# Fatigue resistance of welded joints in aluminium high-speed craft: A total stress concept

Fatigue resistance of welded joints in aluminium high-speed craft: A total stress concept

Author Contributor Faculty DepartmentMaritime and Transport Technology (Ship Hydromechanics and Structures)

Date2015-09-03

AbstractCrew transfers, surveillance duties and {security, rescue, interception} operations at sea typically require high-speed craft. Aluminium is quite often selected as hull structure material because of its weight save potential in comparison to steel. The fatigue strength, however, may become a point of concern because of the decreased Young’s modulus. Bottom slamming is identified as a dominant type of repeated loading, meaning fatigue is a governing limit state in aluminium high-speed craft design. Particular attention in that respect is paid to arc-welded joints connecting the hull structure components, {plates, shells}, since the weld geometry introduces notches; fatigue sensitive locations. Fatigue physics cover an extensive range of scales and modelling may require a multi-scale approach. Adopting a structural response parameter S available at FSS level using global information only, however, seems attractive since S controls plasticity – required to facilitate fatigue damage: crack initiation, growth, propagation and fracture – at macro (structural)- as well as meso and micro (material) scale, but pays off in fatigue resistance data scatter and life time estimate uncertainty. Including physics at smaller scale, local information, improves the accuracy. A continuous increase of the considered scale range of physics as observed in fatigue assessment concepts developed over time – proposed to be classified according to approach, criterion, parameter and process zone – is however typically associated to increased (computational) effort and concept complexity. At the same time, similarity; proper scaling, meaning equal parameter values should yield the same fatigue resistance, seems still incomplete since all concepts available involve multiple fatigue resistance curves rather than one. From {MCF, HCF} design perspective, a local continuum mechanics approach seems sufficient and a total stress concept is proposed to balance accuracy, effort and complexity, improving similarity at the same time to obtain one aluminium arc-welded joint fatigue resistance curve. The weld geometry introduces at least a notch at the weld toe and depending on penetration level another one at the weld root. Cracks may initiate at both fatigue sensitive locations, grow principally in {plate, shell} thickness direction and continue to propagate in general either along or perpendicular to the weld seam through {plate, shell} because of the structure orthotropic stiffness characteristics, suggesting a {plate, shell} thickness based (detectable repair) criterion to be an appropriate fatigue design parameter. The total through-thickness weld notch stress distribution along the expected crack path {??_n?^T,??_nr?^T }, including both the ocean/sea waves induced cyclic remote mechanical loading- and welding process related quasi-constant thermal residual part, is assumed to be a key element. The predominant remote mechanical loading mode-I contribution {?_n,?_nr } has been examined to distinguish the involved stress components. A self-equilibrating weld geometry stress – consisting of a local V-shaped notch- and weld load carrying part – and equilibrium equivalent global structural field stress are identified; a refinement of a well-known definition. The semi-analytical formulations are related to the welded joint far field stress, calculated using a relatively coarse meshed {plate, shell} FE model as typically available for fatigue design purposes. Exploiting (non-)symmetry conditions, a generalised formulation demonstrating stress field similarity has been obtained and extends to the welding induced thermal residual stress distributions {??_n?^r,??_nr?^r }. Fatigue scaling requires both the (zone 1) peak value and (zone 2 notch affected and zone 3 far field dominated) gradient to be incorporated, meaning a damage criterion should take the complete distribution into account. The SIF K seems to meet this criterion, though, the intact geometry related notch stress distributions should be correlated to crack damaged equivalents; fatigue is assumed to be a crack growth (dominated) process. At the same time, hull structure arc-welded joints inevitably contain flaws or crack nuclei (defects) at the weld toe- and root notches, i.e. using the damage tolerant mode-I parameter K_I seems justified since fatigue associated to the {MCF, HCF} life time range at both locations will predominantly be a matter of micro- and macro-crack growth. The zone 3 related equilibrium equivalent stress contribution has been used to obtain a far field factor, distinguishing different type of cracks related to (non-) symmetry conditions for both (quasi) 2D- and 3D configurations. A notch factor incorporates the zone {1, 2} governing self-equilibrating stress. Remote mechanical weld toe- and weld root stress intensities show the zone {1, 2} notch affected- and zone 3 far field dominated parts define a micro- and macro-crack region, turning the stress field similarity into a stress intensity similarity. Each stress component dominates a certain crack length range: the notch stress the micro-crack region, the structural field stress the macro-crack region; the weld load carrying stress determines the transition (i.e. apex) location. The welding induced and displacement controlled mode-I residual stress intensity factor ?K_I?^r is acquired for both weld toe and weld root notches to complete the total weld notch stress intensity similarity factor formulation ?K_I?^T. Cyclic remote mechanical- and quasi-constant thermal residual loading turn ?K_I?^T into a crack growth driving force ??K_I?^T and defects may develop into cracks. The crack growth rate (da/dn) of micro-cracks emanating at notches show elastoplastic wake field affected anomalies, i.e. monotonically increasing or non-monotonic behaviour beyond the material threshold. Modifying Paris’ equation, a two-stage micro- and macro-crack growth law similarity is proposed to include both the weld notch- and far field characteristic contributions, elastoplasticity as well as remote mechanical- and thermal residual mean stress effects. Small/short crack growth data obtained using standard specimens including {SEN, DEN, CEN} in crack configuration – representing weld root notch geometries at the same time – available in literature has been reinvestigated for the alternating material zones in (aluminium) arc-welded joints: WM and HAZ zone containing respectively the weld root- and weld toe notch fatigue damage location, as well as BM for comparison. Fatigue testing series have been developed to identify crack growth behaviour at weld toe notches in aluminium arc-welded joints, adopting a typical fillet weld DS T-joint geometry. Using DIC, the required far field- and notch region parameters are obtained. Spatial displacement fields are estimated on a general kinematic basis using commercial DIC software (Istra4D, Dantec Dynamics). A posteriori, as a mechanical filtering process, the displacement fields are decomposed onto a selected kinematic basis, i.e. an Airy stress function. The displacement amplitudes, least squares solutions, present in a one-to-one correspondence the crack growth governing parameters: linear far field stress distribution, SIF and crack tip location. A sequence of images provides the temporal solution; weld toe crack growth data series showing both far field characteristics and notch affected (non-monotonic) anomalies. Crack growth model integration yields a (MCF) single slope resistance relation, a joint S_T-N curve correlating arc-welded joint life time N and the total stress parameter S_T; a line (equivalent point) criterion to estimate hull structure longevity ensuring {SSS, LSS, FSS} welded joint fatigue resistance similarity. A dual slope (i.e. random fatigue limit) formulation has been adopted to incorporate HCF taking the transition in fatigue damage mechanism (i.e. growth dominant turns into initiation controlled for decreasing load level), a slope change, into account. Regression analysis (i.e. a likelihood approach) is adopted to estimate model parameters, managing both complete- and right-censored data; failures and run-outs. Artificial fatigue test data of DS T-joints is investigated to determine the S_T parameter quality. The fatigue life uncertainty is about a factor 2 (T_S?1:1.2). As-welded SSS (T-T) CA data available in literature has been used to establish a family of (damage tolerant engineering) joint S_T-N fatigue resistance design curves to be able to estimate the fatigue life time N of welded joints (production quality is average) knowing the joint geometry and far field structural response. The MCF life time uncertainty bandwidth increases up to a factor 6, i.e. (T_S?1:1.6). In the hull structure (HCF) design region uncertainty is significant, predominantly because of lacking complete data. Full scale structure representative {T-T literature, T-C} CA LSS data has been examined to verify a SSS data scatter band fit. Since CA {SSS, LSS} fatigue resistance is principally used to estimate a VA FSS value adopting the Palmgren-Miner hypothesis, VA SSS data available in literature is examined and a scatter band fit is observed. The involved equivalent total stress parameter S_(T,eq) is obtained adopting an extended rain flow counting algorithm to capture the damage cube. Last but not least, hourly fatigue damage estimates D_h are obtained for some frame-stiffener connections in the slamming zone of an aluminium high-speed craft, using the FSS response as measured for several trials at the North Sea. The wave (loading) statistics induced D_h uncertainty is about a factor 2.5 comparing the measurement- and simulation structural response based values; quite close to the MCF fatigue design resistance value of 3 (R99Cxx – R50Cxx). The TS concept is implemented in a high-speed craft fatigue design tool, available to all research partners. Using the welded joints geometry- and loading induced far field structural response information, the fatigue damage estimate D(S_T ) of all notch locations is calculated and the governing one identified to obtain life time N.

Subjecthigh-speed craft

aluminium

fatigue

welded joint

total stress concept

978-94-6233-040-5

Part of collectionInstitutional Repository

Document typedoctoral thesis

Rights(c) 2015 Den Besten, J.H.