During the commissioning of a brand-new lost foam castings production line, I was given the responsibility of leading the technical team for the foundry operation. Over six months of trial and error, the line gradually stabilized, but many quality issues persisted. Among them, cracking defects in lost foam castings were the most troublesome. This article summarizes the root causes, experimental investigations, and practical solutions developed to eliminate these cracks. The primary factors identified were carbon pickup, slag inclusion, and gas porosity. Through carefully designed experiments, detailed data collection, and systematic analysis, we confirmed these hypotheses and implemented effective countermeasures. The findings are intended to share practical experience with others working in lost foam castings technology.
Lost foam castings are produced by coating a foam pattern with a refractory layer, embedding it in dry sand or self-hardening sand, and then pouring molten metal directly onto the pattern. The high-temperature metal decomposes the foam, which “disappears” and vacates the cavity. Compared with conventional sand casting, lost foam castings offer higher dimensional accuracy, design flexibility, better environmental conditions, and lower operating costs. However, no process is perfect, and various defects can still occur, especially in steel castings. The key to ensuring high quality in lost foam castings lies in controlling these defects, particularly in the early stages of production when process parameters have not yet been optimized.
The cracking phenomenon in our newly commissioned line was initially observed after heat treatment. Many castings developed visible cracks after quenching and tempering. Preliminary analysis suggested that the cracks could be attributed to three main causes: excessive carbon pickup leading to composition deviation and internal stress, non-metallic inclusions and slag entrapment, and gas porosity. These factors are often interrelated, and a single-defect approach is insufficient. Therefore, we adopted a multivariate experimental strategy to investigate the effects of process variables on the final quality of lost foam castings.
Carbon content is the most critical compositional factor influencing the mechanical properties of steel castings. In lost foam castings, the foam pattern is typically made of expandable polystyrene (EPS). During pouring, if the foam does not gasify completely, residual carbon can dissolve into the steel, increasing the local carbon concentration. This phenomenon, known as carbon pickup, is particularly severe in low-carbon steel grades. The increased carbon changes the hardenability and distortion behavior, leading to high hardness, difficult machining, non-uniform internal stress, and ultimately cracking during tempering.
To quantify the carbon pickup behavior, we selected the low-alloy steel grade ZG27SiMn with a specified carbon range of 0.24–0.32%. We compared the carbon content of the liquid steel at tap with the carbon content measured across the cross-section of the solidified castings. The measurement points were taken from the surface to the core, labeled as points a, b, c, d, and e, with point spacing between 5 and 10 mm. The laboratory samples were ground and analyzed using optical emission spectroscopy.
Experimental Design and Data
Our experimental work was divided into two phases. The first phase focused on the effect of foam pattern density on carbon pickup. We tested three different densities: 18 kg/m³, 14 kg/m³, and 8–10 kg/m³. All castings in this phase were poured with a closed blind riser system, which is a common practice to minimize oxidation. The steel was tapped with carbon content controlled to the lower limit of the specification (0.24–0.27%). The second phase introduced open riser gating systems and also evaluated the effect of empty-shell casting, where the foam pattern is burned out before pouring, leaving only a hollow shell.
Phase 1: Effect of Foam Density
Batch 1 used foam patterns with a density of 18 kg/m³. Three heats (A, B, C) were poured on the same day. The carbon content of the liquid steel at tap for each heat was 0.24%, 0.25%, and 0.22%, respectively. The carbon content measured from the surface to the core of the finished castings is shown in Table 1.
| Heat | Tap C (%) | a (surface) | b | c | d | e (core) |
|---|---|---|---|---|---|---|
| A | 0.24 | 0.46 | 0.38 | 0.33 | 0.28 | 0.26 |
| B | 0.25 | 0.48 | 0.40 | 0.34 | 0.30 | 0.27 |
| C | 0.22 | 0.44 | 0.36 | 0.31 | 0.27 | 0.25 |
The results revealed severe carbon pickup in lost foam castings. The overall carbon content far exceeded the upper specification limit. Importantly, carbon was not uniformly distributed: it was highest at the surface and decreased toward the core. This gradient is explained by the fact that during filling, the molten steel first contacts the foam pattern and dissolves the carbon-rich thermal decomposition products. The surface solidifies first under the negative pressure, trapping the carbon-enriched layer. In some castings, a shiny carbon film was observed on the surface. The core region, which remains liquid longer, has more time for carbon diffusion and mixing, but the hard skin remains brittle and stressed.
Batch 2 used foam patterns with a density of 14 kg/m³. Three heats were poured on a later date, with tap carbon contents of 0.23%, 0.24%, and 0.22%. All other experimental conditions were identical to Batch 1. The carbon data are given in Table 2.
| Heat | Tap C (%) | a (surface) | b | c | d | e (core) |
|---|---|---|---|---|---|---|
| A | 0.23 | 0.39 | 0.33 | 0.29 | 0.26 | 0.24 |
| B | 0.24 | 0.41 | 0.35 | 0.30 | 0.27 | 0.25 |
| C | 0.22 | 0.37 | 0.32 | 0.28 | 0.26 | 0.24 |
The carbon pickup was still significant, though slightly lower than in Batch 1. All surface values exceeded the maximum allowed carbon of 0.32%. Bright carbon wrinkles were still present. The trend from surface to core remained, but the absolute gradient was somewhat reduced. This confirms that decreasing foam density reduces the amount of carbon available for pickup, but it is insufficient to solve the problem.
Batch 3 used foam patterns with a density of 8–10 kg/m³. Three heats were poured, with tap carbon contents of 0.22%, 0.24%, and 0.23%. Table 3 lists the measured carbon values.
| Heat | Tap C (%) | a (surface) | b | c | d | e (core) |
|---|---|---|---|---|---|---|
| A | 0.22 | 0.31 | 0.28 | 0.26 | 0.24 | 0.23 |
| B | 0.24 | 0.33 | 0.29 | 0.27 | 0.25 | 0.24 |
| C | 0.23 | 0.32 | 0.28 | 0.26 | 0.25 | 0.24 |
In this batch, the overall carbon content just met the ZG27SiMn specification for most points. The surface values approached the upper limit, while the core remained well within range. However, the carbon distribution was still non-uniform, with a noticeable increase from core to surface. This non-uniformity itself can create differential transformation behavior during heat treatment, resulting in internal stresses. It is important to note that lower-density foam patterns have lower mechanical strength and are more prone to deformation. They also have more inter-bead voids, which can cause coating to penetrate the pattern and degrade the surface finish. Some coating fragments may even be entrained in the steel, creating inclusions.
After heat treatment of the first three batches, we observed a high incidence of cracking in all batches. Batch 1 and Batch 2 were the most severe, with some castings completely breaking or losing corners. Batch 3 also exhibited cracks, although less frequently. When the cracked areas were sectioned and examined, we found two distinct crack morphologies. Some cracks were network-like on the surface with no internal defects; these were attributed to carbon-induced compositional gradients and the resulting non-uniform martensitic transformation stresses. Other cracks were large, surface-connected cracks with a honeycomb interior, clearly caused by slag inclusions and gas porosity. The closed blind riser design not only worsened carbon pickup but also trapped gases and prevented slag from floating out of the casting. Thus, it became clear that both carbon control and clean metal delivery were essential.
Phase 2: Open Riser and Empty-Shell Casting
Based on the first-phase findings, we developed a two-pronged strategy: accelerate the escape of carbon decomposition products from the mold and minimize slag/gas entrapment. We implemented stricter slagging practices in the furnace and ladle, added a refractory cover over the ladle to retain slag and heat, and assigned a dedicated operator to control the pouring stream and prevent slag carryover. In terms of gating, we replaced closed blind risers with open risers and also tested the empty-shell casting method, in which the foam pattern is removed by burning out before pouring, leaving a hollow refractory shell.
Batch 4 used medium-density foam (14–16 kg/m³) with an open riser and direct pouring on the unbriefed foam (the so-called “full mold” or “real pattern” casting). The tap carbon contents for three heats were 0.23%, 0.23%, and 0.25%, deliberately kept near the lower limit. The results are shown in Table 4.
| Heat | Tap C (%) | a (surface) | b | c | d | e (core) |
|---|---|---|---|---|---|---|
| A | 0.23 | 0.27 | 0.25 | 0.24 | 0.24 | 0.23 |
| B | 0.23 | 0.28 | 0.26 | 0.25 | 0.24 | 0.24 |
| C | 0.25 | 0.29 | 0.27 | 0.26 | 0.25 | 0.25 |
The carbon content in Batch 4 was significantly more uniform than in Batches 1–3. The maximum increase from tap to surface was approximately 0.04 percentage points, and all samples met the specification. The open riser allowed better gas escape and provided a path for carbon-containing gases to leave the mold. Nevertheless, the carbon was still slightly higher at the surface, indicating that some decomposition products were still being absorbed during filling.
Batch 5 was produced using the empty-shell casting process with open risers. The foam pattern was burned out in a furnace before pouring, leaving a hollow ceramic shell. The steel was tapped at a medium carbon level, aiming for 0.27% as the target. The actual tap carbon values for the three heats were 0.28%, 0.27%, and 0.25%. The measured data are given in Table 5.
| Heat | Tap C (%) | a (surface) | b | c | d | e (core) |
|---|---|---|---|---|---|---|
| A | 0.28 | 0.29 | 0.28 | 0.28 | 0.27 | 0.27 |
| B | 0.27 | 0.28 | 0.27 | 0.27 | 0.27 | 0.27 |
| C | 0.25 | 0.26 | 0.25 | 0.25 | 0.25 | 0.25 |
In Batch 5, the carbon content was essentially uniform across the entire cross-section. The tiny variations observed are attributed to the complete removal of the foam pattern before pouring. There was no carbon pickup from the pattern, and the molten steel interacted only with the refractory coating. The carbon content remained almost identical to the tap value, confirming that empty-shell casting is an effective solution to the carbon pickup problem in low-carbon lost foam castings.
The castings from Batch 4 and Batch 5 were all machined, heat-treated, and inspected. We found no cracking after quenching and tempering. Hardness values were within the specified range, and destructive testing of several castings revealed no internal slag or gas defects. The open riser system promoted efficient slag flotation and gas escape, while the empty-shell process eliminated the carbon source. These two factors combined solved the cracking problem in our lost foam castings.
Quantitative Analysis of Carbon Pickup
To better understand the carbon pickup mechanism, we can define the carbon pickup ratio \( R \) for each batch as $$ R = \frac{C_{\mathrm{surface}} – C_{\mathrm{tap}}}{C_{\mathrm{tap}}} \times 100\% $$ where \( C_{\mathrm{surface}} \) is the average carbon content at point a (the surface), and \( C_{\mathrm{tap}} \) is the carbon content of the liquid steel at tapping. The average values across the three heats for each batch are summarized in Table 6.
| Batch | Average tap C (%) | Average surface C (%) | R (%) |
|---|---|---|---|
| 1 (18 kg/m³, blind riser) | 0.237 | 0.460 | 94.1 |
| 2 (14 kg/m³, blind riser) | 0.230 | 0.390 | 69.6 |
| 3 (8–10 kg/m³, blind riser) | 0.230 | 0.320 | 39.1 |
| 4 (14–16 kg/m³, open riser, real pattern) | 0.237 | 0.280 | 18.1 |
| 5 (14–16 kg/m³, open riser, empty shell) | 0.267 | 0.277 | 3.7 |
The dramatic reduction in \( R \) from 94% to less than 4% demonstrates that a combination of lower foam density, open risers, and empty-shell casting can effectively eliminate carbon pickup in lost foam castings. We can also model the carbon concentration profile using a simple diffusion analogy. Assuming the carbon enters from the surface and diffuses inward, the concentration gradient \( \frac{dC}{dx} \) can be approximated by
$$ \frac{dC}{dx} \approx \frac{C_{\mathrm{surface}} – C_{\mathrm{core}}}{\Delta x} $$
where \( \Delta x \) is the distance from surface to core. This gradient creates a stress field in the solid state because different regions transform to different microstructures during quenching. The resulting internal stress \( \sigma_{\mathrm{th}} \) is proportional to the local carbon difference and can be expressed as
$$ \sigma_{\mathrm{th}} = E \alpha \Delta C $$
where \( E \) is the elastic modulus, \( \alpha \) is the coefficient of expansion due to phase transformation, and \( \Delta C \) is the carbon content difference. Even a 0.1% carbon difference can produce significant transformation gradients, leading to micro-cracks.
Another useful parameter is the carbon segregation index \( S = \frac{C_{\mathrm{max}} – C_{\mathrm{min}}}{C_{\mathrm{avg}}} \). For Batch 1, \( S \) was around 0.6, while for Batch 5 it was less than 0.15. High segregation not only affects mechanical properties but also increases the risk of quench cracking. Therefore, minimizing carbon segregation is of utmost importance in lost foam castings.
Slag and Gas Porosity
In addition to carburization, slag inclusions and gas bubbles were major contributors to cracking. In lost foam castings, the decomposition of foam produces gaseous products that must escape through the coating and the sand. If the riser is blind and the mold permeability is low, these gases become trapped in the solidifying metal. Similarly, slag from the melting furnace, especially if the furnace is not properly deoxidized, can be swept into the mold during pouring. These non-metallic inclusions act as stress concentrators and create internal discontinuities that propagate into visible cracks during heat treatment.
The open riser design provides a natural path for slag and gas to float upward and be removed from the casting. In our open riser experiments, we also added an insulating refractory cover on top of the riser after pouring, which slowed the cooling of the riser and improved its ability to feed the solidifying casting. This prevented shrinkage porosity, which is another potential crack initiator. The empty-shell casting process further improved gas evacuation because the foam was already removed, and no extra gas was generated during pouring.
To quantify the cleanliness of the steel, we performed X-ray inspection on selected castings from each batch. In Batch 1, nearly every casting showed indications of slag and gas pores near the surface. In Batch 4, a few small inclusions were still present, but they were well within acceptable limits. In Batch 5, no significant internal defects were found. This correlates well with the cracking incidence.
Practical Challenges of Empty-Shell Casting
Although empty-shell casting solved the carbon pickup problem, it introduced new challenges that must be carefully managed. First, the refractory coating must be thick and strong enough to withstand the mechanical forces of sand filling and the thermal shock during pouring, because once the foam is burned out, the shell alone supports the cavity. If the coating cracks, molten metal can penetrate the sand, leading to burn-on and rough surfaces. We observed some surface network-like defects on Batch 5 castings; these were traced to cracks in the coating layer that formed during the burn-out process. To avoid this, we increased the coating thickness to around 2 mm and ensured complete drying. However, thicker coatings reduce gas permeability, so a balance must be found. Developing an optimized coating composition for empty-shell lost foam castings is an ongoing research topic.
Second, the sand must be compacted with sufficient negative pressure (vacuum) to hold the shell in place. When the foam pattern is removed, the shell has very little structural integrity. The vacuum level in the mold determines whether the sand can support the shell until the metal is poured and solidifies. We found that a vacuum level around 50 kPa (as referenced in our experiments) worked well for simple shapes, but more complex castings required careful pressure control. The holding time after pouring is also critical: too short a time leads to incomplete solidification and distortion; too long a time prevents stress relief and increases the risk of hot tears. The optimal holding time depends on the section thickness and must be determined experimentally for each casting design.
Third, coordination between the burn-out operator, the pouring crew, and the production line staff is essential. Any hesitation between removing the empty shell and pouring can cause the shell to collapse or the sand to collapse into the cavity. We experienced several catastrophic failures during the initial trials, resulting in scrapped castings. Therefore, the process sequence must be meticulously synchronized.
Fourth, open risers have lower thermal insulation capacity than closed risers, and their feeding efficiency is reduced. To compensate, we used top pouring with open risers and added insulating exothermic compounds on top of the risers. This helped maintain a liquid metal reservoir for feeding until the casting had fully solidified. The riser sizes had to be increased, which lowered the casting yield. In our case, the overall yield decreased from about 70% to 60%, but the elimination of cracks justified the extra cost.
Fifth, slag control remains a critical issue. Even with an open riser, excessive slag will accumulate in the riser and may eventually be drawn into the casting if the pouring stream is turbulent. We installed a ceramic foam filter in the gating system for later trials, which further reduced slag-related defects. The combination of good deoxidation in the furnace, clean ladles, filtering, and careful pouring techniques is recommended for all lost foam castings.
Discussion
This investigation confirms that the cracking of low-alloy steel lost foam castings is a multifactorial problem. Carbon pickup from the foam pattern alters the surface chemistry and creates a steep carbon gradient. This gradient, in turn, leads to non-uniform phase transformation and internal stress during heat treatment. The presence of slag and gas pores reduces the effective load-bearing area and provides crack initiation sites. Our experiments showed that reducing foam density alone, while beneficial, is not enough to eliminate the risk. Even at a very low foam density of 8–10 kg/m³, the surface carbon content remained close to the upper limit and the gradient was still significant. This gradient may be acceptable for some applications, but for safety-critical components, the risk of quench cracking remains.
The open riser system was a major improvement because it allowed decomposition gases and slag to leave the mold. However, when the foam pattern is still present during pouring, the gas generation rate is so high that the riser cannot fully compensate for the surface carbon pickup. Only by removing the foam entirely through empty-shell casting did the carbon content remain uniform. The empty-shell method also dramatically reduced gas evolution inside the mold, which contributed to a cleaner casting. Therefore, for low-carbon and low-alloy steel lost foam castings where cracking is a concern, empty-shell casting should be the preferred process.
It is worth noting that empty-shell casting is not suitable for all geometries. Very thin sections may have fragile shells that cannot withstand sand compaction. Complex internal channels may be difficult to coat and burn out uniformly. The cost of the additional burn-out step and the lower yield must be weighed against the benefit of reduced defects. For less critical components, a carefully designed open riser system with medium-density foam and a strong coating may be adequate. The choice of process parameters should be based on the specific casting design and quality requirements.
Conclusion and Outlook
The cracking defects observed in our new lost foam castings production line were successfully traced to two primary sources: carbon pickup and slag/gas porosity. Through a series of controlled experiments using ZG27SiMn steel, we quantified the effect of foam density, riser type, and casting method on carbon distribution and casting integrity. The results clearly show that the empty-shell casting method, combined with an open riser and rigorous slag control, eliminates carbon pickup and removes gas and slag defects, thereby preventing cracks after heat treatment. This approach has since been adopted as the standard for carbon-sensitive steel grades in our foundry.
Nevertheless, future work is required to optimize the coating materials for empty-shell lost foam castings. The interplay between coating thickness, permeability, and mechanical strength is still not fully understood. We are also investigating the use of alternative foam materials that leave less carbon residue, as well as advanced simulation tools to predict carbon diffusion and stress development during solidification and heat treatment. The goal is to make lost foam castings more reliable and allow their use in even the most demanding applications.
In conclusion, the path to defect-free lost foam castings lies in a comprehensive understanding of the physical and chemical interactions during mold filling and solidification. By controlling the carbon source, enhancing gas and slag removal, and optimizing heat treatment, we can unlock the full potential of this promising casting process. The experiments described here provide a practical roadmap for other foundries facing similar challenges with lost foam castings.
We hope that sharing these findings will contribute to the wider acceptance and improvement of lost foam castings technology, especially in steel foundries where quality demands are high.