Project purpose

This project continues the fundamental research effort at LANL to model the dynamic behavior and interactions of these critical infrastructure systems, and develops techniques to analyze, coordinate, and optimize their operations. The objectives are to 1) advance gas pipeline physics and engineering models to improve dynamic pipeline optimization technology; 2) formulate gas and electricity transmission network optimization problems to include improved models, engineering economics, and resilience analysis support; and 3) develop methods for forecasting pipeline loads, state estimation, and management of uncertainty using real-time measurement data. In the short term, this project will develop models of intra-day behavior of coupled gas-electricity networks, to support resilience analysis of response to rapid changes in loads or other disruptions. In the long term, the developed methods will be transitioned to industry to create commercial decision support tools that will improve pipeline operations and wholesale natural gas markets to enable responsive delivery on the time-scales of decision-making for the power grid. Crucially, the developed algorithms and methods will be rooted in solid mathematics and physics foundations, and will be cross-validated using state-of-the-art commercial tools and data from real systems.

 

Technical approach

In past projects, LANL and collaborators have achieved breakthroughs in modeling and optimization of gas pipeline dynamics and coordination of gas pipeline and electric power delivery systems using market-based, optimization-driven approaches. First, these complex dynamics were analyzed to separate the key temporal scales to reveal the cross-infrastructure dynamical interactions of natural gas line-pack and underground storage with the multi-scale, time-linked, electrical constraints resulting from generator operations and control. Further dynamical analysis generated key results for advanced operational optimization and control of these coupled infrastructures: monotonicity of gas flows and gas pressure for both quasi-static and dynamical settings, which can be used for advanced cross-infrastructure security assessments. The pipeline dynamics and coupled gas-grid models were incorporated into optimization formulations beginning with quasi-static optimal gas flow, and advancing to dynamical optimal gas flow and optimal dynamic gas-grid operations and market-based coordination. Together with industry partners, we have examined business practices, operational realities, and regulatory environments of the wholesale gas and electricity transmission sectors. This led to explorations of optimization-based gas market designs and gas-electric coordination mechanisms. The modeling advances, computational methods, and coordination mechanisms that were developed have been validated in a number of studies on synthetic test cases, as well as back cast simulations using real data. The developed models were also adapted for state and parameter estimation, as preliminary work towards the data assimilation tools needed for fielding pipeline control systems. We have also developed methods for resilience analysis of electricity and coupled gas-electric systems.

The current project continues to extend our research to 1) solve fundamental applied mathematics problems arising from gas-electric system modeling, estimation, optimization, and control; 2) develop tools for efficient and computationally tractable gas-electric system resilience analysis that represents ambiguous and non-transparent human factors using optimization; 3) enable future development of pipeline transient optimization tools, gas pipeline market designs, and gas-electric coordination mechanisms that can be demonstrated in pilot projects by improving, extending, and providing solution guarantees on prototype models and algorithms.

 

Advanced gas dynamics modeling to improve pipeline transient optimization

Current pipeline transient models require relatively fine spatial discretization of the pipeline network, resulting in costly computation of solutions to optimization problems. Starting from the basic physics, we will develop models that admit much coarser spatial discretization and linear or quasi-linear representation to enable scaling up of dynamic optimization tools to provide solutions for very-large interconnected pipelines. To advance our progress towards acceptance by our long-term industry collaborators, we will account for how pipeline flows can control using current and emerging technology and business practices. Solutions will be benchmarked and validated on test cases synthesized using models and time-series data of real pipeline systems. The modeling development will be targeted specifically towards formulations that can be used in transient optimization tools.

 

Optimization formulations for gas and electricity transmission with improved models, engineering economics, and resilience support

We will develop new formulations for gas-electricity system intra-day optimization that address realistic decision-making and exchange of information between gas pipeline and electric grid operators within timely decision cycles. These formulations will incorporate our advanced dynamic gas pipeline models to improve computation speed while guaranteeing accurate physical and economic solutions (location-dependent hourly gas prices). The developed optimization tools will provide a guideline of how human operators would manage coupled gas-electricity delivery networks under normal and anomalous conditions, if they were able to do so optimally. These tools will serve as a proxy for operator response to N-k contingencies in resilience analysis. The developed tools will address practical business processes of pipeline markets and operations and contribute to new standards for coordination with the electrical grid. Based on how optimization is used to price wholesale electricity for the power grid, we will advance our theory of gas pipeline engineering economics to guarantee pricing stability, fairness, and market adequacy. We will also formulate transient optimization problems that can be used for pricing intra-day gas delivery nominations. The analysis will serve as a model for price formation in the national gas market, and will be a prototype for optimization tools that could be used by pipeline system operators for pricing gas delivery in the future.

 

Real-time measurements for forecasting loads, state estimation, and mitigation of uncertainty

We will develop models for forecasting gas pipeline loads based on historical data, power grid activity, and weather forecasts. We will then extend our optimal control formulations to moving-horizon, model-predictive methods that use state estimation to re-adjust control setpoints using measurements of the system and look-ahead predictions. Based on our preliminary work, the new estimation schemes will jointly learn the pipeline flow and pressure state and tune parameters using the sparse, noisy time-series data available at instruments on real supervisory control and data acquisition (SCADA) systems of gas pipeline networks. We will derive error bounds on the learned estimates, and apply our work on monotone system theory and stochastic models of dynamic gas flows and pressures to modify our decision support tools to be robust to uncertainty. This will enable recourse for pipeline operators to optimally prepare for sudden changes in gas flows caused by activation of gas-fired generators over the (hour-ahead) short term and respond to more significant power system contingencies, thereby improving the overall resilience of the gas system to extreme conditions arising from events on the power grid. This project component addresses a gap in effective tools that was highlighted by our industry partners.