betaFeedback

WorksMathematics

Counterfactual Screening of Rare Secondary Overloads in Power Grids with a Stochastic Surrogate for Electromagnetic Transient Simulations

Abstract

Power-system operators continually ask what would happen if a component failed now. Single-phase-to-ground faults cleared by single-pole tripping and automatic reclosure usually have no lasting consequences. In rare cases, however, the transient excited during the open-phase interval drives distant lines or transformers into overload that persists after the faulted line is restored. We develop a counterfactual screening framework for these secondary overloads, combining a linear stochastic swing surrogate with cross-entropy importance sampling. The surrogate represents the open-phase corridor through an effective coupling factor and localized multiplicative noise, with both parameters learned from electromagnetic-transient (EMT) simulations rather than prescribed. Closed-form propagation of the mean and covariance permits deterministic calibration without Monte Carlo sampling. On the reduced WECC 240-bus system, a globally calibrated model substantially improves finite-time agreement with the EMT electromechanical state. The global models are calibrated at reclosure and tested on disjoint faulted lines under a longer open-phase interval. Line- and operating-state-specific models use additional reclosure data from the assessment lines and provide smaller further gains at the cost of more separate fits. The fitted coupling is lower than the phase-counting baseline, consistent with the direction anticipated by sequence-network analysis. Cross-entropy importance sampling directs computational effort toward contingencies associated with persistent overload and yields rare-event probability estimates together with a map linking initiating faults to the elements they endanger. By separating offline EMT calibration from repeated surrogate evaluation, the method supports probabilistic contingency screening and identifies where detailed high-fidelity assessment is most valuable.

Introduction

Most of the time, a transmission-system operator is watching a grid in which nothing alarming is happening. The job is nevertheless anticipatory: to know, before anything happens, which plausible events would put the system in danger. Steady-state N−1N-1 security assessment answers this question for post-contingency equilibria and runs continuously in control rooms [2, 3]. It does not answer it for the transient that connects the prefault and post-fault states, even though that transient can drive other lines and transformers above their limits for seconds. This paper develops a tool for that missing, dynamic part of the question. The analysis is counterfactual. It is not a response to an ongoing emergency but an estimate, made in advance and refreshed as conditions change, of what would happen if a given fault occurred now.

Guided by the operating experience of the Israeli transmission system operator (NOGA), we focus on the most common and least threatening transmission fault: a single-phase-to-ground fault on an extra-high-voltage line, cleared by single-pole tripping and automatic reclosure [4, 5]. Opening only the faulted pole keeps the two healthy phases in service. After a dead time of a fraction of a second the pole recloses and, in the overwhelming majority of cases, the event passes without consequence. Occasionally, however, the electromechanical swing excited during the open-phase interval travels through the network and pushes a remote critical element, typically a transformer, into an overload that persists after the faulted line has been restored. We call such a consequence a secondary contingency. It has three features that shape everything that follows. It is triggered by a frequent, benign primary event. It is rare. And when it happens it is costly, since a sustained transformer overload can lead to damage, protective tripping, or the first step of a cascade.

Screening for secondary contingencies is therefore a rare-event problem posed under real-time constraints, and three requirements pull against each other. The first is fidelity. The open-phase interval is an unbalanced three-phase transient, and electromagnetic-transient (EMT) simulation is the industry-accepted description of it [6, 7]. Even a parallel EMT code needs about half a minute of computation per simulated second on the reduced 240-bus model of the Western Electricity Coordinating Council (WECC) system [8]. The second is rarity: resolving a probability of order 10−310^{-3} to within ten percent by direct Monte Carlo requires of order 10510^{5} trajectories. The third is extent and latency: every line is a possible initiating fault, and the whole assessment must be repeated as load, dispatch, and renewable output drift. Direct EMT Monte Carlo meets the first requirement and misses the other two by orders of magnitude.

In previous work [1] we addressed rarity and extent. Following the swing-equation tradition of transient-stability analysis [9, 10, 11], we represented the post-fault dynamics by a linear stochastic swing model in which the faulted corridor has a reduced coupling during the open-phase interval and carries multiplicative noise local-

faulted line has been restored. We call such a consequence a secondary contingency. It has three features that shape everything that follows. It is triggered by a frequent, benign primary event. It is rare. And when it happens it is costly, since a sustained transformer overload can lead to damage, protective tripping, or the first step of a cascade.

Screening for secondary contingencies is therefore a rare-event problem posed under real-time constraints, and three requirements pull against each other. The first is fidelity. The open-phase interval is an unbalanced three-phase transient, and electromagnetic-transient (EMT) simulation is the industry-accepted description of it [6, 7]. Even a parallel EMT code needs about half a minute of computation per simulated second on the reduced 240-bus model of the Western Electricity Coordinating Council (WECC) system [8]. The second is rarity: resolving a probability of order 10−310^{-3} to within ten percent by direct Monte Carlo requires of order 10510^{5} trajectories. The third is extent and latency: every line is a possible initiating fault, and the whole assessment must be repeated as load, dispatch, and renewable output drift. Direct EMT Monte Carlo meets the first requirement and misses the other two by orders of magnitude.

In previous work [1] we addressed rarity and extent. Following the swing-equation tradition of transient-stability analysis [9, 10, 11], we represented the post-fault dynamics by a linear stochastic swing model in which the faulted corridor has a reduced coupling during the open-phase interval and carries multiplicative noise localized at its endpoints. Combined with cross-entropy (CE) importance sampling [12, 13, 14], this model screens many initiating single-phase faults in parallel and concentrates computational effort on the rare scenarios that produce long overloads. The weak point of that construction was its physics. The reduced coupling and the noise amplitude were postulated, not derived from or tested against a model that operators trust. Whether such a light model can stand in for the EMT description of the open-phase transient was left open.

This paper identifies the surrogate’s effective parameters from ParaEMT reference trajectories on the reduced WECC 240-bus system [7]. Calibration matches the electromechanical state at reclosure; subsequent assessment tests the globally fitted model on disjoint faulted lines under a longer open-phase interval. A hierarchy of global, line-specific, and line–operating-state-specific fits examines the value of additional calibration information. The comparison uses the phase-angle and frequency variables retained by the swing model, with parameter identification based on its deterministic moment equations.

The calibrated surrogate is then used to estimate overload probabilities and construct a contingency-to-vulnerable-element map: for each initiating fault, the probability that a monitored line or transformer accumulates more than a prescribed time in overload. This is the operator-facing output of the proposed background screening workflow in Figure 1.

Logic of the paper

Figure 1. Logic of the paper. (a) The counterfactual question. A frequent and usually benign single-phase fault on line ee is cleared by single-pole tripping and reclosed after the open-phase interval [0,Top][0,T_{\mathrm{op}}]. The electromechanical swing it excites can drive a remote critical element into persistent overload within the screening horizon ThT_{\mathrm{h}}. (b) The fast physics model. The open-phase coupling α\alpha and the noise amplitude σ\sigma of a stochastic swing surrogate are learned offline from EMT (ParaEMT) trajectories by closed-form moment matching. The learned coupling is compared with phase-counting and sequence-network estimates. (c) Online, cross-entropy importance sampling over all initiating faults turns surrogate trajectories into the contingency-to-vulnerable-element map Qij(e;γ)Q_{ij}(e;\gamma). The intended tool runs in the background and alerts the operator only when an entry exceeds a threshold. Blue labels mark elements carried over from Ref. [1], orange marks what is new here, and red marks secondary overload and the operator alert. The network and map are schematic; quantitative results are in Sec. IV.

Section II develops the model, calibration, and probability estimation; Section III specifies the experiments. Sections IV and V present the findings and their interpretation, with scope and limitations collected in § 2.4 A.

Methods

Table I collects the notation and time conventions used in the main text and appendices. The method separates representation from estimation: EMT simulations calibrate the reduced dynamics, and importance sampling estimates overload probabilities under those fixed dynamics. Calibration, state assessment, and screening use distinct time settings. The state variables, switching rule, and observation convention remain the same; the fitted physical parameters are held fixed when the screening sampler varies fault location and open-phase duration.

SymbolMeaning
G=(V,E); nE=∣E∣\mathcal{G} = (\mathcal{V}, \mathcal{E});\ n_{\mathcal{E}} = |\mathcal{E}|Network graph, bus and line sets, and number of initiating lines; nE=329n_{\mathcal{E}} = 329 in screening.
e=(i,j); a∈N0(e)e = (i,j);\ a \in\mathcal{N}_0(e)Faulted line and a noise-bearing endpoint; N0(e)={i,j}\mathcal{N}_0(e) = \{i,j\}.
z; χz;\ \chiPre-fault operating state and its dimensionless uniform injection multiplier.
Xt=(ωt,θt)TX_t = (\omega_t,\theta_t)^{\mathsf{T}}Reduced state: frequency deviations and phase angles in the fixed reference.
M; ea\mathbf{M};\ \mathbf{e}_aInertia matrix and the canonical vector for endpoint aa.
A, Ae(α); Ae,IA,\ A_e(\alpha);\ A_{e,\mathrm{I}}Nominal drift, open-phase drift, and equivalent Itô drift.
Ge,a(σ); WtaG_{e,a}(\sigma);\ W_t^aEndpoint-local diffusion matrix and its independent Wiener driver.
α\alphaOpen-phase coupling factor: βe↦αβe\beta_e \mapsto\alpha\beta_e during [0,Top][0,T_{\mathrm{op}}] only.
σ\sigmaCommon amplitude of endpoint-localized multiplicative noise; [σ]=s−1/2[\sigma] = \mathrm{s}^{-1/2}.
Θ\ThetaFitted parameter collection.
M0,…,M4M_0,\ldots,M_4Calibration-hierarchy model names, not inertia matrices.
t=0t = 0Instant the faulted pole opens; fault clearing is treated as instantaneous.
TopT_{\mathrm{op}}Open-phase duration (auto-reclosing dead time).
TevT_{\mathrm{ev}}State-evaluation time: 0.8 s0.8\ \mathrm{s} at reclosure for every fitted hierarchy level; 2.5 s2.5\ \mathrm{s} after a 1.5-s1.5\text{-s} open-phase interval for their common assessment.
ThT_{\mathrm{h}}Upper limit of the overload integrals, fixed at 10 s10\ \mathrm{s} in screening; distinct from the calibration target time.
H; ye,zEMT; Te,zEMTH;\ y_{e,z}^{\mathrm{EMT}};\ T_{e,z}^{\mathrm{EMT}}Observation convention, projected EMT state, and finite-time reference state-transfer map.
μΘ, ΣΘ\mu_{\Theta},\ \Sigma_{\Theta}Surrogate mean and state covariance.
We,H(q); ℓe,a(q)W_{e,H}^{(q)};\ \ell_{e,a}^{(q)}Observation-space neighborhood mask and its binary bus-selection entries.
Σ^eEMT\widehat{\Sigma}_{e}^{\mathrm{EMT}}Covariance target formed across calibration operating states at fixed line.
Σ^Θ,e,z(0), Σ‾Θ,e,z(0)\widehat{\Sigma}_{\Theta,e,z}^{(0)},\ \overline{\Sigma}_{\Theta,e,z}^{(0)}Endpoint-masked predicted covariance and its diagonally regularized version.
Jmap(q); R; ζcov,ζregJ_{\mathrm{map}}^{(q)};\ \mathcal{R};\ \zeta_{\mathrm{cov}},\zeta_{\mathrm{reg}}Calibration objective, scalar regularizer, and objective weights.
C; λ∈spec⁡(Ke,Θ)C;\ \lambda\in\operatorname{spec}(K_{e,\Theta})Matrix argument of the second-moment operator and an eigenvalue of its matrix representation.
ρ=X0/X1\rho= X_0/X_1Zero- to positive-sequence reactance ratio in Appendix B.
Sij; S; γ≥0S_{ij};\ S;\ \gamma\ge0Accumulated line exposure, summed network exposure, and the prescribed strict-exceedance threshold.
Q(γ), Qij(γ)Q(\gamma),\ Q_{ij}(\gamma)Global and line-specific probabilities under the nominal scenario law.
Q(e;γ), Qij(e;γ)Q(e;\gamma),\ Q_{ij}(e;\gamma)Corresponding probabilities conditional on initiating fault ee.
τ; τ0; r\tau;\ \tau_0;\ rSampled open-phase duration, nominal exponential mean, and shared proposal mean, all in seconds.
πe; ϕe; ν\pi_e;\ \phi_e;\ \nuNominal line probability, proposal line probability, and proposal-parameter vector.
NCE, K, NN_{\mathrm{CE}},\ K,\ NSamples per adaptation iteration, number of iterations, and fresh final-batch size.
ρCE; κ; nelite\rho_{\mathrm{CE}};\ \kappa;\ n_{\mathrm{elite}}CE elite fraction, previous-parameter retention factor, and elite count.
ℓ; k; Iℓ\ell;\ k;\ \mathcal{I}_{\ell}Adaptation iteration, sample index, and selected elite index set.
bk(ℓ); sℓeliteb_k^{(\ell)};\ s_{\ell}^{\mathrm{elite}}Binary elite-membership indicator and smallest selected elite score.
wk; w~k; Ykw_k;\ \widetilde{w}_k;\ Y_kRaw likelihood ratio, normalized diagnostic weight, and weighted event contribution.
Q^; s^; RelSE⁡\widehat{Q};\ \widehat{s};\ \operatorname{RelSE}Probability estimate, estimated standard error, and relative sampling standard error.
RSE⁡map\operatorname{RSE}_{\mathrm{map}}Relative finite-time state error; distinct from sampling uncertainty.
P, E, exp⁡\mathbb{P},\ \mathbb{E},\ \expProbability, expectation, and the exponential function.

Table I. Principal notation and time conventions. Physical calibration parameters, scenario-law parameters, and sampling diagnostics have distinct roles. The same time symbol may take different values in different experiments.

Contingency and Reduced Dynamics

The modeled contingency is a single-phase-to-ground fault cleared by single-pole tripping and followed by successful automatic reclosure. The reduced description starts at pole opening and retains the open-phase and post-reclosure electromechanical response.

Fault inception and clearing. Protection opens the faulted pole of line e=(i,j)e = (i,j). Fault inception and breaker operation occur over tens of milliseconds, below the electromechanical time scale retained by the surrogate. Clearing is therefore treated as instantaneous, and t=0t = 0 denotes the instant of pole opening. The reduction begins from the corresponding electromechanical state and does not resolve the preceding fault or breaker transient.

Open-phase interval [0,Top][0,T_{\mathrm{op}}]. One phase is de-energized, but the other two remain in service, so the line continues to transmit power. The interval permits de-ionization of the secondary arc sustained by capacitive and inductive coupling from the energized phases; premature reclosure can restrike the fault. In protection terminology, TopT_{\mathrm{op}} is the auto-reclosing dead time [15]. We use open-phase duration to distinguish it from the evaluation and screening times. During this interval, the effective positive-sequence coupling changes from βe\beta_e to αβe\alpha\beta_e, with α<1\alpha<1.

Reclosure and post-reclosure interval (Top,Th](T_{\mathrm{op}},T_{\mathrm{h}}]. At t=Topt=T_{\mathrm{op}}, the open pole is reclosed and the nominal coupling βe\beta_e is restored [4]. The machines do not instantaneously return to equilibrium: their ensuing swing can cause transient overloads on other elements. The transferred electromechanical state therefore matters even after the network configuration is restored.

Evaluation time TevT_{\mathrm{ev}} and screening horizon ThT_{\mathrm{h}}. The surrogate state is compared with its EMT reference at the experiment-specific time TevT_{\mathrm{ev}}; overload duration is accumulated up to ThT_{\mathrm{h}}. State assessment at or after reclosure has 0<Top≤Tev0<T_{\mathrm{op}}\leq T_{\mathrm{ev}}. Calibration uses the boundary case Top=Tev=0.8 sT_{\mathrm{op}}=T_{\mathrm{ev}}=0.8\ \mathrm{s}, so the target is the state attained at reclosure. The main hierarchy comparison instead uses Top=1.5 sT_{\mathrm{op}}=1.5\ \mathrm{s} and Tev=2.5 sT_{\mathrm{ev}}=2.5\ \mathrm{s}, one second after reclosure. Screening integrates exposure through Th=10 sT_{\mathrm{h}}=10\ \mathrm{s} and does not fix every sampled duration at a calibration or test value. A full-trajectory illustration may also end at 10 s10\ \mathrm{s} without making that endpoint the calibration time. Varying TevT_{\mathrm{ev}} changes the observation time; varying TopT_{\mathrm{op}} changes the switching event. Table I records these roles. Throughout, fault clearing initiates the open-phase interval at t=0t=0; t=Topt=T_{\mathrm{op}} denotes reclosure, not a second fault-clearing event.

Why a stochastic reduced model? The swing variables retain the phase and frequency response used by the screening calculation, but not the three-phase electrical and device states resolved by ParaEMT. Their influence on the retained variables is represented by an effective coupling and a linear stochastic correction, whose parameters are identified from the EMT response.

The two fitted terms address different features of the reduction. The coupling factor represents the change in transfer through the open-phase corridor. The diffusion matrices act only at its endpoints, so the model introduces the correction at the location of the disturbance and retains the network dynamics that transmit its effects elsewhere. Linearity in the state gives the closed moment equations used for calibration.

The stochastic correction represents unresolved variability within the prescribed operating-state family. Its empirical assessment combines the open-phase comparison in Figure 3 with the deterministic–stochastic comparison in the calibration hierarchy. Section IIB specifies the across-operating-state covariance target used to identify its amplitude.

Evolution of phase angles for the removed line 9#4

Figure 3. ParaEMT–surrogate comparison during the open-phase interval. Phase-angle trajectories for the illustrative large-excursion 9#4 contingency over 0≤t≤Top=0.5 s0 \le t \le T_{\mathrm{op}} = 0.5\ \mathrm{s}. Solid curves show ParaEMT; dashed curves show one realization of (2) with (α∗,σ∗)=(0.516,1.5)(\alpha^{*},\sigma^{*})=(0.516,1.5), not an ensemble mean or confidence band. The vertical markers identify pole opening and reclosure. Figure 11 continues this same example to 10 s and displays the corridor separation ∣θ9−θ4∣\lvert\theta_{9}-\theta_{4}\rvert. The full-trajectory endpoint is not the 0.8-s calibration target or the 2.5-s hierarchy-test time.

For a faulted line e=(i,j)e=(i,j) and a pre-fault operating state zz, the surrogate evolves the electromechanical state through the open-phase and post-reclosure intervals of Sec. II A.

The reduced electromechanical state is

Xt=[ωtθt],(1)X_t=\begin{bmatrix}\omega_t\\\theta_t\end{bmatrix}, \tag*{(1)}

where ωt\omega_t and θt\theta_t are the frequency-deviation and phase vectors over the bus set V\mathcal{V}. The formulation preserves the reduced WEC network graph and transmission-line connectivity rather than Kron-reducing the system to generator internal nodes. Thus, the graph G\mathcal{G} used for the neighborhoods in (C1)–(C2) is the graph used to construct the swing dynamics. Phases are absolute coordinates relative to a fixed reference bus, not deviations from the pre-fault phase vector; X0(z)X_0(z) denotes the equilibrium state for operating condition zz in that reference convention.

The symbols α\alpha and σ\sigma denote the coupling and common endpoint-noise amplitude used for a specified scenario. Their sharing across scenarios is determined by the calibration hierarchy in Sec. III A, under Comparison hierarchy, and Figure 2. They are global constants in M1_1/M2_2, line-specific values (αe,σe)(\alpha_e,\sigma_e) in M3_3, and line–operating-state–specific values (αe,z,σe,z)(\alpha_{e,z},\sigma_{e,z}) in M4_4. The baseline M0_0 fixes (α,σ)=(2/3,0)(\alpha,\sigma)=(2/3,0), and M1_1 has σ=0\sigma=0. In every case, one amplitude is shared by the two endpoints of a given faulted line. We suppress these calibration indices in the dynamical equations; the scenario indices on AeA_e and Ge,aG_{e,a} still identify the modified corridor and the endpoint where noise acts.

Calibration hierarchy and common comparison protocol

Figure 2. Calibration hierarchy and common comparison protocol. (A) M0M_0 fixes (α,σ)=(2/3,0)(\alpha,\sigma)=(2/3,0); M1M_1 fits one global deterministic coupling and M2M_2 one global stochastic pair. M3M_3 fits 25 line-specific pairs and M4M_4 fits 1,150 line–operating-state-specific pairs. All fits use Top=Tev=0.8T_{\mathrm{op}}=T_{\mathrm{ev}}=0.8 s. The specialized models additionally use reference data from the 25 assessment lines, which are excluded from the global fits. The arrows describe fitting specificity and effort, not measured runtime ratios. (B) The common protocol fixes the WECC model, observations, and assessment conditions. Figure 4 scores every model on 25 lines and 46 operating states at (Top,Tev)=(1.5,2.5)(T_{\mathrm{op}},T_{\mathrm{ev}})=(1.5,2.5) s. Those line identities are unseen by M1/M2M_1/M_2, but have separate 0.8-s calibration records for M3/M4M_3/M_4. The 25×4625 \times46 count identifies a reference family, not one pool split among all models. The sequence-informed deterministic reference only motivates the coupling correction; it is not implemented or tested in the hierarchy. The line and operating-state indices are ee and zz, respectively.

Across the open-phase and post-reclosure intervals, the dynamics are

{dXt=(Ae(α)Xt+P(z)) dt+∑a∈N0(e)Ge,a(σ)Xt∘dWta,0≤t≤Top,X˙t=AXt+P(z):Top≤t≤Tev,(2)\left\{ \begin{aligned} dX_t&=(A_e(\alpha)X_t+P(z))\,dt+\sum_{a\in\mathcal{N}_0(e)}G_{e,a}(\sigma)X_t\circ dW_t^a,\quad0\leq t\leq T_{\mathrm{op}},\\ \dot{X}_t&=AX_t+P(z):T_{\mathrm{op}}\leq t\leq T_{\mathrm{ev}}, \end{aligned} \right. \tag*{(2)}

where

N0(e)={i,j}(3)\mathcal{N}_0(e)=\{i,j\} \tag*{(3)}

is the endpoint set of the faulted line, equivalently its zero-radius graph neighborhood. During [0,Top][0,T_{\mathrm{op}}], the multiplicative forcing represents variability localized to the affected corridor. The equation is interpreted in the Stratonovich sense, as indicated by ∘dWta\circ dW_t^a, and the WtaW_t^a are independent Wiener processes. The vector P(z)P(z) denotes the power-injection forcing for the pre-fault operating state. At reclosure, the stochastic forcing is removed and the nominal drift AA is restored; the attained electromechanical state supplies the initial condition for the post-reclosure interval. The linear description therefore yields a finite-time map composed across the switching event, rather than one time-invariant faulted-network propagator.

The contingency-dependent matrix Ae(α)A_e(\alpha) is constructed from the swing dynamics by modifying the coupling of the affected line,

βeop=αβe,(4)\beta_e^{\mathrm{op}}=\alpha\beta_e, \tag*{(4)}

where βe\beta_e is its nominal coupling. The modification applies only during the open-phase interval. Reclosure restores βe\beta_e and A=Ae(1)A=A_e(1), so the calibrated parameters refer to the same piecewise network configuration used in trajectory evaluation.

The parameter α\alpha is the effective positive-sequence coupling factor of the open-phase corridor. Counting the two intact phases gives the physically transparent baseline α=23\alpha=\frac{2}{3}, but not an exact sequence-network result. Negative- and zero-sequence contributions modify the positive-sequence transfer under a one-conductor-open condition, providing a physical basis for a coupling below the phase-counting value. The value used by the surrogate is identified from the electromechanical response rather than imposed by phase counting; its comparison with the sequence-network estimate is reported in Sec. IV B.

The noise acts on the phase coordinates of the two endpoints. Let ea\mathbf{e}_a be the aath canonical vector in R∣V∣\mathbb{R}^{|V|}. For a∈N0(e)a\in\mathcal{N}_0(e),

Ge,a(σ)=[000σM−1eaeaT].(5)G_{e,a}(\sigma)= \begin{bmatrix} 0 & 0\\ 0 & \sigma\mathbf{M}^{-1}\mathbf{e}_a\mathbf{e}_a^{\mathsf{T}} \end{bmatrix}. \tag*{(5)}

Here M\mathbf{M} is the inertia matrix of the retained swing model. Each diffusion matrix has a single endpoint contribution in the phase block. Since dWtdW_t has units s1/2\mathrm{s}^{1/2}, the amplitude has units [σ]=s−1/2[\sigma]=\mathrm{s}^{-1/2}; this is not a dimensionless perturbation. Because the multiplicative term uses phase coordinates, its specification is tied to the fixed phase reference introduced above. The same reference must be retained in the EMT projection, initial states, and surrogate evaluation; common-mode cancellation in the overload observable does not make the stochastic state equation reference-invariant.

The two parameters are identified jointly. The coupling α\alpha controls the deterministic open-phase interaction, while σ\sigma controls localized variability and contributes to the equivalent Itô drift through 12∑a∈N0(e)Ge,a2\frac{1}{2}\sum_{a\in\mathcal{N}_0(e)}G_{e,a}^2. The baseline α=23\alpha=\frac{2}{3} provides a warm start, not a constraint on the fit. The stochastic correction can consequently influence both the transferred mean state and its covariance, as made explicit in Sec. II B.

The sequence-network construction in Appendix B provides a physical motivation for the coupling correction. The hierarchy and the status of the sequence-informed control are specified in Sec. III A.

EMT Reference and Moment-Based Calibration

Shared observation space. ParaEMT provides the reference trajectories from which the reduced finite-time map is identified [7]. Its three-phase network and device model supplies the high-fidelity response, which is projected onto the observation space shared with the swing surrogate. Appendix A identifies the reference implementation and case inputs.

Let e=(i,j)e=(i,j) denote the faulted line and zz the pre-fault operating state. ParaEMT generates a high-dimensional trajectory

xe,zEMT(t),0≤t≤Tev,(6)x_{e,z}^{\mathrm{EMT}}(t), \qquad0\leq t\leq T_{\mathrm{ev}}, \tag*{(6)}

where TevT_{\mathrm{ev}} is the evaluation time defined in Sec. II A. Its electromechanical observation is

ye,zEMT(t)=Hxe,zEMT(t),0≤t≤Tev,(7)y_{e,z}^{\mathrm{EMT}}(t)=H x_{e,z}^{\mathrm{EMT}}(t), \qquad0\leq t\leq T_{\mathrm{ev}}, \tag*{(7)}

where HH extracts the phase and frequency variables represented by the surrogate. The same observation convention is used when evaluating the reduced state, so the mismatch concerns corresponding physical variables.

The projection retains the electromechanical response through which the reference model’s fast electromagnetic and device dynamics influence the calibration target. The selected observations determine the dimension of the comparison.

The finite-time reference map is

Te,zEMT:X0(z)⟼ye,zEMT(Tev),(8)\mathcal{T}_{e,z}^{\mathrm{EMT}}:X_0(z)\longmapsto y_{e,z}^{\mathrm{EMT}}(T_{\mathrm{ev}}), \tag*{(8)}

where X0(z)X_0(z) is the reduced electromechanical state associated with the pre-fault equilibrium. Fixing the initial state, switching convention, and observation time makes the map a direct comparison of the reference and reduced responses to the same scenario. Its approximation is the calibration objective. In the reported fit this target is evaluated at Tev=Top=0.8 sT_{\mathrm{ev}}=T_{\mathrm{op}}=0.8\,\mathrm{s}; the subsequent assessment propagates the fitted model to its separately specified observation time.

Deterministic moment propagation. Linearity with multiplicative noise gives a closed deterministic system for the first two moments after conversion to the equivalent Itô form. For parameters Θ\Theta, define the mean

μΘ(t)=E[Xt],(9)\mu_{\Theta}(t)=\mathbb{E}[X_t], \tag*{(9)}

and covariance

ΣΘ(t)=E[(Xt−μΘ(t))(Xt−μΘ(t))T].(10)\Sigma_{\Theta}(t)=\mathbb{E}\left[(X_t-\mu_{\Theta}(t))(X_t-\mu_{\Theta}(t))^{\mathsf{T}}\right]. \tag*{(10)}

The equivalent Itô drift during the open-phase interval is

Ae,I(α,σ)=Ae(α)+12∑a∈N0(e)Ge,a2(σ).(11)A_{e,\mathrm{I}}(\alpha,\sigma)=A_e(\alpha)+\frac{1}{2}\sum_{a\in\mathcal{N}_0(e)}G_{e,a}^2(\sigma). \tag*{(11)}

The second term is the noise-induced drift associated with the Stratonovich convention [16, 17, 18]. Consequently, changing σ\sigma affects the mean as well as the covariance; its role is not restricted to adding fluctuation around an unchanged deterministic trajectory. The mean satisfies

μ˙Θ(t)={[Ae(α)+12∑a∈N0(e)Ge,a2(σ)]μΘ(t)+P(z),0≤t≤Top,AμΘ(t)+P(z),Top≤t≤Tev.(12)\dot{\mu}_{\Theta}(t)= \begin{cases} \left[A_e(\alpha)+\frac{1}{2}\sum_{a\in\mathcal{N}_0(e)}G_{e,a}^2(\sigma)\right]\mu_{\Theta}(t)+P(z), & 0\leq t\leq T_{\mathrm{op}},\\ A\mu_{\Theta}(t)+P(z), & T_{\mathrm{op}}\leq t\leq T_{\mathrm{ev}}. \end{cases} \tag*{(12)}

and the covariance satisfies

Σ˙Θ(t)={Ae,IΣΘ+ΣΘAe,IT+∑a∈N0(e)Ge,a(ΣΘ+μΘμΘT)Ge,aT,0≤t≤Top,AΣΘ+ΣΘAT,Top≤t≤Tev,(13)\dot{\Sigma}_{\Theta}(t)= \begin{cases} A_{e,I}\Sigma_{\Theta}+\Sigma_{\Theta}A_{e,I}^{\mathsf{T}}+\sum_{a\in\mathbb{N}_0(e)}G_{e,a}\left(\Sigma_{\Theta}+\mu_{\Theta}\mu_{\Theta}^{\mathsf{T}}\right)G_{e,a}^{\mathsf{T}}, & 0\le t\le T_{\mathrm{op}},\\ A\Sigma_{\Theta}+\Sigma_{\Theta}A^{\mathsf{T}}, & T_{\mathrm{op}}\le t\le T_{\mathrm{ev}}, \end{cases} \tag*{(13)}

The moments at reclosure initialize their deterministic continuation under AA. In calibration, evaluation stops at reclosure; in the post-reclosure tests, the moments are continued to the specified TevT_{\mathrm{ev}}. This retains the switching event explicitly and avoids Monte Carlo sampling when evaluating candidate parameters [16, 17].

The effective coupling and noise-induced drift jointly determine the mean-based metric in (15). This supplies the mechanism by which M2_2 can improve the finite-time map beyond the deterministic fit M1_1: α\alpha adjusts the corridor interaction, while the Stratonovich correction permits σ\sigma to modify the transferred mean. Their joint interpretation allows part of the effective correction to be represented by noise-induced drift rather than by α\alpha alone. Appendix C 2 records the second-moment operator and the spectral diagnostic for the reduced stochastic dynamics.

Endpoint-local objective. The primary global fit uses only the two endpoints of the faulted line. The mask We,H(q)W_{e,H}^{(q)} selects their observed phase and frequency components, and Σ^eEMT\widehat{\Sigma}_{e}^{\mathrm{EMT}} denotes the scatter of the projected EMT state across the training operating-state family at fixed fault ee, evaluated at Tev=Top=0.8 sT_{\mathrm{ev}}=T_{\mathrm{op}}=0.8\ \mathrm{s} for the training runs. The joint objective is

Jmap(q)(Θ)≐∑(e,z)∈Dtrain∥We,H(q)(HμΘ(Tev;e,z)−ye,zEMT(Tev))∥22+ζcov∑(e,z)∈Dtrain∥We,H(q)(HΣΘ(Tev;e,z)HT−Σ^eEMT)We,H(q)∥F2+ζregR(Θ).(14)J_{\mathrm{map}}^{(q)}(\Theta)\doteq \sum_{(e,z)\in\mathcal{D}_{\mathrm{train}}} \left\|W_{e,H}^{(q)}\left(H\mu_{\Theta}(T_{\mathrm{ev}};e,z)-y_{e,z}^{\mathrm{EMT}}(T_{\mathrm{ev}})\right)\right\|_{2}^{2} +\zeta_{\mathrm{cov}} \sum_{(e,z)\in\mathcal{D}_{\mathrm{train}}} \left\|W_{e,H}^{(q)}\left(H\Sigma_{\Theta}(T_{\mathrm{ev}};e,z)H^{\mathsf{T}}-\widehat{\Sigma}_{e}^{\mathrm{EMT}}\right)W_{e,H}^{(q)}\right\|_{F}^{2} +\zeta_{\mathrm{reg}}R(\Theta). \tag*{(14)}

The first term matches the mean transferred state, the second matches its covariance target, and R(Θ)R(\Theta) regularizes the predicted covariance. The weights are ζcov=1.0\zeta_{\mathrm{cov}}=1.0 and ζreg=0.5\zeta_{\mathrm{reg}}=0.5; the absolute log-determinant regularizer uses the diagonal shift ϵ=0.01\epsilon=0.01. Their explicit definitions and mask construction are given in Appendix C 1.

ParaEMT supplies one deterministic trajectory for each (e,z)(e,z), and Σ^EMT\widehat{\Sigma}^{\mathrm{EMT}} is computed across the training operating states. The fitted σ\sigma therefore represents operating-point variability unresolved by the global reduced description. Only training states from the global calibration-line set enter this target; observations from the global testing lines are not used to identify the shared parameters of M1_1 or M2_2. Models M3_3 and M4_4 instead receive additional calibration observations from those lines at the same 0.8-s reclosure target, as specified in Sec. III A. The later observations used to score Figure 4 are not fitting targets for any model. The restriction q=0q=0 applies to parameter identification, not to the propagation of the disturbance: network-wide responses remain available for validation and screening.

Distribution of Map Errors Across the Calibration Hierarchy

Figure 4. Distribution of RSEmapRSE_{\mathrm{map}} across the WECC 240-bus model hierarchy at Top=1.5 sT_{\mathrm{op}} = 1.5\,\mathrm{s} and Tev=2.5 sT_{\mathrm{ev}} = 2.5\,\mathrm{s}. All models are scored on the same 25 lines and 46 uniformly scaled operating states. Marker shapes identify lines; colors identify injection changes. M1/M2M_1/M_2 were fitted on 25 disjoint calibration lines. M3M_3 instead uses 25 line-specific pairs fitted across the 46 operating states, and M4M_4 uses 1,150 line–state-specific pairs, obtained from additional reference runs on the assessment lines. Every fit uses Top=Tev=0.8 sT_{\mathrm{op}} = T_{\mathrm{ev}} = 0.8\,\mathrm{s} and is held fixed here. Thus, the M3/M4M_3/M_4 results measure duration and observation-time transfer after case-specific calibration, not transfer to unseen lines. Their smaller additional gains beyond M2M_2 require more separate fits. Unlike Figs. 9–10, this comparison includes injection scaling.

Validation metric. The principal validation metric compares the surrogate mean and the projected EMT state at the same evaluation time,

RSE⁡map(e,z)=∥HμΘ(Tev;e,z)−ye,zEMT(Tev)∥2∥ye,zEMT(Tev)∥2.(15)\operatorname{RSE}_{\mathrm{map}}(e,z)= \frac{\left\|H\mu_{\Theta}(T_{\mathrm{ev}};e,z)-y_{e,z}^{\mathrm{EMT}}(T_{\mathrm{ev}})\right\|_{2}} {\left\|y_{e,z}^{\mathrm{EMT}}(T_{\mathrm{ev}})\right\|_{2}}. \tag*{(15)}

The numerator measures discrepancy in the observed electromechanical state; the denominator expresses that discrepancy relative to the reference state’s norm. Both the phase-angle and frequency components selected by HH enter this metric; it is not a phase-only error unless HH selects only phases. Within a given comparison, all hierarchy levels use the same variables, reference convention, switching duration, and observation time. This common scoring protocol does not imply identical fitting data: the global and scenario-specific models use the allocations in Sec. III A. All fitted coefficients are held fixed when the 0.8-s calibration setting is replaced by the longer-duration assessment setting.

In Figure 4, the finite-time state metric is evaluated at Tev=2.5 sT_{\mathrm{ev}}=2.5\ \mathrm{s} after reclosure at Top=1.5 sT_{\mathrm{op}}=1.5\ \mathrm{s}, with the coefficients fitted at the 0.8-s calibration target held fixed. Noise affects HμΘH\mu_{\Theta} through the mean dynamics in Sec. II B. Covariance information enters parameter identification through (14); validation then scores the projected mean state.

Overload Observables and Conditional Risks

Let θ(t)\theta(t) be the phase-angle vector of a sampled surrogate trajectory from (2). For a monitored line {i,j}∈Em\{i,j\}\in\mathcal{E}_{m}, the linearized transfer βij[θi(t)−θj(t)]\beta_{ij}[\theta_i(t)-\theta_j(t)] is compared with its allowable power-flow threshold pˉij\bar{p}_{ij}. The line-specific overload indicator is

Sij(θ0→Th)=∫0Th1{∣βij[θi(t)−θj(t)]∣>pˉij} dt,(16)S_{ij}(\theta_0\to T_h)=\int_{0}^{T_h}\mathbb{1}\left\{\left|\beta_{ij}\left[\theta_i(t)-\theta_j(t)\right]\right|>\bar{p}_{ij}\right\}\,\mathrm{d}t, \tag*{(16)}

where 1{⋅}\mathbb{1}\{\cdot\} equals one when the stated limit is exceeded and zero otherwise. Thus, SijS_{ij} is a cumulative duration: separated periods of overload contribute to the same integral. The monitored element need not be the faulted line. This distinction allows a local initiating contingency to be assessed through the persistent stress it produces elsewhere in the network.

The system-level indicator sums exposure over the monitored set,

S(θ0→Th)=∑{i,j}∈EmSij(θ0→Th),(17)S(\theta_0\to T_h)=\sum_{\{i,j\}\in\mathcal{E}_{m}}S_{ij}(\theta_0\to T_h), \tag*{(17)}

where Em⊆E\mathcal{E}_{m}\subseteq\mathcal{E}. Simultaneous violations on different elements contribute separately, so SS measures accumulated line-overload time rather than the duration for which at least one line is overloaded. It can consequently exceed ThT_h. Neither indicator weights a violation by the amount of its excess above the threshold; their purpose is to distinguish persistent excursions from brief crossings that may have comparable peak phase separations.

Equation (16) uses a linearized-flow screening approximation. Common angular displacements cancel from the phase differences. Figure 11 and Appendix D 4 examine the corridor separations that determine its small-angle interpretation; the scope of the indicator is set out in Sec. V A.

Full illustrative transient and corridor-separation diagnostic

Figure 11. Full illustrative transient and corridor-separation diagnostic. The large-excursion 9#49\#4 example uses Top=0.5T_{\mathrm{op}} = 0.5 s and is observed through 10 s. The in-figure label Tev=10T_{\mathrm{ev}} = 10 s denotes the endpoint of this diagnostic, not the calibration or hierarchy-test time. Panel (a) shows phase-angle trajectories through reclosure; panel (b) shows ∣θ9−θ4∣|\theta_9-\theta_4|. Solid curves show ParaEMT; dashed curves show one surrogate realization with (α∗,σ∗)=(0.516,1.5)(\alpha^\ast,\sigma^\ast)=(0.516,1.5). Figure 3 displays the open-phase portion of the same run. This example illustrates oscillatory post-reclosure behavior but does not use the 1.5-s switching duration of Figure 4. Excursions beyond the shaded 0.5-rad diagnostic band limit the small-angle interpretation of (16); the band is not a protection setting.

For a prescribed overload-duration threshold γ≥0\gamma\ge0, define

Qij(γ)=P[Sij(θ0→Th)>γ].(18)Q_{ij}(\gamma)=\mathbb{P}\left[S_{ij}(\theta_0\to T_h)>\gamma\right]. \tag*{(18)}

This is the probability that monitored line {i,j}\{i,j\} accumulates more than γ\gamma seconds above its flow limit. The corresponding global probability is

Q(γ)=P[S(θ0→Th)>γ].(19)Q(\gamma)=\mathbb{P}\left[S(\theta_0\to T_h)>\gamma\right]. \tag*{(19)}

Here, the event concerns accumulated exposure over the monitored network, not the union of line-specific events at the same threshold. In particular, Q(γ)Q(\gamma) is not obtained by adding the Qij(γ)Q_{ij}(\gamma). The strict comparison also defines Q(0)Q(0) as the probability of any positive overload, rather than the certain event associated with a nonnegative indicator.

The probability law specifies both the initiating scenario and the surrogate uncertainty conditional on it. A scenario comprises a faulted line ee and its open-phase duration TopT_{\mathrm{op}}; the stochastic drivers then determine a realization of the electromechanical response. The nominal scenario law remains fixed when the sampling algorithm changes. The conditional risks defined below answer a different question from averaging over possible fault locations.

Conditional risk scores. For an initiating fault ee and threshold γ\gamma, define

Q(e;γ)=P[S(θ0→Th(e))>γ],(20)Q(e;\gamma)=\mathbb{P}\left[S\left(\theta_{0\to T_h}^{(e)}\right)>\gamma\right], \tag*{(20)}

The probability averages over the nominal open-phase duration and surrogate drivers conditional on ee. Its strict comparison follows (19). Sorting Q(e;γ)Q(e;\gamma) gives a ranking by the probability of exceeding a specified total line-overload time. Because SS includes the monitored set, the score can include exposure on the faulted element as well as secondary exposure elsewhere.

For an individual monitored line, define

Qij(e;γ)=P[Sij(θ0→Th(e))>γ],{i,j}∈Em,(21)Q_{ij}(e;\gamma)=\mathbb{P}\left[S_{ij}\left(\theta_{0\to T_h}^{(e)}\right)>\gamma\right],\qquad\{i,j\}\in\mathcal{E}_m, \tag*{(21)}

The strict comparison is the convention in (18) and (E18). The collection

{Qij(e;γ)}{i,j}∈Em(22)\left\{Q_{ij}(e;\gamma)\right\}_{\{i,j\}\in\mathcal{E}_m} \tag*{(22)}

forms one row of the contingency-to-vulnerable-element map. Entries for {i,j}≠e\{i,j\}\ne e identify secondary exposure.

Cross-Entropy Importance Sampling

As the overload threshold increases, most nominal samples may contribute zero to the event indicator. Importance sampling reallocates the trajectory budget without changing either the event or the calibrated physical model. The stochastic driving law is the same under the nominal and proposal simulations; only fault location and duration are resampled.

Direct Monte Carlo draws the faulted line and open-phase duration from the nominal joint law pZp_Z. CE adaptation instead fits a proposal to high-score contingencies and then freezes the proposal for a separate estimation batch. The WECC implementation retains both the learned categorical line probabilities and a shared duration parameter. With TkT_k denoting the sampled open-phase duration, the estimator is

Q^(γ)=1N∑k=1NI{S(θ0→Th(k))>γ}pZ(ek,Tk)qν∗(ek,Tk).(23)\widehat{Q}(\gamma)=\frac{1}{N}\sum_{k=1}^{N}\mathbb{I}\left\{S\left(\theta_{0\to T_h}^{(k)}\right)>\gamma\right\}\frac{p_Z(e_k,T_k)}{q_{\nu_*}(e_k,T_k)}. \tag*{(23)}

The ratio pZ/qν∗p_Z/q_{\nu_*} corrects both line selection and duration sampling. Positive proposal support on the nominal support, an unchanged stochastic driving law, and a fresh estimation batch preserve the nominal target. The joint law and likelihood ratio are specified in Appendix E [12, 13, 14]. The elite updates and Algorithm 1 are given in the same appendix. The duration parameter rr is an exponential mean, measured in seconds; the nominal mean τ0=0.1 s\tau_0=0.1\,\mathrm{s} remains fixed during proposal adaptation.

During probability estimation, the physical parameters remain fixed while CE adaptation changes the scenario proposal. Appendix E2 records the calibrated parameters and their use in the estimator. Each sampled scenario is evaluated with the specified dynamics and overload criterion before likelihood weighting is applied.

WECC Study

The experiments use the reduced WECC 240-bus system to assess state transfer across fault locations and operating conditions, then evaluate probabilistic screening with the calibrated global model.

Reference Case and Calibration Comparisons

The reduced WECC 240-bus transmission system includes conventional and renewable generation, loads, transmission corridors, and dynamic devices. The ParaEMT implementation supplies the network and dynamic-model inputs [8, 19]; the case description and one-line diagram appear in Appendix A1. The topology and dynamic parameters are fixed throughout the calibration and testing experiments. Only the specified fault, open-phase duration, observation time, and pre-fault operating condition change.

Global calibration data and target. The calibration set for M₁ and M₂ contains 25 faulted lines, each evaluated at 46 pre-fault operating states, giving 25 × 46 = 1,150 training scenarios. The operating states are generated by uniformly scaling the nominal active-power injection vector,

P(zχ)=χP(0),χ∈[0.75,1.20],(24)P(z_\chi) = \chi P^{(0)}, \qquad\chi\in[0.75, 1.20], \tag*{(24)}

with the steady-state equilibrium recomputed for each scale. The nominal case is χ=1\chi= 1; the family spans injection changes from −25% to +20% while retaining the electrical and dynamic-model parameters [20, 21].

Every calibration run, including the additional runs used for M₃ and M₄, uses Top=Tev=0.8T_{\mathrm{op}} = T_{\mathrm{ev}} = 0.8 s. The objective in (14) therefore matches the projected ParaEMT state at the end of the open-phase interval. This state supplies the initial condition for subsequent restored-network evolution. The endpoint-local mask q=0q = 0 selects the observations used for fitting; the covariance target is constructed from the 46 training operating states at the same 0.8-s snapshot. No later testing time enters the global calibration objective.

Line-disjoint testing of the global models. Let Etrain\mathcal{E}_{\mathrm{train}} and Etest\mathcal{E}_{\mathrm{test}} denote the calibration and testing fault-line sets for M₁ and M₂. Each contains 25 distinct lines, and Etrain∩Etest=∅\mathcal{E}_{\mathrm{train}} \cap\mathcal{E}_{\mathrm{test}} = \varnothing. The exact line identifiers and their training/test assignments are specified in the project repository [22]. The split is by faulted-line identity, not a random division of the 1,150 training scenarios. Testing uses a separate 25 × 46 family on the same prescribed operating scales. The global-model records are therefore partitioned as

D=Dtrain∪Dtest,Dtrain∩Dtest=∅.(25)\mathcal{D} = \mathcal{D}_{\mathrm{train}} \cup\mathcal{D}_{\mathrm{test}}, \qquad\mathcal{D}_{\mathrm{train}} \cap\mathcal{D}_{\mathrm{test}} = \varnothing. \tag*{(25)}

For the global fits M₁ and M₂, only Dtrain\mathcal{D}_{\mathrm{train}} enters parameter identification. The fitted coefficients are then fixed when assessing the 25 previously unused faulted lines. The shared operating scales make this a test of transfer to different fault locations within the stated operating-state family, not a test on unseen networks or on an independently held-out range of injection scales.

Scenario-specific calibration. The reference collection is enlarged for M₃ and M₄ by including reclosure observations for the 25 lines in Etest\mathcal{E}_{\mathrm{test}} at all 46 operating states, again with Top=Tev=0.8T_{\mathrm{op}} = T_{\mathrm{ev}} = 0.8 s. These are additional calibration runs, not the Top=1.5T_{\mathrm{op}} = 1.5 s, Tev=2.5T_{\mathrm{ev}} = 2.5 s observations used for assessment. Thus, a line reserved for testing M₁ and M₂ can supply separate fitting observations for the scenario-specific models without entering the global fit.

Model M₃ identifies 25 parameter pairs (αℓ,σℓ)(\alpha_\ell,\sigma_\ell), one for each assessment line, using its 46 calibration operating states jointly. Each pair is then shared across those operating states. Model M₄ identifies 25×46=1,15025 \times46 = 1,150 pairs (αℓ,z,σℓ,z)(\alpha_{\ell,z},\sigma_{\ell,z}), one for each line and operating state. The two specialized models use the same additional reference family but differ in parameter sharing. Their coefficients are fitted at the 0.8-s target and held fixed during the subsequent assessment.

Post-reclosure assessment. Figure 4 uses Top=1.5T_{\mathrm{op}} = 1.5 s and Tev=2.5T_{\mathrm{ev}} = 2.5 s for the 25 testing lines across the 46 scaled operating states. This is a deliberately demanding transfer test: the open-phase interval is longer than in calibration, and the observation is made one second after reclosure, while the response can still exhibit substantial oscillations. It measures how well the calibrated description reproduces the subsequent electromechanical state, rather than how closely it fits the training snapshot. These settings are not asserted to be a worst case over all faults or operating conditions. The full transient in Figure 11(a) illustrates oscillatory post-reclosure behavior under its own, shorter 0.5-s open-phase interval; it is not the trajectory used to define the timing of Figure 4.

Within every reference–surrogate pair, the initial operating state, faulted line, switching times, phase reference, and observation operator are matched. All five models are scored on the same network-wide observations, with no refitting to the assessment response. For M₁ and M₂, the assessment combines transfer to unseen lines with transfer in duration and observation time; for M₃ and M₄, it measures the latter transfer after case-specific calibration. Subsequent probability estimates use only M₂, with its global pair held fixed.

Separate temporal checks. Appendices D2 and D3 vary TevT_{\mathrm{ev}} and TopT_{\mathrm{op}}, respectively, with unscaled injections and the global parameter pair held fixed. These experiments separate observation-time sensitivity from switching-duration sensitivity.

Comparison hierarchy. Figure 2(A) defines the five model levels. The baseline M₀ fixes α=2/3\alpha= 2/3 and σ=0\sigma= 0; M₁ fits a shared deterministic coupling, and M₂ jointly fits the shared coupling and stochastic amplitude. Models M₃ and M₄ add line and operating-state dependence. The comparison tests increasing parameter specificity within this model family.

Panel (B) also shows the sequence-informed deterministic reference αseq≃0.560\alpha_{\mathrm{seq}} \simeq0.560, σ=0\sigma= 0. This reference motivates the coupling correction; it is neither implemented nor tested as part of the calibration hierarchy. The common protocol means matching all model comparisons within each stage; it does not mean using the same fault catalogue or time settings for training and testing.

Probabilistic Screening Setup

CEM and MC use the same calibrated surrogate, nominal contingency law, screening horizon Th=10T_{\mathrm{h}}=10 s, and trajectory-evaluation interval Δt=0.01\Delta t=0.01 s on the reduced WECC system [23]. The tested global thresholds are γ∈{0,0.5,5,10}\gamma\in\{0,0.5,5,10\} s in (19). The global parameters remain fixed during sampling as specified in (E19).

The nominal law selects among nE=329n_{\mathrm{E}}=329 modeled faulted lines uniformly and uses an exponential open-phase duration with mean τ0=0.1\tau_0=0.1 s; see (E2). The probabilities are conditional on a fault occurring within this line set. This 329-line screening catalogue is broader than either of the 25-line calibration and testing sets. Applying the global pair throughout that catalogue extrapolates it beyond the EMT-assessed line sets. The nominal mean 0.1 s specifies the probability law; it is distinct from the fixed 0.8-s calibration duration and 1.5-s hierarchy-test duration. CE adaptation uses 250 trajectories per iteration for 20 iterations, with the 25 largest scores forming the elite set. These 5,000 adaptation evaluations are separate from the final estimation batch and are included in the reported total CEM runtimes.

The duration-resolved assessment in Figs. 12a and 12b uses N=150,000N=150{,}000 surrogate simulations. It varies the open-phase duration rather than marginalizing it under the nominal duration law. Its line-specific probabilities must therefore be distinguished from the durationmarginalized map in Figure 5. The event definitions themselves are those of § 1.1 C; they are not redefined for the WECC results.

Contingency-to-vulnerable-element map for the WECC 240-bus assessment

Figure 5. Contingency-to-vulnerable-element map for the WECC 240-bus assessment at γ=1\gamma= 1 s over Th=10T_{\mathrm{h}} = 10 s. The x-axis shows critical transformer-connected monitored lines; the y-axis identifies the ten displayed initiating faults. The comma-separated axis label Xn,mXn,m denotes the same fault identifier as Xn#m in the experiment records. The row order is not a scalar-risk ranking. Each cell represents Qij(e;γ)=P[Sij(θe)>γ]Q_{ij}(e;\gamma) = P[S_{ij}(\theta^{e}) > \gamma], the probability of more than one second of accumulated overload on element {i,j}\{i,j\} conditional on fault ee, with open-phase duration and surrogate uncertainty marginalized under the specified nominal law. Entries with {i,j}≠e\{i,j\} \ne e identify secondary exposure. The continuous colorbar includes labeled values through 0.015; its colors are not the duration-resolved warning classes.

Results

The results address three questions: what the surrogate retains during the open-phase interval, how much parameter flexibility is needed for state transfer, and what is learned by resolving overload risk across initiating faults and monitored elements. Sampling diagnostics are reported separately from model accuracy.

Open-Phase Trajectory Comparison

Figure 3 compares ParaEMT and one realization of the calibrated stochastic surrogate for the illustrative large-excursion 9#4 contingency during [0,Top][0,T_{\mathrm{op}}], with Top=0.5 sT_{\mathrm{op}}=0.5\,\mathrm{s}. Both calculations show an uneven phase response: many traces remain in a relatively narrow range, while others develop larger excursions before reclosure. The surrogate follows this overall separation of responses, with differences in fluctuation amplitudes. The comparison makes the retained response and the remaining trajectory discrepancies visible during the interval in which the modified coupling and stochastic forcing act.

The full trajectory and corridor-separation diagnostic in Figure 11 continue this example through reclosure. The hierarchy below provides the quantitative comparison of projected mean states.

Calibration Accuracy and Parameter Dependence

Joint optimization of the global stochastic model M2M_2 under the protocol in Sec. III A gives

(α∗,σ∗)=(0.516,1.5).(26)(\alpha^{*},\sigma^{*})=(0.516,1.5). \tag*{(26)}

At Top=1.5 sT_{\mathrm{op}}=1.5\,\mathrm{s} and Tev=2.5 sT_{\mathrm{ev}}=2.5\,\mathrm{s}, the fixed global pair gives M2M_2 errors of approximately 0.0968–0.15 across the 25 testing lines and 46 uniformly scaled operating states, with mean 0.098. The one-sided distribution places the mean near its lower end.

Figure 4 compares the hierarchy under this common assessment. The fixed deterministic baseline M0M_0 has errors of approximately 0.24–0.4, while fitting the deterministic coupling in M1M_1 brings most cases into the 0.17–0.29 range. The dominant reduction occurs from M1M_1 to M2M_2. Because RSEmap_{\mathrm{map}} scores the projected mean, this comparison shows the contribution of the noise-induced drift to state transfer within the tested hierarchy.

Models M3M_3 and M4M_4 produce narrower error distributions and smaller further reductions using the additional case-specific calibration records specified in Sec. III A. Substantial variation in their fitted parameters, reported in (D2) and Figure 8, therefore accompanies relatively modest accuracy gains. This is the main calibration trade-off: the global pair captures most of the improvement, while the specialized fits require additional reference observations and more independent parameter fits. The subsequent screening calculations retain M2M_2 on this basis.

Operating-point sensitivity of the line-dependent calibration

Figure 8. Scenario-specific M4_4 parameters fitted at Top=Tev=0.8 sT_{\mathrm{op}}=T_{\mathrm{ev}}=0.8\,\mathrm{s} for 25 assessment lines and 46 operating states, relative to the global pair (α∗,σ∗)=(0.516,1.5)(\alpha^{*},\sigma^{*})=(0.516,1.5). Each of the 1,150 pairs is held fixed in the later hierarchy assessment. Maximum reported deviations are Δα=0.294\Delta\alpha=0.294 and Δσ=1.07\Delta\sigma=1.07. Ellipses summarize linewise parameter distributions, not confidence regions.

Coupling and temporal transfer. The fitted coupling lies below the phase-counting baseline, in the same direction as the representative sequence-network estimate αseq≃0.560\alpha_{\mathrm{seq}}\simeq0.560. The residual difference distinguishes the line-level sequence calculation from the joint fit to the finite-time network response.

The unscaled-injection studies in Appendices D2 and D3 separate two temporal effects. After a fixed reclosure, discrepancies continue to increase even though the stochastic forcing has ceased: restoring the nominal operator does not erase the mismatch in the attained state. Varying the open-phase duration instead changes how long the modified dynamics act before that state is transferred. These trends complement the hierarchy comparison by showing how error depends on time as well as on the fitted parameterization.

Overload Probabilities and Sampling Efficiency

We evaluate the probabilities defined in (19) using the WECC setup in Sec. III B. Table II reports the probability estimates, final estimation-batch sizes, and total runtimes. Including the adaptation budget specified in Sec. III B, CEM uses 30,000–160,000 surrogate evaluations, compared with 125,000–300,000 for MC. Tables III and IV in Appendix E3 provide the sampling diagnostics and stage-by-stage costs.

Both estimators resolve the decrease in exceedance probability from approximately 0.056 at γ=0\gamma=0 to 0.0077–0.0079 at γ=5 s\gamma=5\,\mathrm{s} and 0.0027–0.0029 at γ=10 s\gamma=10\,\mathrm{s}. The absolute differences between their point estimates range

γ\gamma [s]MethodNfinalN_{\mathrm{final}}Q^\hat{Q}Total [s]
0CEM2.500×1042.500 \times10^{4}5.600×10−25.600 \times10^{-2}4.462×1034.462 \times10^{3}
MC1.250×1051.250 \times10^{5}5.606×10−25.606 \times10^{-2}2.210×1042.210 \times10^{4}
0.5CEM4.000×1044.000 \times10^{4}2.628×10−22.628 \times10^{-2}7.177×1037.177 \times10^{3}
MC1.750×1051.750 \times10^{5}2.636×10−22.636 \times10^{-2}3.185×1043.185 \times10^{4}
5.0CEM1.050×1051.050 \times10^{5}7.743×10−37.743 \times10^{-3}1.970×1041.970 \times10^{4}
MC2.500×1052.500 \times10^{5}7.872×10−37.872 \times10^{-3}4.536×1044.536 \times10^{4}
10.0CEM1.550×1051.550 \times10^{5}2.723×10−32.723 \times10^{-3}2.828×1042.828 \times10^{4}
MC3.000×1053.000 \times10^{5}2.900×10−32.900 \times10^{-3}5.547×1045.547 \times10^{4}

Table 2. Probability estimates and total runtimes for CEM and Monte Carlo (MC) [24]. Here, NfinalN_{\mathrm{final}} is the final estimation-batch size. CEM total runtimes include both the 5,000 adaptation evaluations and final probability estimation.

from 6.0×10−56.0 \times10^{-5} to 1.77×10−41.77 \times10^{-4}. The estimates distinguish any positive overload from larger accumulated network exposure and reach the 10−310^{-3} probability range.

At every reported threshold, CEM uses a smaller final batch and less total runtime than MC. The MC-to-CEM runtime ratios are 4.952, 4.438, 2.303, and 1.961 for increasing γ\gamma. The advantage at these budgets is therefore largest for the lowest threshold and narrows as the target event becomes rarer. At γ=10 s\gamma=10\ \mathrm{s}, CEM reaches an estimate of 0.002723 in 28,281.6 s including adaptation, compared with 0.002900 in 55,468.7 s for MC.

The contribution-variance analysis in Appendix § 2.3.2 favors CEM at γ=0.5\gamma=0.5 and 5 s, and MC at 0 and 10 s. Adaptation selects the largest global scores, so its per-sample benefit varies with the final exceedance threshold.

Contingency-to-Vulnerable-Element Map

The conditional probabilities defined in § 1.1 C resolve the location of the consequences associated with each initiating fault. They are conditional on that fault and the specified operating state, rather than estimates of the fault’s occurrence frequency.

Figure 5 presents the conditional map at γ=1 s\gamma=1\ \mathrm{s} for ten initiating faults and the monitored elements shown. Rows are initiating faults, columns are monitored elements, and each cell is Qij(e;1 s)Q_{ij}(e;1\ \mathrm{s}). The row order is not a descending ranking of the scalar scores Q(e;γ)Q(e;\gamma). The figure instead resolves the fault–element pairs underlying the screening output.

The colorbar has labeled ticks at 0, 0.005, 0.010, and 0.015. Larger entries occur in different columns for different faults. The X2#227 contingency, for example, has a prominent secondary-risk entry on {61,65}\{61,65\}. A prominent entry for X112#155 instead lies on the faulted line {112,155}\{112,155\} itself. These contrasting cases show why a global score alone is insufficient: it does not identify the monitored element or distinguish self-exposure from secondary consequences.

Interpretation of the screening thresholds. The duration-resolved assessment in Figure 12b uses γ=1 s\gamma=1\ \mathrm{s} and probability boundaries 0.025 and 0.04: safe denotes probabilities below 0.025, warning values from 0.025 through 0.04, and emergency values above 0.04. The heatmap marginalizes open-phase duration under the nominal law, whereas the curves in Appendix § 2.12 condition on duration. Its continuous color scale is independent of those classes. § 2.4 A states the interpretation of these screening thresholds.

The duration-resolved curves supply complementary information: an element can have substantial exposure under a long open-phase interval while retaining a small duration-marginalized probability under the nominal law.

Computational Cost and Reuse of Calibration

Replacing repeated EMT simulations with reduced trajectories addresses the largest per-realization computational burden. A direct EMT ensemble requires a high-fidelity trajectory for each sampled contingency and open-phase duration. The reported mean surrogate evaluation time is approximately 0.19 s; the published WECC benchmark reports 29 s on one core and 28 s on eight cores for a 1-s ParaEMT simulation [8]. The benchmark settings are given in Table V. § 2.4 A states the scope of this computational comparison.

CoresHorizon [s]EMT step [μ\mus]Runtime [s]
115029
815028

Table V. Published ParaEMT runtime on the reduced WECC 240-bus system [8]. These external benchmark settings differ from the present surrogate screening configuration.

Reference generation and parameter identification are offline costs. The screening ensemble reuses the calibrated M2M_2 model for every adaptation and estimation trajectory. This reuse is shared by CEM and MC; their sampling costs and precision are compared in Sec. IV C. The additional calibration records used by M3/M4M_3/M_4 belong to the hierarchy assessment, not to the repeated screening calculation.

Discussion

The advance over the earlier screening construction is the identification of its effective parameters from EMT and the assessment of their transferred response. The WECC hierarchy supports a compact global parameterization, while the specialized fits show the value of additional case-specific information. The temporal studies further identify how switching duration and subsequent propagation affect the state mismatch.

Deterministic moment propagation separates parameter identification from trajectory sampling. The calibrated model can then be reused across counterfactual scenarios, with CE adaptation directing evaluations toward large accumulated exposure. The resulting fault–element map identifies where detailed high-fidelity follow-up is most valuable, connecting EMT reference information to probabilistic screening.

Scope and limitations

The assessment covers the reduced WECC network, the prescribed uniform injection-scale family, and the timing protocols in Sec. III A. The fitted coefficients are effective parameters for those conditions. The stochastic correction is an empirical representation of unresolved variability, rather than a white-noise model derived from the EMT equations. The hierarchy establishes its value within the tested family; distinguishing this mechanism from other reduced descriptions with comparable fitting flexibility requires additional comparisons. Applying the global pair to the broader 329-line screening catalogue extends beyond the EMT-assessed line sets. Changes in topology, device models, or operating conditions require further assessment of the calibration domain.

The metric RSEmap\mathrm{RSE}_{\mathrm{map}} evaluates the projected mean state at a specified time. It does not establish agreement of the complete waveform, the path-dependent overload duration, or the associated tail probability. Errors that change whether S>γS > \gamma can affect a rare-event probability substantially, whereas state discrepancies away from the threshold may have little effect on that event. CEM and MC evaluate the same surrogate: their agreement assesses probability estimation under that model, not independent agreement with EMT. Increasing the sampling budget reduces statistical uncertainty without correcting model mismatch. EMT-informed overload, probability, and ranking comparisons under the same scenario law remain the corresponding physical-validation tasks. The state-error ordering of the hierarchy is therefore not an established ordering of probability accuracy. The second-moment spectral diagnostic likewise concerns the reduced stochastic dynamics.

The overload indicator uses a linearized power–angle relation and measures accumulated threshold exposure. The large corridor separations in Figure 11 limit its small-angle interpretation during that transient. Outside this regime, the indicator serves as a reduced-model severity measure for high-fidelity follow-up, rather than an exact AC power-flow or thermal-damage calculation. The duration threshold, probability-class boundaries, and angle-diagnostic band are illustrative; they are not NERC or WECC criteria or relay settings. In particular, safe denotes the low-probability screening class, not a guarantee of security. The resulting map prioritizes assessment rather than prescribing protection actions.

The reported probabilities also depend on the specified exponential open-phase-duration law with mean 0.1 s and on successful reclosure. This law is a duration model for the study, not an empirical distribution of protection outcomes. A failed reclosure requires a different switching sequence. Protection-specific branches and broader operating-state families are extensions of the present formulation.

The runtime comparison uses the reported sample budgets and their associated statistical uncertainties, rather than a matched-precision benchmark. Its contribution-variance advantage is event-dependent. Calibration fit counts describe fitting requirements, not measured wall-clock multipliers. The published ParaEMT timings provide external computational context: the model fidelity, numerical settings, and potential hardware differences preclude a controlled speedup claim or an EMT-ensemble runtime extrapolation from those values. The present screening implementation takes hours per threshold and remains a proof of concept. Complete update latency must include adaptation and final estimation at the required precision. Parallel trajectory evaluation and reuse of a CE proposal across nearby operating states remain options to test before operational deployment.

Data Availability

The project code and WECC calibration materials are available through the PowerGridREserach repository [22] and its calibration-data directory [25]. The model-specific calibration and assessment allocations are specified in Sec. III A. Historical Israel-case materials are stored separately. ParaEMT is available from its upstream repository [26]. and moves in the direction of the EMT-calibrated value α∗=0.516\alpha^* = 0.516. Relative to the 23\frac{2}{3} baseline, this sequence correction accounts for approximately 71% of the observed reduction toward the fitted value. The residual difference, αseq−α∗≃0.044\alpha_{\mathrm{seq}}-\alpha^* \simeq0.044, shows that the line-level sequence equivalent alone does not determine the calibrated coupling.

Figure 7 examines the sensitivity of this result to the unavailable line-specific WECC zero-sequence parameters. Writing ρ=X0/X1\rho= X_0/X_1, neglecting resistance for this sensitivity calculation, and using Z2≃Z1Z_2 \simeq Z_1 gives

Sequence-network prediction of effective open-phase coupling

Figure 7. Sequence-network coupling versus ρ=X0/X1\rho= X_0/X_1, with Z2≃Z1Z_2 \simeq Z_1. Shading marks the representative overhead-line range 2.0–4.6. The representative 230-kV ratio ρ=3.49\rho= 3.49 gives αseq≃0.563\alpha_{\mathrm{seq}} \simeq0.563 (complex-impedance estimate: 0.560); reproducing the fitted α∗=0.516\alpha^* = 0.516 by this correction alone requires ρ≃15.1\rho\simeq15.1.

αseq(ρ)=1+ρ1+2ρ\alpha_{\mathrm{seq}}(\rho) = \frac{1+\rho}{1+2\rho}

Representative overhead-line values ρ≃2.0\rho\simeq2.0–4.6 give αseq≃0.600\alpha_{\mathrm{seq}} \simeq0.600–0.549. The representative value ρ=3.49\rho= 3.49 gives αseq≃0.563\alpha_{\mathrm{seq}} \simeq0.563, consistent with the complex-impedance estimate 0.560 above. In contrast, reproducing α∗=0.516\alpha^* = 0.516 through the line-level sequence correction alone would require ρ≃15.1\rho\simeq15.1, well outside the representative range considered here.

Local Calibration and Moment Diagnostics

Graph-Local Calibration

Calibration uses the electromechanical response at the endpoints of the faulted line, where the stochastic forcing in (2) acts [30]. The endpoint restriction applies to parameter identification, not to propagation through the network.

Let G=(V,E)\mathcal{G} = (\mathcal{V}, \mathcal{E}) denote the structure-preserving network graph introduced in § 1.1 A. For a bus a∈Va \in\mathcal{V} and a faulted line e=(i,j)e = (i,j), define the graph distance from the bus to the line by

d(a,e)=min⁡{d(a,i),d(a,j)},(27)d(a,e) = \min\{d(a,i),d(a,j)\}, \tag*{(27)}

where d(⋅,⋅)d(\cdot,\cdot) denotes the shortest-path distance on G\mathcal{G}. For a locality radius qq, define

Nq(e)={a∈V:d(a,e)≤q}.(28)\mathcal{N}_q(e) = \{a \in\mathcal{V}: d(a,e) \le q\}. \tag*{(28)}

The present analysis uses the endpoint-local case q=0q = 0. The general neighborhood in (C2) then reduces to the endpoint set introduced in (3):

N0(e)={i,j}.(29)\mathcal{N}_0(e) = \{i,j\}. \tag*{(29)}

Only these two buses enter the calibration objective. Broader neighborhoods, including q=1q = 1, are not used in the reported fits.

Let We(q)W_e^{(q)} denote the corresponding diagonal bus mask,

We(q)=diag⁡(ιe,a(q))a∈V,ιe,a(q)={1,a∈Nq(e),0,otherwise.(30)W_e^{(q)} = \operatorname{diag}\left(\iota_{e,a}^{(q)}\right)_{a\in\mathcal{V}}, \qquad \iota_{e,a}^{(q)} = \begin{cases} 1, & a \in\mathcal{N}_q(e),\\ 0, & \text{otherwise}. \end{cases} \tag*{(30)}

When both phase and frequency variables are included in the observation operator, the same bus mask is applied to both components; the resulting observation-space mask is denoted by We,H(q)W_{e,H}^{(q)}. In all results reported here, this mask is evaluated at q=0q = 0.

ParaEMT produces one deterministic trajectory for each faulted line and pre-fault operating state (e,z)(e,z). Consequently, the covariance target used in the calibration is not an ensemble covariance obtained from repeated EMT realizations. Instead, it measures the scatter of the projected finite-time EMT state across the training operating-state family at fixed faulted line. For the global fit, e∈Etraine \in\mathcal{E}_{\mathrm{train}} and all observations in this target use Top=Tev=0.8T_{\mathrm{op}} = T_{\mathrm{ev}} = 0.8 s. Let Ztrain\mathcal{Z}_{\mathrm{train}} denote the 46 scaled training operating states specified in Sec. III A, and define

y‾eEMT=1∣Ztrain∣∑z∈Ztrainye,zEMT(Tev).(31)\overline{y}_e^{\mathrm{EMT}} = \frac{1}{|\mathcal{Z}_{\mathrm{train}}|}\sum_{z\in\mathcal{Z}_{\mathrm{train}}} y_{e,z}^{\mathrm{EMT}}(T_{\mathrm{ev}}). \tag*{(31)}

The corresponding sample covariance is

Σ^eEMT=1∣Ztrain∣−1∑z∈Ztrain(ye,zEMT(Tev)−y‾eEMT)(ye,zEMT(Tev)−y‾eEMT)T.(32)\widehat{\Sigma}_e^{\mathrm{EMT}} = \frac{1}{|\mathcal{Z}_{\mathrm{train}}|-1} \sum_{z\in\mathcal{Z}_{\mathrm{train}}} \left(y_{e,z}^{\mathrm{EMT}}(T_{\mathrm{ev}})-\overline{y}_e^{\mathrm{EMT}}\right) \left(y_{e,z}^{\mathrm{EMT}}(T_{\mathrm{ev}})-\overline{y}_e^{\mathrm{EMT}}\right)^{\mathsf{T}}. \tag*{(32)}

The target Σ^eEMT\widehat{\Sigma}_e^{\mathrm{EMT}} measures variation across training operating states, not numerical randomness in ParaEMT. It constrains the variability represented by σ\sigma in the global surrogate. The fitted parameter distributions are reported in Appendix D.1.

The objective is given in (14) in the main text. The calibration weights are fixed throughout the analysis as

ζcov=1.0,ζreg=0.5.(33)\zeta_{\mathrm{cov}} = 1.0,\qquad\zeta_{\mathrm{reg}} = 0.5. \tag*{(33)}

The mean term in (14) constrains the transferred endpoint response, while the covariance term uses the operating-state scatter in (C6). The parameters are fitted jointly because σ\sigma also enters the mean dynamics. The regularization acts on the endpoint-masked predicted covariance

Σ~Θ,e,z(0)=We,H(0)HΣΘ(Tev;e,z)HTWe,H(0).(34)\widetilde{\Sigma}^{(0)}_{\Theta,e,z}=W^{(0)}_{e,H}H\Sigma_{\Theta}(T_{\mathrm{ev}};e,z)H^{\mathsf{T}}W^{(0)}_{e,H}. \tag*{(34)}

Because the q=0q=0 mask introduces zero directions outside the endpoint subspace, the determinant is evaluated after adding the fixed diagonal regularization

Σ~Θ,e,z(0)=ΣΘ,e,z(0)+ϵI,ϵ=0.01.(35)\widetilde{\Sigma}^{(0)}_{\Theta,e,z}=\Sigma^{(0)}_{\Theta,e,z}+\epsilon I,\qquad\epsilon=0.01. \tag*{(35)}

The regularization term is then

R(Θ)=∑(e,z)∈Dtrain∣log⁡det⁡(Σ~Θ,e,z(0))∣.(36)\mathcal{R}(\Theta)=\sum_{(e,z)\in\mathcal{D}_{\mathrm{train}}}\left|\log\det\left(\widetilde{\Sigma}^{(0)}_{\Theta,e,z}\right)\right|. \tag*{(36)}

The diagonal shift makes the covariance entering the determinant nonsingular. The absolute log-determinant penalizes departures from the reference determinant scale in either direction; it does not bound the condition number or control each covariance eigenvalue separately. This penalty supplements, rather than replaces, the operating-state covariance target in (C6).

The global fit uses only training-line endpoint observations at the 0.8-s reclosure snapshot. Models M3\mathrm{M}_3 and M4\mathrm{M}_4 additionally use 0.8-s observations from the assessment lines, with one pair per line or per line–operating-state combination, respectively. Those additional observations do not enter the globally shared fit. The coupled swing dynamics propagate the disturbance throughout G\mathcal{G}. Network-wide comparison at (Top,Tev)=(1.5,2.5) s(T_{\mathrm{op}},T_{\mathrm{ev}})=(1.5,2.5)\,\mathrm{s} then assesses all models at a duration and observation time not used for identification. The faulted lines themselves are held out only for the global models.

Mean-Square Stability Analysis

The second-moment diagnostic characterizes the reduced stochastic dynamics. Multiplicative noise can destabilize that moment even when the deterministic drift is stable [31]. During 0≤t≤Top0\le t\le T_{\mathrm{op}}, the homogeneous operator, acting on a covariance-matrix argument CC, is

Ke,Θ(C):=Ae,IC+CAe,IT+∑a∈N0(e)Ge,aCGe,aT.(37)\mathcal{K}_{e,\Theta}(C)\mathrel{:=}A_{e,I}C+CA_{e,I}^{\mathsf{T}}+\sum_{a\in\mathcal{N}_0(e)}G_{e,a}CG_{e,a}^{\mathsf{T}}. \tag*{(37)}

The term ∑a∈N0(e)Ge,aμΘμGe,aT\sum_{a\in\mathcal{N}_0(e)}G_{e,a}\mu_{\Theta}^{\mu}G_{e,a}^{\mathsf{T}} in (13) is a nonhomogeneous forcing term and therefore does not enter the spectral stability criterion.

The corresponding mean-square stability margin is defined by

ηms(e,z;Θ):=−max⁡λ∈spec⁡(Ke,Θ)Re⁡(λ).(38)\eta_{\mathrm{ms}}(e,z;\Theta)\mathrel{:=}-\max_{\lambda\in\operatorname{spec}(K_{e,\Theta})}\operatorname{Re}(\lambda). \tag*{(38)}

Thus, ηms>0\eta_{\mathrm{ms}}>0 implies exponential mean-square stability of the homogeneous open-phase stochastic dynamics, whereas ηms<0\eta_{\mathrm{ms}}<0 indicates a noise-induced second-moment instability. For a scalar SDE dX=aX dt+σX dW\mathrm{d}X=aX\,\mathrm{d}t+\sigma X\,\mathrm{d}W, this criterion reduces to 2a+σ2<02a+\sigma^2<0, illustrating why deterministic stability alone is insufficient in the presence of multiplicative noise [32].

We evaluate (C13) at the globally calibrated WECC parameters (α∗,σ∗)=(0.516,1.5)(\alpha^{*},\sigma^{*})=(0.516,1.5) using the same inertia, damping, network-Laplacian, and fault-local noise structure employed by the surrogate. The calculation removes the global-phase mode before evaluating the reported physical spectrum. For the limiting tested contingency X15≠228, the rightmost physical second-moment eigenvalue is

max⁡λ∈spec⁡(Ke,Θ)Re⁡(λ)≃−0.2843 s−1,\max_{\lambda\in\operatorname{spec}(K_{e,\Theta})}\operatorname{Re}(\lambda)\simeq-0.2843\ \mathrm{s}^{-1},

which gives

ηms≃0.2843 s−1>0.\eta_{\mathrm{ms}}\simeq0.2843\ \mathrm{s}^{-1}>0.

The reported operator therefore has a positive margin at σ∗2=2.25 s−1\sigma^{*2}=2.25\ \mathrm{s}^{-1} for the homogeneous second moment. Because the stochastic forcing is removed at t=Topt=T_{\mathrm{op}}, the subsequent interval Top≤t≤TevT_{\mathrm{op}}\le t\le T_{\mathrm{ev}} is governed by the deterministic covariance operator

K0(C)=AC+CAT.(39)\mathcal{K}_0(C)=AC+CA^{\mathsf{T}}. \tag*{(39)}

The post-reclosure covariance therefore evolves under the nominal deterministic operator.

Supporting Calibration and Transient Diagnostics

This appendix reports the parameter distributions, the two temporal sensitivity studies, and the full illustrative large-excursion trajectory. Their interpretation is given with the main calibration results in Sec. IV B.

Parameter Variation Across Scenarios

The scenario-specific M4\mathrm{M}_4 fits use the additional reclosure observations and parameter-sharing allocation described in Sec. III A. They yield 1,150 parameter pairs,

Θe,z=(αe,z,σe,z).(40)\Theta_{e,z}=(\alpha_{e,z},\sigma_{e,z}). \tag*{(40)}
Horizon-dependent relative map error

Figure 9. Horizon-dependent RSEmapRSE_{\mathrm{map}} for the 25 WECC testing contingencies at fixed Top=1.5 sT_{\mathrm{op}}=1.5\,\mathrm{s} and unscaled injections (χ=1)(\chi=1), with (α∗,σ∗)=(0.516,1.5)(\alpha^{*},\sigma^{*})=(0.516,1.5) held fixed. The observation time varies from reclosure to 2.5 s2.5\,\mathrm{s}. Dashed red curves are empirical guides, not universal bounds.

Across the 25 assessment lines and 46 operating states, these 0.8-s calibrated parameters span approximately

0.342≤αe,z∗≤0.810,1.070≤σe,z∗≤2.643,(41)0.342 \le\alpha^{*}_{e,z} \le0.810,\qquad1.070 \le\sigma^{*}_{e,z} \le2.643, \tag*{(41)}

with mean values of approximately

α‾∗≈0.558,σ‾∗≈1.620.(42)\overline{\alpha}^{*}\approx0.558,\qquad\overline{\sigma}^{*}\approx1.620. \tag*{(42)}

Figure 8 displays the variation with fault identity and operating condition in the respective parameter units. The corresponding state-transfer errors and calibration-effort trade-off are analyzed in Sec. IV B.

Horizon Dependence of the Map Error

This experiment varies TevT_{\mathrm{ev}} at fixed Top=1.5 sT_{\mathrm{op}}=1.5\,\mathrm{s} for the 25 testing lines. The injection vector is not rescaled: χ=1\chi=1 in (24). The initial operating condition and global pair Θ∗=(α∗,σ∗)=(0.516,1.5)\Theta^{*}=(\alpha^{*},\sigma^{*})=(0.516,1.5) remain fixed; the pair is not refitted as the observation time changes. In the notation below, the fixed operating state is suppressed. For the projected EMT state yeEMT(Tev)y^{\mathrm{EMT}}_{e}(T_{\mathrm{ev}}), the relative map error is

RSEmap(Tev,e)=∥HμΘ∗(Tev;e)−yeEMT(Tev)∥2∥yeEMT(Tev)∥2,(43)RSE_{\mathrm{map}}(T_{\mathrm{ev}},e)=\frac{\left\|H_{\mu\Theta^{*}}(T_{\mathrm{ev}};e)-y^{\mathrm{EMT}}_{e}(T_{\mathrm{ev}})\right\|_{2}}{\left\|y^{\mathrm{EMT}}_{e}(T_{\mathrm{ev}})\right\|_{2}}, \tag*{(43)}

Figure 9 shows that line-dependent discrepancies continue to grow after reclosure, reaching approximately 0.110.11–0.150.15 at Tev=2.5 sT_{\mathrm{ev}}=2.5\,\mathrm{s}. These are nominal-injection results, whereas Figure 4 aggregates the 46 scaled operating states. The increasing spread indicates that the subsequent propagation retains a dependence on fault location.

Sensitivity to Open-Phase Duration

The complementary experiment fixes Tev=1.5 sT_{\mathrm{ev}}=1.5\,\mathrm{s} and varies TopT_{\mathrm{op}} from 0 to 1.5 s1.5\,\mathrm{s} for the same 25 testing lines. The injections remain unscaled (χ=1)(\chi=1), and the global pair Θ∗=(α∗,σ∗)=(0.516,1.5)\Theta^{*}=(\alpha^{*},\sigma^{*})=(0.516,1.5) is not refitted. This study changes the switching duration, not the operating-state scale. For faulted line ee, the error is

RSEmap(Top,e)=∥HμΘ∗(Tev;e,Top)−yeEMT(Tev;Top)∥2∥yeEMT(Tev;Top)∥2,Tev=1.5 s.(44)RSE_{\mathrm{map}}(T_{\mathrm{op}},e)=\frac{\left\|H_{\mu\Theta^{*}}(T_{\mathrm{ev}};e,T_{\mathrm{op}})-y^{\mathrm{EMT}}_{e}(T_{\mathrm{ev}};T_{\mathrm{op}})\right\|_{2}}{\left\|y^{\mathrm{EMT}}_{e}(T_{\mathrm{ev}};T_{\mathrm{op}})\right\|_{2}}, \qquad T_{\mathrm{ev}}=1.5\,\mathrm{s}. \tag*{(44)}

Unlike (D4), this comparison changes the switching time rather than the observation horizon.

Figure 10 shows error approaching zero as Top→0T_{\mathrm{op}}\to0 and increasing with the open-phase duration. A longer interval extends the action of Ae(α∗)A_e(\alpha^{*}) and the localized multiplicative forcing in (2), changing the state supplied to post-reclosure evolution. The widening spread shows that this sensitivity depends on fault location.

Open-phase-duration dependence of RSEmap for the 25 WECC testing contingencies

Figure 10. Open-phase-duration dependence of RSEmapRSE_{\mathrm{map}} for the 25 WECC testing contingencies at fixed Tev=1.5 sT_{\mathrm{ev}} = 1.5\,\mathrm{s} and unscaled injections (χ=1\chi= 1). The global pair (α∗,σ∗)=(0.516,1.5)(\alpha^*,\sigma^*) = (0.516,1.5) is held fixed while TopT_{\mathrm{op}} varies from 0 to 1.5 s. Dashed red curves are empirical guides, not universal bounds.

The two temporal sweeps use the same random-seed configuration and agree at Top=Tev=1.5 sT_{\mathrm{op}}=T_{\mathrm{ev}}=1.5\,\mathrm{s}, where both report RSEmap∈[0.068,0.086]RSE_{\mathrm{map}}\in[0.068,0.086]. The upper endpoint is also the largest displayed error over the tested duration range. The seed identifies the numerical run; the exact conditional mean in (12) remains seed-independent for fixed inputs.

Illustrative Transient and Large-Angle Diagnostic

The corresponding linearized no-overload region is written as

Πθ={θ∈[−π,π]∣V∣:∣βij(θi−θj)∣≤p‾ij, ∀{i,j}∈Em}.(45)\Pi_{\theta}=\left\{\theta\in[-\pi,\pi]^{|\mathcal{V}|}:|\beta_{ij}(\theta_i-\theta_j)|\le\overline{p}_{ij},\ \forall\{i,j\}\in\mathcal{E}_{m}\right\}. \tag*{(45)}

Because (16) depends on phase differences, uniform common-mode phase motion does not contribute directly to the screening metric. Nevertheless, the linearized relation βij(θi−θj)\beta_{ij}(\theta_i-\theta_j) assumes sufficiently small phase separation. We examine this assumption for the illustrative large-excursion 9#4 contingency in Figure 11. Its 0.5-s open-phase interval and 10-s diagnostic window are independent of the calibration and hierarchy-test timings. Panel (a) shows the absolute phase trajectories obtained from ParaEMT and the calibrated surrogate, while panel (b) reports the corresponding separation ∣θ9(t)−θ4(t)∣\lvert\theta_9(t)-\theta_4(t)\rvert. The phase difference repeatedly exceeds the 0.5-rad diagnostic threshold, particularly during the large initial transient, and remains intermittently above this level during the subsequent response. For this illustrative case, the maximum separation is approximately 3.77 rad in ParaEMT and 4.24 rad in the surrogate. Thus, the large phase excursions in Figure 11 cannot be attributed solely to common-mode drift, and the small-angle approximation is not uniformly valid throughout this severe transient. These corridor-separation results give the diagnostic evidence for the indicator’s interpretation in § 2.4 A.

Sampling Laws, Estimators, and Statistical Records

We estimate the probabilities in (18)–(19) by combining Cross-Entropy Method (CEM) proposal adaptation with a separate importance-sampling estimation stage [12, 13, 14]. Both the learned line probabilities and the learned mean duration are retained for final estimation. Adaptation favors contingencies with large overload scores without changing the nominal probability law or refitting the physical surrogate parameters. Throughout this appendix, rr denotes the mean of the exponential proposal, in seconds, while τ0=0.1\tau_0 = 0.1 s is the mean of the nominal duration law.

Sampling space and distributions. Let nE=∣E∣n_E = \lvert\mathcal{E}\rvert.

The contingency sampling space is

Z=(e,τ)∈Z≔({∅}∪E)×R+,(46)\mathcal{Z} = (e,\tau) \in\mathcal{Z} \coloneqq(\{\varnothing\} \cup\mathcal{E}) \times\mathbb{R}_+, \tag*{(46)}

where ee identifies the faulted line and τ\tau is a realization of the open-phase duration TopT_{\mathrm{op}}. The screening horizon ThT_h is held fixed. The WECC implementation samples only faulted lines, assigning zero probability to e=∅e = \varnothing under both the nominal and proposal laws. Thus, the estimates are conditional on a fault occurring within the modeled line set.

The nominal law uses uniform line selection and an exponential duration,

pZ(e,τ)=1nE1τ0exp⁡(−ττ0),τ0=0.1 s,e∈E.(47)p_Z(e,\tau) = \frac{1}{n_E}\frac{1}{\tau_0}\exp\left(-\frac{\tau}{\tau_0}\right), \qquad\tau_0 = 0.1\,\mathrm{s}, \quad e \in\mathcal{E}. \tag*{(47)}

The adaptive proposal has categorical line probabilities and a single shared exponential mean,

qν(e,τ)=ϕe1rexp⁡(−τr),r>0,e∈E,(48)q_\nu(e,\tau) = \phi_e\frac{1}{r}\exp\left(-\frac{\tau}{r}\right), \qquad r > 0,\quad e \in\mathcal{E}, \tag*{(48)}

with qν(∅,τ)=0q_\nu(\varnothing,\tau) = 0. The mean rr is shared by all faulted lines; no line-specific duration mean is fitted. Both rr and τ0\tau_0 have units s. The nominal expected open-phase duration is therefore 0.1 s, whereas the proposal mean changes during adaptation. The operating-state scaling factor in (24) is a different quantity.

The categorical probabilities satisfy

ϕe≥0,∑e∈{∅}∪Eϕe=1,ϕ∅=0,(49)\phi_e \ge0,\qquad\sum_{e\in\{\varnothing\}\cup\mathcal{E}}\phi_e = 1,\qquad\phi_{\varnothing}=0, \tag*{(49)}

and the adjustable parameters are

ν=((ϕe)e∈E,r).(50)\nu= ((\phi_e)_{e\in\mathcal{E}},r). \tag*{(50)}

The required support condition is

pZ(e,τ)>0⟹qν(e,τ)>0andqν∗(e,τ)>0,(51)p_Z(e,\tau) > 0 \quad\Longrightarrow\quad q_\nu(e,\tau) > 0 \quad\text{and}\quad q_{\nu_*}(e,\tau) > 0, \tag*{(51)}

where qν∗q_{\nu_*} denotes the final estimation proposal defined below. Uniform initialization, positive duration parameters, and positive categorical smoothing preserve support during a finite adaptation stage.

Elite selection and parameter updates. The WECC adaptation uses

NCE=250,K=20,ρCE=0.1,nE=329,nelite=25,κ=0.9,ϕe(0)=1nE,r(0)=1.0 s.(52)\begin{gathered} N_{\mathrm{CE}} = 250,\qquad K = 20,\qquad\rho_{\mathrm{CE}} = 0.1,\\ n_E = 329,\qquad n_{\mathrm{elite}} = 25,\qquad\kappa= 0.9,\\ \phi_e^{(0)} = \frac{1}{n_E},\qquad r^{(0)} = 1.0\,\mathrm{s}. \tag*{(52)} \end{gathered}

Here, NCEN_{\mathrm{CE}} is the number of samples per adaptation iteration, KK is the fixed iteration count, and κ\kappa is the fraction of the previous parameter value retained in each update. The symbol kk distinguishes this numerical smoothing parameter from the physical coupling α\alpha. The initial proposal mean r(0)=1.0r^{(0)} = 1.0 s differs from the nominal mean τ0=0.1\tau_0 = 0.1 s. Updating the proposal does not change the expected duration under the nominal law.

At iteration ℓ\ell, draw NCEN_{\mathrm{CE}} contingencies from qν(ℓ−1)q_{\nu^{(\ell-1)}}, evaluate the corresponding trajectories through (2), and compute Sk(ℓ)=S(θ0(k)→Th)S_k^{(\ell)}=S(\theta_0^{(k)}\to T_{\mathrm{h}}). The elite set contains the indices of the 25 largest scores,

Iℓ=Top⁡25(S1(ℓ),…,SNCE(ℓ)),sℓelite=min⁡k∈IℓSk(ℓ).(53)\mathcal{I}_{\ell}=\operatorname{Top}_{25}\left(S_1^{(\ell)},\ldots,S_{N_{\mathrm{CE}}}^{(\ell)}\right),\qquad s_{\ell}^{\mathrm{elite}}=\min_{k\in\mathcal{I}_{\ell}}S_k^{(\ell)}. \tag*{(53)}

The operator Top⁡25\operatorname{Top}_{25} selects exactly 25 observations; ties do not enlarge the elite set. The empirical threshold sℓelites_{\ell}^{\mathrm{elite}} is distinct from the event threshold γ\gamma used for final estimation.

Elite membership is represented by the binary indicator

bk(ℓ)=I{k∈Iℓ}.(54)b_k^{(\ell)}=\mathbb{I}\{k\in\mathcal{I}_{\ell}\}. \tag*{(54)}

Maximizing ∑kbk(ℓ)log⁡qν(ek,τk)\sum_k b_k^{(\ell)}\log q_\nu(e_k,\tau_k) gives

ϕ^e(ℓ)=125∑k∈IℓI{ek=e},\widehat{\phi}_e^{(\ell)}=\frac{1}{25}\sum_{k\in\mathcal{I}_{\ell}}\mathbb{I}\{e_k=e\},
r^(ℓ)=(125∑k∈Iℓτk).(55)\widehat{r}^{(\ell)}=\left(\frac{1}{25}\sum_{k\in\mathcal{I}_{\ell}}\tau_k\right). \tag*{(55)}

The mean-duration update uses the pooled elite durations, consistent with the shared mean in (E3). The smoothed updates are

ϕe(ℓ)=0.9ϕe(ℓ−1)+0.1ϕ^e(ℓ),r(ℓ)=0.9r(ℓ−1)+0.1r^(ℓ).(56)\phi_e^{(\ell)}=0.9\phi_e^{(\ell-1)}+0.1\widehat{\phi}_e^{(\ell)},\qquad r^{(\ell)}=0.9r^{(\ell-1)}+0.1\widehat{r}^{(\ell)}. \tag*{(56)}

The categorical probabilities are renormalized after each update to remove roundoff. Even when a line has no elite observations, ϕe(ℓ)≥0.9ℓ/nE>0\phi_e^{(\ell)}\ge0.9^\ell/n_E>0 in exact arithmetic.

This is an unweighted elite-based CE proposal-learning rule. Unlike the nominal-density-weighted CE formulation for rare-event estimation, it does not include pZ/qνp_Z/q_\nu in the adaptation objective. Importance weights are applied in the final probability estimate. Adaptation is also independent of the chosen final threshold γ\gamma; it favors large global scores rather than stopping when a prescribed event level is reached.

The stopping rule is a fixed computational budget,

K=20,ν(K)=((ϕe(20))e∈E,r(20)).(57)K=20,\qquad\nu^{(K)}=\left((\phi_e^{(20)})_{e\in\mathcal{E}},r^{(20)}\right). \tag*{(57)}

No parameter-convergence tolerance or early-stopping test is used. The largest score, elite threshold, shared mean duration, and maximum categorical probability are recorded at every iteration.

Fresh-batch probability estimation. After the twentieth iteration, the complete learned proposal is frozen:

ϕe∗=ϕe(20),r∗=r(20),\phi_e^* = \phi_e^{(20)}, \qquad r^* = r^{(20)},
qν∗(e,τ)=ϕe(20)1r(20)exp⁡(−τr(20))=ϕe∗1r∗exp⁡(−τr∗).(58)q_{\nu^*}(e,\tau)=\phi_e^{(20)}\frac{1}{r^{(20)}}\exp\left(-\frac{\tau}{r^{(20)}}\right)=\phi_e^*\frac{1}{r^*}\exp\left(-\frac{\tau}{r^*}\right). \tag*{(58)}

Thus, qν∗=qν(20)q_{\nu^*}=q_{\nu^{(20)}}: the final batch retains both learned components. The probabilities 1/nE1/n_E belong to the nominal law and the initialization, not to the final categorical proposal. In general, ϕe∗≠1/nE\phi_e^*\ne1/n_E. The superscript * denotes the frozen proposal, not a claim of global optimality or verified convergence.

Let ξ∈ΩW\xi\in\Omega_W denote the stochastic driving realization. Matching the strict threshold comparison in the probability-estimation code, define

Aγ={(e,τ,ξ)∈Z×ΩW:S(θ0→Th(e,τ,ξ))>γ},(59)A_\gamma=\{(e,\tau,\xi)\in\mathcal{Z}\times\Omega_W:S(\theta_{0\to T_h}(e,\tau,\xi))>\gamma\}, \tag*{(59)}

for γ≥0\gamma\ge0. The additional coordinate makes explicit that overload is not determined by (e,τ)(e,\tau) alone. The stochastic drivers retain the same conditional law under the nominal and proposal simulations, so their likelihood factors cancel. At γ=0\gamma=0, strict exceedance selects any positive overload; the event S≥0S\ge0 would instead have probability one.

After adaptation, a fresh batch is drawn,

(ek,τk)∼iidqν∗,k=1,…,N,(60)(e_k,\tau_k)\mathrel{\overset{\mathrm{iid}}{\sim}}q_{\nu^*},\qquad k=1,\ldots,N, \tag*{(60)}

with fresh stochastic driving realizations. Conditional on the adaptation history, this estimation batch consists of independent draws from the frozen proposal; no adaptation observations are reused in the estimator.

(23) in the main text gives the estimate. The likelihood ratio corrects both changes of distribution:

pZ(ek,τk)qν∗(ek,τk)=1nEϕek(20) r(20)exp⁡(−τkτ0)τ0exp⁡(−τkr(20))=r(20)nEϕek(20)τ0exp⁡[(1r(20)−1τ0)τk].(61)\frac{p_Z(e_k,\tau_k)}{q_{\nu^*}(e_k,\tau_k)} = \frac{\frac{1}{n_E}}{\phi_{e_k}^{(20)}}\, \frac{r^{(20)}\exp\left(-\frac{\tau_k}{\tau_0}\right)} {\tau_0\exp\left(-\frac{\tau_k}{r^{(20)}}\right)} = \frac{r^{(20)}}{n_E\phi_{e_k}^{(20)}\tau_0} \exp\left[\left(\frac{1}{r^{(20)}}-\frac{1}{\tau_0}\right)\tau_k\right]. \tag*{(61)}

The categorical factor cannot be omitted merely because the nominal fault law is uniform. The denominator must use the probabilities that actually generated the final line samples.

Provided that samples are generated from the density used in this ratio, (E6) and the fresh-batch construction imply E[Q^(γ)∣Fadapt]=Q(γ)\mathbb{E}[\widehat{Q}(\gamma)\mid\mathcal{F}_{\mathrm{adapt}}]=Q(\gamma), where Fadapt\mathcal{F}_{\mathrm{adapt}} denotes the adaptation history. The estimator is therefore unbiased for the specified surrogate probability without self-normalizing the weights [14]. This property does not remove model error relative to ParaEMT.

Algorithm 1 N1Plus with elite-based categorical–exponential adaptation
Require: Surrogate score; nominal law (E2); screening horizon ThT_h; threshold γ≥0\gamma\ge0; final sample size NN.
1: Set NCE=250N_{\mathrm{CE}}=250, K=20K=20, ρCE=0.1\rho_{\mathrm{CE}}=0.1, and κ=0.9\kappa=0.9.
2: Initialize ϕe0=1/nE\phi_e^0=1/n_E and r(0)=1.0 sr^{(0)}=1.0\ \mathrm{s}.
3: for ℓ=1,…,20\ell=1,\ldots,20 do
4: Draw 250 contingencies from qν(ℓ−1)q_{\nu^{(\ell-1)}} and fresh stochastic drivers.
5: Evaluate (2) with Top=τkT_{\mathrm{op}}=\tau_k and compute the global overload scores to ThT_h.
6: Select the 25 largest scores using (E8).
7: Apply (E10)–(E11).
8: Renormalize ϕ\phi and record the adaptation diagnostics.
9: end for
10: Freeze r∗=r(20)r^*=r^{(20)} and ϕe∗=ϕe(20)\phi_e^*=\phi_e^{(20)} for every e∈Ee\in\mathcal{E}.
11: Draw NN fresh line indices from ϕ(20)\phi^{(20)} and durations from the exponential law with mean r(20)r^{(20)}, with fresh stochastic drivers.
12: Compute the global and monitored-line overload indicators.
13: Evaluate (23) using (E16).
14: Report the estimate, sampling uncertainty where applicable, final proposal, and separate adaptation and estimation costs.

Table 1 (PDF p. 21).

For weighted indicators YkY_k in (23), the estimated standard error is

se⁡^(Q^)=[1N(N−1)∑k=1N(Yk−Q^)2]1/2,(62)\widehat{\operatorname{se}}(\widehat{Q})= \left[ \frac{1}{N(N-1)} \sum_{k=1}^{N}(Y_k-\widehat{Q})^2 \right]^{1/2}, \tag*{(62)}

where YkY_k includes the full joint likelihood ratio in (E16). This standard-error interpretation requires N>1N>1 and finite variance. For finitely many line categories with ϕe∗>0\phi_e^*>0, the condition r∗>τ0/2r^*>\tau_0/2 ensures finite second moments of the weights under the mean-parameterized exponential proposal. This follows because the duration factor in pZ2/qν∗p_Z^2/q_{\nu^*} decays only when 2/τ0−1/r∗>02/\tau_0-1/r^*>0. Small categorical probabilities can still produce large weights; a finite empirical standard error alone does not establish reliable tail sampling.

Algorithm 1 summarizes proposal adaptation and the fresh-batch estimation stage using the complete learned joint distribution.

The adaptation budget is 250 × 20 = 5,000 surrogate evaluations. Consequently, a run with NN fresh estimation samples requires N+5,000N+5,000 evaluations; the supplied N=35,000N=35,000 example therefore requires 40,000 evaluations in total. End-to-end computational comparisons must include both stages. The supplied scorer uses Th=10 sT_h=10\ \mathrm{s}. When a sampled duration exceeds ThT_h, re-closure lies beyond the screening window; the original duration must still be retained in the likelihood ratio.

Both stages use the learned fault-location and duration components. The categorical probabilities emphasize particular corridors, while the shared mean duration changes the frequency of long open-phase intervals. Because (E3) factorizes, it does not introduce line–duration dependence. Concentration of ϕe(20)\phi_e^{(20)} is an interpretable indicator of which corridors dominate the adapted elite set, not a change in their physical failure frequencies. Quantitative contributions to the nominal overload probability, and conditional criticality rankings, must be computed with the full joint likelihood ratio rather than identified with ϕe(20)\phi_e^{(20)} itself.

The same estimator applies to a line-specific event by replacing the global score with Sij(θ0→Th(k))S_{ij}(\theta_{0\to T_h}^{(k)}). A global-score proposal can be reused when the support condition holds, although its efficiency need not be uniform across monitored lines. Further justification and definitions for rare-event/instanton sampling are given in Sections VI.A–B of [1]; the connection to static contingency screening is discussed in [33].

Conditional Contingency-to-Element Estimates

Under the uniform nominal fault law πe=1/nE\pi_e = 1/n_E, a fresh CEM batch gives

Q^ij(e;γ)=1Nπe∑k=1NwkI{ek=e}I{Sij(k)>γ},(63)\widehat{Q}_{ij}(e;\gamma)=\frac{1}{N\pi_e}\sum_{k=1}^{N}w_k\mathbb{I}\{e_k=e\}\mathbb{I}\{S_{ij}^{(k)}>\gamma\}, \tag*{(63)}

where nE=∣E∣n_E=|\mathcal{E}| and wkw_k is the full joint likelihood ratio in (E16). Replacing the final indicator by I{S(k)>γ}\mathbb{I}\{S^{(k)}>\gamma\} gives the global conditional estimate. The normalization by πe\pi_e distinguishes conditional risk from a fault’s contribution to the unconditional mixture over fault locations. The learned proposal masses ϕe(20)\phi_e^{(20)} are not substitutes for these weighted conditional probabilities.

Integration of EMT-Informed Calibration

For each sampled scenario (e,τ)(e,\tau), the dynamics in (2) are evaluated with Top=τT_{\mathrm{op}}=\tau and the global M2_2 parameters

Θ∗=(α∗,σ∗)=(0.516,1.5).(64)\Theta^*=(\alpha^*,\sigma^*)=(0.516,1.5). \tag*{(64)}

These physical parameters remain fixed during CE adaptation and final estimation; they are not proposal parameters. The stochastic driving law, pre-fault operating-state prescription, screening horizon, and overload definitions are unchanged from the nominal assessment. Each trajectory is scored using (16) and (17), and its contribution is weighted with the complete likelihood ratio in (E16). The conditional normalization is given in (E18).

WECC Probability Diagnostics and Runtime Records

Assessment protocol. This appendix supplies the statistical diagnostics and timing breakdown behind the main-text comparison in Table II. The nominal law, fixed physical parameters, horizon Th=10T_h=10 s, evaluation interval Δt=0.01\Delta t=0.01 s, and thresholds are those of Sec. IV C. MC samples directly from the nominal law. CEM uses the complete frozen proposal ϕe∗=ϕe(20)\phi_e^*=\phi_e^{(20)}, r∗=r(20)r^*=r^{(20)} and the likelihood ratio in (E16); its final batch is independent of the adaptation draws conditional on the fitted proposal.

The reported probability estimates use the global overload event at the four thresholds specified in the WECC study,

Q(γ)=P[S(θ0→Th)>γ],γ∈{0,0.5,5,10} s.(65)Q(\gamma)=\mathbb{P}[S(\theta_{0\to T_h})>\gamma],\qquad\gamma\in\{0,0.5,5,10\}\,\mathrm{s}. \tag*{(65)}

Probability estimates and sampling diagnostics. Table III reports the probability estimates and final-batch uncertainty and weight-concentration diagnostics. Table IV separately records the evaluation budgets and adaptation and estimation times. The main-text table combines only γ\gamma, method, NfinalN_{\mathrm{final}}, Q^\widehat{Q}, and total runtime; the additional columns are retained here for reproducibility and the variance comparison below.

γ\gamma [s]MethodNfinalN_{\mathrm{final}}Q^\widehat{Q}se^\widehat{\mathrm{se}}RelSEESSw\mathrm{ESS}_{w}w~max⁡\widetilde{w}_{\max}
0CEM2.500×1042.500 \times10^{4}5.600×10−25.600 \times10^{-2}1.678×10−31.678 \times10^{-3}2.997×10−22.997 \times10^{-2}2.009×1042.009 \times10^{4}5.976×10−55.976 \times10^{-5}
MC1.250×1051.250 \times10^{5}5.606×10−25.606 \times10^{-2}6.506×10−46.506 \times10^{-4}1.161×10−21.161 \times10^{-2}1.250×1051.250 \times10^{5}8.000×10−68.000 \times10^{-6}
0.5CEM4.000×1044.000 \times10^{4}2.628×10−22.628 \times10^{-2}5.430×10−45.430 \times10^{-4}2.067×10−22.067 \times10^{-2}3.214×1043.214 \times10^{4}3.817×10−53.817 \times10^{-5}
MC1.750×1051.750 \times10^{5}2.636×10−22.636 \times10^{-2}3.830×10−43.830 \times10^{-4}1.453×10−21.453 \times10^{-2}1.750×1051.750 \times10^{5}5.714×10−65.714 \times10^{-6}
5.0CEM1.050×1051.050 \times10^{5}7.743×10−37.743 \times10^{-3}2.290×10−42.290 \times10^{-4}2.958×10−22.958 \times10^{-2}8.437×1048.437 \times10^{4}1.429×10−51.429 \times10^{-5}
MC2.500×1052.500 \times10^{5}7.872×10−37.872 \times10^{-3}1.767×10−41.767 \times10^{-4}2.245×10−22.245 \times10^{-2}2.500×1052.500 \times10^{5}4.000×10−64.000 \times10^{-6}
10.0CEM1.550×1051.550 \times10^{5}2.723×10−32.723 \times10^{-3}1.940×10−41.940 \times10^{-4}7.125×10−27.125 \times10^{-2}1.245×1051.245 \times10^{5}1.573×10−51.573 \times10^{-5}
MC3.000×1053.000 \times10^{5}2.900×10−32.900 \times10^{-3}9.818×10−59.818 \times10^{-5}3.385×10−23.385 \times10^{-2}3.000×1053.000 \times10^{5}3.333×10−63.333 \times10^{-6}

Table III. Final-batch statistical reporting. Probability estimates, standard errors, relative standard errors, weight-based ESS, and maximum normalized weights are obtained using the WECC simulations.

γ\gamma [s]MethodNfinalN_{\mathrm{final}}NtotalN_{\mathrm{total}}Adaptation [s]Estimation [s]Total [s]
0CEM2.500×1042.500 \times10^{4}3.000×1043.000 \times10^{4}9.241×1029.241 \times10^{2}3.538×1033.538 \times10^{3}4.462×1034.462 \times10^{3}
MC1.250×1051.250 \times10^{5}1.250×1051.250 \times10^{5}0.0002.210×1042.210 \times10^{4}2.210×1042.210 \times10^{4}
0.5CEM4.000×1044.000 \times10^{4}4.500×1044.500 \times10^{4}9.415×1029.415 \times10^{2}6.235×1036.235 \times10^{3}7.177×1037.177 \times10^{3}
MC1.750×1051.750 \times10^{5}1.750×1051.750 \times10^{5}0.0003.185×1043.185 \times10^{4}3.185×1043.185 \times10^{4}
5.0CEM1.050×1051.050 \times10^{5}1.100×1051.100 \times10^{5}9.387×1029.387 \times10^{2}1.876×1041.876 \times10^{4}1.970×1041.970 \times10^{4}
MC2.500×1052.500 \times10^{5}2.500×1052.500 \times10^{5}0.0004.536×1044.536 \times10^{4}4.536×1044.536 \times10^{4}
10.0CEM1.550×1051.550 \times10^{5}1.600×1051.600 \times10^{5}9.502×1029.502 \times10^{2}2.733×1042.733 \times10^{4}2.828×1042.828 \times10^{4}
MC3.000×1053.000 \times10^{5}3.000×1053.000 \times10^{5}0.0005.547×1045.547 \times10^{4}5.547×1045.547 \times10^{4}

Table IV. Separately reported CEM and MC [24] evaluation budgets and stage runtimes. CEM totals include 5,000 adaptation evaluations, and total time is the sum of the supplied stage records.

For the reported estimator, define

wk=r(20)nEϕe(20)exp⁡[(1r(20)−1τ0)τk],w_k=\frac{r^{(20)}}{n_E\phi_e^{(20)}}\exp\left[\left(\frac{1}{r^{(20)}}-\frac{1}{\tau_0}\right)\tau_k\right],
Yk=wkI{Sk>γ},Y_k=w_k\mathbb{I}\{S_k>\gamma\},
w~k=wk∑j=1Nwj,Q^=1N∑k=1NYk.(66)\widetilde{w}_k=\frac{w_k}{\sum_{j=1}^{N}w_j},\qquad\widehat{Q}=\frac{1}{N}\sum_{k=1}^{N}Y_k. \tag*{(66)}

where N=NfinalN=N_{\mathrm{final}}, Sk=S(θ0→Th(k))S_k=S(\theta_{0\to T_h}^{(k)}), and nE=∣E∣=329n_E=|\mathcal{E}|=329. The categorical factor 1/(nEϕek(20))1/(n_E\phi_{e_k}^{(20)}) is retained in every importance weight.

For independent final samples with finite-variance weighted indicators, the estimated standard error and weight diagnostics are

se⁡^(Q^)=[∑k=1N(Yk−Q^)2N(N−1)]1/2,\widehat{\operatorname{se}}(\widehat{Q})=\left[\frac{\sum_{k=1}^{N}(Y_k-\widehat{Q})^2}{N(N-1)}\right]^{1/2},
RelSE⁡(Q^)=se⁡^(Q^)Q^,\operatorname{RelSE}(\widehat{Q})=\frac{\widehat{\operatorname{se}}(\widehat{Q})}{\widehat{Q}},
ESS⁡w=(∑k=1Nwk)2∑k=1Nwk2,w~max⁡=max⁡kw~k.(67)\operatorname{ESS}_w=\frac{\left(\sum_{k=1}^{N}w_k\right)^2}{\sum_{k=1}^{N}w_k^2},\qquad\widetilde{w}_{\max}=\max_k\widetilde{w}_k. \tag*{(67)}

The relative standard error RelSE measures sampling uncertainty in the probability estimate and is distinct from RSEmap\mathrm{RSE}_{\mathrm{map}}. The raw likelihood weight is wkw_k; its normalized diagnostic counterpart is w~k\widetilde{w}_k, with ∑kw~k=1\sum_k\widetilde{w}_k=1. Normalized weights are used only for diagnostics; the probability estimate remains the ordinary importance-sampling estimate in (23). Weight-based ESS and maximum normalized weight describe weight concentration, not event-specific precision by themselves [34].

For independent, unweighted MC samples,

RelSE⁡MC=[1−Q^(N−1)Q^]1/2,ESSw=N,w~max⁡=1N.(68)\operatorname{RelSE}_{\mathrm{MC}}=\left[\frac{1-\widehat{Q}}{(N-1)\widehat{Q}}\right]^{1/2},\qquad\mathrm{ESS}_w=N,\qquad\widetilde{w}_{\max}=\frac{1}{N}. \tag*{(68)}

These relative standard errors require Q^>0\widehat{Q}>0 and N>1N>1. Normalized nonnegative weights also satisfy

1≤ESS⁡w≤N,1 \leq\operatorname{ESS}_{w} \leq N,
1ESS⁡w≤w~max⁡≤1+(N−1)(N/ESS⁡w−1)N.(69)\frac{1}{\operatorname{ESS}_{w}} \leq\widetilde{w}_{\max} \leq\frac{1+\sqrt{(N-1)(N/\operatorname{ESS}_{w}-1)}}{N}. \tag*{(69)}

The entries in Table III satisfy these identities.

Contribution-variance comparison. Comparing different sample budgets requires separating sampling precision from the number of evaluations. Since Ns^e2N\widehat{s}_{\mathrm{e}}^{2} is the sample variance of an individual estimator contribution, define

V^ratio=NMCs^e,MC2NCEMs^e,CEM2.(70)\widehat{V}_{\mathrm{ratio}}=\frac{N_{\mathrm{MC}}\widehat{s}_{\mathrm{e,MC}}^{2}}{N_{\mathrm{CEM}}\widehat{s}_{\mathrm{e,CEM}}^{2}}. \tag*{(70)}

Here, both sample sizes refer exclusively to the final estimation batches. For independent samples, this compares the estimated per-draw variances: values above one favor CEM, whereas values below one favor MC. It is therefore the relevant sample-based comparison of fixed-NN efficiency, rather than the ratio of the selected batch sizes. It does not account for proposal-adaptation cost.

For the results shown in Table III, the ratios are

V^ratio≈{0.7517,γ=0,2.1761,γ=0.5 s,1.4184,γ=5 s,0.4957,γ=10 s.(71)\widehat{V}_{\mathrm{ratio}} \approx \begin{cases} 0.7517, & \gamma=0,\\ 2.1761, & \gamma=0.5\ \mathrm{s},\\ 1.4184, & \gamma=5\ \mathrm{s},\\ 0.4957, & \gamma=10\ \mathrm{s}. \end{cases} \tag*{(71)}

The contribution variance is lower for CEM at γ=0.5\gamma=0.5 and 5 s5\ \mathrm{s}, and lower for MC at γ=0\gamma=0 and 10 s10\ \mathrm{s}. The CEM final batches are smaller, so their standard errors also reflect the different sample counts. Equation (E25) separates that sample-size effect from the variability of individual contributions.

Computational cost. The specified CE adaptation uses

Nadapt=250×20=5,000,NCEM,total=Nfinal+5,000.(72)N_{\mathrm{adapt}}=250\times20=5{,}000,\qquad N_{\mathrm{CEM,total}}=N_{\mathrm{final}}+5{,}000. \tag*{(72)}

Each displayed CEM run includes one adaptation. Reusing a proposal across thresholds incurs that adaptation cost only once. Complete CEM runtime is

CCEM,total=Cadapt+Cestimate.(73)C_{\mathrm{CEM,total}}=C_{\mathrm{adapt}}+C_{\mathrm{estimate}}. \tag*{(73)}

This includes the pilot trajectories, sampling, elite selection, parameter updates, final trajectory evaluations, and weighting.

Table IV separates adaptation time from final estimation and records both final-batch and total evaluation counts. Adding its two time columns gives the totals summarized in Table II; no per-realization extrapolation is used. The comparison includes one adaptation per CEM run and pertains to the reported sampling budgets.

Surrogate runtime and published ParaEMT benchmark. The separately reported mean computation time for one surrogate trajectory is approximately 0.19 s0.19\ \mathrm{s}. The complete stage timings are recorded separately in Table IV.

The published ParaEMT benchmark in [8] reports

CEMT,1 corepub=29 s,CEMT,8 corespub=28 s.(74)C_{\mathrm{EMT},1\ \mathrm{core}}^{\mathrm{pub}}=29\ \mathrm{s},\qquad C_{\mathrm{EMT},8\ \mathrm{cores}}^{\mathrm{pub}}=28\ \mathrm{s}. \tag*{(74)}

These measurements correspond to a 1-s simulated horizon and a 50 μs50\ \mu\mathrm{s} internal integration step on Windows, using a machine equipped with two Intel Xeon Platinum 8280 2.7-GHz processors and 512-GB RAM.

The 0.01-s trajectory-evaluation interval in the screening procedure and the published internal EMT integration step describe different numerical operations. Table IV supplies the measured stage costs for the surrogate ensemble; Sec. V A states the limits of comparisons with the external EMT benchmark.

Duration-Resolved Screening

Line-specific screening. Figure 12 displays line-specific screening quantities as the open-phase duration TopT_{\mathrm{op}} varies over a fixed 10-s observation window. The overload integrals use the screening horizon ThT_{\mathrm{h}}, as defined in Sec. II A. The top panel shows accumulated overload duration; the bottom panel shows the probability of Sij>γS_{ij}>\gamma with γ=1 s\gamma=1\ \mathrm{s}. This line-specific event is distinct from the global event S>γS>\gamma. The faulted element and the monitored element need not coincide.

Accumulated overload versus operating time

(a) (a) Accumulated overload SijS_{ij} versus TopT_{\mathrm{op}}. Green identifies faulted lines (Xn,mX_{n,m}); red identifies critical transformer-connected monitored lines (i,ji,j). The indicator measures overload duration, not peak flow.

Exceedance probability versus operating time

(b) (b) Exceedance probability P[Sij>γ]\mathbb{P}[S_{ij} > \gamma] for γ=1\gamma= 1 s. Illustrative classes: safe (<0.025< 0.025), warning (0.0250.025–0.040.04), and emergency (>0.04> 0.04).

Figure 12. Duration-resolved WECC screening: (a) accumulated line overload; (b) line-specific exceedance probability. Both panels vary TopT_{\mathrm{op}} while accumulating exposure through Th=10T_{\mathrm{h}} = 10 s. The Tev=10T_{\mathrm{ev}} = 10 s notation inside these original graphics denotes that total trajectory window, not a calibration target. The label T(∗)T(*) in panel (b) denotes the exposure threshold γ=1\gamma= 1 s used throughout the text. These duration-resolved quantities are distinct from the global, duration-marginalized estimates in Table II.

These curves condition on duration. Their screening classes are defined in Sec. IV D, with the scope of their interpretation stated in Sec. V A.

Relation to the main-text screening thresholds. The duration-resolved assessment in Figure 12a identifies ten monitored transmission lines with Sij>1 sS_{ij}>1\ \mathrm{s} for the reported open-phase durations Top≥1.0 sT_{\mathrm{op}}\geq1.0\ \mathrm{s}; these lines are distinct from the ten initiating faults displayed as heatmap rows. The associated duration-resolved probabilities in Figure 12b enter the illustrative emergency class for the reported longer-duration cases. These are duration-conditioned results for severe cases within the tested set. The fault-conditioned, duration-marginalized probabilities in Figure 5 instead lie below the first classification boundary and use an independent continuous color scale.

References

  1. [1]Almada, A., Pagnier, L., Goldshtein, I., Kazi, S. R., & al., . et . (2026). Real-time dynamic n-1 screening: Identifying high-risk lines and transformers after common faults. In arXiv preprint. arXiv:2602.12293
  2. [2]ENTSO-E. (2007). Operational handbook and system operation guidelines. <a href="https://doi.org/https://eepublicdownloads.entsoe.eu/clean-documents/pre2015/publications/ce/report_2007_2.pdf" rel="noopener noreferrer" target="_blank">https://doi.org/https://eepublicdownloads.entsoe.eu/clean-documents/pre2015/publications/ce/report_2007_2.pdf</a>
  3. [3]North American Electric Reliability Corporation (NERC), “Reliability standards for the bulk electric systems of north america.” https://www.nerc.com/pa/Stand/Pages/ReliabilityStandards.aspx, 2023. Accessed: 2026-04-27.nerc.com/pa/Stand/Pages/ReliabilityStandards.aspx
  4. [4]A. Montanari, M. Tavares, and C. Portela, “Adaptative single-phase autoreclosing based on secondary arc voltage harmonic signature,” parameters, vol. 6, p. 7, 2009.
  5. [5]Godoy, E., Celaya, A., Altuve, H. J., Fischer, N., & Guzmán, A. (2012). Tutorial on single-pole tripping and reclosing. 39th Annual Western Protective Relay Conference.
  6. [6]Van Cutsem, T., & Vournas, C. (2007). Voltage stability of electric power systems.
  7. [7]M. Xiong, B. Wang, D. Vaidhynathan, J. Maack, M. J. Reynolds, A. Hoke, K. Sun, D. Ramasubramanian, V. Verma, and J. Tan, “An open-source parallel emt simulation framework,” Electric Power Systems Research, vol. 235, p. 110734, 2024.DOI
  8. [8]M. Xiong, B. Wang, D. Vaidhynathan, J. Maack, M. J. Reynolds, A. Hoke, K. Sun, and J. Tan, “Paraemt: An open source, parallelizable, and hpc-compatible emt simulator for large-scale ibr-rich power grids,” IEEE Transactions on Power Delivery, vol. 39, no. 2, pp. 911–921, 2023.DOI
  9. [9]Wood, A. J., Wollenberg, B. F., & Sheblé, G. B. (2013). Power generation, operation, and control.
  10. [10]P. W. Sauer, M. A. Pai, and J. H. Chow, Power system dynamics and stability: with synchrophasor measurement and power system toolbox. John Wiley & Sons, 2017.
  11. [11]Pavella, M., Ernst, D., & Ruiz-Vega, D. (2012). Transient stability of power systems: a unified approach to assessment and control.
  12. [12]R. Y. Rubinstein, “Optimization of computer simulation models with rare events,” European Journal of Operational Research, vol. 99, pp. 89–112, 1997.DOI
  13. [13]P.-T. De Boer, D. P. Kroese, S. Mannor, and R. Y. Rubinstein, “A tutorial on the cross-entropy method,” Annals of Operations Research, 2005.DOI
  14. [14]A. B. Owen, Y. Maximov, and M. Chertkov, “Importance sampling the union of rare events with an application to power systems analysis,” Electronic Journal of Statistics, vol. 13, no. 1, pp. 231–254, 2019.arxiv.org/abs/1710.06965
  15. [15]Commission, I. E. (1984). Iec 60050-441-17-44: dead time (during auto-reclosing). IEC Electropedia.
  16. [16]C. Gardiner, Stochastic Methods: A Handbook for the Natural and Social Sciences. Berlin, Heidelberg: Springer, 4 ed., 2009.
  17. [17]Øksendal, B. (2003). Stochastic Differential Equations: An Introduction with Applications.
  18. [18]R. Mannella and P. V. E. McClintock, “Itô versus stratonovich: 30 years later,” Fluctuation and Noise Letters, vol. 11, no. 1, p. 1240010, 2012.DOI
  19. [19]Rockies, N. L. of . the . (n.d.). ParaEMT Public: 240-Bus WECC PSS/E Model. GitHub Repository. <a href="https://doi.org/https://github.com/" rel="noopener noreferrer" target="_blank">https://doi.org/https://github.com/</a>
  20. [20]P. Kundur and O. P. Malik, Power system stability and control, vol. 7. McGraw-hill New York, 1994.
  21. [21]Chiang, H.-D. (2011). Direct methods for stability analysis of electric power systems: theoretical foundation, BCU methodologies, and applications.
  22. [22]Almada, A. (n.d.). PowerGridREserach: Power-grid dynamic screening code. GitHub Repository. <a href="https://doi.org/https://github.com/AyrtonAlmada/PowerGridREserach" rel="noopener noreferrer" target="_blank">https://doi.org/https://github.com/AyrtonAlmada/PowerGridREserach</a>
  23. [23]J. E. Price and J. Goodin, “Reduced network modeling of wecc as a market design prototype,” in 2011 IEEE Power and Energy Society General Meeting, pp. 1–6, IEEE, 2011.DOI
  24. [24]Kroese, D. P., & Rubinstein, R. Y. (2012). Monte carlo methods. Wiley Interdisciplinary Reviews: Computational Statistics, 4, 48–58.
  25. [25]Almada, A. (2026). WECC240 calibration reference trajectories and provenance. GitHub Repository. <a href="https://doi.org/https://github.com/AyrtonAlmada/PowerGridREserach/tree/main/data/wecc240/calibration" rel="noopener noreferrer" target="_blank">https://doi.org/https://github.com/AyrtonAlmada/PowerGridREserach/tree/main/data/wecc240/calibration</a>
  26. [26]developers, P. (n.d.). ParaEMT: Public source-code repository. GitHub Repository. <a href="https://doi.org/https://github.com/NatLabRockies/ParaEMT_public" rel="noopener noreferrer" target="_blank">https://doi.org/https://github.com/NatLabRockies/ParaEMT_public</a>
  27. [30]G. A. Pavliotis, “Stochastic processes and applications,” Texts in applied mathematics, vol. 60, pp. 41–43, 2014.DOI
  28. [31]Arnold, L. (1974). Stochastic Differential Equations: Theory and Applications.
  29. [32]R. Khasminskii, Stochastic Stability of Differential Equations. Berlin, Heidelberg: Springer, 2 ed., 2012.DOI
  30. [33]M. Chertkov, F. Pan, and M. G. Stepanov, “Predicting failures in power grids: The case of static overloads,” IEEE Transactions on Smart Grid, 2010.arxiv.org/abs/1006.0671
  31. [34]V. Elvira, L. Martino, and C. P. Robert, “Rethinking the effective sample size,” International Statistical Review, vol. 90, no. 3, pp. 525–550, 2022. # EMT Reference, WECC Case, and Event Notation {#sec:A} This appendix identifies the reference implementation and case inputs used in {xref}`Sec. III` and summarizes the switching and observation settings using the time conventions in {xref}`Table I`. ## Reference Model and WECC Case {#sec:A.1} ParaEMT couples three-phase nodal network equations with synchronous-machine, excitation, governing, stabilizer, and converter models. Its differential-algebraic formulation is evaluated using trapezoidal integration and sparse linear algebra {citep}`[7, 27, 28]`. The observation operator in (7) selects the electromechanical variables compared with the surrogate; fast electrical and device states remain in the reference model. The experiments use the public reduced WECC 240-bus case, with conventional and renewable generation, loads, transmission corridors, and approximately 20% inverter-based-resource penetration {citep}`[8]`. The case is distributed in the `models/240bus_psse` directory of ParaEMT, with the `WECC240.raw` network description and `240bus.xls` model data {citep}`[19]`. All calibration levels use this same topology and set of dynamic-model parameters; only the specified contingency and operating condition change. {xref}`Figure 6` shows the network. ::: figure #fig:6 (6) ![One-line representation of the reduced WECC 240-bus calibration case](media/fig6.png) bbox=0.52,0.07,0.92,0.47 caption: One-line representation of the reduced WECC 240-bus calibration case, adapted from [29]. ::: ## Timeline Reference {#sec:A.2} The switching sequence is defined in {xref}`Sec. II A` and illustrated in {xref}`Fig. 1`. {xref}`Table I` collects the time conventions; the experiment-specific settings and calibration allocations are given in {xref}`Sec. III A`. The 9#4 illustration uses $T_{\mathrm{op}} = 0.5$ s and ends at 10 s. These settings remain distinct from the nominal screening-duration distribution. # Sequence-Network Interpretation of the Coupling {#sec:B} A quantitative sequence-network calculation provides a first physical check of the fitted coupling. Among the WECC testing contingencies, line X110#154 has positive-sequence reactance $X_1 = 0.02886$ p.u., equal to the median reactance of the tested line set, and is therefore selected as a representative case. For a one-conductor-open condition, the negative- and zero-sequence networks contribute through $$ Z_{20} = Z_2 \mathbin\Vert Z_0 = \frac{Z_2 Z_0}{Z_2 + Z_0}, $$ so that the effective positive-sequence transfer impedance is $$ Z_{\mathrm{op}} = Z_1 + (Z_2 \mathbin\Vert Z_0). $$ Using the standard transmission-line approximation $Z_2 \simeq Z_1$ together with representative overhead-line sequence ratios $R_0/R_1 = 13.03$ and $X_0/X_1 = 3.49$ gives $$ Z_1 \simeq 0.00202 + \mathrm{j}0.02886,\qquad Z_0 \simeq 0.02628 + \mathrm{j}0.10072, $$ and $$ Z_2 \mathbin\Vert Z_0 \simeq 0.00249 + \mathrm{j}0.02257. $$ Because $\alpha$ scales the line coupling in (4), the corresponding comparison is made at the susceptance level, $$DOI

Paper details

vlnc:W_jw6ug8jt
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents