A professional injection molding supplier does more than run a molding machine. It reviews part geometry, resin behavior, mold construction, cooling, gating, tolerances, inspection methods, and production capacity before volume manufacturing starts. A typical molded part may need 1–2° of draft, while shrinkage behavior can range from roughly 0.2% to above 2% depending on resin and geometry. ISO 20457:2018 also provides a formal framework for tolerancing molded plastic parts. The supplier’s engineering work affects whether the 10th part and the 100,000th part fit, look, and perform the same way.
A CAD model can be dimensionally correct and still be difficult to mold. Thick areas cool more slowly than thin areas, ribs can leave sink marks on the opposite surface, deep vertical walls can resist ejection, and a gate placed several millimeters too far from a thin section can increase filling pressure or create an unwanted weld line.
That is why supplier involvement should begin before mold steel is machined. Common molding guidance starts around 0.5° of draft on vertical surfaces, while 1–2° works for many ordinary geometries; textured areas may require 3° or 5° or more depending on texture depth.
The same review should examine wall thickness rather than treating every polymer alike. Published molding guidelines, for example, place typical ABS wall thickness around 1.14–3.56 mm, nylon around 0.76–2.92 mm, and polycarbonate around 1.02–3.81 mm, although actual limits depend on flow length, resin grade, pressure, and part geometry.
Uneven walls create another issue because cooling does not finish at the same rate throughout the cavity. A supplier may therefore recommend coring out a 6 mm solid section, reducing rib thickness to roughly 40–60% of the adjoining wall, or adding radii where a 90° corner would restrict flow and concentrate stress.
Those changes connect part design directly to mold design. Gate diameter, runner layout, vent depth, cooling-channel position, ejector placement, parting-line location, sliders, lifters, inserts, and cavity count all have to work with the geometry rather than being designed as a separate manufacturing step.
A mold that produces five acceptable samples during a trial is not automatically ready for a 250,000-part annual program. Production suitability depends on repeatability across thousands of heating, filling, packing, cooling, opening, and ejection cycles.
Tool material should therefore match expected volume. Aluminum tooling can suit prototypes and lower-volume programs, while hardened tool steels are commonly chosen where wear, abrasive glass-filled polymers, long production runs, or repeated maintenance cycles justify higher tooling cost.
Production volume also changes the economics of cavity count. A single-cavity mold running a 35-second cycle produces about 103 theoretical cycles per hour before downtime; a four-cavity tool running at the same cycle can produce roughly 412 parts per hour, although runner design, clamp force, cooling capacity, automation, and mold cost also increase.
That makes cooling performance more important as volume rises. If an improved cooling layout reduces a 35-second cycle to 30 seconds, theoretical cycles increase from about 103 to 120 per hour, a gain of roughly 16.5% without adding another molding press.
The mold, however, cannot compensate for a resin chosen without regard to its real operating environment. ABS, PC, PA6, PA66, POM, PP, PBT, TPE, PPS, and glass-filled compounds differ in moisture absorption, shrinkage, stiffness, chemical resistance, temperature performance, flow behavior, and dimensional stability.
A 30% glass-fiber-reinforced PA6 illustrates how much material data matters. BASF’s January 2025 data for one Ultramid B3EG6 grade lists a specific gravity of 1.36, mold shrinkage of 0.003 in./in., dry tensile strength of 172 MPa at 23°C, and moisture absorption of 2.1% at 50% relative humidity.
Material specifications therefore need to cover more than a generic label such as “nylon.” The purchasing file may need the exact grade, glass content, flame rating, color designation, recycled-content rule, drying condition, approved substitutes, and applicable ASTM or ISO test requirements.
| Manufacturing item | Example engineering range or requirement | Why the supplier checks it |
|---|---|---|
| Draft | 0.5° minimum on many vertical faces; 1–2° common | Cleaner ejection and fewer drag marks |
| Textured surfaces | Often 3–5°+ draft | Texture increases release resistance |
| ABS wall | Approx. 1.14–3.56 mm | Fill, cooling and sink control |
| Nylon wall | Approx. 0.76–2.92 mm | Flow and dimensional control |
| Stable-resin shrink allowance | Can be around 0.002 in./in. | Tool dimensions must anticipate shrinkage |
| Less stable elastomer shrink allowance | Can reach about 0.025 in./in. | Greater dimensional change after molding |
| Typical machining capability quoted by one major molder | About ±0.003 in. | Useful reference before assigning part tolerances |
Tolerance planning follows material selection because mold dimensions alone do not determine finished-part dimensions. One published injection molding guideline reports about ±0.003 in. machining accuracy, while shrink allowance may range from about 0.002 in./in. for relatively stable materials such as ABS and PC to 0.025 in./in. for less dimensionally stable TPE materials.
ISO 20457:2018 separates molded-part tolerancing from an unrealistic assumption that every CAD dimension can simply receive a machining-style tolerance. The standard, confirmed as current in 2024, covers dimensional and geometrical tolerance concepts for non-porous molded thermoplastics, thermoplastic elastomers, and thermoset materials.
A supplier should consequently ask which dimensions affect assembly rather than applying the tightest tolerance everywhere. If a 120 mm cosmetic cover only needs ±0.5 mm overall but two locating features require ±0.10 mm, inspection and process work should concentrate on those functional interfaces.
That approach continues into mold trials. A professional supplier records melt temperature, mold temperature, injection speed, transfer position, peak pressure, holding pressure, hold time, cooling time, screw recovery, and cycle time instead of judging the process only by whether the first 10 pieces look acceptable.
A dimensional problem can come from several sources. A 0.3 mm error may originate in tool steel dimensions, resin shrinkage, packing pressure, uneven cooling, fiber orientation, moisture condition, or measurement technique, so changing the mold immediately can create another problem when the actual cause is processing.
This is where Advanced injection mold engineering becomes useful for custom parts with thin walls, deep ribs, multiple slides, insert molding, overmolding, tight interfaces, or high annual volumes. Flow behavior, cooling balance, venting, gate position, steel conditions, and expected shrinkage can be reviewed together before repeated mold modifications consume additional production time.
The same engineering discipline applies to quality inspection. A buyer ordering 100,000 parts does not normally gain much from inspecting every cosmetic dimension on every unit; the inspection plan should separate functional dimensions, cosmetic requirements, material verification, and process-control measurements.
Measurement method matters as tolerances become smaller. Calipers may be sufficient for a ±0.25 mm non-functional dimension, while a ±0.05 mm location, profile, or bore may require a CMM, optical system, fixture, pin gauge, or another method with suitable measurement uncertainty.
Traceability becomes useful when production extends across multiple lots. Resin lot, machine number, mold number, cavity number, operator or shift, production date, process record, inspection result, and any approved deviation can identify whether 2 affected cartons came from a 20-carton lot rather than treating the entire shipment as identical.
Quality work should start with process conditions and measurement planning rather than with sorting finished parts. Finding a dimensional problem after 50,000 pieces have been molded is far more expensive than identifying the same behavior during a controlled mold trial.
Supplier capability also affects secondary operations. Threaded inserts, ultrasonic welding, heat staking, pad printing, laser marking, assembly, overmolding, packaging, and functional testing each add another tolerance or process interface, so a part that costs $0.80 to mold can cost considerably more after several external operations.
Insert molding provides a good example. Metal inserts must remain positioned while plastic flows around them under pressure; poor support can move an insert by fractions of a millimeter, while unsuitable resin or insert geometry can cause weak retention, cracking, or incomplete encapsulation.
Overmolding adds material compatibility to the same discussion. A soft TPE layer may rely on mechanical interlocks, chemical adhesion, or both, and changing from one 60 Shore A compound to another does not guarantee identical bonding even when hardness is nominally the same.
Once secondary work is included, purchase price becomes less informative than total manufacturing cost. A mold quoted 20% cheaper can cost more over 500,000 parts if its cycle is slower, scrap is higher, manual trimming is required, or maintenance interrupts production more often.
Consider a simplified volume calculation: cutting cycle time from 40 to 34 seconds improves theoretical output by about 17.6%. At 300,000 annual parts, that difference affects machine hours every production year, while a one-time tooling saving is paid only once.
Maintenance planning belongs in the same calculation. Slides, lifters, ejector pins, vents, gates, hot-runner components, sealing surfaces, and textured cavity areas wear differently, especially when molding abrasive compounds containing 20–40% glass fiber.
A mature supplier therefore records mold maintenance rather than waiting for flash or dimensional drift to appear. Wear found after 50,000 cycles may require a small insert replacement; wear ignored until 300,000 cycles can affect cavity surfaces, shutoffs, part dimensions, and production availability.
Capacity should also be checked before demand grows. A program using 35% of one molding machine’s available monthly capacity is easier to expand than a program already using 90%, especially when the selected press size, robot, dryer, mold-temperature controller, and inspection equipment are not widely available.
The supplier should explain how additional demand would be handled: more machine hours, another qualified press, a duplicate mold, extra cavities, or automation. A buyer expecting volume to rise from 50,000 to 300,000 parts per year needs that answer before the first production tool is approved.
Communication determines whether all of the engineering work remains consistent across revisions. Drawing revision, resin grade, color, texture, approved samples, inspection method, packaging requirement, and mold modification history should match the current purchase specification rather than relying on email memory.
A change from revision B to revision C that moves a locating boss by 0.4 mm may affect tooling, fixtures, inspection programs, and mating components at the same time. Professional suppliers use controlled drawings and approval records so production in 2027 can still be compared with the specification approved in 2026.
When evaluating suppliers, request a DFM review and examine the questions they ask before issuing tooling approval. A supplier that discusses wall transitions, draft, gate position, parting lines, shrinkage, cooling, cavity count, measurement method, annual volume, resin grade, mold life, maintenance access, and 2–3 likely manufacturing risks is providing engineering input rather than only a molding quotation.
The commercial comparison can then use measurable items: tooling specification, quoted cycle time, cavity count, resin basis, expected scrap allowance, tolerance capability, inspection frequency, mold ownership, maintenance terms, annual capacity, and lead time. For a custom molded part, repeatable production across 100,000 cycles matters more than producing one visually acceptable sample on the first trial.