Cairo University

Faculty of Engineering
Structural Engineering Department

M.Sc. & Ph.D. Courses

2014-2015
2 2

STR 601
Post-Graduate Course
on
Structural Dynamics

Prepared by
Dr. Sameh Samir F. Mehanny,
Professor of Structural Analysis and Mechanics
 PART I

Single Degree of Freedom Systems

 f

It is the angle by which

u(t) lags behind the
function sin (w t) which
means comes after it
or following it
Unstable Stable
r is mass per unit area
rotational

angular acceleration
mass "polar" moment
of inertia; polar since
the rotation is @ an
axis perpendicular to
the disc. N.B.: it is
calculated as the sum
of mass moment of
inertia @ x and y
mass "polar" moment of inertia
(i.e., mass moment of inertia @ x
+ mass moment of inertia @ y)
D = D1 + D2
F/Ke = F/K1 + F/K2

F = F1 + F2
Ke D = K1 D + K2 D
Refer to Chopra
textbook
Fig. 4.7.2
Extracted from Dynamics of Structures by Anil Chopra
Total or Complete
ω [t−τ+(−∆τ )]
C1 C2
since pulse initiates at 0 then τ=0 then

is a constant
since we
integrate with
respect to τ
and not t
Alternative Proof for
the unit impulse response function, h(t)
umax or

umax or
 u
ug
U
Relative
Displacement
To determine the True Total ARS:
EITHER by differentiating u_dot(t,T) then add
to the result the ground acceleration a(t) then
get the max. value of the whole quantity, OR
through the Equ. of motion just above, i.e.,,
max. of w2*u(t)
Then Acceleration RS tends
to zero (asymptotic to hle
axis) as T gets very large
We claim
that
from top of page 9

referring
to end of
page 9
as a ratio of the SE
at the beginning of
that same cycle
. Inertia force is always opposing acceleration direction/sign ; spring in tension
refer to Lecture 2, page 3 means
acceleration has
same sign
Note, accelaration is
the rate of change in
velocity

Both cases below are derived for +ve u

meaning the spring is in tension
mass is moving for a half-cycle from left to right with increasing then
decreasing velocity, but with same velocity sign until it reaches zero value

mass reverses motion, velocity changes sign, and the mass moves from right to
left with increasing then decreasing velocity until a zero value at end of the full
cycle S
.

S
 u(t) R to L

u(t) L to R

From Left
to Right

From Right
to Left
 The natural period for the SDOF with Coulomb damping is the same as that
forthe SDOF with no damping; this is different from the case with Viscous
damping.

 The direction of the friction force opposes motion,

gn of the friction force will change when
and the size
the direction of motion changes.

 This necessitates formulation and solution of 2

differential equations: one valid for motion in one
direction and the other is valid when the motion is
reversed. (i.e., one equation for each half cycle of
vibratory motion)

 Each of the 2 differential equations is linear BUT

the overall problem is nonlinear because the
governing equation changes every half-cycle of
motion.

The envelopes to the displacement-time curves for Coulomb

damping are STRAIGHT lines and not
EXPONENTIAL as in Viscous damping

In each COMPLETE cycle of motion, the amplitude of motion is reduced by 4uF,

where uF = mmg/k.
In each HALF CYCLE of motion (i.e., each T/2 = p /w ) , a new equ. of motion is
written with new values for the particular sol. and the constants of the general sol.

Motion STOPS at the end of the half-cycle for which the amplitude is less than uF;
at that point the spring force acting on the mass is less than the Friction force
and hence the motion STOPS since the velocity at this instant is ZERO. Note that the
velocity is ALWAYS ZERO at end of EACH half cycle.
However, whenever the amplitude is less than uF but within EACH given half cycle (and
not at its end) the motion DOES NOT STOP since at this instant the velocity is NOT ZERO.
Response to harmonic force (sine wave form), ν/ω = 0.2, ζ = 0.05,
I.Cs.: (0) = 0 & (0) = ω.q/k
comparing u(t)=ust rd sin(nt-f) with f(t)=q.sin(nt)
 t

ν/ω = 1
u(t) = ust rd sin(νt – Φ)
f(t) = q sin (νt)
c = ζ1 (2mω), then ζ2 = c / (2mω2) = ζ1 (ω/ω2) = 0.05*(1/√2) = 0.035

1.469
0.035 1.469
 Or Better
dealing with
the exact relationship
and consider n = 2 p 1.487
 /g

if verbally said at resonance

BUT
in the sense that
]
This is how the TR function looks like if the
amplitude of the forcing function, q, is constant
and not a function of the forcing frequency, n
 DAMPING and

not very soft spring for

MUCH in order not
C remains CONST. while Ccr changes
We always
record/measure ur
while we need to
determine us
( - f)

( - f)

( - f)

( - f)

( - f)
This how the TR function looks like in case the
forcing function amplitude, q, is not constant but is
a function of the forcing frequency, n. This is
different from the shape of the TR function for the
constant amplitude forcing function as in page 2 of
this lecture notes.

N.B.: here q = m . us . n^2

For a Seismometer measuring Ground Disp., we need to REDUCE
STIFFNESS and INCREASE MASS

For an ACCELEROMETER measuring Ground Accel., we need to

INCREASE STIFFNESS and DECREASE MASS
In the M-I Eq. it does not matter
the velocity changes from Zero to 5
OR from 2 to 7m/s (ONLY interested in
the amount of CHANGE,
i.e., the Difference)
 -k \J~b'{~hlO"' s. o~'tio 11 :

-!~ (t-'t.")
u.(1:) =e [ c, w-. ~ (t-L) rcz ~wd. Lt-T) J
- I. (, :> : At t =T. (_-9:::: I: ='T 4- .t. t ~ ,,, T ~ o)
:::::> ol\,sf. = 0 ~ cl -=- o
~ ~- == f.lr)J.T ~ -Tv0*~ex&+ciwJ.
c2 =+Cr)Jt
Wl VV)

: . c 2. = i tr) ol r
h->wol
- ~J Jrtt u-\,lor~h.QV\ sotw\-l.'OV\ }'-r a. s.'1yk puJhe
oc..cwc~v; C\f Q \i~ r: te -z'°-rb'l:: ~ ~r~o
-r L\.) ( t - -r>
d Ll (J;) = e. Hr) Ar ~ ~ Ct _-r)
h'> l.()J.
• STEP~~

- A cc."1.>-Mh'J jar ~ f'..&.~ ~ -Ht)

~ [ <A-(1:)= ~ wo\ ~-~~~

j e k. w~ Ct-7.:) Hr:) Jr
J
0

General
 τ

using Duhamel
Integral
τ= t

In this integration we deal with t as if const.

τ= since we tintegrate w.r.t. t t

0
0
EITHER through differentiating u_dot(t;T,zeta), then adding to it the a_g(t),
then get the max. of the combined quantity, OR through the Eq. of motion, i.e.,
the max. value of -omega2*u(t)-2*zeta*w*u_dot(t)
Viscous Damping is PRIMARILY for Elastic Response but is
still present in INELASTIC response; Hysteretic Damping is
ONLY for Plastic (i.e., Inelastic) Response
Viscous damping dissipates LESS energy from the inelastic
system compared to that dissipated by Yielding (i.e., through Hysteretic Damp. for Steel structures
Hysteretic damping). This implies SMALLER Velocities is larger than that for RC structures
in the INELASTIC system relative to the ELASTIC while
system Viscous Damp. for Steel structures
is smaller than that for RC structures
with same Kinitial as that of the
elasto-plastic system but with
ADDITIONAL STRENGTH that
warrants ELASTIC behavior
under m.Sa or k.D calculated
from Response Spectra
implies it will always remain ELASTIC
 Ry m
2 2.5
Increasing Ry 3 5.0
and hence 4 8.5
Increasing 5 13.0
Ductility m
fy
fy
uy
uy
implies it will always remain ELASTIC since by substitution in Eq.
above, you will find ue = uy
Cd < R accounts for 1- Effective Damping
2- Energy Dissipation mobilized during Inelastic
response of Highly Ductile systems

The More DUCTILE the system, the Greater the difference bet. Cd & R
 µ = um/uy
18.5 = 3.95in/uy
uy=0.214in this is corresponding to fy

Note that ue = 1.28in

um = 3.08ue = 18.5uy
N.B.: In this example ud = uy knowing that ud = ue/R
and always uy = ue/R
 Slope of u-t relationship

Curvature
or Change
in the slope

written at each specific time increment

Real I.Cs

will serve as

by substituting in the
Equation of Motion

N.B. Stability of the Solution is different from Accuracy of the Solution

then write disp. u in terms of β and not δ

∆t
Refer to
bottom of
last page

and using
Eqs. 2 and 3 above
Again should remember difference betwee Stability and Accuracy

Tn = 1.0 s
 Dt/Tn = 0.1 < 1/P then STABLE
i
0
1
2
3

As such we will be studying CONVERGENCE

 Order
of Use
for each
time step
Unconditionally STABLE
i
0
1
2
 Dt/Tn = 0.1 < SQRT(3)/P then STABLE

Stability ensures that the solution is not sensitive to small

perturbations either in the input data or in the model or in
the integration time step, while

Convergence ensures that the solution approaches the correct

(or a specific) value as the mesh is refined or the integration time
step is reduced.
q is the GENERALIZED (SINGLE) Degree of Freedom; it is the SINGLE
independent variable needed to describe the displaced position w.r.t intial
position

for a GENERALIZED SDOF

`
is the GENERALIZED SDOF
theoretically