Resistance spot welding clamps two sheets between copper electrodes and passes a large current through the contact interface for a short time. Joule heating at the faying surface, where contact resistance is highest, melts a lens-shaped nugget that fuses the sheets as it cools under continued electrode force:
Q = I² · R(F) · t heat generated per weld
R(F) ∝ 1 / √F contact resistance falls as squeeze force rises
d ≈ k · I · √t · √(F₀/F) nugget diameter (simplified Joule-heating model)
d_min ≈ 5·√(sheet thickness) minimum nugget for a structurally sound weld (AWS D8.1 rule of thumb)
- Current / time — more amps or more cycles pump more heat in; too little and the nugget never reaches
d_min (a "cold" weld that peels apart). Too much and the molten pool blows past the sheet surface — expulsion, weld marked bad.
- Electrode force — squeezing harder lowers contact resistance, which needs more current/time to reach the same nugget, but also suppresses expulsion by containing the molten pool.
- Weld sequence — each weld locally shrinks the sheet as its nugget cools, pulling the panel slightly toward that point. Welding sequentially (row by row) stacks that shrinkage on one side of the panel before the other, building up a lasting bow. Welding center-out symmetrically balances the pulls on both sides of the panel as you go, keeping the running distortion much lower along the way — this is why real automotive body shops plan weld-gun sequence, not just weld quality, on every panel.
This is the exact process (and the exact quality/distortion trade-off) an industrial 6-axis welding robot manages tens of thousands of times a shift on a car body-in-white line. The heat map behind the weld grid below shows the panel's out-of-plane bow accumulating in real time.