Why did my 'fully cured' repair slowly deform? (creep under constant load)

Evidence grade: A1 — First-hand testing (TDS/spec sheets, ASTM/ISO, lab tests)

Answer: "Fully cured" and "creep-safe" are two different milestones. Under a constant, never-relieved load, an epoxy's polymer network keeps flowing — slowly, measurably, forever. A structural epoxy TDS measures 0.010 mm of creep after 1 day growing to 0.25 mm after 3 months under sustained tension (EN 1544) [1]. 3M's own datasheet warns its structural adhesive "may still creep when subjected to continuous stress including bolting and torque application" even after it's cured enough to declamp [2].

Effort: N/A — debunk/reference, no procedure

The milestone everyone confuses

3M's rate-of-strength-buildup test on structural adhesive 9820 makes the trap visible. Overlap-shear joints were tested at increasing dwell times, and each result was marked creep Y/N — whether the load-extension curve showed "movement within adhesive before failure" [2]:

• At 23°C: 2 h → 0.37 MPa, creeps. 3 h → 2.06 MPa, creeps. 4 h → 4.74 MPa, slight creep. 5 h → 9.22 MPa, no creep [2].
• 3M defines handling strength as >2 MPa — enough to declamp (3–4 h at 23°C) [2].
• But for continuous stress, the TDS prescribes 5 h at 23°C (8 h at 15°C, 2 h at 35°C) — a full rung above handling strength [2].

Read that again: at 3 hours the joint holds 2.06 MPa and survives a pull test, yet it creeps. A short pull test cannot see creep; only time under load can. "Hard enough to handle" is not "safe under a permanent load."

How big the slow flow gets

The Fassa Epoxy 100 datasheet — a two-part structural epoxy rated to EN 1504-5/6 — publishes the long-term number most datasheets omit [1]:

• Creep after 1 day (EN 1544): 0.010 ± 0.001 mm
• Creep after 3 months (EN 1544): 0.25 ± 0.01 mm

Same joint, same load — 25× the movement in 90 days. Creep doesn't stop when you stop watching; it decelerates but continues. The same TDS lists the adhesive's glass-transition temperature (Tg) as 50.0 ± 0.2°C [1] — and that number is the reason temperature dominates creep: as service temperature approaches Tg, the polymer network softens and the slow flow accelerates. A joint that's creep-safe in a 20°C workshop can start walking in a 45°C engine bay.

What this means for a real repair

Any repair that holds a constant, never-relieved load is a creep candidate: a bracket permanently carrying weight, a bolt torqued through a bonded joint, a spring-tensioned part, a shelf bracket with books that never come off. The repair passes every short test — pull it, hang on it, it holds — and then deforms over weeks or months.

The defenses, in order of effectiveness:

1. Don't ask adhesive to do a fastener's job. Mechanical fasteners don't creep. The strongest answer to sustained load is a bolt through the repair, with the adhesive as sealant and load-spreader.
2. Give it more cure before loading. 3M's continuous-stress wait times (8 h at 15°C / 5 h at 23°C / 2 h at 35°C [2]) are the minimum, not the ideal. For a DIY repair without a test lab, doubling the datasheet's full-strength time before applying a permanent load is cheap insurance.
3. Increase bond area. Stress is force divided by area — doubling the bonded area halves the stress, and creep rate falls faster than stress does.
4. Keep it cool relative to Tg. Sustained loads belong far below the adhesive's glass-transition temperature, not near it.

Honest limits

• The 3M 9820 sheet is explicitly a preliminary TDS: "This data is not intended to be used for specification purpose" [2]. Its numbers illustrate the mechanism, not your product's rating.
• Creep data is geometry- and stress-dependent. Fassa's 0.25 mm is for its EN 1544 test specimen and load — your joint's numbers will differ, but the shape of the curve (fast early, slowing but continuing) is universal for epoxies.
• DIY has no creep test. If you can't measure it, design around it: fasteners, area, cure time, temperature.
• For the cure-time ladder that feeds this whole problem — pot life, working time, handling strength, full strength — see Repair Lab B10.

Sources

1. Fassa Bortolo — FASSA EPOXY 100 Data Sheet (two-component epoxy resin; complies with EN 1504-5/EN 1504-6). Creep under sustained tension loading (EN 1544): 0.010 ± 0.001 mm after 1 day; 0.25 ± 0.01 mm after 3 months. Tg (EN 12614): 50.0 ± 0.2°C. Minimum curing time: 7 days. URL: https://www.fassabortolo.it/wp-content/uploads/media/documents/Schede_Tecniche/FASSA/FASSA_STE_EN_FASSA-EPOXY-100_2023-09.pdf
2. 3M — 3M™ Two Component Structural Adhesive 9820 — Preliminary Technical Data Sheet (eff. 28/01/2021). Rate of strength buildup: 23°C — 2 h: 0.37 MPa creep Y; 3 h: 2.06 MPa creep Y; 4 h: 4.74 MPa creep Y (slight); 5 h: 9.22 MPa creep N. Handling strength >2 MPa; declamp 6 h at 15°C / 3–4 h at 23°C / 2 h at 35°C; "the adhesive may still creep when subjected to continuous stress including bolting and torque application" — wait 8 h at 15°C / 5 h at 23°C / 2 h at 35°C before continuous stress. "This data is not intended to be used for specification purpose." URL: https://prostech.vn/wp-content/uploads/2025/07/3m-two-component-structural-adhesive-9820-black.pdf

The 24 h mark on a typical structural epoxy is full working strength for normal loads — a joint held under constant, never-relieved stress (bolted, sprung) deserves extra cure time beyond it.

Repair Lab · Release 1 · Evidence last checked September 30, 2026 · A project of AOJEL