I carried out a first-person numerical and experimental study of lost foam casting for a high-pressure valve body. My aim was to reduce shrinkage porosity, shrinkage cavities, hot tearing, and slag entrapment while keeping the process economically suitable for foundry production. I used three-dimensional modeling, finite-element-based casting simulation, response surface methodology, and a small experimental validation program. Throughout the work, I treated lost foam casting as a coupled system involving foam degradation, gas pressure evolution, liquid metal flow, heat transfer, solidification, solute redistribution, stress development, and defect formation. The results showed that the gating system and the process parameters must be designed together if lost foam casting is to produce a sound valve body.
The valve body I studied was a high-pressure component used for flow control in pipelines. Its geometry was complex, non-symmetric, and contained several internal cavities. The casting alloy was cast steel ZG230-450. I selected this material because it offers good toughness, plasticity, and castability. I first constructed the three-dimensional valve body and then added gating and feeding systems. I compared a top pouring design with a side pouring design under identical numerical conditions. After selecting the side pouring system, I optimized pouring temperature, negative pressure, and pouring speed by using a Box-Behnken design. I also explored a secondary development route for the heat-transfer part of ProCAST through Microsoft Visual Studio. Finally, I validated the optimized concept with a simplified casting model and a practical pouring trial.
In lost foam casting, a foam pattern is placed in unbonded sand, compacted by vibration, and poured under vacuum. The liquid metal contacts the foam, degrades it, and takes its place. The process is often called lost foam casting, evaporative pattern casting, or full-mold casting. It is attractive because it reduces machining allowance, improves dimensional accuracy, and lowers production cost. However, the same phenomena that make lost foam casting efficient also create defects. Foam decomposition generates gas, the gas creates back pressure, the back pressure changes filling behavior, and the filling behavior controls temperature distribution. Therefore, I based my work on a combined numerical and experimental framework.
I treated the liquid metal as an incompressible Newtonian fluid during filling. The continuity equation for lost foam casting is written as
$$
\frac{\partial u}{\partial x}+\frac{\partial v}{\partial y}+\frac{\partial w}{\partial z}=0
$$
where \(u\), \(v\), and \(w\) are velocity components in the coordinate directions. The momentum conservation equations are
$$
\frac{\partial u}{\partial t}+u\frac{\partial u}{\partial x}+v\frac{\partial u}{\partial y}+w\frac{\partial u}{\partial z}
=-\frac{1}{\rho}\frac{\partial P}{\partial x}+g_x+\gamma\left(\frac{\partial^2 u}{\partial x^2}+\frac{\partial^2 u}{\partial y^2}+\frac{\partial^2 u}{\partial z^2}\right)
$$
$$
\frac{\partial v}{\partial t}+u\frac{\partial v}{\partial x}+v\frac{\partial v}{\partial y}+w\frac{\partial v}{\partial z}
=-\frac{1}{\rho}\frac{\partial P}{\partial y}+g_y+\gamma\left(\frac{\partial^2 v}{\partial x^2}+\frac{\partial^2 v}{\partial y^2}+\frac{\partial^2 v}{\partial z^2}\right)
$$
$$
\frac{\partial w}{\partial t}+u\frac{\partial w}{\partial x}+v\frac{\partial w}{\partial y}+w\frac{\partial w}{\partial z}
=-\frac{1}{\rho}\frac{\partial P}{\partial z}+g_z+\gamma\left(\frac{\partial^2 w}{\partial x^2}+\frac{\partial^2 w}{\partial y^2}+\frac{\partial^2 w}{\partial z^2}\right)
$$
Here \(P\) is pressure, \(\rho\) is density, \(g_x\), \(g_y\), and \(g_z\) are gravity components, and \(\gamma\) is kinematic viscosity. The energy equation I used for filling and solidification is
$$
\rho C\frac{\partial T}{\partial t}+\rho C\left(u\frac{\partial T}{\partial x}+v\frac{\partial T}{\partial y}+w\frac{\partial T}{\partial z}\right)
=\lambda\nabla^2 T+\rho L\frac{\partial f}{\partial t}
$$
where \(C\) is specific heat, \(T\) is temperature, \(\lambda\) is thermal conductivity, \(L\) is latent heat, and \(f\) is solid fraction. For the gas gap between the liquid front and the foam, I used a pressure balance that accounts for foam gasification, gap thickness, and permeability:
$$
P_{i+1}=\frac{\alpha_p\Delta t V_p T_{i+1}(T_i-T_m)P_0}{L_P T_m\delta_{i+1}}
-\frac{\delta_i F\Delta t K T_{i+1}P_i(P_i+P_0)}{x_c S T_i\delta_{i+1}}
+\frac{\delta_i T_{i+1}(P_i+P_0)}{T_i\delta_{i+1}}-P_0
$$
The gap thickness increment is
$$
\Delta\delta=u\Delta t-\frac{\alpha_p\Delta t(T_i-T_m)}{L_P}-\rho_p
$$
For the free surface boundary, I applied a heat balance between the moving liquid front and the foam:
$$
k\frac{\partial T}{\partial n}=\rho_P L_s u_1
$$
During solidification, I considered heat conduction, convection, and radiation. Fourier’s law is
$$
\vec{q}=-k_T\nabla T
$$
Convective heat transfer follows Newton’s cooling law:
$$
q=\alpha(T_f-T_w)
$$
Radiative heat transfer follows the Stefan-Boltzmann law:
$$
E=\varepsilon\sigma_0 T^4
$$
I also introduced a solute field to describe alloy redistribution during solidification. In the liquid phase,
$$
\frac{\partial C_L}{\partial t}+U\cdot\nabla C_L=D_L\nabla^2 C_L+\bar{\varrho}
$$
In the solid phase,
$$
\frac{\partial C_S}{\partial t}=D_S\nabla^2 C_S+\bar{\varrho}
$$
The source term caused by phase change is
$$
\bar{\varrho}=M\frac{d\phi}{dt}(C_{eq}-C)
$$
At the solid-liquid interface, I used the Gibbs-Thomson condition, heat balance, solute partitioning, solute balance, and tangential velocity continuity:
$$
T_1=T_M\left(1-\frac{2K\gamma}{\Delta H}\right)+mC_L
$$
$$
\Delta H V_I=(u_S\Delta T_S-u_L\Delta T_L)\cdot\vec{n}
$$
$$
C_S=MC_L
$$
$$
\Delta C V_I=C_1V\cdot n+n\cdot(D_S\nabla C_S-D_L\nabla C_L)
$$
$$
U\cdot\tau=0
$$
These equations formed the theoretical basis for the lost foam casting simulations. I then used them to guide the gating design, parameter selection, and defect interpretation.
I designed two gating systems for the valve body: a top pouring system and a side pouring system. In the top pouring system, the liquid entered from above the casting. In the side pouring system, the liquid entered from the side through multiple ingates. I calculated the choke area from hydraulic approximations and then checked the result against practical lost foam casting rules. For the side pouring system, I used a pouring head of 200 mm, a choke area of 19.7 cm2, and three rectangular ingates of 40 mm by 20 mm. The runner was 40 mm by 22 mm, and the sprue was a cylinder 30 mm in diameter. For the top pouring system, I used a pouring head of 160 mm, a choke area of 25 cm2, and one ingate of 48 mm by 28 mm. The sprue diameter remained the same as in the side pouring design.
| Gating item | Top pouring system | Side pouring system |
|---|---|---|
| Pouring head | 160 mm | 200 mm |
| Choke area | 25 cm2 | 19.7 cm2 |
| Ingate count | 1 | 3 |
| Ingate section | 48 mm x 28 mm | 40 mm x 20 mm |
| Runner section | Not dominant | 40 mm x 22 mm |
| Sprue diameter | 30 mm | 30 mm |
| Riser type | Cylindrical | Cylindrical and waist-cylindrical |
| Main concern | Gas escape against flow | Stable side filling |
I used ProCAST modules for meshing, casting setup, and result viewing. The valve body was imported from a solid model. I repaired the geometry, assembled the parts, checked for intersections, and generated tetrahedral meshes. The casting and gating system were meshed more finely, while the sand mold and less critical regions were meshed more coarsely. For the side pouring design, the surface mesh had 41,558 elements and the volume mesh had 529,446 elements. For the top pouring design, the surface mesh had 39,432 elements and the volume mesh had 525,068 elements.
| Model | Surface mesh | Volume mesh |
|---|---|---|
| Side pouring lost foam casting model | 41,558 | 529,446 |
| Top pouring lost foam casting model | 39,432 | 525,068 |
The casting alloy was ZG230-450. I specified its chemical composition so that the software could generate temperature-dependent properties. The solidus and liquidus temperatures were 1467 C and 1511 C. The density, enthalpy, Young’s modulus, and solid fraction were generated as functions of temperature.
| C | Mn | Cr | Mo | Ni | Cu | V |
|---|---|---|---|---|---|---|
| 0.154 | 1.14 | 0.180 | 0.039 | 0.221 | 0.016 | 0.031 |
| Property | Value or behavior |
|---|---|
| Density | Temperature dependent |
| Enthalpy | Temperature dependent |
| Young’s modulus | Temperature dependent |
| Solid fraction | Temperature dependent |
| Solidus temperature | 1467 C |
| Liquidus temperature | 1511 C |
| Viscosity | Temperature dependent |
For the foam pattern, I used EPS. For the sand mold, I used permeable sand. The pattern and sand properties were specified as shown below. These properties are important in lost foam casting because foam degradation and sand permeability control gas evacuation and back pressure.
| EPS property | Value |
|---|---|
| Density | 25 kg/m3 |
| Specific heat | 3.7 kJ/kg/K |
| Latent heat | 100 kJ/kg |
| Thermal conductivity | 0.15 W/m/K |
| Solid temperature | 330 C |
| Liquid temperature | 350 C |
| Sand property | Value |
|---|---|
| Density | 1520 kg/m3 |
| Thermal conductivity | 0.53 W/m/K |
| Specific heat | 1.22 kJ/kg/K |
| Permeability | 1 x 10-7 cm2 |
I defined the initial temperatures and filling states for each component. The pouring cup was filled with liquid metal, while the foam and sand began at room temperature. The interface heat transfer was set according to whether the interface was between metal and foam, foam and sand, or metal and sand. I used equivalent interfaces for continuous fields and coincident interfaces for discontinuous fields.
| Region | Material | Initial fill | Initial temperature |
|---|---|---|---|
| Pouring cup | ZG230-450 | 0% | 1580 C |
| Foam pattern | EPS foam | 100% | 25 C |
| Sand mold | Permeable sand | 100% | 25 C |
| Interface | Heat transfer coefficient |
|---|---|
| Metal to metal | 1000-5000 W/m2/K |
| Metal to sand | 300-1000 W/m2/K |
| Solid to air | 5-10 W/m2/K |
| Solid to cold air | 100-1000 W/m2/K |
| Sand to sand | 200-300 W/m2/K |
For the lost foam casting mode, I set the metal-foam heat transfer coefficient, maximum heat exchange coefficient, burn zone width, and gas fraction. The burn zone width was 1 cm, and the fraction of foam converted to gas was 10%. These values allowed the simulation to represent foam degradation more realistically than a simple sand casting model.
I first simulated the top pouring system. The filling sequence was smooth, and no obvious turbulence appeared. However, the filling direction opposed the upward escape of foam gas. The gas products tended to remain near the liquid front, which increased the risk of gas entrapment and slag inclusion. The total filling time was 17.98 s. After solidification, shrinkage porosity and shrinkage cavities were scattered in the casting, especially near the sprue, riser, and regions adjacent to the riser. The calculated shrinkage volume was 6.276 cc, and the shrinkage rate reached 23.146%. The temperature field was non-uniform, and isolated liquid regions formed. Therefore, the top pouring system did not satisfy the quality requirement.
| Top pouring result | Observation |
|---|---|
| Filling behavior | Stable but gas escape is unfavorable |
| Filling time | 17.98 s |
| Temperature field | Non-uniform; strong gradients near riser |
| Solidification | Isolated liquid zones possible |
| Shrinkage volume | 6.276 cc |
| Shrinkage rate | 23.146% |
| Decision | Not acceptable for high-pressure valve body |
I then simulated the side pouring system. The liquid entered from the side, degraded the foam progressively, and filled the mold in a layered manner. The total filling time was 28.79 s. The filling rate first increased and then decreased because foam gasification and gas escape changed the back pressure. During solidification, the riser and the region near the ingates remained hotter than the casting. This supported directional solidification. The riser provided feed metal to the casting. Shrinkage porosity and cavities were confined to the riser rather than the valve body. The side pouring lost foam casting system therefore produced a much sounder casting.
| Side pouring result | Observation |
|---|---|
| Filling behavior | Stable, layered, and progressive |
| Filling time | 28.79 s |
| Temperature field | Clear boundaries; directional solidification |
| Solidification | No significant isolated liquid zones |
| Shrinkage location | Mainly in riser |
| Casting quality | No internal shrinkage defects observed |
| Decision | Selected for further optimization |
After comparing the two lost foam casting designs, I selected the side pouring system. I then optimized the process parameters. I focused on pouring temperature, negative pressure, and pouring speed because these three variables strongly affect filling, foam degradation, gas evacuation, temperature gradients, and hot tearing. I used a Box-Behnken design because it can fit a second-order response surface with a relatively small number of runs. The response was the hot tearing index calculated by the casting simulation. A smaller hot tearing index means a lower tendency for hot cracks.
| Factor | Symbol | Low level | Center level | High level |
|---|---|---|---|---|
| Pouring temperature | A | 1560 C | 1580 C | 1600 C |
| Negative pressure | B | 0.04 MPa | 0.05 MPa | 0.06 MPa |
| Pouring speed | C | 74 mm/s | 89 mm/s | 104 mm/s |
I ran seventeen simulations according to the Box-Behnken matrix. The response values are listed below. The lowest hot tearing index appeared near the center of the design space, but the interactions showed that the best region was not simply the center point.
| Run | A: temperature | B: negative pressure | C: pouring speed | Hot tearing index |
|---|---|---|---|---|
| 1 | 1580 | 0.06 | 104 | 0.01423 |
| 2 | 1580 | 0.05 | 89 | 0.01174 |
| 3 | 1560 | 0.06 | 89 | 0.01326 |
| 4 | 1560 | 0.05 | 74 | 0.01798 |
| 5 | 1580 | 0.05 | 89 | 0.01124 |
| 6 | 1580 | 0.04 | 74 | 0.01780 |
| 7 | 1580 | 0.05 | 89 | 0.01168 |
| 8 | 1580 | 0.04 | 104 | 0.01701 |
| 9 | 1600 | 0.05 | 74 | 0.01446 |
| 10 | 1600 | 0.04 | 89 | 0.01570 |
| 11 | 1600 | 0.05 | 104 | 0.01671 |
| 12 | 1580 | 0.05 | 89 | 0.01179 |
| 13 | 1560 | 0.05 | 104 | 0.01293 |
| 14 | 1560 | 0.04 | 89 | 0.01633 |
| 15 | 1580 | 0.05 | 89 | 0.01124 |
| 16 | 1600 | 0.06 | 89 | 0.01381 |
| 17 | 1580 | 0.06 | 74 | 0.01408 |
I fitted the results with a second-order polynomial. The regression equation for the hot tearing index \(Y\) was
$$
Y=0.012-2.250\times10^{-5}A-1.432\times10^{-3}B-4.300\times10^{-4}C
+2.950\times10^{-4}AB+1.825\times10^{-3}AC+2.350\times10^{-4}BC
+1.488\times10^{-3}A^2+1.748\times10^{-3}B^2+2.494\times10^{-3}C^2
$$
The analysis of variance showed that the model was highly significant. Negative pressure had the strongest effect on the hot tearing index, followed by pouring speed and then pouring temperature. The order of influence was therefore negative pressure greater than pouring speed greater than pouring temperature. Among the two-factor interactions, the interaction between pouring temperature and pouring speed was especially important. The interaction between pouring temperature and negative pressure was also visible, but it was weaker. The interaction between negative pressure and pouring speed was present but less dominant.
| Source | Sum of squares | Degrees of freedom | Mean square | F value | P value |
|---|---|---|---|---|---|
| Model | 8.548E-005 | 9 | 9.498E-006 | 55.58 | less than 0.0001 |
| A | 4.050E-009 | 1 | 4.050E-009 | 0.024 | 0.8820 |
| B | 1.642E-005 | 1 | 1.642E-005 | 96.07 | less than 0.0001 |
| C | 1.479E-006 | 1 | 1.479E-006 | 8.66 | 0.0216 |
| AB | 3.418E-007 | 1 | 3.418E-007 | 2.04 | 0.1966 |
| AC | 1.332E-005 | 1 | 1.332E-005 | 77.96 | less than 0.0001 |
| BC | 2.209E-007 | 1 | 2.209E-007 | 1.29 | 0.2930 |
| A2 | 9.329E-006 | 1 | 9.329E-006 | 54.59 | 0.0002 |
| B2 | 1.287E-005 | 1 | 1.287E-005 | 75.33 | less than 0.0001 |
| C2 | 2.618E-005 | 1 | 2.618E-005 | 153.20 | less than 0.0001 |
| Residual | 1.196E-006 | 7 | 1.709E-007 | ||
| Lack of fit | 8.941E-007 | 3 | 2.980E-007 | 3.95 | 0.1095 |
| Pure error | 3.021E-007 | 4 | 7.552E-007 | ||
| Total | 8.668E-005 | 16 |
The fit statistics also supported the second-order model. The coefficient of determination was 0.9862, the adjusted coefficient of determination was 0.9685, and the predicted coefficient of determination was 0.8295. The difference between the adjusted and predicted values was 0.139, which is below 0.2. The adequate precision was 19.95. These values indicated that the model was reliable for prediction and optimization.
| Statistic | Value |
|---|---|
| R squared | 0.9862 |
| Adjusted R squared | 0.9685 |
| Predicted R squared | 0.8295 |
| Adequate precision | 19.95 |
The response surfaces showed how the lost foam casting parameters interacted. When pouring speed was held constant, the hot tearing index decreased as pouring temperature and negative pressure moved toward their lower and middle ranges. The contour lines were elliptical, which confirmed interaction between pouring temperature and negative pressure. When negative pressure was held constant, the hot tearing index changed strongly with pouring speed and pouring temperature. The interaction between pouring temperature and pouring speed was clear. The contour lines were dense in the lower-right region, showing that a suitable combination of higher pouring temperature and moderate-to-high pouring speed could reduce hot tearing. When pouring temperature was held constant, the negative pressure and pouring speed interaction was less pronounced than the other two interactions, but the slope of the negative pressure response was steeper than that of pouring speed.
| Interaction | Observation | Preferred region for low hot tearing |
|---|---|---|
| Temperature and negative pressure | Elliptical contours; negative pressure dominant | 1575-1590 C and 0.05-0.06 MPa |
| Temperature and pouring speed | Strong interaction; pouring speed dominant | 1570-1590 C and 85-104 mm/s |
| Negative pressure and pouring speed | Moderate interaction; pressure effect steeper | 0.04-0.05 MPa and 80-95 mm/s |
Using the optimization function of the response surface software, I obtained a recommended process window. After rounding the values to practical settings, I selected a pouring temperature of 1578 C, a negative pressure of 0.05 MPa, and a pouring speed of 91 mm/s. The predicted hot tearing index for this combination was about 0.01122. I then ran a confirmation simulation with these parameters. The simulated hot tearing index was 0.01134, which agreed closely with the predicted value. This confirmed that the optimized lost foam casting process was feasible.
| Parameter | Optimized value | Predicted hot tearing index | Simulated hot tearing index |
|---|---|---|---|
| Pouring temperature | 1578 C | 0.01122 | 0.01134 |
| Negative pressure | 0.05 MPa | ||
| Pouring speed | 91 mm/s |
For the optimized lost foam casting case, the filling time was 28.18 s. The filling behavior was stable and continuous. The foam degraded rapidly because the metal temperature was sufficient, but the gas was able to escape through the permeable sand under the applied negative pressure. No cold shut or misrun appeared. The filling rate followed the typical lost foam casting pattern: slow at first, faster in the middle, and slower again near the end because of gas back pressure and reduced superheat.
| Optimized filling stage | Time | Filled fraction | Behavior |
|---|---|---|---|
| Initial | 2.99 s | 10.5% | Foam degradation begins |
| Early middle | 6.03 s | 21.1% | Stable front movement |
| Middle | 11.99 s | 41.8% | Layered filling |
| Late middle | 18.00 s | 62.7% | Riser begins to fill |
| Final filling | 24.08 s | 83.8% | Gas back pressure slows front |
| Completion | 28.18 s | 98% | Simulation completion criterion |
The solidification sequence of the optimized lost foam casting was consistent with directional solidification. At 32.5% solid fraction, the casting surface and thin sections began to solidify. At 61.8% solid fraction, the riser remained hotter than the casting and continued to supply liquid metal. At 80.2% solid fraction, most of the valve body had solidified, and no isolated liquid region was found in the casting. At 100% solid fraction, the riser and gating system solidified last. The shrinkage porosity and shrinkage cavities were confined to the riser, not the valve body. This was the desired result for a high-pressure valve body produced by lost foam casting.
| Solid fraction | State of the casting | Riser behavior | Defect tendency |
|---|---|---|---|
| 32.5% | Surface and thin sections solidify | Still liquid | Low |
| 61.8% | Interior remains liquid | Feeds the casting | Low |
| 80.2% | Most of valve body solidified | Partially liquid | Low |
| 100% | Complete solidification | Solidified last | Shrinkage in riser |
I also examined the stress field and hot tearing tendency. In the optimized lost foam casting simulation, stress concentrated near the riser-casting junction and near the ingates. These are common hot spots because they remain hot longer and experience restrained contraction. However, the maximum stress remained below the strength limit of the alloy at the relevant temperature. The hot tearing prediction map was mostly purple or blue, indicating a low tendency for hot cracks. The uniformity of the thermal field reduced local stress concentration. This improvement was linked to better control of the lost foam casting parameters and to the heat-transfer behavior of the foam-sand-metal system.
To make the heat-transfer model more realistic, I explored a secondary development of the heat-transfer part of ProCAST using Microsoft Visual Studio. I compiled a user function and placed it in the project folder. The function allowed the interfacial heat-transfer coefficient to vary with position. The heat flux was expressed as
$$
Q=Flux+h(T-T_{\alpha})+\sigma\varepsilon(T^4-T_{\alpha}^4)
$$
where \(Flux\) is heat flux, \(h\) is the film coefficient, \(T_{\alpha}\) is ambient temperature, \(\varepsilon\) is emissivity, and \(\sigma\) is the Stefan-Boltzmann constant. I used a position-dependent function to assign different heat-transfer coefficients to different regions. The logic is summarized below.
| Region condition | Heat-transfer coefficient | Purpose |
|---|---|---|
| \(x \le -0.05\) and \(y \le -0.05\) | 4500 W/m2/K | Represent strong cooling at a selected boundary region |
| \(-0.05 \le x \le -0.01\) and \(-0.05 \le y \le -0.01\) | 500 W/m2/K | Control mold-casting temperature gradient |
| Otherwise | 0 W/m2/K | Reduce heat exchange in non-critical regions |
The user function declaration included project name, dimension, temperature, solid fraction, time, coordinates, and boundary condition number. After compilation, the function was linked with the ProCAST simulation. I then compared the temperature field, stress field, hot tearing index, and slag entrapment before and after the modification. The modified heat-transfer model produced a more uniform temperature distribution. The stress concentration near the riser-casting junction decreased. The hot tearing tendency became less severe. The slag particles moved toward the riser and were captured there rather than remaining in the valve body.
| Result type | Before heat-transfer modification | After heat-transfer modification |
|---|---|---|
| Temperature field | Less uniform | More uniform |
| Stress concentration | Higher near riser junction | Lower and more distributed |
| Hot tearing tendency | Moderate in local regions | Low, mostly purple or blue |
| Slag entrapment | Possible near gating junction | Transferred to riser |
| Overall lost foam casting quality | Acceptable but improvable | Improved and more reliable |
For experimental validation, I used a simplified casting model with the same alloy and similar process conditions. The simplified geometry reduced the cost and time of pattern production, coating, compaction, and pouring while keeping the essential lost foam casting behavior. The simplified casting had overall dimensions of 300 mm by 175 mm by 41 mm and a volume of 1,848,444 mm3. It contained internal passages and a plate-like body. I produced the foam pattern with a computer numerical control foam engraving machine and prepared the sand mold with a three-dimensional vibration table.
| Experimental item | Detail |
|---|---|
| Alloy | ZG230-450 cast steel |
| Pattern material | EPS |
| Pattern density | 16-19 kg/m3 |
| Pattern production | Steam pre-expansion and CNC foam machining |
| Mold medium | Unbonded sand |
| Compaction | Three-dimensional vibration |
| Pouring method | Slow-fast-steady pouring |
| Post-processing | Riser cutting, sand removal, shot blasting |
Before pouring, I simulated the simplified casting with the optimized parameters. The filling time was 26.98 s. The filling sequence remained stable and continuous. The foam degraded progressively, and the gas escaped through the sand under negative pressure. The solidification sequence again showed that the riser remained liquid after the valve body had mostly solidified. Shrinkage defects were confined to the riser. The stress field showed that the highest stresses were near the ingates and the riser junction, but the values were below the critical level. The hot tearing prediction map was again dominated by low-risk colors. Slag particles moved into the riser and did not remain in the casting.
| Simplified casting result | Observation |
|---|---|
| Filling time | 26.98 s |
| Filling behavior | Stable and continuous |
| Solidification | Directional; riser feeds the body |
| Shrinkage location | Riser only |
| Stress | Localized but below critical limit |
| Hot tearing | Low tendency |
| Slag | Captured in riser |
The experimental pouring was performed with a pouring temperature near 1580 C, a negative pressure near 0.05 MPa, and a pouring speed near 91 mm/s. I used a pouring schedule that began slowly, increased in speed, and then became steady. This schedule helped fill the foam pattern without excessive turbulence, allowed gas to escape, and avoided cold shuts. After cooling, I removed the casting from the sand, cut the gating system, cleaned the surface, and inspected the result. The casting was dimensionally complete. It showed no hot tearing. There were only minor surface imperfections, which could be repaired without affecting service performance. This validated the feasibility and reliability of the numerical model and the optimized lost foam casting process.
| Experimental parameter | Value |
|---|---|
| Pouring temperature | 1580 C |
| Negative pressure | 0.05 MPa |
| Pouring speed | 91 mm/s |
| Pouring pattern | Slow-fast-steady |
| Result | Sound casting with no hot tearing |
The main conclusions I draw from this first-person study are as follows. First, the gating system is decisive in lost foam casting of a high-pressure valve body. A top pouring system can fill smoothly, but it opposes gas escape and produces shrinkage defects near the riser and sprue. A side pouring system allows progressive foam degradation, stable filling, and directional solidification. It can confine shrinkage to the riser and produce a sound valve body.
Second, the process parameters of lost foam casting must be optimized together. Negative pressure has the strongest influence on the hot tearing index, followed by pouring speed and pouring temperature. The interaction between pouring temperature and pouring speed is especially important. The optimized combination is a pouring temperature of 1578 C, a negative pressure of 0.05 MPa, and a pouring speed of 91 mm/s. This combination gives the lowest hot tearing tendency and produces a casting with no internal shrinkage defects.
Third, secondary development of the heat-transfer module improves the realism of lost foam casting simulation. A position-dependent heat-transfer coefficient makes the temperature field more uniform, reduces stress concentration, decreases hot tearing tendency, and helps move slag into the riser. This is useful for practical process design because the foam-metal-sand interface is not uniform in real lost foam casting.
Fourth, the simplified experimental validation confirms the numerical predictions. The optimized lost foam casting process produces a complete casting with good dimensional accuracy and no hot cracks. The minor surface defects that remain can be repaired. Therefore, the combination of numerical simulation, response surface optimization, heat-transfer secondary development, and experimental validation is a practical route for improving lost foam casting quality and reducing production cost.
In future work, I would extend the model to include more detailed foam degradation kinetics, three-dimensional gas flow through the coating, sand permeability variation, and microstructure evolution. I would also apply the method to a wider range of valve bodies and other high-pressure castings. Even so, the present study demonstrates that lost foam casting can be made more reliable when the gating system and process parameters are designed with the help of numerical simulation and validated by experiment.