Search NASA⌕ Search

SEARCH · Search NASA

Results for “FLOW - CHANNELS”

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 37 records · Page 2
All sourcesNASA NTRSDOE OSTI
DOE OSTI2020

Multidimensional simulations of Mckenna-driven flow tube configuration: Investigating non-ideality in NO x formation flow tube experiments

Multidimensional simulations have been conducted to simulate atmospheric pressure, flat-flame/McKenna-burner-driven-flow tube experiments targeted to obtain NO x speciation data for predicting/analyzing syngas combustion emissions. In a prior work, we demonstrated the impacts of multidimensional transport on post flame region prediction departures from those assuming unidimensional flow/transport conditions. In this work, we develop and utilize a multidimensional laminar reacting flow solver to simulate the fully coupled flame and post flame regions to further elucidate the impacts of the earlier unidimensional modeling assumptions on interpreting post flame NO x experimental data. The model is used to simulate a lean, premixed syngas/air flame and its associated post flame regions within a cylindrical flow-tube-like arrangement. The combustion process takes place under atmospheric condition with trace amount of NO x seeding fed into the inlet gas stream. The spatial evolution of NO x species (NO and NO 2 ) in the flame and in the post-combustion zone suggests two distinct regions: 1) a region encompassing the flame structure itself; and 2) a post flame region in which the temperature decays due to both axial and radial transport processes. The predictions show that for the conditions studied, a pulsatile flow field exists due to the formation of an expanding and contracting recirculation zone in the outer periphery of the flow tube. By resolving the nature of the flow, the resulting time-averaged temperature and species concentrations show improved agreement with existing experimental measurements. The flow-field interaction results in radial inhomogeneities in the NO 2 profiles with the maximum concentration offset from the flow centerline. The location of the peak in NO 2 is coupled with radial temperature gradients from wall cooling effects and their significant influence on NO/NO 2 interconversion kinetics, producing notable NO 2 accumulation in regions near the wall. Geometrical configurations capable of suppressing/minimizing the pulsatile nature are also investigated and the results are compared. Other experimental configurations could be considered in parametric simulations to determine the optimal configuration that would minimize non-idealities in the observations. The work shows the value in performing such computations in advance of settling on a particular design for flow tube/flow reactor experiments.

42 ENGINEERING↗
DOE OSTI2022

Using Ensemble Data Assimilation to Estimate Transient Hydrologic Exchange Flow Under Highly Dynamic Flow Conditions

Abstract Quantifying dynamic hydrologic exchange flows (HEFs) within river corridors that experience high‐frequency flow variations caused by dam regulations is important for understanding the biogeochemical processes at the river water and groundwater interfaces. Heat has been widely used as a tracer to infer steady‐state flow velocities through analytical solutions of heat transport defined by the diurnal temperature signals. Under sub‐daily dynamic flow conditions, however, such analytical solutions are not applicable due to the violation of their fundamental assumptions. In this study, we developed a data assimilation‐based approach to estimate the sub‐daily flux under highly dynamic flow conditions using multi‐depth temperature observations at a 5‐min resolution. If the hydraulic gradient is measured, Darcy's law was used to calculate the flux with permeability estimated from temperature responses below the riverbed. Otherwise, flux was estimated directly by assimilating multi‐depth temperature data at 1‐ or 2‐hr time intervals assuming one‐dimensional flow and heat transport governing equation. By comparing estimated fluxes with model‐generated synthetic truth, we demonstrated that both schemes have robust performance in estimating fluxes under highly dynamic flow conditions. This data assimilation‐based flux estimation method was able to capture the vertical sub‐daily fluxes using multi‐depth high‐resolution temperature data alone, even in the presence of multi‐dimensional flow. This approach has been successfully applied to real field temperature data collected at the Hanford site, which experiences highly dynamic HEFs. Our study shows the promise of adopting distributed 1‐D temperature monitoring to capture spatial and temporal exchange dynamics in river corridors at a watershed scale or beyond.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

Beyond conventional batteries: a review on semi-solid and redox targeting flow batteries-LiFePO{sub 4} as a case study.

Clean and sustainable energy is becoming increasingly crucial to tackle the current energy crisis. However, the intermittent nature of renewable energy sources presents a challenge for their effective implementation. Redox flow batteries (RFBs) have emerged as a promising solution to this problem, as they can help enhance the stability of grid networks and promote the use of renewable energy sources. RFBs are highly modular and scalable systems that can be customized to meet the power and energy requirements of different renewable energy plants. Moreover, they offer several advantages over conventional battery technologies, including cost and safety concerns. However, conventional RFBs have limited energy densities due to the low solubility of their active species in electrolyte. To overcome this limitation, semi-solid (SSRFBs) and redox targeting (RTFBs) flow batteries have been proposed. These systems feature high concentrations of active species and impressive energy densities, making them highly attractive for renewable energy applications. LiFePO4 (LFP) is a highly promising active material for semi-solid and targeting flow batteries. One of the key advantages of LFP is its low raw materials cost, as it is composed of Earth-abundant elements such as iron and phosphorus. This makes it an attractive option for large-scale battery production. The recent developments in SSRFBs and RTFBs using LFP as catholyte hold great promise for the future of sustainable energy storage. The combination of LFP's low cost, safety, durability, and high energy density with the modularity and scalability of flow battery systems make for a compelling solution to the challenges of intermittent renewable energy sources. Ongoing research and development in this area will likely yield even further improvements in the performance and efficiency of LFP-based flow batteries, opening exciting new possibilities for sustainable energy storage.

El Halya, Nabil↗
DOE OSTI2020

Viscous flow properties and hydrodynamic diameter of phenothiazine-based redox-active molecules in different supporting salt environments

We report viscous flow properties of a redox-active organic molecule, N-(2-(2-methoxyethoxy)ethyl)phenothiazine (MEEPT), a candidate for non-aqueous redox flow batteries, and two of its radical cation salts. A microfluidic viscometer enabled the use of small sample volumes in determining viscosity as a function of shear rate and concentration in the non-aqueous solvent, acetonitrile, both with and without supporting salts. All solutions tested show Newtonian behavior over shear rates of up to 30 000 s−1, which was rationalized by scaling arguments for the diffusion-based relaxation time of a single MEEPT molecule without aggregation. Neat MEEPT is flowable but with a large viscosity (412 mPa⋅s at room temperature), which is ∼1000 times larger than that of acetonitrile. MEEPT solutions in acetonitrile have low viscosities; at concentrations up to 0.5 M, the viscosity increases by less than a factor of two. From concentration-dependent viscosity measurements, molecular information was inferred from intrinsic viscosity (hydrodynamic diameter) and the Huggins coefficient (interactions). Model fit credibility was assessed using the Bayesian Information Criterion. It is found that the MEEPT and its charged cations are “flowable” and do not flocculate at concentrations up to 0.5 M. MEEPT has a hydrodynamic diameter of around 8.5 Å, which is almost insensitive to supporting salt and state of charge. This size is comparable to molecular dimensions of single molecules obtained from optimized structures using density functional theory calculations. The results suggest that MEEPT is a promising candidate for redox flow batteries in terms of its viscous flow properties.

Wang, Yilin (ORCID:0000000316102995)↗
DOE OSTI2021

Convex Optimization of Integrated Power-Gas Energy Flow Model With Applications to Probabilistic Energy Flow

Energy flow calculation is a fundamental problem of the integrated power and gas system (IPGS) operation and planning. However, the nonlinear gas flow model introduces major challenges to the energy flow calculation. In this paper, we propose a tractably convex optimization model to solve the energy flow problem in IPGSs. It is demonstrated that the proposed optimization model has the same optimal solution as the original nonlinear steady energy flow model. Also, piecewise linearization is adopted to tightly linearize the nonlinear objective function of the model, which transforms the formulated convex optimization into a linear program one. Thus, the computation complexity of the proposed energy flow model is significantly reduced as compared with the existing methods. In addition, the proposed model can be extended to probabilistic energy flow estimation. Extensive case studies are conducted to validate the effectiveness of the proposed energy flow model using three IPGSs.

42 ENGINEERING↗
DOE OSTI2020

Characterizing NWP Model Errors Using Doppler-Lidar Measurements of Recurrent Regional Diurnal Flows: Marine-Air Intrusions into the Columbia-River Basin

Ground-based Doppler-lidar instrumentation provides atmospheric wind data at dramatically improved accuracies and spatial/temporal resolutions. These capabilities have provided new insights into atmospheric flow phenomena, but they also should have a strong role in NWP model improvement. Insight into the nature of model errors can be gained by studying recurrent atmospheric flows, here a regional summertime diurnal sea breeze and subsequent marine-air intrusion into the arid interior of Oregon-Washington, where these winds are an important windenergy resource. These marine intrusions were sampled by three scanning Doppler lidars in the Columbia-River Basin as part of the Second Wind Forecast Improvement Project (WFIP2), using data from summer 2016. Lidar time-height cross sections of wind speed identified eight days when the diurnal-flow cycle, (peak wind speeds at midnight, afternoon minima) was obvious and strong. Eight-day composite time-height cross sections of lidar wind speeds are used to validate those generated by the operational NCEP-HRRR model. HRRR simulated the diurnal wind cycle, but produced errors in the timing of onset and significant errors due to a premature nighttime demise of the intrusion flow, producing low-bias errors of 6 m s-1 . Day-to-day and in the composite, whenever a marine intrusion occurred, HRRR made these same errors. The errors occurred under a range of gradient wind conditions indicating that they resulted from the misrepresentation of physical processes within a limited region around the measurement locations. Because of their generation within a limited geographical area, field-measurement programs can be designed to find and address the sources of these NWP errors.

Banta, Robert M.↗
DOE OSTI2021

Linear and Nonlinear Solvers for Simulating Multiphase Flow within Large-Scale Engineered Subsurface Systems

Simulation of multiphase flow in the subsurface is well-known to be computationally challenging. While there have been many studies that have explored approaches to overcoming these challenges, they often utilize relatively simple case studies. In this paper, we focus on the unique numerical challenges posed by modeling large-scale engineered subsurface systems, characterized by discrete features embedded in a heterogeneous natural subsurface setting. The man-made features such as shafts, tunnels, and barriers often cause multiple challenges in modeling the domain for multiphase porous media flow. This flow scenario can have a wide range of applications such as nuclear waste repositories, enhanced recovery of a petroleum reservoir, geothermal engineering, and carbon sequestration. An example of these severe numerical challenges is the case of performance assessment (PA) for Waste Isolation Pilot Plant (WIPP), the only operating deep geological repository in the US, which simulates extreme material properties of bedded salt rock formation and extreme contrast due to open excavation next to the formation. The models have extremes not only of permeability and porosity but also of the constitutive models needed for multiphase flow; additionally, they have process models like salt creep closure reducing porosity over time, fracturing in clay and anhydrite interbeds of the bedded salt, gas generation from the waste materials, and unintentional human borehole intrusions in some scenarios. Numerical simulations require the solution of coupled systems of nonlinear PDEs; in our work, we use the open-source simulator PFLOTRAN which is based on Finite Volume discretization. The solution of the nonlinear equations requires use of the Newton-Raphson iteration at each time step, which entails the solution of the linearized Jacobian system at each iteration. The effects of all the processes (i.e., large number of unknowns, highly nonlinear constitutive relations, large contrasts in material properties in short distances) lead to an ill-conditioned Jacobian matrix that severely challenges traditional linear solver, i.e., stabilized biconjugate gradient with block Jacobi incomplete LU preconditioner (BCGS-ILU) leading to non-convergence for traditional Newton-Raphson nonlinear solver causing unacceptably long computation time for each model. This paper presents linear solvers such as constrained pressure residual (CPR) two-stage preconditioner with alternate-block-factorization (ABF) and quasi- implicit pressure and explicit saturation (QIMPES) decouplers and flexible generalized residual solver (FGMRES). The new general-purpose nonlinear solver, Newton trust-region dogleg Cauchy (NTRDC), is also introduced to resolve extreme nonlinearities in the models. We demonstrate the effectiveness of each method relative to the default BCGS-Newton solver. The two best cases had nearly 50 times speed-up and achieved completion of a simulation in 14 hours that never completed due to non-convergence with the default solver. We also investigate the strong scalability of each method and discuss some of the deficiencies found for Block Jacobi preconditioner using parallel domain decomposition, and node packing effects of modern processor architecture.

Preconditioner, Nonlinear, Porous media, Multiphas↗
DOE OSTI2020

i- flow: High-dimensional integration and sampling with normalizing flows

In many fields of science, high-dimensional integration is required. Numerical methods have been developed to evaluate these complex integrals. We introduce the code i-flow, a python package that performs high-dimensional numerical integration utilizing normalizing flows. Normalizing flows are machine-learned, bijective mappings between two distributions. i-flow can also be used to sample random points according to complicated distributions in high dimensions. We compare i-flow to other algorithms for high-dimensional numerical integration and show that i-flow outperforms them for high dimensional correlated integrals. The i-flow code is publicly available on gitlab at https://gitlab.com/i-flow/i-flow.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2024

Development of a Distribution Optimal Power Flow Federate for Open-Source OEDI-SI Platform

Increasing numbers of distributed generators in the electric power distribution networks require developing a control strategy to optimize solutions in real time. Linearized optimal distribution flow development has seen growth and acceptance in the distribution systems literature for efficiently modeling the \glspl{opf} for distribution systems. This paper examines the implementation and integration procedure for linearized optimal distribution flow federate to \gls{oedisi} platform. Specifically, we discuss i) the usage of the \gls{oedisi} platform, ii) obtaining a tractable solution using developed \gls{opf} federate, and iii) validation of solutions and bench-marking the \gls{oedisi} platform with developed \gls{opf} federate using OpenDSS. In brief, we demonstrate how a general linearized optimal distribution flow federate can be developed and integrated with a co-simulation environment to mimic real-world examples. The efficacy of the proposed method is demonstrated using the IEEE 123-bus test system under different scenarios to obtain a tractable solution and compare its results.

Sadnan, Rabayet↗
DOE OSTI2026

Flow pattern and void fraction characterization in nitrogen/water flows through a diamond-type triply periodic minimal surface lattice

Here, this study presents, to the authors knowledge, the first experimental investigation on the void fraction and flow patterns in two-phase flows through a diamond-type Triply Periodic Minimal Surface (TPMS) lattice. An additively manufactured TPMS structure was tested under upward co-current flow of a water/nitrogen mixture. Superficial velocities varied for a total of 42 test conditions (gas: 0.01–2.4 m/s; liquid: 0.01–2.4 m/s; mass flux: 20–2370 kg/m 2 ·s). High-speed video and X-ray imaging enabled time-averaged void fraction measurements and identified six distinct flow regimes which were used to develop a flow pattern map. Comparison of the void fraction data with correlations from literature demonstrated the Rouhani and Axelsson (1970) [49] correlation modified by Steiner (1993) [52] provided the best agreement, which was improved with empirically fit coefficients. This approach predicted the void fraction with errors < ±20% for 64% of the data, with a mean absolute percent deviation (MAPD) of 23%. The measured frictional pressure drop was compared to correlations from literature which captured the observed trends but did not provide good accuracy. The best agreement was found after optimizing the empirical coefficients of the Muller-Steinhagen & Heck (1986) [56] correlation; this approach captured 33% of the data within ±20% with a MAPD of 46% over the full range and captured 71% of the data within ±20% a MAPD of 13.3% at mass fluxes >1100 kg/m 2 -s. These results provide foundational insight into TPMS two-phase flow behavior and inform modeling and design of advanced heat exchange components incorporating TPMS geometries.

42 - ENGINEERING↗
DOE OSTI2021

Liquid–Liquid Microfluidic Flows for Ultrafast 5-Hydroxymethyl Furfural Extraction

We study the effects of liquid–liquid flow patterns on the extraction of 5-hydroxymethyl furfural from the aqueous phase into an organic phase (methyl isobutyl ketone, ethyl acetate, and 2-pentanol) at ultrashort residence times of 0.2–20 s in a coiled capillary microchannel. Slug, droplet, slug-droplet, dispersed, parallel, annular, and irregular two-phase flow patterns are generated. We evaluate the extraction efficiency and the volumetric mass transfer coefficient via experiments and computational fluid dynamics (CFD). The CFD predictions are in quantitative agreement with the experimental data. The convective and diffusive contributions to the mass transfer of different flow patterns are analyzed. For all solvent pairs, the extraction efficiency varies irregularly with increasing flow rate due to changes in the mass transfer mechanism caused by a transition between flow regimes. Overall, we demonstrate >90% extraction efficiencies with high volumetric mass transfer coefficients ranging from 0.006 to 2.17 s–1. Furthermore, a merit index is defined to assess the trade-off between the fast mass transfer and the pressure drop in biphasic microchannels. In general, the mass transfer rate increases significantly at high flow rates at the expense of a high-pressure drop. Furthermore, the ethyl acetate/water irregular flow pattern is optimal due to a large throughput, extraction efficiency, and mass transfer coefficient, with a minimal pressure drop.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2025

Design and Analysis of the Open-Surface Slow Li Flow Divertor and Comparison to the Fast Li Flow Divertor

In the ongoing U.S. project, “Liquid Metal Plasma Facing Components,” sponsored by the U.S. Department of Energy, efforts have been taken to develop two open-surface divertor designs for the Fusion Nuclear Science Facility using liquid lithium (Li) as a heat and particle flux removal media. The main focus of this study is the design and analysis of a slow (~1 mm/s) and thin (<1 mm) open-surface Li flow divertor with a Li-cooled substrate, which is then compared with an earlier design of a fast (up to 10 m/s) and thick (~0.5 cm) Li flow divertor with the substrate cooled with helium. The slow Li flow divertor design is based on the original LiWall concept developed at the Princeton Plasma Physics Laboratory. Such a thin and slow Li layer can remove the particle flux by reducing the recycling flux, while the heat flux is removed mainly through the heat sink located beneath. In the present study, the heat sink is provided through a Li cooling flow inside the substrate of reduced activation ferritic/martensitic steel. By performing a multiphysics analysis with COMSOL that included liquid-metal magnetohydrodynamics (MHD), heat transfer, and structural mechanics, the impact of various factors on the divertor heat removal capability, such as Li flow velocity, MHD effects, and inlet velocity boundary condition, were examined. Based on comparisons of the two divertor designs, it was shown that the fast-flow divertor significantly outperformed the slow-flow design, whose heat removal capability was limited to ~1 to 2 MW/m2.

Jiang, Yuchen [ORNL] (ORCID:0000000283280284)↗
DOE OSTI2021

Particle charge distributions in the effluent of a flow-through atmospheric pressure low temperature plasma

Atmospheric pressure low-temperature plasmas are often utilized to perform particle synthesis, treatment, and removal. It is well-known that dust particles are highly negatively charged in these plasmas; however, little is known about dust particle charging behavior as particles leave the plasma volume and pass through the spatial afterglow region. In this work, monodisperse particles of various sizes and work functions were introduced into an atmospheric pressure radiofrequency capacitively coupled flow-through plasma. Dust particle electrical mobility distributions downstream of the flow-through plasma were measured utilizing a differential mobility analyzer in conjunction with a condensation particle counter at various gas flow velocities. Charge distributions were determined from the measured electrical mobility distributions. Experiments confirm that particles become less negatively charged, and even net-positively charged after leaving the plasma volume, with a distribution that follows a shifted Boltzmann charge distribution. Additionally, particle charge in the effluent of the flow-through plasma is negligibly dependent on work function but highly size and flow velocity dependent. Larger particles were shown to have a higher magnitude of charge under all studied conditions; however, particle polarity was switchable by varying gas flow velocity. The charging dynamics were simulated utilizing a constant number Monte Carlo model that accounts for electron temperature decay and the transition from ambipolar to free diffusion of electrons and ions in the spatial afterglow. Simulation results also suggest that, at the same flow velocity, larger particles obtain a greater magnitude of charge, negative or positive. Furthermore, the decrease in electron mobility and the difference between ion and electron convective loss rates create an ion-rich region in the plasma effluent that promotes ion–particle collisions and drives particle charge removal and even reversal of polarity. Larger particles more favorably collide with energetic species in these environments, which results in higher charge states.

atmospheric pressure low temperature plasma↗
DOE OSTI2021

Flow battery flow field having volume that is function of power parameter, time parameter and concentration parameter

A flow battery includes a cell that has first and second flow fields spaced apart from each other and an electrolyte separator layer. A supply/storage system is external of the cell and includes first and second vessels fluidly connected with the first and second flow fields, and first and second pumps configured to selectively move first and second fluid electrolytes between the vessels and the first and second flow fields. The flow fields each have an electrochemically active zone that is configured to receive flow of the fluid electrolytes. The electrochemically active zone has a total open volume that is a function of at least one of a power parameter of the flow battery, a time parameter of the pumps and a concentration parameter of the fluid electrolytes.

Smeltz, Andrew↗
DOE OSTI2020

Geochemical narrowing of cement fracture aperture during multiphase flow of supercritical CO2 and brine

For carbon capture and storage operations, wells are used as a necessary conduit for injecting CO2 at depth, but they can also act as leakage conduits if the cement seals are compromised. The specific objective of this research was to investigate the nature of self-healing of a fracture within cement under multiphase flow of CO2 and brine, and to compare this to a newly developed model. This was accomplished through a laboratory investigation, flowing reacting supercritical CO2 and brine through a cement fracture with multiphase flow. The influent end of the fracture was degraded as evidenced by the formation of cracks across the surface. At the effluent end of the fracture, the initial aperture of 137 µm was reduced to 50 µm. This reduction by 87 µm was similar to results from a recently developed model tested in this study, which predicted an aperture reduction of 80 µm. Our results demonstrated that the model accurately predicted the reactions which occurred in this multiphase flow system.

wellbore, cement, fracture, multiphase-flow, super↗
DOE OSTI2024

Riemannian Optimization Applied to AC Optimal Power Flow: Preprint

The nonlinear, nonconvex AC optimal power flow problem is of growing importance as the nature of the power grid evolves. This problem can be difficult to solve for interior point methods. However, the advent of optimization algorithms over smooth Riemannian manifolds presents an alternative approach. The nonlinear, nonconvex constraints in the AC power flow problem form an embedded submanifold of Euclidean space. In this paper, the authors explore the performance of Riemannian optimization algorithms for the ACOPF problem where the optimization is performed directly on the AC power flow manifold. They demonstrate that these are viable computational alternatives to interior point methods. This is done by using Julia and the packages PowerModels.jl and Manopt.jl.

manifold optimization↗
DOE OSTI2023

Anisotropic flow and flow fluctuations of identified hadrons in Pb–Pb collisions at $ \sqrt{s_{\textrm{NN}}}$ = 5.02 TeV

The first measurements of elliptic flow of π ± , K ± , p+$\bar{p}$, K$_{S}^{0}$, Λ+$\bar{Λ}$, $\phi$, Ξ – +$\barΞ$ + , and Ω – +$\barΩ$ + using multiparticle cumulants in Pb–Pb collisions at $\sqrt{s_{NN}}$= 5.02 TeV are resented. Results obtained with two- (v 2 {2}) and four-particle cumulants (v 2 {4}) are shown as a function of transverse momentum, p T , for various collision centrality intervals. Combining the data for both v 2 {2} and v 2 {4} also allows us to report the first measurements of the mean elliptic flow, elliptic flow fluctuations, and relative elliptic flow fluctuations for various hadron species. These observables probe the event-by-event eccentricity fluctuations in the initial state and the contributions from the dynamic evolution of the expanding quark–gluon plasma. The characteristic features observed in previous p T -differential anisotropic flow measurements for identified hadrons with two-particle correlations, namely the mass ordering at low p T and the approximate scaling with the number of constituent quarks at intermediate p T , are similarly present in the four-particle correlations and the combinations of v 2 {2} and v 2 {4}. In addition, a particle species dependence of flow fluctuations is observed that could indicate a significant contribution from final state hadronic interactions. The comparison between experimental measurements and CoLBT model calculations, which combine the various physics processes of hydrodynamics, quark coalescence, and jet fragmentation, illustrates their importance over a wide p T range.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2022

Modeling the AC Power Flow Equations with Optimally Compact Neural Networks: Application to Unit Commitment

Nonlinear power flow constraints render a variety of power system optimization problems computationally intractable. Emerging research shows, however, that the nonlinear AC power flow equations can be successfully modeled using neural networks. These neural networks can be exactly transformed into mixed integer linear programs and embedded inside challenging optimization problems, thus replacing nonlinearities that are intractable for many applications with tractable piecewise linear approximations. Such approaches, though, suffer from an explosion of the number of binary variables needed to represent the neural network. Accordingly, this paper develops a technique for training an "optimally compact'' neural network, i.e., one that can represent the power flow equations with a sufficiently high degree of accuracy while still maintaining a tractable number of binary variables. We demonstrate the use of this neural network as an approximator of the nonlinear power flow equations by embedding it in the AC unit commitment problem, transforming the problem from a mixed integer nonlinear program into a more manageable mixed integer linear program. We use the 14-, 57-, and 89-bus networks as test cases and compare the AC-feasibility of commitment decisions resulting from the neural network, DC, and linearized power flow approximations. Our results show that the neural network model outperforms both the DC and linearized power flow approximations when embedded in the unit commitment problem. The neural network formulation most often selects a feasible unit commitment schedule, and furthermore, it only s

AC power flow↗