A riser is not an accessory added to a casting because the foundry has always added one. It is a reservoir of liquid metal, positioned so that it stays liquid longer than the section it is feeding, and it supplies the contraction that occurs in that section as it solidifies. If the riser freezes before the section does, it has done nothing. If it is in the wrong place, it feeds a region that did not need feeding.
This makes riser position a geometric question before it is a process question. The riser has to be adjacent to the last region to freeze, connected by a path that stays open, and large enough to hold the volume that will be drawn from it. Get the position wrong and no amount of riser size will compensate, because the metal cannot travel through a passage that has already solidified.
The consequence for the drawing is that the location of heavy sections and junctions has to be known before risers can be placed. A designer who specifies the part without indicating where the mass sits has left the foundry to discover it, usually by cutting up the first casting.
The thermal map of a casting is the picture of which regions stay liquid longest. Those regions are the ones that need feeding, and they are usually at junctions, at heavy bosses and at the far end of a thick section. The riser goes there, or as close to there as the geometry allows.
Distance matters because the feeding path has to remain liquid. A riser placed on the opposite side of a thin wall from the hot spot is connected through a passage that freezes early, and the hot spot is isolated. This is why risers are often placed on the heavy section directly rather than on a convenient flat nearby.
Where a hot spot cannot be reached from outside, the answer is often to change the geometry rather than to add a riser. Reducing a heavy section, or moving it so that it can be fed, is frequently cheaper than a complex feeding arrangement that only partly works.
There is also the question of how many. One large riser may not reach the far end of a long section before the path freezes, while several smaller ones can each feed a local region. The choice separates a foundry that understands feeding from one that follows a rule of thumb.
Every riser is metal that will be cut off and remelted. It adds weight to the casting as poured, consumes melting energy and requires a removal operation. It also leaves a mark on the surface where it was attached, which may need grinding. On a part where the riser lands on a machined face, it can leave a region of altered structure that has to be cut back.
This cost is real, which is why risers are sometimes reduced beyond the point where they still work. The saving is immediate; the cost appears later as internal shrinkage found at inspection, or as a leak in service.
The correct approach is to treat riser volume and position as designed features, agreed between designer and foundry, rather than as a foundry detail to be minimised. If the riser has to be large, that is information about the part, and it may be worth changing the part so that it does not.
The drawing does not need to show risers. That is foundry work. But the drawing does need to give the foundry what it needs to place them: the location of heavy sections, the surfaces that must remain sound, and the surfaces that will be machined. Without those, the foundry is guessing at the intent behind the geometry.
It is also useful to state which regions are pressure tight and which are structural only. A riser decision on a region that must hold fluid is different from one on a bracket that only carries load. The acceptance criteria and the feeding plan should be consistent with each other.
None of this removes the foundry's responsibility for the feeding plan. It gives the foundry a basis for it. The difference between a specification and a wish is whether the information needed to act on it is present.
Thin walled components with isolated heavy features are the hardest case. The mass is concentrated in a small volume that is surrounded by metal that freezes first, so the feeder has to reach a target that is thermally isolated. This is where the geometry and the feeding plan have to be designed together, and where a small change to the part can turn an impossible feeding problem into a straightforward one.
Components that will be fabricated instead, such as the parts in the custom parts range, sidestep the feeding question but inherit a welding question in its place. Neither route is free of process risk. The choice is about which risk can be controlled more reliably at the required volume, and that is a decision worth making explicitly rather than by default.
A riser is therefore not optional, but it is also not a solution on its own. It is the last step in a chain that begins with section design. When the chain is sound, the riser is unremarkable and the casting is sound. When the chain is broken, the riser is where the blame lands, which is unfair and also unhelpful.
The useful question at drawing stage is not where the riser goes. It is where the metal will be liquid last, and whether anything can reach it. Answer that, and the riser position follows.
A riser leaves a scar where it was attached, and the scar has to be dressed. Attaching it on a cast surface rather than a machined face keeps that work simple, as does the TIG aluminium welding route, where a bead is dressed rather than cut away.
The first casting tests the plan rather than the part. If shrinkage sits where it was predicted the plan holds, and if not the geometry or the riser position changes before production. aluminium fuel tank range and aluminium water tank range shells are reviewed the same way.
Thin walls give a feeder little margin, and the feeding point is often the only place where mass can be added without distorting the part. The TIG stainless steel welding and titanium fabricated components routes show what the alternative looks like when that margin is absent.
