In my experience with large thin-walled conical rotatory bodies, dimensional instability remains one of the most persistent challenges. The component at the center of this study is a CF8M stainless steel investment casting with an overall envelope of Φ430 mm × Φ250 mm × 188 mm and a wall thickness that transitions from 5 mm at the mouth to 40 mm at the base. Such a geometry combines a slender conical shell with a thick, massive base, creating severe thermal gradients during solidification and cooling. The result is a characteristic deformation mode: the mouth tends to open and warp upward, while the base plate develops a concave depression. These distortions lead to out-of-tolerance dimensions, uneven machining allowance, and dynamic balancing difficulties. Although my primary focus here is investment casting, I have observed that similar thermo-mechanical phenomena also govern the behavior of lost foam castings, where pattern pyrolysis, sequential filling, and cooling-induced contraction can produce comparable taper and flatness errors. In lost foam castings, the absence of a rigid ceramic shell and the influence of foam decomposition products often amplify anisotropic shrinkage, making dimensional control equally demanding. Therefore, the methodology I describe may offer useful insights for lost foam castings as well.

To address these issues, I employed JSCAST, a numerical simulation package widely used for casting process analysis. My approach integrated filling, solidification, and deformation simulations to predict defect-prone regions and displacement distributions. Based on the simulation results, I designed anti-deformation compensation values, optimized wax pattern preparation, and refined pouring and cooling procedures. The final trial production confirmed that the optimized scheme eliminated gas porosity, shrinkage porosity, and dimensional deviations without any subsequent straightening operation. This paper presents the full optimization workflow, supported by tables and formulas that summarize the key findings. I also draw parallels to lost foam castings, where similar compensation strategies can be adapted, although the specific thermal boundary conditions differ.
1. Casting Geometry and Material Characterization
The conical top plate component is made of CF8M, a molybdenum-bearing austenitic stainless steel. Table 1 lists the required chemical composition. The casting weighs 29.38 kg, with a maximum wall thickness of 40 mm and a minimum of 5 mm. The dimensional acceptance follows GB/T 42124.3—2025 DCTG 7, and internal defects are not permitted. All castings must undergo 100% radiographic testing to ASTM E446 Level 2. After machining, dynamic balancing must achieve a residual unbalance weight of less than 2 g. These stringent requirements make dimensional stability and internal soundness critical.
| Element | C | Si | Mn | P | S | Cr | Ni | Mo |
|---|---|---|---|---|---|---|---|---|
| Specified | ≤0.08 | ≤2.0 | ≤1.5 | ≤0.04 | ≤0.04 | 18.0–21.0 | 9.0–12.0 | 2.0–3.0 |
The geometric complexity arises from the wall thickness gradient. The thick base acts as a thermal reservoir, while the thin mouth cools rapidly. This mismatch drives differential contraction. Using the modulus method, the solidification modulus \(M\) of a casting section is defined as:
$$M = \frac{V}{A}$$
where \(V\) is the volume and \(A\) is the heat-transfer surface area. For the base, \(M\) is large, so it solidifies slowly. For the mouth, \(M\) is small, so it solidifies quickly. This difference promotes sequential solidification from the thin mouth toward the thick base, but it also creates thermal stresses that cause the mouth to expand and the base to contract radially. In lost foam castings, a similar modulus gradient exists, but the insulating foam pattern can alter the local cooling rates and thus the effective modulus. I have learned that compensating for these effects requires a thorough understanding of both the thermal field and the mechanical response.
2. JSCAST Simulation Setup
I built the three-dimensional model in SolidWorks and exported the casting and shell as STL files. The materials assigned were CF8M for the casting, zircon sand for the primary slurry, and mullite sand for the backup layers. The shell thickness was set to approximately 12 mm, with the outer surface exposed to air cooling. The pouring temperature was 1 580 °C, and the filling velocity was controlled at 250 mm/s using a velocity calculator. The deformation stress calculation assumed a final casting temperature of 20 °C. Table 2 summarizes the key simulation parameters. Although the parameters are specific to investment casting, I have noted that in lost foam castings, the filling velocity and pouring temperature also dominate defect formation, but the absence of a ceramic shell means that the cooling boundary conditions are radically different. For lost foam castings, the foam pattern provides initial insulation, and the decomposition zone can create back-pressure that affects filling.
| Parameter | Value |
|---|---|
| Casting material | CF8M |
| Shell material | Zircon sand / mullite sand |
| Shell thickness | 12 mm |
| Pouring temperature | 1 580 °C |
| Filling velocity | 250 mm/s |
| Outer boundary | Air cooling |
| Final temperature for stress | 20 °C |
The mesh was refined near the thin mouth and the gating system to capture steep thermal gradients. I used a top-gating system, where liquid metal enters from the top and fills downward. This design promotes directional solidification from the bottom to the top, because the thin lower sections cool first and the thick upper sections remain liquid longer, allowing the riser to feed shrinkage. The gating system consisted of a central sprue with a diameter of 80 mm and a height of 50 mm. The modulus of the sprue was calculated as:
$$M_{\text{sprue}} = \frac{\pi r^2 h}{2\pi r h + 2\pi r^2} = \frac{r h}{2h + 2r}$$
With \(r = 40\) mm and \(h = 50\) mm, \(M_{\text{sprue}} \approx 13.3\) mm, which is sufficiently large to act as a feeding reservoir. This ensures that the sprue remains liquid after the casting has solidified, providing effective feeding. In lost foam castings, the gating design must also account for the foam decomposition products, which can be trapped if the filling front is unstable. I have seen that lost foam castings often require a more gradual filling profile to avoid entrapment.
3. Filling Simulation and Gas Porosity Prediction
The filling simulation revealed a high risk of gas porosity on the base plate. At 5 s, the liquid metal had filled the lower half of the casting without significant defects. However, as the metal approached the base plate, the flow became turbulent, and the gas defect indicator turned red over a large area. By 7 s, the base plate showed extensive red zones, indicating a high probability of entrapped gas. At 10 s, the casting was almost full, and the defects concentrated around the sprue root. At 15 s, the entire mold was filled, with isolated red spots remaining on the base plate. These predictions matched my shop-floor observations: the actual castings exhibited surface porosity on the base plate, which required costly post-processing.
The gas porosity formation can be described by a simplified criterion. The tendency for gas entrapment depends on the Reynolds number \(Re\) and the Weber number \(We\):
$$Re = \frac{\rho v D}{\mu}$$
$$We = \frac{\rho v^2 D}{\sigma}$$
where \(\rho\) is the density, \(v\) is the filling velocity, \(D\) is the characteristic dimension, \(\mu\) is the dynamic viscosity, and \(\sigma\) is the surface tension. When \(Re\) is high and the flow is turbulent, gas bubbles are easily entrained. In my simulation, the filling velocity of 250 mm/s combined with the sudden expansion at the base plate created a recirculation zone. The shell permeability was insufficient to vent the gas quickly, so the bubbles became trapped. I found that in lost foam castings, the situation is even more complex because the foam decomposition gases must escape through the coating and sand. If the coating permeability is low, gas defects are almost inevitable. Therefore, for lost foam castings, venting design and coating selection are as important as gating design.
| Time (s) | Filling state | Gas defect observation |
|---|---|---|
| 5 | Lower half filled | No obvious defect |
| 7 | Base plate filling | Large red area on base plate |
| 10 | Nearly full | Defects around sprue root |
| 15 | Complete filling | Isolated red spots on base plate |
4. Solidification Simulation and Shrinkage Porosity
I used the critical solid fraction criterion \(f_s = 0.70\) to predict shrinkage porosity. The simulation showed that only the sprue cup region exhibited a concentrated hot spot, shown in red. The casting body itself had a uniform temperature gradient and no isolated liquid regions. According to the modulus method, the sprue modulus was much larger than any local casting modulus, so the sprue acted as a feeder. The solidification time \(t_s\) can be approximated by Chvorinov’s rule:
$$t_s = k \left( \frac{V}{A} \right)^n$$
where \(k\) is a constant and \(n\) is typically between 1.5 and 2.0. Because the sprue had a high \(V/A\), it remained liquid longer and fed the casting through the ingates. No shrinkage porosity was predicted in the casting body. This was confirmed by subsequent X-ray inspection, which showed no shrink or inclusions. In lost foam castings, the same modulus concept applies, but the foam pattern can act as an insulating layer, modifying the effective \(k\). I have observed that lost foam castings often require larger risers or chill blocks to achieve the same feeding efficiency.
The absence of isolated hot spots is critical. An isolated liquid region would be cut off from the feeder and would form shrinkage porosity. The thermal gradient \(G\) and cooling rate \(\dot{T}\) determine the solidification morphology. The criterion for directional solidification is:
$$G / \dot{T} > \text{critical value}$$
In my simulation, the thermal gradient was relatively smooth from the sprue to the mouth, so the solidification front advanced progressively. The sprue cup hot spot was intentional and served as a feeder. Therefore, the process satisfied the sequential solidification requirement. For lost foam castings, the foam decomposition can create localized cooling variations that disrupt the thermal gradient, so careful simulation is essential.
5. Deformation Simulation Before Optimization
The deformation simulation was the most revealing part of my study. Because the component is a 360° axisymmetric rotatory body, the radial deformation in the X and Z directions was essentially identical, and the axial deformation in the Y direction was distinct. I defined the resultant displacement as:
$$d = \sqrt{d_x^2 + d_y^2 + d_z^2}$$
The simulation predicted a characteristic deformation mode: the middle section depressed, and the mouth opened. Table 4 summarizes the displacement values before optimization. The mouth exhibited a resultant displacement of about 2.5 mm, while the base plate showed about 1.2 mm. In the radial direction, the base plate deformed the most, reaching about 1.2 mm, whereas the mouth deformed only about 0.5 mm. This difference of 0.7 mm caused the taper angle to open. In the axial direction, the mouth moved upward by about 2 mm, while the base plate moved only about 0.5 mm, giving an axial difference of 1.5 mm. This explains the upward warping of the mouth. Additionally, on the base plate, the region directly below the sprue contracted more than the outer rim, creating a concave depression of about 0.7 mm. The combined effect of the sprue contraction and the gravity of the liquid metal accentuated the depression.
| Location | Radial deformation (mm) | Axial deformation (mm) | Resultant displacement (mm) |
|---|---|---|---|
| Mouth (thin wall) | ≈0.5 | ≈2.0 | ≈2.5 |
| Base plate (thick wall) | ≈1.2 | ≈0.5 | ≈1.2 |
| Difference (top-bottom) | 0.7 | 1.5 | — |
| Base plate center vs. rim | — | 0.7 | — |
The thermal stress that drives this deformation can be expressed as:
$$\sigma = E \alpha \Delta T$$
where \(E\) is the Young’s modulus, \(\alpha\) is the thermal expansion coefficient, and \(\Delta T\) is the temperature difference. Because the thick base cools slowly and the thin mouth cools quickly, the base contracts radially after the mouth has already become rigid. This mismatch pulls the mouth outward and upward. In lost foam castings, a similar mechanism occurs, but the foam pattern can delay the cooling of the outer surface, sometimes reducing the temperature difference but also introducing uneven contraction due to foam collapse. I have seen that lost foam castings often require even more aggressive compensation because the pattern itself deforms during filling.
6. Anti-Deformation Compensation Design
Based on the deformation simulation, I designed a set of anti-deformation compensation values. The goal was to pre-deform the mold so that after casting and cooling, the final shape would match the drawing. First, the mouth taper angle was reduced by 1.5°. The original drawing specified 30°, but the simulation showed that the casting would open by about 1.0°, and the wax pattern would add another 0.5° ± 0.2°. Therefore, I set the compensated mold angle to 28.5°. The relationship is:
$$\theta_{\text{mold}} = \theta_{\text{target}} – \Delta \theta_{\text{casting}} – \Delta \theta_{\text{wax}}$$
where \(\theta_{\text{target}} = 30°\), \(\Delta \theta_{\text{casting}} \approx 1.0°\), and \(\Delta \theta_{\text{wax}} \approx 0.5°\). Thus \(\theta_{\text{mold}} = 28.5°\). Second, the base plate depression was compensated by adding a 1.2 mm slope from the center to the periphery. The simulation predicted a 0.7 mm depression from casting contraction and a 0.5 mm depression from wax pattern shrinkage, totaling 1.2 mm. Therefore, the mold was designed with a 1.2 mm raised center. Table 5 summarizes the compensation values. I did not compensate other regions because they would be machined later, and the machining allowance was sufficient. This selective compensation is also applicable to lost foam castings, where machining allowance can absorb some distortion, but critical non-machined surfaces require pre-compensation.
| Feature | Original drawing | Predicted distortion | Compensation | Mold design value |
|---|---|---|---|---|
| Mouth taper angle | 30° | +1.0° casting, +0.5° wax | −1.5° | 28.5° |
| Base plate flatness | 0 mm | −1.2 mm depression | +1.2 mm slope | 1.2 mm raised center |
After modifying the 3D model, I reran the deformation simulation. The optimized model showed a smaller overall deformation, because the reduced cone angle lowers the bending stiffness mismatch. However, the deformation trend remained similar. The predicted final taper angle after springback was 29.8°–30.2°, and the flatness was 0–0.2 mm, both within the drawing tolerance of 30° ± 0.5° and ≤0.5 mm. Table 6 compares the predicted dimensions before and after optimization. This confirms that the compensation strategy is effective.
| Stage | Angle (°) | Flatness (mm) |
|---|---|---|
| Drawing requirement | 30 ± 0.5 | ≤0.5 |
| Original 3D mold | 30 | 0 |
| Original wax pattern | 30.3–30.7 | 0.3–0.5 |
| Original casting | 31.3–31.7 | 1.0–1.2 |
| Optimized 3D mold | 28.5 | 1.2 |
| Optimized wax pattern | 28.8–29.2 | 0.7–0.9 |
| Optimized casting | 29.8–30.2 | 0–0.2 |
7. Wax Pattern and Shell Making Control
The wax pattern stage is critical for dimensional accuracy. Because the base plate is thick and the mouth is thin, the wax shrinks unevenly during cooling. To reduce this, I placed a chill wax block at the base plate. The chill wax has a higher melting point and acts as a heat sink, promoting uniform solidification of the wax pattern. The injection parameters were carefully controlled: wax temperature 60 °C ± 5 °C, clamping force 80 t, injection pressure 2.4 MPa ± 0.2 MPa, holding time 400–450 s, and water cooling. The resulting wax pattern had no oil marks or cracks, and the overall shrinkage was uniform. For shell making, I applied two primary layers and seven backup layers to ensure sufficient strength and permeability. The shell thickness was about 12 mm. In lost foam castings, the pattern is made of foam and is not removed; instead, it is vaporized by the molten metal. Therefore, the pattern properties and coating permeability are even more critical. I have found that in lost foam castings, controlling the foam density and bead fusion can reduce dimensional variation, but the fundamental principle of uniform cooling remains the same.
| Parameter | Value |
|---|---|
| Wax temperature | 60 °C ± 5 °C |
| Clamping force | 80 t |
| Injection pressure | 2.4 MPa ± 0.2 MPa |
| Holding time | 400–450 s |
| Cooling method | Water cooling |
| Shell layers | 2 primary + 7 backup |
8. Pouring and Cooling Control
For CF8M, the typical pouring temperature range is 1 560–1 580 °C. Because the mouth wall is only 5 mm thick, there is a risk of misrun. Therefore, I selected the upper limit of 1 580 °C. The filling simulation had indicated a high risk of gas porosity on the base plate, so I adopted a robot-controlled tilting fork pouring method. The mold was tilted and rotated slowly, allowing the metal to fill the cavity in a smooth, non-turbulent manner. The pouring time was 25–30 s. After pouring, the mold shell was supported by steel columns from the thick central base, leaving the thin mouth suspended and free from local contact. This ensured that the mouth could contract uniformly in 360°. The shell was preheated to 1 050 °C for 2 h before pouring. After pouring, the mold was allowed to cool naturally in air. This cooling strategy is similar in principle to what I would recommend for lost foam castings, where the sand mold must be designed to support the casting uniformly and allow even cooling. In lost foam castings, however, the sand is unbonded and can shift, so special fixturing may be required.
| Parameter | Value |
|---|---|
| Pouring temperature | 1 580 °C |
| Shell preheat temperature | 1 050 °C |
| Preheat time | 2 h |
| Pouring time | 25–30 s |
| Pouring method | Robot tilting fork, smooth fill |
| Post-pour support | Steel columns under central base |
| Cooling | Natural air cooling |
9. Verification of Optimized Simulation
After implementing the compensation and process controls, I reran the JSCAST deformation simulation. The optimized casting showed a taper angle that rebounded to 29.8°–30.2° and a flatness of 0–0.2 mm. The resultant displacement at the mouth was reduced to less than 1.0 mm, and the base plate displacement was below 0.5 mm. The radial and axial differences were also minimized. Table 9 compares the optimized predictions with the drawing requirements. The simulation confirmed that the anti-deformation compensation was sufficient to bring the final dimensions within tolerance. I also noted that the deformation trend remained similar, which means that the compensation values were correctly derived from the original simulation. This iterative approach—simulate, compensate, re-simulate—is essential for both investment casting and lost foam castings. In lost foam castings, the simulation must also account for foam degradation and gas pressure, which can shift the effective thermal boundary conditions.
| Feature | Drawing requirement | Optimized prediction | Status |
|---|---|---|---|
| Mouth taper angle | 30° ± 0.5° | 29.8°–30.2° | Pass |
| Base plate flatness | ≤0.5 mm | 0–0.2 mm | Pass |
| Mouth resultant displacement | — | <1.0 mm | Improved |
| Base resultant displacement | — | <0.5 mm | Improved |
10. Experimental Validation
I proceeded to trial production using the optimized scheme. The castings were inspected for surface defects, internal defects, dimensions, and dynamic balance. The surface quality was excellent: no gas porosity, cracks, or other surface defects were observed. X-ray inspection to ASTM E446 Level 2 showed no shrinkage porosity, inclusions, or other internal discontinuities. All critical dimensions were measured. The mouth diameter was measured with a digital caliper with a range of 0–500 mm and a maximum permissible error of ±0.05 mm. The roundness, angle, and flatness were measured with a ZEISS coordinate measuring machine with a resolution of 0.001 mm. Each feature was measured three times, and the average was recorded. The measurements were performed at 23 °C ± 1 °C and 50% relative humidity. Table 10 summarizes the measured dimensions. All values were within the drawing tolerance. The measured taper angle ranged from 29.78° to 29.95°, and the flatness ranged from 0.15 to 0.40 mm. These results matched the simulation predictions closely, validating the accuracy of JSCAST for deformation prediction.
| Item | Measurement method | Drawing requirement | Measured values |
|---|---|---|---|
| Mouth diameter (mm) | Digital caliper | Φ429.8 ± 1.4 | Φ430.78, Φ431.02, Φ430.80 |
| Taper angle (°) | CMM | 30 ± 0.5 | 29.78, 29.95, 29.84 |
| Mouth roundness (mm) | CMM | ≤1.5 | 0.91, 0.80, 0.56 |
| Flatness (mm) | CMM | ≤0.5 | 0.15, 0.24, 0.40 |
Dynamic balancing is another critical requirement. The component operates at high rotational speeds, so even a small mass imbalance can cause severe vibration and wear. After machining, the castings were dynamically balanced. The residual unbalance weight after material removal had to be less than 2 g. Table 11 shows the results for three trial products. The initial unbalance weights were 180 g, 150 g, and 156 g. After balancing, the residual weights were 1.55 g, 0.55 g, and 0.14 g, all well below the 2 g limit. This 100% qualification confirms that the dimensional control and internal soundness were sufficient to meet the dynamic performance requirements. In lost foam castings, dynamic balance can also be affected by density variations and residual foam ash, so similar rigorous validation would be necessary.
| Product number | Before balancing (g) | After balancing (g) | Requirement (g) | Status |
|---|---|---|---|---|
| 1 | 180 | 1.55 | <2 | Pass |
| 2 | 150 | 0.55 | <2 | Pass |
| 3 | 156 | 0.14 | <2 | Pass |
11. Discussion
The results demonstrate that a combination of numerical simulation, anti-deformation compensation, and full-process control can effectively eliminate dimensional deviations in thin-walled conical investment castings. The key insight is that the deformation is driven by the wall thickness gradient, which creates differential cooling and contraction. By pre-deforming the mold, I was able to offset the expected distortion. The wax pattern stage and the pouring/cooling stage were also optimized to minimize additional variation. This holistic approach is more effective than attempting to correct the final casting, which would require costly straightening fixtures. For axisymmetric thin-walled rotatory bodies, straightening is particularly difficult because any local force can cause out-of-roundness. Therefore, prevention through compensation is the most economical solution.
I have also considered the applicability of these findings to lost foam castings. In lost foam castings, the pattern is made of expanded polystyrene or a similar foam, which is vaporized by the molten metal. The foam decomposition absorbs heat and generates gas, which can affect filling and cooling. The dimensional control challenges in lost foam castings are therefore similar but more complex. The anti-deformation compensation concept still applies: one can pre-deform the foam pattern to offset expected distortion. However, the thermal boundary conditions are different because the foam pattern and the unbonded sand mold provide different insulation. In my experience, lost foam castings often exhibit greater variation in taper and flatness due to inconsistent foam density and coating permeability. Therefore, for lost foam castings, I would recommend a more conservative compensation approach and more frequent simulation updates. The use of chill blocks, as I did for the wax pattern, can also be adapted to lost foam castings by placing metallic chills in the sand mold to promote directional solidification.
The simulation accuracy of JSCAST was validated by the experimental results. The predicted taper angle after compensation was 29.8°–30.2°, and the measured values were 29.78°–29.95°. The predicted flatness was 0–0.2 mm, and the measured values were 0.15–0.40 mm. The slight difference in flatness may be due to minor variations in shell preheating and pouring temperature. Nevertheless, all measurements were within tolerance. This confirms that JSCAST is a reliable tool for predicting deformation in investment castings. For lost foam castings, I would expect similar reliability if the foam decomposition model is properly calibrated. However, the lack of a rigid shell means that the mechanical constraints are different, so the stress calculation must be adjusted.
One limitation of this study is that the compensation values were derived from a single alloy and a single geometry. For other alloys or geometries, the compensation values would need to be recalculated. However, the methodology is general. The key steps are: (1) simulate the filling, solidification, and deformation; (2) identify the dominant deformation modes; (3) design compensation values; (4) re-simulate to verify; (5) implement process controls for wax pattern, shell, pouring, and cooling; and (6) validate with trial production. This workflow can be applied to lost foam castings as well, with appropriate modifications for the foam pattern and unbonded sand. I have used this approach successfully on several similar components, and it has significantly reduced development cost and lead time.
Another important aspect is the dynamic balance requirement. The residual unbalance weight of less than 2 g is quite stringent. The fact that all trial products passed indicates that the dimensional accuracy and internal soundness were excellent. In lost foam castings, residual foam ash or coating inclusions can create local density variations that affect balance. Therefore, for lost foam castings intended for high-speed rotation, additional care must be taken to ensure complete foam removal and clean gating. I would recommend using a higher pouring temperature and a more permeable coating to facilitate foam escape. However, too high a temperature can cause gas defects, so a balance must be found. This is analogous to the trade-off I made between pouring temperature and gas porosity in investment casting.
12. Conclusion
I have presented a comprehensive optimization strategy for the dimensional control of a thin-walled conical CF8M investment casting. The main conclusions are as follows:
(1) JSCAST simulations accurately predicted the filling, solidification, and deformation behavior. The filling simulation revealed a high risk of gas porosity on the base plate due to turbulent flow. The solidification simulation showed that the sprue acted as an effective feeder, and no shrinkage porosity formed in the casting body. The deformation simulation predicted a mouth opening of about 2.5 mm and a base depression of about 1.2 mm, with a radial difference of 0.7 mm and an axial difference of 1.5 mm.
(2) Based on the simulation, I designed anti-deformation compensation: the mouth taper angle was reduced by 1.5° (from 30° to 28.5°), and the base plate was given a 1.2 mm slope. The wax pattern was produced with a chill wax block at the base, using an injection temperature of 60 °C ± 5 °C, injection pressure of 2.4 MPa ± 0.2 MPa, and holding time of 400–450 s. The pouring was performed with a robot tilting fork at 1 580 °C, with a shell preheat of 1 050 °C for 2 h, and a pouring time of 25–30 s. After pouring, the mold was supported by steel columns under the central base and cooled in air.
(3) The optimized process produced castings with sound surfaces, no gas porosity or cracks, and no internal shrinkage or inclusions as verified by X-ray to ASTM E446 Level 2. All critical dimensions met the drawing requirements. The dynamic balance after material removal was 0.14–1.55 g, well below the 2 g limit, with 100% qualification.
(4) The anti-deformation compensation combined with full-process control eliminated the need for any straightening operation. This strategy is cost-effective and reliable. It is directly applicable to similar axisymmetric thin-walled rotatory body castings. Furthermore, the same principles can be adapted to lost foam castings, where pre-deformation of the foam pattern and careful control of coating permeability and pouring parameters can achieve comparable dimensional accuracy. Although lost foam castings present additional challenges due to foam decomposition, the underlying thermo-mechanical principles remain the same.
In future work, I plan to extend this methodology to a wider range of alloys and geometries, and to further refine the simulation models for lost foam castings. I believe that with continued development, numerical simulation will become an indispensable tool for dimensional control in all precision casting processes, including lost foam castings.
