Most decisions to split a part in two are not aesthetic. They are about reach. A torch has to arrive at a joint at an angle, with the cup clear of the work and the gas flowing where it is needed, and there is a limit to how far into a closed shape a hand can go before none of that is true.
So the first question on any closed form is not how to make it in one piece. It is whether a person can physically weld the inside of it. When the answer is no, the part gets a seam, and the seam becomes the most important feature on the drawing.
A box, a can, a tank or an enclosure has an inside. If it is formed from sheet, the forming operation can produce a cup or a channel, but it cannot produce a closed volume without a joint somewhere. If it is machined, the tool can only reach from the directions the machine allows, and deep internal cavities are either impossible or disproportionately expensive.
That leaves two halves as the standard solution: two formed or machined pieces joined along a perimeter. It is a compromise, and like all compromises it moves cost from one operation to another. You gain access and lose a continuous surface.
The seam position decides almost everything downstream. Put it where it can be welded comfortably. Put it away from the highest stress. Put it where the finished part will hide it, if the part is visible. And put it where the two halves can be held in alignment while it is welded.
Those four requirements often conflict, and the resolution is usually to prioritise access and alignment, because a seam in a slightly less convenient place can still be made well, while a seam nobody can reach will be made badly. On a visible part, the seam is often moved to an edge or a face that faces away from the viewer, and the finishing operation then has to make the join disappear.
A long seam is unforgiving because any inconsistency accumulates over its length. The joint preparation has to be consistent, which usually means a machined or carefully formed edge rather than a hand-ground one. Fit-up has to be controlled along the whole length, not just at the ends. And the two halves have to be held so the gap does not open as the weld progresses.
On thin material the preparation matters even more, because there is very little thickness to absorb a variation. A seam that is tight at one end and open at the other will either be welded with a changing technique along its length or will be forced closed and distorted in the process.
Two halves welded along a perimeter will pull toward the weld. The material beside the seam contracts as it cools, and the part curves. The standard countermeasures are sequence and restraint: tack in a balanced order, weld in short runs that alternate sides, and let the part cool between passes rather than chasing the heat around the perimeter.
Restraint is the other half. Fixtures, temporary braces and tabs hold the part in the intended shape while it is welded, and are removed afterwards. The judgement is how much restraint to use, because over-constrained parts crack and under-constrained parts move.
Alignment is a separate problem from distortion. Two halves can be flat and still be offset by a small amount where they meet, which produces a step at the seam that is visible on a finished part and difficult to remove without thinning the wall.
The usual solutions are simple: dowels, a machined register, interlocking tabs, or a step formed into one half so the other sits into it. Each of these makes the part self-locating before the first tack, which is worth more than any amount of care afterwards.
A closed part welded from outside still has an inside, and that inside needs shielding on stainless and titanium. It also collects whatever the welding process leaves behind: oxide, spatter, discolouration. On a tank or a can, that residue is inside a volume that may later hold fluid.
So a split part usually needs a cleaning plan as well as a welding plan. Access for a brush or a wipe, a purged volume that can be reached, and a final check through an opening before the last piece goes on. The last closure is the one that seals everything you can no longer see.
Splitting is the general answer, not the only one. A spun or flow-formed shape can produce a closed profile without a seam. A casting or a moulding can produce internal features that fabrication cannot reach. A deep drawing can make a cup that needs only a lid rather than two halves.
The trade is tooling against labour and against the seam. For one part, a seam is usually cheaper. For many, the tool that removes the seam often pays for itself, and the part gets a continuous surface as a bonus.
If a part has to be made in two halves, the drawing should say so, and it should say where. Leaving the split to the fabricator is sometimes the right call, but only when the buyer genuinely does not care where the seam lands. When they do care, the drawing is the only place to express it.
A handful of details are worth stating outright, because none of them can be inferred from an outline.
A drawing that carries those is buildable as drawn. One that does not will be interpreted, and interpretation varies by shop.
It also helps to show the joint preparation and the register, rather than leaving both to be invented at the bench. Two shops given the same outline and no seam instruction will produce two different parts, both of which fit the drawing.
Parts that end up as two halves in our own work are usually the enclosures and housings listed under Tigindustry Swap, and the same construction appears at small scale in aluminium catch cans and at large scale in aluminium fuel tanks. The welding discipline a long seam demands is described at titanium fabricated components, the inputs we need in order to place a seam are on the customization process page, and further fabricated parts shows the smaller split components.
