Thermal cycling and heat checking: where the cracks come from and how to slow them

The thermal-fatigue literature consistently identifies three variables that set how fast a heat-check network develops on a given die. Peak surface temperature on the contact spike, ΔT per cycle, and cycle frequency. Each one acts independently. A shop has different room to move on each one.

Peak surface temperature. Set primarily by the billet temperature, the die material's heat conduction, and the contact time. For a 1100°C carbon-steel billet on H13, the peak surface spike sits in the 500-700°C range depending on contact time and die-bulk temperature. The shop usually cannot move billet temperature. The forge spec calls for the steel to be in its hot-working window, and dropping billet temperature to protect the die changes the forge load, the part fill, and the downstream microstructure. Peak surface temperature is mostly fixed by the process.

ΔT per cycle. The temperature swing between the contact spike and the inter-hit equilibrium. This is set by preheat (Lesson 2), by cooling spray management (Lesson 3), by lube application (Lessons 5 and 6), and by the die-bulk temperature trend across the shift. A shop has real room to move here. A cold die between hits has a larger ΔT than a die held in its operating window. A lube spray that drops the surface 150°C on every cycle drives a larger ΔT than one that drops it 70°C. The Hawryluk 2020 paper on preheating methods showed induction preheat reaching working temperature in roughly 20 minutes against more than 50 minutes for traditional methods, which means the die spends less of its early service in the high-ΔT regime.

Cycle frequency. The number of thermal cycles per unit time. Set by the production rate. A press running 8 hits per minute drives the surface through 8 cycles per minute, regardless of what each cycle looks like. The shop usually cannot move cycle frequency without changing the job economics. Cycle frequency is mostly fixed by the production schedule.

The conclusion from the literature is the conclusion the shop floor reaches independently if it pays attention. Of the three contributors, ΔT per cycle is the one that is realistically movable. Peak temperature is the billet's problem. Frequency is the schedule's problem. ΔT is the operator's problem.