Collective effects of link failures in linear flow networks

Dirk Witthaut; Franz Kaiser; Julius Strake

REVIEW 3 major objections 10 minor 48 references

Collective effects of link failures in linear flow networks

T0 review · 3 major / 10 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Simultaneous line outages in flow networks have genuinely collective effects, and only in regular lattices do they reduce to a clean sum of dipoles.

desk verdict A genuinely useful N-2 screening paper with one real gap in the continuum-limit derivation that is likely fixable; deserves review, not desk rejection. read the letter →

arxiv 1909.00774 v1 pith:4PBTYQYG submitted 2019-09-02 physics.soc-ph

Franz Kaiser , Julius Strake , Dirk Witthaut This is my paper · ORCID

classification physics.soc-ph

keywords powergridslineoutagedistributionfactorsmultiplelinkoutagescollectiveflowreroutingdipoleapproximationBraessparadoxlinearnetworkscontinuumlimit

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Simultaneous failure of several links in a linear flow network is not the sum of the individual failures: the failed lines interact through the flows they reroute onto each other. The paper derives the exact coupling formula for two outages and shows concrete grids where the pair amplifies a link beyond its limit, relieves the grid like Braess's paradox, or even reverses a flow's direction even though each single failure had pushed it the same way. For many outages it derives a compact nodal equation whose sources are K effective dipoles, and in the continuum limit of an infinite homogeneous square lattice it claims these dipoles decouple completely, so the potential change is exactly $\psi(\mathbf r)=\sum_k \mathbf q_k\cdot(\mathbf r-\mathbf r_k)/\|\mathbf r-\mathbf r_k\|^2$. It also introduces a topology-only predictor $\Lambda(o,k)=\sqrt{L_{o,k}L_{k,o}}$ that tracks the strength of collective effects with near-perfect rank correlation on several test grids. The practical stakes are N-1-secure grids: the second failure is where intuition from single-outage theory breaks down.

What carries the argument

The carrying object is the generalized outage equation $\psi=B^\dagger D(1_K-P)^{-1}F_{\rm out}^{(0)}$, with $B^\dagger$ the Moore-Penrose inverse of the graph Laplacian, $D$ the node-edge vectors of the failing links, $P$ the projected Power Transfer Distribution Factor matrix among the failing links, and $F_{\rm out}^{(0)}$ the pre-outage flows on them. The inverse $(1_K-P)^{-1}$ encodes all collective interactions: its off-diagonal entries decide whether failing links amplify or cancel each other, and its diagonal would leave the naive superposition of single failures. In the continuum derivation, the paper identifies the projected PTDF entries as mixed second derivatives of the Green's function and concludes they vanish off-diagonally, which turns the inverse into the identity and produces the clean dipole sum. For two failures it also introduces the mutual-LODF predictor $\Lambda(o,k)=\sqrt{L_{o,k}L_{k,o}}$ used to forecast when those off-diagonal entries matter.

What would settle it

Compute the discrete entry $P_{ki}=b_k\,d_k^\top B^\dagger d_i$ for two distinct links in a large periodic square lattice and test whether it is nonzero. If it is nonzero for $k\ne i$, the projected PTDF matrix is not diagonal and the exact dipole superposition fails; equivalently, numerically compare the flow change after two simultaneous outages in a large homogeneous lattice with the sum of two dipole fields and check whether the difference tends to zero with the lattice spacing.

Watch

Extended reading notes

Core claim

The paper's central claim is that multiple link failures have an intrinsically collective component, controlled by the mutual line-outage distribution factors between the failing lines, and that this component vanishes only under special symmetry. In arbitrary networks the exact flow change after two outages contains the prefactor $(1-L_{o,k}L_{k,o})^{-1}$; when the mutual factors are large, collective effects can dominate. The paper shows that the strength of these effects is already captured by the single parameter $\Lambda(o,k)=\sqrt{L_{o,k}L_{k,o}}$, and it proves $\xi(o,k)\ge \Lambda(o,k)$ for the collectivity measure it defines. For $K$ simultaneous outages it derives the closed form $\psi=B^\dagger D(1_K-P)^{-1}F_{\rm out}^{(0)}$, then shows that in the continuum limit of a regular square lattice the matrix $(1_K-P)^{-1}$ becomes the identity, leaving $\psi(\mathbf r)=\sum_k \mathbf q_k\cdot(\mathbf r-\mathbf r_k)/\|\mathbf r-\mathbf r_k\|^2$ and the corresponding dipole formula for flow changes. On the paper's own account, collective effects in homogeneous lattices are therefore exactly the superposition of single-outage effects, while in heterogeneous real grids they are common and predictable.

Load-bearing premise

The continuum-limit superposition rests on the premise that, in a large regular grid, the mutual influence between two distinct failing lines vanishes exactly as the grid spacing goes to zero; if that mutual influence does not disappear, the dipole formula is at best an approximation.

Editorial extensions

If this is right

N-1 secure operation does not guard against N-2 events: two failures whose mutual LODFs are large can overload a line that the two single outages leave safe.
The predictor $\Lambda(o,k)$ lets operators rank line pairs by collective risk using topology alone, narrowing the search space for N-2 contingency screening.
In large regular grids, multiple outages become analytically tractable: flow changes are sums of independent dipole fields with no interaction correction.
Because a second outage can reduce the maximum loading below either single outage, controlled disconnection of a carefully chosen line is a possible overload mitigation strategy.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Editorial inference: if the diagonalization premise holds, the residual collective effect in a finite but large homogeneous grid should decay with lattice spacing; the paper does not compute this finite-size correction.
Editorial inference: the success of $\Lambda$ suggests a graph invariant analogous to mutual effective resistance that could rank dangerous N-2 pairs without enumerating all outage scenarios.
Editorial inference: the same continuum reasoning might extend to other periodic lattices whose Green's-function second derivatives are localizable, but the paper only claims the square lattice.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Here’s my take on arXiv:1909.00774. I largely agree with the reader’s conditional verdict. The core two-failure algebra is correct, the lambda predictor is a useful screening tool, and Theorem 1’s lower bound is real. But the continuum-limit derivation in Section IV.B has a genuine gap: the claim that off-diagonal projected PTDF entries are exactly zero because delta functions have different arguments is wrong. The inverse Laplacian is nonlocal; the off-diagonal terms are actually O(h^2) and do not vanish identically. The superposition of dipoles might still hold at leading order for fixed K and fixed physical separation, but the paper does not provide that scaling argument. This is the load-bearing weak spot and it needs fixing, not just polishing.

What is genuinely new: the collectivity parameter xi, the lambda predictor with its lower-bound theorem, and the K-dipole continuum picture. The examples of attenuation and flow sign inversion are striking and correctly computed from Eq. (9). The empirical part shows lambda beating distance measures, which is credible even though part of the correlation is built in, since lambda and xi are both functions of the same LODF matrix. I would not call that circular in a damning sense—Theorem 1 gives the bound—but the 0.998 correlation should be framed as a consistency check rather than as independent confirmation.

Soft spots in proportion: the continuum gap is substantial; the predictor validation is against a derived proxy rather than actual flow changes; and the DC approximation plus homogeneous lattices limit practical reach, though that is standard for this literature. The literature review is adequate, and earlier LODF work is cited properly.

Who gets value: power-systems people working on N-2 screening, and the network-physics crowd interested in dipole analogies. It deserves a serious referee despite the gap, because the main contributions are usable and the flaw is localized. I would send it out and ask the authors to supply the missing scaling argument. My recommendation: accept for peer review with heavy revision; my own verdict is conditional, not reject.

Referee Report

3 major / 10 minor

Summary. The manuscript analyses simultaneous failures ('N−K outages') of several links in linear flow networks in the DC-approximation of power flow. It derives the exact two-outage flow-change formula (Eq. 9), which contains a coupling term compared to naive superposition; demonstrates with small explicit grids three counterintuitive effects (amplification, attenuation via Braess paradox, and sign inversion of flow changes); generalizes to arbitrary K outages using the Woodbury identity and the projected PTDF matrix, obtaining the nodal equation Bψ = D(1_K−P)^{-1}F^{(0)}_{out} (Eqs. 16–19); and proposes a collectivity parameter ξ(o,k) with a predictor Λ(o,k)=√(L_{o,k}L_{k,o}), proving a lower bound (Theorem 1) and reporting Pearson correlations ρ≈0.998 on test grids. Finally, Section IV.B claims that in the continuum limit for homogeneous square lattices the projected PTDF matrix becomes diagonal, so that K simultaneous failures reduce exactly to a superposition of K independent dipole fields (Eqs. 21–24), with collective effects completely vanishing.

Significance. The algebraic core of this paper is sound and useful: the two-outage formula and the Woodbury-based K-outage expression are correct, Theorem 1 is a genuine lower bound with a complete proof, and the elementary examples are explicit and instructive. If the continuum-limit claim in Section IV.B were established, it would be a valuable analytic tool for large regular grids, reducing N−K contingencies to a superposition of K dipole fields, and would meaningfully extend the authors' single-line work [17]. The numerical comparison across five test grids (Table I) is a clear asset. However, two load-bearing points currently undercut the headline claims: the derivation of the continuum diagonalization is invalid as written, and the near-perfect predictor correlation is largely a consequence of the algebraic construction rather than an independent validation.

major comments (3)

[Section IV.B, Eqs. (21)–(24)] The claim that 'all off-diagonal entries are zero due to the delta functions' different arguments' is not correct. The matrix element P_{ki} = b_k d_k^T B† d_i contains the nonlocal inverse Laplacian (two-point Green's kernel G(r,r')); after the continuum substitution d ≈ h∇δ it evaluates to a mixed second derivative of G at the two link positions, b ∂²G/∂x_k∂x'_i(r_k,r_i), which is generically nonzero for k≠i. The derivation instead writes a single local function b†(x,y) inside the integral, which amounts to replacing B† by a multiplication operator. The resulting exact diagonalization of (1_K−P)^{-1} is therefore unsupported. The leading-order dipole superposition may still be true, but only as an asymptotic statement for fixed K and fixed macroscopic separations r_k−r_i = O(1) as h→0, where the off-diagonal entries are O(h²); this scaling argument is absent. Worse, for failing links separated by O(h) (neighboring lattice links), the off-diagonal entries are O(1), the mixing term does not vanish in the continuum limit, and the dipole formula Eq. (23) becomes singular as r_k→r_i. This is precisely the regime in which Section III.B shows collective effects to be strongest, so the abstract's unconditional claim that 'collective effects completely vanish in the continuum limit' is not established.
[Section IV.B, Eqs. (21)–(24)] The same continuum calculation mishandles the diagonal normalization. The entries P_{kk}=b_k d_k^T B† d_k are O(1) (the self-PTDF, which for an interior link of a homogeneous square lattice is a positive constant near 1/2), not O(h²), so (1−P_{kk})^{-1} ≠ 1; the dipole-source strength in Eq. (21) should be the renormalized flow [(1_K−P)^{-1}]_{kk}F^{(0)}_k, as already contained in Eqs. (16)–(18), rather than the bare 'unperturbed current field' F^{(0)}. In addition, the stated solution of Eq. (22) is quantitatively incomplete: for constant b the 2D Green's function gives ψ(r)=Σ_k q_k·(r−r_k)/(2π b |r−r_k|²) up to orientation sign, whereas Eq. (23) has no 1/b or 1/(2π); the same factor affects Eq. (24). Please verify these factors against the known K=1 lattice result (Ref. [17]), where the LODF denominator (1−PTDF_{kk})^{-1} is known to renormalize the single-outage dipole strength.
[Section III.A, Eqs. (12)–(13), Fig. 5, Table I] The claim that Λ(o,k) predicts ξ(o,k) with correlation ρ=0.998 is presented as the validation of a new quantifier, but the correlation is largely built into the construction. As the authors' own Appendix B shows, ξ(o,k) factors as Λ(o,k) × R(o,k), where R involves only the same LODF entries L_{l,o}, L_{l,k}, L_{o,k}, L_{k,o}; the log-log scatter in Fig. 5 therefore mostly confirms the algebraic reduction already derived, not an independent match. The non-circular content is Theorem 1 (the lower bound ξ ≥ Λ), the narrow observed spread of R(o,k) in the tested grids, and the comparison with distance-based predictors (Fig. 6). I recommend reframing the claim accordingly, and ideally testing Λ against the flow-dependent quantity ΔF−ΔF^{naive} for sampled injection vectors P, since ξ itself is built purely from LODFs and drops the initial flows F^{(0)}.

minor comments (10)

[Section IV.B, Eq. (22)] The upper limit of the sum in Eq. (22) is printed as M but should be K, the number of failed links; with M the equation ranges over all links of the infinite lattice and is not the intended dipole-source sum.
[Section IV.B, Eq. (21)] The right-hand side of Eq. (21) refers to the position x_{s_i} although the left-hand side is q_k; the index should be k.
[Section IV.B, Eqs. (19)–(23)] The symbol q_k is overloaded: it denotes the nodal vector d_k F^{(K)}_k in Eq. (19), the continuum dipole moment in Eqs. (21)–(22), and the field value F^{(0)}(x_{s_k},y_{s_k}) in the sentence after Eq. (22); distinct symbols would clarify the scaling argument.
[Appendix A, first paragraph] 'the extend to which' should read 'the extent to which'.
[Section II.B] 'to the same extend' should read 'to the same extent'.
[Section V] 'occurence' should read 'occurrence'.
[Figure 7 caption] 'degreee' should read 'degree'.
[Appendix C, proof of Theorem 1] The '!' placed above the second inequality of the proof is unexplained; use a numbered target inequality instead. The theorem statement should also mention that the proof uses L_{o,o}=L_{k,k}=−1 and the same-sign property of mutual LODFs (Appendix A).
[Figure 5 caption] The statement that 'the slope of the curve indicates a linear relationship on the normal scale' is only meaningful if the slope of the log-log fit is stated, and it should be specified whether the reported Pearson ρ is computed on raw or log-transformed variables.
[Table I caption] 'all possible pairs of inks' should read 'all possible pairs of links'.

Circularity Check

1 steps flagged · score 6.0 of 10

Continuum-limit derivation makes the off-diagonal PTDF entries vanish by representing the nonlocal Green's function as a one-point local field, so the claimed disappearance of collective effects is put in by construction.

self definitional [Section IV.B, paragraph immediately before Eq. (21)]
"All off-diagonal entries are zero due to the delta functions’ different arguments. Importantly, this observation is independent of the orientation of the two links under consideration. The inverted matrix is thus diagonal and can be calculated as [(1K− P)−1]ki = (1− Pki)−1 = (1−O (h2))−1"

The matrix element P_ki is defined as b_k d_k^T B† d_i, where B† is the nonlocal inverse Laplacian. The continuum calculation replaces B† by a one-point function b†(x,y), so the integral contains two delta distributions over the same coordinate at different link positions; disjoint supports force every k≠i element to zero. That diagonality of P, and therefore of (1_K−P)^−1, is precisely what makes the collective effects vanish and yields the dipole superposition. The actual continuum limit of d_k^T B† d_i is a mixed second derivative of the two-point Green's function, which is generally nonzero. Thus the paper does not derive the vanishing of off-diagonal couplings from the discrete equations; it inserts that vanishing through the local representation chosen for B†.

full rationale

The exact discrete multi-link formalism in Eqs. (16)–(19) is non-circular: it derives the Poisson-like equation Bψ = D(1_K−P)^−1 F_out from the Woodbury identity and standard LODF algebra. The predictor Λ(o,k) is a simplified proxy for the collectivity parameter ξ(o,k); it is not a fitted parameter, and Theorem 1 gives a genuine lower bound, while Appendix B explicitly factors ξ into Λ times an ‘other term’ that is empirically weak. The high correlation is therefore partly algebraic but not a fitted-input circularity. The self-citation to the authors' earlier single-link dipole work [17] is also not load-bearing in a circular sense: that work treats single-link failures with parameter-free methods and does not assume the multi-link result being tested. The serious circular step is in the continuum limit: the paper assumes a one-point continuum version b† of the Green's function, which makes the off-diagonal PTDF entries vanish by construction, and from that diagonality concludes that collective effects completely vanish. Because the true Green's function is nonlocal, the off-diagonal entries are generally nonzero, though they are O(h^2) and may vanish in the h→0 limit. The conclusion may be salvageable with a proper scaling argument, but as written the central continuum derivation reduces to an assumption of the locality it claims to prove, warranting a score of 6.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

No constants are fitted to data, so the free-parameter list is empty. The numerical examples choose susceptances and injections by hand to realize the sign-inversion phenomenon, but the general formulas do not depend on these values. The axioms listed are the standard linear-network assumptions plus the specific continuum diagonalization step that the paper does not rigorously justify.

assumptions (5)

domain assumption DC approximation: flows are linear in phase differences; link susceptances are constant and nodal injections are unchanged after an outage.
Standard power-grid modeling assumption (Ref [18]) used throughout Section II; all LODF formulas and examples rest on it.
standard math Moore-Penrose pseudoinverse B-dagger of the weighted Laplacian is used in place of the non-invertible Laplacian, and standard identities such as the Woodbury formula apply.
Invoked in deriving LODF Eq. (6), the K-failure formula Eq. (16), and the generalized Poisson equation Eq. (18).
domain assumption LODFs are bounded by one and mutual LODFs L_o,k and L_k,o have the same sign, so Lambda is real and nonnegative.
Used to define the predictor Lambda in Eq. (13), to drop higher-order LODF terms, and in the proof of Theorem 1; argued in Appendix A and Ref [6].
ad hoc to paper In the continuum limit, the projected PTDF matrix P becomes diagonal because off-diagonal elements vanish; consequently (1_K - P)^(-1) can be treated as diagonal.
Central step of Section IV.B leading to the superposition result Eqs. (22)-(24); the offered justification using delta-function supports is not valid for a nonlocal Green's function and needs a proper scaling argument.
domain assumption A continuum version b-dagger of the lattice Green's function B-dagger exists and can be used in the dipole approximation.
Assumed in Section IV.B before the calculation of P_ki; standard for continuum limits of lattice Laplacians.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Collective effects of link failures in linear flow networks." pith.science (2026). https://pith.science/paper/4PBTYQYG

@misc{pith2026190900774,
  author       = {Pith},
  title        = {Pith review of: Collective effects of link failures in linear flow networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4PBTYQYG}},
  note         = {Machine review of arXiv:1909.00774}
}

read the original abstract

The smooth operation of supply networks is crucial for the proper functioning of many systems, ranging from biological organisms such as the human blood transport system or plant leaves to man-made systems such as power grids or gas pipelines. Whereas the failure of single transmission elements has been analysed thoroughly for power grids, the understanding of multiple failures is becoming more and more important to prevent large scale outages with an increasing penetration of renewable energy sources. In this publication, we examine the collective nature of the simultaneous failure of several transmission elements. In particular, we focus on the difference between single transmission element failures and the collective failure of several elements. We demonstrate that already for two concurrent failures, the simultaneous outage can lead to an inversion of the direction of flow as compared to the two individual failures and find situations where additional outages may be beneficial for the overall system. In addition to that, we introduce a quantifier that performs very well in predicting if two outages act strongly collectively or may be treated as individual failures mathematically. Finally, we extend on recent progress made on the understanding of single link failures demonstrating that multiple link failures may be treated as superpositions of multiple electrical dipoles for lattice-like networks with collective effects completely vanishing in the continuum limit. Our results demonstrate that the simultaneous failure of multiple lines may lead to unexpected effects that cannot be easily described using the theoretical framework for single link failures.

Figures

Figures reproduced from arXiv: 1909.00774 by the authors.

**Figure 3.** FIG. 3. Collective effects can lead to a complete reversal of [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

**Figure 4.** FIG. 4. Two different network topologies are used to demon [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗

Figures from the paper (4 more)

**Figure 5.** Figure 5: FIG. 5. The predictor Λ( [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]

**Figure 6.** Figure 6: FIG. 6. Distance performs moderately in predicting the overall collective effects of a double link failure of two links [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]

**Figure 7.** Figure 7: FIG. 7. With increasing degree of sparsity [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]

**Figure 8.** Figure 8: FIG. 8. (a) The predictor Λ( [PITH_FULL_IMAGE:figures/full_fig_p015_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

48 extracted references · 46 canonical work pages

[17]

C. M. Davis and T. J. Overbye, IEEE Transactions on Power Systems 26, 1294 (2011)

work page 2011
[1]

An elementary example of a network where this is happening is shown in Fig

Amplifying single outages To start with, we present a case where the naive overlay of two individual outages underestimates the collective eﬀects such that |∆Fl|≫| ∆F naive l | for some link l. An elementary example of a network where this is happening is shown in Fig. 1 where the topology is given by a network consisting ofN = 9 nodes and M = 12 links co...

work page
[2]

A minimal example is shown in Figure 2

Eﬀect of individual outages exceeding simultaneous outage In addition to the eﬀect presented in the last section, it may also happen that an additional outage is beneﬁ- cial for the maximally loaded link. A minimal example is shown in Figure 2. In this setup, we have a unit inﬂow of power at the centre left node (coloured yellow) P = +1 and a unit outﬂow ...

work page
[3]

Sign inversion through double outages In this section, we will discuss a highly surprising phe- nomenon that appears in the case of multiple interacting 5 0.41 0.16 0.59 0.410.48 0.36 0.16 0.36 0.16 0.23 a) 0.4 0.0 0.6 0.40.6 0.4 0.0 0.4 0.0 0.2 b) 0.52 0.13 0.48 0.520.39 0.48 0.13 0.48 0.13 0.0 c) 0.5 0.0 0.5 0.50.5 0.5 0.0 0.5 0.0 0.0 d) 0.0 0.1 0.2 0.3...

work page
[4]

Ll,oLo,kF (0) k +Ll,kLk,oF (0) o <−(Ll,kF (0) k +Ll,oF (0) o ),

work page
[5]

Energy System 2050 – a contribution of the research ﬁeld energy

Ll,oF (0) o > 0. To study this condition in detail, we assume that both the initial ﬂows on the failing links and the LODFs be- tween the failing links and the reference link are positive, F (0) o ,F (0) k ,L l,o,L l,k > 0, without loss of generality – this can always be accomplished by redeﬁning the orientation of one or both of the initial ﬂows. We can ...

work page
[6]

Atputharajah and T

A. Atputharajah and T. K. Saha, in 2009 Interna- tional Conference on Industrial and Information Systems (ICIIS) (2009) pp. 460–465

work page 2009
[7]

Pourbeik, P

P. Pourbeik, P. Kundur, and C. Taylor, IEEE Power and Energy Magazine 4, 22 (2006)

work page 2006

Show all 48 references

[8]

A. Y. Shih, P. Blinder, P. S. Tsai, B. Friedman, G. Stan- ley, P. D. Lyden, and D. Kleinfeld, Nature Neuroscience 16, 55 (2013)

2013
[9]

C. B. Schaﬀer, B. Friedman, N. Nishimura, L. F. Schroeder, P. S. Tsai, F. F. Ebner, P. D. Lyden, and D. Kleinfeld, PLoS Biology 4, e22 (2006)

2006
[10]

T. J. Brodribb, D. Bienaim´ e, and P. Marmottant, Pro- ceedings of the National Academy of Sciences 113, 4865 (2016)

2016
[11]

A. J. Wood, B. F. Wollenberg, and G. B. Shebl´ e, Power Generation, Operation and Control (John Wiley & Sons, New York, 2014)

2014
[12]

Weather-related power outages and electric system resiliency,

R. J. Campbell, “Weather-related power outages and electric system resiliency,” Congressional Research Service, https://fas.org/sgp/crs/misc/R42696.pdf (2012)

2012
[13]

Top- 004–2: Transmission operations,

North American Electric Reliability Corporation, “Top- 004–2: Transmission operations,” https://www.nerc. com/files/TOP-004-2.pdf (2007)

2007
[14]

L. A. Clarfeld, P. D. H. Hines, E. M. Hernandez, and M. J. Eppstein, arXiv:1901.03304 [physics] (2019), 1901.03304

2019 arXiv
[15]

K. S. Turitsyn and P. A. Kaplunovich, in 2013 46th Hawaii International Conference on System Sciences (2013) pp. 2161–2166

2013
[16]

P. A. Kaplunovich and K. S. Turitsyn, in 2013 IEEE Power Energy Society General Meeting (2013) pp. 1–5

2013
[18]

Weckesser and T

T. Weckesser and T. V. Cutsem, in 2018 Power Systems Computation Conference (PSCC) (2018) pp. 1–8

2018
[19]

Guler, G

T. Guler, G. Gross, and M. Liu, IEEE Transactions on Power Systems 22, 879 (2007)

2007
[20]

J. Guo, Y. Fu, Z. Li, and M. Shahidehpour, IEEE Trans. Power Syst. 24, 1633 (2009)

2009
[21]

Soltan and G

S. Soltan and G. Zussman, in 2016 IEEE Power and En- ergy Society General Meeting (PESGM) (2016) pp. 1–5

2016
[22]

Strake, F

J. Strake, F. Kaiser, F. Basiri, H. Ronellenﬁtsch, and D. Witthaut, New Journal of Physics 21, 053009 (2019)

2019
[23]

Purchala, L

K. Purchala, L. Meeus, D. V. Dommelen, and R. Bel- mans, in IEEE Power Engineering Society General Meet- ing (2005) pp. 454–459 Vol. 1

2005
[24]

Van Hertem, J

D. Van Hertem, J. Verboomen, K. Purchala, R. Belmans, and W. Kling, in The 8th IEEE International Conference on AC and DC Power Transmission (2006) pp. 58–62

2006
[25]

Bollob´ as, Modern graph theory , Graduate texts in mathematics No

B. Bollob´ as, Modern graph theory , Graduate texts in mathematics No. 184 (Springer, 1998)

1998
[26]

D ¨orﬂer, J

F. D ¨orﬂer, J. W. Simpson-Porco, and F. Bullo, Proceed- ings of the IEEE 106, 977 (2018)

2018
[27]

Kirchhoﬀ, Annalen der Physik 148, 497 (1847)

G. Kirchhoﬀ, Annalen der Physik 148, 497 (1847)
[28]

Hwang and R

N. Hwang and R. Houghtalen, Fundamentals of Hydraulic Engineering Systems (Prentice Hall, Upper Saddle River, NJ, 1996)

1996
[29]

D´ ıaz, J

S. D´ ıaz, J. Gonz´ alez, and R. M´ ınguez, Journal of Wa- ter Resources Planning and Management 142, 04015071 (2016)

2016
[30]

Katifori, G

E. Katifori, G. J. Sz ¨oll˝ osi, and M. O. Magnasco, Phys. Rev. Lett. 104, 048704 (2010)

2010
[31]

M. E. J. Newman, Networks – An Introduction (Oxford University Press, Oxford, 2010)

2010
[32]

Merris, Linear Algebra and its Applications 197-198, 143 (1994)

R. Merris, Linear Algebra and its Applications 197-198, 143 (1994)

1994
[33]

E. H. Moore, Bull. Amer. Math. Soc. 26, 394 (1920)

1920
[34]

Biggs, Bulletin of the London Mathematical Society 29, 641 (1997)

N. Biggs, Bulletin of the London Mathematical Society 29, 641 (1997)

1997
[35]

Labavic, R

D. Labavic, R. Suciu, H. Meyer-Ortmanns, and S. Ket- temann, Eur. Phys. J. ST 223, 2517 (2014)

2014
[36]

Witthaut, M

D. Witthaut, M. Rohden, X. Zhang, S. Hallerberg, and M. Timme, Phys. Rev. Lett. 116, 138701 (2016)

2016
[37]

Braess, Unternehmensforschung Operations Research 12, 258 (1969)

D. Braess, Unternehmensforschung Operations Research 12, 258 (1969)

1969
[38]

Witthaut and M

D. Witthaut and M. Timme, New Journal of Physics 14, 083036 (2012)

2012
[39]

A. E. Motter, Physical Review Letters 93, 098701 (2004)

2004
[40]

Witthaut and M

D. Witthaut and M. Timme, Phys. Rev. E 92, 032809 (2015)

2015
[41]

H ¨orsch, F

J. H ¨orsch, F. Hofmann, D. Schlachtberger, and T. Brown, Energy Strategy Reviews 22, 207 (2018)

2018
[42]

R. D. Zimmerman, C. E. Murillo-Sanchez, and R. J. Thomas, IEEE Trans. Power Syst. 26, 12 (2011)

2011
[43]

C. Josz, S. Fliscounakis, J. Maeght, and P. Panciatici, arXiv:1603.01533 [math] (2016), 1603.01533

2016 arXiv
[44]

Kettemann, Phys

S. Kettemann, Phys. Rev. E 94, 062311 (2016)

2016
[45]

Jung and S

D. Jung and S. Kettemann, Phys. Rev. E 94, 012307 (2016)

2016
[46]

M. A. Woodbury, Memorandum report 42, 106 (1950)

1950
[47]

Klein and M

D. Klein and M. Randic, Journal of Mathematical Chem- istry 12, 81 (1993)

1993
[48]

Xiao and I

W. Xiao and I. Gutman, Theoretical Chemistry Accounts 110, 284 (2003)

2003

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [17]

C. M. Davis and T. J. Overbye, IEEE Transactions on Power Systems 26, 1294 (2011)

work page 2011

[2] [1]

An elementary example of a network where this is happening is shown in Fig

Amplifying single outages To start with, we present a case where the naive overlay of two individual outages underestimates the collective eﬀects such that |∆Fl|≫| ∆F naive l | for some link l. An elementary example of a network where this is happening is shown in Fig. 1 where the topology is given by a network consisting ofN = 9 nodes and M = 12 links co...

work page

[3] [2]

A minimal example is shown in Figure 2

Eﬀect of individual outages exceeding simultaneous outage In addition to the eﬀect presented in the last section, it may also happen that an additional outage is beneﬁ- cial for the maximally loaded link. A minimal example is shown in Figure 2. In this setup, we have a unit inﬂow of power at the centre left node (coloured yellow) P = +1 and a unit outﬂow ...

work page

[4] [3]

Sign inversion through double outages In this section, we will discuss a highly surprising phe- nomenon that appears in the case of multiple interacting 5 0.41 0.16 0.59 0.410.48 0.36 0.16 0.36 0.16 0.23 a) 0.4 0.0 0.6 0.40.6 0.4 0.0 0.4 0.0 0.2 b) 0.52 0.13 0.48 0.520.39 0.48 0.13 0.48 0.13 0.0 c) 0.5 0.0 0.5 0.50.5 0.5 0.0 0.5 0.0 0.0 d) 0.0 0.1 0.2 0.3...

work page

[5] [4]

Ll,oLo,kF (0) k +Ll,kLk,oF (0) o <−(Ll,kF (0) k +Ll,oF (0) o ),

work page

[6] [5]

Energy System 2050 – a contribution of the research ﬁeld energy

Ll,oF (0) o > 0. To study this condition in detail, we assume that both the initial ﬂows on the failing links and the LODFs be- tween the failing links and the reference link are positive, F (0) o ,F (0) k ,L l,o,L l,k > 0, without loss of generality – this can always be accomplished by redeﬁning the orientation of one or both of the initial ﬂows. We can ...

work page

[7] [6]

Atputharajah and T

A. Atputharajah and T. K. Saha, in 2009 Interna- tional Conference on Industrial and Information Systems (ICIIS) (2009) pp. 460–465

work page 2009

[8] [7]

Pourbeik, P

P. Pourbeik, P. Kundur, and C. Taylor, IEEE Power and Energy Magazine 4, 22 (2006)

work page 2006