Fatigue of Welded Pressure Vessels

Fatigue of Welded Pressure Vessels CCOPPS Webinar Wednesday 21 st May 2008 15:00 16:00 BST Steve Maddox Jim Wood The Welding Institute (TWI) University of Strathclyde

Agenda Introduction and Relevant CCOPPS Activity Jim Wood Fatigue of Welded Pressure Vessels Steve Maddox Q&A Session Steve Maddox & Jim Wood Closing Remarks Jim Wood

Industry Needs Survey Findings relevant to this webinar: FEA use is increasing as is complexity of models Interfacing with codes of practice seen as issue Happy with facilities in commercial codes in general (exception is weld modelling and assessment + automation of the analysis process) Non-European Codes are used often by most respondents. P li i t il bl f d l d f Preliminary report available for download from http://www.ccopps.eu/

Fatigue design of welded pressure vessels BS PD 5500 Annex C, Detailed fatigue assessment EN 13445-3 and - use of design curves ASME VIII, Division 2 - classification of weld details (new structural - stress concentrations stress approach) - fatigue life improvement Fatigue failure in Simplified assessment methods pressure vessels Fatigue assessment of welding Fatigue design data imperfections Stresses used in Future needs fatigue design

Effect of welding on fatigue resistance Stress range, MPa 400 300 200 100 50 R=0 Grade 50 structural steel 10 10 5 10 6 10 7 10 8 Cycles

Fatigue cracking in a welded joint Fatigue cracking from weld root Fatigue failure in plate at toe (from pre-existing sharp flaw at X) X

Key features of welds and welded joints Sharp section changes - High SCF Local discontinuities - Crack initiation sites High tensile residual stresses - Maximum mean stress effect, compressive Consequences: stresses damaging Relatively low fatigue strength, dominated by fatigue crack growth and controlled by full applied stress range. Fatigue life not increased by use of higher strength material

Fatigue failure in pressure vessels Fatigue loading Pressure fluctuations Temperature changes Temperature differentials External mechanical loading Vibration Sources of stress concentration and hence fatigue cracking Joints (welds, bolts) Geometric discontinuities (openings, nozzles, ends, supports) Temporary attachments

Fatigue failure from weld details Fatigue cracking from inside of butt weld between a cylindrical shell and a flat end Fatigue cracking from nozzle weld toe

Fatigue design process Compare number of repetitions (n i ) of stress range Sr i which vessel or part of vessel must withstand in its service life with the number (N i ) withstood by representative specimens at same stress in fatigue tests, such that: n n n n 1 + 2 + 3 + etc = Σ i 1. N N N N 0 1 2 3 N values obtained from relevant design S-N curves i

Basis of fatigue design data and assessment method for welded joints BS PD 5500 and EN 13445 adapted from nominal stress-based fatigue rules for welded structures (bridges, offshore structures, etc) Thus, grid of S-N curves each applicable to one or more particular weld detail, chosen on the basis of a classification system In general, curves related to structural stress range For potential fatigue failure from weld toe, structural stress range at toe (hot-spot structural stress range) may be used New ASME VIII uses only hot-spot structural stress calculated in a specific way and converted to a fracture mechanics based parameter called the Equivalent structural stress range parameter

Stress range S r N/mm 2 World Centre for Materials Joining Technology Fatigue design curves for weld details in BS PD 5500 and other UK Standards S r 3 N = constant for N < 10 7 cycles Applicable to any of metals covered 5 S r N = constant by PD 5500, on for N > 10 7 cycles basis of relative E values For thickness e > e ref, where e ref =22mm, allowable stress 0. 25 e 10 7 ref = Sr. y e E = 2.09 x 10 5 N/mm 2 Class D E F F2 G W Endurance N, cycles

Weld detail classification in BS PD 5500 Classification depends on: welded joint geometry direction of loading crack initiation site methods of manufacture and inspection

Stresses used in BS PD 5500 for fatigue design Structural (primary + secondary) stress range Principal stress (not stress intensity) used directly Nominal stress for simple details (e.g. attachments, seam welds) Nominal stress x SCF for structural details, or Hot-spot structural stress at any weld toe (Higher design curve than those used with nominal stress) Net section of load-carrying fillet welds

Fatigue design S-N curves for weld details in EN 13445 Stress range S r N/mm 2 Class 100 90 80 71 63 56 50 45 40 32 Class = fatigue strength in N/mm 2 at 2 x 10 6 cycles S r 5 N = constant for N > 5x10 6 cycles Currently applicable only to steel For thickness e > 25mm, allowable stress 3 S r N = constant for N < 5 x 10 6 cycles 2x10 6 5x10 6 Endurance N, cycles = S r 25mm. e 0.25

Stresses used in EN 13445 for fatigue design Structural (primary + secondary) stress range Option to use principal or equivalent stress range Nominal stress for simple details (e.g. attachments, seam welds) Structural stress at any weld toe (Hot-spot stress) Net section of load-carrying fillet welds

Vessel shell World Centre for Materials Joining Technology Weld detail classification in EN 13445 Ring stiffener Class 71 EN 13445 offers choice of using equivalent stress or principal stress. Since equivalent stress has no direction, a consequence is that the lowest detail Class must Hoop stress be assumed. Longitudinal Thus, the joint shown stress would be designed as Class 71. Class 80

ΔS ess World Centre for Materials Joining Technology Fatigue design curves from ASME VIII 1500 1000 MPa/mm -0.222 500 400 300 200 100 ASME 'Master' fatigue design curves Mean - 3SD (normal design curve) Mean - 2 SD 50 10 3 10 4 10 5 10 6 10 7 10 8 Endurance, cycles Applicable to any of metals covered by ASME on basis of relative E values Stress parameter depends on plate thickness, membrane/bending stress ratio and applied stress ratio (Not units of stress) No other thickness correction required

Basis of fatigue design curves Regression analysis of S-N data obtained from tests on actual welded joints 2.5% probability of failure (mean 2 standard deviations of log N [SD]):- BS PD 5500 design curves; included in ASME. 0.1% probability curves (mean - 3SD):- BS PD 5500 simplified design methods; all EN 13445 design curves for weld details; generally required for ASME

Application of design curves for weld details No effect of applied mean stress in PD 5500 and EN 13445; mean stress correction in ASME No effect of material tensile strength No effect of welding process May need to be reduced to allow for corrosive environment; no specific guidance in BS PD 5500 or EN 13445 but penalty factors specified in ASME

Stresses used in ASME VIII Div. 2 for fatigue design of welded joints Structural (primary + secondary) stress range based on through-thickness stress distribution obtained by numerical analysis Structural equivalent stress range parameter (a function of material s fatigue crack propagation properties (m=3.6), applied structural stress range (Δσ), material thickness (t), membrane to bending stress ratio and applied stress ratio): ΔS ess = t 2 2 m m ess Δσ.I m 1.f M

Generalized stress parameter (Maddox, 1974) da / dn = C( Δ K ), ΔK = YΔσ πa, Y = a f t a i t (Y d ( a t πa t ) ) m m = I = m C Δσ m t ( m 2 1) N f ( a / t m m ( 1) 1 2 t i.e Δ σ.n = 1 = cons tant I C or where or Δσ * ( * ) Δσ m.n = a cons tant, ' Generalized stress Δσ = (cf MASME ΔS 2 m m t 2 m.i 1 parameter' Δσ ess ) * & crack t = Δσ front m 2 I 1 m 1 shape a / 2c )

ASME Equivalent structural stress range parameter ΔS ess Based on: Structural stress (presumably assumed to allow for SCF factor M k normally applied to stress intensity factor) Fatigue life mainly growth of a pre-existing crack Implicit assumptions made about initial flaw size and shape Specific fatigue crack growth rate relationship

Hot-spot structural stress approach Fatigue crack Hot-spot Hot-spot stress = structural stress at weld toe. It includes all stress concentrating effects except the local notch effect of the weld toe.

Hot-spot structural stress from through-thickness stress distribution Actual stress Structural stress σ hot-spot t τ(y) σ x (y) τ m σ m σ b

Methods for calculating the hot-spot structural stress Using surface stresses (e.g. measured): Surface stress extrapolation (SSE) Using through-thickness stress distribution (e.g. from numerical analysis): - Through-thickness integration (TTI) - Nodal forces (NF) method

Hot-spot stress developments in British Standards Current TWI joint-industry project aims to produce guidance on hot-spot structural ral stress approach for inclusion in British Standard fatigue design rules (initially BS 7608) Comparison of the three methods of calculating hot-spot stress (SSE, TTI and NF) from FEA Solid and shell elements, examination of sensitivity to mesh size Case studies on range of structural components All methods mesh sensitive but mesh sensitivity least for simple welded joints in plates under unidirectional loading Findings so far indicate that nodal force method most mesh sensitive of the three when applied to structural component Mesh insensitivity of the ASME method may be for a restricted set of possible mesh types and weld meshing options.

Fatigue data expressed in terms of hotspot stress range (SSE method) 500 400 300 Hot-spot stress range at weld toe 200 N/mm 2 Class E 100 90 60 Pressure vessels - nozzles Plate welded components 95% confidence intervals enclosing data 10 4 10 5 10 6 10 7 3x10 7 Endurance, cycles

Hot-spot stress range at weld toe Comparison with mean 2SD S-N curves derived from ASME VIII 500 400 300 200 N/mm 2 BS PD 5500 Pressure vessel data Structural specimen or component data Mean ± 2SD Class E 100 90 ASME VIII, Div.2 R 0, t 16mm lower: all membrane upper: all bending 60 10 4 10 5 10 6 10 7 3x10 7 Endurance, cycles

Comparison with mean 3SD S-N curves derived from ASME VIII Hot-spot stress range at weld toe 500 400 300 Pressure vessel data Structural specimen or component data Mean ± 3SD 200 N/mm 2 EN 13445 Class 71 Class 63 100 90 ASME VIII, Div.2 R 0, t 16mm lower: all membrane upper: all bending 60 10 4 10 5 10 6 10 7 3x10 7 Endurance, cycles

Misalignment as a source of stress concentration Bending stresses due to misalignment Same guidance on calculation of SCF is given in PD5500 and EN 13445

Treatment of weld details subject to combined or multi-axial loading Δσ L = Maximum change in σ L σ H σ 45 σ L Δσ H = Maximum change in σ H Δσ 45 = Maximum change in σ H45 Principal stresses then calculated from Δσ L, Δσ H, and Δσ 45 Same approach in PD5500 and EN 13445. Current research should provide better method for out-of-phase loading.

ASME treatment t t of weld details subject to multi-axial loading ΔS = 1 Δσ ess F( ) 2 2 m m 2 2 m m m 1 m 1 δ tess.i.fm tess. I Where: 2 + 2 3 Δλ For in-phase loading (principal i stress directions remain constant t throughout loading cycle), F(δ) = 1 while For out-of-phase loading (principal stress directions change during loading cycle), F(δ) = function of applied normal and shear stresses and out-of-phase angle, or conservative value of 1/ 2 2 0.5

Elastic-plastic (E-P) conditions BS PD 5500 & EN 13445: Stress range For stress ranges > twice yield: E-P strain obtained directly from analysis and converted to stress or Stresses from design S-N curves reduced by factor that depends on load source (mechanical or thermal) and stress range / yield. ASME VIII, Div. 2: For every case: Corrected for plasticity Design S-N curve 10 3 10 4 10 5 10 6 10 7 10 8 Endurance N, cycles E-P strain obtained directly from analysis (Neuber s rule & material s cyclic stress-strain strain properties) and converted to stress range

Fatigue life improvement methods BS PD 5500 and EN 13445 allow weld toe grinding and corresponding increase in design classification ASME VIII accepts weld toe grinding, TIG dressing or hammer peening. Equivalent structural stress parameter design curves increased accordingly, with greatest benefit in high-cycle regime

Assessment of welding flaws Guidance given in BS PD 5500 and EN 13445 on flaw assessment based on fitness-for-purpose: Reference to BS 7910 Specific recommendations on: 1. Planar flaws not acceptable 2. Buried flaws - inclusions,,porosity; 3. Deviations from design shape - misalignment, peaking, ovality. ASME seems to accept planar flaws since equivalent structural stress parameter can be calculated for cracks up to 10% of section thickness

Fatigue design of pressure vessels - future needs Parametric hot-spot SCFs for pressure vessel details Elastic-plastic fatigue Effect of environment (corrosive, elevated temperature, hydrogen) Closer link between design and fabrication quality Experimental methods for design (draft for EN 13445 now available)

Thank you for your attention