Master of Science Thesis in Lightweight Structures Dept of Aeronautical and ehicle Engineering Stoc!holm" Sweden #$%#
EFFECT OF MATERIAL GRADE ON FATIGUE STRENGTH AND RESIDUAL STRESSES IN HIGH STRENGTH STEEL WELDS Wasim Asgher A Master Thesis Report written in collaboration with Department of Aeronautical and Vehicle Engineering Royal Institute of Technology Stocholm! Sweden And Vol"o #onstruction E$uipment %ra&s! Sweden
April! '()'
Preface This thesis wor was performed at Vol"o #E! %ra&s and di"ision of lightweight structures! "ehicle and aeronautical engineering department! *T+! Stocholm, I am greatly thanful to my super"isors Dr, -uheir %arsoum .*T+/ and %ertil 0onsson .V#E/ for their continuous support and guidance during the thesis wor, Especially! I am highly indebted to Dr, -uheir %arsoum for teaching me a lot and spending his precious time, I would also lie to than Ru 1eng and Annethe %illenius from 2in3ping 4ni"ersity for pro"iding assistance in measurement of residual stresses, 2astly! I want to than my parents who ha"e been a great source of moral support and encouragement to me during this wor,
Stockholm, April 2012 Wasim Asgher,
Abstract. This thesis wor is concerned with effect of material grade on fatigue strength of welded 5oints, 6atigue strength e"aluation of welded 5oints in as welded and post weld treated condition was carried out with effecti"e notch method, Results of pea stress method ha"e also been compared with those of effecti"e notch method for as welded 5oints, In addition! using the results of effecti"e notch method! the effect of important weld and global geometry factors on notch stress concentration factor has been studied with '7le"el design of e8periment and a mathematical relation among stress concentration factor and the geometric factors has been proposed, 9"erall! thicness of the base plate and toe radius is found to be the most important factors determining fatigue strength of the 5oint, Welding induced residual stresses ha"e also been predicted using 'D and :D 6EM analysis to see their effect on fatigue strength of the 5oints, Also! trans"ersal residual stresses were measured using ;7ray diffraction method to assess the accuracy of predicted results, %ased on simulation results! effect of geometric factors on ma8imum "alue of trans"ersal residual stress was also in"estigated, Keywr!s. 6atigue Strength! 6illet Welded 0oints! Effecti"e <otch Method! Residual Stress Simulation,
Tab"e f C#te#ts
1reface ' Abstract, : *eywords, : Table of #ontents = Table of 6igures > ) I<TR9D4#TI9< = ),) %acground = ),' Wor Approach = ' 2ITERAT4RE REVIEW ? ',) 6atigue Assessment of Welded 0oints ? Effecti"e <otch Stress Method ? 1ea Stress Method > ',' Residual Stresses > : MET+9DS @ :,) Model Description @ :,' 6atigue 2ife #alculations @ Effecti"e <otch Method @ 1ea Stress Method A :,: Residual Stresses B 1rediction and Measurement A Test Specimens A 1rediction of Residual Stresses C Measurement of Trans"ersal Residual Stresses )( = RES42TS A<D DIS#4SSI9< )) =,) 6atigue Strength E"aluation )) #omparison between Effecti"e <otch Method and 1ea Stress Method ): =,' Influence of Deometric 6actors on 6atigue Strength ): As Welded 0oints )= 1ost Weld Treated 0oints )= =,: Welding Residual Stresses )? Simulation of Welding Residual Stress )? Measurement of Trans"ersal Residual Stress )@ ? #9<#24SI9<S ') > RE6ERE<#ES '' @ Appendi8 ': @,) Temperature dependent material properties ': @((M1a SteelE ': C>(M1a SteelE '>
Tab"e f F$%&res
Figure 1 *itigawa diagram B An illustration of the effect of material grade on fatigue strength, ......................................................................................................... 4 Figure 2 - Schematic illustration of the relation between accuracy! comple8ity and wor effort re$uired for fatigue analysis of welded structures FrefG .............................. 5 Figure 3 An illustration of creating effecti"e notches .......................................... 6 Figure 4 Model of the fillet 5oint ........................................................................ 7 Figure 5 Modeling with effecti"e notch method ................................................. 7 Figure 6 Modelling with pea stress method ...................................................... 8 Figure 7 6EM model for residual stress simulations, a/ 'D Model with boundary conditions, b/ boundary conditions for :D model for first weld, c/ boundary condition for weld on other side, ............................................................................................... 9 Figure 8 Mo"ing heat source for :D analysis ...................................................... 9 Figure 9 Temperature dependent material properties, a/ conducti"ity and specific heat b/ yield stress! tangent modulus and thermal e8pansion coefficient FrefG ............ 10 Figure 10 Distribution of first principal stress in the 5oint ................................ 11 Figure 11 6irst principal stress along longitudinal direction .............................. 11 Figure 12 Illustration of different load lengths ................................................. 12 Figure 13 6atigue li"es of the 5oints mapped on pea stress scatter bands ........ 13 Figure 14 Influence of main factors on stress concentration factor .................. 14 Figure 15 Influence of main factors on stress concentration factor for post weld treated 5oints, ........................................................................................................... 15 Figure 16 Temperature distribution in molten Hone, a/ +A- Hone b/ Temperature history ................................................................................................. 15 Figure 17 - Stresses in 6mm 5oints, S87Trans"ersal! Sy7A8ial! SH72ongitudinal, ..... 16 Figure 18 Measurement of trans"ersal residual stresses, a/ 6mm 5oints b/ 10mm 5oints ....................................................................................................................... 17 Figure 19 Trans"ersal residual stress ................................................................ 18 Figure 20 - #omparison of measurements with 6EM results ............................... 19 Figure 21 6actors affecting pea "alue of trans"ersal residual stress, a/ Main factors b/ Interaction of parameters ......................................................................... 20
- 4 -
' INTRODUCTION '.' (ac)%r&#! Vol"o construction e$uipment is one of the largest manufacturers of construction e$uipment in the world, In Sweden! articulated haulers and wheel loaders are manufactured where welding is used primarily for 5oining steel structures, According to many regulations of fatigue in welded structures! high strength steels offer no ad"antage for fatigue loaded welded structures, This is because of crac lie imperfections formed by welding processes which go"ern the fatigue life of the 5oint to a greater e8tent, It is howe"er recently established by the researchers that a comparati"ely higher fatigue life can be obtained by using high strength steels for fatigue loaded welded structures pro"ided that weld $uality is controlled, The same is suggested by a typical *itigawa diagram! see figure 1, Therefore it is re$uired to formulate an acceptability criterion for defect siHe so that high strength steels can be used for achie"ing better fatigue life in as welded conditions,
Figure 1 *itigawa diagram B An illustration of the effect of material grade on fatigue strength, '.* Wr) A++rac, The wor is di"ided in two sectionsE 1. Fatigue assessment Life prediction ith F!" and parametric anal#sis. 6atigue life assessment of fillet welded 5oints was carried out with effecti"e notch stress method and pea stress method, A design of e8periment was performed to identify the factors ha"ing ma5or influence on fatigue strength, #onse$uently a mathematical model of stress concentration factor was proposed through regression analysis, 2. $esidual Stresses "easurement and F!" simulations Trans"ersal residual stresses induced by welding process were measured using ;7ray diffraction method, 6inite element modeling and se$uentially coupled thermo7mechanical analysis of the welded 5oints was performed to predict the residual stress state, 6urthermore a comparison of 6EM results was made with e8perimental measurements and effect of different parameters on residual stresses was studied through finite element simulations,
7 ? 7
* LITERATURE RE-IEW *.' Fat$%&e Assess.e#t f We"!e! /$#ts Apart from e8perimental in"estigation! a number of methods ha"e been de"eloped for fatigue e"aluation of welded 5oints, %ased on their approach! they are categoriHed as global method or local method, Dlobal methods lie nominal stress method and structural hot spot method do not tae into account the stress concentration effects due to notches produced by welds and therefore are not good for critical e"aluations, 9n the other hand! local methods e.g. effecti"e notch stress method and linear elastic fracture mechanics .2E6M/ address these issues "ery well and gi"e comparati"ely accurate results, +owe"er woring effort for these methods is relati"ely greater and is usually a trade7off between accuracy of the results and comple8ity of the structure! see figure 2, A detailed description of these can be found in F)G,
Figure 2 - Schematic illustration of the relation between accuracy! comple8ity and wor effort re$uired for fatigue analysis of welded structures F=G Iuite recently another method! called pea stress method! has been proposed for fatigue assessment of welded 5oints, This is an efficient method in terms of calculation time howe"er it is restricted in applications, In this thesis wor fatigue assessment of fillet welded 5oints was performed with effecti"e notch stress method and pea stress method, Effect$0e Ntc, Stress Met,! Effecti"e notch stress method was introduced by International Institute of Welding .IIW/ in )CC>, According to +obbacher F'G! the stress at the root of the notch! obtained by assuming linear elastic material beha"ior represents effecti"e notch stress, The statistical "ariation of weld shape and non7 linear material beha"ior at the notch is accounted for by replacing the actual notch with an effecti"e notch of 1mm radius as shown in figure %,
7 > 7
Figure 3 An illustration of creating effecti"e notches Pea) Stress Met,! 1ea stress method is based on notch stress intensity factor approach and can be utiliHed for fatigue assessment of welded 5oints by using the scatter bands proposed by Meneghetti F? J >G, The scatter bands ha"e been "alidated on a number of different types of welded 5oints for steel and aluminium structures and are termed as uni$ue based on their consistency to gi"e sufficiently accurate results for different geometries and loading conditions, This method is much more efficient than other contemporary methods of fatigue assessment because it re$uires significantly less computational resources, +owe"er this efficiency has come with certain limitations on its applicability, It is currently "alid for sharp notches restricted to an opening angle of 1%& degrees and therefore canKt be used for fatigue assessment of post weld treated 5oints, 6urthermore it has only been "alidated for fillet 5oints, *.* Res$!&a" Stresses Due to rapid heating and cooling of welds! residual stresses are produced in the 5oints, Depending upon weld shape and boundary conditions these stresses may be tensile or compressi"e in the weld, While compressi"e residual stresses may be good for fatigue loaded structures! tensile stresses are detrimental and may significantly reduce the endurance limit of the structure, It has been demonstrated in se"eral te8ts that residual stresses due to welding can be of the order of material yield strength,
7 @ 7
1 METHODS 1.' M!e" Descr$+t$# Model consists of a fillet welded T75oint! see figure =, The trans"ersal dimension of the 5oint is taen long so that important residual stresses are present in the 5oint, It was howe"er! limited by the en"elope of the fatigue testing machine with a ma8imum load capacity of 2&0'( in dynamic loading,
Figure 4 Model of the fillet 5oint Two le"els were chosen for base plate thicness! t 1 and stiffener plate thicness! t 2 with magnitudes 6 mm and 10 mm, 2ength of both plates .out of plane dimension/ was selected to be 1%0 mm, 9ther parameters in the figure ) and are defined belowE Width of base plate L 1 L %00 mm Width of stiffener plate L 2 L &0 mm Material properties of steel were used for fatigue life calculations i.e. MoungKs modulus ! * 20& +,a and 1oisson ratio - L (,:, 6or residual stress analysis! thermal and temperature dependent mechanical properties were also used for three different material grades corresponding to yield strength %&0! .00 and /60 ",a, These properties are outlined in section %.% and appendi8 F)G, 1.* Fat$%&e L$fe Ca"c&"at$#s A 'D parametric analysis was performed with different weld geometry parameters for fatigue life calculation of the selected model using effecti"e notch stress method and pea stress method, These parameters were selected for as welded and post weld treated 5oints and are outlined in section ).2. Effect$0e Ntc, Met,! Due to symmetry! half of the geometry was modeled with plain strain condition, 6inite element mesh was generated with $uadratic $uadrilateral elements and sufficiently refined mesh at notches! see figure &, A 10'( load was applied in four point bending arrangement with a load span of 100mm,
Figure 5 Modeling with effecti"e notch method Symmetry Mesh at Root Mesh at Toe
7 A 7
6atigue li"es were calculated using the following relationE N = C o m
+ere 0 and m are material parameters and o represents stress range, 0 is calculated using recommended 6AT "alue of ''?",a at 2 million cycles F'G, Pea) Stress Met,! 9nly half of geometry was modelled due to symmetry with 12A<E=' elements in A<SMS with free mesh ha"ing an element siHe of )! as was done by Meneghetti F>G, With this element type and mesh generation code! ' 1 F! * 1.%2 was used F>G and notch stress intensity factor ' 3 was calculated using K PL
= K I o pcuk J 1-x 1.S8 +ere K I is the stress intensity factorN d is element siHe and o pcuk represents stress at weld toe determined through pea stress method, K PL
and z are constants which depend on element type!
mesh siHe and the software used, 6atigue life was estimated by using the lower scatter band as reference cur"e. 6igure 6 shows mesh and loading condition,
Figure 6 Modelling with pea stress method 1.1 Res$!&a" Stresses 2 Pre!$ct$# a#! Meas&re.e#t It is well established from e8periments that fatigue crac at weld toes usually propagate into base metal under tensile or bending loads, Therefore trans"ersal residual stresses in the selected T75oint model become important, Residual stress analysis of the 5oint was performed with finite element based uncoupled thermal and mechanical analysis to study the stress state in the 5oint, Moreo"er! eeping in "iew the failure location suggested by fatigue calculations! trans"ersal residual stresses were measured in the 5oints to see the actual stresses and "erify the predicted results, Test S+ec$.e#s T75oints were constructed from D9ME; steels with yield strength :??! @(( and C>( ",a, Deometry of the 5oints is described in section %.1, The specimens were MAD welded! manually! in a single pass on each side with following welding conditionsE Tab"e ' B Welding condition for T75oints, Welding 1rocess #urrent FAG Voltage F4G Welding speed Fcm5minG 6iller wire diameter FmmG MAD 'C( :(,? ?( ),= Weld toe is shared by only two elements, See Meneghetti FrefG, Symmetry
7 C 7
Pre!$ct$# f Res$!&a" Stresses Welding simulations were performed in 'D and :D by a se$uentially uncoupled thermal and mechanical analysis, The temperature distribution predicted by transient thermal analysis was used as load for subse$uent mechanical analysis to estimate stresses due to welding, Same mesh was used for mechanical analysis as was used in thermal analysis and is shown in the figure ., %oundary conditions shown in figure were used, Since base plate was tac welded on the bottom side before welding of stiffener plate! it was held fi8ed for first weld and stiffener plate was allowed to deform, 6or welding on the other side! boundary conditions on the base plate were remo"ed and stiffener plate was constrained instead,
Figure 7 6EM model for residual stress simulations, a3 'D Model with boundary conditions, b3 boundary conditions for :D model for first weld, c3 boundary condition for weld on other side,
+eat source modeling described by %arsoum F)( J ))G was used for 'D simulations while a mo"ing heat source with constant "olume flu8 was implemented for :D analysis, Mo"ing heat source was implemented by di"iding the weld "olume into a finite number of "olumes and then se$uentially acti"ating elements contained by each "olume during welding, It was assumed that heat distribution due to welding arc was uniform and arc stayed on each "olume for a specific time before mo"ing on to ne8t "olume,
Figure 8 Mo"ing heat source for :D analysis
a b c
7 )( 7
Temperature dependent material properties were used for the simulations and are shown in figure / for D9ME;:?(, Material properties for other material grades were taen from 0mat1ro,
Figure 9 Temperature dependent material properties, a3 conducti"ity and specific heat b3 yield stress! tangent modulus and thermal e8pansion coefficient F)(G
Meas&re.e#t f Tra#s0ersa" Res$!&a" Stresses Symmetrical distribution of trans"ersal residual stress was assumed based on restraint conditions during welding and measurements were taen only on one side of the stiffener, ;7ray diffraction method was used and measurements were taen on top surface of base plate, Measurements were also compared with results obtained through finite element simulations,
7 )) 7
4 RESULTS AND DISCUSSION 4.' Fat$%&e Stre#%t, E0a"&at$# 6atigue e"aluation was carried out on the basis of first principle stress whose ma8imum "alue occurs at load carrying weld toe for all 5oints and mars the position of fatigue failure! see figure 10, :D analysis also gi"es the same results and is shown in figure 10.
Figure 10 Distribution of first principal stress in the 5oint <otch stress was e"aluated along the longitudinal direction with &00mm 5oint length to figure out the possibility of a distinct failure location, It was obser"ed that stress remains nearly constant in the middle e8cept for the start and stop position where it declines sharply, 6igure 11 shows the distribution of principle stress along longitudinal direction for a particular set of weld geometry,
Figure 11 6irst principal stress along longitudinal direction The distribution shows no distinct location of fatigue failure and thus failure location is more go"erned by distribution of residual stress and welding defects,
7 )' 7
It was further in"estigated whether this longitudinal distribution of ma8imum principal stress could be affected by different lengths of applied load and boundary condition! as shown in figure 12. Results show that longitudinal distribution of first principal notch stress remains nearly indifferent with different load lengths as modeled with parameter L d ! table 2,
Figure 12 Illustration of different load lengths Table 2 Distribution of first principal stress in longitudinal direction for different load lengths, L!5'6.. L!516..
L!576.. L!586..
7 ): 7
C.+ar$s# betwee# Effect$0e Ntc, Met,! a#! Pea) Stress Met,! 6atigue strength data obtained through effecti"e notch stress method has been mapped on scatter bands for steel! proposed by pea stress method F? J >G, The plot in figure 1% shows that pea stress method is more conser"ati"e than effecti"e notch method for prediction of fatigue life for welded 5oints, This method is more efficient than other contemporary methods of fatigue assessment for welded 5oints because it re$uires significantly less computational resources, +owe"er this efficiency has come with certain limitations on its applicability, 6or instance! it is currently "alid for sharp notches restricted to an opening angle of 1%& degrees and therefore canKt be used for fatigue assessment of post weld treated 5oints,
Figure 13 6atigue li"es of the 5oints mapped on pea stress scatter bands 4.* I#f"&e#ce f Ge.etr$c Factrs # Fat$%&e Stre#%t, <on7load carrying cruciform and T75oints are critical to fatigue failure at weld toes when sub5ected to tensile andOor bending loads, Stress raising effects of the welded 5oint and thereby fatigue strength of the 5oint considerably depends on weld geometry, Therefore optimum weld geometry can be de"eloped if a mathematical model correlating fatigue life and weld geometry parameters is nown, To accomplish this! a '7le"el full factorial design of e8periment was performed to identify the weld and global geometry parameters affecting fatigue life of the 5oint, The fatigue li"es and stress concentration factors were determined through effecti"e notch stress method, Since fatigue failure at toe was suggested by 6EM! the effect of weld penetration was not included in the study, Table % and ) summariHe the weld geometry parameters in"estigated, Tab"e 1 2 Weld parameters for as welded 5oints, 1arameter Symbol 2ow le"el +igh le"el Toe radius P FmmG r ) ' Throat thicness FmmG a = > 6lan Angle FdegG Q =? >( : )
7 )= 7
Tab"e 4 2 Weld parameters for post weld treated 5oints, 1arameter Symbol 2ow le"el +igh le"el Toe radius P FmmG r = ? Throat thicness FmmG a = > 6lan Angle FdegG Q =? >( P Toe radius here indicates effecti"e notch radius, As We"!e! /$#ts Stress concentration factor ' t was calculated for different simulated cases and the effect of different geometric factor on stress concentration factor is shown in the plots belowE
Figure 14 Influence of main factors on stress concentration factor The plot shows that weld toe radius and base plate thicness are the most dominating factors affecting fatigue strength of the load carrying 5oint, While a higher toe radius tends to reduce the stress concentration factor significantly! thicness of the base plate does the opposite, Therefore a thinner fatigue loaded 5oint would be safer compared to a thicer one pro"ided that nominal stress in the 5oint is same, This is because of thicness effect and has its e8planation in sharp stress gradients associated with plate thicness FCG, 9ther factors and their interactions ha"e a little impact on K t ! as indicated by figure 1), %ased on these obser"ation! following mathematical model establishes a relationship among K t and different geometric factors for as welded 5ointsE K t = 1.Su7S +u.u176r + u.uS1S1t 1 - u.u2SS rt 1 + u.uu274o Pst We"! Treate! /$#ts Design of e8periment in"estigation for post weld treated 5oints show results similar to those of as welded 5oints, +owe"er! one notable thing is that the effect of weld geometry parameters! particularly weld toe radius becomes less pronounced, The relationship of K t with geometric factors! for post weld treated 5oints isE 2 1 2.1 2.0 1.9 1.8 1.7 10 6 10 6 6 4 2.1 2.0 1.9 1.8 1.7 60 45 Toe Radius M e a n Base Plate Thickness Stiffner Thiickness Throat Thickness lank !n"le Main Effects Plot for Stress Concentration Factor, Kt #ata $eans
7 )? 7
K t = 1.u9SS2 - u.u1S7Sr + u.u6927t 1 -u.uu6S4 rt 1
Figure 15 Influence of main factors on stress concentration factor for post weld treated 5oints, 4.1 We"!$#% Res$!&a" Stresses S$.&"at$# f We"!$#% Res$!&a" Stress During thermal analysis! "olume heat flu8 was ad5usted with a body factor to obtain a reasonable siHe of molten weld pool F)(G and is shown in figure 16, Temperature history of a point )?mm away from weld toe is also shown,
Figure 16 Temperature distribution in molten Hone, a/ +A- Hone b/ Temperature history
6ollowing diagrams show different stresses in 6mm thic 5oints, Results show that longitudinal stresses can be of the order of material yield strength or e"en more howe"er trans"ersal and a8ial stresses donKt, 5 4 1.45 1.40 1.%5 1.%0 1.25 10 6 10 6 6 4 1.45 1.40 1.%5 1.%0 1.25 60 45 Toe Radius M e a n Base Plate Thickness Stiffner Thickness Throat Thickness lank !n"le Main Effects Plot for Stress Concentration Factor, Kt #ata $eans a b
7 )> 7
Figure 17 - Stresses in 6mm 5oints, S87Trans"ersal! Sy7A8ial! SH72ongitudinal,
7 )@ 7
Meas&re.e#t f Tra#s0ersa" Res$!&a" Stress 6ollowing figure show the measurements for trans"ersal residual stressE
Figure 18 Measurement of trans"ersal residual stresses, a/ 6mm 5oints b/ 10mm 5oints The results show that trans"ersal residual stresses ha"e a pea "alue near weld toe and tend to rela8 sharply, %eing tensile in nature! this can be detrimental for fatigue loaded 5oints, It can also been seen from the plots that irrespecti"e of material yield strength and plate thicness! trans"ersal residual stresses ha"e a pea "alue at the same distance from toe and tend to rela8 at a same distance for all 5oints, This indicates that distribution of residual stresses is independent of mechanical properties and depends solely on thermal properties of the material, Measurements on both sides of stiffener were also performed on another set of specimens and are shown in figure 12, They show nearly symmetrical distribution of trans"ersal stress on both sides, 6or the second sample of /60",a steel! plot shows a big "ariation which is possibly due to a different constraint condition during welding, a b M:??t> represents the 5oint with material yield strength %&&",a and thicness 6mm, Same applies to other legends,
7 )A 7
Figure 19 Trans"ersal residual stress
9177t: 9177t'6 9;:6t: 9<66t:
7 )C 7
C.+ar$s# f FEM Res&"ts w$t, Meas&re.e#ts 6igure 1/ shows comparison of measurements with 6EM results for 6mm 5oints, 6EM results agree "ery well for :?(M1a 5oint, They donKt agree well for other 5oints! howe"er show similar distribution pattern,
Figure 20 - #omparison of measurements with 6EM results S.e Factrs I#f"&e#c$#% Pea) 0a"&e f Tra#s0ersa" Res$!&a" Stress A design of e8periment in"estigation was carried out to study different factors that may ha"e an effect on pea "alue of trans"ersal residual stresses, 6actors studied include yield strength! plate thicness and throat thicness and are described in table 6, Table 5 1arameters for DoE in"estigation of Trans"ersal Residual Stress 1arameter 2ow 2e"el Intermediate le"el +igh le"el Mield Strngth F",aG %&& @(( C>( 1late thicness FmmG 6 7 )( Throat Thicness FmmG = ? >
1ea "alue of trans"ersal stress for described boundary condition was! obtained through 6EM simulations! was used as response, 6ollowing results were obtainedE 9177t: 9<66t: 9;:6t:
7 '( 7
a
b Figure 21 6actors affecting pea "alue of trans"ersal residual stress, a/ Main factors b/ Interaction of parameters 1lots show that only yield strength of the material has a significant effect on pea "alue of trans"ersal residual stress, With higher yield strength! pea "alue of residual stress increases! figure 12a! and the effect is more pronounced as thicness of the 5oint is increased! figure 126, It should be noted! howe"er! that these results are "alid for a particular set of parameters and a specific boundary condition and does not represent fillet welded 5oints in general,
% 2 1 %20 280 240 200 % 2 1 2 1 %20 280 240 200 &ield Stress M e a n Throat Plate Thickness Main Effects Plot for Peak Stress #ata $eans % 2 1 2 1 400 %00 200 400 %00 200 Yield Stress Throat Plate Thickness 1 2 % Stress &ield 1 2 % Throat Interaction Plot for Peak Stress #ata $eans
7 ') 7
7 CONCLUSIONS Based on above results, following conclusions can be made:
[1] Fatigue strength of fillet welded joints is considerably affected by plate thickness and toe radius. Other weld geometry parameters dont play an important role for fatigue life enhancement. [2] Notch stress, as calculated by effective notch method, remains almost constant along longitudinal direction of the joint and therefore no distinct failure point can be assumed. [3] Longitudinal stresses due to welding can be of the order of material yield strength or even more however transversal and axial stresses remain much below yield limit. [4] Transversal residual stresses tend to have peak value at same distance from toe location irrespective of material yield strength and plate thickness. They also tend to relax at same distance from toe. [5] Material yield strength as well as constraint conditions during welding govern the peak value of transversal residual stresses in the joint.
7 '' 7
: REFERENCES
[1] Radaj D. and Sonsino C.M., Fatigue assessment of welded joints with local approaches, Abington publishing, Cambridge, 1998. [2] Hobbacher A., Fatigue design of welded joints and components, IIW doc. XIII-1539-96, 1996. [3] Fricke W., Guidelines for the fatigue assessment by notch stress analysis for welded structures, IIW Doc. XIII-WG3-03r6-08, 2008. [4] Barsoum Z., Residual stress analysis and fatigue assessment of welded structure, Ph.D. thesis, Dept. of Aeronautical and Vehicle Engineering, KTH, Sweden, 2005. [5] Barsoum Z. and Samuelsson J., Fatigue assessment of cruciform joints with different methods, Steel Research International 77, No. 12, 2006. [6] Barsoum Z. and Jonsson B, Fatigue assessment and LEFM analysis of cruciform welded joints fabricated with different welding processes, IIW, 2007. [7] Meneghetti G., The peak stress method applied to fatigue assessment of steel and aluminum fillet welded joints subjected to Mode-I loading, Fatigue and Fracture of Engineering Materials and structures, 2008. [8] Meneghetti G. and Lazzarin P., The use of peak stress method for fatigue strength assessments of welded lap joints and cover plates with toe and root failures, Engineering Fracture Mechanics, 2011. [9] Mats G., A study of thickness effect on fatigue in thin welded high strength steel joints, Steel Research International, ISSN 1611-3683, 2006. [10] Barsoum Z. and Lunbck A., Simplified FE Welding simulation of fillet welds 3D effects on the formation of residual stresses, Engineering Failure Analysis, 2009. [11] Barsoum Z.and Barsoum I., Residual stress effects on fatigue life of welded structures using LEFM, Engineering Failure Analysis, 2009. [12] Martinsson J., Fatigue assessment of complex welding structures, Ph.D. thesis, Dept. of Aeronautical and Vehicle Engineering, KTH, Sweden, 2005. [13] Hansen J.L., Numerical modeling of welding induced stresses, Ph.D. thesis, Technical university of Denmark, ISBN 87-90855-52-3, 2003. [14] Barsoum Z., Residual stress analysis and fatigue of multipass welded tubular structures, Engineering Failure Analysis, 2008. [15] ANSYS guide, Ansys Release 13.0, Swanson analysis systems: Housten.