Closing the loop to develop optimal compounds, goods.

Do you know that early applications of a material which came to be known as natural rubber (NR), with [s.sub.5][H.sub.8] as a basic monomer unit, involved a product derived from the Hevea Brazieliensis tree? Other varieties of NR came from balata, guayule and gutta-percha.

The super heat dissipation properties under cyclic loading, resilience, electrical insulation, high tensile strength and wear resistance make NR an attractive choice in golfball interiors, cable insulation, tires, etc.

However, the desire to improve properties like resistance to environmental factors (ozone and ultraviolet rays), and protection against industrial oils, led to the discovery of synthetic rubber, especially with the advent of World War II. Commonly known synthetic rubbers are polychloroprene, isoprene, styrene-butadiene, butyl, nitrile, acrylic, butadiene and urethanes. The basis of modern synthetic rubbers lies in macromolecules synthesis by way of step- or chain-growth polymerization.

Rubber products are thus manufactured through some vulcanization process. In a non-vulcanized state, rubber does not have the desired tensile strength, is sticky and deforms permanently under large deformations. Rubber gets vulcanized at high temperatures and pressures, with addition of sulfur, accelerators and curatives. The sulfur and carbon atoms, along with metal ions and organic radicals, form crosslinks between polymer chains. A crosslink network measures physical properties and is controlled by vulcanization time and temperature. Mechanically, such process manifests itself by an increase of retractile forces and a possession of "rubbery" properties, such as increased elasticity.

These unique properties of rubber make it indispensable to many industries (bio-medical/dental professions, highway and flight safety, packaging, sports and consumer industries). Still, rubber applications are mainly derived through "trial and error." (Prototypes are built for testing, results of which are used to alter designs, as new prototypes are further tested until a satisfactory outcome is reached).

Such a sequential development approach to developing compounds and goods causes time delays in bringing products to market. Furthermore, any analysis after prototyping adds to the budgets of programs. In the end, the limited integration of computer-aided engineering (CAE) results in inefficiencies and a lack of design creativity.

Today product development integrates materials testing to computer-aided design (CAD), computer-aided manufacturing (CAM) and CAE. Still, products need to go through "what-if' scenarios. Though these operations take place in a virtual world, they are computer generated through a technology known as finite element analysis (FLEA).

Design using FEA has helped large firms improve the performance and quality of their finished products. It also helped them reduce the timetable to bring ideas to market, while optimizing the use of materials. By and large, design by analysis reduces weights, specifies the structural integrity of goods before prototyping and reduces costs of development and production.

In technical terms, design by analysis requires the amalgamation of:

* Material characterization;

* computing power (hardware and software); and

* technical know-how.

Still, characterizing rubber for FEA is at the research and development (R&D) stage. Required tests include modes of deformation beyond the uniaxial (specified in ASTM guidelines D595 and E412), minimum pressure to seal, and decay in physical properties in given environments. Besides, software for rubber analysis is highly specialized to tackle such complex issues as:

* Large strains and deformations;

* interfaces (bounding or contact);

* material incompressibility, nonlinear response to loads and viscoelasticity (time and temperature dependence); and

* mesh-refinement, during the course of an analysis, etc.

Rubber products analysis can obviously be acquired if price and the learning curve are properly assessed. An alternative is to rely on the exclusive services of a third party company on a contract basis.

Now catering to rubber manufacturers' needs in design by analysis requires mastery of two main areas:

* Compound development and characterization; and

* computer modeling (CAD/CAM and CAE).

These areas split into three branches, namely:

* Quasi-static evaluation;

* viscoelastic analysis or time effects: and

* frequency domain assessment.

Still, each of the branches outlined require special experimental and computing tools and know-how. Indeed, from a laboratory standpoint, one needs first to be able to assess compounds for processing, and compression or injection mold samples for testing after conditioning in controlled temperature and humidity conditions. Then, one needs to run mechanical and servo-hydraulic load frames and accessories, e.g., environmental chamber, non-contact video extensometry, etc. Additional testing equipment falls into time-domain evaluation, e.g., aging, viscoelasticity, fluids diffusion, for-which equipment such as heat and humidity chambers, ovens, relaxation fixtures, and so on is utilized.

In the sector of modeling, one needs workstations or high-end Pentiums for number crunching. Software with a reputation in CAD include Pro|Engineer, with its Mechanica; in CAE, MARC is very well respected for rubber analysis; and in CAM, codes such as FillCalc, CadPress and Polyflow are popular.

It is well recognized that companies have limited resources to allocate in functions that are not essential to their processes; outsourcing is attractive to complement core competencies. This, in fact, is not new, as companies have used outside suppliers for years. Indeed, 46% of the firms Purchasing Magazine recently surveyed, say they have increased outsourcing activities versus just 4% who say they are moving the other way around. The rise in outsourcing popularity appears tightly linked to the quest for corporate growth and market agility. Companies attempting insulation from future economic bust cycles are outsourcing to endow current growth, while avoiding large investments in permanent work forces or capital equipment.

When would one think of outsourcing? A company should outsource when:

* It is growing faster than investment resources;

* a third party firm can do the job cheaper;

* changes in technology make investment difficult to overcome;

* the company needs more flexibility;

* it faces investment in a non-core process;

* laws and statutory requirements changes lead to noncompliance; or

* employees are stretched thin, performing duties that compromise their main function.

There are three steps to outsourcing. First, identify your core competence, i.e., what differentiates your business from the competition? Maintain those functions within the company. Second, identify those processes in which your company does not have expertise: Identify any constraints, or bottleneck processes. Finally, identify the requirements for such processes, and select a source to assist in different levels of requirements.

In fact, one of the items outsourced today is rubber finite element analysis. Rubber is hyper and viscoelastic, incompressible, deforms significantly under load or straining, and is often used in contact with fluids. A prerequisite to any rubber product design is "quasi-static" analysis, tot which the material needs to be characterized in four basic modes of deformation: Uniaxial, equi-biaxial, planar and volumetric, in tension or compression. Test data under the deviatoric regime (uniaxial, planar and equi-biaxial) is to fit energy density functions (polynomials in strain invariants or nonlinear series of stretch ratios). Compressibility derives from "pressure vs. volume-ratio" charts.

Still, rubber applications generally need definition of least and maximum material conditions (LMC and MMC). The former ensures a minimum performance, such as the onset of sealing or retention in place. The latter is bound by the strength of rubber or mating components. Macroscopicalty, sealability follows leak testing of ring gaskets of simple cross-section under increased strain and stress levels. Material strength taps into principles of fracture mechanics under deviatoric modes of deformation.

Besides, seals should further look at friction effects and abrasion, particularly in dynamic situations. They should also account for degradation of rubber in operation (chemical and mechanical effects).

Moreover, some applications involve bonding rubber to metals or plastics. Others include reinforcing fabrics, in which case, any reinforcement-yarn is to test under quasi-static conditions, along with rubber testing. Resulting stress-strain data allow the determination of mechanical properties of bundle to combine with the geometry, boundary conditions, constraints and periodicity of rubber-matrix composite for homogenization, in the end, orthotropic elasticity constants are destined for computer modeling of the whole rubber product. Other applications subject seals to thermal cycling. Rubber is known to swell and degrade when exposed to fluids and temperature. Still, fluids diffusion in rubber is often coupled to mechanical effects (e.g., oil weeping through compressed gaskets). Hence, mass-transfer analysis requires testing for solubility and diffusivity at various levels of temperature and compression.

Additional rubber applications are dynamic. Impact loading is also a frequent acquaintance. Characterizing rubber for transient loading requires servo-hydraulic machinery, sized for frequency and stroke. Besides, transient loads are to program, as generated data include out of phase response, dampening characteristics, loss-angle and so on. Strain sensitivity is an issue in impact analysis too, from hyperelastic and dynamic standpoints. Testing and modeling of such testing are thus to conduct until satisfaction with test jig, sample and procedure.

Last but not least, modeling, especially at the model building stage, requires validation testing. Still, validation depends on product requirements. This topic is unfortunately too broad to cover in this article; it is therefore better to discuss with clients, case by case. Again, computer-analysis of rubber components requires software that handles hyper and viscoelastic material behavior, contact (between deforming bodies or deforming bodies and rigid surfaces), and large deformations. As calculations are nonlinear, thus iterative, any simplification in geometry speeds considerably virtual "what if'?" scenarios leading to optimal designs. In most applications, only sections are analyzed under plane strain or plane stress conditions (typical to long or thin geometries). Parts can also take advantage of axial or cyclic symmetry where only half a section, or a portion often referred to as a "piece of pie," needs analysis.

Reducing computing requirements in analyzing rubber products should not only look at the geometry, but material distribution, boundary conditions (restraints or contacts) and loads. Examples where axial symmetry fails include the crimping of hose assemblies (as a result of gaps between dies) or diaphragms (because of orthotropy of the reinforcing layer); Fabric-reinforced rubber should consider homogenization schemes based on properties of the yarn and rubber matrix. Though non-symmetric loads can still use axial or cyclic symmetry through decomposition in Fourier series. any event, rubber products are not expected to work upon assembly (inspection and evaluation at the factory level), but over a specified life. Viscoelastic analysis allows monitoring the degradation of rubber products in their working environment. Still, concerns should not end at optimal designs (i.e., nominal dimensions and "time zero"). Sensitivity studies ensure parts functioning in time, while tolerating variations in manufacturing, materials and loads. Additionally, parts need to meet assembly requirements, a problem that often emerges at OEMs levels when gathering supplied components.

In particular, design for assembly is important for sealing systems. Today's pressures to come up with further better products, at reduced schedules and costs, compel industries to shift to polymer-based composites and modeling. Rubber is known to be slightly compressible: its hydrostatic reaction out-weighs its shearing behavior. Over-compressing a gasket can thus break plastic housings, especially in the presence of knit lines. Therefore, assessing the onset of sealing is crucial both at "time zero" (leak testing) and during product life (in order to account for creep and stress relaxation phenomena).

In the end, while FEA has over the past three decades proven to be instrumental in designing mechanical parts, confidence comes from "closing the loop" between test material data, computer simulation and validation testing.