SEARCH · Search NASA

Results for “distributed load sharing”

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.

SourceResearch connectionFrom yearThrough year

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

At least 55 records · Page 3

All sources NASA NTRS DOE OSTI

DOE OSTI2018

COOLR: A New System for Dynamic Thermal-Aware Computing

Extreme-scale systems need to walk a fine line between the amount of cooling they receive and the thermal-induced performance and reliability degradation they can sustain. System managers are extremely motivated to battle cooling-energy cost: the largest line item in their total operating cost. On the other hand, pushing the nodes to peak performance in tightened cooling regimes places the burden on the dynamic thermal management (DTM) to protect the hardware. DTM schemes throttle performance to relieve heat accumulation within nodes when cooling cannot mitigate the problem. These interventions prevent fatal failures but introduce inevitable performance degradations and variations. Such variations can have drastic consequences on the performance of extreme-scale systems. Furthermore, distribution of heat across different system nodes depends not only on the amount of workload. Even if a strictly equal share of computational load is assigned to all nodes, there are physical attributes and topological features of each system that inherently cause uneven accumulation of heat. These can cause one subcomponent to trigger DTM prematurely and penalize the overall system. Our goal in this project is to create a holistic thermal-aware view of the system, capturing both inherent physical attributes and dynamic system state. We developed COOLR, a dynamic system with the ability to evaluate the interplay between management of computation, data, power dissipation, and thermal state. Thereby, COOLR will achieve higher overall performance at the same energy and cooling cost. First, we performed a systematic power and thermal modeling and of high performance computing architectures. Analytical and empirical techniques were blended to generate a light-weight thermal model of target systems. Second, we developed a thermal-aware OS and runtime system. Policies for managing resources and schedules for processes has been designed to co-optimize the thermal state and performance of the system. The main outcome of our proposed thermal-aware dynamic computing system are scheduling and allocation of processes and memory accesses leading to an appropriately “skewed” distribution of activity across the system. The ultimate result of this new arrangement resulted in reduced occurrences of hotspots and reduced episodes of DTM intervention, leading to improvements in performance. The overall thermal characterization and modeling methodology (thermal instrumentation techniques, model building, and systematic model reduction) resulting from this project can be applied to systems beyond our immediate scope and will be applicable to future scaling. The thermal-aware dynamic computing paradigm will impact the efficiency of extreme-scale systems. The gained “thermal slack” can be given back to further tighten the cooling budget, benefiting the management cost of next generation extreme-scale systems.

97 MATHEMATICS AND COMPUTING↗

NASA NTRSDate not supplied

Updating the Thermal Vacuum Chambers at the NASA Johnson Space Center

Chambers A and B are two large thermal vacuum chambers at Johnson Space Center which enable space simulation for unmanned and human-rated missions, respectively. With the resurgence in deep space missions for scientific research and various private commercial ventures, these chambers are expected to be used frequently for at least the next decade. For qualifying the James Webb Space Telescope, upgrades to Chamber A were performed which included the addition of a 12.5 kW refrigeration system with helium shrouds capable of simulating deep space environment and an efficient and reliable LN2 natural flow thermosiphon system for the thermal shield. Continuous improvements since then have focused on ensuring operational readiness by modernizing the data acquisition, recording, controls, and visualization systems for both chambers and clean room. These upgrades will be the focus of this paper. Controlling the Cryochambers was enhanced by moving from a 32-bit SCADA system to a 64-bit architected system. Infrastructural changes involved installing redundant power circuits, adding new servers, network switches, and including load balancing with fail-over between servers to minimize downtime. Instead of distributed servers, 3 redundant servers are used to share configurations. Configurations are now kept in a shared SQL instance making it easy to deploy and maintain. During this upgrade process many sub-systems (PLCs and NI PXI interfaces to sensors) were upgraded from the prior OPC-DA to the more secure OPC-UA protocol. Cryo-system PLCs were updated to allow Ganni cycle floating pressure calculations from any cold box to be sent to any compressor. During the project, the team recreated over 70,000 live data points, 50000 historical data points, and 4000 alarms. Finally, the graphical interfaces were upgraded to support HTML 5 in conjunction shared pages were implemented reducing the total number of webpages by over 75%. These system updates reinvigorated the previous SCADA system which had reached its end of life. The same look and feel was maintained while providing operators with an updated interface to control, troubleshoot, and record. The new system architecture is more robust and easier to maintain creating a path forward to address remaining problem points and implement additional features.

Cody Schaefer↗

NASA NTRSDate not supplied

Updating the Thermal Vacuum Chambers at the NASA Johnson Space Center

Cody Schaefer↗

DOE OSTI2026

FitCache: A Transparent Drop-In Framework for Multi-Tier Caching to Accelerate Distributed Deep Learning Workloads

Training in Deep learning (DL) remains highly compute- and data-intensive, with I/O becoming a critical bottleneck as models and datasets scale. Recent studies report that data loading can dominate training time, especially on large-scale HPC systems with shared parallel file systems (PFS). Existing caching approaches either rely on single-tier designs or require intrusive modifications to training pipelines, limiting their portability and effectiveness. In this work, we present FitCache, a transparent drop-in framework for multi-tier caching to accelerate distributed DL training by coordinating fast local memory (e.g., DRAM, Persistent Memory (PMem)) and NVMe as hierarchical caches atop PFS. Our design adapts to hardware diversity, i.e., if NVMe is missing, memory transparently acts as a caching tier, ensuring stable performance. FitCache transparently intercepts I/O requests and issues concurrent fetches across all tiers, returning data from the fastest responder without centralized metadata or static redirection paths. FitCache adapts to dynamic workloads and heterogeneous clusters while maintaining POSIX compatibility. Experiments on Frontier (2048 GPUs) and smaller research clusters show that FitCache reduces training time by up to 40% and per-batch I/O latency by up to 71.6% compared to Lustre Orion PFS, offering a drop-in solution for scalable DL training.

Hu, Guangxing [ORNL] (ORCID:0009000283203614)↗

NASA NTRS1989

The role of graphics super-workstations in a supercomputing environment

A new class of very powerful workstations has recently become available which integrate near supercomputer computational performance with very powerful and high quality graphics capability. These graphics super-workstations are expected to play an increasingly important role in providing an enhanced environment for supercomputer users. Their potential uses include: off-loading the supercomputer (by serving as stand-alone processors, by post-processing of the output of supercomputer calculations, and by distributed or shared processing), scientific visualization (understanding of results, communication of results), and by real time interaction with the supercomputer (to steer an iterative computation, to abort a bad run, or to explore and develop new algorithms).

Levin, E.↗

DOE OSTI2020

Learning-Accelerated ADMM for Distributed DC Optimal Power Flow

We suggest a novel data-driven method to accelerate the convergence of Alternating Direction Method of Multipliers (ADMM) for solving distributed DC optimal power flow (DC-OPF) where lines are shared between independent network partitions. Using previous observations of ADMM trajectories for a given system under varying load, the method trains a recurrent neural network (RNN) to predict the converged values of dual and consensus variables. Given a new realization of system load, a small number of initial ADMM iterations is taken as input to infer the converged values and directly inject them into the iteration. We empirically demonstrate that the online injection of these values into the ADMM iteration accelerates convergence by a significant factor for partitioned 14-, 118-and 2848-bus test systems under differing load scenarios. The proposed method has several advantages: it maintains the security of private decision variables inherent in consensus ADMM; inference is fast and so may be used in online settings; RNN-generated predictions can dramatically improve time to convergence but, by construction, can never result in infeasible ADMM subproblems; it can be easily integrated into existing software implementations. While we focus on the ADMM formulation of distributed DC-OPF in this paper, the ideas presented are naturally extended to other distributed optimization problems.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DOE OSTI2021

Learning-Accelerated ADMM for Distributed DC Optimal Power Flow

We propose a novel data-driven method to accelerate the convergence of Alternating Direction Method of Multipliers (ADMM) for solving distributed DC optimal power flow (DC-OPF) where lines are shared between independent network partitions. Using previous observations of ADMM trajectories for a given system under varying load, the method trains a recurrent neural network (RNN) to predict the converged values of dual and consensus variables. Given a new realization of system load, a small number of initial ADMM iterations is taken as input to infer the converged values and directly inject them into the iteration. We empirically demonstrate that the online injection of these values into the ADMM iteration accelerates convergence by a significant factor for partitioned 14-, 118- and 2848-bus test systems under differing load scenarios. The proposed method has several advantages: it maintains the security of private decision variables inherent in consensus ADMM; inference is fast and so may be used in online settings; RNN-generated predictions can dramatically improve time to convergence but, by construction, can never result in infeasible ADMM subproblems; it can be easily integrated into existing software implementations. While we focus on the ADMM formulation of distributed DC-OPF in this paper, the ideas presented are naturally extended to other distributed optimization problems.

alternating direction method of multipliers↗

DOE OSTI2026

Roadmap and Benchmarking: Privacy in Federated Load Forecasting

Data-driven techniques for energy demand forecasting continue to emerge with promising impacts on distribution grid planning. However, the development of robust and generalizable machine learning models requires that representative high quality training data are available. Distributed energy resources have begun to embed intelligence, gathering large amounts of data on customer demand, behavior, and household devices that are connected to the grid. Though utilities aggregate meter-level demand data for load shaping, demand response, outage management, reliability planning, and billing applications, there lies an inherent privacy concern in sharing consumption data that may identify individual consumer behavioral patterns. Hence, while sharing the data is crucial, the private sensitive customer data must be safeguarded from being exposed or manipulated. In this study, we propose a roadmap for implementing a based privacy preserving framework to support the advancement of data-driven analytics in data-sensitive distributed energy resources environments. The roadmap incorporates federated learning–a distributed training framework, differential privacy–a statistical framework that provides guarantees to safeguard the leakage of sensitive data, secure multiparty computation and homomorphic encryption– techniques for encrypting model gradients and applying secure aggregation on the server. Moreover, we perform baseline experiments on the federated short-term load forecasting (STLF) task using open-source residential load profile datasets, offering insights into the challenges of integrating differential privacy into federated learning.

Abebe, Waqwoya [Oak Ridge National Laboratory (ORN↗

NASA NTRS1990

Dynamic load-sharing using predicted process resource requirements

Heuristics which use predicted process resource requirements to make scheduling decisions are proposed. Four heuristics are presented. The first two, MINQ and SMPL, employ centralized scheduling and the remaining two, DMINQ and FDMINQ, use distributed scheduling. These heuristics are first compared against random scheduling and then against two conventional heuristics, CENTEX and DISTED, which schedule processes solely based on system state information. Results based on trace-driven simulations show that the proposed centralized heuristics offer significantly improved mean response time and they require fewer status update messages. In experiments using the same status update rates, SMPL response times were, on the average, 22 percent lower than those for CENTEX; MINQ response times were, on the average, 18 percent lower. The simulations also showed that MINQ and SMPL can perform as well as, or better than, CENTEX while using up to 70 percent fewer status update messages. The use of fewer status update messages imposes less overhead on the system. The use of prediction for distributed scheduling produced similar results. When prediction was used to filter small processes and execute them locally a 50 percent improvement in response times was obtained.

Goswami, Kumar K.↗

DOE OSTI2022

Harmonic Modeling, Data Generation and Analysis of Power Electronics-Interfaced Residential Loads

The share of electronics-based residential load is expected to rise as devices such as variable frequency drives (VFDs), electric vehicle chargers, and inverter-based distributed energy resources (DERs), e.g., photovoltaic (PV) systems become more common. These loads may introduce significant harmonics into power networks that need to be closely studied in order to perform accurate load modeling and forecasting. However, it can be difficult to obtain harmonic-rich voltage and current data - necessary for identifying accurate load models - for residential electrical loads. Recognizing this need, we identify and model a set of electronics-based end-use loads and DERs in an electromagnetic transients program (EMTP) tool for a residence with a single- phase split-phase supply. Further, a procedure is developed to model harmonic interactions between end-use loads connected to the same non-ideal supply voltage in a residential setting. Finally, a harmonic-rich dataset produced via the proposed procedure is utilized to identify frequency coupling matrix (FCM) based load model. Numerical results demonstrate the accuracy of the model, and explore model identifiability with limited data points.

harmonics, power quality, load modeling↗

DOE OSTI2024

Community-Based Transactive Coordination Mechanism for Enabling Grid Edge Systems

The changing landscape of the electricity industry, characterized by a surge in distributed energy resources (DERs) and proactive customers, necessitates practical solutions for coordinated operations especially at the distribution-level. This paper introduces a community-based transactive coordination mechanism designed to incentivize customers for providing localized and system-level services reflected through real-time prices. The work presents a bidding approach for communities with DERs, such as solar photovoltaic (PV) and battery energy storage systems (BESS), to formulate their price-responsiveness for retail energy coordination, emphasizing a community-centric model. By sending bidding curves to a third-party, the mechanism enables customers with DER assets to actively participate in localized coordination with the load serving entity (LSE), thereby supplementing each other’s and even the utilities needs through a shared energy economy. The proposed transactive mechanism is implemented leveraging a co-simulation framework that integrates a distribution grid simulator and Python-based agents for performance evaluation. Collaboration with a local utility to access real distribution feeder models and consumption profiles yields simulation results demonstrating the potential to reduce costs by 12\% for communities with DERs like PV and BESS.

Community-based coordination, grid-edge systems, R↗

DOE OSTI2026

U.S. Distribution Transformer Demand Phase III - Key Drivers and Managing Demand [Slides]

This presentation demonstrates a significant analysis, on forecasting the demand for distribution transformers. The analysis is conducted for the United States, estimating the initial in-service capacity of these assets, and forecasting demand for these assets through 2050 with several sensitivities conducted. It examines not only demand for these assets, but importantly, how utility planning practices can impact the demand for theses assets in time. Under load growth scenarios, whether utilities practice like-for-like replacement strategies as failures occur, or whether they practice proactive up-sizing, anticipating electric load growth, can have major impacts on future demand. It also examines several other growth factors, such as the increasing demand for step-up transformers, which share many of the same characteristics as distribution transformers, and the demand for specific transformers for large project growth from data centers.

24 POWER TRANSMISSION AND DISTRIBUTION↗

NASA NTRSDate not supplied

Establishing a Lunar Surface Power Grid

The National Aeronautics and Space Administration (NASA) is seeking to develop a lunar surface power system for its Artemis missions as well as for planned future commercial lunar surface activities. This power system is composed of a number of assets which contain electrical loads and many of these contain energy resources. These assets can be connected to form a grid, which allows power to be shared between the system’s distributed energy resources. This grid can provide backup power to existing assets as well as primary power to future assets. A trade study was conducted for an Artemis scale grid, which has identified 3 kV AC as an ideal transmission voltage. Development work has also begun for a universal modular interface converter (UMIC) which connects assets and microgrids operating at 120 V DC space power to the grid. Products from this grid definition and UMIC development work will enable highly reliable and available power for Artemis and future planetary surface missions, which is key to achieving NASA’s long term vision of space exploration.

Planetary surface power↗

DOE OSTI2023

Investigation on Tuning Power–Frequency Droop for Improved Grid–Forming Inverter and Synchronous Generator Transient Load Sharing

Poor transient load sharing has been observed during the parallel operation of synchronous generators (SGs) and droop-controlled grid-forming (GFM) inverter-based resources (IBRs) in islanded microgrids. This can result in overcurrent protections, causing the inverters to trip, which leads to the collapse of voltage and frequency. In this article, small-signal analysis supported by electromagnetic transients (EMTs) simulation is performed on a detailed model of a microgrid containing IBRs and an SG to identify underdamped modes and their sensitivities to the power–frequency (P-F) droop parameter. Further EMT simulation of a more complex microgrid model with multiple GFM IBRs and SGs is performed to determine whether trends identified from the modal analysis extend to cases where the number of SGs in operation may vary. The objective is to investigate the effectiveness of tuning this parameter to reduce the likelihood of overcurrents in response to load and generation disturbances. The work uncovers both fast and slow modes of concern. Primarily, the findings show that increasing the P-F droop value improves the damping of the slower mode while degrading the damping of the faster mode, leading to an upper limit on the damping improvement of the GFM IBR output current. In EMT simulations of the more complex microgrid, the greatest damping improvement still exhibits significant maximum overshoot. Furthermore, the optimal value of the P-F droop parameter is sensitive to the number of SGs in operation. The conclusion is that tuning the P-F droop parameter is not an effective means to avoid the overcurrent tripping of GFM IBRs during response to large load or generation disturbances when operating in parallel to SGs in islanded microgrids.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DOE OSTI2021

A Novel Multiagent Resource Sharing Algorithm for Control of Advanced Energy Systems

This paper implements a novel resource sharing control strategy on a fuel cell–gas turbine hybrid power system at the National Energy Technology Laboratory’s Hybrid Performance Facility (Hyper). In a fuel cell–gas turbine hybrid power system, the simultaneous interaction of the gas turbine and the fuel cell creates a tightly coupled environment characterized by conflicting dynamics. In this paper, a model-free control approach is applied to solve the tightly coupled control problem posed by this challenging environment. Specifically, this control problem is presented as a resource sharing problem that can be solved using a resource sharing algorithm that is defined based on the distribution construction concept. This algorithm creates computational agents and solves the problem through the distribution and redistribution of shared resources defined as blocks. Furthermore, two agents were created; the first agent (agent 1) controls the gas turbine speed by adjusting the electric load, and the second agent (agent 2) controls the cathode mass flow through the fuel cell using the cold-air bypass valve. A parametric study was performed over the course of 15 experimental tests for both agents 1 and 2 through an evaluation of the responses based on setpoint changes. The algorithm was shown to have behavior comparable to a previously implemented multi-input multioutput state-space controller, which was designed through a model-based control approach. The resource sharing algorithm was able to find stable performance during run-time operations without any prior system knowledge identification on the power plant and without creating models used in traditional control strategies.

25 ENERGY STORAGE↗

DOE OSTI2026

Safe Deep Reinforcement Learning for Robust Frequency and Voltage-Constrained Networked Microgrid Restoration

Here, this paper proposes a safe soft actor-critic reinforcement learning (RL) algorithm–based controller for networked microgrid restoration. It formulates the post black-start start as a finite-horizon constrained Markov decision process. The RL agent co-optimizes real and reactive power set-points for both grid-forming and grid-following inverters under explicit voltage and frequency constraints, while enforcing proper power sharing via the Mean Active Power Sharing Index (MPSI) and Mean Reactive Power Sharing Index (MQSI). Numerical results obtained on the IEEE 123-bus distribution system show that the proposed method achieves a mean voltage build-up time of 0.01 s without breaching the 5% sharing-violation budget under various load scenarios, considering MPSI and MQSI indices. These findings demonstrate that the proposed method yields fast and safe black-start schedules without resorting to heuristic penalties.

Selim, Alaa [Dartmouth College, Hanover, NH (Unite↗

NASA NTRS1980

Thermal elastohydrodynamic lubrication of spur gears

An analysis and computer program called TELSGE were developed to predict the variations of dynamic load, surface temperature, and lubricant film thickness along the contacting path during the engagement of a pair of involute spur gears. The analysis of dynamic load includes the effect of gear inertia, the effect of load sharing of adjacent teeth, and the effect of variable tooth stiffness which are obtained by a finite-element method. Results obtained from TELSGE for the dynamic load distributions along the contacting path for various speeds of a pair of test gears show patterns similar to that observed experimentally. Effects of damping ratio, contact ratio, tip relief, and tooth error on the dynamic load were examined. In addition, two dimensionless charts are included for predicting the maximum equilibrium surface temperature, which can be used to estimate directly the lubricant film thickness based on well established EHD analysis.

Wang, K. L.↗

DOE OSTI2018

Simultaneous Switching of Multiple GaN Transistors in a High-Speed Switch

broadband travelling wave kicker operating with 80 MHz repetition rates is required for the new PIP-II accelerator at Fermilab. We present a technique to drive simultaneously four series-connected enhancement mode GaN-on-silicon power transistors by means of microwave photonics techniques. These four transistors are arranged into a high voltage and high repetition rate switch. Using multiple transistors in series is required to share switching losses. Using a photonic signal distribution system is required to achieve precise synchronization between transistors. We demonstrate 600 V arbitrary pulse generation into a 200 Ohm load with 2 ns rise/fall time. The arbitrary pulse widths can be adjusted from 4 ns to essentially DC.

Frolov, Daniil↗