Lost foam castings are widely recognized for their near-net-shape capability, excellent dimensional accuracy, flexible design, and clean production environment. I have personally found, however, that lost foam castings also demand a deeper understanding of gas evolution and metal-front behavior than conventional sand casting. Among the most troublesome defects in steel lost foam castings are irruptive gas holes. These defects are not always caused by hydrogen or nitrogen pick-up; they are often generated by the decomposition products of the foam pattern itself. In this article, I discuss the formation mechanism of such gas holes, using a steel sliding rotor ring body as a production example, and I explain the preventive measures that I have successfully applied in practice.
1. Production Background and Defect Observations
The component under investigation is a steel sliding rotor ring body produced for water-turbine generator equipment. Its material grade is ZG270-560. The outer diameter is 550 mm, the height is 670 mm, the main wall thickness is 90 mm, and the mass is approximately 690 kg. I used a polystyrene foam pattern with a density of about 0.024 g/cm3. The pattern was cut with a hot wire by hand. A self-prepared bauxite water-based coating was brushed on in three layers. The average coating thickness was 1.5 to 3.0 mm, and the coated pattern was dried at 40 to 50°C for 36 hours.
In the original process, the pattern was placed horizontally. The sprue pattern size was 50 mm × 50 mm. One runner was 50 mm × 50 mm, and two ingates were each 40 mm × 45 mm. The designed gating area ratio was \(S_{\mathrm{sprue}} : \Sigma S_{\mathrm{runner}} : \Sigma S_{\mathrm{gate}} = 1:1:1.3\). Each flask contained one casting. The gating system was a bottom gate with reverse tangential entry. Three top blind risers were used to supply shrinkage. The steel was melted in a medium-frequency induction furnace. During pouring, the vacuum was maintained at 0.055 MPa, the pouring temperature was 1520 to 1570°C, and the average pouring time was 19 to 26 s. After pouring, vacuum was stopped after 15 min, and shakeout was carried out after 4 h.
After machining, many dispersed gas holes appeared on the upper part of the outer cylindrical surface and at the top of the sliding rotor ring. The largest gas holes had a diameter of about 6 mm and a depth of 12 mm. Based on their shape, size, and distribution, I classified them as irruptive gas holes rather than shrinkage porosity or metallurgical gas porosity. In my view, these gas holes were caused by gas generated during thermal decomposition of the foam pattern, which penetrated into the liquid metal and could not escape before solidification.
2. Conditions for Gas Intrusion in Ordinary Sand Castings
The classical criterion for gas intrusion in ordinary sand castings is well known:
\[
P_{I} > P_{C} + P_{S} + P_{R} \tag{1}
\]
where:
\(P_{I}\) is the gas pressure at the mold/metal interface,
\(P_{C}\) is the gas pressure acting on the free surface of the liquid metal in the cavity,
\(P_{S}\) is the metallostatic pressure of the liquid metal, and
\(P_{R}\) is the resistance to gas penetrating into the liquid metal.
In ordinary green sand or water-glass sand molds, gas is generated by moisture evaporation and by combustion of organic materials in the mold. If the mold cannot vent this gas quickly enough, \(P_{I}\) increases. When inequality (1) is satisfied, gas can enter the liquid metal. If the gas bubbles cannot float out before solidification, they remain as irruptive gas holes.
However, I have found that this equation alone is not sufficient to explain the behavior of lost foam castings. In lost foam castings, the foam pattern is not removed before pouring. Instead, it is vaporized by the advancing liquid metal. The gas pressure at the metal/foam interface and the gas pressure at the metal/coating interface can vary independently. Therefore, two different types of pressure conditions must be considered.
3. Two Gas-Intrusion Mechanisms in Lost Foam Castings
In lost foam castings, the liquid metal advances against a solid foam pattern. The heat of the liquid metal causes the foam to decompose into gas, liquid residue, and solid carbonaceous residue. The decomposition gas forms a gas film in the gap between the metal front and the unvaporized foam. I call this gas-film pressure \(P_{F}\).
At the same time, some liquid residue may be pushed against the coating layer. This residue is retained on the coating surface because the coating has a certain roughness and adhesion. Under the high steel temperature, the residue continues to gasify. This creates a gas pressure \(P_{I}\) at the interface between the liquid metal and the coating.
The two important pressure conditions for lost foam castings are:
Case 1 — Gas-film-driven intrusion:
\[
P_{F} > P_{I} + P_{S} + P_{R} \tag{2}
\]
In this case, the gas film between the metal and the foam is the driving pressure. If the metal front advances faster than the foam can gasify, the gas film is compressed and its pressure rises until it forces gas into the liquid metal.
Case 2 — Interface-gas-driven intrusion:
\[
P_{I} > P_{C} + P_{S} + P_{R} \tag{3}
\]
In this case, the liquid residue accumulated on the coating gasifies and raises \(P_{I}\). The gas at the coating interface then penetrates through the liquid-metal/coating interface into the steel. This mechanism is particularly important on vertical side surfaces where residue tends to accumulate.
Table 1 summarizes the variables used in these equations.
| Symbol | Meaning | Role in lost foam castings |
|---|---|---|
| \(P_{F}\) | Gas pressure in the gas film between liquid metal and foam | Driving pressure in Eq. (2) |
| \(P_{I}\) | Gas pressure at the liquid-metal/coating interface | Back-pressure in Eq. (2); driving pressure in Eq. (3) |
| \(P_{C}\) | Pressure above the liquid free surface in the mold cavity | Opposing pressure in Eq. (1) and Eq. (3) |
| \(P_{S}\) | Metallostatic pressure of the liquid steel | Opposing pressure in all equations |
| \(P_{R}\) | Resistance of liquid metal to gas entry | Increases with surface tension and viscosity |
The metallostatic pressure can be written as
\[
P_{S} = \rho_{l} g h
\]
where \(\rho_{l}\) is the density of the liquid steel, \(g\) is gravitational acceleration, and \(h\) is the vertical depth below the free surface. At the top surface \(h = 0\), so \(P_{S}\) becomes zero. This is why gas intrusion is easier near the top of a casting.
The resistance term \(P_{R}\) is related to the surface tension of the liquid steel and the radius of a gas bubble embryo:
\[
P_{R} = \frac{2\sigma}{r}
\]
where \(\sigma\) is the surface tension and \(r\) is the radius of the gas bubble embryo. A larger surface tension and a smaller bubble radius both make gas entry more difficult.
4. Effect of Filling Speed and Foam Pyrolysis
In a lost foam casting, the gas generation rate depends on the filling speed and the pattern density. I can express the gas generation rate \(\dot{m}_{g}\) at the decomposition front as
\[
\dot{m}_{g} = \rho_{\mathrm{EPS}} A_{F} v_{F} \alpha_{g}
\]
where:
\(\rho_{\mathrm{EPS}}\) is the density of the foam pattern,
\(A_{F}\) is the area of the decomposition front,
\(v_{F}\) is the linear filling velocity, and
\(\alpha_{g}\) is the fraction of decomposition products that become gas.
At the same time, the liquid residue generation rate \(\dot{m}_{r}\) is
\[
\dot{m}_{r} = \rho_{\mathrm{EPS}} A_{F} v_{F} (1 – \alpha_{g})
\]
If the pouring time is too short, \(v_{F}\) is high. The foam does not have enough time to gasify completely. Thus \(\alpha_{g}\) decreases and \(\dot{m}_{r}\) increases. More liquid residue is carried to the coating interface. This directly increases the risk of Case 2 gas intrusion.
The gas-film pressure \(P_{F}\) can be evaluated by considering the balance between gas generation and gas escape. A simplified conservation equation for the gas film is
\[
\frac{dP_{F}}{dt} = \frac{RT}{V_{F}} \left( \dot{m}_{g} – \dot{m}_{\mathrm{out}} \right)
\]
where \(V_{F}\) is the gas film volume, \(T\) is the gas temperature, and \(\dot{m}_{\mathrm{out}}\) is the mass rate of gas escaping through the coating. If \(\dot{m}_{g}\) exceeds \(\dot{m}_{\mathrm{out}}\), \(P_{F}\) rises rapidly. Once Eq. (2) is satisfied, gas enters the liquid steel.
The escape of gas through the coating follows Darcy’s law for flow through a porous layer:
\[
\dot{m}_{\mathrm{out}} = \frac{k_{t} A_{c} \rho_{g}}{\mu_{g} d_{c}} \left( P_{F} – P_{I} \right)
\]
where:
\(k_{t}\) is the coating permeability,
\(A_{c}\) is the area through which gas escapes,
\(\rho_{g}\) is the gas density,
\(\mu_{g}\) is the gas viscosity, and
\(d_{c}\) is the coating thickness.
From this equation, I can see that an increase in coating permeability \(k_{t}\) or a decrease in coating thickness \(d_{c}\) will increase the gas escape rate. This lowers \(P_{F}\) and reduces the chance of gas-film-driven intrusion.
5. Analysis of the Defects in the Steel Sliding Rotor Ring
In the original process, the average pouring time was 19 to 26 s. The same component produced in a water-glass sand mold had a pouring time of 33 to 38 s. Although lost foam castings generally require a somewhat faster pour than sand mold castings, a pouring time of 19 to 26 s was too short for this heavy steel casting. The linear filling velocity \(v_{F}\) was therefore too high.
This high filling velocity produced two effects. First, the foam could not gasify fast enough, so the gas film pressure \(P_{F}\) increased. Second, the incomplete gasification produced a large amount of liquid residue. This residue accumulated at the coating interface, especially on the upper side surfaces, and continued to gasify there, increasing \(P_{I}\).
The gating system was a bottom gate. At the beginning of filling, the steel entered the cavity through the lower ingates and rose upward. In the lower part of the casting, the metallostatic pressure \(P_{S}\) at the metal front was relatively large because there was already a considerable height of liquid steel above the point of gas entry? Actually, in bottom filling, the metal front is the free surface; \(P_{S}\) at the front is zero. However, along side walls below the front, gas entering from the coating interface must overcome the full height of metal above it. Therefore, lower side regions are protected by a larger \(P_{S}\). As the metal front rises toward the top, the static pressure at the top region is very small. When the front reaches the top, \(P_{S}\) becomes nearly zero, so even a moderate \(P_{F}\) can satisfy Eq. (2). Gas from the gas film then enters the liquid steel at the top.
In the upper side region, the situation is different. Liquid residue has been scraped from the foam and attached to the coating. This residue continues to decompose, producing gas at the coating interface. Because the upper side surface is close to the final metal level, \(P_{S}\) is small. The gas pressure \(P_{I}\) can therefore become high enough to satisfy Eq. (3). Gas enters the steel from the coating interface and is trapped as gas holes on the upper outer surface.
The temperature at the top of the casting is also important. The top is the last region to fill, so the liquid steel there has lost more superheat. The steel becomes more viscous. The rise velocity of a gas bubble in liquid steel can be estimated from Stokes’ law:
\[
v_{b} = \frac{2 r_{b}^{2} (\rho_{l} – \rho_{g}) g}{9 \mu_{l}}
\]
where:
\(v_{b}\) is the bubble rise velocity,
\(r_{b}\) is the bubble radius,
\(\rho_{l}\) is the liquid steel density,
\(\rho_{g}\) is the gas density, and
\(\mu_{l}\) is the dynamic viscosity of the liquid steel.
Because the top steel is cooler, \(\mu_{l}\) increases, and \(v_{b}\) decreases. Gas bubbles entering the top region cannot float upward and escape quickly enough. They are trapped near the top surface and on the upper side wall. After machining, these gas holes are exposed. This explains exactly where the defects were found on the sliding rotor ring.
Table 2 summarizes the relationship between defect location and gas intrusion mechanism in this steel casting.
| Defect location | Dominant mechanism | Reason |
|---|---|---|
| Top of the casting | Eq. (2) — gas-film pressure | \(P_{S}\) is near zero; top steel is cold and viscous; gas film pressure is high due to fast filling |
| Upper outer side surface | Eq. (3) — coating-interface gas pressure | Liquid residue accumulates on coating; gasifies and raises \(P_{I}\); \(P_{S}\) at upper side is small |
| Lower part of the casting | Usually no gas holes | \(P_{S}\) is larger; gas cannot enter; gas escapes through the coating |
6. Preventive Measures
Based on the above analysis, I introduced the following measures into my lost foam casting process. All of them are intended to reduce \(P_{F}\), reduce \(P_{I}\), increase gas escape through the coating, or help gas bubbles float out of the liquid steel.
6.1 Increase the Pouring Time
The most important measure was to increase the pouring time. I reduced the cross-section of the gating system. The sprue was changed from 50 mm × 50 mm to 45 mm × 45 mm. The runner was changed from 50 mm × 50 mm to 45 mm × 40 mm. Each ingate was changed from 40 mm × 45 mm to 35 mm × 30 mm. As a result, the pouring time was increased from 19 to 26 s to about 40 s.
This longer pouring time reduced the linear filling velocity \(v_{F}\). It allowed the foam to gasify more completely at the decomposition front. The gasification fraction \(\alpha_{g}\) increased, and the liquid residue fraction \((1 – \alpha_{g})\) decreased. Therefore both \(P_{F}\) and \(P_{I}\) were reduced. The gas that formed had enough time to escape through the coating before the steel solidified.
I also changed the designed gating area ratio from
\[
S_{\mathrm{sprue}} : \Sigma S_{\mathrm{runner}} : \Sigma S_{\mathrm{gate}} = 1:1:1.3
\]
to
\[
S_{\mathrm{sprue}} : \Sigma S_{\mathrm{runner}} : \Sigma S_{\mathrm{gate}} = 1.3:1.1:1.4
\]
This produced a front-sealed and rear-open gating system. The sprue filled quickly and remained full, which reduced aspiration of air. The rear part of the gating system allowed the steel to enter the mold cavity calmly, avoiding turbulence and gas entrainment.
| Parameter | Original process | Modified process |
|---|---|---|
| Sprue cross-section | 50 mm × 50 mm | 45 mm × 45 mm |
| Runner cross-section | 50 mm × 50 mm | 45 mm × 40 mm |
| Ingate cross-section each | 40 mm × 45 mm | 35 mm × 30 mm |
| Designed gating area ratio | 1 : 1 : 1.3 | 1.3 : 1.1 : 1.4 |
| Average pouring time | 19–26 s | about 40 s |
6.2 Use Stepped Gating Instead of Bottom Gating Only
I changed the original side-bottom gating system to a side-stepped gating system. The steel first enters the cavity through the lower ingate. This preserves a calm initial filling stage and prevents oxidation, splashing, and turbulence. When the steel level rises to the upper ingate, the main flow shifts to the upper ingate.
This is very beneficial for lost foam castings because it delivers hot liquid steel to the upper part of the casting. The upper region remains hotter, the viscosity remains lower, and gas bubbles can float out more easily. It also improves feeding of the top risers and reduces the chance of the last-filled region becoming too cold. In my experience, this measure is especially effective for thick steel castings such as the sliding rotor ring.
6.3 Reduce Foam Pattern Density and Use Hollow Sections
Another effective measure is to reduce the density of the foam pattern. In the original process, the pattern density was about 0.024 g/cm3. I reduced it to 0.018 to 0.020 g/cm3 while maintaining sufficient pattern strength and rigidity. This reduced the total amount of foam material that must be decomposed during pouring. Therefore \(\dot{m}_{g}\) and \(\dot{m}_{r}\) were both reduced.
For thick sections, such as the sprue, risers, and heavy regions of the casting, I used hollow foam patterns. Hollow patterns greatly reduce the quantity of polystyrene foam in the regions where gas generation would otherwise be highest. This lowers \(P_{F}\) and also reduces the amount of liquid residue that can accumulate on the coating.
6.4 Increase Coating Permeability and Reduce Coating Thickness
The coating is the main escape path for decomposition gas in lost foam castings. If the coating is too dense or too thick, gas cannot escape easily. I therefore chose a high-temperature permeable coating with good wetting behavior. I also reduced the coating thickness compared with the original process.
The objective is to make the gas escape capacity equal to or greater than the gas generation rate:
\[
\frac{k_{t} A_{c}}{\mu_{g} d_{c}} \left( P_{F} – P_{I} \right) \ge \rho_{\mathrm{EPS}} A_{F} v_{F} \alpha_{g}
\]
This condition helps to keep \(P_{F}\) below the critical value required by Eq. (2). A more permeable coating also reduces the pressure rise at the coating interface, making Eq. (3) less likely to be satisfied.
I emphasize that the coating must still be strong enough to support the foam pattern and resist erosion by liquid steel. The reduction in thickness should be balanced with adequate strength and refractory performance.
6.5 Follow the “Slow-Fast-Slow” Pouring Principle
I also adopted a strict pouring procedure for lost foam castings. At the beginning, I use a small stream of steel to open and preheat the sprue. Then I increase the stream quickly to fill the pouring cup and keep the sprue full. This prevents air aspiration and provides a stable metal front. Near the end of pouring, when the liquid steel reaches the upper part of the casting, I reduce the pouring speed again.
The final slow-pouring stage is important because it allows gas in the upper part of the cavity to escape through the coating and risers. It also gives gas bubbles that may have entered the steel more time to float out before the top region solidifies.
| Measure | Primary effect on pressure | Why it prevents gas holes |
|---|---|---|
| Increase pouring time | Reduces \(v_{F}\), lowers \(P_{F}\) and \(P_{I}\) | Allows foam to gasify completely; residue formation is reduced |
| Stepped gating | Keeps upper steel hot; lowers \(\mu_{l}\) | Bubbles float out more easily; feeding is improved |
| Lower pattern density and hollow sections | Reduces \(\dot{m}_{g}\) and \(\dot{m}_{r}\) | Less gas and less liquid residue are produced |
| Higher coating permeability and thinner coating | Increases \(\dot{m}_{\mathrm{out}}\); lowers \(P_{F}\) and \(P_{I}\) | Gas escapes through coating instead of entering steel |
| Slow-fast-slow pouring | Stabilizes \(P_{F}\); avoids excessive peak pressure | Prevents air aspiration and gives final gas time to escape |
7. Production Verification
After applying these measures, I produced the same steel sliding rotor ring body using the modified lost foam casting process. The pouring time was approximately 40 s, the pattern density was reduced, the gating system was changed to stepped side gating, and the coating permeability was increased. The pouring temperature and vacuum level remained essentially unchanged.
I found that the dispersed irruptive gas holes were essentially eliminated. Machined surfaces no longer showed the large gas holes at the top or upper outer side. The improvement was consistent over multiple castings. This confirmed to me that the original defect was indeed caused by excessive gas-film pressure and coating-interface gas pressure, not by ordinary gas content in the steel.
The results can be understood by re-examining the pressure equations. By reducing \(v_{F}\), I reduced both \(\dot{m}_{g}\) and \(\dot{m}_{r}\). By increasing coating permeability, I increased \(\dot{m}_{\mathrm{out}}\). By using stepped gating, I kept the top region hot enough for gas bubbles to float out. Each measure attacks a different part of the gas intrusion mechanism, and together they provide a robust solution for lost foam castings.
8. Conclusions
From my analysis of this steel sliding rotor ring and from production experience, I draw the following conclusions.
First, gas holes in steel lost foam castings are often irruptive gas holes caused by foam decomposition products. They appear especially on the top and upper side surfaces because the metallostatic pressure is small there and the steel is colder and more viscous.
Second, the conventional sand-casting gas intrusion criterion \(P_{I} > P_{C} + P_{S} + P_{R}\) is not sufficient for lost foam castings. For lost foam castings, two conditions must be considered:
\[
P_{F} > P_{I} + P_{S} + P_{R}
\]
and
\[
P_{I} > P_{C} + P_{S} + P_{R}
\]
The first condition describes gas entering from the gas film at the metal/foam front. The second condition describes gas entering from liquid residue gasifying at the metal/coating interface.
Third, the most effective preventive measures are increasing pouring time, using top or stepped gating, lowering the density of the foam pattern, making thick pattern sections hollow, increasing coating permeability, reducing coating thickness, and applying the slow-fast-slow pouring principle. These measures reduce gas generation, increase gas escape, and promote bubble removal from the liquid steel.
I believe that these principles are generally applicable to steel lost foam castings, not only to the sliding rotor ring described here. A successful lost foam casting process must balance the rate of metal filling with the rate of foam decomposition. When that balance is achieved, the risk of irruptive gas holes in lost foam castings can be dramatically reduced.