Why a Window Beats a Lucky Setup

Most molds can make a good part once. That proves little. Resin lots change. Ambient temperature shifts. A heater band ages. A process with no margin turns each small change into flash, sinks, or short shots. A tested window absorbs those changes and still ships good parts.

Mold-flow simulation predicts where a part should fill and where it should cool. It does not prove the press on the floor can repeat that result. Scientific molding is the floor test that follows the prediction. Our earlier guide on mold flow simulation explains what simulation can tell you before steel is cut. This article covers the next step: proving the process after the mold arrives.

Think of the window as four walls. Fill speed sets one wall. Pack pressure sets a second. Hold time sets a third. Cooling sets the fourth. A buyer does not need to run these tests. A buyer does need to know whether the molder ran them, and what numbers came out.

Start With the Machine, Not the Mold Settings

A process built on a tired press will not transfer to a healthy one. Good shops test the press before they test the mold. They check injection-speed linearity, check-ring repeatability, load sensitivity, and pressure response. Each test shows whether the press does what its controller claims.

One practical rule follows from those baselines. Build the process on the least capable press that will run the job. A process that works on the weakest press transfers upward with little drama. A process built on the strongest press may fail on every other press in the shop. That failure shows up as a late delivery, not as a theory debate.

Ask one question at tool approval. Which press ran the trial, and was it the limiting press? If the molder cannot answer, the trial data may not travel. Press data belongs on the same setup sheet as mold data.

Viscosity Curve: Find the Flat Spot

Plastic thins as it is pushed faster. Push slowly and the melt stays thick. Push fast and it flows with less resistance. That shear-thinning behavior is normal for most resins. It also means fill speed changes part quality in ways that look random when speed is guessed.

A viscosity curve removes the guesswork. The molder molds fill-only shots at a series of injection speeds. Each speed gives a peak pressure. Plotting pressure against speed shows a curve that falls, then flattens. The flat region is the prize. In that region, a small speed change barely moves pressure or part weight.

Set fill speed inside that flat region when part appearance allows it. Appearance sometimes forces a slower start at the gate to avoid a halo or jetting mark. That exception still starts from the curve. The molder slows only the gate-crossing portion, then returns to the flat-region speed for the rest of the fill.

A fill speed chosen from the flat region gives the process room to breathe. Resin viscosity will drift from lot to lot. A process parked on a steep part of the curve feels every drift. A process parked on the flat part barely notices it.

Cavity Balance: Prove Every Cavity Fills Alike

Multi-cavity molds hide a trap. One cavity can fill first and flash while another still shows a short shot. The press readout looks fine. The parts tell a different story. A cavity-balance study catches that split before production starts.

The test uses short shots at several fill levels. The molder weighs or measures each cavity at each level. A balanced mold shows cavities filling in step. An unbalanced mold shows one cavity racing ahead. Runner size, gate size, vent depth, and wall thickness can each cause the split. Shear heating in a cold-runner layout can also tilt the balance.

Balance matters most when the window is narrow. If one cavity flashes before the others fill, no single pack pressure fixes both problems. The molder must fix the mold or the runner balance first. Pack pressure cannot rescue a mold that fills unevenly.

For a buyer, the deliverable is simple. Ask for cavity-by-cavity weights from the balance study. Similar weights across cavities mean the mold shares melt fairly. A wide spread means trouble that will return with every resin lot change.

Pressure Drop: Leave Headroom on the Press

Pressure falls as melt travels from the nozzle through the runner, the gate, and the cavity. Each section takes a share. A pressure-drop study maps that loss at the chosen fill speed. It shows where pressure goes, and whether the press has enough left to finish the job.

A process that needs nearly all of the press pressure has no reserve. A slightly thicker resin lot then slows the screw. Fill becomes inconsistent. Short shots, sinks, and warp follow. Shops that watch this study keep fill pressure well below the press maximum. The exact margin depends on the press and the part. The habit matters more than any single number.

Pressure drop also explains why two molds behave differently on the same press. A long runner or a small gate eats pressure before melt reaches the cavity. A hot-runner layout with balanced drops behaves differently again. The study turns those differences into numbers instead of opinions.

When a molder quotes a tight-tolerance part, pressure headroom is part of the quote. A press running at its limit cannot hold dimensions across lots. A press with reserve can. Shrinkage compensation explains how mold dimensions are derived from part targets.

Gate Seal Study: Set Hold Time by Weight

Hold pressure keeps melt packed in the cavity while the part cools. Hold time must last until the gate freezes. Stop too early and melt flows back out of the cavity. Part weight drops and dimensions drift. Hold too long and the cycle grows for no gain.

A gate seal study finds the freeze point by weighing parts. The molder steps hold time upward in small increments. Part weight rises, then stops rising. The plateau marks gate freeze. The set hold time sits just past that plateau, with a small safety margin.

That plateau is the only honest way to set hold time. Guessed hold times either waste cycle time or invite weight variation. A gate that seals late also warns about gate size or mold temperature. Both deserve a look before production locks in. A worked gate seal study example shows the weight plateau in practice.

Note the difference between cold-runner and hot-runner tools here. A cold-runner gate must freeze before hold pressure drops. A valve-gated hot runner closes on command, so the study answers a different question. Either way, the buyer should see a weight-versus-time curve, not a round number pulled from habit.

Map the Scientific Molding Process Window

With fill, balance, pressure, and gate seal known, the molder maps the window itself. The usual test varies two variables at once. Hold pressure is one. Melt temperature or mold temperature is the other. The molder molds parts across a grid of settings and grades each part.

Good parts form a region on that grid. Bad parts sit outside it. Flash marks one edge. Short shots mark another. Sinks and warp mark the rest. Sink marks follow a diagnostic path of their own, covered in our sink-mark flowchart for engineers. The production setting belongs near the center of the good region, not near any edge.

Center matters because real production drifts. Melt temperature moves with screw speed and back pressure. Hold pressure moves with resin viscosity. A centered process absorbs drift in any direction. An edge process fails at the first small shift.

Cooling deserves a place in this map too. Cooling often takes most of the cycle time, so rushing it costs dimensions. A cooling-time study shortens cooling step by step and checks dimensions and distortion at each step. The set cooling time sits where parts stay stable, with margin left for warm days and warm water.

Cooling Time and the Setup Sheet

Cooling starts the moment melt touches steel. It does not wait for the hold stage to end. That fact explains why cooling dominates cycle time on many parts. A mold with poor water flow or blocked circuits cannot be saved by faster fill or higher pack pressure.

Good shops record coolant flow and mold surface temperature per circuit. Each circuit gets an identity on a watering diagram. Flow meters beat guesswork. When a circuit scales up or plugs, flow drops and parts warp. A recorded baseline shows the change at once.

The setup sheet ties these studies together. It records melt temperature measured at the nozzle, not just barrel setpoints. Barrel setpoints lie. Screw speed and back pressure add shear heat that setpoints never show. A nozzle melt reading tells the truth.

The sheet also records fill time, peak pressure, cushion behavior, part weight with and without pack, and coolant flow. Those are outputs, not inputs. Inputs live on one press only. Outputs travel with the mold to any press that can match them. Material moisture also belongs on that sheet, as our drying guide for hygroscopic resins shows.

Decoupled molding gives this sheet its structure. Fill runs on velocity control to a set transfer point. Pack and hold run on pressure control after transfer. Keeping those stages apart lets each study change one thing at a time. Mixed stages mix causes, and mixed causes waste trial days.

Checks a Buyer Can Ask For

A buyer need not attend the trial to benefit from it. Ask for four items at tool approval. First, the viscosity curve with the chosen speed marked. Second, cavity weights from the balance study. Third, the gate seal curve with the plateau marked. Fourth, the window grid with the production setting near its center.

Watch for warning signs in the answers. A molder who sets hold time by habit has no gate seal curve. A molder who fills to 100 percent on first stage has no decoupled process. A molder who cannot name the limiting press has no transfer plan. Each gap predicts a repeat trial later, at your schedule cost.

Also ask what changed since the last good run when a problem appears. A documented process answers that question fast. Fill time slower than record points at temperature or material. Cushion drift points at a check ring or barrel wear. Part weight down with pack removed points at fill. Data narrows the search before anyone touches a setting.

Scientific molding does not promise a mold that never varies. It promises a process whose variation stays inside a tested region. That region is the product you are buying, along with the parts.

When Simulation Pays Before Steel Is Cut

Simulation pays when a wrong choice costs a mold rework. A moved gate, a thicker rib, or a second cavity can each force welding, re-cutting, or a new insert. Those fixes cost weeks and real money. A simulation run costs a fraction of one rework loop. That comparison is the whole business case.

Two moments matter most. The first is a new mold design, before any steel is ordered. The second is an existing mold that keeps producing rejects. Plastics Technology puts it plainly: both uses save time and money when the work is done right. The catch sits in those last four words.

One veteran analyst estimates that nearly 80% of hot runners from major vendors are designed with flow simulation. He puts the share for molds built with flow analysis at about 50%. Those are estimates, not census data. They still tell you where the industry puts its confidence: feed systems first, full molds second.

A simulation also needs a decision attached to it. A report that nobody reads is wasted money. Before you commission one, write down the choice it must settle. Gate location, cavity count, cooling layout, or material grade all count. “Check the design” does not.

Fill Analysis: What It Actually Shows

Fill analysis models molten plastic flowing into the cavity up to switchover. Plastics Technology’s mold filling simulation guide notes that a fill run predicts flow to 99% filled. Packing is a separate run. That boundary matters because buyers often expect one report to cover everything.

A fill plot shows where plastic arrives first and where it arrives last. It marks weld lines where two flow fronts meet. It flags air traps where air cannot escape before the melt arrives. It gives the peak pressure needed to fill the cavity. Each output maps to a shop decision: vent placement, gate count, or machine size.

Fill alone suits narrow questions. A cosmetic part with weld lines on a visible face is one example. A single-cavity tool with no tolerance tighter than ±0.5 mm is another. MoldMinds’ Moldflow simulations guide draws the same line: fill-only fits projects where weld placement is the main risk.

Fill plots also carry a warning. Green does not mean good everywhere. Autodesk’s Moldflow results guide colors confidence of fill green, yellow, and red. Yellow marks regions that fill with difficulty or low quality. Red marks regions that barely fill at all. Ask your analyst to walk you through every yellow patch, not just the green headline.

Pack Analysis: Where Dimensions Come From

Pack analysis starts where fill ends. Extra plastic is forced into the cavity as the part cools and shrinks. This phase sets dimensions, density, and sink risk. A part that fills fine can still fail here.

Three outputs deserve your attention. Volumetric shrinkage shows how much each region shrinks during cooling. Pack pressure distribution shows whether pressure still reaches far regions before the gate freezes. Gate freeze-off time marks the moment pressure can no longer reach the cavity at all.

Once the gate freezes, more pack pressure changes nothing. That single fact explains many sink marks. The thick boss behind a cosmetic face starves while the thin wall nearby already froze. Raising pack pressure after freeze-off cannot feed it. Only a bigger gate, a closer gate, or a thinner boss can.

Pack analysis matters most on parts with thick sections. One vendor uses 4 mm in unfilled resins and 3 mm in fiber-filled resins as the threshold. It also matters on multi-cavity tools, where pack must stay consistent from cavity to cavity. If your drawing holds a tight dimension across cavities, pack belongs in the scope.

Cooling Analysis: Cycle Time and Mold Temperature

Cooling analysis models heat moving from molten plastic into steel and out through cooling channels. It is the most computation-heavy of the four runs. It answers two buyer questions: how fast can this mold run, and will the mold surface stay even?

Its key output is time to ejection temperature. That number sets the floor for cycle time. No process tweak beats it without risking distortion at ejection. Autodesk adds a second useful output: cooling time variance. Regions that freeze much later than average set the cycle. Regions that freeze much earlier can feed warpage.

Cooling analysis earns its fee on high-volume tools. A few seconds per cycle, multiplied across a million shots, pays for the run many times over. It also earns its fee before a conformal cooling investment. Straight channels cannot reach every hot spot. Simulation shows which spots stay hot and by how much.

Ask for the mold surface temperature map, not just a cycle time. A mold that runs fast but holds a 15-degree difference across the cavity will fight warpage on every shot. Cycle time without temperature balance is half a report.

Warpage Analysis: The Hardest Output to Trust

Warpage analysis combines fill, pack, and cooling data, then adds post-mold cooling and stress relaxation. It predicts the final part shape after ejection and cooling to room temperature. It is the most valuable output and the least forgiving of bad inputs.

A real case shows both sides. A Tier 1 automotive molder ran Moldflow on a minivan appliqué panel. The first run predicted an 11 mm gap in the middle, tapering to 2 mm at the corners. Process tweaks could not fix it. After 49 runs, sequenced gating cut warpage to 4.1 mm in the critical direction. That is a 63% reduction, and the molded parts matched the prediction.

Note what the case actually proves. Simulation found the fix, but only after dozens of iterations and a gating redesign. It did not hand over one answer. Budget for iterations, not for a single run. A quote that allows one run and no rework of the model is a quote for a picture, not an answer.

A mold flow simulation report also splits warp by cause. Autodesk separates differential shrinkage, differential cooling, and orientation effects. That split tells you where to act. Cooling-driven warp points at the cooling layout. Orientation-driven warp in a glass-filled part points at gate location and flow direction. Demand the split in your report.

What Mold Flow Simulation Cannot Tell You

Mold flow simulation is a model, and every model has edges. Wrong resin data is the sharpest edge. If the software library lacks your exact grade, the analyst may substitute an “equivalent” material. Glass content, filler level, and viscosity all shift the result. Ask which grade was modeled and whether it matches your purchase spec.

Mesh quality is the second edge. Thin walls and small ribs need a fine mesh to capture flow and cooling. A coarse mesh smooths away the very feature that causes your defect. Ask about mesh density on thin sections before you accept a clean-looking plot.

Some physics stays outside the report. Plastics Technology warns that flow simulation generally does not predict clamp forces acting on slides over heel blocks. Multi-cavity balance is another weak spot: a geometrically balanced runner does not always fill evenly. Treat those two areas as questions for trials, not for plots.

Finally, simulation cannot replace a trial. It predicts under assumed conditions. If the shop runs a different melt temperature, fill time, or pack profile, the prediction no longer describes the part. Insist that the trial sheet carries the simulated conditions. Compare trial data against the prediction and keep both on file.

How to Scope and Brief a Simulation

Mold flow simulation scope follows risk, not habit. A cosmetic single-cavity tool needs fill only. A multi-cavity production tool with tight dimensions needs fill plus pack. A high-volume tool with cycle pressure needs cooling added. A fiber-filled part with assembly tolerances needs the full set, including warp.

Vendor pricing gives a rough scale. One US consultancy lists fill-only work at $1,000 to $2,500 and the full package at $4,500 to $8,500. A warp add-on to an existing model runs $2,000 to $3,500 at the same firm. Prices vary by part size and analyst, so treat these as a budgeting band, not a quote.

Brief the analyst with five items. Send the current CAD and drawing, with the resin grade and supplier. Mark cosmetic faces and no-go zones for gates. List the tolerances that affect assembly. State the machine you plan to run, with its shot size and pressure limit. Name the decision the report must settle.

When the report lands, check four things before you pay the balance. The modeled grade matches your spec. The gate locations match the tool design. Yellow and red zones carry an explanation and a fix. The trial sheet will use the simulated fill time and pack profile. If any check fails, the report is not finished.

For buyers weighing a new tool, the question is not whether simulation works. The question is which run answers your decision, and whether anyone will act on the answer. Get those two right and the fee looks small next to one avoided rework.

Gate choice drives most fill results. Compare gate styles in the interactive guide below before you lock a gate location.

Interactive Injection Molding Gate Types guide by

Why Moisture in the Pellet Becomes a Defect

Water hides in the pellet. Melt exposes it.

PA6 can hold 8 to 9% water when saturated. That is roughly a hundred times what slightly polar resins like HIPS can hold. Hygroscopic resins pull water deep into the pellet, not just onto the surface. PP and PE behave the opposite way. They are non-polar, and water has no chemical affinity for them, so they pick up almost nothing from humid air.

Inside the barrel the trapped water flashes to steam. The steam pushes silver streaks (splay) across the surface, traps bubbles under it, and leaves foamy, rough flow marks. The silver streaks wipe off the surface. The lost mechanical strength does not. In PA, PC and PET, water also attacks the polymer chain itself. That is hydrolysis. It breaks long chains into shorter ones, so molded parts come out weaker and more brittle than the datasheet promises.

PET is the sharpest example. Hydrolysis begins around 150°C, far below its 250°C melting point. By the time the resin is fully molten, wet PET is already losing molecular weight. Buyers blame the mold or the press. The real culprit is usually the bag the resin arrived in.

Which Resins Need Drying

Not every resin earns its drying time. Sort them first.

Drying hygroscopic resins is not optional for PA, PC or PET. For these materials, skipping the dryer is a guaranteed defect, not a calculated risk. Resin choice sets the drying burden before steel is cut. Our injection molding part design guide covers how material selection shapes the whole project.

Watch out for one exception. Some additives are hygroscopic. A non-hygroscopic resin carrying a wet additive or filler still needs drying. If the pellet blend changed and splay appeared, check the additive’s moisture behavior before blaming the base resin.

Moisture Limits by Resin

PET at 0.1% moisture is a ruined batch. PET at 0.004% is a good bottle.

The allowable limit depends on the resin and the part. These are the numbers molders check against with a rapid moisture analyzer or a dew-point test on the dried pellets.

Some shops run tighter in-house specs. One Chinese molder holds PA under 0.05% and PC and PET under 0.02% before every run. Buyers do not need to copy that number. They need to know their supplier has a number, measures against it, and can show the reading.

Drying Hygroscopic Resins: Temperature and Time

Temperature drives moisture out. Time lets it reach the surface. Both must fit the resin.

The table blends two reputable sources, and the ranges reflect real practice. Hybster dries PA at 80°C for 4 to 8 hours and allows 12 hours for resin that was never dried before. A Chinese molding shop publishes PA at 90–100°C for 8 to 12 hours. Both are correct for their equipment. The resin decides the window. Your dryer decides where inside it you land.

Two warnings travel with these numbers. Never exceed the resin’s heat tolerance. Too much temperature causes caking, clumping and color change. And keep the material layer 20 to 50 mm thick in the dryer. Plastic conducts heat poorly. A layer that is too thick dries unevenly: the surface dries while the center stays wet, and the moisture reading lies.

Why a Hot-Air Dryer Fails on Nylon

Hot air moves moisture around. Dry air removes it.

An ordinary hot-air dryer raises temperature but keeps feeding humid shop air to the pellets. For PA the moisture sits deep inside the pellet. Hot air alone cannot pull it out. Hybster states this plainly: a heated ambient-air dryer is ineffective for polyamide.

The right equipment is a dehumidifying dryer. It feeds air at a dew point of −30 to −40°C. The dry air gives the water somewhere to go, and the desiccant bed keeps the air that dry. Demand a dew-point reading from your supplier. The dryer’s temperature setpoint says nothing about its real performance. A spent desiccant bed can run at 90°C all day and still deliver air at a −10°C dew point. The parts still come out with splay.

Ask two questions on a factory audit. What dew point does your drying air hold? When did you last regenerate the desiccant? A supplier who answers in numbers owns the process. A supplier who answers in adjectives does not.

What Undried Resin Does to Parts

Splay announces the problem on the surface. Hydrolysis hides it inside the polymer chain.

The visible damage is familiar. Silver streaks fan out from the gate. Bubbles sit just under the surface. Flow marks look foamy and rough. On PET preforms, poor drying shows as foggy crystallization and yellowing. The invisible damage is worse. Hydrolyzed PA and PET lose impact strength and tensile strength. Shrinkage turns uneven, so dimensions drift. One PET machinery source holds the intrinsic-viscosity drop during drying and molding to no more than 0.02 dl/g. Above that, bearing strength and impact resistance fall off, and preform transparency degrades.

These defects sit next to sink marks in the defect family tree. Our sink-mark diagnostic flowchart covers the geometry-driven side of surface quality. For the process-driven side, run your symptoms through the interactive troubleshooter below.

Moisture also changes how the part shrinks. Hydrolyzed PET crystallizes faster, which shifts final dimensions away from the values the mold steel was cut for. Our shrinkage compensation guide explains how mold dimensions are really derived from resin behavior, including why crystallinity changes matter.

The Clock Starts After Drying

Dried resin reabsorbs moisture the moment it cools in humid air. The drying job is not done when the timer ends.

Move dried pellets into production within 2 hours. If the press stops for more than an hour, seal the hopper and keep it heated. Never let the resin cool in damp air. Store unused pellets sealed, off the floor, in a dry warehouse. Re-dry any resin that sat overnight before it goes back in the machine. Recycled runner material dries separately and mixes back with new resin at no more than 20%.

The old purge test still works as a quick check. Melt a small sample and watch. If it foams with continuous bubbles, the resin is not ready, whatever the dryer display says. When the sample runs clean and smooth, the moisture content meets the target.

One Formula Drives Shrinkage Compensation

The cavity is always bigger than the part. How much bigger is the whole job.

Every mold dimension starts from the same calculation. You take the target part size and scale it up by the resin’s expected shrinkage.

Mold cavity dimension = Part target dimension ÷ (1 − S), where S is the shrinkage rate as a decimal.

A 100 mm part in a resin with 1.8% shrinkage needs a cavity of 100 ÷ 0.982, or about 101.8 mm. That is the entire trick. There is no second formula hiding behind it.

You can also run it backwards. Measure the cavity and the molded part, then S = (cavity − part) ÷ cavity. T1 sampling works this way. The first parts tell you the real S for that resin, that wall, and that process.

The formula is exact. The input is the problem. Everything below is about getting S right.

Why the Datasheet Range Is a Starting Point, Not an Answer

A datasheet range is a menu, not a measurement.

Suppliers publish shrinkage as a range because the number moves with conditions. ABS might list 0.4–0.7%. The midpoint is a reasonable first pick, but the edges are 0.3 points apart.

On a 100 mm dimension, that spread is 0.3 mm. Many engineering drawings call ±0.05 mm on the same feature. Pick the wrong end of the range and you miss the tolerance six times over.

So treat the range as a screening value. Then narrow it with conditions: packing level, melt and mold temperature, wall thickness, and gate behavior. A shrinkage control reference from a working mold shop shows how these inputs interact in practice.

If the drawing tolerance is tighter than half the datasheet spread, do not guess. Run a trial tool or insert test before cutting production steel. Our sink mark diagnostic flowchart covers the packing-side checks that also set your real shrinkage.

Material Shrinkage Rates You Will Actually Use

Crystalline resins shrink roughly two to five times more than amorphous ones.

The split is structural. Amorphous resins like ABS, PC, and PS only contract thermally. Semi-crystalline resins like PP, PE, and nylon also fold into crystal structures as they cool, which pulls the volume down further.

Two rows deserve a second look. PA66 with 30% glass fiber drops from roughly 1.5–2.0% to 0.3–0.6%, but the shrinkage turns anisotropic. A 15% glass-filled ABS shrinks about 0.18% along flow and 0.35% across it. One number per dimension is no longer enough.

HDPE is the wild card at 2.0–6.0%. Wide range, thick sections, and heavy post-mold contraction make it the resin where buyers most often get surprised. If your part is HDPE and the tolerance is tight, ask one question. How will the supplier prove the rate before you approve steel?

Process Levers That Move Shrinkage After Steel Is Cut

Packing is the only lever that adds material after the cavity is full.

Higher packing pressure and longer holding time push more plastic into the shrinking core. That is why a small gate can sabotage everything. The gate freezes early, the pressure never reaches the thick section, and adding more packing time does nothing.

Temperature works the other way. Hotter melt and hotter molds both raise shrinkage. The part starts cooling from a higher point, so it contracts more on the way down. Cooler molds freeze the skin faster and resist contraction, but they can hurt surface finish and flow.

Wall thickness multiplies all of it. Thicker sections cool slower and shrink more, and crystalline resins keep shrinking after ejection. This is one reason uniform walls matter so much. Our conformal cooling guide explains how even cooling keeps differential shrinkage from turning into warpage.

To see how these levers trade off, try the interactive shrinkage simulator. Watch how resin choice and process settings change the result.

One concrete anchor: ABS typically runs melt at 220–250°C with the mold at 50–80°C, and it must be dried two to four hours at 80–90°C. Wet ABS does not just cause splay. It makes the viscosity drift, and drifting viscosity makes shrinkage drift too.

The Steel-Safe Rule Every Mold Maker Follows

You can always cut steel off. Putting it back costs weeks.

Good mold makers bias the first cut toward safety. For the cavity, they pick the small end of the shrinkage range. For the core, they pick the large end. If the part comes out slightly big, they grind or polish steel away. If it comes out small, the tool needs welding, stress relief, and re-machining.

This is why experienced shops talk about keeping steel safe rather than about hitting the number on the first try. T1 tells the truth. The first sampling run measures the real S for your resin, your wall, and your process, and the tool is adjusted to it.

Resin batches move the number too. Lot-to-lot variation can shift dimensions by 0.02–0.05 mm on a typical part. Tight-tolerance programs sometimes qualify a specific resin lot for the whole run, or at least re-check dimensions after a material change.

The part design decides how much freedom the shop has. Uniform walls, generous draft, and cored-out thick sections are covered in our part design guide. A part that fights physics costs more in tooling adjustments than a part that works with it.

Where Compensation Usually Goes Wrong

Most bad shrinkage jobs share one cause: one number applied everywhere.

A single shrinkage value across all dimensions ignores wall differences. The 1.2 mm wall and the 4 mm boss do not shrink the same amount. Neither do flow direction and cross-flow in glass-filled grades. Uniform scaling of a non-uniform part gives non-uniform dimensions.

Another trap is lifting values from the standard test plaque. ASTM D955 plaques are measured under fixed lab conditions after 24 or 48 hours. Your part is not a plaque. The plaque number is a material index, not your steel allowance.

Post-mold shrinkage bites thick crystalline parts. A PP housing can keep moving for days after ejection as crystals finish growing. Measuring parts at the press and approving them there can mean approving parts that are out of tolerance by the time they reach your dock.

And resin grade matters more than resin family. PA6 at 0.5% and PA66 at 2.5% are both nylon. Swapping grades without re-cutting steel is how a running tool suddenly makes bad parts. The Plastics Industry Association publishes industry references that can help buyers sanity-check material claims.

Gate choice belongs on this list too. Hot runners hold pressure longer and shrink less than cold runners in many cases, which changes the S the tool was cut for. Our hot runner versus cold runner guide walks through the trade-offs.

Four Questions to Ask Your Supplier

The answers tell you whether shrinkage was engineered or guessed.

• Which shrinkage rate did you use for each resin and each major dimension, and why that value within the range?
• Did you use one rate for the whole part, or different rates for flow and cross-flow on filled grades?
• How much steel did you keep safe on the critical dimensions, and where will you adjust after T1?
• Can you show measured shrinkage from the last similar part you ran in this resin?

If the supplier cannot name the rate per dimension, the tool was cut on hope. Walk away or budget for a re-cut.

How Sink Marks Form After the Skin Freezes

The skin does not fail. The core does.

Plastic shrinks as it cools. A rib, boss, or heavy wall holds a pocket of molten material long after the surrounding skin has frozen solid. As that pocket cools, it contracts. Nothing replaces the lost volume, so the skin above it dips inward. That dip is the sink mark.

This is why sinks almost always sit opposite ribs, bosses, internal fillets, or other heavy sections. The defect confuses people because the surface can look fine at ejection and the dip shows up later. It looks like a molding problem. Often it is a geometry problem wearing a process disguise. Diagnose in the right order and you stop chasing the wrong cause.

Step 1: Map the Wall-Thickness Ratio

Start with geometry, not the press.

Find the exact feature under each sink. Measure the nominal wall, then the thick section behind the mark, and compare the two. The standard diagnostic sequence runs wall-thickness ratio first, then gate freeze, packing, cooling, and material. A rib nearly as thick as the wall it stands on will sink on any press at any setting.

Record the ratio before touching any machine setting. These limits are part of basic injection molding part design. Check them before the mold is cut. If the sink sits over a 1:1 rib-to-wall section, you already know the ending. The steps below still matter: they tell you whether the process is masking the geometry or magnifying it.

Step 2: Check Whether the Gate Froze Too Early

Packing cannot cross solid plastic.

This is the most skipped step in sink diagnosis. The gate must stay open long enough for packing to feed the thick zone. Gate type and runner system decide how long that is. A small gate freezes much faster than an edge gate. Once the gate seals, the cavity is on its own. Raising holding pressure after that point changes nothing in the part.

If the gate froze early, more holding pressure only loads the machine harder.

Run a gate seal study. Raise holding time in steps and weigh the parts. Part weight climbs while the gate is open, then flattens. The flat point is gate freeze. Any packing beyond it is wasted cycle time, not extra quality.

Step 3: Read the Packing Profile

Packing feeds shrinkage. Holding stops backflow.

These are two different jobs, and mixing them up means tuning the wrong one. Packing pressure compensates for shrinkage by pushing extra material into the cavity as the plastic cools. Holding pressure, the lower of the two, keeps the melt from flowing back out before the gate freezes.

A short packing phase leaves more shrinkage in the part, which means more or larger sink marks.

Watch the pattern. If the sink fades when you raise packing pressure but returns when the cycle speeds up, your profile is marginal. Fix the window, not the setpoint. And know the limit: packing alone may never pull a bad geometry into spec.

Step 4: Check Cooling Balance on Both Sides

Cooling rarely causes sink. It decides where it shows.

The dip appears on the side whose skin stayed soft longest. A hot mold surface opposite a rib lets that skin yield while the core contracts. Check whether the cooling circuit behind the cosmetic face is weaker than the circuit behind the rib side. A few degrees of mold temperature difference across the wall can move or erase a marginal sink.

Lowering mold temperature on the sink side is the cheapest experiment in this whole flowchart.

If one side of the tool runs hot, find out why before blaming the process. The cause is often a blocked, scaled, or missing cooling channel. Balanced cooling gives the biggest single improvement for surface defects on parts with thick features.

Step 5: Rule Out Material and Regrind

Some resins shrink harder than others.

Semi-crystalline resins like PP, nylon, and PE shrink more than amorphous resins like ABS or PC. The same mold can run sink-free on ABS and sink badly on PP. If the defect appeared after a resin lot change or a higher regrind ratio, suspect the material before the tool. Coarse or excessive regrind changes flow and shrinkage behavior.

When a sink appears mid-production on a mold that ran clean for months, ask what changed first. Check the resin lot, regrind ratio, and mold temperatures before suspecting the steel.

Barrel temperature set too high adds volumetric shrinkage. Dropping melt temperature slightly can flatten a marginal sink. It also raises viscosity, so watch fill at the same time.

When the Process Cannot Fix It

Some sinks are born in CAD, not on the press.

If all five steps check out and the sink stays, stop tuning. The geometry itself defeats packing. The durable fixes are steel fixes, not process tweaks. Core out solid bosses and replace thick sections with ribbed structures. Move the gate closer to the heavy zone so packing reaches it before freeze, or enlarge an undersized gate. Keep wall transitions gradual instead of abrupt.

When packing alone cannot pull a bad geometry into spec, the fix is steel, not settings.

Walk through the same six steps on your own part with the interactive tool below. Then decide whether your answer is a process window or a design change.

Interactive Injection Molding Defect Troubleshooter by

FAQ

What Conformal Cooling Actually Is

Most buyers picture cooling as drilled holes. That picture is twenty years old.

A conventional circuit is a set of straight drilled lines. The drill only goes straight, so each channel sits wherever the drill can reach, not where the heat sits.

A conformal channel bends with the cavity and core walls. It holds a steady distance from the plastic along the whole part, including ribs, bosses, and deep cores that drills cannot reach.

You cannot drill a curve. Shops build these channels with metal 3D printing (DMLS), vacuum brazing, or diffusion bonding instead of a drill press.

The payoff is even heat removal. Hot spots shrink, warpage drops, and the cycle shortens. A conformal channel goes where the heat is; a drilled channel goes where the drill can reach.

Filling gets the attention, but cooling decides the economics of the injection molding process.

Why Cooling Owns Your Cycle Time

Cooling takes 50–80% of a typical molding cycle. Fill, pack, and eject take seconds. Cooling takes the rest.

Cooling time grows with the square of wall thickness. A 4 mm wall needs roughly four times the cooling of a 2 mm wall.

That square law is why thin-wall parts fly and thick bosses drag. It is also why no press upgrade beats a better cooling layout. A faster machine cannot cool plastic sooner than physics allows.

Cooling time grows with the square of wall thickness, so the wall decides the cycle before the press is even chosen. This is where part design rules pay off: uniform walls cool evenly and fast.

Try the interactive cooling simulator below. Change the wall thickness and watch the cooling time move.

Interactive Cooling Layout & Heat Transfer Simulation by Topworks Plastic Mold

The Design Rules That Hold Up

Three dimensions control a conformal layout: diameter, pitch, and wall distance. The table below gives the working rules.

A published design-of-experiments study found the best cooling at 6 mm diameter, 8 mm pitch, and 4 mm wall distance. When mold strength counted too, the optimum shifted to 6 mm, 12 mm, and 6 mm.

Start the circuit at the gate area. That zone runs hottest, so it deserves the coldest water.

Prefer series layouts over parallel splits. Parallel branches invite uneven flow, and one starved branch ruins the whole circuit.

Pitch at three to four diameters and wall distance at two to three diameters is the layout every simulation then refines.

How the Channels Get Built

A beautiful channel design means nothing if the shop cannot build it. Three methods dominate: DMLS metal printing, vacuum brazing, and diffusion bonding.

DMLS prints the insert layer by layer in maraging steel, often grade 1.2709. The shop then machines and polishes the working surfaces to standard mold finish.

Vacuum brazing joins machined plates with the channels milled into the joint faces. Diffusion bonding fuses the plates without filler metal. Both suit larger inserts where printing gets expensive.

Most shops build conformal cooling into the insert only, then fit it into a conventional mold base. Ask your supplier which method they run in-house. A subcontracted print adds weeks and a second margin.

Most suppliers print or braze only the core or cavity insert. The rest of the mold stays conventional, which bounds the cost premium.

Cycle-Time Gains With Names Attached

NyproMold cut a high-cavitation tool from 23 seconds to 16. That is over 40%, confirmed by flow simulation before any steel was cut (MoldMaking Technology).

Cimatron’s car reflector case went from 40.3 to 33.4 seconds, down 17%. Warpage fell 62% and sink-mark area fell 50% (MoldMaking Technology).

Star Rapid’s transparent parts saw cooling time drop 60%, about 38 seconds. Manual air cooling was no longer needed (engineering.com).

Consultant Bob Beard puts the typical range at 20–40% cycle reduction across programs. Treat vendor claims above 50% with care. They usually describe cooling time, not total cycle time.

A 40% cut on a 20-second cycle saves about 7 seconds a shot, and that compounds across millions of shots. Use these numbers when comparing mold quotes, not the brochure headline.

Doing the ROI Math

Conformal cooling pays back in press hours, not in tool price. The tool costs more. Beard estimates roughly 20% extra on cavity and core work.

Take a worked example with round numbers. A 4-cavity cap mold runs 2 million shots a year. The cycle drops from 30 to 22.5 seconds, a 25% cut. Press time costs $35 an hour. The conformal work adds $18,000 to the tool.

That program runs 500,000 cycles a year. Each cycle saves 7.5 seconds, which totals about 1,040 press hours, worth roughly $36,000. Payback lands near six months.

Payback (months) = tool premium ÷ monthly press-hour savings.

Below about 500,000 shots a year, the same $18,000 premium takes over two years to repay. Volume decides.

Conformal cooling is a volume game: high shots per year repay the premium in months, while low volumes may never repay it. Pair it with the hot runner vs cold runner decision, since both target cycle economics.

When Conformal Cooling Does Not Pay

The most expensive cooling channel is the one the part never needed.

Flat parts with uniform walls cool fine on straight drilled lines. Curves earn the premium, not flat plates.

Low-volume programs cannot amortize the extra tool cost. The math in the last section shows why.

Conformal channels are harder to descale and flush than drilled lines. Poor water treatment will choke them first, so ask about your plant’s water quality before approving.

I’ve seen buyers spec conformal cooling on a flat cover where four drilled lines would have matched it. If straight drilled channels already cover the part evenly, conformal cooling buys cost without benefit.

Before you approve the premium, ask your supplier to show the simulated cycle with and without it.

Frequently Asked Questions