I optimized the investment casting process for a forklift head used as part of a fork assembly. The part has a hollow, side-enclosed, thin-wall structure and is joined to other components by welding. Because welding performance is critical, I selected a low-carbon alloy steel with the mechanical-property window shown in Table 1. The low carbon content gives high toughness, which is beneficial for service but difficult for gate removal: when I tried hammering off the gating system, the inner gate did not fracture reliably, and repeated hammer blows produced impact marks. For similar low-carbon steel parts, I had previously used flame cutting to remove the gating system, but that route is costly and slow. I therefore used casting CAE simulation to redesign the gating system so that hammer-off removal would become possible. The same logic is useful when I evaluate lost foam castings, where gate and runner removal can also dominate finishing cost.
Table 1. Mechanical-property requirements for the low-carbon steel.
| Carbon content, ω(C) (%) | Yield strength (MPa) | Tensile strength (MPa) | Elongation (%) |
|---|---|---|---|
| 0.18–0.22 | 240 | 450–700 | 22 |
From the three-dimensional model, I confirmed that the wall thickness is relatively uniform. I chose a single-runner, inner-gate gating system and calculated the component sizes by the sprue-cup feeding-capacity method. Although my production route is investment casting, I often compare the feeding behavior with lost foam castings, because both processes demand that the liquid metal path remain open long enough to feed the last solidifying regions.
Inner Gate Sizing
The sprue-cup feeding-capacity method treats a casting as a set of cylindrical equivalent hot spots. Each hot spot has an equivalent hot-spot diameter. When the inner gate length is 10–15 mm, I related the inner-gate equivalent hot-spot diameter to the casting hot-spot equivalent diameter by:
$$D_g = k D_c$$
where Dg is the inner-gate equivalent hot-spot diameter in mm, Dc is the casting hot-spot equivalent hot-spot diameter in mm, and k is a weight factor. For a rectangular hot-spot section a × b, I read Dc from the equivalent-hot-spot chart. The weight factor k depends on the weight of the casting zone that the individual gate must feed. When several hot spots require several gates, I divide the casting into feeding zones and use the zone weight G for each gate.
At the two critical hot spots, I measured a section of approximately 15 mm × 75 mm. The weight of the casting feeding zone assigned to one inner gate was about 600 g. Using the equivalent-hot-spot chart, I obtained:
$$D_c = 23 \ \text{mm}$$
Using the weight-factor chart, I obtained:
$$k = 0.93$$
Therefore:
$$D_g = k D_c = 0.93 \times 23 = 21.39 \ \text{mm}$$
Because the local casting structure did not allow a round inner gate of 21.39 mm diameter, I converted the gate to a rectangular section of 22 mm × 20 mm. This change kept the feeding area close to the calculated requirement while fitting the available space.
Table 2. Initial inner-gate calculation.
| Parameter | Value | Unit |
|---|---|---|
| Hot-spot section | 15 × 75 | mm |
| Feeding-zone weight, G | 600 | g |
| Equivalent hot-spot diameter, Dc | 23 | mm |
| Weight factor, k | 0.93 | — |
| Inner-gate equivalent diameter, Dg | 21.39 | mm |
| Adopted inner-gate section | 22 × 20 | mm |
To make hammer-off removal reliable, I designed a sharp-angle neck in the direction of the breaking force. This neck creates stress concentration, so the gate deforms locally and then fractures completely at the neck. I did not add a neck in the other direction because it has little effect on hammer-off removal and would add mold inserts. Based on prior experience, I selected a neck angle of 75°. To reduce the risk of a missing-metal defect on the casting body after fracture, I left a 1.5 mm root allowance below the inner-gate neck. This design principle is also relevant to lost foam castings, where a controlled neck can localize fracture and reduce finishing damage.
Runner and Sprue-Cup Sizing
In the sprue-cup feeding-capacity method, the runner does not provide feeding; it only acts as a feeding channel. As long as the channel remains open, a smaller runner section is preferred. I used:
$$D_k \ge 1.15 D_g \quad \text{to} \quad 1.25 D_g$$
where Dk is the equivalent hot-spot diameter of the runner section in mm. With the available rod-core sizes and the inner-gate weld-face dimensions, I selected a runner equivalent hot-spot diameter of 35 mm.
Because the sprue cup is the main feeding source in this single-runner system, I had to provide sufficient feeding volume without making the cup unnecessarily large. For a square runner section, I used the standard size groups for the runner and sprue cup. With Dk = 35 mm, I selected the third group. The inner-gate distance limited the minimum length L to 105 mm, and I used D = 70 mm.
Table 3. Selected square-runner and sprue-cup sizes.
| Dk (mm) | D1 (mm) | D2 (mm) | K (mm) | H (mm) | L (mm) | D (mm) |
|---|---|---|---|---|---|---|
| 25 | 25 | 20 | 25 | 25 | 2Dk, 3Dk, 4Dk | 50, 60, 70, 80 |
| 30 | 30 | 25 | 30 | 30 | 2Dk, 3Dk, 4Dk | 60, 70, 80, 90 |
| 35 | 35 | 30 | 35 | 35 | 2Dk, 3Dk, 4Dk | 70, 80, 90, 100 |
| 40 | 40 | 35 | 40 | 40 | 2Dk, 2.5Dk, 3Dk | 80, 90, 100, 110 |
| 45 | 45 | 40 | 45 | 45 | 2Dk, 3Dk | 90, 100, 110, 120 |
| 50 | 50 | 45 | 50 | 50 | 2Dk, 2.5Dk, 3Dk | 100, 110, 120, 130 |
After assembling the wax pattern tree, I used the layout as the initial simulation model. This is the stage at which I also compare the thermal mass of the gating system with lost foam castings, because excessive gating mass can delay solidification and alter feeding behavior.
Initial Simulation and Problem Identification
The casting CAE software included a pre-processing module, a solver, and a post-processing module. I used it for solidification analysis rather than full filling analysis because the purpose was to verify the calculated gating dimensions. The solidification model showed only a small overall shrinkage tendency, but it also showed a very small shrinkage pore at the inner-gate location. More importantly, the solidification map showed that the inner gate solidified before the hot spot, creating a clear flow blockage. The hot spot therefore could not receive enough liquid feeding.
Based on prior experience, I predicted that the physical part would show a certain proportion of shrinkage porosity at the hot spot after gate blockage. That porosity could affect subsequent machining of the hole. The initial design therefore needed optimization. In lost foam castings, I have seen the same warning sign: a frozen gate can isolate a hot spot even when the sprue cup still contains liquid metal.
Table 4. Initial simulation diagnosis.
| Observation | Cause | Consequence |
|---|---|---|
| Small shrinkage pore at inner gate | Local thermal mass insufficient | Possible machining defect |
| Flow blockage before hot-spot solidification | Inner gate cools too quickly | Hot spot cannot be fed |
| Runner channel adequate | Runner section acceptable | No major runner-side blockage |
| Liquid metal surplus between gates | Central runner region over-sized | Lower process yield |
Optimization of the Inner Gate
The blockage showed that the inner-gate feeding channel was not sufficient. The feeding channel depends not only on the neck section area but also on heat loss at the inner gate. At the same neck area, a gate that loses heat more slowly remains liquid longer and feeds more effectively. Heat loss from the inner gate is related to the neck section size and shape, the gate shape, the gate height, and the gate weld-face area. In the initial simulation, the gate was almost completely solid when blockage occurred. Without increasing the neck section, I could reduce heat loss by increasing the gate volume, increasing the weld-face area, and reducing the gate height.
I also wanted to reduce the risk of wax-pattern breakage during shell building, which can occur when the neck section is too small. If I increased the gate size, the runner weld face became insufficient, so I had to increase the runner size accordingly. The original gate height was 15 mm. Lowering the height directly would affect sand filling around the gate, so I introduced an angled weld face. This change reduced the bottom clearance of the pattern tree and allowed the shell to fill the gap, increasing overall shell strength. I placed the tree horizontally during knockout, with the sprue cup at an angle to the horizontal, so hammering required less effort.
The optimized inner-gate geometry can be summarized by the following relationships:
$$A_{weld,opt} = 2850 \ \text{mm}^2$$
$$\theta_{neck} = 75^\circ$$
$$t_{root} = 1.5 \ \text{mm}$$
Here Aweld,opt is the optimized weld-face area, θneck is the neck angle, and troot is the root allowance. I found that increasing the weld-face area from 1950 mm² to 2850 mm² was sufficient to eliminate the flow blockage and remove the shrinkage tendency. This result is consistent with my experience with lost foam castings: a larger local thermal mass near the gate can keep the feeding path open, but it must be balanced against knockout and cleaning requirements.
Optimization of the Runner Tree
The initial simulation showed that the sprue cup contained enough liquid metal for feeding, and the runner channel was smooth. However, the central portion between the two inner gates contained excess liquid metal. I therefore removed the central part of the runner to reduce runner weight and improve process yield. As the inner-gate weld face increased, the runner weld face also increased, but the runner thickness remained unchanged.
To improve pattern-tree strength and shell strength, I added a thread-like interlocking structure at the sprue cup and its connection to the runner. I connected the sprue cup and runner with a web. The optimized tree reduced the bottom gap and made the shell more robust. These details are easy to overlook in lost foam castings as well, where pattern assembly stiffness can influence dimensional accuracy and coating integrity.
Table 5. Optimization actions and expected effects.
| Optimization action | Reason | Expected effect |
|---|---|---|
| Increase inner-gate volume | Reduce heat loss | Longer feeding time |
| Increase weld-face area | Improve thermal mass | Eliminate flow blockage |
| Use angled weld face | Reduce gate height without sand-fill loss | Better shell filling and strength |
| Remove central runner section | Reduce excess metal | Higher process yield |
| Add interlocking web | Increase pattern-tree stiffness | Less handling damage |
Verification by Solidification Simulation
After optimization, I repeated the solidification simulation. The optimized inner-gate weld-face area was 2850 mm². At this condition, no flow blockage occurred at the inner gate, and the shrinkage tendency disappeared. The hot spot remained connected to the liquid feeding path until it solidified. The optimized gating dimensions are therefore the basis for tooling and trial production.
The optimization can be expressed as a simple feeding criterion:
$$t_{gate} > t_{hot spot}$$
where tgate is the local solidification time of the inner gate, and thot spot is the local solidification time of the casting hot spot. When this inequality is satisfied, the gate remains open while the hot spot is still liquid. In my initial design, the inequality was reversed; after increasing thermal mass at the gate, the inequality was satisfied. I use the same criterion when I review lost foam castings, because the gate must remain permeable to feeding until the last liquid region freezes.
Table 6. Initial versus optimized simulation.
| Feature | Initial design | Optimized design |
|---|---|---|
| Inner-gate weld-face area | 1950 mm² | 2850 mm² |
| Flow blockage | Present | Absent |
| Shrinkage pore tendency | Small pore at gate | No significant porosity |
| Runner excess metal | Central section present | Central section removed |
| Knockout method | Flame cutting preferred | Hammer-off at neck |
Trial Production and Practical Results
I conducted a first trial of 100 pieces. No wax patterns dropped during shell building. After pouring, I hammered the gate tree, and the casting separated easily at the inner-gate neck. The fracture surface was smooth and had no obvious sharp corners. I then inspected the rough castings by X-ray nondestructive testing. The hot-spot regions showed no internal defects. A few local corners showed slight shrinkage porosity, which matched the simulation result. I machined 95 pieces, and the high-precision holes had no shrinkage porosity or shrinkage cavities.
The trial results confirmed three points. First, the calculated gating dimensions required simulation validation because the theoretical calculation alone can deviate from actual local cooling. Second, increasing the inner-gate volume and weld-face area can compensate for a small neck section by reducing heat loss. Third, a controlled neck with a root allowance can make hammer-off removal practical for low-carbon, high-toughness steel. I also apply these lessons to lost foam castings, where knockout and cleaning costs often determine whether a process is economically competitive.
Table 7. Production comparison with a similar casting.
| Metric | Similar casting | This product | Improvement |
|---|---|---|---|
| Knockout and cleaning output per shift (pieces) | 260 | 600 | 130.77% |
| Process yield (%) | 55 | 68 | 23.64% |
Discussion
The forklift head is a thin-wall, side-enclosed structure with welding requirements. The low-carbon alloy steel gives the required toughness and weldability, but it also makes mechanical gate removal difficult. Flame cutting solves the removal problem but adds cost and reduces efficiency. My optimized design converts the removal method from flame cutting to hammering. The key is not simply reducing the neck section; the key is to control the thermal balance of the gate so that the neck is weak enough to fracture but the feeding channel remains open long enough to feed the hot spot.
The equivalent hot-spot method gave me a starting point:
$$D_g = k D_c$$
but the simulation showed that the local cooling condition dominated the result. When the gate solidified before the hot spot, the theoretical feeding path was lost. Increasing the gate volume and weld-face area delayed gate solidification and restored feeding. The neck angle and root allowance then allowed controlled fracture. I regard this as a coupled design problem involving feeding, solidification, and knockout. It is similar to the trade-offs I examine in lost foam castings, where the gating system must feed the part while also permitting clean separation.
I also improved the runner tree by removing the central excess runner section. This reduced metal weight and increased process yield from 55% to 68%. The interlocking web and thread-like connection improved pattern-tree stiffness, which reduced handling damage and shell cracking. These improvements are not limited to investment casting. In lost foam castings, pattern assembly stiffness and gate design can similarly influence dimensional accuracy, coating quality, and final cleaning cost.
Conclusions
I drew the following conclusions from the optimization and trial production:
(1) The theoretical calculation of a gating system can deviate from actual behavior because of local casting structure and cooling conditions. Simulation-assisted verification quickly identifies defect distribution and root causes, clarifies the optimization direction, and improves process reliability. This is true for investment casting and for lost foam castings.
(2) When the inner-gate neck section is too small and the feeding channel is insufficient in a low-carbon steel investment casting, increasing the inner-gate volume and weld-face area can reduce gate heat loss and strengthen feeding capacity. The optimized weld-face area of 2850 mm² eliminated flow blockage and shrinkage tendency.
(3) For low-carbon, high-toughness steel, the inner-gate structure can be designed according to whether the feeding channel remains open. A smaller neck size can then be used to achieve hammer-off removal. The 75° neck angle and 1.5 mm root allowance produced a clean fracture surface without missing metal on the casting body.
(4) The optimized process raised knockout and cleaning output from 260 pieces per shift to 600 pieces per shift, an increase of 130.77%. Process yield increased from 55% to 68%, an increase of 23.64%. The production cost decreased, and production efficiency improved. I continue to apply the same feeding and knockout principles when I evaluate lost foam castings and other low-carbon steel casting routes.