Shrinkage in Nodular Iron

With increasing complexity in casting geometry and continued stringent requirements for completely sound castings, understanding and predicting the shrinkage behavior of ductile cast iron parts is all the more crucial for successful foundry operations.

Four distinct regions can be isolated when observing ductile iron solidify.

1. Liquid contraction from the superheat temperature to the liquidus. This contraction is very predictable since it is dependent on the coefficient of expansion of the alloy (generally around 1.5% by volume per 100^oC).
2. Liquid shrinkage through the liquidus temperature. A phase change takes place at this juncture. A portion of the liquid iron transforms to solid austenite. Occasionally for highly hypereutectic irons graphite precipitates at the liquidus instead of austenite, resulting in expansion rather than contraction.
3. Eutectic expansion follows the liquidus. The remaining liquid transforms into austenite and graphite. Expansion always occurs during the eutectic transformation and it is very significant. This is because all of the carbon in the liquid iron minus the carbon dissolved in the austenite precipitates as graphite during the eutectic. The volume fraction of graphite (in the eutectic) that precipitates can be calculated using the lever rule. For an iron with a typical 3.65% carbon (Co =3.65%) the fraction percent of graphite in the eutectic is as follows:
   G/G+g = Co-Cg/CG-Cg = (3.65-1.90)/(100-1.90) = 1.78%
   The eutectic consists of 98.22% austenite and 1.78% graphite by weight. The amount of carbon dissolved in the austenite is roughly 1.90%. Therefore of the 3.65% compositional carbon, 1.87% is dissolved in the austenite and 1.78% precipitates, hopefully, as graphite.
   Graphite has a much higher specific volume compared to iron causing the expansion that is observed. The density of graphite is 2.2 g/cc compared to 7 g/cc for that of iron. 
4. Solid contraction is also dependent on the expansion coefficient.

These changes are depicted schematically in Fig.1 for three different irons.

The following should be noted:

1. All three irons undergo a net expansion during solidification. The volume of the solidified iron at the end of solidification (before solid contraction) is greater than the volume of the liquid poured into the mold!
2. Hypereutectic ductile irons have been measured to exhibit volumetric expansion as high as 4%.
3. For the same carbon equivalent ductile will expand more than gray.
4. Feed metal must be supplied by risers and/or the gating system for all cast irons in zone A. Additional feed metal must be provided in zone B for hypoeutectic irons.
5. The reason eutectic expansion cannot be effectively utilized to compensate for earlier contraction and shrinkage is that green sand mold walls dilate (move outward) when subject to the enormous expansion forces. Note (in Fig. 2) that at the end of solidification when the metal contracts the mold wall stays at its maximum dilated position.

Solidification Mechanisms: Cast iron solidification is very different from that of a pure metal. Pure metals solidify with a solidification front that is very well defined and a clearly delineated solid liquid interface. Ductile cast iron solidification, on the other hand, is characterized by a very thin solidified skin and if conditions are not optimal a large mushy zone. The outer skin formed during gray cast iron solidification is much heavier than that of ductile. Flake graphite is a better conductor of heat compared to nodular. The heavier skin prevents the transmission of the eutectic expansion forces to the mold walls. This is the reason why gray irons need less risering than ductile even though ductile iron solidification results in a larger net expansion.

The width of the mushy zone and the aspect ratio of the austenite dendrites have been linked to the feeding capability of the riser. Generally short stubby dendrites in a narrower mushy zone will produce better feeding characteristics. Narrower mushy zones are obtained when nodular iron solidifies as a eutectic with very little separation between the liquidus and eutectic temperatures. Austenite that precipitates during the liquidus tends to grow much larger in size. Finer eutectic austenite is also believed to improve feeding capability and to be associated with higher nodule counts. Most foundry engineers have to rely on experience or guess at how far a particular riser will feed. Even though research has produced test patterns that can evaluate feeding distances, very few foundries take the time to evaluate this key variable. The problem is compounded particularly since the mushy zone changes from tap to tap depending on the metallurgy and quality of the iron. Therefore the feeding distance itself is a function of the metallurgical integrity of the iron.

Comparative Solidification Schematic - Fig. 3

 For the purposes of this paper shrinkage will be divided into four categories:

1. Pull downs or suckins.
2. Macro shrink larger than 5 mm
3. Micro shrink or shrinkage porosity less than 3 mm
4. Microscopic grain boundary shrinkage. Generally only visible under a microscope at a magnification greater than 100X.

Fig. 4

The current paper will focus on the first three types only. These defects occur at very different and distinct times during solidification as depicted in Fig.5.

Thermal analysis is probably the strongest tool available in the foundry man's arsenal to understand and combat shrinkage defects.

For example a high value for the area S1 is associated with a lot of primary austenite and a large mushy zone and therefore with an iron that is more likely to produce pull down and macro shrinkage upon solidification.

In fact large variations in S1 have been observed from treatment batch to batch (before post inoculation) in the same foundry on the very same day. Base iron holding time appears to be the single most dominant variable contributing to this deviation. Strong post inoculation appears to mitigate the variance in S1.

Pull downs or suckins are produced very early in solidification. The skin formed at the top cope surface is extremely thin. If feed metal is not provided then contraction will cause a negative pressure just below the skin. The atmospheric pressure then pushes the wall inward producing the "pull down" or "outer sunk" defect

Macro shrink generally appears a little later. The skin formed is thick enough and will not cave in. The negative pressure consequently produces rather large shrink holes. If this defect appears at the riser contact or inside the casting cavity relatively close to the riser (as it generally does) proper risering technique can and should be utilized to solve the problem.

The first observation when trouble-shooting macro shrink should be "Did the riser pipe?" The remedies applied are very different depending on whether the riser piped or not.

If the riser piped properly then possible solutions are:

1. Increase riser size
2. Check carbon equivalent. It may be too low 
3. Lower pouring temperature

However, if the riser did not pipe then the analysis is not as straight forward and the following are recommended:

1. Reduce riser contact modulus. The contact modulus may be too large keeping the contact open during the casting eutectic expansion leading to back feeding.
2. Reduce the modulus of the ingate feeding the riser. If the ingate stays open too long initial feed metal will be delivered to the casting cavity from the gating system rather than the riser. The top of the riser will then freeze off preventing proper piping. Conical risers are particularly vulnerable to this phenomenon. 
3. Check carbon equivalent. It may be too high
4. There may be too many risers present
5. Pouring temperature may be too cold

If macro shrink appears infrequently and intermittently (comes and goes) and still within the known limit of the risers feeding capability, then variations in metallurgical integrity (larger mushy zone and S1 inhibiting feeding) or poor sand compaction with soft molds are more than likely the culprits particularly if the chemistry checks out OK. From a chemistry point of view, hypoeutectic irons (both gray and ductile) are far more susceptible to macro shrink and outer sunks. A large separation between liquidus and eutectic (as would be expected with hypoeutectic irons) produces a lot more primary austenite thereby reducing the riser's ability to feed. In ductile irons, which tend to be hypereutectic except when pouring very heavy sections, it is desirable for the casting to freeze as a eutectic alloy i.e. with the liquidus arrest as close as possible to the eutectic. Generally when the liquidus appears at a much higher temperature from that of the eutectic, primary austenite is precipitating from the melt even though the chemical composition is hypereutectic. In ductile irons this happens because of the strong undercooling effects of elements such as magnesium and rare earths. Furthermore, highly oxidizing conditions in the melt coupled with high melting temperatures and long holding times reduce the carbon activity causing a chemically hypereutectic iron to solidify as if it were hypoeutectic.

Micro shrinkage porosity appears very late in solidification. At this stage feed paths are well closed. This type of shrink commonly appears on isolated bosses or outside the riser's ability to feed. The only possibility to obtain sound castings is to rely on late eutectic graphite precipitation, with its inherent expansion, to "fill in" the shrinkage voids. Eutectic solidification patterns where most of the graphite comes out early are undesirable.
A uniform precipitation pattern is preferred. A good thermal analysis program can help measure such variables.

Since it is helpful to have graphite come out late then, by definition, a microstructure with varying nodule sizes (nodule bifurcation) or a bi- modal nodule size distribution will be less likely to produce micro shrink. Graphite that comes out early in the eutectic will grow to a larger size when compared to that of graphite that precipitates toward the end of the eutectic, since the late graphite will not have sufficient time for growth. 
Care must be taken when evaluating structures since one is viewing a three-dimension picture in 2D. The size of any given nodule will not only depend on the nodule size but also where the nodule happened to be sectioned. Furthermore, great care should be taken, when making such analysis, that the bimodal distribution is not due to pre-eutectic graphite precipitation. Pre-eutectic arrests associated with exceedingly hypereutectic irons can also exhibit a bi-modal distribution. Graphite that precipitates during the liquidus generally ends up much larger in size than the eutectic graphite. This is generally an undesirable outcome. Therefore thermal analysis curves should be viewed concurrently with the microstructure. Furthermore, several late solidification phenomena can also be evaluated from the cooling curves. These will not be discussed in this paper other than to add that they are invaluable in determining the amount of graphite that precipitates late in the eutectic and therefore the susceptibility of the iron to micro shrinkage defects.

General Foundry Practice: There can be no substitute for good common sense foundry practice. Avoid super heating, long holding times, oxidized charge materials and poorly compacted soft molds. Keep carbon as high as possible, silicon maintained at the lower end of normal operational ranges, appears to reduce shrink defects. Residual magnesium should be maintained at levels to ensure proper nodularity and no higher. Rare Earth elements should be optimized depending on the level of tramp elements such as sulfur, oxygen and bismuth (if added). Inoculant addition should be precisely controlled and the type and quantity should be optimized. Clamping cope and drag molds will help reduce shrink defects. For flask less molding ensure that mold weighting is sufficient.