Directives de conception 1TP15 T : épaisseur des parois, tolérances et DFM

What design engineers need to know before locking geometry — first-party tolerance data and decision rules from a Class 7 LSR cell producing 1,000+ tons per year.

These LSR design guidelines answer the questions that matter before a single mold cavity is cut: how thin a wall can be without porosity, how much shrinkage to expect after the part leaves the press, when draft is non-negotiable and when it can be skipped, and where flash will appear if the parting line lands in the wrong place. Liquid silicone rubber behaves nothing like the thermoplastics most product designers learn first, and carrying thermoplastic instincts into an LSR program is the most expensive mistake we see at the prototype stage.

At-a-Glance: LSR Design Specs

Before working through the individual sections, this Quick Specs card bundles the eight numbers most frequently asked by mechanical and product designers starting an LSR program. Use them for fast reference; the full reasoning lives in the H2 sections below.

Material Selection: Durometer, Grades, and Additives

Material decisions come before geometry decisions in a working LSR program. Durometer drives wall feasibility, gate sizing, and demolding behavior — pick the wrong Shore value first, and you redesign the part later.

LSR is a two-part addition-cured (platinum-catalyzed) thermoset, injected cold into a heated mold where it crosslinks irreversibly. Standard commercial durometer covers 30, 40, 50, 60, and 70 Shore A. Engelhardt’s custom LSR injection molding services hold a wider 5–80 Shore A range across 1,000+ tons of annual production, with the lower-durometer grades reserved for soft seals and the higher end for industrial gaskets.

For medical applications, Shore selection is only the beginning. Material grades branch into four working categories, each tied to a different qualification path:

• Standard LSR — industrial gaskets, seals, kitchen and consumer products. No biocompatibility documentation required.
• Medical-grade LSR — meets FDA biocompatibility evaluation per ISO 10993 and often USP Class VI. Usually also produced under ISO 13485.
• Optical-grade LSR — for lenses, light pipes, and translucent housings; demands controlled cleanliness during processing.
• Fluorosilicone — fuel and oil resistance; lower elongation, higher cost.

Implantable devices add a regulatory dimension. FDA guidance distinguishes short-term implantation (≤29 days) from long-term implantation (>29 days or permanent), and the testing profile escalates accordingly. A peer-reviewed 2024 review on advances in silicone implant characterization documents how host-tissue interactions vary with surface preparation, durometer, and fillers — a reminder that “medical-grade” is a starting point rather than a guarantee.

Wall Thickness, Ribs, and Section Transitions

Wall thickness drives almost every downstream cost: cavity count, cycle time, flow balance, and the rejection rate from porosity. The published industry floor sits at 0.010 in (0.254 mm), but treating that number as a casual minimum is a recipe for trial-molding pain.

How thin can LSR walls be?

LSR injection can fill walls as thin as 0.010 in (0.254 mm), and specialized programs reach below that. Wacker’s ELASTOSIL LR processing manual describes the platinum-catalyzed cure system that makes thin-section flow practical, with viscosity dropping under shear in the runner and gate. Practical floor depth depends on three things: the size of the thin region, the location of adjacent thicker sections, and how aggressively the cavity is vented.

“Our 150-ton LSR press routinely fills 0.020-inch walls on 64-cavity tools. Below 0.015 inch we run a flow-balance trial in the prototype mold before committing to the production cavity layout. That trial — about a week of press time — costs trivially compared with cutting a 64-cavity tool that won’t fill evenly.”

Industry practitioners report that thin-wall LSR parts are normally molded in 30–50 cavity tools rather than smaller experimental layouts, because production economics demand high cavitation once the geometry is proven.

Rib rules come next, and they surprise designers crossing over from thermoplastic. In thermoplastic injection, ribs commonly run 50–60% of the adjoining wall to control sink. In LSR, the working range is 0.5–1.0 × the adjoining wall — wider, because LSR’s cure-driven shrinkage profile differs and because oversized ribs stiffen the part in ways that change demolding force. Inside fillets should approximate the wall thickness; sharper or rounder than that risks porosity at the transition.

Section transitions matter as much as the wall thickness itself. Going directly from a 1.5 mm wall to a 4 mm boss creates a cure imbalance: the thin region has finished vulcanizing while the thick zone is still curing, and the part comes out with localized warp or surface porosity. Step the transition over at least 3× the wall delta, or use a tapered fillet rather than a stepped change. The rubber molding tolerances reference details how transition geometry feeds directly into the achievable dimensional class.

Tolerances, Shrinkage, and Dimensional Accuracy

LSR tolerance discussions get muddled because two distinct mechanisms — mold-temperature expansion and post-cure cooling shrinkage — both move the dimensional outcome, and most design references collapse them into a single “shrinkage rate.” Untangling the mechanism is the difference between a part that comes off the press in spec and one that drifts during the first production run.

What actually happens dimensionally is counter-intuitive. LSR expands inside the hot mold rather than shrinks. Volumetric cavity fill in a working LSR program targets 98–99% so that material expansion completes the fill, and holding pressure exists to prevent the expanding rubber from being pushed back through the gate, not to compensate for shrinkage. Measurement of this expansion behavior is governed by ISO 17744 (specific volume as a function of temperature and pressure — pvT diagrams).

Shrinkage happens after the part leaves the mold. As the part cools to room temperature it contracts roughly 2.5–3%, and a post-cure cycle (often required for medical and high-temperature grades) adds another 0.5–0.7% on top. A Momentive Silopren LIM 6061 datasheet reports linear shrinkage of 0.027 in/in (2.7%) after a 10-second vulcanization at 200 °C without post-cure — landing inside the same band. Shrinkage in the flow direction runs higher than perpendicular, and thicker sections shrink less than thin ones, which is why parts with mixed wall thickness need cavity scaling tuned per region instead of a single uniform shrink factor.

If a critical sealing dimension lands within ±0.05 mm of a parting line, request a prototype run before production tool steel is cut. Parting-line behavior under pressure changes the dimensional outcome in ways that cavity-only measurement won’t predict.

For programs evaluating LSR against thermoplastic elastomers on a tolerance basis, the LSR vs TPE comparison walks through the trade-offs at scale.

Draft Angle: When You Need It and When You Can Skip It

Draft is the LSR rule designers most often over-apply. Carrying over the thermoplastic instinct of “always add 1° everywhere” creates unnecessary cosmetic toolmarks on shallow surfaces and complicates seal geometry where draft would compromise sealing function.

What is the minimum draft angle for LSR?

One degree is the common working draft for LSR injection. Vertical faces tolerate 0.5°. Shut-off conditions — where two mold halves seal against each other on the part surface — generally need 3°. Lightly textured surfaces start at 3°, and medium textures at 5° or more. Shallow geometries can sometimes be molded with zero draft because LSR’s high elongation makes elastic release possible across small features.

These values converge across multiple LSR design references and have held stable for two decades. They are working defaults rather than first-principles outputs — your geometry, durometer, and surface finish drive the final choice. Where a sealing feature genuinely cannot tolerate draft (an O-ring groove face, a flat sealing rim), treat that as a documented exception and pay for it in either a more complex tool action or a slightly longer cycle.

Undercuts, Parting Lines, and Gating Strategy

These three decisions share one root: where the mold splits. Get the split right and undercuts, parting flash, and gate vestige all become manageable; get it wrong and every downstream issue compounds.

Undercuts are one of LSR’s structural advantages. Because cured LSR has elongation reaching several hundred percent, simple undercuts on the order of 30% of part depth typically release manually or with light air assist — no slides, lifters, or cams needed. Beyond that, mechanical actions become necessary, and those add roughly 25–40% to tool cost depending on count and complexity. If an undercut is purely an assembly convenience, redesign it out. If it is functionally required (a snap retention, a captive seal lip), evaluate the system cost — labor, validation effort, automation, and tool maintenance — not just the upfront tooling delta.

LSR overmolding over a thermoplastic substrate further constrains undercut design because the substrate cannot deform during release. Plan undercut depth around the rigid substrate’s tolerance, not the LSR’s elasticity.

Parting lines reward simplicity. Three rules from production experience:

1. Avoid running a parting line through critical seal surfaces. Parting-line flash, even when trimmed, leaves a witness that compromises seal integrity.
2. Place the line so the part predictably stays on one mold half during opening. LSR rarely tolerates ejector pins (the pin pad becomes a permanent cosmetic scar), so part retention is a layout problem, not an ejection problem.
3. Hide the line on non-cosmetic geometry where possible. A line on a flange edge is invisible; a line across a Class A face is a callout for life.

Gates for LSR run smaller than thermoplastic equivalents. Sub-gates of 0.5–1.0 mm diameter are common where thermoplastic would use 1.5–3.0 mm, because LSR’s viscosity drops sharply under shear and a smaller gate gives better cosmetic outcomes. Rubber-engineering forums consistently report production-floor practice of vents around 0.0002 inch deep and gates around 0.002 inch — figures that have held since at least 2002 and align with the flash threshold discussed in the next section.

Flash, Cure Behavior, and Ejection Planning

Flash and ejection get treated as production problems, but most of these issues are designed in months earlier at the geometry-and-tooling-strategy stage. Solving them then is cheap; solving them in production is expensive.

When does flash occur in LSR molding?

Flash appears when mold gaps reach approximately 0.0002 inch (0.005 mm) or larger. LSR’s low viscosity at injection temperatures means it flows into gaps a thermoplastic would not. Tooling tightness, parting-line geometry, and venting strategy are the three controls; all three need attention from the first design review onward.

Cure-window timing is similarly narrow. LSR cures in molds running 250–400 °F (121–204 °C), with most production programs sitting in the 320–450 °F (160–232 °C) band documented in Plastics Technology’s LSR tooling reference. Press barrels and nozzles stay water-cooled to prevent premature crosslinking from shear heating; only the mold delivers the cure energy. This is why LSR tooling is mechanically more demanding than thermoplastic tooling: tight, flat, well-vented, and thermally controlled with water and oil circuits.

Ejection rarely uses traditional ejector pins. Pin pads telegraph through cured LSR as small flat circular impressions that survive every secondary process. Production tools instead rely on part retention (the part stays on one half by design), manual removal, air assist, or robotic pickers. For programs scaling above ~250,000 parts per year, plan the automation pickoff path during initial design — a part that releases cleanly is easier to validate and run at throughput than one that fights the press operator.

DFM Workflow: From Prototype Tool to Production Tool

Most LSR programs that hit problems in production were missing a workflow rather than data. A five-step path below is the one that consistently catches issues while they are still cheap to fix.

1. Material lock. Resolve grade, durometer, and additives against service temperature, sterilization method, and patient-contact duration. Lock before any geometry is finalized.
2. Geometry freeze. Apply the wall, rib, draft, parting-line, and gate rules from the H2s above. Run the design through an internal checklist before requesting a quote.
3. Prototype tool (mild tool steel). Cut a 1- or 2-cavity prototype mold. Validate flow, cure profile, parting-line behavior, and dimensional outcome under realistic processing conditions. This week or two of press time is the cheapest insurance available.
4. Pilot run. Produce 100–500 parts under quasi-production conditions. Measure tolerance distribution, characterize flash, document any cosmetic anomalies. Modify the prototype tool as needed.
5. Production tool (420 pre-hardened stainless steel). Cut the production cavity layout — typically 8 to 64 cavities depending on volume. Industry-quoted tool life for 420 SS in LSR service is on the order of 1 million cycles, with actual life depending on part complexity, maintenance discipline, and how aggressively the tool is run.

For most programs this sequence is non-negotiable. A “skip prototype” branch is allowed only when three conditions all hold: tolerance class is C (commercial), Shore A sits in the well-proven 30–60 range, and the parting line is simple. Outside that envelope, skipping prototype tooling means cutting expensive cavity steel against unverified assumptions — and the rework cost dwarfs whatever calendar time the shortcut saved. Run the geometry through Engelhardt’s LSR cost estimator for a feasibility number before deciding.

Engelhardt delivers new LSR molds in 30–50 days from kickoff, with the longer end reserved for high-cavity-count tools requiring three-plate or hot-runner construction.

A 3-question gate sits before cutting any production tool steel: are tolerances proven on the prototype, has flash been characterized at production cycle time, and is the cure profile stable across the planned cavitation? A “no” to any of the three sends the program back to step 4, not forward to step 5.

Foire aux questions