Update from our side — we're now at 3.5 years on the A-pillar line and some things have changed.
Die life has been the biggest surprise. We're getting 180K hits on our trim steels now vs the 120K we started with. The difference wasn't the steel grade (still DC53) — it was the approach velocity. We dropped from 60mm/s to 35mm/s on the trim stations specifically and the edge chipping basically stopped. The cycle time hit was only 0.3 seconds per stroke because the trim section is such a small portion of the total stroke.
sarahJ_mfg's point about die temperature is dead on. We added thermocouples in the die (4 per station, embedded 3mm below the working surface) and log them continuously. What we found: it's not the absolute temperature that correlates with springback variation — it's the temperature GRADIENT across the part. If one side of the die is 45C and the other is 62C, you get asymmetric springback that looks like a twist defect. Took us 6 months to figure that out.
Our fix was adding targeted cooling only to the hot spots rather than flooding the whole die. Saved water, saved energy, and the part consistency improved because we're controlling the gradient, not just the peak temp.
One more thing for anyone starting UHSS on servo: your blank holder force profile matters as much as your punch motion profile. We run a decreasing blank holder force through the draw (start at 100%, ramp down to 70% by end of stroke). This lets the material flow more freely as the draw deepens and eliminated the edge splits we were getting on the trailing flange.