Search for: All records

Extreme weather events and weather anomalies are on the rise, creating unprecedented struggles for the electrical power grid. With the aging of the United States power grid, the status quo for maintaining the transmission and distribution system, demand, generation, and operations will no longer suffice under the current and future conditions. Such conditions will require a shift in thinking and operating the power grid toward a weather-driven power system. This paper conducts a comprehensive review of each component of the power grid regarding the current leading weather events related to major power outages in the United States. For each event, contemporary issues and possible adaptions are presented, following a parallel comparison of the power grid development and knowledge of global climate modeling. Further, a background in global climate modeling is provided through the lens of an energy professional to aid in emission scenarios used in future studies. Overall, this paper works toward bridging the gap between weather and climate-related studies and operating the power grid in an uncertain climatic landscape while offering possible adaptions and solutions at a short-term and long-term scale.

 

Optimal Power Flow (OPF) is a challenging problem in power systems, and recent research has explored the use of Deep Neural Networks (DNNs) to approximate OPF solutions with reduced computational times. While these approaches show promising accuracy and efficiency, there is a lack of analysis of their robustness. This paper addresses this gap by investigating the factors that lead to both successful and suboptimal predictions in DNN-based OPF solvers. It identifies power system features and DNN characteristics that contribute to higher prediction errors and offers insights on mitigating these challenges when designing deep learning models for OPF. 

This paper describes the development of a phasor-based campus microgrid model utilizing the Modelica language and the OpenIPSL library. The phasor-based modeling approach was chosen because the resulting microgrid model would yield faster simulation run times when compared to models developed using electromagnetic transient (EMT) methods. Beyond the benefits of simulation performance, this becomes necessary when attempting to understand dynamic phenomena arising under emergency conditions across time scales ranging from milliseconds to hours, which will aid in developing resiliency improvement plans for the real-world campus microgrid that the model represents. Considering the increasing number of distributed energy sources (DERs) being added to power grids across the world and the paradigm shift on how electrical grids can operate with more DERs, the implementation of such a microgrid campus model can help in the development and testing new control strategies to support new operational approaches while guaranteeing system stability and resiliency. The added benefit of having the microgrid model in Modelica is that it can be simulated in any Modelica complaint tool (both proprietary or not), preserving an open-source code, unlocked for the user to explore and adjust the implementation as well as observe and edit the mathematical formulation. This enables not only nonlinear time simulation, but also linear analysis techniques and other approaches to be applied. 

Increasing levels of renewable generation motivate a growing interest in data-driven approaches for AC optimal power flow (AC OPF) to manage uncertainty; however, a lack of disciplined dataset creation and benchmarking prohibits useful comparison among approaches in the literature. To instill confidence, models must be able to reliably predict solutions across a wide range of operating conditions. This paper develops the OPF-Learn package for Julia and Python, which uses a computationally efficient approach to create representative datasets that span a wide spectrum of the AC OPF feasible region. Load profiles are uniformly sampled from a convex set that contains the AC OPF feasible set. For each infeasible point found, the convex set is reduced using infeasibility certificates, found by using properties of a relaxed formulation. The framework is shown to generate datasets that are more representative of the entire feasible space versus traditional techniques seen in the literature, improving machine learning model performance. 

Learning mappings between system loading and optimal dispatch solutions has been a recent topic of interest in the power systems and machine learning communities. However, previous works have ignored practical power system constraints such as generator ramp limits and other intertemporal requirements. Additionally, optimal power flow runs are not performed independently of previous timesteps - in most cases, an OPF solution representing the current state of the system is heavily related to the OPF solution from previous timesteps. In this paper, we train a recurrent neural network, which embeds natural relationships between timesteps, to predict the optimal solution of convex power systems optimization problems with intertemporal constraints. In contrast to traditional forecasting methods, the computational benefits from this technique can allow operators to rapidly simulate forecasts of system operation and corresponding optimal solutions to provide a more comprehensive view of future system states. 

Optimal Power Flow (OPF) is a fundamental problem in power systems. It is computationally challenging and a recent line of research has proposed the use of Deep Neural Networks (DNNs) to find OPF approximations at vastly reduced runtimes when compared to those obtained by classical optimization methods. While these works show encouraging results in terms of accuracy and runtime, little is known on why these models can predict OPF solutions accurately, as well as about their robustness. This paper provides a step forward to address this knowledge gap. The paper connects the volatility of the outputs of the generators to the ability of a learning model to approximate them, it sheds light on the characteristics affecting the DNN models to learn good predictors, and it proposes a new model that exploits the observations made by this paper to produce accurate and robust OPF predictions.