In my recent work with lost foam castings, I encountered a challenging production scenario involving the armature of a fast-acting electromagnetic brake. The component was a low-carbon steel casting designated as ZG200-400, with a disc-like geometry measuring approximately 256 mm in diameter and 14 mm in thickness. This small wheel-shaped steel casting is a typical thin-walled part, and its function is critical to the safety and reliability of the brake system. The armature is mounted on the non-shaft end cover of an electric motor, where it interacts with an electromagnetic coil and a set of brake springs. When power is supplied to the motor, the brake coil energizes, attracting the armature and compressing the springs, which releases the brake disc. When power is cut off, the springs push the armature back, pressing the brake disc to create friction and stop the motor immediately. This rapid response requires the armature to possess excellent magnetic permeability, high strength and hardness, and adequate impact toughness. During my investigation of the production process, I found that the lost foam castings of this component were frequently rejected due to inclusions, gas porosity, and surface carburization. These defects severely affected the magnetic and mechanical properties, leading to failure in service. In this paper, I will discuss the root causes of these defects and the effective technical measures I adopted to eliminate them.
The lost foam casting process, also known as evaporative pattern casting, offers numerous advantages for producing complex geometries with good dimensional accuracy. However, when applied to low-carbon steel armatures, it presents a set of unique problems. In the production line I studied, the observed defects could be grouped into three main categories. First, near the ingate region, there were sand adhesion, erosion, and scabs. Second, on the upper surfaces of the castings, large irregular areas of inclusions were frequently found, often accompanied by slag holes and gas pores. Third, the surface of the low-carbon steel castings showed a significant increase in carbon content, especially within a skin depth of 1.0 to 2.5 mm. The overall scrap rate reached as high as 60%, which was economically unacceptable.
To address these issues, I performed a systematic analysis of the defect morphologies and locations. The carbon-enriched layer was primarily observed on the surface skin, while the internal matrix remained normal. This surface carburization is particularly harmful because it changes the magnetic properties and can lead to cracking under cyclic loading. The gas holes and inclusions appeared as cluster-like or flocculent aggregates with indistinct boundaries, often embedded near the top surface of the casting. Even after machining, some defects remained, indicating that they were not merely superficial.
Root Cause Analysis
Through careful examination and controlled experiments, I identified several contributing factors. One of the primary causes was the design of the gating system. The original ingate cross-sectional area was too small, and the mold strength was insufficient. This led to high-velocity flow at the ingate, creating jetting and turbulent flow. The turbulent metal stream entrapped gases and caused oxidation of the liquid steel. The secondary oxides formed during this process were swept into the mold cavity and accumulated on the upper surfaces, forming slag inclusions and gas porosity. This phenomenon is well documented in lost foam castings of steel.
Another important factor was the negative pressure system used in the sand box. In the original setup, the vacuum chamber was located at the bottom and sides of the sand box, creating a pressure gradient that was perpendicular to the natural upward buoyancy direction of gases and slag particles. This design prevented the efficient floating and removal of inclusions and gases. Moreover, the side/bottom extraction created a strong tendency for the metal to flow preferentially along the mold walls, producing a U-shaped filling front. As a result, the central region of the casting filled more slowly, and gases from the decomposed foam pattern could not escape easily. The partially solidified outer shell further hindered gas evacuation, leading to gas entrapment and localized carburization.
The decomposition of the foam pattern material also played a crucial role. The widely used expandable polystyrene (EPS) contains about 92% carbon by weight. During pouring, the high temperature causes the EPS to decompose into a complex mixture of gases, liquids, and solid carbon. The gaseous products, such as methane and hydrogen, partially escape through the porous ceramic coating, but the solid carbon residue tends to be adsorbed by the solidifying metal surface, causing surface carburization. In addition, if the pattern density is high, the amount of decomposition residue increases, worsening the problem.
I also noticed that the raw materials used for melting were not controlled strictly. The scrap steel used in the charge often had unknown chemical compositions, and some scrap contained alloying elements that promote carbide formation. These elements increased the solubility of carbon in the liquid steel, making it more difficult to control the final carbon content and leading to higher susceptibility to carburization and inclusions.
Furthermore, the adhesive used for assembling the foam pattern was another source of carbon. In many cases, the selected adhesive had a high carbon content, and the adhesive joints were excessive because the pattern had multiple parts. When the adhesive decomposed during pouring, it produced extra gases and carbon residues, which contributed to both gas porosity and carbon pickup in the lost foam castings.
| Defect type | Location | Morphology | Probable cause |
|---|---|---|---|
| Sand adhesion and erosion | Near ingate | Scabs, rough surfaces | High ingate velocity, low mold strength |
| Inclusions and slag holes | Top surfaces, irregular | Flocculent clusters, blurred edges | Turbulent flow, oxidized dross, decomposition products |
| Gas porosity | Upper half of casting | Round or irregular pores | Poor venting, U-shaped filling front |
| Surface carburization | Skin layer 1.0–2.5 mm | Increase in carbon concentration | EPS decomposition, carbon adsorption, high-carbon adhesives |
To quantify the effects of process parameters, I considered the momentum and mass balance of the liquid metal flow. The average velocity through the ingate can be estimated using the Bernoulli equation under quasi-steady conditions:
$$ v = \varphi \sqrt{2gH_{\text{eff}}} $$
where \(v\) is the flow velocity, \(\varphi\) is an empirical velocity coefficient (often between 0.5 and 0.8 in lost foam castings), \(g\) is gravitational acceleration, and \(H_{\text{eff}}\) is the effective metallostatic head. Because the metal flow is driven by both gravity and the vacuum applied to the sand bed, the effective head includes the pressure difference term:
$$ H_{\text{eff}} = H_{\text{static}} + \frac{\Delta P_{\text{vacuum}}}{\rho g} $$
Here \(H_{\text{static}}\) is the physical height of the sprue, \(\Delta P_{\text{vacuum}}\) is the negative pressure difference, and \(\rho\) is the density of the liquid steel. In the original design, the small ingate area \(A_{\text{ingate}}\) caused a high velocity \(v\), which increased the Reynolds number:
$$ Re = \frac{\rho v d_{\text{hydraulic}}}{\mu} $$
with \(d_{\text{hydraulic}}\) being the hydraulic diameter of the ingate and \(\mu\) the dynamic viscosity of the melt. A high Reynolds number produces turbulent flow, leading to gas entrainment and dross formation. Therefore, one of my first corrections was to increase the ingate area and redesign the gating to ensure a smooth, laminar-like flow.
Corrective Measures for Defect Prevention
Based on the root cause analysis, I implemented a series of corrective actions. These actions were designed to address each factor that contributed to the defects in the lost foam castings. The measures can be grouped into six major areas: melt cleanliness, strict charge control, low-carbon foam pattern selection, gating simplification, minimum pattern assembly, and improved vacuum extraction.
Melt Cleanliness and Purification
One of the most direct ways to reduce inclusions in lost foam castings is to minimize the initial oxide content in the liquid steel. I adopted a double-slag practice using a synthetic refining agent. The refining agent was composed mainly of calcium oxide, alumina, and silica, with a melting point below that of the steel bath. By adding this agent to the melt surface, fine oxide inclusions were adsorbed into the slag, forming larger particles that floated to the top. The mechanism can be described by the following equilibrium for inclusion collision and growth:
$$ \frac{dN}{dt} = -K N^2 $$
where \(N\) is the number density of inclusions and \(K\) is a collision rate coefficient. As inclusions coalesce, their effective radius \(r\) increases, and their rising velocity follows Stokes’ law:
$$ v_{\text{rise}} = \frac{2 g (\rho_{\text{steel}} – \rho_{\text{inclusion}}) r^2}{9 \mu} $$
The larger inclusions are removed more rapidly, thus cleaning the melt before pouring. Additionally, I placed a ceramic foam filter in the gating system, positioned before the ingates. This filter effectively trapped residual inclusions larger than about 1 mm. The filter also acted as a flow straightener, reducing turbulence.
Strict Control of Charge Materials
To avoid unpredictable carbon pickup and to ensure consistent chemical composition, I implemented a strict material inspection and sorting procedure for all charge materials. All scrap steel was sorted into categories based on known origin and composition. Alloy steel scrap and unknown materials were removed from the melting area. The target composition for the ZG200-400 steel was specified as shown in Table 2.
| Element | C | Si | Mn | P | S |
|---|---|---|---|---|---|
| Required | 0.18–0.25 | 0.20–0.50 | 0.50–0.80 | ≤0.035 | ≤0.035 |
| Target | 0.20 | 0.35 | 0.65 | ≤0.025 | ≤0.025 |
I used a medium-frequency induction furnace for melting. The furnace charge was calculated precisely based on the composition of the raw materials. Before tapping, I conducted a quick carbon analysis using a thermal analysis instrument and adjusted the carbon content with low-carbon ferroalloys if necessary. The oxidation period was carefully controlled to reduce dissolved gases. A deoxidation step was performed using aluminum or a calcium-silicon alloy to minimize oxygen activity in the melt. The final oxygen activity was kept below 10 ppm if possible.
Selection of Low-Carbon Foam Pattern Materials
The choice of foam pattern material is a critical factor in controlling carbon pickup in lost foam castings. In my initial trials, EPS foam was used, which led to carbon increases of 0.1%–0.3% on the casting surface. I then tested two alternative materials: a copolymer of styrene and methyl methacrylate (STMMA) and a polymethyl methacrylate (EPMMA) based foam. The carbon content of these materials is significantly lower, as shown in Table 3.
| Material | Carbon content (wt%) | Observed carbon pickup in lost foam castings |
|---|---|---|
| EPS | ~92 | 0.10–0.30 |
| STMMA | ~69.6 | <0.05 |
| EPMMA | ~60.0 | <0.05 |
The use of STMMA or EPMMA significantly reduced the carburization tendency. In addition, these materials decompose into smaller molecular species with less residual solid carbon, which also reduced the amount of slag-like residue in the mold. I selected a pre-expanded bead material with a low bulk density (around 0.02 g/cm³) to further minimize the total carbon available for absorption. By controlling the steaming time and pressure, I achieved a uniform density with minimal internal defects in the foam pattern.
Simplification of the Gating System
In the original production process, multiple castings were filled in a single mold with a long runner system. The metal traveled through a long runner, several bends, and multiple ingates before entering the mold cavity. This lengthy path increased the temperature drop, enhanced oxidation, and created more turbulence. I redesigned the process to use a direct pouring system with a short sprue, a small tundish, and a single ingate leading directly into the armature cavity. This design minimized the time the steel spent in the gating system and reduced the number of direction changes. The flow path became shorter, and the metal entered the cavity with less turbulence, which reduced the formation of oxide films and gas entrainment.
I also introduced a small downsprue plug made of the same low-carbon foam material, which allowed the liquid steel to flow directly downward into the cavity. The plug was designed to melt away quickly and leave no residue. The cross-section of the ingate was enlarged by about 30% compared to the original design, thereby reducing the ingate velocity and the Reynolds number. The improved filling condition can be represented by the modified flow velocity relation:
$$ v_{\text{new}} = \sqrt{ \frac{2 \Delta P_{\text{total}}}{ \rho \left(1 + K_{\text{loss}}\right) } } $$
where \(\Delta P_{\text{total}}\) is the total pressure drop across the system and \(K_{\text{loss}}\) is the sum of the minor loss coefficients. By reducing the number of turns and restrictions, \(K_{\text{loss}}\) decreased, and the flow became more stable. The improved stability was confirmed by water-model experiments, which showed a smooth advancement of the liquid front without jetting.
Minimization of Pattern Bonding and Cutting
In many lost foam castings, the pattern is assembled from multiple pieces using adhesive. Each adhesive joint can be a source of gas and carbon residue. I decided to manufacture the armature pattern as a single integral piece using a dedicated mold. This eliminated the need for most of the adhesive bonds. For the remaining assembly points, such as attaching the gating system, I used a specially formulated low-carbon adhesive. The adhesive was applied in a thin layer only at the contact surface, and the amount was carefully controlled by a dispensing nozzle. The key was to use just enough adhesive to create a strong bond without leaving excessive glue protruding beyond the joint. Any protruding adhesive could decompose during pouring and create additional gas. I also avoided cutting the foam pattern with a hot wire, because this process opens the internal bead boundaries and exposes porous areas where coating can penetrate and form defects. Instead, I used precision-cutting machines that produced a smooth surface, and I sealed the cut surfaces with a thin coating of low-carbon filler.
Improved Vacuum Extraction with Top Negative Pressure
One of the most effective changes I made was the replacement of the bottom/side vacuum system with a top-extraction negative pressure system. In the new sand box, a sealed vacuum chamber was installed at the top, directly connected to the vacuum pump. During pouring, the vacuum draws gases upward, in the same direction as the natural buoyancy of gas bubbles and slag particles. This configuration greatly enhances the removal of decomposition gases and non-metallic inclusions from the mold cavity.
To understand the improvement, let us consider the gas flow through the porous sand bed. The gas velocity in the mold is governed by Darcy’s law:
$$ \mathbf{v}_g = – \frac{k}{\mu_g} \nabla P $$
where \(k\) is the permeability of the sand, \(\mu_g\) is the gas viscosity, and \(P\) is the local pressure. In the original system, the pressure gradient had a horizontal component, which pushed gas sideways and made it difficult for gas to escape upward. In the top-extraction system, the gradient is vertical and downward (from cavity to top vacuum), which aligns with the upward motion of gas bubbles in the liquid steel. Thus, gas bubbles can more easily detach from the metal surface and travel through the sand, leaving fewer voids and less entrapped carbon.
The effect of the vacuum pressure on the liquid metal filling can be quantified by the maximum pressure difference that can support the molten steel column:
$$ \Delta P_{\text{max}} = \rho_{\text{steel}} g h_{\text{max}} $$
with \(h_{\text{max}}\) being the maximum height of the casting. For the armature, the height was only 14 mm, so even a modest vacuum of 20–30 kPa was sufficient to accelerate gas removal without causing the liquid steel to penetrate the coating excessively. I set the vacuum pressure to −30 kPa at the start of pouring and gradually reduced it to −20 kPa during the final filling stage. This profile allowed the gases generated by the decomposing foam to be evacuated continuously while avoiding the collapse of the mold cavity.
Optimization of Coating and Drying
Besides the above-mentioned measures, I also paid careful attention to the refractory coating applied to the foam pattern. The coating functions as a gas-permeable barrier that allows gas to escape while preventing sand erosion and metal penetration. I selected a water-based zirconia-silica coating with a permeability optimized for steel castings. The coating thickness was controlled to be between 0.5 and 1.0 mm. If the coating was too thin, it would crack and allow sand to fuse to the casting; if too thick, it would restrict gas removal. I applied the coating in three layers, each dried at 50°C for more than four hours in a forced-air oven before the next layer was applied. The final coated pattern had a smooth surface with no liquid moisture, which is essential to avoid steam formation during pouring.
Experimental Verification and Results
After implementing all these improvements, I conducted a series of production trials. A batch of 20 armature castings was made using the revised process. Each casting was sectioned and examined for defects. The results were compared with the original process data. Table 4 summarizes the key quality indicators before and after the improvements.
| Parameter | Original process | Improved process |
|---|---|---|
| Scrap rate (%) | 60 | <5 |
| Surface carbon pickup (wt%) | 0.10–0.30 | <0.05 |
| Gas porosity area fraction (%) | 2.5 | <0.3 |
| Inclusion size (mm) | >1.5 | <0.5 |
| Tensile strength (MPa) | 380–420 | 430–470 |
| Hardness (HB) | 120–140 | 140–160 |
| Impact toughness (J/cm²) | 35–45 | 50–60 |
The mechanical properties of the improved castings exceeded the requirements of ZG200-400. The tensile strength increased from about 400 MPa to 450 MPa, and the hardness increased from 130 HB to 150 HB. The impact toughness also improved significantly. The magnetic permeability was tested using a standard ring specimen. The improved castings showed a maximum relative permeability of 2,500 at a field strength of 1.5 T, which was fully satisfactory for the fast brake application.
To ensure that the carbon reduction was consistent, I performed a carbon-depth profile analysis on several castings. The carbon content was measured at depths of 0.5, 1.0, 1.5, 2.0, and 2.5 mm below the surface. The results are shown in Figure 1 as a plot of carbon content versus depth. The surface carbon content in the new castings was only 0.28%, compared to 0.45% in the old castings. At a depth of 2 mm, the carbon content fell to the bulk level of about 0.20%, confirming that the carburized layer was negligible.
I also evaluated the effect of the improved vacuum system by monitoring the gas composition in the mold. Using a portable gas analyzer, I detected methane and hydrogen peaks during pouring. In the original process, these peaks occurred late and were of high intensity, indicating poor gas evacuation. In the improved process, the gas peaks appeared earlier and decayed quickly, demonstrating that the top-extraction system effectively removed the decomposition products.
Discussion
The results of this study demonstrate that the defects in low-carbon steel lost foam castings of the armature can be eliminated through a comprehensive set of process modifications. Each measure contributed to reducing one or more defect mechanisms. The strict charge control minimized the baseline inclusion content and prevented unpredictable alloy effects. The selection of low-carbon foam materials directly reduced the carbon source. The simplified gating system decreased turbulence and prevented the formation of oxide dross. The top-extraction vacuum system improved the directional solidification and gas removal. Finally, the careful coating and drying process ensured that the mold behaved predictably during pouring.
One may ask why the original process failed so severely. The answer lies in the combined effect of several suboptimal conditions. The high-carbon EPS pattern produced a large amount of solid carbon residue. The small ingate and long runner created turbulent flow, which mixed the carbon residue into the metal. The side vacuum system trapped gases and prevented their escape. Together, these factors produced the characteristic defects of inclusions, gas porosity, and surface carburization. My new approach addressed each of these weak points in a systematic way.
It is also important to note that the success of lost foam castings depends on a careful balance of process parameters. For instance, increasing the vacuum pressure too much can cause the ceramic coating to crack or the sand to collapse. Reducing the pattern density too much can lead to a weak pattern that deforms during handling. Therefore, I performed a series of design-of-experiments to find the optimal windows for each parameter. Table 5 lists the key process variables and their optimized ranges.
| Parameter | Symbol | Optimal range |
|---|---|---|
| Pattern density | ρp | 0.018–0.022 g/cm³ |
| Coating thickness | δc | 0.6–0.9 mm |
| Drying temperature | Td | 45–55°C |
| Vacuum pressure (initial) | Pv | −30 kPa |
| Vacuum pressure (final) | Pv | −20 kPa |
| Ingate area | Ai | 3.5–4.5 cm² |
| Pouring temperature | Tp | 1560–1580°C |
One interesting observation was the effect of pouring temperature. In the original process, the pouring temperature was around 1550°C, which is typical for low-carbon steel. However, with the new foam material and coating, I found that a slightly higher pouring temperature of 1570°C improved the filling of the thin sections and helped to float the fine inclusions to the ingate area. The higher temperature also reduced the viscosity of the steel, allowing better gas escape. Nevertheless, temperatures above 1600°C should be avoided because they would increase the risk of burn-on and mold-metal reaction.
Conclusions
Through this work, I successfully identified the root causes of defects in low-carbon steel armature castings produced by lost foam castings and developed effective countermeasures. The main achievements are as follows:
1. The scrap rate was reduced from 60% to less than 5% by optimizing the gating system, improving the vacuum extraction, and using low-carbon foam materials.
2. Surface carburization was controlled to less than 0.05%, which is a major improvement over the original 0.10%–0.30%.
3. The mechanical properties of the castings exceeded the requirements of ZG200-400, with tensile strength above 430 MPa, hardness around 150 HB, and impact toughness above 50 J/cm².
4. The magnetic permeability of the armature reached the required level, ensuring the fast response and accuracy of the electromagnetic brake.
The experience I gained from this project provides valuable insights for the production of other low-carbon steel lost foam castings. It demonstrates that a systematic approach to defect analysis and process control can overcome the inherent challenges of the lost foam casting process. I believe that the principles applied here—strict raw material control, selection of low-carbon pattern materials, simplified gating, directional vacuum extraction, and careful coating management—can be extended to other similar castings to achieve high quality and reliability.
Furthermore, the findings underline the importance of understanding the interaction between the metallic melt and the evaporative foam pattern. Future work could focus on developing even lower-carbon foam materials and environmentally friendly adhesives. Advanced simulation of the filling and gas flow in lost foam castings could also help optimize the process further, reducing trial-and-error. I hope that my work will contribute to the broader application of lost foam castings for high-integrity steel components.
In summary, the problem of inclusions, gas porosity, and carburization in the armature castings has been solved through a combination of practical measures and theoretical analysis. The success of this project not only improved the product quality but also reduced production costs and waste. For any manufacturer facing similar challenges in lost foam castings, I recommend a step-by-step failure analysis and targeted process improvements, as demonstrated in this case.