I carried out this investigation to reduce sand hole defects in DPR valve seat castings manufactured by investment casting. The DPR valve seat is a critical sealing component in a DPR series control valve. Its structural integrity, dimensional accuracy, and surface quality directly control sealing performance, regulation precision, and service reliability. Because the component has a complex geometry, tight dimensional tolerances, and strict surface requirements, investment casting is a natural manufacturing route. Investment casting gives high forming accuracy, low surface roughness, and near-net shaping capability for complex stainless steel valve seat parts. However, in actual production, DPR valve seat castings still reveal sand hole defects after machining, and the scrap rate remains high. I found that sand hole defects accounted for approximately 35 percent of total rejected parts, making them one of the main obstacles to quality improvement and productivity. Previous work indicates that sand holes in stainless steel investment castings are closely related to entrapment, limited flotation, and capture of non-metallic inclusions during solidification. Therefore, I systematically studied the formation mechanism of sand holes in this specific product and optimized the investment casting process to control them effectively.
Material Requirements and Product Context
I selected a low-carbon 304-series stainless steel, designated 1.4308, for the DPR valve seat. This material provides a balanced combination of corrosion resistance, weldability, and formability under complex working conditions. In investment casting, the alloy must remain fluid enough to fill thin sections, resist oxidation during pouring, and solidify without excessive shrinkage or inclusion capture. The valve seat geometry includes a thin bottom wall, an upper sealing face, and edge regions that are sensitive to local cooling and inclusion accumulation. Because the part is used in a regulating valve, even small sand holes can interrupt the sealing surface and create leakage paths. I therefore treated the sand hole issue as both a metallurgical cleanliness problem and a gating and solidification problem.
The chemical and physical requirements of the material can be summarized in a general form. For a low-carbon austenitic stainless steel, the corrosion resistance depends on chromium, nickel, and molybdenum equivalents, while the low carbon content reduces sensitization risk. I used the following common equivalent expressions to describe the alloy balance:
$$ \mathrm{Cr_{eq}} = \mathrm{Cr} + 1.37\mathrm{Mo} + 1.5\mathrm{Si} + 2\mathrm{Nb} + 3\mathrm{Ti} $$
$$ \mathrm{Ni_{eq}} = \mathrm{Ni} + 22\mathrm{C} + 0.31\mathrm{Mn} + \mathrm{Cu} + 14.2\mathrm{N} $$
These expressions helped me confirm that the selected grade remained in a stable austenitic region after investment casting and heat treatment. The important point for defect control was not the bulk chemistry but the local capture of exogenous inclusions. The sand holes were not caused by the base steel composition alone. Instead, they were caused by aluminosilicate non-metallic inclusions that entered or remained in the investment casting mold and were then trapped in the final solidifying regions.
Initial Investment Casting Process Design
I first reviewed the original investment casting process. The original method used a flat two-bar tree pattern assembly. Each tree could cast two DPR valve seat parts at the same time. Each wax pattern was connected to the central bar through four internal gates. The main process parameters are listed in Table 1. Under these conditions, the cast tree weight after pouring was 5.75 kg, and the process yield was only 20.87 percent. This low yield increased metal consumption and cost per acceptable part.
| Parameter | Value |
|---|---|
| Material | 1.4308 low-carbon 304 stainless steel |
| Parts per investment casting tree | 2 |
| Single casting weight | 0.6 kg |
| Weight of poured tree | 5.75 kg |
| Process yield | 20.87 percent |
| Gating style | Flat two-bar tree, four internal gates per part |
I calculated the process yield using the standard investment casting yield expression:
$$ \eta = \frac{n m_c}{m_t} \times 100\% $$
where eta is the process yield, n is the number of castings per tree, m_c is the single casting weight, and m_t is the total poured tree weight. Substituting the original values gives:
$$ \eta_{\mathrm{original}} = \frac{2 \times 0.6}{5.75} \times 100\% = 20.87\% $$
This result confirmed that the original investment casting tree contained a large proportion of non-product metal. The central bar, runner, and gates occupied most of the poured weight. Although this design provided sufficient feeding in some regions, it did not provide favorable conditions for inclusion flotation and removal. The sand hole problem was therefore linked to both the gating layout and the inclusion transport path in the investment casting mold.
Initial Production Performance and Defect Statistics
I analyzed the production record for the original investment casting process. A total of 500 DPR valve seat castings were produced. Among them, 65 parts were rejected, giving a total rejection rate of about 13 percent. A detailed inspection showed that 46 rejected parts contained sand hole defects, which represented 71 percent of all rejected parts. Sand holes usually became visible only after machining. They were concentrated on the upper surface and bottom edge regions of the valve seat. These locations are critical because the upper surface participates in sealing and the bottom edge experiences thin-wall solidification and high cooling rates.
| Production Item | Value |
|---|---|
| Total castings produced | 500 |
| Rejected castings | 65 |
| Overall rejection rate | 13.0 percent |
| Sand hole rejected castings | 46 |
| Sand hole share of total rejects | 71 percent |
| Sand hole defect rate based on total production | 9.2 percent |
| Major defect locations | Upper surface and bottom edge |
I used the following expressions to quantify the defect distribution:
$$ R_{\mathrm{total}} = \frac{N_r}{N_t} \times 100\% $$
$$ R_{\mathrm{sand}} = \frac{N_s}{N_t} \times 100\% $$
$$ S_{\mathrm{sand}} = \frac{N_s}{N_r} \times 100\% $$
Here, R_total is the total rejection rate, R_sand is the sand hole defect rate, S_sand is the sand hole share among all rejects, N_r is the number of rejected castings, N_s is the number of sand hole castings, and N_t is the total number produced. Substituting the original data gave:
$$ R_{\mathrm{total}} = \frac{65}{500} \times 100\% = 13.0\% $$
$$ R_{\mathrm{sand}} = \frac{46}{500} \times 100\% = 9.2\% $$
$$ S_{\mathrm{sand}} = \frac{46}{65} \times 100\% = 70.8\% \approx 71\% $$
These values showed that sand holes were not an isolated defect. They were the dominant failure mode in the original investment casting process. Because the defects appeared after machining, they also caused additional value loss. The casting had already consumed wax, ceramic shell, melting energy, labor, and machining time before the defect was detected. Reducing sand holes was therefore essential for improving both quality and production economics.
Defect Sampling and Experimental Approach
I designed the experimental work to identify the composition, morphology, and phase structure of the sand hole defects. To avoid conclusions based on a single abnormal sample, I selected four DPR valve seat finished parts with clear sand hole defects. The selected defects had typical sizes and distributions. I avoided defect-free regions and edge damage so that the samples represented the general production problem. I used wire cutting to prepare sample blocks containing complete sand hole regions. The detection face of each block was controlled to a square with a side length of about 10 mm. This size provided a flat, consistent surface while preserving the original defect morphology as much as possible.
After cutting, I ground, cleaned, and dried the samples. These steps removed surface oxide, oil, and cutting residue that could interfere with later analysis. I then used scanning electron microscopy, SEM, to examine the microscopic morphology and composition of the defect regions. I also used energy-dispersive spectroscopy, EDS, to measure local elemental concentrations. For phase identification, I selected one defect-free product and two sand hole products. I prepared cubic samples with a side length of 10 mm and analyzed them by X-ray diffraction, XRD. The combination of SEM, EDS, and XRD allowed me to distinguish exogenous inclusions from base metal and to infer the likely source of the sand holes.
SEM and EDS Analysis of Sand Hole Regions
I found that the defect centers contained complex non-metallic inclusions. In one region, the inclusion was mainly composed of oxygen at 46.07 percent, silicon at 11.94 percent, and aluminum at 5.45 percent. These elements correspond to aluminosilicate oxides such as SiO2 and Al2O3. I interpreted this phase as residual slag, refractory erosion product, or deoxidation product. Carbon at 28.19 percent was also detected, which may have come from carbonaceous material residue or contamination. Manganese at 7.56 percent may have formed MnO. Therefore, this inclusion was a foundry slag mixture composed of aluminosilicate oxide, carbonaceous impurity, manganese oxide, and a small amount of metallic elements.
In another region, the oxygen content reached 54.09 percent, with small amounts of aluminum at 0.30 percent and silicon at 0.95 percent. These values correspond to iron, aluminum, and silicon oxides such as Fe2O3, Al2O3, and SiO2. I inferred that this inclusion formed from secondary oxidation products or residual slag during pouring. Iron at 26.14 percent and nickel at 1.38 percent were also detected. These elements likely came from small particles of stainless steel base metal that were entrained into the slag. The inclusion was therefore a mixture of iron-based oxide, base metal particles, carbonaceous impurity, and trace silicon-aluminum or potassium compounds.
A third region showed oxygen, silicon, and aluminum at 46.07 percent, 11.94 percent, and 5.45 percent, respectively. This again corresponded to aluminosilicate oxides such as SiO2 and Al2O3. I attributed it to slag, refractory erosion, or deoxidation products. Carbon at 28.19 percent and manganese at 7.56 percent appeared as well, indicating a mixture of stainless steel oxidation slag, carbonaceous refractory impurity, and auxiliary material residue. A fourth region had oxygen at 61.93 percent. Together with iron, chromium, aluminum, and silicon, this indicated iron oxide such as Fe2O3, chromium oxide such as Cr2O3, and aluminosilicate oxides such as Al2O3 and SiO2. I inferred that this inclusion formed from oxidation of the molten steel and residue from charge or flux. Iron at 9.36 percent, chromium at 5.11 percent, and nickel at 0.97 percent came from entrained stainless steel base metal particles. This inclusion was therefore a mixed foundry inclusion made of Fe, Cr, Al, and Si oxides, base metal particles, and carbonaceous impurity.
| Defect Region | Major Elements Detected | Approximate Concentrations | Likely Phase or Source |
|---|---|---|---|
| Region 1 | O, Si, Al, C, Mn | O 46.07, Si 11.94, Al 5.45, C 28.19, Mn 7.56 percent | Aluminosilicate oxide, carbonaceous residue, MnO, casting slag |
| Region 2 | O, Al, Si, Fe, Ni | O 54.09, Al 0.30, Si 0.95, Fe 26.14, Ni 1.38 percent | Iron-aluminum-silicon oxide, secondary oxidation, entrained base metal |
| Region 3 | O, Si, Al, C, Mn | O 46.07, Si 11.94, Al 5.45, C 28.19, Mn 7.56 percent | Aluminosilicate oxide, carbonaceous refractory impurity, slag mixture |
| Region 4 | O, Fe, Cr, Al, Si, Ni | O 61.93, Fe 9.36, Cr 5.11, Al and Si present, Ni 0.97 percent | Fe-Cr-Al-Si oxide, base metal particles, carbonaceous impurity |
From these EDS results, I concluded that the sand holes were not caused by a single simple inclusion. They were caused by a family of exogenous and oxidation-derived inclusions. The common feature was a high oxygen content combined with aluminum and silicon. This pointed strongly toward aluminosilicate compounds. The presence of iron, chromium, and nickel confirmed that some base metal particles were also entrained. The carbon signal suggested that carbonaceous material from the investment casting process or auxiliary materials may have contributed. The combination of these phases created irregular defects that became visible after machining.
XRD Phase Identification
I used XRD to determine the crystalline phases in the sand holes. I compared a defect-free sample, a first defect sample, and a second defect sample. The core detected phases were Al2SiO5, which is an aluminosilicate phase related to andalusite, kyanite, and sillimanite polymorphs, and Fe, which is the metallic matrix. The defect-free sample was dominated by the Fe matrix phase. A strong diffraction peak appeared near 2 theta of about 45 degrees, indicating a high content of well-crystallized Fe matrix. Only weak Al2SiO5 diffraction peaks appeared, and the peak shape was slightly broad. This showed that the defect-free sample contained a low amount of aluminosilicate phase.
In the two defect samples, characteristic peaks near 2 theta of about 60 degrees increased significantly in intensity. The peak shape became sharper, and the Al2SiO5 phase became a major phase. This demonstrated that the inclusion mixed into the sand hole defect products was Al2SiO5. The XRD results therefore supported the EDS interpretation. The sand holes were associated with aluminosilicate inclusions rather than with a purely metallic inclusion or a gas porosity mechanism.
| Sample Type | Major Phase | Minor Phase | XRD Observation |
|---|---|---|---|
| Defect-free sample | Fe matrix | Al2SiO5, weak | Strong Fe peak near 2 theta 45 degrees; weak broad aluminosilicate peaks |
| Defect sample one | Al2SiO5 | Fe matrix | Stronger aluminosilicate peaks near 2 theta 60 degrees; sharper peak shape |
| Defect sample two | Al2SiO5 | Fe matrix | Aluminosilicate phase becomes dominant; Fe matrix still present |
I also considered the stoichiometry of the identified phase. Al2SiO5 can be written as a combined oxide:
$$ \mathrm{Al_2SiO_5} \rightarrow \mathrm{Al_2O_3} + \mathrm{SiO_2} $$
Using standard molar masses, I estimated the oxide mass fractions:
$$ M_{\mathrm{Al_2SiO_5}} = 2M_{\mathrm{Al}} + M_{\mathrm{Si}} + 5M_{\mathrm{O}} $$
$$ M_{\mathrm{Al_2SiO_5}} \approx 2(26.98) + 28.09 + 5(16.00) = 162.05 \ \mathrm{g/mol} $$
$$ w_{\mathrm{Al_2O_3}} = \frac{M_{\mathrm{Al_2O_3}}}{M_{\mathrm{Al_2SiO_5}}} \approx \frac{101.96}{162.05} \approx 62.9\% $$
$$ w_{\mathrm{SiO_2}} = \frac{M_{\mathrm{SiO_2}}}{M_{\mathrm{Al_2SiO_5}}} \approx \frac{60.08}{162.05} \approx 37.1\% $$
These values helped me explain why the EDS spectra showed high oxygen, aluminum, and silicon. The inclusion was consistent with an aluminosilicate refractory or slag phase. The presence of carbon and manganese indicated that other residual materials were mixed into the same defect. I concluded that the sand holes were mainly caused by aluminosilicate non-metallic inclusions, with contributions from oxidation products and entrained base metal particles.
Formation Mechanism of Sand Holes in the Original Investment Casting Process
After identifying the inclusion chemistry, I analyzed the original investment casting flow path. The steel was poured through the pouring cup and first entered the lowest region of the DPR valve seat. Because the bottom wall of the DPR valve seat is thin, the steel filled this region rapidly and cooled quickly. The rapid cooling reduced the time available for inclusions to float. Inclusions that were mixed into the steel therefore tended to stagnate and accumulate at the bottom boundary. This produced sand holes at the bottom edge location.
In addition, each DPR valve seat was connected through four gates. After the steel filled the entire valve seat cavity, it continued upward to the region above the gates. During this stage, inclusions that had floated to the upper surface needed to enter the gate channel and continue upward to be discharged. If they failed to enter the gate channel in time, they remained on the upper surface of the valve seat. This produced sand holes on the upper surface. The two defect locations were therefore controlled by two related phenomena: insufficient flotation time in the thin bottom region and incomplete inclusion discharge from the upper surface.
I summarized the mechanism as a sequence:
| Stage | Physical Event | Defect Consequence |
|---|---|---|
| Pouring and initial filling | Steel enters the bottom of the investment casting cavity | Inclusions are transported into the thin bottom region |
| Rapid bottom cooling | Thin wall solidifies quickly; viscosity increases | Inclusions cannot float before solidification |
| Upward filling | Steel fills the valve seat and moves toward the gates | Some inclusions rise to the upper surface |
| Gate discharge | Floating inclusions must enter the gate channel | Inclusions that miss the gate remain at the upper surface |
| Final solidification | Remaining liquid pockets solidify | Sand holes become fixed in the casting |
The defect mechanism can be expressed as a competition between flotation time and solidification time. If the available flotation time is less than the time required for an inclusion to reach a discharge path, the inclusion is captured. I used the following criterion:
$$ t_f \ge \frac{h}{v_s} $$
Here, t_f is the time available before the local region solidifies or before the gate freezes, h is the vertical distance the inclusion must travel to reach a discharge path, and v_s is the inclusion flotation velocity. In the original investment casting process, the effective flotation distance was short, but the thin bottom region also solidified quickly. The combination produced a high capture probability. On the upper surface, the inclusions had to compete with the gate entry. If the gate entry was not favorable, the inclusions remained at the surface. This explained the observed defect concentration.
Inclusion Flotation Model
I used Stokes law to estimate the flotation velocity of aluminosilicate inclusions in the stainless steel melt. For a spherical particle in laminar flow, the terminal velocity is:
$$ v_s = \frac{2 r_p^2 (\rho_l – \rho_p) g}{9 \mu_l} $$
where r_p is the inclusion radius, rho_l is the liquid metal density, rho_p is the inclusion density, g is gravitational acceleration, and mu_l is the dynamic viscosity of the liquid metal. I used typical values for low-carbon stainless steel and aluminosilicate inclusions. The liquid steel density was about 7.8 g/cm3, the inclusion density was about 3.16 g/cm3, and the dynamic viscosity was about 0.006 Pa s. I calculated the flotation velocity for several inclusion radii.
| Inclusion Radius, micrometers | Density Difference, kg/m3 | Calculated Flotation Velocity, mm/s | Time to Rise 20 mm, s | Time to Rise 133 mm, s |
|---|---|---|---|---|
| 10 | 4640 | 0.167 | 119.8 | 796.4 |
| 25 | 4640 | 1.045 | 19.1 | 127.3 |
| 50 | 4640 | 4.180 | 4.8 | 31.8 |
The calculation showed that small inclusions rise slowly. A 10 micrometer aluminosilicate particle has a very low flotation velocity. Even a 25 micrometer particle requires about 19 seconds to rise 20 mm. In a thin-wall investment casting region, such a time may not be available before solidification. Larger inclusions rise faster, but they are also more likely to be captured at surfaces and in narrow sections. Therefore, the original investment casting design was unfavorable for inclusion removal because the flotation path and solidification sequence did not provide enough time for the smaller inclusions to escape.
I also considered the effect of local cooling using the Chvorinov rule:
$$ t_s = B \left( \frac{V}{A} \right)^n $$
where t_s is solidification time, V is volume, A is surface area, B is a mold constant, and n is an exponent usually close to 2. In the thin bottom wall, the modulus V/A is small, so t_s is short. This short solidification time reduces the window for flotation. The original process therefore had a fundamental mismatch between inclusion flotation requirements and local solidification behavior. To reduce sand holes, I needed to change the investment casting tree and gating system so that the melt path provided more favorable inclusion transport and discharge.
Redesign of the Investment Casting Tree and Gating System
I redesigned the investment casting process using a round bar tree instead of the original flat two-bar tree. Each DPR valve seat still used four internal gates. The material and the number of parts per tree remained unchanged. I adjusted the tree structure, bar mass, and wax tree weight. The optimized process parameters are shown in Table 6. The poured tree weight decreased from 5.75 kg to 4.55 kg. The process yield increased from 20.87 percent to 26.37 percent.
| Parameter | Original Investment Casting Process | Optimized Investment Casting Process |
|---|---|---|
| Material | 1.4308 low-carbon 304 stainless steel | 1.4308 low-carbon 304 stainless steel |
| Parts per tree | 2 | 2 |
| Single casting weight | 0.6 kg | 0.6 kg |
| Poured tree weight | 5.75 kg | 4.55 kg |
| Process yield | 20.87 percent | 26.37 percent |
| Tree style | Flat two-bar tree | Round bar tree |
| Internal gates per part | 4 | 4 |
I recalculated the optimized process yield:
$$ \eta_{\mathrm{optimized}} = \frac{2 \times 0.6}{4.55} \times 100\% = 26.37\% $$
The yield improvement reduced unit raw material consumption and production cost. More importantly, the redesign changed the inclusion flotation and discharge conditions. In the optimized investment casting process, the steel was poured through the pouring cup and filled the DPR valve seat from the bottom toward the gate direction. Compared with the original process, the flotation distance from the valve seat bottom to the gate increased from 20 mm to 133 mm. I interpreted this as an increase in the effective vertical path over which inclusions could rise and be discharged. The longer path and modified flow pattern provided more opportunity for inclusions to move into the discharge channel before final solidification. This reduced the probability that inclusions would be captured at the upper surface and bottom edge.
I used a simplified inclusion removal index to compare the original and optimized investment casting layouts:
$$ I_r = \frac{h_{\mathrm{eff}}}{v_s t_s} $$
Here, I_r is an inclusion removal index, h_eff is the effective flotation path, v_s is the flotation velocity, and t_s is the local solidification time. A higher I_r indicates a greater tendency for inclusion removal before solidification. In the optimized investment casting design, h_eff increased substantially, and the gating arrangement promoted a more favorable temperature gradient. The round bar tree also reduced dead zones and improved metal flow stability. These changes were consistent with the observed reduction in sand holes.
Comparison of Original and Optimized Investment Casting Flow Behavior
I compared the inclusion flotation behavior in the original and optimized investment casting trees. In the original process, the flow entered the thin bottom region and rapidly cooled. The flotation path to the gate was short, and the gate entry was not well positioned to capture floated inclusions. In the optimized process, the flow path was reorganized. The steel filled the DPR valve seat from the bottom toward the gate direction, and the vertical distance to the gate increased. This gave inclusions a longer rising path and a better chance to enter the discharge channel. The round bar tree also changed the thermal field. The central bar provided a more uniform thermal mass, which helped maintain liquid fluidity in the gate region.
| Flow Feature | Original Investment Casting | Optimized Investment Casting | Effect on Sand Holes |
|---|---|---|---|
| Tree style | Flat two-bar tree | Round bar tree | More stable flow and thermal field |
| Effective flotation distance | 20 mm | 133 mm | Longer inclusion rise path |
| Bottom filling behavior | Rapid fill and fast cooling | Controlled fill along gate direction | Less inclusion stagnation at bottom edge |
| Upper surface discharge | Inclusions often missed gate | Improved gate entry and discharge | Fewer surface sand holes |
| Process yield | 20.87 percent | 26.37 percent | Lower cost and less remelt |
I also considered the Reynolds number and flow regime in the investment casting gating system. For a simplified circular channel, the Reynolds number is:
$$ Re = \frac{\rho v D}{\mu} $$
where rho is density, v is flow velocity, D is channel diameter, and mu is dynamic viscosity. In investment casting, the flow is often turbulent during filling. Turbulence can help transport inclusions but can also entrain gas and slag. The round bar design helped reduce abrupt expansion and dead zones, which reduced the probability of inclusion entrapment. I did not attempt to eliminate turbulence completely; instead, I aimed to create a flow path that allowed inclusions to reach the gate and riser before solidification.
Verification Production and Quality Improvement
I validated the optimized investment casting process by batch production. I strictly followed the optimized process parameters and tree assembly method. I produced 100 DPR valve seat castings. All products passed through the complete pouring, cooling, cleaning, and quality inspection sequence. The inspection results showed that only one of the 100 products had a sand hole defect. The sand hole defect rate was therefore 1 percent. This was a large reduction compared with the original investment casting process, which had a sand hole defect rate of 9.2 percent. The verification test confirmed that the optimized tree and gating system effectively improved inclusion flotation and reduced sand hole formation.
| Verification Item | Original Investment Casting | Optimized Investment Casting |
|---|---|---|
| Production quantity | 500 | 100 |
| Sand hole defect count | 46 | 1 |
| Sand hole defect rate | 9.2 percent | 1.0 percent |
| Total rejection rate | 13.0 percent | Reduced in the validation batch |
| Process yield | 20.87 percent | 26.37 percent |
| Main defect location | Upper surface and bottom edge | Greatly suppressed |
I calculated the relative reduction in sand hole defect rate as follows:
$$ \Delta R = \frac{R_{\mathrm{original}} – R_{\mathrm{optimized}}}{R_{\mathrm{original}}} \times 100\% $$
$$ \Delta R = \frac{9.2 – 1.0}{9.2} \times 100\% \approx 89.1\% $$
This calculation showed that the optimized investment casting process reduced the sand hole defect rate by approximately 89 percent relative to the original process. The process yield increased by:
$$ \Delta \eta = \eta_{\mathrm{optimized}} – \eta_{\mathrm{original}} $$
$$ \Delta \eta = 26.37\% – 20.87\% = 5.50\% $$
The combined effect was a significant improvement in quality stability and production economy. The optimized investment casting process not only reduced the number of rejected parts but also reduced the amount of metal that had to be melted and recycled. This is important because remelting increases energy consumption and can introduce additional inclusions if the melt is not carefully controlled.
Discussion of Inclusion Sources and Prevention
I concluded that the sand holes were primarily caused by Al2SiO5 inclusions. The sources of these inclusions can be grouped into several categories. First, refractory erosion can release aluminosilicate particles from the ceramic shell or ladle lining. Second, slag and deoxidation products can remain in the melt if the holding and pouring practices are not optimized. Third, secondary oxidation during pouring can form Fe, Cr, Al, and Si oxides. Fourth, carbonaceous residue from wax or auxiliary materials can contribute to the complex inclusion structure. Fifth, entrained base metal particles can be captured together with these inclusions. These sources are consistent with the EDS results, which showed high oxygen, aluminum, and silicon in all defect regions.
| Inclusion Source | Typical Phase | Evidence from Analysis | Preventive Direction |
|---|---|---|---|
| Ceramic shell erosion | Al2SiO5, Al2O3, SiO2 | High Al, Si, O in defect centers | Improve shell refractory quality and slurry control |
| Slag and deoxidation products | Al2O3, SiO2, MnO | Mn and O detected with Al and Si | Optimize melting and slag removal |
| Secondary oxidation | Fe2O3, Cr2O3 | Fe, Cr, O rich regions | Control pouring atmosphere and turbulence |
| Carbonaceous residue | C-rich impurity | Carbon up to 28.19 percent | Improve wax removal and shell firing |
| Entrained base metal | Fe, Ni, Cr particles | Fe, Ni, Cr detected in inclusions | Reduce flow impact and dead zones |
I also noted that the sand hole defects were not caused by poor steel cleanliness alone. The base steel composition was acceptable. The problem was the interaction between exogenous inclusions and the investment casting process. If the gating system cannot discharge inclusions, even a relatively clean melt can produce sand holes because inclusions form during pouring and shell interaction. Therefore, process optimization must address both inclusion generation and inclusion removal. The optimized round bar tree improved removal, but further work should also reduce inclusion generation.
Further Process Improvement Opportunities
I identified several additional measures that can further reduce non-metallic inclusions in DPR valve seat investment casting. Improved shell refractory materials can reduce ceramic erosion. Pouring filters can capture large inclusions before they enter the mold cavity. Better wax removal and shell firing can reduce carbonaceous residue. Melt treatment and holding practices can promote slag flotation before pouring. Numerical simulation can help visualize inclusion motion and flotation behavior, allowing the gating system to be refined further. I consider these measures complementary to the tree redesign. The current optimization solved the dominant flotation and discharge problem, but continuous improvement should target the inclusion source as well.
| Improvement Measure | Expected Benefit | Implementation Consideration |
|---|---|---|
| Improved shell refractory | Less Al2SiO5 and aluminosilicate erosion | Maintain shell strength and thermal stability |
| Pouring filter | Capture larger non-metallic inclusions | Avoid excessive flow restriction and cold shut |
| Optimized wax removal | Lower carbonaceous residue | Control heating rate and atmosphere |
| Melt holding and slag removal | Fewer slag and deoxidation products | Balance superheat and inclusion flotation time |
| Numerical simulation | Predict inclusion trajectories and capture zones | Validate with production trials |
I also recommend monitoring the following process variables in future investment casting production:
| Variable | Reason for Monitoring | Target Direction |
|---|---|---|
| Pouring temperature | Controls fluidity and solidification time | Maintain within validated window |
| Shell preheat temperature | Controls cooling rate and inclusion flotation time | Uniform and sufficient for thin sections |
| Melt superheat | Affects viscosity and inclusion removal | Avoid excessive superheat and oxidation |
| Gating velocity | Controls turbulence and inclusion entrainment | Reduce abrupt flow changes |
| Tree arrangement | Controls thermal gradient and discharge path | Use optimized round bar design |
| Shell quality | Controls refractory particle release | Reduce erosion and cracks |
Quantitative Summary of the Optimization
I summarized the quantitative results of the investment casting optimization in Table 11. The original process produced 500 parts with 46 sand hole defects, giving a sand hole defect rate of 9.2 percent. The optimized process produced 100 parts with 1 sand hole defect, giving a sand hole defect rate of 1 percent. The process yield increased from 20.87 percent to 26.37 percent. The poured tree weight decreased from 5.75 kg to 4.55 kg. The optimized tree used a round bar design while keeping two parts per tree and four gates per part.
| Metric | Original Investment Casting | Optimized Investment Casting | Change |
|---|---|---|---|
| Parts per tree | 2 | 2 | Unchanged |
| Single part weight | 0.6 kg | 0.6 kg | Unchanged |
| Poured tree weight | 5.75 kg | 4.55 kg | Reduced by 1.20 kg |
| Process yield | 20.87 percent | 26.37 percent | Increased by 5.50 percentage points |
| Sand hole defect rate | 9.2 percent | 1.0 percent | Reduced by 8.2 percentage points |
| Relative sand hole reduction | Reference | 89.1 percent lower | Substantial quality improvement |
| Effective flotation distance | 20 mm | 133 mm | Longer inclusion rise path |
The optimization also reduced the amount of metal that must be remelted per acceptable casting. I estimated the metal efficiency improvement using the yield values:
$$ E_m = \frac{\eta_{\mathrm{optimized}} – \eta_{\mathrm{original}}}{\eta_{\mathrm{original}}} \times 100\% $$
$$ E_m = \frac{26.37 – 20.87}{20.87} \times 100\% \approx 26.4\% $$
This means the optimized investment casting process improved the metal utilization efficiency by about 26.4 percent relative to the original process. The economic benefit is significant because stainless steel melting and remelting are energy-intensive. The reduction in sand holes also lowers machining waste and inspection cost. Because sand holes are often exposed only after machining, preventing them at the casting stage avoids the loss of already added value.
Practical Conclusions from the Investment Casting Study
I reached several practical conclusions. First, the sand holes in DPR valve seat investment castings were mainly caused by Al2SiO5 aluminosilicate inclusions. SEM and EDS showed high oxygen, silicon, and aluminum in the defect centers, with additional carbon, manganese, iron, chromium, and nickel from auxiliary phases and entrained base metal. XRD confirmed that Al2SiO5 was the dominant defect phase in the sand hole products. The inclusions were not simply metallic inclusions, and their formation was not controlled solely by the bulk steel cleanliness.
Second, the original investment casting process had unfavorable inclusion flotation and discharge conditions. The flat two-bar tree caused rapid filling and cooling in the thin bottom region, leaving little time for inclusions to float. The upper surface also allowed inclusions to remain if they did not enter the gate channel. The defect locations on the upper surface and bottom edge were consistent with this mechanism.
Third, redesigning the investment casting tree and gating system using a round bar improved the process. The effective flotation distance from the valve seat bottom to the gate increased from 20 mm to 133 mm. The process yield increased from 20.87 percent to 26.37 percent. The sand hole defect rate decreased from 9.2 percent to 1 percent in the verification batch. These results confirmed that the optimized investment casting process is effective and practical.
Fourth, further improvement should focus on reducing inclusion sources. Improved shell refractory materials, pouring filters, better wax removal, cleaner melt handling, and numerical simulation of inclusion motion can further reduce defects. I consider the current optimization a robust foundation for continued quality improvement in DPR valve seat investment casting.
Final Remarks
I found that the sand hole problem in DPR valve seat investment casting could be controlled by understanding the inclusion chemistry and then changing the process flow. The key was not simply to increase pouring temperature or add more metal. The key was to create a gating and tree design that gave inclusions a feasible path to float and discharge before solidification. The round bar investment casting tree achieved this goal. It improved process yield, reduced sand hole defects, and increased production stability. I believe the same logic can be applied to other complex stainless steel valve seat components produced by investment casting, especially where thin walls, sealing surfaces, and high surface quality requirements make inclusion capture a dominant failure mode.
The optimized investment casting route is therefore recommended for DPR valve seat production. It provides a better balance between casting quality, material utilization, and manufacturing cost. The results also show that systematic defect analysis using SEM, EDS, and XRD can directly guide process improvement. By linking inclusion composition to fluid flow and solidification behavior, I was able to convert a persistent sand hole problem into a controlled and largely resolved manufacturing issue.