You are on page 1of 6

International Journal of Computational Engineering Science ^ ^ .

Vol. 5, No. 2 v(2004) 445-450 ® Imperial College Press


' ' yqgp www.icpress.co.uk
© Imperial College Press

DYNAMIC ANALYSIS OF STRUCTURE BY USING WILSON-6 METHOD

DE MEN HOU , YIBING WANG and MIN YIN


School of Civil Engineering and Mechanics, Xi'an Jiaotong University,
by UNIVERSITY OF ILLINOIS AT URBANA CHAMPAIGN on 11/26/15. For personal use only.

Xi 'an, 710049, P. R. China


*
dmhoe @ mailxjtu. edu. en
Int. J. Comp. Eng. Sci. 2004.05:445-450. Downloaded from www.worldscientific.com

XIAOYAN MA
Mathematics Division, Xi'an Electrical Power College,
Xi'an, 710032, P. R. China

In this paper, the Wilson-9 integration method is thoroughly studied. By using Wilson-9 integration
method and single particle shear model, the dynamic response of brick-concrete structure is given.
The simulation results of dynamic response of brick-concrete structure under sine wave load are
compared with analytical results. From this study, it demonstrated that the present approach of
combining Wilson-9 integration method and single particle shear model method is a useful
technique in the dynamic and vibration analysis of brick-concrete structure. Furthermore, by
optimizing and selecting suitable parameters of Wilson-9 method, this approach also can be used to
analyze the seismic dynamic response analysis of the brick-concrete structure.

Keywords: Wilson-9; Precision; Stability; Shear Model.

1 Introduction
There are many dynamical problems in engineering application that cannot be solved by
analytical method of mathematics, such as, building response under complex earthquake
wave loading[l]. In this case, numerical method, such as step integral, is needed. As a
consequence, the problem of precision and stability emerges. In this paper the precision
and stability of Wilson-9 method, which is used to study dynamic response of building
under a sine wave loading, was studied.

2 Wilson-9 Method

2.1 Hypotheses and Mathematical Model x


The dynamical shear model of single freedom is illustrated j^—>f m
in Fig l.[3]. m is the mass of the particle, k is the lateral
stiffness of rod, xg is the horizontal acceleration of
earthquake and x is the horizontal displacement of particle.
The dynamical equation of particle can be expressed in Eq. 1
(no damping case). -4"±
mX + kx = -mX (1) Figl Shear model

445
446 D. M. Hou et al.

where x and x are horizontal displacement and acceleration of particle relative to the
earth. According to the hypothesis of the linear linear acceleration method, a linear
variation of acceleration x from time t to time t + At is assumed. Let T denote the
increase in time, where 0 < T < At; then for the time interval t to t + At, it is assume
that
x^=xt+T(xt+At-xt)/At (2)
The Wilson 0 method is essentially an extension of the linear acceleration method, in
which the acceleration is assumed to be linear from time t to t + OAt ,where 0 > 1.0.
by UNIVERSITY OF ILLINOIS AT URBANA CHAMPAIGN on 11/26/15. For personal use only.

Let T denote the increase in time, where 0 < T < OAt; therefore, the acceleration of
particle can be expressed at the time t + T
Int. J. Comp. Eng. Sci. 2004.05:445-450. Downloaded from www.worldscientific.com

For unconditional stability we need to use 0 > 1 . 3 7 . Obviously it is the linear


acceleration method in case of 8=1, usually we employ 8=1.40[4].

2.2 Integration Equations


If the state of the particle is known at the time t, the state of particle can be described by
Eq. 4 at time t + At .
X = a
t+At 4\Xt+6&t ~Xt)+a5Xt +a X
6t
' X
t+At =X
t +a X
l( t+At +X
t) (4)
X
t+At = X, + XtAt + d% (Xt+&t + 2'Xt)

where xt+eAt can be calculated by Eq.l at time t + OAt [4]


At represents the increment of integral time for one step.
a0 = 6/(dAt)2; ax = 3/(0Af); a2 = lax; a3 = dAt/2; a4 = a0 /0;
(5)
a5=-a2jd; a6=l-3/0; a7 = Af/2; ag = Af2/6

2.3 Analytical Solution


If the acceleration of earth is a sine wave the dynamical equation of particle can be
expressed as in Eq. (6).
mx + kx = -mxg = A sin cot (6)
where, CO is the frequency and A is the amplitude of the sine wave. Furthermore, the
state of particle can be expressed in Eq. 7 at the moment t,
- Aco m . Ik
2 — sin J — t + -sin cot
k -mco k V m k - mco'
- Aco k Aco (7)
k -mco'
-cos —1 + k -mco' -cos cot
Aco ACQ2
sin -sin cot
k -mco2 k -mco'
Dynamic Analysis of Structure by Using Wilson-9 Method 447

3 Simulations
By using Wilson-9 method (Eq. 4) and also analytical method ( Eq. 7), the numerical
results of dynamic response of a single freedom system, are showed below. Moreover,
the comparisons of errors between the two methods are given. The parameters of the
model were selected as: m = 20kg , k = 1000N Im , xg= Bsinox , B = 0.25m/s 2 ,
CO = Inf , f = 0.2 Hz. A comparison between the beginning and the end of a
calculation with 9=2 and total integral time =15 seconds is showed in Fig 2. It could be
seen that the errors increased when the integral was lasting several seconds, at the
beginning of the simulation the two curves were very close (Fig 2a) and at the end of the
by UNIVERSITY OF ILLINOIS AT URBANA CHAMPAIGN on 11/26/15. For personal use only.

simulation the two curves were separate (Fig 2b).


Int. J. Comp. Eng. Sci. 2004.05:445-450. Downloaded from www.worldscientific.com

Acceleration(m/s) O.O6-1
Aeceleration(rrVs2)
Wilson
Wilson 004-
Analytical// o
Analytial
002-

/
/
/ 0.00-

002-
\ 1
004-
\\ ;7

006- Time(s)
1
1
n'- ' n', ' n'o ' n'o ' .'n ' .'- ' .'. ' .'o ' ' l ' l 1 • 1
00 02 04 06 08 10 12 14 16 1. ' o '
13 0 135 140 14 5

(a) Response of acceleration at the beginning (b) Response of acceleration at the end
Fig 2 Acceleration of response with 8=2.0

3.1 Error due to 0 Value


In order to examine the error caused by the choice of 9, calculations were carried out
when 9 was changed from 0.5 to 2.0. All the calculations had the total integral time
/ = 15 seconds and an integral increment At = 0.01 second. In the discussion, the
following subscriptions are used: 'w' for Wilson-9 method, 'p' for analytical method, 'i'
for the number of integral step and 'n' for the total number of integral steps. \xw • x„

is the absolute error of acceleration at i step and J J £ ( x w - ^ | / « is the Mean

Square Root (MSR) error of acceleration and MAX \x - x is the absolute


MAXimum (MAX) error of acceleration. The statistical error of simulative calculation is
illustrated in figure 3. Furthermore, \(xw — x )/x X l 0 0 % represents the relative

error of acceleration at i step and - j ^ \(xw — xp)/x\ >/n represents the average

relative error of acceleration. Zero value exists in analytical solution so that the error is
448 D. M. Hon et al.

added up if ABS (X ) > MAX (x ) x 5 % • The result of average relative error is


showed in figure 4. Because the errors of displacement and velocity have the same figure
with acceleration, so only the results of acceleration are illustrated in this paper(see
Fig4). As shown in figure 3 and figure 4, the error is minimal when 9 is about 0.79. In
order to show it clearly, the error of acceleration is illustrated and the figure is magnified
by 6 value is from 0.75 to 0.85, in figure 5.

0.01 4

0.013
Acceleration(m/s
by UNIVERSITY OF ILLINOIS AT URBANA CHAMPAIGN on 11/26/15. For personal use only.

0.012

0.01 1

0.010 -MSR
Int. J. Comp. Eng. Sci. 2004.05:445-450. Downloaded from www.worldscientific.com

0.009 MAX
0.008

0.007

0.006

0.005

0.004

0.003 V
0.002

0.001

0.000 , i^^, r ,iprr, • , . , . , • , • ,


0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2

Fig 3 The absolute error of acceleration between Wilson 0 method and analytical

26-]percentag e

- • — D isplacement
"-• Velocity
A cceleration

0.4
-*=¥=
0 .6
.p-y-y-p-y-
0.8 1.0 1.4 1 .6 1.8 2.0

Fig 4 The average relative error of response between Wilson method and analytical

0 00026-
Acceleration(m/s)
0 00024. 05-,
percentage
0 00022-

0 00020-
«' 04- /
000018-
—•—MSR / /
—®~~ MAX
000014- / 03-

/ /
000010- \ 8 / 02-
/
0 00008- \
\ -' ^
0 00004-
"^ • ^ L X * ^ 01-

0 00002-
- ® " /
^
0 78 0 78 00-
0 74 0 76 0 78 0 80 0 82 0 84 0 88

(a) Absolute error of acceleration (b) Average relative error of acceleration


Fig 5 The error of acceleration is magnified by Q value is from 0.75 to 0.85
Dynamic Analysis of Structure by Using Wilson-9 Method 449

3.2 Errors about Increment


In order to examine the influence of increment, calculation was made at 6 = 1.4 with
At changed from 0.001 to 0.05. the error of acceleration is considered below. The
statistical absolute error of simulation is illustrated in figure 6 and the average relative
error is illustrated in figure 7.

'_ Error(m/s) Error(m/s ]

-MSR
by UNIVERSITY OF ILLINOIS AT URBANA CHAMPAIGN on 11/26/15. For personal use only.

MAX
Int. J. Comp. Eng. Sci. 2004.05:445-450. Downloaded from www.worldscientific.com

Increment Increment

(a) Absolute error of acceleration (b) Increment value is from 0.001 to 0.01 second

Fig 6 Absolute error of acceleration and is magnified by increment value is from 0.001 to 0.01 second

Percentage
Percentage

Increment Increment

(a) Relative error of acceleration (b) Increment value is from 0.001 to 0.01 second

Fig 7 Average relative error of acceleration and is magnified by increment value is from 0.001 to 0.01 second

4 Stability of Wilson-0
In most cases, an earthquake persists about 60 seconds, therefore the stability of Wilson-
9 method is only needed to be checked in this period for this class of problems. From the
simulation results, the optimized selection of parameter of Wilson-9 method is
6 = 0.79 and At < 0.004 . The result of simulation is expressed in table 1 and Fig 8
in the case of 9 = 0.79; At = 0.004 and total integral time = 60 seconds. The errors
between Wilson-9 method and analytical method are very small even if the integral time
reaches to 60 seconds. The maximum average relative error of acceleration is 0.0120%,
the errors of velocity and displacement are more smaller (Table 1). The two curves of
response of acceleration are almost same (Fig 8).
450 D. M. Hou et al.

Table 1 . 0 = 0.79; At = 0.004; Integral time=60

MAX value MSR error MAX error Average relative error


Displacement 6.078E-3 3.780E-8 9.927E-8 0.0013%
Velocity 1.296E-2 2.611E-7 6.528E-7 0.0055%
Acceleration 5.400E-2 3.632E-6 7.135E-6 0.0120%
by UNIVERSITY OF ILLINOIS AT URBANA CHAMPAIGN on 11/26/15. For personal use only.

0.04-
Acceleration(m/s)
Int. J. Comp. Eng. Sci. 2004.05:445-450. Downloaded from www.worldscientific.com

0.02-
\ - - - -Analytic

0.00-

0.02-

0.04-

0.06.
Time(s)
58 .0 58.5 59.0 59.5 60.0

Fig 8 Response of acceleration with 9=0.79 and Increment=0.004 second, integral time reaches 60 seconds

5 Conclusions
A numerical analysis of Wilson-9 method is unconditional stability to use 0 > 1.37, but
in some cases it is not the optimized choice. In this paper it is obvious that the choice of
9 and increment is very important to reduce error. For example, if 9=0.79 and
increment=0.004 are chosen and the simulation reaches 60 seconds, the average relative
error of acceleration is only 0.012%, whereas the total integral step is 15000 steps. It is
possible that error will be small even if the integral time is longer.
(1) Wilson-9 method is adopted that 9 is better to be chosen as 0.79 and the result of
simulation is more accurate when the integral time is not more then 60 seconds.
(2) The integral increment of Wilson-9 method is better to less than 0.004 second.
(3) For seismic analysis, it is stable when 9=0.79 and At < 0.004 by using Wilson-9
method.

References
1. N.Mostaghel and M.Khodaverdian, Seismic Response of Structures Reported on
R-FBI System, Earthquake Eng. Struct,Dyn., 16,3 (1983).
2. L.P.Xing, Effective location of active control devices for building vibrations caused
by periodic excitation acting on intermediate storey, Earthquake Eng. Struct, Dyn.,
2,29 (2000).
3. M.H.Yu,Y.B.Wang,etc. Seismic analysis of the fundamental isolation of brick
masonry building, Learned Journal of Construction, 4,17 (1996).
4. KJ.Bathe, Finite element procedure, Prentice-Hall, Englewood Cliffs,NJ. p777-778.

You might also like