I have conducted a systematic investigation into the formation mechanism of the underside shrinkage—a detrimental sand casting defect observed in ZA27 alloy castings produced in sand molds. This defect manifests as irregular upward cavities on the bottom surface of castings, often resembling a honeycomb pattern, and in severe cases leads to scrapping of the casting. My research aimed to elucidate the process and root causes of this sand casting defect through targeted experiments and theoretical analysis.
In my experimental work, I used a ZA27 alloy with a nominal composition of 25–28% Al, 2.0–2.5% Cu, 0.01–0.015% Mg, balance Zn. Melting was carried out in a graphite crucible within a resistance furnace using a Al–50%Cu master alloy and pure Zn, Al, and Mg ingots. The test casting was a stepped block with dimensions enabling evaluation of solidification behavior along the height. Two variants were employed: one with no chill and one with a 60-mm-thick aluminum chill placed at the bottom of the mold. Clay-bonded silica sand molds were used. Temperature profiles were recorded using NiCr–NiAl thermocouples placed at the center of the casting at various heights, monitored by a six-channel automatic recorder. After solidification, the castings were visually inspected and sectioned for chemical analysis (using spectroscopy), density measurement (by Archimedes method), and microstructural examination. The effective density was calculated to account for compositional variations using the formula:
$$P_{\text{eff}} = \frac{M}{M_c} \cdot P_{\text{measured}}$$
where \(M = 2.7 \times \%\text{Al} + 8.9 \times \%\text{Cu} + 7.1 \times \%\text{Zn}\) (based on pure densities of 2.7, 8.9, and 7.1 g/cm³ for Al, Cu, and Zn respectively), \(M_c\) is the density coefficient calculated from the actual local composition, and \(P_{\text{measured}}\) is the measured density.
The image above illustrates a typical sand casting defect; my investigation focused specifically on the underside shrinkage defect that appears in ZA27 alloy castings.
The cooling curves I recorded for the no-chill casting are shown conceptually in my analysis. The curve at the bottommost location (height 23 mm) consistently remained below the curves at higher positions throughout solidification, indicating slower cooling at the bottom. Chemical analysis revealed severe gravity segregation. Table 1 presents the chemical composition and density data for the no-chill case.
| Height (mm) | Al (wt%) | Cu (wt%) | Zn (wt%) | Measured Density (g/cm³) | Effective Density (g/cm³) |
|---|---|---|---|---|---|
| 230 (top) | 29.0 | 2.12 | 68.88 | 4.74 | 5.12 |
| 212 | 29.2 | 2.28 | 68.52 | 4.80 | 5.16 |
| 187 | 27.4 | 2.06 | 70.54 | 4.81 | 5.33 |
| 173 | 27.2 | 2.14 | 70.66 | 4.78 | 5.22 |
| 137 | 26.8 | 2.16 | 71.04 | 5.12 | 5.55 |
| 123 | 25.2 | 2.30 | 72.50 | 5.16 | 5.68 |
| 87 | 22.4 | 2.35 | 75.25 | 5.33 | 5.92 |
| 73 | 23.4 | 2.22 | 74.38 | 5.22 | 5.80 |
| 37 | 23.6 | 2.17 | 74.23 | 5.22 | 5.75 |
| 23 (bottom) | — | — | — | — | — |
Note: The 23 mm location could not be sampled due to severe underside shrinkage cavity.
The data clearly show that the top sections are richer in aluminum and leaner in zinc, while the bottom sections are zinc-rich. Copper remains relatively uniform. This gravity segregation is a direct consequence of the wide freezing range of ZA27 (487–375°C) and the low density of the primary α phase, which is enriched in aluminum. The primary α crystals have a composition of approximately Zn 5.6%, Al 42.5%, Cu 1.9% (reported from energy-dispersive spectroscopy analysis of quenched samples), yielding a density about 85% that of the liquid alloy. In sand casting, slow cooling allows ample time for these light α dendrites to float upward, while zinc-rich liquid descends, establishing a compositional gradient. This sand casting defect is fundamentally linked to the alloy’s solidification mode.
Using the Scheil equation for non-equilibrium solidification, I estimated the solid fraction as a function of temperature:
$$f_s = 1 – \left( \frac{T_M – T}{T_M – T_L} \right)^{\frac{1}{k-1}}$$
where \(T_M = 660.3^\circ\text{C}\) (melting point of aluminum), \(T_L\) is the liquidus temperature, \(k = 0.865\) is the equilibrium distribution coefficient for Zn in α-Al, and \(f_s\) is the solid fraction. Literature indicates that bulk feeding (macro-feeding) ceases when the solid fraction reaches about 75% (\(f_s = 0.75\)). The corresponding temperature, \(T_{\text{stop}}\), can be calculated for different locations based on their local liquidus temperature. Table 2 lists the liquidus temperature (derived from the Al-Zn phase diagram for the measured Al content) and the calculated feeding stop temperature for both no-chill and chill cases.
| Height (mm) | Al (wt%) (No Chill) | \(T_L\) (℃) (No Chill) | \(T_{\text{stop}}\) (℃) (No Chill) | Al (wt%) (With Chill) | \(T_L\) (℃) (With Chill) | \(T_{\text{stop}}\) (℃) (With Chill) |
|---|---|---|---|---|---|---|
| 230 | 29.0 | 500 | 467 | 29.1 | 500 | 467 |
| 187 | 27.4 | 490 | 455 | 28.2 | 493 | 459 |
| 137 | 26.8 | 488 | 453 | 24.1 | 484 | 448 |
| 73 | 23.4 | 482 | 445 | 24.0 | 484 | 448 |
| 23 | — | — | — | 23.5 | 482 | 445 |
For the no-chill casting, the liquidus temperature at the bottom (23 mm) is substantially lower than at the top by about 25°C. During solidification, the actual temperature difference between top and bottom never exceeded 20°C (as observed from cooling curves). This means that the bottom begins to solidify later than the top, reversing the normal bottom-to-top solidification sequence typical of most alloys. The consequence is that the upper regions, which contain more aluminum, solidify first and contract, generating a negative pressure that can only be relieved by feeding from the still-molten lower regions—a reversed feeding direction upward against gravity. This abnormal feeding direction is the direct cause of the sand casting defect, as the liquid metal must flow upward through the mushy zone to compensate for shrinkage in the upper solidified regions. The rough, honeycomb surface of the cavity results from interdendritic feeding under capillary forces.
In contrast, when a 60-mm chill was placed at the bottom, the temperature gradient was drastically altered. The bottom cooled faster, and the solidification sequence returned to normal: bottom-to-top. I evaluated the feeding stop temperatures for the chill case using the cooling curves. At the moment the bottom (23 mm) reached its \(T_{\text{stop}}=445^\circ\text{C}\), the temperature at 73 mm was still 487°C, well above its \(T_{\text{stop}}=448^\circ\text{C}\). Thus, the upper regions remained capable of feeding the lower ones via gravity-assisted flow. No sand casting defect appeared on the bottom surface of the chill casting. The microstructures also differed: the chill casting showed finer dendrites near the bottom due to rapid cooling, and the gravity segregation was less pronounced, although still present.
These findings underscore that the underside shrinkage defect in sand casting of ZA27 is a unique solidification-driven phenomenon. The wide freezing range promotes a pasty solidification mode where the entire casting remains in a two-phase region for long periods. Weight segregation produces a composition gradient that inverts the liquidus temperature profile, leading to a top-down solidification sequence. The solidification contraction from the top regions then draws liquid from the bottom, creating upward feeding that cannot fully compensate, resulting in the sand casting defect. The defect occurs late in solidification when feeding becomes interdendritic, explaining the rough cavity surface. By introducing a chill, the thermal gradient is restored, the solidification order is corrected, and the sand casting defect is eliminated.
I also note that insufficient chill thickness, while preventing underside shrinkage, could lead to internal shrinkage porosity in the midsection of the casting. This is because the bottom solidifies too quickly, but the upper part still retains reversed solidification, creating a zone of conflicting feeding directions. Thus, proper chill design is critical for defect-free castings.
In conclusion, my research demonstrates that the sand casting defect known as underside shrinkage in ZA27 alloy originates from the interplay of pasty solidification and gravity segregation. The abnormal solidification sequence and reversed feeding are the root causes. Practical measures to counteract the defect include the application of chills to enforce a favorable temperature gradient. These insights provide a scientific basis for process optimization to produce sound ZA27 castings by sand casting.