Journal of Mechanical Science and Technology 25 (5) (2011) 1159~1165

www.springerlink.com/content/1738-494x

DOI 10.1007/s12206-011-0221-6

Study on determination of durability analysis process and fatigue damage parameter for rubber component
Seong-In Moon1,*, Il-Je Cho1, Chang-Su Woo2 and Wan-Doo Kim2
1

Korea Atomic Energy Research Institute, 1045 Daedeok-daero, Yuseong-gu, Daejeon-si, 305-353, Korea 2 Korea Institute of Machinery and Materials, 104 Sinseong-ro, Yuseong-gu, Daejeon-si, 305-343, Korea (Manuscript Received May 31, 2010; Revised August 5, 2010; Accepted January 6, 2011)

----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Abstract
Rubber components, which have been widely used in the automotive industry as anti-vibration components for many years, are subjected to fluctuating loads, often failing due to the nucleation and growth of defects or cracks. To prevent such failures, it is necessary to understand the fatigue failure mechanism for rubber materials and to evaluate the fatigue life for rubber components. The objective of this study is to develop a durability analysis process for vulcanized rubber components, that can predict fatigue life at the initial product design step. The determination method of nonlinear material constants for FE analysis was proposed. Also, to investigate the applicability of the commonly used damage parameters, fatigue tests and corresponding finite element analyses were carried out and normal and shear strain was proposed as the fatigue damage parameter for rubber components. Fatigue analysis for automotive rubber components was performed and the durability analysis process was reviewed.
Keywords: Fatigue life; Fatigue damage parameter; Rubber component; Suspension bush ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

1. Introduction
Rubbers ability to withstand very large strain without permanent deformation or fracture makes it ideal for many applications, including tires, vibration isolators, seals, hoses, belts, impact bumpers, medical devices and structural bearings to name a few [1, 2]. These rubber components subjected to fluctuating loads often fail due to the nucleation and growth of defects or cracks. To prevent such failures, it is necessary to understand the fatigue failure mechanism for rubber materials and to evaluate the fatigue life for rubber components. For these reasons, not only the rubber component manufacturers but also their customers like automotive makers perform a series of strict fatigue test on the components such as component fatigue tests and driving fatigue tests. Currently, designers rely on their own trial-error based experiences for the fatigue design. Thus, those designs depending on only experience may result in disqualification from the fatigue test during final product evaluation. Those fatigue failures of any new designs are prohibitive for automotive manufacturers. To avoid this problem, many researchers [3-5] are focusing on evaluation of fatigue life using CAE techniques that could
This paper was recommended for publication in revised form by Associate Editor Chongdu Cho Corresponding author. Tel.: +82 42 868 8485 E-mail address: simoon21c@kaeri.re.kr KSME & Springer 2011
*

supplement drawbacks of evaluation through tests and could significantly reduce time for fatigue-proof design. However, there are still some problems. First, the rubber materials show particular mechanical properties according to compounding ingredients and manufacturing conditions [6, 7]. Therefore, to evaluate the fatigue life of designed rubber components, the material properties of components should be obtained. It is practically impossible to measure the material properties for the whole component. Second, some parameters like stress, strain, SED (strain energy density) and so on are generally used to estimate fatigue life of rubber components [8-10], but the question remains how we should use these parameters to estimate the component life and what their limitations of the parameters are. The objective of this paper is to develop a durability analysis process for vulcanized rubber components that is applicable to predicting fatigue life at initial product design stage. Fig. 1 is a schematic diagram showing the fatigue life evaluation procedure. The fatigue damage parameters such as principal strain, SED and so on, which are calculated by FEA (finite element analysis), and the fatigue properties of the rubber material are used in order to estimate the fatigue life of rubber components. This paper proposes a methodology to extract the material properties for finite element analysis input data from basic test results. Fatigue tests and corresponding finite element analyses were carried out to investigate the applicability

1160

S.-I. Moon et al. / Journal of Mechanical Science and Technology 25 (5) (2011) 1159~1165

Table 1. Basic material properties of suspension bush. Shore Hardness 67.8 IRHD Hardness 70.5 Ultimate Tensile Strength (MPa) 26.3 Elongation (%) 436

Table 2. Non-linear material constants of suspension bush. Strain Range 25% 50% 100% C10 (MPa) 0.901 0.781 0.727 C01 (MPa) 0 0 0.001

Fig. 1. Fatigue life evaluation procedure.

(a)

(b)

(c) (a)

Fig. 2. Nonlinear material constant tests: (a) simple tension; (b) equibiaxial; (c) pure shear.

of commonly used fatigue damage parameters, and optimum fatigue damage parameter was selected. Also, a fatigue analysis for automotive rubber components was performed and the durability analysis process was reviewed.

2. Measurement of material property and fatigue test of component

2.1 Mechanical material property The basic material properties of the rubber component, a suspension bush, were measured through hardness tests and simple tension tests. Hardness was measured in Shore A hardness and IRHD (International Rubber Hardness Degree). For the mechanical tests, UTM (Universal Test Machine) was used at a speed of 100 mm/min, and deflection was measured using a laser extensometer in Fig. 2. Table 1 shows the measured basic material properties of the suspension bush. 2.2 Non-linear material constant for hyper-elastic material For general elastic materials, spring-back is shown within the elastic limit but rubber materials are characterized by hyper-elasticity which is valid for materials that exhibit elastic response up to large strain. Hyper-elastic materials are described in terms of a strain energy potential W, which defines the strain energy stored in the material per unit of reference volume as a function of the strain. In this paper, the MooneyRivlin form expressed with strain invariant was used to characterize the mechanical behaviors and the strain energy potential can be expressed as the following Eq. (1): W = C10 ( I1 3) + C01 ( I 2 3) , (1)
(c) Fig. 3. Stress-strain curves: (a) simple tension; (b) equibiaxial; (c) pure shear. (b)

where I1 and I2 are the first and second strain invariant of the deviatoric component of the left Cauchy-Green deformation tensor, respectively, and C10 and C01 are material constants empirically determined by curve fitting the stress-strain relationship to simple tension test, equibiaxial tension test and pure shear test results. Three basic tests for strain states, simple tension, equibiax-

S.-I. Moon et al. / Journal of Mechanical Science and Technology 25 (5) (2011) 1159~1165

1161

Z Y X

ial tension and pure shear test, were carried out and the strain levels were progressively increased up to the maximum value, 1.0. Fig. 3 shows the stress-strain curves for each stress state and equibiaxial tension, pure shear and simple tension arrange in order of stiffness magnitude. The non-linear material constants were determined through curve fitting for these stressstrain curves and presented in Table 2. 2.3 Fatigue material property The three-dimensional (3D) dumbbell specimens shown in Fig. 4(a) were used for fatigue tests. To determine a fatigue damage parameter of the rubber material and fatigue life, fatigue tests using 3D dumbbell specimens were conducted in ambient temperature under the displacement controlled condition with a sine waveform of 3.0 Hz. The mean displacement varies from 0 mm to 5 mm and the amplitude displacement ranges from 3 mm to 12 mm. With increase of cycles in the fatigue test, the maximum load decreased little by little. When the crack grew over the critical size, the maximum load suddenly decreased, and the specimen was finally fractured. In this paper, we assumed that fatigue failure occurs when the maximum load drops up to 20% level of initial maximum load. Fig. 5 shows the relationship between the mean displacement and the fatigue life for 3D dumbbell specimens. The fatigue life was reduced as the mean displacement and amplitude displacement increased. To determine the relationship between fatigue damage parameters and the fatigue life, the deformation behavior of the dumbbell specimen was estimated by 3D finite element analysis and ABAQUS Ver. 6.8 [11] package was used for the finite element analysis. Fig. 4(b) shows a finite element mesh of the 3D dumbbell specimen. An eighth of the specimen was modeled by considering the symmetry condition and the mesh was constructed with 8node linear brick, hybrid elements. The number of nodes and elements of the model is 20,157 and 4,340, respectively. The nodes on the lower surface of the model were constrained, and those on the upper surface were controlled with displacement according to the loading condition. It was assumed that fatigue life was determined by the maximum value of fatigue damage parameters [3]. From a series of finite element analysis results and test results, the relationship between the fatigue life and the typical fatigue damage parameters, principal stain and SED, was obtained and presented in Fig. 6. 2.4 Characteristic and durability tests for rubber component To verify the validity of finite element analyses and durability analyses, static stiffness and fatigue life were measured by using a suspension bush for vehicles and the results presented in section 3. Fig. 7(b) shows the suspension bush being used in the tests. The static stiffness was measured in two directions, void direction (P-direction) and bridge direction (Q-direction), and a durability test was performed under the condition presented in Table 3. For the durability tests, the outer pipe of the

(a)

(b)

Fig. 4. Fatigue specimen: (a) dumbbell specimen; (b) FE model.

Fig. 5. Displacement-fatigue life curve.

Log(SED)

Mean Displacement, mm

(a)

Log(Strain)

(b) Fig. 6. Damage parameter-fatigue life curves: (a) strain energy density; (b) 1st principal strain.

1162

S.-I. Moon et al. / Journal of Mechanical Science and Technology 25 (5) (2011) 1159~1165

Table 3. Loading condition for fatigue test. Direction Radial P Torsion Load P5.5P T1.2T Frequency (Hz) 3.3 3.3

Table 4. Basic material properties searched from RubPRO. Material Name NR60-01 NR60-02 NR65-01 NR65-02 NR65-03 NR70-01 NR70-02 IRHD Hardness 66 65 66 72 65 75 70 Tensile Strength (MPa) 21.2 20.9 20.8 19.7 24.9 26.3 24.0 Elongation (%) 400 456 460 412 471 452 495

(a)

(b)

Fig. 7. Suspension bush specimen: (a) loading condition; (b) tested specimen.

bush was constrained, and the inner pipe was controlled by loading.

3. Development of durability analysis procedure

In this section, the methodology to extract the material properties for FEA from the rubber material database of Korea Institute of Machinery and Materials, RubPRO, and basic test results and then optimum fatigue damage parameter was determined. Also, a fatigue analysis for automotive rubber components was performed and the durability analysis process was reviewed. 3.1 Determination of non-linear material constant for hyperelastic material To perform finite element analyses for newly designed rubber components, the non-linear material constants have to be conformed in advance because the rubber materials show different mechanical behaviors according to compounding ingredients and manufacturing conditions. However, it is practically impossible because there is no standard for compounding condition. Therefore, a methodology to determine the nonlinear material constants for finite element analyses from RubPRO was proposed by using the basic mechanical properties. A total of 39 data set was searched from RubPRO on condition of 60 ~ 75 IRHD similar to the hardness of the suspension bush, 70 IRHD. Among these data, seven sets of rubber material were selected on the basis of similarity of ultimate tensile strength and elongation and these were presented in Table 4. It can be known that NR70-01 is the most similar material among these materials, and the ultimate tensile strength and elongation is 26.3 MPa and 452%, respectively. The finite element analyses for the suspension bush were performed by using the non-linear material constants presented in Table 4, and then the load-displacement curves were produced. Fig. 8 illustrates that NR70-01 material presents the
(a)

(b)

(c) Fig. 8. Load-displacement curves of dumbbell specimen: (a) 25% strain range; (b) 50% strain range; (c) 100% strain range.

most similar mechanical behavior with suspension bush material with respect to load-displacement curve.

S.-I. Moon et al. / Journal of Mechanical Science and Technology 25 (5) (2011) 1159~1165

1163

Z Y X

Fig. 9. Finite element mesh of suspension bush.

(a)

(b)

Fig. 11. Contour plots of SED and principal strain: (a) SED; (b) 1st principal strain.

1 2
0.72

1 2 0.56

0.81
(a) (a)

0.43
(b)

0.64

0.68

Fig. 12. Mohrs circle for strain at (a) MAX. SED point; (b) MAX. strain point.

experimental results within maximum difference of 15%. 3.3 Determination of fatigue damage parameter For fatigue life evaluation, not only crack nucleation but also crack growth is considered. However, in the case of rubber components in vehicles, crack initiation is defined as failure because the mechanical characteristic of rubber can be changed, and the estimation for crack nucleation was focused on in this study. The two widely used fatigue damage parameters for crack nucleation prediction in rubber components are maximum principal strain (or stretch) and SED. Stress has rarely been used as a fatigue damage parameter in rubber components [8, 9]. The contour plots of strain energy density and first principal strain at maximum load are shown in Fig. 11. In this figure, we can see that the maximum points of the damage parameters are different from one another. The maximum point of SED was identical to crack nucleation point observed from the fatigue life tests for the suspension bush. Consequently, it is thought that SED as fatigue damage parameter can be used for predicting the crack nucleation point. Fig. 12 shows the Mohrs circles for strain at the maximum points of SED and first principal strain, respectively. In this figure, it is known that the maximum shear strain appears at the maximum SED point but different from the maximum principal strain point. Hence, it was thought that to only use the first principal strain is not sufficient to estimate the failure of rubber components in case where shear deformation pre-

(b) Fig. 10. Load-displacement curves of suspension bush: (a) P-direction; (b) Q-direction.

3.2 Finite element analysis The deformation behavior of the suspension bush was estimated by using 3D finite element analysis and ABAQUS Ver. 6.8 explicit code was used for the analysis. Fig. 9 shows the finite element mesh for the suspension bush constructed with 10-node quadratic tetrahedron elements (C3D10M) without inner and outer pipe of bush. The number of nodes and elements of the model is 152,601 and 101,707, respectively. The nodes on the inner pipe were constrained, and those on the outer pipe were controlled by loading. It was assumed that the initial ratio of bulk elastic modulus to shear elastic modulus, K0/0, was 100. Fig. 10 depicts the load-displacement curves for the suspension bush. The estimated results show a good agreement with

1164

S.-I. Moon et al. / Journal of Mechanical Science and Technology 25 (5) (2011) 1159~1165

dominantly influences the failure of rubber components. 3.4 Fatigue life prediction for rubber component The fatigue life was calculated by using SED and first principal strain obtained from finite element analyses. The fatigue properties measured by using 3D dumbbell specimens were used, and the relationship between fatigue life and fatigue damage parameters, SED and first principal strain, was presented as follows:
N f = 190,546 ( SED )
2.063 4.521

(4) To estimate more accurate fatigue life, not only normal strain but also shear strain should be considered, and fatigue properties for shear deformation are needed.

References
[1] Korea Institute of Machinery and Materials, Development of Integrated Design System for Mechanical Rubber Components, M1-9911-00-0014 (2004). [2] A. N. Gent, Engineering with Rubber, Hanser Gardner (2001). [3] C. S. Woo, W. D. Kim and J. D. Kwon, A Study on the Fatigue Life Prediction and Evaluation of Rubber Components for Automobile Vehicle, Transaction of KSME, 13 (6) (2005) 56-62. [4] C. H. Kim, K. J. Kim, H. T. Jeong, C. W. Kim, I. S. Sohn and J. B. Kim, Prediction of Durability, Static and Dynamic Properties on Rubber, Transaction of KSME, 14 (6) (2006) 17-23. [5] C. S. Woo, W. D. Kim and J. D. Kwon A Study on the Material Properties and Fatigue Life Prediction of Natural Rubber Component, Material Science and Engineering (2008) 367-381. [6] J. H. Kim and H. Y. Jeong, A Study on the Material Properties and Fatigue Life of Natural Rubber with Different Carbon Blacks, Internal Journal of Fatigue, 27 (2005) 263-272. [7] B. J. Roberts and J. B. Benzies, The Relationship between Uniaxial and Equibiaxial Fatigue in Gum and Carbon black filled Vulcanizates, Proceedings of Rubbercon, 2.1 (1997) 2.1-2.13. [8] W. V. Mars and A. Fatemi, A Literature Survey on Fatigue Analysis Approaches for Rubber, International Journal of Fatigue, 24 (2002) 949-961. [9] GM, Standard Test Methods for Vulcanized Rubber and TPE for Use in finite Element Analysis Modeling. [10] N. Andre, G. Cailletaud and R. Piques, Haigh Diagram for Fatigue Crack Initiation Prediction of Natural Rubber Components, Kautschuk Und Gummi dunstoffe, 52 (1999) 120123. [11] Dassault Systems, ABAQUS Version 6.8 Users manual (2008).

(2) (3)

N f = 21,314 ( Strain )

where Nf is fatigue life. The fatigue life calculated by using SED and first principal strain was underestimated, and the fatigue life normalized with fatigue test results was 0.33 and 0.52, respectively. As stated above, it is thought that these underestimated results are caused by shear deformation. SED is a scalar quantity, so it does not predict crack nucleation in a specific orientation. Also, equibiaxial tension fatigue life was longer than simple tension fatigue life by a factor of approximately 4, when compared based on equal strain energy density [8]. Therefore, it is not appropriate to evaluate the fatigue life of rubber components by using SED. Also, if normal deformation is dominant in failure mode, the fatigue life can be estimated by using maximum principal strain as fatigue damage parameter [3, 5]. Thus, for the purpose of more accurate fatigue life, both normal strain and shear strain should be considered. Moreover, fatigue properties for shear deformation are needed as well.

4. Conclusions
To develop a durability analysis process for vulcanized rubber components, that is applicable to predicting fatigue life at initial product design stage, a methodology is proposed to extract the material properties for FEA from basic test results. Also, to investigate the applicability of commonly used fatigue damage parameters, fatigue tests and corresponding finite element analyses were carried out and optimum fatigue damage parameter was selected. The key findings and results are as follows: (1) A methodology to extract the non-linear material properties for FEA from the rubber material database of Korea Institute of Machinery and Materials, RubPRO, and basic test results was proposed. (2) Strain energy density as a fatigue damage parameter accurately predicted the crack initiation point, but the predicted fatigue life showed a big discrepancy of 77% with test results. (3) First principal strain as a fatigue damage parameter did not predict the crack initiation point and the fatigue life. The predicted fatigue life showed a big discrepancy of 52% with test results.

Seong In Moon received his B.S., M.S. and Ph. D. in Mechanical Design from Sungkyunkwan University in 1999, 2001 and 2005 in Korea. He worked for Hyundai Motors as a senior researcher from 2005 to 2009. He is currently a senior researcher in Korea Atomic Energy Research Institute. His research interests are durability evaluation of mechanical system and nuclear fuel cycle facility design.

S.-I. Moon et al. / Journal of Mechanical Science and Technology 25 (5) (2011) 1159~1165

1165

Il Je Cho received his B.S. and M.S. in Nuclear Engineering from Hanyang University in 1996 and 1998 in Korea. He is currently a senior researcher in Korea Atomic Energy Research Institute. His research interests are nuclear fuel cycle facility design (hot-cell), management of spent fuel, and safety issues of nuclear fuel cycle. Chang Su Woo received his M.S from Seoul National University in 1987 and Ph.D. in Mechanical Engineering in 2003 from Youngnam University. He was recruited in 1989 by the Korea Institute of Machinery and Materials (KIMM) to work on fatigue analysis of mechanical components. He has published approximately 50 papers relating to research on rubber engineering components, and their applications, such as automotive mounts and bushings, suspension of railway vehicles, shock & vibration isolators, and structure energy dissipation systems.

Wan Doo Kim received his B.S, M.S and Ph.D. in Mechanical Engineering from Seoul National University in 1980, 1982 and 1993. He is currently a tenured researcher at Korea Institute of Machinery and Materials, and a steering committee member of National Science and Technology Council. His research interests are nature-inspired technology, rubber mechanics and sustainable mechanical design.