Why Molded TPU Parts Keep Changing Size: Shrinkage, Conditioning, and Dimensional Validation

Quick answer

TPU molding shrinkage is not a single transferable percentage. It depends on the grade, part geometry, flow direction, mold and melt conditions, packing, cooling, and the time and environment of measurement. Qualify the real material–tool–process combination: identify parts by cavity, measure the same critical dimensions at fixed intervals, condition them under a written protocol, and approve tolerances only after the dimensions reach the application’s defined decision point.

A part that passes beside the press but fails two days later does not automatically indicate a bad resin lot. The first question is whether both results represent the same molding history, cavity, elapsed time, temperature, humidity, fixture, and measurement method.

Dimensional validation starts with a stable molding process and records tied to the actual machine, tool, and cavity. The robotic engineer1, CC BY-SA 4.0 (license/status). Contextual image; not proof of GOAFLEX material, equipment, endorsement, chemistry, or grade performance.

Molding shrinkage and post-molding change are not the same event

Molding shrinkage is the dimensional difference between the mold cavity and the molded part after cooling, reported at an agreed time and in an agreed direction. A simple calculation is:

Shrinkage = (cavity dimension − part dimension) / cavity dimension × 100%

That expression is only useful when the cavity and part dimensions correspond, the measurement direction is recorded, and the measurement time is fixed. A gate-to-end length, a transverse width, and a thickness can behave differently.

Post-molding dimensional change happens after ejection as the part continues cooling and internal stresses relax. BASF describes total shrinkage for its Elastollan® TPU as molding shrinkage plus a smaller post-shrinkage contribution during annealing or prolonged storage. It lists product shape, wall thickness, gate design and position, material, melt temperature, injection pressure, hold pressure, and mold temperature as influences.[1] These are useful principles—not universal values.

Covestro likewise cautions that article shape, wall thickness, and processing conditions make TPU shrinkage definable only within limits. Its guide reports different post-molding behavior among its own soft and hard grades and between thin and thick sections.[2] Do not convert those supplier-specific charts into a blanket mold-compensation rule for another grade.

Define the dimensional endpoint before cutting steel

“Final dimension” may mean the dimension:

• after handling, a fixed hold, or laboratory conditioning;
• after a specified annealing cycle; or
• after assembly or simulated service exposure.

The drawing and inspection plan should identify which state controls acceptance. Otherwise, the mold shop may optimize one time point while quality inspects another.

What controls TPU dimensional stability?

Think in four connected groups rather than searching for one cause.

1. Material

TPU chemistry, hardness, morphology, additives, color package, moisture condition before processing, and regrind policy can affect flow, solidification, stress, and subsequent change. A hardness label alone does not define shrinkage. Two nominally similar grades can need different processing and validation.

Drying is also distinct from conditioning. Drying prepares hygroscopic pellets for processing according to the grade supplier’s instructions. Conditioning brings molded specimens or parts through a defined environment before testing. Confusing the two makes records ambiguous and can hide a process problem.

Exact TPU grade, lot, moisture preparation, and additive package belong in every dimensional study record. Luigi Chiesa, CC BY-SA 3.0 (license/status). Contextual image; not proof of GOAFLEX material, equipment, endorsement, chemistry, or grade performance.

2. Mold

Gate location and size influence pressure transmission and flow orientation. Cooling-channel layout and actual cavity-surface temperature influence how uniformly the part freezes. Venting, cavity finish, draft, and ejection can add distortion or handling strain. In a family or multi-cavity tool, distance from the sprue and local cooling can create cavity-specific populations.

BASF calls mold-temperature control essential because mold temperature influences surface, shrinkage, and warpage.[1] The correct setting remains grade- and part-specific.

Cavity position, gate layout, cooling, and local geometry should remain traceable in a dimensional study. Britiju at English Wikipedia, Public domain (released by creator; status identified by Commons as PD-user) (license/status). Contextual image; not proof of GOAFLEX material, equipment, endorsement, chemistry, or grade performance.

3. Process

Melt temperature, fill speed, transfer point, hold pressure and time, cushion, cooling time, and demolding temperature interact. Covestro identifies injection and holding pressure as major influences on dimensional stability and demoldability in its TPU processing guidance.[2] That does not mean “more pack is always better”: insufficient packing can leave sinks, while excessive pressure can increase stress or make demolding difficult. Establish a stable window rather than chasing one nominal setting.

Before blaming dimensional drift on material, confirm that the process was at steady state and that logged values—not only machine setpoints—were consistent.

4. Geometry and constraint

Wall-thickness transitions, ribs, bosses, long flow paths, holes, inserts, and asymmetric cooling can turn uniform material contraction into local warpage. A thin flexible wall may also deflect under caliper force or a poorly designed fixture. An assembled part can be constrained differently from a free part.

Why direction, wall thickness, and cavity position matter

Polymer flow creates directionality. Therefore, record whether each dimension is parallel to flow, transverse to flow, or through thickness. If the gate or flow path changes, the comparison may no longer represent the same orientation.

Wall thickness changes both packing and cooling history. Thick regions generally remain hot longer; thin regions may freeze before hold pressure can fully act. The practical lesson is not that one thickness always shrinks more. It is that plaque data cannot replace measurements on the real geometry.

Multi-cavity averages can conceal the problem. Record each cavity separately and map dimensions to the part. If cavity 6 trends differently, pooling it with seven stable cavities produces a reassuring average and a continuing field risk.

For a broader defect-by-defect review, see GOAFLEX’s TPU injection molding problems guide. If grade selection is still open, begin with how to choose the right TPU grade for injection molding.

Measurement timing and conditioning discipline

ASTM D618-21 is the active ASTM practice for conditioning plastics for testing. Its scope says temperature and relative humidity can materially affect plastics test results and should be standardized for reliable comparisons; the relevant material specification takes precedence where it modifies the procedure.[3] It is not a universal conditioning schedule for every TPU part.

A useful protocol states:

• elapsed time starts at ejection;
• agreed immediate, intermediate, and release checkpoints;
• conditioning temperature, relative humidity, duration, and allowed tolerance;
• whether parts are free, supported, stacked, bagged, or assembled;
• the stabilization rule before measurement;
• instrument, calibration, method, fixture, contact force, and operator;
• cavity, shot, batch, machine, and process-window identifiers.

Use noncontact metrology or a defined low-force fixture when a soft wall deforms under contact. Confirm the measurement system is capable relative to the tolerance. NIST’s Dimensional Metrology Group emphasizes traceable dimensional measurement and the effects that limit measurement accuracy; this supports controlling the measurement method rather than treating every displayed digit as part variation.[4]

Separate three questions

1. Repeatability: Can the same operator and method reproduce the reading?
2. Reproducibility: Do operators, instruments, and sites agree?
3. Part evolution: With the measurement system held constant, does the same part change over time?

A marked subset measured repeatedly can reveal part evolution. Decide the sampling method before the trial.

A controlled, calibrated measurement method is necessary before time-dependent part change can be separated from gauge variation. Senior Airman Grace Lee, Public domain (U.S. Air Force work; status identified by Commons) (license/status). Contextual image; not proof of GOAFLEX material, equipment, endorsement, chemistry, or grade performance.

When annealing may help—and what it can change

Annealing can accelerate the development of properties for some TPU grades. Both BASF and Covestro publish post-treatment guidance for specific product families, and both show that thermal history matters.[1][2] But their times and temperatures are supplier/product recommendations, not universal TPU settings.

Annealing may change dimensions or deform a poorly supported flexible part. BASF advises supporting products with low dimensional stability during annealing.[1] Therefore:

• use only the selected grade supplier’s current recommendation;
• validate the actual part, not just a standard plaque;
• define support, air circulation, ramp/cool conditions, and oven uniformity;
• measure before and after annealing at fixed times;
• recheck dimensions and the properties that matter to the application;
• do not use annealing to mask an unstable molding process.

If production parts will not be annealed, an annealed laboratory value should not become the acceptance basis. If annealing is part of production, include it in dimensional qualification.

Diagnostic matrix: what a dimensional pattern can indicate

Observed pattern	First hypotheses to check	Evidence to collect	Avoid concluding
All cavities change similarly over time	Normal post-molding evolution; conditioning or measurement timing mismatch	Same-part time series; temperature/RH log; grade guidance	“The resin lot is defective” from one late reading
One cavity is consistently different	Local cooling, venting, gate, cavity dimension, or pressure-transfer difference	Cavity map; cavity-surface temperatures; tool inspection	A global material change
Flow-direction length and transverse width move differently	Orientation, gate/flow path, asymmetric geometry or cooling	Flow-direction map; fill study; dimension-by-direction data	One compensation factor fits every axis
Drift changes after a process adjustment	Packing, transfer, melt/mold temperature, cooling, or demolding interaction	Actual process traces and a controlled before/after run	The adjusted variable acted alone
Operators or sites disagree	Fixturing, force, datum, environment, calibration, or method definition	Measurement-system study; correlated samples	The parts changed without checking the gauge
Annealed parts move or warp	Unsupported heating, nonuniform section/cooling, unsuitable cycle	Before/after map; oven record; support method	Annealing always improves dimensional stability
Parts pass free-state inspection but fail assembled	Assembly constraint, creep/relaxation, datum mismatch, service exposure	Free and assembled measurements; load/time/environment	Mold shrinkage is the only mechanism

The matrix generates hypotheses; it does not diagnose a part remotely. Change one controlled factor at a time in troubleshooting, or use a designed experiment when interactions are expected.

A practical dimensional-validation plan

Step 1: Freeze the question

List CTQ dimensions, tolerance, datums, flow direction, acceptance time/environment, and whether free-state or assembled dimensions govern.

Step 2: Establish traceability

Record resin producer, exact grade, lot, color/additive package, drying record, regrind content, machine, tool revision, cavity, and shot sequence. Keep startup and steady-state samples separate.

Step 3: Verify process stability

Use the grade supplier’s current drying and processing instructions. Record actual melt and mold temperatures, fill/transfer behavior, peak and hold pressure, hold time, cushion, cooling time, and cycle time. Do not begin the dimensional study while settings are still being tuned.

Step 4: Sample by cavity and time

Use balanced quantities from every cavity. Mark parts without deforming them and randomize measurement order.

Step 5: Measure a time series

Measure the selected dimensions at fixed elapsed times and after any specified conditioning or annealing step. Preserve individual values; report mean, spread, and change from baseline rather than only a pooled average.

Step 6: Challenge the useful process window

After the baseline is repeatable, test justified low/nominal/high levels for relevant variables without exceeding supplier or equipment limits. Include cavity as a block where appropriate. A qualified window is better than one “golden” shot.

Step 7: Delay irreversible tooling decisions

Compare the stable, final-condition result with the drawing and assembly need. Use steel-safe tooling strategy where feasible. Do not cut every cavity from an immediate press-side measurement or a single lot.

Release checklist

• [ ] Exact grade and current supplier processing guidance confirmed
• [ ] CTQ dimensions, datums, flow direction, and acceptance state defined
• [ ] Measurement system and soft-part fixturing validated
• [ ] Temperature, humidity, elapsed time, and storage/support controlled
• [ ] Every cavity represented and analyzed separately
• [ ] Stable process records retained with each sample set
• [ ] Post-molding and any annealing change quantified
• [ ] Multiple lots or production runs included when risk justifies it
• [ ] Assembly/service simulation completed where relevant
• [ ] Tool correction based on final-condition evidence

Common mistakes to avoid

1. Copying one shrinkage percentage into the mold design. Supplier figures belong to specified grades and test conditions.
2. Comparing press-side parts with laboratory-aged parts. Different clocks create an artificial lot difference.
3. Mixing cavities in one bag. Cavity evidence is lost permanently.
4. Changing several process variables without a trial plan. The team cannot identify the cause or interaction.
5. Using excessive caliper force on soft features. The gauge can become the largest source of “variation.”
6. Treating annealing as a universal fix. It can alter dimensions and requires product-specific validation.
7. Confusing pellet drying with molded-part conditioning. They have different purposes and records.
8. Correcting steel before dimensions stabilize. A fast correction can lock in the wrong compensation.

For bonded or insert-molded designs, dimensional behavior must be reviewed alongside adhesion and substrate constraints; see GOAFLEX’s TPU overmolding validation guide.

What to send GOAFLEX for troubleshooting

Send enough context to separate material, mold, process, and measurement effects:

• application and service temperature, humidity, load, and expected life;
• drawing with CTQ dimensions, tolerances, datums, and measurement state;
• exact TPU grade, hardness, lot, color/additive package, and compliance needs;
• drying equipment, setpoint, duration, dew point if monitored, and material handling;
• part weight, wall-thickness map, gate and runner layout, cavity count, and tool material;
• machine size and screw details plus the full process sheet and actual trends;
• cavity-marked measurements at each elapsed-time/conditioning checkpoint;
• measurement method, fixture, contact force, instrument, resolution, and calibration status;
• photos of parts, warpage, sinks, ejection marks, and packaging/support orientation;
• any annealing, assembly, aging, or service-simulation protocol and before/after data.

GOAFLEX can help review grade fit and structure a part-specific trial, but final tolerances should be confirmed on the production material, tool, machine, process, and inspection method.

Planning a TPU dimensional trial or investigating parts that move after molding? Email GOAFLEX at hello@goaflex.com with your drawing, grade, process sheet, cavity map, and time-based measurements for a sample review.

Image credits

• Image 1: Injection Molding Machine with a robot arm. Photo: The robotic engineer1 / Wikimedia Commons / CC BY-SA 4.0. Source: https://commons.wikimedia.org/wiki/File:Injection_Molding_Machine_with_a_robot_arm.jpg. License/status: CC BY-SA 4.0 — https://creativecommons.org/licenses/by-sa/4.0/.
• Image 2: Precision measurement equipment laboratory Airman 141117-F-NQ441-075. U.S. Air Force photo by Senior Airman Grace Lee, via Wikimedia Commons (public domain). Source: https://commons.wikimedia.org/wiki/File:Precision_measurement_equipment_laboratory_Airman_141117-F-NQ441-075.jpg. License/status: Public domain (U.S. Air Force work; status identified by Commons) — https://commons.wikimedia.org/wiki/Template:PD-USGov-Military-Air_Force.
• Image 3: Chips TPU. Photo: Luigi Chiesa / Wikimedia Commons / CC BY-SA 3.0. Source: https://commons.wikimedia.org/wiki/File:Chips_TPU.jpg. License/status: CC BY-SA 3.0 — https://creativecommons.org/licenses/by-sa/3.0/.
• Image 4: Standard two plate injection molding tool. Photo: Britiju, via Wikimedia Commons (public domain). Source: https://commons.wikimedia.org/wiki/File:Standard_two_plate_injection_molding_tool.jpg. License/status: Public domain (released by creator; status identified by Commons as PD-user) — https://commons.wikimedia.org/wiki/Template:PD-user.
• Context note: These photographs are illustrative. They do not prove GOAFLEX ownership, endorsement, exact TPU chemistry, exact commercial grade, processing conditions, or performance. The featured crop is an adaptation of Image 1 under CC BY-SA 4.0.

References

1. BASF, Elastollan® Thermoplastic Polyurethane Elastomers (TPU): Processing Recommendations, especially “Post-treatment,” “Mold Temperature Control,” and “Shrinkage.” PDF. Accessed October 20, 2026.
2. Covestro, Processing of TPU by Injection Molding, especially sections 2.4, 3.2.3, and 3.4.6. PDF. Accessed October 20, 2026.
3. ASTM International, ASTM D618-21, Standard Practice for Conditioning Plastics for Testing, active edition shown by ASTM. Accessed October 20, 2026.
4. National Institute of Standards and Technology, Dimensional Metrology Group. Accessed October 20, 2026.

Technical note: processing and post-treatment conditions must be confirmed against the current technical data and safety documentation for the exact commercial grade. This article intentionally provides no universal TPU shrinkage percentage, conditioning duration, annealing cycle, or tolerance capability.