Riser-Free Casting Optimization for a Ductile Iron Machine Tool Stand

Heavy-section ductile iron castings have always presented one of the most stubborn challenges in foundry engineering. Because the solidification interval of spheroidal graphite iron is wide and its freezing mode tends toward mushy rather than skin-forming, the graphite nodules that form during eutectic solidification are coarse, sparse, and unevenly distributed when the wall thickness becomes large. The consequence is a pronounced tendency toward shrinkage cavities and shrinkage porosity that far exceeds what is observed in comparable grey iron castings. In the production of thick-walled ductile iron castings, foundries have traditionally relied on two competing philosophies: the conventional riser-fed approach, in which large feeding heads are placed on the thermal nodes and are expected to supply liquid metal throughout the whole solidification period, and the riser-free approach, in which the graphite expansion accompanying eutectoid and eutectic solidification is deliberately harnessed to compensate for liquid and crystallization contraction inside a rigid mould cavity that is tightly constrained by a strong sand system and a bolted flask.

In my own work on a large machine tool stand made of ductile iron, I began with the riser-fed philosophy because it is the safer default. Two large risers were placed above the thick support brackets of the shaft holes at the top face of the casting, and the entire gating system was designed around the assumption that these risers would feed the last-solidifying regions. What I observed in serial production, however, contradicted that assumption. Slight shrinkage porosity appeared repeatedly at the root of the risers, exactly at the junction between the feeding head and the casting, and the affected castings had to be rejected. Rather than simply enlarging the risers, which would have lowered the casting yield still further and complicated the moulding operation, I decided to investigate the solidification behaviour numerically and then to test a riser-free route under otherwise identical casting parameters. The result of that investigation is the subject of this article: a systematic comparison of the riser-fed and riser-free strategies for ductile iron castings, the justification of the riser-free route through thermal-field simulation, and the practical measures — wedge-shaped vent holes, thinner ingates, external and internal chills, and a controlled pouring temperature — that made the riser-free process reliable in the foundry.

The stand under discussion is a single-piece, small-batch component with an overall envelope of 2 230 mm × 1 750 mm × 550 mm and a poured weight of approximately 2 483 kg. Its internal architecture is highly cavity-dominated, with a number of internal cores forming hollow sections, which makes the moulding operation labour-intensive and makes any additional moulding feature, such as an enlarged riser neck, disproportionately expensive. The maximum wall thickness is 80 mm while the minimum is 20 mm, giving a wall-thickness ratio of four to one and therefore a very wide distribution of thermal moduli across the same casting. The top face carries a thick support bracket pierced by a nominal Φ70 mm hole, and this bracket is the region where the highest metallurgical quality is demanded because it will eventually carry the load of the moving column. The dimensional and quality requirements for the component are summarized in Table 1.

Parameter Requirement Remark
Material grade QT400-18 Spheroidal graphite iron
Overall envelope 2 230 mm × 1 750 mm × 550 mm Complex internal cavities
Poured weight ≈ 2 483 kg Single-piece, small batch
Maximum wall thickness 80 mm Thermal node at shaft-hole bracket
Minimum wall thickness 20 mm Ratio of 4 : 1
Quality level EN 12890, level H2 Heavy-section castings
Dimensional tolerance DIN ISO 8062-CT12 Machining allowance applied
Minimum surface roughness Ra 12.5 μm Achieved by machining
Internal defects No sand inclusion, shrink hole or porosity Radiographic and visual inspection
Moulding method Furan resin self-hardening sand, manual Metal flask, bolted
Gating system Open stepped system with multiple ingates Balanced filling and feeding

For the melting and treatment route I specified a charge based mainly on steel scrap with carburizer addition, so that the base iron chemistry could be adjusted precisely before tapping. Control of carbon and silicon is the single most important lever for suppressing shrinkage in ductile iron castings, because the two elements together determine the carbon equivalent and hence the amount of graphite that can precipitate during solidification. The carbon equivalent was calculated using the standard expression

$$CE = \%C + \frac{\%Si + \%P}{3}$$

and was maintained at a value slightly above the eutectic composition so that the primary austenite shell would not develop excessively before the eutectic reaction began. The target chemistry of the base iron and of the spheroidized iron is listed in Table 2. The residual magnesium was deliberately kept low and below 0.05%, because an excessive residual magnesium promotes carbide formation, increases the tendency toward shrinkage porosity, and raises the risk of dross and slag inclusions in ductile iron castings.

Element Base iron (%) After spheroidization (%) Function
C 3.6 – 4.0 3.5 – 3.9 Graphite formation, expansion
Si 1.3 – 1.6 2.2 – 2.5 Eutectic shift, matrix
Mn ≤ 0.5 ≤ 0.45 Pearlite control
S ≤ 0.03 ≤ 0.02 Must remain low for Mg efficiency
P ≤ 0.04 ≤ 0.04 Prevents embrittlement
Mg (residual) — ≤ 0.05 Nodularization, kept low
CE value ≈ 4.1 ≈ 4.4 Slightly hypereutectic

The spheroidizing treatment used a purchased calcium-bearing rare-earth magnesium alloy, added at approximately 1.4% of the iron weight, and the treated iron was tapped at about 1 450 °C. Pouring was carried out in the window of 1 310 °C to 1 330 °C. This pouring temperature range is markedly lower than what is often specified for thin-walled ductile iron castings, and the choice was deliberate. A lower superheat reduces the liquid contraction that must be compensated before the eutectic begins, shortens the total solidification time, and therefore narrows the interval during which the mould cavity must remain hydraulically connected to a feed source. For thick-section ductile iron castings, this reduction in superheat is worth more than any increase in fluidity that a higher pouring temperature could provide.

The gating system was an open stepped design with several ingates distributed over the height of the mould cavity, so that the iron rises through the cavity in relatively calm layers and the temperature field remains as uniform as possible at the end of filling. The ingate cross-section was a flat rectangle rather than a circle, because a flat section distributes the incoming stream over a wider area of the mould wall, reduces the local velocity of the metal, and lowers the risk of sand erosion and mould-wall impingement. The basic filling parameters were estimated from the standard relations

$$t_{pour} = \frac{W}{\rho \, v \, A_g}$$

where \(t_{pour}\) is the pouring time, \(W\) is the poured weight, \(\rho\) is the density of liquid iron, \(v\) is the average velocity through the ingates and \(A_g\) is the total ingate area. The total ingate area was in turn sized from the specific pouring rate, which for a casting of this weight and wall thickness was taken in the range typical for heavy ductile iron castings. The stepped arrangement also allowed the lower ingates to complete the filling of the bottom cavities while the upper ingates fed the top bracket region, thereby postponing the solidification of the ingate channels relative to the last-solidifying casting sections in the original design. This postponement was later identified as one of the root causes of the riser-root porosity, because it prevented the risers from being isolated at the correct moment.

The first numerical simulation was carried out with a commercial finite-difference casting simulation package, using the measured thermophysical properties of the alloy and the actual moulding-sand properties. The thermal field was solved with the transient heat-conduction equation

$$\rho c_p \frac{\partial T}{\partial t} = \nabla \cdot (k \nabla T) + q_{lat}$$

in which the latent-heat term \(q_{lat}\) was modelled with an equivalent specific-heat method over the freezing interval. The simulation predicted that the last region to solidify in the whole casting was the thick support bracket around the shaft holes at the top face. The solidification time in that region reached approximately twice the average solidification time of the surrounding walls. With a thermal modulus defined in the classical way as

$$M = \frac{V}{A}$$

where \(V\) is the volume of the section and \(A\) is its heat-transfer surface area, the modulus of the shaft-hole bracket was significantly larger than the modulus of any other part of the stand, and by Chvorinov’s rule

$$t_s = B \left( \frac{V}{A} \right)^{n}$$

with \(B\) the mould constant and \(n\) close to 2 for a sand mould, the local solidification time in the bracket was far longer than elsewhere. This confirmed that the bracket was the natural thermal node of the casting and that any shrinkage defect would be concentrated there.

Following this result, two large risers were placed above the bracket. The intention was straightforward: the risers would remain liquid longer than the bracket, and the hydrostatic pressure of the liquid column in the risers would force feed metal into the contracting bracket. The simulation of the riser-fed version, however, produced a solidification sequence that was the opposite of what had been intended. Because the riser bodies were located above the top face, they lost heat from five faces to the surrounding sand and, more importantly, they were connected to the casting through relatively narrow necks whose modulus was close to that of the bracket itself. The result was that the risers began to freeze at nearly the same time as the bracket, and in some areas the simulation showed the riser neck freezing before the bracket had completed its own solidification. Once the neck is frozen, the feed path is interrupted, and any subsequent contraction inside the bracket must be compensated locally. Because the graphite expansion generated inside the bracket cannot easily be pushed back into a solidifying riser, the net effect is a localized deficit of liquid metal and the appearance of shrinkage porosity exactly at the junction between the riser and the casting.

Simulation output Riser-fed design Observation in production
Last solidifying region Shaft-hole support bracket Confirmed by sectioning
Riser freezing sequence Riser neck freezes before bracket Feed path interrupted
Predicted defect location Riser root / neck junction Slight shrinkage porosity found
Local modulus of bracket ≈ 2.1 × mean wall modulus Thermal node confirmed
Casting yield Reduced by riser weight Extra melting and cleaning cost
Moulding complexity High Longer cycle time
Acceptance rate Below target Rejection of affected castings

The production evidence matched the simulation closely. Shrinkage porosity was found at the riser roots, and the porosity was of the fine dispersed type rather than a large open cavity, which is characteristic of mushy freezing in ductile iron castings. This is an important diagnostic detail. In a skin-forming alloy, inadequate riser feeding usually produces a single concentrated pipe or cavity, and the remedy is to enlarge the riser or improve its insulation. In a mushy-freezing alloy such as spheroidal graphite iron, inadequate feeding produces distributed microporosity that is difficult to eliminate by riser enlargement alone, because the resistance to interdendritic flow is high and the last liquid is trapped in small isolated pockets. Simply making the riser bigger would have increased the amount of liquid metal available but would not necessarily have improved the accessibility of that metal to the last pockets, while it would certainly have increased the riser weight, reduced the casting yield, and added to the cleaning and fettling burden. I therefore rejected the enlargement route and decided to test whether the casting could be produced with no risers at all.

The physical basis of riser-free production of ductile iron castings rests on the volume changes that occur during solidification. When the iron freezes, three competing volume effects act simultaneously: liquid contraction, which removes volume; crystallization contraction of the austenite shell, which also removes volume; and graphite precipitation, which adds volume because the density of graphite is far lower than that of liquid iron. The volume balance per unit mass can be written as

$$\Delta V_{net} = \Delta V_{liquid} + \Delta V_{\gamma} + \Delta V_{graphite}$$

and the process is self-feeding when the graphite expansion term is at least as large in magnitude as the sum of the two contraction terms. Expressing the terms as fractional changes referred to the initial volume, the self-feeding condition becomes

$$\alpha_V \Delta T + \varepsilon_{c} \le \varepsilon_{g} \, f_{g}$$

where \(\alpha_V\) is the volumetric coefficient of thermal expansion of the liquid, \(\Delta T\) is the superheat above the liquidus, \(\varepsilon_{c}\) is the crystallization contraction, \(\varepsilon_{g}\) is the volumetric expansion associated with the precipitation of one unit volume fraction of graphite, and \(f_{g}\) is the graphite volume fraction formed during eutectic solidification. For a hypereutectic or slightly hypereutectic iron, \(f_{g}\) is high and \(\Delta T\) is deliberately kept small; both changes push the inequality toward satisfaction. This is exactly why a low pouring temperature and a high carbon equivalent are the two most powerful process levers in riser-free production of ductile iron castings.

For the expansion to be useful, however, it must act inward, that is, it must be directed toward the interior of the casting and must not be allowed to escape by deforming the mould wall or by pushing metal back through the gating system. Two conditions must therefore be satisfied simultaneously. First, the mould must be sufficiently rigid. This was achieved by using furan resin self-hardening sand, which develops high compressive strength after curing, and by placing the sand mould inside a heavy metal flask that was bolted shut so that the cope and drag could not separate under the internal pressure. Second, the gating system must be isolated at the correct moment. If the ingates remain open while the casting is expanding, the expansion is wasted by pushing metal back into the sprue; if the ingates freeze too early, the cavity may not be completely filled. The correct sequence is for the cavity to be completely filled first, and then for the ingates and the vent openings to freeze before the casting itself completes solidification. Once a solid skin has formed over the whole casting surface and all the gates and vents are sealed, the interior behaves as a closed autoclave in which graphite expansion raises the internal pressure and suppresses the formation of porosity.

Based on this reasoning, I modified the original design as follows. The two large risers were removed entirely and replaced by flat wedge-shaped vent openings, one above each of the two support brackets and several more distributed along the top face of the casting. These vent openings serve a dual purpose: they allow the mould gases and the initial expansion of the metal to escape during filling, and because their cross-section is thin and flat, they freeze very early and seal the cavity immediately after filling is complete. The ingates were redesigned as flat rectangles with a thickness of 18 mm, which is thinner than the local wall thickness of the casting, so that the ingate channels also freeze before the casting sections they feed. The sprue and the bottom ingates were formed with pre-embedded ceramic tubes to avoid erosion and to give a clean, reproducible channel geometry. The pouring parameters, including the pouring temperature window of 1 310 °C to 1 330 °C, were kept unchanged from the earlier trials, so that the comparison would be a fair one. The revised parameters are collected in Table 3.

Feature Original riser-fed design Revised riser-free design
Risers Two large top risers None
Top openings Riser necks Flat wedge-shaped vent holes
Vent location Above brackets only Above brackets and along top face
Ingate thickness Equal to wall thickness 18 mm flat rectangle
Ingate freezing relative to casting After casting Before casting
Vent freezing relative to casting After casting Before casting
Sprue and bottom ingates Sand channels Pre-embedded ceramic tubes
Chills None External or internal chills at brackets
Pouring temperature 1 310 – 1 330 °C 1 310 – 1 330 °C
Moulding sand Furan resin self-hardening Furan resin self-hardening
Flask clamping Bolted Bolted, with increased preload
Graphite expansion utilization Partly lost Deliberately exploited

The simulation of the revised design showed a very different solidification pattern. The thermal field at the moment when the casting was about to complete its solidification indicated that the ingates and the wedge-shaped vent holes had already frozen completely, while the interior of the shaft-hole brackets was still liquid. This is precisely the condition required for autoclave self-feeding. The simulation also showed that the temperature gradients within the casting were more uniform than in the riser-fed design, because the removal of the risers eliminated the localized hot spots that the riser necks had created and because the thin vent openings acted as small cooling fins that extracted heat from the top face in a controlled manner. The solidification sequence, in other words, had been changed from a pattern dominated by two isolated hot spots to a pattern in which the whole casting solidified progressively from the outside inward and from the thin sections toward the thick sections.

For a more quantitative assessment, the local solidification time and the local cooling rate were estimated for the principal sections of the stand using the modulus method, and the results are collected in Table 4. The cooling rate was approximated as

$$\dot{T} = \frac{T_{pour} – T_{solidus}}{t_s}$$

where \(T_{pour}\) is the pouring temperature, \(T_{solidus}\) is the solidus temperature of the alloy and \(t_s\) is the local solidification time. The table makes clear that the bracket region has both the longest solidification time and the slowest cooling rate, and it is therefore the region in which the microstructure will be coarsest and the mechanical properties lowest unless additional cooling is applied. This is the reason why the introduction of chills at the bracket was considered an essential part of the riser-free design and not an optional refinement.

Section Wall thickness (mm) Modulus M (mm) Relative solidification time Relative cooling rate
Thin outer rib 20 ≈ 9 1.0 1.0
Side wall, typical 35 ≈ 15 ≈ 2.8 ≈ 0.36
Base plate 50 ≈ 21 ≈ 5.4 ≈ 0.19
Heavy internal rib 65 ≈ 27 ≈ 9.0 ≈ 0.11
Shaft-hole support bracket 80 ≈ 33 ≈ 13.4 ≈ 0.075
Bracket with external chill 80 ≈ 24 ≈ 7.1 ≈ 0.14
Bracket with internal chill 80 ≈ 22 ≈ 6.0 ≈ 0.17

Two cooling strategies were evaluated for the bracket region. The first was an external chill placed against the outer surface of the bracket on both sides, which increases the effective heat-transfer area and thereby reduces the apparent modulus. The second was an internal chill placed inside the mould cavity at the bracket, in the region that would otherwise become the thermal core. Because an internal chill must be completely fused into the casting and must not leave a cold shut or an unbonded interface, its weight and position must be chosen so that it is fully melted by the superheat of the surrounding iron and then dissolves into the matrix. Both approaches accelerate the local cooling rate, refine the graphite nodule count, and reduce the tendency toward intercellular porosity. Their effect on the local modulus is shown in the last two rows of Table 4.

The simulation of the riser-free casting with external chills showed a noticeably faster cooling of the bracket, a more balanced overall solidification pattern, and a further reduction of the volume of the last-solidifying region. The temperature field in the partial cross-section through the bracket indicated that the thermal centre had shifted away from the geometric centre of the section and toward the chill face, which is the intended effect. This shift is beneficial in ductile iron castings because it changes the direction of the last liquid flow, moving the final feeding distance from the surface toward the interior, where a limited amount of porosity is less damaging than porosity connected to the machined surface.

Several production variants were then manufactured and compared under the same melting, pouring and inspection conditions. The variants differed in whether the shaft hole was cast to shape or left solid for subsequent machining, and in whether external chills, internal chills or no chills were used at the bracket. All variants used the riser-free gating scheme with thin ingates and wedge-shaped vent holes. The comparison is summarized in Table 5.

Variant Shaft hole Chill type Moulding effort Bracket soundness Acceptance
A Cast to shape External chills Moderate Sound Qualified
B Cast to shape No chills Low Minor porosity Marginal
C Left solid External chills Moderate Sound Qualified
D Left solid Internal chills Moderate to high Sound Qualified
E Left solid No chills Low Minor porosity Marginal

The results confirmed that the chills were essential. Variants A, C and D all produced fully sound brackets after machining verification, whereas variants B and E showed minor porosity in the last-solidifying region of the bracket even though the surface appearance was satisfactory. An important practical observation was that leaving the shaft hole solid, as in variants C and D, made it easier to position the chills, because the solid hole region provided a convenient location for an internal chill without any risk of core distortion, and it also simplified the core assembly. When the hole was cast to shape, the core that formed the Φ70 mm hole acted as a local insulator and made the internal chill option more difficult to implement. After machining, the hole produced from the solid variant showed a clean, dense surface with a refined microstructure, and ultrasonic inspection revealed no internal defects in the surrounding material.

The mechanism by which the chills improve the graphite morphology deserves some comment. In a thick section of ductile iron castings, the cooling rate at the thermal centre is very low, sometimes on the order of a fraction of a degree per second. Under such conditions the number of graphite nodules that nucleate is small, the nodules grow to a large size, and the last liquid to solidify is concentrated in the intercellular regions where it cannot be fed. By increasing the local cooling rate, the chill raises the nucleation density of graphite nodules, which both refines the nodule size and distributes the expansion more uniformly throughout the freezing section. The nodule count \(N\) scales approximately with the cooling rate according to a relation of the form

$$N \propto \dot{T}^{\,m}$$

with an exponent \(m\) typically between 0.5 and 1.0 for cast irons. A higher nodule count means a larger number of expansion centres, and the expansion that each nodule produces has only a short distance to travel before it fills the adjacent intercellular pocket. The result is a lower level of microporosity and a more uniform matrix. This is why chills are so effective in heavy ductile iron castings, even though they do not change the total amount of graphite that precipitates.

The casting yield was calculated for the original and revised designs using the usual definition

$$\eta = \frac{W_{casting}}{W_{casting} + W_{gating} + W_{risers} + W_{feeders}} \times 100\%$$

The removal of the two large risers increased the yield substantially, and the reduction in fettling and cleaning work that followed from the elimination of the riser necks was an additional and even more valuable benefit, because fettling labour is one of the largest cost components in heavy ductile iron castings. The comparison is given in Table 6.

Quantity Riser-fed design Riser-free design Change
Casting weight (kg) ≈ 2 483 ≈ 2 483 0
Riser weight (kg) ≈ 480 0 −480
Gating weight (kg) ≈ 320 ≈ 300 −20
Total poured weight (kg) ≈ 3 283 ≈ 2 783 −500
Casting yield η ≈ 75.6% ≈ 89.2% +13.6 percentage points
Moulding cycle Long Short Reduced
Fettling work High Low Reduced
Acceptance rate Marginal Qualified Improved

The wedge-shaped vent holes deserve particular attention because they are the element that replaces the risers in controlling the internal pressure. Their geometry is chosen so that their cross-sectional area is small and their perimeter-to-area ratio is high, which means that they lose heat rapidly and freeze very early. At the same time, their shape allows gas to escape freely during filling, so that the gas pressure inside the cavity does not rise and oppose the incoming metal. Once they freeze, they seal the cavity and allow the graphite expansion to pressurize the interior. If the vent holes were made too thick, they would remain open after the cavity was filled, and the expansion would be partly lost through them; if they were made too thin, they would freeze before the cavity was completely filled, and the casting would suffer from misruns and cold shuts. The design window is therefore narrow, and it was determined by combined thermal simulation and trial pours. The optimum thickness was found to lie between one quarter and one half of the adjacent casting wall thickness for this particular component.

The thin ingates play a complementary role. In the original riser-fed design, the ingates were as thick as the surrounding wall, so they provided a long-lived hydraulic connection between the sprue and the casting. During the first stage of solidification, this connection allowed the casting to absorb liquid metal from the sprue and the risers, which is what the feeding theory requires for the contraction phase. However, it also allowed the metal to be pushed back into the sprue during the graphite expansion phase, which is exactly the opposite of what is wanted. By reducing the ingate thickness to 18 mm, I ensured that the ingates froze shortly after the cavity was filled. The sequence of events then became: complete filling, early sealing of ingates and vents, solid skin formation over the casting, internal graphite expansion, and finally solidification of the thick sections under internal pressure. This is the classic riser-free sequence for ductile iron castings, and it worked reliably in the trials.

The role of the pouring temperature can be quantified through the superheat term in the volume balance. The liquid contraction before the start of eutectic solidification is proportional to the superheat, so reducing the pouring temperature from a typical 1 380 °C to the range of 1 310 °C to 1 330 °C reduces the superheat by approximately 50 °C to 70 °C. If the volumetric contraction coefficient of liquid iron is taken as roughly \(1.0 \times 10^{-4}\ \mathrm{K^{-1}}\), the reduction in the liquid contraction volume per unit mass of iron is on the order of

$$\Delta \varepsilon = \alpha_V \Delta T \approx 1.0 \times 10^{-4} \times 60 \approx 6.0 \times 10^{-3}$$

which corresponds to about 0.6% of the volume of the liquid. For a casting of 2.5 tonnes this represents a substantial quantity of liquid metal that no longer has to be supplied by feeding. The lower pouring temperature therefore directly reduces the feeding requirement and brings the process closer to the self-feeding condition. The penalty is a reduction in fluidity, which was compensated by the improved gating design and by the fact that the minimum wall thickness of 20 mm is still well above the fluidity limit for this alloy at 1 310 °C.

Another factor that contributed to the success of the riser-free route was the rigidity of the mould. The internal pressure generated by graphite expansion in a well-fed casting can reach several bar, and if the mould wall yields, the expansion is absorbed by mould deformation instead of being used to compensate contraction. Furan resin self-hardening sand has a high compressive strength when properly cured, and the heavy bolted flask prevents the cope from lifting. I also increased the flask clamping preload and verified that the mould did not open during pouring. In addition, the sand was cured to a controlled strength so that it was strong enough to resist the expansion pressure but still sufficiently permeable to allow the mould gases to escape. The interaction between sand strength and expansion pressure can be expressed as a simple stability criterion

$$p_{g} \le \frac{\sigma_{m}}{k_{s}}$$

where \(p_{g}\) is the internal expansion pressure, \(\sigma_{m}\) is the compressive strength of the moulding sand and \(k_{s}\) is a safety factor that accounts for the geometry of the mould and the non-uniformity of the pressure distribution. For the sand and the flask used here, the criterion was satisfied with an adequate margin.

After the trials, the riser-free process was adopted for the stand and for several other heavy ductile iron castings in the same product family. The key parameters that were fixed in the process specification are summarized in Table 7. These values were validated over a production campaign, and no shrinkage-related rejections occurred during that campaign.

Process parameter Specified value Reason for the choice
Carbon equivalent ≈ 4.4 Maximizes graphite expansion
Residual Mg ≤ 0.05% Suppresses carbides and dross
Pouring temperature 1 310 – 1 330 °C Reduces superheat and contraction
Ingate thickness 18 mm flat rectangle Freezes before the casting
Vent hole type Wedge-shaped Freezes early, seals the cavity
Vent hole thickness 1/4 to 1/2 of wall Balances venting and sealing
Chill position Both sides of the bracket Refines graphite, speeds cooling
Mould rigidity Furan sand, bolted flask Directs expansion inward
Casting yield ≈ 89% Economic benefit

The metallographic examination of the bracket region confirmed the expectations from the thermal analysis. In the chill-free variants the graphite nodules were coarse and the nodule count was low, while in the chilled variants the nodule count was substantially higher and the nodules were smaller and more evenly distributed. The matrix in the chilled region was predominantly ferritic, which is the required matrix for the QT400-18 grade, and no free carbide was detected at the chill faces, which confirms that the chill weight and position had been selected correctly and that the chill had been fully fused. The hardness distribution across the bracket was uniform, with no hard spots that would have made machining difficult.

In terms of the mechanical properties, test bars cut from the bracket region of the riser-free castings satisfied the requirements of the grade, with tensile strength above 400 MPa, elongation above 18%, and a nodularity above 85%. The ultrasonic inspection of the full casting showed no indications exceeding the acceptance threshold. The machined shaft holes showed a clean surface with no exposed porosity, which is particularly important because these holes carry the moving column of the machine and any surface porosity would act as a stress raiser.

The economic benefit of the change was evaluated by comparing the total cost per acceptable casting before and after the change. The main contributions were the reduction in melting cost associated with the eliminated riser weight, the reduction in fettling and cleaning labour, the reduction in moulding time, and the improvement in the acceptance rate. The elimination of the risers alone reduced the poured weight by about 15%, and since melting energy and charge materials scale with the poured weight, this reduction propagated through the entire cost structure. The reduction in fettling work was equally significant, because the riser necks on a casting of this size require careful arc-air gouging and grinding, and any error in that operation can damage the casting surface. The quantitative comparison is given in Table 8.

Cost element Riser-fed (relative) Riser-free (relative) Direction
Charge material 100 ≈ 85 Decrease
Melting energy 100 ≈ 85 Decrease
Moulding labour 100 ≈ 80 Decrease
Core assembly 100 ≈ 90 Decrease
Fettling and cleaning 100 ≈ 60 Decrease
Rejection and rework 100 ≈ 40 Decrease
Overall cost per good casting 100 ≈ 78 Decrease

Several general lessons can be drawn from this work and applied to other heavy ductile iron castings. The first is that the riser-fed approach is not automatically safer than the riser-free approach; in a mushy-freezing alloy, a poorly designed riser can actually create the defect that it was intended to prevent, because the thermal node created by the riser neck can become the last region to solidify. The second is that the decision between the two approaches should be based on a quantitative volume balance rather than on tradition. If the graphite expansion term is large enough and the superheat is small enough, the self-feeding condition can be satisfied and the risers become unnecessary. The third is that the moulding system, the gating system and the chilling practice must all be designed as a single integrated system. A rigid mould without sealed ingates will leak expansion, and sealed ingates without a rigid mould will simply deform the mould. The fourth is that the pouring temperature is a process variable of the first order in the riser-free production of ductile iron castings, not a secondary parameter to be adjusted for convenience.

It is also worth noting that the riser-free approach is not universally applicable. It requires a carbon equivalent that is high enough to generate sufficient graphite expansion, a mould that is strong enough to contain that expansion, and a section that is not so massive that the cooling rate falls below the level at which graphite nucleation becomes inadequate. For extremely heavy sections, chills or internal cooling may be needed to raise the nucleation density, as was the case for the bracket in this study. For thin sections, the risk of misruns at low pouring temperature must be balanced against the benefit of reduced contraction. The design window for the riser-free production of ductile iron castings is therefore real but finite, and it must be established by a combination of simulation, calculation and trial pours.

Looking back at the sequence of decisions, the most important step was the recognition that the porosity at the riser root was not a feeding-capacity problem but a feeding-access problem. Enlarging the risers would have addressed the wrong variable. Once the problem was reformulated in terms of thermal sequence and volume balance, the solution followed naturally: remove the risers, seal the cavity early with thin ingates and wedge-shaped vents, increase the local cooling rate with chills, keep the pouring temperature low, and ensure that the mould is rigid enough to direct the graphite expansion inward. Each of these measures supports the others, and together they convert the casting from a shape that requires external feeding into a shape that feeds itself.

In conclusion, the riser-free casting process was successfully applied to a heavy ductile iron machine tool stand. The original riser-fed design produced slight shrinkage porosity at the riser roots and required a low casting yield. The revised design replaced the risers with wedge-shaped vent holes, reduced the ingate thickness, and introduced external or internal chills at the shaft-hole support brackets, while keeping all other casting parameters unchanged. The revised process produced fully sound castings with a higher nodule count and better mechanical properties in the critical bracket region, and the casting yield rose from about 76% to about 89%. Both the variant in which the shaft hole was cast to shape with external chills and the variant in which the hole was left solid for subsequent machining with external or internal chills were found to give qualified castings, and the solid-hole variant was the more convenient one for moulding and for chill placement. The work demonstrates that the riser-free route is a viable and economically attractive option for heavy ductile iron castings when the moulding, gating, chilling and pouring parameters are designed as an integrated system, and it shows that numerical simulation of the thermal field is an effective tool for identifying the true thermal node and for predicting whether the self-feeding condition can be met.

Conclusion item Finding
Original riser performance Risers did not prevent porosity; defect at riser root
Cause of defect Riser neck froze before the bracket, feed path interrupted
Riser-free viability Confirmed by simulation and production trials
Key measures Wedge vents, 18 mm ingates, chills, low pouring temperature
Chill necessity Required to obtain sound brackets
Hole strategy Cast-to-shape or left solid both acceptable
Yield improvement From ≈ 75.6% to ≈ 89.2%
Cost reduction Overall cost per good casting reduced by ≈ 22%
General applicability Applicable when volume balance and mould rigidity permit
Scroll to Top