Design for injection moulding (DFM) comes down to four disciplines that a 3D-printed part never forces you to think about: give every vertical face draft so the part ejects, keep walls uniform so plastic cools evenly, size ribs at roughly 40–60% of the nominal wall to add stiffness without sinking the surface, and cored-out bosses so screw towers do not become thick blobs that suck in a visible dimple. Get those four right and the tool runs; get them wrong and you pay for it in sink marks, warp, short shots and reworked steel. At Layer X we produce hybrid metal injection mould inserts from our Ahmedabad facility with a ±0.05mm cavity tolerance, and the cavities that reach us clean are almost always the ones whose CAD respected these rules before a single shot was fired.
This guide is deliberately mould-specific. Our additive-manufacturing DFM library covers overhangs, supports and layer orientation — none of which apply once you are cutting a steel cavity. Injection moulding is governed by melt flow, differential cooling and a physical draw direction, and those constraints reshape how you draft, wall, rib and boss a part from the ground up.
Why moulding DFM is a different discipline
An injection-moulded part is formed by forcing molten polymer into a closed steel cavity, holding pressure while it solidifies, then opening the tool and ejecting the part along a single axis — the draw direction. Three physical facts drive every design rule that follows. First, the part shrinks as it cools, and it shrinks more where there is more material. Second, the melt must reach every extremity before it freezes off, so flow length and wall section are linked. Third, the part has to leave the steel, which means every surface parallel to the draw must taper.
The reference points here are well established. The ISO 294 series governs the moulding of plastics test specimens and the reporting of shrinkage, and material suppliers publish mould-shrinkage figures on every datasheet. The Boothroyd Dewhurst DFMA methodology — the origin of the modern "design for manufacture and assembly" term — quantifies how feature choices drive tool cost. You do not need to memorise a standard to design a good part, but you do need to design as though cooling and ejection are real forces, because in the tool they are.
Draft angles: the non-negotiable taper
Draft is the taper applied to any face that runs parallel to the draw direction, and without it the part grips the core as it shrinks and either scuffs, stress-whitens or refuses to eject. The widely used engineering convention is a minimum of one degree of draft per side on smooth surfaces, with more added as depth and texture increase.
Practical starting points we apply when reviewing a cavity:
- Smooth, shallow walls: 1° per side is the accepted floor; 1.5–2° ejects more reliably and is cheap to add in CAD.
- Deep ribs and tall cores: increase draft with depth — a deep rib on 1° can bind, so 1.5–2° is safer.
- Textured surfaces: add roughly 1° of extra draft for every 0.025mm of texture depth. A moulded grain that looks fine on a flat sample will drag badly on an under-drafted wall.
- Shut-off faces and appearance A-surfaces: draft the whole face consistently so the parting line sits where you intend, not where the mouldmaker is forced to put it.
The cost of missing draft is not abstract. An under-drafted cavity means polishing the steel in the draw direction, hand-benching to break the grip, or in the worst case re-cutting the insert. On our injection tooling service we flag zero-draft faces before we quote, because it is far cheaper to add a degree in CAD than to correct it in H13.
Uniform wall thickness: the master rule
If you remember one thing about moulding DFM, make it this: keep the wall uniform. Plastic shrinks as it cools, thick sections cool slower than thin ones, and any change in section creates a race between two cooling rates. That race produces the three classic defects — sink marks over thick spots, warp as the part pulls toward the section that cooled last, and internal voids where a thick core solidifies from the outside in.
Nominal wall thickness is material-dependent, driven by melt viscosity and flow length. The figures below are the standard design ranges published across polymer supplier DFM guides and are a sound starting point before you tune to a specific grade and flow length:
| Polymer | Typical wall range | Notes for the mould |
|---|---|---|
| Polypropylene (PP) | 0.8–3.0 mm | Forgiving flow; tolerates thin walls well |
| ABS | 1.0–3.5 mm | Good surface finish; watch sink on bosses |
| Polycarbonate (PC) | 1.0–4.0 mm | Needs generous radii; sensitive to stress |
| Nylon (PA6 / PA12) | 0.8–3.0 mm | High shrinkage; uniformity is critical to control warp |
| POM (Acetal) | 0.8–3.0 mm | High, consistent shrinkage — avoid thick sections |
| PC/ABS blend | 1.0–3.5 mm | Balanced; a common enclosure default |
Two working rules keep walls honest. Keep every transition within about 15% of the nominal wall wherever you can. And where a section change is unavoidable, blend it with a gradual ramp rather than a step, so the flow front and the cooling gradient both stay smooth. All of the tooling materials we run — H13 tool steel for 10,000-plus shot production inserts, Aluminium 7075 for short-run soft tooling, and copper alloy for fast-cycle high-conductivity inserts — reward a uniform wall, because uniform cooling is exactly what conformal cooling channels are built to deliver.
Coring out: turning thick into uniform
The instinct to draw a part as a solid block is the single biggest cause of sink and long cycle times. The fix is coring — removing material from the underside of a thick feature so the wall around it stays uniform. A solid handle, knob or thick base becomes a shelled feature with ribs for stiffness. Coring does more than kill sink: thinner walls cool faster, so the cycle time drops, and less polymer per shot lowers the piece part cost. Every thick region on a part should prompt the same question — can this be cored to the nominal wall and stiffened with ribs instead?
Rib design: stiffness without sink
Ribs are how you recover the stiffness you lost by coring out thick sections, and their geometry is governed entirely by the wall they attach to. Make a rib too thick and it reads through the opposite face as a sink line; too tall and thin and it will not fill or will buckle on ejection.
The standard rib proportions, relative to the nominal wall thickness (T):
- Rib base thickness: 40–60% of T. Stay at the lower end for high-shrinkage, gloss-surface materials where any sink shows; 60% is acceptable on textured or hidden faces.
- Rib height: up to about 3× T. Taller ribs need more draft and risk filling problems.
- Rib draft: at least 0.5–1° per side, more on deep ribs so they release from the core.
- Base fillet: a radius of roughly 0.25–0.5× T at the rib root to reduce the stress concentration — but no larger, because an over-generous fillet rebuilds the thick section you were trying to avoid.
- Spacing: space ribs at least 2× T apart so the steel between them is not a thin, fragile blade that is hard to cool and easy to damage.
A grid of correctly proportioned ribs will out-stiffen a thick wall at a fraction of the material, the cycle time and the sink risk. This is the same stiffness-per-gram thinking behind the topology work in our DMLS metal printing service — different process, identical principle: put material only where the load path needs it.
Bosses: cored towers, not solid pegs
A boss is a cylindrical stand-off for a screw, pin or press-fit, and it is the feature designers most often get wrong by drawing it as a solid post. A solid boss is a thick mass hanging off a thin wall — a guaranteed sink mark on the opposite face and a local hot spot that slows the cycle. Design bosses as cored towers instead.
The proven boss recipe:
- Wall thickness: the boss wall should follow the same 40–60% of nominal-wall rule as a rib, so it does not become a thick section in its own right.
- Cored hole: core the centre to suit the fastener — a self-tapping screw wants a hole a defined amount under the thread diameter per the screw supplier's boss chart, so the plastic forms the thread.
- Support ribs, not thick walls: stiffen a tall boss with gussets or connecting ribs to a nearby wall rather than by fattening the boss.
- Standoff from walls: if a boss sits against a wall, connect it with a rib and leave a small gap rather than merging the two into one thick lump.
- Base radius: a small fillet at the boss base for strength, kept small enough not to recreate a thick junction.
Corners, gates and the details that finish the part
A few remaining rules tie the discipline together. Never leave a sharp internal corner: sharp corners concentrate stress and choke melt flow, so radius every internal corner to about 0.5× the wall thickness and let the outer radius follow so the wall stays uniform through the bend. Think about the gate position early — plastic should flow from thick to thin, never thin to thick, or it will freeze off and short-shoot. And design the parting line deliberately, because if you do not choose where the tool splits, the mouldmaker will choose for you, and it may land across an appearance surface.
When these rules conflict — and on a real part they will — uniform wall wins. A slightly awkward rib layout that preserves a uniform wall beats an elegant one that introduces a thick junction. Cooling is the force you are designing against, and uniform section is how you keep it under control. It is the same first-principles discipline we apply when validating a geometry with bridge tooling before committing to a production mould, exactly the workflow described in our guide to choosing between FDM, SLA and SLS for prototyping ahead of moulding.
From DFM to a running tool
Good moulding DFM is not about hitting every number on this page perfectly. It is about designing with cooling and ejection in mind so that draft, wall, rib and boss all pull in the same direction. Do that and the tool fills cleanly, ejects without drama, and holds tolerance shot after shot — which is exactly what bridge tooling is meant to prove before a production commitment. At Layer X our hybrid inserts run at 60% of the cost of CNC-machined tooling with a 7–14 day lead time, and first-run injection-moulded samples are included in every order, so you see real parts in your specified material before scaling.
Ready to move from CAD to a running cavity? Upload your CAD file for a 24-hour quote and our team will flag any draft, wall or rib issues in the same review — before we cut steel.