Professional Documents
Culture Documents
Use of Taguchi Method To Develop A Robust Design For The Magnesium Alloy Die Casting Process
Use of Taguchi Method To Develop A Robust Design For The Magnesium Alloy Die Casting Process
Abstract
This study applies the Taguchi method to optimize the process parameters for the die casting of thin-walled magnesium alloy parts in
computer, communications, and consumer electronics (3C) industries. The objectives of the Taguchi method for robust parameter design are
to establish the optimal combination of design parameters and to reduce the variation in quality from a minimum number of experiments.
This study investigates the effects of various die casting control parameters, including the die temperature, injection velocity, and cooling
time upon the surface warping of magnesium alloy die cast components. The present results concerning the die casting of magnesium alloy
personal data assistant (PDA) cover surfaces confirm the effectiveness of this robust design methodology.
© 2004 Elsevier B.V. All rights reserved.
0921-5093/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.msea.2004.03.006
D.H. Wu, M.S. Chang / Materials Science and Engineering A 379 (2004) 366–371 367
parameters (e.g. the controllable factors) to be established In the Taguchi method, a design parameter is considered
from a minimum number of experiments. It has been shown to be significant if its influence is large compared to the
that a robust measurement system will; (1) minimize vari- experimental error as estimated by the analysis of variance
ability as the input signal changes, (2) provide consistent (ANOVA) statistical method. If this is the case, the design
measurements for the same input, (3) continue to give ac- parameter is a fundamental factor in determining the optimal
curate readings as the input values change, (4) adjust the solution to the design problem.
sensitivity of the design, and (5) be robust when subjected r
n
to noise. SStotal = η2ij − nrη2m (3)
The main objective of the present study is to apply i=1 j=1
the Taguchi method to establish the optimal set of con-
L
trol parameters for the pressure die casting of thin-walled nr
magnesium alloy (AZ91D) parts. The Taguchi method for SSfactor = (ηk − ηm )2 (4)
L
parameter design is employed to determine the optimal k=1
combination of design parameters, including the die tem- DOF = L − 1 (5)
perature, the injection velocity, and the cooling time, such
that the warping of the cast components is minimized. SSfactor
Vfactor = (6)
Specifically, the effectiveness of the Taguchi method is DOF
demonstrated by considering the fabrication of personal Vfactor
data assistant (PDA) cover components. Ffactor = (7)
Verror
where SStotal is the total sum of squares, SSfactor the factorial
2. Taguchi method sum of squares, n the number of experiments, r the number
of specimens taken, DOF the number of degrees of freedom,
The Taguchi method which is one of the best robust design Vfactor the variance of the factor, Ffactor the F ratio of the
methods. The basic principle of this method is to develop an factor, ηm the mean S/N ratio of the factor in level m, and η
understanding of the individual and combined effects of var- is the mean of the factor.
ious design parameters from a minimum number of experi-
ments. The objectives of the Taguchi method for parameter
design are to establish the optimal combination of design 3. Robust design of magnesium alloy die
parameters and to reduce variations in the product quality. casting process
The Taguchi method employs a generic signal-to-noise
(S/N) ratio to quantify the present variation. Depending on The major principle of the robust design methodology is
the particular type of characteristics involved, different S/N to optimize the product quality and the process designs such
ratios may be applicable, including “lower is better” (LB), that they become insensitive to various causes of variation
“nominal is best” (NB), and “higher is better” (HB). without actually eliminating these causes. Robust design in-
The S/N ratio for the LB characteristics related to the volves two key tools, namely a S/N ratio to assess the qual-
present study is given by ity, and an OA table to facilitate the study of many design
n parameters simultaneously. The principal characteristics of
1 2
S/N = −10 log y (1) robustness for the magnesium alloy die cast PDA surface
n cover considered in the present study include a reduction in
i=1
the warping of the die cast product and a reduced sensitivity
where n is the number of simulation repetitions under the
to external noise.
same design parameter conditions, y represents the results
of measuring, and subscript i indicates the number of simu-
lation design parameters in the orthogonal array (OA) table. 3.1. Robust design procedures
The analysis of mean (ANOM) statistical approach is
used to obtain the optimal combination of design parame- In the thin-walled die casting process considered in the
ters. Eq. (1) is applied to construct the S/N response table present study, the warping of the PDA cover surface is treated
and response graph, which then enable the robust design for as the characteristic value, and the design intent is to reduce
the current problem to be derived. this value such that the warping is minimized. Therefore,
the current problem is defined as a static, smaller-the-better
L
1 A
problem, and the principal objective of the design problem
MAi = (S/N)Ai (2) is to maximize the S/N ratio. Previous studies have identi-
nA
i=1 fied the die cast process parameters which contribute to die
where MAi is the mean S/N ratio of factor A in level i, nA casting defects. These factors are presented in the form of
the number of appearances of factor A in the OA table, and a cause-effect diagram in Fig. 1. As illustrated in Table 1,
(S/N)Ai is the S/N ratio of factor A in level i. this study specifies eight principal design parameters, each
368 D.H. Wu, M.S. Chang / Materials Science and Engineering A 379 (2004) 366–371
Table 1
Design parameters and levels
Factors Description Level 1 Level 2
Table 2
S/N ratio of PDA warping defects
Experiment Factor (from Table 1) Measurement points (mm) (shown in Fig. 5) Average S/N
number
A B C D E F G H y1 y2 y3 y4 y5 y6
Table 3
Factors response table
Control factor A B C D E F G H
Level 1 (L1) 13.848 12.795 13.420 12.724 12.204 12.822 12.783 13.185
Level 2 (L2) 11.598 12.651 12.026 12.722 13.242 12.624 12.663 12.261
l L1–L2 l 2.2494 0.1434 1.3935 0.0022 1.0377 0.1979 0.1194 0.9235
Rank 1 6 2 8 3 5 7 4
Table 4
Analysis of variance (ANOVA)
Factors Description Level (S/N) DOF SS Variance Contribution σ (%) Confidence (F ratio)(%)
1 2
contribution and variance of factors A, C, E, and H indicate turbances. The present robust study reveals several useful
that outlet temperature of temperature controller, cooling results, namely that outlet temperature of temperature con-
method of casting, cooling time and plunger speed are troller, cooling method of casting, cooling time and plunger
significant. Meanwhile, the F ratio reveals that all factors speed have a more significant influence upon product qual-
other than die temperature are of high confidence. Fi- ity than when break the runner, die temperature, delay time
nally, the standard deviation of the system is 0.0385 mm. of ejection and die temperature distribution. Furthermore,
The results of this study reveal that design factors when the outlet temperature of the temperature controller is the
break the runner (B), die temperature (D), delay time of most important factor influencing the warping of the cover
ejection (F), and die temperature distribution (G) have a plate, while the die temperature has the least influence. In
less significant influence upon the warping of the PDA other words, a high-quality die cast product can be pro-
cover. duced by accurately controlling outlet temperature of tem-
perature controller, cooling method of casting, cooling time
4.3. Confirmation of improvement and discussions and plunger speed.