Guidelines for handling contact in simulations

Dec 5, 2007 10:47 AM, By Paul Dvorak

It's not uncommon to analyze a mechanism in which components initially touch or later come into contact.

Bob Williams
Product Manager
Algor Inc.
Pittsburgh, Pa.

Impact Engineers at Stern Pinball Inc., Melrose Park, Ill., model the impact of steel pinballs striking drop targets to find stresses and distortions in the targets.

But simulating contact is not always simple, even with cutting-edge software. Many parameters must be specified to properly simulate contact including the type of contact, stiffness, tolerance, interaction distance, time-step size, and more. And although the latest FEA software selects most of these parameters, some models benefit from user adjustments.

How you set up a contact analysis can help quickly and easily get a converged and accurate solution. But beware: Less-than-optimal setups can make the software run longer than necessary, compromise results, and even fail.

WHAT CAN GO WRONG

A few common difficulties encountered in contact analysis include:

Nonsmooth surfaces. When surfaces in contact have irregular edges, their roughness can cause convergence problems. For instance, the software will repeatedly try to solve the contact problem by adjusting parameters until satisfactory values are found, thus lengthening the analysis run-time. Nonsmooth surfaces can also lead to inaccurate results. For example, fitting a small cylinder inside a larger one should produce a uniform stress contour along the contact surface. But insufficiently smooth surfaces can produce stress variations.

The shrink-fit problem lets a chilled shaft expand in a close-tolerance hole. Stresses are calculated in the surrounding material.

A soft material contacting a harder one can cause a default contact-stiffness calculation that leads to poor convergence, an effect called "chattering." Or, the software may generate an unrealistic penetration in which one surface passes through another.

Fast-moving impacts produce high-frequency loads that force the solver to take smaller time steps that lengthen run times. A high-energy loss may also lead to an overdamped behavior. Finally, parts might penetrate each other incorrectly because one moved too fast for the solver to "see" contact.

Large displacements or deformations, more than 50% strain, mean the basic assumptions used in calculating contact parameters may no longer be valid due to changes in geometry.

BASIC STEPS OF CONTACT ANALYSIS

Avoiding these difficulties means understanding at least seven basic steps for contact analysis. For instance:

Surface-to-surface contact in the planetary gears drives the motion of four planet gears about the sun gear.

Create the geometry and mesh. Reasonably smooth surfaces with a uniform mesh work best in contact analyses. Thus, it's important to know how to use software tools to generate and refine the mesh.

Identify possible contact areas. Before starting a contact analysis, specify where contact might occur. A part could touch itself in large-deformation problems such as rubber-elasticity analyses. In such cases, make certain that multiple contact areas are defined (either automatically or by you) to cover all potential contact situations.

Determine the contact parameters. FEA software often calculates contact-parameter values based on geometry, mesh, and material constants. But you may need to adjust values for anticipated contact problems, such as large deformation. For example, you might try increasing the contact interaction distance so the software can better detect the nodes that are in contact. Determine the solution method. FEA software typically provides several options for the numerical-solution technique used in contact problems. The full Newton-Raphson method is often recommended for nonlinear contact analyses.