Search NASA⌕ Search

SEARCH · Search NASA

Results for “distributed control and optimization”

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 1,657 records · Page 92

A Fast Dynamic Internal Predictive Power Scheduling Approach for Power Management in Microgrids: Preprint

This paper presents a Dynamic Internal Predictive Power Scheduling (DIPPS) approach for optimizing power management in microgrids, particularly focusing on external power exchanges among diverse prosumers. DIPPS utilizes a dynamic objective function with a time-varying binary parameter to control the timing of power transfers to the external grid, facilitated by efficient usage of energy storage for surplus renewable power. The microgrid power scheduling problem is modeled as a mixed-integer nonlinear programming (MINLP-PS) and subsequently transformed into a mixed-integer linear programming (MILPPS) optimization through McCormick's relaxation to reduce computational complexity. A predictive window window with 6 data points is solved at an average of 0.92s, a 97.6% improvement over the 38.27s required for the MINLP-PS formulation, implying the numerical feasibility of the DIPPS approach for real-time implementation. Finally, the approach is validated against a static objective using real-world load data across three case studies with different time-varying parameters, demonstrating the ability of DIPPS to optimize power exchanges and efficiently utilize distributed resources while shifting the external power transfers to specified time durations.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Hydrogen Dispersion Modeling for Development of Smart Distributed Monitoring

Studying hydrogen dispersion is crucial for ensuring the safe and effective deployment of hydrogen as an energy carrier. This study presents a comprehensive CFD modeling framework for simulating hydrogen dispersion at a real-world hydrogen production, storage, and utilization facility. Utilizing the Hydrogen Research Facility under the Advanced Research on Integrated Energy Systems (ARIES) at the National Renewable Energy Laboratory's (NREL) Flatirons campus, controlled hydrogen releases at 27 kg-H2/hr were simulated. The model incorporated site-specific atmospheric conditions, including hourly wind speeds and temperatures recorded between 8 AM and 8 PM from October to December 2023. To reduce computational demands, a statistical reduction technique was applied to condense the dataset to 100 representative scenarios, validated by statistical tests for wind speeds and power law coefficients. Simulations were conducted using the Reynolds-Averaged Navier-Stokes equations. Results demonstrated that wind speed substantially influences hydrogen dispersion, with low wind conditions forming concentrated clouds and higher wind speeds stretching the plume. Additionally, clustering analysis informed optimal sensor placement at various elevations with up to 10 sensor locations on each elevation. This framework offers a robust approach for understanding hydrogen behavior in ambient conditions and informing detection strategies.

08 HYDROGEN↗

Mastering Uncertainty and Risk at Multiple Time Scales in the Future Electrical Grid

Today's electrical grids enjoy a relatively clean separation of spatio-temporal scales yielding a compartmentalization of grid design, optimization, control and risk assessment allowing for the use of conventional mathematical tools within each area. In contrast, the future grid will incorporate time-intermittent renewable generation, operate via faster electrical markets, and tap the latent control capability at finer grid modeling scales; creating a fundamentally new set of couplings across spatiotemporal scales and requiring revolutionary advances in mathematics techniques to bridge these scales. One example is found in decade-scale grid expansion planning in which today's algorithms assume accurate load forecasts and well-controlled generation. Incorporating intermittent renewable generation creates fluctuating network flows at the hourly time scale, inherently linking the ability of a transmission line to deliver electrical power to hourly operational decisions. New operations-based planning algorithms are required, creating new mathematical challenges. Spatio-temporal scales are also crossed when the future grid's minute-scale fluctuations in network flows (due to intermittent generation) create a disordered state upon which second-scale transient grid dynamics propagate effectively invalidating today's on-line dynamic stability analyses. Addressing this challenge requires new on-line algorithms that use large data streams from new grid sensing technologies to physically aggregate across many spatial scales to create responsive, data-driven dynamic models. Here, we sketch the mathematical foundations of these problems and potential solutions.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Implementation of a Publish-Subscribe Protocol in Microgrid Islanding and Resynchronization with Self-Discovery

Communications systems and protocols are becoming second nature to utilities operating distribution systems. Traditionally, centralized communication approaches are often used, while recently in microgrid applications, distributed communication and control schema emerge offering several advantages such as improved system reliability, plug-and-play operation and distributed intelligence. Still, operation and control of microgrids including distributed communication schema have been less of a discussion in the literature. To address the challenge of multiple-inverter microgrid synchronization, a publish-subscribe protocol based, Data Distribution Service (DDS), communication schema for microgrids is proposed in this paper. The communication schema is discussed in details for individual devices such as generators, photovoltaic systems, energy storage systems, microgrid point of common coupling switch, and supporting applications. In conclusion, islanding and resynchronization of a microgrid are demonstrated on a test-bed utilizing this schema.

25 ENERGY STORAGE↗

Chance Constrained Optimal Power Flow: Risk-Aware Network Control under Uncertainty

When uncontrollable resources fluctuate, Optimum Power Flow (OPF), routinely used by the electric power industry to re-dispatch hourly controllable generation (coal, gas and hydro plants) over control areas of transmission networks, can result in grid instability, and, potentially, cascading outages. This risk arises be cause OPF dispatch is computed without awareness of major uncertainty, in particular fluctuations in renewable output. As a result, grid operation under the OPF with renewable variability can lead to frequent conditions where power line flow ratings are significantly exceeded. Such a condition, which is borne by simulations of real grids, would likely resulting in automatic line tripping to protect lines from thermal stress, a risky and undesirable outcome which compromises stability. Smart grid goals include a commitment to large penetration of highly fluctuating renewables, thus calling to reconsider current practices, in particular the use of standard OPF. Our Chance Constrained (CC) OPF corrects the problem and mitigates dangerous renewable fluctuations with minimal changes in the current operational procedure. Assuming availability of a reliable wind forecast parameterizing the distribution function of the uncertain generation, our CC-OPF satisfies all the constraints with high probability while simultaneously minimizing the cost of economic re-dispatch. CC-OPF allows efficient implementation, e.g. solving a typical instance over the 2746-bus Polish network in 20s on a standard laptop.

Energy Planning, Policy, & Economy(29)↗

Distributed Coordination of Energy Storage with Distributed Generators

With a growing emphasis on energy efficiency and system flexibility, a great effort has been made recently in developing distributed energy resources (DER), including distributed generators and energy storage systems. This paper first formulates an optimal coordination problem considering constraints at both system and device levels, including power balance constraint, generator output limits, storage energy and power capacity and charging/discharging efficiencies. An algorithm is then proposed to dynamically and automatically coordinate DERs in a distributed manner. With the proposed algorithm, the agent at each DER only maintains a local incremental cost and updates it through information exchange with a few neighbors, without relying on any central decision maker. Simulation results are used to illustrate and validate the proposed algorithm.

Distributed control↗

High-Performance All-Polymer Solar Cells and Photodetectors Enabled by a High-Mobility n-Type Polymer and Optimized Bulk-Heterojunction Morphology

All-polymer solar cells (all-PSCs) and all-polymer photodetectors (all-PPDs) are promising for application in flexible electronics, but their performance is greatly limited by the lack of polymer acceptors and the difficulties in morphology control. Herein, we report the use of a high-mobility n-type polymer semiconductor, PNDI-DTBT, for fabricating high-performance all-PSCs and all-PPDs. By pairing PNDI-DTBT with a matched polymer donor, a prominent power conversion efficiency of 8.5% was obtained in all-PSCs, demonstrating the great potential of high-mobility n-type polymers with a conventional donor–acceptor skeleton for photovoltaic application. Moreover, all-PPDs with a low dark current density of 1.32 × 10 –8 A cm –2 at –0.1 V bias and a high specific detectivity of 4.77 × 10 12 Jones were achieved, which belong to the best results of organic photodetectors. Morphology investigations revealed that the formation of a bicontinuous interpenetrating network with optimal phase separation scale, high domain purity, and preferential vertical composition distribution in the bulk-heterojunction active layer contribute to the remarkable device performance. Furthermore, these results suggest that combining high-mobility n-type polymers and morphology optimization is fruitful to achieving high-performance all-PSCs and all-PPDs.

36 MATERIALS SCIENCE↗

Photovoltaic frequency–watt curve design for frequency regulation and fast contingency reserves

When renewable energy resources are installed in electricity grids, they typically increase generation variability and displace thermal generator control action and inertia. Grid operators combat these emerging challenges with advanced distributed energy resource (DER) functions to support frequency and provide voltage regulation and protection mechanisms. This paper focuses on providing frequency reserves using autonomous IEC TR 61850-90-7 pointwise frequency-watt (FW) functions that adjust DER active power as a function of measured grid frequency. The importance of incorporating FW functions into a fleet of photovoltaic (PV) systems is demonstrated in simulation. Effects of FW curve design, including curtailment, deadband, and droop, were analyzed against performance metrics using Latin hypercube sampling for 20%, 70%, and 120% PV penetration scenarios on the Hawaiian island of Lanai. Finally, to understand the financial implications of FW functions to utilities, a performance function was defined based on monetary costs attributable to curtailed PV production, load shedding, and generator wear. An optimization wrapper was then created to find the best FW function curve for each penetration level. Lastly, it was found that in all cases, the utility would save money by implementing appropriate FW functions.

14 SOLAR ENERGY↗

Deep Reinforcement Scheduling of Energy Storage Systems for Real-time Voltage Regulation in Unbalanced LV Networks with High PV Penetration

The ever-growing higher penetration of distributed energy resources (DERs) in low-voltage (LV) distribution systems brings both opportunities and challenges to voltage support and regulation. This paper proposes a deep reinforcement learning (DRL)-based scheduling scheme of energy storage systems (ESSs) to mitigate system voltage deviations in unbalanced LV distribution networks. The ESS-based voltage regulation problem is formulated as a multi-stage quadratic stochastic program, with the objective of minimizing the expected total daily voltage regulation cost while satisfying operational constraints. While existing voltage regulation methods are mostly focused on onetime- step control, this paper explores a day-horizon systemwide voltage regulation problem. In other words, the size of action and state spaces are extremely high-dimensional and need to be delicately handled. Furthermore, in order to overcome the difficulty of modeling uncertainties and develop a realtime solution, a learn-to-schedule feedback control framework is proposed by adapting the problem to a model-free DRL setting. The proposed algorithm is tested on a customized 6-bus system and a modified IEEE 34-bus system. Simulation results validate the effectiveness and near-optimality of voltage regulation by ESS in comparison with a deterministic quadratic program solution.

Wang, Shengyi↗

Physical Constraints on Geologic CO2 Sequestration in Low-Volume Basalt Formations

Deep basalt formations within large igneous provinces have been proposed as target reservoirs for carbon capture and sequestration on the basis of favorable CO2-water-rock reaction kinetics that suggest carbonate mineralization rates on the order of 102–103 d. Although these results are encouraging, there exists much uncertainty surrounding the influence of fracture-controlled reservoir heterogeneity on commercial-scale CO2 injections in basalt formations. This work investigates the physical response of a low-volume basalt reservoir to commercial-scale CO2 injections using a Monte Carlo numerical modeling experiment such that model variability is solely a function of spatially distributed reservoir heterogeneity. Fifty equally probable reservoirs are simulated using properties inferred from the deep eastern Snake River Plain aquifer in southeast Idaho, and CO2 injections are modeled within each reservoir for 20 yr at a constant mass rate of 21.6 kg s–1. Results from this work suggest that (1) formation injectivity is generally favorable, although injection pressures in excess of the fracture gradient were observed in 4% of the simulations; (2) for an extensional stress regime (as exists within the eastern Snake River Plain), shear failure is theoretically possible for optimally oriented fractures if Sh is less than or equal to 0.70SV; and (3) low-volume basalt reservoirs exhibit sufficient CO2 confinement potential over a 20 yr injection program to accommodate mineral trapping rates suggested in the literature.

99 GENERAL AND MISCELLANEOUS↗

NVMalloc: Exposing an Aggregate SSD Store as a Memory Partition in Extreme-Scale Machines

DRAM is a precious resource in extreme-scale machines and is increasingly becoming scarce, mainly due to the growing number of cores per node. On future multi-petaflop and exaflop machines, the memory pressure is likely to be so severe that we need to rethink our memory usage models. Fortunately, the advent of non-volatile memory (NVM) offers a unique opportunity in this space. Current NVM offerings possess several desirable properties, such as low cost and power efficiency, but suffer from high latency and lifetime issues. We need rich techniques to be able to use them alongside DRAM. In this talk, we present a novel approach for exploiting NVM as a secondary memory partition so that applications can explicitly allocate and manipulate memory regions therein. More specifically, we propose an NVMalloc library with a suite of services that enables applications to access a distributed NVM storage system. We have devised ways within NVMalloc so that the storage system, built from compute node-local NVM devices, can be accessed in a byte-addressable fashion using the memory mapped I/O interface. Our approach has the potential to re-energize out-of-core computations on largescale machines by having applications allocate certain variables through NVMalloc, thereby increasing the overall memory capacity available. Our evaluation on a 128-core cluster shows that NVMalloc enables applications to compute problem sizes larger than the physical memory in a cost-effective manner. It can bring more performance/efficiency gain with increased computation time between NVM memory accesses or increased data access locality. In addition, our results suggest that while NVMalloc enables transparent access to NVM-resident variables, the explicit control it provides is crucial to optimize application performance.

99 GENERAL AND MISCELLANEOUS↗

In Situ Sample Environment for Studying Corrosion of Nuclear Materials

Currently there is lack of fundamental understanding of kinetics and reaction mechanisms that occur at nuclear cladding-steam interface under extreme environments in the nuclear reactors. Such understanding is vital to design and develop advanced cladding materials for more economical and safer nuclear power. We present advanced in situ sample environment for probing solid-fluid interfaces using high resolution synchrotron methods for studying early stages of corrosion including development of oxide layer as corrosion proceeds. Both X-ray diffraction and X-ray fluorescence methods will be discussed to probe structural and elemental composition changes under steam environment using corrosion resistant cell at high temperatures. High intensity and brightness of synchrotron coupled with advanced in situ sample environment developed, presents an opportunity to understand the fundamental interfacial processes involved in degradation of materials. The in situ sample environment is optimized for XRD and XRF data collection at NSLS-II and will be available to all users through the proposal process. The in situ sample environment is built to study the interfacial interactions at temperatures and pressures up to 400 deg. C and 1500 psi. It consists of three main parts; the distribution manifold, the pressure vessel, and the reaction cell as shown in Figure 1. Water is allowed to heat up to 350 deg. C inside the pressure vessel to produce steam which is delivered to reaction cell using insulated delivery system. The cell design is adapted from J. Diefenbacher, et. al. However, it is modified for corrosion applications in steam environments. The reaction cell and sample holder are made from hastelloy due to its high corrosion resistance. The reaction cell is surrounded by copper heating block to control its temperature independently from the rest of the system. Lab view based platform is developed for remote monitoring of pressure and temperature. The sample environment was commissioned at 6-IDD beamline in the Advanced Photon Source, for grazing incidence XRD corrosion studies. Metallic Zr sample (200 μm thickness) was used due to its faster oxidation kinetics as compared to Zircaloy. Sample was exposed to steam at 350 deg. C and 50 psi for a total time of 64 hours. Results showed the presence of α-Zr as the dominant phase. After 32 hours of exposure to steam, three oxide peaks in the low diffraction angle side started to develop with time. These peaks are attributed to the growth of substoichiometric zirconium oxide phases (Zr{sub 3}O and Zr{sub 3}O{sub 1-x}). (authors)

36 MATERIALS SCIENCE↗

Process and feedstock driven microstructure for laser powder bed fusion of 316L stainless steel

Here in the pursuit of improving additively manufactured (AM) component quality and reliability, fine-tuning critical process parameters such as laser power and scan speed is a great first step toward limiting defect formation and optimizing the microstructure. However, the synergistic effects between these process parameters, layer thickness, and feedstock attributes (e.g. powder size distribution) on part characteristics such as microstructure, density, hardness, and surface roughness are not as well-studied. In this work, we investigate 316L stainless steel density cubes built via laser powder bed fusion (L-PBF), emphasizing the significant microstructural changes that occur due to altering the volumetric energy density (VED) via laser power, scan speed, and layer thickness changes, coupled with different starting powder size distributions. This study demonstrates that there is not one ideal process set and powder size distribution for each machine. Instead, there are several combinations or feedstock/process parameter ‘recipes’ to achieve similar goals. This study also establishes that for equivalent VEDs, changing powder size can significantly alter part density, GND density, and hardness. Through proper parameter and feedstock control, part attributes such as density, grain size, texture, dislocation density, hardness, and surface roughness can be customized, thereby creating multiple high-performance regions in the AM process space.

316L stainless steel↗

Mechanically activated and deactivated ion transport across nanopores with heterogeneous surface charge distributions

To mimic the intricate and adaptive functionalities of biological ion channels, electrohydrodynamic ion transport has been studied extensively, albeit mostly, across uniformly charged nanochannels. Here, we analyze the ion transport under coupled electric field and pressure across heterogeneously charged nanopores with oppositely charged sections on their lateral surface. We only consider such pores with symmetric hourglass-like and cylindrical shapes to focus on the effects of the non-uniform surface charge distribution. Finite-element simulations of a continuum model demonstrate that a pressure applied in either direction of the pore-axis equally suppresses or amplifies the ionic conductance, depending on the electric field polarity, by distorting the quasi-static distribution of ions in the pore. The resulting anomalous mechanical deactivation and activation of ionic current under opposite voltage biases exhibit the functional modularity of our setup, while their intensities are highly tunable, substantially greater than those of analogous behaviors in other nanochannels, and fundamentally correlated to ionic current rectification (ICR) in our pores. A detailed study of ICR subsequently reveals its counterintuitive non-monotonous variations, in the pores, with the magnitude of applied voltage and the pore length, that can help optimize their diode-like behavior. We further illustrate that while the hourglass-shaped nanopores yield the more efficient mechanical suppressors of ion transport, their cylindrical analogs are the superior rectifiers and mechanical amplifiers of ion conduction. Therefore, this article provides a blueprint for the strategic design of nanofluidic circuits to attain a robust, modular, and tunable control of ion transport under external electrical and mechanical stimuli.

Physics↗

Evaluation of Hydrogen Production Feasibility for a Light Water Reactor in the Midwest

Increased electricity production from renewable energy resources coupled with low natural gas prices has caused existing light-water reactor (LWRs) to experience ever diminishing returns from the electricity market. Via partnership among Idaho National Laboratory (INL), The National Renewable Energy Laboratory (NREL), Argonne National Laboratory (ANL), Exelon, and Fuel Cell Energy a techno-economic analysis on the viability of retrofitting existing pressurized water reactors (PWRs) to produce hydrogen (H2) via high temperature steam electrolysis (HTSE) has been conducted. Such integration would allow nuclear facilities to expand into additional markets that may be more profitable in the long-term. To accommodate such an integration, a detailed analysis of HTSE process operation, requirements, and flexibility was conducted. Technical analysis includes proposed nuclear system control scheme modifications to allow dynamic operation of the HTSE via both thermal and electrical connection to the nuclear plant. High fidelity Modelica simulations showcase the viability of such control schemes. However, due to limited knowledge of solid oxide fuel cell (SOFC) stack degradation due to thermal gradients; thermal cycling of the HTSE was not included. Therefore, the control schemes proposed are only utilized to re-distribute steam at startup and only the portion of electricity utilized in the electrolyzers is cycled. From the detailed analysis of the nuclear integration and the HTSE process design, a comprehensive cost estimation was conducted in the APEA and H2A models to elucidate capital and operational costs associated with the production, compression, and distribution of hydrogen from a nuclear facility. Alongside this costing analysis, market analyses were conducted by NREL and ANL on the electric and hydrogen markets respectively in the PJM interconnect. Utilizing the electricity data market projections in the PJM interconnect from NREL and hydrogen demand/pricing projections from ANL a five variable sweep over component capacities, discount rates, and hydrogen pricing was completed using the stochastic framework RAVEN (Risk Analysis Virtual ENVironment) through its resource dispatch plugin HERON (Heuristic Energy Resource Optimization Network). Each combination of variables was evaluated stochastically four times over a seventeen year timespan from 2026-2042 (inclusive) to determine the most economically advantageous solution. Following the five variable sweep an optimization was conducted about the best sweep point to determine optimal component sizing and setpoints. Results suggest positive gain is achievable at all projected hydrogen market pricing levels and at all discount rates. However, exact component sizing and net returns vary based on these values and if incorrect sizing is selected; major net losses can occur. The optimal result occurred with set points as follows: high hydrogen prices, the largest possible HTSE unit in the sweep set at 7.47 kg/sec [645.4 tpd], a contractual hydrogen market agreement 7.29 kg/sec [629.8 tpd], and a hydrogen storage size 115,188 kg. The analysis suggested that with a discount rate of 8%, a ΔNPV = 1.2 billion over the seventeen year span can be achieved. The results illuminate that by operating in multiple markets the nuclear facility is capable of avoiding sale of electricity during times of low electricity market pricing, while maintaining the ability to capitalize on the high electricity market pricing. It should be noted that the analysis conducted in this report is a differential cash flow analysis and as such does not present profit levels. Instead it highlights the net benefit between building and competing in the hydrogen market utilizing nuclear facilities and conducting business as usual in the electricity markets. Additionally, results presented in this report exhibit conservatism due to four key assumptions: Given the limited knowledge on SOFC stack degradation due to thermal gradients the high temperature steam electrolysis plant is not allowed to thermally cycle. This limitation decreases electricity generation capacity in the nuclear plant, reducing capacity payments and the maximum output to the electrical grid. This analysis considers building a separate hydrogen pipeline for use in the nuclear facility. Research is currently being conducted to determine if existing natural gas pipelines can accommodate direct hydrogen injection. If hydrogen can be integrated into existing natural gas pipelines then a capital savings of ~$19,000,000 per kg/sec [1 kg/sec = 86.4 tpd] of installed HTSE capacity could be realized. Fast transients in the electric grid are typically served by flywheels and electric batteries. With electrolysis only operation the nuclear plant can flex its electricity just as quickly. However, for this analysis, the ancillary services market has been neglected. Should the nuclear plant have the ability to operate in the ancillary services market during periods when it is producing hydrogen, an additional substantial income stream that would increase overall system profitability could be considered. Subsidies for non-emitting generation technologies are not introduced in the cash flow analysis, but instead are completed assuming zero subsidies. Profitability is based solely on electricity pricing. Electric pricing is assumed to never go negative. Instead, it is assumed overall system curtailment by the Independent System Operator (ISO) will occur in such a manner that pricing remains at or above zero. This assumption inflates the “business as usual” scenario, thus decreasing the ΔNPV calculated. If negative pricing scenarios continue to occur as they do today, one would expect a further increase in ΔNPV. Overall, results advocate that, through market diversification, nuclear power plants have the potential to substantially increase current profit margins, increase market penetration, and ultimately solidify their place as a mainstay in energy production.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Carbon doping in GeTe drives differences in local structure and properties

Advances in low-power, energy-efficient information storage and computing require understanding and controlling the atomic and nanoscale structures of functional materials, such as phase-change materials. Phase-change memory technology enables nonvolatile, low-power memory in devices by storing information through reversible changes in a phase-change material's atomic structure (i.e., transformations between amorphous and crystalline phases) that have corresponding changes in properties, including electronic resistivity and optical reflectivity. Here, we apply complementary X-ray absorption spectroscopy and X-ray pair distribution function analyses to experimentally identify the local- and medium-range atomic structure differences of GeTe and C-doped GeTe thin films. Upon controlled heating, composition- and temperature-dependent atomic structure evolution in GeTe and C-doped GeTe films shows differences in bonding behavior and local structure that directly influence crystallization onset temperature. We find that the introduction of C interrupts Ge–Ge bonds in amorphous GeTe, altering the as-deposited structure to be more similar to the distorted rocksalt structure of crystalline α–GeTe. The change alters the response of the amorphous atomic structure to heating and also lowers the crystallization onset temperature, from 230 °C in GeTe to 220 °C in the C-doped film. The combined insights from both X-ray techniques provide understanding of structural transformations that enables the development and optimization of next-generation memory and computing materials.

36 MATERIALS SCIENCE↗

NVMalloc: Exposing an Aggregate SSD Store as a Memory Partition in Extreme-Scale Machines

DRAM is a precious resource in extreme-scale machines and is increasingly becoming scarce, mainly due to the growing number of cores per node. On future multi-petaflop and exaflop machines, the memory pressure is likely to be so severe that we need to rethink our memory usage models. Fortunately, the advent of non-volatile memory (NVM) offers a unique opportunity in this space. Current NVM offerings possess several desirable properties, such as low cost and power efficiency, but suffer from high latency and lifetime issues. We need rich techniques to be able to use them alongside DRAM. In this paper, we propose a novel approach for exploiting NVM as a secondary memory partition so that applications can explicitly allocate and manipulate memory regions therein. More specifically, we propose an NVMalloc library with a suite of services that enables applications to access a distributed NVM storage system. We have devised ways within NVMalloc so that the storage system, built from compute node-local NVM devices, can be accessed in a byte-addressable fashion using the memory mapped I/O interface. Our approach has the potential to re-energize out-of-core computations on large-scale machines by having applications allocate certain variables through NVMalloc, thereby increasing the overall memory capacity available. Our evaluation on a 128-core cluster shows that NVMalloc enables applications to compute problem sizes larger than the physical memory in a cost-effective manner. It can bring more performance/efficiency gain with increased computation time between NVM memory accesses or increased data access locality. In addition, our results suggest that while NVMalloc enables transparent access to NVM-resident variables, the explicit control it provides is crucial to optimize application performance.

Computer Science↗

Reinforcement Learning to Enhance Optimal Operation of Resilient Community Energy Systems

This paper presents a novel model-free multi-agent Reinforcement Learning (RL) control method to enhance the resilience of community energy systems in island mode, which coordinates multiple objectives without the necessity of identifying system models that require expert knowledge. Specifically, a community-level coordinator agent is designed to allocate renewable energy resources among different buildings, and multiple building-level agents are developed to optimize load schedules based on limited energy resources and requirements of building loads and occupants’ comfort. In a two-day evaluation, our RL approach demonstrated a similar performance against MPC without requiring system models and formulation of optimization problems as required in MPC.

ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATION↗