttests were continued until 50,000 cycles in order to examinethe development of the crack network. At the end of thetests, the maximal crack surface length is nearly the samefor the three tests but the density of cracks in the networkis different.After the tests, the test-tubes were cut to study crackpropagation in the thickness. In this aim, different loca-tions for cutting were selected in order to determine thecrack depth along the cutting line.In the case of welded tubes, a short time after the begin-ning of the test (14,000 cycles), initiation of small cracks isdetected (several millimetres length). However, the initia-tion location differs depending on the orientation of theweld joint:   For the FAT3D No. 7 test (circumferential weld), thefirst cracks appeared above the weld joint, at the topof the cooling zone (Fig. 5a). The cracks are longitudinaland perpendicular to the weld joint.   For the FAT3D No. 8 test (longitudinal weld) the firstcrack appeared at the interface of the weld joint at thetop of the cooling zone (Fig. 5b).After initiation, the continuation of the FAT3D No. 7test resulted into the initiation of many cracks at the topof the cooling zone. During this time, the first cracks prop-agate until a length of 25 mm in 50,000 cycles (Fig. 6a). Thecrack orientation is longitudinal and the presence of theweld joint does not change this orientation. However, adecrease of the propagation rate is noticed when the crackgrows through the weld joint. The crack depth is impor-tant, (Fig. 6b) up to 4.5 mm in depth.The crack propagation in the FAT3D No. 8 test isdifferent: Boundary of the cooling zone a zz Boundary of the cooling zone b 5 mm5 mm r   e  t    r   a n s  c r  i        p t    i      on Fig. 3. Observation of the internal skin of the test-tube at the end of the experiment: (a) experiment FAT3D No. 5 after 30,093 cycles; (b) experimentFAT3D No. 6 after 48,147 cycles.956  O. Ancelet et al. / International Journal of Fatigue 30 (2008) 953–966     The first crack, initiated on the interface of the weld,rapidly propagates (Fig. 7a). It is worth noticing thatin comparison with the FAT3D Nos. 6 and 7 tests,the crack density near the weld is low. The examina-tion of the cracks through the thickness (Fig. 7b)shows a propagation of 3 mm. However, as mentionedearlier this crack depth depends on the cutting locationwhich is particularly difficult to select for this test.Therefore, this value of depth might not be consideredas a maximum.   Far from the weld, the initiation of a network of crackswith a high density similar to the FAT3D No. 6 test isobserved.However, as shown in Fig. 7a, the weld is not at themiddle of the cooling zone, and therefore not in themaximal loading zone. Therefore, a larger propagationof the cracks initiated on the interface of the weld anda smaller propagation of the network of cracks mightbe observed if the weld is located at the maximum ther-mal load area.The analysis of these observations lead to the followingconclusions:   The presence of a weld is not harmful since the numberof cycles to initiation is reduced only by a factor 2;   the orientation of the weld strongly influence both initi-ation and propagation:If the weld is parallel to the prin-cipal stress direction, cracks initiate above the weld andpropagate perpendicularly to the weld.If the weld is per-pendicular to the principal stress direction, a crack initi-ates at first on the interface of the weld. Then, a fastpropagation of this crack is observed. Meanwhile acrack network initiates far from the weld. As the weldwere not located at the maximal loading zone, the prop-agation of the cracks initiated on the interface of theweld is very important in comparison of the othercracks. 10mm75 ° 75 ° a b 75˚231 pass Tungsten InertGazwelding11mm10mm c 4passes Manual Metal Arc weldingMachining after welding46.7mm1 Fig. 4. Geometry of the welded test-tube for the test FAT3D No. 7 (a) and for the test FAT3D No. 8 (b); geometry of the welding joint (c).Table 2Synthesis of the FAT3D experiments on welded pipeFAT3DNo. 6FAT3D No. 7 FAT3D No. 8Duration of the cycle  t c (s)91 91 91Duration of the cooling t f   (s)11 11 11 D T   in external skin (  C) 220 220 220Weld orientation NoweldCircumferential LongitudinalNumber of cycles at theinitiation14,000– 23,0007462–14,296 7240–14,593Surface length on of thefirst cracks observed(mm)6 6 2Total number of cycles 48,147 48,182 50,999Final surface length atthe end of theexperiment (mm)20–25 20–25 20–25Crack depth (mm) >3.2 >4.5 >2.7Mean length betweentwo adjacent cracks(mm)1.5 2.5 3 near from theweld 1.1 far fromthe weld O. Ancelet et al. / International Journal of Fatigue 30 (2008) 953–966   957
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!