Jtam.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 49, 1, pp. 83-95, Warsaw 2011 THE INFLUENCE OF LOADING PROGRAM ON THE COURSE OF FATIGUE DAMAGE CUMULATION Stanisław Mroziński University of Technology and Life Sciences in Bydgoszcz, Faculty of Mechanical Engineering, Bydgoszcz, Poland; e-mail: stanislaw.mrozinski@utp.edu.pl In the paper, results of a comparative analysis of cyclic properties of specimens made of 30HGSA steel under constant-amplitude and pro- grammed loading were presented. The analysis was carried out with the use of parameters of the characteristic hysteresis loop in function of the degree of fatigue damage. The analysis showed that courses of cyclic properties changes for chosen strain levels are very similar and do not depend on the loadingprogram. Itwas shown thatduringdamage cumu- lation it is possible to take into account the changes of cyclic properties. Key words: fatigue life, low cycle fatigue, damage accumulation, pro- grammed loading 1. Introduction Fatigue life calculations of construction elements are inseparably connected with the problemof the fatigue damage cumulation. Since 1924 till the present day, there have appeared about 40 various fatigue damage cumulation hypo- theses. A comparative analysis of the fatigue damage cumulation hypotheses was performed among others in the papers by Manson and Halford (1986), Fatemi and Yang (1998), Szala (1998). The oldest one is the Palmgren-Miner linear hypothesis (Palmgren, 1924; Miner, 1945). According to the Palmgren- Miner rule, fatigue damage cumulation can be performed with the use of va- rious fatigue descriptions (stress, strain, energy).Analysis of particular fatigue descriptions allows one to conclude that the energy description of the fatigue is more complete than the stress or strain description. It takes into account themutual interactions between stress and strain. For the energy description, numerous proposals of cumulation hypotheses, alternative to the linear hypo- thesis were formulated (Kujawski and Ellyin, 1984; Gołoś and Ellyin, 1988; 84 S. Mroziński Leis, 1988; Duyi and Zhenlin, 2001; Mroziński and Topoliński, 1999). Despi- te good agreement of the experimental results with the fatigue life calculation ones obtainedwith the use of newhypotheses, they have not foundwidespread application in fatigue life analysis so far. One of the reasons for lasting popularity of the linear hypothesis is its simplicity and still satisfactory agreement of the obtained calculation and test results, as far as the engineers’, opinion is concerned. In the linear hypothesis, it is accepted that in the case of a constant-amplitude loading each loading cycle, independently of the phase of the fatigue process, contributes to damage in the same degree. For example, after carrying out ni cycles of the constant- amplitude loading, the degree of fatigue damage can be determined from the relation Di = ni Ni (1.1) where ni is the number of cycles of the constant-amplitude loading, Ni – number of cycles until failure at the given level of loading. In the case of a multistep loading program, the failure will occur if the following condition is met Di =λ k∑ i=1 ni Ni =1 (1.2) where k denotes the number of steps in the program, λ – number of program iterations until failure. The course of damage cumulation according to the linear hypothesis and found with the use of plastic strain energy ∆Wpl dissipated in the material during one loading cycle as the criterion value, is shown in Fig.1. At Apoint, after carryingout nb1 cycles of the constant-amplitude loading with energy ∆Wpl3, the damage degree is Di(s) = nb1 N3 (1.3) However, in the case of multistep loading, after carrying out n1 + n2 +n3 cycles on the successive levels of the program, the degree of fatigue damage Di(p) at A point will be Di(p) = n1 N1 + n2 N2 + n3 N3 (1.4) According to the linear hypothesis of damage cumulation, the damage degrees Di(s) and Di(p) at A point should be the same Di(s) =Di(p) (1.5) The influence of loading program... 85 Fig. 1. Damage cumulation according to the Palmgren-Miner hypothesis in the energy approach: (a) loading program, (b) course of damage cumulation In the context of assumptions of the fatigue life calculation method, in which material data were determined in the low-cycle fatigue area, it should also mean the same criteria for ∆Wpl at A point after nb1 cycles of the constant- amplitude loading and after n1+n2+n3 cycles of the programmed loading. Such a case was shown at the diagram in Fig.1. In order to make it simpler, it was accepted that during the tests changes of the criterion value ∆Wpl at individual levels of the loadingprogramarenotobserved. Suchacase, however, takes place only when the energy ∆Wpl is the controlling value during the test (∆Wpl = const). Because of problems concerning the strength machines control, such tests are hardly ever performed, however. They were presented, for example, in the papers by Boroński and Mroziński (2007), Kasprzyczak andMacha (2006). Most often, the energy ∆Wpl is the resultant value calculated after reali- sation of tests under controlled stress (σa = const) or strain (εa = const, εap = const). As a result of changes of cyclic properties (weakening or har- dening of the material) there occur changes of the criterion value ∆Wpl in function of the number of loading cycles at individual levels of the constant- amplitude loading (Koh, 2002; Li et al., 1997; Mroziński, 2008). The changes of cyclic propertiesmay be one of the reasons of the discrepancy between fati- gue life calculations and experimental results if they are not considered in the calculations. The basic aim of this paper is valuation of the possibility of taking into account changes of cyclic properties of the material during fatigue damage cumulation obtained with the use of the linear method. An additional aim is to determine the influence of the loading program on the course of damage cumulation. 86 S. Mroziński 2. Tests description Low-cycle fatigue tests were carried out under constant-amplitude and pro- grammed loadings. Constant-amplitude loadings were applied at five levels of total strain (εac = 0.5, 0.6, 0.8, 1.0, 1.2%). The tests were performed un- der controlled strain (εac = const) according to the guidelines defined in the standard (ASTME606-92). Programmed loadingswere set in formof repeated blocks with an irregular sequence of steps. The block of programmed loading was obtained on the base of schematization of the loading with a random sequence of cycles (Fig.2a). Level εac [%] ni other 1 0.15 2 2 0.3 8 3 0.45 9 4 0.6 15 5 0.75 14 n0 =100 6 0.9 17 ζ =0.56 7 1.05 9 8 1.2 19 9 1.35 5 10 1.5 2 Fig. 2. Loading programs: (a) methodology of program elaboration, (b) loading programparameters Each cycle in the block of programmed loading (Fig.2b) was an oscilla- tory cycle (R = −1). The loading program was described with the value of maximum total strain amplitude εacmax and with the coefficient of spectrum density ζ. For the used loading program, the applied values of these parame- ters were: ζ = 0.56 and εacmax = 1.5%. The values of strain amplitudes εac and numbers of cycles on each step of the program are presented in Fig.2b. The specimens used in the fatigue tests were made of 30HGSA alloy steel ac- The influence of loading program... 87 cording to theASTMstandard [1]. The strength parameters of 30HGSA steel are: Rm = 1030MPa, E = 207000MPa, A5 = 9.5%. A general view to the applied specimen is presented in Fig.3 Fig. 3. The specimen used in the tests The fatigue tests were performedwith the use of the Instron 8501 strength machine. During the tests, a constant growth rate of relative strain for the measuring part equal to 1%/s was accepted. The controlling parameter both during theprogrammedandconstant-amplitude loadingwas the total strain of themeasuring part obtainedwith the use of an extensometer. During the tests under the constant-amplitude loading momentary values of the loading force and strain for chosen loading cycles were recorded. In the case of programmed loading, values of these parameters for the whole loading blocks (100 cycles) were recorded. 3. Test results 3.1. Constant-amplitude loading Momentary values of the loading force and strain recorded during the te- sts at respective strain εac levels were used for calculation of hysteresis loop parameters, i.e. stress amplitude σa, plastic strain amplitude εap and plastic strain energy ∆Wpl described with the hysteresis loop area. In Fig.4, exam- plary diagrams of changes of these parameters in function of the loading cycles number are shown. Basing on the courses of parameters σa, εap (Fig.4a and 4b), it was found that the steel applied in the tests undergoes cyclic weakening. Confirmation of that fact is gradual (in function of the number of cycles) decreasing the stress amplitude σa and increasing the plastic strain amplitudewith a constant level of the total strain amplitude εac. The weakening refers to all strain levels. 88 S. Mroziński Fig. 4. Hysteresis loop parameters for the constant-amplitude loading: (a) σa, (b) εap, (c) ∆Wpl The result of mutual stress and strain interactions which occur in the energy description is that in the case of specimenmade of 30HGSA steel this description not always reflects cyclic properties observedwith the use of stress or strain description. Basing on the analysis of the courses of energy changes ∆Wpl (Fig.4c), it can be found that cyclic properties of 30HGSA steel in the energy approach depend on the level of total strain. For the levels of εac = 0.5%, εac = 0.6% and εac = 0.8%, the energy ∆Wpl slightly increases with the number of loading cycles, which indicates cyclic weakening of the material. For the remaining two strain levels, the ∆Wpl energy decreases, which indicates slight hardening of the tested steel. Basing on the test results, a fatigue diagram in the bilogarithmic co- ordinates system: plastic strain energy ∆Wpl – number of cycles until fa- ilure N wasmade. The fatigue diagramwas approximatedwith a straight line described with the equation log∆Wpl =α log(N)+Kp (3.1) Thevalues of energy ∆Wpl obtainedatfive strain levels duringthe last realised cycle before fatigue failure were approximated with the diagram line. The influence of loading program... 89 3.2. Programmed loading Similarly like during the constant-amplitude loading, for the recorded suc- ceeding blocks of loading, values of the basic hysteresis loop parameters, i.e. stress amplitude σa, plastic strain amplitude εap and plastic strain energy ∆Wpl were defined. In Fig.5, examplary diagrams of stress σa changes in cho- sen blocks of the programmed loading realised in various periods of fatigue life are shown. Fig. 5. Changes of stress σa in the block of loading program Fig. 6. Changes of stress σa during programmed loading: (a) at the step with amplitude εac =0.6%, (b) at the step with amplitude εac =1.2% Basing on the analysis of the course of stress σa in the succeeding blocks of loading, it can be found that independently of the loading level, 30HGSA steel also undergoes cyclic weakening. This is proved by the decreasing stress σa on the same steps in succeeding block iterations of the loading program. In the paper, detailed analysis of the courses of σa, εap and ∆Wpl changes at indi- vidual steps of the realised programs was carried out. Because of the limited 90 S. Mroziński volume of this paper, the presentation of the obtained results is limited only to changes of the stress amplitude σa and two steps of the program (εac =0.6% and 1.2% – Fig.6). Analysis of diagrams presented in Fig.6 shows that changes of σa at in- dividual steps depend in little degree on the loading program. Changes of the strain amplitude from lower to higher ones and vice versa lead to momen- tary weakening of the material at the succeeding step and the occurence of a new level of momentary stress stabilization σas. This stress is lower than stress stabilization obtained at the given step in the former block of loading program. 4. Analysis of test results A comparative analysis of the course of changes of the basic hysteresis loop parameters under the constant-amplitude and programmed loading was car- ried out in function of the fatigue damage degree Di. The analysis referred to these levels of strain εac which were realised during the constant-amplitude and programmed tests (εac = 0.6% and εac = 1.2%). Values of the obtained plastic strain energy ∆Wpl(i) in succeeding cycles of the constant-amplitude and programmed loading enabled one to define from the equation of fatigue diagram (6), the corresponding numbers of cycles until specimen failure Ni, and then to carry out damage cumulation according to relations (1.1) and (1.2) for succeeding cycles of the constant-amplitude andprogrammed loading. The presented procedure during damage cumulation is explained in Fig.7. An Fig. 7. Procedure during fatigue damage cumulation exemplary course of energy changes at one strain level in function of the num- ber of cycles and the fatigue diagram in the energy approach are presented The influence of loading program... 91 there. The idea of the presented procedure is to connect the process of damage cumulation with the course of changes of cyclic properties. The obtained calculation results are presented in form of diagrams of hy- steresis loop parameters in function of the fatigue damage degree Di. Exam- plary diagrams of changes of the loop parameters obtained at two strain levels (εac =0.6% and 1.2%) were shown in Fig.8 and Fig.9. Fig. 8. Changes of σa (a) εap (b) and ∆Wpl (c) under the constant-amplitude and programmed loading at the strain level εac =0.6% Basing on the comparative analysis of diagrams presented in Figs.8 and 9, one can observe the qualitative and quantitative similarity in the course of analysed parameters under the constant-amplitude and programmed loading. Independently of the kind of loading, momentary values of the hysteresis lo- op parameters (σa, εap, ∆Wpl) for the same steps of fatigue damage Di are comparable. Thehysteresis loop parameters obtained in the terminal step cyc- les of the programmed loading with amplitudes εac = 0.6% and εac = 1.2% reach a level comparable to that observed under the constant-amplitude lo- ading for the same steps of damage. Moreover, it results from the diagrams that in spite of the disturbances in the stabilization process due to strain changes at successive steps, the steel 92 S. Mroziński Fig. 9. Changes of σa (a) εap (b) and ∆Wpl (c) under the constant-amplitude and programmed loading at the strain level εac =1.2% seems to remember the course of this process observed under the constant- amplitude loading. In the diagrams of changes of three parameters under the programmed loading, a trend of changes of cyclic properties is very clearly visible. It is similar to the course of changes of cyclic properties under the constant-amplitude loading. The above observation has capital practical con- sequences since it shows the possibility of predicting the course of changes of cyclic properties of a material under operating loading on the basis of known course under the constant-amplitude loading.Theabovementioned conclusion confirms the test results presented byMroziński (2008). Basing on the determined diagrams, it can also be found that the degree of fatigue damage for the moment of failure in little degree depends on the loading program. The values of the total damage Di close to one for both loading programs (constant-amplitude and programmed) are a proof of good agreement between the fatigue life found in calculations and experimentally. The influence of loading program... 93 5. Conclusions The carried out analysis of the obtained experimental results allows one to formulate the following conclusions: • Both under the programmed block loading and constant-amplitude lo- ading of the specimens made of the alloy steel no period of cyclic pro- perties stabilizationwas observed.Having considered this problem, some doubts concerning the results of fatigue life calculations based uppon the constant material data determined during the tests under constant- amplitude loading have arisen. • The process of cyclic weakening which proceeds under a constant- amplitude and programmed loading of 30HGSA steel analysed with the use of such hysteresis loop parameters as σa, εap and ∆Wpl showed qu- alitative similarity concerning the nature of changes of cyclic properties and quantitative similarity concerningmomentary values of thementio- ned parameters for the same degree of fatigue damage. • Connection of the changes of cyclic properties with the process of da- mage cumulation may lead to improvement of the agreement between analytical and experimental results. Such an approach to the problem of fatigue life calculation can be of special importance in the case of fatigue life determination of construction elementsmade ofmaterials characteri- sed with the lack of stabilization period (aluminium and copper alloys). Realisation of such an approach in fatigue life calculations is possible if the courses of cyclic properties changes in function of the fatigue dama- ge degree are known. The proposal of a method of determination of the course of cyclic properties changes under a constant-amplitude loading was presented byMroziński (2008). Acknowledgement This paper was realized in the framework of research grantNo. N50104031/2563 funded by Ministry of Science and Higher Education in the years 2006-2009. References 1. ASTME606-92: Standard Practice for Strain-Controlled Fatigue Testing 2. Boroński D., Mroziński S., 2007, Metal tests in conditions of controlled strain energy density, Journal of Theoretical and Applied Mechanics, 45,4, 773-784 94 S. Mroziński 3. Duyi Y., Zhenlin W., 2001, A new approach to low cycle fatigue damage based on exhaustion of static toughness and dissipation of cyclic plastic strain energy during fatigue, International Journal of Fatigue, 23, 679-687 4. FatemiA.,YangL., 1998,Cumulative fatiguedamageand life prediction the- ories: a survey of the state of the art for homogeneousmaterials, International Journal of Fatigue, 20, 1, 9-34 5. GołośK.,EllyinF., 1988,Atotal strainenergydensity theory for cumulative fatigue damage,ASME, Journal of Pressure Vessel Technology, 110, 36-41 6. Kasprzyczak L., Macha E., 2006, Selection of the PID controller structure for control of stress, strain and energy parameter at the hydraulic fatigue test stand, II International Conference Mechatronic Systems and Materials, 1-10 7. Koh S.K., 2002, Fatigue damage evaluation of high pressure tube steel using cyclic strain energy density, International Journal of Pressure Vessels and Pi- ping, 79, 791-798 8. Kujawski D., Ellyin F., 1984, A cumulative damage theory of fatigue crack initiation and propagation, International Journal of Fatigue, 6, 2, 83-88 9. Leis B.N., 1988, A nonlinear history-dependent damage model for low cycle fatigue, In: Low Cycle Fatigue, ASTM STP 942, H.D. Solomon, G.R. Halford, L.R.KaisandandB.N.Leis (Edit.),AmericanSociety forTesting andMaterials, Philadelphia, PA, 143-159 10. Li D.M., Kim K.W., Lee C.S., 1997, Low cycle fatigue data evaluation for a high-strength spring steel, International Journal of Fatigue, 19, 8/9, 607-612 11. Manson S.S., Halford G.R., 1986, Re-examination of cumulative fatigue damage analysis – an engineering perspective,Engineering FractureMechanics, 25, 5/6, 539-571 12. MinerM.A., 1945,Cumulative damage in fatigue,Transactions of the Ameri- can Society of Mechanicals Engineers, Journal of Applied Mechanics, 67, 159- 164 13. Mroziński S., 2008, Stabilizacja własności cyklicznych metali i jej wpływ na trwałość zmęczeniową,WydawnictwaUczelnianeUniwersytetuTechnologiczno- Przyrodniczego w Bydgoszczy, RozprawyNr 128 [in Polish] 14. Mroziński S., Topoliński T., 1999, New energy model of fatigue damage accumulationand its verification for 45-steel,Journal ofTheoretical andApplied Mechanics, 37, 2, 223-239 15. PalmgrenA., 1924,Die Lebensdauer vonKugellagem, VerfahrenstechnikBer- lin, 68, 339-341 [in German] 16. Szala J., 1998,Hipotezy sumowania uszkodzeń zmęczeniowych, Wydawnictwa Uczelniane ATRwBydgoszczy [in Polish] The influence of loading program... 95 Wpływ programu obciążenia na przebieg kumulacji uszkodzeń zmęczeniowych Streszczenie Wpracyprzedstawionowyniki analizyporównawczejwłaściwości cyklicznychpró- bek ze stali 30HGSAwwarunkachobciążenia stałoamplitudowego i programowanego. Analizę prowadzonozwykorzystaniemcharakterystycznychparametrówpętli histere- zy w funkcji stopnia uszkodzenia zmęczeniowego. Przeprowadzona analiza wykazała, że przebiegi zmianwłaściwości cyklicznychnawybranychpoziomachodkształcenia są bardzo podobne i nie zależą od programu obciążenia.W pracywykazano, że podczas sumowania uszkodzeń istnieje możliwość uwzględniania zmian własności cyklicznych. Manuscript received February 24, 2010; accepted for print June 7, 2010