We've talked about runout, warp, waviness, and form error on V-band flanges. Now let's cover parallelism — the geometric relationship between the sealing face and its reference datum (usually the back face of the flange or the pipe centerline). Parallelism is angular error: does the sealing face lie in a plane truly perpendicular to the pipe axis, or is it tilted?

Think of it this way: axial runout measures how much a single point on the sealing face moves as you rotate the flange. Parallelism measures whether the entire face is cocked at an angle relative to the pipe centerline. A face can have zero axial runout at four measured points but still be tilted 0.3° if the whole plane is skewed. Parallelism catches the tilt that spot-checks miss.

Why it matters for V-bands: The clamp's 20° wedge angle assumes both mating flanges present their sealing surfaces on planes perpendicular to the pipe axis. If one flange is tilted 0.5°, the clamp band pulls the flanges together but the sealing surfaces don't meet flush — they contact heavily on one side and gap on the opposite side. The clamp can't correct angular error; it just amplifies the local contact stress on the high side while leaving the low side to leak.

Real-world example: A fabricator TIG-welds a stainless V-band flange onto a 2.5" downpipe. Heat input from a single-pass weld pulls the flange toward the hot side as it cools, tilting the sealing face by 0.4° relative to the pipe centerline. On a 2.5" flange (roughly 3.5" OD sealing ring), that tilt produces about 0.024" of gap on the cold side — enough that even properly torqued V-band bolts can't close it. Result: an audible boost leak that shows up under load, disappears at idle (thermal expansion closes the gap temporarily), and drives the tuner crazy.

Rule of thumb: Parallelism tolerance for turbo-side V-band sealing surfaces should be held to 0.05mm (0.002") per 25mm of flange diameter. For a 3" (76mm) flange, that's roughly 0.006" total. Anything looser and the wedge-clamp geometry can't reliably close the joint. Precision-machined billet flanges hold this easily; welded-on flanges need post-weld facing on a lathe to guarantee it.

Measurement: Set the flange back face on a granite surface plate, sweep the sealing face with a dial indicator on a height gauge. The difference between the highest and lowest reading across the full face is the parallelism error — not just runout at a single radius.