Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Model Predictive Control”

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 1,225 records · Page 68
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

Data-Driven Approaches for Bead Geometry Prediction Via Melt Pool Monitoring

In the realm of additive manufacturing, the selection of process parameters to avoid over and under deposition entails a time-consuming and resource-intensive trial-and-error approach. Given the distinct characteristics of each part geometry, there is a pressing need for advancing real-time process monitoring and control to ensure consistent and reliable part dimensional accuracy. Here, this research shows that support vector regression (SVR) and convolutional neural network (CNN) models offer a promising solution for real-time process control due to the models’ abilities to recognize complex, non-linear patterns with high accuracy. A novel experiment was designed to compare the performance of SVR and CNN models to indirectly detect bead height from a coaxial image of a melt pool from a single-layer, single bead build. The study showed that both SVR and CNN models trained on melt pool data collected from a coaxial optical camera can accurately predict the bead height with a mean absolute percentage error of 3.67% and 3.68%, respectively.

36 MATERIALS SCIENCE↗
DOE OSTI2022

Development of a Flotation Simulator That Can Predict Grade vs. Recovery Curves from Mineral Liberation Data

Flotation is a kinetic process designed to separate mineral fines by control of hydrophobicity. In general, both the recovery and selectivity of a flotation process increase with contact angle. It is difficult, however, to model flotation kinetics using the thermodynamic parameter. In the present work, we consider flotation as heterocoagulation so that the bubble-particle interactions occurring in the pulp phase of a flotation cell can be modeled using the hydrophobic force as a kinetic parameter that can represent the role of contact angles in flotation. Here, the pulp-phase model developed in this manner has been combined with a froth-phase model developed earlier by considering the effect of particles in film thinning kinetics and bubble coarsening (Park et al, 2018). The combined model has been used to develop a flotation simulator that can be used to construct grade vs. recovery curves from feed characteristics (e.g., mineral liberation and particle size) and various operating parameters. The simulator has been tested successfully against a set of bench-scale continuous flotation test results reported in the literature (dos Santos and Galery, 2018).

58 GEOSCIENCES↗
DOE OSTI2020

Time-Sequenced Flow Field Prediction in an Optima Spark-Ignition Direct-Injection Engine Using Bidirectional Recurrent Neural Network (bi-RNN) with Long Short-Term Memory

To further improve the energy conversion efficiency of internal combustion engine, the transient and complex air flow movement inside the cylinder needs to be better understood and controlled. Although the in-cylinder flow fields are highly stochastic with strong cycle-to-cycle fluctuations, machine learning can still provide an efficient way to learn and regress the complex flow movement process inside the cylinder. In this work, a bidirectional recurrent neural network (bi-RNN) model with long short-term memory was applied to predict the in-cylinder flow fields at different time steps using training data from mull-cycle particle image velocimetry (PIV) measurements. To evaluate the agreement between the true and predicted flow fields, structure and magnitude comparison indices are calculated both globally and locally. The comparison results show that the bi-RNN model can accurately predict the bulk flow and vortex motions from early intake stroke to compression stroke. This work demonstrates that the machine learning model has the potential to predict the underlying dynamics of the interaction between in-cylinder flows and provides a reliable way to improve temporal resolution in PIV flow data to better reveal transient in-cylinder flow features.

Bi-RNN model↗
DOE OSTI2024

Understanding Processes Controlling the Temporal and Spatial Variations of PBL Structures Over the ARM SGP Site

The surface heat, moisture, and momentum fluxes are transferred to the atmosphere above through the planetary boundary layer (PBL), where vertical mixing due to turbulent eddies of different sizes plays critical roles. Therefore, reliably representing PBL processes in numerical models is critical for weather, climate, and air quality prediction. Currently, there are over ten PBL schemes that are selectable within the advanced research version of the Weather Research and Forecasting (WRF) model, indicative of the challenges in capturing the impacts of turbulence within the PBL in models. Further improvements in PBL parameterizations are needed for both weather and climate models, as emphasized in many recent national reports, but require an advanced understanding of the underlying boundary layer processes from observations. This project takes advantage of Department of Energy (DOE) Atmospheric Radiation Measurement (ARM) investments in the atmospheric boundary layer observations and Large-Eddy Simulation (LES) ARM Symbiotic Simulation and Observation (LASSO) simulations to characterize PBL structures, understand key physical processes controlling the mixed layer development, and to evaluate PBL parameterization.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2021

Comparative and Cost Analysis of a Novel Predictive Power Ramp Rate Control Method: A Case Study in a PV Power Plant in Puerto Rico

One of the most important aspects that need to be addressed to increase solar energy penetration is the power ramp-rate control. In weak grids such as the one found in Puerto Rico, it is important to smooth power fluctuations caused by the intermittence of passing clouds. In this work, a novel power ramp-rate control strategy is proposed. Additionally, a comparison with some of the most common power ramp-rate control methods is performed using a proposed model and real solar radiation data from the Coto Laurel photovoltaic power plant located in Ponce, Puerto Rico. The proposed model was validated using one-year real data from Coto Laurel. The power ramp-rate control methods were compared in real-time simulations using the OP5700 from Opal-RT Technologies considering power ramp rate fluctuations, power ramp-rate violations, fluctuations in the state-of-charge, among other indicators. Moreover, the proposed power ramp-rate control strategy, called predictive dynamic smoothing was explained and compared. Results indicate that the predictive dynamic smoothing produced a considerably reduced Levelized Cost of Storage compared to other power ramp-rate control methods and provided a higher lifetime expectancy for lithium batteries.

36 MATERIALS SCIENCE↗
DOE OSTI2020

Atomic-Scale Structural Mapping of Active Sites in Monolayer PGM-Free Catalysts by Low-Voltage 4D-STEM

Two-dimensional (2D) materials have attracted a large amount of attention in both basic and applied fields, and scanning transmission electron microscopy (STEM) is often uniquely well-suited for characterizing the atomic-scale structure of these materials [1-4]. As a result, STEM is poised to significantly impact progress on platinum group metal (PGM)-free catalysts, which are currently under intense development to enable low-cost, commercially viable hydrogen fuel cells [5]. While recent advancements have resulted in fuel cell performance comparable to Pt catalysts by some measures [6], cell durability remains a significant challenge, limiting practical applications [7]. Catalytically active sites in PGM-free materials are proposed to be FeN4 complexes embedded in a graphene lattice (Fig. 1b) within layered or other larger materials, but this is still under debate largely due to the range of potential actives sites predicted by computational methods and lack of methods for directly validating these models [5]. Fundamental insights into the atomic structure and resulting degradation pathways of proposed active sites are therefore needed to fully understand and control cell performance and durability [6].2D materials typically make ideal samples for STEM, but those within PGM-free catalysts present additional challenges since these materials are often defect-rich, with a high density of edges, dopant atoms, etc., which significantly increase susceptibility to beam damage at standard operating voltages. This makes analysis of potential FeN4 active sites particularly challenging, since a large proportion of Fe exists at edge sites where beam-induced atomic displacements can prohibit high-resolution structural characterization [6]. Conventional dark-field imaging compounds this problem by producing less signal for a given dose and being less sensitive to light elements than dose-efficient phase contrast imaging techniques such as those enabled by four-dimensional (4D)-STEM [8-10] (Fig. 1a). Consequently, active site structural analysis is often left to methods such as low-resolution imaging combined with quantum chemical calculations [6], which hinders accurate determination of reaction and degradation mechanisms.Here, we demonstrate direct atomic-scale structural mapping of FeN4 sites by performing low-voltage 4D-STEM on a model PGM-free catalyst system with many exposed monolayer regions. To accomplish this, we pair a 30 keV aberration-corrected probe with a fast pixelated detector that has optimal performance at low beam voltages [11]. This enables us to simultaneously image light and heavy elements with high signal-to-noise by center-of-mass analysis (Fig. 1c) while minimizing beam-induced atomic displacements at sensitive sites. The monolayer nature of these materials additionally allows for experimental validation by direct comparison with multislice simulations [12] of model structures (Fig. 1d-e). This work demonstrates how low-voltage 4D-STEM will provide new insights into the atomic-scale structure and degradation mechanisms of active sites in PGM-free catalysts, facilitating the development of low-cost hydrogen fuel cells and other energy conversion technologies in the future [13].

Zachman, Michael↗
DOE OSTI2023

Integrated Process Optimization for Biochemical Conversion

This research is motivated by the challenges faced during biomass processing in bioenergy plants. It has been observed that variations in biomass characteristics, such as moisture, ash, and carbohydrate contents cause variations in feeding of the system which led to underutilization of equipment and the reactor. The objective of this research is to ensure a continuous flow of biomass to the reactor in plants that use the biochemical conversion process to generate liquid fuels. The overall goal is to lower the cost of producing biofuels, which could lead to improving US’s energy independency and growing US’s rural economy. The research team developed analytical models, such as discrete element method (DEM) models and mathematical models. The DEM models are unit-level models that explicitly capture biomass characteristics and quantify the impacts of biomass characteristics on bulk material properties and the performance of specific equipment. The mathematical models are system-level models that capture the impacts of system infeed rate, equipment processing rate, storage location and capacity, and biomass characteristics on system throughput. The functional relations predicting the bulk material properties from DEM models are incorporated to the mathematical models. The models developed were validated and evaluated using data collected at Idaho National Laboratory’s biomass processing facility. Via these models, we identified process control strategies that ensure a continuous flow of biomass to the reactor, while meeting the requirements of biochemical conversion process. Our analysis indicates that sequencing of biomass bales based on moisture level, and carbohydrate contents could have a positive impact on reducing processing time and inventory level and increasing throughput rate. Short bale sequences that repeat frequently, seem to have the greatest impact on improving system’s performance. Based on our experiments, the total annual system operating costs reduced by 20-30%, and the maximum inventory level reduced by 3 to 4 times. The operating costs include the annual equipment amortization cost and processing cost. The implementation of the models developed requires the use of standardized bale format, Radio Frequency Identification technology, sensing and real time monitoring of material attributes, automated material handling equipment, and automated process control. The scope of the model proposed can be extended to include the whole supply chain. The supply chain models help identify how many bales of different biomass feedstock to purchase given biomass availability in the region, biomass price and quality, and the biomass processing capabilities of the biorefinery. Thus, the outcomes of supply chain models can be used to inform the design of long-term contracts among farmers and the biorefinery.

09 BIOMASS FUELS↗
DOE OSTI2023

Isolating the Source Region of Infrasound Travel Time Variability Using Acoustic Sensors on High-Altitude Balloons

High-altitude balloons carrying infrasound sensor payloads can be leveraged toward monitoring efforts to provide some advantages over other sensing modalities. On 10 July 2020, three sets of controlled surface explosions generated infrasound waves detected by a high-altitude floating sensor. One of the signal arrivals, detected when the balloon was in the acoustic shadow zone, could not be predicted via propagation modeling using a model atmosphere. Considering that the balloon’s horizontal motion showed direct evidence of gravity waves, we examined their role in infrasound propagation. Implementation of gravity wave perturbations to the wind field explained the signal detection and aided in correctly predicting infrasound travel times. Our results show that the impact of gravity waves is negligible below 20 km altitude; however, their effect is important above that height. The results presented here demonstrate the utility of balloon-borne acoustic sensing toward constraining the source region of variability, as well as the relevance of complexities surrounding infrasound wave propagation at short ranges for elevated sensing platforms.

47 OTHER INSTRUMENTATION↗
DOE OSTI2023

Tech-X contributions to the Center for Tokamak Transients (CTTS) Scientific Discovery through Advanced Computing (SciDAC) project (Final Report)

The Center for Tokamak Transients Simulations (CTTS) consists of computational plasma physicists, computer scientists, and applied mathematicians who are together developing and applying simulation software to improve our understanding of and ability to predict and control transient events in tokamak discharges, particularly those that lead to violent disruptions. We have major simulation activities concerning disruptions that are caused by violating ideal MHD stability criteria, vertical displacement events, resistive wall modes, and tearing modes. We are performing comprehensive modeling of disruption mitigation techniques, particularly by shattered pellet injection (SPI). The Center is built around two mature extended MHD codes: NIMROD and M3D-C1. These codes are being extended in a number of ways as required by the demands of disruption modeling, and are being optimized to work well for larger problem size and modern computer architectures as required to model tokamak plasmas with realistic parameters over the required timescales. The codes will be coupled to the FronTier code for state-of-the-art modeling of SPI. Validation with experimental data on DIII-D, NSTX, and JET is being facilitated by rigorous uncertainty quantification (UQ). The high-level objectives of the center are to: (1) Develop better understanding and improved predictive capability of when crossing a linear stability boundary will likely lead to a disruption, (2) Provide a tool for calculating vessel forces, both axisymmetric and non-axisymmetric, for worst case VDE and other disruptions in tokamaks, (3) Develop increased understanding and improved predictive capability for locked mode disruptions and how best to avoid them, (4) Develop, verify, and validate 3D models for disruption mitigation by shattered pellets that can be used to design and optimize a system for future tokamaks, and (5) Increase the efficiency and scalability of the M3D-C1 and NIMROD codes so that they can model disruptions in ITER scale devices using the most powerful DOE HPC hardware, and (6) Demonstrate code-coupling via FronTier coupling with NIMROD and M3D-C1 and by interfacing these codes within a WDM framework.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Toxicology and Biodegradability of Blendstocks for Mixing Controlled Compression Ignition Combustion: Literature Review of Available Data

The goal of the U.S. Department of Energy’s Co-Optima project is to develop an understanding of potential biomass-derived blendstocks for use in the transportation sector. This report documents the estimated environmental and toxicological impact of potential bioblendstocks for mixing controlled compression ignition (MCCI) engines. The candidate molecules were: 2-nonanol; n-undecane; 2,6,10-trimethyldodecane; 5-ethyl-4-propylnonane; hexylhexanoate; methyldecanoate; dibutoxymethane; 4-butoxyheptane; dipentylether; renewable diesel; and soy biodiesel. To provide a robust comparison, diesel surrogate molecules were also considered: α-methylnaphthalene; decahydronaphthalene; 2,2,4,4,6,8,8-heptamethylnonane; n-butylcyclohexane; n-hexadecane; tetralin; and n-dodecylbenzene. The intent of this work was not to provide absolute answers and remove any molecules from consideration, but to provide more robust data for each molecule that can be used in subsequent evaluations. Little literature data was available for many of the molecules, both potential blendstocks and diesel surrogates, so a quantitative structural activity relationship (QSAR) approach was used to provide input into the relevant models. This assessment included an evaluation of compartment partitioning in the event of an accidental chemical spill, fate and transport indicators, estimated biodegradability, and human and environmental toxicology. The higher molecular weight and longer hydrocarbon chain lengths of these MCCI molecules reduce water solubility and mobility. Generally, the oxygenated molecules improved biodegradability metrics. The hydrocarbons farnesane and 5-Et-4-PrNonane are not predicted to be biodegradable. Human acute toxicity was found to be slightly to non-toxic for the MCCI blendstocks and the diesel surrogates. 2-Nonanol, undecane, MeDecanoate, dibutoxymethane, and dipentylether were likely developmental toxicants. Although poorly water soluble, all the compounds, whether Co-Optima blendstocks or diesel surrogates, showed significant toxicity to D. magna. The oxygenated blendstocks were less likely to bioaccumulate than the non-oxygenated blendstocks, which behaved similar to the diesel surrogates. Overall, this predicted analysis showed that the MCCI blendstocks are very similar to their diesel counterparts. No significant showstoppers were found that would indicate concern moving forward with additional research into these potential fuels. The results presented were based solely on predictive models and serve to provide information going forward for continued evaluation of these molecules.

09 BIOMASS FUELS↗
DOE OSTI2022

Effect of injected flux and current temporal phasing on self-organization in the HIT-SI3 experiment

The HIT-SI3 device at the University of Washington uses three oscillating inductive helicity injectors to form and sustain spheromak plasma equilibria. By adjusting the temporal phase of the injector waveforms with respect to each other, the toroidal spectrum of the imposed perturbations can be controlled. Using a recently implemented GPU-based control system, the available mode spectra were explored experimentally by scanning the space of relative injector phasing. In this space, significant variation in the toroidal mode spectrum ($n$ = 1, 2, 3) of the perturbations was observed. Additionally, variation in characteristics of driven equilibria was also observed, including a ≈ $30$% range in toroidal current gain ($I$ $Φ$ / $I$ $Inj$ ). Experimental results are compared with both a composite-equilibria and nonlinear dynamic model, including extended MHD simulations using the NIMROD code and composite Taylor state equilibria computed using the PSI-Tet code. In conclusion, qualitative agreement is seen with the nonlinear models, but not with composite-equilibria models, suggesting the use of nonlinear models to better capture observed plasma dynamics and provide predictive use for future experiments.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2022

Evaluating performance of different generative adversarial networks for large-scale building power demand prediction

We report as an unsupervised-learning data-driven model, Generative Adversarial Networks (GANs) have recently attracted a lot of attention for various applications. There is potential to apply GANs for large-scale building power demand prediction, which is needed for power grid operation. However, there are many GAN variations and it is unclear which GAN is suitable for this application. To answer this question, this paper identifies five promising GANs (Original GAN, cGAN, SGAN, InfoGAN, and ACGAN) and evaluates their performance for predicting building power demand at a large scale. Physics-based building energy models are developed to generate training and reference data. A new evaluation indicator that combines accuracy and reproducibility is proposed to evaluate the performance of different GANs in predicting building power demand. The results show that SGAN and InfoGAN are not suitable because they cannot control the number of generated building samples for different building types. The prediction performance among the Original GAN, cGAN, and ACGAN can vary depending on training sample sizes and number of building types. If the training sample size is sufficiently large, Original GAN and cGAN can predict building power demand more accurately than ACGAN with the same number of samples. If training samples are limited, Original GAN provides better accuracy than cGAN and ACGAN. When the number of building types increase, the prediction accuracy increases for cGAN, decreases for ACGAN, and remains the same for Original GAN. As a result, cGAN and Original GAN are recommended for large-scale building power demand prediction.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2020

Using Machine Learning Tools to Predict Compressor Stall

Clean energy has become an increasingly important consideration in today’s power systems. As the push for clean energy continues, many coal-fired power plants are being decommissioned in favor of renewable power sources such as wind and solar. However, the intermittent nature of renewables means that dynamic load following traditional power systems is crucial to grid stability. With high flexibility and fast response at a wide range of operating conditions, gas turbine systems are poised to become the main load following component in the power grid. Yet, rapid changes in load can lead to fluid flow instabilities in gas turbine power systems. These instabilities often lead to compressor surge and stall, which are some of the most critical problems facing the safe and efficient operation of compressors in turbomachinery today. Although the topic of compressor surge and stall has been extensively researched, no methods for early prediction have been proven effective. This study explores the utilization of machine learning tools to predict compressor stall. The long short-term memory (LSTM) model, a form of recurrent neural network (RNN), was trained using real compressor stall datasets from a 100 kW recuperated gas turbine power system designed for hybrid configuration. Two variations of the LSTM model, classification and regression, were tested to determine optimal performance. The regression scheme was determined to be the most accurate approach, and a tool for predicting compressor stall was developed using this configuration. Overall, results show that the tool is capable of predicting stalls 5–20 ms before they occur. With a high-speed controller capable of 5 ms time-steps, mitigating action could be taken to prevent compressor stall before it occurs.

42 ENGINEERING↗
DOE OSTI2021

Thermodynamic modeling of hydrogen fueling process from high-pressure storage tank to vehicle tank

This study develops a hydrogen fueling station (HFS) thermodynamic model that simulates the actual fueling process in which hydrogen is supplied from a high-pressure (HP) storage tank into a fuel cell electric vehicle (FCEV) tank. To make the model as accurate as possible, we use the same components and specifications as in actual HFSs, such as a pressure control valve, a pre-cooling system, and an FCEV tank. After the components and their specifications are set, pressure and temperature profiles are set as the HP tank supply conditions. Based on the pressure and temperature profiles, the model solves for the temperature, pressure, and mass flow rate of hydrogen at each downstream position, including the inside of the vehicle tank. The values predicted by the model are compared with experimental data, and we show that the developed model makes it possible to accurately simulate those values at any position during the fueling process.

08 HYDROGEN↗
DOE OSTI2023

Evaluation of Digital Twin Modeling and Simulation

A digital twin has intelligent modules that continuously monitor the condition of the individual components and the whole of a system. Digital twins can provide nuclear power plants (NPP) operators an unprecedented level of monitoring, control, supervision, and security by contributing a greater volume of data for more comprehensive data analysis and increased accuracy of insights and predictions for decision making throughout the entire NPP lifecycle. NPP operators and managers have historically relied on limited, second hand or incomplete data. With proper implementation, digital twins can provide a central hub of all intel that allows for a multidisciplinary view of an NPP. This equips operators and managers with the ability to have more information, context, and intel that can be used for greater granularity during planning and decision making. Digital twins can be used in many activities as the technology has many different concepts surrounding it. From the various definitions of a digital twin within the industry, digital twins can be differentiated by levels of integration/automation. The three main models include digital model, digital shadow, and digital twin. Digital twins offer many potential advancements to the nuclear industry that could reduce costs, improve designs, provide safer operation, and improve their overall security.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2024

Equivariant graph convolutional neural networks for the representation of homogenized anisotropic microstructural mechanical response

Composite materials with different microstructural material symmetries are common in engineering applications where grain structure, alloying and particle/fiber packing are optimized via controlled manufacturing. In fact these microstructural tunings can be done throughout a part to achieve functional gradation and optimization at a structural level. To predict the performance of particular microstructural configuration and thereby overall performance, constitutive models of materials with microstructure are needed. In this work we provide neural network architectures that provide effective homogenization models of materials with anisotropic components. These models satisfy equivariance and material symmetry principles inherently through a combination of equivariant and tensor basis operations. We demonstrate them on datasets of stochastic volume elements with different textures and phases where the material undergoes elastic and plastic deformation, and show that the these network architectures provide significant performance improvements.

anisotropy↗
DOE OSTI2021

Autoignition study of iso -cetane/tetralin blends at low temperature

iso-Cetane and tetralin are the two fuel components commonly considered in literature diesel surrogates, and are hydrocarbon classes representative of iso-alkanes and naphthoaromatics, respectively. Since both surrogate components are involved in the key reactions controlling the autoignition reactivity of diesel surrogates, autoignition studies of iso-cetane, tetralin, and their blending behavior at low temperatures can be helpful in developing/refining chemical kinetic models of diesel surrogates, in order to better predict diesel ignition response under low temperature combustion conditions. In this investigation, autoignition experiments of iso-cetane, tetralin, and their binary mixtures at varying blending ratios have been conducted in a rapid compression machine (RCM) in the temperature/pressure range of 630–930 K and 10–20 bar. For the neat fuel components, the newly-acquired RCM data have been compared with the literature experimental results at some overlapping conditions. In addition, a chemical kinetic model describing the oxidation of iso-cetane and tetralin compiled/updated from the diesel surrogate model developed by Lawrence Livermore National Laboratory is used to simulate the experimental datasets. It is shown that this model exhibits good agreement with the experimental results of tetralin and also predicts the total ignition delay times of iso-cetane at lower temperatures reasonably well. However, the current model fails to fully capture the increase of total ignition delay time with increasing amount of tetralin in the binary blends observed in the present RCM experiments. Therefore, chemical kinetic analyses of the current model are conducted to help identify possible reasons leading to the discrepancies between simulations and experiments. Furthermore, the chemical kinetic interactions between iso-cetane and tetralin are explored by including additional cross-reactions among iso-cetane, tetralin, and their fuel fragments in the current model to assess their effects on model predictions. As a result, based on the chemical kinetic analysis results, future direction for model refinements is discussed.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2020

Point defect evolution under irradiation: Finite size effects and spatio-temporal correlations

The evolution of point defect concentrations under irradiation is controlled by their diffusion properties, and by their formation and elimination mechanisms. The latter includes the mutual recombination of vacancies and interstitials, and the elimination of point defects at sinks. Two models are traditionally used to predict the defect concentration evolution, the standard rate theory (SRT) and the atomistic kinetic Monte Carlo (AKMC). We show in this work a large discrepancy in the defect concentrations between both methods when the average number of defects in the AKMC simulation box is close or lower than one. The reason is that AKMC naturally captures strong space and time correlations between vacancies and interstitials generated by the finite size of the periodic simulation box. These correlations strongly affect the recombination rate and the point defect concentrations. SRT fails to predict such correlations, and the corresponding solution deviates from the more accurate solution given by the AKMC simulation under similar conditions. These finite size correlation effects are strong when the elimination of point defects occurs by recombination only, but can still be significant in the presence of sinks. In order to account for the spacio-temporal correlation in a continuum framework, we introduce a Correlated Pair Theory (CPT). This theory fully takes into account the correlations between vacancy and interstitial pairs and predicts point defect concentrations in good agreement with AKMC simulations. Inversely, if the goal is for the AKMC to reproduce defect concentrations in bulk using small simulation boxes, the naturally occurring correlations need to be corrected. We show here that the CPT can be used to modify the elimination rates in the AKMC simulations, so as to yield point defect concentrations in agreement with SRT.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗