Modeling thermal fatigue in copper alloys: where Coffin-Manson breaks down

Copper alloys are the workhorse materials of the mining industry. Bus bars, motor windings, heat exchangers, and electrical contacts all depend on copper’s combination of conductivity, ductility, and corrosion resistance. And they all fail eventually.

The standard approach to predicting fatigue life is the Coffin-Manson relationship, which relates plastic strain amplitude to the number of cycles to failure. It works well in controlled laboratory conditions: isothermal, uniaxial, constant amplitude. The problem is that real components in mining applications experience none of these conditions.

The real loading environment

A copper bus bar in a smelter experiences:

• Thermal cycling from ambient to 200+ degrees Celsius as furnace loads change
• Superimposed mechanical vibration from connected equipment
• Current-induced Joule heating that varies with process demand
• Oxidation and environmental degradation that changes material properties over time
• Bolt relaxation at connections that changes the mechanical boundary conditions

The Coffin-Manson relationship sees none of this complexity. It sees a single number: plastic strain amplitude. Everything else is collapsed into empirical fitting constants that were measured under conditions nothing like a smelter bus bar.

A different approach

Our models start from the coupled thermomechanical equations. We solve the heat equation to get the temperature field, feed that into the constitutive model to get the stress and strain fields, and track the accumulated damage cycle by cycle. The material model accounts for cyclic hardening, mean stress effects, and oxidation-assisted crack initiation.

This is more work than looking up a Coffin-Manson curve. But the result is a model that can answer questions the empirical approach cannot:

• What happens if we change the cooling schedule?
• What happens if we increase the operating current by 15%?
• Where will the crack initiate, and in which direction will it propagate?
• How much life do we gain by switching from C11000 to C18200?

These are the questions that matter to the engineer responsible for keeping the smelter running.

When to use empirical models

Coffin-Manson is not wrong. It is a powerful tool when used within its domain of validity: well-characterized materials, well-defined loading, conditions similar to the test data. If your operating conditions match the test conditions, use the empirical model. It is faster and cheaper.

The physics-based approach earns its place when conditions are outside the empirical envelope: unusual loading histories, extreme temperatures, multi-axial stress states, or new material variants without extensive test data. In those situations, extrapolating an empirical curve is guessing. Solving the governing equations is engineering.

What we are working on

Our current research focuses on developing validated computational models for copper alloy fatigue under realistic mining conditions. The goal is not to replace empirical methods but to extend the engineer’s toolkit to conditions where empirical data is scarce, expensive, or unavailable.

If you are dealing with unexpected copper alloy failures in a mining or industrial application, we would like to hear about it.