Influence of the Constitutive Law in the Superplastic Forming of In718

O. Gonzalo1, A. Mugarra2, K. Ostolaza2, and J.L. Alcaraz1

Dpt. of Mechanical Engineering, University of the Basque Country, 48013-Bilbao (SPAIN) Dpt. of Materials, Industria de Turbo Propulsores I.T.P. , S.A., 48170-Zamudio, Vizcaya (SPAIN)


Abstract: Superplastic forming of In718 alloy sheets allows the manufacturing of geometrically complex shape components for the aircraft industry. This work presents some improvement in the material simulation of the forming process, at a stage prior to the implementation on a production scale. The constitutive law is considered in detail, as well as its repercussion in the results stemming from the numerical analysis. The material model is obtained based on several lab scale tests and is introduced in ABAQUS through user subroutines. The maximum strain and thickness values are compared with the experimental ones at the same strain rate conditions. After the material model validation process, better adjustment is derived between the simulation results and the experimental tests.

1. Introduction
Superplastic forming is a recently developed technique applied to obtaining geometrically complex parts in a simple way and at relatively low cost (see for example Sherby, 1985). On the other hand, the finite element analysis makes it possible to derive the most adequate geometrical and material parameters for this process. In this work an Inconel 718 alloy is selected based on its high mechanical strength and corrosion resistant properties at high temperature (Huang, 2000; Ceschini, 1994). This alloy can be used in the hottest parts of aircraft engines and in pipelines with gas at high temperature. The paper is focused on the finite element model of thin sheets superplastic forming (Bonet, 1988; Xing, 1997). The aim is to improve the fitting of numerical results with lab-scale tests by means of a better material simulation. This has been accomplished by a feedback from several lab-scale results applied to the same alloy into the ABAQUS input.

2. Analysis
Some simulation analyses previously performed for Inconel 718 (Mugarra, 2000) will be the basis of the present analysis centered in the alloy material modeling. The die for the tests is shown in Figure 1. Results from the press machine provide the pressure-time variation during the test and the geometry of the formed sheet. These results can be used to readjust the initial geometrical parameters introduced in the simulation. Thus other more realistic simulations will then be carried out. 2002 ABAQUS Users’ Conference 1

On increasing the strain level a material softening can be deduced from a typical simulation. Furthermore the material softening is derived from changes in parameter n. Actually the material is quite sensitive to the grain size growth. so small values for C as 10-25 are derived. However. Reference pressure in the loading process (*DLOAD option).The following parameters are taken into account in order to improve the simulations: a. Results are compared at the lowest point of the deformed dome. These parameters have been made variable. Contact pressure and clearance (*SURFACE c. [ 1 b b +1 ] (2) There is obviously a relation between the material parameters in the two expressions. as the superplastic behavior is based on the grain size fineness. The constitutive equation used in the analysis is the following: σ = Kε n ε m A quite noticeable fluctuation in parameter m is obtained for considerable strain level variations. Parameters in the material constitutive law. e. This feature gives rise to a higher pressure obtained in ABAQUS than that from the actual process. Residual values for convergence (*CONTROLS option). The behavior just described has motivated the task of attaining a better fitting of the material parameters included in the constitutive equation.f in the first analysis and creep. Sheet cracks can therefore be produced during the forming operation because the higher pressure implies a higher strain rate and the stress levels increase accordingly so as to cause fracture in some cases. f. Two types of material behavior analyses have been carried out: 1. 2 2002 ABAQUS Users’ Conference . The second analysis is motivated by the scattered results in the material parameters obtained from Equation 2. from Equation 1 the parameters fluctuate less. Material parameter variations are introduced by means of user subroutines in FORTRAN (usdfld. By using the ABAQUS expression: (1) ε = Cσa ((b + 1)ε ) 2. Friction coefficient (*FRICTION option) d. For example.f in the second). These subroutines are briefly commented below. By Equation 1. b. CETOL BEHAVIOR option) parameter and allowed time in the analysis. Both variations will be implemented by user subroutines in ABAQUS.

the *CREEP option is used based on the parameter values provided by USD.f subroutine This subroutine allows to introduce a variable that assumes user defined values. the following values result: 2002 ABAQUS Users’ Conference 3 . The following three parameters are required by the subroutine: ∂∆ε ∂∆ε DECRA (5) = ∂ε ∂σ They are obtained from the creep strain increment. 3) Once the constitutive law and the strain rate have been adjusted. Two tests have been carried out. Once the user defined variable (USD) has been calculated. reversed cycles are performed to select the reference pressure parameter. This parameter will be very useful to derive the pressure-time variations in more complex geometry components.2. The material parameters are included in a tabulated form at different strains for the uniaxial test. in our case-. Consequently the equivalent creep strain is associated with the uniaxial test strain to obtain the material parameters. 2.f subroutine This subroutine allows to create original creep laws (although special care should be taken to avoid numerical problems). tabulated parameter values are taken as a function of some variable –the equivalent creep strain. By measuring the sheet thickness at the highest strain point and assuming a constant strain rate. The strain rate for the analysis and a comparable geometry of the deformed sheet are thus obtained. ABAQUS performs this task automatically.2 Creep. Moreover the tabulated parameters are introduced in an external data file (not in the input or in the subroutine). which can be expressed as: DECRA (1) = ∆ε DECRA (2) = (3)  1 m  1 m ∆ε = ε∆t = σ     ∆t  K ε 1 m 1 n (4) 3. Material parameters will change according to this creep strain. As in the previous one. 3. Model validation For the validation of the material model the following steps will be taken: 1) Estimation of the actual strain rate of the test. Equation 1 has been defined by using this subroutine. The values obtained within the analysis can be used for this task.1 Strain rate estimation.1 Usdfld. 2) Direct simulations with ABAQUS assuming the pressure variations applied to the press and the new data for the constitutive law from experimental results. In our case the equivalent creep strain has been selected as the user defined variable.

different pressure-time cycles will result. This behavior can mainly be attributed to the influence of the strain rate in the material parameters (although it may also affect the availability of parameter data only up to a maximum strain of 0. Again the sheet conforms more than expected. The difference between the actual cycle and the analysis is motivated by the different strain rate of the process.7⋅10-4 s-1 versus 1. Later verification will be performed for the two tests. and a strain increase results. Adjusting this parameter. 1984). The applied cycle is plotted in Figure 3.a) Test I: ε = 1. as deduced from Figure 2 (Chandler.1 MPa. the material will show a higher flow stress than that defined in the constitutive law for 1⋅10-4 s-1.8). It is observed that the sheet has been conformed more than expected. The most relevant parameter is the reference pressure in the *VISCO option. If the reference pressure changes. at the strain rates originated by the applied pressure. but the sheet geometry seems to be unaffected.3 Inverse analyses Once the material constitutive law has been adjusted according to the geometry obtained from the tests. For this cycle. The pressure-time cycle is adjusted to the actual one by using the reference pressure that provides a similar maximum pressure. 3. On the other hand the value of the maximum strain rate lies above that previously estimated (3.2 Direct simulations of the tests A temperature of 950ºC and strain rates in the range of 10-4 s-1 are assumed. direct simulations provide higher strains than those obtained in the tests.17 ⋅ 10 s 3. with a maximum in the strain rate above the one estimated (6. the Von Mises stress.19 ⋅ 10 s −4 −1 −4 −1 b) Test II: ε = 3. will lead to controlled pressure-time variations. Consequently a softener material definition is being used. sheet thickness and strain rate contours are shown in Figure 4. Higher pressures will then be applied and as a result the sheet will conform more than obtained in the test. The Von Mises stress. equivalent creep strain. Test II has been selected to do the adjustment. b) Test II: For this test the estimated strain rate is about three times that of the tension test. 4 2002 ABAQUS Users’ Conference . the strain levels and the thickness contours are quite similar.6⋅10-4 s-1 versus 3. the next step will be to select ABAQUS parameters so as to obtain pressure cycles similar to those applied in the superplastic forming tests. The higher strains are explained because the flow stress increases at higher strain rates. This value is 1.17⋅10-4 s-1). sheet thickness and strain rate contours are obtained and shown in Figure 6. along with the initial amplitude in *AMPLITUDE. a) Test I: In this test conditions were approximately those used in the tension test. In conclusion. The strain rate in the superplastic forming tests differs from the tension test rate but the flow stress and parameter m do not change significantly within the 1⋅10-4 to 3⋅10-4 s-1 range. Comparing different simulations. equivalent creep strain. In other words. The applied cycle is plotted in Figure 5.19⋅10-4 s-1).

P. vol. To attain this. vol.. b) Test II: The target strain rate is 3⋅10-4 s-1 and the reference pressure 1. Results are shown in Figure 7. Bonet. 4. material parameter variation has been allowed by implementing user subroutines in ABAQUS. “Finite Element Modeling of the Superplastic Forming of Thin Sheets”. 21. 2000. A. 1.99 (versus 1. Strangwood. in Metal Forming.13 (versus 1. vol. n. as the former was selected for the adjustment. and Mahoney. Failure is obtained in the test because the pressure value is too high. “Superplastic Forming and Diffusion Bonding of INCO 718”.D. G. “Superplastic behaviour of Inconel 718 sheet”.M. the maximum strain is 1.. 1988. 291-295. “Finite element results of sheet superplastic forming”. A very good agreement is thus obtained. edited by M.27 mm (versus 0. 1994. Journal Spacecraft. “Superplastic Behaviour of Fine-Grained In-718 Superalloy”.24 mm (versus 0. pp. Results for the geometry.L. and Afrikantov. 2002 ABAQUS Users’ Conference 5 . M.C. Chandler. Materials Science Forums.22 mm in the test).K. In Figure 10 the two pressure-time cycles are compared. The dome height is about 88 mm (versus 94 mm obtained in the test). In the actual cycle a time of 1120 s has elapsed without pressure. The maximum pressure is similar.T. Wood. As commented before the material softening at high strains can be the main reason for this behavior. pp.1 MPa.20 in the test). F. Materials Science and Technology. in Superplasticity and Superplastic Forming. P. A. References 1.20 in the test) and the final thickness is 0. n. A material constitutive law adjusted to actual results from several experimental tests has been obtained. R. The difference between the obtained contours and the original ones can be explained by noting the different pressure-time cycles applied (see Figure 8).. Huang. Hamilton and N. 1984. A. Pietrzyk et al.1 MPa.L. 4. 351-356. the reference pressure has played an important role in the validation of the model. 2. pp. edited by C..E. W. and the final thickness in the lowest point is 0. Ghosh.H.. Ceschini. J. Mugarra. 2000. On the other hand. 16. Conclusion In this paper the superplastic forming of Inconel 718 alloy sheets has been presented. The dome reaches a height of 95 mm (versus 91 mm in the test).Next these results are verified in the two tests: a) Test I: The target strain rate is 1⋅10-4 s-1 and the reference pressure 1. and Blackwell. G. Persiani. 11. 3. O. W. Cammarota. The pressure fall is due to the material softening in the calculated cycle.22 mm in the test). the maximum strain in the sheet is 0. and Alcaraz. Paton.. J. 5. Y. Garagnini. 707-712. thickness and strain level provide a verification of the improvement obtained after the material adjustment.L.. 170-172. L. and Zienkiewicz. This test gives rise to better results than Test I (see Figure 9).

Sherby. 1-7. J. 1985.R. 1.D.6. vol. 7. and Wadsworth.L. Die model for the tests. O. pp. “Finite-element analysis and design of thin sheet superplastic forming”. pp. 925-936. 68. Flow stress and parameter m as a function of the strain rate. v. 6 2002 ABAQUS Users’ Conference . Z. Figures Figure 1. Materials Science and Technology. Figure 2. H. “Superplasticity and superplastic forming processes”. Xing. Journal of Materials Processing Technology. 1997. and Wang.

0.7 0.1 0 0 2000 4000 6000 Time (sec) 8000 10000 12000 Figure 3.5 0.8 0.3 0.4 0. Pressure-time variation for the press machine in Test I. a) b) 2002 ABAQUS Users’ Conference 7 .6 0.9 0.2 0.

b) Equivalent creep strain.c) d) Figure 4. Results for Test I: a) Von Mises stress. c) Sheet thickness. d) Strain rate. 8 2002 ABAQUS Users’ Conference .

2 0 0 500 1000 1500 2000 Time (sec) 2500 3000 3500 4000 Figure 5. Pressure-time variation for the press machine in Test II. a) b) 2002 ABAQUS Users’ Conference 9 .6 0.4 0.1.8 0.2 1 0.

b) Equivalent creep strain. d) Strain rate. c) Sheet thickness. 10 2002 ABAQUS Users’ Conference . Results for Test II: a) Von Mises stress.c) d) Figure 6.

a) b) 2002 ABAQUS Users’ Conference 11 .

4 0.3 0. Results after the material law adjustment for Test I: a) Von Mises stress.8 0. Comparison between two pressure-time cycles. Test I.5 0.6 0. Press.0 Ref.: 1.7 0. Pressure (MPa) 0.c) Figure 7. c) Sheet thickness. b) Equivalent creep strain.2 0.1 Original Cycle 0 200 400 600 800 Time (s) 1000 1200 Figure 8. 12 2002 ABAQUS Users’ Conference .1 0.

a) b) 2002 ABAQUS Users’ Conference 13 .

2 0. Test II.4 0.7 0.0 Ref . Press.8 0. 14 2002 ABAQUS Users’ Conference .0 0.9 0. b) Equivalent creep strain.1 Original Cycle 0 100 200 300 Time (s) 400 500 Figure 10. Results after the material law adjustment for Test II: a) Von Mises stress. Pressure (MPa) 1.c) Figure 9. Comparison between two pressure-time cycles.3 0.6 0.1 0.: 1.5 0. c) Sheet thickness.