A more advanced way to designing rubber products.

For most of the history of rubber as an engineering material, its applications have largely been developed by trial and error. Prototypes are made for testing, results of which are used to alter designs, as modified prototypes are further tested until acceptable products emerge. Despite spending time and money up-front on compounds and processes, resulting goods may not work satisfactorily and are certainly not optimal if they do. An alternative to trial and error is virtual testing - the computer testing of products and processes. This requires a good understanding of materials and the art of modeling. This article describes testing rubber towards building computer models. It also presents several studies completed on behalf of rubber molders in Canada and the United States.

Crimping composite hose

WIDL developed a quasi-static finite element model to simulate leakage of crimp on a composite hose for Aeroquip (figure 1). Yarn reinforces hose, withstanding pressure. The hose fits on the nipple and the socket crimps outer face prior to testing, an exercise Aeroquip repeats until the assembly seals.

[Figure 1 ILLUSTRATION OMITTED]

Iso-butyl rubber was tested as collected data were fit to the models for calculations. Bares of alloys making the nipple and socket - received from Aluminum Association Ltd. (England) and Pichnet (France) - were tested to elastic-plastic straining. The yam was pulled under quasi-static conditions and data were post-processed for elasticity modulus and Poisson's ratio. Mechanical properties of bundle were calculated based on those of fiber and matrix. These were used for non-linear orthotropic (material properties vary with direction [x, y or z]) modeling. The onset of sealing on computer was correlated to testing rings from hose. Predictions were within 5% of testing crimped assemblies. The project took four weeks of testing and analysis. Aeroquip is adopting the techniques developed within this study in-house.

Cracked seal in piping system

A T-pipe joint in a system to protect buildings against fire cracked in service and Central Sprinkler called WIDL for investigation (figure 3). A T-joint presents two plans of symmetry; only a quarter was analyzed. High shear strains on the pipe opening and the base of the housing were clear. These locations initiated cracks on seals in service. Sealing pressures were low, the gasket slipped between the housing and the pipe, and the reduced height lip did not compress.

[Figure 3 ILLUSTRATION OMITTED]

WIDL eliminated contact seal-sharp openings. Beads reduced rotation of the gasket, thus slippage. Sealing derived from testing the rings and monitoring degradation of the rubber in boiling water. A design emerged in three weeks of material characterization and computer simulation; cost of the entire project was less than the expenses of trial and error.

Closure of electrical connector

Framatome Connectors Interlock (FCI) called upon WIDL to lower forces to assemble while seal a 14-way connector. A seal in production for a 10-way connector started the study.

Blue prints of the production seal and plastic mating components were used in building the quarter assembly. Self-lubricating silicone was tested for analysis. Friction of the self-lubricating silicone on plastic substrates was monitored for inclusion in Coulomb's model.

The female connector was split to expand the seal prior to inserting the male connector. The latter was found to hit on rubber, explaining the forces FCI complained about. Sealing pressures adopting the 10-way seal in a 14-way application were low.

In the new design, contact was reduced to three beads, improving sealing. Insertion forces depended on friction and the first contact male connector-seal. These dropped to a fifth of existing values with the re-design to the male connector.

Collapse of underhood latch

Ford was supplied latches meant to mushroom on rings upon pulling cables mounted to inner inserts in an underhood application. The original latch buckled inward; one molder proposed for further testing kinked with imperfect loading (figure 5). Simulations suggested include:

* Reducing height and optimizing the wall to control initiation and monitor propagation of collapse;

* adding ribs to the support roof and wall from collapsing inward, while the former deforms to a flat;

* increasing width and depth of cuts so that the wall travels outward the most; and,

* adding a peripheral nib to constantly the hold latch at the same location prior to mushrooming its top.

[Figure 5 ILLUSTRATION OMITTED]

Rubber was tested in uniaxial tension and compression and both equi-biaxial and planar tension. It was also tested under volumetric compression conditions.

The proposed latch emerged in four weeks of testing and analysis. Axi-symmetric models and "pieces of pie" were used in optimization; half latches confirmed the findings. The proposed design was compact and 13% lighter and cost less than the mold that was about to be built for testing.

Dampening tractor cab

Frequencies between 18 and 23 Hz raised complaints as the cab on tractor by John Deere shacked, questioning the dampening characteristics of the mount.

Modes within such a frequency range amplified the acceleration via the frame rattling the cab in the showroom. Modeling investigated:

* The use of a lower durometer rubber while;

* weakening the mount by adding holes to its body; and

* a combination of the first two.

Rubber was characterized in the hyper-elastic regime. It was also characterized for hysteresis under the weight of cabs over operating frequencies. Also, John Deere supplied vibration data for the frame, input to the viscoelastic finite element model built. Simulation resulted in a better dampening mount.

Design of a pad for rail crossings

Performance Polymers Inc. (PPI) requested virtual testing to design pads for railway-road crossings. In addition to improving dampening over the current solid pads, PPI recommended that (i) pads absorb mis-match between ties and (ii) cells deform under the weight of concrete panels.

EPDM was characterized for modeling. A step-profile with collapsing cells was developed. The ends were lighter and the width accommodated expansion upon compression.

Extruded pads were tested to indicate perfect agreement with analytical predictions. The entire project took three weeks.

Optimizing a seal for a dishwasher

Poly-Nova called WIDL to improve the performance of a dishwasher seal for Whirlpool. The existing seal had a reversed U shape with on and off cleats and ribs holding a gap between the walls. Due to cyclic symmetry, only a "piece of pie" was analyzed.

Normal stress was higher on the pump than the sump. Shield by the sump was under tension. Its geometry promoted stresses in addition to collecting residues. Initial analyses suggested:

* Shaping the outside bore not to collect dust, etc., while modifying the housing to protect the interface gasket-pump;

* eliminating the wall below the sealing lines, balancing the latter and adding beads at the interface seal-pump; and

* re-designing the cleats, as it was possible to slightly lower the pump.

Ribs between the walls were not altered; waviness in the sealing pressure was thought to increase if these were sparse. The new design emerged in three weeks of material testing and computer analysis; savings in rubber alone were above 25%.

Conclusion

Trends in all industries to develop products faster, lighter and at a lower cost create a need for "virtual" (computer) rather than actual testing. To succeed in such an endeavor, two items need combination: (1) material science and (2) computer technology.

This article demonstrates the usefulness and accuracy of virtual manufacturing. It reviews some tests on rubber needed for building models that would simulate the performance of finished rubber products. It also presents testing geared at ensuring the accuracy of virtual prototypes. While design through analysis is seeing wide acceptance in several industries, it has only recently addressed non-linearities associated with rubber-like materials. Virtual prototypes can be developed in-house if cost and the learning effort have been assessed adequately. Alternatively, they can derive from a third party company on an as needed basis.

[Figures 2 and 4 ILLUSTRATION OMITTED]

References

[1.] "ASTM Designation D638 - 91 (Reapproved 1992) - Standard Test Method for Tensile Properties of Plastics" 1994 Annual Book of ASTM Standards, Vol. 08.01.

[2.] "ASTM Designation E132 - 86 (Reapproved 1992) - Standard Test Method for Poisson's Ratio at Room Temperature" 1994 Annual Book of ASTM Standards, Vol. 08.01.

[3.] Trealor, L.R.G., "The physics of rubber elasticity," Second Edition, Oxford University Press, England (1958).

[4.] Trealor, L.R.G., "Stress-strain data for vulcanized rubber under various types of deformation," Trans. Faraday Soc., Vol. 40, pp. 59-70 (1940).

[5.] Gregory, M.T., "The stress-strain behavior of filled rubbers at moderate strains," Plastics and Rubber: Material and Applications (1979).

[6.] Ogden, R.W., "Recent advances in the phenomenological theory of rubber elasticity," Rubber Chemistry and Technology, Vol. 59, 361 (1986).

[7.] Rivlin, R.S., "Large elastic deformations of isotropic materials, IV. Further developments of the general theory," Phil Trans. Royal Soc. London, A 241 (1948).

[8.] "ASTM Designation E412-87 - Test Methods for Rubber Property in Tension" 1994 Annual Book of ASTM Standards, Section 9 Rubber, Vol. 09.02.

[9.] "ASTM Designation D575-88 - Test Methods for Rubber Property in Compression" 1994 Annual Book of ASTM Standards, Section 9 Rubber, Vol. 09.01.

[10.] "MARC User's Manual," version 6.2, MARC Analysis Research Corporation, Palo Alto, CA (1996).

[11.] "ABAQUS/Standard User's Manual," version 5.9, Hibbitt, Karlsson & Sorensen, Inc., Pawtucket, RI (1998).

[12.] "Pro\Engineer User Manuals," version 18, Parametric Technology Corp., 28 Technology Drive, Waterham MA (1998).

[13.] "Pro\Mesh/Pro\FEM-Post User Guide," version 18, Parametric Technology Corp., 28 Technology Drive, Waterham MA (1998).

[14.] Twizell, E.H. and Ogden, R.W., "Non-linear optimization of the material constants in Ogden's stress-deformation function for incompressible isotropic elastic materials," J. Austral. Math. Soc., Series B, Vol. 24, pp. 424-434 (1983).

Dr. Ben Chouchaoui completed his engineering education at Polytechnic School of Algiers and graduate studies at Polytechnic School of Montreal and the University of Waterloo. He currently runs WIDL, a lab specializing in materials and process testing and simulations to aid in product design and manufacturing.