See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/337290470

Response Surface Methodology based Multi- Objective Optimization of Stock

Bridge Damper for Pump Induced Floor Vibration

Conference Paper · January 2020

DOI: 10.1109/ICCAKM46823.2020.9051456

CITATION READS

1 104

3 authors:

Mohammad Sabbir Rahman Abul Hasnat

The University of Asia Pacific American International University-Bangladesh
26 PUBLICATIONS   569 CITATIONS    16 PUBLICATIONS   26 CITATIONS   

SEE PROFILE SEE PROFILE

Musawer Ahmad Saqif

University of Michigan
2 PUBLICATIONS   1 CITATION   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Tannery Industries in Bangladesh View project

National Research Foundation , South Korea View project

All content following this page was uploaded by Abul Hasnat on 15 November 2019.

The user has requested enhancement of the downloaded file.

 Response Surface Methodology based Multi-
Objective Optimization of Stock Bridge Damper for
Pump Induced Floor Vibration
Mohammad Sabbir Rahman Abul Hasnat Musawer Ahmad Saqif
Lecturer, Deptt. of CE Asst. Prof, CE, Faculty of Engineering Lecturer, Deptt. of CE
University of Asia Pacific AIUB University of Asia Pacific
Dhaka, Bangladesh Dhaka, Bangladesh Dhaka, Bangladesh
[email protected] [email protected] [email protected]

Abstract—This paper presents a review on getting a simultaneously. In most real engineering problem, all
Weighted Multi-Objective Optimization (WMO) of objectives may have the respective relative importance. This
Stockbridge Damper parameters based on Response Surface paper has concerns about the multi-objective optimization
Methodology (RSM) coupled central composite design and with relatively weighted objectives of Stock-bridge damper
Weighted Desirability Function (WDF) to attenuate the pump
based on RSM coupled central composite design (CCD) and
vibration of a steel floor. To optimize the parameters of the
damper, the frequency ratio and mass ratio were considered as desirability function considering the relative importance of
a design variable and the top displacement and frequency each objective. The multi-objective optimization of stock
response were considered as objective functions. The bridge damper system would be difficult due to the different
optimization has been carried out under pump which can functions having different units and different orders of
generate different frequency with different RPM (0-3000). The magnitude. Thus, it needs to make these objective functions
optimal Stockbridge Damper based on RSM shows better be on the same scale by applying desirability function. In
response among the different cases. It is concluded that the addition to this, each of different objective functions must
proposed response model offers an efficient approach have different importance in designing Stock bridge damper
regarding Stockbridge Damper optimization.
parameters depending on applications, requirements and so
Keywords: weighted multi-objective optimization; on. The multi-objective optimization problem is structurally
Stockbridge damper; response surface methodology; weighted similar to multi-criteria decision making (MCDM) process
desirability function; floor vibration. dealing with the problem of choosing an option from a set
of alternatives, which are characterized in terms of their
I. INTRODUCTION attributes. One of the most outstanding multi-criteria
Floor vibration of a structure can be happened in many decisions making (MCDM) approaches is the analytical
ways like as machine vibration or for any other mechanical hierarchy process (AHP) developed by L. Saaty (1990),
instruments, even people’s walking on floor may cause which has its roots in obtaining the relative weights among
floor vibration. Floor vibration often causes crack on the the factors and the total values of each alternative based on
floor. Long span column designed for architectural purpose these weights. This study constructed the relative weight of
is another major cause of such type of crack. Several design two objective functions by the AHP.
guide lines and standards are crated to evaluate such types
of susceptibility of floors in terms of general step The proposed optimization method is also expected to
frequencies. The acceleration response of the floor is find more accurate optimum properties of Stockbridge
usually compared with an acceptable limit to assess the damper. To find an optimal messenger cable length and tip
vibration and the design of the structure. This paper presents mass, the mass ratio and frequency ratio of the damper has
a work to introduce a stock bridge damper design and been considered as design variables while the response of
optimization of damper parameters to control the machine the structure has been taken as the objective function. The
generate floor vibration. To optimize the damper parameters optimization of Stockbridge damper is configured with
Response surfaced methodology based multi-objective pump vibration on the floor system.
optimization has been performed.
II. STOCKBRIDGE DAMPER
The response surface methodology (RSM) is one of the Stockbridge damper is a tuned mass damper which has
most widely-used statistical approaches that is useful for been invented in the 1925s by George H. Stockbridge
analyzing the data and optimizing the processes. L. Soto- (Stockbridge, 1925) used to suppress wind-induced
Pérez (2015) applied this to the optimization of cement vibrations on slender structures such as overhead power
paste mix design. A. Khan et al. (2016) investigated multi- lines (Markiewicz, 1995) and long cantilevered signs
objective optimization to optimize the cost-effective mix (Garlich and Thorkildsen, 2005) The dumbbell-shaped
proportioning of high strength self-compacting concrete by device consists of two masses at the ends of a short length
approximating material model and cost model by means of of cable or flexible rod, which is clamped at its middle to
RSM. Like other engineering problem, Stockbridge damper the main cable. The damper is designed to dissipate the
design is a multiobjective optimization problem. Thus, energy of oscillations in the main cable to an acceptable
Stockbridge design has to optimize multiple objectives level (Kiessling et al., 2014). Its distinctive shape gives it
the nickname “dog-bone damper” (Garlich and Thorkildsen, response, . The coefficients indicate the
2005). Fig. 1 illustrates the patent closeup of Stockbridge behavior of different factors on the response where various
damper. Stockbridge damper has been considered as models can be described. The interaction model:
continuous hysteresis system damper for vibration reduction
(Sauter and Hagedorn, 2002; Verma, 2002).
where = predicted response; a1, and a2 = coded
factors; b = coefficients; k = offset term.
For a design with three or more levels, a curvature is
observed. The response surface plots were generated using
Minitab to enable modeling of the response over an entire
range of varying factors with limited number experimental
runs. Analysis of Variance (regression ANOVA) was
employed to determine the model fitness and significance of
Fig.1. Stockbridge patent closeup (Stockbridge, 1925) various factors to minimize the floor acceleration and
displacement considering the sum of squares and residual
III. EXPERIMENTAL DESIGN sum of squares. ANOVA was in addition performed by
The CCD of two factors involving mass ratio (a1) and differentiating the variances of linear regressions. All
frequency ratio (a2) is shown in Fig. 2. Where four corners statistical analysis was performed using Minitab.
of the square represent the factorial (+/- 1) design points
Four-star points represent the axial (+/- ) design points and V. MODEL QUALITY
0 represent replicated center point. The design consists of The regression coefficients, and adjusted , were
the factorial design with centre points and the star/axial calculated to determine the variation between the model and
points used to estimate the curvature. This allows the the experimental data (Tekindal et.al,2012). According to
quadratic terms to interfere in the model for estimation of Weinberg and Abramowitz (1973), often overestimates
the optimum combination of factors. The total number of the model fit with excessive predictors therefore adjusted
experiments is calculated based on n = 2f + 2f +n where f is is recommended. The adjusted takes all independent
the number of factors and n is the replication of center point predictors into consideration and fixes the more
[24]. A value is defined as the distance between axial points accurately according to the number of predictors present in
and center point, α=(2f)1/4. As inscribed CCD is a scaled the model. usually increases with increasing predictors;
down of circumscribed CCD, the factorial design is further however, adjusted will only increase when that
divided by a, demonstrating a 1/α coded level in factorial
particular predictor has positive effects on the response.
design.
Thus, the latter is particularly useful in comparing models
with different numbers of predictors (Frost J,2013). Both R2
and adjusted indicate the quality of the proposed models;
values closer to 1.0 demonstrate a good agreement between
the experimental and predicted responses (Demirel M,
Kayan B, 2012)

Where, n=number of experiments and p=predictor

Fig.2. Central Composite Design with 2 factors
VI. NUMERICAL MODEL AND ANALYSIS
IV. RESPONSE SURFACE METHODOLOGY (RSM) One steel 10 mm thickness slab with 3000 mm length
and 3000 mm width (Fig.3) has been considering to perform
The design matrix involving interaction terms are
the structural analysis. One pump has been installed in the
employed to develop the equations for estimation of the
middle of the slab to vibrate the slab. To get the modal
adsorption performance. The variable coefficients were
frequency of slab modal analysis is performed and for the
derived based on multiple linear regression: ,
pump vibration results (floor acceleration and displacement)
where the predicted response is expressed as the product of structural analysis is performed by SAP200. After analysis,
the design matrix, D, and the corresponding coefficients, b the first modal frequency of the slab is 2.5 Hz. Steady-
[25]. The vector of the corresponding coefficients is the State Vibration forced is applied as a pump vibration
pseudo inverse of D and the experimental considered the rotational speed from 0 to 3000 RPM which
covered the frequency range from 0 to 50 Hz which can also Stockbridge damper, Response surface methodology based
generate the resonance frequency with some different mode. Center Composite Design(CCD) approached is perform. To
perform the CCD analysis point has been created by
Minitab software considered the range of mass ratio (a1) is
1% to 3% of structural mass and frequency ratio(a2) is .8 to
1. 13 of first modal frequency of slab which includes lower
axial, higher axial and the center point which is shown in
Table 1, where 0 indicate center point (n=5) and 1 and -1
indicate the axial points (2f=4) individually. Mass ratio and
frequency ratio is considered as analysis point. And
acceleration and displacement amplitude are considered as a
response for each analysis point to develop the statistical
model. The minimum acceleration is shown under the
combination of mass ratio is 1% and frequency ratio is 1. In
case of displacement that is 2% mass ratio and .75
frequency ratio. The structural analysis result also shown in
Fig 4(a) and 4(b).

Fig. 3. Numerical models of a slab

To design the Stockbridge damper first modal frequency
of slab is taken. To find out the optimum design of

TABLE I. ANALYSIS POINT AND CORRESPONDING COST FOR MINITAB

Analysis Point Stockbridge damper Response

Run Pt- Mass Tip mass Length Acceleration Displacement
Order Type Ratio(a1) Freq. Ratio(a2) (kg) (mm) (mm/s2) (mm)
1 -1 3.41421 0.90000 0.503 329.57 1.24591 0.00296
2 1 3.00000 0.80000 0.503 299.01 1.35113 0.00341
3 1 1.00000 0.80000 0.503 329.57 1.35092 0.00433
4 0 2.00000 0.90000 0.805 304.80 1.20044 0.00325
5 0 2.00000 0.90000 0.201 416.95 1.34702 0.00359
6 1 1.00000 1.00000 0.503 329.57 1.04526 0.00321
7 0 2.00000 0.90000 0.201 483.83 1.23335 0.00310
8 1 3.00000 1.00000 0.503 369.35 1.15832 0.00312
9 -1 0.58579 0.90000 0.805 262.66 1.20615 0.00353
10 0 2.00000 0.90000 0.503 329.57 1.29062 0.00344
11 -1 2.00000 1.04142 0.076 618.26 1.12554 0.00313
12 -1 2.00000 0.75858 0.503 329.57 1.15215 0.00294
13 0 2.00000 0.90000 0.930 268.54 1.23335 0.00310

Fig. 4(a). Frequency vs floor acceleration under different Stockbridge damper

 Fig. 4(b). Frequency vs floor displacement under different Stockbridge damper

Table 2 shows the regression equation, R2 and R2adj for both of them are one and the first priority is to minimize
a full quadratic model of concrete and reinforcement cost better performance of Stockbridge damper.
response variables. The R2 values of the response surface
models for acceleration and displacement cost are 89.94% TABLE III. BEST GLOBAL SOLUTION FOR OBTAINING THE MINIMUM
FREQUENCY AND MASS RATIO
and 87.91% for the full quadratic equation. The R2adj
values of the response surface models for acceleration and Items Goal Lower Target Upper Weight Importance
displacement are 89.90% and 87.84% for the full quadratic
equation. It is seen that R2 and R2adj shows a high Acceleration Minimize 1 Min 1.5 1 1
Displacement Minimize 0.001 Min 0.0035 1 1
correlation coefficient for the responses i.e. model shows a
good correlation between the measured values. Therefore, it
is evident that the established model is adequate and thus
can be applied for the predicted models. For the quadratic
model equation RSM, there are two factors (mass ratio,
frequency ratio) in the first equation frequency ratio is a
greater effect than the mass ratio on acceleration. In the
second equation is mass ratio has a greater effect on
displacement than frequency ratio. For both equation R2and
values are near to 90% which means the statistical
model quality is very good.

TABLE II. QUADRATIC MODELS FOR ACCELERATION AND DISPLACEMENT

AMPLITUDE

Items Model Equation R2

a) Surface plot of Displacement
89.94 89.90%
%
87.91 87.84%
%

Table III shows, the global solution to optimize the

frequency and mass ratio of stockbridge damper. This
research goal is to minimize the acceleration and
displacement amplitude of vibration floor. To find the
optimum tip mass (mass ratio) and corresponding
messenger cable length( frequency ratio), a lower and
higher value for the mass ratio is 1% to 3% and frequency
ratio is 0.8 to 1. For the best global solution of
b) Contour plot of Displacement
acceleration(mm/s2) and displacement(mm) importance
 Fig. 5 Surface and Contour plot of floor displacement

Fig. 7. Overlaid contour plot of multiple responses

a) Surface plot of Accleration Fig 8 shows the Optimization plot of acceleration and
displacement is minimum with very good desirability value.
For the acceleration desirability 1, displacement desirability
is 0.96 and composite desirability is 0.976, this value
indicates that desirability is quite good for
optimization.Also show the optimum mass ratio and
frequency ratio which is 1.5286% and 1.0414
correspondingly.

b) Contour plot of Acceleration

Fig. 6 Surface and Contour plot of floor acceleration

Fig 5 is Surface and Contour plot of displacement where

frequency ratio and mass ratio is increased than
displacement is minimum. Another Fig 6 is Surface and
Contour plot of floor acceleration where frequency ratio is
increase and mass ratio of the damper is decrease than
acceleration decrease.
To get the complete picture of the feasible region for
multiple responses, overlaid contours are plotted which
shown on Fig 7. Fig 7 is an Interaction Diagram of
Fig. 8. Optimization plot and desirability
frequency and mass ratio which represent the optimum
region and variation of displacement and acceleration At Table IV, the optimum mass ratio and frequency ratio
amplitude. The gray portion means the optimum are displayed and also shows the predicted and analysis
displacement and acceleration region for mass and responses with considering the optimum design of
frequency ratio. In this diagram for mass ratio vary 0-3.5% Stockbridge damper. Analysis responses are very close to
and 0.75 to 1.5. Final optimum mass ratio is 1.5286 % and predicted response and the percentage of error is almost
frequency ratio is 1.0414. negligible.

TABLE IV. OPTIMIZED VARIABLES AND PREDICTED COST

Mass From Error
Frequency Predicted
Ratio(%) Analysis %
Ratio
0.0030 0.0032 6.6%
Displacement 1.5286
1.0414
Acceleration 1.0358 1.0384 0.25%
 acceleration reduction rate is 32.95 %. The optimum
Stockbridge damper-controlled value is satisfied to mitigate
the slab vibration responses.
In conclusion, the optimization of Stockbridge damper
by the multi-objective optimization based on RSM has been
found as rational and beneficial.
REFERENCES
[1] Soto-Pérez, L., López, V. and Hwang, S. S. (2015), “Response
Surface Methodology to optimize the cement paste mix design: Time-
dependent contribution of fly ash and nano-iron oxide as
admixtures”, Materials and Design, 86: 22-29.
[2] Khan, A., Do, J. and Kim, D. (2016), “Cost effective optimal mix
proportioning of high strength self-compacting concrete using
response surface methodology”, Computers and Concrete, 17(5),
629-638.
[3] Saaty, T. L. (1990). Decision Making for Leaders: The Analytic
Hierarchy Process for Decisions in a Complex World. RWS
publications.
[4] Stockbridge, G., H. (1925), “Vibration Damper,” Patent No.
Fig. 9(a) . Comparison of uncontrolled and controlled acceleration (mm/s2) 1675391, USA Patent Office.
[5] Aslan, N., and Cebeci, Y. (2007), “Application of Box–Behnken
design and response surface methodology for modeling of some
Turkish coals”, Fuel, 86(1): 90-97.
[6] Barry, O. R. (2014), “Vibration Modeling and Analysis of a Single
Conductor with Stockbridge Dampers”, Ph.D. Dissertation,
University of Toronto, Canada.
[7] Barbieri, N. and Barbieri, R. (2012), “Dynamic analysis of
stockbridge damper”, Advances in Acoustics and Vib., 2012, Article
ID: 659398.
[8] Bayramov, F., Taşdemir, C., and Taşdemir, M. A. (2004).
“Optimisation of steel fibre reinforced concretes by means of
statistical response surface method.” Cement and Concrete
Composites, 26(6): 665-675.
[9] Bezerra, M. A., Santelli, R. E., Oliveira, E. P., Villar, L. S., and
Escaleira, L. A. (2008), “Response surface methodology (RSM) as a
tool for optimization in analytical chemistry”, Talanta, 76(5): 965-
977.
[10] Chang, S., Sun, W., Cho, S. G. and Kim, D. (2016), “Vibration
Control of Nuclear Power Plant Piping System Using Stockbridge
Damper under Earthquakes”, Sci. Technol. Nuclear Install., 2016,
Article ID. 5014093.
[11] Ferreira, S. C., Bruns, R. E., Ferreira, H. S., Matos, G. D., David, J.
M., Brandao, G. C., and Dos Santos, W. N. L. (2007), “Box-
Behnken design: an alternative for the optimization of analytical
methods”, Analytica chimica acta, 597(2): 179-186.
[12] Kasap, H. (2012), “Investigation of Stockbridge dampers for
Fig. 9(b). Comparison of uncontrolled and controlled displacement (mm) vibration control of overhead transmission lines”, Ph.D. Dissertation,
Middle East Technical University.
Figure 9 represents the uncontrolled and controlled [13] Khan, A., Do, J. and Kim, D. (2016), “Cost effective optimal mix
responses; optimized Stockbridge damper has been installed proportioning of high strength self-compacting concrete using
to get the controlled response whereas no damper is response surface methodology”, Computers and Concrete, 17(5),
629-638.
installed in uncontrolled structural. Uncontrolled and [14] Khuri, A. I. (1996). Multiresponse Surface Methodology. Handbook
controlled acceleration amplitude is 1.54 mm/s2 and 1.0384 of Statistics, 13:377-406.
mm/s2, the reduction percentage is 32.95 % . in case of [15] Leung, A. Y. T., and Zhang, H. (2009), “Particle swarm optimization
displacement reduction is 48.30% where uncontrolled and of tuned mass dampers”, Engineering Structures, 31(3): 715-728.
[16] Myers, R. H., Montgomery, D. C., & Anderson-Cook, C. M. (2016).
controlled displacement is 0.0062 mm and 0.0032 mm Response Surface Methodology: process and product optimization
respectively. using designed experiments. John Wiley and Sons.
[17] Navarro Canales, C., Lara López, A., Colín Venegas, J., Razo-García,
J., and AguileraCortés, L. A. (2008), “Optimal Design of Stockbridge
VII. CONCLUSION
Dampers”, Ingeniería mecánica, tecnología y desarrollo, 2(6): 193-
199.
The optimization scheme that utilizes the response [18] Rajmohan, T. and Palanikumar, K. (2013). “Application of the
surface methodology based on the weighted multi-objective central composite design in optimization of machining parameters in
optimization to find an optimum Stockbridge damper that drilling hybrid metal matrix composites”, Measurement, 46(4),1470-
1481.
was applied to a steel slab and to trade off the conflicting [19] Sepehri, A., Daneshmand, F. and Jafarpur. K. (2012), “A modified
between two objective functions depending on the design particle swarm approach for multi-objective optimization of
variables of frequency ratio and mass ratio. laminated composite structures”, Struct. Eng. Mech., 42(3): 335-352.
The reduction rate of displacement is 48.30%, whereas the
[20] Soto-Pérez, L., López, V. and Hwang, S. S. (2015), “Response
Surface Methodology to optimize the cement paste mix design: Time-
dependent contribution of fly ash and nano-iron oxide as
admixtures”, Materials and Design, 86: 22-29.
[21] Tang, H., Zhang, W., Xie, L. and Xue, S. (2013), “Multi-stage
approach for structural damage identification using particle swarm
optimization”, Smart Structures and Systems, 11(1): 69-86.
[22] Tekindal, M. A., Bayrak, H., Ozkaya, B. and Genç, Y. (2012), “Box-
Behnken experimental design in factorial experiments: the
importance of bread for nutrition and health”, Turkish Journal of
Field Crops, 17(2): 115-123.
[23] aeolian conductor vibration with a stockbridge-type damper”, J.
Fluids Struct., 14(4): 489-509.
[24] Wagner, H., Ramamurti, V., Sastry, R. V. R. and Hartmann, K.
(1973), “Dynamics of Stockbridge dampers”, J. Sound Vib., 30(2):
207-220.
[25] Frost J (2013) Multiple regression analysis: Use adjusted
R-squared and predicted R-squared to include the correct number
of variables. In: Minitab Blog.
[26] Demirel M, Kayan B (2012) Application of response surface
methodology and central composite design for the optimization of
textile dye degradation by wet air oxidation. Int J Ind Chem 3:1–10

View publication stats