You are on page 1of 14

Paper 05-006(02).

pdf, Page 1 of 14
AFS Transactions 2005 American Foundry Society, Schaumburg, IL USA

Effect of Artificial Cooling on the Metallurgical Quality


of Al-Si-Mg Investment Castings
F. Chiesa
Collge de Trois-Rivires, Trois-Rivires, Qubec, Canada
G. Morin, J. Rousseau
Centre Intgr de Fonderie et de Mtallurgie, Trois-Rivires, Qubec, Canada
Copyright 2005 American Foundry society
ABSTRACT
The experimental results obtained from 6 heats poured into 12 investment shells have been used to assess the influence of
forced air cooling on the solidification and the metallurgical quality of a casting geometry corresponding to solidification
times varying from 1 to 5 minutes in the natural cooling condition. Also, a more conducting shell material was used to verify
whether the shell conductivity was a factor. Tensile properties were measured at 3 locations in the casting, in sections
varying from 6mm in diameter, up to 30x30mm. Secondary dendrite arm spacing and levels of microporosity were measured
at these three locations and compared to the results predicted by solidification modeling.
The Quality Index was used to relate the metallurgical quality of the casting to the requirements of ASTM B686 standard
applying to premium quality aluminum castings. The prediction of the Quality index Q by solidification modeling has been
shown to be a useful practical tool to evaluate the capacity of given casting conditions to deliver the expected level of
Quality for instance, Class 10, 11 or 12 in terms of ASTM B686.
It was found that artificial air-cooling has a notable impact on the metallurgical quality of the solidified casting. It typically
brought about an upgrade of one level of quality in all the sections considered.
INTRODUCTION
In a Silver anniversary paper of the AFS aluminum division, (JACOB, 1999) the concept of the Quality index, Q, and its
practical interest was revisited. Q is calculated from the ultimate tensile strength (UTS) of the alloy and its elongation (E) via
the relationship:
Q = UTS + 150 log E
where Q and UTS are expressed in MPa, and E in percent.
Q is directly tied to the metallurgical quality of the alloy; namely its microstructure fineness (secondary dendrite arm spacing,
DAS), and the level of microporosity present, expressed in % by volume. Q does not depend on the temper conditions of the
heat treated alloy, making it a true marker of the metallurgical quality. On the contrary, the yield strength is not a good
indicator of the metallurgical quality as it depends mainly on the aging time and temperature of the heat treatment and on the
magnesium content of the alloy.
In a recent paper ( CHIESA, 2003), it was shown that the tensile properties required by ASTM standards applying to
aluminum castings could be expressed in terms of the Quality index, Q. For instance, it was shown that the tensile
requirements of ASTM B618-03 Standard Specification for Aluminum Alloy Investment Castings demand a value of Q
above 316 MPa. On the other hand, meeting the tensile requirements of ASTM B686-03 Standard Specification for
Aluminum Alloy Castings, High Strength demands Quality index values of:
Q > 266 MPa everywhere in the casting for class 12 to be achieved
Q > 300 MPa everywhere in the casting for class 11 to be achieved
Q > 367 MPa everywhere in the casting for class 10 to be achieved

Paper 05-006(02).pdf, Page 2 of 14


AFS Transactions 2005 American Foundry Society, Schaumburg, IL USA

Over the past 10 years, inroads have been made into the prediction the microstructure of alloys from the thermal parameters
obtained by solidification modeling. It is now possible to predict the local value of DAS and the level of microporosity from
the knowledge of local parameters such as the solidification time and solidus velocity (for a given gas level in the alloy
poured). It is thus possible to model the Quality index distribution and by way of consequence, to determine whether a
casting poured under known conditions is likely to meet the requirements of a given standard.
METALLURGICAL QUALITY IN Al-Si-Mg CASTING ALLOYS
DROUZY (1980) and al. determined experimentally that, for a clean Al Si7Mg03-06 (A356, 357) melt with a low iron
content (<0.1%), the Quality index Q = UTS + 150 log E is related to the density, d, and dendritic index Id by the family
of lines shown in Figure 1.

Fig. 1. Quality index Q in AlSiMg (A356, 357) alloys


as a function of density and dendritic index.

These lines correspond to the following mathematical relationship:


Q=1830 d 1.04 Id 4430 [1]
Since the density of compact A356 is 2.675, d can be expressed in terms of por%, percent volume microporosity.
d=2.675*(100-por%)/100
The dendritic index Id, as defined by Centre Technique des Industries de la Fonderie (CTIF, 1976), has been shown to be
equal to 1.5 DAS. This allows to re-write relationship [1] in terms of the secondary dendrite arm spacing DAS and the
volume percent microporosity por%.
Q (MPa) = 465 49 por% - 1.56 DAS [1b]
PREDICTION OF DENDRITE ARM SPACING AND MICROPOROSITY
FROM LOCAL THERMAL PARAMETERS
When the local solidification conditions are known, the secondary dendrite arm spacing DAS and the microporosity por%
can be estimated by empirical relationships.
LEE (1990) has shown that the DAS depends on the local solidification time tsl only:
DAS (m) = 38 tsl1/3 [2]
where tsl is the time elapsed between the beginning and end of solidification, in minutes.

Paper 05-006(02).pdf, Page 3 of 14


AFS Transactions 2005 American Foundry Society, Schaumburg, IL USA

As far as the microporosity is concerned, a criteria function FCC has been determined for a reasonably degassed melt, i.e. a
melt containing 0.1 ppm of dissolved [H], or 0.13 ml/100g of Al. This empirical relationship (CHIESA, 1999) was
established for local solidification times between 1 and 13 minutes and is expressed in terms of local solidification time tsl in
min, and solidus velocity, Vs:
por% (FCC) = 0.21 tsl 2/3 Vs1/5 [3]
Vs, expressed in cm/min, increases towards infinity as the gradient approaches zero. FCC has been validated (MAMMEN)
against experimental results published by various researchers on a wide variety of geometrical shapes such as bars (FUOCO,
1996), plates (KAO, 1995 and PAN, 1991) , cubes (MAMMEN, 1999) and unidirectional solidification (EMADI, 1994).
FCC criterion has been introduced as a standard feature in one of the most widely used foundry solidification modeling
software (see SOLIDCast workbook in reference list).
PREDICTING THE QUALITY INDEX Q AND
RELATING IT TO ASTM STANDARDS
From the above relationships [1b] [2] and [3], it can be derived that, for a typical degassed melt (0.1 ppm of dissolved [H], or
0.13 ml/100g of Al) and for local solidification times between 1 and 13 min:
Q (MPa) = 465 10 tsl2/3 Vs1/5 60 tsl1/3 [4]
where tsl is expressed in min, and Vs in cm/s.
On the other hand, the following standards apply to aluminum investment castings:
ASTM B618-03 Standard Specification for Aluminum-Alloy Investment Castings
ASTM B686-03 Standard Specification for Aluminum Alloy Castings, High-Strength
ASTM B618 standard applies to all investment cast parts; the minimum tensile properties required are measured on a
separately cast bar, (6.4mm) in diameter. ASTM B686 standard applies to premium quality castings; they must meet
minimal tensile requirements :
1) Either in designated area of the casting (Class 1,2, 3)
2) Or anywhere in the casting (Class 10, 11, 12)
In all of these cases, the minimum tensile properties are expressed in terms of minimum yield strength YS, ultimate tensile
strength UTS and elongation E. These minimal properties may be expressed in terms of the Quality index (Q MPa = UTS +
150 log E). It has also been shown that the yield strength YS is tied to UTS and E by the following relationship (JACOB,
1999):
YS (MPa) = UTS -60 log E 13 [5]
Q only depends on the metallurgical quality of the alloy, while YS can easily be adjusted by changing the aging conditions
(temper temperature, or time). So that when Q is above the required minimum value for the standard at hand, it is only a
matter of adjusting the aging time and temperature to satisfy that standard.
Table 1 indicates the minimum tensile properties required by the above mentioned ASTM B618 and B686 standards. The
probable yield strength YS* shown in Table 1 has been calculated from relationship [5]. It is noteworthy that the probable
yield strength YS* is always substantially higher than the required yield strength YS, except for Class 11 where YS* and YS
are virtually identical.
It is interesting to plot these requirements in a UTS versus E graph, as shown in Figure 2. The value of Q corresponding to a
(UTS, E) couple of values can be read on the lines sloping downward. The round dot indicates the required minimum values
of the (UTS, E) couple whilst the probable yield strength can be read from the lines sloping upwards. The tip of the arrow
indicate the required yield strength. Thus, an arrow pointing downwards indicates that the required yield strength is lower
than the probable yield strength by an amount equal to the length of that arrow.

Paper 05-006(02).pdf, Page 4 of 14


AFS Transactions 2005 American Foundry Society, Schaumburg, IL USA

Table 1. Tensile Requirements for Al-Si Mg (A356-357) Investment Castings (ASTM B618, ASTM B686)

Required UTS
INVESTMENT CASTING
ASTM B618
Separately cast bar

Req.
El.
%

ksi
34.0

MPa
235

3.5

MPa
316

Required YS

Probable YS
YS*=f(UTS, El)

ksi
24.0

MPa
166

ksi
27.3*

MPa
189*

ASTM B686

Class 1

38.0

262

5.0

367

28.0

193

29.9*

207*

Designated area
In casting

Class 2

40.0

276

3.0

348

30.0

207

33.8*

234*

Class 3

45.0

310

3.0

382

34.0

235

38.8*

269*

ASTMB686

Class 10

38.0

262

5.0

367

28.0

193

29.9*

207*

Anywhere in casting

Class 11

33.0

228

3.0

300

27.0

186

26.8*

186*

Class 12

32.0

221

2.0

266

22.0

152

24.7*

190*

C la s s 3

C la s s 2
C la s s 1

C la s s 1 0

S e p a r a te ly c a s t b a r (A S T M B 6 1 8 )
C la s s 1 1
C la ss 1 2

Fig. 2. ASTM B618 and ASTM B686 (Class 1, 2,3 10,11,12) tensile requirements
as plotted in terms of the Quality index Q.

It can be seen that


ASTM B618 requires a minimum Q of 316 MPa in separately cast tensile bars.
ASTM B686 class 1, 2, 3 require minimum Qs of 367, 348 and 382 MPa in designated areas of the casting.
ASTM B686 class 10, 11, 12 require minimum Qs of 367, 300 and 266 MPa everywhere in the casting.
It is interesting to note that, in spite of the slightly higher tensile strength required (but lower elongation), a
Class 2 designated area does not require a higher metallurgical quality than a Class 1 would; the very contrary.

Paper 05-006(02).pdf, Page 5 of 14


AFS Transactions 2005 American Foundry Society, Schaumburg, IL USA

The requirement of class 2 could be obtained on a Class 1 quality by changing the temper condition of the precipitation
hardening heat treatment (by using a longer aging time or a higher aging temperature); this would result in a higher tensile
strength, to the detriment of elongation.
PREDICTING COMPLIANCE WITH ASTM STANDARDS
BY SOLIDIFICATION MODELING
Since solidification modeling provides local thermal parameters in the whole casting, it will be possible to predict the local
values of the Quality index, Q; this will allow to assess whether the current casting process and casting conditions provide
the necessary level of Q to meet a given standard, i.e. ASTM B618-01 or B686-02 class xx (or MIL-A-2110-CL xx).
In investment casting, the main factors which will affect the Quality index will be:
a) the feeding of the designated areas of the casting. Directional solidification will enhance metallurgical quality and
increase Q.
b) the solidification time which in turn will depend on:
- the initial temperature of the shell (550C in our experiments)
- the nature and thickness of the shell (5.5mm average thickness in our experiments)
- the artificial cooling, if any, used to speed up solidification and/or promote directional solidification.
- the pouring temperature of the alloy.
PURPOSE OF THE EXPERIMENTAL STUDY
Our purpose is to investigate the influence of artificial cooling (forced air) on the Quality index of a casting at different
locations corresponding to different casting thickness. We wish to also assess whether the use of a more conducting shell
containing 50% SiC will improve the metallurgical quality under forced cooling conditions.
The geometry of the casting is shown in Figure 3. It consists of a square 30x30mm downsprue and two 20x30mm vertical
upflow channels to which four 6.4mm diameter tensile test bars have been attached.

Fig. 3.

Wax patterns

Alumino-silicate shell

SiC containing shells are black

Paper 05-006(02).pdf, Page 6 of 14


AFS Transactions 2005 American Foundry Society, Schaumburg, IL USA

In order to achieve this experimental study, 6 pours were carried out, each producing 2 castings; they allowed to investigated
3 conditions:
1) 4 Alumino silicate shells, naturally cooled.
2) 4 Alumino silicate shells, forced air cooled by the blower shown in Figure 4.
3) 4 shells containing 50% of SiC, forced air cooled by the same blower.
For each of the 3 conditions, the following tensile test bars were produced:
- 16 as cast 6.4mm diameter test bars attached to the channels.
- 8 half an inch diameter (12,8mm) test bars machined in the channels
- 4 half an inch diameter (12,8mm) test bars machined in the sprues
The velocity of the wind coming out of the blower (15kW turbine) was evaluated at 8 m/s using an anemometer.

.
Fig. 4. Blower used for artificially cooling the shells (wind velocity 8 m/s - heat tranfer coef. = 110 W/m2/K).

EXPERIMENTAL PROCEDURE AND CONDITIONS


PRODUCING THE SHELLS
Shells were produced by building the following successive layers of ceramic:
1) A primary coat with 125 mesh zircon particles; 0.1mm typical thickness.
2) An intermediate coat produced in a fluidized bed from a 60 mesh, 48% Al2O3 alumino-silicate; 0.3mm typical
thickness.
3) 4 main coats using coarser alumino-silicate (or with 50% silicon carbide content ), 20-30 mesh in size; 1mm per coat
typical thickness.
THICKNESS, PHYSICAL AND THERMAL PROPERTIES OF THE MOLD
An extra mold of uniform thickness (4 main coats) was produced with the sole purpose of determining the average thickness
of the shell, its apparent density and its porosity. The thickness of the shell was measured in 42 points of that mold; it was
found to average 5.5mm in value, with a standard deviation of 0.7mm.
The apparent density was determined to be 2.1 kg/dm3 by Archimedes method, on first immersion of a piece of the shell ;
then the porosity of the ceramic was found to be equal to 30%, by weighing a complete mold before and after immersion in
water for 10 hours. Since the thermal conductivity of the solid alumino-silicate is 3.5 W/m/K, the actual conductivity of the
mold was calculated to be 1.8 W/m/K when porosity was taken into account. When 50% SiC was added to the ceramic
particles, a resulting conductivity of 5.9 W/m/K was determined using the rule of mixtures (VAN VLACK); also, the
density of the shell was increased from 2.1 to 2.7 by the denser SiC. When modeling solidification, a heat capacity of 0.84
J/kg/K was used for all conditions.

Paper 05-006(02).pdf, Page 7 of 14


AFS Transactions 2005 American Foundry Society, Schaumburg, IL USA

The density values were crossed checked by using the solidification software to mesh a 5.5mm shell around the model of
the wax assembly. When this was done for a shell density of 2.1, the predicted weight of the molds were found to be 2.05 and
2.59 kg for the alumino-silicate and SiC containing shells respectively; this agreed reasonably well with actual measured
weights of 1.98 and 2.53 kg .
MELTING AND POURING
Six 8 kg heats of pre-modified primary Al Si Mg03 alloy (A356) allowed to produce 12 castings into shells preheated at
550C in an electric furnace. The composition of the melt was:
7.22% Si, 0.36% Mg, 0.054% Fe, 0.13% Ti, 0.017% Sr, 0.0002%Li, 0.0000% Na
The alloy was melted in a gas fired furnace, while making sure the melt temperature never exceeded 730C to avoid
hydrogen pick up. No degassing was performed. The pouring temperatures were measured and the gas levels of the melts
determined by the density of a sample solidified under 0.1 atm (Reduced pressure test, RPT). An empirical correlation from
MULAZIMOGLU (1989) was used to approximate the hydrogen content of melt in ml/100g. The conditions for the 6 heats
are summarized in Table 2. It is clear that the conditions of the 6 heats are very similar; they corresponds to a fairly high gas
level in the melt.
Table 2. Pouring Temperatures and Gas Levels of the Melts for the Six Heats

Pouring temperature
Alumino-silicate natural cooling
Heat #1 (2 castings)
Heat #2 (2 castings)
Alumino-silicate with forced air
Heat #3 (2 castings)
Heat #4 (2 castings)
Silicon carbide with forced air
Heat #5 (2 castings)
Heat #6 (2 castings)

RPT sample density

ml H2/100g

708 C
715 C

2.31
2.42

0.42 ml/100g
0.32 ml/100g

710 C
701 C

2.39
2.40

0.35 ml/100g
0.34 ml/100g

728 C
709 C

2.42
2.40

0.32 ml/100g
0.34 ml/100g

The 6 sections of the RPT samples are shown in Figure 5; they indicate that all heats were poured under similar conditions of
gassed melt. It took 7s to pour the 3 kg of metal necessary to fill one mold.

Fig. 5. Sections of the reduced pressure test samples for heats 1 to 6.

A standard T6 heat treatment was performed on 48 as-cast tensile samples (these 6.4mm diameter samples are shown in
Figure 6), the 24 channels and the 12 sprues. The solution heat treatment lasted 12 hours at 540C, followed by a quench in
water at 65C. After a minimum of 8 hours, the samples were aged for 4 h at 155C.

Paper 05-006(02).pdf, Page 8 of 14


AFS Transactions 2005 American Foundry Society, Schaumburg, IL USA

Fig. 6. As cast tensile specimens ready for T6 heat treatment 16 samples for each of 3 conditions.

SIMULATION OF THE SOLIDIFICATION


Using a commercial modeling software, simulations were performed for the 3 conditions investigated, namely:
1) Alumino silicate shell naturally cooled,
2) Alumino silicate shell forced air cooled and
3) 50%SiC shell forced air cooled.
The following thermal parameters were used in the simulation:
Thermal conductivity of shell :
kshell = 1.8 W m-1 K-1 for the alumino-silicate shell
kshell = 5.9 W m-1 K-1 for the 50% SiC shell
Heat capacity of shell: Cp = 0.84 J kg-1 K-1
Density of shell : = 2100 and 2700 kg m-3 for the alumino-silicate and 50% SiC shells respectively.
The heat transfer coefficient between external surface of shell (initially at 550) and ambient, h, was determined by inserting
3 thermocouples in one of the casting and calibrating the simulation by adjusting the value of h; the result was cross
checked by measuring the evolution of the surface temperature at one point of the surface of the shell during cooling. The
retained values were:
h=52 W m-2 K-1 for natural cooling.
h=110 W m-2 K-1 for air blowing at a velocity of 8 m/s in the two 200mm diameter ducts shown in Figure 7.
The blower noise level was 60 dB 3m.

Fig. 7. Position of a shell in the forced air cooling device.

Paper 05-006(02).pdf, Page 9 of 14


AFS Transactions 2005 American Foundry Society, Schaumburg, IL USA

Figure 8 presents a typical result from a simulation where the times of beginning and end of solidification are mapped for a
naturally cooled alumino-silicate shell poured at a temperature of 710C with a filling time of 7 s.

Fig. 8. Distribution of liquidus and solidus times for a naturally cooled alumino-silicate shell poured at 710C.

By finely analyzing the results summarized in Figure 8, it was possible to determine the predicted values of the solidus and
liquidus times, ts and tl respectively; and hence the local solidification time tsl = ts tl.
Under natural convective cooling, it was determined that for the alumino-silicate shell; the predicted solidification times
were :
ts = 1.62, tl = 0.26 min
in the 6.4mm diameter specimens
or tsl = 1.36 min
ts = 0.59, tl = 4.16 min
in the runners, 200mm below the top
or tsl = 3.57 min
ts = 1.40, tl = 6.95 min
in the downsprues, 100mm below the cup
or tsl = 5.55 min
It was found that under natural cooling conditions, the predicted solidification times using SiC containing shells were only
slightly reduced ( 3%).
Under forced air cooling conditions the predicted solidification times for the alumino-silicate shells were:
ts = 0.18 tl = 0.84 min
in the 6.4mm diameter specimens
or tsl = 0.66 min
ts = 0.43, tl = 2.52 min
in the runners, 200mm below top
or tsl = 2.09 min
ts = 0.93, tl = 4.10 min
in the downsprue, 100mm below cup
or tsl = 3.17 min
It is predicted that forced air cooling of the shells reduces the solidification times by a factor of a little more than 2 in the
6.4mm specimens, and a little less than 2 in the runners and downsprues. For shells containing SiC, the computed local
solidification times are substantially reduced by artificial cooling (22%).
PREDICTION OF METALLURGICAL PROPERTIES
As explained in a previous section, the level of microporosity, por%, secondary dendrite arm spacing DAS and Quality Index
(Q) may be predicted from the thermal parameters available when modeling solidification. This was done for the 3 conditions
experimented, namely : 1) alumino-silicate shell, naturally cooled; 2) alumino-silicate shell, forced air cooled and; 3) SiC
containing shell, forced air cooled.
In the following sections, these predicted metallurgical properties will be compared to those measured in our experimental
campaign.

Paper 05-006(02).pdf, Page 10 of 14


AFS Transactions 2005 American Foundry Society, Schaumburg, IL USA

Predicted distributions of microporosity ( Equation [3])

alumino-silicate, natural cooling

alumino-silicate, with ventilation

50% SiC shell with ventilation

Fig. 9. Predicted distribution of microporosity for different shell materials and cooling conditions.

The predictions mapped in Figure 9 (Obtained from the model using equation [3]) correspond to a relatively well degassed
melt containing 0.13ml of hydrogen /100g Al. Predicted values vary from about 0.5% to 1.5% in the naturally cooled shells.
The prediction is that forced air cooling substantially decreases the microporosity in all parts of the casting. Using a shell
containing SiC has only a small effect on microporosity.
Predicted distributions of secondary dendrite arm spacing DAS (Equation [2])

alumino-silicate, natural cooling alumino-silicate, with forced air

50% SiC shell with forced air

Fig. 10. Predicted distribution of DAS for different shell materials and cooling conditions.

In natural cooling, Figure 10 predicts that the DAS should vary from 40 to 70m; quite expectedly, the structure is the finest
in the 6.4mm specimens, and the coarsest in the sprue. Again, the prediction is that forced air cooling substantially refines the
DAS whilst the use of a SiC containing shell has a much lesser beneficial influence.

Paper 05-006(02).pdf, Page 11 of 14


AFS Transactions 2005 American Foundry Society, Schaumburg, IL USA

Predicted distributions of the Quality Index (Equation [4])

alumino-silicate, natural cooling

alumino-silicate, with forced air

50% SiC shell with forced air

Fig. 11. Predicted distribution of the Quality index for different shell material and cooling conditions.

In natural cooling, Figure 11 predicts that the Quality index Q will vary from 380 MPa in the 6.4mm tensile specimen to 270
MPa at the top of the 30x30mm sprue. It is also predicted that forced air cooling will improved the Quality index by about
40 MPa in all parts of the casting while adding SiC to the shell has only a small influence.
In terms of Class level per ASTM B686, let us recall that:
Q must be greater than 266, 300 and 367 MPa everywhere in the casting for class 12, 11 and 10 respectively.
Q must be greater than 367, 348 and 382 MPa in designated areas of the casting for Class 1, 2 and 3.
Looking at Figure 11 (left picture), the naturally cooled casting (considered as one single casting) is predicted to satisfy Class
12 requirements; it satisfies Class 11 requirements if the sprue is excluded from the casting. Only the 6.4mm diameter tensile
specimens are predicted to satisfy Class 1 requirements.
Looking at Figure 11 again (center picture), the forced air cooled casting (considered as one single casting) is predicted to
satisfy Class 11 requirements; it satisfies Class 10 requirements if the sprue is excluded from the casting. Again, only the
6.4mm diameter tensile specimens are predicted to satisfy the requirements of Class 1, 2 or 3. With a forced air cooled SiC
containing shell, the channels are predicted to exhibit a sufficient metallurgical quality to satisfy Class 2 tensile
requirements, as long as the a properly adjusted heat treatment is applied.
To summarize, it is predicted that:
- Forced air cooling improves the Quality index (hence tensile properties) by about 10%
- Using SiC containing shells has a moderate beneficial effect when forced air cooling is used
- Forced air cooling allows to upgrade the casting quality from Class 12 to Class 11 per ASTM B686
EXPERIMENTAL RESULTS AND DICUSSION
As-cast 6.4mm diameter tensile specimens (16 per condition), 12.8mm diameter specimens machined in the channels (8 per
condition) and sprues (4 per conditions) were tested in tension in the T6 temper.
The DAS were measured on a minimum of 80 secondary arm spacings in all sections. The levels of microporosity were
measured by image analysis on the whole section of the samples (test bars, channel and downsprue).
Table 3 summarizes these results and compares them to the predictions obtained by solidification modeling.

Paper 05-006(02).pdf, Page 12 of 14


AFS Transactions 2005 American Foundry Society, Schaumburg, IL USA

Table 3. Experimental Results Compared to Predictions of Solidification Modeling at Three Locations

Quality index, MPa


predicted

Microporosity,
%vol

experimental

predicted

rounded to closest
multiple of 5

379 (4)
445 (3)
414 (6)

0.4
0.2
0.2

0.73 (0.29)
0.15 (0.10)
0.09 (0.04)

41
32
30

44.5 (5.6)
27.0 (4.7)
28.2 (4.3)

1.0
0.7
0.6

1.32 (0.21)
1.04 (0.22)
1.07 (0.13)

58
46
43

62.0 ( 2.7)
47.3 ( 1.3)
41.9 ( 0.6)

68
56
52

62.7 ( 7.6)
55.2 ( 5.0)
49.2 ( 5.0)

Average of 8 tests

300
340
350

283 (5)
340 (8)
365 (6)
Average of 4 tests

30x30mm sprues
Al.silicate
Al silicate ventilated
SiC ventilated

experimental

measured on
80 arm spacings

(0.12ml/100g) (0.35 ml/100g)

380
425
435

20x30mm Channels
Al.silicate
Al silicate ventilated
SiC ventilated

predicted

Average of 16 tests

6.4mm specimens
Al.-silicate
Al silicate ventilated
SiC ventilated

experimental

Dendrite fineness DAS,


m

257 (9)
311 (16)
335 (11)

280
325
335

1.5
1.2
1.1

1.44 (0.50)
0.91 (0.02)
1.37 (0.36)

For an easier comparison, the results have been plotted graphically in Figures 12, 13 and 14.
Prediction versus experiment, Q

Quality index, MPa

500
400
300
200
100
pr

exp

0.25"

pr

ex

channels

pr

ex
downsprues

SiC forced air


Alum.silicate forced
Alum.silicate natural

Fig. 12. Comparison of predicted and experimental values of the


Quality index at 3 locations for 3 conditions.

On Figure 12, it is clear that forced air cooling increases the Quality index by more than 10%, and more when the high
conducting SiC shell is used. The difference between the predicted values of the Quality index and the experimental ones is
always less than 6%, except in the downsprue of the naturally cooled alumino-silicate shells were experimental results are
8% lower than the predicted ones. The average absolute difference between predictions and experiments for all 3 conditions
is 4%. It is -0.2%, 5.9% and -3.7% in the 6.4mm diameter specimens; -5.6%, 0.0% and 4.3% in the channels and -8%,-5.7%
and -1.4% in the sprues.

Paper 05-006(02).pdf, Page 13 of 14


AFS Transactions 2005 American Foundry Society, Schaumburg, IL USA

Prediction vs experiment, microporosity

Microporosity, %vol.

1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
pr

ex

6.4mm"

pr

ex

channels

pr ex
downsprues

SiC forced air


Alum.sil. forced
Alum.sil. natural

Fig. 13. Comparison of predicted and experimental values


of the microporosity at 3 locations in casting.

Figure 13 shows that the measured microporosity levels are always about 50% higher than the predicted ones, except in the
downsprues. This can be explained in part because the predicted value is based on an empirical relationship obtained on a
degassed melt (0.13 ml [H]/100g Al) while the gas content of our 6 melts was much higher. However, for very low
microporosity levels, the prediction overestimates the actual level of microporosity. On the other hand, we notice that the
level of microporosity in the downsprues is similar to that in the channels in spite of the slower solidification. In a previous
project involving a casting of similar geometry, relatively higher levels of microposity had already been observed in the
channels (vs the sprues), in spite of the finer dendritic structure observed. This confirmed phenomenon might be attributed to
the natural degassing of the melt under the pouring cup after the end of filling; the time to reach the liquidus at that location is
of the order of 1 to 2 minutes, leaving ample time to the melt to expulse part of the hydrogen dissolved.
Prediction versus experiment, DAS

70
60
DAS, m

50
40
30
20
10
0
pr.

exp

6.4mm

pr

ex

channels

pr
ex
downsprues

SiC forced
Alum.sil forced
Alum.sil natural

Fig. 14. Comparison of predicted and experimental values


of the dendrite arm spacing at 3 locations in casting.

Figure 14 indicates that forced air cooling markedly refines the dendritic structure everywhere in the casting. Using a SiC
containing shell has a notable effect, bigger than what was predicted by modeling, except in the small 6.4mm tensile
specimens.
As far as the DAS are concerned, the predicted values are, on average, 6% off the measured values (8.5,-15,-6%; 6.9,2.8,2.6% and -7.8,1.4,-5.4%); these differences is always within the dispersion typical of DAS measurements.

Paper 05-006(02).pdf, Page 14 of 14


AFS Transactions 2005 American Foundry Society, Schaumburg, IL USA

CONCLUSIONS
1) The Quality index can be reliably predicted by solidification modeling. The Quality index has been shown to be closely
tied to the level of Quality per ASTM B686. Consequently, it is possible to know beforehand what level of quality can
reasonably be expected from a given casting practice.
2) Using forced air convection to cool the shell after pouring substantially improves the metallurgical quality in the whole
casting. For instance, the DAS is reduced from 41 to 32m in the 6.4mm diameter tensile test bar. In the 20x30mm
channels and 30x30mm downsprue, the DAS is reduced from 58 to 46m and from 68 to 56 m respectively. The
microporosity is reduced from 0.7 to 0.15% in the 6.4mm diameter bars, from 1.3 to 1.0% in the channels and from 1.4
to 1.0% in the downsprue. This improvement in metallurgical quality results in the Quality index rising from 379 to 425
MPa in the 6.4mm tensile test bars, from 283 to 340 MPa in the channels and from 257 to 311 MPa in the downwsprue.
Overall, forced air cooling allows to upgrade the casting by one Class level (from Class 12 to Class 11).
3) Adding 50% more conducting SiC in the shell does not bring about a worthwhile improvement in the metallurgical
quality.
ACKNOWLEDGMENTS
The authors wish to acknowledge the financial support of the Qubec Ministry of Education Programme dAide la
Recherche Technologique . They are also indebted to Alcan of Canada for procuring the primary A356 alloy (Beauharnois
plant) and to Alphacasting (Montreal) for providing the waxes and the tensile specimen die.
REFERENCES
Chiesa, F., Duchesne, B., Jacob, S., AFS Transactions, Vol 111, p 185-191 (2003).
Chiesa, F., Regimbal, P.; Light Metals 1999, Canadian Institute of Mining and Metallurgy, pp 331-342 (1999).
CTIF, Atlas de micrographie quantitative des A-S7 G, Editions Techniques, www.ctif.com, pp 24-29 (1976).
Drouzy, M., Jacob, S., Richard, M.; AFS International Cast Metals Journal, June 1980, pp 43-50 (1980).
Fuoco R., H. Goldenstein R.H., Gruzleski J.E.; AFS Transactions, vol. 102, p 302 (1994) .
Emadi D., J.E., Gruzleski J.E.; AFS Transactions, vol. 102, p 307 (1994).
Jacob, S., Silver Anniversary Paper, Aluminum division, AFS Transactions, Vol 106, pp 811-818 (1999).
Kao S.T., E. Chang, L.C. Chan; AFS Transactions, vol 103, p 533 (1995) .
Lee, Y.W., Chang, E.,Chieu, C.F.; Metallurgical Transactions, Vol.21B, p 717 (1990).
Mammen J., Chiesa F., AFS Transactions, Vol 107, pp 103-111 (1999).
Mulazimoglu H.M., Handiak N., Gruzleski J.E., AFS Transactions, Vol 97, p 225 (1989).
Pan E.N., Chiou H.S., Liao G.J.; AFS Transactions, vol. 99, p 605 (1991).
SOLIDCastTM workbook, www.finitesolutions.com, p 27-1.
Van Vlack L.; Material Sciences Second Edition, p 302 (1967).

You might also like