Direct answer: TPU that cracks or peels along printed layer lines usually has marginal interlayer fusion, an inconsistent feed, or a part orientation that repeatedly pries those bonds apart. First confirm the split follows the layer stack. Then restore the exact TPU baseline and test heat, cooling, feed consistency, spool condition, orientation, and geometry in that order.
This page covers one exact failure: a TPU part separates on a visible layer boundary during bending, flex cycling, assembly, or normal use. It does not cover a random tear through solid material, a cut started during support removal, a single raised seam, or cosmetic strings between features.
Editorial scope: This workflow uses current manufacturer guidance and does not claim hands-on testing. TPU hardness, formulations, extruders, and slicer labels vary. Keep temperature, cooling, speed, and drying changes inside the exact filament and printer maker's supported limits.
Disclosure: GoodPrints may earn from qualifying purchases through the clearly labeled affiliate link on this page, at no added cost to you.
Confirm it is a layer-bond failure
Mark the first crack before pulling the part farther apart. A true interlayer failure follows a printed plane, often looks like a peel, and may repeat at the same height or transition. A random jagged tear through several layers points more toward overload, a notch, or the material itself. Damage beside a support cut or tool mark may have started during cleanup rather than printing.
Record when and where the split starts
Note whether it opens on the first bend, after repeated flexing, during press-fit assembly, or only in one thin corner. Photograph the fracture direction relative to the build plate and mark the loaded side. This separates weak fusion everywhere from one geometry that is peeling the layer stack apart.
If the defect is a zipper-like raised restart line, use the TPU seam-bump guide. If the material is torn or gouged where supports were removed, use the TPU support-scar branch. Keep this page for a clean split along a printed layer.
Match the evidence to the first test
| What you observe | First hypothesis | Clean proof |
|---|---|---|
| Several areas peel easily on the first bend | Low fusion or excessive cooling | Repeat one flex coupon from the supported baseline with one bond-related change |
| Walls look thin or the split follows a visibly starved band | Feed or flow instability | Observe extrusion and compare a slow, short coupon without changing heat |
| The spool also became fuzzy, poppy, rough, or stringier after exposure | Spool-condition drift | Run a same-spool dry-versus-exposed comparison |
| Only one hinge, tab, or loaded corner peels | Orientation or stress concentration | Print the same geometry with a load-aware orientation or gentler transition |
| The crack begins beside a seam, void, or cleanup nick | Local initiator | Remove the initiator while holding the global profile fixed |
Check 1: restore the exact supported TPU baseline
Verify the printer, nozzle, nozzle size, extruder path, TPU product and hardness, and material preset. Remove forgotten overrides before tuning. Record nozzle and bed temperature, cooling, outer-wall speed, volumetric limit, flow adjustment, wall count, and orientation. A profile copied from a different TPU or extruder can make every later result ambiguous.
Do not treat all flexible filament as one profile
Ordinary 95A TPU, softer grades, high-speed TPU, foaming flexible materials, and filled flexibles can need different handling. Prusa's current flexible-materials guide emphasizes careful flexible-filament loading and slower printing. Use the exact maker's starting range as the baseline, not a remembered setting from a different spool.
Check 2: test interlayer heat without chasing appearance
When the fracture is clean across multiple regions, test whether the existing profile is producing enough bond. Use a small representative flex coupon and make one supported temperature change. Keep speed, cooling, flow, spool, walls, and orientation fixed. Do not jump beyond the filament maker's published range.
Read the fracture, not only the surface
A glossy or smooth wall is not proof of strong fusion. Compare the bend angle at which the coupon first opens, the location of the first crack, and whether the new fracture still follows one clean plane. If bond behavior improves repeatably without new feed problems, heat was probably part of the failure.
Check 3: separate excessive cooling from low heat
Cooling can improve shape while shortening the time available for one strand to merge with the previous layer. If the supported profile uses strong part cooling and the split is broad, run the same coupon with one supported cooling adjustment. Do not change fan and nozzle temperature together; either can change the same symptom.
Keep bridges and small features out of the first comparison
Automatic bridge fan, short-layer behavior, and overhang rules can override the normal fan setting. Choose a coupon whose failing wall uses ordinary perimeter moves, then inspect the sliced preview. If only overhangs curl while ordinary walls bond, use the TPU overhang-curl diagnostic instead of weakening cooling everywhere.
Check 4: look for feed or under-extrusion evidence
Flexible filament can buckle, compress, drag, or slip before a rigid filament would. Watch the inlet, gears, tube path, and nozzle flow during the failed band. Look for clicking, changing line width, thin walls, gaps, or a chewed section of filament. Confirm the spool turns freely and the path does not add a sharp bend.
Slow the proof, not the whole diagnosis
Use one short coupon at the exact material's supported speed baseline, or test a single lower speed if the current path is visibly unstable. If broader starvation appears on walls and infill, move to the under-extrusion workflow. Do not hide a feed problem by increasing global flow until the good regions become overfilled.
Check 5: isolate spool-condition drift
Move moisture higher only when the timeline supports it: the same spool printed better before exposure, extrusion now pops or looks foamy, gloss changes randomly, or stringing and roughness rose at the same time. Follow the exact TPU maker's drying guidance and keep the recovered spool protected during the repeat.
Use the same slice before and after drying
Print the coupon before the approved drying process and again afterward with the same printer, nozzle, profile, orientation, and room setup. If random extrusion noise falls but the same loaded layer still peels, moisture was noise rather than the root cause. The TPU-after-sitting-out guide covers that narrower timeline.
Check 6: inspect orientation against the real load
Draw the bend, peel, pull, or twist direction on the part. If normal use opens the printed planes like pages in a book, tuning alone may not create enough margin. Rotate the part so the main flex runs through continuous roads where practical, while checking support scars, dimensional fit, seam placement, and build stability.
Use the real flex direction in the coupon
A coupon bent parallel to its layers proves something different from one peeled across them. Label orientation on every sample and bend them the same way. Use the functional-part orientation guide when the load path, surface finish, and support tradeoff need to be decided together.
Check 7: remove geometry stress concentrators
A sharp inside corner, sudden wall-thickness change, notch, thin tab root, or seam placed at a loaded bend can start a peel even when the rest of the part is acceptable. Inspect the first crack under neutral light. Add a supported radius, gentler transition, locally appropriate wall structure, or a less exposed seam, then keep the proven material settings fixed.
More walls are not a substitute for fusion
Wall count can add load area, but extra underbonded walls can still peel. Prusa's layers and perimeters guide explains the separate roles of perimeter and solid-layer controls. Prove basic fusion first, then add structure only when the failed part needs more section strength.
Apply only the fix your evidence supports
- Broad clean peel: restore the supported TPU profile, then test one bond-temperature or cooling variable.
- Starved band or thin walls: remove spool drag, path friction, buckling, or speed-driven feed instability before changing global flow.
- Timeline-linked fuzzy extrusion: follow the exact TPU drying guidance and protect the same spool during the repeat.
- One loaded hinge or tab: reorient the layer stack or redirect the flex so ordinary use does not peel a printed plane.
- Crack beside a notch or seam: soften the transition or move the initiator while preserving the proven profile.
- Generally strong part with too little section: add appropriate wall structure only after fusion is stable.
Run one controlled flex test
Use a short coupon that includes the failed wall thickness and bend direction. Print at least two unchanged baseline coupons before choosing a fix. Then change one variable and print two more. Mark the build direction, bend every sample the same way, and record whether the first crack is immediate, delayed, or absent.
Define acceptance before tuning
Accept the change when repeated coupons no longer peel prematurely along a printed plane and the part still meets fit, flexibility, surface, and dimensional requirements. This is a workshop comparison, not a standardized mechanical test. Safety-critical or regulated parts need an appropriate engineering validation plan beyond a hand-flex coupon.
Avoid these common failed fixes
- Changing heat, fan, flow, speed, and walls together: a stronger sample will not reveal which control helped.
- Adding infill first: infill does not repair a weak outer layer bond or a peel-loaded hinge.
- Drying every spool automatically: drying is a useful proof only when the timeline or extrusion evidence supports it.
- Maxing temperature outside guidance: this can create degradation, dimensional, and feed problems without solving orientation.
- Turning cooling off universally: bridge, overhang, and small-feature behavior may need a different branch.
- Blaming TPU hardness alone: hardness choice matters, but first prove fusion, feed stability, orientation, and geometry.
Use limits: qualify the actual TPU part, not only the flex coupon
A coupon that stops peeling proves one controlled comparison, not the finished part's load capacity, fatigue life, assembly margin, temperature range, chemical resistance, or safety. Release only the cooled production geometry printed with the saved spool, orientation, nozzle, profile, feed path, and post-processing sequence. A change in TPU formulation or hardness, printer, wall layout, layer height, or load direction needs a new proof.
- Flexed, stretched, compressed, or cycled use: test several complete parts through the real direction and travel. Reject whitening that grows, a visible opening layer, permanent set that breaks fit, or a crack that advances from the same plane.
- Press fit, screw, insert, snap, or clamp: assemble with the actual mating hardware and controlled force or torque. A print that survives hand bending but peels during installation has not passed.
- Dimensional or sealing job: measure the cooled critical features, mate the real counterpart, and repeat the intended bounded leak or seal check. A closed coupon does not prove a housing or gasket interface.
- Heat, outdoor, wet, oil, cleaner, or other chemical exposure: condition representative parts for the intended environment, then repeat fit and load checks. The TPU label alone does not qualify every formulation for every exposure.
- Repeated or batch production: verify multiple parts across the intended spool-handling and cooldown workflow. Hold the batch if the failure returns intermittently or moves between print regions.
- Pressure, lifting, restraint, protective, electrical-safety, vehicle-control, food-contact, medical, or other consequence-heavy use: this diagnostic is not a design standard or certification method. Use the applicable engineering controls, standards, inspection, and qualified process.
Stop tuning the same build orientation when the full part still opens at the same printed plane after fusion, feed, spool condition, and geometry checks are stable. Reorient the layer stack, soften the transition, split the assembly, change the load path, or choose another material or manufacturing process; then repeat qualification from the saved baseline.
Choose the next troubleshooting branch
If the whole part has weak or missing bands, use the general weak-layers guide. If the spool is newly stringy without a structural split, use the TPU stringing diagnostic. If the first layer alone has separated roads, use the TPU first-layer-gap branch. If the nozzle is rubbing a raised part, use the nozzle-scraping workflow.
Common questions
Can TPU split between layers even though it is flexible?
Yes. Flexibility does not guarantee strong interlayer fusion. Repeated bending can peel a marginal bond, especially when the printed planes face the wrong way for the real load.
Should I raise nozzle temperature first?
Only after restoring the exact supported profile and confirming a true broad interlayer peel. Test one supported temperature change while holding cooling, speed, flow, spool, and orientation fixed.
Can wet TPU cause layer separation?
It can contribute when the same spool also became poppy, fuzzy, rough, or stringier after exposure. Prove that with a same-spool comparison; moisture should not explain a repeatable crack in one peel-loaded hinge.
Will more walls stop TPU from cracking?
More wall structure can help a well-bonded part carry load, but it cannot make underfused layers reliable. Fix fusion and orientation first, then decide whether the cross-section still needs more support.
When is orientation the real fix?
Orientation becomes the leading fix when the part bonds acceptably elsewhere but normal use repeatedly opens one printed plane. Reorient only after checking the new support, fit, seam, and dimensional tradeoffs.