A two degree-of-fredom primary/secondary damped oscillator

We consider a two degree-of-fredom primary-secondary damped oscillator. This system is composed of a two-stage oscillator characterized by a mass, a stiffness and a damping ratio for each of the two oscillators. This system is submitted to a white-noise excitation. The limit-state function is highly nonlinear, mainly due to the interactions between the two stages of the system, and presents one failure zone.

use case oscillator

Two stage oscillator

The limit state function is defined as follows:

G = F_s - 3 k_s \sqrt{\frac{\pi S_0}{4 \zeta_s \omega_s^3} \left[\frac{\zeta_a \zeta_s}{\zeta_p \zeta_s (4 \zeta_a^2 + \theta^2)+\gamma \zeta_a^2}\frac{(\zeta_p \omega_p^3 + \zeta_s \omega_s^3)\omega_p}{4 \zeta_a \omega_a^4}\right]}

The natural frequency of the first oscillator is equal to:

\omega_p = \sqrt{\frac{k_p}{m_p}}

The natural frequency of the secondary oscillator is equal to:

\omega_s = \sqrt{\frac{k_s}{m_s}}

The average natural frequency of the system is equal to:

\omega_a = \frac{\omega_p+\omega_s}{2}

The average damping ratio of the system is equal to:

\zeta_a = \frac{\zeta_p+\zeta_s}{2}

The mass ratio is equal to:

\gamma = \frac{m_s}{m_p}

The tuning parameter of the system is equal to:

\theta = \frac{\omega_p - \omega_s}{\omega_a}

Eight uncertainties are considered in the system:

  • on the masses of the primary and secondary systems (m_p and m_s),

  • on the spring stiffeness of the primary and secondary oscillators (k_p and k_s),

  • on the damping ratios of the primary and secondary systems (\zeta_p and \zeta_s),

  • on the loading capacity of the secondary spring (F_s),

  • on the intensity of the white noise excitation (S_0).

We consider the following distribution fResponse of Two-Degree-of-Freedom Systems to White-Noise Base Excitation:

  • m_p \sim \text{LogNormalSigmaOverMu}(  \mu_{m_p} = 1.5, \delta_{m_p} = 0.1)

  • m_s \sim \text{LogNormalSigmaOverMu}(  \mu_{m_s} = 0.01,  \delta_{m_s} = 0.1)

  • k_p \sim \text{LogNormalSigmaOverMu}(  \mu_{k_p} = 1,  \delta_{k_p} = 0.2)

  • k_s \sim \text{LogNormalSigmaOverMu}(  \mu_{k_s} = 0.01,  \delta_{k_s} = 0.2)

  • \zeta_p \sim \text{LogNormalSigmaOverMu}(  \mu_{\zeta_p} = 0.05, \delta_{\zeta_p} = 0.4)

  • \zeta_s \sim \text{LogNormalSigmaOverMu}(  \mu_{\zeta_s} = 0.02,  \delta_{\zeta_s} = 0.5)

  • F_s \sim \text{LogNormalSigmaOverMu}(  \mu_{F_s} = 27.5,  \delta_{F_s} = 0.1)

  • S_0 \sim \text{LogNormalSigmaOverMu}( \mu_{S_0} = 100,  \delta_{S_0} = 0.1)

The failure probability is:

P_f = \Prob{G(m_p,m_s,k_p,k_z,\zeta_p,\zeta_s,F_s,Z_0) \leq 0}.

The value of P_f is:

P_f = 3.78 \times 10^{-7}.

References

  • Der Kiureghian, A. and De Stefano, M. (1991). Efficient Algorithm for Second-Order Reliability Analysis, Journal of engineering mechanics, 117(12), 2904-2923

  • Dubourg, V., Sudret, B., Deheeger, F. (2013). Metamodel-based importance sampling for structural reliability analysis. Probabilistic Engineering Mechanics, 33, pp. 47–57

API documentation

class Oscillator

Data class for the oscillator example.

Examples

>>> from openturns.usecases import oscillator
>>> # Load the oscillator
>>> osc = oscillator.Oscillator()
Attributes:
dimint

dim = 8, dimension of the problem

modelSymbolicFunction

The limit state function

muMpfloat

muMp = 1.5, mean of the mass of the primary system

sigmaOverMuMpfloat

sigmaOverMuMp = 0.1, coefficient of variation of the mass of the primary system

distributionMpLogNormal distribution of the mass of the primary system

distributionMp = ot.LogNormalMuSigmaOverMu(muMp, sigmaOverMuMp).getDistribution()

muMsfloat

muMs = 0.01, mean of the mass of the primary system

sigmaOverMuMsfloat

sigmaOverMuMs = 0.1, coefficient of variation of the mass of the primary system

distributionMsLogNormal distribution of the mass of the secondary system

distributionMs = ot.LogNormalMuSigmaOverMu(muMs, sigmaOverMuMs).getDistribution()

muKpfloat

muKp = 1, mean of the spring stiffness of the primary system

sigmaOverMuKpfloat

sigmaOverMuKp = 0.2, coefficient of variation of the spring stiffness of the primary system

distributionKpLogNormal distribution of the spring stiffness of the primary system

distributionKp = ot.LogNormalMuSigmaOverMu(muKp, sigmaOverMuKp).getDistribution()

muKsfloat

muKs = 0.01, mean of the spring stiffness of the secondary system

sigmaOverMuKsfloat

sigmaOverMuKs = 0.2, coefficient of variation of the spring stiffness of the secondary system

distributionKsLogNormal distribution of the spring stiffness of the secondary system

distributionKs = ot.LogNormalMuSigmaOverMu(muKs, sigmaOverMuKs).getDistribution()

muZetapfloat

muZetap = 0.05, mean of the damping ratio of the primary system

sigmaOverMuZetapfloat

sigmaOverMuZetap = 0.4, coefficient of variation of the damping ratio of the primary system

distributionZetapLogNormal distribution of the damping ratio of the primary system

distributionZetap = ot.LogNormalMuSigmaOverMu(muZetap, sigmaOverMuZetap).getDistribution()

muZetasfloat

muZetas = 0.02, mean of the damping ratio of the secondary system

sigmaOverMuZetasfloat

sigmaOverMuZetas = 0.5, coefficient of variation of the damping ratio of the secondary system

distributionZetasLogNormal distribution of the damping ratio of the secondary system

distributionZetas = ot.LogNormalMuSigmaOverMu(muZetas, sigmaOverMuZetas).getDistribution()

muFsfloat

muFs = 27.5, mean of the loading capacity of the secondary spring

sigmaOverFsfloat

sigmaOverFs = 0.1, coefficient of variation of the loading capacity of the secondary spring

distributionFsLogNormal distribution of the loading capacity of the secondary spring

distributionFs = ot.LogNormalMuSigmaOverMu(muFs, sigmaOverFs).getDistribution()

muS0float

muS0 = 100, mean of the intensity of the white noise

sigmaOverS0float

sigmaOverS0 = 0.1, coefficient of variation of the intensity of the white noise

distributionS0LogNormal distribution of the intensity of the white noise

distributionS0 = ot.LogNormalMuSigmaOverMu(muS0, sigmaOverS0).getDistribution()

distributionJointDistribution

The joint distribution of the input parameters

Examples based on this use case

Using the FORM - SORM algorithms on a nonlinear function

Using the FORM - SORM algorithms on a nonlinear function