DFM is what separates a mold that runs from a mold that fights you
Injection molding is unforgiving about geometry. A part that looks perfectly reasonable on a screen can be impossible to eject, sink in the thick spots, warp as it cools, or short-fill at the far end of a long thin wall. None of those failures show up until the steel is cut and the first shots come off the machine - by which point fixing them means reworking the tool. The discipline that prevents this is design for manufacturability (DFM): designing the part so the plastic can actually flow, pack, cool and release the way the process needs. A molding partner such as DAYIN, a plastic injection molding and OEM/ODM manufacturer that states 30 years of mold manufacturing experience, will usually flag these issues in a DFM review before cutting steel - but the buyer who arrives with a mold-ready design gets a cheaper tool, fewer iterations and a shorter path to stable production.
This note covers the five geometry decisions that cause the majority of molding problems: draft, wall thickness, ribs and bosses, undercuts and shutoffs, and gating and flow. Get these five right and the molding machine becomes the easy part of the project. Get them wrong and no amount of machine tuning rescues the part.
Draft angle: the one dimension that decides if the part ejects
Draft is a slight taper applied to every wall that runs parallel to the direction the mold opens, so the cooled part slides off the core or out of the cavity as the halves separate. Without it, the plastic grips the steel: the part scuffs, sticks, or needs ejection force high enough to distort it. The instinct to make side walls perfectly vertical 'for accuracy' is exactly what creates a stuck part every cycle. As a general rule, every wall that demolds straight needs draft, and the cost of adding it is usually a few tenths of a degree on a wall that nobody will measure.
How much depends on the surface. A polished, untextured wall can often get by on 1-2 degrees; a textured surface keyed into the mold needs more, because the texture increases grip and the part will tear or stick if the draft is too tight. Deep walls need more draft than shallow ones, simply because the contact area is larger. The table below is a working guide, not a specification - confirm the exact value with the molder against the resin and the surface finish you actually chose.
| Surface / wall type | Typical draft | Why |
|---|---|---|
| Polished, shallow wall | ~1 deg | Minimal grip; easiest ejection |
| Untextured, standard wall | 1.5-2 deg | General default for most parts |
| Lightly textured | 2-3 deg | Texture increases grip on the steel |
| Deep wall (high draw) | 3-5 deg | Large contact area needs more release |
| Knurled / heavy texture | 5 deg+ | Strong grip; generous draft required |
Read this as a starting point, not a rule carved in stone. A molder with the actual resin, tool steel and finish in hand may ask for more draft on a deep textured wall and less on a shallow polished one. The point is that draft is a design decision you make on every wall, not a number the molder invents later.
Wall thickness: keep it uniform, and never thicker than you need
Molten plastic enters at the gate, fills the cavity, and then freezes from the outside in. A thick section stays molten longer than the thin sections around it; when it finally cools it pulls in on itself and leaves a sink mark on the surface, or it traps a void inside. Thick sections also lengthen the cycle, because the part has to cool all the way through before it can be ejected - so a 'beefier' wall costs you both appearance and time. The fix is to keep walls as thin as the part's stiffness allows and as uniform as the geometry permits, because uniformity is what lets the whole part cool at the same rate.
When a part needs local strength, add a rib instead of thickening the wall - that is the next section. Typical nominal wall-thickness ranges by resin are shown below as general DFM guidance; the right number for your part depends on its size, the load it carries, and the resin's flow length. A supplier offering custom plastic injection molding services will size the wall against the resin's flow and shrink data, but the drawing still has to state which walls are functional and which are cosmetic.
| Resin | Typical nominal wall (mm) | DFM note |
|---|---|---|
| PP (polypropylene) | 1.5-2.5 | Flows well; thin walls are practical |
| ABS | 1.8-3.0 | Common; moderate flow length |
| PC (polycarbonate) | 2.0-3.5 | Higher viscosity; prefers thicker walls |
| PA (nylon) | 1.5-3.0 | Absorbs moisture; watch warp |
| POM (acetal) | 1.5-3.0 | Stiff, low and predictable shrink |
| PMMA (acrylic) | 2.0-4.0 | Brittle; thicker walls reduce brittle failure |
| TPE (elastomer) | 1.5-3.0 | Soft; watch tear at thin gates |
These ranges are starting points, not permission to pick the thin end blindly. A large part needs a thicker wall than a small one simply to carry the melt to the far end before it freezes; a heavily loaded part needs more than a cosmetic one. The discipline is to set the thinnest wall the part's size and load allow, then keep every other wall within reach of it so the part cools evenly and sinks nowhere.
Ribs and bosses: stiffness without thickening the wall
A rib is a thin wall standing proud of the main wall, perpendicular to it, that adds bending stiffness for almost no extra material and with far less sink risk - provided it follows the same draft and thickness rules as everything else. The classic mistake is a rib as thick as the wall it supports; that simply recreates the sink problem on the back side of the part. The general rule of thumb is to keep rib thickness around 40-60 percent of the nominal wall, give the rib its own draft, and radius the root where it meets the wall so the melt flows in cleanly.
Bosses - the standoffs that receive screws or locate pins - are the same idea turned into a post. A boss is a thick feature by nature, so it is the prime suspect for sink; relieve it with a cored hole down the centre, keep the surrounding wall near 60 percent of the nominal, and give it draft. Tie bosses to a neighbouring wall with a gusset rather than letting them stand alone: the gusset cuts sink, improves ejection, and stops the post from flexing. Done this way, a part gains stiffness exactly where it needs it without paying the sink and cycle-time penalty of a thick section.
Undercuts and shutoffs: when the tool has to move
An undercut is any feature that stops the part from being pulled straight out of the mold - a snap fit, an inward lip, a thread, a reverse taper. A straight two-plate mold cannot release an undercut; something has to move out of the way. That 'something' is a side action (slider), a collapsing core, or a hand-loaded insert, and each one adds cost, a maintenance item, and usually a parting line or witness mark on the part. Undercuts are where a cheap mold quietly becomes an expensive one.
The DFM move is to challenge every undercut before the quote: can the part be split differently so the undercut becomes a straight wall? Can a snap fit be replaced by a feature that demolds straight? If the undercut must stay, design it so a single, short-travel slider clears it rather than two mechanisms working at once. A molder will tell you in the quote which features force side actions, and a cleaner parting line is almost always cheaper than the mechanism it removes. This is also why an early DFM review saves money: it surfaces the undercut while it is still a free change on the drawing.
Gates, flow and the defects that announce bad geometry
The gate is where the melt enters the cavity, and its size and position decide whether the part fills, where weld lines form when two flow fronts meet, and whether there is enough pressure at the far end to pack the part solid. Geometry interacts with gating constantly: a long thin section far from the gate may short-fill, a thick section near the gate may flash, and a wall that suddenly thickens mid-span invites a sink exactly where the section grows. A good gate location is chosen against the part's wall layout, not bolted on afterwards.
Many problems blamed on 'the material' are really geometry problems. Sink, warp, short shots and weld lines are frequently solved by changing wall thickness, adding a rib, moving a gate, or balancing wall sections - not by switching resin. This is the heart of why a DFM review before tooling pays for itself: it moves the conversation from 'the mold is bad' to 'the geometry asked for trouble,' and geometry is far cheaper to change on a screen than in steel.
Shrinkage and tolerances: the molded dimension is not the drawn dimension
Plastic shrinks as it cools, and the shrink is directional - the flow direction shrinks differently from across the flow, because the melt orients as it travels. That means the as-molded dimension is the drawn dimension minus shrinkage, and it varies feature by feature depending on how the plastic reached that feature and how it was packed. Tight, uniform tolerances across a molded part are more expensive than the same tolerances on a machined part, because shrinkage and warp must be controlled at every functional feature rather than just at the cutter.
The practical rule is to set a moderate general tolerance and tighten only the functional features - sealing faces, bearing bores, snap-fit engagements - while telling the molder the exact resin and grade so shrink can be compensated in the tool. A molding operation running 40 injection molding machines (90-1200T) at its Vietnam plant established in 2024, such as DAYIN's molding operation, sizes tools against the resin's published shrink rate, but the drawing still has to state which dimensions are functional and which are not. Treat the tolerance scheme as part of the DFM, not an afterthought on the title block.
A DFM checklist to run before you cut steel
Before the file is released for tooling, run this list: every wall that demolds straight carries draft; walls are uniform and as thin as stiffness allows; ribs are at or below 60 percent of the nominal wall and have their own draft; bosses are cored and tied to a wall with a gusset; undercuts are challenged and, where kept, cleared by one short slider; wall-thickness changes are gradual rather than abrupt; the gate location is set against the wall layout; and functional tolerances are called out while the general tolerance stays moderate. Each item is a few minutes on the drawing and potentially a tool rework if skipped.
Run the list and a molding partner quotes a tool that runs from the first week, not a tool that needs three rounds of correction. The geometry - not the machine, not the resin, not the molder's skill - is where molding cost is won or lost. A part designed to mold is cheaper to make, faster to stabilize, and far less likely to surface a sink or a warp at the worst possible moment, which is the entire point of doing DFM before steel is cut.
Frequently asked
How much draft do I really need on a molded part?
As a general rule, 1-2 degrees on a polished, untextured wall is enough; textured or deep walls need more because the surface grips the steel. The cost of adding draft is usually a fraction of a degree on a wall nobody measures, while omitting it causes sticking, scuffing or distortion at ejection. Confirm the exact value with the molder against your resin and finish rather than treating a single number as universal.
Why does my thick plastic part sink even though the resin is fine?
A thick section cools last and pulls in on itself as it solidifies, leaving a sink mark or an internal void - this is a geometry problem, not a material one. The fix is to keep walls thin and uniform and to add ribs for stiffness instead of thickening the wall. Thick sections also lengthen cycle time because the part must cool through before ejection.
Can I mold a thread or a snap-fit feature?
Yes, but a thread or inward lip is an undercut that a simple mold cannot release, so it forces a slider, a collapsing core or an insert - each adding cost and a maintenance item. Challenge the undercut first: often the part can be split or the feature redesigned so it demolds straight. Keep any necessary undercut cleared by a single, short-travel slider to limit cost.
My molded part warps - is that the machine's fault?
Usually not. Warp is most often directional shrinkage from wall sections that cool at different rates, or from a tolerance scheme that did not account for shrink. Balance wall thickness, call out only functional tolerances tightly, and let the molder compensate shrink in the tool against the resin's published rate. Geometry and tolerancing, not machine tuning, resolve most warp.
When should I involve the molder in the design?
Before the steel is cut. A DFM review at the drawing stage catches draft, wall, rib, undercut and gate problems while they are free to change; after tooling, the same fixes mean reworking the mold. Bringing a mold-ready design to the molder also produces a tighter, cheaper quote because the tool's risk is already designed out.