Early warning and proactive control strategies for power blackouts caused by gas network malfunctions

The paper addresses growing interdependence between gas networks and electric power systems during the transition to low-carbon energy, where gas turbines provide flexible generation to complement variable renewables. While gas infrastructure may later carry low-carbon fuels (e.g., hydrogen), current and future reliance on gas networks exposes power systems to gas-related disturbances. Historical events (e.g., February 2021 Texas/South Central U.S. cold-weather event) and other incidents show gas malfunctions can trigger large power deficits and load shedding. The authors hypothesize that the inherent propagation delay of gas network failures (buffered by pipeline line pack) relative to the rapid detection/communication and power system redispatch can be leveraged for early warning and proactive control to prevent or reduce power outages. They propose a gas-electric early warning system that provides the electric power control center (EPCC) with concise indicators of gas failures so that proactive, coordinated redispatch can occur before gas-turbine inlet pressures fall below protection thresholds, thereby mitigating gas-electric cascading failures.

The study situates itself within resilience and security research from institutions such as LANL, NREL, and EMPA, which analyze system responses to internal (e.g., leaks) and external (e.g., extreme weather, cyberattacks) disruptions across prepare–absorb–adapt–recover phases. Work on integrated energy systems (e.g., hurricane analysis, disaster avoidance) has advanced, but real-time cross-infrastructure coordination to prevent gas-electric cascading failures remains limited. Existing approaches often overlook the exploitable delay between gas system malfunctions and their impact on gas-fired generators. Inspired by early-warning successes in geohazards (earthquake/tsunami), the authors propose a cross-energy early warning that uses minimal, privacy-preserving indicators to coordinate timely power system actions.

Concept and information flow: Upon a major gas network failure, adjacent valves isolate the fault, potentially severing supply to certain gas turbines (impacted turbines). Due to line pack in the terminal pipeline segment between the nearest isolation valve and the turbine, turbine inlet pressure declines over minutes, providing an opportunity window. The gas dispatch centre (GDC) detects the failure, computes early warning indicators, and transmits them to the EPCC, which proactively redispatches generation before turbine protection thresholds are reached. Early warning indicators: Two indicators are defined and sent from GDC to EPCC:

• Available Escape Time (AET): The remaining time before the impacted turbine’s inlet pressure drops to its protection threshold given current gas loads and the turbine’s ramp-down trajectory. AET(t) decreases with time, and under turbine ramp-down with constant other loads, AET(t) − AET(t+Δt) < Δt. The initial AET at failure onset, SAET = AET(0), measures urgency: smaller SAET implies a more urgent event.
• Available Line Pack (ALP): Spatial indicator quantifying the portion of line pack in the terminal area that can be utilized before the turbine is forced to reduce generation or trip. Let LP be the current line pack and LP_last the line pack corresponding to the lowest inlet pressure that still avoids forced reduction; ALP = LP − LP_last. Larger ALP implies more buffer to enable controlled ramp-down and substitution. Illustrative example: A small network with one impacted gas turbine and one fixed gas load demonstrates how different turbine generation ramp-down rates affect mass flow, ALP, and AET. A critical declining ramp rate maximizes operation time: slower ramps exhaust ALP before shutdown (forced trip), faster ramps stop the turbine early, leaving unused ALP but limiting operation time. Hydraulic simulation and indicator computation: Dynamic gas hydraulics are modeled with an isothermal pipeline PDAE set, reduced to a telegraph-like wave equation system and solved using a finite-difference time-domain (FDTD) scheme to obtain pressure and mass flow trajectories after failure. For steady-state checks, the Weymouth equation provides a simple pressure-distance relation. For fast computation of indicators, the inlet pressure is approximated by the average pressure of the terminal segment; then ALP is computed from initial LP (integral of density over terminal segment) minus LP at the protection-threshold average pressure. SAET across multiple impacted turbines is computed by ordering turbines by initial ALP and accumulating flows from turbines and other loads. Proactive control formulation: With AET/ALP, EPCC solves a linear programming redispatch to minimize total control cost over the proactive control horizon (minutes), prioritizing avoiding load shedding (generator adjustment cost much smaller than load-shedding cost). Decision variables are generator and load powers over time. Constraints include: power balance, DC power flow approximation, generator capacity and ramp-rate limits, load adjustment bounds, nonnegative ALP, and control-time constraints for impacted turbines. Static proactive control uses each impacted turbine’s control time set to its SAET (i.e., reduce to zero within SAET). If residual deficit remains, dynamic proactive control iteratively extends the control time within remaining AET to further exploit ALP and reduce deficits. Communication latencies are negligible relative to gas pressure propagation times, enabling timely action.

• Concept feasibility: Pressure propagation in gas pipelines occurs over minutes, while detection, communication, and power redispatch occur over seconds, enabling early warning and proactive mitigation.
• City-level case (Zhejiang, China): Under gas source (GS) failure, without control, turbines XS and HZ would be forced offline at 30 min 25 s and 13 min 21 s, yielding deficits of 1047.2 MW and 675.4 MW. Static proactive control, using SAET and ALP, fully eliminates the aggregated 1722.6 MW deficit. For certain pipeline failures, static control cannot fully eliminate deficits; dynamic control that extends proactive control time reduces or eliminates deficits by exploiting additional ALP (e.g., extending XS from 4 min 16 s to 7 min 53 s reduces a pipeline-induced deficit to zero leaving 913.32 kg unused ALP; extending HZ from 1 min 07 s to 1 min 38 s reduces the maximum deficit to 438.4 MW with ALP exhausted).
• Provincial case (multi-fault winter storm scenario): With simultaneous gas source/pipeline failures and two wind farm malfunctions, six of twelve gas turbines are impacted with SAETs ranging from minutes to hours. Compared to passive control (max deficit 1577.0 MW; total energy deficit 345.5 MWh), static proactive control reduces deficits to 1023.6 MW and 213.2 MWh; dynamic proactive control further reduces them to 372.0 MW and 13.8 MWh.
• Operational insights: Proactive control leverages longer adjustment windows for normal generators, especially gas turbines with higher ramp rates than coal-fired units, allowing substitution before turbine trips. Sharing only AET and ALP suffices for EPCC optimization, preserving cross-entity privacy.

The findings show that exploiting the temporal lag inherent in gas network failure propagation can materially reduce or eliminate power deficits that would otherwise trigger frequency declines and load shedding, especially as system inertia decreases with higher renewable penetration and traditional N−1 criteria become insufficient under common-mode fuel disruptions. By transmitting only AET and ALP, the GDC enables EPCC to coordinate timely, system-wide redispatch without exposing detailed gas network data, addressing privacy and institutional separation. Static proactive control can often avoid deficits when SAET is adequate; when SAET is short or ALP limited, dynamic proactive control iteratively extends control time to utilize more line pack, further shrinking deficits. The approach is particularly valuable under extreme weather and multi-contingency scenarios, and aligns with broader resilience strategies by shifting from passive, late-stage emergency actions to proactive, cost-minimizing control that maintains service continuity.

This work proposes a gas-electric early warning system that quantifies gas network malfunctions using two concise indicators—AET and ALP—and enables the EPCC to implement proactive redispatch before gas turbine protection thresholds are reached. The method includes a tractable linear programming formulation with static and dynamic variants to minimize total control cost and energy deficit while respecting operational constraints. Case studies on a real coupled gas-electricity system in Zhejiang Province demonstrate significant reductions in maximum and total energy deficits, and in several scenarios complete avoidance of power shortfalls. The contributions include: (i) defining privacy-preserving early warning indicators that capture temporal and spatial urgency of gas failures; (ii) integrating gas hydraulic dynamics with power system proactive control; and (iii) validating effectiveness at city and provincial scales. Future work includes integrating the early warning framework with more detailed cascading failure models, assessing and enhancing gas storage configurations to ensure sufficient buffering, and extending cross-energy early warning concepts to other coupled infrastructures (e.g., transportation–electricity).

The effectiveness of the early warning approach depends on sufficient pipeline line pack to buffer pressure declines; if the available ALP is small or the SAET is very short (e.g., failure near the turbine), proactive control time may be insufficient to fully substitute generation, limiting benefit. Not all contingencies provide adequate SAET for complete escape, necessitating dynamic control or accepting residual deficits. Practical deployment requires configuring typical failure scenarios and probabilities to verify that existing gas storage and pipeline capacities meet early warning needs; if not, additional gas storage may be required. The method also presumes timely detection, communication from GDC to EPCC, and reliable execution of generator redispatch within ramp and capacity constraints.