International Journal of Advanced Computer Science, Vol. 2, No. 11, Pp. 441-448, Nov., 2012.

Manuscript
Received:
8, Feb., 2012
Revised:
19, Jun., 2012
Accepted:
26, Aug., 2012
Published:
15, Dec., 2012

Keywords
I nterface slip
damping,
Welded joint,
Surface
roughness,
Amplitude,
Tack length,
Response
surface
methodology.



Abstract Response surface
methodology with BoxBenhken (BB)
design of experiment approach has been
utilized to study the mechanism of
interface slip damping in layered and
jointed tack welded beams with varying
surface roughness. The design utilizes the
initial amplitude of excitation, tack length
and surface roughness at the interfaces to
develop the model for the logarithmic
damping decrement of the layered and
jointed welded structures. Statistically
designed experiments have been
performed to estimate the coefficients in
the mathematical model, predict the
response, and check the adequacy of the
model. Comparison of predicted and
experimental response values outside the
design conditions have shown good
correspondence, implying that empirical
model derived from response surface
approach can be effectively used to
describe the mechanism of interface slip
damping in layered and jointed tack
welded structures.


1. Introduction
Over the years, the researchers have emphasized their
studies on the development of mathematical models for the
mechanism of damping and techniques to improve the
damping capacity of laminated structures to control the
adverse effects of vibrations. Following the requirements of
modern technology, there has been significant increase in
demand to design, develop and fabricate machine tools,
space structures, high speed automobiles, etc. to meet the
global demand. The manufacture of such structures also
requires high damping capacity and stiffness with light
weight for its effective use. Such requirements necessitated
and popularized the use of welded, bolted and riveted
layered beams as structural members with high damping
capacity. In the alternative, cast structures can be used, but
unfortunately, these are more expensive to manufacture and
as a result, the deployment of welded, bolted and riveted
multi-layered beam structures is becoming increasingly
common in such industries.
Joints which contribute significantly to the inherent


B. Singh is with the National Institute of Technology, Rourkela, and
Odisha, India (E-mail: bhagat.mechanical@gmail.com)
B.K. Nanda, National Institute of Technology, Rourkela, and Odisha,
India (E-mail: bknandanit@gmail.com)
damping are present in most of the fabricated structures.
Joints have a great potential for reducing the vibration
levels of a structure and have attracted the interest of many
researchers such as; Earles and Mott [1], Earles and
Mansoor [2], Beards [3], Beards and Williams [4], Beards
and Imam [5], Beards and Neroutspoulos [6], Beards and
Robb [7], Beards and Woowat [8].
The earlier workers such as; Goodman and Klumpp
[9], Masuko et al. [10], Motosh [11], Nishiwaki et al.
[12]and Nanda and Behera [13] have presented different
techniques of improving the damping capacity of layered
and jointed bolted structures without considering the
effects of surface irregularities and asperities.
Many comprehensive review papers on joints and
fasteners have appeared in recent years. Berger [14] has
studied the effect of microslip on the passive damping of a
jointed structure. Mayer and Gaul [15] have discussed the
segment- to-segment contact elements of a structure having
both linear and nonlinear constitutive contact behavior in
normal and tangential directions, including nonlinear
micro-slip behavior. Mayer and Gauls contact element
included a massing element (a variant of the Jenkins and
Iwan models) to represent the tangential contact behavior.
Csaba [16] proposed a microslip friction model with a
quadratic normal load distribution based on the model
developed by Menq et al. [17] where the shear layer has
been neglected for simplicity. Song et al. [18] have devised
a model to estimate the friction in the joints by adding a
parallel spring to the parallel-series Iwan model. Bowden
and Tabor [19] have developed a friction model to compute
the effective coefficient of friction resulting from
deformation of asperities. A similar strategy has been
adopted by Olofsson [20] to implement an asperity based
friction model for assessing the frictional energy dissipation
due to micro-slip. Berger et al. [21] have formulated a new
model based on microslip approach for accurate
determination of system dynamic response under a variety
of contact conditions.
Although a lot of work has been carried out on the
damping capacity of bolted structures, but a little amount of
work has been reported till date on the mechanism of
damping in layered and jointed welded structures. Anno et
al. [22] have established that the steel plates welded with
plug joints exhibit higher damping compared to other forms
of welded joints.
RSM approach has so far not been used by any
researcher to study the mechanism of damping of layered
and jointed welded structures. Response surface
methodology (RSM) explores the relationships between
several control variables to develop a mathematical model
Damping mechanism in welded structures
B.Singh & B.K.Nanda
International Journal of Advanced Computer Science, Vol. 2, No. 11, Pp. 441-448, Nov., 2012.
International Journal Publishers Group (IJPG)

442
for the response. However, an experimental design involves
choosing the appropriate combination of various factors and
the levels of each factor for developing a model. Since
experimental runs cost time and money, it is pertinent to
minimize the number of runs without compromising the
desired goals. In order to achieve this, some strategies such
as; full factorial (FF), BoxBenhken (BB), central
composite designs (CCD) etc. are frequently used.
The BoxBenhken (BB) design of experiment (DOE)
allows the designer to utilize 3 levels of each factor (with
each factor placed at one of each equally spaced value to
ensure orthogonality and near rotatability) to adequately
quantify second-order response models in15 runs, inclusive
of 3-replicated center points of a cubical design region.
However, full factorial (FF) designs use different levels of
different factors, with every level of each factor combining
with those of other factors. They are good for first-order
response models, enabling the estimation of main and
interaction effects. However, as the number of factors and
levels increase the number of requisite runs becomes cost
and time prohibitive, and therefore the Taguchi designs, and
fractional factorial design are utilized for product
improvement and cost reduction. However, the Taguchi
designs suffer a major inadequacy of handling interaction
and confounding effects. Hunter [23], Montgomery [24],
and Sukthomya and Tannock [25] have highlighted other
weaknesses of the Taguchi designs such as; (a) unnecessary
complication using inner and outer arrays,(b)
non-recognition of randomized experiments to save the cost
of changing level settings,(c) non- applicability of
orthogonal arrays to processes involving factors that vary
with time and cannot be quantified exactly, and noise
factors may not always be independent of one another and
(d) the techniques requires the designer to be aware of all
control and noise factors affecting a product or process.
However, a 2-level FF- design with center points can detect
individual quadratic effects but cannot estimate them
effectively. On the other hand, a 3-level FF-design is
appropriate for quadratic effects and hence optimization
(using 3
3
= 27 runs) requires a higher cost and time. The
central composite designs (whether central composite
circumscribed (CCC), central composite inscribed (CCI) or
central composite face centered (CCF) will require 5 levels
of each factor (3 levels for CCF) with a total of 20 runs, and
hence higher cost and time. The BB-design is known to give
about 98% efficiency for predicting second-degree models
in 3 factors, a level that can only be achieved in FF- design
in1-factor.
In the present work response surface methodology
with BoxBenhken (BB) design of experiment approach has
been utilized to study the mechanism of interface slip
damping in layered and jointed tack welded beams with
varying surface roughness. The design utilizes the initial
amplitude of excitation, tack length and surface roughness
at the interfaces to develop the model for the logarithmic
damping decrement of the layered and jointed welded
structures. Statistically designed experiments have been
performed to estimate the coefficients in the mathematical
model, predict the response, and check the adequacy of the
model. Furthermore, the effect of the aforesaid parameters
on the damping rate of such structures is investigated and
discussed in detail.
2. Experimental Details
A. Mathematical Models and Experimental Design
The logarithmic damping decrement of layered and
jointed welded structures may be affected by operational
control variables such as; tack length, initial amplitude of
excitation and surface roughness. The relationship of the
logarithmic damping decrement with respect to the first two
could be estimated by a first-degree model but not with
respect to the others. Hence, a second-degree model
becomes expedient. A suitable second-order polynomial
involving main, interaction and quadratic components have
been selected based on the estimation of statistical
parameters, such as; coefficient of determination (R
2
),
adjusted R
2
, standard error of regression and analysis of
variance (ANOVA). In the present work, the input variables
are tack length (L), initial amplitude of vibration (a), and
surface roughness (Ra) and the output response is the
logarithmic damping decrement (). A second order
response surface model is usually expressed as:
2
0
1 1 1 1
k k k k
i i i i ij i j
i i i j
y x x x x | | | | c
= = = =
= + + + +

(Equ. 1)

where,
0
,
i
(i = 1, 2 . . . k) and
ij
(i = 1, 2 . . . k, j = 1, 2 . . .
k) are the unknown regression coefficients to be estimated
by using the method of least squares. In this expression, is
experimentally random errors and x
1
, x
2
. . . x
k
are the
input variables that influence the response (y), and k is the
number of input factors. The method of least square is used
to estimate the coefficients of the second order model. The
response surface analysis is then done in terms of the fitted
surface. The general form of the polynomial expression (1)
could be simplified to expression (2) for the logarithmic
damping decrement () response as given by;
0 1 2 3 12
2 2 2
23 13 11 22 33
L a Ra L a
a Ra L Ra L a Ra
y A A X A X A X A X X
A X X A X X A X A X A X
= + + + + +
+ + + +
(Equ. 2)

where,
0
A is the intercept,
1
A ,
2
A ,
3
A are the linear (main)
effects,
12
A ,
23
A ,
13
A are the cross product (interaction)
effects,
11
A ,
22
A ,
33
A are the quadratic effects.
L
X ,
a
X ,
Ra
X are the coded control variables corresponding
to the tack length (L), initial amplitude of excitation (a) and
surface roughness (Ra) respectively. The transforming
equations for each coded control variable are given as:
30
,
10
L
L
X

=
0.2
,
0.1
a
a
X

=
1.47
0.55
Ra
Ra
X

= (Equ. 3)
Hence, three orthogonal coded factor levels -1, 0, -1
with tack length L = 20, 30, 40 mm coded as
L
X = -1, 0, 1;
initial amplitude of excitation a = 0.1, 0.2, 0.3 mm coded as
B. Singh et al.: Damping mechanism in welded structures.
International Journal Publishers Group (IJPG)

443
a
X = -1, 0, 1; surface roughness Ra = 0.92, 1.47, 2.02 m
coded as
Ra
X = -1, 0, 1 respectively, has been used in this
system. The important coded and uncoded factors and their
levels are shown above in Table 1.
TABLE 1
IMPORTANT FACTORS AND THEIR LEVELS

TABLE 2
LOGARITHMIC DAMPING DECREMENT RESPONSE FOR BB
DESIGN OF EXPERIMENT

The lack of fit and the degree of significance of the model
have been tested by analysis of variance (ANOVA) using
the software MINITAB-14. The BB design of experiment
runs with independent control variables in coded, uncoded
forms and response are shown in Table 2.
B. Surface Roughness (SR) Measurements
The specimens are prepared from commercial mild
steel flats and the roughness measurement at the interfaces
have been carried out using a portable stylus type
profilometer, Talysurf (Taylor Hobson, Surtronic 3+). The
profilometer has been set to a cut-off length of 0.8 mm,
filter 2CR with traverse speed of 1mm/second having
evaluation length of 4 mm. Roughness measurements, in the
transverse direction, on the specimens have been repeated at
least for five times and the average of these measurements
has been recorded. The measured profile is digitized and
processed through the advanced surface finish analysis
software Talyprofile for the evaluation of the roughness
parameters. Surface roughness is expressed as the
irregularities resulted from the various machining
operations and is usually denoted as Ra. Theoretically, Ra
is the arithmetic average value of the departure of the
profile from the mean line throughout the sampling length.
C. Logarithmic Damping Decrement()Measurements
An experimental set-up as shown in Fig. 1 has been
fabricated to conduct the experiments.


Fig. 1 Experimental set-up
The specimens are prepared from the stock of mild
steel flats with different surface roughness by tack welding
two layers of various thickness and cantilever length. The
cantilever specimens are then excited transversally at the
amplitudes of 0.1, 0.2, 0.3 mm at their free ends with the
help of a spring arrangement. The free vibration is sensed
with a vibration pick-up and the corresponding signal is fed
to a digital storage oscilloscope to measure the amplitudes
of the first cycle (a
1
), last cycle (a
n+1
) and the number of
cycles (n) of the steady signal. The logarithmic damping
decrement is then evaluated from the expression = ln
(a
1
/a
n+1
)/n. The structure considered in the present work is a
lightly damped structure where the initiation of slip is
delayed and reduced at the higher modes of vibration
resulting in lower damping rate. As established by
Nishiwaki et al. [12], the logarithmic damping decrement is
lower at the higher modes compared to first mode.
Moreover, the present analysis is based on experimental
results where the effect of the higher modes is automatically
accounted for in the recorded time domain curves which is
further used for calculating the logarithmic damping
decrement.
Runs Factors
Coded
Factors
Response

L(mm) A(mm) Ra(m) X
L
X
a
X
Ra

1 30 0.1 2.02 0 -1 1 0.00348
2 20 0.1 1.47 -1 -1 0 0.00408
3 30 0.1 0.92 0 -1 -1 0.00350
4 30 0.2 1.47 0 0 0 0.00189
5 40 0.2 2.02 1 0 1 0.00182
6 30 0.2 1.47 0 0 0 0.00191
7 20 0.2 2.02 -1 0 1 0.00226
8 20 0.2 0.92 -1 0 -1 0.00220
9 30 0.2 1.47 0 0 0 0.00193
10 30 0.3 0.92 0 1 -1 0.00130
11 40 0.3 1.47 1 1 0 0.00125
12 40 0.2 0.92 1 0 -1 0.00180
13 30 0.3 2.02 0 1 1 0.00131
14 40 0.1 1.47 1 -1 0 0.00312
15 20 0.3 1.47 -1 1 0 0.00152

Factor Notation Levels
uncoded coded
Tack
length

L 20 30 40 -1 0 1
Amplitude a

0.1 0.2 0.3 -1 0 1
Surface
roughness
Ra

0.92 1.47 2.02 -1 0 1

International Journal of Advanced Computer Science, Vol. 2, No. 11, Pp. 441-448, Nov., 2012.
International Journal Publishers Group (IJPG)

444
3. Results and Discussion
A. Development of Response Surface Model for
Logarithmic Damping Decrement ()
The results from the experimental runs performed as
per the experimental plan are shown in Table 2. These
results are used as the input data to the Minitab 14 software
for further analysis following the steps outlined in Section 2.
Without performing any transformation on the response,
examination of the Fit Summary output have revealed that
the quadratic model is statistically significant for the
response and therefore it can be used for further analysis.
The following equations are the final empirical models in
terms of coded factors for the logarithmic damping
decrement as given by;
2 2
2
0.00191 0.000259 0.0011
0.000009 0.000103 0.00048
0.000007 0.000173
0.00001 0.000007
L a
Ra L a
Ra L a
L Ra a Ra
X X
X X X
X X X
X X X X
o =
+ + +
+ +
+
(Equ. 4)
B. Analysis of Variance (ANOVA) for ()
The tests for significance of the regression model,
significance of individual model coefficients and lack-of-fit
are to be performed to ensure the adequacy of the model.
An ANOVA table is commonly used to summarize the tests
performed. Table 3 shows the ANOVA table for response
surface quadratic model for logarithmic damping
decrement.
TABLE 3
ANALYSIS OF VARIANCE FOR LOGARITHMIC DECREMENT
(FULL MODEL)
Estimated Regression Coefficients for are shown in
Table 4 which depicts both the significant and
insignificant parameters.

TABLE 4
ESTIMATED REGRESSION COEFFICIENTS FOR LOGARITHMIC
DECREMENT (FULL MODEL)

TABLE 5
ANALYSIS OF VARIANCE FOR LOGARITHMIC DECREMENT
(REDUCED MODEL)

The value of P in Table 3 for model is less than 0.05
which indicates that the model is significant, which is
desirable as it indicates that the terms in the model have a
significant effect on the response. The value of P for the
lack of fit in the Table 3 is more than 0.05 indicating that it
is insignificant. This shows that further improvement is
required in the model by eliminating the insignificant
interaction terms from the model. Furthermore, the
significance of each coefficient in the full model has been
Source DF Seq SS
10
-4
Adj SS
10
-4


Adj MS
10
-4

P
Regression 5 0.11 0.11 0.02 0.000
Linear 2 0.10 0.10 0.05 0.000
Square 2 0.01 0.01 0.00 0.000
Interaction
1 0.00 0.00 0.00 0.000
Residual
Error
9 0.00 0.00 0.00
Lack of Fit 3 0.00 0.00 0.00 0.005
Pure Error 6 0.00 0.00 0.00
Total 14 0.11

Term Coef SE Coef P
Constant 0.001910 0.000038 0.000
L
X -0.000259 0.000023 0.000
a
X -0.001100 0.000023 0.000
Ra
X 0.000009 0.000023 0.719
L
X
L
X 0.000103 0.000034 0.029
a
X
a
X 0.000480 0.000034 0.000
Ra
X
Ra
X 0.000007 0.000034 0.836
L
X
a
X 0.000173 0.000033 0.003
L
X
Ra
X -0.000010 0.000033 0.771
a
X
Ra
X 0.000007 0.000033 0.827

Source DF Seq SS
10
-4
Adj SS
10
-4


Adj MS
10
-4

P
Regression 9 0.11 0.11 0.01 0.000
Linear 3 0.10 0.10 0.03 0.000
Square 3 0.01 0.01 0.00 0.000
Interaction 3 0.00 0.00 0.00 0.017
Residual
Error
5 0.00 0.00 0.00
Lack of Fit 3 0.00 0.00 0.00 0.056
Pure Error 2 0.00 0.00 0.00
Total 14 0.11

B. Singh et al.: Damping mechanism in welded structures.
International Journal Publishers Group (IJPG)

445
examined by the P-values and the results are listed in Table
4.
The resulting ANOVA table for the reduced quadratic
model for logarithmic damping decrement has been shown
in Table 5. Results from Table 5 indicate that the model is
still significant. The response regression coefficients of the
terms in the reduced model are shown in Table 6.
TABLE 6
ESTIMATED REGRESSION COEFFICIENTS FOR LOGARITHMIC
DECREMENT (REDUCED MODEL)


C. Surface and Contour Plots for Logarithmic Decrement
The effects of the parameter interactions in the form of
response surfaces and contour plots on the logarithmic
damping decrement are shown in Figs. 24.

Fig. 2-a Effect of tack length and surface roughness on the logarithmic
damping decrement (): Response surface plot

Fig. 2-b Effect of tack length and surface roughness on the logarithmic
damping decrement (): Contour plot

Fig. 3-a Effect of amplitude and surface roughness on the logarithmic
damping decrement (): Response surface plot

Fig. 3-b Effect of amplitude and surface roughness on the logarithmic
damping decrement (): Contour plot

Fig. 4-a Effect of amplitude and tack length on the logarithmic damping
decrement (): (a) Response surface plot

Fig. 4-b Effect of amplitude and tack length on the logarithmic damping
decrement (): Contour plot.
Term Coef SE Coef T P
Constant 0.001915 0.000024 79.530 0.000
L
X
0.000259 0.000018 -14.597 0.000
a
X
0.001100 0.000018 -62.079 0.000
L
X
L
X
0.000103 0.000026 3.925 0.003
a
X
a
X
0.000480 0.000026 18.443 0.000
L
X
a
X
0.000173 0.000025 6.895 0.000
S = 0.00005011 R-Sq = 99.8% R-Sq(adj) = 99.7%

International Journal of Advanced Computer Science, Vol. 2, No. 11, Pp. 441-448, Nov., 2012.
International Journal Publishers Group (IJPG)

446
The initial amplitude of excitation of free vibration is
an important parameter influencing the logarithmic damping
decrement of layered and jointed welded structures. The
logarithmic damping decrement of such structures decreases
with an increase in amplitude of excitation. This decrease is
due to introduction of higher strain energy into the system
compared to that of the dissipated energy due to interface
friction.
The logarithmic damping decrement increases with the
decrease in the length of the tack. The frequency of
vibration depends on the stiffness and mass. With the
decrease in the length of the tack weld, the static bending
stiffness (k = 3EI/l
3
) remains same but the overall mass
decreases because of the less deposition of weld material.
The frequency of vibration being directly proportional to the
square root of the static bending stiffness and inversely
proportional to the square root of the mass increases due to
decrease in mass deposition in case of tack welded joints.
Hence the product o is enhanced as established by
[26] resulting in an increase in the logarithmic damping
decrement as evident from Figs. 2-4.
D. Main Effects of Interaction on Logarithmic Decrement
The main effects plot for has been shown in Fig. 5.
These plots are used to compare the changes in the level
means to see which factors influence the response the most.
The surface roughness effect line is almost parallel to the
X-axis which denotes that the effect of surface roughness on
is almost negligible. Furthermore, the slope of amplitude
is more than the tack length with respect to the X-axis
which shows that the effect of amplitude is more
pronounced than tack length on the response as evident
from Fig .5.

Fig. 5 Main effects plot for logarithmic decrement ()
E. Residual Plots for Logarithmic Decrement
The regression model is used for determining the residuals
of each individual experimental run. The difference between the
measured values and predicted values are called residuals. The
residuals are calculated and ranked in ascending order. The normal
probabilities of residuals for both the responses are shown in Fig. 6.
The normal probability plot is used to vary the normality
assumption. As shown in Fig. 6, the data are spread roughly
along the straight line for . Hence, it is concluded that the data
are normally distributed.

Fig. 6 Normal probability plot of the residuals
Fig. 7 Residual versus order of the data (Response is )

Fig. 8 Residuals versus the fitted values (Response is ).
Figure 7 shows the residuals against the observation
order. This figure is used to show the correlation between the
residuals. From the Figure 7, it is emphasized that the tendency
to have runs for positive and negative residuals indicate the
existence of a certain correlation and the plots also show that the
residuals are distributed evenly in both positive and negative
directions along the run. Hence, the data can be said to be
independent.
Figure 8 indicates the residuals versus fitted values, which
shows the maximum variation in the range of -0.000075 to
0.000075 for the response between the measured and the
B. Singh et al.: Damping mechanism in welded structures.
International Journal Publishers Group (IJPG)

447
fitted values. These plots do not reveal any obvious pattern and
therefore the fitted model is ample
F. Checking the Adequacy of Mathematical Models
The goodness of fit for the mathematical models has
also been tested by coefficient of determination (R
2
) and
adjusted coefficient of determination (R
2
adj
). The R
2
is the
proportion of the variation in the dependent variable
explained by the regression model. On the other hand, R
2
adj

is the coefficient of determination adjusted for the number
of independent variables in the regression model. Unlike R
2
,
the R
2
adj
may decrease if the variables are incorporated in
the model that does not add significantly to the model fit.
The R
2
and R
2
adj
values of reduced mathematical model for
is found to be 99.8 and 99.7% respectively which clearly
indicate the excellent correlation between the experimental
and the predicted values of the responses.
G. Validity of the Models
The performance of the developed model has been tested
using five experimental data which are not used in the modeling
process. The results for the response predicted by the
developed model in expression 5 have been used to evaluate the
theoretical values of the logarithmic damping decrement and
compared with the experimental ones. Time history plot as
recorded in storage oscilloscope for two sample experimental
readings has been presented in the Figures 9 and 10.
Furthermore, average percentage deviation between the
experimental and theoretical values of logarithmic damping
decrement has been calculated and presented in the Table 7. The
results indicate that the model predicting the values of
logarithmic damping decrement has good validity with
acceptable percentage deviation of 7.56 %. Moreover, the
Figure 11 has been plotted between the theoretical and
measured logarithmic damping decrement where the points
are very closer to form a straight line, which implies that the
data is normal and thus validates the model developed.
TABLE 7
COMPARISON OF THE THEORETICAL AND EXPERIMENTAL
LOGARITHMIC DAMPING DECREMENT

Fig. 9 Typical time history plot for amplitude (a) 0.2 mm, tack length (L)
35mm and surface roughness (Ra) 1.2 m
Fig. 10 Typical time history plot for amplitude (a) 0.1 mm, tack length (L)
50mm and surface roughness (Ra) 1.76 m

Fig. 11 Theoretical versus experimental logarithmic decrement.
4. Conclusion
This paper presents the findings of an experimental
investigation into the effect of tack length, initial amplitude
of excitation and the surface roughness on the logarithmic
damping decrement of layered and jointed tack welded
structures. In this study, BoxBenhken (BB) design of
experiments has been employed to develop a second-order
polynomial expression for predicting the values of
logarithmic damping decrement of layered and jointed tack
welded joints having various lengths of tack, surface
roughness and initial amplitude of excitation.
This has been done by performing statistically
designed experiments, estimating the coefficients in the
mathematical models, predicting the response, and thereby
Parameters Logarithmic damping decrement
L a Ra Theoretical Experimental Deviation (%)
25 0.5 1.52 0.000959 0.00104 8.51
35 0.2 1.24 0.002018 0.00221 9.53
50 0.1 1.76 0.002758 0.00297 7.68
45 0.3 1.47 0.001028 0.00108 5.31
35 0.4 1.98 0.000856 0.00091 6.81

International Journal of Advanced Computer Science, Vol. 2, No. 11, Pp. 441-448, Nov., 2012.
International Journal Publishers Group (IJPG)

448
checking the adequacy of the model. The RSM approach
provides a wide range of information on the
interrelationships of control and response variables with a
relatively small number of test runs.
The relationship of with interaction parameters has
been successfully obtained by using RSM at 95%
confidence level. Furthermore, the response regression and
variance analysis of the second order model for the response
shows that surface roughness parameter is statistically
insignificant and the logarithmic damping decrement is
constant for a jointed interface of same material irrespective
of the surface roughness. From the analysis of variance
(ANOVA), it is concluded that the logarithmic damping
decrement decreases with the increase in amplitude and tack
length. The logarithmic damping decrement remains almost
constant with the varying surface roughness at the interfaces
of the jointed and welded cantilever beams of same material.
The accuracy of the response surface model has been
verified with five sets of other experimental data and the
average percentage deviation has been found to be 7.56 %.
As the deviation is within the acceptable limit, the model
developed can be utilized effectively to evaluate the
logarithmic damping decrement for the layered and jointed
welded structures with other configurations.
References
[1] S.W.E. Earles & N. Mott, A Response Predication and
Optimization of a Frictionally Damped Structure, (1972)
Proceedings of 13th International (MTD&R) Conference,
Manchester University, pp. 43.
[2] S.W.E Earles & F.S. Mansoor, Frictional damping applied
to a Cantilever-beam Structure, A Theoretical and
Experimental Response Comparison, (1974) Int. J. Mach.
Tool. Des. Res., vol. 14, pp. 111-124.
[3] C.F. Beards, Some Effects of Interface Preparation on
Frictional Damping in Joints, (1975) Int. J. Mach. Tool.
Des. Res., vol. 1, no. 15, pp. 77-83.
[4] C.F. Beards & J.L. Williams, The Damping of Structural
Vibration by Rotational Slip in Joints, (1977) Journal of
Sound and Vibration, vol. 53, no. 3, pp. 333-340.
[5] C.F. Beards & I.M.A. Imam, The Damping of Plate
Vibration by Interfacial Slip between Layers, (1978)
International Journal of Machine Tool Design and Research,
vol. 18, pp. 131-137.
[6] C.F. Beards & A. Neroutspoulos, The Control of Structural
Vibration by Frictional Damping in Electro discharge
Machined Joints, (1979) ASME, no.79-DET-79.
[7] C.F. Beards & D.A. Robb, The Use of Frictional Damping
to Control the Vibration of the Plates in Structures, (1980)
Proceedings of International Conference on Recent
Advances in Structural Dynamics, Southampton, England,
pp. 749-760.
[8] C.F. Beards & A. Woowat, The control of frame vibration
by friction damping in joints, (1983) ASME, 83-DET-76.
[9] L.E. Goodman & J.H. Klumpp, Analysis of slip damping
with reference to turbine blade vibration, (1956) Journal of
Applied Mechanics, vol. 23, pp. 421-429.
[10] M. Masuko, I. Yoshimi & Y. Keizo, Theoretical
analysis for a damping ration of a jointed cantibeam, (1973)
Bull. JSME, vol. 16, no. 99, pp. 1421-1433.
[11] N. Motosh, Stress distribution in Joints of bolted or
riveted connections, (1975) Journal of Engineering for
Industry, ASME, vol. 97, Ser B (1), pp. 157-161.
[12] N. Nishiwaki, M. Masuko, Y. Ito & I. Okumura, A
study on damping capacity of a jointed cantilever beam (1st
report; experimental results), (1978) Bulletin of JSME, vol.
21, no. 153, pp. 524-531.
[13] B.K. Nanda & A.K. Behera, Study on damping in
layered and jointed structures with uniform pressure
distribution at the interfaces, (1999) Journal of Sound and
Vibration, vol. 226, no. 4, pp. 607624.
[14] E.J. Berger, Friction modeling for dynamic system
simulation, (2002) Applied Mechanics Reviews, vol. 55, no.
6, pp. 535577.
[15] M.A. Mayer & L. Gaul, Segment-to-segment contact
elements for modelling joint interfaces in finite element
analysis, (2007) Mechanical Systems and Signal
Processing, vol. 21, pp. 724734.
[16] Csaba, Forced response analysis in time and frequency
domains of a tuned bladed disk with friction dampers,
(1998) Journal of Sound and Vibration, vol. 214, pp.
395412.
[17] C.H. Menq, J. Bielak, & J.H. Griffin, The influence of
microslip on vibratory responsepart I: a new microslip
model, (1986) Journal of Sound and Vibration, vol. 107,
pp. 279293.
[18] Y. Song, C.J. Hartwigsen, D.M. McFarland, A.F.
Vakakis & L.A. Bergman, Simulation of dynamics of beam
structures with bolted joints using Iwan beam elements,
(2004) Journal of Sound and Vibration, vol. 273, pp.
249276.
[19] Bowden F.P. & Tabor D., The Friction and Lubrication
of Solids, Clarendon Press, Oxford, (1950), ISBN
978-0-19-850777-2.
[20] U. Olofsson, Cyclic micro-slip under unlubricated
conditions, (1995) Tribology International, vol. 28, no. 4,
pp. 207217.
[21] E. J. Berger, M. R. Begley & M. Mahajani, Structural
dynamic effects on interface response formulation and
simulation under partial slipping conditions, (2000) Journal
of Applied Mechanics, ASME, vol. 67, pp. 785-792.
[22] Y. Anno et al., (1970) Transactions of Japan Society of
Mechanical Engineers (in Japanese), vol. 36, no. 284, pp.
663.
[23] J.S. Hunter, Statistical design applied to product
design (1985) J. Quality Technology, vol. 17, no. 4, pp.
210-221.
[24] Montgomery D.C., Introduction to statistical quality
control, 2nd edition, Wiley, Canada.
[25] W. Sukthomya & J.D.T. Tannock, Taguchi
experimental design for manufacturing process optimization
using historic data and a neural network process model, Int.
J. Quality Reliability Management, vol. 22, no. 5, pp.
485502.
[26] B. Singh & B.K. Nanda, Effect of Welding on the Slip
Damping of Layered and Jointed Structures, (2010) Journal
of Engineering Mechanics, ASCE, vol. 136, no. 7, pp.
928932.