Mold Design for Injection Molding: Four Decisions Buyers Should Review, from Gate Position to the Weld Line on Your Logo

An injection mold design is settled in a room the buyer is rarely in. The tooling engineer opens a 3D model, the salesperson says the design is standard, and somewhere in the attachments sits a 2D layout with six dimensions on it. Nine weeks later a tool arrives that makes parts, and a set of decisions nobody on the buying side ever looked at has become permanent.

Four of those decisions are worth reviewing, because they are the ones that cannot be corrected afterwards without cutting steel: where the gate sits, how much draft the walls carry, where the two halves part, and where the weld line lands. Everything else on the drawing is either derived from these or can be changed later with a polish, an insert or a process setting. These four change the tool.

This page is about mold design for injection molding from the buyer's side: it is for whoever signs the tooling purchase order, not the person who runs the press. It assumes a polypropylene food container or lunch box, and stays with what can be written down before the first cut.

The Drawing You Approve Is Not the Mold You Get

A 2D layout usually shows the part outline, the cavity count, the overall size and a handful of dimensions. It does not show the gate unless somebody dimensioned it. It does not show draft unless somebody drew the taper. It does not show the parting line at all, because the parting line lives in the toolmaker's 3D model and only surfaces on the drawing as a note, if it surfaces at all.

That gap is where the money goes. Moving a gate after the first trial means welding the old gate shut and sinking a new one in a hardened cavity, then re-polishing the face around it. Adding draft to a wall that has none means re-cutting that wall, and if the wall sits on a shut-off, re-cutting the shut-off too. Moving a parting line means rebuilding the shut-offs on both halves. On paper each of those is a line on a screen; in steel each is a week and an invoice.

The quote is a separate conversation: what the tool costs, how many cavities it should have and who owns it are settled in the seven lines that decide a tooling quote. This page starts one step later, where the price is agreed and the geometry is still open.

What the Four Decisions Actually Control

Before the detail, the map: each decision fixes something specific in the part, and each lands on a different desk when it goes wrong.

Three of the four are geometry. The fourth is an outcome of the first: the weld line forms where the flow finishes, and the flow finishes where the gate put it. That is why gate position carries more leverage than the other three combined — and why it is the decision most often left to the toolmaker's default.

Whether the product needs a new tool at all is a different decision, worked through in what can change on an existing mold and what needs a new one. The four below assume the tool is being cut.

Gate Position: Where the Plastic Enters Decides Where the Mark Stays

Every mold has a gate; it is the opening through which melt enters the cavity. In injection molding, gate location is the decision that places that opening on one face of the part rather than another. In the most common construction for a container, an edge or tab gate attaches to the part along its perimeter at the parting line of the mold, and the point where it is trimmed off leaves a small witness mark — the vestige.

Two rules pull in opposite directions on a food container.

The first is mechanical. Beaumont Technologies' molding glossary states it plainly: gates "should be located at the thickest area of the part, so that the part does not have defects, such as hesitation." The same logic runs the other way: too small a gate restricts packing. A gate that sits at a thin wall also freezes before the thick section downstream has finished packing, which is how a sink mark appears on the face opposite a boss.

The second is cosmetic. On a storage box the thick area is often a base corner, a foot or a boss; the visible area is the outer side wall and the top of the lid. They are rarely the same place, and the decision is which one you are willing to trade.

There are two answers that work in practice, and both are worth putting in writing rather than leaving to the toolmaker:

• Gate into the base, so the vestige faces the shelf rather than the customer. The base is also usually the thickest section of a container like this, which satisfies the mechanical rule at the same time.
• Gate from inside a recess. Eastman's molding design guide notes that, where part size and cooling allow, "pinpoint gating may be from inside the part or into a recess to hide the vestige." The recess is designed for the purpose, and it is a decision made on the drawing, not after the trial.

What goes wrong is the third answer, which is no answer: the gate is placed wherever the runner was convenient, and it lands on the lid's show face — the one surface the retail photograph lights from above.

Gate Size: A Ratio to the Wall, Not a Number on Its Own

Gate size is where a buyer can most easily turn an adjective into a number. Eastman's guide gives figures that are usable as written: edge gates and tab gates "should be 50%–80% of the nominal wall thickness," typical tunnel gates are at least 1.6 mm in diameter, and typical pinpoint gates in three-plate molds run 0.75–1.1 mm to keep the vestige small.

The guide is written for Eastman's own copolyesters. These ratios are also the conventional starting figures for unfilled polypropylene, but they are starting figures: the toolmaker still has to confirm them against the wall the gate feeds and the resin grade.

Read against a container wall, the first figure is the useful one: the gate is a fraction of the wall it feeds, so when the wall changes, the gate changes with it. A 3 mm wall with a 2 mm gate is a 67% gate; a 2 mm wall with a 2 mm gate is a 100% gate. Both drawings say "gate 2 mm," and only one of them sits inside the band above.

Size cuts both ways. An undersized gate freezes early, which stops packing and leaves the far thick section short or sunk. Covestro's weld-line paper adds the second cost: "undersized gates can limit the filling speed and reduce weld line strength." Too large a gate solves the fill and leaves a bigger mark to trim, which is why the gate type and the gate size are one decision rather than two.

Write the type, the size and the face it sits on. "Appropriate gate" is not a specification; "edge gate, 60% of nominal wall, on the base underside" is one, and it can be checked against the 2D layout before steel is ordered.

Draft Angle: The Number That Decides Whether the Box Releases Without a Scratch

Draft angle in injection molding is the taper on every surface parallel to the direction the mold opens. Without it, the part has to be dragged off the core, and the drag shows up as scuffing, whitening or a distorted rim — on a container, a distorted rim is a leak.

The draft guidance is short and specific, and it comes from the same copolyester source. From Eastman's injection molding design guide:

In polished molds, 1° draft per side is suggested to aid ejection. However, 0.5° draft can be used to obtain reasonable dimensions in ribs, bosses, or other design features. Avoid using zero draft in any part of the mold. Texturing requires increasing the draft angle by 1°–1.5° for every 0.025 mm (0.001 in.) of texture depth.

Three details decide how it is applied.

The 1° is per side, not per wall. A 60 mm deep wall at 1° per side leans about 1.05 mm over its height on each face — enough to release, small enough that nobody notices it on the shelf.

The core side is the hard side. Plastic shrinks onto the core as it cools, so the inner walls grip the steel that forms them while the outer walls pull away from it. A container with 1° outside and 0.5° inside is drafted on paper and still drags in production. Ask for the number on the inside face specifically.

Texture changes the arithmetic. A matte or leather-grain finish on a lid is a series of micro-undercuts, and the draft has to absorb them: at 1°–1.5° per 0.025 mm of depth, a 0.1 mm texture on a semi-crystalline resin adds 4°–6° to the base angle. That is enough to change the look of the part, so it is a decision for the drawing stage, not a note added after the texture is chosen.

Whether the seal itself is molded as a separate ring, overmolded into the lid or formed as a plug changes the tooling bill and the ejection problem, which is the subject of the liquid silicone versus compression-molded comparison. What does not change is the draft requirement on the walls around it.

Parting Line: The One Line Every Customer Will See

The parting line in injection molding is where the cavity half and the core half of the tool meet. It leaves a witness line, and if the shut-off is not tight, flash — a thin fin of plastic along that line.

For a food container there is a right place and several wrong ones. The right place is a shoulder or an edge the eye already reads as a change of direction: under the lid skirt, along a step in the base, or on the outer corner radius. The wrong places are the seal seat, the inner food-contact rim, and the flat top of the lid. Flash on a seal seat is a leak path with a cosmetic excuse; on the inner rim it is a cleaning problem and a food-contact argument.

Parting line and gate interact, so they are reviewed together. Beaumont's glossary notes that edge gates attach to the part "along its perimeter at the parting line of the mold" — so the gate is at the line, and the mark from degating sits on the same feature that carries the witness line. Move one and the other moves with it, usually after the trial.

The line is also a tolerance problem in its own right. ISO 20457:2018, the standard for tolerances on molded parts, gives parting line and tool offset a clause of its own (7.3), separate from the tolerance grades that apply to dimensions inside one half of the tool. That separation is the point: a dimension that crosses the parting line is manufactured by two pieces of steel meeting, and it cannot be held to the same band as a dimension cut into one.

ISO 20457:2018 is the edition quoted here. A second edition, ISO 20457:2026, has cancelled and replaced it: the scope sentence quoted further down is unchanged in the new edition, but the clause number above and the title-block example are the 2018 ones, so check them against the edition your supplier will hold before you paste either into a purchase order.

Weld Lines: Where Two Flow Fronts Meet, and Why Your Logo Cannot Cover It

A weld line in injection molding — also called a knit line — is what happens when two flow fronts meet inside the cavity. Covestro's technical note on weld lines describes the mechanism: as the melt advances, it moves by fountain flow, with a rounded front; when two fronts meet head on, "the rounded leading edges form a V-shaped groove along the interface," and that notch "can act as a mechanical stress concentrator and reduce mechanical performance in properties such as impact strength and elongation to break."

Two consequences follow, and both land on a lunch box.

It is visible. The V-groove catches light differently from the surface around it. On a translucent or light-colored polypropylene part it reads as a hairline; on a molded logo pad it reads as a scratch across the brand. That is not a finish problem that polishing the tool will fix, because the line is formed by the flow, not by the steel.

It is weaker. Covestro notes that weld lines "can also be weaker than the surrounding area." On a latch root, a hinge or a clip that flexes twice a day for three years, a weld line in the wrong place is a crack starter.

The geography is predictable, and that is the useful part. A weld line typically runs from an opening or a divider until the angle between the merging fronts exceeds about 90 degrees, per the same note — so on a compartmented box, the line lands behind the divider walls and along the far wall from the gate, not in the middle of the floor.

Those patterns are typical rather than universal; the point of the table is to give the review a checklist, not to replace the simulation. Covestro is explicit that molding simulation "can predict weld line locations, weld line severity and the flow front temperature at the weld line" — so the question to the factory is not whether weld lines exist, but where the mold flow analysis puts them and whether that is inside a zone you agreed to.

Keep their closing line in mind when a design promises no weld lines at all: "Weld lines are a normal part of injection molding and will appear in nearly any part with holes, openings, or multiple gates." A factory that promises none is either gating sequentially, which is a cost decision, or has not run the analysis.

Where the line lands on the logo is the decision buyers most often discover too late. The three routes — molded into the tool, printed on the surface, or applied as a film — behave differently over a weld line, and the printed routes are the ones that show it, because the ink sits on a surface that is already slightly different from the surface beside it. The ten coloring methods used in practice matter less here than the position: a molded logo on a pad that the weld line crosses will read badly in every color, and no print method fixes geometry.

Wall Thickness and Tolerance: What Makes the Four Decisions Checkable

The four decisions are geometry, and geometry is only enforceable if the drawing carries a tolerance. That is where ISO 20457 comes in.

The standard sets out general tolerances for all geometrical features plus direct tolerances for the functional ones, and it expects them to be stated on the drawing rather than assumed — the worked example in the standard is a title-block note reading "General tolerances ISO 20457:JJJJ – TG6." It also separates general from direct tolerancing on purpose: general tolerances apply to non-functional features and "can be as large as practically possible," while functional dimensions are directly toleranced and kept few, for cost reasons.

The scope then says something worth quoting to a supplier who treats tolerance as a catch-all. It covers dimensional and geometrical tolerances, acceptance conditions and limits of technical manufacturability — then adds: "Moulded part surface imperfections such as sink marks, undesired flow structures and roughness, as well as joint lines are not addressed in this document."

In other words, meeting the tolerance says nothing about the weld line. The line is a separate agreement, which is why it belongs in the purchase order as a named zone rather than in the drawing as a dimension.

Injection molding wall thickness sits underneath all of it. A container works when the nominal wall is held constant and the transitions are gradual; a step from a thick wall to a thin rib creates a thermal mass difference, and the thick side shrinks last, which is how a flat lid becomes a dished lid and a flat rim stops sealing.

There is a cheap way to police this in production, and it does not need a CMM. Part weight tracks wall thickness closely enough to be a drift alarm. The BX1051 lunch box in our own records is listed at 182 g and 72 pieces to a carton (both figures carry a "confirm basis" note on the record). A uniform 10% thickening of every wall would add roughly 18 g per box — 1.3 kg to the carton, and about 728 kg across a 40,000-piece order. Nobody notices 18 g on one box; everybody notices a carton that no longer matches its declared gross weight — a check a buyer can run on arrival without opening a drawing.

The same logic applies to a molded set with metal and silicone in it, such as the five-compartment 6521: each material carries its own shrinkage, so the wall that matters for the seal is the polymer one, and its weight is the number that moves first when the process drifts.

How to Review a Mold Design for Injection Molding: Six Things to Ask the Factory

A mold design review is only real if it produces artifacts. These six are worth asking for, in this order, because each narrows the next.

Two are commonly missing. Draft analysis gets skipped because the CAD model already had draft applied and nobody re-checked it after the last change; the simulation gets skipped because it costs money and the part "looks simple" — on a four-compartment container with three latches, that is exactly where the weld-line prediction earns its fee.

If a printed prototype is on the table at this stage, be clear about what it can and cannot prove before signing anything against it; that boundary is covered in what a printed prototype does and does not verify.

Five Clauses for the Tooling Purchase Order

The review ends in a document. Five clauses carry most of the weight:

1. Design sign-off is a milestone, and its deliverable is the six artifacts above. Not a meeting, not an email saying the design is fine — the files, dated, attached to the PO.
2. Gate position, gate size, draft and parting line are recorded as dimensions and named faces, not as adjectives. "Standard gate" is not a record.
3. The weld line is excluded from named zones: not on the seal seat, not through the logo pad, not across a latch root. This is a cosmetic and structural agreement, separate from the dimensional tolerance.
4. Change control names who pays when a decision moves after steel. Moving a gate after the first trial is a cost; the question is whose. The deposit and balance structure in a Chinese tooling order usually decides that in practice, which is why payment terms are worth settling early.
5. First-article acceptance is measured against the directly toleranced dimensions plus the general tolerance grade — and, separately, against the cosmetic zones in clause 3, because a part can be dimensionally perfect and still fail on a visible weld line.

Clause 5 is where the two kinds of evidence meet. The first-article and golden-sample gates describe what happens after the tool is cut; the four decisions on this page are what determine whether that first article has a chance.

Where This Sits in the Sequence

The order is: annual volume justifies a dedicated tool; the tool is quoted and bought; the prototype proves the shape, not the process; the mold design is reviewed and signed; the first article is measured; the golden sample freezes what mass production has to match.

This page covers the fourth step, and it is the cheapest one to get wrong because it is the only one where a decision costs nothing to change. Once the steel is cut, every one of the four becomes a repair.

If you are at that step now, the custom mold development program is where the design review runs on our side, and the four-compartment 6522 shows how a body, its lid and its seal split across a tool. The conversation starts with the drawing, the annual volume and the four decisions — gate before steel.