Shrinkage Defects in Castings

In my own foundry work, I have repeatedly seen how shrinkage cavities and shrinkage porosity can destroy the performance of an otherwise well-made casting. I have studied these defects in investment castings, and I have also followed their behavior in evaporative pattern casting, where the same fundamental rules of liquid and solidification shrinkage still govern the final soundness of the part. My purpose here is to explain the features, locations, mechanisms, causes, and prevention methods for shrinkage defects in a practical, first-person way, while summarizing the most useful relationships with formulas and tables. The discussion applies directly to investment castings and is also highly relevant to evaporative pattern casting, because in both processes the mold must remove heat, the metal must feed the last solidifying regions, and the final part must be free of internal voids if it is to meet service requirements.

How I Classify Shrinkage Defects

I divide shrinkage defects into concentrated shrinkage cavities and distributed shrinkage porosity. A concentrated shrinkage cavity is a relatively large, irregular void formed where the metal is the last to solidify and cannot receive enough feed metal. Its internal surfaces are rough, and under magnification I can often see well-developed dendritic crystals. When the cavity breaks through the casting surface, I call it an external shrinkage cavity. When it remains inside the casting, I call it an internal shrinkage cavity. When it forms in a re-entrant corner or a sharp thermal junction, I call it a corner shrinkage cavity. Although these names describe different appearances, their root cause is essentially the same: the last liquid cannot be fed during cooling and solidification.

Distributed shrinkage porosity is different in scale. It consists of many small holes scattered among dendrites or grains. I often describe it as microporosity when the voids are very fine, but in practice shrinkage cavity and shrinkage porosity form a continuum. A large cavity can be surrounded by a halo of porosity, and a region that looks dense on a machined surface may still contain fine internal porosity that appears after nitric acid etching or radiographic inspection.

Defect type Typical appearance Usual location Main feeding condition
External shrinkage cavity Open, rough, irregular depression or hole Top surface, feeder contact area, hot spot exposed to air No liquid feed at the final stage
Internal shrinkage cavity Closed, rough, dendritic void Thermal center, thick section, isolated hot spot Feeding path blocked or exhausted
Corner shrinkage cavity Void at a re-entrant corner or junction Concave corner, rib junction, local hot zone Poor local heat extraction and late solidification
Shrinkage porosity Fine scattered pores among dendrites Last solidified interdendritic regions Insufficient interdendritic feeding and gas evolution

Where I Expect Shrinkage to Appear

I expect shrinkage defects in regions that solidify last and cannot be fed. These are usually hot spots, thick sections, isolated heavy bosses, junctions of ribs and walls, and areas remote from the gate or riser. The exact position depends on where I place the internal gate, how the casting is oriented, how heat is extracted by the mold, how the gating system affects cooling, and how multiple parts on the same tree or cluster influence one another. In my experience, the gating layout is never a neutral factor; it changes the temperature field and therefore changes where the last liquid remains.

I use several methods to predict shrinkage locations. I draw solidification isotherms or use the inscribed circle method to estimate thermal centers. I dissect sample castings to confirm where cavities and porosity occur. I also use computer simulation, especially a casting process analysis system, to visualize mold filling and solidification and to predict shrinkage cavity and shrinkage porosity. In evaporative pattern casting, simulation is equally valuable because the pattern decomposition, gas pressure, coating permeability, and sand compactness can alter the local cooling rate and feeding resistance.

Why Shrinkage Happens: The Three Stages of Contraction

When liquid metal is poured into a hot mold, it cools from the pouring temperature to room temperature and passes through three contraction stages: liquid contraction, solidification contraction, and solid contraction. Liquid contraction and solidification contraction are the stages that create the volume deficit responsible for shrinkage cavities and porosity. Solid contraction mainly affects final dimensions, residual stress, distortion, and cracking, although it can also slightly reduce the size of an existing cavity.

I express liquid contraction as:

$$ \epsilon_{V,L} = \alpha_{V,L} (T_p – T_L) \times 100\% $$

where $$\epsilon_{V,L}$$ is the liquid volumetric contraction, $$\alpha_{V,L}$$ is the average volumetric contraction coefficient of the liquid from the pouring temperature $$T_p$$ to the liquidus temperature $$T_L$$, and $$T_p – T_L$$ is the superheat. For carbon steel, I commonly take $$\alpha_{V,L}$$ as approximately $$1.6 \times 10^{-4} \, ^\circ C^{-1}$$.

Solidification contraction occurs as the metal changes from liquid to solid between the liquidus and solidus temperatures. I write it simply as:

$$ \epsilon_{V,C} = \frac{\Delta V_C}{V_0} \times 100\% $$

where $$\epsilon_{V,C}$$ is the solidification volumetric contraction. For medium-carbon steel, I often use a value near $$3\%$$, but the value depends strongly on carbon content and alloy composition.

Solid contraction occurs after solidification is complete and continues to room temperature:

$$ \epsilon_{V,S} = \alpha_{V,S} (T_S – T_R) \times 100\% $$

and the corresponding linear contraction is:

$$ \epsilon_L = \alpha_L (T_S – T_R) \times 100\% $$

where $$\epsilon_{V,S}$$ is solid volumetric contraction, $$\epsilon_L$$ is linear contraction, $$\alpha_{V,S}$$ and $$\alpha_L$$ are volumetric and linear contraction coefficients, $$T_S$$ is the solidus temperature, and $$T_R$$ is room temperature. For many steels, I approximate the solid linear contraction as one third of the solid volumetric contraction, often around $$2.4\%$$, although it again depends on carbon and other elements.

The total volumetric contraction from pouring to room temperature is:

$$ \epsilon_{V,total} = \epsilon_{V,L} + \epsilon_{V,C} + \epsilon_{V,S} $$

For one steel example with a superheat of about $$100^\circ C$$, I have used:

$$ \epsilon_{V,L} = 1.6\%,\quad \epsilon_{V,C} = 3.0\%,\quad \epsilon_{V,S} = 7.2\% $$

so that:

$$ \epsilon_{V,total} = 1.6\% + 3.0\% + 7.2\% = 11.8\% $$

This total is a useful first estimate, but it is not a direct prediction of cavity volume. Gas evolution, mold restraint, core restraint, casting geometry, and actual feeding all reduce or redistribute the final shrinkage. In evaporative pattern casting, the decomposition products of the pattern can also influence gas pressure and local feeding, so I always treat the theoretical contraction as a starting point rather than a final answer.

Carbon content, wt% Approximate solidification contraction, %
0.10 2.0
0.35 3.0
0.45 4.3
0.70 5.3
Carbon content, wt% Solid linear contraction, %
0.08 2.47
0.14 2.46
0.35 2.40
0.45 2.35
0.55 2.31
0.90 2.18

The Mechanism of a Concentrated Shrinkage Cavity

I explain the formation of a concentrated cavity with a simple cylindrical casting. When metal enters the mold, most heat is extracted through the mold wall. This creates a temperature gradient from the surface to the center and from the bottom to the top, depending on gating and feeding. The outermost liquid reaches the liquidus first, and a thin solid skin forms. As cooling continues, this skin thickens inward. The liquid inside contracts, so the liquid level falls. Because liquid and solidification contraction together are greater than the solid contraction of the already frozen skin, the liquid pulls away from the top surface, especially under gravity. A cavity begins to form. At the end of solidification, a concentrated cavity remains in the upper part or in the last-fed region. Final solid contraction then shrinks both the casting and the cavity slightly.

The same mechanism occurs in evaporative pattern casting, although the pattern is replaced by liquid metal as the foam decomposes. The mold cavity is not empty at the start; the pattern occupies space and then disappears. If the decomposition gas is not evacuated properly, local pressure can oppose feeding. If the coating or sand permeability is poor, the effective feeding path can be restricted. Therefore, I regard evaporative pattern casting as a process in which the classical shrinkage rules still apply, but with an additional gas and pattern-decomposition factor.

Solidification Time and the Thermal Modulus

To control shrinkage, I must control solidification sequence. The most useful geometric quantity is the thermal modulus:

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

where $$V$$ is the volume of the casting region and $$A$$ is its heat-dissipating surface area. Chvorinov’s rule gives the solidification time as:

$$ t_s = B M^n $$

where $$t_s$$ is solidification time, $$B$$ is a mold constant, and $$n$$ is usually close to $$2$$ for many castings. If I want directional solidification toward a feeder, I need:

$$ t_{s,1} < t_{s,2} < t_{s,3} < \cdots < t_{s,n} $$

or equivalently:

$$ M_1 < M_2 < M_3 < \cdots < M_n $$

where region 1 is farthest from the feeder, region 2 is closer, and region n is the feeder itself. In practice, I try to make the feeder modulus larger than the modulus of the feeding region. A common starting condition is:

$$ M_f > M_c $$

where $$M_f$$ is the feeder modulus and $$M_c$$ is the casting hot-spot modulus. For many steels and long-freezing-range alloys, I prefer a feeder modulus ratio that accounts for the alloy’s feeding efficiency, gas content, and solidification range. For short-freezing-range alloys, a smaller ratio may work; for wide-freezing-range alloys, I often need a larger ratio and additional chilling.

Region Modulus relation for sound feeding Practical meaning
Thin outer section Smallest $$M$$ Solidifies first and should not cut off feed paths
Intermediate section Intermediate $$M$$ Acts as a feed channel toward the hot spot
Hot spot Largest casting $$M$$ Last casting region to solidify
Feeder or riser Larger than hot spot Supplies liquid metal until the hot spot is frozen

How I Diagnose the Causes

I group the causes of shrinkage cavities and porosity into seven practical categories: gating system design, casting structure, alloy shrinkage, pouring conditions, mold or shell cooling capacity, local heat extraction, and feeding pressure head. Each category interacts with the others. For example, a poor gating design can be partly compensated by a higher pouring temperature, but that same higher temperature increases liquid contraction and may enlarge the cavity. In evaporative pattern casting, the mold cooling capacity and gas evacuation are especially important because the pattern decomposition changes the local thermal environment.

Cause 1: Unreasonable Gating System Design

An unreasonable gating system prevents directional solidification and adequate feeding. Directional solidification means that I deliberately control the freezing sequence so that the metal solidifies progressively toward the feeder or riser. The last liquid should freeze inside the gating system, not inside the casting. If the gate freezes too early, the casting hot spot cannot be fed. If the gate remains liquid too long, a cavity may form at the junction between the gate and the casting because that region becomes the last place to solidify and no additional liquid remains to feed it.

I think of the ideal sequence as a relay of liquid metal. The thin sections freeze first. The intermediate sections freeze next. The hot spot is fed by the feed channel. The feeder freezes last. Then the feeder and gate are removed, leaving a dense casting. If that relay is interrupted, a shrinkage cavity appears. In evaporative pattern casting, I also need to consider that the foam pattern burns away progressively, so the filling pattern can influence the temperature distribution and the effective feeding path.

Cause 2: Unreasonable Casting Structure

I frequently find that the casting geometry itself is the main problem. A large change in cross-section causes the thin section to freeze quickly while the thick section remains liquid and isolated. An isolated heavy section cannot be fed. A fillet that is too small creates a sharp transition and a feeding bottleneck. A fillet that is too large creates a new thermal center. Both extremes can promote shrinkage. For parts converted from forgings or weldments, I must redesign the geometry for casting, not simply copy the original shape. I need to use gradual transitions, uniform wall thickness, and properly placed feeding aids.

Geometry issue Effect on solidification Shrinkage risk
Abrupt thickness change Thin section freezes before thick section High
Isolated heavy section No feed path from feeder Very high
Too-small fillet Restricted feed channel High
Too-large fillet New hot spot High
Uniform wall with gradual transition Controlled directional freezing Low

Cause 3: High Liquid and Solidification Shrinkage

The larger the liquid contraction and solidification contraction, the greater the potential cavity volume. For a given alloy, liquid contraction depends on superheat:

$$ \epsilon_{V,L} \propto \alpha_{V,L} (T_p – T_L) $$

If the gating system has enough feed metal, reducing superheat can reduce liquid contraction. However, if the pouring temperature is too low, the gating system may freeze early and lose its feeding ability. If I increase superheat, liquid contraction increases, but the feeder may stay liquid longer and feed more effectively. This is why I never judge pouring temperature by itself. I judge it together with feeder size, gate modulus, mold preheat, and alloy freezing range. The same trade-off exists in evaporative pattern casting, where a lower pouring temperature may reduce gas generation but can also reduce feed ability.

Solidification contraction also depends on alloy composition. A higher solidification contraction means a larger volume deficit. Gas evolution during solidification can offset some volume loss by forming gas pores, but this is not a desirable remedy because it replaces one defect with another. If the gas cannot float or enter an existing cavity, it remains as dispersed porosity and expands the porosity zone.

Cause 4: Improper Pouring Conditions

Pouring speed strongly affects shrinkage. If the casting does not freeze locally during pouring, a faster pour generally produces a larger cavity. The reason is that a fast pour gives the metal less time to lose heat. When pouring ends, the liquid in the mold is hotter, so liquid contraction is greater. In addition, the first metal poured has less time to be fed by later metal, so the final cavity volume increases. I have seen this clearly in both investment castings and evaporative pattern casting. In evaporative pattern casting, a fast pour can also generate more decomposition gas and turbulence, which may further disturb feeding.

I use pouring practices such as high-temperature tapping with low-temperature pouring, and for large clusters I sometimes use a fast-slow-refill sequence. For top pouring, I tend to use a higher pouring temperature and a slower pouring speed. For bottom pouring, I tend to use a lower pouring temperature and a faster pouring speed. These rules are not absolute, but they help me balance filling, gas evacuation, and feeding.

Pouring condition Effect on shrinkage Preferred practice
Very fast pour Higher final liquid temperature, larger cavity Control speed; avoid excessive turbulence
Very slow pour Early freezing of feed paths Maintain enough superheat for feeding
Low pouring temperature Lower liquid contraction but risk of cold shut and blocked feed Use with adequate feeder and mold preheat
High pouring temperature Higher liquid contraction but better feed fluidity Use with larger feeder and controlled gas evacuation

Cause 5: Poor Mold or Shell Cooling Capacity

If the mold or shell removes heat poorly, the liquid cools and solidifies slowly. The earlier liquid contraction is then less compensated by later liquid, so the final cavity can be larger. If I can make the solidification time equal to the pouring time, the casting can freeze continuously and avoid a final cavity, which is the principle of continuous casting. In investment casting, the shell thickness, shell material, preheat temperature, and insulation control cooling. In evaporative pattern casting, the sand, coating, pattern decomposition, and gas evacuation all influence cooling. I therefore treat mold cooling as a controllable process variable, not a fixed condition.

Cause 6: Poor Local Heat Extraction

A re-entrant corner is a classic example of poor local heat extraction. The mold volume near the corner is shared by several surfaces, so heat accumulates and the corner remains liquid longer. If it cannot be fed, a corner shrinkage cavity forms. I sometimes place chills in the mold or shell to improve local cooling. For investment casting shells, I prefer chills made of brass or stainless steel because carbon steel can oxidize during shell firing and may not bond well. For evaporative pattern casting, internal or external chills can also be used, but I must ensure that they do not interfere with pattern decomposition or coating application.

Cause 7: Insufficient Feeding Pressure Head

If the pressure head is too low, the liquid metal flows too slowly and feeding is incomplete. Increasing the pressure head usually improves feeding, but there is a limit. For thick sections, the effective feeding distance does not increase indefinitely with thickness. A very tall feeder may only feed the lower part of the casting, leaving a secondary shrinkage cavity higher up. In such cases, I use insulating risers, exothermic covers, and feeding aids. If necessary, I change the process to centrifugal casting or pressure-assisted pouring. In evaporative pattern casting, the feeding pressure head must also overcome the resistance of the decomposing pattern and the coating, so the gating design must be even more carefully balanced.

Cause category Main mechanism Typical remedy
Gating design No directional freezing or feed path Correct gate position, modulus, and area ratios
Casting structure Hot spots and isolated heavy sections Uniform walls, gradual transitions, padding
Alloy shrinkage Large liquid and solidification contraction Select suitable alloy; control superheat
Pouring conditions Excess heat or early freezing Controlled pouring speed and temperature
Mold cooling Slow heat extraction Optimize shell, sand, coating, and preheat
Local cooling Corner or junction stays liquid Use chills and local cooling
Pressure head Insufficient feed flow Increase head; use insulating or exothermic risers

How I Prevent Shrinkage: The Principle of Sequential Solidification

My basic prevention principle is simple: make the casting solidify sequentially and leave the shrinkage cavity in the gating system, not in the part. I want the metal to freeze progressively toward the feeder, with no isolated liquid regions. I achieve this by controlling geometry, gating, thermal gradients, and pouring practice. The goal is to create a temperature gradient that favors feed flow:

$$ \frac{dT}{dx} < 0 \quad \text{toward the feeder} $$

where $$x$$ is distance from the far end of the casting toward the feeder. The feeder must remain liquid longer than the casting hot spot, so I require:

$$ t_{s,f} > t_{s,c} $$

and therefore, for similar mold conditions:

$$ M_f > M_c $$

I also need an open feed channel. If the channel freezes before the hot spot, the feeder cannot work regardless of its size. I therefore design the gate and feed path with a modulus equal to or greater than the adjacent hot spot, and I avoid sharp changes that cut off the feed path.

Correct Gating System Design

In investment casting, the gating system usually includes a pouring cup, sprue, runner, and internal gate. Sometimes I add a slag trap, buffer, or vent. The pouring cup receives the liquid and establishes pressure for filling and feeding. Its top is normally kept at a suitable distance above the pattern, often in the range of $$70$$ to $$100 \, \text{mm}$$. The sprue supports the shell and often acts as a feeder. To ensure feeding, I commonly make the sprue cross-sectional area about $$1.4$$ times the total internal gate area. The runner distributes metal, traps slag, and can assist feeding. If it acts as a feeder, its area is often $$1.1$$ to $$1.3$$ times the sprue area. If it does not act as a feeder, its area may be $$0.7$$ to $$1.0$$ times the sprue area.

Gating element Main functions Typical area relationship
Pouring cup Receive metal and establish pressure Height above pattern: $$70$$ to $$100 \, \text{mm}$$
Sprue Support shell; often acts as feeder Area about $$1.4 \times$$ total internal gate area
Runner as feeder Distribute metal, trap slag, feed Area about $$1.1$$ to $$1.3 \times$$ sprue area
Runner not as feeder Distribute metal and trap slag Area about $$0.7$$ to $$1.0 \times$$ sprue area
Internal gate Control filling and feeding into casting Based on hot-spot modulus or diameter

Internal Gate Design

The internal gate is the most critical part of the gating system. It influences mold filling, solidification, feeding, residual stress, shrinkage, hot tearing, distortion, cleaning, machining, and surface quality. I consider its position, number, shape, length, and cross-sectional size. From the filling viewpoint, I avoid directing metal against a core or a weak mold section. I avoid jetting, splashing, and eddy currents. I avoid local overheating that can soften or deform the mold. From the feeding viewpoint, I place the gate at the casting hot spot so that directional solidification can occur toward the feeder.

For the number of internal gates, I generally prefer one gate per feeding region. If the casting is complex, has dispersed hot spots, or requires high soundness, I may use two or more gates. For shape, I use rectangular, circular, square, or fan-shaped gates depending on the local geometry. For length, I often use $$8$$ to $$12 \, \text{mm}$$ for easy-break gates and $$10$$ to $$15 \, \text{mm}$$ for gas-cut gates. For cross-sectional size, I use two main methods. The first is the hot-spot circle ratio method:

$$ D_i = K D_h $$

where $$D_i$$ is the internal gate diameter or equivalent diameter, $$D_h$$ is the hot-spot circle diameter, and $$K$$ is usually $$0.7$$ to $$0.9$$. The second is the equivalent hot-spot method. I convert the casting hot spot into one or more simple units, such as a cylinder, make the unit match the casting weight and hot-spot solidification modulus, and then use the same ratio method to determine the gate size.

If the internal gate is too large, its neck remains liquid until the very end. At that time there is no extra liquid to feed it, so a shrinkage cavity forms at the junction between the gate neck and the casting. If the internal gate is too small, it freezes early and cannot feed the hot spot. I therefore treat internal gate design as a balance between feed path life and final solidification location.

Internal gate parameter Practical rule I use Reason
Position At hot spot or feeding direction Promotes sequential solidification
Number Usually one; more if hot spots are dispersed Avoids interference and improves feeding
Shape Rectangular, circular, square, or fan Matches local filling and feeding needs
Length $$8$$ to $$12 \, \text{mm}$$ easy-break; $$10$$ to $$15 \, \text{mm}$$ gas-cut Balances cleaning and feeding
Cross-section $$D_i = K D_h$$, with $$K = 0.7$$ to $$0.9$$ Relates gate size to hot-spot modulus

Improving Casting Structure

When I review a casting design, I first analyze the structure and then add machining allowance and draft as a preliminary design. Next I identify the thermal centers and divide the casting into feeding regions. If feeding is difficult, I adjust the process allowance by locally thickening or thinning the casting. I try to make wall thickness uniform and use generous transitions and fillets, but not so generous that they create new hot spots. When these measures are not enough, I use padding. Padding means adding thickness near the feeder to establish a temperature gradient toward the feeder. It increases the feed channel and improves the feeding ability of the sprue or feeder. For top-poured castings, the padding width usually increases with height. For a given height, a thinner wall requires a wider feeding channel. However, padding increases machining allowance, and excessive padding reduces the economic advantage of investment casting. I also consider padding in evaporative pattern casting, but the pattern and coating must be adjusted accordingly.

Reducing Alloy Shrinkage

I try to select an alloy with low liquid and solidification contraction when the service conditions permit. Once the alloy and casting structure are fixed, the freezing range, thermal conductivity, crystallization temperature, latent heat, and wall thickness are largely set. However, I can still control mold material, mold preheat, pouring conditions, and post-pour cooling. These variables change the temperature gradient and the feeding behavior. Choosing the correct pouring temperature is a balance:

$$ \epsilon_{V,L} = \alpha_{V,L} (T_p – T_L) \times 100\% $$

If the pouring temperature is too high, liquid contraction increases and the cavity can grow if feeding is insufficient. If the pouring temperature is too low, the gating system freezes early and feeding stops, which also increases the cavity. I therefore determine pouring temperature together with mold temperature. A higher mold temperature can improve feeding by keeping the gate liquid longer, but it also slows solidification and may increase porosity in wide-freezing-range alloys. In evaporative pattern casting, mold and sand temperature also affect pattern decomposition and gas pressure, so I keep the pouring window narrow and well controlled.

Variable Increase effect Decrease effect My control strategy
Pouring temperature More liquid contraction, better fluidity Less liquid contraction, risk of early freezing Optimize with feeder design and mold preheat
Mold preheat temperature Slower cooling, longer feed time Faster cooling, shorter feed time Balance for alloy and section thickness
Superheat Higher liquid contraction Lower liquid contraction Use minimum superheat that still fills and feeds
Gas content More porosity, less apparent shrinkage Cleaner metal, more concentrated shrinkage if unfed Deoxidize and control melt cleanliness

Deoxidation and Gas Control

I pay close attention to deoxidation because gas evolution during solidification changes the shrinkage behavior. If deoxidation is incomplete, gas may be released during freezing. If the gas cannot float or enter an existing cavity, it remains as small pores between dendrites. This reduces the apparent solidification volume contraction, but it expands the porosity zone. The result is not a sound casting; it is a casting with distributed porosity. I therefore deoxidize thoroughly and control gas content, especially in steels. In evaporative pattern casting, gas from the pattern adds another source of gas that must be managed by coating permeability, sand permeability, and venting.

Pouring Practice and Filling Control

I select pouring practice based on alloy, casting geometry, mold type, and feeding design. I use high-temperature tapping and low-temperature pouring when possible. For large clusters, I sometimes use a fast-slow-refill sequence. For top pouring, I use higher pouring temperature and slower pouring speed. For bottom pouring, I use lower pouring temperature and faster pouring speed. These practices help fill the mold without excessive turbulence while preserving enough heat for feeding. In evaporative pattern casting, the pour must also manage pattern decomposition, so I often use a controlled pour rate that matches gas evolution and coating permeability.

Local Cooling and Chills

I improve local heat extraction when a corner, junction, or heavy section stays liquid too long. I position chills in the mold or shell to accelerate local cooling and reduce the hot-spot modulus. For investment casting shells, I prefer brass or stainless steel chills because carbon steel may oxidize during shell firing and bond poorly. For evaporative pattern casting, I can use external chills in the sand or internal chills in the pattern, but I must ensure that the pattern decomposition and coating process are not disrupted. The goal is not to cool the whole casting uniformly, but to create a favorable thermal gradient toward the feeder.

Pressure Head and Feeding Distance

Increasing the pressure head accelerates liquid flow and improves feeding. However, the effective feeding distance does not increase without limit. As section thickness increases, the relative effective feeding distance becomes shorter. A very tall feeder may only feed the lower part, leaving a secondary cavity higher in the casting. I therefore use insulating feeders, exothermic covers, and feeding aids. An exothermic cover reduces or delays the freezing of the top surface of the feeder and prevents a secondary cavity. When necessary, I switch to centrifugal casting or pressure-assisted pouring. In evaporative pattern casting, the same principles apply, but the pressure head must also overcome the resistance caused by pattern decomposition and coating.

Prevention method What it changes Best application
Correct gating design Feed path and solidification sequence All castings
Uniform wall design Hot-spot size and location Complex structures
Padding Feeding channel and thermal gradient Castings with isolated heavy sections
Low-shrinkage alloy Liquid and solidification contraction When service conditions allow
Controlled pouring Superheat, fill time, gas evolution All processes, especially evaporative pattern casting
Deoxidation Gas porosity and solidification behavior Steel and reactive alloys
Chills Local cooling rate and modulus Corner and junction hot spots
Insulating or exothermic feeders Feeder life and pressure head Large hot spots and thick sections

A Practical Case from My Experience

I once worked on a critical investment-cast component weighing about $$16 \, \text{kg}$$. The service life requirement was more than ten thousand meters of advance, but in actual use the part failed after only three thousand to four thousand meters. The failure was not caused by the base alloy alone. Inspection showed severe shrinkage cavities and axial porosity on the wear surface. These defects reduced wear resistance and caused early failure. This is a common pattern I have seen: a small internal shrinkage defect can control the life of a heavily loaded part.

I decided to modify the alloy with a rare-earth addition. Rare-earth elements can improve liquid fluidity, increase feeding ability, reduce shrinkage cavities and porosity, and improve casting process performance and mechanical properties. They also purify, modify, and alloy the steel. I selected a rare-earth-modified steel with the following composition range.

Element Content, wt%
C 0.15 to 0.25
Si 0.40 to 0.75
Mn 0.90 to 1.20
Cr 1.10 to 1.40
Mo 0.20 to 0.30
P ≤0.04
S ≤0.04
Re addition 0.15

I tested three addition methods in a $$500 \, \text{kg}$$ coreless induction furnace with an acid lining. The results showed that the ladle-injection method gave the best recovery and the most stable rare-earth residual. I recorded the following comparison.

Addition method Rare-earth recovery, % Rare-earth residual, %
Furnace insertion Decreases with pouring time Decreases with pouring time
Furnace addition 15 to 25 0.02 to 0.04
Ladle injection 50 to 80 0.08 to 0.12

I estimated the addition amount using a mass-balance approach. A simplified expression is:

$$ \text{Re addition} = 5.85([O]_{initial} – [O]_{residual}) + 2.93([S]_{initial} – [S]_{residual}) + \text{Re residual} $$

This expression is useful, but it does not include rare-earth losses. I therefore consider four requirements: deoxidation, desulfurization, loss, and residual. To find the optimum, I used the same heat and added different amounts of rare-earth alloy by ladle injection: $$0$$, $$0.15\%$$, $$0.30\%$$, and $$0.40\%$$. The test bars were normalized at $$900$$ to $$920^\circ C$$ for one hour, tempered at $$650$$ to $$680^\circ C$$ for two hours, carburized at $$920$$ to $$940^\circ C$$ for fourteen hours, quenched at $$830$$ to $$850^\circ C$$ for twenty minutes, and tempered at $$180$$ to $$200^\circ C$$ for one hour. The mechanical property results are summarized below.

Rare-earth addition, % Tensile strength, MPa, condition 1 Tensile strength, MPa, condition 2 Impact toughness, J/cm², condition 1 Impact toughness, J/cm², condition 2
0 217 212 1.08 3.07
0.15 236 312 1.58 5.17
0.30 243 319 1.63 5.73
0.40 234 288 1.50 3.80

Considering casting quality, cost, and benefit, I selected the ladle-injection method with $$0.15\%$$ rare-earth alloy addition. The rare-earth recovery was about $$50$$ to $$80\%$$, and the residual rare-earth content was about $$0.08$$ to $$0.12\%$$. I then poured test castings and test bars from the same heat, comparing the unmodified steel and the rare-earth-modified steel. During pouring of the modified steel, the rare-earth elements reacted with gases in the liquid to form rare-earth compounds. This reduced gas evolution before solidification and increased feeding ability. The result was fewer gas-related rejects and better density, with much less shrinkage porosity.

Bearing-body type Number poured Gas-related rejects Microstructure after nitric acid etching
With rare-earth addition 15 0 Slight shrinkage porosity visible
Without rare-earth addition 10 2 Severe shrinkage cavity and porosity visible

I also compared chemical composition. The modified steel showed a controlled rare-earth residual after addition, while the unmodified steel had none. The mechanical property tests showed that the rare-earth-modified steel had higher bending strength, much higher impact toughness, and better wear resistance. The true-carburized case and the false-carburized core both improved. The wear test results are shown below.

Material Hardness, HRC Revolutions Mass loss, g Dimensional reduction, mm
Unmodified 59 to 61 5,000 0.2087 0.0825
Unmodified 59 to 61 20,000 1.7770 0.3360
Unmodified 59 to 61 40,000 3.7872 0.8810
Rare-earth modified 60 to 62 5,000 0.1557 0.0500
Rare-earth modified 60 to 62 20,000 1.2368 0.2250
Rare-earth modified 60 to 62 40,000 3.2678 0.7170

The wear test showed that the rare-earth-modified material improved wear resistance by about $$10$$ to $$20\%$$. In field tests, two modified bearing bodies both exceeded ten thousand meters of advance, more than twice the life of the original unmodified part. Later, I further reduced the wall thickness and machining allowance. The combination of rare-earth modification, optimized feeding, and reduced machining stock increased the service life to more than three times the original. This experience reinforced my belief that shrinkage control is not only a foundry issue; it is a life-cycle and economic issue.

How Evaporative Pattern Casting Changes My Shrinkage Strategy

Although my main examples are investment castings, I always keep evaporative pattern casting in mind because the same shrinkage mechanisms appear in a different thermal and gas environment. In evaporative pattern casting, the foam pattern is decomposed by the liquid metal. The decomposition products must escape through the coating and sand. If they do not, gas pressure can build up and oppose liquid feeding. This can create porosity that looks like shrinkage porosity but has a gas component. I therefore adjust my strategy in several ways. I increase coating permeability where needed. I control sand compaction so that gas can escape without mold collapse. I avoid excessive pouring temperature, which increases gas generation. I design the gating system to provide a stable feed path. I use vents and, when necessary, chills to control the final solidification location. In other words, evaporative pattern casting requires the same sequential solidification principle as investment casting, but with stricter gas management.

I summarize the comparison between investment casting and evaporative pattern casting as follows.

Factor Investment casting Evaporative pattern casting
Mold filling Empty preheated shell Pattern decomposes as metal enters
Gas source Mostly mold and metal gas Pattern decomposition gas plus mold and metal gas
Cooling control Shell thickness, preheat, insulation Sand, coating, pattern, gas evacuation
Feeding resistance Gate and shell geometry Gate, coating, sand, decomposition gas
Shrinkage risk Hot spots and blocked feed paths Hot spots, gas pressure, and blocked feed paths
Main remedy Sequential solidification and gating design Sequential solidification, gating design, and gas control

My Calculation Framework for Feeding Design

When I design feeding for a new casting, I start with geometry and then refine with thermal calculations. For a simple rectangular section, I calculate the modulus as:

$$ M = \frac{abc}{2(ab + bc + ac)} $$

where $$a$$, $$b$$, and $$c$$ are the dimensions of the rectangular region. For a cylinder of diameter $$D$$ and length $$L$$, ignoring the ends, I use:

$$ M \approx \frac{D}{4} $$

For a plate of thickness $$t$$, if the plate is large compared with its thickness, I use:

$$ M \approx \frac{t}{2} $$

These simple estimates help me compare the hot spot with the feeder. I then apply Chvorinov’s rule:

$$ t_s = B M^2 $$

and require the feeder to freeze after the casting hot spot. If the feeder modulus is not large enough, I increase the feeder size, add insulation, or add a chill to the casting hot spot. I also check the feed path modulus. If the feed path modulus is smaller than the hot-spot modulus, the path freezes first and the feeder is isolated. I therefore design the path with a modulus equal to or greater than the adjacent hot spot.

Geometry Modulus formula Practical note
Rectangular block $$M = \frac{abc}{2(ab+bc+ac)}$$ Use for simple hot-spot estimates
Cylinder $$M \approx \frac{D}{4}$$ Use when length is much greater than diameter
Large plate $$M \approx \frac{t}{2}$$ Use for wall-like sections
Feeder $$M_f > M_c$$ Feeder must freeze after the hot spot
Feed path $$M_p \ge M_c$$ Path must remain open during feeding

My Rules for Avoiding Shrinkage in Production

I reduce shrinkage defects by following a set of production rules that I have refined over many trials. First, I identify every hot spot before I design the gating system. Second, I place the internal gate at the hot spot or in a position that promotes feeding toward the feeder. Third, I make the feeder modulus larger than the casting modulus. Fourth, I keep the feed path open until the hot spot has solidified. Fifth, I avoid abrupt section changes and isolated heavy sections. Sixth, I control pouring temperature and speed so that the metal fills completely but does not overheat the mold. Seventh, I deoxidize the melt thoroughly. Eighth, I use chills for local hot spots that cannot be fed easily. Ninth, I use insulating or exothermic feeders for large hot spots. Tenth, I verify the design by dissection, radiography, or simulation. In evaporative pattern casting, I add an eleventh rule: I manage pattern decomposition gas by controlling coating permeability, sand permeability, and venting.

Why Sequential Solidification Can Create Other Problems

I must also remember that sequential solidification can create large temperature differences within the casting. These differences can produce residual stresses, distortion, or cracks. A strong directional gradient is good for feeding, but it is not always good for stress control. I therefore balance feeding and stress management. Sometimes I use a less aggressive gradient, a stress-relief heat treatment, or a modified cooling schedule. In evaporative pattern casting, the sand can provide some support and slow cooling, which may reduce stress but can also increase porosity if the alloy has a wide freezing range. I always evaluate the whole system rather than optimizing one defect at the expense of another.

Using Simulation and Experimental Validation

I use simulation as a guide, not as a replacement for experience. A good casting process analysis system lets me observe mold filling and solidification, predict shrinkage cavity and porosity, and test gating designs before cutting tooling. I still validate the simulation with trial castings. I cut sections, inspect with nitric acid etching, use radiographic testing, and measure density. In evaporative pattern casting, I also inspect gas defects and coating performance because the pattern decomposition can change the apparent shrinkage pattern. Simulation is especially useful for comparing alternatives, but the final proof is always a real casting that meets service requirements.

Validation method What I learn Limitation
Solidification simulation Hot spots, last liquid, feeding paths Depends on accurate properties
Sectioning and etching Actual cavity and porosity distribution Destructive
Radiographic inspection Internal voids and porosity Resolution and orientation effects
Density measurement Overall soundness Not a full map of defects
Service testing Real performance and life Expensive and time-consuming

My Summary of Causes and Remedies

I see shrinkage cavity and shrinkage porosity as predictable consequences of poor feeding during the last stages of solidification. The main causes are poor gating design, unfavorable casting geometry, high alloy shrinkage, improper pouring conditions, poor mold cooling, poor local heat extraction, and insufficient pressure head. The main remedies are directional solidification, correct gating and feeder design, improved casting structure, suitable alloy and pouring practice, thorough deoxidation, local chilling, and adequate feeding pressure. These remedies are interconnected. If I change one, I must check the others. This is true in investment casting, and it is equally true in evaporative pattern casting, where gas evolution and pattern decomposition add another layer of complexity.

Key Relationships I Keep in Mind

I keep the following relationships close at hand when I evaluate a shrinkage problem:

$$ \epsilon_{V,L} = \alpha_{V,L} (T_p – T_L) \times 100\% $$

$$ \epsilon_{V,S} = \alpha_{V,S} (T_S – T_R) \times 100\% $$

$$ \epsilon_L = \alpha_L (T_S – T_R) \times 100\% $$

$$ \epsilon_{V,total} = \epsilon_{V,L} + \epsilon_{V,C} + \epsilon_{V,S} $$

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

$$ t_s = B M^n $$

$$ t_{s,f} > t_{s,c} $$

$$ M_f > M_c $$

$$ D_i = K D_h,\quad K = 0.7 \text{ to } 0.9 $$

$$ \text{Re addition} = 5.85([O]_{initial} – [O]_{residual}) + 2.93([S]_{initial} – [S]_{residual}) + \text{Re residual} $$

These equations are simple, but they capture the core of the problem. The liquid contracts before freezing, the solidification event creates a volume deficit, and the solid then contracts as it cools. The only way to avoid a cavity in the casting is to provide liquid metal to the last solidifying region until freezing is complete. If I cannot do that, I must move the last solidifying region into the gating system or feeder. This is the essence of sequential solidification, and it is the foundation of my prevention strategy in both investment casting and evaporative pattern casting.

Final Practical Observations

In my experience, the most successful shrinkage-control programs combine design, process, and material control. Design control means hot-spot identification, uniform walls, gradual transitions, and proper feeding. Process control means gating, pouring, mold cooling, and local chilling. Material control means alloy selection, deoxidation, gas control, and sometimes rare-earth modification. I have seen rare-earth modification improve fluidity, feeding, and wear resistance, but I have also seen that it cannot compensate for a fundamentally bad gating design. I have seen simulation predict a hot spot correctly, but I have also seen actual pouring produce a different result because of mold permeability or gas pressure. Therefore, I always combine calculation, simulation, trial casting, and service testing.

When I work with evaporative pattern casting, I add gas management to every step. I check coating permeability, sand permeability, pattern density, pouring rate, and venting. I know that the decomposition products can create back pressure and that this back pressure can stop feeding even when the modulus calculations say the feeder is adequate. I also know that the same back pressure can cause gas porosity that resembles shrinkage porosity. My solution is to keep the filling and solidification path open, control gas evolution, and maintain a clear feed path to the last liquid region.

I have learned that shrinkage defects are not random. They follow heat flow, feeding paths, and solidification time. If I control those three factors, I can control shrinkage. If I ignore them, I will find cavities and porosity in the last places to freeze. This rule has held true for me in investment castings, and I have no doubt that it holds true in evaporative pattern casting as well. The details differ, but the physics does not.

Scroll to Top