Every cutting operation that pushes metal instead of shearing it cleanly leaves something at the exit edge. The material does not vanish where the tool leaves the part. It folds over, tears or extrudes, and it stays attached or falls off depending on how thin the remaining web was. That is a burr, and it is not a mistake that someone made. It is the default outcome of the cut.
This matters because deburring is treated as the last five minutes of a job, done by whoever is free, with whatever tool is nearest. That treatment is fine on a bracket nobody will look at. On an exhaust component it is not fine, because the burr is inside a path that carries hot gas and a loose fragment at speed, and because the same edge is often the one an inspector runs a finger along and calls good.
A burr that stays attached is a surface irregularity. It disturbs flow locally, it can act as a stress raiser at the edge, and in a high-temperature stainless part it is a thin section that heats and cools faster than the material around it. A burr that detaches is a different problem entirely, because it becomes a small hard particle travelling through the system with nowhere to go but into a turbocharger wheel, a catalyst substrate or a valve seat.
Nobody designs for that particle, and it is rarely traced back to the bench where it came from. The part passed inspection, the weld was good, the fit was right, and something failed downstream later. The most expensive defects in exhaust work are usually the ones that were never visible on the part at all, and a burr inside a tube is exactly that kind of defect.
A third effect is on the joint itself. A burr left at the mouth of a slip joint changes the fit, so the tube either will not enter fully or enters with a gap that is not uniform around the circumference. That gap becomes a leak, and the leak is blamed on the clamp.
A rolled burr is the common one on a drilled hole: a thin lip that folds over and stays attached on one side. It is a sign the drill was sharp enough to push rather than cut, and it comes off easily with a hand tool. A torn burr is rougher and heavier, and it usually means a dull edge, a feed that was too light, or a material that work hardened under the tool.
A pooched exit is different again. Where a drill breaks through into a cavity, the last of the material is pushed out rather than cut, and the edge around the hole is pushed outward with it. On thin-wall tube that can mean the wall around the hole has been deformed, not just burred, and deburring will not put the wall back.
Reading the burr tells you what happened in the cut. A shop that looks at burrs as information rather than as cleanup work catches tool wear early, and catches the wall deformation before the part is welded.
A deburring blade, a scraper and a flap wheel will handle most external edges quickly, and there is nothing wrong with them. Their limit is consistency. Two people with the same tool will remove different amounts, and the same person will remove different amounts on a Friday afternoon than on a Monday morning. On an external cosmetic edge that does not matter. On an edge that has to meet a mating part, it does.
Their second limit is access. A hand tool cannot reach the far side of a cross-hole in a tube, or the inside of a small bore, or the root of a groove. Those edges are where the burrs that matter actually live, and they are the ones that get left because they cannot be seen from the operator's position.
A cross-hole is the classic case. The drill enters the outside of the tube, breaks through the inner wall, and leaves a burr on the inside surface that no external tool can reach. If the hole is for a sensor boss that will be welded on, the burr sits inside the gas path behind a weld. If the hole is a drain or a take-off, the burr sits in the flow.
There are only a few honest answers. Deburr from the inside with a tool that can reach, which requires access from an open end. Use a cutter geometry that produces a cleaner exit, which helps but does not eliminate the burr. Or design the hole so it does not break through into a critical volume at all. You can see the kind of small welded and machined hardware where this decision comes up on our aluminium welding pages.
Inspection looks at what it can see and measure. External edges appear on a drawing as a note, and a note is checked by eye. Internal edges appear on no drawing at all, because the drawing shows the hole and not the material that was pushed through it. So the burr is outside the inspection plan by default rather than by decision.
The practical fix is not more inspection, it is a process step with a name. If deburring is a line on the route card with a specified method and a defined check, it happens. If it is a sentence in a general note, it happens when the bench is quiet. The difference between those two states is organisational, and it costs nothing to change.
The strongest version of the step is a rule about when it must happen, not how. Deburr before welding, not after, wherever the edge will be inaccessible later. Deburr before assembly, not at dispatch, because at dispatch the part is packed and the decision is already made. And never deburr a sealed assembly through a hole, because the debris that goes in does not come out.
The second part of the rule is about the tool. A fixed scraper, a blade holder set to a consistent angle, or a dedicated deburring tool for each hole size removes the operator variable without removing the operator. None of that requires capital equipment, and all of it produces a part that behaves the same way twice.
Deburring is invisible in a quotation and visible in service, which is exactly the wrong way round. It is worth naming explicitly.
You can see the hardware these edges belong to on our machined and fabricated parts pages, the second group of small components, the brackets and adapters, our titanium fabrication components and the titanium welded parts that follow the same route.
