An open joint can move. Weld one edge of a bracket and the far end lifts, the part accepts a new shape, and the job continues. A closed shape cannot do that. When the joint runs all the way around a tube, a flange or an end cap, the shrinkage has to resolve inside the ring, so the ring either closes in, opens out, or goes oval. Nothing escapes.
That is why tacking a closed shape is a different exercise from tacking a seam. On a seam you are holding two edges together long enough to weld. On a ring you are also setting the final diameter, the final roundness and the final position of every feature that is measured from that ring. The tacks are the last chance to influence all three at once.
The order of those tacks matters as much as their size. A ring tacked straight across from one point to the opposite point, then closed up in sequence, pulls itself into an oval before the first full bead is laid. The same ring, tacked in a balanced pattern, stays round enough that the weld only has to close a small gap.
A tack is a small weld with a large job. It has to carry the fit-up load while the part is handled, resist the shrinkage of the first full pass, and then be consumed by that pass without leaving a defect. Those are three different requirements, and the common failure is to satisfy the first and ignore the other two.
A tack that is too small cracks when the part is lifted or when the first bead pulls against it. A tack that is too large becomes a local lump of extra metal that the final pass has to melt through, and it distorts the ring at the exact point where you were trying to hold it. Both failures look the same afterwards: a mark on the finished surface and a repair.
The practical target is a tack that is long enough to resist the pull and thin enough to be fully remelted. On thin-wall tube that means a short, smooth tack with no crater. On heavier flanges it means a tack with enough throat to survive handling but still small relative to the finished leg.
Symmetry is the usual advice and it is only half right. Two opposite tacks balance the pull across the diameter, but they leave the other axis free, so the ring can still go oval at right angles to the tack line. What actually holds a circle is a pattern that constrains both axes early.
A workable habit is to place the first pair on one axis and the second pair on the other, then fill in between the four. Each new tack is placed where the gap is largest, not where the torch happens to be pointing. The welder walks the ring, checks the gap at the next position, and tacks the widest opening first.
On a large flange with bolt holes, the pattern follows the holes. Tacking between holes rather than on them keeps the hole circle true, and it means any local pull is shared between two bolt positions instead of dragging one hole out of line. That single habit removes most of the bolt-hole alignment complaints that arrive after welding.
A welder who has spent years on heavy plate will tack a thin tube the way he tacks plate, and the ring will move. The tack has to be scaled to the material it sits on, because the same deposit that is a small dot on thick stainless is a stiff lump on thin titanium.
Length and throat are the two variables worth controlling. Length sets how much the tack resists bending along the seam. Throat sets how much it resists the seam opening. On thin wall both can be modest, because the section itself is flexible and a stiff tack simply becomes a hard spot that the bead has to fight.
Every tack adds metal to a ring that is already closed, so the circumference grows slightly as you tack around it. If the part was fitted with a tight butt gap, the last tacks have to close a gap that has been consumed. Welders experience this as the ring getting stubborn near the end of the tack sequence, and they compensate by forcing the last edges together.
Forcing them is where the ovality starts. A better approach is to fit with a slightly generous gap and let the tacks take it up, checking the diameter at two axes after every few tacks. The diameter reading is the early warning; once the ring is oval, the full pass locks the shape in.
On a long body with several rings, tack one ring at a time and verify before moving to the next. Tacking every ring loosely and then welding all of them in one run removes the chance to correct anything, and it turns a series of small controllable movements into one combined error.
The first failure is a tack that bridges a gap instead of closing it. A tack laid across an open gap has no root fusion on one side, and the full pass simply buries the defect. It passes a visual and fails later under vibration, which is the worst possible combination for an exhaust part.
The second is a tack placed on a faying surface that has not been cleaned. Titanium and stainless both punish this immediately with discolouration, and the contaminated tack then contaminates the pass that consumes it. Cleaning the joint before tacking is cheaper than cleaning a repair.
The third is tackling the ring on the bench and then bolting it to a fixture that pulls it out of shape before welding. The tacks were set to hold one geometry and are now holding a different one under load. If the part is going into a fixture, tack it in the fixture.
On a flange-to-tube joint the tube is the flexible member and the flange is the stiff one, so the tube moves toward the flange. Tacking the flange flat to the bench and letting the tube find it produces a straighter assembly than tacking the tube first and pulling the flange to it.
For an end cap on a canister, the cap edge is usually the thinnest section in the joint and the most likely to wave. Tacks spaced evenly around the cap, each pulled tight before it is laid, keep the cap edge in contact. Tacking one side fully and then working around leaves a wave that no amount of welding will flatten.
When the joint includes an internal baffle or a tube entering the side, tack the baffle first. It stiffens the ring before the outer seam is closed, and a stiffened ring moves less. The order of operations inside a closed shape is a stiffness decision, not a convenience decision.
Most tacks are meant to be consumed by the final pass, and that is the cleanest outcome. A tack that sits outside the finished bead, on the visible face of a titanium part, has to be dressed back, and dressing thin-wall titanium is a good way to thin the section you were trying to protect.
Where a tack cannot be consumed, place it where it will be hidden or where the finished surface allows blending. On polished work this is decided at the drawing stage. On a part that will be brushed, a tack on the reverse face is invisible and costs nothing.
Never remove a tack by grinding it flush and welding over the scar without cleaning. The ground area holds abrasive and oxide, and the pass that follows will show it. If a tack must come out, cut it, clean the site, and re-tack.
For a closed welded assembly, the specification has more influence on roundness than the welder's hand. A few points settled at the quotation stage prevent most of the arguments that follow a dimensional check.
Tube and pipe sections we produce are listed under exhaust pipe, and the way a custom ring or flange is drawn and approved is set out on the customization page. Welded assemblies that combine rings, brackets and fittings appear under other parts and TIG titanium welding parts, with material options on shop by material. The plant that builds them is described on about us.
