Same AFS, Different Results: How Thermal-Expansion Curves Drive Veining and Dimensional Drift

1) Why AFS alone is not predictive

AFS (or D50) is a convenient particle-size midpoint, but it says little about:

• Temperature-dependent expansion of the sand skeleton

• Elastic vs. inelastic strain accumulation during thermal shock

• PSD tails (D10/D90) that change packing and stress transfer

• Binder and coating interactions that amplify micro-cracking

Two sands with AFS = 50 may show 2–3× differences in micro-strain through the 600–900 °C window, yielding starkly different veining incidence even under identical pouring conditions.

2) The physics in one picture

Imagine a simple band-mold constrained by a core box. As liquid metal raises the surface temperature, the near-face sand tries to expand while the interior is cooler and restrains it. The resulting tensile stress peaks when the sand’s thermal expansion rate (dε/dT) is steepest. If that peak coincides with binder softening or coating micro-cracks, stress relief occurs as veins—thin fissures that print into the metal.

What matters most is curve shape:

• Low(ish) overall expansion is useful, but

• A flatter slope (reduced Δε/ΔT from 650–900 °C) is what suppresses crack initiation, all else equal.

3) Measuring it correctly: from CTE to full curves

Method: Use dilatometry on compacted sand specimens (or bonded cores) to record ε(T) from ambient to ~1,000 °C at 5–10 °C/min.
Report not just a single CTE but the full set:

• ε(T) at 100 °C increments

• dε/dT (expansion slope) and its peak value

• T* (inflection temperature) where slope changes sign or drops sharply

• Hysteresis on cooling (irreversible strain, Δε_irrev)

Recommended acceptance line (example for aluminum core work; refine locally):

• Peak dε/dT ≤ defined threshold in 650–900 °C

• Δε_irrev after one heat cycle ≤ X% of total ε

• Curve repeatability (R&R across lots) within ±10%

4) A practical test matrix you can run next week

A. Materials

• Sand A and Sand B, both AFS ≈ 50 (or your standard)

• Same core recipe, same coating thickness and dry-out

B. Measurements

1. Dilatometry: ε(T), dε/dT, T*

2. Core mechanicals: shell strength at 200/300 °C

3. Casting trial (paired molds): record veining % area, penetration score, Ra, release force

4. PSD tails: D10/D90 growth after 3/5 reclaim cycles, LOI vs. gas porosity

C. Analysis
Plot veining area vs. peak dε/dT; you’ll typically see a monotonic trend even when AFS and D50 are identical.

5) Reading the curve: what to look for

• Steep mid-range slope (650–900 °C) → high tensile stress, higher veining probability

• Early inflection (T* below ~700 °C) → stress capped sooner, often fewer veins

• Large irreversible strain on cool-down → dimensional drift on subsequent cycles

• PSD tail growth after reclaim → tighter packing near the face, amplifying stress

Rule of thumb

If two sands share AFS, choose the one with lower, broader peak in dε/dT across the 600–900 °C band—even if total ε differs only slightly.

6) Line controls that actually move the needle

1. Sand temperature (pre-mold)

   • Keep ΔT by shift < 5 °C; warm sand expands less abruptly and reduces micro-crack initiation.

2. Coating dry-out profile

   • Under-dried coatings store moisture and promote spall; over-dried, brittle coats crack under the same stress. Use weight-loss or IR checks.

3. Minimum viable binder

   • Excess binder increases LOI, gas, and local softening; reduce until gas porosity rises, then back off 5–10%.

4. Gating/velocity

   • Limit local impingement where the expansion slope peaks; match metal speed to coating and sand face strength.

5. Reclaim stability

   • Track PSD tails and LOI at 3/5/8 cycles; cap fines or blend new/add-back to hold dε/dT repeatable.

6. Cooling path symmetry

   • Uneven cooling turns expansion into bending; add local chills or adjust risers to avoid hot-spot asymmetry.

7) Case vignette (illustrative)

• Two sands: both AFS = 50, identical D50; Sand-A shows peak dε/dT 1.3× higher between 720–860 °C than Sand-B.

• Trial: same core design and coating; 40 paired pours on an aluminum pump housing.

• Results: Sand-A veining area median 1.8%, Ra 6.4 µm; Sand-B veining 0.6%, Ra 5.1 µm. After 5 reclaim cycles, Sand-A fines ↑1.5 %, dε/dT peak shifts +20 °C; Sand-B stable.

• Conclusion: the curve shape and its stability across reclaim cycles, not AFS, predicted the defect delta.

8) What to put in your weekly dashboard

• Peak dε/dT (from periodic lab runs) with lot traceability

• Sand temperature by shift, target band and alarms

• PSD tails (D10/D90) and fines % after reclaim

• LOI, Ra, veining % area, machining minutes per part

• Out-of-window count (number of specs exceeded per 100 molds)

9) Suggested visuals for your article/page

• Overlay of ε(T) and dε/dT for two sands with AFS = 50

• Box plots of veining % area by sand type and reclaim cycle

• Diagram of stress formation near the mold face with coating micro-cracks

• Control-chart snippets for sand temperature and PSD tails

10) Key takeaways

• AFS is a necessary but insufficient descriptor for casting behavior.

• Thermal-expansion curve shape—especially the slope peak in the 600–900 °C band—tracks closely with veining and dimensional stability.

• Bring ε(T) data, PSD tails, LOI, and sand temperature into one control stack; that is how foundries turn “measured” into managed.

Mini-FAQ

Q: Do I need a dilatometer to start?
A: It helps, but you can begin by correlating veining area, Ra, and reclaim-cycle PSD tails; when patterns emerge, lab curves explain the “why.”

Q: Is lower total expansion always better?
A: Not always. Slope and inflection timing matter more than the absolute endpoint.

Q: How often should we re-characterize curves?
A: At least per new supplier lot or when reclaim mix changes beyond a set threshold (e.g., ±10% fines).