Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “dynamic scheduling”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 73 records · Page 4
All sourcesNASA NTRSDOE OSTI
DOE OSTI2026

The relative influences of hydrologic information and dams’ hydropower scheduling decisions on electricity price forecasts

Price dynamics in wholesale electricity markets are driven by supply and demand. In markets with hydroelectric dams, the timing and amount of hydropower offered can influence prices in similar ways to wind and solar power. Unlike variable renewable energy, however, the supply of hydropower in wholesale markets is a function of both water availability and operational decisions at dams. Dam operators maximize revenues in wholesale markets by aligning generation with the periods of highest expected prices, and these scheduling decisions may in turn influence prices. Here, we examine the relative importance of two types of information in predicting forward electricity prices: a) water availability at dams, in the form of short-to-medium-range hydrological forecasts; and b) hourly scheduling decisions at dams. Using softly coupled hydrologic, hydropower scheduling, and power systems models spanning the U.S. Western Interconnection, we quantify the importance of hydrologic forecast accuracy in correctly predicting wholesale electricity prices and compare this with the influence of dam operators’ own hourly scheduling decisions on realized market prices. We find that aligning hydropower generation schedules with the periods of high forecasted prices causes larger, inadvertent price forecast errors than imperfect hydrologic forecasts. This suggests that knowledge of how water is managed by dam operators within the week is more important than weekly inflow forecast errors when predicting forward electricity prices. Our findings have implications for optimal hydropower scheduling by region. Specifically, accounting for price effects is critical in markets dominated by hydropower capacity.

Electricity markets↗
DOE OSTI2022

DRAS: Deep Reinforcement Learning for Cluster Scheduling in High Performance Computing

Cluster schedulers are crucial in high-performance computing (HPC). They determine when and which user jobs should be allocated to available system resources. Existing cluster scheduling heuristics are developed by human experts based on their experience with specific HPC systems and workloads. However, the increasing complexity of computing systems and the highly dynamic nature of application workloads have placed tremendous burden on manually designed and tuned scheduling heuristics. More aggressive optimization and automation are needed for cluster scheduling in HPC. In this work, we present an automated HPC scheduling agent named DRAS (Deep Reinforcement Agent for Scheduling) by leveraging deep reinforcement learning. DRAS is built on a hierarchical neural network incorporating special HPC scheduling features such as resource reservation and backfilling. An efficient training strategy is presented to enable DRAS to rapidly learn the target environment. Once being provided a specific scheduling objective given by the system manager, DRAS automatically learns to improve its policy through interaction with the scheduling environment and dynamically adjusts its policy as workload changes. We implement DRAS into a HPC scheduling platform called CQGym. CQGym provides a common platform allowing users to flexibly evaluate DRAS and other scheduling methods such as heuristic and optimization methods. Furthermore, the experiments using CQGym with different production workloads demonstrate that DRAS outperforms the existing heuristic and optimization approaches by up to 50%.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

Inertia estimation for power grids: A review of methods, challenges, and future prospects

The electric power grid is undergoing a significant transformation, shifting from traditional synchronous generators to inverter-based resources (IBRs) such as solar photovoltaics, wind turbines, and energy storage systems. This evolution leads to a reduction in system inertia, a critical attribute for maintaining frequency stability in response to disturbances. Consequently, the ability to monitor and estimate system inertia has become increasingly essential. This paper provides a comprehensive review of existing inertia estimation methodologies, analyzing them from multiple perspectives, including the types of data utilized, underlying estimation principles, operational modes, and system-wide applicability. A comparative summary table is included to distill commonalities and key characteristics across various studies. In addition, the paper examines practical implementations of inertia estimation across several major power systems worldwide, including the U.S. interconnections, the Nordic power system, and the U.K. grid. Key challenges are identified, particularly in estimating contributions from virtual inertia sources and load-induced inertia in increasingly converter-dominated networks. To address these emerging challenges, the paper proposes an integrated framework for real-time inertia estimation and monitoring. This framework encompasses critical components such as data acquisition, inertia estimation from both synchronous and non-synchronous sources, load-induced effects, optimization techniques, forecasting, and virtual inertia scheduling. Collectively, these elements enable dynamic, system-wide monitoring and adaptive control of grid inertia.

Inertia estimation↗
DOE OSTI2020

Data-Driven Security Assessment of Power Grids Based on Machine Learning Approach: Preprint

Data-driven security assessment provides key indicators on power system stability using simulations on scheduling models, as opposed to dynamic simulations that are more time-consuming. This paper investigates data-driven security assessment of power grids based on machine learning. Multivariate random forest regression is used as the machine learning algorithm due to its high robustness to the input data. Three stability issues are analyzed using the proposed machine learning tool, including transient stability, frequency stability and small signal stability. The estimation values from machine learning tool are compared with those from dynamic simulations. Results show that the proposed machine learning tool can effectively predict the stability margins for the three stability metrics.

14 SOLAR ENERGY↗
DOE OSTI2020

A direct transcription-based multiple shooting formulation for dynamic optimization

The growing need for fast and efficient solution techniques for solving dynamic optimization problems is driven by a broad spectrum of applications in scheduling and control. We suggest a novel framework for dynamic optimization that utilizes a multiple shooting “backbone” with discrete rather than continuous subproblems, thereby eliminating need for repeated time-integration. A Lagrangian relaxation (LR)-based decomposition scheme is proposed, which dualizes the state continuity requirements between subproblems and enables parallel solution of the problem. We demonstrate the applicability of our method on two case studies: the Van der Pol oscillator and a batch reactor.

42 ENGINEERING↗
DOE OSTI2021

Optimal scheduling for profit maximization of energy storage merchants considering market impact based on dynamic programming

This paper analyzes how electricity merchants' market impact affects merchants' profit. Energy storage has long been studied for its role in maximizing profit, and merchant decisions are assumed to have no impact on market prices. However, the trading decisions of large-scale energy storage merchants (e.g., pumped storage hydro) will affect the market prices. This paper employs dynamic programming theory to investigate merchants' optimal economic dispatch considering the market impact and physical characteristics of storage systems. Our findings show that the State-of-Charge (SOC) based analytical solution significantly facilitates energy storage merchants' decision-making. The SOC range is segmented into three regions by two optimal SOC reference points, which depend on the available energy in storage, given prices, and market impact. By comparing the current storage SOC with the reference points, the merchant can get the corresponding optimal actions. We analytically show that if the merchant neglects the market impact on the power market, she will exaggerate her expected profit when the price-taker and price-maker merchants have the same generating and pumping upper limits offered to Independent System Operators (ISOs). Furthermore, the profit-maximizing merchant must, therefore, assay to balance the trade-off correctly between the intensity of market impact and the dispatched power. Our findings are verified by numerical simulation, and results demonstrate the ramifications for electricity merchants in energy arbitrage decisions.

25 ENERGY STORAGE↗
DOE OSTI2022

Koopman-based Differentiable Predictive Control for the Dynamics-Aware Economic Dispatch Problem

The dynamics-aware economic dispatch (DED) problem embeds low-level generator dynamics and operational constraints to enable near real-time scheduling of generation units in a power network. DED produces a more dynamic supervisory control policy than traditional economic dispatch (T-ED) that reduces overall generation costs. However, the incorporation of differential equations that govern the system dynamics makes DED an optimization problem that is computationally prohibitive to solve. In this work, we present a new data-driven approach based on differentiable programming to efficiently obtain offline parametric solutions to the underlying DED problem. In particular, we employ the recently proposed differentiable predictive control (DPC) for offline learning of explicit neural control policies based on identified Koopman operator (KO) model of the system dynamics. We demonstrate the high solution quality and five orders of magnitude computational-time savings of the DPC method over the original optimization-based DED approach on a 9-bus test power grid network.

King, Ethan↗
DOE OSTI2021

Solving the Dynamics-Aware Economic Dispatch Problem with the Koopman Operator

The dynamics-aware economic dispatch (DED) problem embeds low-level generator dynamics and operational constraints to enable near real-time scheduling of generation units in a power network. DED produces a more dynamic supervisory control policy than traditional economic dispatch (T-ED) that reduces overall generation costs. However, in contrast to T-ED, DED is a nonlinear, non-convex optimization problem that is computationally prohibitive to solve. We introduce a machine learning-based operator-theoretic approach for solving the DED problem efficiently. Specifically, we develop a novel discrete-time Koopman Operator (KO) formulation that embeds domain information into the structure of the KO to learn high-fidelity approximations of the generator dynamics. Using the KO approximation, the DED problem can be reformulated as a computationally tractable linear program (abbreviated DED-KO). We demonstrate the high solution quality and computational-time savings of the DED-KO model over the original DED formulation on a 9-bus test system.

King, Ethan↗
DOE OSTI2022

EJFAT: Towards Intelligent Compute Destination Load Balancing

To handle increased data flow, Jefferson Lab (JLab) is partnering with ESnet for development of an AI/ML directed compute work Load Balancer (LB) of UDP streamed data. The LB is FPGA based featuring dynamically configurable, low latency and high throughput destination address switching. The LB provides integration of edge and core computing to support JLab experimental programs, the Electron-Ion Collider, as well as data centers of the future. In the ESnet/JLab FPGA Accelerated Transport (EJFAT) initiative, the function of the LB Data Plane (DP) is to redirect data streams to selectable (but unknown to sender) destination hosts based on current worload and within that host to destination ports as a function of sub- stream id. This effects hierarchical scaling, first across compute machines for processing over a series of events and second, across ports so different data source sub-streams may be assigned to different processors for further parallelization. The LB Control Plane (CP) programs the DP using compute farm telemetry to direct and balance workloads across a compute cluster as the operating conditions require. While Proportional/Integrative/Derivative (PID) controllers are often seen in similar applications, here we investigate the feasibility of a Reinforcement Learning (RL) based schedule manager running in the CP to provide dynamic updates to the DP scheduling policy.

Lawrence, David↗
DOE OSTI2021

Shared automated vehicle fleet operations for first-mile last-mile transit connections with dynamic pooling

Shared automated vehicles (SAVs) have the potential to promote transit ridership by providing efficient first-mile last-mile (FMLM) connections through reduced operational costs to fleet providers and lower out-of-pocket costs to riders. To help plan for a future of integrated mobility, this paper investigates the impacts of SAVs serving FMLM connections, as a mode that provides flexibility in access/egress decisions and is well coordinated with train station schedules. To achieve this objective, a novel dynamic pooling algorithm was introduced to match SAVs with riders while coordinating the riders' arrival times at the light-rail station to a known train schedule. Microsimulations of SAVs and travelers throughout two central Austin neighborhoods show how larger service areas, higher levels of SAV demand, and longer arrival times between successive trains require larger SAV fleet sizes and higher SAV utilization rates to deliver close traveler wait times. Four-person SAVs appear to perform similar to 6-seat SAVs but will cost less to provide. Using a dynamic pooling algorithm tightly coordinated with train arrivals (every 15 min) delivers 87% of travelers to their stations in time to catch the next train, whereas uncoordinated assignments deliver just 58% of travelers in time.

33 ADVANCED PROPULSION SYSTEMS↗
DOE OSTI2025

Dynamical logical qubits in the Bacon-Shor code

The Bacon-Shor code is a quantum error correcting subsystem code composed of weight-2 check operators that admits a single logical qubit, and has distance 𝑑 on a 𝑑×𝑑 square lattice. We show that when viewed as a Floquet code, by choosing an appropriate measurement schedule of the check operators, it can additionally host several dynamical logical qubits. Specifically, we identify a period-4 measurement schedule of the check operators that preserves logical information between the instantaneous stabilizer groups. Such a schedule not only measures the usual stabilizers of the Bacon-Shor code, but also measures and promotes gauge operators of the parent subsystem code to additional temporary stabilizers that protect the dynamical logical qubits against errors. We show that the code distance of these Floquet-Bacon-Shor codes scales as Θ⁢(𝑑/√𝑘) on an 𝑛=𝑑×𝑑 lattice with 𝑘 dynamical logical qubits, along with the logical qubit of the parent subsystem code. Unlike the usual Bacon-Shor code, the Floquet-Bacon-Shor code family introduced here can therefore saturate the subsystem bound 𝑘⁢𝑑=𝑂⁡(𝑛). Moreover, several errors are shown to be self-corrected purely by the measurement schedule itself. This work provides insights into the design space for dynamical codes and expands the known approaches for constructing Floquet codes.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2026

Machine learning-enhanced MPC for demand flexibility in small commercial buildings: An experimental study

Small- and medium-sized commercial buildings (SMCBs) represent the majority of U.S. commercial building stock and a significant share of peak electricity demand, yet they often lack centralized building automation systems, representing a significant untapped resource for urban energy management. This infrastructure gap makes advanced control implementation challenging, limiting the potential for widespread demand flexibility. Model Predictive Control (MPC) has shown strong potential for load shifting, peak demand reduction, and cost savings, but its effectiveness is hindered by unmeasured disturbances such as internal heat gains. This paper presents a Hybrid MPC framework that integrates a physics-based gray-box building thermal model, identified using a lumped disturbance (LD) approach, with a machine learning (ML) model for forecasting unmeasured disturbances. The hybrid approach is designed for buildings with multiple individually controlled heat pump and thermostat pairs, common in SMCBs, and aims to optimize coordinated scheduling of multiple heat pumps under dynamic electricity pricing while respecting comfort constraints. The methodology is validated through both simulations of case study buildings and experimental studies at a highly-instrumented test facility. Simulation results show that the Hybrid MPC achieves substantial load shifting and peak demand reduction, approaching the performance of an ideal MPC with perfect disturbance knowledge, and outperforming a conventional MPC without disturbance forecasting. In experiments, the Hybrid MPC reduced daily HVAC energy costs by 8.7%, peak-price time load (load shifting) by 41.7%, and peak demand by 29.2% compared to baseline control, demonstrating comparable benefits to the 11.6% cost savings, 42.9% load shifting, and 23.2% peak reduction of the ideal MPC. These results demonstrate that the proposed hybrid modeling approach can significantly improve MPC performance in real-world SMCB applications without requiring additional disturbance measurements.

Demand Flexibility↗
DOE OSTI2022

Deadlock prediction via generalized dependency

Deadlocks are notorious bugs in multithreaded programs, causing serious reliability issues. However, they are difficult to be fully expunged before deployment, as their appearances typically depend on specific inputs and thread schedules, which require the assistance of dynamic tools. However, existing deadlock detection tools mainly focus on locks, but cannot detect deadlocks related to condition variables. This paper presents a novel approach to fill this gap. It extends the classic lock dependency to generalized dependency by abstracting the signal for the condition variable as a special resource so that communication deadlocks can be modeled as hold-and-wait cycles as well. It further designs multiple practical mechanisms to record and analyze generalized dependencies. In the end, this paper presents the implementation of the tool, called UnHang. Experimental results on real applications show that UnHang is able to find all known deadlocks and uncover two new deadlocks. Overall, UnHang only imposes around 3% performance overhead and 8% memory overhead, making it a practical tool for the deployment environment.

Zhou, Jinpeng↗
DOE OSTI2023

A Task Based Approach for Co-Scheduling Ensemble Workloads on Heterogeneous Nodes

Scientific workflows consist of multiple, connected applications, with data and results flowing from one to another in a pipeline. Traditionally, such workflows are executed in sequential order, storing intermediate data in storage disks. Co-scheduling application workflows concurrently on the same compute nodes would greatly reduce the cost of moving data to/from storage and allow real-time analysis of intermediate results. Nevertheless, most parallel programming runtimes do not allow seamless integration of various applications in a scientific workflow, in part due to the complexity of managing data and resources. The situation is even more complicated for heterogeneous systems. In this work we extend the Minos Computing Library (MCL) runtime to accelerate pipe-lined and parallel workloads where multiple applications are running in the same system. MCL’s asynchronous task library and runtime dynamically manages resources to allow co-scheduling of multiple processes sharing heterogeneous resources. In addition, we design a custom ex- tension of the Open Compute Language (OpenCL) to enable multiple processes to share device memory. We enable MCL to coordinate these shared buffers to allow for easy, fast data sharing between applications. Using malleable micro-benchmarks and two application workflows that combine scientific simulation and AI-based analysis, we show that our method outperforms traditional approaches.

Index Terms—Parallel systems, Scheduling and Task ↗
DOE OSTI2023

ALTRIOS (Advanced Locomotive Technology and Rail Infrastructure Optimization System) [SWR-22-54]

The Advanced Locomotive Technology and Rail Infrastructure Optimization System (ALTRIOS) is a unique, fully integrated, open-source software tool to evaluate strategies for deploying advanced locomotive technologies and associated infrastructure for cost-effective decarbonization. ALTRIOS simulates freight-demand driven train scheduling, mainline meet-pass planning, locomotive dynamics, train dynamics, energy conversion efficiencies, and energy storage dynamics of line-haul train operations. Because new locomotives represent a significant long-term capital investment and new technologies must be thoroughly demonstrated before deployment, this tool provides guidance on the risk/reward tradeoffs of different technology rollout strategies. An open, integrated simulation tool is invaluable for identifying future research needs and making decisions on technology development, routes, and train selection. ALTRIOS was developed as part of a collaborative effort by a team comprising the National Renewable Energy Laboratory (NREL), University of Illinois Urbana-Champaign (UIUC), Southwest Research Institute (SwRI), and BNSF Railway. Python Package: https://pypi.org/project/altrios/ Rust Crate: https://crates.io/crates/altrios-core

Baker, Chad↗
DOE OSTI2023

Topology, criticality, and dynamically generated qubits in a stochastic measurement-only Kitaev model

Here we consider a paradigmatic solvable model of topological order in two dimensions, Kitaev's honeycomb Hamiltonian, and turn it into a measurement-only dynamics consisting of stochastic measurements of two-qubit bond operators. We find an entanglement phase diagram that resembles that of the Hamiltonian problem in some ways, while being qualitatively different in others. When one type of bond is dominantly measured, we find area-law entangled phases that protect two topological qubits (on a torus) for a time exponential in system size. This generalizes the recently proposed idea of Floquet codes, where logical qubits are dynamically generated by a time-periodic measurement schedule, to a stochastic setting. When all types of bonds are measured with comparable frequency, we find a critical phase with a logarithmic violation of the area law, which sharply distinguishes it from its Hamiltonian counterpart. The critical phase has the same set of topological qubits, as diagnosed by the tripartite mutual information, but protects them only for a time polynomial in system size. Furthermore, we observe an unusual behavior for the dynamical purification of mixed states, characterized at late times by the dynamical exponent z = $\frac{1}{2}$ a superballistic dynamics made possible by measurements.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2025

An Agent-Based Modeling Approach for Spatiotemporal Optimization of Electric Vehicle Fast-Charging Station Demand

With increasing electric vehicle (EV) adoption, managing public fast-charging demand effectively is crucial to avoid grid strain. This study investigates the potential of using dynamic pricing schemes to address this challenge. Presented in this study is a scalable agent-based simulation framework, which is applied to a case study in Richmond, Virginia, that assumes a 50% EV adoption rate in 2040. Two pricing schemes are compared: (1) a dynamic-pricing scheme based on station utilization and (2) a dynamic-pricing scheme based on peak power at the station. These schemes are compared to two baseline scenarios: (1) unscheduled first-come, first-served and (2) scheduled with constant price. The study’s results suggest that dynamic pricing has the potential to influence EV charging behavior, inducing both spatial and temporal shifts, but does so at the cost of inducing inconvenience to EV drivers. The results suggest the peak-power dynamic pricing scheme has the potential to mitigate peak demand pressures on the grid with minimal inconvenience, offering a promising approach for sustainable EV charging infrastructure expansion.

33 - ADVANCED PROPULSION SYSTEMS↗