You are on page 1of 8

International Journal of Fatigue 53 (2013) 67–74

Contents lists available at SciVerse ScienceDirect

International Journal of Fatigue


journal homepage: www.elsevier.com/locate/ijfatigue

Effect of the casting process on microstructure and lifetime of the Al-piston-alloy


AlSi12Cu4Ni3 under thermo-mechanical fatigue with superimposed high-cycle
fatigue loading
Alexander Humbertjean ⇑, Tilmann Beck
Research Centre Juelich, IEK-2, 52425, Germany

a r t i c l e i n f o a b s t r a c t

Article history: For the present research work, the well-known Al-piston-alloy AlSi12Cu4Ni3 was manufactured in three
Received 11 May 2011 different processes (gravity die casting – GDC, low pressure die casting – LPDC, and high pressure die
Received in revised form 9 September 2011 casting – HPDC) and T5 heat treated. The microstructure of the material from each process was analyzed,
Accepted 26 September 2011
specimens were tested in OP-TMF loading with and without superimposed high-frequency fatigue and
Available online 18 October 2011
lifetimes were compared.
The microstructure of GDC specimens shows a homogeneous distribution of primary Si and intermetal-
Keywords:
lic phases. The LPDC material also shows a homogeneous microstructure over the whole sample. How-
Aluminum alloys
Piston
ever, the Al-mixed-crystal formed bigger dendrite arms compared to the GDC material. The HPDC
Thermo mechanical fatigue material shows a gradient in the microstructure getting finer from the center to the outer shape. In
High cycle fatigue the very fine microstructure in the outer regions of the specimen no primary Si was formed and the
Al-mixed-crystal built a globular–dendritic structure, surrounded by a eutectically solidified melt. Poros-
ity was higher in the LPDC and HPDC compared to the GDC material.
To simulate the thermally induced loading at the ‘‘hot side’’ of a piston during start–stop, strain con-
trolled out-of-phase thermal–mechanical fatigue (TMF) tests with superimposed high-cycle fatigue
(HCF) loading were performed. The TMF-cycles were carried out with a minimum temperature
Tmin = 200 °C and a maximum temperature Tmax = 440 °C. The mechanical strain amplitude eme a;t of the
TMF cycles was kept equal to 50% of the thermal strain amplitude eth a and the amplitude of the superim-
posed HCF cycles eHCFa;t was varied between 0.03% and 0.05%.
The lifetime of the specimens produced in the GDC-process represents the current state of the art. Com-
pared to that, the specimens manufactured in LPDC and HPDC reached nearly the same TMF lifetimes as
the GDC samples. The maximum stress of the LPDC specimens is approximately equal to that observed at
the GDC material. However, during the very first TMF cycles, HPDC-specimens show higher maximum
stress than the reference material. This is attributed to the very fine, nearly defect free microstructure
at the outer shape of the HPDC material. Afterwards, the maximum stress of the HPDC samples is
decreasing faster than that of the GDC material due to early formation of crack networks starting from
fine pores in the HPDC microstructure. In TMF/HCF-testing the HPDC material shows the same effect. Fur-
thermore, the HPDC samples show pronounced swelling during temperature cycling and TMF testing. The
root cause was identified as the high internal pressure of air encased in the pores formed during HPDC.
Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction ing, for instance, the content of alloying elements, especially Si, Cu,
Ni and Fe, is difficult [1,2]. Therefore, it is essential to explore the
To develop more efficient diesel engines, lightweight high per- applicability of other manufacturing processes. A first option is to
formance pistons are essential. The current manufacturing process use low-pressure-die-casting (LPDC) to smoothen the melt flow,
of Al-base pistons, gravity-die-casting (GDC), has been optimized which minimizes casting defects. Furthermore, the cooling rate
for many years together with the applied alloys. Due to process can be increased to obtain a microstructure with finely dispersed
parameter limits for temperature, filling time, and cooling rate, a high-temperature stable second phases [1]. In LPDC, the mold fill-
further improvement of the high temperature strength by increas- ing is controlled by air pressure on the melt. The ladle is enclosed
in an air-tight chamber such that by applying a given air pressure
⇑ Corresponding author. the liquid metal flows through a refractory pouring tube into the
E-mail address: a.humbertjean@fz-juelich.de (A. Humbertjean). mold. Turbulences in the melt during the filling are minimized

0142-1123/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijfatigue.2011.09.017
68 A. Humbertjean, T. Beck / International Journal of Fatigue 53 (2013) 67–74

Fig. 1. (a) Sketch of specimen with adapters, taken from a piston, used for GDC, LPDC, and HPDC, (b) photograph and sketch of separately cast specimen with (left) and
without (photo right) adapters, used for HPDC. (c) Test setup for big specimen geometry with induction coil, thermocouple type K, extensometer, and cooling sleeves. (d) Test
setup for separately cast specimen with induction coil, thermocouple type K, extensometer with glass holder, and cooling sleeves. (e) Sketch of TMF/HCF cycle temperature
(red) and mechanical strain (blue) against time. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
A. Humbertjean, T. Beck / International Journal of Fatigue 53 (2013) 67–74 69

Fig. 1 (continued)

state [5]. The effects of the manufacturing process on microstruc-


120
ture and on the behavior in thermo-mechanical fatigue testing
100 with superimposed high-cycle fatigue loading were analyzed. The
80 results of this research are given in the present paper.

60
σ [MPa]

40 2. Test material and experimental details

20
In order to analyze the differences between the three chosen
0 casting processes AlSi12Cu4Ni3 pistons were manufactured by
-20 GDC, LPDC, and HPDC. The TMF specimens were machined from
cast pistons (Fig. 1a). Additionally, HPDC samples were cast sepa-
-40
rately with a different geometry (Fig. 1b). The diameter in the par-
-60 allel length l0 is dpiston = 9 mm for the samples machined from
-0.4 -0.3 -0.2 -0.1 0 0.1
pistons and dseparate = 5 mm for the separately cast HPDC samples.
ε tme [%] To the samples of the bigger geometry, adapters with a diameter of
dadapter,piston = 16 mm were electron beam welded; the separately
Fig. 2. ra —eme
t hysteresis of separately cast HPDC sample in TMF/HCF test.
NTMF = 100.
cast samples were equipped with adapters of a diameter of
dadapter,separate = 12 mm by the same process, in order to fit the
specimens into the testing device. The TMF- and TMF/HCF-tests
and, consequently, the content of oxides and inclusions is reduced. were carried out on an Instron 8000 servo hydraulic testing ma-
The second option is to use high-pressure-die-casting (HPDC), chine. The specimens were heated by induction (Fig. 1c). Addi-
where higher cooling rates than in GDC and LPDC are possible. tional cooling sleeves were applied at the specimen shoulders to
Consequently, a higher amount of alloying elements can be added, assure sufficient cooling by heat conduction with well-defined
which increases the amount of sufficiently finely dispersed high- thermal boundary conditions. The temperature was measured with
temperature stable second phases in the microstructure, especially a ribbon-type thermocouple type K in the middle of l0. The exten-
at the surface zone of the molded product [3,4]. In HPDC the melt is someter with l0 = 12 mm was attached by two alumina rods with
first charged in a shot chamber before it is shot into the mold with rounded endings. This shape avoids premature crack initiation at
high pressure. the contact point of extensometer rod and samples. The TMF-tests
Prior to developing new alloys for LPDC and HPDC, the effects of were carried out in strain-control and an 180°. Out-of-phase cycle
the processing route were investigated at a reference material, the with a mechanical strain amplitude eme a;t equal to 50% of the thermal
well-known Al-piston-alloy AlSi12Cu4Ni3 in T5 heat treatment strain amplitude eth, a minimum temperature Tmin = 200 °C and a
70 A. Humbertjean, T. Beck / International Journal of Fatigue 53 (2013) 67–74

120 glass holder was developed, which supports the sample from the
100 minimum stress opposite side (Fig. 1e).
The pre-tests for each TMF test include, according to the Euro-
80 maximum stress pean code of practice for strain controlled TMF-testing [6–8]: The
60 static evaluation of Young’s modulus, 10 zero-stress stabilization
mean stress
40 cycles, 5 zero-stress cycles to measure the thermal strain, and 10
σ [MPa]

20 zero-stress verification cycles to optimize the temperature path.


The data acquisition is time-based and for TMF tests without
0
superimposed HCF loading one data point per second, for TMF tests
-20 with superimposed HCF loading one hundred data points per sec-
-40 ond was chosen (Fig. 1d). The TMF- and TMF/HCF data were eval-
-60 uated with an MS-Excel Macro. In the case of TMF/HCF tests, the
-0.4 -0.3 -0.2 -0.1 0 0.1 huge amount of data of the HCF hysteresis within a TMF cycle
ε tme [%] (Fig. 2) was condensed to a ‘‘maximum stress hysteresis’’ and
‘‘minimum stress hysteresis’’ of the TMF cycle (Fig. 3a and b). Using
(a) these data, the mean stress hysteresis was calculated (Fig. 3a and
b).
120
For characterization of the microstructure light-optical micros-
100 minimum stress copy was used. The sample preparation included cutting by a dia-
80 maximum stress mond saw, hot embedding, grinding, and polishing. The polishing
60 was carried out ‘‘dry’’, which means that oil, instead of a water-
mean stress based solution, was used as lubricant because water-based polish-
40
σ [MPa]

ing solutions caused stain on porous material such as the LPDC and
20 HPDC samples.
0
-20 3. Results and discussion
-40
-60 3.1. Microstructure in the as-cast state
-0.4 -0.3 -0.2 -0.1 0 0.1
ε tme [%] Microscopic analysis of AlSi12Cu4Ni3 was carried out for all
manufacturing processes to identify their influence on microstruc-
(b) ture. The microstructure of AlSi12Cu4Ni3, GDC (Fig. 4) shows a
homogeneous distribution of primary Si and intermetallic phases.
Fig. 3. (a) ra  eme hysteresis of GDC sample in TMF/HCF test. eHCF a;t ¼ 0:05%,
t
NTMF = 100. (b) ra  eme hysteresis of separately cast HPDC sample in TMF/HCF test.
The morphology and size of primary Si is similar to that of the
t
eHCF
a;t ¼ 0:05%, N
TMF
= 100. same alloy manufactured by LPDC (Fig. 5). However, compared to
GDC the Al mixed crystal formed larger dendrite arms in LPDC.
In Fig. 6, a cross-section of separately cast HPDC specimen is dis-
maximum temperature Tmax = 440 °C (Fig. 1d) was applied. The played. Compared to the GDC and LPDC material, the microstruc-
TMF cycle time is tTMF = 121 s, with a heating- and cooling rate of ture is much finer. Additionally a pronounced gradient of the
DT = 10 K/s and a dwell time at Tmax of tTmax = 60 s and at Tmin of microstructure within the sample is visible with very fine grains
tTmin = 5 s, respectively. In the TMF/HCF tests, a superimposed close to the surface and bigger grains in the center regions. Com-
HCF-loading was applied with a frequency of 10 Hz and strain pared to this, the microstructure in the interior of HPDC pistons
amplitudes between eHCF a;t = 0.03% and 0.05%. All tests were termi- shows distinctly bigger grains and pores (Fig. 7). However, the
nated at a maximum stress drop to 50% of the value measured dur- microstructure in the interior of the HPDC piston is still finer than
ing the first cycle, or after fracture of the specimen. The small that observed in the GDC or LPDC material. In the very fine surface
samples from the HPDC-process required a special solution for zone of the HPDC material no primary Si was formed and the Al
the extensometer attachment: To avoid bending deformation of mixed crystals have a globular dendritic structure, surrounded by
the test piece due to the contact pressure of the extensometer, a a eutectically solidified melt (Fig. 6). The porosity of HPDC and

Fig. 4. Light optical micrographs of GDC AlSi12Cu4Ni3, taken from a piston; white: a Al-matrix; blue: primary Si, different gray tones: multi-component phases containing
the elements Al, Mg, Cu, Ni, Si, Fe, Mn.
A. Humbertjean, T. Beck / International Journal of Fatigue 53 (2013) 67–74 71

Fig. 5. Light optical micrographs of LPDC AlSi12Cu4Ni3, taken from a piston.

Fig. 6. Light optical micrographs of HPDC AlSi12Cu4Ni3, separately cast sample.

Fig. 7. Light optical micrographs of HPDC AlSi12Cu4Ni3, taken from a piston.

LPDC specimens was higher compared to the GDC manufactured Furthermore, the HDPC process leads to high gas pressure in the
material. In the LPDC pistons, pores were mainly found in the cen- pores which supports the formation of a crack network during
ter of the thickest areas. The pores in the HPDC samples were finely TMF/HCF testing (see Section 3.3) [9]. The LPDC material shows lar-
dispersed. However, they also concentrate in the center of the ger dendrites than the GDC material, which refers to the smoother
areas with higher thickness, i.e. in the shoulders of the separately filling of the mold and a more directed crystal growth [10]. The rel-
cast samples and in the pistons up from 5 mm below the surface. atively high porosity is caused by suboptimal feeding of the mold.
The homogeneous morphology of the GDC microstructure is the Future process optimization will result in distinctly lower porosity.
result of alloy and process optimization for many years. For the In case of HPDC, the process itself has to be changed to reduce the
same reason, the porosity of GDC cast pistons is very low. The amount of pores and of the gas pressure within them. This can be
porosities of LPDC and HPDC cast AlSi12Cu4Ni3 are high, since realized by evacuating the mold before casting [11].
each process needs to be further improved together with the ap-
plied alloys to get sufficiently low porosity for application in pis- 3.2. Lifetime TMF and TMF/HCF
tons for car engines. The HPDC porosity is, as the microstructure,
finer compared to the LPDC cast pistons. This is due to the faster The TMF test results are presented in Fig. 8 as a Woehler
filling and solidification in the HPDC process compared to GDC. diagram, in which the HCF amplitude was plotted against lifetime
72 A. Humbertjean, T. Beck / International Journal of Fatigue 53 (2013) 67–74

0.06 demonstrates the big influence of the fine grained microstructure


GDC in the surface zone, and of the lower porosity of the separately cast
0.05 HPDC
HPDC specimen. The GDC samples reached TMF lifetimes between
ε a,t me, HCF [%]

HPDC separate
0.04 2000 and 4500 cycles. The LPDC and separately cast HPDC samples
LPDC
reached TMF-lifetimes within the same range. Nevertheless, the
0.03
highest lifetimes were reached by the GDC material. In the TMF
0.02 tests with superimposed HCF loading of eHCF a;t ¼ 0:05%, the GDC

0.01 samples reached lifetimes between 500 and 1000 TMF cycles.
The separately cast HPDC specimen achieved 520 TMF cycles for
0 the same eHCF HCF
a;t . For ea;t ¼ 0:03% the lifetime was increased up to
1 10 100 1000 10000
3000 TMF cycles.
lg NfTMF Fig. 9 shows the minimum-, maximum, and mean-stress plotted
against the TMF cycle number for pure OP-TMF-loading of all
Fig. 8. Superimposed HCF amplitude vs. number of TMF cycles to failure for investigated materials. The GDC sample reached the highest life-
AlSi12Cu4Ni3 manufactured in GDC, LPDC, and HPDC (from piston and separately
cast specimen).
time and continuous softening of the material until failure was
monitored: The maximum stress at the beginning of the test was
about 80 MPa. During the 4500 TMF cycles the maximum stress
decreased to approximately 60 MPa before failure. The LPDC sam-
GDC
120 ple started at 80 MPa as well. The subsequent softening was similar
HPDC piston
to the GDC material until macro crack growth occurred which is
100 HPDC separate
indicated by the stress-drop at N  2000. The separately cast HPDC
σ max, σ m, σ min [MPa]

LPDC
80 had a similar lifetime of about 2000 TMF cycles, but the maximum
stress in the first cycles was 100 MPa. Even though softening pro-
60
ceeded faster compared to the GDC and LPDC material, the higher
40 initial stress demonstrates the potential of the fine grained HPDC
20 microstructure for application under service conditions inducing
high peak stress. The r  eme hysteresis of maximum, mean and
0 t
minimum stress (see Section 2) for the GDC (Fig. 3a) and HPDC
-20 (Fig. 3b) material underline this potential: In both diagrams, the
-40 TMF cycle number 100 is plotted. The strain range of the GDC is
1 10 100 1000 10000 slightly lower than that of the HPDC-material, indicating a lower
lg N TMF thermal expansion coefficient of the GDC alloy which is caused
by its higher content of primary Si. On the other hand, the stress
Fig. 9. Minimum-, maximum- and mean-stress vs. TMF cycles for AlSi12Cu4Ni3 range induced by the superimposed HCF loading at a given temper-
manufactured in GDC, LPDC, and HPDC (from piston and separately cast specimen),
ature, e.g. of the TMF cycle is higher for the GDC-material than for
TMF tests without superimposed HCF loading.
the HPDC material. The lower stress range of the HPDC material
might be caused by the earlier and stronger compressive plastic
represented by the number of TMF cycles to failure N TMF
f . Therefore, deformation during heating which leads to lower minimum stress
the results of TMF tests without superimposed HCF are located on compared to the GDC specimen. The additionally tested HPDC
the N TMF
f -axis. Note that in case of the HPDC material separately samples from pistons show very short lifetimes in TMF and TMF/
cast specimens as well as samples taken from pistons were tested, HCF testing, due to high porosity and missing of the fine grained
whereas all GDC and LPDC specimens were taken from pistons. The surface zone which is present in the separately cast test pieces.
number of cycles to failure of the HPDC samples machined from The LPDC specimens were not tested in TMF/HCF, as limited mate-
pistons was in TMF and TMF/HCF tests significantly lower than rial was available and the best samples with lowest porosity were
that of the GDC and the separately cast HPDC samples. This used for TMF tests without superimposed HCF.

load axis load axis

Fig. 10. Light optical micrographs longitudinal section of GDC AlSi12Cu4Ni3 after TMF testing, N TMF
f ¼ 2021; left: crack propagation through the second phases; right: broken
primary Si phase.
A. Humbertjean, T. Beck / International Journal of Fatigue 53 (2013) 67–74 73

load axis load axis

changed eutectic
structure

Fig. 11. Light optical micrographs longitudinal section of HPDC AlSi12Cu4Ni3, separately cast sample, after TMF testing, N TMF
f ¼ 2200; left: overview of the cross section with
fracture surface on the right side; right: details of the microstructure in the center of the specimen.

load axis load axis

Fig. 12. Light optical micrograph longitudinal section of HPDC AlSi12Cu4Ni3, separately cast sample, after TMF/HCF testing, eHCF
a;t ¼ 0:05%, Nf = 3040; left: overview of the
cross section with fracture surface on the right side; right: details of the microstructure in the center of the specimen.

3.3. Damage mechanisms alteration of the fine globular dendritic structure is caused by the
time duration at high temperatures which exceeded in these tests
In the GDC material, crack initiation was found in primary Si or (Figs. 11 and 12) 30 h at Tmax = 440 °C.
at its interface to the Al matrix (Fig. 10). Crack propagation mostly
occurred along phase boundaries as well as through intermetallic
second phases and primary Si as also observed at Al–Si–Mg alloys
4. Summary and conclusions
under superimposed-mechanical fatigue and high cycle fatigue
loading [12] and at near eutectic Al–Si alloys under three various
1. The LPDC and HPDC processes are shown to be competitive to
fatigue tests [13]. Especially bigger primary Si particles close to
GDC, since in TMF and TMF/HCF lifetimes of AlSi12Cu4Ni3 pro-
the surface were already broken before they were reached by the
duced by these processes are within the same range of the sam-
main crack. The separately cast HPDC samples, with their fine
ples produced by GDC. However, further optimization of the
porosity and much smaller primary Si particles even in the interior
processes, especially considering porosity, is essential.
of the specimens, showed a completely different failure mecha-
2. In GDC and LPDC samples, crack initiation starts from the sur-
nism which may be explained from the microstructure after TMF
face. In the separately cast HPDC specimens this is not the case,
testing (Figs. 11 and 12) as follows: During TMF and TMF/HCF load-
due to the fine grained, nearly defect free surface zone. Conse-
ing cracks initiate at the pores while expansion of the enclosed gas
quently, the HPDC samples reached considerably good TMF
during the high temperature phase of the cycle accelerates crack
and TMF/HCF results even though a crack network is formed
propagation and leads to a pronounced swelling of the sample
at the internal pores during early stages of testing.
[9,14]. Only the very fine and pore-free surface zone is not affected
3. The enclosed air in the HPDC material caused swelling and for-
by this process. This zone is, due to its lower porosity, not affected
mation of a crack network during TMF and TMF/HCF testing,
by the internal degradation for a relatively long time while in the
mainly due to the high internal pressure of the pores, which
later stages of the damage process internal cracks penetrate into
is originated from the HPDC process.
the surface zone and finally lead to a massive stress-drop (Fig. 9)
4. For small specimens, a special glass holder avoiding bending
followed by fracture. Furthermore, the microstructure of the HPDC
due to the force of the extensometer on the specimen was
samples is altered during the TMF and TMF/HCF tests by transfor-
invented. This glass holder prevented an undesired deformation
mation of the initial eutectic structure (Fig. 6) into Al mixed crystal,
of the specimen during testing and could be used for similar
coarser Si particles, and intermetallic phases (Figs. 11 and 12). This
applications.
74 A. Humbertjean, T. Beck / International Journal of Fatigue 53 (2013) 67–74

Acknowledgements [7] Beck T, Hahner P, Kuhn HJ, Rae C, Affeldt EE, Andersson H, et al. Thermo-
mechanical fatigue – the route to standardisation (‘‘TMF-Standard’’ project).
Mater Corros-Werkstoffe Und Korrosion 2006;57:53–9.
The present research work is funded by the Federal Ministry of [8] Beck T, Henne I, Lohe D. Lifetime of cast AlSi6Cu4 under superimposed
Education and Research, Germany (BMBF) within the WING pro- thermal–mechanical fatigue and high-cycle fatigue loading. Mater Sci Eng A –
Struct Mater Prop Microstruct Process 2008;483:382–6.
gram. The authors appreciate the support of Federal Mogul, MONE-
[9] Lu Y, Taheri F, Gharghouri MA, Han HP. Experimental and numerical study of
VA and KURTZ, who provided the AlSi12Cu4Ni3 test material. the effects of porosity on fatigue crack initiation of HPDC magnesium AM60B
alloy. J Alloys Compd 2009;470:202–13.
[10] Hou LG, Cai YH, Cui H, Zhang JS. Microstructure evolution and phase
References
transformation of traditional cast and spray-formed hypereutectic
aluminium–silicon alloys induced by heat treatment. Int J Miner Metall
[1] Haque MM, Maleque MA. Effect of process variables on structure and Mater 2010;17:297–306.
properties of aluminium–silicon piston alloy. J Mater Process Technol [11] Niu XP. Vacuum assisted high pressure die casting of aluminium alloys. J Mater
1998;77:122–8. Process Technol 2000, pp. 119–27.
[2] Woodthorpe J, Thomson R, Daykin C, Reed P. Cast Al–Si piston alloys. Ser Mater [12] Beck T, Lohe D, Luft J, Henne I. Damage mechanisms of cast Al–Si–Mg alloys
Sci Engng 2003:69–76. under superimposed thermal–mechanical fatigue and high-cycle fatigue
[3] Yamagata H, Kurita H, Aniolek M, Kasprzak W, Sokolowski JH. Thermal and loading. Mater Sci Eng A – Struct Mater Prop Microstruct Process
metallographic characteristics of the Al-20% Si high-pressure die-casting alloy 2007;468:184–92.
for monolithic cylinder blocks. J Mater Process Technol 2008;199:84–90. [13] Joyce MR, Styles CM, Reed PAS. Elevated temperature short crack fatigue
[4] Chen ZW. Skin solidification during high pressure die casting of Al–11Si–2Cu–Fe behaviour in near eutectic Al–Si alloys. Int J Fatigue 2003;25:863–9.
alloy. Mater Sci Eng A – Struct Mater Prop Microstruct Process 2003;348:145–53. [14] Lu Y, Taheri F, Gharghouri M. Study of fatigue crack incubation and
[5] Zeren M. The effect of heat-treatment on aluminum-based piston alloys. Mater propagation mechanisms in a HPDC AM60B magnesium alloy. J Alloys
Design 2007;28:2511–7. Compd 2008;466:214–27.
[6] Hahner P, Rinaldi C, Bicego V, Affeldt E, Brendel T, Andersson H, et al. Research
and development into a European strain-controlled thermo-mechanical code-
of-practice for fatigue testing. Int J Fatigue 2008;30:372–81.

You might also like