The usual way to think about a leak in an intake is that some air escapes. That is the wrong picture for most of the leaks that actually exist on a car. A leak is a parallel path, and air takes it because it is easier than the path through the filter, the duct and the turbo. The total quantity entering the engine barely changes. What changes is where it came from.
That distinction matters because it changes what the symptom looks like. A leak between the filter and the compressor does not usually produce an obvious whistle or a dramatic loss. It produces an engine that is slightly down on power, slightly worse on fuel, and drawing a proportion of its air from the hot, dirty engine bay instead of from the cold source the duct was built to reach.
A gap between two clamped parts is a short passage with sharp edges at both ends, and flow through it contracts twice. The stream narrows as it enters and again as it leaves, so the area that actually passes air is a fraction of the area you can see with the clamp loosened. A gap that looks generous is often a genuine restriction, which is why a small unsealed joint can matter more than its apparent size suggests.
The reverse is also true, and it is the more common mistake. A joint that looks well closed can flow a surprising amount if its edges are clean and the passage is short and straight. Judging a seal by eye, with the parts assembled and nothing flowing, tells you almost nothing about either case.
They are rarely in the middle of a straight run. Leaks concentrate at joints, at transitions between materials, at penetrations through panels, and at anything that has been removed and refitted. A silicone hose pushed onto a pipe that is slightly too small seals on the clamp line but not beyond it, so the seal depends entirely on distortion of the rubber rather than on contact along the spigot.
The other cluster is at the ends of the system: the lid of a housing that has been opened, the clip that was reused, the grommet that was stretched over a connector, the port that was left open after a modification. Those are the places where a build acquires an air path that nobody designed and nobody can see.
A worm-drive or T-bolt clamp does not seal by friction. It seals by compressing the elastomer against the spigot until the contact pressure exceeds the pressure trying to escape. Everything about the joint follows from that. The clamp needs to sit squarely over the section it is compressing, the hose needs enough wall thickness to deform, and the spigot needs a bead or a shoulder for the hose to seat against.
Over-tightening is the common error because it feels like the safe direction. Past a point the clamp cuts into the hose or distorts it into an oval, and an oval section cannot maintain contact pressure all the way round. The result is a joint that is tighter than it should be and leaking at two points on the circumference.
Under-tightening is easier to find, because the joint moves. A hose that can be rotated by hand after the clamp is done up is not sealed against pressure, whatever it does at idle. Checking that with the system at rest costs nothing and catches the majority of assembly errors.
Where two pipes of different diameter are joined by a hose, the hose has to seal on two surfaces at once and bridge the step between them. On the smaller pipe the hose is not compressed against anything, so the clamp has to compress it far enough to close the gap, which produces an uneven section and a poor seal at the transition.
The fix is a spigot, an insert or a swaged end that brings both pipes to the same outside diameter at the joint. That is a fabrication decision, not an assembly one, and it is the kind of detail that separates a duct that stays sealed from one that needs attention every time it is disturbed.
Upstream of the turbo, the duct runs at a slight depression and at ambient temperature. Joints there tend to fail by sucking in, which pulls the hose tighter onto the spigot and can make a leaking joint look sealed. The symptom is a whistle at small throttle or a fuel trim that drifts as the engine bay heats up.
Downstream, the duct runs at boost pressure and at high temperature, and the failure is the opposite. Pressure pushes the hose off the spigot, and heat softens the elastomer so it creeps and loses compression over time. A joint that was sound when built can become a leak months later without anything being touched, which is why the boosted side deserves a check at service rather than only at build.
Modern engine management compensates. A small unmetered air path shows up as a fuel trim correction, and the correction keeps the engine running well enough that the driver notices nothing. The leak is still there, still admitting hot air, and still reducing the margin the calibration was written with.
That is why the leak has to be found by a method that does not rely on the engine running badly. Pressure or smoke introduced into a sealed section, with the openings capped, shows the path directly. Watching live trim data while spraying a suspected joint is the cheaper version, and it works because the trim moves the moment the joint is wetted.
Sealing assumptions break down wherever the parts move relative to each other. An engine rocks on its mounts, a chassis flexes, and a duct that spans between the two has to absorb that movement somewhere. A joint that is perfectly sealed in a static test can open under load, and no amount of clamp torque fixes a geometry that is asking the hose to be a structural member.
They also break down at very low restriction. Once a duct is properly sealed, the remaining loss is distributed along its length rather than concentrated at a joint, and further work on the joints returns nothing. Knowing when that point has been reached is what stops the job.
Sealing is decided at the drawing stage more often than at the assembly stage.
The flexible sections we make are on our air intake hoses pages, the rigid path around them under air intakes, and the filter end under air filters and prefilters. What the sealed air then passes through is shown under intercooler and intercooler airflow, with the measuring method under intercooler pressure testing.
