-
Analysis and Mitigation of Cascading Failures Using a Stochastic Interaction Graph with Eigen-analysis
Authors:
Zhenping Guo,
Xiaowen Su,
Kai Sun,
Byungkwon Park,
Srdjan Simunovic
Abstract:
In studies on complex network systems using graph theory, eigen-analysis is typically performed on an undirected graph model of the network. However, when analyzing cascading failures in a power system, the interactions among failures suggest the need for a directed graph beyond the topology of the power system to model directions of failure propagation. To accurately quantify failure interactions…
▽ More
In studies on complex network systems using graph theory, eigen-analysis is typically performed on an undirected graph model of the network. However, when analyzing cascading failures in a power system, the interactions among failures suggest the need for a directed graph beyond the topology of the power system to model directions of failure propagation. To accurately quantify failure interactions for effective mitigation strategies, this paper proposes a stochastic interaction graph model and associated eigen-analysis. Different types of modes on failure propagations are defined and characterized by the eigenvalues of a stochastic interaction matrix, whose absolute values are unity, zero, or in between. Finding and interpreting these modes helps identify the probable patterns of failure propagation, either local or widespread, and the participating components based on eigenvectors. Then, by lowering the failure probabilities of critical components highly participating in a mode of widespread failures, cascading can be mitigated. The validity of the proposed stochastic interaction graph model, eigen-analysis and the resulting mitigation strategies is demonstrated using simulated cascading failure data on an NPCC 140-bus system.
△ Less
Submitted 12 March, 2025;
originally announced March 2025.
-
Recursive Schur Decomposition
Authors:
Rahul S. Sampath,
Bobby Philip,
Srikanth Allu,
Srdjan Simunovic
Abstract:
In this article, we present a parallel recursive algorithm based on multi-level domain decomposition that can be used as a precondtioner to a Krylov subspace method to solve sparse linear systems of equations arising from the discretization of partial differential equations (PDEs). We tested the effectiveness of the algorithm on several PDEs using different number of sub-domains (ranging from 8 to…
▽ More
In this article, we present a parallel recursive algorithm based on multi-level domain decomposition that can be used as a precondtioner to a Krylov subspace method to solve sparse linear systems of equations arising from the discretization of partial differential equations (PDEs). We tested the effectiveness of the algorithm on several PDEs using different number of sub-domains (ranging from 8 to 32768) and various problem sizes (ranging from about 2000 to over a billion degrees of freedom). We report the results from these tests; the results show that the algorithm scales very well with the number of sub-domains.
△ Less
Submitted 23 October, 2012;
originally announced October 2012.
-
Anomalous Roughness of Fracture Surfaces in 2D Fuse Models
Authors:
Phani K. V. V. Nukala,
Stefano Zapperi,
Mikko Alava,
Srdjan Simunovic
Abstract:
We study anomalous scaling and multiscaling of two-dimensional crack profiles in the random fuse model using both periodic and open boundary conditions. Our large scale and extensively sampled numerical results reveal the importance of crack branching and coalescence of microcracks, which induce jumps in the solid-on-solid crack profiles. Removal of overhangs (jumps) in the crack profiles elimin…
▽ More
We study anomalous scaling and multiscaling of two-dimensional crack profiles in the random fuse model using both periodic and open boundary conditions. Our large scale and extensively sampled numerical results reveal the importance of crack branching and coalescence of microcracks, which induce jumps in the solid-on-solid crack profiles. Removal of overhangs (jumps) in the crack profiles eliminates the multiscaling observed in earlier studies and reduces anomalous scaling. We find that the probability density distribution $p(Δh(\ell))$ of the height differences $Δh(\ell) = [h(x+\ell) - h(x)]$ of the crack profile obtained after removing the jumps in the profiles has the scaling form $p(Δh(\ell)) = <Δh^2(\ell)>^{-1/2} ~f(\frac{Δh(\ell)}{<Δh^2(\ell)>^{1/2}})$, and follows a Gaussian distribution even for small bin sizes $\ell$. The anomalous scaling can be summarized with the scaling relation $[\frac{<Δh^2(\ell)>^{1/2}}{<Δh^2(L/2)>^{1/2}}]^{1/ζ_{loc}} + \frac{(\ell-L/2)^2}{(L/2)^2} = 1$, where $<Δh^2(L/2)>^{1/2} \sim L^ζ$.
△ Less
Submitted 14 April, 2008;
originally announced April 2008.
-
Crack Roughness in the 2D Random Threshold Beam Model
Authors:
Phani K. V. V. Nukala,
Stefano Zapperi,
Mikko Alava,
Srdjan Simunovic
Abstract:
We study the scaling of two-dimensional crack roughness using large scale beam lattice systems. Our results indicate that the crack roughness obtained using beam lattice systems does not exhibit anomalous scaling in sharp contrast to the simulation results obtained using scalar fuse lattices. The local and global roughness exponents ($ζ_{loc}$ and $ζ$, respectively) are equal to each other, and…
▽ More
We study the scaling of two-dimensional crack roughness using large scale beam lattice systems. Our results indicate that the crack roughness obtained using beam lattice systems does not exhibit anomalous scaling in sharp contrast to the simulation results obtained using scalar fuse lattices. The local and global roughness exponents ($ζ_{loc}$ and $ζ$, respectively) are equal to each other, and the two-dimensional crack roughness exponent is estimated to be $ζ_{loc} = ζ= 0.64 \pm 0.02$. Removal of overhangs (jumps) in the crack profiles eliminates even the minute differences between the local and global roughness exponents. Furthermore, removing these jumps in the crack profile completely eliminates the multiscaling observed in other studies. We find that the probability density distribution $p(Δh(\ell))$ of the height differences $Δh(\ell) = [h(x+\ell) - h(x)]$ of the crack profile obtained after removing the jumps in the profiles follows a Gaussian distribution even for small window sizes ($\ell$).
△ Less
Submitted 14 April, 2008;
originally announced April 2008.
-
Time Parallel Scalable Multiphysics/Multiscale Framework
Authors:
George Frantziskonis,
Krishna Muralidharan,
Pierre Deymier,
Srdjan Simunovic,
Sreekanth Pannala
Abstract:
We propose a new computational framework that combines the recently developed time-parallel (TP) and the compound wavelet matrix (CWM) methods. The framework, termed tpCWM, offers significant computational acceleration by making multiscale/multiphysics simulations computationally scalable in time and space domains. We demonstrate the accuracy and the scalability of the method on a prototype prob…
▽ More
We propose a new computational framework that combines the recently developed time-parallel (TP) and the compound wavelet matrix (CWM) methods. The framework, termed tpCWM, offers significant computational acceleration by making multiscale/multiphysics simulations computationally scalable in time and space domains. We demonstrate the accuracy and the scalability of the method on a prototype problem with oscillatory trajectory. The method corrects the coarse solution by iterative use of the CWM, which compounds the fine and the coarse solutions for the processes. Computational savings, over the fine solution as well as the TP method, in terms of the real time required to perform the simulations, can reach several orders of magnitude. We believe that this method is general enough to be applicable to a wide-class of computational physics problems. Tendency towards large number of cores and processors in parallel computers is compatible with the computational scalability of the algorithm.
△ Less
Submitted 7 April, 2008; v1 submitted 31 March, 2008;
originally announced April 2008.
-
Effect of Disorder and Notches on Crack Roughness
Authors:
Phani K. V. V. Nukala,
Stefano Zapperi,
Mikko J. Alava,
Srdjan Simunovic
Abstract:
We analyze the effect of disorder and notches on crack roughness in two dimensions. Our simulation results based on large system sizes and extensive statistical sampling indicate that the crack surface exhibits a universal local roughness of $ζ_{loc} = 0.71$ and is independent of the initial notch size and disorder in breaking thresholds. The global roughness exponent scales as $ζ= 0.87$ and is…
▽ More
We analyze the effect of disorder and notches on crack roughness in two dimensions. Our simulation results based on large system sizes and extensive statistical sampling indicate that the crack surface exhibits a universal local roughness of $ζ_{loc} = 0.71$ and is independent of the initial notch size and disorder in breaking thresholds. The global roughness exponent scales as $ζ= 0.87$ and is also independent of material disorder. Furthermore, we note that the statistical distribution of crack profile height fluctuations is also independent of material disorder and is described by a Gaussian distribution, albeit deviations are observed in the tails.
△ Less
Submitted 22 August, 2007;
originally announced August 2007.
-
Scaling of Fracture Strength in Disordered Quasi-Brittle Materials
Authors:
Phani Kumar V. V. Nukala,
Srdjan Simunovic
Abstract:
This paper presents two main results. The first result indicates that in materials with broadly distributed microscopic heterogeneities, the fracture strength distribution corresponding to the peak load of the material response does not follow the commonly used Weibull and (modified) Gumbel distributions. Instead, a {\it lognormal} distribution describes more adequately the fracture strengths co…
▽ More
This paper presents two main results. The first result indicates that in materials with broadly distributed microscopic heterogeneities, the fracture strength distribution corresponding to the peak load of the material response does not follow the commonly used Weibull and (modified) Gumbel distributions. Instead, a {\it lognormal} distribution describes more adequately the fracture strengths corresponding to the peak load of the response. Lognormal distribution arises naturally as a consequence of multiplicative nature of large number of random distributions representing the stress scale factors necessary to break the subsequent "primary" bond (by definition, an increase in applied stress is required to break a "primary" bond) leading up to the peak load. Numerical simulations based on two-dimensional triangular and diamond lattice topologies with increasing system sizes substantiate that a {\it lognormal} distribution represents an excellent fit for the fracture strength distribution at the peak load. The second significant result of the present study is that, in materials with broadly distributed microscopic heterogeneities, the mean fracture strength of the lattice system behaves as $μ_f = \frac{μ_f^\star}{(Log L)^ψ} + \frac{c}{L}$, and scales as $μ_f \approx \frac{1}{(Log L)^ψ}$ as the lattice system size, $L$, approaches infinity.
△ Less
Submitted 30 March, 2005;
originally announced March 2005.
-
An Efficient Block Circulant Preconditioner For Simulating Fracture Using Large Fuse Networks
Authors:
Phani Kumar V. V. Nukala,
Srdjan Simunovic
Abstract:
{\it Critical slowing down} associated with the iterative solvers close to the critical point often hinders large-scale numerical simulation of fracture using discrete lattice networks. This paper presents a block circlant preconditioner for iterative solvers for the simulation of progressive fracture in disordered, quasi-brittle materials using large discrete lattice networks. The average compu…
▽ More
{\it Critical slowing down} associated with the iterative solvers close to the critical point often hinders large-scale numerical simulation of fracture using discrete lattice networks. This paper presents a block circlant preconditioner for iterative solvers for the simulation of progressive fracture in disordered, quasi-brittle materials using large discrete lattice networks. The average computational cost of the present alorithm per iteration is $O(rs log s) + delops$, where the stiffness matrix ${\bf A}$ is partioned into $r$-by-$r$ blocks such that each block is an $s$-by-$s$ matrix, and $delops$ represents the operational count associated with solving a block-diagonal matrix with $r$-by-$r$ dense matrix blocks. This algorithm using the block circulant preconditioner is faster than the Fourier accelerated preconditioned conjugate gradient (PCG) algorithm, and alleviates the {\it critical slowing down} that is especially severe close to the critical point. Numerical results using random resistor networks substantiate the efficiency of the present algorithm.
△ Less
Submitted 30 March, 2005;
originally announced March 2005.
-
An Efficient Algorithm For Simulating Fracture Using Large Fuse Networks
Authors:
Phani Kumar V. V. Nukala,
Srdjan Simunovic
Abstract:
The high computational cost involved in modeling of the progressive fracture simulations using large discrete lattice networks stems from the requirement to solve {\it a new large set of linear equations} every time a new lattice bond is broken. To address this problem, we propose an algorithm that combines the multiple-rank sparse Cholesky downdating algorithm with the rank-p inverse updating a…
▽ More
The high computational cost involved in modeling of the progressive fracture simulations using large discrete lattice networks stems from the requirement to solve {\it a new large set of linear equations} every time a new lattice bond is broken. To address this problem, we propose an algorithm that combines the multiple-rank sparse Cholesky downdating algorithm with the rank-p inverse updating algorithm based on the Sherman-Morrison-Woodbury formula for the simulation of progressive fracture in disordered quasi-brittle materials using discrete lattice networks. Using the present algorithm, the computational complexity of solving the new set of linear equations after breaking a bond reduces to the same order as that of a simple {\it backsolve} (forward elimination and backward substitution) {\it using the already LU factored matrix}. That is, the computational cost is $O(nnz({\bf L}))$, where $nnz({\bf L})$ denotes the number of non-zeros of the Cholesky factorization ${\bf L}$ of the stiffness matrix ${\bf A}$. This algorithm using the direct sparse solver is faster than the Fourier accelerated preconditioned conjugate gradient (PCG) iterative solvers, and eliminates the {\it critical slowing down} associated with the iterative solvers that is especially severe close to the critical points. Numerical results using random resistor networks substantiate the efficiency of the present algorithm.
△ Less
Submitted 30 March, 2005;
originally announced March 2005.
-
Statistical properties of fracture in a random spring model
Authors:
Phani Kumar V. V. Nukala,
Stefano Zapperi,
Srdan Simunovic
Abstract:
Using large scale numerical simulations we analyze the statistical properties of fracture in the two dimensional random spring model and compare it with its scalar counterpart: the random fuse model. We first consider the process of crack localization measuring the evolution of damage as the external load is raised. We find that, as in the fuse model, damage is initially uniform and localizes at…
▽ More
Using large scale numerical simulations we analyze the statistical properties of fracture in the two dimensional random spring model and compare it with its scalar counterpart: the random fuse model. We first consider the process of crack localization measuring the evolution of damage as the external load is raised. We find that, as in the fuse model, damage is initially uniform and localizes at peak load. Scaling laws for the damage density, fracture strength and avalanche distributions follow with slight variations the behavior observed in the random fuse model. We thus conclude that scalar models provide a faithful representation of the fracture properties of disordered systems.
△ Less
Submitted 3 February, 2005;
originally announced February 2005.
-
Crack avalanches in the three dimensional random fuse model
Authors:
Stefano Zapperi,
Phani Kumar V. V. Nukala,
Srđan Šimunović
Abstract:
We analyze the scaling of avalanche precursors in the three dimensional random fuse model by numerical simulations. We find that both the integrated and non-integrated avalanche size distributions are in good agreement with the results of the global load sharing fiber bundle model, which represents the mean-field limit of the model.
We analyze the scaling of avalanche precursors in the three dimensional random fuse model by numerical simulations. We find that both the integrated and non-integrated avalanche size distributions are in good agreement with the results of the global load sharing fiber bundle model, which represents the mean-field limit of the model.
△ Less
Submitted 14 January, 2005;
originally announced January 2005.
-
Crack roughness and avalanche precursors in the random fuse model
Authors:
Stefano Zapperi,
Phani Kumar V. V. Nukala,
Srdan Simunovic
Abstract:
We analyze the scaling of the crack roughness and of avalanche precursors in the two dimensional random fuse model by numerical simulations, employing large system sizes and extensive sample averaging. We find that the crack roughness exhibits anomalous scaling, as recently observed in experiments. The roughness exponents ($ζ$, $ζ_{loc}$) and the global width distributions are found to be univer…
▽ More
We analyze the scaling of the crack roughness and of avalanche precursors in the two dimensional random fuse model by numerical simulations, employing large system sizes and extensive sample averaging. We find that the crack roughness exhibits anomalous scaling, as recently observed in experiments. The roughness exponents ($ζ$, $ζ_{loc}$) and the global width distributions are found to be universal with respect to the lattice geometry. Failure is preceded by avalanche precursors whose distribution follows a power law up to a cutoff size. While the characteristic avalanche size scales as $s_0 \sim L^D$, with a universal fractal dimension $D$, the distribution exponent $τ$ differs slightly for triangular and diamond lattices and, in both cases, it is larger than the mean-field (fiber bundle) value $τ=5/2$.
△ Less
Submitted 21 July, 2004;
originally announced July 2004.
-
Percolation and localization in the random fuse model
Authors:
Phani Kumar V. V. Nukala,
Srdan Simunovic,
Stefano Zapperi
Abstract:
We analyze damage nucleation and localization in the random fuse model with strong disorder using numerical simulations. In the initial stages of the fracture process, damage evolves in an uncorrelated manner, resembling percolation. Subsequently, as the damage starts to accumulate, current enhancement at the tips of the microcracks leads eventually to catastrophic failure. We study this behavio…
▽ More
We analyze damage nucleation and localization in the random fuse model with strong disorder using numerical simulations. In the initial stages of the fracture process, damage evolves in an uncorrelated manner, resembling percolation. Subsequently, as the damage starts to accumulate, current enhancement at the tips of the microcracks leads eventually to catastrophic failure. We study this behavior quantifying the deviations from percolation and discussing alternative scaling laws for damage. The analysis of damage profiles confirms that localization occurs abruptly starting from an uniform damage landscape. Finally, we show that the cumulative damage distribution follows the normal distribution, suggesting that damage is uncorrelated on large length scales.
△ Less
Submitted 9 August, 2004; v1 submitted 12 November, 2003;
originally announced November 2003.
