You are on page 1of 3

with a simple theoretical derivation given in the Appen-

Discussion dix of their article. The derivation assumed that shrink-


age formation was controlled by the pressure drop
Discussion of ‘‘Prediction of Shrinkage associated with the flow of liquid through the semisolid
portion of the casting. Carlson and Beckerman made a
Pore Volume Fraction Using a similar, more sophisticated derivation, using the same
Dimensionless Niyama Criterion’’* fundamental assumption.
No one is questioning the validity of the results
plotted in Figure 1 and represented by Eq. [2]. However,
GEOFFREY K. SIGWORTH several important questions remain:
(1) What is the physical significance of the Niyama
DOI: 10.1007/s11661-009-0056-2 criterion?
Ó The Minerals, Metals & Materials Society and ASM (2) To what extent can the Niyama criterion, or a
International 2009 modified form of it, be used in other alloy systems
having different thermal and solidification charac-
teristics?
(3) Can we use the Niyama criterion to predict the
Carlson and Beckermann[1] recently proposed a mod- amount of porosity that forms at any particular
ified, ‘‘dimensionless’’’ form of the Niyama criterion. location in commercial castings?
Although this appears to be a useful approach, there are
problems with their analysis. These are important questions, which formed the
It will be convenient to first consider the original impetus for the study by Carlson and Beckermann and
experiments of Niyama et al.[2] Their results were which they attempted to answer in their article.
obtained from steel castings of varying sizes, up to Regarding the first point, both Niyama et al.[2] and
68 tons in weight. The soundness at various locations Carlson and Beckerman assumed that formation of
was compared to the gradient at the end of solidifica- shrinkage porosity was governed by the pressure drop
tion, as calculated from computer simulations. When associated with metal flow inside the semisolid casting.
the thermal gradient in the feeding direction was greater Is this correct? Probably not.
than a certain value, no shrinkage was observed in steel Sigworth and Wang conducted a detailed theoretical
castings. That is, the castings were ‘‘sound.’’ However, study of this problem[3,4] and showed that the pressure
the gradient required to avoid shrinkage depended on drop associated with metal flow through the semisolid
the square root of the local solidification time. Figure 1 region is extremely small. The plot in Figure 2 is taken
summarizes their results and has been reproduced from from Reference 4. The top, nearly horizontal line shows
their article. the ‘‘mechanical’’ pressure inside the casting during
From pffiffiffiffi
the preceding plot, we see that when solidification. At the beginning of solidification, the
G  32= tS ; there was no shrinkage in their castings. pressure in the liquid is equal to the prevailing atmo-
The solidification time (tS) may be approximated by spheric pressure, plus a small additional pressure pro-
 vided by metal height in the riser of the casting (a total
tS ffi DTf T_ ½1 of 1.1 atm.). At 99 pct solidification, this pressure
decreased very slightly, to 1.04 atmospheres. The diag-
where onal lines show the ‘‘chemical’’ pressure of hydrogen
tS = local solidification time (in minutes), dissolved in the melt, for different gas contents at the
DTf = freezing range of the alloy (in degrees beginning of solidification. The chemical pressure
Celsius), and increases because of the segregation of dissolved hydro-
T_ = cooling rate at the end of solidification (in gen into the liquid as the metal freezes.
o
C/min). Gas or shrinkage porosity may form when the lines
for the chemical and mechanical pressure cross. This
Thus, the Niyama criterion (Ny) becomes plot shows that the pressure drop associated with metal
pffiffiffiffi flow in solidifying castings cannot explain the results
32 T_ G 32
G  pffiffiffiffiffiffiffiffi or Ny ¼ pffiffiffiffi  pffiffiffiffiffiffiffiffi ½2 found by Niyama et al.[2]
DTf T_ DTf Instead, the Niyama results are better described by
using a ‘‘geometric’’ model of the solidification process.
where G is the thermal gradient in the feeding direc-
This model is shown in Figure 3, which illustrates
tion (having units of degrees Celsius per centimeter).
schematically the cross section of a solidifying plate
This was the form of the equation offered originally by
casting. The dark area in the center of the plate
Niyama et al.[2] They justified this functional dependence
represents a tapered pool of liquid metal. The taper is
described by the angle h. As long as this angle is greater
than some critical value, feeding occurs.
GEOFFREY K. SIGWORTH is with GKS Engineering Services, Equation 19 of Reference 4 describes the functional
Rockdale, TX 76567. Contact e-mail: gksigworth@gmail.com
*K.D. CARLSON and C. BECKERMANN: Metall. Mater. Trans. dependence of the geometric model on the angle h and
A, 2009, vol. 40A, pp. 163–75. the relevant thermal constants for the mold and the
Discussion submitted July 22, 2009. metal. If one assumes that the critical angle for feeding

METALLURGICAL AND MATERIALS TRANSACTIONS A


100
50

Sound
G (deg C/cm)

10
5

Shrinkage
1
0.5
Low C steel

0.5 1 5 10 50 100 1000


Solidification Time
(minutes)
Fig. 3—Schematic illustration of the geometric model.
Fig. 1—Thermal gradient required to avoid shrinkage in steel cast-
ings (reproduced from Ref. 2).
Table I. Dimensionless Freezing Range (/) of Sand
Castings[4]

Alloy Value of /
0.3 pct C steel 1.8
85Cu-5Sn-5Pb-5Zn 28
Al-7Si-Mg 50
88Cu-8Sn-4Zn 92
Al-4.5 pct Cu 133
Al-7Mg 140
Mg-6Al-3Zn 144

Fig. 2—Plot of mechanical and chemical pressures during the direc-


tional solidification of A356 alloy.[4]

(h) is 1 deg, it is possible to derive a curve that falls


exactly on top of the line shown in Figure 1.
The logic of the pressure drop calculations has been
so compelling that it has seduced scientists (myself
included) for many years. However, when one makes
exact numerical calculations, as in Figure 2, it is seen
that this approach cannot explain the formation of
shrinkage porosity in steel castings.[4] The apparent
explanation, a simple geometric criterion, was over-
looked.
We now pass onto the second question: ‘‘To what
extent can the Niyama criterion, or a modified form of
it, be used in other alloy systems having different
thermal and solidification characteristics?’’
This question was considered at length in Reference 4 Fig. 4—Volume fraction porosity found in A356 alloy castings
and in a follow-on experimental study.[5] It was con- (reproduced from Ref. 6).
cluded that Niyama’s criterion and the geometric model
do not apply to alloys having a long freezing range. This
is best illustrated by considering the dimensionless the average temperature difference between the surface
freezing range (/) given by Eq. [21] of Reference 4. and center of the casting. It is also equal to the effective
This is equal to the freezing range of the alloy divided by length of the mushy region compared to the thickness of

METALLURGICAL AND MATERIALS TRANSACTIONS A


the mold. Values of / are given in Table I for sand porosity. In my experience, porosity formation is
castings of several different alloys. controlled primarily by freezing rate and, to a lesser
The Niyama criterion describes the tendency for extent, by gas content. This was discussed at some
shrinkage formation, by calculating only the local length in References 3 and 4; however, it will also be
thermal conditions at each point in the casting. This is useful to consider some of the results given by Fang and
reasonable in steel castings because of their narrow Granger.[6] From Figure 4, we see that the solidification
freezing range. In alloys that have a long freezing range rate has an overriding influence on the amount of
and higher thermal conductivity, as in many aluminum- porosity in a casting. Castings that freeze quickly can
and copper-based alloys, the situation is vastly different. tolerate high contents of gas. Slowly cooled castings,
Consider the popular Al-7Si-Mg casting alloy. If we however, easily form significant amounts of porosity.
have a sand casting 1-in. thick, the ‘‘mushy’’ semisolid
portion of the casting will be 50-in. long. Thus,
questions such as the following become important:
(1) Is the riser sufficiently liquid to ‘‘feed’’ the casting?
(2) Is there a continuous feed path between the riser
and the portion of casting under consideration? REFERENCES
In other words, we need to consider conditions 1. K.D. Carlson and C. Beckermann: Metall. Mater. Trans. A, 2009,
throughout the entire casting, not just at local condi- vol. 40A, pp. 163–75.
2. E. Niyama, T. Uchida, M. Morikawa, and S. Saito: AFS Int. Cast
tions existing at any single point. Met. J., 1982, vol. 7 (3), pp. 52–63.
It should be noted that the values for dimensionless 3. G.K. Sigworth and C. Wang: AFS Trans., 1992, vol. 100, pp. 979–
freezing range (/) depend on the casting process used. In 87.
castings made with metal molds, the value of / may be 4. G.K. Sigworth and C. Wang: AFS Trans., 1992, vol. 100, pp. 989–
10 to 100 times less than the values given in Table I. 1004.
5. G.K. Sigworth, C. Wang, H. Huang, and J.T. Berry: AFS Trans.,
Regarding the third and final question posed previ- 1994, vol. 102, pp. 245–61.
ously, I do not believe the Niyama criterion (in any 6. Q.T. Fang and D.A. Granger: AFS Trans., 1989, vol. 97, pp. 989–
form) should be used to estimate the amount of 1000.

METALLURGICAL AND MATERIALS TRANSACTIONS A

You might also like