In my work on the production of large-diameter rotatory bodies for centrifugal separation equipment, I have repeatedly observed that the dimensional stability of investment casting components is governed not by a single process step but by the coupled interaction of pattern geometry, shell rigidity, filling behaviour, solidification sequence and post-pouring cooling restraint. The component examined here is a top disc of conical geometry produced by investment casting in CF8M duplex-austenitic stainless steel. Its envelope dimensions are Φ430 mm × Φ250 mm × 188 mm, its wall thickness varies from 5 mm to 40 mm, and its single-piece mass is 29.38 kg. Because the wall thickness decreases progressively from the bottom plate towards the thin-walled mouth, the component exhibits a strongly non-uniform thermal field during cooling, and this non-uniformity translates directly into taper opening, plate dishing and out-of-roundness of the mouth. In the present study I combined numerical simulation with full-process parameter control in order to eliminate the need for any post-casting straightening operation, which for an axisymmetric thin-walled rotatory body is expensive, slow and difficult to implement uniformly over 360°.
1. Introduction and Engineering Context
The filling and solidification of liquid metal inside a ceramic shell is accompanied by complex physical and chemical changes. When the process scheme or the structural design is unfavourable, the resulting investment casting is prone to shrinkage cavities, porosity, gas holes, misruns and dimensional distortion. For the family of top disc products considered here, the metallurgical quality requirements, the dynamic balancing accuracy and the dimensional stability requirements have all been tightened in recent years. The combination of a 40 mm maximum wall thickness and a 5 mm minimum wall thickness in a single Φ430 mm envelope creates a situation in which the deformation potential is intrinsically high and the mouth roundness is difficult to hold.
Conventional approaches to this problem fall into two broad groups. The first group relies on numerical simulation to predict stress and distortion trends and then adjusts the process accordingly. The second group relies on local process optimisation, such as split-die pressing, combined mould assembly, or dedicated straightening fixtures. Straightening fixtures are effective for many compact geometries, but for a thin-walled conical rotatory body the fixture must apply a perfectly uniform 360° load, which makes the tooling expensive, slow to manufacture, physically large and difficult to assemble and operate on the shop floor. Consequently I decided to attack the problem at source: predict the distortion with JSCAST, compensate the die geometry by the predicted amount in the opposite direction, and then hold every downstream process step within a narrow window so that the compensation survives the whole manufacturing chain.
The industrial value of this approach is significant. The investment casting route for this product family covers eleven top disc variants, and any reduction in the number of process steps—particularly the elimination of a straightening operation—reduces both the development cost of new variants and the recurring cost of series production. The strategy described in the following sections was validated on a representative member of the family and then transferred to the remaining variants.
2. Casting Description and Process Analysis
The component is a conical top disc cast in CF8M stainless steel. I first established the material specification, the geometric envelope and the acceptance criteria, because the allowable dimensional deviation determines how much compensation must be built into the die and how tightly the downstream process windows must be controlled.
| Element | Specification (wt.%) | Function in the alloy |
|---|---|---|
| C | ≤ 0.08 | Controls carbide precipitation and corrosion resistance |
| Si | ≤ 2.0 | Deoxidation and castability |
| Mn | ≤ 1.5 | Deoxidation and sulphur control |
| P | ≤ 0.04 | Impurity control |
| S | ≤ 0.04 | Impurity control |
| Cr | 18.0–21.0 | Passive film formation |
| Ni | 9.0–12.0 | Austenite stabilisation |
| Mo | 2.0–3.0 | Pitting and crevice corrosion resistance |
| Parameter | Value | Remark |
|---|---|---|
| Material | CF8M | Duplex-austenitic cast stainless steel |
| Envelope diameter | Φ430 mm | Mouth region |
| Overall height | 188 mm | Bottom plate to mouth |
| Reference cone angle | 30° ± 0.5° | Critical non-machined feature |
| Minimum wall thickness | 5 mm | Mouth region |
| Maximum wall thickness | 40 mm | Bottom plate region |
| Single-piece mass | 29.38 kg | As-cast |
| Dimensional tolerance grade | DCTG 7 | Per GB/T 42124.3—2025 |
| Internal defect requirement | No porosity, shrinkage or slag inclusion | 100% radiography |
| Radiographic acceptance | ASTM E446 Level 2 | Full casting volume |
| Residual unbalance after trimming | < 2 g | Dynamic balance test after machining |
The gating system is a top-gating arrangement in which liquid metal enters the mould cavity from the top and flows downward under gravity. Because the section is thick at the top and thin at the bottom, the top-gating configuration provides an adequate feeding path to the thin lower wall, and during cooling the thin lower wall solidifies first. This establishes a directional solidification sequence from the bottom of the casting upward towards the gate, which is the desired sequence for an investment casting of this type.
Nevertheless, top gating also means that the incoming stream impinges directly on the cavity floor, which promotes turbulence and gas entrainment. The combination of a thick, slowly cooling bottom plate and a thin, rapidly cooling mouth generates a pronounced differential contraction that manifests as taper enlargement and plate dishing. These two effects—gas entrapment at the plate and differential thermal contraction—formed the two technical targets of my optimisation work.
3. Numerical Simulation Methodology
I built the three-dimensional model in SolidWorks and exported the casting and mould as STL files, which were then imported into JSCAST. The material database entries for CF8M, zircon sand and mullite sand were imported first, followed by mould definition, mesh generation and the specification of the flow, solidification and deformation calculation conditions. The key boundary and initial conditions are summarised below.
| Simulation parameter | Setting |
|---|---|
| Solidification and flow solver | JSCAST |
| Casting alloy | CF8M |
| Face coat / backup sand | Zircon sand / mullite sand |
| Pouring temperature | 1 580 °C |
| Shell thickness | 12 mm |
| Shell outer surface condition | Air cooling |
| Filling velocity (velocity calculator) | 250 mm/s |
| Final temperature for stress calculation | 20 °C |
| Critical solid fraction for shrinkage criterion | 0.70 |
3.1 Solidification Time and Feeding Criterion
The solidification time of a section can be estimated from the classic Chvorinov relationship, which I used as a first-order check on the gate and riser design before running the full transient calculation:
$$t_{s} = B\left(\frac{V}{A}\right)^{n}$$
where $t_{s}$ is the local solidification time, $V$ is the volume of the section, $A$ is its heat-transfer surface area, $B$ is a mould constant that depends on the mould material and the initial mould temperature, and the exponent $n$ is close to 2 for a simple chill-free shell. The geometric modulus of the section is therefore:
$$M = \frac{V}{A}$$
The shrinkage criterion used by the solver is based on the local solid fraction, defined as:
$$f_{s} = \frac{V_{s}}{V_{s} + V_{l}}$$
where $V_{s}$ and $V_{l}$ are the solid and liquid fractions of the local volume element. Regions that reach $f_{s} = 0.70$ while still being surrounded by liquid constitute isolated hot spots and are flagged as shrinkage-risk zones. Regions that remain connected to a liquid reservoir throughout solidification are fed and are not flagged.
3.2 Deformation Calculation
The distortion calculation solves the thermo-mechanical problem in which the total strain is decomposed into an elastic, a plastic and a thermal contribution. The thermal strain increment at each step is:
$$\Delta\varepsilon_{th} = \alpha\,\Delta T$$
where $\alpha$ is the temperature-dependent coefficient of thermal expansion and $\Delta T$ is the local temperature change. When the thermal contraction is restrained by neighbouring material, a thermal stress develops according to:
$$\sigma_{th} = \frac{E\,\alpha\,\Delta T}{1 – \nu}$$
where $E$ is Young’s modulus and $\nu$ is Poisson’s ratio. The resultant displacement at any node is evaluated as the vector sum of the three Cartesian components:
$$d = \sqrt{d_{x}^{2} + d_{y}^{2} + d_{z}^{2}}$$
Because the component is a body of revolution about the vertical axis $Y$, the $X$ and $Z$ components are nominally identical; I therefore refer to them collectively as the radial deformation, while the $Y$ component is referred to as the axial deformation. Radial and axial deformation differences between the top and bottom of the component were extracted as the two principal indicators of taper change:
$$\Delta d_{r} = d_{r,top} – d_{r,bottom}$$
$$\Delta d_{a} = d_{a,top} – d_{a,bottom}$$
4. Filling Simulation Results
The filling simulation was run with the pouring temperature, shell thickness and filling velocity listed above, and the progress of the metal front was recorded at several time steps. The colour scale in the gas-defect display ranges from blue, representing liquid metal, to red, representing a high-risk gas defect, with darker shades indicating a higher risk. The essential observations were as follows.
During the first part of the fill, while the lower half of the component is being filled, no significant gas defect is predicted. As the cavity approaches completion, however, a large red area appears on the bottom plate, indicating that gas has been entrapped in this region and that a gas hole is likely to form there. Once the cavity is essentially full, the gas defect zone concentrates around the gate on the bottom plate. The mechanism is that the permeability of the ceramic shell is finite, and the gas carried into the cavity by the liquid stream cannot escape quickly enough through the shell wall, so it is trapped against the relatively cold plate surface. In the final stage, when the casting, the gate and part of the runner are filled, only isolated red spots remain on the bottom plate.
The conclusion I drew from the filling analysis is that the bottom plate surface is the principal gas-hole risk zone in this investment casting. This prediction agreed with what I had observed in trial pours, in which gas holes on the plate surface made post-processing difficult and increased cost. Reducing turbulence and slowing the metal front in the plate region therefore became a mandatory requirement of the optimised process.
5. Solidification Simulation Results
The gate dimensions were calculated by the modulus method and set to a diameter of 80 mm with a height of 50 mm. The solidification simulation was then evaluated against the critical solid fraction criterion defined in Section 3.1.
The predicted shrinkage map shows a single red region located in the pouring cup at the top of the sprue. This is a concentrated hot spot, that is, a region in which solidification lags locally and in which an isolated liquid pool is likely to form. In the present case, however, the casting body itself exhibits only a small variation in solidification time and a shallow temperature gradient, so no isolated hot spot and no closed liquid pool are formed inside the component. As a result, no shrinkage cavity or shrinkage porosity risk exists in the casting body. The hot spot in the pouring cup is consistent with the intended feeding design, acting as a reservoir that supplies liquid metal to compensate the contraction of the casting during solidification.
I therefore concluded that the 80 mm × 50 mm gate satisfies the directional solidification requirement and does not need to be enlarged. Increasing the gate size would add metal to be removed and would increase the thermal load on the shell without any benefit to internal soundness.
6. Deformation Simulation Results
The deformation simulation is the core of this study, because dimensional deviation rather than internal porosity was the dominant rejection cause for this component. The predicted displacement field, examined both in the as-calculated state and with the deformation magnified ten times for visibility, revealed a consistent pattern: the component opens at the mouth, dishes in the middle and exhibits the smallest displacement at the bottom plate.
| Deformation indicator | Bottom plate region | Mouth region | Difference |
|---|---|---|---|
| Resultant displacement | ≈ 1.2 mm | ≈ 2.5 mm | ≈ 1.3 mm |
| Radial (X, Z) displacement | ≈ 1.2 mm | ≈ 0.5 mm | ≈ 0.7 mm |
| Axial (Y) displacement | ≈ 0.5 mm | ≈ 2.0 mm | ≈ 1.5 mm |
| Axial variation within the plate plane | ≈ 0.7 mm between the point beneath the gate and the farthest point | ≈ 0.7 mm | |
In the radial direction, the bottom plate exhibits the largest displacement, approximately 1.2 mm, while the region near the mouth exhibits a relatively small displacement of approximately 0.5 mm. The reason is that the plate is thick, so its volumetric contraction is large and its radial contraction is pronounced, whereas the thin mouth contracts much less. The radial difference of 0.7 mm between the top and the bottom of the component produces the observable opening of the included cone angle.
In the axial direction, the region near the mouth shows the largest displacement, approximately 2.0 mm, whereas the plate shows a displacement of approximately 0.5 mm, giving an axial top-to-bottom difference of 1.5 mm. The mechanism is that the thick plate contracts predominantly in the radial direction, while the thin mouth cools rapidly; under the dragging action of the contracting plate, the thin mouth tends to expand outward and to curl upward.
In addition, the plate itself is predicted to dish. Within the plane of the plate, the axial displacement directly beneath the gate differs from that at the point farthest from the gate by approximately 0.7 mm. This arises from the superposition of two effects: the large contraction of the gate region and the gravitational load of the liquid metal in the pouring cup and sprue. The combined result is the characteristic deformation mode of this product—mouth opening, central dishing and gradual reduction of the displacement towards the mid-height of the cone.
It is worth noting that the deformation mode is essentially a global, low-order mode rather than a local buckle. This is important because it means the distortion can be compensated by a global geometric correction of the die rather than by local stiffening, and it also means that the distortion is reproducible from part to part, which is a precondition for effective compensation.
7. Process Scheme Optimisation
On the basis of the simulation results and the trial production experience, I designed a process scheme that addresses the deformation of this specific structural family from the die model through the wax pattern, the shell and the pouring and cooling stages, with the objective of removing the straightening operation entirely. In parallel, the scheme addresses surface gas holes and internal defects.
7.1 Anti-Deformation Compensation of the Die Geometry
The deformation simulation indicated a radial mouth deformation of approximately 1.2 mm and a plate deformation of approximately 0.5 mm. Converting this radial offset into an angular change using the component height gives:
$$\Delta\theta_{cast} \approx \arctan\!\left(\frac{\Delta r}{H}\right)$$
The corresponding increase in the included cone angle is of the order of 1°. In addition, the wax pattern itself distorts during injection and cooling. Based on my accumulated experience with components of the same family, the wax pattern angle increases by 0.5° ± 0.2° relative to the die cavity. The total angular deviation to be compensated is therefore:
$$\Delta\theta_{total} = \Delta\theta_{cast} + \Delta\theta_{wax}$$
The die model angle is then set as:
$$\theta_{mold} = \theta_{nominal} – \Delta\theta_{total}$$
With $\theta_{nominal} = 30°$ and $\Delta\theta_{total} = 1.5°$, the die angle becomes 28.5°. This value was applied uniformly around the circumference, preserving the axisymmetric character of the component.
For the plate dishing, the simulation indicated an axial difference of 0.7 mm between the region directly beneath the gate and the farthest region, and the wax pattern distortion contributes a further 0.5 mm of dishing, giving a total as-cast dishing of approximately 1.2 mm. I therefore introduced a compensating slope on the plate plane, running from the centre outwards, expressed as a linear function of the radial coordinate:
$$z(r) = z_{c} + \frac{\delta_{s}}{R}\,r$$
where $z_{c}$ is the centre height, $R$ is the outer radius of the plate and $\delta_{s} = 1.2$ mm is the compensating rise at the periphery. In practice this is a 1.2 mm taper superimposed on the plate plane in the direction opposite to the predicted dishing.
The compensated model was compared with the original model in the SolidWorks assembly environment using a hidden-line-removed display mode, in which the compensated outer contour and the original outer contour can be superimposed and compared directly. The comparison confirms that the compensated model is radially contracted at the mouth by the full angular compensation, and that the plate carries the compensating slope.
It is important to state clearly which surfaces were compensated and which were not. The compensation was applied only to the non-machined surfaces that must be formed to final dimension directly from the casting. All other surfaces retain a generous machining allowance and are finished by machining after casting, so no pre-compensation is required for them; their deformation is absorbed by the allowance. The selective application of compensation is therefore a deliberate design decision rather than an omission.
7.2 Verification of the Compensated Model by Simulation
The compensated model was re-simulated under identical boundary conditions. The overall deformation magnitude predicted for the optimised geometry is smaller than that predicted for the original geometry, which confirms that the deformation behaviour of the conical component is closely related to the cone angle: the smaller the cone angle, the smaller the total deformation. The deformation trend, however, remains the same, and the magnitude of the displacement change at corresponding locations is essentially unchanged. This consistency is what makes the compensation predictable: because the deformation mechanism is unchanged, the deformation of the 28.5° die geometry can be expected to rebound to the target 30° after casting.
| Stage | Cone angle, original route | Flatness, original route | Cone angle, optimised route | Flatness, optimised route |
|---|---|---|---|---|
| Drawing requirement | 30° ± 0.5° | ≤ 0.5 mm | 30° ± 0.5° | ≤ 0.5 mm |
| Three-dimensional die model | 30° | 0 | 28.5° | 1.2 mm compensating slope |
| Wax pattern | 30.3°–30.7° | 0.3–0.5 mm | 28.8°–29.2° | 0.7–0.9 mm |
| As-cast component | 31.3°–31.7° | 1.0–1.2 mm | 29.8°–30.2° | 0–0.2 mm |
The predicted as-cast angle of 29.8°–30.2° and the predicted as-cast flatness of 0–0.2 mm both fall inside the drawing tolerance. On that basis the compensated model was released for die manufacture and the predictions were subsequently checked against production.
7.3 Wax Pattern Preparation and Shell Building
Because the wall thickness of the plate differs greatly from that of the mouth, a chill wax block was placed at the plate position in order to reduce the shrinkage of the plate section of the wax pattern and to equalise the contraction of the pattern as a whole. The injection parameters were held in a narrow window so that the pattern geometry remains repeatable.
| Wax injection parameter | Setting |
|---|---|
| Wax injection temperature | 60 °C ± 5 °C |
| Die clamping force | 80 t |
| Injection pressure | 2.4 MPa ± 0.2 MPa |
| Holding time | 400–450 s |
| Cooling method | Water cooling |
| Plate chill | Chill wax block at the bottom plate |
| Pattern surface acceptance | No oil lines, no cracks |
Under these conditions the wax pattern exhibits a uniform overall contraction with no surface oil lines and no cracks. Shell building was carried out with two face coats and seven reinforcement coats, which provides both the strength required to resist the metallostatic pressure of a 29.38 kg pour and the permeability required to vent gas through the shell wall during filling. The shell build also directly serves the gas-hole problem identified in the filling simulation, because a more permeable backup layer allows entrapped gas to escape rather than being driven back into the metal.
7.4 Pouring and Solidification Control
The nominal pouring temperature range for CF8M is 1 560–1 580 °C. Because the mouth wall is only 5 mm thick, there is a misrun risk during pouring, and I therefore selected the upper end of the range in order to maintain superheat without introducing excessive overheating. The selection of 1 580 °C also keeps the metal sufficiently fluid to fill the thin mouth before the front freezes.
Because the filling simulation predicted an extensive gas-hole risk zone on the plate surface, pouring is performed by a robot using a tilting fork-shell technique. The mould is tilted and rotated so that the metal front advances smoothly and progressively, and the pouring rate is controlled so that turbulence and gas entrainment are suppressed. After pouring, the cooling arrangement is deliberately designed to preserve roundness of the mouth.
| Pouring and cooling parameter | Setting |
|---|---|
| Pouring temperature | 1 580 °C |
| Shell preheating temperature | 1 050 °C |
| Shell preheating time | 2 h |
| Pouring time | 25–30 s |
| Pouring method | Robot tilting fork-shell, smooth and stable pour |
| Post-pour cooling | Shell supported on steel columns at the thick central bottom, suspended and air cooled |
The post-pour support arrangement deserves particular emphasis. The whole shell is lifted and supported from the thick central bottom region by steel columns, so that the shell hangs freely and is not in contact with the floor or with any flat surface at the mouth. This prevents the thin mouth wall from being loaded locally during cooling. Because the mouth is supported only through the stiff central bottom, it contracts uniformly over its full 360° circumference. Since the final dynamic balance requirement of less than 2 g and the mouth roundness requirement of less than 1.5 mm both depend on circumferential uniformity, this single measure has a disproportionate effect on the final acceptance rate. If the shell were allowed to rest on a flat surface, the weight of the assembly would flatten the mouth locally, producing an ovality that no downstream machining operation could correct without removing an excessive amount of material.
8. Verification Results
8.1 Surface and Internal Defect Verification
Following the trial production run, the castings were inspected for surface quality and internal soundness. The surfaces were found to be sound, with no gas holes and no cracks. Radiographic inspection of the trial castings showed no shrinkage porosity, no shrinkage cavity and no slag inclusion, which confirms both the feeding design and the gas-venting strategy. The elimination of the surface gas holes on the plate is attributed jointly to the smooth tilting pour and to the more permeable shell build.
| Inspection item | Method | Requirement | Result |
|---|---|---|---|
| Surface gas holes | Visual inspection | None permitted | None found |
| Surface cracks | Visual and penetrant inspection | None permitted | None found |
| Internal shrinkage | Radiography, ASTM E446 | Level 2 | Level 2 or better |
| Slag inclusions | Radiography, ASTM E446 | Level 2 | Level 2 or better |
8.2 Dimensional Verification
Four key control dimensions were selected for verification: the mouth diameter, the roundness measured 5 mm above the mouth, the included cone angle, and the flatness of the dished region. The mouth diameter was measured with a 0–500 mm digital vernier calliper having a maximum permissible error of ±0.05 mm. Roundness, angle and flatness were measured on a coordinate measuring machine with a resolution of 0.001 mm, collecting coordinate points on the component surface. Each feature was measured three times and all measurements were performed in a controlled environment at 23 °C ± 1 °C and 50% relative humidity.
| Item | Method | Drawing requirement | Measured value, sample 1 | Measured value, sample 2 | Measured value, sample 3 |
|---|---|---|---|---|---|
| Mouth diameter | Digital calliper | Φ429.8 ± 1.4 mm | Φ430.78 mm | Φ431.02 mm | Φ430.80 mm |
| Included cone angle | CMM | 30° ± 0.5° | 29.78° | 29.95° | 29.84° |
| Mouth roundness | CMM | ≤ 1.5 mm | 0.91 mm | 0.80 mm | 0.56 mm |
| Flatness | CMM | ≤ 0.5 mm | 0.15 mm | 0.24 mm | 0.40 mm |
All key dimensions fall inside the acceptance range after the combination of anti-deformation compensation, wax-stage control and pouring and solidification control. The measured values also agree closely with the simulation predictions, which supports the accuracy of the JSCAST deformation prediction for this class of thin-walled conical investment casting.
8.3 Machining and Dynamic Balance Verification
The component operates at high rotational speed. Even a small mass imbalance produces a large centrifugal force, which can cause severe vibration, wear and loss of precision, and in the worst case can lead to machine damage or a safety incident. After machining, the components were therefore subjected to a dynamic balance test, and material was removed at the locations specified on the drawing until the residual unbalance fell below the limit.
| Component number | Unbalance before trimming (g) | Unbalance after trimming (g) | Requirement (g) | Verdict |
|---|---|---|---|---|
| 1 | 180 | 1.55 | < 2 | Pass |
| 2 | 150 | 0.55 | < 2 | Pass |
| 3 | 156 | 0.14 | < 2 | Pass |
All three sampled components achieve a residual unbalance well below 2 g after trimming, and the batch achieved a 100% pass rate. The low residual unbalance values are a direct consequence of the circumferential uniformity that the suspended cooling arrangement provides, because a mouth that is round to within approximately 0.6–0.9 mm requires only a small and well-distributed correction.
9. Discussion of the Control Strategy
The results allow several general principles to be stated for the dimensional control of axisymmetric thin-walled rotatory bodies produced by investment casting.
First, the dominant deformation mode of this geometry is a global, low-order mode rather than a random local distortion. The radial top-to-bottom difference and the axial top-to-bottom difference are consistent in sign and magnitude from part to part, which is precisely what makes geometric compensation viable. If the distortion were stochastic, compensation in the die would be ineffective.
Second, the deformation is strongly dependent on the cone angle. Reducing the die angle from 30° to 28.5° not only shifts the final angle back to the nominal value but also reduces the absolute magnitude of the deformation. This is a favourable coupling, because the compensation is self-reinforcing rather than self-defeating.
Third, the wax pattern contributes a non-negligible share of the total deviation. A die compensation derived only from the casting simulation would have been insufficient because the pattern angle grows by approximately 0.5° before the metal is ever poured. The total compensation must therefore be the sum of the casting contribution and the wax contribution:
$$\Delta\theta_{total} = \Delta\theta_{cast} + \Delta\theta_{wax} + \Delta\theta_{shell}$$
where the shell term accounts for any additional deviation introduced by shell expansion and restraint during preheating and pouring.
Fourth, thermal boundary conditions during post-pour cooling are as important as the die geometry. The shell support arrangement converts a potentially uncontrolled contact condition into a well-defined, rotationally symmetric cooling condition, which preserves roundness and protects the dynamic balance characteristics of the finished part.
Fifth, filling-related gas defects and solidification-related shrinkage defects have different remedies and must be addressed separately. In this case the shrinkage risk was eliminated by a properly sized gate designed by the modulus method, whereas the gas-hole risk was eliminated by a smooth tilting pour and a permeable shell build. Treating both problems with a single measure, for example by simply raising the pouring temperature, would have worsened one while attempting to cure the other.
10. Conclusions
The following conclusions can be drawn from this study of a Φ430 mm × Φ250 mm × 188 mm CF8M conical thin-walled rotatory body produced by investment casting.
JSCAST numerical simulation was used to predict defects in the filling, solidification and deformation stages of the investment casting process, and the predictions were compared with trial production. The filling simulation showed that a metal stream impinging directly on the cavity surface, or a filling rate that is too high, entraps gas and produces surface gas holes on the bottom plate of the casting. The solidification simulation, evaluated against a critical solid fraction of 0.70, showed that the gate and riser dimensions of Φ80 mm × 50 mm calculated by the modulus method provide sufficient feeding, that the casting body carries no shrinkage cavity or porosity risk, and that the process satisfies the directional solidification requirement. The deformation simulation showed that the conical rotatory body deforms by opening at the mouth, curling upward and dishing in the middle, with a radial top-to-bottom difference of 0.7 mm, an axial top-to-bottom difference of 1.5 mm, and an in-plane axial variation of 0.7 mm across the bottom plate.
Based on the simulation results, a complete process scheme was established. In the die design stage, an anti-deformation compensation was applied: the included cone angle was reduced by 1.5° overall, and a 1.2 mm slope compensation was added to the bottom plate plane. The compensated model was re-simulated to verify the predicted final dimensions. In the wax pattern stage, a chill wax block was placed at the bottom plate so that the pattern contracts uniformly and its distortion is reduced. In the pouring stage, a robot-driven tilting fork-shell pour was used, with a smooth and progressive filling action that suppresses gas entrainment. After pouring, the whole shell was suspended on steel columns beneath the thick central bottom so that the thin mouth wall is not loaded locally and contracts uniformly around its full circumference.
Series production with the optimised scheme produced components with good surface quality and no gas holes or porosity. Radiographic inspection showed no defects in either the bottom plate or the mouth. All four key control dimensions met the acceptance criteria, and the batch achieved 100% qualification in both full-dimensional inspection and dynamic balance testing, with residual unbalance values of 1.55 g, 0.55 g and 0.14 g against a limit of 2 g. The anti-deformation compensation combined with full-process control therefore eliminates the dimensional deviation of thin-walled conical investment castings without any subsequent straightening operation. This strategy is directly applicable to the dimensional accuracy control of similar axisymmetric thin-walled rotatory body castings and removes a costly, tool-intensive and difficult-to-control step from the production route.