You've learned about radial runout (wobble side-to-side) and axial runout (warp front-to-back). But there's a third geometric error that kills V-band seals: circumferential runout — the flange isn't round. It's slightly oval, egg-shaped, or lobed around its perimeter. The 20° sealing cone stays perpendicular to the axis and centered on the shaft, but the radius from center to sealing surface varies as you sweep around the flange.

Why it matters: a V-band clamp applies uniform radial clamping force through its retaining ring. If the flange radius varies by 0.15mm around the circumference, the clamp's V-groove contacts the high spots first, deflects the retaining ring outward there, and leaves the low spots under-clamped. You get a leak path at the "valleys" of the out-of-round profile, even though the axial and radial runout measure perfectly on a dial indicator sweep.

Where circumferential runout comes from:

• Welding distortion in a non-uniform pattern — a header primary that welds to the flange at 3 and 9 o'clock pulls those regions inward, leaving 12 and 6 o'clock proud. Result: oval flange.
• Chuck jaw imprinting during machining — a three-jaw chuck grips at 120° intervals. Grip too tight, machine the sealing cone, release the part, and the flange springs back into a three-lobed profile (trilobal).
• Thermal cycling after installation — uneven cooling from a nearby turbo or heat shield can plastically deform thin flanges over time.

Real-world example: A Garrett G-series turbo inlet V-band flange (3.0" nominal, ~76mm sealing diameter) spec'd at ±0.05mm circumferential roundness. Aftermarket flanges from budget suppliers routinely measure 0.20-0.30mm out-of-round because they're plasma-cut from plate and stress-relieved poorly. Guys blame V-band clamps for "always leaking" — the clamp is fine, the flange is oval.

Measurement rule of thumb: Roundness error should stay under 0.1% of the sealing diameter. For a 76mm flange, that's 0.076mm max total indicated reading (TIR) sweeping a dial indicator around the sealing cone at a fixed axial position. Anything over 0.15mm and you're chasing leaks with more clamp torque, which just distorts the flange further.

Fix: Post-weld machining on a lathe with a soft-jaw chuck (or expanding mandrel) that grips without imprinting. Cut the sealing cone after all welding is complete, not before. Pros machine flanges 0.5mm oversized, weld, then finish-machine to spec.