Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Paige

Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

FEA in the Design Process of Rubber Bushings

Ryan E. Paige
Cooper Standard Automotive NVH Control Systems Division 207 West Street Auburn, IN 46706 Abstract: The development of the rubber bushing as a vibration control system component can be very complex. This makes finite element analysis (FEA) a valuable and often necessary tool in the design process as it can include many of the bushing design variables and accurately predict resulting behaviors. This paper demonstrates how FEA can be incorporated into the design process to capture complexities such as pre-compression, voids, simple friction effects and other complex geometry. Comparisons of numerical, analytical and test data are shown for some specific bushing designs and then the numerical method is used to demonstrate the effective range of behavior resulting from modifying various design parameters individually.

1. Introduction
Filled natural rubber bushings are widely used in automotive engineering for vibration control. It has the ability to be used in a simple packaging space yet can be easily modified to exhibit complex and variable behaviors. In today's industry the bushing has become an intricate part of the structure as it is expected to perform more elaborate functions and meet increasingly precise requirements. The resulting development process of the bushing as a component can range from a somewhat simple to a complicated design. The finite element method provides a powerful tool to quickly assess complicated bushing designs and vary the necessary parameters to meet the given requirements. The results of the analyses make it possible to achieve product requirements quickly for very complex designs. Rubber bushings come in many shapes and sizes but typically can be described as consisting of an outer metal sleeve and an inner metal tube with a molded annular rubber element (see Figure 1 for example). The rubber geometry is characteristically described by an inner and outer radius and an axial length or height. The rubber often incorporates some sort of void configuration and ends or sides can be tapered, filleted and so forth. For the purpose of this paper it is necessary to classify bushings in three general categories describing the production processes that must be incorporated to accurately predict behavior. 1. 2. Mold bonded bushing The cross-linking (vulcanization) and cross-bridging (bonding between rubber and adhesive layers applied to metals) occur simultaneously in the mold. Post bonded bushing The cross-bridging is post vulcanization and occurs separately from the mold process. The post bonded bushing includes those bushings that may be utilizing both a mold bond and post bond, one to the inner metal and one to the outer metal. The post

2002 ABAQUS Users Conference

bond usually involves radial pre-compression to the rubber by the inner or outer metal before cross-bridging. 3. Shot bushing The rubber is molded and then subjected to significant radial pre-compression by an under sized outer metal and over sized inner metal relying on friction to prevent slip between the metal and rubber under loading instead of an actual bond. The shot bushing includes those that combine a shot inner or outer with a mold bonded outer or inner.

Bushings perform a variety of functions usually pertaining to vibration isolation and load absorption. The main characteristics of interest for the bushing are stiffness and durability, and the stiffness of the bushing can be characterized both statically and dynamically. However, only static stiffness is considered for the scope of this paper since it is the most fundamental design characteristic, and can be related directly to the dynamic stiffness as well. Characterization of bushing stiffness can be quite involved. Static stiffness can be non-linear varying with load or deflection and is described by a load vs. deflection curve, or rate curve (and may be referred to as just the rate). The bushing stiffness characterization can be further divided into the following directional rates: radial, axial, torsional and conical (or tilting) and may include static preloads. Adding voids to a bushing creates multiple radial and conical rates, for example, the radial rate can now be described as both radial solid and radial void rates. The void direction rate (and even solid direction rates for large voids) is then comprised of a primary rate, and a secondary rate referring to the stiffness before and after the void is closed. In industry, mold bonded bushings normally include a radial pre-compression accomplished by a swaging process of the outer metal as part of assembly. This is typically done to eliminate stresses due to rubber shrinkage upon cooling but the amount of precompression can be varied to affect rates as well. As is evident from the explication and characterization of the bushing, it is not reasonable to oversimplify behavior in the design process by ignoring all the differences and variations achievable with a bushing. The remainder of the document will briefly review classical bushing theory and then provide more detailed discussion of numerical and experimental results to demonstrate the value, and often necessity of FEA in bushing design.

2. Classical Bushing Theory


Much of the classical theory refers to a very simple bushing, one that consists of inner and outer metal tubes that are bonded to an annular rubber wall of uniform length with planar ends and does not include any pre-compression, pre-load, voids or other geometric complexities. The behavior of these simple bushings can be approximated by analytical methods developed for the various modes of deflection. 2.1 Radial Stiffness Approximation

The methods most commonly used to approximate the radial stiffness of a simple bushing analytically are based on the work of (Adkins and Gent, 1954) which considers two special cases for simplification. One is a long bushing where plain strain is assumed, and the other is a short bushing where plain stress is assumed. These conditions yield the following analytical expression for the radial stiffness of a simple bushing.
Kr = LG

(1)

2002 ABAQUS Users Conference

where

K r = radial stiffness G = shear modulus L = bushing length = numerical factor corresponding to a long or short bushing as follows

2 2 80 D + d = 2 2 2 2 25 D + d ln (D / d ) 9 D d

) ( ) ( ) 2 2 4 (D + d ) = (D 2 + d 2 )ln(D / d ) (D 2 d 2 )

where

D = outer diameter of rubber d = inner diameter of rubber

Many bushings in industry are intermediate in length resulting in a behavior that is between these two extremes. A more detailed discussion of exact and approximate expressions of radial stiffness for a simple bushing of finite length is presented by (Horton, 1999). However, a typical interpolation (Lindley, 1992) used in the natural rubber engineering community is
S + 0.15( L S )L

(D d )

(2)

which works well for simple bushings where d / D = 0.5 , similar to the bushing shown in Figure 1. 2.2 Axial Stiffness Approximation

Commonly used formulae to approximate the axial stiffness of a simple bushing analytically are based on the work of (Rivlin, 1949). A commonly used analytical approximation for axial stiffness is given by
Ka = log10 (D / d ) 2.73GL

(3)

where

Ka = axial stiffness.

3. Bushing FEA
In industry, classical bushing theory is useful for only a fraction of bushing concept development but the potential of FEA for bushings was pointed out by (Morman and Pan, 1987). As the bushing requirements and design become more complex it is evident why the benefits of FEA for bushings is tremendous. The focus of the remainder of the document is to demonstrate the value and necessity of FEA to accurately predict and evaluate the behavior of not only the simpler bushings but also those of very complicated design. Furthermore, FEA is used to capture a range of behaviors possible from individual design parameter and assembly process parameter changes. 2002 ABAQUS Users Conference 3

ABAQUS was used for all FEA discussed herein. Before proceeding, it must also be understood the importance of good material models for accurate FEA. Some of the fundamentals are briefly discussed here, however, it is not the intent of this paper to address this subject in detail. All analyses employed common hyperelastic material models based on curve fits of experimental stress strain data for a particular compound. The rubber compound in bushings typically undergo very complex multiple states of deformation. For this reason and good practice, the test data utilizes three modes of deformation, each cyclically loaded to specific strain levels until stabilized. The three deformation modes used are uniaxial tension, planar tension and equibiaxial tension, which correspond to tension, pure shear and compression. (Miller, 2000) discusses elastomer testing for hyperelastic material models in much more detail. The material for these bushings is also assumed to be isotropic, perfectly elastic, slightly compressible and operating at a specific temperature. Moreover, the behavior of natural rubber components is very dependent on overall geometry including the small details. Thus, changes produced by varying a specific design parameter may have larger or smaller effects on similar but non-identical parts. Unique designs need to examined individually to quantify the amount of change in behavior. 3.1 Simple Mold Bonded Bushing Static Rate Correlation at Small Strain

The bushing on the left in Figure 1 is an example of a very simple mold bonded bushing of natural rubber compound. The bushing design is at its most basic with planar ends, no voids and no radial precompression, or sizing of the bushing. As expected, this fundamental design at small strains yields favorable predictions for both classical theory and FEA. Figure 2 shows the initial (small strain) static radial rate of the bushing comparing experimental data to FEA and classical theory based on Equations 1 and 2. There is no common standard defined for calculating the rates of these bushings but for this example the initial radial rates are calculated linearly between 0.5mm and 1.0mm of deflection. Figure 3 makes this same comparison for axial rates with the classical theory results based on Equation 3. The axial rates for this example are calculated between 2.0mm and 3.0mm. Note here, a different rubber compound was used in the bushings for the axial data and for the radial data such that they have different values of G and different material models. However, a mold bonded bushing this simple is not common in industry as the design would typically incorporate sizing or pre-compression. 3.2 Sizing Effects on Initial Static Rates and Correlation at Small Strain

Routinely, even the simple mold bonded bushings used in industry are sized as shown for the bushing in Figure 1. The sizing is typically accomplished by a swaging process that shrinks the outer metal radially inward. A 2-3% radial reduction of the rubber wall thickness is generally used for the elimination of residual material stresses and larger amounts typically ranging from 3-6% are used to introduce radial precompression. This process parameter will affect rates, especially radial rates so that it is necessary to include in behavior predictions. The amount of the effect of the sizing process depends somewhat on the overall geometry of the bushing. The same simple bushings used for acquisition of the non-sized experimental data were the same ones radially pre-compressed and used for the sized bushing data acquisition. This eliminates any mold process variation or vulcanization differences in experimental data. Figure 4 shows the initial (small strain) static radial rate of the same simple bushing in Figure 1 after it has been sized 4.5%. Experimental data is compared to FEA and classical theory based on Equations 1 and 2. The initial rates in this example were calculated linearly between 0.5mm and 1.0mm of deflection. Even at 4 2002 ABAQUS Users Conference

small strains, this particular bushing has an initial radial rate increase of nearly 20%. As demonstrated in this example, FEA proves useful for predicting radial rate changes due to sizing. Figure 5 shows the resulting initial axial static rates for the same simple bushing, again sized roughly 4.5%, comparing experimental data to FEA and classical theory based on Equation 3. This example shows that the axial rate in a mold bonded bushing is much less sensitive to sizing than radial rate when considering typical amounts of sizing as discussed previously. FEA correlates well with changes in this design parameter simulating the effects of sizing on behavior. The radial rates are sensitive to various amounts of radial pre-compression so not only is sizing considered here, but also post bonding the outer metal. Although typically post bonding the outer metal will not be done to a bushing with this simple of geometry, it is used here for relative comparison since it will allow for larger amounts of pre-compression. The methods yield a difference in boundary conditions during the assembly phase of the bushing. For the post bonded case the rubber can expand axially at the outer metal interface during radial compression and is then fixed, or bonded for loading. Figure 6 shows the radial rate of the simple bushing versus pre-compression from sizing 2%, 4% and 6%, and post bonding to the outer metal after 6%, 9% and 12%pre-compression. The radial rates are calculated at a strain of approximately 25%, and show a rate variation of about 40%. 3.3 Complex Bushing Correlation at Large Strains

Today, the majority of mold bonded bushings in industry will incorporate not only pre-compression but also complex geometry such that the bushing can be tuned to different rates along all three axes. Adding voids to a bushing is an easy way to introduce multiple and variable radial and conical. Of course, classical theory does not account for this but FEA will make it possible to predict necessary shapes and sizes for the voids. This will allow not only the primary rates to be determined but secondary as well. Additionally, when and how quickly the changes in rate occur can be assessed with FEA. Another geometric design variable for bushings is the end condition, or the fact that the ends of a bushing often are not planar. The ends may incorporate scallops or grooves to varying degrees or conical build-ups of rubber at the inner or outer metal. Figure 7 shows an example of a more complex mold bonded bushing design used for NVH control in the automotive industry referred to hereafter as the baseline complex bushing design. The left side of Figure 7 shows the bushing cut in half in its as molded free state. This bushing incorporates various design parameters including sizing (about 7.5%), small voids and a non-planar end condition, all of which are necessary to include to predict behavior well. The right side of Figure 7 shows this bushing after being sized and then subjected to a large radial load in the void direction. Figure 8 compares experimental results versus FEA results in the radial void direction showing very good correlation to experimental data from five individual bushings for both primary rates and secondary rates which occur after the void is in full contact. The radial solid direction rate curves are shown in Figure 9 which again shows excellent correlation between FEA and experimental data taken from five individual bushings. Even with this small void the radial rates will vary significantly. It takes almost 10,000N to achieve 7mm of static deflection in the radial solid direction, but less than 7000N for the void direction. This small void also yields roughly a 2:1 secondary to primary rate ratio in the void direction. This amount of loading combined with the significant pre-compression put the bushing at levels over 40% strain.

2002 ABAQUS Users Conference

3.4

Void Size Effect

The size and geometry of the void are very important for determining rates. Many bushings with voids are designed such that the void will close under radial load in the void direction and continue to carry load yielding a specific secondary rate. Often predicting the secondary rate, when it occurs and whether the rate change is abrupt or gradual may be required. Figure 10 shows the complicated baseline bushing design previously discussed with significantly larger voids, the left side showing the bushing in its as molded free state and the right side showing the bushing after radial loading. Common sense dictates that changing the void size will change rate characteristics, however FEA can be used to predict how significant the changes will be. Figure 11 numerically compares the radial void direction rate curves for the same bushing design, varying only the void size. Note how the bushing with the large void carries the initial radial void rate of about 105N/mm nearly 6mm and then transitions abruptly, while the baseline bushing design has an initial rate of about 600N/mm that gradually changes after 2mm of displacement. The large void design also achieves about an 11:1 secondary to primary radial void rate ratio. Figure 12 numerically compares the radial solid direction rate curves for the same bushing design, again with the only difference being the void size. 3.5 Friction Effect

Accurate void geometry has the greatest impact on characterization of primary rates for the radial void direction but friction also plays a role in the secondary rates. As the bushing void closes, the surfaces of the void come into contact invoking both normal and shear forces. Rubber friction behavior is very complex, exhibiting non-linearity in both the static and kinetic friction interactions which change relative to different amounts of compression in the rubber. Examining the intricacies of friction in detail is beyond the scope of this paper, however, including simple static Coulomb friction in the model is considered. Figure 13 shows numerical results used to compare the difference in secondary (after void closure) radial rates for frictionless contact, using friction coefficients of 0.1 and 0.3, and a no slip interaction. The large void bushing model is used here due to the clarity of the void being fully open and fully closed. This example shows a distinction between the frictionless case and incorporating friction, yielding a 15% difference in radial load at 2mm deflection after the void has closed. However, the sensitivity to the amount of friction is only relevant for very small coefficients of friction when using the basic Coulomb friction model. As demonstrated for this case, a coefficient of 0.3 yields results very close to the no-slip condition. 3.6 End Condition Effect

Adding or removing rubber at the ends of a bushing can have a significant effect on stiffness rates. Figure 14 shows the baseline complex bushing with planar ends in its as molded free state on the left and loaded state on the right. Figure 15 uses the numerical results to compare the resulting radial rates due to this particular change in the bushing geometry. In this example there is almost a 30% difference in load at 7mm deflection in both the radial solid and radial void directions. This demonstrates how variations in the end geometry for a bushing geometry must also be considered to predict behavior accurately. Often changes in end geometry are more subtle than this example, but still must be considered. 3.7 Sizing Effect at Large Strain for Radial Void Rates

Previously it was shown how much influence various amounts of sizing have on the initial radial rates of a simple mold bonded bushing. The design sensitivity of this parameter for complex bushings can be 6 2002 ABAQUS Users Conference

enormous at large strains, even in radial void rates. The baseline complex bushing is sized 7.5% based on rubber wall thickness to which the model shows very good agreement with experimental data. The extent to which variations in the amount of sizing will influence the behavior of the baseline complex bushing is further demonstrated here numerically. The void direction static radial rate curves are shown in Figure 16 for the same bushing sized 1.5% and 4%, and compared to the baseline of 7.5% sizing. The radial force necessary for a 7mm static deflection for the baseline is about 6800N, and drops to 5650N when sized 4%, and to 4930N when sized only 1.5%.

4. Conclusion
A brief history and theory were provided to show the relevance and enormous benefits of utilizing FEA in the development process of natural rubber bushings. FEA allows one to evaluate the sensitivity of many design parameters in the detail necessary to accurately predict complex bushing stiffness rate behavior. Examples were given to show correlation with experimental data and numerical results were used to illustrate examples of design parameter sensitivity.

5. References
1. Adkins, J. E. and A. N. Gent, Load-Deflexion Relations of Rubber Bush Mountings, British Journal of Applied Physics, vol. 5, pp. 354-358, 1954. 2. Horton, J. M., M. J. C. Gover, and G. E. Tupholme, Stiffness of Rubber Bush Mountings Subjected to Radial Loading, Rubber Chemistry and Technology, vol. 73, pp. 253-264, 1999. 3. Lindley, P. B., Engineering Design with Natural Rubber, 5th ed., MRPRA, (revised by K. N. G. Fuller and A. H. Muhr), 1992. 4. Morman, K. M., and T. Y. Pan, Application of FEA in the Design of Automotive Components, Rubber Chemistry and Technology, vol. 61, pp. 503-533, 1987. 5. Miller, K., Testing Elastomers for Hyperelastic Material Models in Finite Element Analysis, Axel Products, Inc., 2000. 6. Rivlin, R. S. and D.W. Saunders, Philos. Trans. R. Soc. London, Ser. A 242, 173, 1949.

2002 ABAQUS Users Conference

Figure 1. Simple mold bonded bushing before sizing and after sizing.

3000 2500 2000 Load (N) 1500 1000 500 0 0 0.5 Deflection (mm) 1 1.5 expt. (k = 1370 N/mm) FEA (k = 1395 N/mm) cl. theory (k = 1670 N/mm)

Figure 2. Initial static radial rate curves of a simple bushing (k stiffness between 0.5 and 1.0mm).

2002 ABAQUS Users Conference

2500 expt. (k = 350 N/mm) 2000 FEA (k = 370 N/mm) cl. theory (k = 340 N/mm)

Load (N)

1500

1000

500

0 0 1 2 3 Deflection (mm) 4 5 6

Figure 3. Initial static axial rate curves of a simple bushing (k stiffness between 2.0 and 3.0mm).

3000 expt. (k = 1625 N/mm) 2500 2000 FEA (k = 1635 N/mm) cl. theory (k= 1840 N/mm)

Load (N)

1500 1000 500 0 0 0.5 Deflection (mm) 1 1.5

Figure 4. Initial static radial rate curves of a simple bushing sized 4.5% (k stiffness between 0.5 and 1.0mm).

2002 ABAQUS Users Conference

2500 expt. (k = 330 N/mm) 2000 FEA (k = 350 N/mm) cl. theory (k = 350 N/mm)
Load (N)

1500

1000

500

0 0 1 2 3 Deflection (mm) 4 5 6

Figure 5. Initial static axial rate curves of a simple bushing sized 4.5% (k stiffness between 2.0 and 3.0mm).

2500 sizing process 2250 post bonding outer metal

Radial Rate (N/mm)

2000

1750

1500 0 2 4 6 8 10 12 % Radial Pre-Compression

Figure 6. Radial rate comparison of simple bushing with various levels of precompression. 10 2002 ABAQUS Users Conference

Figure 7. Mold bonded baseline complex bushing design in free state on left and deformed shape after radial void loading on right (ID=25mm and OD=60mm).

7000 6000 5000 expt.(specimen_1) expt.(specimen_2) expt.(specimen_3) expt.(specimen_4) expt.(specimen_5) radial_void_FEA secondary rate

Load (N)

4000 3000 2000

void closure primary rate

1000 0 0 1 2 3 4 Deflection (mm) 5 6 7

Figure 8. Void direction radial rate comparison of experimental data versus FEA for mold bonded bushing with small void at large strain. 2002 ABAQUS Users Conference 11

10000 9000 8000 7000 expt.(specimen_1) expt.(specimen_2) expt.(specimen_3) expt.(specimen_4) expt.(specimen_5) radial_solid_FEA

Load (N)

6000 5000 4000 3000 2000 1000 0 0

3 4 Deflection (mm)

Figure 9. Solid direction radial rate comparison of experimental data versus FEA for mold bonded bushing with small void at large strain.

Figure 10. Mold bonded baseline complex bushing design with larger void (free state on left and deformed shape after radial void direction loading on the right). 12 2002 ABAQUS Users Conference

6000 radial_rate_curve (small_void) 5000 4000 radial_rate_curve (large void) k2 = 1260 N/mm

Load (N)

3000 2000 k1 = 600 N/mm 1000 k1 = 105 N/mm 0 0 2 4 Deflection (mm) 6 k2 = 1150 N/mm

Figure 11. Radial void direction rate curves for same mold bonded bushing with different void geometries and size at large strain.

10000 radial_rate_curve (baseline) 8000 radial_rate_curve (large_void)

Load (N)

6000

4000

2000

0 0 1 2 3 4 Deflection (mm) 5 6 7

Figure 12. Radial solid direction rate curves for same mold bonded bushing with different void geometries and size at large strain. 2002 ABAQUS Users Conference 13

3000

frictionless mu = 0.1 mu = 0.3 no-slip

2500

Load (N)

2000

1500

1000

500 5 6 Defelection (mm) 7 8

Figure 13. Effect of simple constant friction coefficient for void surfaces on secondary radial rate (numerical results).

Figure 14. Mold bonded bushing with small void and planar ends in free state and deformed shape after radial void loading.

14

2002 ABAQUS Users Conference

10000 9000 8000 7000


Load (N)

radial_void_rate (baseline) radial_void_rate (planar_ends) radial_solid_rate (baseline) radial_solid_rate (planar_ends)

6000 5000 4000 3000 2000 1000 0 0 1 2 3 4 Deflection (mm) 5 6 7

Figure 15. Comparison of radial rates for the complex baseline bushing design with different geometric end conditions at large strain.

8000 sized 1.5% 7000 6000 sized 4.0% sized 7.5%

Load (N)

5000 4000 3000 2000 1000 0 0 1 2 3 4 Deflection (mm) 5 6 7

Figure 16. Comparison of sizing effects on static radial void direction rate curves for the complex baseline bushing design at large strain. 2002 ABAQUS Users Conference 15

You might also like