Predicting Fatigue Life With ANSYS Workbench

Predicting Fatigue Life with ANSYSWorkbenchHow To Design Products That Meet Their Intended DesignLife RequirementsRaymond L. Browell, P. E.Product Manager New TechnologiesANSYS, Inc.Al HancqDevelopment EngineerANSYS, Inc. 2006 ANSYS, Inc.ANSYS, Inc. Proprietary

Fatigue Agenda General Fatigue ReviewFatigue Analysis in WorkbenchWorking with Legacy ModelsOptimization with FatigueFatigue Within Workbench Fits IntoEveryone’s Process! Questions and Answers 2006 ANSYS, Inc.ANSYS, Inc. Proprietary

General Fatigue Review Why Fatigue Analysis?– While many parts may work well initially,they often fail in service due to fatiguefailure caused by repeated cyclic loading– In practice, loads significantly below staticlimits can cause failure if the load isrepeated sufficient times– Characterizing the capability of a materialto survive the many cycles a componentmay experience during its lifetime is theaim of fatigue analysis 2006 ANSYS, Inc.ANSYS, Inc. Proprietary

General Fatigue Review Common Decisionsfor Fatigue Analysis– Fatigue AnalysisType– Loading Type– Mean Stress Effects– Multiaxial StressCorrection– Fatigue ModificationFactorFatigue Analysis TypeLoading TypeMean Stress EffectsMultiaxial StressCorrectionFatigue Modifications 2006 ANSYS, Inc.ANSYS, Inc. Proprietary

General Fatigue Review Fatigue Analysis Types– Strain Life(Available in ANSYS Fatigue Module)– Stress Life(Available in ANSYS Fatigue Module)– Fracture Mechanics 2006 ANSYS, Inc.ANSYS, Inc. Proprietary

General Fatigue Review Strain Life– Strain can be directly measured and has been shown tobe an excellent quantity for characterizing low-cyclefatigue– Strain Life is typically concerned with crack initiation– In terms of cycles, Strain Life typically deals with arelatively low number of cycles and therefore addressesLow Cycle Fatigue (LCF), but works with high numbersof cycles as well– Low Cycle Fatigue usually refers tofewer than 105 (100,000) cycles– The Strain Life approach is widely usedat present 2006 ANSYS, Inc.ANSYS, Inc. Proprietary

General Fatigue Review Strain LifeThe Strain Life Relation equationis shown below:Δε σ (2N ) ε (2N )2EThe two cyclic stress-strain parametersare part of the equation below:'bffΔε2ΔσEN'Δεcff Total Strain Amplitudeσ 2 x the Stress AmplitudeB Modulus of ElasticityεfΔσE 21/ n ' Fatigue Strength Coefficient Fatigue Strength Exponent(Basquin’s Exponent)'f Fatigue Ductility Coefficient Number of Cycles to FailurefC2N' Δσ ' Kf Number of Reversals to FailureKn 2006 ANSYS, Inc. Fatigue Ductility Exponent' Cyclic Strength Coefficient' Cyclic Strain Hardening ExponentANSYS, Inc. Proprietary

General Fatigue Review Strain Life– For Strain Life, the total strain (elastic plastic) is the required input– But, running an FE analysis to determinethe total response can be very expensiveand wasteful, especially if the nominalresponse of the structure is elastic– So, an accepted approach is to assume anominally elastic response and then makeuse of Neuber’s equation to relate localstress/strain to nominal stress/strain at astress concentration location 2006 ANSYS, Inc.ANSYS, Inc. Proprietary

General Fatigue Review Strain LifeTo relate strain to stress we use Neuber’s Rule, which is shown below:εσ2K eStε Local (Total) Strainσ Local StressK 2006 ANSYS, Inc. t Elastic Stress Concentration Factore Nominal Elastic StrainS Nominal Elastic StressANSYS, Inc. Proprietary

General Fatigue Review Strain Life– Simultaneously solving Neuber’s equation alongwith cyclic strain equation, we can calculate thelocal stress/strains (including plastic response)given only elastic input– Note that this calculation is nonlinear and is solvedvia iterative methods– ANSYS fatigue uses a value of 1 for Kt, assumingthat the mesh is refined enough to capture anystress concentration effects 2006 ANSYS, Inc. Note: This Kt is not be confused with the StressReduction Factor option which is typically used in Stresslife analysis to account for things such as reliability andsize effectsANSYS, Inc. Proprietary

General Fatigue Review Stress Life 2006 ANSYS, Inc.– Stress Life is based on S-N curves (Stress – Cyclecurves)– Stress Life is concerned with total life and does notdistinguish between initiation and propagation– In terms of cycles, Stress Life typically deals with arelatively High number of cycles– High number of cycles is usually refers to more than105 (100,000) cycles– Stress Life traditionally deals withrelatively high numbers of cycles andtherefore addresses High Cycle Fatigue(HCF), greater than 105 cycles inclusiveof infinite lifeANSYS, Inc. Proprietary

General Fatigue Review Fracture Mechanics– Fracture Mechanics starts with an assumed flaw of knownsize and determines the crack’s growth– Facture Mechanics is therefore sometimes referred to as“Crack Life”– Facture Mechanics is widely used to determine inspectionintervals. For a given inspection technique, the smallestdetectable flaw size is known. From this detectable flaw sizewe can calculate the time required for the crack to grow to acritical size. We can then determine our inspection intervalto be less than the crack growth time.– Sometimes, Strain Life methods are used to determine crackinitiation with Fracture Mechanics used to determine thecrack life therefore:Crack Initiation Crack Life Total Life 2006 ANSYS, Inc.ANSYS, Inc. Proprietary

Fatigue Analysis in Workbench ANSYS Fatigue Module– Integrated into the ANSYS Workbench Environment– ANSYS Fatigue Module is able to further leverageadvances that the ANSYS Workbench offers such as: 2006 ANSYS, Inc.CAD support including Bi-Directional ParametersSolid ModelingVirtual TopologyRobust MeshingHex-Dominant MeshingAutomatic Contact DetectionOptimizationDesign for Six SigmaRobust Design that the ANSYS Workbench offersANSYS, Inc. Proprietary

Fatigue Analysis in Workbench Strain Life Decisionsfor Fatigue Analysis– Loading Type– Mean Stress Effects– Multiaxial StressCorrection– Fatigue ModificationFactor 2006 ANSYS, Inc.ANSYS, Inc. Proprietary

Fatigue Analysis in Workbench Stress LifeDecisions forFatigue Analysis– Loading Type– Mean Stress Effects– Multiaxial StressCorrection– Fatigue ModificationFactor 2006 ANSYS, Inc.ANSYS, Inc. Proprietary

Fatigue Analysis in Workbench Loading Types– Constant amplitude, proportional loading– Constant amplitude, non-proportionalloading– Non-constant amplitude, proportionalloading– Non-constant amplitude, non-proportionalloading 2006 ANSYS, Inc.ANSYS, Inc. Proprietary

Fatigue Analysis in Workbench Constant amplitude, proportional loading– Loading is of constant amplitude because onlyone set of FE stress results along with aloading ratio is required to calculate thealternating and mean values 2006 ANSYS, Inc.ANSYS, Inc. Proprietary

Fatigue Analysis in Workbench Constant amplitude, proportional loading– Is the classic, “back of the envelope”calculation describing whether the load has aconstant maximum value or continually varieswith time– The loading ratio is defined as the ratio of thesecond load to the first load (LR L2/L1)– Common types of constant amplitude loadingare: 2006 ANSYS, Inc. Fully reversed (apply a load, then apply an equaland opposite load; a load ratio of –1) Zero-based (apply a load then remove it; a load ratioof 0)ANSYS, Inc. Proprietary

Fatigue Analysis in Workbench Constant amplitude, proportional loading– Loading is proportional since only one set of FEresults are needed (principal stress axes do notchange over time)– Since loading is proportional, looking at a singleset of FE results can identify critical fatiguelocations.– Since there are only two loadings, no cyclecounting or cumulative damage calculationsneed to be done 2006 ANSYS, Inc.ANSYS, Inc. Proprietary

Fatigue Analysis in Workbench Constant amplitude, non-proportional loading– Looks at exactly two load cases that need not berelated by a scale factor– Analyses where loading is proportional but resultsare not– The loading is of constant amplitude but nonproportional since principal stress or strain axesare free to change between the two load sets– This happens under conditions where changingthe direction or magnitude of loads causes achange in the relative stress distribution in themodel. This may be important in situations withnonlinear contact, compression-only surfaces, orbolt loads 2006 ANSYS, Inc.ANSYS, Inc. Proprietary

Fatigue Analysis in Workbench Constant amplitude, non-proportional loading– This type of fatigue loading can describe commonfatigue loadings such as: Alternating between two distinct load cases (like abending load and torsional load) Applying an alternating load superimposed on a staticload– No cycle counting needs to be done– But since the loading is non-proportional, thecritical fatigue location may occur at a spatiallocation that is not easily identifiable by looking ateither of the base loading stress states 2006 ANSYS, Inc.ANSYS, Inc. Proprietary

Fatigue Analysis in Workbench Constant amplitude, non-proportional loading– Fatigue tools located under asolution branch are inherentlyapplied to that single branchand thus can only handleproportional loading 2006 ANSYS, Inc. In order to handle nonproportional loading, the fatiguetool must be able to spanmultiple solutions This is accomplished by adding afatigue tool under the solutioncombination folder that canindeed span multiple solutionbranchesANSYS, Inc. Proprietary

Fatigue Analysis in Workbench Non-constant amplitude, proportional loading– Instead of using a single load ratio to calculate alternating andmean values, the load ratio varies over time– Think of this as coupling an FE analysis with strain-gauge resultscollected over a given time interval– Since loading is proportional, the critical fatigue location can befound by looking at a single set of FE results 2006 ANSYS, Inc.ANSYS, Inc. Proprietary

Fatigue Analysis in Workbench Non-constant amplitude,proportional loading 2006 ANSYS, Inc.– However, the fatigue loadingwhich causes the maximumdamage cannot easily be seen– Thus, cumulative damagecalculations (including cyclecounting such as Rainflow anddamage summation such asMiner’s rule) need to be doneto determine the total amountof fatigue damage and whichcycle combinations cause thatdamage– Cycle counting is a means toreduce a complex load historyinto a number of events, whichcan be compared to theavailable constant amplitudetest dataANSYS, Inc. Proprietary

Fatigue Analysis in Workbench Non-constant amplitude, proportional loading– The ANSYS Fatigue Module uses a “quickcounting” technique to substantially reduceruntime and memory– In quick counting, alternating and mean stressesare sorted into bins before partial damage iscalculated– Without quick counting, data is notsorted into bins until after partialdamages are found– The accuracy of quick counting isusually very good if a proper numberof bins are used when counting 2006 ANSYS, Inc.ANSYS, Inc. Proprietary

Fatigue Analysis in Workbench Non-constant amplitude, proportional loading– Bin size defines how many divisions the cyclecounting history should be organized into Strictly speaking, bin sizespecifies the number ofdivisions of the rainflow matrix A larger bin size has greaterprecision but will take longer tosolve and use more memory Bin size defaults to 32, meaningthat the Rainflow Matrix is 32 x32 in dimension. 2006 ANSYS, Inc.ANSYS, Inc. Proprietary