Simulation of casting solidification

2.5 Shrinkage
2.5.1 General Shrinkage behaviour
There are three quite different contractions to be dealt with, as Figure 2.9
illustrates. As the temperature reduces, the first contraction to be experienced is that in
the liquid state. This is the normal thermal contraction observed by everyone as a
mercury thermometer cools; the volume of the liquid metal reduces almost exactly
linearly with falling temperature. In the casting situation the shrinkage of the liquid
metal is usually not troublesome; the extra liquid metal required compensating for this
small reduction in volume is provided without difficulty, usually without even being
noticed, by a slight extension to the pouring time, or by a slight fall in level in the
feeder. The contraction on solidification is quite another matter, however. This
contraction occurs at the freezing point, because (in general) of the greater density of
the solid compared to that of the liquid. Contractions associated with freezing for

Fig. (2.9) Schematic Illustration of General Shrinkage Behavior [15].

7
 Table 2.1 Solidification shrinkage for some metals [15]

a number of metals are given in Table 2.1 [15].

The contraction causes a number of problems. These include 'feeding', which is
defined as any process which will allow for the compensation of solidification
contraction by the movement of either liquid or solid, and 'shrinkage porosity', which is
the result of failure of the feeding processes to operate effectively. All these are dealt
with at length in this chapter. The final stages of shrinkage in the solid state can cause
quite different problems. As cooling progresses, and the casting attempts to reduce its
size in consequence, it finds itself constrained to some extent by the mould. This
constraint always leads to the casting being somewhat larger than would be expected
from free contraction alone because of a certain amount of plastic stretching which the
casting necessarily suffers. It leads to difficulties in predicting the size of the pattern
since the degree to which the pattern is made oversize (the 'contraction allowance' or

8
'patternmaker's allowance') is not easy to quantify. The mould constraint during the
solid-state contraction can also lead to more localized problems such as hot tearing or
cracking of the casting. [16]
2.5.2 Solidification shrinkage
In general, liquids contract on freezing because of the arrangement of atoms
from a rather open 'random close-packed' arrangement to a regular crystalline array of
significantly denser packing. The densest solids are those which have cubic close-
packed (face-centred-cubic and hexagonal close-packed) symmetry. Thus the greatest
values for contraction on solidification are seen for these metals. Table 2.1 shows the
contractions to be in the range 3.2-7.2 per cent. The solidification shrinkage for the less
closely packed body-centred-cubic lattice is in the range 2-3.2 per cent. The exceptions
to this general pattern are those materials which expand on freezing. These include
water, silicon and bismuth (Table 2.1). [15]
For the majority of materials which do contract on freezing it is important to have
a clear idea of what happens in a poorly fed casting. As an ideal case of an unfed
casting, it is instructive to consider the freezing of a sphere. We shall assume that the
sphere has been fed via an ingate of negligibly small size up to the stage at which a
solid shell has formed of thickness x (Figure 2.10). The source of supply of feed metal
is then frozen off. Now as solidification continues with the freezing of the following
onion layer dx, the reduced volume which the layer dx occupies compared to that of the
original liquid means that either a pore has to form, or the liquid has to expand a little,
and the solid correspondingly has to contract a little. If we assume for the moment that
there is no favourable nucleus available for the creation of a pore, then the liquid has to
accommodate this by expanding, creating a state of tension, or negative pressure. It is in
mechanical equilibrium with the enclosing solid shell, effectively sucking it inwards.
As more onion layers form, so the tension in the liquid increases, the liquid expands,
and the solid shell is drawn inwards. [17]

9
 Fig.2.10. The Solidification Model For An Unfed Sphere [15].

2.6 Solidification Modeling

Solidification modeling can be divided into three separate models, where each
model is identified by the solution to a separate set of equations: heat transfer
modelling which solves the energy equation; fluid-flow modelling which solves the
continuity and momentum equations; and free-surface modelling which solves the
surface boundary conditions (of a surface that is free to move). For a complete
description of solidification modelling, all of these equations need to be solved
simultaneously, but under special circumstances, the equations can be decoupled and
modeled independently. This is the case for heat-transfer modelling which has been
widely used and their applications have significantly improved casting quality. On the
other hand, fluid flow, and in particular free-surface modelling, has until recently been
neglected. Although both have been recognized for their importance in the casting
process, they have been too complex to model commercial castings . Solidification
modelling also requires a great deal of thermophysical data, and since solidification is a
dynamic process, most of this data changes in both time and space. Although in many

10
cases this data can be averaged, or functions estimated, improvements can be made by
modelling the dynamic processes in detail. For example, in cases where the fluid metal
is an alloy it may be important to keep track of the relative compositions by solving the
solute equation. This is known as solute redistribution modelling. In cases where the
solidification of metal is dendritic, it may be important to model the nucleation and
growth of these dendrites. This is called microstructural modelling. Lastly, for metals
that contain dissolved gases, it may be important to model the partial pressure of the gas
in the metal. This is achieved by modelling the concentration of the gas in the metal as
the gas is rejected from the solid. This is called microporosity modelling. [18], [19],
[20].

11