You are on page 1of 9

International Journal of Fatigue 24 (2002) 19

www.elsevier.com/locate/ijfatigue

Casting defects and fatigue strength of a die cast aluminium alloy:


a comparison between standard specimens and production
components
M. Avalle, G. Belingardi, M.P. Cavatorta *, R. Doglione
Department of Mechanical Engineering, Politecnico di Torino, Corso Duca degli Abruzzi 24, 10129 Torino, Italy
Received 9 April 2001; received in revised form 26 July 2001; accepted 31 July 2001

Abstract
The influence of casting defects on static and fatigue strength is investigated for a high pressure die cast aluminium alloy. Defects
exist in gas and shrinkage pores as well as cold fills, dross and alumina skins. For the three batches of specimens, differing for
the spruerunner design, the influence was straightforward, while no significant variation in the fatigue strength was observed when
looking at batches of acceptable and non-acceptable components, as judged within the foundry quality control. In this case,
defects count for their size and location, while quality control often takes no account for component working conditions. The Haigh
diagram shows a good matching between the specimen reference material and the component fatigue data. 2001 Elsevier Science
Ltd. All rights reserved.
Keywords: Die casting; Porosity; Casting defects; Fatigue; Aluminium

1. Introduction
High pressure die casting is widely used for the possibility of obtaining net to shape components of complex
geometry at high production rates. A disadvantage of the
technology is the almost inevitable presence of shrinkage
cavities, often coupled with other defects: cold fills, alumina skins, dross, entrapped air bubbles [1]. The influence of casting defects on the material properties of cast
aluminium alloys has been investigated by a number of
authors [210], with porosity being the most common
defect encountered. In literature data [25], the influence
of casting defects on static strength appears to be generally modest (i.e. of the order of some percentage points).
More important variations are observed for the elongation at rupture [3].
The influence of casting defects on fatigue strength
has been studied under constant-amplitude, variableamplitude and simulated in-service conditions. It is gen-

* Corresponding author. Tel.: +39-11-5646933; fax: +39-115646933.


E-mail address: cavatorta@polito.it (M.P. Cavatorta).

erally acknowledged that the fatigue strength of


materials containing defects is lower than that of a defect
free material. It is also generally accepted that fatigue
life is determined not only by the size of the defect but
also by its distance from a free surface [9,10]. Recent
data also showed the importance of the distribution and
morphology of phases in the microstructure. In Ref. [11],
fatigue strength is observed to vary with solidification
parameters and hence location in the casting.
However, results reported in the literature are almost
exclusively concerned with the effect of porosity, as
investigations are carried out on samples obtained by
sand or permanent mould casting. Moreover, often the
difference between the samples consist of the adjustment
of the melt gas content which is likely to influence the
porosity level alone. Actual criteria for quality control on
castings subjected to fatigue loading assume a maximum
acceptable level of porosity. Standards too [12,13] focus
mainly on this kind of defect. On the other hand, the
other aforementioned defects, more commonly encountered in die casting components, are known to adversely
affect the mechanical properties [1], but quantitative data
are still lacking.
The paper investigates the influence of casting defects

0142-1123/01/$ - see front matter 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 2 - 1 1 2 3 ( 0 1 ) 0 0 1 1 2 - 8

M. Avalle et al. / International Journal of Fatigue 24 (2002) 19

on the static and fatigue strength of a high pressure die


cast aluminium alloy. The material is EN AC-46000UNI EN 1706 (ex GD-Al Si8.5 Cu 3.5 Fe-UNI 5075),
widely used in load-bearing components in the automotive field. Static and fatigue tests are run on standard
specimens in as cast conditions as well as production
components. For the specimens, three batches are tested.
The batches differ in the geometry and dimension of the
spruerunner and, consequently, in the content and type
of defects. By changing the feeding conditions, it was
possible to reproduce in a more realistic way the assortment of defects typical of a die casting, whereas a mere
increase in the gas content, often adopted in experimental studies, simply gives rise to a more important
porosity. For the components, two batches are analysed,
corresponding to acceptable and non-acceptable
components as judged within the foundry on-line standard quality control.
As part of the study, a comparison between fatigue
data for the specimens and the components was
attempted with the purpose of investigating whether data
obtained for the standard specimens can be meaningful
for the evaluation of the fatigue behaviour of a production component. For this reason, constant amplitude
fatigue tests were run for the specimens and the components. A conventional and sufficiently high number of
cycles was chosen for sake of comparison.

2. Materials and methods


Table 1 shows the chemical composition of the alloy
used in the study. Reported values refer to several
measurements taken on different specimens and production components. Scattering was quite limited. The
percentage in weight of the different elements fall within
the standard chemical composition of the alloy (UNI EN
1706), also reported in Table 1.
One series of static and fatigue tests was run on standard specimens (Fig. 1). The specimens were obtained
directly from die casting and were tested in as cast
conditions (i.e. with as cast surface). The only modification to the cast surface was the trimming of flash.
Three batches of specimens were analysed. The batches
differ for the geometry and dimensions of the sprue
runner. As previously mentioned, batches are likely to
differ for the porosity level as well as for the presence
of other casting defects due to a non-optimisation of the

Fig. 1. Geometry and dimensions of the standard specimen used in


the study.

feeding channel. However, casting defects as dross, cold


fills and alumina skins are not detectable a priori through
X-ray examination. An initial classification for the three
batches was therefore made with regards to the porosity
level alone. The level of porosity of the three batches
was assessed through X-ray examination and supported
by density measurements and scanning electron
microscopy (SEM) observation of the fracture surface.
The three batches were classified with respect to their
porosity level as range 0 (reference material), 2 and 4
(Category A) according to ASTM E 505 [12]. In what
follows, the three batches of specimens will be referred
as A0, A2 and A4 according to their porosity level.
Table 2 reports density measurements for the three
batches. Data are average of several measurements on
different specimens of the same batch. The void volume
fraction f is also reported on Table 2, with f being calculated from the well-known formula:
r0r
f
r0

(1)

where r0 is the density for the reference material (A0)


and r the density of the considered batch.
Fig. 2 shows representative fracture surfaces for the
different batches. The distribution and size of pores may
be evaluated. The cavities are rounded and more concentrated in the core of the castings. The pore shape and
Table 2
Density measurements on the standard specimens

r (kg/m3)
f

A0

A2

A4

2730
0

2570
0.059

2230
0.183

Table 1
Chemical composition of the aluminium alloy used in the study (% by weight)

EN AC 4600

Si

Cu

Fe

Zn

Mn

Mg

Ni

Ti

Pb

9.83
811.1

2.99
24

0.98
0.61.1

1.11
1.2

0.19
0.55

0.04
0.05

0.064
0.55

0.052
0.25

0.19
0.35

M. Avalle et al. / International Journal of Fatigue 24 (2002) 19

It is commonly acknowledged that the concept of endurance limit loses its significance when aluminium and
other non-ferrous materials are considered as well as
when components are subjected to variable amplitude
loading, as it is always the case in real components. In
view of evaluating the influence of porosity for specimens and production components, fatigue strength was
evaluated at a conventional and sufficiently high number
of cycles for sake of comparison. Moreover, preliminary
tests run on the A0 batch showed a limited difference
between the fatigue strength at 107 cycles and the fatigue
strength at 2106 cycles [14]. This result is in accordance
with the broadly accepted view that a nearly flat SN
curve is to be expected for materials for which there is
little difference between yield strength and endurance
limit [4], as is the case for the alloy used in this study.
A second series of fatigue tests was run on industrial
components taken from the foundry production line. The
components were tested in as cast conditions (i.e. with
as cast surface). The only modification to the cast surface
was the trimming of the overpress corresponding to the
half mould. Fig. 3 shows the geometry and shape of the
alternator support used in the study. Two batches of
components were tested. The two batches group acceptable and non-acceptable components respectively, as
judged from the on-line radioscopic inspection made
within the foundry quality control. The inspection is
made by a skilled technician who decides whether the
level of porosity in the component is acceptable or nonacceptable for the foundry standards. A total of 42 supports were tested in fatigue, 18 for the acceptable batch
and 24 for the non-acceptable batch.
Prior to testing, all components were subjected to Xray examination. One component of each batch was also
examined through computed tomography (CT scan).
Shrinkage and gas pores were observed for components
of both batches. For components classified as non-

Fig. 2. Representative fracture surfaces for the three batches (A0,


A2, A4 from top to bottom).

limited size of the samples indicate that porosity arises


mainly by entrapped air bubbles during the feeding of
the die and eventually by hydrogen development during
solidification. However, shrinkage is also present at the
centre of the samples for the high porosity range.
Fatigue tests were run on a cantilever rotating bending
testing machine. Test frequency was 100 Hz. The fatigue
strength at 2106 cycles was estimated for the three
batches by means of the standard stair-case procedure.

Fig. 3.
study.

Geometry and shape of the alternator support used in the

M. Avalle et al. / International Journal of Fatigue 24 (2002) 19

acceptable by the foundry quality control, isolated large


pores in the larger sections of the component were
observed together with a finer porosity. The larger pores,
whose size is in the millimetre range, are not present in
the acceptable components. Density measurements,
taken after failure on pieces of the components, showed
no difference between the two batches. A value of about
2720 kg/m3 was found in both cases. This value of density is nevertheless an average value. Lower values of
the density are likely to be present locally, in the thicker
sections of the components, but could not be measured.
For fatigue testing of the industrial components, an
universal hydraulic testing machine was used. Special
fixtures were designed to hold the component during the
test. Two supports were tested simultaneously to ensure
a limited transversal component of the force. The loadcontrolled constant-amplitude tests were carried out at a
test frequency of 25 Hz. The fatigue strength at 2106
cycles was estimated for the two batches by means of a
modified stair-case procedure. The minimum load
applied (Pmin=0.5 kN) was kept constant for all tests,
while the maximum load was varied according to the
stair-case rule. Since two components were tested simultaneously, the adopted stair-case procedure was modified
as follows: a step increase in the maximum load was
applied when there was a run-out for both components.
If one of the two components failed, then a step decrease
in the maximum load was applied and a new pair of
components was tested. At last, the whole set of results
(failure and run-out data) was also processed with a
maximum likelihood methodology [15] to assess that the
procedure changes have not induced large errors in the
calculated fatigue strength (mean and standard deviation values).

3. Results and discussion


3.1. Data on specimens
Owing to important variations in porosity (Table 2),
static tensile tests showed a notable decrease in the yield
and ultimate strength with the degree of porosity. Data
reported in Table 3 are averages of at least three tensile
tests for each batch. As expected, the elongation at rupture also decreased for increasing levels of porosity. The

decrease in the static characteristics is progressive with


the porosity range (Fig. 4).
Cantilever rotating bending fatigue tests were then
performed. Bending tests were run against tensile tests
considering that die casting enhances material properties
at the surface, due to the local fineness of the solidification structure.
For comparison of the three specimen batches, the
fatigue strength at 2106 cycles was estimated. The
fatigue tests showed a decrease in the fatigue strength for
increasing degrees of porosity (Table 4). The decrease in
fatigue strength was most significant when the degree
of porosity was increased from A0 to A2, while it was
distinctly lower from A2 to A4. This trend of results is
in agreement with literature data published by Sonsino
and Ziese on siliconmagnesium aluminium cast alloys
[4].
Finally, the fracture surfaces were observed through
SEM. For the A0 batch, the fatigue fracture was seen to
originate from an area with no visible defects. For the
A2 and A4 batches, gas cavities were observed to act as
initiation sites for the fatigue fracture (Fig. 5). However,
the changes in the spruerunner design made by the
foundry to increase the porosity of the die cast alloy
inevitably introduced other defects due to a non-optimisation of the feeding channel. In particular, cold fills (Fig.
6) and alumina skins (Fig. 7) appeared on the specimen
fracture surfaces. The cold fills originate during the filling when fine drops of the cast alloy get into contact with
the mould and solidify. No coherence therefore exists
between the drops and the cast component and fatigue
fractures can easily initiate.
Cold fills and alumina skins are not visible through
either radiography or CT scan. The initial classification
of the specimens, which reflects only the difference in
the porosity level due to a change in the spruerunner
design, was the only one possible by means of nondestructive techniques. However, it is worthwhile noticing that the observed decrease in fatigue strength for the
three batches is to be considered in view of the combination of the increased porosity level and above mentioned metallurgical defects. It is probably the combination of the different casting defects that is the cause
of the large decrease in the fatigue strength characteristics when compared with literature data [25], where
porosity is the main investigated casting defect. In this

Table 3
Results for the static tensile tests run on specimens. Reported standard data are minimum values (UNI EN 1706)

Elastic modulus (GPa)


Yield strength (MPa)
Ultimate strength (MPa)
Elongation %

EN AC 46000

A0

A2

A4

140
240
1

70
150
275
2.1

65
140
250
1.4

50
110
150
0.85

M. Avalle et al. / International Journal of Fatigue 24 (2002) 19

Fig. 4.

Static strength characteristics for the three batches.

Table 4
Fatigue strength data at 2106 cycles for standard specimens

Mean value fatigue strength (MPa)


Standard deviation (MPa)

Fig. 5.

A0

A2

A4

152
4.9

96
6.5

76
5.8

A gas cavity as fatigue fracture initiation site (80).

respect, in the case of this study, pores are mainly concentrated in the middle part of the resisting section and
scarcely provide sites for fatigue crack initiation. On the
contrary, defects as alumina skins, dross and above all
cold fills are effective weak points at the surface.
3.2. Data on production components

Fig. 6.
980).

Cold fills at the edge of the specimen (top: 170; bottom:

A finite element model of the component, with load


and restraint conditions reproducing those of the experimental testing, was run to get information on regions of
high stress and hence of possible failure. Two regions
of high tensile equivalent stress were found: one close

to the bottom hole (Fig. 8(a) and (b)) and one inside the
central hole (Fig. 8(c)).
Experimental fatigue tests resulted mainly in one
region of failure, corresponding to the central hole (Fig.
9(a)). In the following, this region will be referred to as

M. Avalle et al. / International Journal of Fatigue 24 (2002) 19

3.3. Specimens versus components

Fig. 7.

An alumina skin as fatigue fracture initiation site (470).

the region of failure. In a limited number of tests, three


other regions of failure occurred, one being the bottom
hole where high tensile equivalent stress were predicted
by the FE analysis (Fig. 9(b)). The other two regions of
failure were on the contrary unexpected (Fig. 9(c) and
(d)). For the these two regions of failure, voids or oxide
layers of significant dimensions were observed on the
fracture surface (Fig. 10).
Tables 5 and 6 show results of the stair-case procedure
for the acceptable and non-acceptable batch,
respectively. The load level shown in the first column
refers to the maximum load applied to each component.
No significant variation in the fatigue strength can be
observed between the two batches (Table 7).
In the authors opinion, this result can be explained
considering that in a component, defects count for their
size as well as for their location. Porosity is to lower the
fatigue strength of the component if located in highly
stressed regions. The on-line radioscopy inspection looks
for pores of significant dimensions, generally taking no
account of the component working conditions. Moreover, oxide layers, which are known to constitute a severe defect, were present in components of both batches,
being not detectable through radioscopy.

Fig. 8.

A question the authors tried to investigate was


whether data obtained for the standard specimens could
be somehow related to the study of the industrial component.
As a first step, the finite element model of the alternator support was used to express results on fatigue
strength in terms of stress instead of load. The fatigue
load evaluated for acceptable components (Table 6) was
used as input load in the FE analysis. Data for the stress
in the region of failure were then obtained. Values of
stress were averaged in the area where fatigue cracks
seem to propagate.
Being the state of stress in the component triaxial, a
criteria for multiaxial fatigue was applied [16]. Eqs. (2)
and (3) allow calculation of the value of the equivalent
mean and alternate stress, respectively, knowing the
mean and alternate principal stresses in the region of
failure from the FE analysis.
smeqs1ms2ms3m
saeq

(2)

(s as a) +(s as a) +(s as a)

2
1

(3)

Values obtained for the equivalent mean and alternate


stress were plotted in the Haigh diagram traced with
static and fatigue data obtained from the specimens [17].
The diagram is plotted for a duration of 2106 cycles. A
quite good correlation exists in the case of the A0 specimens (Fig. 11). Reported data points for the alternator
support fatigue strength are for a probability of survival
of 10, 50 and 90%.
The result seems to indicate that the A0 specimens
are metallurgically equivalent to the regions of the
component where fatigue cracks nucleate. It also shows
that fatigue data obtained through rotating bending tests
may be employed to predict the fatigue strength of cast
complex components, where the state of stress is triaxial,

Equivalent stress maps from the FE analysis.

M. Avalle et al. / International Journal of Fatigue 24 (2002) 19

Fig. 9.

Experimental regions of failure for the alternator support.

provided that no surface defects (like cold fills, alumina


skins and dross) are present, irrespective of important
shrinkage cavities in the thicker sections of the component. In addition, the Goodman relationship appears to
be the most appropriate tool to link specimen rotating
bending fatigue tests to component fatigue data.
4. Conclusions
The effect of porosity and other casting defects on
static and fatigue properties of specimens and components obtained by high pressure die casting on a Al
SiCu alloy has been assessed. The experimental and
numerical results point out the following conclusions:
Fig. 10.

Oxide layers of significant dimensions on a fracture surface.

as expected, static characteristics are sensitive to the

Table 5
Modified stair-case sequence for acceptable components
Maximum load
(kN)
11.5
11.0
10.5
10.0

9
XO

XO
XO

OO
OO

OO
XX

OO
OO

M. Avalle et al. / International Journal of Fatigue 24 (2002) 19

Table 6
Modified stair-case sequence for non-acceptable components
Maximum load
(kN)
11.0
10.5
10.0
9.5

XO

XO
OO

10

11

12

XO
OO

XX

XO
OO

XO
XO

OO
OO

Table 7
Comparison between the fatigue maximum load at 2106 cycles found for acceptable and non-acceptable components

Acceptable components
Non-acceptable components

R(50%) (kN)

R(10%) (kN)

R(90%) (kN)

10.84
10.50

11.22
10.83

10.25
10.00

Fig. 11. Fatigue strength data for the production component. The Haigh diagram is traced with data obtained for the A0 specimens (square data
points). The diamond data points are for the production component and refer to a probability of survival of 10, 50 and 90%.

defects; in particular, the tensile strength decreases


linearly with the porosity range
the constant amplitude fatigue strength at 2106 scales
with the porosity range and is slightly lower than the
yield strength
for the specimens, the three batches were classified
prior to testing through non-destructive techniques,
i.e. with regards to the porosity range. However, the
changes made in the spruerunner design introduced
other metallurgical defects, such as alumina skins and
cold fills
it is the combination of pores and metallurgical
defects that account for the observed decrease in the
fatigue strength for the three batches. In particular,
cold fills being a discontinuity in the material at the
edge of the specimen represent severe casting defects
owing to the previous results, no difference in fatigue
strength was found between acceptable and non-

acceptable components, as judged within the foundry


on-line radioscopic quality control
for this alloy, the Goodman relationship appears to
be an appropriate tool to relate the specimen rotating
bending fatigue tests to component fatigue data
for castings subjected to fatigue loading, criteria for
quality control based on a maximum acceptable level
of porosity are not appropriate if the average density
is not affected and the cavities are located at the
centre of thick sections.

Acknowledgements
This research was supported by the European Union
under contract no BRST-CT98-5328. The authors would
like to thank Ing. M. Albertinazzi for helping with the

M. Avalle et al. / International Journal of Fatigue 24 (2002) 19

fatigue tests. Valuable discussion with Prof. M. Rossetto


is also acknowledged.
References
[1] Metals Handbook. Casting, vol. 15, 9th edition, ASM Ohio: International, Metals Park, 1988:294295.
[2] Caceres CH, Selling BI. Casting defects and tensile properties of
an AlSiMg alloy. Mat Sci Engng, Part A 1996;220:10916.
[3] Garat M. The respective effects of fineness of structure and compactness on the static and dynamic mechanical properties of AS7G06. Fondeur dAujourdhui, 1989:20 (in French).
[4] Sonsino CM, Ziese J. Fatigue strength and applications of cast
aluminium alloys with different degrees of porosity. Int J Fat
1993;15(2):7584.
[5] Dabayeh AA, Xu RX, Du BP, Topper TH. Fatigue of cast aluminium alloys under constant and variable amplitude loading. Int
J Fat 1996;18(2):95104.
[6] Odegard JA, Hafsas JE, Peterson K. Fatigue crack initiation and
growth in a DC-Cast AlSi7Mg alloy. Proc Fat 1990;90:2738.
[7] Dabayeh AA, Berube AJ, Topper TH. An experimental study of
the effect of a flaw at a notch root on the fatigue life of cast Al
319. Int J Fat 1998;20(7):51730.
[8] Mayer HR, Lipowsky HJ, Papakyriacou M, Ro sch R, Stich A,
Stanzl-Tschegg S. Application of ultrasound for fatigue testing
of lightweight alloys. Fat Fract Engng Mat Struct 1999;22:5919.

[9] Seniw ME, Fine ME, Chen EY, Meshli M, Gray J. Relationship
of defect size and location to failure in Al alloy A356 cast specimens, Paris International Symposium Fatigue of Materials. TMSASM Fall Meeting, 1997.
[10] Couper MJ, Nesson AE, Griffiths JR. Casting defects and the
fatigue behaviour of an aluminium casting alloy. Fat Fract Eng
Mat Struct 1990;13(3):21327.
[11] Seniw ME, Conley JG, Fine ME. The effect of microscopic
inclusion locations and silicon segregation on fatigue lifetimes of
aluminum alloy A356 castings. Mat Sci Engng Part A
2000;285:438.
[12] Standard Reference Radiographs for Inspection of Aluminum and
Magnesium Die Casting, ASTM E 505-96. Annual book of
ASTM standards. American Society for Testing and Materials.
[13] Standard Reference Radiographs for Inspection of Aluminum and
Magnesium Casting, ASTM E 155-79. Annual book of ASTM
standards. American Society for Testing and Materials.
[14] Avalle M, Belingardi G, Cavatorta MP. Static and fatigue
strength of a die cast aluminium alloy under different feeding
conditions. Presented at EUROMAT 2001, Rimini 1014 June
2001.
[15] Goglio L, Rossetto M. Elaborazione e confronto di dati di fatica,
Quaderno AIAS n. 1, Milano, 22 November 1995:3549.
[16] Fuchs HO, Stephens RI. Metal fatigue in engineering. New York:
J. Wiley and Sons, 1980.
[17] Collins MS. Failure of materials in mechanical design. New
York: J. Wiley and Sons, 1981.

You might also like