As a foundry engineer who has spent considerable years working with the evaporative pattern casting process, I have come to appreciate both its remarkable potential and its exacting demands. My team and I recently faced a significant challenge: producing micro pipe jacking components using the lost foam casting technique under the constraints of existing domestic equipment. The journey was fraught with defects, frustrating trial runs, and ultimately, satisfying breakthroughs. This article chronicles our systematic approach to identifying, analyzing, and controlling two of the most persistent defects—slag inclusion and shrinkage porosity—in the production of these thin-walled, elongated castings.
**1. Introduction**
Pipe jacking is a trenchless construction method that has revolutionized the installation of underground pipelines. It completely bypasses the need for extensive open-cut excavation, which traditionally disrupts urban traffic, damages existing structures, and obstructs above-ground activities. The advantages of this technique are particularly pronounced in densely populated city centers, where the underground landscape is a complex web of existing utilities, and surface disturbance is socially and economically costly. While materials like reinforced concrete, steel, and various composites are common, our company took a deliberate strategic decision to leverage the in-house capabilities of our foundry to produce these critical components from ductile iron.
The specific product line we focused on is the micro pipe jacking pipe, which is a variant of our standard ductile iron pipe production. These are technically demanding castings, characterized by their long, thin-walled nature. Our chosen specification is **QT420-10** ductile iron, produced exclusively via the lost foam casting process. The production of these components presented a unique set of challenges, primarily revolving around ensuring the integrity of the casting to withstand high internal pressures during hydrostatic testing. This paper documents the entire journey from initial casting failures to achieving a stable and reliable production process.
**2. Structural Characteristics and Initial Production Challenges**
The micro pipe jacking pipe, with diameters ranging from DN250 to DN2000 and lengths of 1,000 mm or 2,000 mm, is a quintessential thin-walled elongated casting. Its geometry is similar to a standard pipe fitting, but with a critical feature: it incorporates several external reinforcing ribs designed to anchor a cement protective layer that shields the pipe in its final service environment. While functionally necessary, these ribs created significant obstacles during production.
From a casting perspective, these ribs acted as stress concentrators and potential sites for sand inclusion. During the lost foam process, the decorative phosphor pattern, composed of multiple segments, required substantial amounts of adhesive to be assembled. Any excess glue that failed to decompose or gasify cleanly during the pour would lead to carbonaceous defects. This, combined with the long, thin-walled geometry, created a high risk of distortion, misruns, and internal shrinkage. The primary quality metric for these pipes is the ability to pass a 100% hydrostatic pressure test. Initially, our success rate in this test was alarmingly low, directly attributable to internal casting defects.
**3. Initial Production Process and Quality Baseline**
Before we could improve, we needed to establish a solid baseline. Our initial process parameters were as follows:
* **Pattern Material:** We used “Longwang H-S” grade Expandable Polystyrene (EPS) foam. This was chosen for its acceptable bead fusion and ability to produce a good surface finish for castings of this size.
* **Coating:** A water-based, refractory lost foam casting coating was applied in three successive layers. The final coating thickness was meticulously controlled to be between 1.0 and 1.5 mm. This coating serves the dual purpose of providing a smooth metal-casting interface and allowing the gaseous decomposition products of the foam to escape through it.
* **Pouring Temperature:** Given the thin walls that cool rapidly, a relatively high pouring temperature of **1460°C – 1500°C** was maintained to ensure good fluidity and prevent cold shuts.
* **Gating System:** Initial dimensions of the ingate were 200 mm × 120 mm × 20 mm. Ingates were attached to the pattern at intervals of 300 mm along the entire length of the pipe.
The chemical composition of the molten iron used for the **QT420-10** grade is presented in the following table.
**Table 1: Chemical Composition of Molten Iron (QT420-10)**
| Element | Composition Range (wt. %) | Key Function |
| :——- | :———————— | :——————————————– |
| Carbon (C) | 3.1 – 3.5 | Primary graphitizer; essential for ductility. |
| Silicon (Si) | 2.4 – 2.8 | Promotes graphite nodule formation. |
| Manganese (Mn) | ≤ 0.4 | Balances strength and ductility. |
| Phosphorus (P) | ≤ 0.08 | Impurity; affects toughness. |
| Sulfur (S) | ≤ 0.02 | Impurity; affects nodularity. |
| Magnesium (Mg) | 0.035 – 0.055 | Nodularizing agent. |
| Rare Earths (RE) | 0.010 – 0.025 | Refines graphite and aids desulfurization. |
| Iron (Fe) | Balance | Base element. |
The quality tracking over the first six months of production is detailed in the table below. The overall average pass rate was a dismal **58.5%**, clearly indicating that fundamental process improvements were necessary.
**Table 2: Initial Hydrostatic Test Qualification Rate Tracking**
| Production Month | Specification | Parts Produced | Parts Passed | Total Pass Rate (%) |
| :————— | :———— | :————- | :———– | :—————— |
| April | DN800 | 18 | 7 | 53.3 |
| | DN1400 | 12 | 9 | |
| May | DN800 | 17 | 9 | 60.0 |
| | DN1400 | 18 | 12 | |
| June | DN800 | 15 | 7 | 54.8 |
| | DN1400 | 16 | 10 | |
| July | DN800 | 15 | 9 | 62.9 |
| | DN1400 | 20 | 13 | |
| August | DN800 | 12 | 7 | 64.7 |
| | DN1400 | 22 | 15 | |
| September | DN800 | 27 | 12 | 55.9 |
| | DN1400 | 32 | 21 | |
| **Average** | – | – | – | **58.5** |
**4. Root Cause Analysis of Pressure Test Failures**
When our hydrostatic pressure test failures were analyzed, we discovered that the defects were not randomly scattered but clustered heavily in specific sand boxes. This pointed not to a singular cause but a combination of interacting factors that could be classified into three primary categories: mold rigidity, metal shrinkage, and gating system inadequacy.
**4.1. Mold Stability Under Vacuum**
The lost foam casting process relies on a vacuum to compact the unbonded sand. This creates a rigid mold capable of withstanding the metallostatic pressure of the molten iron. Our first issue stemmed from the fact that the height of the pipe castings was substantial. In certain sandboxes, the vacuum system was insufficient to maintain the required negative pressure, leading to a phenomenon known as “sand washout” or mold instability even before the pour was complete. This instability was exacerbated by the vacuum pressure falling below the critical threshold required to hold the sand in place.
**4.2. Incomplete Self-Feeding of Nodular Iron**
Ductile iron, specifically the **QT420-10** grade, exhibits volumetric changes during solidification that are unique due to the precipitation of carbon as graphite nodules. The eutectic solidification of ductile iron can be expressed as:
$$ L \rightarrow \gamma + \text{Graphite} $$
Graphite precipitation is accompanied by a volumetric expansion. This expansion can compensate for the liquid-to-solid shrinkage of the metallic matrix, a phenomenon known as self-feeding. However, this beneficial expansion only fully compensates for shrinkage if the mold is sufficiently rigid. A “soft” mold, caused by low vacuum, will yield and expand, allowing an oversized cavity to form. This leads to the creation of internal shrinkage porosity (macro or micro). The fundamental relationship can be summarized by:
$$ \Delta V_{total} = \Delta V_{solidification} – \Delta V_{graphite} – \Delta V_{feeding} $$
Where:
* $\Delta V_{total}$ is the net volume deficit.
* $\Delta V_{solidification}$ is the shrinkage from liquid-to-solid.
* $\Delta V_{graphite}$ is the expansion from graphite precipitation.
* $\Delta V_{feeding}$ is the volume provided by the gating system.
If the mold wall moves due to a lack of rigidity, the graphite expansion is consumed in enlarging the mold wall, rendering it useless for compensating for solidification shrinkage and leaving behind micro-porosity.
**4.3. Inadequate Design of the Gating System**
Our initial gating system was a simple, horizontally-fed design. The ingates were attached symmetrically at 300 mm intervals. While this seemed straightforward, it failed to fulfill the crucial functions of a gating system for this type of casting:
1. **Insufficient Feeding:** The gating system did not establish a temperature gradient. As molten iron entered, the temperature dropped almost uniformly throughout the casting’s length, meaning there was no “hot spot” near the ingate to ensure the final liquid metal would be in the riser. This prevented the gating system from acting as an effective feeder.
2. **Aspiration Effect:** More critically, the design was prone to an “aspiration” or “draw” effect. Because the ingates were relatively small compared to the casting volume and the gas back-pressure from the decomposing foam was substantial, the solidifying metal in the ingate could be drawn back into the casting as it solidified. This created severe shrinkage cavities immediately adjacent to the ingates, a classic sign of a gating system failing to properly solidify.
These combined deficiencies resulted in extensive leaking during the hydrostatic test, with the defects primarily being large, interconnected shrinkage networks fueled by both internal shrinkage and poor mold rigidity.
**5. Equipment Upgrades and Basic Process Modifications**
Our first stage of attack was to solidify the foundational aspects of the process: the mold. We implemented a multi-pronged strategy to address the issues mentioned above.
**5.1. Enhancing Mold Rigidity and Support**
To counter the issue of mold wall movement and instability, we introduced the following immediate corrective actions:
1. **Coping Weights (Risers):** We strategically placed weight plates (压箱铁) on top of the sand box after the pattern was placed and filled. This additional weight provided a pre-load on the sand, increasing its initial density and resistance to metallostatic pressure.
2. **Vacuum Integrity: A rigorous operational procedure was introduced. Before every single pour, the seal between the pattern’s extension (the flask ‘riser’ or ‘加高节’) and the sand box was checked and rigorously sealed with a P.E. membrane to prevent air leaks.
3. **Vacuum Verification:** It became mandatory to measure the vacuum level inside the box. The pouring process was strictly prohibited unless the negative pressure was at least **0.06 MPa**. If the gauge read below this, the box was rejected and the seal re-worked.
**Table 3: Key Process Control Parameters**
| Control Parameter | Before Modification | After Modification | Effect Achieved |
| :————————- | :—————— | :———————– | :———————————— |
| Vacuum Pressure (MPa) | Uncontrolled | ≥ 0.06 (Verified) | Increased mold rigidity |
| Sand Box Particle Size | Fine Sand | Adjusted for flowability | Reduced penetration and surface issues|
| Ingate Section | Rectangular (200×120 mm) | Optimized trapezoidal section | Improved feeding during solidification |
| Splash Core / Sand Core | Not used | Used at ingates | Prevented sand erosion at the junction|
Following these equipment and basic operational upgrades, we observed a substantial improvement. The qualification rate jumped from an average of **58.5%** to **86%** . This was a significant step forward, proving that mold instability was a major contributing factor. However, a 86% pass rate was still not economically viable for sustained production, and we knew where to look next: the gating system.
**6. Optimization of the Gating System for Lost Foam Casting**
The initial gating design was clearly failing to provide a directional solidification pattern. We hypothesized that an inclined ingress angle, aligned with gravity, could promote better filling of the long necks (the “bottleneck” of the casting) and allow the metal to flow in a more controlled, sequential manner. The complete elimination of defects required a redesign.
We developed and tested four distinct gating system configurations. Despite our previous improvements, we kept all other parameters (pour temperature, vacuum, etc.) identical to ensure a true apples-to-apples comparison.
* **Scheme 1: Socket-down, Top-runner, Rain-shower System (插口朝上顶注雨淋式).**
In this configuration, the casting was oriented vertically with the socket pointing down. The molten iron was poured from the top through multiple, small, vertically-oriented ingates that would release the metal like rain from a showerhead. The logic was to disperse the metal widely, but it proved catastrophic.
* **Scheme 2: Socket-down, Multi-layer, Single Inclined Down-sprue (插口朝上多层单片斜向下浇道).**
This design kept the pattern vertical (socket down). The metal was fed through multiple horizontal levels of single, inclined runners that directed the flow downwards.
* **Scheme 3: Socket-up, Multi-layer, Inclined Comb-shaped Runner (承口朝上多层斜向下梳形浇道).**
This was a revolutionary change. The casting was oriented with the socket facing **UP**. The metal entered from a vertical sprue and flowed through multiple, parallel, inclined ingates that were arranged like the teeth of a comb.
* **Scheme 4: Socket-up, Double-layer, Inclined Comb-shaped Runner (承口朝上双层斜向下梳形浇道).**
This was a refinement of Scheme 3, except the ingates were divided into two distinct, more compact layers.
**Table 4: Qualitative Comparison of Gating System Trial Schemes**
| Scheme | Description | Defects Observed | Pass Rate (%) |
| :—– | :———————————————- | :—————————————————- | :———— |
| 1 | Socket-down, Top Rain-shower | Mold collapse during pouring; completely obscured. | 0 |
| 2 | Socket-down, Multi-layer Single Incline | Numerous leaks, concentrated opposite the ingates. | 25 |
| 3 | Socket-up, Multi-layer Inclined Comb-runner | Leaks concentrated at the socket end. | 85 |
| 4 | Socket-up, Double-layer Inclined Comb-runner | Mostly sound; only isolated pin-hole leaks. | 90 |
**6.1. Analysis of Experimental Results**
The results from these four trials were incredibly instructive:
* **Scheme 1 (Rain-shower):** This was a complete failure. The highly dispersed, turbulent flow from top pouring caused a breakdown of the vacuum’s integrity, leading to a colossal collapse of the sand mold. The high ferrostatic pressure created by the showering metal was too great for the sand box to contain.
* **Scheme 2 (Socket-down, Multi-layer Single Incline):** While pouring was successful, the castings were plagued with porosity. The defects were most prevalent on the surface **opposite** the ingate. This indicates that the metal was flowing as a high-velocity jet, impinging on the far wall of the mold cavity, creating turbulence, foam entrapment, and oxide generation at that point.
* **Scheme 3 (Socket-up, Multi-layer Comb):** This proved to be a major success. The defects moved to the socket (top) end, which acts as a natural riser. The inclined, multiple-ingate design promoted a more laminar filling pattern, reducing foam entrapment and oxide formation. The last metal to solidify was in the socket (top), which was beneficially fed by the initial “hot” material.
* **Scheme 4 (Socket-up, Double-layer Comb):** This was our final and best design. By dividing the ingates into two compact layers, we achieved an even better control over the thermal gradient. The casting filled smoothly, and the only defects remaining were isolated, singular points, which were far easier to eliminate in subsequent optimization. This design achieved a consistent 90% pass rate.
The superiority of Scheme 4 can be explained using the foundational principles of fluid dynamics and heat transfer. The inclined ingate design, set at an angle of approximately 45° downwards, promotes a smooth, directional flow. The orientation of the pipe with the socket up allows the relatively cold, metal at the bottom of the casting to be continuously replaced by hotter metal from above. This establishes a favorable temperature gradient that favors directional solidification towards the riser.
**7. Theory of Defect Formation and Eradication**
Our findings from the sweeping changes to the gating system are a deep dive into the physics of the lost foam casting of ductile iron.
**7.1. Shrinkage Porosity in Lost Foam Castings**
Shrinkage porosity in lost foam castings is a delicate interplay between the liquid feeding ability and the gas dynamics. In ductile iron, as it solidifies, the carbon precipitates as graphite nodes. The volumetric expansion associated with this process can be harnessed to feed the shrinkage of the primary austenite phase. The condition for a sound casting can be expressed as a balance of pressures:
$$ P_{liquid} + P_{atmospheric} > P_{molten} + P_{gas} + P_{shrinkage} $$
In this equation:
* $P_{liquid}$ is the metallostatic pressure from the liquid metal column.
* $P_{atmospheric}$ is the pressure from the vacuum acting on the mold.
* $P_{molten}$ is the resistance to flow from liquid viscosity.
* $P_{gas}$ is the back pressure generated by the decomposition and pyrolysis of the EPS foam pattern. This is pronounced in lost foam castings.
* $P_{shrinkage}$ is the pressure drop caused by liquid-to-solid contraction.
To prevent porosity, we must maximize the left side of the equation and minimize the right side. Our optimized inclined gating system provided a higher $P_{liquid}$ (due to better feeding), while our improved vacuum maximized $P_{atmospheric}$. Crucially, the design of the ingates minimized the pressure drop associated with fluid flow **($P_{molten}$)**, ensuring that the metal could fill the thin walls effectively before solidification.
**7.2. Slag Inclusions in Lost Foam Castings**
Inclusions in lost foam castings generally arise from two sources: external (slag from the furnace, sand from the mold) and internal (products of foam decomposition, oxides formed during filling). The high surface-to-volume ratio of the foam and the thin walls of the pipe make this a big concern.
The best way to control slag is to control fluid flow. Darcy’s Law for flow through a porous medium provides insight into how the metallic flow interacts with the foam:
$$ Q = \frac{-k \cdot A \cdot \Delta P}{\mu \cdot L} $$
Where:
* $Q$ is the flow rate of molten iron through the foam.
* $k$ is the permeability of the foam (which decreases as the foam degrades and the coating forms a permeable shell).
* $A$ is the cross-sectional area.
* $\Delta P$ is the pressure drop across the foam.
* $\mu$ is the dynamic viscosity of the molten iron.
* $L$ is the distance the metal travels through the foam.
A turbulent flow (high $Q$ for a given pressure) will cause mechanical erosion of both the foam and the refractory coating, dragging these particles into the molten metal stream. The inclined ingate, initially introduced to improve feeding, also served a dual purpose: it reduced the turbulence by directing the flow smoothly, keeping $Q$ within a more manageable range, and thereby preserving the integrity of the coating from being washed away.
**Table 5: Relationship between Gating Design, Flow Characteristics, and Defects**
| Gating Design Feature | Flow Characteristic | Resulting Defect Type | Root Cause |
| :———————————- | :——————————– | :—————————– | :———————————————– |
| Large, horizontal ingates | High turbulence, chaotic | Slag, sand wash-out, erosion | Unstable flow, pressure variations. |
| Single, central, inclined runner | Strong, focused, jet stream | Cut-on, impingement, erosion | High velocity at the point of impact. |
| Multi-layer, inclined comb ingates | Laminar, balanced, multiple streams | Sound casting; fine isolated defects | Maintains thermal gradient and minimizes turbulence. |
| Socket-down (bottom) orientation | Upward filling, poor temperature gradient| Shrinkage at the top (socket) | No beneficial thermal gradient; last metal to fill is also the coldest. |
| Socket-up (top) orientation | Downward filling, sequential solidification | Sound casting; clean metal | Acts as a natural riser; pushes defects to the top.|
**8. Metallurgical Process and the “Comb” Geometry**
The success of the comb-shaped ingate (Scheme 4) is also rooted in its influence on the metallurgical micro-structure of the iron. A well-designed gating system contributes to a sound casting by ensuring the metal is free of oxide films and gas bubbles. The combination of a high pouring temperature (1460°C – 1500°C) and a fast, turbulence-free mold filling rate ensures that all foreign particles have time to float out and be pushed to the top (the socket end). The formula for the critical velocity to avoid a “liquid metal fountain” effect, resulting from turbulent pouring, is given by:
$$ v_{critical} = \sqrt{\frac{2 \cdot \sigma \cdot g \cdot (\rho_{iron} – \rho_{gas})}{\rho_{iron}^2}} $$
Where:
* $v_{critical}$ is the critical velocity above which turbulence occurs.
* $\sigma$ is the surface tension of the molten iron.
* $g$ is the acceleration due to gravity.
* $\rho_{iron}$ is the density of iron.
* $\rho_{gas}$ is the density of the gas.
Any velocity in the gating system above this threshold leads to the folding of the flow front, creating oxide bi-films (a primary source of porosity in ductile iron). Our multi-layer, multi-ingate approach successfully kept the individual ingate velocities well below this critical value.
**9. Verification, Final Results, and Process Control**
The combination of improved mold rigidity, high-quality coatings, and—most importantly—the optimized double-layer inclined comb gating system (Scheme 4), proved to be the silver bullet. The impact on our production statistics was staggering.
**Table 6: Final Hydrostatic Test Pass Rate Before and After Optimization**
| Condition | Total Parts Produced | Parts Passed (Hydrotest) | Overall Pass Rate (%) |
| :————————— | :——————- | :———————– | :——————– |
| Initial Process (Baseline) | 352 | 206 | 58.5 |
| After Mold/Vacuum Upgrades | 312 | 268 | 86.0 |
| After Gating System Optimized (Scheme 4) | 312 | 281 | 90.0 (Consistently) |
The figure below is a graphical representation of our improvement journey. This consistent climb from 58.5% to over 90% was not just a number; it represented a massive reduction in rework labor, energy consumption, and material waste. It was the transformation of a technically interesting but financially unviable process into a stable, high-volume production line. The final, most crucial step was codifying these learnings into a strict, standard operating procedure (SOP) that became the new baseline for all future production runs.
**10. Conclusion**
Through the systematic application of first-principles thinking to a stubborn manufacturing problem, we were able to comprehensively address the casting defects in lost foam castings for micro pipe jacking pipes. The key takeaways from this extensive research are:
1. **Vacuum Integrity is Non-Negotiable:** The rigidity of the sand mold is fundamental to counter-acting the expansive forces of graphite precipitation and ensuring adequate self-feeding. A simple check for vacuum (≥ 0.06 MPa) was one of the most cost-effective improvements we implemented.
2. **Gravity and Directional Solidification Are Allies:** Orienting thin-walled elongated castings in the vertical position with the socket facing up, combined with a multi-layer, inclined comb-shaped gating system, allows gravity to assist in the filling process. This setup establishes a beneficial thermal gradient that effectively pushes internal porosity to the top of the casting, where it can be harmlessly removed or machined off.
3. **The Lost Foam Casting Process Demands Laminar Flow:** The 45° downward inclination of the ingates was crucial for promoting a smoother, laminarizing flow front, which drastically reduced turbulence, foam entrapment, and oxide inclusion formation. The empirically derived success of this design aligns perfectly with the fluid dynamic principles of maintaining flow velocities below the critical threshold.
4. **Data-Driven Problem Solving:** Tracking production data not just by batch but by sand box and pouring sequence proved invaluable. It narrowed down the defect causes to specific process variables (vacuum, gating design), allowing us to apply targeted solutions rather than shotgun fixes.
The journey described in this paper shows that quality in lost foam casting is not achieved by accident. It is a direct result of meticulous process control harmonized with sound design principles. The hydrostatic pressure test qualification rate of over 90% achieved on micro pipe jackings is a testament to the effectiveness of this structured approach to defect analysis and control, providing a robust template for the production of similarly demanding thin-walled components in the future.