In this work I set out to establish a complete and reproducible sand casting route for a guide connector, a load-bearing and connecting component that acts as the crank-arm support inside a transmission system. The part must carry complex, alternating loads, restrain deformation and damage of the crank arm, and simultaneously provide a stable locating datum for the surrounding components. These functional duties translate directly into severe requirements on structural integrity and quality consistency, and they are the reason I treated the sand casting design as an integrated problem of gating, feeding, solidification control and numerical verification rather than as a sequence of independent decisions.
Precision casting routes for similar guide and support components have been reported, where the die head was optimized and the mould structure simplified in order to raise the product yield. For the specific geometry studied here, however, no systematic investigation of a sand casting route existed. I therefore approached the problem from the standpoint of classical sand casting process design, combined it with a full three-dimensional numerical simulation performed in ProCAST, and used the simulation results as the feedback signal for successive iterations of the feeding system. My objective was to obtain a sand casting process that produces a sound casting free of internal shrinkage cavities and porosity, while remaining economical for a small batch of one casting per mould.
1. Casting Structure and Process Analysis
The guide connector is a compact but geometrically complex part. Its overall envelope is roughly 383.8 mm in length, 271 mm in width and 399.4 mm in height, and its net mass is 58.695 kg. Although the mass places it in the category of small castings, its wall thickness distribution is highly non-uniform, ranging from a minimum of 15 mm to a maximum of 70 mm. This contrast is the dominant technological difficulty: the thick sections behave as thermal centres that remain liquid long after the thin walls have solidified, so any sand casting route that does not deliberately control the solidification sequence will inevitably generate shrinkage cavities and shrinkage porosity in the heavy regions.
All transitions and junctions between walls are generously filleted, which satisfies good sand casting practice and reduces stress concentration in the solidified shell. Because the total mass is moderate, the mechanical strength demanded of the mould cavity is relatively low, so I selected an ester-cured alkaline phenolic resin sand for both moulding and core making. This binder system offers high dimensional accuracy, good collapsibility and excellent surface finish after the mould and cores are coated with an alcohol-based coating. The component material is ZG20CrNiMo, a low-alloy cast steel whose casting properties are favourable, with adequate fluidity and a comparatively narrow solidification interval that makes directional solidification achievable when proper feeding is applied.
| Parameter | Value |
|---|---|
| Envelope size (mm) | 383.8 × 271 × 399.4 |
| Net mass (kg) | 58.695 |
| Material | ZG20CrNiMo cast steel |
| Minimum wall thickness (mm) | 15 |
| Maximum wall thickness (mm) | 70 |
| Production mode | Small batch, one casting per mould |
| Moulding method | Manual, ester-cured alkaline phenolic resin sand |
| Moulding principle | Two-part mould, middle-plane parting |
| Number of cores | 2 |
The ratio between the mass of liquid steel poured and the geometric envelope volume gives a relative density indicator that is used, in conventional sand casting practice, to estimate the pouring time and the filling behaviour:
$$\rho_r=\frac{G_L}{V_c}$$
with the poured mass of liquid steel taken as $G_L = 86.32$ kg and the envelope volume calculated as $V_c = 30\,735.7\ \mathrm{cm^3}$, which yields:
$$\rho_r=\frac{86\,320\ \mathrm{g}}{30\,735.7\ \mathrm{cm^3}}=2.81\ \mathrm{g/cm^3}$$
A relative density of 2.81 g/cm³ corresponds to a compact, relatively dense casting geometry, which means that the mould cavity is filled quickly and that the thermal load on the mould is concentrated. This value is a useful indicator for the sand casting designer because it directly influences the choice of the empirical coefficient in the pouring-time equation and the pressure head used for the choke-area calculation.
2. Pouring Position and Parting Plane
The pouring position defines the orientation of the casting inside the mould during pouring, and its selection must simultaneously satisfy several constraints: the casting must fill completely, the quality requirements must be met, and the operations of moulding, core setting and fettling must remain convenient. After consulting standard sand casting design data and comparing several candidate orientations, I adopted a position in which the two large concave regions of the component are located on the upper and lower sides of the mould cavity respectively.
This orientation offers several advantages. It allows the cores to be located and supported securely, so that no cantilevered core is required and the core prints can be designed with adequate bearing area. Gas evolved from the cores and from the mould surface can escape freely through the upper mould, which reduces the risk of blowholes. It also permits the feeding paths to be arranged so that the last-solidifying regions remain in contact with the risers.
The parting plane is placed at the middle plane of the casting, indicated by the plane separating the upper and lower mould halves. With this choice the casting is distributed symmetrically between the cope and the drag, which balances the centre of gravity, reduces the tendency for core displacement or mould-wall deformation under the ferrostatic pressure, and provides a more uniform temperature field during cooling. In addition, the two recessed regions can be formed by sand cores rather than by complicated moulding operations, which reduces core complexity, lowers tooling cost, and shortens the cleaning and finishing work after shakeout.
| Design aspect | Chosen solution | Expected benefit |
|---|---|---|
| Orientation during pouring | Concave features on upper and lower faces | Stable core support, no cantilever core |
| Gas removal | Venting through the cope | Reduced blowhole and gas-defect risk |
| Parting plane | Middle plane of the casting | Balanced centre of gravity, symmetric cooling |
| Moulding complexity | Two-part mould with two cores | Fewer cores, simpler core boxes |
| Feeding access | Risers placed on the thermal centres | Directional solidification toward risers |
3. Core Design
Core design begins with an analysis of which internal or re-entrant features cannot be formed directly by the pattern. For the guide connector I identified two such features: an internal concave pocket and an external recessed region. Each was therefore assigned a dedicated sand core. Core prints were proportioned so that each core is supported on at least two surfaces, ensuring that the core cannot float or shift when the liquid metal is introduced, and so that the core gases can be led out through the core prints into the mould cavity and thence to the atmosphere.
Because the moulding sand and the core sand are both ester-cured alkaline phenolic resin bonded, the cores possess sufficient green and cured strength for manual handling while still collapsing readily during shakeout. The alcohol-based coating applied to both the mould cavity and the cores improves the surface finish of the sand casting and provides a barrier that limits sand burn-on and metal penetration, particularly in the heavy sections where the local thermal load is highest.
| Core | Feature formed | Support condition | Venting route |
|---|---|---|---|
| Core 1 | Internal concave pocket | Two core prints, fully seated | Through core print to cope vent |
| Core 2 | External recessed region | Two core prints, fully seated | Through core print to drag vent |
4. Casting Layout and Sand Allowance
The position and number of castings inside the moulding box are governed by the sand allowance, that is the distance between the casting surface and the walls of the moulding box, which must be large enough to prevent mould-wall collapse and to allow adequate venting, but not so large that the box volume and the sand consumption become excessive. Using standard data for cast steel, and adopting the one-casting-per-box arrangement, I selected the allowances listed in Table 4. The distance from the casting to the runner was kept greater than 70 mm in order to avoid local overheating of the mould wall, and the distance from the sprue to the box edge was kept greater than 50 mm to ensure a sound sprue wall.
| Direction / distance | Value (mm) |
|---|---|
| Vertical sand allowance | 90 |
| Horizontal sand allowance | 50 |
| Casting to runner | > 70 |
| Sprue to box edge | > 50 |
| Simulation box sand allowance | 50 |
5. Gating System Design
Gating systems are classified by the ratio of the cross-sectional areas of their elements into open, closed, semi-closed and closed–open types. For this sand casting I adopted a semi-closed, bottom-pouring system. Bottom pouring keeps the filling process smooth, encourages slag and gas to float to the top of the mould cavity, and avoids the violent jetting that would otherwise entrain air and oxide films into the metal stream. The area ratio between the total ingate, runner and sprue cross-sections was fixed as:
$$\sum A_{\text{ingate}} : \sum A_{\text{runner}} : \sum A_{\text{sprue}} = 1.0 : 0.8 : 1.1$$
The sprue is therefore the largest element and the runner the smallest, which makes the runner the controlling or choke section and produces a slightly pressurized system that suppresses aspiration of air through the sprue–runner junction.
5.1 Pouring Time
The pouring time for cast steel was estimated with the empirical relation:
$$t = C\sqrt{G_L}$$
in which $C$ is a coefficient determined by the relative density $\rho_r = G_L / V_c$ of the casting. With $\rho_r = 2.81\ \mathrm{g/cm^3}$ the standard tables give $C = 1$, and the calculation returns a pouring time of approximately 9.7 s. This relatively short pouring time is consistent with the compact geometry and the need to fill the cavity before the thin walls begin to freeze.
5.2 Choke Area
The choke cross-sectional area was calculated with the choke-area design method:
$$A_{\text{choke}}=\frac{G_L}{\rho\,\mu\,t\sqrt{2gH_p}}$$
where $G_L = 86.32$ kg is the mass of metal passing through the choke, $\rho = 7.8\ \mathrm{g/cm^3}$ is the density of the liquid cast steel, $\mu = 0.6$ is the empirical flow-loss coefficient, $t = 9.7$ s is the pouring time, $g = 981\ \mathrm{cm/s^2}$ is the gravitational acceleration, and $H_p$ is the average static pressure head. The average pressure head was obtained from:
$$H_p = H – \frac{P^2}{2H_C}$$
with $H = 238.25$ mm as the height from the metal surface in the pouring cup to the ingate level, $H_C = 279.5$ mm as the cavity height in the pouring orientation, and $P = 148.25$ mm as the height of the cavity above the ingate level. Substituting these values gives:
$$H_p = 238.25 – \frac{148.25^2}{2 \times 279.5} \approx 198\ \mathrm{mm}$$
Introducing $H_p$ into the choke-area equation yields a total choke area of approximately 10.2 cm². Applying the prescribed area ratio, the remaining elements follow directly, as summarized in Table 5.
| Element | Area ratio | Total area (cm²) | Configuration |
|---|---|---|---|
| Ingates | 1.0 | 12.75 | Five circular ingates, each 2.55 cm², Ø18 mm |
| Runner (choke) | 0.8 | 10.20 | Annular runner beneath the casting |
| Sprue | 1.1 | 14.03 | Circular, Ø42 mm, ceramic tube |
The circular sprue is embedded as a ceramic tube so that the mould sand is not eroded by the incoming stream, and the runner is arranged as an annulus below the casting from which five ingates rise into the lower face of the cavity. This arrangement distributes the incoming metal over a wide area of the casting, reduces the local temperature rise at any single entry point, and provides a symmetrical filling pattern that helps to avoid core displacement.
5.3 Numerical Simulation Setup
The three-dimensional solid model was constructed and transferred to the simulation environment, where the mesh was generated with a non-uniform density to balance accuracy against computation cost. A fine mesh length of 2 was assigned to the casting, 5 to the gating system, and 20 to the moulding box. Boundary and initial conditions were then prescribed together with the thermophysical data, as listed in Table 6.
| Item | Setting |
|---|---|
| Pouring temperature | 1 550 °C |
| Pouring time | 9.7 s |
| Casting material | ZG20CrNiMo |
| Mould material | Resin self-hardening sand |
| Casting / mould heat transfer coefficient | 5 000 W/(m²·K) |
| Casting / chill heat transfer coefficient | 2 000 W/(m²·K) |
| Chill / mould heat transfer coefficient | 900 W/(m²·K) |
| Cooling condition | Air cooling |
| Casting mesh length | 2 |
| Gating mesh length | 5 |
| Moulding box mesh length | 20 |
5.4 Flow Field Analysis of the Bare Casting
In the first simulation I studied the bare casting, that is the casting with the gating system only and no feeding or chilling elements. During filling, the metal entered the cavity through the five ingates and initially advanced in a fountain-like pattern, splashing along the inner walls. Such splashing is undesirable in any sand casting because it can trap slag and oxide films and can entrain gas into the stream. As filling progressed, the bottom of the casting filled first, but the free surface continued to advance as a wave rather than as a flat front. Once roughly five to six seconds of pouring had elapsed, the filling process became noticeably smoother and the remainder of the cavity filled in a more orderly manner.
| Time (s) | Observed behaviour | Process implication |
|---|---|---|
| 1.07 | Metal enters through ingates, fountain-like jet against inner wall | Risk of slag entrapment and gas entrainment |
| 2.28 | Bottom of casting nearly filled, free surface wavy | Moderate surface disturbance |
| 5.69 | Filling becomes smooth and progressive | Stable cavity filling |
| 8.51 | Upper regions filling steadily | No premature freezing detected |
| 9.59 | Cavity essentially complete | Consistent with the 9.7 s design value |
5.5 Solidification Analysis of the Bare Casting
The solidification field of the bare casting showed that heat was concentrated in the upper thick section. Metal solidified inward from the surrounding thinner walls toward this central heavy region, and at approximately 106 s two isolated liquid pools remained, disconnected from any feed path. Once such an isolated liquid region forms, it cannot be fed by any riser, and it inevitably contracts into a shrinkage cavity or a dispersed shrinkage porosity. In the later stages, at around 646 s, defects also appeared on the upper planar region of the casting, indicating that a surface depression or sink could develop there as well.
The defect prediction confirmed the diagnosis. A large depressed region appeared on the upper plane of the casting, and severe shrinkage porosity and shrinkage cavities developed in the region of the largest thermal node. The bare casting therefore failed the soundness requirement, which is exactly the situation that a properly designed sand casting feeding system must prevent.
| Time (s) | Solidification state | Predicted defect |
|---|---|---|
| 11.33 | Solid shell forming, heavy section still liquid | None |
| 46.52 | Inward solidification toward the heavy upper section | None |
| 76.52 | Thermal centre persisting | Potential shrinkage site |
| 106.52 | Two isolated liquid pools | Shrinkage cavity and porosity |
| 176.52 | Isolated pools contracting | Shrinkage cavity and porosity |
| 646.52 | Upper planar region solidifying last | Surface depression |
6. Feeding System Design
Because the bare casting exhibited isolated liquid regions and shrinkage defects, I proceeded to design a feeding system. The classical criterion for a sound sand casting is that the modulus of the feeder must exceed the modulus of the region it feeds:
$$M_{\text{riser}} > M_{\text{casting section}} = \frac{V}{A}$$
where $V$ is the volume and $A$ is the heat-dissipating surface area of the section under consideration. In addition to the modulus criterion, the feeder must remain liquid longer than the casting section, and a continuous feed channel must exist between the feeder and the last-solidifying liquid.
6.1 Riser Design
Guided by the simulated location of the thermal nodes, I first placed three risers: two top open risers and one waisted blind riser. The two open risers were positioned directly above the principal thermal centres so that gravity and the metallostatic head assist feeding, while the blind riser was placed on a side thermal node. The dimensions were derived from the shrinkage-modulus method for cast steel risers and checked against standard empirical proportions. The nominal sizes are collected in Table 9.
| Riser | Type | Nominal dimension (mm) | Neck dimension (mm) | Outcome |
|---|---|---|---|---|
| No. 1 | Top open | Ø90, height 186, 45° taper | Ø70 | Effective |
| No. 2 | Top open | Ø147, height 186, 6° taper, R2 | Ø89 | Effective |
| No. 3 | Waisted blind | 50 × 75 section | — | Removed |
With the risers added and all other parameters unchanged, I repeated the simulation. The result was informative rather than satisfactory. The recessed region on the right-hand side of the casting solidified prematurely, leaving two voids, and at 138 s the recessed region on the rear face also solidified prematurely. The waisted blind riser positioned at that location failed either to feed the region or to modify the local solidification sequence, so it contributed nothing while adding metal consumption and fettling work. I therefore removed it entirely.
The revised simulations still showed shrinkage porosity and shrinkage cavities in the lower part of the casting. Comparing this defect location with the solidification field showed that the defective zone coincided exactly with an isolated liquid region, that is, a region that had lost hydraulic connection with every feeder. Adding more riser metal to such a location is ineffective, because the problem is not a shortage of feed metal but the absence of a feed path. The correct remedy is local cooling, which re-establishes the temperature gradient and reconnects the liquid region to an existing feeder.
6.2 Chill Design
The lower part of the casting is thick, whereas the central wall is relatively thin, so the heavy bottom region naturally solidifies later than its surroundings and becomes isolated from the risers. To extract heat from this region and reverse the local gradient, I placed chills on the lowest part of the casting. The chill thickness was taken as one half of the local casting wall thickness, in accordance with standard sand casting practice, and four chills were distributed over the heavy bottom area.
| Chill | Geometry | Dimension (mm) | Quantity |
|---|---|---|---|
| Type 1 | Cylindrical | Ø35 × 40 | As required by layout |
| Type 2 | Rectangular block | 35 × 66.5 | As required by layout |
| Total | — | — | 4 |
6.3 Verification of the Final Design
With both the two top open risers and the four chills in place, and with all other parameters unchanged, I ran the simulation once more. The solidification sequence now behaved as intended. The solidification time of each riser exceeded that of the casting section it fed, satisfying the modulus requirement. Throughout solidification a continuous feed channel was maintained between the risers and the casting body, and liquid metal from the risers was able to compensate the contraction of the casting continuously until the casting itself had fully solidified. No isolated liquid region remained anywhere in the casting. The predicted defect field was clean, with no shrinkage cavity and no shrinkage porosity in the interior of the casting.
| Condition | Isolated liquid region | Shrinkage cavity | Shrinkage porosity | Surface depression |
|---|---|---|---|---|
| Bare casting, gating only | Present (two pools) | Severe | Severe | Large area |
| Two open risers, no chill | Partially present | Moderate | Moderate | Reduced |
| Two open risers plus four chills | Eliminated | None | None | None |
An additional benefit of the chill design deserves emphasis. Chills do not merely accelerate local cooling; they reshape the isotherm pattern so that the thermal gradient points continuously from the thin sections toward the risers. In this sand casting the four chills converted an unfavourable gradient into a favourable one, and the risers could then operate under the modulus criterion that classical sand casting design assumes.
7. Process Yield and Discussion
The final process yield was evaluated from the ratio of the net casting mass to the total mass of metal poured:
$$\eta = \frac{G_{\text{casting}}}{G_{\text{poured}}} \times 100\%$$
With a net casting mass of 58.695 kg and a total poured mass of approximately 94 kg, the yield is about 62.4 %. This value is reasonable for a small-batch, manually moulded sand casting of a component with strongly non-uniform wall thickness, where the risers must necessarily be large in order to satisfy the modulus requirement at the thermal nodes.
| Item | Value |
|---|---|
| Net casting mass (kg) | 58.695 |
| Riser and gating mass (kg) | Approximately 35.3 |
| Total poured mass (kg) | Approximately 94.0 |
| Process yield (%) | 62.4 |
The complete design can be summarized as a sequence of coupled decisions, each of which was validated by simulation before the next was introduced. This stepwise strategy proved far more efficient than attempting to optimize all elements simultaneously, because each simulation isolated a single physical cause and its corresponding remedy.
| Design element | Selected solution |
|---|---|
| Production mode | Small batch, one casting per mould |
| Moulding method | Manual moulding, ester-cured alkaline phenolic resin sand |
| Moulding principle | Two-part mould, middle-plane parting |
| Core count | 2 |
| Gating type | Semi-closed, bottom pouring |
| Area ratio | 1.0 : 0.8 : 1.1 (ingate : runner : sprue) |
| Number of ingates | 5, each Ø18 mm |
| Sprue diameter | Ø42 mm, ceramic tube |
| Pouring temperature | 1 550 °C |
| Pouring time | 9.7 s |
| Average pressure head | 198 mm |
| Feeding elements | 2 top open risers |
| Chilling elements | 4 chills |
| Process yield | 62.4 % |
8. Limitations and Further Work
The process presented here has a clear scope of validity. It is built around manual moulding and a one-casting-per-mould arrangement, which is well suited to small-batch production but whose efficiency and economy in large-volume production have yet to be demonstrated. Increasing the number of castings per mould would raise the metal utilization and reduce the moulding cost per part, but it would also change the thermal environment, because neighbouring cavities interact through the shared mould sand and can shift the solidification sequence. Any move to a multi-cavity layout would therefore require a fresh simulation study to re-establish the feeding design.
A second direction for further work is the substitution of manual moulding by machine moulding. Machine moulding would improve dimensional repeatability and surface quality and would shorten the moulding cycle, but it imposes different constraints on pattern draft, core print design and sand properties. These constraints would need to be reconciled with the gating and feeding solution established here, since the chill and riser positions are already tightly coupled to the thermal field.
A third consideration concerns the pouring temperature and pouring time. The combination of 1 550 °C and 9.7 s was shown by simulation to fill the cavity completely without premature freezing, but pouring temperature also affects shrinkage behaviour and the susceptibility to hot tearing. A parametric study covering a modest range of pouring temperatures would clarify how much tolerance the process possesses in practice, which is valuable information for a sand casting that must be produced reproducibly.
9. Conclusions
The following conclusions can be drawn from the design work described above.
First, a complete sand casting process route was established for the guide connector. The route uses manual moulding with an ester-cured alkaline phenolic resin sand for both the mould and the cores, a small-batch one-casting-per-mould layout, a two-part mould with a middle-plane parting plane, and two sand cores to form the internal and external re-entrant features.
Second, a semi-closed bottom-pouring gating system was designed, with a cross-sectional area ratio of ingate to runner to sprue of 1.0 : 0.8 : 1.1. The choke area was calculated as 10.2 cm², the total ingate area as 12.75 cm² distributed over five circular ingates, and the sprue area as 14.03 cm². The pouring temperature was set at 1 550 °C and the pouring time at 9.7 s.
Third, the solidification simulation of the bare casting revealed two isolated liquid pools and a large depressed region on the upper plane, confirming that gating design alone is insufficient for this geometry. Through feeding system design, two top open risers combined with four chills were introduced, and the sequential solidification of the casting was successfully achieved. The verified process yield is 62.4 %.
Fourth, the results demonstrate the practical value of coupling classical sand casting design calculations with numerical simulation. The modulus criterion identified where feeders were required, the solidification simulation explained why an additional blind riser failed to work, and the defect prediction justified the introduction of chills at the correct locations. This combination allowed the sand casting to be brought to a sound condition without repeated trial pours, which is a significant advantage for small-batch production of a component with strongly non-uniform wall thickness.