Documente Academic
Documente Profesional
Documente Cultură
STUDENT:
FRANCESCO PUGLIESE
𝐻𝑋 (𝜔) = (∑ 𝑎𝑗 (𝑖𝜔)𝑗 )
𝑗=0
For F(t) being a covariant stationary process, we can use the equation as written above to write
down the autospectral density of the response of the SDOF system:
Usually the values of damping ratioo are about 5%, i.e. reinforced concrete and masonry
structures, while the exercise gives the value 2%. For low frequency, or better when the value ω
is close to zero, the autospectral density of response is close to the autospectral of the excitation
at that frequency divided by k2(Stiffness square). Conversely, if the frequency is very high, or
better the ω is bigger than the natural frequency, SXX(ω) is approximated by SFF(ω)/m2ω4
For small values of ζ the SDF system acts as a bandpass filter, giving substantial amplification
only to components of {F(t)} that are near ±ω0. In fact, unless SFF(ω) is much larger for some
other frequencies than it is for ω =ω0, it is clear that the SXX (ω) response autospectral density
will be dominated by the frequencies near ±ω0. This leads to the common situation in which the
stochastic response of the SDF system can be considered a narrowband process, which
sometimes results in significant analytical simplifications. The excitation {F(t)}, on the other
hand, can often be considered a broadband process or even approximated by an “equivalent”
white noise.At this stage if we want to justify that F(t) can be approximated by equivalent white
noise , we need to do some considerations. Since SXX (ω) is the product of SFF(ω) and |Hx (ω| 2
and the most significant portion of SXX (ω) comes from the near resonant frequencies with ω ≈
±ω0, we can see that SXX (ω) would not be changed very significantly if the actual SFF(ω) were
replaced by some other broadband function that agreed with SFF(ω) for ω ≈ ±ω0. The simplest
such approximation is the white noise with constant autospectral density S0 = SFF(ω0) . The most
common usage for this approximation is in the computation of the response variance, for which
the approximation gives
Remeber that:
𝜎𝑋2 = 𝐸 {𝑋 2 (𝑡 )} − 𝜇𝑥
But:
𝜇𝑥 = 0
So:
And hence,
𝜋𝑆0
𝜎𝑋2 = 𝐸{𝑋̇ 2 (𝑡 )} − 𝜇𝑥̇ = 𝐸{𝑋̇ 2 (𝑡)} = =
2𝜁𝜔𝑚2
We have a Gaussian and stationary process with zero-mean. For stationary process X and Ẋ, two
random variables, are statistically uncorrelated, and we know for definition they are also
independent, which means that the expected value E[X(t) Ẋ(t)]=0. At this point, it is easy to
compute the value of the joint probability density function:
(𝑥−𝑚)2 𝑥2
1 (− ) 1 (− )
𝑝𝑋 (𝑥) = 𝑒 2𝜎𝑋 2 = 𝑒 2𝜎𝑋 2
√2𝜋𝜎𝑋 √2𝜋𝜎𝑋
𝑥2 (𝑥̇ )2
1 (− ) 1 (−
2𝜎𝑋̇ 2
)
𝑝𝑋,𝑋̇ (𝑥, 𝑥̇ ) = 𝑒 2𝜎𝑋 2 × 𝑒
√2𝜋𝜎𝑋 √2𝜋𝜎𝑋̇
𝐱𝟐 𝒙̇ 𝟐
𝟏 (− −
𝟐𝛔𝐗 𝟐 𝟐𝛔𝐗̇ 𝟐
)
( )
𝐩𝐗,𝐗̇ 𝐱, 𝐱̇ = 𝐞
𝟐𝛑𝛔𝐗 𝛔𝐗̇
𝐟𝐭𝟐 𝒇𝒕 𝒎
𝛔𝐗̇ = √ 𝝈𝟐𝑿̇ = √𝟐. 𝟓 = 𝟏. 𝟓𝟖 = 𝟎. 𝟒𝟖
𝐬𝐞𝐜𝟐 𝒔𝒆𝒄 𝒔
𝐱𝟐 𝒙̇ 𝟐
𝟏 (− 𝟐− 𝟐) 𝟐 𝟐
𝐩𝐗,𝐗̇ (𝐱, 𝐱̇ ) = 𝐞 𝟐𝛔𝐗 𝟐𝛔𝐗̇ = 𝟑. 𝟗𝟗𝟕𝒆(−𝟕𝟖𝟖.𝟔𝟒𝐱 −𝟎.𝟎𝟖𝒙̇ )
𝟐𝛑𝛔𝐗 𝛔𝐗̇
Assuming a Rayleigh distribution, the probability density function for the maxima of response
X(t) can be written as follows:
𝛏 𝛏𝟐
−
𝐩𝚵 (𝛏) = 𝒆 𝟏𝟐.𝟔𝟖×𝟏𝟎−𝟒
𝟔. 𝟑𝟒 × 𝟏𝟎−𝟒
Rayleigh distribution
30
Probability density function
25
20
15
10
0
0 1 2 3 4 5 6 7 8 9
η
Where γ it the Euler’s constant whose value is 0.5772, while T is the duration whose value is 30
sec and finally v is the rate of zero up-crossing of the process so calculated:
𝜔0 62.8
𝜐= = ≅ 10𝑠𝑒𝑐 −1
2𝜋 2𝜋
𝟎. 𝟓𝟕𝟕𝟐
𝑬[𝑿𝒎 ] = {√𝟐 𝐥𝐧(𝟏𝟎 × 𝟑𝟎) + } 𝟎. 𝟎𝟐𝟓𝟐
√𝟐 𝐥𝐧(𝟏𝟎 × 𝟑𝟎)
𝟎. 𝟓𝟕𝟕𝟐
𝑬[𝑿𝒎 ] = {𝟑. 𝟑𝟕𝟕𝟓𝟏 + } 𝟎. 𝟎𝟐𝟓𝟐 = 𝟎. 𝟎𝟖𝟗
𝟑. 𝟑𝟕𝟕𝟓𝟏
6th QUESTION
The standard deviation of the extrema value of PDF as written in the first part of exercise can
be computed by the following expression:
𝜋
𝜎𝑌𝑒 = σX = 0.0095 𝑓𝑡
√6𝛼
Where:
α = √2 ln(υT) = 3.37751